The oxidation behaviour of amorphous alloy Fe40Ni40P14B6

The oxidation behaviour of amorphous alloy Fe40Ni40P14B6

Aclo metall. Vol. 36. No. 1, pp. 167-180, 1988 Printed in Great Bntain. All rights reserved oool-61643/88 $3.00+ 0.00 Copyright C 1988Pergamon Journa...

2MB Sizes 0 Downloads 85 Views

Aclo metall. Vol. 36. No. 1, pp. 167-180, 1988 Printed in Great Bntain. All rights reserved

oool-61643/88 $3.00+ 0.00 Copyright C 1988Pergamon Journals Ltd

THE OXIDATION BEHAVIOUR OF AMORPHOUS ALLOY Fe,Ni,,P,,B, GAO WEIt and B. CANTOR Department of Metallurgy and Science of Materials, University of Oxford, Parks Road, Oxford OX] 3PH, England (Received

28 February 1987)

Abstmaet-A combination of thennogravimetry, optical microscopy, scanning and transmission electron microscopy, electron probe microanalysis and differential scanning calorimetry has been used to investigate the oxidation behaviour of amorphous Fe,Ni,P,,B,. The oxide layer formed on amorphous Fe,,,Ni,,,P,,B, has a whisker-like or-Fe,O, structure, which grows very rapidly to build up a thick layer of oxide. Kinetic data indicate that the oxidation of amorphous Fe,,,N&P,,B, obeys a parabolic rate law as long as the alloy remains amorphous and the rate controlling process is diffusion of iron in the amorphous alloy matrix. However, the oxidation rate drops sharply if crystallization of amorphous Fe9Ni4P,,B, takes place during oxidation annealing. Crystalline Fe,N&,P,,B, also obeys a parabolic rate law but with as much smaller rate constant than the amorphous alloy. The rate controlling process for oxidation of crystalline Fe,Ni,P,,B, is diffusion of iron and nickel in the multiphase oxide layer, which consists of a fine scale mixture of NiO, Fe,O,, F%O, and NiF%O, crystals. The difference in oxidation behaviour between amorphous and crystallized FeoNi,J,,B, is caused by the different alloy microstructures. R&m&--Nous avons itudie l’oxydation de Fe,Ni,P,,B, en utilisant conjointement les techniques de la thermogravimttrie, de la microscopic optique, de la microscopic electronique par transmission et par balayage, de la microanalyse par sonde electronique et de la calorimetric differentielle en balayage. La couche d’oxyde formee sur FesNi,P,,B, amorphe presente une stucture FqO,a de type whisker qui croit tres rapidement pour former une tpaisse couche d’oxyde. Les donnCs de cinitique indiquent que l’oxydation de Fe,Ni,P,,B6 oMit a une loi de vitesse parabolique tant que l’alliage demeure amorphe, et que le mecanisme qui contrBle la vitesse est la diffusion du fer dans la matrice de l’alliage amorphe. Cependant, la vitesse d’oxydation chute brutalement si Fe&Ni,P,,B, amorphe cristallise pendant le recuit d’oxydation. Fe,Ni,P,,B, cristallin ob&it egalement a une loi de vitesse parabolique, mais avec une constante de vitesse bien plus faible que celle de l’alliage amorphe. Le mecanisme qui controle la vitesse d’oxydation de Fe,Ni,P,,B, cristallin est la diffusion du fer et du nickel dans la couche d’oxyde multiphasee qui est constituee d’un melange P petite echelle de cristaux de NiO, de Fe,O,, de F%O, et de NiFe,O,. La diffkrence que l’on observe entre l’oxydation de Fe,N&,P,,B, amorphe et cristallise a pour origine les microstructures differentes de l’alliage. Zusammenfnssung-Das

Oxidationsverhalten von amorphem Fe,Ni,P,,B, wird mit einer Kombination von Megmethoden untersucht: Thennogravimetrie, Lichtmikroskopie, Raster- und Durchstrahlungselektronenmikroskopie, Mikroanalyse und differentielle Kalorimetrie. Die auf diesen Proben gebildete Oxidschicht be&t eine Whisker-artige or-Fe,O,-Stuktur, die sehr rasch wachst und eine dicke Oxidschicht bildet. Kinetische Daten zeigen, daB die Oxidation so lange einem parabolischen Geschwindigkeitsgesetz gehorcht, wie die Legierung amorph bleibt, und daD der geschwindigkeitsbestimmende Schritt die Diffusion des Eisens in der amorphen Matrix ist. Die Oxidationsgeschwindigkeit fillt allerdings scharf ab, wenn die amorphe Legierung wlhrend der Oxidations-Auslagerung kristallisiert. Die Oxidation der kristallinen Legierung folgt such einem parabolischen Gesetz, allerdings mit einer vie1 kleineren Ratenkonstanten. In diesem Falle ist der geschwindigkeitsbestimmende Schritt die Diffusion von Eisen und Nickel in der vielphasigen Oxidschicht, welche aus einer feinen M&hung von NiO-. Fe,O,-, Fe,O,und NiFe,O,-Kristallen besteht. Der Unterschied im Oxidationsverhalten zwischen amorpher und kristallisierter Legierung wird durch die unterschiedliche Mikrostruktur der Legierungen verursacht.

1. INTRODUCTION

been given to the crystallization

behaviour

of amor-

phous FeaNimP,,B,; [l-4] there has been almost no Amorphous alloys are intrinsically unstable materials invetigation of its oxidation behaviour [S], even and can decompose when exposed to moderate temthough a preliminary study indicates that amorphous peratures. The two main potential causes of deFe,Ni,P,,B, has a particularly poor oxidation regradation are crystallization and oxidation. Thus, sistance compared with several other amorphous investigating the oxidation behaviour of amorphous alloys [6]. In the present work, a combination of alloys is important for both understanding and prethermogravimetry, optical microscopy, scanning and venting of this type of degradation. Fe40Ni*OP,4B6 transmission electron microscopy, electron probe mi(METGLAS 2826) is a well known amorphous alloy croanalysis and differential scanning calorimetry has with structure and properties which have been studbeen used to investigate the oxidation kinetics, and ied in considerable detail. While much attention has oxide morphology, structure and composition in exposed to moderate temton leave from Chengdu University of Science and amorphous Fe,Ni,P,,,B, Technology, China. peratures in air. 167

