Iron suicide formation from fe thin-film electrodeposition on hydrogen-terminated si(111)

Article abstract of DOI:10.1149/1.2097607

We report on the surface morphology, crystalline, and electronic properties of Fe thin films potentiostatically eletrodeposited on hydrogen-terminated Si(111) substrates. Spontaneous reaction between Fe deposits and hydrogen-terminated Si surfaces have been observed at room temperature with formation of polycrystalline suicides Fe₂Si and Fe₅Si₃, together with metallic Fe having a body-centered cubic structure. X-ray photoemission spectroscopy and X-ray diffraction measurements have been performed to investigate the thermal stability between Fe and Si under high-vacuum thermal treatments up to 300°C.

Full text of DOI:10.1149/1.2097607

C808
Journal of The Electrochemical Society, 152 ͑12͒ C808-C812 ͑2005͒
0013-4651/2005/152͑12͒/C808/5/$7.00 © The Electrochemical Society, Inc.
Iron Silicide Formation from Fe Thin-Film Electrodeposition
on Hydrogen-Terminated Si(111)
J. Zarpellon, H. F. Jurca, J. J. Klein, W. H. Schreiner,
,z
*
N. Mattoso, and D. H. Mosca
Departamento de Física, Universidade Federal do Paraná, 81531-990 Curitiba PR, Brasil
We report on the surface morphology, crystalline, and electronic properties of Fe thin films potentiostatically eletrodeposited on
hydrogen-terminated Si͑111͒ substrates. Spontaneous reaction between Fe deposits and hydrogen-terminated Si surfaces have been
observed at room temperature with formation of polycrystalline silicides Fe2Si and Fe5Si3, together with metallic Fe having a
body-centered cubic structure. X-ray photoemission spectroscopy and X-ray diffraction measurements have been performed to
investigate the thermal stability between Fe and Si under high-vacuum thermal treatments up to 300°C.
© 2005 The Electrochemical Society. ͓DOI: 10.1149/1.2097607͔ All rights reserved.
Manuscript submitted January 3, 2005; revised manuscript received July 12, 2005. Available electronically October 18, 2005.
Metal deposition on semiconductor surfaces has been extensively
investigated primarily because of the technological importance of
Schottky contacts. Notably, the direct metal deposition onto Si is of
interest to the microelectronics industry, for example, to improve
interconnects in ultralarge-scale integration metallization,1 and to
build Schottky diodes.2,3 The integration of magnetic materials and
nanostructures with semiconductor electronics is an increasingly in-
teresting approach to explore spin injection from metal/
semiconductor interfaces4 for a new generation of electronic
technology.5 In spite of this, surprisingly little work has been done
on direct electrochemical deposition of ferromagnetic materials on
Si surfaces, e.g., Co6 and Ni7 nanoclusters, Co films,8-10 Ni,11,12
NiCu alloys films,13 Co–Ni–Cu/Cu multilayers,14,15 NiFe Permalloy
films,15,16 and Fe on porous Si.17,18 In this work, we investigated the
specific case of Fe deposition on hydrogen-terminated n-type
Si͑111͒ surfaces and our attention was focused on the morphology
as well as the crystalline and electronic structures of the Fe deposits.
mately 10 nm/h. The spectra were collected using Mg K␣ radiation
and the overall energy resolution was approximately 0.8 eV. Atomic
force microscopy ͑AFM͒ were performed using a scanning probe
microscope ͑Shimadzu SPM-9500J3͒ operating in the contact mode.
Results and Discussion
Figure 1 shows a typical CV obtained at a sweep potential rate of
10 mV/s in the solution containing 10 mM Fe͑NH4͒2͑SO4͒2 onto
hydrogen-terminated Si͑111͒ surface. The potential scan in the ca-
thodic direction reveals that the reduction of the Fe2+ ions has onset
potential at −1.05 V, leading to a Fe nucleation wave. Superim-
posed on the reduction wave is seen a rapid increase of the cathodic
current due to hydrogen evolution. We believe that the spread of
reduction potential between −1.05 and −1.30 V associated with Fe
deposition is related to a charge-transfer layer with complexed and
solvated Fe ions as well as different activated sites for nucleation at
the silicon surface. Evidence of a complex mechanism in the charge
exchange processes appears when the potential scan is performed in
the anodic direction.
