Nonthermal Plasma Synthesis and Oxidation Mechanisms of
Photoluminescent Silicon Nanocrystals from Silicon Tetrachloride
Ryan Gresback, Shunsuke Kanegae, Tomohiro Nozaki and Ken Okazaki
Departments of Mechanical and Control Engineering
Tokyo Institute of Technology
Abstract: A nonthermal very-high-frequency plasma is used to synthesize
luminescent silicon nanocrystals from silicon tetrachloride. The average
nanocrystal diameter can be varied between 3 to 12 nm with narrow size
distributions and the size can be adjusted independently of the relative hydrogen
and argon concentrations in the plasma. The photoluminescence of oxidized
nanocrystals can also be varied from 650 nm to 900 nm. The surface of the
nanocrystals is terminated with hydrogen and chlorine. The nanocrystals oxidize
at significantly faster rates than purely hydrogen terminated nanocrystals, but
slower than fluorine terminated nanocrystals indicating the electronegativity of
surface terminating atoms strongly influence the oxidation process.
Keywords: Silicon, Nanocrystals, Nonthermal Plasma, Photoluminescence
1. Introduction
2. Results and Discussion
Recently silicon nanocrystals (NCs) have
gained attention as a material for third-generation
photovoltaics [1], optoelectronics [2], and
biomedical applications [3]. Due to confinement of
electronic charge carriers within NCs, the optical
and electronic properties are dependent on NC size,
surface termination, and shape [4, 5]. Silicon is a
desirable material because it is non-toxic, abundant,
and compatible with existing microelectronic and
photovoltaic fabrication techniques. In contrast to
many of the well developed NC systems, such as IIVI materials, that include heavy metals, which are
toxic and less abundant.
Free-standing, quantum-confined Si NCs are
synthesized with a flow-through, nonthermal veryhigh-frequency (70 MHz) plasma. Silicon
tetrachloride vapor (SiCl4), hydrogen, and argon are
introduced into the plasma which operates at
pressures of 1-12 Torr. Hydrogen is used as a
scavenger of the chlorine and to partially terminate
the NCs, while argon sustains the plasma. The
process yields NCs with mean diameters between 312 nm with size distributions of less than 20%.
These
NCs
show
nearly
Gaussian
photoluminescence (PL) with peak positions from
650-900 nm and full-width half-maximums
(FWHM) of 100-150 nm. Mangolini et al. have
described a similar method to synthesis Si NCs with
silane (SiH4) [7]. While here, SiCl4 is chosen as it
provides a cheaper and safer alternative to silane and
yields NCs with different possible surface
termination which may be advantageous to postsynthesis processing.
Many methods to synthesize Si NCs have
been developed, including liquid phase synthesis [6],
growth in Si rich oxides and nitrides [2], and
nonthermal plasmas [7]. Nonthermal plasmas have
been shown to synthesis high quality, free-standing
NCs without the need for a secondary material, such
as organic ligands as in the case of liquid phase
synthesis or an oxide or nitride matrix which limit
possible applications. In this work we focus on the
controlled synthesis of Si NCs using a nonthermal
plasma and investigate the resulting optical and
oxidation properties of Si NCs.
TEM of the Si NCs indicate highly
crystalline and spherical material. Additionally the
particles grain boundaries appear distinct. Figure 1
is an example TEM for NCs with a residence time of
60 ms, with a pressure of 3 Torr and flow rates of 2,
40, and 40 SCCM of SiCl4, Ar, and H2, respectively.
Figure 2. Required power for production of crystalline material
for different hydrogen concentrations and pressure. Total flow
rate of 82 SCCM, reactor pressure 3 Torr and 1:40:40
(SiCl4:Ar:H2), unless otherwise noted.
Figure 2. TEM of silicon nanocrystals.
Nanocrystal surface termination influences
the optical and electrical properties and can also
have a key role in the fabrication of devices.
Nanocrystals synthesized with this system allows for
three different surface terminations: dangling bonds,
hydrogen, and chlorine. The plasma chemistry plays
an important role in the surface termination and
crystallinity of synthesized material. Experimental
parameters to control the crystallinity, size, and
surface termination of Si NCs determined. Nominal
input power is adjusted to ensure crystallinity over
different gas ratios, pressures, and reactor volumes.
The results are summarized in Figure 2 for a reactor
with an inner diameter of 22 mm. The error bars
indicate the range where predominately crystalline
material is measured, while to the left of the error
bars (lower power) significant amorphous material is
found and to the right of the error bars mass yield
drops significantly. As the hydrogen concentration is
increased in the plasma, the required power to
produce crystalline increases. This is likely due to
the amount of power required to dissociate hydrogen
as compared to argon. Similarly as the pressure
increases the required power similarly increases.
However the required power increases beyond the
reduced electric field due to increasing thermal
losses.
The size of NCs is controlled by the
residence time of the gas in the plasma. The surface
termination can be partially adjusted through the
fraction of hydrogen introduced into the plasma,
with atomic fractions from 0.1 to 0.7 yielding Si
NCs with various degrees of hydrogen terminated
surfaces, however above or below these conditions
yield very little material. This is likely due to an
etching mechanism which prevents stable nucleates
from forming and growing. In the case of too little
hydrogen, the chlorine is not scavenged and
therefore allowed to recombine with silicon clusters
and effectively break apart the clusters. While in the
case of excess hydrogen, the abundance of hydrogen
allows for a strong pathway to silicon hydride
radicals and not the formation of clusters. From
FTIR, chlorine and hydrogen appear to varying
degrees for all NCs synthesized under the conditions
investigated.