168

GAO and CANTOR:

OXIDATION BEHAVIOUR OF AN AMORPHOUS ALLOY

2. EXPERIMENTAL DETAILS The Fe40Ni40P,dB6amorphous alloy was supplied by Allied Chemical Corp. in the form of ribbon 12.8 mm wide and approx. 45 pm thick. The chemical composition of the ribbon was measured by using a Cameca CAMEBAX wavelength dispersive electron probe microanalyser and the results are shown in Table 1. Samples of amorphous Fe40Ni90P14B6with measured surface area of about 5 cm* were cut directly from the as-supplied ribbons and cleaned thoroughly in an ultrasonic cleaner before exposure to oxidation testing in air. The furnaces which were used for oxidation annealing had an 8 cm hot zone over which the temperature profile was constant to within + 1 K. Oxidation annealing experiments were all performed in air for times ranging from 0.1-500 h at eight 150-500°C. A the range temperatures in Perkin-Elmer AD-2 microbalance wilth a sensitivity of 0.1 pg and an accuracy of 1 pg was used for the thermogravimetric mass gain measurements. For each test, three specimens were used to calculate an average mass change as the result. Pre-crystallized samples of Fe,,,Ni,,,P,4B6 were subjected to a similar series of oxidation annealing and thermogravimetric experiments in order to compare the oxidation behaviour of the amorphous and crystallized alloy. The precrystallized specimens were prepared by annealing in a vacuum tube at 500°C for 30 min. After annealing, the pre-crystallized specimen surfaces remained as bright as the as-supplied ribbon, indicating that no premature oxidation had taken place. To determine oxidation annealing conditions under which the alloy structure changes from amorphous to crystalline, a DuPont 1090 Thermal Analyzer was used in differential scanning calorimetry mode with the oxidized samples. Continuous heating from 350 to 500°C with a heating rate of 10 K/min was used and the sample cell was protected by a dynamic argon flow. The exothermic heat of crystallization was monitored during the continuous heating differential scanning calorimeter experiments, to determine whether any crystallization had taken place during the prior oxidation anneal. Details of the oxide microstructure were examined in an Olympus BMH optical microscope and a Hitachi S-530 scanning electron microscope by direct observation of the oxidized specimen surfaces, and by preparing taper cross-section specimens for throughthickness investigation. The cross-section samples were prepared by nickel plating the oxide surface, Table 1. Nominal and measured compositions Fe,,,Ni,P,,B, Ekment Nominal composition Measured composition

of amorphous

(at.%)

FC

Ni

P

Bt

40 42.2

40 39.0

14 12.9

6 5.9

tBoron content measured by subtraction

from 100%.

followed by mounting in epoxy resin and then polishing with a specimen surface/polishing surface angle of about 3”. With this technique, the thin oxide layers had a visual thickness in the microscope which was approx. 20 times greater than the true thickness. Transmission electron microscope specimens were prepared in two ways. Firstly, the thin oxide films were removed from oxidized amorphous and precrystallized samples for subsequent direct examination, by dissolving the underlying metal in a saturated iodine-methanol solution [7l. Secondly, thin foil samples of the amorphous and precrystallized ribbons were prepared by ion beam thinning, and then subjected to a suitable oxidation anneal, in order to examine the early stages of oxidation with the oxide still attached to the underlying metal. Transmission electron microscope examination was carried out by using JEOL IOOC and 2OOOFXtransmission electron microscopes with Link Systems energy dispersive electron probe microanalysis attachments. Examination of the stripped oxide films gave details of the oxide structure and composition without any influence from the matrix, while examination of the oxidized thin foils gave direct information about the mechanism of oxide nucleation and growth. Oxide microstructures were investigated by bright and dark field imaging, oxide crystal structures were determined by indexing selected area electron diffraction patterns, and oxide compositions were determined from electron probe X-ray microanalysis spectra. Selected area diffraction information was obtained from regions of about 0.5 pm in diameter and composition information was obtained from regions of about 10 nm in diameter. 3. RESULTS 3.1. Oxidation kinetics Figure l(a) shows the measured mass gain of amorphous Fe40Ni.,,,P,4B6as a function of the square root of oxidation annealing time over a range of oxidation annealing temperatures. Figure l(b) shows the same results plotted as In mass gain vs In time. Best fit linear slope and intercepts from the data in Fig. 1 are given with correlation coefficients in Table 2. As shown in Fig. 1, the mass gain was proportional to the square root of oxidation time at the lower oxidation temperatures of 200, 250 and 300°C. In other words, the oxidation reaction obeyed a parabolic rate law: y = Kt’/*

(1)

where y is mass gain, t is oxidation annealing time, and K is the parabolic reaction rate constant. As shown in Table 2, the parabolic rate constant K increased from 1.8 x 10T3 to 2.9 x lo-* g/m* hln with increasing oxidation annealing temperature from 200 to 300°C. At the higher oxidation temperatures of 450 and 5OO”C, the oxidation reaction also obeyed the parabolic rate law in equation (I), with even

GAO and CANTOR:

““1(a)

FdON4Ci’1426

OXIDATION BEHAVIOUR OF AN AMORPHOUS ALLOY

rate law in equation (1) again with a rate constant of 1.8 x lo-* g/m2 h12 much smaller than in the first stage of oxidation annealing. Because of this unusual deceleration in oxidation rate, Fig. 1 shows that the total oxidation mass gain after 500 h at 400°C was much less than that at 350 or even 300°C. The change in oxidation behaviour as a function of oxidation time at 350 and 400°C was found to correspond to a change in the Fe,Ni,P,,B, alloy structure from amorphous to crystalline. Figures 2(a) and (b) show differential scanning calorimetry results for the exothermic heat of crystallization as a function of prior oxidation time. Crystallization of amorphous Fe40Ni4$,4Bs was complete within 10 min during oxidation at 400°C; crystallization began after about 24 h and was complete after about 70 h of oxidation at 350°C; however no crystallization was detected even after 700 h of oxidation at 300°C. These results indicate that the onset of crystallization causes the sharp drop in oxidation rate after 0.1 h at 400°C and 20 h at 35O”C, which is shown in Figs l(a) and (b). In order to investigate further the effect of crystallization on the oxidation behaviour of amorphous FeMNi,,,P,,Bs, comparison oxidation tests were performed on amorphous and pre-crystallized samples. Figures 3(a) and (b) show the measured mass gains for both types of Fe40Ni,P,,B, alloys structure as a function of oxidation time in air at 300 and 350°C respectively. The results show verly clearly that amorphous FeMNi,P,,B, oxidized very much more rapidly than pre-crystallized Fe,Ni,P,,B,. As shown in Fig. 3, the amorphous alloy had a mass gain about eight times greater than the pre-crystallized alloy after 700 h oxidation at 3OO”C,and about six times greater after 625 h oxidation at 350°C.

