Interdiffusion and reaction in the FeAl bilayer: III: Phase characterization of furnace-annealed samples

Materials Science and Engineering, 96 (1987) 285-293

285

Interdiffusion and Reaction in the Fe-AI Bilayer: III: Phase Characterization of Furnace-annealed Samples S. R. TEIXEIRA Instituto de Fisica, Universidade Federal do Rio Grande do Sul, 90049 Porto Alegre (Brazil) C. A. DOS SANTOS Departamento de F[sica TeSrica e Experimental, Universidade Federal do Rio Grande do Norte Natal, 59000 Natal (Brazil) P. H. DIONISIO, W. H. SCHREINER and I. J. R. BAUMVOL Instituto de F[sica, Universidade Federal do Rio Grande do Sul, 90049 Porto Alegre (Brazil) (Received February 9, 1987; in revised form May 12, 1987)

ABSTRACT

This is the third of a series o f articles concerning the experimental investigation o f the interdiffusion processes and solid phase reactions occurring in Fe-Al bilayers subjected to furnace annealing, ion beam mixing and the simultaneous combination o f these two types o f me tallurgical treatment. In this article, we describe the characterization o f the phases formed during the solid phase reaction between the two components o f the bilayer under different thermal annealing conditions, by means o f conversion electron MSssbauer spectroscopy and X-ray diffraction. It is demonstrated that there are two metallurgical transformations occurring in this bilayered system. The first occurs on annealing at about 600 K with the formation o f the Fe2Al~ phase. The thickness o f this intermetallic layer increases, consuming iron and aluminium from the original films, with increase in the temperature and time o f annealing. The second takes place on annealing at and above 650 K, producing the metastable phase FeAls, which is seen in this work to remain stable up to at least 930K. 1. INTRODUCTION In the two previous articles Part I [ 1] and Part II [2] of this series, we have determined the annealing conditions for interdiffusion and reaction in the Fe-A1 bilayer to occur. The discussion of several experimental aspects 0025-5416/87/$3.50

revealed in these articles, such as the holdup effect of the diffusion process, the reaction kinetics, the drastic increase in the sheet resistivity and the nature of the grains of the precipitates formed in the outermost surface, required a characterization of the intermetallic phases of the Fe-A1 system that are formed as interdiffusion and solid phase reactions under furnace annealing proceed. In the present article, we report on the results of the phase characterization at the surface layers of the same samples that were previously subjected to RBS, sheet resistivity and scanning electron microscopy (SEM) analyses. The methods used were STFe conversion electron MSssbauer spectroscopy (CEMS) and X-ray diffraction. It will be demonstrated that these two analytical techniques, combined with the concentration vs. depth profiles obtained in Part I [ 1 ] and the sheet resistivity measurements and SEM images in Part II [2], give a consistent characterization of the phases, whether solid solutions or intermetallic compounds. The characterization of the intermetallic phases formed by thermal annealing at temperatures between 870 and 940 K of an iron thin film evaporated on an aluminium substrate has been performed previously by Preston [3], who identified the Fe4A113 phase. This result is completely different from those of the present work. Another paper by Gaboriaud and Jaouen [4], in which an Fe-A1 bilayer annealed at 770 K was characterized, gives evidence of an intricate mixture of Fe3Al, © Elsevier Sequoia/Printed in The Netherlands

286

Fe-A1 and Fe2A15, which also disagrees with the results of the present work. A third paper concerning the annealing at temperatures between 770 and 870 K for 20 min of a previously ion-beam-mixed Fe-A1 bilayered system [5] shows the precipitation of a metastable FeaA1 phase at 770 K, which decomposes into phase-separated systems of iron and aluminium at 8 7 0 K .

