crystalline soft-magnetic alloy

4 downloads 241 Views 259KB Size Report
D. R. dos Santos a,b. , I. L. Torriani b*. , F. C. S. Silva b and M. Knobelb. aLaboratório Nacional de Luz Síncrotron - LNLS, Brazil, bInstituto de. Fisica, Unicamp ...
conference papers In-situ small-angle scattering study In-situ small angle scattering study on the the formation of a nanocrystalline on formation of a nano- softmagnetic alloy crystalline soft-magnetic alloy D. R. dos Santos Knobel

a,b

b*

conventional furnace isothermal treatments. The ability to give important information such as the size of the scattering objects on a nanometer scale, irrespective of their internal structure, makes this technique very appropriate for the study of the first stages of crystallization. In conventional annealing the heating rate of the samples is sufficiently low for time resolved scattering experiments using synchrotron radiation.

b

, I. L. Torriani , F. C. S. Silva and M.

b

a

2. Experimental procedures b

Laboratório Nacional de Luz Síncrotron - LNLS, Brazil, Instituto de Fisica, Unicamp, Brazil. Email:[email protected]

Abstract A detailed study is presented on the nanocrystallization of the amorphous alloy Fe86Zr7Cu 1B6 (indices indicate at. %). Melt-spun ribbons were rapidly annealed by Joule heating, and the electrical resistance showed strong variations during thermal treatment. X-ray diffraction patterns indicate that these variations are related to the nucleation and growth of α-Fe nanocrystals, and from peak profile analysis we obtained the average grain size and crystalline volume fraction for different annealing currents. The disorder-order transition was studied by in-situ small-angle X-ray scattering during conventional furnace treatments. SAXS intensity evolution for different temperatures, both below and above the crystallization temperature of the alloy, showed that a fast atomic rearrangement leads to the formation of atomic clusters before crystallization. The evolution of the size distribution function of these clusters as a function of time and temperature was obtained assuming a polydisperse system of spherical particles.

1. Introduction Studies on the production and properties of soft magnetic materials have become widespread since 1988 when the first reports on nanocrystalline iron based alloys appeared, e.g. on Fe-Si-B with small additions of Cu and Nb, which after thermal treatment (1 h at 813 K) presents a fine metallic grain structure and a remarkably soft magnetic behaviour (Yoshizawa et al., 1988). More recent studies on alloys with different atomic compositions show similar or even better magnetic properties, e.g. Fe-Zr-B (Suzuki et al., 1991; Kim et al., 1995). For large scale production, Joule heating of amorphous metallic ribbons is faster and less expensive than the conventional isothermal treatment. Joule heating is highly reproducible and results in samples with somewhat better mechanical properties, owing to a faster heat transfer to the sample that partially suppresses stress relaxation before crystallisation (Allia et al., 1994). In the present work, amorphous samples of Fe-Zr-Cu-B produced by the melt-spinning technique were submitted to DC Joule heating applying different annealing currents. Structural transformations were monitored via electric resistance changes. X-ray diffraction (XRD) was used to analyze the structural evolution of the samples, revealing the nucleation and growth of α-Fe nanocrystals. Important structural parameters like average grain size, crystalline volume fraction and lattice parameter were obtained from XRD peak profile analysis. Furthermore, the disorder-order transition was studied by small angle X-ray scattering (SAXS), measured in-situ during

J. Appl. Cryst. (2000). 33, 473±477

The amorphous alloy of nominal composition Fe86Zr7Cu 1B6 (at. %) was rapidly quenched from the melt by planar flow on a fast rotating cylinder under a controlled Ar atmosphere, producing high quality -3 metallic ribbons of average width (2.7 ± 0.1) × 10 m and thickness -5 (2.5 ± 0.1) × 10 m. The average values of the ribbon cross section σ -8 2 = (6.7 ± 0.3) × 10 m and of the initial resistance at room temperature Raq = (2.62 ± 0.05) Ω were determined for the set of samples that were to be submitted to DC Joule. This treatment was done in vacuum, with an effective length of 0.1 m between electrical contacts. A constant electrical current was applied for a fixed time of 50 s, for current intensities ranging from 0.5 A to 2.0 A to obtain a series of samples with different stages of crystallization. During Joule treatments, the resistance variations were determined using two couples of electrical contacts to measure simultaneously the voltage across the sample and the electrical current flowing through it. XRD measurements were performed using Cu Kα radiation from a Rigaku RU200 rotating anode generator, equipped with a LiF curved crystal diffracted beam monochromator, which eliminated the fluorescence from Fe in the samples. Diffraction from a standard Al2O3 powder with controlled grain size (10-6 m) was measured in the same experimental conditions, for correction of the instrumental broadening affecting the peak profile of the nanocrystalline samples. SAXS experiments were performed using synchrotron radiation, at the SAXS beam-line of the LNLS - National Synchrotron Light Laboratory, Brazil (Kellermann et al., 1997). Scattering intensities were measured in transmission geometry, with sample-to-detector distance 1.870 m and incident wavelength λ = 1.760 Å. The corresponding incident energy is 50 eV below the K absorption edge of Fe atoms, and was chosen in order to avoid fluorescent radiation and reduce the attenuation factor of the samples. The dimensions of the X-ray spot over the sample were approximately 1 mm × 5 mm. Due to the small size of the incident-beam cross section at the detection plane, no mathematical desmearing of the experimental curves was necessary. Two scintillation detectors were used to determine the attenuation factor of the samples. The intensities were corrected for incident beam intensity variation during acquisition and linear position-sensitive gas detector homogeneity. Several parasitic scattering curves were registered for the appropriate corrections. SAXS curves of the amorphous samples were measured at room temperature during t = 300 s. The scattering evolution during conventional furnace treatments at 633, 723 and 813 K was measured in-situ. The amorphous sample was rapidly inserted in a furnace of high heat capacity with pre-stabilized temperature, and the acquisition of the scattering curves was immediately initiated. Individual SAXS curves were registered during short times (t = 20 s, 40 s or 60 s), and their evolution was measured during about 30 min of thermal treatment. XRD patterns of the furnace annealed samples were measured in the conditions described above.

