Surface oxidation behaviour of amorphous and crystallized Fe40Ni40P16B4 alloys

Journal of Non-Crystalline North-Holland, Amsterdam

SURFACE OXIDATION AND CRYSTALLIZED Jean FUSY

Solids 89 (1987)

BEHAVIOUR Fe,Ni,P,,B,

131

OF AMORPHOUS ALLOYS

and Pierre PARPJA

CNRS, Luhomroire Maurice L.&m 54600 Villers les Nmq? France Received

131-142

10 June

associ.! 6 I’lJnioersitP

de Nancy

I, BP 104.

1986

In situ oxidation at low pressure of amorphous and crystallized Fe,Ni,P,,B, was investigated using AES and XPS. The influence of crystaIIinity on oxidation was not clearly demonstrated because the initial superficial content of phosphorus inhibits the oxygen up&&e. The main difference in the composition of the oxidized layer lies in the presence of nickel for crystallized samples.

1. Introduction The chemical properties of amorphous alloys that have been studied are mainly concentrated in the field of corrosion which was reviewed recently [l-4]. Other chemical properties such as catalysis were also studied and various rapidly quenched alloys were found active catalysts particularly for the Fischer-Tropsch reaction of CO + Hz. By comparing systematically the activities of amorphous Fe-Ni-base alloys with their crystalline counterparts. Yokoyama et al. [5,6] demonstrated that except for F%,Ni,P,,, amorphous ribbons are significantly more active than the crystallized ribbon of identical composition. A similar result was reported by Kisfaludi et al. [7] for a FeB amorphous ribbon used in the same reaction. Related to this, interaction of carbon monoxide with the surface of several glassy alloys including FeB [8], NiZr [9,10], NiBSi and FeNiMoB [ll] was studied using electron spectroscopy. Surface oxidation of Metglas 2826A (Fe,,Ni,,Cr,,P,,$) was studied by Baer and Thomas [12-141 who showed the ambiguous role of phosphorus which lies at the alloy-oxide layer interface. A comparative study of surface oxidation of amorphous and crystallized Metglas 2826A and of Fe,,B,, Si, alloys [15-161 showed a different behavior of the two samples especially as far as oxygen uptake is concerned. The goal of this work was to investigate the possible change in composition and reactivity of the surface due to in situ crystallization. The sample chosen in this study was Fe,Ni,P,,B,, a composition for which a stable elevated activity was reported at 320°C as compared with the corresponding crystallized state [5,6]. 0022-3093/87/$03.50 0 Elsevier Science Publishers (North-Holland Physics Publishing Division)

B.V

132

J. Fuy,

P. Porejo / Surface

oxidorion

hehouiour

of Fe,, Ni 1o P, n B, allqa

The test for surface reactivity was the oxygen uptake. Special attention was paid to the effect of phosphorus in the upper most layer so as to distinguish between the influence of crystallization and that of metalloid segregation. 2. Experimental The amorphous alloy used in this study was Fe,Ni,P,,$, manufactured by Allied Chemical Corporation and sold under the trade name Metglas 2826. They were obtained as ribbons 20~ thick and cut into strips of 13 x 6.5 mm2. The samples were cleaned first with acetone in an ultrasonic cleaner and then rinsed with alcohol and dried before being introduced in an UHV chamber (vacuum - 10-l’ Torr) for Auger and XPS analysis. The temperature of the sample can be varied from room temperature to 600°C. Various gases can be introduced into the chamber through an adjustable leaking valve, particularly oxygen used for in situ surface oxidation. An argon ion gun was operated at 3 kV, O-100 mA, for cleaning or depth profiling analysis. Auger spectra were recorded in A E/E = const. mode using a 2 V peak-to-peak modulation voltage. The XPS analysis was done with an ESCA system, using magnesium Kol X-ray radiation. Calibration of the electron binding energy was performed so that the Au 4f,,, XPS peak was referenced to 84.0 f 0.1 eV. 3. Results and discussion 3. I. Composition When analyzed “as received” the sample appeared largely covered with an oxidized layer polluted with carbon. The sample was then cleaned by ion sputtering at a temperature of 280°C. This temperature was chosen so that the amorphous state of the ribbon was preserved. We decided to consider the cleaning as finished when the surface composition did not change after the ion etching had been stopped for 30 min. The superficial content of each element of the ahoy was determined according to Palmberg [17], using the following Auger transitions P 120 eV, B 179 eV, Fe 651 eV and Ni 848 eV. The XPS analysis was performed by using the sensitivity factors published by Scofield [18]. The compositions determined in the two ways are given in table 1 along with the bulk composition by the analysis of a solid sample. The XPS and AES analysis of the ribbon were found to differ significantly from one another and from the bulk composition. Such substantial differences were previously discussed by Baer et al. [14] whose explanation is that elemental concentrations can vary rapidly in depths comparable to the electron escape depths. Moreover, different sputter yield of the various elements can be suspected as pointed out by McHugh [19].

