Conversion electron Mössbauer spectroscopy and xray photoemission
spectroscopy studies of thermal oxidation of Fe67Co18B14Si1 metallic
glass
Y. V. Bhandarkar, S. M. Kanetkar, S. B. Ogale, S. V. Ghaisas, V. G. Bhide et al.
Citation: J. Appl. Phys. 59, 4158 (1986); doi: 10.1063/1.336675
View online: http://dx.doi.org/10.1063/1.336675
View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v59/i12
Published by the American Institute of Physics.
Related Articles
Continuous-annealing method for producing a flexible, curved, soft magnetic amorphous alloy ribbon
J. Appl. Phys. 111, 07A309 (2012)
Fabrication of nanoscale glass fibers by electrospinning
Appl. Phys. Lett. 100, 063114 (2012)
Thermodynamic, Raman and electrical switching studies on Si15Te85-xAgx (4 ≤ x ≤ 20) glasses
J. Appl. Phys. 111, 033518 (2012)
Synthesis of single-component metallic glasses by thermal spray of nanodroplets on amorphous substrates
Appl. Phys. Lett. 100, 041909 (2012)
A simple criterion to predict the glass forming ability of metallic alloys
J. Appl. Phys. 111, 023509 (2012)
Additional information on J. Appl. Phys.
Journal Homepage: http://jap.aip.org/
Journal Information: http://jap.aip.org/about/about_the_journal
Top downloads: http://jap.aip.org/features/most_downloaded
Information for Authors: http://jap.aip.org/authors
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
Conversion electron Mc;ssbauer spectroscopy and x-ray photoemisSBon
spectroscopy studies of thermal oxidation of Fe67Co18814Si1 metallic glass
Y. V. Bhandarkar, S. M. Kanetkar, S. B. Ogale, S. V. Ghaisas, and V. G. Bhide
Department of Physics, University ofPoona, Pune 41 I 007, India
P. D. Prabhawalkar
Regional Sophisticated Instrumentation Centre, Indian Institute of Technology, Powal; Bombay 400 076,
India
(Received 6 May 1985; accepted for pUblication 28 January 1986)
The oxidation kinetics and associated structural and chemical transformations in the
Fe67CoIHBI4Sil metallic glass have been investigated by using the techniques of conversion
electron Mossbauer spectroscopy (CEMS) and x-ray photoemission spectroscopy (XPS). The
as-quenched as well as vacuum-annealed samples of the metallic glass have been SUbjected to
thermal treatment in air at 100, 300, and 390 ·C for! h in each case. The information regarding the
structural transformations and chemical reactions has been obtained by deconvoluting the
complex CEMS spectra and analyzing the contributions of various elements to the XPS spectra. It
is shown that oxygen incorporation in the glass initiates the process of recrystallization of the
glassy state at a temperature which is lower as compared to its temperature of recrystallization in
vacuum. The oxygen-coordinated phases in the oxide scale and the other phases precipitated due
to the reactive thermal treatment have been identified and the physical mechanisms underlying
the observed effects have been commented upon. Experiments have also been performed to
examine the transformations occurring in the metaUic glass upon its prolonged thermal treatment
in air at a typical temperature ( -100 ·C) within the recommended temperature range (25125 'C) for its use in applications. It has been demonstrated that the surface layers of the asreceived glass are affected by such a treatment, though the sample preannealed at 300 'C remains
relatively unaffected. The observations made on the basis of CEMS and XPS studies have been
supported by x-ray diffraction and resistance-annealing measurements.
I. INTRODUCTION
In recent years there has been a considerable growth of
interest and activity in the field of disordered and metastable
materials because of the novel and interesting physical properties of these systems and their application potential. 1-9 In
spite of considerable efforts spent in synthesizing and characterizing the new metastable materials, however, the physics of these systems remains far from being understood.
Further experimental. work, expecially oriented towards developing an understanding regarding the microscopic
aspects of such materials is therefore required to bring out
the correlation between their structural and compositional
aspects and their mechanical, electrical, optical, and other
properties. Metallic glasses form an important and perhaps
the most interesting class of metastable materials which has
been the focus of scientific activity in recent years because
these materials exhibit such peculiar physical and chemical
properties which cannot be realized in other material systems. 10,11 However, the special properties of metallic glasses,
which emerge as attributes of their specific structural state,
can be exploited in novel applications only as long as the
temperature of the glass is not raised beyond -100-150 DC.
Even at such low temperatures if the glass is held for long
durations of time, irreversible structural transformations
can occur in these glasses, depriving them of their novel
properties. It has been argued that structural transformations in metallic glasses below their glass transition tempera4158
J. Appl. Phys. 59 (12),15 June 1986
tures occur via atomic transport mediated by transient embryonic vacancies generated by statistical thermal processes.
However, there is not enough evidence for such a model;
hence, it is necessary to perform additional experiments
which could throw Ijght on the diffusive processes in metallic glasses, leading to chemical transformation therein. Oxidation offers an easy and perhaps a useful mode of in vesti gating the diffusive reactions in their near surface region and
hence a study of the oxidation of metallic glass is expected to
yield valuable information regarding these materials. When
such a study is carried out over a wide temperature range,
i.e., from below the recommended operating temperature for
the metanic glass to its glass transition temperature, one may
expect to obtain new information of scientific importance
and technological. relevance. This indeed has been the objective of our work.