Y-t”’

i.e -9.8

-2.e

i.0 In

I

a.0

5.0

7.8

houn

Fig. 1. (a) Mass gain y vs square root of oxidation for amorphous Fe,Ni,,,F’,,B( over the temperature 20&5CWC in air; (b) In mass gain y vs In oxidation for amorphous Fe,Ni>,,B, over the temperature 200-500°C in air.

169

time t range time r range

higher values of 6.7 x lo-* and 2.0 x 10-l g/m2 h”’ for the parabolic rate constants as shown in Table 2. However, the amorphous alloy oxidation behaviour was considerably more complicated at the intermediate oxidation temperatures of 350 and 4OO”C, where the oxidation rate law clearly changed as a function of time during the oxidation anneal. At 35O”C, the oxidation reaction obeyed the parabolic 3.2. Oxide morphology rate law of equation (1) for the first 20 h of annealing Figure 4 shows typical taper cross section optical with a high parabolic constant of 1.2 x 10-l g/m2 hII*, and then the reaction rate dropped to almost zero, micrographs of both amorphous and pre-crystallized Fe40Ni,P,,B, after 700 h oxidation at 300°C in air. with little further mass gain up to 625 h of annealing. At 4OO”C,the oxidation reaction obeyed equation (1) Figure 5 shows typical scanning electron micrographs of the oxide surface on both amorphous and prefor the first 0.1 h of annealing, with a high parabolic constant of 4.0 x 10-l g/m2 h”2; then the reaction rate crystallized Fe@Ni,P,,B, after 120 h oxidation at dropped suddenly to almost zero, with no mass gain 300°C. As shown in Fig. 4, amorphous Fe,Ni,P,,B, from 0.1 h to about 50 h; finally the reaction rate had a layer thickness of approx. I pm, much greater than the oxide layer thickness of approx. 0.05 pm on increased slowly after 50 h to follow the parabolic Table 2. The best 61linear slopes and intercepts from the mass gain measurements as a function of square root of oxidation annealing lime in Fig. I(a). The linear slopes are qua1 to the parabolic rate constant for the oxidation reaction Temperature (“C) 200 ;o” :z 400 400 450 500

Slope

Oxidation period (h)

(g/m2 h”?)

O-500 O-500 O-500 D-20 75-620 O-Q.1 IO&500 O-250 O-41

1.8 x 9.8 x 2.9 x 1.2 X 1.5 X 4.0 X I.8 x 6.7 x 2.0 x

IO-’ IO-’ IO-* IO ’ IO_’ IO ’ IO ’ 10m2 10-I

Intercept (g/m*) 2.6 x IO-’ 3.1 X 10-2 3.7 X 10-2 -5.8 x IO ’ 7.7 X IO ’ 1.8~10’ I.2 X IO_’ 1.0x IO ’ 7.5 x 10-Z

Correlation coefficient 0.999 0.994 0.999 0.999 0.985 0.996 0.997 0.996 0.998

GAO and CANTOR

170

OXIDATION

BEHAVIOUR

OF AN AMORPHOUS

ALLOY

DSC

26..

Fr40Ni40P 1496 350 EC

*’

24.. $

20..

E -

.. A,

16..

II. In

Unoxidized +

72h

4.:

1.

66h

r 0, 350

360

370

300

390

400

410

Temperature

420

430

440

450

460

(‘C 1

DSC

z 0 < z ;

lOi

I II _& -409.2 Yz 156 J/Q I :___-

0 .:

-10 ‘1

Unoxidized

300

409.7 lc 163 J/g

-20 i -3oA-----350

C,7OOh

400C,

0.2 h I

1 360

370

360

390

400

410

Temperature

420 (*c

430

440

450

460

1

Fig. 2. Differential scanning calorimetry (DSC) results showing exothermic heat of crystallization for amorphous Fe,Ni,P,4B, as a function of oxidation time at (a) 35O”C, (b) 300 and 400°C.

the pre-crystallized Fe,Ni,P,,B,. These results are in agreement with the mass gain measurements shown in Figs 1 and 3. In addition, Figs 4 and 5 show a sharp difference in oxide morphology between the amorphous and pre-crystallized FedONi40P,4B6speci-

mens. The oxide layer formed on amorphous Fe,Ni,P,,B, was not very compact and had a whisker structure towards the top surface, as can be seen very clearly in the scanning electron micrographs in Fig. 5(a). In contrast, the oxide layer formed on 100

1

(b)

y-t

(a)

Fe4ON14OP1480 A

_

0

P 60 ;

i?

26

1

0

160

600 TIME

460 60”~~

600

760

0

0

200

200

400 TIME

200

how*

Fig. 3. (a) Mass gain y vs oxidation time showing comparison of amorphous (A) and pre-crystallized (C) Fe,Ni,P,,B, at 300°C in air. (b) Mass gain y vs oxidation time showing comparison of amorphous (A) and pre-crytallized (C) Fe,Ni,P,,B, at 350°C in air.

GAO and CANTOR:

OXIDATION

BEHAVIOUR

OF AN AMORPHOUS

ALLOY

171

Fig. 6. Brighht field TEM image of the whisker-like oxide structure on amorphous Fe,,,Ni,&‘,,B6 oxidized in air at 300°C for 100 h.

I% (b) Fig. 4. Taper cross-section optical micrographs of the through thickness structures of oxide layers on (a) amorphous and (b) pre-crystallized Fe,Ni,P,,B6 oxidized at 300°C in air for 700 h.