J. Fwy. Table 1 Clean sample

Bulk analysis AES XPS ‘) Laboratoire

P. Porejo

/ Sur/oce

oxidation

behooiour

of Feel0 Ni do P, o B4 alloys

133

composilion

‘I

Central.

Fe

Ni

P

40

40

14

20 52

11

47 8.5

CNRS.

34

Solaize,

B 6 22 5.5

France.

Although no trace of P was found in the gas phase by mass spectrometry analysis - even at higher temperatures - it must be noted that P is the most volatile element. Therefore it is not surprising that the surface appeared rich in that element by Auger analysis since the observed electron has an escape depth of 2 monolayers for P LMM (120 ev). 3.2. Segregation When annealing, a change in the superficial concentrations as measured by AES could be observe. As a matter of fact the only noticeable change is that of P while Fe and B decreased slightly and Ni remained almost constant. In fig. 1 is shown the variation with temperature of the ratio HP/HA” where H, is the peak height of P measured on the spectrum relative to a given temperature and HA” the peak height of Au measured at the same time from the golden part of the sample holder which constitutes an internal reference. It can be observed for two different samples that enrichment of the surface begins at about 380°C (recrystallization temperature). It means a temperature at which atom mobility is sufficient to allow crystallization of the sample. In other words, a change of the surface composition is observed at the same time with a change of the crystallinity. 3.3. Oxygen uptake 3.3.1. Amorphous The oxidation

200

300

sample behavior

400

of an amorphous

P

500 T C-Cl

sample when submitted

to oxygen

Auger signal Fig. 1. Variation of phosphorus. with temperature. 0, sample 1; 0. sample 2.

134

J. Fuyv, P. Pomja

MO

1000

/ Surjuce

oxidation

1500 EXPOSURE CL1

Fig. 2. Oxygen uptake on amorphous Metglas at 200°C. Variation of Auger signal of the various elements with exposure of oxygen in langmuir.

behouiour

of Fe,,Ni,o

PIa B4 olloy~

5

10 EXPOSURE

15 x lo-’

CL)

Fig. 3. Oxygen uptake on amorphous Mctglas at 200°C. Variation of Auger signal of the various elements for higher exposures.

under low pressure at 2OO“C is illustrated in figs. 2 and 3 where the variation of the Auger signal for each element has been plotted as a function of the exposure in langmuir (1 L = 10e6 Torr s). Very large and rapid changes in the surface composition were observed. First of all, P dropped very rapidly and disappeared for exposures as low as for example 150 L in the experiment reported here. This phenomenon might be attributed to the volatility of P40,,. although we were not able to detect traces of this compound in the gas phase by using mass spectrometry. Concerning B, the first change took place within a few langmuir. A second peak appeared on the Auger spectrum at 170-171 eV which can be assigned to B-oxide. According to Baer et al. (141 it was assumed that the same sensitivity factor applied to both peaks and that peak heights were additive. Note that the experithe ratio Boxide/Btolnl kept a constant value of about f throughout ment. The value plotted in figs. 2 and 3 is that of BtoId and we can observe that after a significant increase, it decreased again and returned to its initial value above 4000 L and did not change very markedly for the rest of the experiment. For the two metals of the alloy, we observed a rapid decrease of the corresponding signals, although less marked than that of P, very likely because of a larger escape depth. For exposures between 1000 and 1500 L, the variation of both Fe and Ni diverged. Then while the signal of Ni decrease and disappeared for exposures between 5000 and 10000 L, that of Fe rose slowly and eventually saturated. The XPS spectra - P 2p, Fe 2p and Ni 2p - were recorded for a clean sample and for several exposures corresponding to different steps of the change evidenced by using Auger spectrometry: - 50 L: an exposure for which all the elements are still present on the surface; - 500 L: that is a situation where the P-pea- had disappeared;

