Tetrakis(trimethylsilyloxy)silane for nanostructured SiO2-like ﬁlms
deposited by PECVD at atmospheric pressure
J. Schäfer a,
⁎, J. Hnilica b
, J. Šperka b,c
, A. Quade a
, V. Kudrle b
, R. Foest a
, J. Vodák d
, L. Zajίčková b,c
a
Leibniz Institute for Plasma Science and Technology e.V., Felix-Hausdorff-Straße 2, 17489 Greifswald, Germany
b
Department of Physical Electronics, Masaryk University, Kotlářská 2, CZ-61137 Brno, Czech Republic
c
CEITEC — Central European Institute of Technology, Masaryk University, Kamenice 753/5, CZ-62500 Brno, Czech Republic
d
Institute of Physical Engineering, Faculty of Mechanical Engineering, Brno University of Technology, Technická 2896/2, CZ-61669 Brno, Czech Republic
a b s t r a c ta r t i c l e i n f o
Article history:
Received 16 May 2015
Revised 21 August 2015
Accepted in revised form 26 September 2015
Available online 3 October 2015
Keywords:
Tetrakis(trimethylsilyloxy)silane
Tetrakis(trimethylsiloxy)silane
Plasma jet
Silicon dioxide
Plasma enhanced chemical vapor deposition (PECVD) from tetrakis(trimethylsilyloxy)silane (TTMS) has been
studied at atmospheric pressure. TTMS has been chosen because of its unique 3D structure with potential to
form nano-structured organosilicon polymers. Despite the widespread surveying of various silicon-organic molecules
for PECVD, the use of TTMS in AP-PECVD has not been investigated deeper yet. PECVDs have been performed
with two different plasma jets. While they are alike regarding the geometry and injection of TTMS,
they differ in input power and excitation frequency. The radiofrequency plasma jet operates at lower power densities
as compared to the microwave plasma jet. Despite this all the deposited ﬁlms exhibit similar chemical properties
resembling that of silicon dioxide (Si:O = 1:2) with carbon content below 5%. The ﬁlms demonstrate a
broad variety of morphologies from compact smooth ﬁlms to nano-dendritic 3D structures depending on the particular
process.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
If complex organosilicon precursors are used for PECVD, the chemistry
and properties of the resulting ﬁlms are heavily inﬂuenced by the
operational conditions. Depending on temperature, power density or
gas mixture (e.g. oxygen content) inorganic SiO2 [1,2] ﬁlms are observed
as well as branched and/or cross-linked structures of
polymethylsiloxanes [3]. Often, the incorporation of a high content of
organic functional groups, in particular methyl groups is causing a porous
or less dense ﬁlm structure with inferior chemical and mechanical
stability. Hence, the independent adjustment of the morphological and
chemical ﬁlm properties by variation of the process parameters at atmospheric
pressure is a challenging task. Much of current effort is dedicated
to this issue [4,5]. In particular, a stability against chemical or
mechanical impact is desirable for coatings in a wide range of morphologies.
For instance, permeation barriers rely on compact pin-hole free
coatings [6] whereas surfaces for heterogeneous catalysis require hierarchically
nano-structured ﬁlms with large surface area and their catalysts
centered in chemically stable matrices [7,8]. The importance of
hybrid materials production from molecular precursors leads to investigation
of organosilicates from complex precursors. For example, precursors
of the general formula (CH3O)3SiRSi(OCH3)3 are under
investigation for the preparation of hybrid materials by sol-gel polycondensation
[9].