In this paper, we report the results of our experimentaJ
studies on the oxidation behavior of Fe67Co 1sB 14Si 1metallic
glass (MetgJ.as:"J 2605 CO, Allied Corp., USA) over the
temperature range between 100 and 390 DC. We have used
three powerful characterization techniques viz. conversion
electron Mossbauer spectroscopy (CEMS),12-14 x-ray photoemission spectroscopy (XPS), and x-ray diffraction
(XRD) to unfold the microscopic features of the oxidation
process and associated chemical transformations. In order
to distinguish clearly between the transformations caused by
the incorporation of oxygen in the glassy metal from the
transformations induced by pure thermal effects, an oxida-
0021-8979/86/124158-09$02.40
© 1986 American Institute of Physics
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4158
tion study has also been performed on the metallic glass samples vacuum-annealed at various temperatures for different
durations of time. This study has thus brought out a number
of interesting features concerning the atomic transport as
well as structural and chemical transformations in
Fe67Co'8B'4Si, metallic glass.
II. EXPERIMENT
The metallic glass samples used in the present studies
were Metglas® 2605 CO (Allied Corp. USA) having composition ofFe67Co,sB,4Si" The oxidation of the samples was
carried out in air having a relative humidity of 55% ± 3%,
at various values of temperature, below the glass transition
temperature (Tg = 430°C) for the glass. A certain number
of samples were annealed in a clean vacuum of 10- 6 Torr
prior to oxidation. The vacuum system used to perform the
annealing was equipped with adequate liquid nitrogen and
molecular sieve traps to avoid contamination due to oil mist.
The techniques of conversion electron Mossbauer spectroscopy (CEMS),14-16 x-ray photoemission spectroscopy
(XPS), and x-ray diffraction were employed to characterize
the as-received, vacuum-annealed, as well as the oxidized
samples. The CEMS spectra of the samples were recorded by
using a constant acceleration Mossbauer setup with CO S7 :Rh
as the source. The 7.3-keV K-sheH conversion electrons
emitted within - 100 nm below the sample surface were detected in a continuous gas flow (helium + 4% ethanol) proportional counter. Transmission Mossbauer spectrum of the
as-received sample was also recorded to bring out the difference in the nature of atomic ordering in the bulk and in the
surface layers. In addition to the use of the conventional
MOssbauer fitting procedure the Fourier decomposition
method suggested by Window '5 was also used to obtain the
distribution of hyperfine field. This was found to be necessary in the present case in view of the presence of random
arrangement of atomic constituents in the Metglas® . Since
the reliability of results obtained by Windows's method depends on the correctness of the input parameters viz. the
linewidth, the line shapes,and calibration, care was taken to
vary the input parameters to obtain the minimum value of
mean-square deviation between the experimental and fitted
spectra. The I CL 1904 S computer was used to carry out the
Mossbauer fitting and to obtain the hyperfine field distributions.
In order to support the resu:lts obtained by CEMS measurements, depth profiling XPS studies were carried out on
the samples by using the V. G. Scientific Mark 11 machine
with twin anode x-ray source. In these experiments aluminum Ka line (1486.6 eV) was used and the measurements
were carried out in a clean vacuum ofbetter than 5 X 10- 10
Torr. The sputtering rate, at the time of depth profiling with
lO-keV Ar+ ions at a current density of 1J1.A/cm 2 , was kept
to be - 2 Nmin for all the metallic glass samples. The phenomenon of preferential sputtering and the inaccuracies in
the value of the sputtering coefficient often make compositional analysis difficult and questionable. Hence, only gross
features of results were considered in the analysis. To study
the occurrence of crystallization, x-ray diffraction measurements were also carried out on the vacuum-annealed as well
4159
J. Appl. Phys., Vol. 59, No. 12, 15 June 1986
as the oxidized samples. The Cu Ka radiation was employed
in these measurements to obtain the diffraction data. All the
results are obtained from the shiny (away from the wheel)
surface of metallic glass.
III. RESULTS AND DISCUSSION
Metallic glasses are extremely complex systems to understand on the basis of any reasonably simple theoretical
model because these systems are characterized by the presence of structural disorder which is generally dressed with
two or more elemental constituents having widely different
chemical properties. Yet, in order to discuss the results of the
experimental studies performed on such a system, it becomes
necessary to work within the framework of a certain suitable
structural modeL In the case of amorphous Fe-B alloys,
there is an evidence that Fe3 B-like short-range order exists 16;
the Fe3 B structural unit being isostructural to Fe 3 P for compositions richer in Fe above -75 at. %. In this model, the
hyperfine field distribution obtained for Fe3 + x B, _ x amorphous alloys is explained by assuming random substitution
of B atoms by Fe atoms. This model is questioned on the
basis of the observed Zeeman splitting and it is suggested
that in the case of Fe82 B ,s amorphous alloy,17 the quasicrystalline structure may be orthorhombic cementite, rather
than tetragonal-like Fe3 P. In both these experiments, however, a short-range order is indicated, while its exact nature
is disputed. In the present case the composition
Fe67Co,sB'4Si, can be looked upon as (Fe,Co) 85BI4Si" Crystallization study by Klein et al. '8 on amorphous Fe-Co-B
alloys shows that the as-quenched metallic glass can be fitted
with the dense random packing of hard sphere (DRPH)
model, which takes into account stronger interaction
between metal-metanoid atoms as compared to the metaImetal atoms. Their study shows that in the process of crystallization, a-(Fe,Co) and (Fe,CohB phases are formed as
precursor states, which are subsequently converted into a(Fe,Co)and (Fe,Co)2B phases. These observations indicate
that the metallic glasses can be modeled in terms of the presence of a short-range order and that such an order could be
arbitrarily close to the one in the structural state, which is
precursor to crystallization. Thus, in order to analyze the
observations of the oxidation behavior we have assumed the
presence of a short-range order in the Metglass® along with
other assumptions which are related to the influence of Co,
B, and Si atoms on the definition of the hyperfine field at the
Fe nucleus. The latter assumptions are (i) when substitutional boron coordination increases, the hyperfine field at
the Fe site decreases '9 ; (ii) interstitial boron tends to decrease the hyperfine field at the Fe site more as compared to
substitutional boron 19; (iii) presence of Co tends to increase
the hyperfine field in amorphous Fe-Co-B combination 18.20;
(iv) a sman amount ofSi does not change the effect ofB on
the hyperfine field at Fe, when the composition is like
Fes5 B ,4 Si]. Here we assume that the effect of Co, when substituted for Fe,21 does not change the behavior of the alloy
appreciably as a function of Si concentration, because the
effect of Si has been explained to appear primarily by the
virtue of the atomic size dependence rather than due to any
stronger chemical interaction. 2 I
Bhandarkar et 8/.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4159
(e)
(a)
1520
(a )
12
6
(t)
:1
';-120.6
$2
x
-...