J. Ftq.

P. Porejo

/ Surface

oxidarion

behoviour

o/ Fed0 NiJo P,n El, alloys

135

Fig. 4. Oxygen uplake on amorphous Metglas at 200°C: XPS spectra of P 2p at different exposures.

- 1500 L: which corresponded

roughly to the minimum of the B-total signal. It corresponds also to the point-where the oxygen uptake began to saturate; - 5000 L: a large exposure corresponding to a complete saturation of the surface. An examination of fig. 4 shows that even for a long sputtered amorphous sample the P 2p spectrum consisted of two peaks: one at about 129.5 eV corresponds to the phosphorus in the metallic state (probably covalent), the second peak at about 133.5 eV must be assigned to the phosphorus present as phosphate. The fact that this shoulder on the second peak did not disappear after a long sputtering time leads to think that the P in the alloy was slightly oxidized during its preparation. It must be noted that the difference between XPS and AES electron escape depth is clearly demonstrated by the dramatic change that occurred on the height of P in the AES spectrum after an exposure of 50 L while the XPS spectrum underwent a very slight modification for the same exposure. It can be noted that P was still observable by XPS for exposures as large as 1500 L. Moreover, it is remarkable that the contribution of oxidized P increased by increasing exposure to oxygen but covalent P remained the major species even for an exposure of 5000 L. Therefore we are led to conclude that P was present below the oxidized layer and that its oxidation did not result from direct contact with the gas phase but occurred at the interface between the superficial oxidized film and the underlying alloy. Concerning Ni 2p XPS, we can observe in fig. 5 that Ni 2p,,, and 2p,,z lines progressively decrease without any shift of the photo peaks. This fact demonstrates that Ni remained insensitive to oxidation. This was previously noted in oxidation of alloys Fe-Ni by Wandelt [20] and Greco [21]. On the contrary, the Fe 2p XPS change with exposure by oxygen (fig. 6). For an

136

J. Fwy,

P. POE@ / Surjace

oxidation

N(E) cts/s

behoviour

oj Fe,, Ni4, PI0 B4 alloys

N CEI cts/s

870

860

850

E,$evI

Fig. 5. Oxygen uptake on amorphous Metglas at 200°C: XPS spectra of Ni 2p at different exposures.

730

720

710

EbCW

Fig. 6. Oxygen uptake on amorphous Metglas at 2OO’C: XPS spectra of Fe 2p at different exposures

exposure of 50 L both Fe 2p,,, and Fe 2p,,, lines exhibited a shoulder on the side of higher energy. At 500 L the entire spectrum was shifted by 3.3 eV and the peaks were very broad showing the existence of a mixture of oxides. At 5000 L the peaks were sufficiently narrow to allow unambiguous characterization of Fe III [22]. No conclusion on the oxidation state of B was derived from B XPS line because B 1s and P 2p photopeaks have about the same energy. But assuming that it is almost entirely due B Is, the surface composition was roughly estimated from peak are measurements of 0 Is, Fe 2p and B 1s corresponding to an exposure of 8000 L. The concentrations corresponded to Fe,,$0s6. Now, if we take into account the ratio of the two peaks observed in AES for B and %xi&~ we can assume that 4 of the boron was oxidized - likely as B,O, and this led to a molecular ratio Fe/O equal to 0.72 corresponding very closely to the oxide Fe,O,. Moreover, the AES analysis allowedan estimation of the thickness of the oxidized layer. For example, we can observe that P disappeared from the AES spectrum at an exposure of about 150 L. At 110 L, it was about 5% of its initial height. With an escape depth of 5 A, we can calculate a thickness of about 15 A at this exposure. A similar calculation with Ni (escape depth of 12 A) shows that the thickness was about 36 A at 3300 L. This means that the thickness was multiplied by about 3 when the exposure was 30 times larger and that it then probably reached its limit. This last point was confirmed by sputter depth profiles as exemplified in fig. 7. Supposing that Ni and P were hidden below the oxidized layer, its