In the present study it is demonstrated that such different morphological
structures can be obtained using tetrakis(trimethylsilyloxy)silane
(TTMS, C12H36O4Si5) as thin ﬁlm precursor for PECVD, while the chemical
composition remains essentially SiO2–like. Among other complex silicon
organic molecules commonly in use for PECVD, the TTMS molecule (see
Fig. 1) exhibits several advantages, with regard to forming structured
coatings at suitable conditions. The molecular structure of TTMS is characterized
by the presence of ﬁve rigid tetrahedral sub-units: the central
unit SiO4 and four peripheral units OSi(CH3)3. The central unit represents
the elementary cell of quartz-like structures in dense SiO2 ﬁlms. The average
valence angle of SiOSi in TTMS is 146° [10] and hence wider than in
usually used hexamethyldisiloxane (HMDSO, (CH3)3Si–O–Si(CH3)3)
where 130° is found [11]. Thus, the rotation around the Sic
–O bonds1
in
TTMS is virtually more unrestricted than in HMDSO. This causes a relatively
ﬂexible conformation of TTMS [12]. Consequently, this ﬂexible conformation
can play a crucial role for the self-adapted growth of nanostructured
ﬁlms. In addition, this precursor is non-ﬂammable, nontoxic,
and has reasonably high vapor pressure. After studies [13] and
[14], the present work is the ﬁrst utilizing TTMS in atmospheric pressure
high-frequency PECVD processes.
Surface & Coatings Technology 295 (2016) 112–118
⁎ Corresponding author.
E-mail address: jschaefer@inp-greifswald.de (J. Schäfer). 1
The index c
denotes the central atom of the molecules.
http://dx.doi.org/10.1016/j.surfcoat.2015.09.047
0257-8972/© 2015 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
Plasma enhanced chemical vapor deposition of silicon-organic ﬁlms
at atmospheric pressure (AP-PECVD) is commonly performed with a
wide variety of precursors [15–18]. The most prominent class of precursors
in use is alcoxysilanes (e.g. hexamethyldisyloxane, HMDSO,
tetramethyldisiloxane, TMDSO, tetraethoxysilane, TEOS, triethoxysilane,
TriEOS) or hexamethyldisilazane, HMDSN. Their widespread usage is
lessening their purchase price which further alleviates their application
for technological processes. Furthermore they are in use because of
their uncomplicated handling, non-toxicity and simple vaporization at
room temperature. Often the studies of the ﬁlm deposition processes
are directed to obtain chemically and morphologically homogeneous
coatings [19]. From this point of view, the investigation of structural variations
in AP-PECVD using complex organosilicon precursors represents a
rare objective with an interesting technological potential [4,20].
2. Experimental set-up
In this paper, we present a comparative study of AP-PECVD using
two different plasma jets: a MW plasma jet and a radiofrequency (RF)
plasma jet described in previous experiments [19,21]. The MW jet
does not allow the operation at low power ranges (below 100 W),
whereas the main operating range of the RF jet is typically below
20 W. Here, we disregard the differences in the plasma kinetics of
both devices caused by their different excitation schemes. Despite
being operated at different excitation frequencies their experimental arrangement
is quite similar, see Fig. 2. We keep the geometry of the jets
and precursor injection similar in both experiments and consider the
differing power densities as the most signiﬁcant disparity in the ﬁrst approximation.
Using both jets then effectively covers a wider power
range. The usage of an extensive surface and material analytics allows
discussing the relationship of homogeneity, type of nanostructuring
and chemical composition of the deposited ﬁlms.
2.1. Microwave jet
The atmospheric pressure microwave electrode-less jet uses a
surfatron [22] as excitation source (SAIREM Surfatron 80 type with integrated
matching). This source is powered by a magnetron generator
(SAIREM GMP 20 KED, 2.45 GHz, 0–2 kW) via ferrite circulator, waveguide
and coaxial cable. The cable and coaxial connectors limit the supplied
power to 300 W. During this study the microwave power was
235 W.
The surfatron excitation source consists of a tuned coaxial resonant
cavity through which a fused silica discharge tube (80 mm long, 6 mm
and 8 mm inner and outer diameter, respectively) is inserted. There is
another silica coaxial capillary with 1 mm diameter in the center of
this discharge tube. The supplied gas – argon – passes through the annulus
between the capillary and the discharge tube. Its ﬂow is maintained
at 2.6 slm by a mass ﬂowmeter.
Liquid TTMS is placed in a bubbler cylinder ﬁlled with glass beads.
After passing the bubbler cylinder the auxiliary argon carrier gas ﬂow
(90 and 435 sccm) is enriched with TTMS vapors at laboratory temperature.
