On the thermal oxidation kinetics of two metallic glasses

Corrosion Science, Vol. 22, No. 2, pp. 135-146. 1982

0010-938X/82/020135-11 $03.00/0

Printed in Great Britain.

ON

THE

Pergamon Press Ltd.

THERMAL OXIDATION KINETICS TWO METALLIC GLASSES*

OF

OLA HUNDERI a n d ROALD BERGERSEN Department of Physical Metallurgy, Norwegian Institute of Technology, Trondheim, Norway Abstract---The oxidation kinetics of two metallic glasses have been studied by ellipsometry and Auger Electron spectroscopy The presence of chromium in Fe-based alloys has a marked influence on the corrosion film formed and on the kinetics of film formation. The chromium-free alloy shows piecemeal parabolic oxide growth; this is transformed into a logarithmic growth law for the chromium concentration studied. The influence of moisture on the kinetics of film formation has also been studied. INTRODUCTION

SEVERAL authors have reported that amorphous iron alloys containing more than 8 ~o chromium exhibit extremely high corrosion resistance in neutral and acid solutions. Assami eta[. have shown that the passive protective film in solution is mainly composed of CrOx(OH)z 2.,. nH20.1 The high corrosion resistance is only partly related to the properties of the passive film since the composition of the film on the corresponding crystalline alloys is essentially the same as on the amorphous alloys. The crystalline alloys show considerable corrosion. XPS studies show that the concentration of hydrated chromium oxy-hydroxide in the passive film and the values of x and n are somewhat different fl'om that on the crystalline iron-chromium alloys and 18-8 stainless steels. In general, the concentrations of OH and H20 in the passive film are also higher than in the crystalline stainless steels. These effectively capture metal ions in the tilm, thereby increasing the passive film thickness. They also make the passive film more ductile, thereby preventing mechanical film breakdown. The high corrosion resistance of the anaorphous alloys is therefore partly due to the properties of the passive film and partly due to the low density of crystal defects in the amorphous alloy -no grain boundaries, scgregates, etc. The medium and high temperature atmospheric oxidation of these amorphous alloys have, however, never been studied, and it is the aim of the present investigation to study the oxidation kinetics of these glasses anti to see how these are influenced by the presence of chromium in the metallic glass. EXPERIMENTAL METHOD The kinetics of film growth were measured by ellipsometry. With such instrumentation, it is possible to observe changes in film thickness --~ 0.1,~ when measured in situ and 1-2,~ changes when measured otherwise."- The amorphous metals used in this investigation were Allied Chemical Metglas 2826 and 2862A, with nomimal composition Fe~uNi,oPt2B, and Fea~Ni:~,~Crl~P~.BGrespectively. The amorphous ribbons were mechanically polished, thoroughly cleaned and the ellipsometric parameters immediately measured? The "clean" sample is thus covered by roughly a monolayer of oxide, as was verified by Auger electron spectroscopy. The samples were then placed in a temperaturestabilized furnace and the ellipsometric parameters were again measured after various exposures to °Manuscript received 10 April 1981; in revised form I June 1981. 135

136

OLA HUNDERI and ROALD BERGERSEN

the high-temperature atmosphere. At each temperature, 5 parallels were measured to improve the precision of the e x s i t u technique employed. This gave precision in the thickness determination of a few per cent, which is more than sufficient for the present investigation. Since the high corrosion resistance of the chromium-containing alloy has been attributed to the formation of the oxy-hydroxide, we have also studied the corrosion kinetics in a moist atmosphere. Water vapour was introduced into the furnace atmosphere from the vapour over boiling water connected to the furnace through a tube. This is a rather unsatisfactory manner of controlling the water vapour pressure, and we plan to repeat this part of the experiment under better controlled conditions. The experimental results were, however, quite reproducible, and as such the crude method served our purpose. After exposure for 48 h, the surface films of a representative selection of samples were studied by Auger electron spectroscopy. The instrument used was a Varian Automated Auger Microprobe. Concentration profiles of elements of interest were taken, and the thickness of the film was determined approximately from the sputtering times necessary to remove the reaction products. Although the sputtering rates are only approximately known, the two methods yield values for the thickness which are in satisfactory agreement. EXPERIMENTAL