j'
-.••_ _.........-.05
--_.",.4
III
r-
Z
o :::c
III
:J
-e
::s
III
a:
12
o
t
( b)
<{
~
148
':';'f/ ~4 /i\;~V ~<:/{i_. I-+-\--'rd--++--JH~--?!
6:::
::I:
>
r-
....·,""""...........".._5
III
Z
.........~-"'---"'--.... L.
w
150
.
':-.j
1
(d)
r-
z
::: ::;/::;'''1':/ '~\~;;;'H-~-#-H-----:'~h1
-8 -4
0
4
2
rJ'_...,....... . . . . _ . . . . .___
8
FIG. 1. Conversion electron Miissbauer spectra of (a) the shiny (away
from the wheel) surface ofthe as-quenched metallic glass and (b), (c), (d)
the metallic glass samples oxidized at 100,300, and 390 ·C, respectively, for
! h each. The hyperfinefieJd distributions [P(H) curves] corresponding to
the CEMS spectra in (a), (b), (c~, and (d) are shown in (e), (0, (g), and
(h), respectively. The dashed line in (e) shows the P( H) curve corresponding to the transmission spectrum of the as-received metallic glass.
3
2
X)
180
190
BINDING
The Mossbauer spectra and the corresponding hyperfine field distributions for the case of the as-quenched metallic glass are given in Fig. 1. The CEMS spectrum of the asquenched glass appears as shown in Fig. 1 (a), and its field
distribution [Fig. 1(e)] shows a broad peak centered at
~ 288 kOe, which can be fitted with sextets corresponding to
four values ofintemal magnetic field viz. 215, 250, 278, and
305 kOe. The DRPH or (Fe,Co)3B-like ordering'S predicts
that (Fe,Co)ssB ,4 Si, should have a significant contribution
of the magnetic components corresponding to 330 kOe or
higher field values due to the zero and one coordination of B
neighbors; when Co atoms are randomly substituted for Fe.
However, absence of this sextet suggests that, in the nearsurface region the metal (Fe,Co) concentration is less than
85 % while the metalloid (B,Si) concentration is higher than
that corresponding to the bulk. The field distribution corresponding to the bulk of the metallic glass is al,so shown by a
dotted line in Fig. 1 (e). This distribution is obtained from
the transmission M6ssbauer measurement. It is clearly seen
that in this case there is a definite contribution of a field value
of 330 kOe. Apart from the variation in quenching rates
across the sample width during the formation process and
SUbsequent atomic relaxations, other factors such as incorporation of oxygen could be responsible for the observed
differences in the coordinations in the bulk and in the surface
layers.
4160
J. Appl. Phys., Vol. 59, No. 12, 15 June 1986
)
........~---""\...-.---4
Velocity (llYTl/sec.)
A. Analysis of as-quenched metallic glass
~~c
210
ENERGY (eV) ___
FIG. 2. Characteristic boron [IS] contribution in the results of depth profiling x-ray photoemission studies on the as-quenched and oxidized metallic
glass samples. The curves labeled (I), (2), (3), (4), and (5) correspond to
depths ofO, 25, 50,100, and 200 A below the shiny surface. (a) is for the asquenched metallic glass while (b) and (c) are metallic glass samples oxidized at 100 and 300 ·C, respectively, for ~ h each.
The boron contributions to the XPS spectra for the case
of the as-quenched metallic glass are shown in Fig. 2. Figure
2(a) shows that on the surface of the as-quenched glass boron shows a positive shift of3.1 eV of its IS level with respect
to the value of 187.7 eV reported for pure boron. 22 - 24 Oxidation states of boron are known to lead to positive shift in the
case of H 3 B0 3 (192.8 eV)23 and B 2 0 3 (193.4 eV).23 The
iron, cobalt, and oxygen contributions to the XPS spectra for
the case ofthe as-quenched samples appear as given in Fig. 3,
4, and 5, respectively. In the case of Fe on the surface [Fig.