Using the coaxial capillary this mixture is injected downstream
the surfatron excitation gap into the active microwave discharge. The
diffusion losses in the thin capillary prevent ignition of a discharge inside
the capillary. TTMS vapor is dissociated in the discharge and
forms a deposit on a sample placed 6 mm from discharge tube end.
The amount of TTMS vapors introduced (0.02–0.07 g/h i.e. 0.02–
0.07 sccm) is controlled by the auxiliary argon ﬂow and veriﬁed using
gravimetry. The adjusted rates of argon and TTMS provide the precursor
concentration of 7 ppm to 23 ppm in order to study deposition conditions
under variation of the plasma-chemically relevant parameter energy per
molecule. Note that the only way to vary the speciﬁc energy in this particular
microwave experiment is to adjust the gas management (ﬂows).2
Nevertheless, for both set-ups (MW and RF), the resulting relative concentrations
of TTMS in the total argon ﬂux have the same order of
10−5
. The duration of a deposition experiment is 5 min.
2.2. Radiofrequency jet
The atmospheric pressure radiofrequency plasma source consists of
a fused silica tube (i.d. 4 mm, o.d. 6 mm) with two external ring electrodes.
The electrodes are 5 mm wide and their distance is 5 mm. A coaxial
thin-wall capillary with diameter of 1.9 mm is located in the center
of the discharge tube. Similar to the MW jet, the dimension of the inner
capillary does not impede the generation of a discharge inside the tube
and no discharge is ignited inside the inner capillary due to diffusion
losses. The substrate for ﬁlm deposition is located 6 mm below the nozzle
of the discharge tube.
Argon ﬂow rates through the central capillary (inner Ar ﬂow rate)
and the annulus between the capillary and the discharge tube (outer
Ar ﬂow rate) are ﬁxed at 1.0 slm and 0.5 slm, respectively. The coaxial
ﬂows are dimensioned such that the gas velocities in each of the capillaries
ﬁt to a simple parabolic velocity proﬁle in the tube as without coaxial
ﬂux separation. Thus lateral ﬂuxes are avoided. The setting of the
optimal ﬂow rates required to adapt the dosage of TTMS by means of
a blower for a better comparability with the MW set-up. Instead of
using a conventional bubbler arrangement, TTMS vapors are admixed
to the inner Ar ﬂow by ﬂowing Ar above the liquid TTMS in the blower
ﬂask before the mixture enters the capillary at laboratory temperature
of 23 °C. Indeed, the Ar ﬂow of 1 slm through the blower carries the
same ﬂow rate of TTMS vapors, 0.02 sccm, as 0.09 sccm of argon through
the bubbler in the MW jet set-up. This was checked by measuring the
TTMS ﬂow rates beforehand for various Ar ﬂow rates in both the setFig.
1. Molecule of tetrakis(trimethylsilyloxy)silane (TTMS C12H36O4Si5). Fig. 2. Comparison of two plasma jet set-ups.
2
The variation of applied power would be feasible in principle; the energy support depends
crucially on the resonance conditions which would be infringed by changing the applied
power.
113J. Schäfer et al. / Surface & Coatings Technology 295 (2016) 112–118
ups by gravimetry, i.e. weight loss of liquid TTMS after longer operation
at constant conditions. The accuracy of the TTMS ﬂow rates given in
Table 1 is estimated as 10%.
The RF power (27.12 MHz, DTG2710, Dressler) supplied to the plasma
was 6 or 15 W. This, along with the given ﬂow ensures a source operation
in a distinctive discharge mode with four azimuthally rotating
discharge ﬁlaments dubbed Locked Mode 4 (LM4) and described in detail
in [23]. Here, the time needed to deposit a ﬁlm is longer: 15 min for
one experiment.
3. Materials and methods
All experiments have been performed with a similar gas mixture,
Argon (9.5, Linde or Messer Griesheim, resp.) to which a small amount
of TTMS (Sigma-Aldrich, purity 97%) vapor was added. Because the
usage of TTMS is uncommon for PECVD, we summarized the basic properties
in Table 2. Thin ﬁlms have been deposited on double side polished
(DSP) silicon wafers, that are transparent in the middle infra red (IR)
spectral range.