RESULTS

Film thickness determination It is not possible to determine the film thickness and complex refractive index from the measurement of one set of ellipsometric parameters alone. Knowledge of the eUipsometric parameters for two values of the film thickness is, however, sufficient to determine both the refractive index of the film and the two thickness values, provided that the refractive index of the clean surface is measured and that films do not change with thickness. The value of the film refractive index is most conveniently solved graphically by plotting the contours of the index satisfying experimental data at one value of the film thickness. This is exemplified in Fig. 1 which shows contours of film dielectric constants satisfying measured data for five values of the film thickness. All curves intersect in one point which thus uniquely determine the dielectric constant of the corrosion film. The film thickness can be read off as a parameter along the curves (not shown in the graphs). These countours represent experimental data for the alloy 2826A exposed to dry air at 25ff~C. The measured ellipsometric parameters

I

5

! 0

FIG. 1.

1 1

I

I

I 2

EO 2

Contours of dielectric constants satisfying measured ellipsometric parameters after various times exposure.

On the thermal oxidation kinetics of Iwo mcmltic glasses

137

were first plotted against time and then smoothed by a non-linear least-square fitting procedure to minimize the effect of scatter in the raw data on the refractive index determination. The various contours in Fig. 1 satisfy these smoothed data. The value of the dielectric constant so obtained is 6.2 + i • 1.85. The dielectric constant of the film on pure Fe in pure O,., has been determined by Kruger and Yolken 4 and was found to be 6.15 -I i • t.5. The alloy 2826 at high temperatures did not give a unique value of the film dielectric constant for temperatures above 150~C. In these cases, the constant determined at 150'C was used in the film thickness determination. This does not introduce serious errors in the thickness determination. The thickness of the corrosion fihn was determined as outlined above as a function of time at 100, 150, 200 and 250°C, for the two alloys 2826 and 2826A. The experimental results are summarized in Figs. 2-5. Figure 2 shows the thickness of the film on alloy 2826 in dry air against time at the temperatures indicated, Fig. 3 is the corresponding curve in moist air. Figures 4 and 5 show the corresponding curves for the alloy 2826A. The kinetics of film growth are not the same for the two alloys. The variation in film thickness with time on alloy 2826A shows a direct logarithmic law which in a moist atmosphere is transformed into a rather unusual asymptotic growth. The film thickness on alloy 2826 shows piecemeal parabolic growth. This is shown in Fig. 6. Piecemeal parabolic growth is a well known phenomenon and is generally caused by film fracture. Film fracture is indeed found in micrographs of the sample surface as illustrated in Fig. 7. Alloy 2826A in Fig. 7(a) shows a homogeneous film, alloy 2826 in Fig. 7(b) shows film fracture. Usually, however, fracture leads to a successive increase in the slopes of the straight sections in Fig. 6. The cracking must therefore be accompanied by some blocking mechanism leading to a reduction of the effective area where growth can occur. dial -o

35e

3oc

25G

20C _

150:

1001

i /

°I

F x c , 2.

i

I

G

10~

"

100°¢_ Z_

,

[

,

GO

102

500

,L 1"--

10

, 5000

t i m e [mi nl

Film thickness on alloy 2826 in dry air against time for various temperatures.

138

OLA HUNDERI and ROALD BERGERSEN

'

I

~

l

I

'

/ zsQ'c

300

200

100

I

5

FIG. 3.

10 +

5'0

I

10 2

I

I 500

,

10 s

5000 time[mini

Film thickness on alloy 2826 in moist air against time for various temperatures.

'

i

'

I

'

I

. j ,

70 60 50 40

30 20 10 0

FIG. 4.

T

l

I

I

,

I

5

10;

50

102

500

103

,

5000 t i m e [ml n'~ d

Film thickness on alloy 2826A in dry air against time for various temperatures.

On the thermal oxidation kinetics of two metallic glasses

'

1

'

I

,

139

f

v

7~

6~

2

5

0

~

50

40

30

20

10

0

Fit]. 5.

,

I

S

101

L, 50

I

,

I

102

SO0

103

, 5000 time [mi.]

Film thickness on alloy 2826A in moist air against time for various temperatures.

dI

160000 Air 140 000 2SOeC-Mellt Air 120 000

100 000

BOO00

Air

- Melst Air

a 300

I 600

FIG. 6.

! ioo

I 12oo

I lsoo

I 10oo

I lloo

I 14oo

I 17oo

Film thickness squared on alloy 2826 against time.