J. Fuy,

P. Poreja

/ Surface

oxidarion

30 Sf’UTTERlNG Fig.

7. Depth

profile

of an oxide

layer

hehoviour

of Fe4, Ni,,

60 TiME

amorphous sample. elements with time.

P,, B, alloys

137

t Cmin) Variation

of the Auger

signals

of the

thickness was estimated at various stages of the sputtering experiment from the way their heights increased. If the sputter rate for the film is assumed to be constant, the thickness for an exposure of 21000 L had a value in the range 30-40 A. Moreover, we noticed that the sputter time was about the same for an exposure as large as 10” L. This clearly confirms the existence of a limit for the thickness of the oxidized layer. 3.3.2. Crystallized sample Fig. 8 and 9 are typical of the variation in the AES peak heights as a function of the exposure to oxygen in the case of a sample previously crystalbzed in situ under vacuum. The sample was annealed for about 10 h at 550°C under UHV, then cleaned at this temperature by ion etching. It was then left in vacuum for i h at the same temperature to prevent possible

coo

800

1200 EXPOSURE

Fig.

8. Oxygen

uptake

1600 CL1

at 200°C on crystallized Metglas. Variation of the Auger various elements with exposure of oxygen in Langmuir.

signal

of the

J. Fury. P. Parq’a

/ Surjace

oxidation

5

10

hehoviotrr

oj Fe,, Ni,oP,,

B4 ollqs

J

EXPOSURE Fig. 9. Oxygen

uptake

15 x lo-’

20 CL)

at 2OO“C on crystallized Me~glas. Variation various elements for higher exposures.

of the Auger

signal

ol the

amorphization of the surface due to bombardment. Finally the sample was allowed to cool to 200°C. As previously observed, the P signal in the AES spectra decreased most abruptly. But it must be noted that P disappeared from the spectrum at an exposure close to 1000 L (instead of 150 L). The change in the B signal was very different and if we observed a maximum at about 1000 L, it was after a significant decrease between 0 and 400 L, that is the reverse of the initial behavior. The second B peak is observed again but the ratio B,,,,/B increased continuously, it was close to i for an exposure of 1000 L and about unity from 7500 L on. The rate of oxygen uptake will be discussed in the next section, but note that the ratio Fe/O corresponding to the saturation was very close to its previous value showing that the Fe,,,, formed was the same. The major difference with the amorphous sample is that the AES Ni peak underwent a variation similar to that of Fe and reached a limit instead of disappearing. Figures 10 and 11 give the XPS spectra of Fe 2p and Ni 2p for a crystallized sample before and after exposure to oxygen. Comparison of the Fe 2p and Ni 2p XPS lines shows the similarity with the case of the amorphous sample. The only peaks shifted were those of Fe 2p while the energies of Ni 2p did not change demonstrating that Ni did not undergo any oxidation. Concerning P 2p (fig. 12), it must be noted that after annealing and ion etching, there was only one peak at 129.5 eV corresponding to the covalent phosphorus. We might think that the oxidized phosphorus was allowed to segregate towards the surface during annealing and was then removed by ion bombardment. It must also be noted that exposure to oxygen caused the appearance of a peak at 133.5 eV larger than that observed at 129.5 eV which is consistent with the fact that phosphorus disappeared by a larger exposure from the AES spectrum and then was oxidized by contact with the gas phase.