3(a)] the 2P3/2 peak occurs at 709 eV, signifying the presence ofthe Fe 2 + charge state. The 2P312 level corresponding
to Co on the surface (Fig. 4(a) 1 shows a peak at 782.2 eV
while the oxygen IS level is found to appear at 529.8 eV [Fig.
5 (a) 1. All peaks in the XPS results are broad as may be
expected and the oxygen peak is asymmetric. The asymmetry can be attributed to the presence of the contributions at
- 53 I and 532 eV. These data indicate that at the surface no
known stoichiometric oxides of Co, S, and Fe exist. As the
deeper layers are probed, however, changes appear in the
Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4160
(al
(a)
5
4
......"""":...:...-........ 3
................;..-" 2
1
1
( b)
~---5
_ _ _- - 4
>-
I<J)
z
UJ
--..:.~- 3
I-
Z
_ ............. 2
775
785
BINDING
3
2
700
710
720
795
805
ENERGY (ell) _
FIG. 4. Characteristic cobalt (2P)f2) contribution in the results of depth
profiling x-ray photoemission studies on as-quenched and oxidized metallic
glass samples. The curves labeled ( 1), (2), (3), (4), and (5) correspond to
depths 0[0,25,50,100, and 200 A below the shiny surface. (a) is for the asquenched metallic glass while (b) and (e) are for metallic glass samples
oxidized at 100 and 300 'C, respectively. for ~ h each.
730
BINDING ENERGY (eV)_
FIG. 3. Characteristic iron (2P3/2) contribution in the results of depth profiling x-ray photoemission studies on the as-quenched and oxidized metallic
glass samples. The curves labeled (1), (2), (3), (4), and (5) correspond to
depths of 0, 25, 50, 100, and 200 A below the shiny surface. (a) is for the asquenched metallic glass while (b) and (e) are for metallic glass samples
oxidized at 100 and 300 ·C, respectively, for ~ h each.
nature of the chemical environment. In the present case,
depth profiling was carried out by using lO-keV Ar+ ions at
glancing incidence. The results of depth profiling studies
which have been compiled in Fig. 6 indicate that oxygen is
present in the as-quenched sample up to a depth of - 300 A
below the sample surface [Fig. 6 (a) ]. As seen from the figure, the ratio of Fe/O increases with depth, showing less
incorporation of oxygen in the deeper layer of the surface of
the metallic glass. The ratio of FelB for the as-quenched
glass [Fig. 6(a)] also increases in the deeper layers. This
shows that in the as-quenched sample boron segregates near
the surface. At greater depth, the ratio ofFelB remains more
or less constant as a function of depth. The chemical states of
4161
J. Appl. Phys., Vol. 59. No. 12. 15June 1986
Fe, Co, and B also show changes along the depth below the
surface. The chemical state of boron is changed after profiling up to 50 A below the surface and it shows a broad peak
located around 186.3 eV which shifts to 187.5 eV for deeper
layers [Fig. 2 (a) ]. It is interesting to point out that in the
case of the crystalline phase corresponding to the metallic
glass, no such shift is observed. It has been observed that in
amorphous FeB alloy22 lSI/2 binding energy of boron does
not change appreciably with composition. However, boron
in the B4 0 7 radical does show a negative shift. Therefore, the
negative shift of the transition to lS 1/2 level of boron can be a
result of an interplay of the dissolved oxygen, the amorphous
structure, and the presence of Co. In the case of Fe [Fig.
3(a)], the peak at 705.7 eV starts appearing as the deeper
layers are probed and the area under this spectral contribution increases with depth while the contribution of the peak
at 709.0 eV decreases. Further analysis confirms that the
peak appearing at 709.0 eV corresponds to the coordination
of Fe with oxygen. Since such a peak is absent in the case of
Fe-B amorphous alloys the possibility of the occurrence of
any coordination of Fe and B can be eliminated. In the case
of Co, a broad peak centered around 782.2 eV is seen for the
Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4161
c)
7
6
l.3
09
0.5
O.t
3
(b)
2.9
2.5
t
(/)
9
....
<!
cr
2.1
1.7
~3
0.9
05
7
6
!P
0::
«
1.
(al
200
25
--DEPTHIA) -
FIG. 6. The plot of variation in the area ratios of peaks in x-ray photoemission studies of (a) as-quenched metallic glass and (b) and (e) the metallic
glass samples oxidized at 100 and 300 'C, respectively, for! h each. The
curve (-e-) denotes the area ratio variations for Fe/B. while (-.-) and
(-.-) denote the area ratio variation for Fe/O and Fe/Co, respectively.
(c )
525
535
545
5'55
BINDING ENERGY (eV) _
FIG. 5. Characteristic oxygen [IS) contribution in the results of depth profiling x-ray photoemission studies on aso{juenched and oxidized metallic
glass samples. The curves labeled (1), (2), (3), (4), (5), (6), and (7) correspond to depths of 0, 25, SO, 100, 200, 300, and 500 A below the shiny
surface. (a) is for the as-quenched metallic glass while (b) and (c) are for
metallic glass samples oxidized at l()() and 300 'C, respectively, foq h each.
surface of the as-quenched metallic glass, while its shifts
towards 781.5 eV with the appearance of another peak at
776.9 eV, as deeper layers are probed [Fig. 4(a)]. Beyond
-100 A, the ratio of the areas corresponding to the peaks at
776.9 to 781.2 eV increases. The peak corresponding to
776.9 eV can be associated with Co in an oxygen-free environment, whil.e the other to the presence of oxygen coordinated with Co. This indicates that both Fe and Co are in an
oxidized state near the surface region. Thus, the depth profiJ.ing of as-received Metglas (0) shows that native oxide is pres4162
J. Appl. Phys., Vol. 59, No. 12. 15 June 1986
ent on the surface almost up to a depth of - 300 A. Beyond
300 A, core level spectra do not indicate confirmedly the
presence of any phases or compounds. Thus, XPS and
CEMS results show that the surface of the as-quenched glass
is metanoid-rich as compared to the bulk and the Fe, Co, and
B atoms are coordinated with oxygen in the near-surface
region.
