Structural study of Fe40Ni40B20 amorphous alloy

Physica B 156 & 157 (1989) North-Holland, Amsterdam

STRUCTURAL

220-222

STUDY OF Fe,,Ni,,B,,

AMORPHOUS

R. CACIUFFO’, M. STEFANON’, W.S. HOWELLS”, S. MELONE’ and F. RUSTICHELLIS

ALLOY A.K.

‘Dipartimento di Scienze dei Materiali e della Terra, Universitti di Ancona. ‘ENEA, C. R. E. “E. Clementel”, Bologna, Italy ‘Neutron Division, Rutherford Appleton Laboratory, UK 41stituto Elettrotecnico Nazionale Galileo Ferraris, Torino. Italy ‘Istituto di Fisica Medica, Universitri di Ancona, Ancona. Italy

SOPER’,

P. ALLIAJ,

Via Brecce Bianche.

F. VINAIJ,

I-60131 Ancona,

Italy

We present the results of a neutron diffraction study of the Fe,,Ni,,,Bz, amorphous alloy. From the split first peak of the pair correlation function, the parameters of the TM-TM and TM-B (TM = Transition Metal) distributions are derived. Small changes of the chemical short-range order parameter are induced by the variation of the rate of quenching.

1. Introduction Among the ternary transition metal-metalloid glasses, the (Fe,,,,Ni,,,),,,,,.,B, alloy is one of the most studied, because of its interest for the technology of electrical devices [l]. Both its macroscopic properties and its atomic-scale structure are, as a consequence, fairly well known [2-61. In the as quenched state the (Fe,,,Ni,,,),,,,,.,B,, like every other amorphous alloy prepared by rapid melt quenching, is thermodynamically unstable. Structural relaxation accompanied by strong modification of several physical properties may then be obtained by isothermal annealing [7,8]. On the other hand, similar effects could be produced by reducing the rate of the melt quenching process. It is known, in fact, that certain macroscopic physical properties of amorphous metallic alloys containing transition metals are variously modified by the quenching rate from the melt. The effect is most evident in the case of the aftereffect, or relaxation, of the magnetic premeability [9], which is observed to substantially increase with increasing quenching rate. The differential scanning calorimetry signal and the electrical resistivity are significant, if less strongly influenced by the rate of quenching [lo]. These observations suggest the possibility of inducing modifications of the atomic scale structure by varying the quench rate. The high resolution neutron diffraction study presented in this 0921-4526/89/$03.50 0 (North-Holland Physics

Elsevier Science Publishers Publishing Division)

paper was undertaken in order to verify this possibility by carefully determining and comparing the features of the radial distribution function of Fe,,Ni,,,B,, samples prepared with different hyperquench velocities. 2. Experimental

results and discussion

The investigated samples of the Fe,,Ni,,,B,,, amorphous alloy were prepared by melt spinning at 32 m/s, 36 m/s and 40 m/s, respectively. The approximately 20 km thick and 2 mm wide ribbons were wound on a Al frame to give a planar sample of about 40 x 40 x 0.6 mm’ dimensions. Natural boron was used in spite of its high absorption cross section. The experiment was carried out on the LAD time of flight diffractometer at the U.K. spallation neutron source ISIS of the Rutherford Appleton Laboratory. Each sample was run for about 600 FA hrs, together with a vanadium standard and background, and the structure factors S(Q) were measured over a range of scattering vectors from 1 A -’ up to 40 A-‘. The transmission decreased to 30%-40% over the wavelength range 0.1 to 7 A, due to the absorption from the boron. The structure factors obtained from the measured data, after absorption and multiple scattering corrections, are shown in fig. 1, for the three samples. It is interesting to notice B.V.

oscillations

extending

out to about 25 A- ‘.

R. Caciuffo et al. I Fe,,Ni,,B,,

3.0 (a)

20

10

0.01: 30 (b)

20 E ul 1.0

)q 00

30IC)

2.0 -

lo-

0.0 0

I 5

1

" 10

MOMENTUM

/ 15

I

I 20

TRANSFER

25 Cd’,-

Fig. 1. Structure factors S(Q) vs momentum transfer Q obtained for Fe,Ni,,B,, samples prepared with a quench rate of (a) 32m/s, (b) 36m/s and (c) 40m/s.

amorphous

alloy

221

sity. The g(r) distributions, for the three samples investigated in our experiment, are shown in fig. 2. The high r-space resolution given by the large Q-range explored (truncation Ar = 21r/Q,,, = 0.16 A) enables the observation of a 0.43 8, splitting of the first peak, corresponding to the nearest neighbour distances. It is evident, from eq. (l), that the TM-TM correlation function is the dominant term in g(r). Thus, the first sub-peak at 2.118, may be attributed to the TM-B first neighbour pair and the second at 2.54 8, may be related to the TM-TM distribution. The second peak of g(r) represents a distance at about 1.63 times the nearest neighbours distance, a typical value for metal-metalloid amorphous alloys. Some of the parameters characterizing the g(r) are given in table I. They compare favorably with the previous determination on Fe,,Ni,,B,,, but very small differences, if any at all, are to be noticed for the different samples. The slope (l/ rl) (dr,ldj) of the straight line obtained by plotting the relative peak position rjlrl of the coordination shells vs their number j, is the same in the three cases. This fact suggests that the packing

The atomic scale structure may be described by the atomic distribution function G(r), which is a linear combination of several terms, resulting from the partial atom pairs. A previous experiment [5] has shown that the distribution of iron and nickel atoms on the transition metal (TM) sites in the Fe,oNi,,,B,, is almost random. An “average” TM atom can then be considered, leading to G(r) = 0.776G,,_,,(r)

‘~“tl

+ 0.209G,,_,(r)

+ O.O14G,_,(r).