# 2000 International Union of Crystallography



Printed in Great Britain ± all rights reserved

473

conference papers 3.

110 reflection grows superimposed to the amorphous band, which remains nearly unchanged.

Results and discussions

3.1 Resistance variation during Joule heating and XRD analysis Fe 110

The variations of the electric resistance relative to the initial room temperature values are shown in Fig. 1 as a function of time, for a set of annealing currents. 8 6

2

0.90 1.00 0.60 0.50

0

1.10

-2

2.00

R / Raq (%)

4

2.00 1.50 1.40 1.30 1.20 1.10 1.00 0.60 as-quenched

-4 -6

40

-8

1.50

-10

1.40 1.20 1.30

-12 -14 0

10

20

30

40

42

44

46

48

50

2 (degrees)

Fig. 2 Diffraction patterns for the samples Joule heated with different annealing currents.

50

time (s)

Fig. 1 Resistance variation measured on-line during DC Joule thermal treatments. The variations are relative to the initial room temperature value, Raq. Electrical currents of annealing varied in the range 0.50 A ≤ I ≤ 2.00 A.

A theoretical model that describes the temperature variation in this kind of treatment has been successfully applied to estimate the crystallization temperature of other nanocrystalline alloys like Fe40Ni40B20 (Allia et al.,1993) or Fe73.5Cu1Nb3Si13.5B9 (Allia et al.,1994). In the case of our alloy, Fig. 1 shows that for treatments with currents of up to 0.60 A the samples reach a steady-state temperature after approximately 15 s of heating, and this temperature remains approximately constant indicating that crystallization does not occur. Increasing the electrical current between 0.90 and 1.00 A a broad bump appears. This maximum moves towards shorter times when the current is increased, indicating an enhancement in the heating rate of the sample and the start of thermally activated microstructural changes. For currents equal to 1.10 A or larger the structural transformations are started before a steady-state temperature can be attained, and the main transformations occur after approximately 15 s of treatment with 1.20 A. It is worth noting that this sample, which is in the lower limit of annealing currents suited to cause a drastic reduction in the resistance, is the one that displays the best soft magnetic properties. Also, although the shapes of the curves are quite different from the ones observed in FINEMET alloys (Allia et al., 1993) they are consistent with the theoretical description and with reported experimental results (see e.g. Holzer et al., 1997). The analysis of the XRD patterns (shown in Fig. 2) furnishes the structural evolution as a function of annealing current, allowing a quantitative association between the microstructure and the resistance behaviour. The as-quenched samples presented typical amorphous spectra, that can be fitted by a gaussian band with FWHM ≅ 6°. Very similar spectra were observed for samples treated with currents up to 0.90 A. The first traces of crystallization appear for currents in the range 1.00 A ≤ I ≤ 1.10 A, where the crystalline

474

D. R. dos Santos et al.

A small increase in the current of annealing to I = 1.20 A leads to a drastic decrease in the amorphous intensity. The amorphous halo also becomes broader and is displaced towards lower angles (from 43.9° to 42.2°), indicating an increase in the average interatomic distance. This may be due to a Zr enrichment of the residual amorphous component. These facts suggest that 1.20 A is a threshold current for the massive precipitation of α-Fe particles. This can be explained assuming an atomic diffusion between the crystalline and amorphous phases which would cause the large Zr atoms to migrate from the Ferich regions to the interface region, thus hindering the grain growth and enhancing the thermal stability of the amorphous phase. From 1.20 to 1.50 A the XRD results are nearly the same, but a further increase in the annealing current to 2.00 A leads to the formation of other crystalline phase in the residual amorphous component: zirconium dioxide. Assuming pseudo-Voigt line profiles, the 110 reflection was decomposed into crystalline and amorphous contributions, and the crystalline volume fraction was determined from the integrated intensity of the crystalline component relative to the total intensity. The line broadening of the crystalline component was then used to determine the average grain size of the α-Fe particles using the well known Scherrer formula (Klug & Alexander, 1974): = 0.94λ / β cosθ