2.2. Experimental results The CEMS spectra for the Fe-A1 bilayers subjected to isochronal anneals for 60 min at a temperature between 270 and 870 K are shown in Fig. 1. The corresponding fitting parameters are displayed in Table 1. The CEMS spectrum obtained from a pure iron foil is given in Fig. 1, spectrum a, for reference. In Fig. 1, spectrum b, we show the spectrum

2. E X P E R I M E N T A L D E T A I L S A N D R E S U L T S

% 2.1. Experimental details The Fe-A1 bilayered samples were obtained by sequential thermal evaporation of aluminium and iron in a high vacuum chamber (p = 2 X 10 -5 Pa), using oxidized silicon as the substrate. The iron film was 100 nm thick and the aluminium film 500 nm. The sequential deposition of iron and aluminium was performed without breaking the vacuum in the chamber, in order to keep the Fe-A1 interface as free as possible from iron or aluminium oxide layers. The substrate temperature during deposition was kept at 290 K. The thermal annealing of the bilayered samples was performed in a high vacuum (p = 2 X 10 -5 Pa) furnace at temperatures that varied between 290 and 870 K. The CEMS data were obtained in a backscattering geometry. A proportional counter through which He-5%CH 4 was allowed to flow was added to a conventional constantacceleration MSssbauer spectrometer. A source of 57Co in a rhodium matrix was used. All the CEMS measurements were performed at room temperature. Further experimental details and data analysis can be found in recent reviews [6-8]. The X-ray diffractograms were obtained using a Siemens diffractometer, with a DebyeScherrer camera and a rotating-plate sampleholder system perpendicular to the incidence plane, in order to minimize the effects of the crystallographic preferential orientations of the samples. Co K s radiation was used. The effects of texture cannot be eliminated completely, however, and as a consequence the intensities of the diffraction lines cannot be used to calculate the proportions of the phases formed. Apart from this problem, the X-ray patterns show the transitions occurring in the Fe-A1 system as the thermal anneals proceed.

10.0 5.0 0

as-deposited

:

0.2 0 0.2 (h

_ :':

.. 570K

•

' " °"

•

: %':':.

, " • ' :o • ~ "..% o.% .~.-...-~,,~:~. ,.~.:':.'.',

.

". ..:

...:.(cl

" s:

"°"

".. * • >-.'~':<~ :~:..',:::. • ° .,.°. : .•"'." .'....

-6.0-4.0-2.0 0 2.0 4.0 VELOCITY mm/s)

6.0

Fig. 1. 57Fe CEMS spectra: s p e c t r u m a, a pure iron foil; s p e c t r u m b, an a s - d e p o s i t e d Fe-A1 bilayer prep a r e d as described in Part I [ 1 ]; spectra c-g, bilayered samples a n n e a l e d at various d i f f e r e n t t e m p e r a t u r e s for 60 rain. See t e x t a n d Table 1.

287 TABLE 1 MSssbauer parameters from the fittings of the spectra shown in Figs. 1-3 AEQ 1 (rams-)

6a

(mms -1)

F (mms-1)

Relative area

(kOe)

As deposited As deposited As deposited

330 0 0

0 0 0

0 0.02 0.46

0.28 0.39 0.47

82 14 3

(~-Fe 9'-Fe Solid solution

600 600 600

330 0 0

0 0 0.52

0 0.03 0.25

0.28 0.35 0.33

23 18 59

cv-Fe 7-Fe Fe2Al 5

620

.

Annealing temperature

H

Phases

(%)

(K)

620. 620 650

.

650 650 870 870 870

.

0 0

.

0 0.50 .

0 0 .

.

0 0.37 .

0 0

.

0.02 0.25 .

0.03 0.21 .

0 0.30

.

0.35 0.33 .

0.35 0.44 .

0.01 0.22

.

15 85 .

11 89 .

0.31 0.38

")'-Fe Fe2A15 ~/-Fe FeAl 6 .

9 91

~'-Fe FeAI6

H is the hyperfine magnetic field at the site of 57Fe nuclei; AEQ is the electric quadrupole splitting; 6 is the isomer shift and l" is the linewidth of each component of the fitting [6-9 ]. The last column indicates the corresponding phases of the Fe-AI system described in the literature. Typical errors for the fitting parameters are 5% in H, AEQ and 6. a Relative to (~-Fe.