The resulting ﬁlms have been analyzed with several methods in
order to characterize their structuring and chemical composition. A
morphological characterization has been carried out using atomic
force microscopy (AFM, Dimension Icon microscope, Bruker) and high
resolution scanning electron microscopy (HR-SEM). The SEM (JSM
7500F, JEOL) employs a ﬁeld-emission gun. The secondary electron inlens
detector enables to observe structural features of the deposited
ﬁlms at a maximum speciﬁed resolution of 1.0 nm at 15 keV. The device
can operate without any preparative coatings which could lead to morphological
artifacts at nanometer scale. For the analysis of the vertical
ﬁlm structure, a cross-section ion polisher (CIP, JEOL) has been used.
The chemical composition has been characterized by means of X-ray
spectroscopy (EDX, Bruker X-Flash spectrometer, 30 mm2
silicon drift
droplet detector).
The elemental surface composition was analyzed with X-ray photoelectron
spectroscopy (XPS) using an AXIS Ultra DLD electron spectrometer
(Kratos Analytical). The spectra were recorded by means of
monochromatic Al–Kα excitation (1486.6 eV) with a medium magniﬁcation
(ﬁeld of view 2) lens mode and by selecting the slot mode, providing
an analysis area of approximately 25 μm in diameter. Charge
neutralization was implemented by low energy electrons injected into
the magnetic ﬁeld of the lens from a ﬁlament located directly atop the
sample. Data acquisition and processing were carried out using CasaXPS
software, version 2.14dev29 (Casa Software Ltd.). To remove the uppermost
layer of contaminating hydrocarbons the samples were sputtered
for 5 min with 4.5 keV Ar-ions.
The thickness distribution of smooth coatings has been measured by
means of spectral imaging reﬂectometry (ISR) [24]. The ISR spectra
were measured at normal incidence of light in the spectral range between
270 and 900 nm with pixel size (spatial resolution) of 37 × 37 μm2
. They
were ﬁtted using the PJDOS dispersion model for SiO2-like materials [25]
in custom software as described in a previous work [24].
Fourier transform infra-red spectroscopy (FTIR) was conﬁgured for
ATR (Attenuated Total Reﬂection, Spectrum One microscope, Perkin
Elmer) and for transmission measurements (Bruker Vertex 80v).
4. Results
From 20 produced and analyzed samples, four representative samples
were chosen in order to demonstrate the general semi-quantitative
changes of signiﬁcant material properties and reveal the apparent relation
to the speciﬁc energy per molecule of TTMS and on the used plasma
source. We consider them representative of the trends and suitable for
the discussion. The overview of experimental parameters for these
four samples is given in Table 1. The Yasuda parameter i.e. speciﬁc energy
per unit volume of TTMS precursor is calculated for each condition
and plotted in the right column.
4.1. Morphology and structure
For the investigation of the structural features in the ﬁlms a combined
application of scanning electron microscopy and atomic force microscopy
has been chosen. By this complementary approach apparent
disproportional differences in the ﬁlm morphology can be described depending
on the precursor amount and speciﬁc energy for the PECVD.
While SEM is suited preferably for the investigation of deeply and
sharply structured forms, AFM is appropriate for the topographic description
of more or less smooth samples. Here, the ﬁlms deposited
using the RF jet form uniform layers and they exhibit a comparably
low surface roughness which has been characterized by AFM. On the
other hand, the ﬁlms produced by the MW jet demonstrated a ﬁne-textured
topography which has been investigated with SEM.
The texture of two SEM images (MW1 and MW2) is shown in Fig. 3.