Fig. 5. Scanning electron micrographs of the oxide layer surfaces formed on (a) amorphous and (b) pre-crystallized Fe,Ni4P,,B, oxidized in air at 300°C for 120 h.

pre-crystallized Fe40Ni4$,4B6 had a compact structure of microcrystalline grains, as shown in Fig. 5(b). Whisker-like oxide structures usually grow very fast and have poor protective ability due to the brittleness of the oxide whiskers. Previous investigations have found that whisker-like oxide structures can grow under certain conditions during the oxidation of copper, iron, molybdenum, tantalum and even stainless steel [&lo]. In the present experiments on amorphous Fe,,,,Ni40P,4B6rthe oxide whiskers grew typically to a length of 1-2pm, a width of O.l-Q).2pm and a thickness of 6-10nm as shown in the transmission electron micrograph in Fig. 6 after 100 h at 300°C. At the higher temperatures of 450 and SOo”C, the oxide layer on amorphous FemNiaP,,Bs was found to break away after a period of oxidation time. This is shown in the scanning electron micrograph in Fig. 7 after oxidation annealing for 200 h at 500°C. In Fig. 7, the top layer of oxide has broken away and a new whisker-like oxide layer has begun to form and grown on the freshly exposed alloy surface. This type of oxide layer fracture and re-growth on freshly exposed metal may account for the slight acceleration in oxidation rate which can be seen in Fig. 1 after about 25 h at 500°C.

Fig. 7. Scanning electron micrograph of the oxide layers on Fe,Nid,,B, oxidized in air at 500°C for 200 h showing oxide fracture and whisker-like oxide re-growth on freshly exposed alloy.

172

GAO and CANTOR:

OXIDATION BEHAVIOUR OF AN AMORPHOUS ALLOY

Fig. 9. ‘IBM micrographs of the oxide layers removed from pre-crystallized Fe,Ni$,,B, oxidized in air at 300°C for 24Oh. (a) Bright field image: (b) selected area diffraction pattern from (a).

Fig. 8. TEM micrographs of the oxide layers removed from amorphous Fe,NiloP,,B, oxidized in air at 300°C for 240 h. (a) Bright field image, (b) selected area diffraction pattern from (a), (c) indexing of (b). 3.3- ~~~~rn~sion electron microscopy Figures 8 and 9 show bright field transmission electron micrographs and selected area electron diffraction patterns of the oxide film stripped from ~o~hous and pre-crystallized Fe,N&Pi,B, respectively after 240 h oxidation at 300°C. The oxide grain size formed on the pre-crystallized Fe~Ni.$i4Bb was much smaller than on the amorphous alloy. This can be seen directly on the bright field images of Figs 8(a) and 9(a), and is also shown by the spot diffraction pattern of Fig. 8(b) compared with the ring pattern of Fig. 9(b). Even using the smallest available selected area aperture, of 0.2 pm dia, the oxide diffraction patterns from precrystallized FeurNirOpi,Bs still contained many rings as shown in Fig. 9(b). Indexing selected area diffraction patterns showed that the oxide layer on amorphous Fe,,,NiloP,,B, consisted mainly of a-FezO,, while the oxide layer on pre-crystallized Fe40Ni,J’,,B, consisted of a mixture of NiO, Fe,O,,

Fe,O,, NiFe,O, and possibly some other compounds. The detailed results obtained from indexing selected area electron diffraction patterns are given in Tables 3 and 4 for amorphous and pre-crystallized Fe,,,Ni,$i,B, respectively. Crystal structures found in the two types of oxide are given in Table 5, together with composition measurements obtained from electron probe X-ray microanalysis. Figure 10 shows typical electron probe X-ray microanalysis spectra from the oxide layers from both amorphous and pre-crystallized Fe,Ni,,,P,,B,, and the resulting oxide compositions are included in Table S. The measured compositions were found to be quite different for the stripped oxide layers from amorphous and pre~~stalli~d Fe~Ni~P,,B~, as shown in Fig. 10 and Table 5. The oxide formed on pre-crystallized Fe,ONi,P,,B, had a variable composition in the range 52-83 at.%Fe, 11_43%Ni and 4-15%P, similar to the composition of the matrix alloy (boron cannot be measured by X-ray microanalysis). In contrast, the oxide formed on amorphous Fe,Ni,,,P,,B, contained only iron, with no nickel or phosphorus detected in the X-ray spectra. As shown in Table 5, the X-ray microanalysis results were in agreement with the oxide crystal structures obtained by analysis of selected area electron diffraction patterns. The oxide layer formed on the pre-crystallized Fe,,,NiWPL,B6 clearly had a much more complicated structure and composition than the oxide layer formed on amo~hous Fe~Ni~P,~~~

GAO

and

CANTOR

OXIDATION

BEHAVIOUR

OF AN AMORPHOUS

ALLOY

Table 3. Cakulated d spacings for typical diffpxctiott Mytorn obtained from an oxidctilm stripped from amorphous Fe,NioP,,B, after 240 h oxidatioh at WC (d,). Ynd an oxidized amo’hhous Fe,Ni,P,,B, thin foil specimen after 50 h at 3OOC (4) together with A.S.T.M. powder card index d spacings of a-Fe,O,

(A)

4 R, (mm)

K = 27.45 mm

7.5 10.2 10.9 12.0 12.4 13.2 15.0 16.25

3.66 2.69 2.52 2.29 2.21 2.08 1.83 1.69

18.4

1.49

20.8

1.32

22.4 23.3 23.8 24.2 25.0 26.5 27.7

1.23 1.18 1.15 1.13 1.10 1.04 0.99

30.4

0.0903

A

4 (mm)

(4

4

K=l7.6mmA

4.8 6.6 7.0

3.66 2.67 2.51

8.0 8.5 9.5 10.5 11.0 11.9 12.1

2.20 2.07 1.85 I .68 1.60 1.48 1.45

14.0

1.26

15.6

1.13

16.5 17.8 18.5 19.3

1.06 0.99 0.95 0.91

(W 3.66 2.69 2.51 2.285 2.201 2.070 1.838 1.690 1.596 I .484 1.452 1.310 1.258 1.226 1.189 1.162 1.141 1.102 I.055 0.989 0.951 0.908