Experimental
Fe deposits were potentiostatically deposited from aqueous
acidic solutions containing 10 mM Fe͑NH4͒2͑SO4͒2 chemical grade
directly on n-type Si͑111͒ wafers with resistivity of 50–80 ⍀ cm.
All electrochemical experiments were performed with a commercial
galvanostat/potentiostat ͑EG&G model 273A͒ using a conventional
cell with three stationary electrodes. A platinum disk was used as
counter electrode, whereas an Ag/AgCl reference electrode was em-
ployed. Rectangular slices of Si wafers with area of 2.8 cm2 were
used as working electrodes after removal of the native oxide by
chemical HF-etching prior to electrodeposition experiments. The
HF-etching is known to lead to hydrogen-terminated Si͑111͒ sur-
faces rather stable to oxygen exposure at room temperature.19
Gallium-aluminum alloying was used on the rear of the Si wafers to
form a good ohmic contact. The pH was kept at 2.5 by addition of
an appropriate amount of H2SO4. All potentials of this work are
given with respect to the Ag/AgCl reference electrode. Samples
grown from 10-mM Fe͑NH4͒2͑SO4͒2 solutions at cathodic potential
of −1.2 V for 10 min give rise to homogeneous Fe deposits which
are adherent with metallic shine and low conductivity. These depos-
its become grayish with an insulating character when stored under
atmospheric conditions.
For the potential scan in the anodic direction, the nucleation
cycle starting from the potential scan in the cathodic direction fin-
ishes at around −1.0 V after a strong hydrogen evolution. At the end
of the voltammogram, i.e., potential region of −1.0–0.0 V, two
peaks of anodic current are observed. Even if it is plausible that the
anodic oxidation of Si could occur at these potentials,17,20 the two
anodic peaks are likely the oxidation of Fe to Fe2+ and Fe2+ to Fe3+
͑or Fe to Fe3+ in some part͒, as indicated by XRD measurements of
specimens ͑not shown͒ prepared at these anodic currents. A simple
visual inspection clearly confirms that anodic peaks promote the
dissolution of Fe deposits by exposing bare Si surface. Charge ex-
change processes with coupling between the anodic dissolution of
silicon and cathodic metal deposition from the metallic ion solutions
have been recently discussed.20 However, a fluoride solution has
been used and therefore the discussion cannot simply apply to our
present experiments. The time evolution of the cathodic current for
potentiostatic deposition at −1.2 V is shown in the inset of Fig. 1.
After the initial transient, the current decreases and then reaches a
stationary value associated with diffusion limit. Certainly, the sta-
tionary current coincides with the maximum rate of ͑solvated and
complexed͒ ions arriving at the surface layer, and current transient
strongly depends on the applied overpotential between electrodes of
the cell and the ion concentration in the electrolyte. As we discuss
below, the initial growth seems to occur by random nucleation with
an interfacial silicide formation assisted by diffusion-limited aggre-
gation of the ions.21
X-ray diffraction ͑XRD͒ analyses were performed in the conven-
tional Bragg-Brentano geometry using Co K␣ radiation to avoid
fluorescence. X-ray photoelectron spectroscopy ͑XPS͒ measure-
ments were carried out using a commercial VG ESCA3000 system
with a base pressure of 3 ϫ 10−8 Pa, while XPS profiles were
obtained by sputtering with Ar+ ions ͑3 keV, 5 ␮A͒. A previous
calibration of the sputtering rate for pure iron films gives approxi-
We can only speculate how the iron deposition could occur
through the displacement of surface hydrogen atoms which termi-
nate the Si͑111͒ surface. Calculations using Nerst equations for the
splitting of Si–H bonds at pH 2 and 25°C by assuming that
the dislocation of the equilibrium balance arises only from the
increase of the H+ active ions results in the reaction ϵSi–H →
*
Electrochemical Society Active Member.
z E-mail: mosca@fisica.ufpr.br
Downloaded on 2014-05-06 to IP 169.236.37.160 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑12͒ C808-C812 ͑2005͒
C809
Figure 1. Cyclic voltammogram obtained from 10-mM Fe͑NH4͒ ͑SO4͒
2
2
solution. The inset shows a plot of the current as a function of deposition
time.