B. MetaUic glass oxidized at 100·C
When the metallic glass is oxidized at l00'C for! h the
field distribution corresponding to its CEMS spectrum [Fig.
1(f) J indicates that the low-field components are reduced,
while the high-field components are increased as compared
to the case of the as-quenched metallic glass. These changes
are ofsmaU magnitude. In view of the fact that Co increases
the hyperfine field at the Fe site, this result shows that in the
region of the sample which contributes to the CEMS spectrum, the coordination of Fe and Co exhibits an overall increase. This CEMS result brings out that the process of local
reordering is initiated in the near surface region, even at
l00·C. The XPS spectra also show changes in their features
as compared to those of the as-quenched metallic glass: The
chemical shifts remain almost the same as those in the case of
the as-received metallic glass as seen in Fig. 2-5. The area
ratio ofCo/O decreases in the near surface region while the
ratio Fe/B increases [Fig. 6 (b) J. This indicates the onset of
the oxidation process in which oxygen combines mainly with
Fe and Co atoms diffusing out toward the surface.
Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4162
c. Metallic glass oxidized at 300 and 390 ·C
-
t
r--
•
VI
-..
.."
........
t118.6
.1,17.0 ".
I
o
X
-;165.6
C
~
o
....... 'Pt
..---""""""" .--...
.....
~'62.1,
o
.,...
.s:>
~ 139
'.
\.,
z
\13.5 "~or "
-8 -I,
0
I,
8
Velocity (mm/sec.)
l.8
21,0
432
624
H(kOe)
FIG. 7. Conversion electron Miissbauer spectra of shiny surface of the metallic glass annealed at (a) 300, (b) 390, and (c) 44O'C for 10 h each. The
hyperfine field distributions [P(H) curves] corresponding to the CEMS
spectra in (a), (b), and (c) are shown in (e), (f), and (g), respectively.
at 300 "C in vacuum for 10 h. The field distribution is very
much like that of the as-quenched metallic glass [Fig. 1 (e)].
Thus, it is clear that the cause of the observed crystallization
process in the case of the sample oxidized at 300·C is the
incorporation of oxygen from air in the near surface region.
It has indeed been reported by Buravikhin et al. 25 that in the
case of FeCo alloys the crystallization of a - (Fe,Co) occurs at a temperaure which is lower in the presence of oxygen
---------
.... .-.."'...
(d) 12
(a)
When the samples are oxidized at 300 and 390 ·C for! h
in each case, the CEMS spectra and the field distributions
[Figs. 1(g) and 1(h)] appear to be significantly different as
compared to the cases of the as-quenched sample and the
sample oxidized at l00·C for! h. Specifically, the spectral
lines appear to be considerably sharper and the field distribution curves exhibit resolved contributions of hyperfine field
components. These features are signatures of the occurrence
of a crystallization process, at least in the near-surface region
of the sample which is probed by the CEMS technique. The
peak at - 348 kOe in the field distribution can be attributed
to the presence of crystalline a - (Fe,Co) phase with an
average Co concentration of -12 at. %20; while the peaks
situated at 216 and 265 kOe signify the presence of an F e3 Clike structure of Co-rich (Fe,Co)3B. These results can be
compared with those of Klein et al.,)~ who have reported
that during recrystallization of Co-rich (Fe,Co,B) glasses,
crystalline (Fe,CohB structure appears as a precursor metastable phase prior to precipitation of a-(Fe,Co) and (Fe,
CO)2B phases. 9 In addition to these observations which are
common to the samples oxidized at 300 and 390·C in our
case, the sample oxidized at 390 ·C also shows the presence
of some high-field components in the field distribution
which correspond to oxide phases. Since these oxides are
likely to contain two or more elements of the glass, it is not
expected that the observed hyperfine field would match exactly with the values which correspond to any of the phases
of pure iron oxides.
It is now interesting to compare the Mossbauer results
on the samples oxidized at different temperatures with the
samples annealed at the corresponding temperatures in
vacuum. Figures 7(a) and 7(d) show CEMS spectrum and
the corresponding field distribution for the sample annealed
J a1
..
( b)
,"'IMroow\. .....
~-.A •
( c)
~
(d)
~--"---
)\
VI
C
c
(e)
~
FIG. 8. X-ray diffraction patterns of (a)
as-quenched metallic glass and (b), (c),
(d), (e), and (0 the metallic glass samples annealed at 300, 390, 440 'C, and
oxidized at 300 and 390 'C, respectively,
for! h each.
(1)
~--""-,~ .J"-.
84
82
44
66
46
64
--"'42
28
26
2
29 (deg.)