The NC size and hence PL can be adjusted
from the residence time of the NCs in the plasma.
Shown in Figure 3 the peak PL wavelength can be
adjusted from 900 to 650 nm for oxidized NCs. The
Gaussian profile of the PL is also narrow indicating
monodisperse NCs.
The PL history of a sample produced with
conditions: total flow rate of 328 SCCM; gas
composition of 2:40:40 (SiCl4:Ar:H2), and a pressure
of 3 Torr can be seen in Figure 4. The
photoluminescence was monitored after oxidation
for on the time scale of seconds to weeks. The
triangles are the measured peak of the PL and the
diamonds are the oxide thickness as determined by
the change and PL peak with the core size
determined by Zunger and Wangs method [8, 9].
The line is from the Elovich equation with the
parameters in Table 1.
Figure 3. Photoluminescence of oxidized silicon nanocrystals.
3. Oxidation of Silicon Nanocrystals
The size of nanocrystals correlates well with
the peak position of the photoluminescence if the
photoluminescence is due to confinement of
electronic carriers (quantum confinement) and not
related to defects or size independent emission sites
[8, 9]. As Si NCs oxidize surface atoms are
consumed by oxidation which effectively reduces
the size of the NC core. By monitoring the peak
photoluminescence position the nanocrystal size can
be determined and hence as oxidation occurs the
shift in photoluminescence can be used to determine
the core size and oxide thickness. The oxidation rate
for Si NCs has been shown by Liptak et al [10] be
driven by the Cabrera-Mott and follows the Elovich
equation. The Elovich equation describes the oxide
thickness, ox, as a function of time, t.
The fitting parameters of monolayer growth rate, ro,
and characteristic growth time, tm, describe the
growth rate.
Table 1 also shows the values reported by
Liptak et al. for hydrogen and fluorine terminated
silicon. While the values reported for hydrogen
terminated silicon are 3-4 orders of magnitude
slower than those found here. However the values
for chlorine terminated Si NCs are 1 order of
magnitude slower than those found for fluorine
termination. These differences likely are related to
the electronegativity of the surface termination.
Since the Cabrera-Mott mechanism relies on
electrons reaching the interface which can ionize
oxidizing species (oxygen and water), the dipole
which is created by highly electronegative surface
atoms (Cl and F) significantly increases the rate of
this process.
Figure 4. Time dependent behavior of silicon nanocrystals after
exposure to air (triangles) photoluminescence and (diamonds
and line) oxide thickness.
Table 1. Oxidation rate of silicon nanocrystals for different
surface coverage.
Surface
coverage
Characteristic
growth time:
tm (sec)
Monolayer
growth rate:
ro (nm/hr)
Si-H [10,11]
7200–19800
0.07
Si-Cl
(our results)
3
384
Si-F [10]
0.1–0.6
915–5000
4. Conclusions
This work reports on the synthesis of silicon
nanocrystals with a nonthermal flow-through veryhigh-frequency plasma using silicon tetrachloride
and the resulting optical and oxidation properties of
the nanocrystals. Crystalline material can be
produced over a wide range of plasma pressures and
gas compositions. Photoluminescence is observed
for with peak emissions varying between 650 and
900 nm for oxidized NCs. The peak emission can be
adjusted through the residence time of the gas in the
plasma region with shorter residence times leading
to more quantum confined (blue-shifted) emission.
Additionally the oxidation rate of Si nanocrystals
synthesized was studied using the change in
photoluminescence peak position. The oxidation rate
was found to be 3-4 orders of magnitude faster than
hydrogen terminated Si nanocrystals but only one
order of magnitude slower than fluorine terminated
nanocrystals. These results indicate are consistent
with a Cabrera-Mott oxidation process.
4. Acknowledgments
This project is supported by the Funding
program for Next Generation World-Leading
Researchers. R.G. acknowledges Japan Society for
Promotion of Science for providing a research
fellowship (JSPS, DC1).
We thank Dr. Y.
Murakami for his assistance with PL measurements
and the Center for Advanced Materials Analysis for
TEM.
5. References
1. M.C. Beard, K.P. Knutsen, P. Yu, J.M. Luther, Q.
Song, W.K. Metzger, R.J. Ellingson, and A.J.
Nozik, Nano Letters 7, 2506-2512 (2007).
2. L. Pavesi, L. Dal Negro, C. Mazzoleni, G. Franzo,
and F. Priolo, Nature 408, 440-444 (2000).
3. Z.F. Li and E. Ruckenstein, Nano Letters 4, 14631467 (2004).
4. A.J. Read, R.J. Needs, K.J. Nash, L.T. Canham,
P.D.J. Calcott, and A. Qteish, Phys. Rev. Lett. 69,
1232 (1992).
5. M.V. Wolkin, J. Jorne, P.M. Fauchet, G. Allan, and
C. Delerue, Phys. Rev. Lett. 82, 197 (1999).
6. J.P. Wilcoxon and G.A. Samara, Appl. Phys. Lett.
74, 3164 (1999).
7. L. Mangolini, E. Thimsen, and U. Kortshagen,
Nano Lett. 5, 655-659 (2005).
8. A. Zunger and L. Wang, Applied Surface Science
102, 350-359 (1996).
9. L. Wang and A. Zunger, Phys. Rev. Lett. 73, 1039
(1994).
10. R.W. Liptak, U. Kortshagen, and S.A. Campbell,
J. Appl. Phys. 106, 064313 (2009).
11. M. Morita, T. Ohmi, E. Hasegawa, M.
Kawakami, and M. Ohwada, J. Appl. Phys. 68,
1272 (1990).