ϵ Si*͑active͒ + H+ + e− is −1.1 V at pH 2 and 25°C.12,22 Our ex-
periments were performed under potential −1.2 V vs saturated
Ag/AgCl, i.e., −1.42 V vs a normal hydrogen reference electrode
͑NHE͒. Although the mechanism of the chemical deposition and
compound formation is certainly more complex and involves vari-
ous other factors, the above estimated potential value indicates that
iron deposition by galvanic displacement between Fe2+ and ϵSi–H
is thermodinamically feasible in our experiments. It is worthy of
note that a narrow electrochemical window was found for homoge-
neous and metallic Fe deposition. In our experiments, nonhomoge-
neous and insulating deposits are found when solutions with con-
centrations smaller than 10 mM Fe͑NH4͒2͑SO4͒2 or pH values
higher than 3.2 are used. We have used the ammonium sulfate trying
to increase the Fe ion discharging ability, but it does not eliminate
the hydroxide formation as shown below. In addition, the formation
of more homogeneous coatings was obtained by illumination of the
silicon electrolyte junctions.
The surface morphology was determined by AFM analysis. Fig-
ure 2a and b exhibits AFM images of Fe eletrodeposits grown on
Si͑111͒ surfaces after 5 and 10 min of Fe deposition from 10-mM
Fe͑NH4͒2͑SO4͒2 solution under cathodic potential of −1.2 V. Fe
electrodeposits onto Si͑111͒ surfaces exhibit a surface morphology
having submicrometer grains with slightly flat terraces. This micro-
structure is a possible reason for the brightness of these deposits.
Such a surface morphology is not surprising, because the nucleation
and growth of metals on semiconductor surfaces from liquids gen-
erally proceeds by the Volmer–Weber mechanism, i.e., three-
dimensional growth of metal nucleii. In our case, the Si surface
stability is promoted by hydrogen passivation; and thus even for
Si͑111͒ surfaces which are not coordinatively saturated, these sur-
faces tend to have a low surface energy, which favors a Volmer–
Weber growth mode.23
Figure 2. AFM images of Fe deposits on silicon surfaces obtained in contact
mode. Fe deposit after ͑a͒ 5 and ͑b͒ 10 min of deposition from 10-mM
Fe͑NH4͒ ͑SO4͒ solutions at cathodic potential of −1.2 V.
2
2
annealing and could be largely attributed to the high chemical activ-
ity of fresh iron surfaces under exposure to the atmosphere. The
relative amount of oxygen along the film thickness was followed by
XPS depth profiling analyses. Spontaneous formation of Fe-rich sil-
icides ͑Fe2Si and Fe5Si3͒ is also observed from XRD patterns. As in
the bulk Fe–Si system, where a continuous range of solid solutions
are formed, Fe–Si͑111͒ interface has also revealed the formation of
a large variety of Fe silicides. Bulk Si-rich silicide phases ␤–FeSi2
XRD was performed to determine the crystallographic structure
and phase formation in the electrodeposits. XRD patterns ͑␪−2␪
scans͒ obtained for the freshly as-deposited sample and two different
annealing temperatures of individual pieces of the same sample are
shown in Fig. 3. In the plot the diffraction intensity is arbitrary, but
all the XRD patterns were normalized with respect to the intensity
of the Si͑111͒ Bragg reflection positioned around 2␪ = 33.2° ͑not
shown͒, which was used to orient the samples, maximizing the dif-
fracted intensity in the experiments. For higher annealing tempera-
tures, significative amounts of iron oxyhydroxide ͑FeOOH͒, iron
oxide ͑Fe2O3͒, and SiO2 from the Si substrate are observed. These
findings are not surprising because the oxidation is known to easily
occur at polycrystalline Fe surfaces exposed to air. A fraction of
these oxide phases does not show any significant changes under
Figure 3. XRD patterns of Fe deposits. As-prepared and annealed samples
are indicating alongside normalized diffractograms.