I sooo

I t i m e F w~,- - ji . 1

140

OLA HUNDERI and ROALD BERGERSEN

Figures 2-5 shows that there is a dramatic difference between the corrosion rates of the two alloys. The addition of chromium has reduced the film thickness by a factor of typically 5 after 48 h; most importantly, it has changed the kinetics from a slowly growing logarithmic film thickness into an asymptotic growth when moisture is present.

Auger profiles The composition of the corrosion film was studied by Auger electron spectroscopy. After 48 h exposure, a representative number of samples were studied in a Varian Automated Auger Spectrometer. Profiles of the elements Fe, Ni, Cr, O, C and P were taken. By a mistake boron was not included in the profile, but this is of no consequence to the subsequent discussion. Some results of this profiling are shown in Figs. 8-10. Figure 8 shows the profile in alloy 2826 exposed to dry air at 150°C for 48 h. The film consists mainly of an iron oxide; from the values of the dielectric constant of the film, it appears to be mainly the Fe304 phase. This is also consistent with the Auger data using standard sensitivity factors. It seems clear that the Fe content is higher than in y-Fe203, although FezO3 cannot be completely ruled out from standard sensitivity factors alone. The optical can, however, rule out the ~-Fe~Oz phase since its dielectric constant is considerably higher. The profiles are broadened by some 20/~ due to escape depth and knock-on mixing, and it is, therefore, difficult to draw any conclusions as to the sharpness of the interface. By mentally deconvoluting the observed profiles, it appears, however, that the Ni content in the oxide is low. By further profiling, an enrichment of P and Ni just at and below the metal/film interface is found, as might be expected. Figures 9 and 10 show the corresponding profiles for the alloy 2826A in dry and humid air respectively. The outer part of the film consists mainly of an iron oxide while the film near the metal is rich in chromium. Unfortunately the broadening

Ni

D

lllllllll[ll till IIJ[llllilll o ~ o 2

ti I I I I l l l t l [ l l l l l 03 o.a

III1[11 1 1 1 1 II I[111111 I I l [ l l l l l l l l 0.5 06 0 7

08

Sputtering time ~min]

Fro, 8.

Auger profiles for alloy 2826 after 48 h exposure in dry air at 150°C,

FI(3. 7.

Micrographs of the sample surface after 48 h exposure at 150~C in dry air (a) alloy 2826A, (b) alloy 2826.

O n the thermal oxidation kinetics of two metallic glasses

143

,Ni O

,Fe

i

~c~

V ~ \

tt~}lll

~fll~l[~IIlllltlllll]li~lj[Itlllt~I~]111~

O.L

0.2

03

04

05

06

07

0.8

0.9

1.0

Sputtering time [mini Fte. 9.

A u g e r profiles for alloy 2826 after 48 h exposure in dry air at 150~C.

Ni O

-"-.,P

/ / ' V ~

c>,,,\

c

C .........

0

I I.

'

'

O

. . . . . . . . . . . . . . . . . . . . . . . .

02

..

0.3

i

........

0 4

0. 1o,

.......

o, 6

Sputtering time [mini FIG. 10.

A u g e r profiles for alloy 2826A after 48 h exposure in moist air at 150°C.

of the profile renders a further description of the film composition impossible. By comparison of the two profiles, however, it can be seen that the chromium content in the sample exposed to humid atmosphere is higher than in the corresponding film lbrmed in dry air. DISCUSSION

The simplest kinetics are found for the chromium containing alloy in dry air. By inspection of Fig. 4 the film thickness follows the relation d--- K l l n t + C~.

(1)

144

OLA HUNDERI and ROALD BERGERSEN

This direct logarithmic law is characteristic for the oxidation of a number of metals below 300-400°C. Several theories based on various rate-determiningmechanisms have been put forward to account for the logarithmic laws. 5 The different logarithmic rate equations often prove difficult to distinguish. The parameters in the rate equations are not easily accessible by independent means, and this complicates the interpretation. The rate constant K1 is not strongly temperature-dependent showing little change at all from 150° to 200°C. This is consistent with the model where electron transport via tunneling is rate controlling. This leads, as shown by Mott, 5 to a direct logarithmic rate law with the rate constant Ka independent of temperature. In our experiments, Ka increases from 2.7A at 100°C to 7.0 A at 250°C. Compared to what would be predicted on the basis of a diffusion-controlled mechanism with activation energy of (say) 100 kJ mol -~, this is, for all practical purposes constant, as predicted on the basis of the theory of Mott. When plotting the film thickness squared against time for alloy 2826, a curve consisting of several linear sections is observed, as shown in Fig. 6. The film thicknesses on many metals are found to follow a parabolic time dependence (2)

