APPLIED PHYSICS LETTERS 88, 113111 共2006兲
Magnetic iron silicide nanowires on Si„110…
S. Lianga兲 and R. Islam
Science and Engineering of Materials Program, Arizona State University, Tempe, Arizona 85287-1704
David J. Smith
Department of Physics and Astronomy, and Center for Solid State Science, Arizona State University, Tempe,
Arizona 85287
P. A. Bennett
Department of Physics and Astronomy, Arizona State University, Tempe, Arizona 85287
J. R. O’Brien
Quantum Design, San Diego, California 92121
B. Taylor
Physics Department, University of California, San Diego, La Jolla, California 92093-0350
共Received 30 October 2005; accepted 16 February 2006; published online 16 March 2006兲
Self-assembled iron silicide nanowires were formed by depositing 1 ML of Fe onto Si共110兲 at
700 ° C in ultrahigh vacuum. The nanowires have average dimensions of 5 nm high ⫻10 nm wide
⫻ ␮m long, as measured with ex situ atomic force microscopy. High-resolution electron microscopy
identifies the crystal structure as cubic FeSi2 with orientation FeSi2共1̄11兲 / / Si共11̄1兲,
FeSi2具110典 / / Si具110典. Magnetometer measurements show a magnetic moment of 0.3 Bohr
magneton per iron atom at 2 K. This magnetic property in metastable cubic FeSi2 nanowires opens
up the possibility for high-density data storage and logic applications. © 2006 American Institute of
Physics. 关DOI: 10.1063/1.2185610兴
Iron silicides have attracted much attention recently. Different phases of interest are shown in Table I. These include:
␣ 共metallic兲, ␤ 共semiconducting兲, ␥ and s phases 共metastable兲. The semiconducting phase is particularly interesting,
since the band gap of 0.87 eV1,2 is favorable for optical fiber
communication systems operating at a 1.5-␮m wavelength.
Among the transition-metal silicides, FeSi2 is the only reported light emitter.3,4 Iron disilicide is also the only
transition-metal silicon compound reported to occur in both
semiconducting and metallic phases. Metastable phases are
observed in thin films, and depend on the film thickness and
other growth conditions.5 The ␥ phase is particularly interesting since calculations suggest that it is ferromagnetic.6
The structure and properties of Fe-Si disilicide phases are
shown in Table I;5,7 metastable ␥, and s phases are denoted
by star.8
Transition-metal silicides are widely used as contacts
and interconnects in current microelectronics9 and potential
optoelectronic devices.10 As continuing miniaturization encounters the limits of lithography, self-assembled structures
become more and more attractive. Silicide nanowires 共NWs兲
in particular may find application as low-resistance interconnects, or as nanoelectrodes for attaching small electrically
active structures. Silicide NWs were reported for rare-earth
metals on Si共100兲11–13 The NW shape is thought to result
from an anisotropic lattice mismatch that is small
共⬍1 % 兲 in the long direction and large 共⬎5 % 兲 in the short
direction. Silicide NWs have since been observed in other
systems, such as Pt/ Si共100兲,14 Ti/ Si共111兲,15,16 and
Yb/ Si共100兲.17 He et al. have described an “endotaxial”
growth mechanism that applies to a range of transition meta兲
Electronic mail: [email protected]
als on Si共100兲, Si共110兲, and Si共111兲.18 This mechanism involves growth of the silicide into the substrate and does not
involve anisotropic lattice mismatch. The FeSi2 NWs reported here appear to be examples of this type of growth.