012 104 110 006 113 202 024 116 018 214 300 IOF, 119 220 036 128 0210 134226 2110 232;318 140,0213 1310 - -

Table 4. Calculated d spacings for typical diffraction patterns obtained from an oxide film stripped from pre-crystallized Fe,Ni,P,,B, after 240 h oxidation at 3OO”C, together with A.S.T.M. powder card index d spacings of several Fe and Ni oxides R (mm)

d (A) K =28.5mm 4.19 3.70 2.97 2.94 2.69 2.52 2.42 2.32 2.21 2.10 I .98 1.85 1.72 1.70 1.61 1.55 I .48 1.41 1.32 1.27 1.21 1.145 1.109

6.8 ;:: 9.7 10.6 11.3 11.8 12.3 12.9 13.6 14.2 15.4 16.6 16.8 17.7 18.4 19.2 20.2 21.6 22.5 23.5 24.9 25.7 26.0 26.3 27.2 29.4 30.4 31.4 32.3 33.3 35.1

1.096 1.085 1.048 0.969 0.938 0.908 0.882 0.856 0.812

A

Fe@,

(Il-614)

Fe@, (13-534)

NiO (22-l 189)

NiFe,O,

(l&325)

3.66 2.966 2.948 2.69 2.51

2.530 2.419

2.513 2.412

2.20 2.096

2.088

2.085

1.84

1.712 1.690 1S% 1.483 1.327 1.279, 1.264 1.211

I.484 I.452 1.310 1.258

1.702 1.605

I.477

1.476

1.259 1.2064

1.271

1.260,

1.141 1.092

1.102

I .049

1.055 0.9601

1.0857 0.9692 0.9386

1.0458 0.9592 0.9345.0.9337

0.9080 0.8794 0.8565 0.8113

0.8529,0.8519

Table 5. Results of selected area electron diffraction patterns and electron probe microanalysis from oxide layers removed from amorphous and pre-crystallized Fe,Ni,P..B.

Alloys Amorphous Fe,Ni,P,,B, Pre-crystallixed FeaNi,P,,B,

Oxidation conditions

Oxide structure

Oxide compositiont (at.%)

3OO”C,240 h in air 300°C. 240 h in air

a-FerO,

Fe > 99%

Fe,O,, NiO, FerO,, NiFexO,

Fe: 52-83% Ni: 1143% P: 415%

tElements with atomic number < analysis.

I1 cannot be measured by electron probe micro-

0.9630 0.9324

173

174

GAO and CANTOR: TN-5500

0.000

OXIDATION

BEHAVIOUR

OF AN AMORPHOUS

CHEMICAL CRYSTALLOGRAPHY OXFORD TUE 09-SIP-86

23A-240

VFS = 2046

HTA 9/9/06

ALLOY

12.41

10.240

(a) TN-5500 CHEMICAL CRYSTALLOGRAPHY OXFORD TUE 09-SEP-86 (2) 7.250: 7.730 Curror : 0.000 keV = 0 ROI

O.OOD 102

230-240

HTA 9/9/W

VFS = 1024

11.05

10.240

(b) Fig. 10. Typical electron probe X-ray microanalysis spectra from the oxide layers removed from (a) amorphous and (b) pre-crystallized Fe,Nig,,B, oxidized at 300°C for 240 h.

Fig. II. TEM micrographs of unoxidized amorphous Fe,NQ,,B, foils. (a) Bright field image; (b) Selected area diffraction pattern.

Figures 11-14 show bright field transmission electron micrographs and selected area electron diffraction patterns from as-prepared and oxidized thin foils of amorphous and pre-crystallized Fe,ONi,r,P,,B, respectively. With no oxidation annealing, the amorphous Fe,Ni,P,,B, exhibited a featureless bright field micrograph and a diffraction pattern consisting of two broad haloes, as shown in Fig. 11. After only 0.5 h oxidation at 3OO”C,the oxide microstructure which formed on the amorphous alloy thin foil could easily be detected, as shown in Fig. 12(a). Bright field images showed many dark stripes, which were whisker-like oxide growing on the amorphous thin foil. Near the edges of the thin foils as shown in Fig. 12(a), a different oxide structure was found, with no stripes but small oxide grains instead. The d spacings calculated from the diffraction patterns of the dark stripe areas were confirmed to be those of a-FerO,, as given in Table 3, in agreement with the results obtained from diffraction patterns from oxide layers stripped from the amorphous alloy. The pre-crystallized Fe,Ni,,,P,,B, thin foils produced very different results from the amorphous thin foils, as can be seen by comparing Figs 11 and 12 with Figs 13 and 14. According to Watanabe and Scott [2], amorphous Fe,Ni,PI,B6 crystallizes by a eutectic

GAO and CANTOR:

OXIDATION

BEHAVIOUR

OF AN AMORPHOUS

ALLOY

175

Fig. 13. TEM micrographs of unoxidized pre-crystallized Fe,,,Ni,P,,B, foils. (a) Bright field image; (b) selected area diffraction pattern.

Fig. 12. TEM micrographs of amorphous Fe,NiJ’,,B, foils oxidized at 300°C for (a) 0.5 h and (b) 50 h. (c) Selected area diffraction pattern for (b). mechanism to form an alternating two phase lamellar structure of f.c.c. (FeNi) y-austenite and b.c.t.

(FeNi), (PB). In the present experiments, the lamellae coarsened because of the higher annealing temperature and longer annealing time which were used

(SOOC, 30min). Figure 13(a) shows the typical resulting two-phase structure of pre-crystallized Fe40Ni4$,4B6, which consisted of small particles of the f.c.c. y-austenite phase, about 10-30 nm in size, embedded in a matrix of b.c.t. (FeNi),( Typical X-ray microanalysis spectra of the two phases are shown in Fig. 15 and the resulting composition analyses are given in Table 6. The discrete f.c.c. y-phase particles contained a high concentration of Fe, while the matrix phase was depleted in Fe and contained a high concentration of Ni and P. As shown in Figs 13(a) and 14(a), very little change was detected in the transmission electron micrographs from pre-crystallized FeMNiMP,,B, thin foils

Fig. 14. TEM micrographs of pre-crystallized Fe,,,Ni,$,,B6 foils oxidized at 300°C for 240 h (a). (b) Selected area diffraction pattern for (a).