Downloaded on 2014-05-06 to IP 169.236.37.160 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C810
Journal of The Electrochemical Society, 152 ͑12͒ C808-C812 ͑2005͒
Figure 4. Si 2p3/2-1/2 core-level spectra of thin Fe deposits for some selected
sputtering times as indicated. The dark bars drawn on the top of the panel
correspond to the binding energy range for Si 2p photoelectrons for each
compound type.23
Figure 5. Fe 2p3/2 core-level spectra of Fe deposits for some selected sput-
tering times indicated alongside spectra. The dark bars drawn on the top of
the panel correspond to the binding energy range for Fe 2p photoelectrons
for each compound type.23,24
͑orthorhombic structure͒ and ␧–FeSi ͑B20 structure͒ as well as me-
taestable Fe͑Si1−xFex͒ silicides ͑1 Ͻ x Ͻ 0͒ with a local cubic CsCl
structure have been obtained by different techniques.24,25
Fe͑Si1−xFex͒ thin films on Si͑111͒ are ferromagnetic at room tem-
perature for x values ranging from 0.15 to 0.50, and their stoichi-
ometries are found by introducing Fe vacancies in the CsCl-type
lattice. The spontaneous formation of the Fe2Si is well known, but
the silicide Fe5Si3 ͑x = 0.25͒ formation at room temperature is not
expected. Nevertheless, the XRD of our films reveals the formation
of this silicide phase with no significant changes from freshly as-
deposited samples to annealed ones. The electrochemical activity is
the agent responsible for the stability of Fe-rich silicide at the inter-
face between Fe and Si. The electrodeposition kinetics is, therefore,
different from the thermodynamics of physical vapor deposition
techniques. We believe that the large values of growth rate imposed
by us through solution concentration and large overpotential privi-
lege a random deposition with local surface diffusion to neighboring
sites, which privilege metastable silicide formation. Whereas in a
simple random deposition only metastable and nonstoichiometric
silicides could be formed, our specific growth conditions lead to
stoichiometric and thermally stable compounds.
XPS analyses indicate that all deposits do not contain impurities
and exhibit an elemental composition consisting of Fe, Si, and O
with a residual layer of N and C adsorbates. Figure 4 exhibits the
depth evolution of the Si 2p core-level spectra for a freshly depos-
ited Fe thin film. Because iron silicide bonding is based on the
rehybridization of the Fe d bands with Si sp3 bands, an appreciable
asymmetry in the Si 2p and Fe 2p core-level photoemission spectra
is expected. Oxygen has an electronegativity value approximately
twice that of those values observed for iron and silicon, so it is
expected that the Fe–Si bonds should account for a small chemical
shift toward higher binding energies with respect to bulk silicon.
Thus, the Si 2p core-level spectra are decomposed into three com-
ponents and backgrounds; i.e., pure Si component positioned at
99.5 eV binding energy, Si–O bond component centered at about
103 eV binding energy, and a component at around 100-eV binding
energy which is correlated to Fe–Si bonds. The Si–O component
drastically decreases as a function of the sputtering time, whereas
the broadening of the pure Si component for higher binding energies
indicates that a Si–Fe component is preserved and that silicides are
predominantly localized close to the interface between Fe and Si.
Figure 5 shows the depth profiling of the Fe 2p3/2 core-level
spectra for the same sample. Together with a metallic Fe component
centered at around 706.8 eV binding energy, two other components
peaking at around 710 and 713.5 eV binding energies are observed.