In the MW jet, the deposition mechanism leads to the growth and
branching of nanoscopic auxiliary buds. Thus, three dimensional
nanodendrites are formed. The spreading structure takes up a speciﬁc
volume with a large free surface similar to a fractal. The fractalized
parts of the top texture indicate narrow trenches reaching fully down
to the initial points of the branched dendrites. Therefore, two detectors
were utilized during the SEM characterization of the MW samples. The
detector of the secondary emitted electrons (SEI) collects signals of top
areas (focus on the lateral structure), while the detector of back
scattered electrons (BEI) extracts information from deeper parts of the
surface (focus on the vertical structure). Fig. 3 combines both images
(SEI and BEI) by color-coding thus visualizing the degree of spatial
spreading. In the case of lower energy per TTMS molecule (sample
MW1), bigger rugged structures take a speciﬁc lateral dimension exceeding
5 μm. If the energy increases (sample MW2), more dense dendrites
with lateral diameters between 1 to 3 μm can be observed. The
effect can be explained with a faster deposition kinetics at higher specific
energy. Indeed, a shorter reaction time limits the surface diffusion
length. Therefore dimensions of the rugged topography are rescaled
down for higher speciﬁc energy.
The cross-sections of the deposited ﬁlms have been investigated
deeper in order to explore the progression of the growth from the substrate
to the top of the dendrites. The comparison between ﬁlms deposited
by the MW jet (sample MW2) and by the RF jet (sample RF2) is
shown in Fig. 4. The ion beam polishing of the ﬁlm cross-sections reveals
the presence of an amazing vertical structure of the MW samples. The
branched structure grows over several micrometers in vertical direction
fully intact, see Fig. 4 and exhibits a fractal behavior. Moreover, at the
bottom of the ﬁlm an initial layer with a thickness of approximately
Table 1
Deposition conditions for the ﬁlms prepared by RF (27.12 MHz) and MW (2.45 · 103
MHz)
plasma jets.
Sample Power
[W]
Ar outer
[slm]
Ar inner
[slm]
TTMS
[sccm]
W/F
[kJ cm−3
]
RF1 6 0.5 1.0 0.02 20
RF2 15 0.5 1.0 0.02 50
MW1 235 2.60 0.4 0.07 200
MW2 235 2.60 0.09 0.02 900
Table 2
Fundamental properties of TTMS (CAS Nr. 3555-47-3) [www.sigmaaldrich.com,
www.arb.ca.gov].
Quantity Unit Value
Molecular mass AMU 384.85
Boiling temperature at 267 Pa °C 103.0–106.0
Flash point °C 76
Density of liquid g/cm3
0.87
Vapor pressure at 25 °C Pa 8.96
114 J. Schäfer et al. / Surface & Coatings Technology 295 (2016) 112–118
200 nm has been observed. Because of the constant external conditions
during the deposition, we assume that the transition from the initial
layer to the dendrites results from a self-organized competition between
parallel nucleation processes [26]. The continuous growth at
points with a high nucleation dominates over points with a smaller nucleation.
Such disturbance can occur at locally increased surface temperature,
in particular at a higher speciﬁc energy in the MW experiment.
The disturbance consequently stimulates an increasing roughness of
the initial layer and leads to a growth of dendrites if sufﬁcient energy
is supplied. This is not the case for the deposition with the RF jet,
where signiﬁcantly less energy per TTMS molecule is coupled and evidently
no structured morphology can be observed. Finally, the thickness
of the deposited ﬁlms (RF1 and RF2) is comparable to the dimensions of
the initial dense layers of samples from the MW jet, while the total ﬁlm
thickness of MW samples is signiﬁcantly higher (see Table 3).
The lateral nanostructuring of the smooth ﬁlms was studied using
AFM (see Fig. 5). An extremely low surface roughness of 0.3 nm
(RMS) has been obtained. This property of the ﬁlms permitted the application
of ISR for a two-dimensional characterization of the local deposition.
The ISR results, shown in Fig. 5, reproduce the maximum ﬁlm
thickness obtained with SEM of the ﬁlm cross-sections (compare the
image RF2-a in Fig. 5 with the right image in Fig. 4). The refractive
index of the RF1 ﬁlm, determined by ISR, was only slightly lower than
for SiO2, speciﬁcally 1.45 at wavelength λ = 500 nm. However, the refractive
index of the RF2 ﬁlm was considerably lower, about 1.38 at
500 nm. Since the carbon content was almost identical in both ﬁlms,
this indicates that RF2 has a lower density than RF1, which is in agreement
with the results from IR spectroscopy below. Note, that the application
of ISR on the samples deposited by means of the MW plasma jet is
not possible due to extremely high surface roughness. The results from
the morphological analysis of the coated samples are summarized along
with the results of the chemical characterization in Table 3.