In this paper, we report the formation of epitaxial iron
silicide NWs with nominal dimensions: 5 nm high ⫻10 nm
wide ⫻ ␮m long. The morphology, crystal structure, and epitaxial orientation are determined using atomic force microscopy 共AFM兲 and high-resolution transmission electron microscopy 共HRTEM兲. The NWs are identified as cubic phase
FeSi2 and are shown to have a magnetic moment of 0.3 Bohr
magneton per iron atom at 2 K. This result opens up the
possibility for self-assembled single-crystal magnetic NWs
on silicon. Ordered arrays of such magnetic NWs could facilitate fundamental studies of nanoscale magnetism as well
as possibly enabling high-density information storage.19
Si共110兲 samples 共phosphorous-doped, ⬃10 ⍀ cm, miscut angle ⬍0.5°兲 were cleaned in ultrahigh vacuum by degassing for 12 h, followed by flashing multiple times at
1250 ° C for several seconds, using resistive heating. Iron
TABLE I. Structure and properties of Fe-Si disilicide phases.5,7
Phase
␣
Lattice parameter
共Å兲
␤
a = 2.684, c = 5.128
a = 9.863, b = 7.791
c = 7.833
␥a
5.389
sa
2.7
a
8
Lattice
Tetragonal
P4 / mmm
Orthorhombic
Cmca
Cubic 共fcc兲
Fm3m
Cubic 共bcc兲
Pm3m
Property
Metallic
Semiconducting
Metallic
Metallic
Metastable phases.
0003-6951/2006/88共11兲/113111/3/$23.00
88, 113111-1
© 2006 American Institute of Physics
Downloaded 24 Jun 2010 to 129.219.247.33. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
113111-2
liang et al.
Appl. Phys. Lett. 88, 113111 共2006兲
FIG. 2. 共Color online兲 Arrhenius plot showing variation of nanowire density.
shown in Fig. 2 as an Arrhenius plot. The strong temperature
dependence suggests that the NWs nucleate homogeneously
with a critical cluster size larger than one atom. We have
shown elsewhere that classical nucleation theory may be applied to a two-component reactive system such as this.20,21
According to conventional nucleation theory;20,21 the density
of stable nuclei nx for two-dimensional 共2D兲 island growth in
the complete condensation regime is given by
nx = ␩共F/D0兲共i/i + 2兲exp共− E*/KT兲,
FIG. 1. 共Color online兲 AFM images of nanowires grown at different temperatures: 共a兲 590, 共b兲 640, 共c兲 680, 共d兲 710, 共e兲 750, and 共f兲 790 ° C. Z scales
are 10 nm except 共f兲 of 20 nm.
was deposited by sublimation from a 4N-purity iron wire
onto the heated silicon substrate; this is so-called reactive
deposition epitaxy 共RDE兲. Coverage was determined in situ
using a crystal thickness monitor, with an accuracy of 20%.
Coverage values are given in monolayer 共ML兲 units, where
1 ML= 4.78⫻ 1014 iron atoms/ cm2. Temperature was determined using an optical pyrometer 共emissivity= 0.4兲, with an
accuracy of ±20 ° C. AFM images were obtained using a
Digital Instruments NanoScope III in tapping mode. A JEOL
4000EX high-resolution electron microscope was used for
electron diffraction and imaging. Magnetic properties were
measured using an MPMS XL5-AC superconducting quantum interference device 共SQUID兲 magnetometer at Quantum
Design, San Diego, CA.
The morphology of the NWs was shown to depend on
several parameters, including silicon surface steps, growth
temperature, metal deposition rate, and time. The dependence on growth temperature is shown in Fig. 1, with the
coverage and deposition time held approximately constant at
1 ML and 2 min, respectively. The NWs have a single orientation, along Si具110典. Over the temperature range from
640 to 750 ° C, the NW width is in the range of 10– 20 nm.
The length varies with growth time and nucleation density.
From the pattern of step flow, it appears that the NWs are not
impeded by encounters with steps — they simply push the
step ahead at both ends. Lengths in excess of 5 ␮m are
readily attained. At 590 ° C, the NWs are short and indistinct
due to high nucleation density; at 790 ° C, the NWs are approximately 25 nm wide and have more distinct trenches
around them due to the increased consumption of Si during
growth at this temperature.
The temperature dependence of the NW number density,
holding the deposition rate constant at 0.5 ML/ min, is
共1兲
where ␩ is a dimensionless number near unity, F is the deposition flux, D0 is the surface diffusion prefactor, i is the number of atoms in the critical cluster, and E* = 共iEd + Eb兲 / 共i + 2兲
is a weighted sum of diffusion activation energy Ed and cluster binding energy Eb. The diffusion prefactor may be written
as D0 = 共a20 ␷0兲 / 4, where a2 is the unit-cell area and ␷0 is an
attempt frequency. A cluster containing more than i atoms is
regarded as stable on the surface. Such stable clusters can
increase their numbers of atoms by capturing other single
atoms on the surface or by direct condensation. Thus, the
density of stable islands will follow Eq. 共1兲. The data have
been fitted to this equation, which yields an activation energy
of 1.8± 0.15 eV. This energy corresponds to a weighted sum
of diffusion and cluster binding energies of the moving
species.