The large energy shifts of these two extra components toward higher
binding energies indicate that they are probably related to Fe oxide
͑FeO–Fe2O3͒ and oxyhydroxide ͑FeOOH͒ compounds.26 The Fe sil-
icides appear at slightly higher binding energy with respect to me-
tallic Fe25 and is buried in the spectra taken far from the Fe and Si
interface. As a function of the depth, the Fe 2p3/2 photoemission
spectra significantly change. Decomposition of spectra using a me-
tallic Fe component and gaussian profiles for the oxide and oxyhy-
droxide components show that the integrated fraction corresponding
to metallic Fe component increases from 10 to 42.9% after 30 min
of argon-ion bombardment. Concomitantly, Fe–O components de-
crease from 71 to 48.9% and the Fe-oxyhydroxide component di-
minishes from 20 to 8.2% for the same sputter time. This clearly
indicates that Fe oxyhydroxyde is a superfical layer resulting from
the moist air exposure after electrodeposition, whereas the Fe oxide
is more intrinsically related to the deposition mechanism. Beyond
this superficial oxygen-rich layer, metallic Fe and Fe silicide make a
predominant contribution in the XPS spectra.
The depth profile of the Fe 2p3/2 and Si 2p photoemission spectra
were also analyzed by decomposition of raw spectra in three com-
ponents after a background subtraction. Figure 6 exhibits Si 2p and
Fe 2p3/2 core-level spectra for an as-deposited sample collected after
90 min of argon ion bombardment, which results in a thin layer with
photoelectron signals probing the interface between Fe and Si. Ac-
cording to Fig. 6a, a resulting pure silicon component is accom-
paned by a Si–Fe component centered at 100.2 eV binding energy
͑labeled as A in the spectra͒ and a small Si–O component ͑labeled as
Downloaded on 2014-05-06 to IP 169.236.37.160 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑12͒ C808-C812 ͑2005͒
C811
Figure 7. Si 2p core-level spectra of Fe deposits after annealing at 300°C
during 3 h ͑bottom͒ and after sputter time given alongside spectra.
Figure 6. Si 2p3/2-1/2 and Fe 2p3/2 core-level spectra of as-prepared Fe de-
posits collected after 90 min of argon-ion bombardment. The solid curves are
measured spectra and dotted lines are resolved components after background
subtraction.
27.2% after 30 min of exposure to Ar-ion bombardment. A compari-
son between the integrated fraction of Si–Fe and Si–O components
of the sample before and after annealing for equivalent sputter time
of 30 min shows that the Si–Fe component remains practically con-
stant at ϳ27%, whereas the Si–O component, which is negligible
before annealing, increases to 34.7%. Thus, the annealing effect
does not significantly change the Fe silicide contribution but seems
to indicate that oxygen atoms diffuse toward the interface between
B in the spectra͒ centered at 103.3 eV binding energy. Concomi-
tantly, a pronounced metallic Fe component at ϳ707 eV binding
energy remains superimposed to two extra components, one cen-
tered at 708 eV binding energy ͑labeled as C͒ and another centered
at 710.2 eV ͑labeled as D͒. These two peak energies associated with
components C and D are quite close to the values reported by Ber-
ling et al.25 for silicides Fe2Si and Fe5Si3 for X-ray absorption ex-
periments. However, the FeO compound cannot be completely dis-
carded for the component labeled as D. The Fe silicide contribution
is not evidenced from the Fe 2p photoemission signal. In spite of a
low-energy resolution in our experiments, we believe that Fe elec-
trodeposition with Fe silicide intermediate compounds is demon-
strated. The effect of the oxygen on the Fe deposition and Fe silicide
formation is not completely clear in this work, but there is no evi-
dence of their accumulation at the interface Fe/silicide as reported
by Swart and Berning.27 Our results are qualitatively in agreement
with sputter depth profiles of the XPS oxygen signal shown by
Ronkel et al.17 for porous silicon substrates, i.e., during Fe deposi-
tion process the sample is oxidized. It is worth noticing that the
complete removal of the Si oxide layer with formation of a stable
H-terminated Si͑111͒ is less evident in the case of porous silicon
instead of flat Si͑111͒ surfaces. The oxygen concentration in our Fe
deposits could be related to the statement that Fe oxide formation
prevents any silicide formation.