t ak en f r o m the as-deposited sample. The fitting o f this spectrum is b e t t e r explained in Fig. 2, where the so-called "stripping p r o c e d u r e " [9] is used to show that, apart f r o m ~-Fe, the o th er c o m p o n e n t s of t he fitting are two singlets, one due to solid solution of iron in aluminium, and the o t h e r due to 7-Fe precipitates. The identification o f the solid solution of iron in aluminium by its MSssbauer parameters was p e r f o r m e d by comparison with the almost c o m p l e t e list o f MSssbauer parameters for the defects and phases of t he Fe-A1 system given by Nasu e t al. [10]. We interpreted the central singlet in Fig. 2, spectrum b, as being due to small precipitates o f 7-Fe c o h e r e n t with the aluminium lattice. The idea was originally p r o p o s ed by L o n g w o r t h and Jain [11] for 57Fe implanted into a c o p p e r substrate. T h e y observed t he same singlet in the CEMS spectrum, and t h e y ascribed it t o 7-Fe precipitates with a grain size of m or e t han 10 atoms which f o r m e d at t he intersection of the dislocation loops, grain boundaries and ot her damaged regions p r o d u c e d by the bombarding 57Fe ions. The similarity bet w e e n their system and that in the present w or k arises because the

surface o f the vacuum-deposited aluminium layer has a very large density of the abovem e n t i o n e d e x t e n d e d defects, which makes the near-interface region favourable for 7-Fe precipitation. The central singlet observed in the CEMS spectra in ref. 5 is certainly also due to 7-Fe, although t hat was n o t t he interpret at i on given t o it by G o d b o l e e t al. In Fig. 1, spectrum c, we show t he CEMS spectrum for the sample annealed at 570 K for 60 min. There is a noticeable decrease in the relative spectral area of ~-Fe. However, it is rather difficult to identify the ot her comp o n e n t s of the spectrum. In view of this, we did n o t proceed in fitting this spectrum. In Fig. 1, spectrum d, we show the CEMS spectrum for the sample annealed at 600 K, where we can see a m u c h larger decrease in the ~-Fe c o m p o n e n t and an increase in a quadrupole d o u b l e t whose MSssbauer parameters are given in Table 1. These parameters resemble those obtained by Stickels and Bush [12] for Fe2A15. The c o n t r i b u t i o n from the 7-Fe singlet remains, and we cannot rule o u t a solid solution c o n t r i b u t i o n as well. At 620 K (Fig. 1, spectrum e) t he MSssbauer spectrum

288

%

(a)

0.75

!

0.2

_ 62oK '.-.'

(a)'

650K :"%"

(b)

8- oK

(c)

0.50

0

.L."

0.25

• :: : ~ .

0 I

]5

sfrip I

I

I

I

.

I

(b)-

0.75

0.2

o ":':":':':-"'"-~

~.,,':"-';'z""f"~;~ '

0

I

I

I

I

I

I

-6.0 -4.0 -20 0 2.Ù 4.0 VELOClTY(mm/s)

I

6.0

Fig. 2. Example of the "stripping" procedure for data reduction, used in the analyses of some of the CEMS spectra of this work: spectrum a, the same spectrum as in Fig. 1, spectrum b, with only the ~-Fe component fitted; spectrum b, the spectrum obtained by subtracting the ~-Fe component from spectrum a, fitted by two singlet components. See text and Table 1.

' "

0.2 0

A

consists o f the Fe2A15 q u a d r u p o l e , t h e 7-Fe and e v e n t u a l l y solid s o l u t i o n singlets. In Fig. 3, we show t h e CEMS spectra f r o m samples isochronally annealed f o r 60 min at 620, 6 5 0 and 8 7 0 K, m e a s u r e d o n a r e d u c e d v e l o c i t y scale. This t y p e o f m e a s u r e m e n t allows us t o see in b e t t e r detail t h e n a t u r e o f t h e spectra s h o w n in Fig. 1, spectra e-g. So, Fig. 3, s p e c t r u m a, shows the Fe2A15 q u a d r u pole d o u b l e t and t h e ~-Fe singlet w i t h relative spectral areas o f 85% and 15% respectively. In Fig. 3, s p e c t r u m b, we can observe a change in t h e m a g n i t u d e o f t h e q u a d r u p o l e d o u b l e t f r o m 0.50 t o 0.37 m m s-1, r e p r e s e n t i n g a new phase also r e p o r t e d b y Stickels and Bush [ 12 ], n a m e l y the m e t a s t a b l e FeA16 phase. Finally, in Fig. 3, s p e c t r u m c, we can see t h a t t h e FeAI 6 is still p r e s e n t a f t e r annealing at 8 7 0 K f o r 60 min. A CEMS s p e c t r u m t a k e n f r o m a bilayered sample annealed at 930 K f o r 60 min gave a result identical w i t h t h a t in Fig. 3,