4.2. Chemical composition
The pronounced differences in the ﬁlm structure requires a careful
combination of analytic methods in order to characterize the chemical
composition of ﬁlms correctly considering the ﬁlm thickness, surface
roughness and physical limitations of the methods. We applied XPS
analysis for the characterization of the elemental composition present
on the surface of the deposited ﬁlms. EDX analysis was carried out for
the elemental composition in the bulk, particularly in the case of thicker
samples (MW1 and MW2). The EDX analysis of thinner ﬁlms (samples
RF1 and RF2) is inﬂuenced by the substrate because of the higher signal
generation depth (here: 200–500 nm). Therefore the analysis of the
samples RF1 and RF2 was conducted using XPS and in situ sputtering
of the ﬁlms. Here, 50 nm of the ﬁlm were removed by sputtering, before
the chemical composition of the bulk was measured. Note that the information
depth of XPS is limited to only several nm. Finally, the combination
of EDX, XPS, and the sputtering technique provides the elemental
composition on the surfaces and in the bulk separately. The infra-red
spectroscopy of all samples has been applied as a complementary method
to the previous technique in order to identify chemical functional
groups, and to characterize the stoichiometry (SiOx), porosity and
trends concerning impurities (carbon content). Again, two approaches
have been applied: transmission FTIR for smooth thin ﬁlms (RF1 and
RF2) and ATR-FTIR for thicker, absorbing samples (MW1 and MW2).
The compilation of the main ﬁndings is shown in Table 3.
FTIR spectra of all ﬁlms deposited from TTMS are compared in Fig. 6.
The spectra show characteristic peaks of silica at about 800 and
Fig. 3. SEM images taken by combined detection of back-scattered (SEI — green color) and secondary electrons (BEI — red color) from surfaces of MW1 (left image, lower speciﬁc energy)
and MW2 (right image, higher speciﬁc energy) ﬁlms.
Fig. 4. Cross-sectional SEM micrographs of the ﬁlms on silicon substrates: nanostructured MW2 ﬁlm deposited in MW jet (left), compact smooth RF2 ﬁlm deposited in RF jet (right).
115J. Schäfer et al. / Surface & Coatings Technology 295 (2016) 112–118
1070 cm−1
corresponding to the bending and antisymmetric stretching
mode (‘bulk mode’, AS) of the Si–O–Si groups, respectively [27]. A transverse
optical mode of Si–O–Si (‘surface mode’, TO) appears between
1130 and 1200 cm−1
. While the AS mode can be interpreted as diagnostics
for the ﬁlm stoichiometry, the characteristic of the TO peak can indicate
the inner surface or porosity of ﬁlms qualitatively. Indeed, the
samples MW1 and MW2 exhibit a strong absorption of the TO peak
(left ﬂank of the AS mode) as expected from the SEM analysis of these
samples. They are characteristic with their dendrite structure causing
a large effective surface of the samples and thus an intensive excitation
of the surface TO mode in the IR spectrum. The transmittance measurements,
that could be performed down to 370 cm−1
for RF1 and RF2
ﬁlms, conﬁrm the existence of the peak at 450 cm−1
that is associated
with the Si–O–Si rocking. Finally, absorption peaks at 1279 cm−1
and
920 cm−1
are related to the Si–CH3 and silanol groups (Si–OH), respectively
[28,29]. They indicate an (in)sufﬁciency of decomposition and oxidation
of TTMS molecules by the plasma. Essential characteristic peaks
are listed in Table 4.