4163
J. Appl. Phys., Vol. 59, No. 12, 15 June 1986
Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4163
than the temperature of crystallization for thermal treatment in vacuum. The field distribution of the sample oxidized at 390·C [Fig. 1(h) J appears to be quite similar to
that of the completely recrystallized (annealed at 44O·C for
10 h) sample shown in Fig. 7(f). The only important difference is the observation of certain high-field oxide components in the case of the sample thermally treated in air.
In order to confirm that crystallization has occurred in
the surface layers of the samples oxidized at 300 and 390 ·C
for! h in each case and that no observable crystallization has
occurred in samples annealed at the same temperature for
the same duration of time in vacuum, we performed x-ray
diffraction studies on these samples (Fig. 8). Also, since the
CuKa emission line is highly absorbed in the 2605 CO metallic glass, the diffractometer scans are sensitive to the top few
microns on!y.26 Thus, it is extremely interesting to observe
that the samples annealed in vacuum do not show any sharp
lines in the diffraction spectra while the samples heated in air
do show well defined diffraction patterns. In fact the sample
oxidized at 300·C exhibits a line pattern superimposed on
the background intensity corresponding to a glasslike structure, indicating the onset of crystallization, while the sample
oxidized at 390 ·C shows a well-defined diffraction pattern
without an appearance of a broad structure characteristic of
an amorphous state. Contrary to this, the diffraction patterns corresponding to the samples annealed in vacuum at
300 and 390·C for! h almost resemble the pattern corresponding to the as-quenched metallic glass itself. In the light
of the fact that the oxidized samples are crystall.ized in the
near surface region, it is interesting to compare the diffraction patterns of the oxidized samples with the pattern corresponding to the metaUic glass sample annealed at 440 ·C (i.e.
above Tg ) for! h in vacuum. It is observed that the spectral
lines appear to have the same positions in both the cases
except for the appearance of a very small additional contribution of a diffraction peak at 2(} value of 28. 70 in the sample
oxidized at 390°C, The three x-ray diffraction peaks [viz.
the ones at 2(} value of 44.8· (110),65.02° (200), and 82.18·
(211) J in all of the above cases correspond to the a-Fe
phase. Considering that the atomic structure factors for cobalt and Fe atoms are only slightly different, a substitutional
solid solution of cobalt in Fe would indeed lead to the same
x-ray pattern. This indicates that in the physico-chemical
processes associated with the thermal treatment of metallic
glass in air, the oxygen atoms play the role of agents which
disturb the chemical balance of the glassy state leading to
crystallization.
In the XPS results for the sample oxidized at 300·C
[which are shown in Figs. 3(c), 4(c), and 5(c)] for ~ h,
boron is seen to be absent up to a depth of 300 A, while the
main contribution to the spectrum is seen to emerge from
Fe,Co, and 0 atoms. This suggests that the oxidation process
takes place by preferential out-diffusion of Fe and Co atoms.
It may be noted that all the peaks in the XPS spectra are
broad., which indicates that there is a considerable degree of
inhomogeneity in the near surface region of the sample. The
peak corresponding to Fe is situated at 710 eV for the case of
the top surface, while it shifts towards - 709 eVas the depth
up to - 500 A is probed. This shows that the average coordi4164
J. Appl. Phys., Vol. 59, NO. 12, 15June 1966
TABLE I. Binding energies of several core levels.
Binding energy (eV)
Iron
Fe
FeB
a-Fe2O J
a-FeOOH
Co
CoO
Co 3O.
Co(OH)2
B,03
H 3B03
2P3/2
Cobalt
2P 3 / 2
706.82
706.65
710.97
711.44
Boron IS
Oxygen IS
187.65
529.98
531.4
778.2
779.5
779.6
781.3
193.4
192.8
531.2
530.8
531.6
533.0
533.0
nation of Fe with 0 decreases with increasing depth. Similarly, as seen from Fig. 4(c), the binding energy corresponding
to Co 2P312leveJ starts appearing at - 783 eV for the surface
and it reaches an energy value of - 781 eV for the 500-A
deep layer below the surface. Figure 5(c) shows the variation in the oxygen IS / 2 binding energy, from 528.3 eV for
the surface to - 529.7 'eV for the layer at a depth of 500 A.
Typically, the peak structure for oxygen is broad and asymmetric showing the presence of a contribution from the higher side of the binding energy. Table I reproduces the data
from Ref. 27 and 28 for Co and Fe oxides to anow a fruitful
comparison with the present observations. The comparison
indicates that one or more of the following oxide phases
could be present on the surface: a-FeOOH,
a-Fe2 0 3 ,Fe30 4 ,"7 CoO,CO(OH)2' and Co3 0 4 -like2H complexes. Since hydroxides tend to increase the binding energy
of Fe and Co atoms, while decreasing that of the oxygen
atom, the trend of binding energy as a function of depth
suggests that the hydroxide contribution decreases in content as one goes away from the surface.
O. Resistivity measurements at 1OO·C in air
In order to understand the changes in the composition
and hence the electronic properties of the near surface layers
due to thermal treatment in air, we performed measurements of the time dependence of resistance at fixed values of
temperature. An osciHatory variation was observed in the
resistance value as a function of time and these oscillations
were seen to damp out for observations performed over an
extended period. When averaged over a number of such
measurements, the characteristic of the resistivity time curve
appears as shown in Fig. 9. Such variations can be attributed
to an interplay of the diffusion of metal ions toward the surface, leading to a decrease in resistance, and the oxidation of
the near surface layers, leading to an increase in the same. On
the basis of this data as wen as the results of CEMS and XPS
studies some comments can now be made regarding the nature of the oxidation process in the glassy metal.