II

(b)

1

(1)

The G(r) function is obtained by a Fourier transformation of the measured structure factor, QM.W

G(r) = (2/n)

1 0

Q (S(Q) - 1) sin( Qr) dQ

. (2)

Then, the reduced pair correlation function g(r) = 1 + G(r)/4mpo may be calculated. p,g(r) being the local atom density and p. = 94.5 nme3 being the average number of atom per unit volume, calculated from the measured mass den-

Fig. 2. Reduced pair correlation functions for Fe,,Ni,,B,, metallic glass prepared at (a) 32m/s, (b) 36m/s and (c) 4Om/s.

R. Caciuffo el al. I Fe,,,Ni,,,B?,,

222

Table I Selected parameters characterizing the atomic velocities of the rotating drum. The estimated r2

32 m/s 36 m/s 40mis

scale structure of Fe,,,Ni,,,Bz,, metallic errors are quoted in parenthesis.

1 dr,

(i)

(A)

rl

di

2.11 (2) 2.11 (2) 2.11 (2)

2.54 (2) 2.54 (2) 2.54 (2)

0.78 0.78 0.78

Q,,, g(r,)

R(r?)

1.08 (3) l.lO(3) 1.05 (3)

2.74 (8) 2.69 (8) 2.65 (8)

efficiency in the glass is not influenced by the quench rate. The widths Ag of the nearest neighbour TM-B and TM-TM distributions were calculated from the measured ones according to a Gaussian approximation and considering a truncation broadening of 3.8/Q,,, [ll]. A sharper TM-B distribution is obtained for all the samples. The first neighbour partial coordination numbers Z were calculated from the area under the resolved sub-peaks of the radial distribution function RDF (I) = 4Tr*p,g(r), using the weighting factors in eq. (1). The variations of these parameters are inside the error band, but they seem to indicate that small changes of the chemical short-range ordering (CSRO), i.e. of the local surrounding of a given atom, could occur. The CSRO may be quantitatively described by the parameter 71 defined by Cargill and Spaepen [12]. The values of q reported in fig. 3 have been calculated supposing Z,., = 0. This is justified not only by the small weighting factors of the partial B-B pairs in the total distribution function, but also by the previous observation of a

o’2r--i 32

36

LO

amorphous

Y cm/s,-+

Fig. 3. Chemical short-range order parameter of the transition metal-boron distribution, 7rM_B, measured in Fe,,,Ni,,,B,,, samples prepared by melt quenching at different quench rates.

I,

(A) 0.27 0.27 0.25

h&h,

alloy

glass prepared

by melt spinning

at different

Ih,

(A) 0.38 0.38 0.38

z,,

I$

1.8(2) 1.X(2) 1 .h (2)

&, It,,

z,,,

Ih,

1.2 (5) 7.3 (5) 6.5 (5)

9.4 (9) 9.2 (Y) Y.1 (9)

IllhI II 0.16(2) 0.15 (2) 0.13 (2)

negligible B-B correlation within the first coordination shell of the Ni,, B ,4 metallic glass [ 131. As shown in fig. 3. the CSRO parameter appears to decrease as the quench rate increases. Actually, the observed variation is comparable with the estimated errors. As a consequence, this result should only be considered as a possible indication for an increase of the preference for unlike neighbours when the quench rate is slowed down. A check of this fact could be achieved by extending the investigation to samples having considerably larger differences in the hyperquench velocities. References [l] P. Fournier,

in: Les Amorphes Mktalliques, Ecole d’Hiver d’Aussois. 13-22 Janvier 1983, (Editions de Physique, Paris, 1984), pp. 595-615. [2] J. Wong, Topics in Applied Physics. vol. 46 (Springer. Berlin, 1981). pp. 45-77; and refs. quoted therein. [3] S. Steeb and P. Lamparter, J. Non-Cryst. Solids 61-62 (1984) 237; and refs. quoted therein. Rapidly Quen141 B.J. Thijsse and I. Majewska-Glabus, ched Metals, S. Steeb and H. Warhmont, eds. (NorthHolland, Amsterdam. 1985). p. 435. [51 J. Sietsma. C. van Dijk. ibid., p. 436. [61 E. Svib, R. Bellisant and Gy. M&szBros, ibid.. p. 467. W. Chambron and J. Hillairet, in: Les [71 A. Chamberod, Amorphes M&talliques, Ecole d’Hiver d’Aussois. 13-22 Janvier 1983 (Editions de Physique. Paris. lY84), pp. 329-401. Solids 83 (1986) 134. 181 A. van der Beukel. J. Non-Cryst. PI P. Allia and F. Vinai, Phys. Rev. B 26 (1982) 6141. F. Vinai and G. Riontino, [l~)l P. Allia. R. Sate-Turtelli, Solid State Commun. 43 (1982) 821. [Ill E. SvBb, N. Krob, S.N. Ishmaev. I.P. Sadikov and A.A. Chernyshow. Solid State Commun. 46 (1983) 351. [I21 G.S. Cargill III and F. Spaepen. J. Non-Cryst. Solids 43 (1981) 91. W. Sperl, S. Steeb and J. Bletry. 2. 1131 P. Lamparter, Naturforsch. 37a (1082) 1223.