I

.95

I

I

-l.O0 0 l.O0 VELOCITY(mm/s)

1.95

Fig. 3. 57Fe CEMS spectra taken at a reduced velocity scale for the same samples as in Fig. 1: spectrum a, the same spectrum as in Fig. 1, spectrum e; spectrum b, the same spectrum as in Fig. 1, spectrum f; spectrum c, the same spectrum as in Fig. 1, spectrum g. See text and Table 1.

s p e c t r u m c. A n o t h e r aspect revealed in Fig. 3 is t h a t t h e relative spectral area d u e t o ~/-Fe decreases w i t h increasing isochronal annealing t e m p e r a t u r e (see T a b l e 1). This can be interp r e t e d as an effective decrease in t h e c o n c e n t r a t i o n o f 7 - F e p r e c i p i t a t e s d u e t o t h e decrease in t h e d e n s i t y o f e x t e n d e d d e f e c t clusters in t h e a l u m i n i u m film capable o f a c c o m m o d a t i n g ~/-Fe precipitates.

289

The isothermal anneals at 600 K of the bilayered samples gave the CEMS spectra in Fig. 4. These are complemented b y Fig. 1, spectrum d, where the spectrum for the sample

r

r

i

i

i

i

i

AS- DEPOSITED (a)

. . . .

,

..

.-'..

.

00 :i

:'.'.:.

-

600K,12Omin

. ".. •

I

1

-6

-4

(o)

..

iii°i

(d)

• ..,

,

. ..~"

• ... : ::.:. :..

[

I

::'

I

I

-2 0 2 4 VELOCITY (rnm/s)

6

%

600K,]2Omin

0.2-

.

:. /,

?. •

J

-].95 (b )

"._

°

I

.

~

o

I

I

..%

l

I

eOo•

•••

I

0 VELOCITY (ram s-1)

Fig. 4. (a) 57Fe CEMS spectra for the samples isothermally annealed at 600 K; (b) a spectrum taken a reduced velocity scale for the same sample as in P a r t I, F i g . 4 , s p e c t r u m d.

[95

at

annealed at 600 K for 60 min is shown. The evolution of the reaction leading to the formation of the Fe2A15 intermetallic layer is clearly seen, especially the progressive disappearance of the ~-Fe c o m p o n e n t of the spectrum, as the iron film is consumed in the reaction. The a s y m m e t r y of the spectra, however, indicates the presence of 7-Fe precipitates in the remaining aluminium layer for all annealing times (see also Part I, Figs. 4 and 5). The reaction kinetics at 600 K could in principle be determined from the fittings of Fig. 4 and Fig. 1, spectrum d. However, this would involve several correction factors for the recoilfree fraction of 7-ray absorption b y the different phases, which are not well defined. One relevant aspect of the reaction kinetics at 600 K is the confirmation, given in Fig. 4(b), that the final composition of the surface (after annealing at 600 K for 120 min) is Fe2A15 and ~-Fe, with spectral areas in the approximate proportions of 85% and 15% respectively. This result will be discussed in Section 3 in the c o n t e x t of the previous RBS and sheet resistivity analyses performed on the system. X-ray diffraction analyses of the isochronally annealed samples (t -- 60 min) are presented in Figs. 5 and 6. The as
290

__

A

550K

570 K

I

>-

o

O9 Z I11 t-Z

..id tD

v

v

>I-U3 Z 1,1 I'-Z

650K 620K

. . . . . .

O4

Cq

__jilL__

'

~

A

o

o v

L

~

o

"

I

650 K

I

58

54

I

.,,

50

I

1

I

30

25

20

2e 870K

Fig. 6. X-ray diffraction patterns showing the details of the metallurgical transformations which occurred as a result of annealing the bilayered samples for 60 min at three different temperatures, namely 6 0 0 , 6 2 0 and 650 K. See text.