GAO and CANTOR:

176

TN-5500 Cursor:

OXIDATION.BEHAVIOUR

OF AN AMORPHOUS ALLOY

CHEMICAL CRYSYALLO~AP~Y OXFORD TU2 26-OCt-66 0.000 KeV =0 ROS (2) .?.mo: 7.700

11*x, If

,..__, _.I,. I/

.,_,_..__ _-...._I I

_-

.-..

-1..-

T

bet

(a) TN-S!500

-

CHEhWAL

cursor : 0.000

VFS = 64

10.240

CRYSTALLOGMPHY OXFORO WE 28-OCT-66 12) 7.280: 7.700 ROI

tl.35

kSV = 0

VFS = 64

$0.241

(b) Fig. 15. Typical electron probe X-ray microanalysis spectra from the two phases in pre-crystallized

Fe,Ni$P,,B6 foils. (a) From b.c.t. phase; (b) from f&c. phase. even after 240 h oxidation at 300°C when compared with the same sample before oxidation. However, the diffraction patterns of pre-crystallized FewNi4$,,B, after 240 h oxidation at 300°C contained three continuous rings which did not exist before oxidation [Fig. 14(b)]. These extra three rings corresponded very well to the three strongest lines of NiO as can be seen in Table ‘7,which indicated that a layer of very fine nickel oxide (grain size c 100A) had formed on the surface of the pre-crystallized alloy. Wavelength dispersive probe ~croanal~is was used to measure iron concentration profiles through the alloy/oxide interface. Typical results are shown in Figs 16(a) and (b) for taper cross section samples of amorphous and precrystallized Fe,N&,P,,B6 after 700 h oxidation at 3OO”C, and for ~o~ho~ Fe,MNi,,,P,,B6after 80 h oxidation at 350°C reTable

6.

Microanalysis of the Fe~NLP,.B,t

Pbnsb.c.L phase f.c.c. phase

two phases kt.K)

ia

crystallii

Fe

Ni

P

37-40 H-62

45-48 29-36

15-18 9-10

tEight point analyses for each phase; results for the small f.c.c. particles may have been affected by the matrix; boron cannot be measmed.

spectively. As shown in Fig. 16(a), the arno~ho~ alloy was depleted in iron over a region of approx. 15 pm from the alloy/oxide interface. However, there was no equivalent depletion of iron in, the case of pre-crystallized Fe~Ni,,,P,,B,. A greater thickness of Fe depletion was detected in the case of amo~hous FemN&,PlrB6after 80 h oxidation at 350°C as shown in Fig. 16(b). 4. DISCUSSION 4.1. Uxidafion kinetics Figures 1 and 2 shows that the oxidation behaviour of Fel(;Nid,,B, is strongly affected by structural changes in the alloy. At relatively low temperatures in the range 20%3OO”C,the oxidation of amorphous FeMNi4$,,B6obeys a parabolic rate law, since during oxidation annealing at these temperatures the alloy remains amorphous. At relatively high temperatures such as 450 and SWC, crystallization takes place almost imm~ia~iy at the canning of the oxidation anneal, and the oxidation reaction again obeys a parabolic rate law. However, there are significant variations in oxidation kinetics during the annealing at intermediate temperatures such as 350 and 400°C. At these inte~ediate temperatures, c~talli~tion is

GAO and CANTOR:

OXIDATION

BEHAVIOUR

Table 7. The calculated d,spacin.gs

R (mm)

d(A)

7.9 9.2 13.0

2.42 2.08

DISTANCE

-I 8B.B (urn)

2.88

(b) 3%c. m

Fig. 16. Typical profiles of the ratio of Fe to Ni measured by electron probe microanalysis in the taper cross section samples for amorphous (A) and pre-crystallized (C) Fe*Ni,P,,B, after 700 h oxidation at 300°C (a), and for amorphous Fe,Ni,$,,B, after 80 h oxidation at 350°C (b). The thickness r shown here is about l/20 of the true thickness due to the taper section.

A.M. 3611-L

three rings ih Fig.

60% 100% 70%

taking place relatively slowly, after the amorphous alloy has already oxidized to a certain extent, and the crystallization process clearly interferes with the otherwise steady growth of the oxide film. The three stages of oxidation at 400°C can be explained as follows, as a consequence of simultaneous crystallization. The first stage of oxidation up to 0.1 h of annealing represents the oxidation behaviour of amorphous Fe40Ni40P,4B6,similar to that found at the lower temperatures of 200-300°C. The third stage represents oxidation of crystallized FeMNi,,,,P,4B6r similar to that found at the higher temperatures of 450 and 500°C. The second intermediate stage is a period of transition. The alloy has already crystallized but the oxide is still characteristic of that formed on the amorphous alloy. For reasons discussed further below, this causes the oxidation rate to decrease sharply and remain at a very low level for some period of oxidation time. The oxidation kinetics

48.0

tht

Relative intensity

1.47

,

!Wii

OF AN AMORPHOUS

ALLOY

177

14(b) W0

101 012 110.104

of amorphous FeaNiMP,,B, over the whole temperature range of 200-500°C can be explained on the same basis as that just outlined for oxidation at 400°C. This is shown schematically in Fig. 17. At all temperatures, there are the same three stages in oxidation of the amorphous alloy, but there is only a limited region of accessible experimental results, as shown by the rectangular box in Fig. 17. The three oxidation stages are only seen clearly at 400°C. The parabolic rate law found in fully amorphous and fully crystalline Fe,N&P,,B, suggests that the oxidation reaction obeys Wagner’s mechanism, i.e. the rate controlling process is the diffusion of ions through the oxide layer brought about by a concentration gradient across the oxide layer [ll]. For a thermally activated process like diffusion, the parabolic rate constant K is given by an Arrhenius equation: K = K,,exp(-Q/RT)