The XPS results show that on the shiny surface (away
from the wheel) of the as-quenched metallic glass, boron is
coordinated with oxygen in a configuration close to B20 3 Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4164
(a)
3S
t 112~
12,
...g
., :", " :: " ,,::;':
:··/.~'·;:F ,;.: \;''/.:,:
109.6
'j
<:;\}.~~A.+.--Jh-:::-A:.-rr-;
(b)
x
vi
.ci
....
..
<{
U
E 13'18
<{
Z
W
z 17
t e)
:::J
~
II)
Vi
134.6
W
0:::
!
131J.
11
~-~8~~-4~~O~~4~~8-L~~~,*,~~r
Velocity (mm/sec.)
90
210
330
450
570
- - Time (min) _
FIG. 9. Sheet resistance variation with time for oxidation of metallic glass in
air at a fixed temperature of lOO·C.
like23 structure, although distorted stoichiometrically. The
boron-oxygen bonds have a more covalent character as
compared to the Fe-O or C<r-O bonds (the electronegativity
difference being smaUer between boron and oxygen) and
hence it is difficult to break the oxygen coordinated boron
compound on the surface by low temperature thermal treatment. When oxidation is carried out at 100·C the process
occurs in the range of logarithmic growth 29 of the oxide. In
this case the migration of meta'! ions takes place under the
influence of the electric field created by the absorbed oxygen
atoms, which acquire negative charge from the metal due to
their strong electron affinity. Thus, preferential oxidation of
Fe and Co can take place in the near surface layers. The Fe
and Co atoms are also coordinated with oxygen in the asquenched metallic glass at the surface; however, this coordination is not complete and a higher concentration of absorbed oxygen can be effective in pulling out the metal
cations. The observation of a systematic variation in the surface resistance of the metallic glass oxidized at a fixed temperature is consistent with the above explanation. It is
known that in the logarithmic growth region,29 microscopic
cavities are created at the oxide-metal interface. The rate of
growth ofthe oxide film is limited by the rate of diffusion of
the meta] atoms across such cavities leading to a considerably slower oxide growth. Further, for higher thickness, the
electric field reduces and limits the diffusion of the cations.
This explains the damping of oscillation in the resistancetime curve at a fixed temperature and observation of a constant value of resistance subsequent to oxidation in air for
about 6 h.
In addition to the experiments discussed in the earlier
paragraphs, we also performed some studies on the influence
4165
::f/t// ;~~/:},:?>~. :':'\/:N1I'tI':\cr.--""~~cl----i
J. Appl. Phys., Vol. 59, No.12,15 June 1986
FIG. 10. Conversion electron Mossbauer spectra of (a) as-quenched metallic glass oxidized at 100 'C, (b) metallic glass annealed at 300'C and oxidized at 100 'C, and (c) metallic glass annealed at 390'C and oxidized at
100 'C, for 10 h each. The hyperfine field distributions [P(B) curves 1corresponding to the CEMS spectra in (a), (b), and (cl are shown by dashed
lines in (d). (e). and (fl, respectively. The continuous lines in (d). (e). and
(f) show the PCB) curves corresponding to CEMS spectra of the asquenched metallic glass and the metallic glass samples annealed at 300 and
390 'c, respectively, for 10 h each.
of the thermal treatment of the metallic glass in air within
the recommended temperature range for its use in applications. Typically, the as-quenched glass was heated in air at
l00·C for 10 h and the CEMS spectrum of the sample was
recorded subsequent to its natural cooling to room temperature. Itmaybec1earlyobservedfromFig.l0(d)-(f) (which
show the field distribution curves for the as-quenched and
the thermally treated samples) that the thermal treatment of
metallic glass even within its recommended operating temperature range leads to transformations in the near surface
layers. Specifically the high-field components tend to increase at the cost of the low-field components indicating a
buildup of Fe-Co coordination.
Considering the fact that the metallic glasses used in
actual applications are generally annealed at a temperature
of approximately (Tg - 100 ·C)3,4 for removal of deposition-induced stresses and for improvements in magnetic
properties, we also performed CEMS studies on the metallic
glass samples vacuum-annealed at - 300·C for 10 hand
subsequently heated in air at l00·C for 10 h. It is very interesting to note from Fig. 10(e), which shows the field distribution curves corresponding to the as-annealed and subsequently thermally treated samples, that there is very little
influence of the thermal treatment in air on the hyperfine
field distribution and hence presumably on the atomic arrangements in the surface layers of the as-annealed
Metglas® . This brings out that the annealing treatment of
the metallic glass at (Tg - ] 00 ·C) prior to its use in applications is a useful method to render a thermal stability to its
surface against environmental reactions in addition to its
importance in improving the other desired properties of the
Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4165
metaUic glass. Finally, from the stand point of academic interest, it is useful to bring out the influence of preannealing
treatment of metallic gl.ass in vacuum at 390·C [which is
closer to Tg than (Tg - lOO·C) in the present case] on the
behavior of the surface layers under thermal treatment in air
at 100·C for 10 h. Figure 10(f) shows the field distribution
curves for the two cases and it is very clearly seen that preannealing at this temperature leads to significant transformations during the low-temperature oxidation process, with
emergence of some high-field components (440, 480, and
520 kOe) which are representative of oxide phases.