Both series of experiments (RF and MW) exhibit the same trends regarding
the absorption peak at 1279 cm−1
: with increasing speciﬁc energy
a reduced Si–CH3 signal at 1279 cm−1
can be observed. Please note
that for MW1 the overall concentration of TTMS is two times higher
(≈20 ppm) than for RF1, RF2, and MW2, therefore Si–CH3 can appear
in the spectra of MW1 easily, even if the speciﬁc energy at MW1 is
higher than in the case of RF2. The deposition rate follows the same
trend as the change of the Si–CH3 peak: with the increase of the speciﬁc
energy within the particular series (RFx and MWy) the deposition rate
decreases. The positive correlation between the change of the deposition
rate and the Si–CH3 peak let us conclude that the increasing speciﬁc
energy contributes to better oxidation – indeed, the Si–CH3 peak disappears
– improving the ﬁlm density and retarding the growth. However,
the presence of silanol termination groups suggests that the density of
the deposited ﬁlms (even the compact ones from RF process) is lower
than SiO2 glass. Further conclusion on the properties of the SiO2 ﬁlms
can be drawn from the position of the AS mode in the spectrum. Its
shift to lower wavenumbers can be related to several phenomena like
carbon contamination [30], oxygen deﬁciency [31,32] and material densiﬁcation
[33]. Finally, these effects result in lower stoichiometry of the
SiOx. Moreover, the dependence of x on the position of the AS mode can
be derived for the studied class of ﬁlms precisely [19,34,35]. The obtained
values are between 1.8 and 2.2. The samples RF1 and MW1 are stoichiometrically
similar (1.8), while the ratio [O]/[Si] determined for
samples RF2 and MW2 is higher, 2.1–2.2, than the stoichiometric ratio
of pure SiO2. This can be explained by the higher content of -OH terminating
groups in RF2 and MW2. Indeed, the Si–OH absorption is higher
for RF2 and MW2 ﬁlms compared to RF1 and MW1, which on the other
hand exhibit a higher concentration of carbon (Si–CH3 peaks) and lower
stoichiometry. Additionally, the –OH group is bond on a residual carbon,
too. Water absorption peaks were not detected.
Table 3
Summary of the ﬁlm properties.
RF1 RF2 MW1 MW2
Speciﬁc energy W/F [kJ cm−3
] 20 50 200 900
Structurea
SL SL SL + D SL + D
Deposition rate [nm/s] 0.1b
0.4b,a
17a
10a
Maximum height [μm] 0.13b
0.35b,a
5a
3a
Roughness [nm] 0.3c
0.3c
– SL: 100d
SiOx stoichiometry [O]/[Si] 1.9e
/ 1.8f
2.1e,f
1.6g
/ 1.8f
2.1g
/ 2.2f
Residual carbon in bulk [%] 5e
4e
3g
1g
Total carbon on surface [%]h
22 7 12 6
SL — single dense layer; D — structure of dendrites.
a
SEM analysis.
b
Imaging spectroscopic reﬂectometry.
c
AFM analysis.
d
SEM analysis using cross-section ion polishing.
e
XPS analysis using the sputtering of the surface.
f
FTIR analysis.
g
EDX analysis.
h
XPS analysis.
Fig. 5. Topography of the local deposition from TTMS using the RF jet. Film thickness of samples RF1 and RF2 is imaged using spectroscopic reﬂectometry (images RF1 and RF2-a), the
surface roughness in the middle of the samples RF2 is visualized by AFM, RMS = 0.3 nm (image RF2-b).
Fig. 6. Infrared spectra of the deposited ﬁlms. Transmission FTIR (top), ATR-FTIR (bottom).
The shift of the main absorption peak (1050–1080 cm−1
) to higher wavenumber indicates
a higher stoichiometry of SiOx and a broad absorption between 1200 and 1130 cm−1
reveals
a large top or inner surface (roughness and/or porosity) of MW samples.