An HRTEM micrograph showing the cross section at a
typical NW is presented in Fig. 3. From this image, we identify lattice planes with spacings of 2.68 and 1.90 Å 共±1 % 兲
separated by an angle of 90± 2°. These match well with
共002兲 and 共220兲 of ␥-FeSi2, which have theoretical unstrained bulk spacings of 2.69 and 1.91 Å, respectively. The
measured spacings are also consistent with s-phase FeSi2,
but not with the tetragonal or orthorhombic phases. The ␥
and s phases are both cubic 共fcc vs bcc, respectively兲 and
have lattice parameters that differ by only 0.2%. This difference cannot be resolved in the experiment, and the silicide
FIG. 3. HRTEM cross-section image of a NW, viewed along Si关110兴. The
inset is the electron-diffraction pattern obtained from Fourier transformation
of this image. Only cubic FeSi2 is indexed.
Downloaded 24 Jun 2010 to 129.219.247.33. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
113111-3
Appl. Phys. Lett. 88, 113111 共2006兲
liang et al.
FIG. 4. Magnetic response of the nanowires, measured at 2 K. Applied
magnetic field H is parallel to the NWs’ long axis.
lattice might easily be strained by this amount. Theoretically,
the two phases can be distinguished, however, based on their
magnetic response. Calculations suggest that the ␥ phase is
magnetic with 0.3 Bohr magneton per iron atom.6 There are
no direct measurements of the magnetic response so far,
but Hall measurements in thin films are consistent with
this calculation.22 The s phase 共CsCl structure兲 is confirmed
as nonferromagnetic by x-ray magnetic circular dichroism
and magneto-optic Kerr effect measurement.23 But strain
and nonstoichiometry can affect magnetism, so we only
concluded this NW as metastable cubic FeSi2 with an
epitaxial relationship of FeSi2共1̄11兲 / / Si共11̄1兲 and
FeSi2关110兴 / / Si关110兴. The NW is unstrained, within the uncertainty of 1%. This NW, like all other NWs observed, is
mostly buried in the substrate, a signature of endotaxial
growth.18 In this image, one side of the NW has a welldefined habit plane, while the other side does not.
The ␥ phase was discovered and named by Onda et al.22
It is metastable, but can be formed in thin films on Si共111兲
using molecular-beam epitaxy. The cubic silicide/silicon interface for the NWs is the same as for thin films on Si共111兲,
where the lattice mismatch is −0.8% and isotropic. We infer
that a good epitaxial match stabilizes the cubic phase in the
NWs. For films of 1.5 nm thickness, the cubic phase converts irreversibly to the semiconducting ␤ phase 共5% lattice
mismatch兲 upon annealing to 500 ° C.22 The same phase
transition occurs in the NWs, but at a higher temperature,
which we will report elsewhere.
The magnetic response for a sample with NWs grown at
700 ° C is shown in Fig. 4. The magnetic field was applied
parallel to the NW long axis. The diamagnetic response of a
clean Si sample has been subtracted. This M-H curve shows
a clear hysteresis at 2 K with a saturation magnetization of
1.2⫻ 10−6 emu and a coercive field of 30 Oe. Based on the
average coverage of 1 ML and the sample area of 0.5 cm2,
this corresponds to a magnetic moment of 0.3± 0.1 Bohr
magneton per iron atom. This measurement agrees closely
with the theoretically calculated value of 0.3␮B for unstraind
␥-FeSi2.6 This should be regarded as a fortuitous agreement,
since the crystal structure 共e.g., ␥ vs s phase, in this case兲
and magnetic response in a thin film and/or nano-
structured materials can be strongly affected by strain and
nonstoichiometry.