d 2 = Kpt + C o

As a rule, high temperature parabolic oxidation signifies that a thermal diffusion process is rate determining. When plotting the logarithm of the rate constant Kp for the first straight section (uncracked film) in Fig. 6 against I/T, an activation energy of 81 kJ tool -1 is found as shown on Fig. 11. The activation energy for diffusion of Fe in FesO4 is 180 kJ mol -x.

K.I 8

6

"

!

2

FIG. 11.

_1

~2

I

I

I

[

2h

2,6

~e

3,0

100^/.rK_, 1 ~'L

J

T h e logarithm o f the parabolic rate c o n s t a n t plotted against inverse temperature.

On the thermal oxidation kinetics of two metallic glasses

145

The observed activation energy is lower than values given in the literature. However, the number of defects in the film is high and the temperature is rather low. The diffusion is therefore perhaps predominantly along easy paths and the activation energy for such diffusion is typically half the normal value. The presence of water vapour does not have any marked influence on the film formation on alloy 2826. The kinetics still show a piecemeal parabolic behaviour, the activation energy being about the same as in dry air. For the chromium-containing alloy, the presence of water in the atmosphere changes the direct logarithmic kinetics into rather unusual asymptotic kinetics. The asymptotic kinetics are often found to follow a relation of the form d

::

C. - Kae .t

13)

Our data however can be fitted to a relation of the form d = C a .... Kae_, I.t

(4)

if the temperature T > 100~C. A least square fit of this relation to the experimental data is shown in Fig. 12. The value of the limiting thickness so obtained is for obvious reasons rather uncertain but the fits should still yield values of the correct magnitude. For T = 100°C, the thickness still follows a direct logarithmic law. It appears that at 100°C and below, at the vapour pressure used in the experiment, the formation of the chromium oxy-hydroxide is too slow compared to the formation rate of other components in the film, to build up a dense protective layer. At higher temperatures, the oxy-hydroxide formation is fast enough in relation to competing formation processes to build up a protective film near the metal/film interface. The kinetics

120 I10 I00 90 80 70 60 50 40 30 20

]O-

_ _ iii ~

150 °C.....e~ ,......e

k '~'" io °

I0 ~

I 10 2

III

•

I 10 3

I IO 4

L

1

l

I 0 ';

10 6

IO 7

T i m e Emin]

FIG, 12, Non-linear least square fit of equation (4) to the data in Fig. 5.

I0 ~

146

OLA HUNDERIand ROALD BERGERSEN

observed are therefore the result of a complicated interplay between competing processes. From the Auger profiles it can be concluded that the outward diffusion of iron and inward diffusion of oxygen are both taking place. In all cases a nearly pure iron oxide in the outer layer is observed formed by the outward diffusion of Fe ions and a chromium/oxy-hydroxide near the metal formed by the inward diffusion of 0 2- and OH±-ions. SUMMARY

The presence of chromium has a marked influence on the oxidation kinetics of Fe-based metallic glasses. It prevents surface film fracture and transforms the parabolic oxide growth law observed in the chromium-free alloy into a logarithmic growth law in dry air and into an asymptotic law through the formation of a passive oxyhydroxide film near the metal/film interface. The temperature independence of the rate constants in the logarithmic law is consistent withtheelectron tunnelling model of Mutt while the rate-limiting mechanism in the film formation on the chromium-free alloy is defect-assisted diffusion.

1. 2. 3. 4. 5.

REFERENCES K. ASAMI,K. HASHIMOTO,T. MASAMUTOand S. SHIMODAIRA,Corrosion Sci. 12, 909 (1976). J. KRUGER, Adv. Electrochem. 9, 227 (1973). O. HUNDERI, Appl. Opt. 16, 3012 (1977). J. KRUGER and H. YOLKEN, Corrosion 20, 29t (1964). U. R. EVANS, The Corrosion and Oxidation of Metals, E. Arnold, London (1960).