_%~
100

5O

40

30

2e Fig. 5. X-ray diffraction patterns for the Fe-Al bilayered samples as deposited and after annealing at various temperatures for 60 min.

for the identification of the Fe2A15 and FeA16 compounds, namely between 20 = 20 ° and 20 = 32 °, and between 20 = 48 ° and 20 = 58 °, are shown in better detail in Fig. 6, for the

anneals at 6 0 0 , 6 2 0 and 650 K. We note the decrease in the Fe2A15 lines and the simultaneous increase of the (301}, (123}, {200), (020) and (110) lines of FeA16 as the isochronal annealing temperature is increased from 600 to 650 K. The evolution of the solid phase reaction generated by the isothermal anneals at 600 K is followed with the X-ray diffractograms shown in Fig. 7. The decrease in the intensities of the iron and aluminium lines, and the increase after annealing at 600 K for 120 min in the Fe2A15 lines should be noted. Again here, as in the CEMS analyses of the isother-

291

_

o 600K,2Omin~l

-i

_o -II q

;/~

.,',

~ ,

--'L 600K,5Omin I

-I' %1],~,

II

.,J !i

i

!

il 1

,'

,

I Ii

I

u')

600K,4Omin

E

!

_ci v

o

>CO

z W

600K,6Omin

z

U

i 600K,12Omin o,,,o oo c~

cq

2@ Fig. 7. X-ray diffraction patterns showing the kinetics of growth of the intermetallic compound layer of Fe2A15 at 600K.

mal anneals at 600 K, the reaction kinetics could be extracted b y using appropriate scattering factors. Since the reaction kinetics were measured rather precisely b y RBS in Part I, and since there is not any qualitative inconsistency with the CEMS or X-ray results, we prefer to keep the reaction kinetics as determined by RBS, as they need less correction.

3. DISCUSSION AND CONCLUSIONS

Identification of the phases and solid solutions formed in the Fe-A1 bilayered samples subjected to thermal annealing can be carried out as follows.

(i) The as
292 which leads to precipitation (e.g. of 7-Fe) at an iron concentration in aluminium much lower than that necessary for the formation of Fe2A15. However, the solid solubility of aluminium in iron is much higher (about 35 wt.%) and so only the diffusion of aluminium in iron allows the achievement of the relative proportions of iron and aluminium atoms necessary for the formation of the FegAI 5 intermetaUic c o m p o u n d . This conclusion is in contradiction to the interpretation given in ref. 12, in which the driving force for the nucleation and growth process is attributed to iron diffusion. Moreover, the sheet resistivity measurements after annealing at 570 K, described in Part II, display only a slight increase in R s compared with those for previous annealing temperatures, confirming the depth limitation of the Fe2A15 nucleation. The occurrence of strong iron and aluminium interdiffusion across the original bilayer interface, as for instance observed in Part I, Fig. 1, spectrum c, for annealing at 600 K for 60 min, causes the formation of an appreciable amount of Fe2A15, detected here by CEMS and X-ray diffraction. Whether or not this is an individual layer of intermetallic c o m p o u n d is hard to judge, even with all the methods used here. However, all these analytical techniques agree that the iron t o p layer is not completely consumed in the interdiffusion and reaction processes after this annealing condition. By extending the annealing time at 600 K to 120 min, we arrive at a situation in which CEMS indicates that the iron top layer may be completely consumed (see Part I, Fig. 4, spectrum d), whereas RBS, sheet resistivity and X-ray diffraction indicate that it is not. The almost saturated behaviour of the sheet resistivity curve at a value of a b o u t 0.5 ~2/[::] for anneals at 600 K for times longer than 60 min (see Part II, Fig. 2) can now be attributed to the observations b y RBS, CEMS and X-ray diffraction of the remaining aluminium layer with 7-Fe precipitates on it (see Part I, Figs. 4 and 5, and Figs. 4 and 7 of the present article). The complete consumption of the iron top layer to form the Fe2A15 intermetallic layer (with some 7-Fe remaining) is unambiguously observed by all the five analytical techniques used in the present work for the Fe-A1 bilayered sample annealed at 620 K. The fiat

plateau in the RBS spectrum in Part I, Fig. 1, spectrum d, corresponds to the complete disappearance of the ~-Fe sextet from Fig. 1, spectrum e, of the present article. The increase in the sheet resistivity seen in Part II, Fig. 1, from 600 to 620 K annealing is also the result of the formation of a plain layer of Fe2A15 by the complete consumption of the iron film. So, the CEMS and X-ray diffraction analyses identify the outermost surface structure seen in the SEM image in Part II, Fig. 3(c), as being Fe2A15 grains. Finally, the observation of the metallurgical transformation following annealing of the bilayered sample at 650 K, which was indicated by sheet resistivity and SEM analyses, is now seen as corresponding to the formation of an intermetaUic layer of the FeA16 phase. This transformation was not detected by RBS analyses. In Part I, we hoped that the identification of the phases and solid solutions formed by thermal annealing of the Fe-A1 bilayered sample would throw some light on the observation of a holdup effect of the diffusion of iron into aluminium only up to 570 K, and not up to 900 K as described b y Preston [3], or up to 800 K as in the work by H o o d [15]. Indeed, the as
293