The postgrowth temperature effect on the deposits has been in-
vestigated by depth profiles of the Si 2p and Fe 2p3/2 core-level
spectra. The typical sequence of photoemission spectra as a function
of depth for a sample annealed at 300°C for 3 h is shown in Fig. 7
and 8. The same decomposition procedure as described above was
used. The examination of the depth evolution of the Si 2p core-level
spectra shown in Fig. 7 revealed that while pure Si component re-
mains between 39.5 and 38.1%, the Si–O component decreases from
47.3 to 34.7% and the Si–Fe component increases from 16.6 to
Figure 8. Fe 2p3/2 core-level spectra of Fe deposits after annealing at 300°C
during 3 h ͑bottom͒ and after sputter times given alongside spectra.
Downloaded on 2014-05-06 to IP 169.236.37.160 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C812
Journal of The Electrochemical Society, 152 ͑12͒ C808-C812 ͑2005͒
Fe and Si. As pointed out previously, the Fe silicides formed at the
interface are thermodynamically stable even if their oxygen environ-
ments have been enhanced. Analyses of the Fe 2p3/2 core-level spec-
tra shown in Fig. 8 were also performed. For equivalent sputtering
times of 30 min, the samples before and after annealing exhibit a
reduction of the integrated area of the metallic Fe components of
42.9 and 37.2%, whereas the fraction of the Fe–O component in-
creases from 48.9 to 60.8% and the FeOOH component decreases
from 8.2 to 2.0%. We believe that these results corroborate to con-
clude that the increase of oxygen close to the Si interface is not an
oxidation process of the sample coming from the residual base pres-
sure of the quartz tube during annealing, but it could be understood
as a dissociation of the Fe oxyhydroxide, which leads to the oxida-
tion of the Fe together with the diffusion of oxygen atoms toward Si
substrate.
References
1. L. Santinacci, T. Djenizian, and P. Schmuki, Appl. Phys. Lett., 79, 1882 ͑2001͒.
2. G. Oskam, D. E. Van Heerden, and P. C. Searson, Appl. Phys. Lett., 73, 3241
͑1998͒.
3. T. Zambelli, M. L. Munford, F. Pillier, M. C. Bernard, and P. Allongue, J. Elec-
trochem. Soc., 148, C614 ͑2001͒.
4. S. Datta and B. Das, Appl. Phys. Lett., 56, 665 ͑1990͒.
5. S. A. Wolf, D. D. Awschalom, R. Buhrman, J. M. Daughton, S. von Molnár,
M. L. Roukes, A. Y. Chtchelkanova, and D. M. Treger, Science, 294, 1488 ͑2001͒.
6. M. V. Rastei, R. Meckenstock, J. Bucher, E. Devaux, and Th. Ebbsen, Appl. Phys.
Lett., 85, 2050 ͑2004͒.
7. A. Imanishi, K. Morisawa, and Y. Nakato, Electrochem. Solid-State Lett., 4, C68
͑2001͒.
8. M. L. Munford, M. L. Sartorelli, L. Seligman, and A. A. Pasa, J. Electrochem. Soc.,
149, C274 ͑2002͒.
9. Y. Souche, J. P. Levy, E. Wagner, A. Lienard, L. Alvarez-Prado, and R. T. Collins,
J. Magn. Magn. Mater., 242, 578 ͑2002͒.
10. M. L. Munford, L. Seligman, M. L. Sartorelli, E. Voltolini, L. F. O. Martins,
W. Schwarzacher, and A. A. Pasa, J. Magn. Magn. Mater., 226–230, 1613 ͑2001͒.
11. P. Gorostiza, M. A. Kulandainathan, R. Diaz, F. Sanz, P. Allongue, and
J. R. Morante, J. Electrochem. Soc., 147, 1026 ͑2000͒.
Conclusion
12. N. Takano, N. Hosoda, T. Yamada, and T. Osaka, J. Electrochem. Soc., 146, 1407
͑1999͒.