(2)

where Q is the activation energy for the diffusion process, T is the absolute temperature, R is the gas constant and K, is a constant. Figure 18 shows In K vs inverse T for the oxidation of Fe,Ni,J’rpB,. Three sets of data are included in Fig. 18. Curve A represents the In K - l/T variation for amorphous Fe40Ni,$,4B6, using the full range of data in Fig. 1 at 20@-300°C but taking only the first stage of oxidation mass gain at 350 and 400°C. Curve Cl represents the In K - l/T variation for amorphous Fe,,,Ni,$,,B, specimens which crystallized during oxidation annealing, using the full range of data in Fig. 1 at 450 and 5OO”C,but taking only the last stage of oxidation at 350 and 400°C. Curve C2 represents the In K - l/T variation for pre-crystallized specimens of FeMNi.,r,P,4B,. As shown in Fig. 18, C, and C, give very similar results, confirming the interpretation given above of the amorphous alloy oxidation behaviour at 450 and 500°C and in the third stage at 400°C. Activation energies for the three curves in Fig. 18 were calculated by using linear regression analysis. The results are given in Table 8 together with other relevant diffusion data from the literature. As shown in Table 8, the measured activation energy for oxidation of crystallized Fe@N&,P,.,B, is close to the activation energies for diffusion of Fe ions in Fe,O, and Ni ions in NiO. This is in good agreement with the parabolic kinetics of oxidation of crystallized Fe.,,,Ni,P,,B, shown in Fig. 1, and confirms that oxidation takes place by Wagner’s mechanism with diffusion of metal ions (Fe and Ni) through the oxide layer (Fe,O, and NiO) as the rate controlling process of the oxidation reaction.

178

GAO and CANTOR

OXIDATION

BEHAVIOUR

OF AN AMORPHOUS

ALLOY

Lnt Fig. 17. Schematic lny-lnt diagram for the general oxidation kinetics of Fe,Ni,P,,B, over the temperature range of 2OWOO”C. The rectangular box shows the accessible area for obtaining experimental results.

The results are quite different for oxidation of amorphous Fe,Ni,,,P,,B,, which has an oxide layer consisting mainly of a-Fe,O,. As shown in Table 8, the measured activation energy for oxidation of amorphous Fe40Ni,P,,B6 is much smaller than the activation energy for diffusion of Fe ions in Fe*O,. This can be explained as follows. Whisker-like oxide growth takes place when a fast diffusion mechanism can operate in some of the oxide grains to feed the rapid advance of the whisker tip (8, 141. Unfortunately, there is very little current understanding of how this fast diffusion takes place. However, fast diffusion in the oxide implies that the oxidation reaction will become limited by the rate of supply of metal ions into the oxide at the oxide/metal interface. An iron depletion layer builds up in the underlying amorphous alloy, as shown in Fig. 16, and the parabolic reaction rate constant has an activation energy as shown in Table 8 near to that measured previously [13] for diffusion of iron in amorphous

-4

1 1.2

I 1.4

Fe,Ni,,,P,,B,.

The

coefficient D can be from the depletion layer

diffusion

calculated approximately thickness z using z=zfi

(3)

From Fig. 16(a), z z 0.5 pm for r = 2.5 x lo6 s at 3OO”C, and equation (3) gives D = 2.5 x 10-20m2/s. This agrees reasonably with the value of 3 x lo-*’ m*/s calculated from the previous measurements [ 131 of activation energy and frequency factor in Table 8 for diffusion of iron in amorphous Pe,Ni,P,,B,. 4.2. Oxide growth mechanism Oxidation of amorphous Fe,r,N@,,B6 takes place by rapid growth of or-Fe20, whiskers, However, the crystalline alloy oxidizes much more slowly by forming a compact microcrystalline multiphase oxide layer of Fe,O,, NiO, Fe,O, and NiFe,O,. It seems difficult to explain this strong effect of alloy structure, when oxidation is a chemical reaction and the amor-

I 1.6 l/T

I 1.8 K”

I

2.0

: 2

Fig. 18. The Arrhenius variation of In parabolic rate constant vs inverse temperature (ln K - l/T) for amorphous (A), oxidationcrystalli (C,) and pre-crystallixed (Cd Fe,Ni,$,,B,.

GAO and CANTOR:

OXIDATION BEHAVIOUR OF AN AMORPHOUS ALLOY

Table 8. Activation energies for oxidation of amorphous and crystallized Fe&NhP,,Bs diffusion data obtained from the literature 112, 131 Temperature range (“C)

Diffusion ions

Diffusion matrix

Fe Fe + Ni

a-Fe.0, 1 .’ Fe,O,, WO, Fe,Ol, NiFc,O,, Fe,Oj,NiFe,O,, FeO,.OU

Fe+Ni Fe

Fe,‘& Fez03 NiFe*O, NiO Fe 40Ni 40P IIB6

Ni Fe

2tIO-400 400-500 350-500 69C-1010 750-loo0 930-1270 850-I 190 740-1400 268-344

phous and crystalline alloy have identical chemical composition. However, an explanation for the strong effect of microstructure on oxide growth can be given as follows. Amorphous FeMN&P14B6 has a very uniform chemical composition which can supply a reasonable concentration of Fe continuously to feed the rapid

I_I b)

3

PI

e)

Frequency factor (m’ls)

ayst-

d)

Fig. 19. Schematic oxide growth on amorphous (a + b + c -) d), oxidation-crystallized (a + b --*c + e) and pre-crystallized (f + g) Fe,Ni,J,,B,, (h) shows the oxide growth near the edges of amorphous thin foil specimens.