The reaction kinetics at 600 K measured b y RBS in Part I can now be described as the kinetics for the nucleation and growth of the Fe2AI5 intermetallic c o m p o u n d layer. The reaction kinetics for the formation of the FeAls c o m p o u n d layer could not be determined because of the extremely high velocity of this reaction; according to the RBS measurements described in Part I, the iron film is completely consumed for the formation of FeAls after annealing for only 10 min at 650 K, which was seen in the present work to be the threshold annealing temperature for the formation of FeA16. This fact, namely that the velocity of the reaction leading to the formation of an intermetallic c o m p o u n d layer at 650 K is much higher than the velocity for the formation of a c o m p o u n d layer at 600 K, is essentially the explanation that emerges from the present work for the formation of Fe2A15 at annealing temperatures such as 600 or 620 K, and FeAls for all temperatures above 650 K; according to the w o r k of Stickels and Bush [12], a high precipitation rate leads to the formation of FeA16, whereas a moderate precipitation rate leads to the formation of FeAls. Although the work of Stickels and Bush was performed with bulk specimens, we believe that the basic facts a b o u t the precipitation kinetics remain the same for thin film structures. In conclusion, the characterization of the phases formed in the Fe-A1 bilayered system allowed us to identify the possible causes for several of the experimental facts observed so far in the present work, such as the holdup of diffusion up to 570 K, the t y p e of intermetallic c o m p o u n d resulting from the reaction kinetics measured at 600 K, the type of phase transformation responsible for the increase in the sheet resistivity between 620 and 650 K

annealings and the final composition of the outermost surfaces observed by SEM. ACKNOWLEDGMENTS

This work was supported in part b y Conselho Nacional de Desenvolvimento Cientifico e Tecnoldgico and by Financiadora de Estudos e Projetos.

REFERENCES 1 S. R. Teixeira, P. H. Dionisio, E. F. da Silveira, F. L. Freire, Jr.,W. H. Schreiner and I. J. R. Baumvol, Mater. Sci. Eng., 96 (1987) 267-277. 2 S. R. Teixeira, P. H. Dionisio, M. A. Z. Vasconcellos, E. F. da Silveira,W. H. Schreiner and I. J. R. Baumvol, Mater. Sci. Eng., 96 (1987) 279-283. 3 R. S. Preston, Metall. Trans., 3 (1972) 1831. 4 R. J. Gaboriaud and C. Jaouen, Appl. Phys. A, 41 (1986) 127. 5 V.P. Godbole, S. M. Chandhari, S. V. Ghaisas, S. M. Kanetkar and S. B. Ogale, Phys. Rev. B, 31 (1985) 5703. 6 M. J. Tricker, Adv. Chem., 194 (1981) 63. 7 G. Longworth and R. Atkinson, Adv. Chem., 194 (1981) 101. 8 J. A. Sawicki and B. D. Sawicka, Hyperfine Interact., 13 (1983) 199. 9 A. H. Muir, Jr., in J. J. Gruverman (ed.), Mossbauer Effect Methodology, Vol. 4, Plenum, New York, 1968, p. 75. 10 S. Nasu, U. Gonser and R. S. Preston, J. Phys. (Paris), Colloq. C1, 41 (1980) 385. 11 G. Longworth and R. Jain, J. Phys. F, 8 (1978) 351,363. 12 C. A. Stickels and R. J. Bush, Metall. Trans., 2 (1971) 2031. 13 E. H. Hollingsworth, G. R. Frank, Jr., and R. E. Willett, Trans. Metall. Soc. AIME, 224 (1962) 188. 14 R. Wang, J. Gui, S. Yao, Y. Cheng, G. Lu and M. Huang, Philos. Mag. B, 54 (1986) 188. 15 G. M. Hood, Philos. Mag., 21 (1970) 305.