(1)

where is the average grain size in the [110] direction, λ is the wavelength of the incident radiation, θ is the Bragg angle and β is the full width at half maximum (FWHM) of the 111 peak of the crystalline component, corrected with respect to the instrumental broadening, and measured in radians. The average grain sizes determined by this method have an associated error band of about 10%. A narrower error band (6%) was estimated for the crystalline volumetric fraction, since peak area is less affected by the fitting procedure than peak width. Similarly, it was estimated that the uncertainty in the determination of peak positions will cause an error of 4% in the lattice parameter calculated. The results are presented in Table 1.

J. Appl. Cryst. (2000). 33, 473±477

conference papers Table 1 – Crystalline volume fraction Vcr, lattice parameter a0 and average grain size for different annealing currents. 4000

I (A)

Vcr (%)

a0 (Å)

(Å)

1.00

8.0 (2)

2.876 (6)

90 (5)

3000

1.10

16.0 (5)

2.875 (6)

110 (6)

2000

1.20

54 (2)

2.878 (6)

90 (5)

1.30

63 (2)

2.880 (6)

100 (5)

1.40

59 (2)

2.876 (6)

100 (5)

0 1500

813 K

as quench 1 min 2 min 3 min 4 min 5 min 10 min 20 min 40 min

65 (2)

2.876 (6)

100 (5)

1.75

90 (3)

2.871 (6)

170 (9)

2.00

≅ 100 (3)

2.872 (6)

320 (16)

In the earlier stages of crystallization (1.00 ≤ I ≤ 1.10 A) the crystalline volume fraction increases from 8% to 16%, and the grains have average sizes of about 100 Å. For I = 1.20 A the crystalline volume fraction reaches 54 % in a rapid process, as indicated by the resistance stabilization after 15 s (see Fig. 1). This is the threshold current to cause a fast atomic rearrangement between amorphous and crystalline phases, leading to the massive crystallization of α-Fe particles in the samples. The large crystalline fraction in the bulk material is the cause of the drastic reduction in the on-line resistance measurements (-12 %). Treatments with I = 1.30 to 1.50 A resulted approximately in the same microstructure, with slightly larger grain sizes and crystalline fractions when compared to the optimum annealing (I = 1.20 A). After treatments with 1.75 A a significant increase in the grain size from 100 to 170 Å is observed, and the crystalline volume fraction reaches 90%. The grains grow to about 320 Å after treatments with 2.00 A, probably due to coalescence after the formation of ZrO2 in the residual amorphous component. Even in the final stages of crystallization the lattice parameter in the samples is larger than the value for pure crystalline iron (a0 = 2.866 Å). This result could indicate that B and/or Zr atoms are randomly arranged in the nanocrystals. This quantitative information on the structural parameters and the correlation established with the changes observed in the room temperature resistance, reinforce the use of the ribbons resistance as a reliable guide to the microstructure developed upon heating.

723 K 1000 500 0 1960

633 K 980

0 0,010

0,015

-1

q (Å )

20 min

10000

q

1000 100

-4

813 K

10 1 0,1 1000

Intensity (a.u.)

1.50

Intensity (a.u.)

1000

723 K

100 10 1 0,1

3.2 Furnace annealing and SAXS analysis The previous results properly characterize the structural properties of a set of partially crystallized samples, and Joule heating proved to be a fast and reproducible technique for controlled production of nanocrystalline metallic alloys. In conventional furnace treatment, however, the crystallization process is significantly slowed down, and this feature turns out to be appropriate for the study of the very initial steps of microstructural transformation. To study this disorder-order transformation, SAXS intensity curves were measured in-situ during isothermal treatments at 633, 723 K (bellow the crystallization temperature of the alloy, Tx) and 813 K (above Tx). The curves were obtained with acquisition times as short as 20 s, and their evolution was followed during about 30 minutes of thermal treatment. The range of scattering vectors covered in the -1 -1 experiments was between qmin = 0.005 Å and qmax = 0.160 Å (where q = 4π sinθ /λ, θ being half the scattering angle and λ the wavelength of the incident radiation). The in-situ measurements were performed

J. Appl. Cryst. (2000). 33, 473±477

100

633 K 10 1 0,1 0,01 0,01

-1

q (Å )

0,1

Fig. 3 Top: Time evolution of the scattered intensity for small values of q measured in-situ during conventional isothermal annealing at 813, 723 and 633 K. The scattering from the amorphous sample (as quenched) measured at room temperature is shown for comparison. Bottom: Scattering intensities for the samples with 20 minutes of annealing in the complete q range. We observe that the theoretical curves (solid lines) calculated for a polydisperse system of spherical particles has a fairly good fit, except for deviations for values of q