In summary, we grew FeSi2 NWs with cubic structure on
silicon 共110兲 by RDE at 700 ° C. The self-assembled nanowires have average dimensions of 5 nm high ⫻10 nm wide
⫻ ␮m long. Since lattice mismatch between cubic FeSi2 and
silicon is isotropic, endotaxial growth mechanism applied,
which is similar to CoSi2, NiSi2 NWs. Temperaturedependent growth of NWs shows it follows conventional
nucleation theory. SQUID measurements show a magnetic
moment of 0.3 Bohr magneton per iron atom at 2 K. This
result opens up the possibility for self-assembled singlecrystal magnetic NWs on silicon. Ordered arrays of such
magnetic NWs could facilitate fundamental studies of nanoscale magnetism as well as possibly enabling high-density
information storage.
This work was supported by NSF NIRT Grant No.
ECS-0304682. We acknowledge use of facilities in the John.
M. Cowley Center for High Resolution Electron Microscopy,
and the MPMS XL5-AC SQUID magnetometer at Quantum
Design, San Diego, CA.
Q. Wan, T. H. Wang, and C. L. Lin, Appl. Phys. Lett. 82, 3224 共2003兲.
M. C. Bost and J. E. Mahan, J. Appl. Phys. 58, 2696 共1985兲.
3
A. G. Birdwell, R. Glosser, D. N. Leong, and K. P. Homewood, J. Appl.
Phys. 89, 965 共2001兲.
4
D. Leong, M. Harry, K. J. Reeson, and K. P. Homewood, Nature 共London兲
387, 686 共1997兲.
5
I. Berbezier, J. Chevrier, and J. Derrien, Surf. Sci. 315, 27 共1994兲.
6
N. E. Christensen, Phys. Rev. B 42, 7148 共1990兲.
7
M. Han, M. Tanaka, M. Takeguchi, and K. Furuya, Thin Solid Films 461,
136 共2004兲.
8
T. B. Massalski, H. Okamoto, P. R. Subramanian, and L. Kacprzak, Binary
Alloy Phase Diagrams, 2nd ed. 共ASM International, Materials Park, Ohio,
1990兲.
9
J. A. Kittl and Q. Z. Hong, Thin Solid Films 320, 110 共1998兲.
10
K. J. Reeson, J. Sharpe, M. Harry, D. Leong, C. McKinty, A. Kewell, M.
Lourenco, Y. L. Chen, G. Shao, and K. P. Homewood, Microelectron. Eng.
50, 223 共2000兲.
11
C. Preinesberger, S. Vandre, T. Kalka, and M. Daehne-Prietsch, J. Phys. D
31, L43 共1998兲.
12
B. Z. Liu and J. Nogami, Surf. Sci. 540, 136 共2003兲.
13
Y. Chen, D. A. A. Ohlberg, and R. Stanley Williams, Mater. Sci. Eng., B
87, 222 共2001兲.
14
K. L. Kavanagh, M. C. Reuter, and R. M. Tromp, J. Cryst. Growth 173,
393 共1997兲.
15
Z. He, M. Stevens, D. J. Smith, and P. A. Bennett, Surf. Sci. 524, 148
共2003兲.
16
P. A. Bennett, B. Ashcroft, Z. He, and R. M. Tromp, J. Vac. Sci. Technol.
B 20, 2500 共2002兲
17
M. Kuzmin, P. Laukkanen, R. E. Perala, R.-L. Vaara, and I. J. Vayrynen,
Appl. Surf. Sci. 222, 394 共2004兲.
18
Z. He, D. J. Smith, and P. A. Bennett, Phys. Rev. Lett. 93, 256102 共2004兲.
19
J. I. Martín, J. Nogués, K. Liu, J. L. Vicent, and I. K. Schuller, J. Magn.
Magn. Mater. 256, 449 共2003兲.
20
J. A. Venables, Surf. Sci. 299–300, 798 共1994兲.
21
T. H. McDaniels, J. A. Venables, and P. A. Bennett, Phys. Rev. Lett. 87,
176105 共2001兲.
22
N. Onda, J. Henz, E. Muller, K. A. Mader, and H. von Kanel, Appl. Surf.
Sci. 56–58, 421 共1992兲.
23
D. Berling, P. Bertoncini, M. C. Hanf, A. Mehdaoui, C. Pirri, P. Wetzel,
G. Gewinner, and B. Loegel, J. Magn. Magn. Mater. 212, 323 共2000兲.
1
2
Downloaded 24 Jun 2010 to 129.219.247.33. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp