Growth of silicon nanowires on UV structurable glass using self organized nucleation centres

Physica E 38 (2007) 40–43
Growth of silicon nanowires on UV-structurable glass
using self-organized nucleation centres
K. Tonisch
a,Ã
, F. Weise
a
, M. Stubenrauch
a
, V. Cimalla
a
, G. Ecke
a
, F. Will
a
, H. Romanus
a
,
S. Mrotzek
a
, H. Hofmeister
b
, M. Hoffmann
a
,D.Hu
¨
lsenberg
a
, O. Ambacher
a
a

Institute of Micro and Nanotechnologies, Technical University of Ilmenau, P.O. Box 100565, 98684 Ilmenau, Germany
b
Max-Planck-Institut fu
¨
r Mikrostrukturphysik, Weinberg 2, 06120 Halle, Germany
Available online 11 January 2007
Abstract
We report on the growth of silicon nanowires on photostructurable glass by low-pressure chemical vapour deposition. Thereby, no
additional catalyst was needed to stimulate the growth process. Instead, a self-organized crystallization process leads to the formation of
metallic clusters and seed crystals within the glass, which are supposed to initialize the nanowire growth. The nanowires were contacted
by direct deposition of Pt using a focussed ion beam system and characterized electrically.
r 2007 Elsevier B.V. All rights reserved.
PACS: 85.35.ÀP; 81.07.Àb; 61.43.Fs
Keywords: Nanowires; UV-structurable glass; FIB
1. Introduction
Due to its predominant role in semiconductor technol-
ogy, the bulk properties of silicon are well understood.
Therefore, low-dimensional silicon nanowires offer an ideal
basis to study nano-size effects and their possible applica-
tions. Low-pressure chemical vapour deposition (LPCVD)
is an established method to grow silicon nanowires.
Usually different catalyst materials, most common is gold,
are used to stimulate the growth, which have to be
deposited and patterned prior to the nanowire grow th. In
this paper, a low-temperature growth of silicon nanowires
was accomplished on UV-structurable glass without pre-
deposition of any catalyst. These kinds of glasses already
contain different ions like Ce, Ag, Sb, Sn (Li, Na, and K).
By a heat pre-treatment of the UV-structurable glass, the
Ag ions segregate into pure metallic clusters, which then

serve as nucleation centres. A succeeding heating step leads
to the formation of crystallized phases within the glass
matrix, whereby the shape and density of the crystals is
controlled by a self-organized process depending on the
process parameters. As we describe in this communication,
the growth of nanowires can be stimulated on the surface
of such partially crystallized glasses without adding a metal
catalyst.
Various methods have been reported for contacting
nanowires for electrical measurements. In the most
common approach, the nanowire is removed from its
growth substrate and placed on an additional sample by
suspension in solvents and spin-on deposition [1]. Metal
contacts can be defined by optical lithography [2], electron
beam lithography [1] or by focussed ion beam (FIB) [3].We
present the growth of nanowires without additional
catalyst and their chemical and electrical characterization.
For the latter, the nanowires were contacted by direct
writing platinum leads using a FIB equipment.
2. Experimental
The p hotostructurable glass fro m the LiO
2
–Al
2
O
3
–SiO
2
system wa s pretreated by a n exposure t o UV light and a
crystallization ste p at 570 1C. The surface of th e glass was

cleaned using only organic solvents. No sp ecific m odifications
ARTICLE IN PRESS
www.elsevier.com/locate/physe
1386-9477/$ - see front matter r 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.physe.2007.01.001
Ã
Corresponding author.
E-mail address: (K. Tonisch).
were necessary for the succeeding growth o f nanowires.
Silicon nanowires were grown on by low-pressure CVD using
silane as precursor. A small fraction of phosphine was added
to supp ly the phosph or necessary for n-type doping. No
additional catalyst was deposited prior to the growth process,
but the glass c ontained s o-called micro-d opants su ch as A gO
2
which influences the crystallization behaviour of the glass, as
will be shown i n the following section. An additional
substrate of silicon covered with 800 nm thermal oxide has
been used for electrical characterization. Large Ti/Au contact
pads were patterned via optical lithography and lift-off
technique. The nanowires were removed from the growth
substrate and deposited on the thus prepared secondary
substrate. A focussed ion beam system (FIB 200 ODP) with a
liquid Ga source was used to connect the nanowires with the
contact pads by writing Pt-connections. We used 30 k V Ga
+
at 350 p A t o decomp ose the metal-organic P t-precursor
trimethylcyclopentadienyl-platinum [(CH
3
)

3
CH
3
C
5
H
4
Pt] into
predefined bridging leads.
3. Results
The substrate was found to be covered homogeneously
by a dense net of nanowires; ‘‘sea urchin’’-like bundles
grow only where large inhomogeneities occurred at the
surface (Fig. 1). The diameter of the nanowires varied from
50 to 300 nm, their length reached 10–20 mm. The photo-
structurable glass was exposed to UV light with an energy
flux of 60 J/cm
2
and tempered at 570 1C for several hours
prior to the nanowire growth, which leads to the formation
of Ag-clusters which init ialize the growth of lithium meta
silicate (LMS) crystals within the glass [4]. As a result of
this pretreatment, the glass partially crystallizes into
dendrite like LMS crystals within the glass matrix. The
Ag clusters and LMS crystals were analysed using
transmission electron microscopy (Fig. 2). Whereas the
Ag cluster itself is too small to initiate the vapour–liquid–
solid growth (VLS-growth) of such large nanowires, the
LMS crystals exhibit a size which seems to correspond to
the diameter of the nanowires. Since the eutectic tempera-

ture of the Ag–Si system of about 1100 1C is well above the
nanowire growth temperature, the VLS growth mode can
be excluded as growth model in this case. The dendrite
LMS crystals lead to a mechanical distortion of the
surrounding glass matrix [5], which might in turn change
the surface free energy such that the growth of silicon
nanowires is favoured. However, the init ialization of the
nanowire growth requires additional investigations. The
chemical composition of the nanowires was checked by
energy dispersive X-ray analysis (EDX) in order to exclude
the possibility of having grown silica or silicon nitride
instead of pure silicon nanowires. However, the nanowires
and the photostructurable glass were expected to contain
mainly silicon and oxygen. Therefore, a direct measure-
ment would not give a reliable composition for the
nanowires. Additional nanowires were grown by the same
LPCVD process on a substrate consisting of GaN on
sapphire with the aid of an evap orated gold layer of 2 nm
which served as catalyst. Though the growth of these
nanowires follows the VLS growth regime, which is not the
case on the photostructurable glass, the chemical composi-
tion is expected to be the same since the gas composition,
which supplies the material for the nanowire growth, did
not change. As expected, the EDX analysis (Fig. 3)
revealed high amounts of silicon and oxygen. Additional
peaks for Ga, N and Au occurred due to the substrate and
growth process. A carbon peak could be associated with a
contamination of the surface due to the handling under
ambient conditions. The quantitative evaluation of the
EDX analysis gives a content of approximately 90.4 and

4.9 at% for silicon and oxygen, respectively. Though the
growth process leads to a pure silicon nanowire, a thin
layer of natural silicon oxide is formed soon after the
removal of the substrate from the growth chamber. No
ARTICLE IN PRESS
Fig. 1. Scanning electron microscopy (SEM) images of homogeneously grown nanowires (right) and sea urchin-like bundles (left).
K. Tonisch et al. / Physica E 38 (2007) 40–43 41
trace of the phosphor doping could be found due to the
measurement range of EDX analysis.
Fig. 4 shows a representative SEM image of a FIB-Pt
connected nanowire in a two-probe geometry. The nominal
Pt thickness was 500 nm, and the width of the leads was
4 mm. The FIB-deposition conditions must be caref ully
chosen in order to avoid damage or even destruction of
delicate structures by bombardment with massive Ga
+
ions. Therefore, the direct scanning of the nanowires with
the ion beam prior to the contact writing was reduced as
much as possible. We found no evidence for such damage.
On the contrary, a light sputtering at the contact areas on
the nanowires might even improve the contact resistance by
removing the native oxide. The conductivity of the
nanowires was calculated from their electrical resistance
which was determined by current–voltage measurements.
The linear I–V curves of several nanowires with a diameter
of 200–300 nm is displayed in Fig. 5. The different slope is
mainly caused by the different wire geometry resulting in a
different value of the electrical resistance.
Though the thickness of the SiO
2

-shell is not known, it
was neglected, because the diameter of the nanowires was
ARTICLE IN PRESS
Fig. 2. High resolution transmission electron microscopy (HRTEM) image of LMS crystals (marked with solid arrows) and Ag cluster (marked with open
arrows) in the glass matrix (left); the outline of one of the dendrite-shaped LMS crystals is roughly marked by a line. Detailed HRTEM image of a single
Ag-cluster (right; inset: fast Fourier transformation of the image).
Fig. 3. EDX analysis of silicon nanowires grown on a GaN/sapphire
substrate with Au as catalyst.
Fig. 4. SEM image of a two-probe Si-nanowire-based device with contacts
patterned by FIB-Pt deposition.
K. Tonisch et al. / Physica E 38 (2007) 40–4342
large compared to the thickness of a typical native oxide
layer. The specific resistance was calculated to be about
3.5 Â 10
À3
to 9.4 Â 10
À3
O cm using the wire geometries
which were estimated from SEM pictures. This is in the
range of typical LPCVD-grown polycrystalline silicon
which was grown with the same SiH
4
/PH
3
- ratio and had
a specific resi stance of 9.6 Â 10
À4
to 8 Â 10
À3
O cm.

4. Discussion and conclusions
Silicon nanowires were grown successfully on photo-
structurable glass without the need of an additional
catalyst and were analysed with regard to their chemical
composition and their conductivity. Though the LMS
crystals are a prominent candidate for initializing the
nanowire growth, an extended study of the first stages of
the growth process is necessary in order to gain a valid
model. Further research will also focus on a more detailed
characterization of the nanowires concerning their elec-
trical and structural properties. The contamination of the
substrate surface with carbon and gallium is expected to be
high due to low background pressure of the FIB system
and the necessity to use the Ga
+
-beam for visualizing,
which might lead to an incorporation of gallium. This
might result in an uncertainty of the conductivity
measurements due to possible leakage currents across the
substrate surface. Though the agreement of the specific
resistance of the nanowires with that of polycrystalline
silicon supports the correctness of the electrical measure-
ments, additional analysis with Auger electron spectro-
scopy or other surface-sensitive measurements are
necessary. Another way to eliminate the uncertainty of
the conductivity measurements is the undercutting of the
nanowires. Additionally, a freestanding structure will
enable a mechanical characterization concerning the
resonant frequency of such nanowire-based NEMS and
Young’s modulus of the nanowire itself.

References
[1] Y. Cui, Z. Zhong, D. Wang, W.U. Wang, C.M. Lieber, Nano Lett. 3
(2003) 149.
[2] B. Lei, C. Li, D. Zhang, Q.F. Zhou, K.K. Shung, C. Zhou, Appl. Phys.
Lett. 84 (2004) 4553.
[3] C.Y. Nam, J.Y. Kim, J.E. Fischer, Appl. Phys. Lett. 86 (2005) 193112.
[4] S. Mrotzek, A. Harnisch, D. Hu
¨
lsenberg, U. Brokmann, Glass
Technol. 45 (2004) 97.
[5] S. Mrotzek, Ph.D. Thesis, TU Ilmenau, Ilmenau, Germany, 2005.
Further Reading
[6] D.A. Dikin, X. Chen, W. Ding, G. Wagner, R.S. Ruoff, J. Appl. Phys.
93 (2003) 226.
[7] E.C.M. Silva, L. Tong, S. Yip, K.J. van Vliet, Small 2 (2006) 239.
ARTICLE IN PRESS
Fig. 5. Two-probe I–V data from FIB-Pt contacted several Si nanowires
with a diameter of 200–300 nm and a length of 1.5–3 mm.
K. Tonisch et al. / Physica E 38 (2007) 40–43 43