Influence of preannealing on crystallization kinetics of some metallic glasses

Influence of preannealing on crystallization kinetics of some metallic glasses

Journal of Non-Crystalline Solids 44 (1981) 287-295 North-Holland Publishing Company 287 INFLUENCE OF PREANNEALING ON CRYSTALLIZATION KINETICS OF SO...

410KB Sizes 0 Downloads 70 Views

Journal of Non-Crystalline Solids 44 (1981) 287-295 North-Holland Publishing Company

287

INFLUENCE OF PREANNEALING ON CRYSTALLIZATION KINETICS OF SOME METALLIC GLASSES A. LUCCI, L. BATTEZZATI, C. ANTONIONE and G. RIONTINO [stituto di Chimica Generale ed lnorganica, Facolt~ di Farmacia, Universit~ di Torino, Via Ptetro Giuria, 9-10125 Torino, Italy

Received 1 August 1980 Revised manuscript received 11 February 1981

DSC measurements and X-ray analysis have been carried out on Fe4oNi38Mo4B18 (Metglas ® 2826 MB) and Fe29Ni49P14B6Si2 (Metglas ® 2826 B) glassy alloys after quenching at increasing temperatures in correspondence with significant points of the thermogram. The activation energy of crystallization and crystalline phases present after such annealings have been determined to reveal the effect of the latter on crystallization. A possible insight in the dimensional nature of the growth process has been looked for through the determination under nonisothermal conditions of the n exponent of the Avrami rate equation.

1. Introduction Two main objects may be assigned to the study of the transition from the amorphous to the crystalline phase in metallic glasses: first the evaluation of the temperature field within which the glassy phase is stable and then the determination o f the crystalline structure produced, together with the relative kinetic trend o f its formation. It has been previously shown [1] that a partial crystallization due to a prior annealing of the as-quenched material is able to condition the transformation during a successive heating, as proved by the features o f DSC thermograms. Such an observation led one to surmise that the activation energy of crystallization cannot be defined univocaUy, as is generally done. An amorphous alloy produced by Allied Chemical Corporation, displaying a two-step crystallization [2], namely Metglas 2826 MB (Fe4oNiaaMo4B~8), has been examined to verify this hypothesis and to establish the nature of the crystalline phases after heat treatments. The effect of partial transformations of the amorphous phase on the subsequent crystallization kinetics has been also tested on another glassy alloy o f the same firm (Metglas 2826 B: Fe29Ni49P14B6Si2) which exhibits a single-step crystallization. The ability of a non-isothermal method [3,4] to describe the crystallization process in terms of the Avrami kinetic law has been checked on both alloys. 0022-3093/81/0000-0000/$02.50 © North-Holland

288

A. Lucci et al. / Crystallization kinetics of metallic glasses

2. Experimental The as-received metallic glasses were in the form of ribbons ~0.03 mm (2826 MB) and ~0.06 mm (2826 B) in thickness. Heat treatments at a constant rate (20°C/min) under a protective argon atmosphere (4 ~/h flow rate) were carried out on specimens of the two amorphous alloys in a DSC apparatus. Annealings were stopped by quenching (>160°C/min cooling rate) at different temperatures through a quick admission of liquid nitrogen into the DSC cell. In this way the structural condition of various specimens was frozen at successive steps of the crystallization process recorded by the thermograms. Quenched samples were then submitted to a DSC analysis in the same 910 DSC cell connected to a Du Pont 990 Thermal Analyzer. Five heating rates were adopted (2, 5, 10, 20, and 50°C/min) on samples weighing between 0.7 and 3.5 mg. The temperature scale of the instrument has been calibrated for the different heating rates on the basis of the onset temperatures of the endothermal peaks recorded during melting of a few mg of In, Zn and A1. The crystallization progress of Metglas 2826 MB has also been followed on quenched specimens by X-ray diffractometry using the Co Ks radiation. 3. Results 3.1. D S C curves

Original DSC recordings of the 2826 MB alloy are reported in fig. 1 (20°C/min heating rate). The upper thermogram refers to the as-received metallic glass, while the others pertain to specimens quenched at increasing temperatures within the range of the first exothermal peak (arrows 1-4). After heat treatments interrupted at higher temperatures in correspondence with the second DSC peak (arrows 5-7), only this latter exothermal effect is still present with the features shown in fig. 2. As is seen, partial transformations occurring within the lower temperature DSC peak not only reduce the peak in height up to its disappearance, but also displace the peak towards higher temperatures, while the second peak is unaffected. Displacements towards lower and lower temperatures occur on the contrary for the second DSC peak when a previous annealing has reached the temperature range of the peak (fig. 2). This same behaviour has also been found for the single exothermal effect displayed by the 2826 B alloy after quenching beyond the glass transition temperture of the alloy (fig. 3). 3.2. X-ray analysis

The diffraction pattern of the as-received 2826 MB alloy is a spread halo that is practically unchanged after the first quenching of point 1 in fig. 1. Quenching from

A. Lucci et al. / Crystallization kinetics o f metallic glasses

289

q

g_

I

I

I

t

650

700

750

I

I

800 850 Calibrated temperature, K Fig. 1. DSC recording at 20°C/min of Metglas 2826 MB. The upper curve refers to the starting material and the arrows indicate the quenching temperatures. The other curves pertain to specimens previously submitted to annealing.

point 2, on the contrary, makes a diffraction spike develop. This latter is displaced in the field of diffraction angles when the heating temperature is increased (quenching points 3 and 4). Heating to higher temperatures, within the range of the second DSC peak, leaves the diffraction pattern of this first crystalline phase unchanged. In the meantime, however, further reflections of a new lattice appear, while the halo is reduced and lastly eliminated when heating overcomes the second DSC peak. After quenching at points 2, 3 and 4 the diffraction signal may be attributed to the (111) reflection of fcc N i - F e - M o solid solutions. The lattice parameter of such a phase (evaluated with an error of-+7 × 10 -4 nm) is 0.3537 nm after quenching 2, 0.3544 nm after quenching 3 and 0.3565 nm after quenching 4. After heating at higher temperatures the residual amorphous part changes into the second crystalline phase, (Fe, Ni, Mo)23B6, a metastable boride [5] of a form individuated on crystallized F e - B glassy systems by Herold and K6ster [6]. The introduction of low Mo contents in F e - N i based metallic glasses makes the

A. Lucci et al. / Crystallization kinetics o f rnetallic glasses

290

After quenching at :

A4

(~) 772K rng 1.35

/~/

(E) 800K rng 1.25

I!

~I

(~) 813K mg 1.85 / ~ 5 ~ /

~uW400_[

a_

.750

i

i

800

850

Calibrated temperature ,K Fig. 2. DSC traces recorded at 20°C/rain on Metglas 2826 MB previously submitted to heating up to the indicated temperatures.

4,

1~

mg 1.00 After quenching: (~) mg 1.05 1~) mg 0.95 t_

(~)

As received :

As received

2,3,4 quenching temperatu r e s

mg 0.90

o 13_

i

~2mW

I

I

I

I

650

S I

I

650

700

I

L

I

I

700 K

I

I

750 800 Calibrated temperature, K

Fig. 3. DSC traces recorded at 20°C/min on the Metglas 2826 B. Curve 1 refers to the starting material, the other curves to preannealed samples. The quenching temperatures after preheating are shown by arrows on an extended part of curve 1 reported in the inset.

A. Lucci et al. / Crystallization kinetics of metallic glasses

291

crystallization temperature increase [2,7-9]. Since both DSC peaks of 2826 MB alloy take place at temperatures around 70°C higher than those of amorphous alloys of similar composition, but without molybdenum, it seems reasonable to infer that Mo is present in both crystalline phases. 3.3. Activation energy o f crystallization

DSC curves like those of figs. 1 - 3 (q~ heating rate of 20°C/min) have also been recorded at 2, 5, 10 and 50OC/min. The various maximum temperatures T M of DSC peaks have been introduced in the Ozawa relationship [2,10] in order to evaluate the apparent activation energy connected to crystallization. As for the 2826 MB alloy, the activation energy of the low temperature peak after quenching 2 and 3 of fig. 1 cannot be determined in a precise way, due to the difficult T M determination. Only after the first quenching temperature (curve 1) may it be stated with an error of about 4% that the original activation energy of the unannealed metallic glass (296 kJ/mol) is increased by 44 kJ/mol. Further erdlancements of the activation energy after quenching at higher temperatures could be hypothesized on the basis of the observed displacements of TM towards higher temperatures. The high-temperature DSC peak is, on the contrary, sharper and well reproducible, so that activation energy may be evaluated with an error within -+4 kJ/mol. These data, deduced from the Ozawa plot of log • (heating rate) vs. 1/TM, are reported in table 1 ; the coefficients of linear correlation show the good agreement between the experimental points and the calculated straight line. Heat treatments up to the range of the high-temperature peak thus lower the activation energy for the crystallization of the boride phase. Such behaviour does not seem to be peculiar for a two-step crystallizing glassy alloy: the activation energy for crystallization of as-received Metglas 2826 B (single peak curve 1 in fig. 3) is 429 kJ/mol [2,11], but quenching at 712 K (curve 4 in fig. 3) reduces it to 402 kJ/mol.

Table 1 Effect of preannealing on the activation energy of boride crystallization in Metglas 2826 MB Quenching temperature (K)

Activation energy (kJ/mol)

Coefficient of linear correlation

as~eceived 800 813 820

384 375 367 352

-0.9999 -0.9995 -0.9998 -0.9994

292

A. Lucci et aL / Crystallization kinetics of metallic glasses

3.4. Avrami kinetic law: the n parameter

Crystallization of metallic glasses has been recognized to follow the nucleation and growth kinetics described by Avrami [12] and Johnson and Meb_l [13] for phase changes in metals. This is what emerges from the literature in all the cases examined hitherto [11,14-18]. The Avrami law relating the (1 - a) untransformed fraction to the time of isothermal annealing t is of the form: 1 - a = exp(-ktn), where the rate constant k,

~,-~'~'-0.425

2D 1,5

I _/-

I -

mr~Ceived~

I I I I - n E = 5 8 6 k J , mol -I n=1.53

"092

~467

~

nE= 928kg mo1-1 n=242

E

2,0

d, =0014

>" (_ ~1

at 772 K ~

.~

nE=567kJ.mo1-1

°

L.

n =1.48

,o

.043

t~

~-<1" 1,50

quenched at 8 0 0 K

-

1.0-

~

- ~ , ~ _

nE=570k,.T.mo1-1

n=1.54

~

-1.5

quenched

n E = 4 2 8 k J m o l -1 n=117

-

:

-1.0

~ -05 I

1.22

'

I

1,24

i

I

1,26 1000 / T, K -1

'

I

1,28

Fig. 4. Logarithmic plot of the AT deflection of DSC traces as a function of the reciprocal of the absolute temperature for Metglas 2826 MB. The a transformed fractions on the DSC curves corresponding with the extremes of the linear plots are indicated, as well as the calculated n Avramiexponent.

293

A. Lucci et al. / Crystallization kinetics of metallic glasses

is usually put in the Arrhenius form k = ko e x p ( - E / R T ) , with ko constant, E activation energy, R gas constant and T absolute temperature. From a theoretical standpoint the n exponent may vary within three ranges of values ( 1 - 2 , 2 - 3 , 3 - 4 ) following the one-, two-, or three-dimensional nature of crystal growth. The nucleation rate fixes the limits of such a variation: a zero nucleation rate corresponds to the lower value within each range and a constant nucleation rate to the upper value; this latter may be overcome when the nucleation rate is increasing. Recent developments of non-isothermal kinetics [3,4] enable one to determine the n exponent through an analysis of the first part of a DSC or DTA thermogram, using the method of Piloyan et al. [19], if the activation energy of the process is known. A logarithmic plot of da/dt (or proportional quantities, like the differential temperature AT) as a function of the reciprocal of the absolute temperature should give in fact a straight line with a nE/R slope. Figs. 4 and 5 report such plots for 2826 MB (high-temperature peak) and 2826 B metallic glasses in the as-received form and after heat treatments. Straight lines are obtained for a transformed fractions up to around 40% or more (higher than those theoretically computed [4] for different kinetic parameters). A slope change is observed only on the curve pertaining to the as-received 2826 MB alloy. As is seen from both figures, quenching from the higher temperatures makes the n exponent decrease.

'

I

as received

'

I

'

1

quenched at 712K

1.5¢-

~nE=788k3. '~=

1.0 .0 L t~

E= 2,108 k3.mol -~ n = 491

<] Q5 _o

mo1-1 -1.96

~ _

\

0 1.40

1.42 1.44 1,000 / T, K-1

Fig. 5. Logarithmic plot of the AT deflection vs. I/T for Metglas 2826 B. The n values of the Avrami exponent are indicated.

294

A. Lucci et al. / Crystallization kinetics of metallic glasses

4. Discussion

The increase in the lattice parameter of the y-phase after annealing stopped on the first DSC peak of the 2826 MB alloy shows that the composition of the solid solution is changing. In particular at the lower temperatures it is richer in nickel (~90%) while the Fe + Mo contents increase with the quenching temperature (~70% Ni after quenching at 499°C). Higher amounts of less diffusive elements (Fe and Mo) in the solid solution could account for the observed increasing values of the activation energy for crystallization of the first crystalline phase. Even if such an explanation seems reasonable, some preliminary tests on a Fea6B14 glassy alloy cause some doubts on its validity to arise. This metallic glass has in fact a DSC pattern very similar to the one of Metglas 2826 MB and the first exothermal effect of this glass corresponds to crystallization of mere a-iron. In the absence of any enrichment from other metals (like Ni and Mo) this peak shows some shift towards higher temperatures after heat treatments within the temperature range of the peak. A different mechanism should have to be invoked, in any case, to explain the reduction of the activation energy for the crystallization of the boride (2nd peak in 2826 MB alloy) and phosphoboride (Metglas 2826 B) after heating to the region of their corresponding DSC peak. K6ster and Herold [20] and Scott and Ramachandrarao [17] claim that the energy barrier for nucleation is definitely higher than the one for crystal growth in crystallizing metallic glasses. If this is true, it is not difficult to admit that preannealed samples, containing a given population of preformed nuclei, can crystallize in an easier way than the completely amorphous metal. Straight lines of figs. 4 and 5 enable one to determine the n exponent of the Avrami equation under non-isothermal conditions. An initial two-dimensional growth of the boride in Metglas 2826 MB (n = 2.42) tends to become one-dimensional at higher a transformed fractions (n = 1.53). Successive annealings do not modify the geometrical feature of crystal growth; the lowest n value (1.17) at the highest quenching temperature should indicate a decreasing nucleation rate. The experimental TEM evidence of a rod form in a boride of similar composition [21] supports such findings. The high n value (4.91) in the as-received Metglas 2826 B is in agreement with previous isothermal measurements on this same alloy [ 11 ]. These latter measurements, in fact, give values of n increasing with temperature and becoming larger than 4 at temperatures lower than those of the present tests, thus indicating an increasing nucleation rate. Also in this alloy a previous annealing is able to reduce both the process dimension and the nucleation rate. A n = 1.96 figure should indicate an one-dimensional crystal growth with a constant nucleation rate, or (n ~--2) a two-dmensional growth with a constant number of nuclei (zero nucleation rate).

A. Lucci et al. / Crystallization kinetics of metallic glasses

295

5. Conclusions Partial transformations from the amorphous to the crystalline condition in F e - N i based metallic glasses are able to modify the successive crystallization behaviour o f these materials. In the Fe4oNi38Mo14Bls alloy heating up to the range of the first crystallization stage produces a N i - F e - M o 3' solid solution with decreasing Ni contents at increasing annealing temperatures. A contemporary enhancement o f the activation energy for crystallization is observed. A reduction o f the DSC peak temperature and the relative activation energy o f crystallization occurs on the contrary after heating up to the stage o f (Fe, Ni, Mo)23B 6 crystallization in the same alloy and o f (Fe, Ni)a(P, B. Si) crystaUization in the Fe29Ni49PI4B6Si2 metallic glass. Such behaviour could be explained admitting that the energy barrier for nucleation is higher than that for crystal growth, so that the presence of preformed nuclei enhances the crystallization rate. In the meantime the value o f the " n " exponent of the Avrami rate equation, determined under non-isothermal conditions, indicates that the nucleation rate tends to decrease and that crystals grow in a one-dimensional way, or possibly a two-dimensional way, in 2826B metallic glass.

References [1] C. Antonione, L. Battezzati, A. Lucci, G. Riontino and G. VentureUo, Proc. JCAT 78 (Torino, 1978) B18, 141. [2] C. Antonione, L. Battezzati, A. Lucci, G. Riontino and G. VentureUo, Scripta Met. 12 (1978) 1011. [3] A. Marotta and A. Bud, Thermochim. Acta 25 (1978) 155. [4] J. Colmenero, J. llarraz and J.M. Barandiar~in, Thermochim. Acta 35 (1980) 381. [5] C. Antonione, L. Battezzati, A. Lucci, G. Riontino, M.C. Tabasso and G. VentureUo, Proc. 4th Int. Conf. on Liquid and Amorphous Metals (Grenoble, July 1980); J. de Phys. 41 (1980) C8, 131. [6] U. Herold and U. K6ster, Z. Met. 69 (1978) 326. [7] F.E. Luborsky, Mat. Sci. Eng. 28 (1977) 139. [8] F.E. Luborsky and H.H. Liebermann, Appl. Phys. Lett. 33 (1978) 233. [9] J.L. Walter, Mat. Sci. Eng. 39 (1979) 95. [10] T. Ozawa, J. Therm. Anal. 2 (1970) 301. [11] M.G. Scott, J. Mat. Sci. 13 (1978) 291. 12] M. Avrami, J. Chem. Phys. 7 (1939) 1103; 8 (1940) 212; 9 (1941) 177. 13] W.A. Johnson and R.F. Mehl, Trans. Met. AIME 135 (1939) 416. 14] T. Masumoto and R. Maddin, Acta Met. 19 (1971) 725. 15] J.M. Vitek, J.B. van der Sande and N.J. Grant, Acta Met. 23 (1975) 165. 16] E. Coleman, Mat. Sci. Eng. 23 (1976) 161. 17] M.G. Scott and P. Ramachandrarao, Mat. ScL Eng. 29 (1977) 137. 18] E. Coleman, Mat. Sci. Eng. 39 (1979) 261. 19] F.O. Piloyan, I.O. Ryabchikov and O.S. Novikova, Nature 212 (1966) 1229. 20] U. K6ster and U. Herold, Scripta Met. 12 (1978) 75. 21] J.L. Walter, S.F. Bartram and R.R. Russell, Met. Trans. 9A (1978) 803.