1.4 X 10-e 5.2 x IO-’ 4.0 5.2 x IO-’ 1.7 X 10-6 1.0 X lo-’

179

together with other relevant

Activation energy (kJ/mol)

Correlation coefficient

Source

142 226

0.997 0.999

Curve A Curve C,

209 126.4 230 469 298 234 192

0.997

Curve Cl Ref. 1121 Ref. (121 Ref. (131

whisker-like Fe,O, growth. Whisker-like oxide grows on the surface, and builds up to form a thick layer of a-Fe,O,. Thus, amorphous Fe,Ni,P,,B, has a very high oxidation mass gain. On the other hand, crystalline Fe,Ni,P,,B, consists of a two-phase structure with discrete particles of f.c.c. (FeNi) y austenite embedded in a matrix b.c.t. (FeNi),PB phase. The f.c.c. y phase contains a high concentration of Fe, while the matrix is rich in Ni and dilute in Fe, as shown in Table 6. During oxidation annealing, therefore, a layer of small grains of NiO grows on the surface of the matrix, together with a certain amount of iron oxide. However, the fast growing whisker-like Fe,O, and a subsequent thick layer of Fe*O, cannot form because the formation mechanism requires a higher concentration of Fe than is present in the matrix. The small NiO and other oxide grains make a compact oxide layer which grow relatively slowly and can protect the metal effectively from further oxidation. Unlike the matrix phase, the f.c.c. y-phase does contain a high concentration of iron, but is in the form of discrete particles with a size from 10 to 30 nm. Small grains of iron oxide (Fe,O, and Fe,O,) grow on the y particle surfaces. Once again, however, no whisker-like Fe,O, oxide can form because the discrete y-phase particles become rapidly depleted in iron. A similar effect also happens near the edges of amorphous thin foil specimens, where no whisker-like oxide structure is found even after 240 h oxidation at 300°C as shown in Fig. 12(a). Thus, crystalline Fe40Ni1$,4B6, has a much lower oxidation reaction rate than the amorphous alloy. During oxidation of amorphous FeMNi,P,,B, at 350 and 4OO”C, there is a transition period of exceptionally slow oxidation. The mechanism of oxide growth in this transition period must be quite complicated. Since a thick layer of Fe,O, has already developed on the amorphous sample during the first stage of oxidation, there is an amorphous alloy region adjacent to the oxide layer which is depleted in Fe as shown in Fig. 16(b). When crystallization takes place in this depleted region, the crystallization reaction as well as crystallized structure is likely to be different from those in the unoxidized samples, probably with little or no formation of f.c.c. (FeNi) y-phase. This inhibits not only the fast growing

180

GAO and CANTOR:

OXIDATION BEHAVIOUR OF AN AMORPHOUS ALLOY

whisker-like iron oxide but also other iron oxides. The oxidation reaction is dominated by formation of NiO and the oxidation rate becomes exceptionally low. Figure 19 shows schematically the different types of oxide growth discussed above.

5. CONCLUSIONS From the present investigation, clusions can be drawn: 1. Oxidation

the following con-

of amorphous Fe,Ni,P,*B6 in air rate law over the temperature range 2oo-4oO”C as long as no crystallization takes place. The parabolic rate constant for amorphous Fe&&P1,B6 increases with the oxidation temperature according to an Arrhenius law, with an activation energy of 142 KJ/mol. 2. Oxidation of amorphous Fe,Ni.,$,,B, takes place by rapid growth of whisker-like a-F%Or and a subsequent build up of a thick Fe,O, layer. The rate controlling process for oxidation of amorphous FeaNi,,,P,,B6 is the diffusion of iron ions in the amorphous alloy matrix. 3. ~dation of crystalline Fe,Ni,P,,B, in air also obeys a parabolic rate law over the temperature range 350-500°C. The parabolic rate constant increases with the oxidation temperature and again obeys an Arrhenius law, with a higher activation energy of 209 KJ~mol. 4. Oxidation of crystallized Fe,Ni,P,,B, takes place by relatively slow growth of small grains of NiO, Fe,O,, FezOl and NiFe,Ol, to form a thin, compact multiphase oxide layer. The oxidation rate is much smaller than that of amorphous FeMNi,P,,B6. The rate controlling process for oxidation of crystalline Fe,Ni,P,,B, is the diffusion of iron and nickel ions in the multiphase oxide layer. 5. The oxidation rate drops sharply when crystallization of amorphous Fe~Ni~P,~B~ takes place during oxidation annealing. After crystallization, there is a period of time during which the oxidation

obeys a parabolic

rate is ex~ptional~y low. This effect is particularly significant during oxidation at 350 and 400°C. 6. The different oxidation kinetics and oxide growth mechanisms for amorphous and crystallized Fe4(;Ni,P,,B6 are caused by the different alloy microstructures. The two-phase microstructu~ of crystalline Fe,NiaPI,B6 consists of discrete particles of Fe-enriched f.c.c. (FeNi) y austenite embedded in a matrix of N&enriched b.c.t. (FeNi),( This microstructure prevents rapid formation of whisker-like a-Fe*O, and the subsequent thick oxide layer of Fe,O, which takes place on amorphous Fe,Ni40P&,. Acknowledgemenrs-We would like to thank Dr L. A. Davis of Allied Chemicals for supplying the amorphous ribbon. We would also like to thank Professors Sir Peter Hirsch and J. W. Christian for provision of laboratory facilities. One of the authors (GW) would like to thank the Royal Society, British Council and the Education Committee of China for financial support of this research programme.

REFERENCES 1. M. G. Scott, Amorphous Metallic Alloys, (edited by F. E. Luborsky), pp. 144-168. Buttenvorths, London (1983). 2. T, Watanabe and M. Scott, J. Muter. Sci. 15, 1131 (1980). 3. J. L. Walter et al., Mefall. Trans. $A, 1141 (1977). 4. D. G. Morris, Acta metali. 29, 1213 (1981). 5. K. Hashimoto, Rapidly Quenched Metals Y, (edited by

S. Steeb and H. Warlimont), Vol. 2, pp. 1449-1456. Elsevier, Amsterdam (1985). 6. Gao Wei and B, Cantor, Proc. 1stInternationalWork-

shop in con-Crystalline So&&, Spain (1986). 7. 0. Kubaschewski and B. E. Hokins, oxidation of Mefats and Alloys, 2nd edn, p. 166. Buttenvorths, London (1962). 8. 0. Kubaschewski and B. E. Hopkins, ibid, p. 55. 9. J. Paidassi, Acta metalf. 6, 718 (1958). 10. Gao Wei and Zhao Xiancun. Beiiina Central Iron and Steel Research Institute Tech&a; Bulletin, No. 4 (1983). 11. N. Birks and G. H. Meier, Inrroductionto High Temperature Oxidation of Metals, p. 42. Edward Arnold, London (1983). 12. 0. Ku~~ewski and B. E. Hopkins, ibid. p. 33. 13. P. Valenta er al., Phys. Stat. Sol. 105,537 (f981), 106, 129 (1981). 14. H. K. Hardy,_ Prog. - Metal Phys. 6, 45 (1956).