J. Fwy,

P. Porejo

/ Surface

oxidaiion

hehaurour

of Fee,, Ni,,

139

P,, B, ullqa

NCE> cts/s

20000

1000 CLEAN 4; 12000 730

720

710

L 870

EbCe’4

Fig. 10. Oxygen uptake al 2oO°C 0” crystallized Met&u: XPS spectra of Fe 2p.

850

EbCW

Fig. 11. Oxygen uptake at 200°C on crystallized Metglas: XPS spectra of Ni 2p.

As regards the thickness of the pxidized film, a calculation similar to the one described above (derived from a decrease of an AES peak during oxidation or from sputtering) leads to a value of about 35 A identical to that of the amorphous sample in spite of the difference in composition. 3.3.3. Initial rate of oxygen uptake In order to better understand the differences in reactivity towards of amorphous and crystallized samples, we determined the variation

oxygen in the

N CE> cts/s

Fig. 12. Oxygen

uptake at 200°C on crystallized glas: XPS spectra of P 2p.

Met-

1

Fig.

13. Variation sample

200

300

400

500

b TC”CI

of initial oxygen uplake with temperature for increasing temperature: 1: 0. sample 2. and after crystallization: @, sample 1: 0, sample 2.

t.

initial rate of oxygen uptake between 200 and 560°C. This was done by measuring the increase of the AES 0 peak with time during the first five minutes of an exposure to a pressure of 3 x lo-” Torr oxygen. Because of accuracy considerations, the 0 signal was compared with two measurements of the Au signal mentioned above and recorded before and after each experiment. Then, for each temperature, the rate given in fig. 13 is the rate of increase of the ratio Ho/HA,. Note that the sample was first cleaned by ion etching and then left under vacuum for $ h at the temperature of the experiment without any change in the AES concentrations. Although the accuracy of the measurement is rather poor, the curves in fig. 13 obtained for two different samples by increasing temperature show that the rate varied continuously between two limits. One can think that the higher value was related to the amorphous sample, the lower to the crystallized sample. This is in agreement with the fact that the rate oxygen uptake was lower in the second case. Nevertheless, if the amorphous or crystallized states of the sample were determining the rate of oxygen uptake, we would have expected a steep change around 380°C. Besides, this rate of oxygen uptake would have kept its lower value when the temperature was allowed to decrease. In fact, we observed that after treatment at a temperature of 510°C this rate measured at 200°C is lower than in the case of the amorphous sample but significantly higher than expected for a crystallized sample (fig. 13(a)). For a second sample which presented a variation similar to the first one between 200 and 560°C we observed that after treatment at 560°C the rates of oxygen uptake measured not only at 200 but also at 350 or 450°C respectively (fig. 13 (b-d)), were surprisingly higher than those measured in the initial amorphous sample. It must be noted that if the amorphous or crystallized state is not the key factor of the oxygen uptake, the only other factor which underwent a significant change was the superficial concentration of phosphorus. As a matter of

J. Fusy. P. Port@

/ SurJuce

oxidation

behauiour

of Fe,, Ni,,

Hp/Hau Fig. 14. Variation

of initial

oxygen

uptake

(values

of fig. 13) with initial

PI6 B, alloys

141

C,=-O phosphorus

Auger

signal.

fact, phosphorus segregation was very difficult to rationalize when the sample was not new. Segregation depended on the ‘history’ of the sample and particularly on the temperature and duration of ion etching to which it was submitted after crystallization. Because of this phenomenon, we obtained cleaned crystallized samples of different phosphorus concentrations for apparently similar treatments. Figure 14 displays the same results as fig. 13 except that they are no more plotted vs temperature but vs the initial concentration of phosphorus as measured by AES at the beginning of the experiment. It confirms the point of view according to which the superficial concentration in P influences more the rate of oxygen uptake that the state - amorphous or crystallized - of the sample. Indeed, the rate of oxygen uptake and the initial superficial concentration of phosphorus vary in a reversed way.