IV. CONCLUSiONS
The oxidation kinetics and associated structural and
chemical transformations in the Fe67Co,sB,4Si, metallic
glass have been investigated by using the techniques of conversion electron Mossbauer spectroscopy (CEMS), x-ray
photoemission spectroscopy (XPS), x-ray diffraction
(XRD), and resistivity measurements. It has been shown
that oxygen incorporation in the surface layers due to thermal treatment at temperature below Tg leads to recrystallization as weH as precipitation of complex oxide phases of the
constituents of the metallic glass. The phases have been identified by analyzing the CEMS and XPS results, and possible
mechanism responsible for the observed effects are discussed. By subjecting the metallic glass samples in their asquenched as well as annealed forms to prolonged thermal
treatment in air at a temperature of l00·C (which is within
the recommended operating temperature range for the
glass), it has been shown that a preanneal treatment of the
metallic glass at a typical temperature of - Tg - l00·C
renders thermal stability to its surface layers.
ACKNOWLEDGMENTS
The authors are grateful to the Department of Science
and Technology and Department of Atomic Energy, India
for financial support for the work. One of us (YVB) is
thankful to the Council of Scientific and Industrial Research, (India) for awarding research fellowship. The auth-
4166
J. Appl. Phys., Vol. 59, No. 12, 15 June 1986
ors are also thankful to P. M. Raole of the Indian Institute of
Technology, Bombay, and the authorities of the National
Chemical Laboratory, Pune, for help in obtaining the x-ray
photoemission and x-ray diffraction data, respectively.
'H. N. Ok and A. H. Morrish, Phys. Rev. B 22.3471 (1980).
2A. G. Cullis. J. M. Poate, and J. A. Borders, Appl. Phys. Lett. 28, 314
(1976).
3J. J. Gilman and H. J. Leamy, eds., Metallic Glasses (American Society of
Metals, Cleveland, OH, 1978).
4H. S. Chen, Rep. Prog. Phys. 43. 23 (1980).
'F. E. Luborsky, ed., A.morphous Metallic Alloys (Butterworths, London,
1983).
6A. K. Bhatnagar and N. Ravi, Phys. Rev. B 28,359 (1983).
7L. A. Davis, R. Ray, C. P. Chou, and R. C. O'Handley, Scr. Metal!. 10,
541 (1976).
"B. Bhanu Prasad, A. K. Bhatnagar, D. Ganesan, R. Jagannathan, and T.
R. Anantharaman, J. Non-Cryst. Solids 61"'()2, 391 (1984).
'13. Bhanu Prasad and A. K. Bhatnagar, J. Magn. Magn. Mater. 31-34,
1479 (1983).
'0c. A. Pampillo and H. S. Chen, Mater. Sci. Eng. 30,181 (1974).
IIF. E. Luborsky, Proceedings of lInd International Conference on Amorphous Magnetism, edited by R. A. Levis and Hasegawa (Plenum, New
York, 1982).
'2U. Gonser, ed., Mossbauer Spectroscopy (Springer, Berlin, 1975), Vol. 5.
'3V. G. Bhide, Mossbauer Effect and Its Applications (Tata-McGraw-Hill,
New Delhi, 1973).
14D. C. Champeney, Rep. Prog. Phys. 42. 1017 (1979).
"B. Window, J. Phys. E 4,401 (1971).
'61. Vincze, D. S. Boudreaux, and M. Tegze. Phys. Rev. B 19, 4896 ( 1979).
17M. Eibschutz, M. E. Lines, and H. S. Chen, Phys. Rev. B 28, 425 (1983).
'8H._P. Klein, M. Ghafari, M. Ackermann, U. Gonser, and H.-G. Wagner.
Nucl. Instrum. Methods 199, 159 (1982).
'''R. Oshima and F. E. Fujita, Jpn. I. Appl. Phys. 20, 1 (1982).
20C. E. Johnson, M. S. Ridout, and T. E. Cranshaw, Proc. Phys. Soc. 81.
1079 (1963).
2'U. Gonser, M. Ghafari, M. Ackermann, H.-P. Klein, J. Bauer, and H.-G.
Wagner, Proceedings of IVth International Conference on Rapidly
Quenched Metals, Vol. II, edited by T. Masumoto and K. Suzuki (Japan
Institute of Metals, Sendai, 1981), p. 639.
22E. Cartier, Y. Bauer, M. Liard, and H.-I. Guntherodt, J. Phys. FlO, L21
(1980).
23 D. J. Joyner and D. M. Hercules, J. Chem. Phys. 72, 1095 (1980).
2'0. Johnson, D. J. Joyner, and D. M. Hercules, J. Chern. Phys. 84, 542
(1980).
25V. A. Buravikhin, A. T. Ushakov, V. V. Litvintsev, and A. T. Titov, Fiz.
Met. Metalloved 36, 62 ( 1973 ) .
26M. Choi, D. M. Pease, W. A. Hines, J.1. Budnick, G. H. Hayes, and L. T.
Kabacotf, J. Appl. Phys. 54. 4193 (1983).
27K. Asami and K. Hashimoto, Corros. Sci. 17, 559 (1976).
28c. R. Brundle, T. J. Chuang, and D. W. Rice, Surf. Sci. 60. 286 (1976).
290. Hunderi and R. Bergersen, Corros. Sci. 22, 135 (1981).
Bhandarkar et al.
Downloaded 27 Feb 2012 to 14.139.97.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
4166