13. R. G. Delattore, M. L. Sartorelli, A. Q. Schervenski, A. A. Pasa, and S. Guths, J.
Appl. Phys., 93, 6154 ͑2003͒.
14. A. A. Pasa and W. Schwarzacher, Phys. Status Solidi A, 173, 73 ͑1999͒.
15. A. P. O’Keeffe, O. I. Kasyutich, W. Schwarzacher, L. S. de Oliveira, and
A. A. Pasa, Appl. Phys. Lett., 73, 1002 ͑1998͒.
16. L. G. Gao, P. Ma, K. M. Novogradecz, and P. R. Norton, J. Appl. Phys., 81, 7595
͑1997͒.
17. F. Ronkel, J. W. Schultze, and R. Arens-Fischer, Thin Solid Films, 276, 40 ͑1996͒.
18. C. Renaux, V. Scheuren, and D. Flandre, Microelectron. Reliab., 40, 877 ͑2000͒.
19. X. Zhang, Y. J. Chabal, S. B. Christman, E. E. Chaban, and E. Garfunkel, J. Vac.
Sci. Technol. A, 19, 1725 ͑2001͒.
In conclusion, we have shown that Fe deposition from aqueous
sulfated solution onto hydrogen-terminated Si͑111͒ surfaces could
be obtained. An iron silicide-rich layer is formed at the interface
between Fe and Si, which is buried below a metallic Fe-rich layer
and covered by a layer consisting of iron oxide and oxyhydroxide.
XRD measurements indicate the formation of polycrystalline Fe2Si
and Fe5Si3 phases concomitantly with the metallic Fe having a poly-
crystalline character with body-centered cubic structure. Fe silicide
formation was confirmed by XPS analyses of the Fe 2p and Si 2p
core-level photoemission signals. XPS experiments using depth pro-
files reveal that both metallic Fe and Fe silicide contributions in-
crease as a function of depth after removal of a surface layer con-
sisting of iron oxyhydroxides. Finally, post-growth thermal
treatments indicate that Fe silicides are stable even if diffusion of
oxygen atoms toward Si substrate occurs.
20. P. Gorostiza, M. A. Kulandainathan, R. Diaz, F. Sanz, P. Allongue, and
J. R. Morante, J. Electrochem. Soc., 147, 1026 ͑2000͒.
21. B. Scharifker and G. Hills, Electrochim. Acta, 29, 879 ͑1983͒.
22. P. Allongue, V. Costa-Kieling, and H. Gericher, J. Electrochem. Soc., 140, 1018
͑1993͒.
23. A. Zangwill, Physics at Surfaces, Cambridge University Press, Cambridge, U.K.
͑1988͒.
24. U. Kafader, P. Wetzel, C. Pirri, And G. Gewinner, Appl. Phys. Lett., 63, 2360
͑1993͒, and references therein.
25. D. Berling, G. Gewinner, M. C. Hanf, K. Hricovini, S. Hong, B. Loegel,
A. Mehdaoui, C. Pirri, M. H. Tuilier, and P. Wetzel, J. Magn. Magn. Mater., 191,
331 ͑1999͒, and references therein.
26. Handbook of X-Ray Photoemission Spectroscopy, J. Chastain and R. C. King, Jr.,
Editors, Physics Electronics Div., Perkin-Elmer, Eden Prairie, MN ͑1995͒.
27. H. C. Swart and G. L. P. Berning, Appl. Surf. Sci., 78, 77 ͑1994͒.
Acknowledgment
This work is partially financed by CNPq/PIBIC/UFPR and
PRONEX/Fundação Araucária—CNPq.
Universidade Federal do Paraná assisted in meeting the publication
costs of this article.
Downloaded on 2014-05-06 to IP 169.236.37.160 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).