4. Conclusions The present study of the oxygen uptake on the surface of the alloys Fe,,Ni,P,,B, - amorphous and in situ crystallized Metglas 2826 - demonstrates how difficult it would be to assign to the modification of the crystalline structure a clear influence on the reactivity of the surface toward oxygen. As a matter of fact, this difference - if any - is completely hidden by the change in the superficial concentration of phosphorus. The only noticeable difference in behavior between the two amorphous and crystallized samples is the presence of Ni - otherwise undergoing no oxidation - in the saturated oxide layer formed on the alloy. For an amorphous sample, this layer essentially consists of Fe,O, and a bit of B with an overall estimated composition Fe,,$O,, while for a crystalline sample Ni is about one half of Fe and B is unmeasurable.

In this context, it appeared interesting to carry out the same experiments with different samples in order to decide wether the amorphous or crystallized state has a determining influence upon oxygen uptake. These experiments are in progress with two iron-metalloid alloys, one containing phosphorus and the other only boron.

References [l] Y. Waseda and K.T. Aust, J. Mater. Sci. 16 (1981) 2337. [2] K. Hashimoto. in: Amorphous metallic alloys. ed. F.E. Luborsky (Buthxworths. London, 1983) p. 471. [3] R.B. Die& N.R. Sorensen, T. Tsur and R.M. Latanison. Treatise on Marc&l Science and Technology. Vol. 23. ed. J.C. Scully (Academic Press, New York, 1983) p. 29. [4] R.B. Diegle, J. Non-Cryst. Solids 61 (1984) 601. [5] A. Yokoyama. K. Komiyama, H. Inoue, T. Masumoto and H.M. Kimura. J. Catal. 5X (1981) 355. [6] A. Yokoyama. K. Komiyama. H. Inoue. T. Masumoto and H.M. Kimura. in: Proc. 4th Int. Cool. on Rapidly Quenched Metals. Vol. II. eds. Masumoto and Suzuki (Japan Inst. Metals. Sendai. 19Rl) p. 1419. [7] G. Kisfaludi. K. Lazar. 2. Schay. L. Guczi. C. Fewer. G. Konnos and A. Lovas. Appl. Surf. Sci. 24 (1985) 225. 181 W.E. Brewer Jr., K.M. Simon, W. Kowbel and E.E. Alp. in: 8th Int. Congress on Catalysis. Berlin (Dechema, Frankfurt/Main, 1984). [9] R. Hauert. P. Oelbafen, R. Schlagl and H.J. GUntherodt, in Proc. 5th Int. Conf. on Rapidly Quenched Metals, Wurzburg, 1984 (North-Holland. Amsterdam. 1985) p. 1493. [lo] R. Hauert. P. Oelhafen and H.J. Guntherodt, EMRS Conf. Proc.. Strasbourg (1984). [ll] K. Prabhakaran and C.N.R. Rao, Surf. Sci. 163 (1985) L 771. [12] M.T. Thomas and D.R. Baer, Proc. 4th Int. Conf. on Rapidly Quenched Metals. Vol. II. eds. Masumoto and Suzuki (Japan Insr. Metals. Sendai, 1981). [13] D.R. Baer and M.T. Thomas, J. Vat. Sci. Technol. 18 (1981) 722. [14] D.R. Baer. D.A. Petersen. L.R. Pedersn and M.T. Thomas. J. Vat. Sci. Tcchnol. 20 (1982) 957. [15] P.P. Karve. SK. Kulkami and A.S. Nigavekar, Solid. St. Commun. 49 (1984) 719. [16] P.P. Karve, M.G. Thube. S.K. Kulkami and AS. Nigavekar, Solid SI. Sommun. 50 (1984) 1027. [17] T.W. Palmberg. J. Vat. Sci. Technol. 13 (1976) 214. [18] J.H. Scofield. J. Electron Spectr. Rel. Phen. 8 (1976) 129. [19] J.A. McHugh, Method of Surface Analysis 223. cd. A.W. Czandema (Elsevier. Amsterdam. 1975). (201 K. Wandelt and G. Ertl, Surface Sci. 55 (1976) 403. [21] S.E. Greco, J.P. Roux and J.M. Blakely. Surface Sci. 120 (1982) 203. [22] C.R. Brundle, E. Silverman and R.J. Madix. J. Vat. Sci. Technol. 16 (1979) 474.