116 J. Schäfer et al. / Surface & Coatings Technology 295 (2016) 112–118
The elemental analysis conﬁrms a chemical similarity of all SiOx
samples. The oxygen atomic content of the ﬁlms is between 60 and
67%, while the silicon atom percentage is 32–37%. The rest includes carbon
(1–5%) and hydrogen atoms. However, hydrogen cannot be quantiﬁed
by methods utilized here. The stoichiometric ratio x = [O]/[Si]
obtained from the elemental analysis (1.6–2.1) correlates very well
with the prediction from FTIR analysis (1.8–2.2). Note, that the considerations
on the correlation are based on the comparison of analytic
methods speciﬁed by similar measurement depth (EDX vs. FTIR and
sputtering XPS vs. FTIR). In contrast, the elemental composition of the
ﬁlm surfaces differs notably (see Table 3), e.g. the carbon atom percentage
varies signiﬁcantly between 6–22%. This deviation is mainly caused
by two reasons: (1) The immanent difference between bulk and surface
elemental composition. At the surface, a rearrangement of functional
groups (particularly –OH) occurs, and a higher contamination by carbon
and hydrocarbons (containing additional oxygen, too) is built up. (2)
Evident differences in the ﬁlm morphology as described in the previous
subsection. For example, the carbon concentration of sample MW2 is six
times higher at the surface than in the bulk, while the oxygen concentration
on the surface changes only slightly (66–67%).
5. Discussion
5.1. On speciﬁc energy and temperature conditions
Primarily morphological properties and – despite the chemical similarity
of studied ﬁlms – small differences of the chemical composition
are related to speciﬁcs of the plasma process, which shall be discussed
here. Even with apparent different excitation scheme, we consider the
differing power densities as the most signiﬁcant disparity between the
two set-ups. The geometry of the jets and precursor injection is similar
in both experiments leading to comparable hydrodynamics and chemical
kinetics. The power input in the MW jet was substantially higher
than in the RF jet. Hence, qualitatively different PECVD processes
could be expected. Indeed in a previous work [13], the atomic lines of
Si (from dissociated TTMS) were observed in the optical emission spectra
of the MW jet, while they could not be detected in the RF jet. Both
set-ups, the MW jet and the RF jet differ considerably in numerous technical
parameters: e.g. gas ﬂow, average velocity, excitation frequency
resulting in largely differing plasma physical characteristics (e.g. charge
carrier kinetics, energies and concentrations, transport mechanisms and
subsequently in different plasma-surface interactions). For instance, the
observed growth rates differ essentially (up to 25–100 times higher in
the MW jet). In an attempt to overcome the complexity of all these parameters
we consider the energetic characterization and implement the
concept of the Yasuda parameter for the comparison of the different deposition
experiments and complement this concept by considering the
different surface temperatures for the discussion of our results. The calculated
Yasuda parameter (W/F) is shown in Tables 1 and 3. Note, that in
the RF jet the variation of speciﬁc energy W/F (20–50 kJ cm−3
) is caused
by the variation of the electric power, while in the MW jet (200–
900 kJ cm−3
) it is due to the changing TTMS concentration.
It can be assumed, that the high speciﬁc energy input into the process
initiates material synthesis already in the volume of the plasma,
at least partially. Coalescence of nucleation centers in the plasma has
potentially a coordination effect at the surfaces. They exhibit a high
chemical afﬁnity and low surface mobility. Considering the higher reaction
temperatures and shorter reaction times at the higher electric
power, the growth of nucleation centers can consequently explain the
formation of dendrites at energies 200 and 900 kJ cm−3
in PECVD
from TTMS (MW).
Above mentioned effects are deﬁnitely dependent on the temperature
conditions at the plasma-surface interface during the deposition,
too. However, a synchronous temperature diagnostics during the deposition
has not been performed in this investigation. Nevertheless, the
characteristics of both jets regarding the neutral gas temperature have
been published in the previous studies [20,36,37]. The results suggest
a maximum temperature of 420 K and 320 K during the deposition for
the MW and RF jet, respectively. Thus, the ratio of temperature differences
from the laboratory temperature is 3:1 (MW:RF) and hence
smaller than the ratios of speciﬁc energies (in this study they are in
the range from 4:1 to 45:1). In order to prove the speciﬁc effects of
the surface temperature and its correlation with the Yasuda parameter,
in situ thermography of the deposition would be desirable as well as the
investigation of nanoparticle formation in the plasma. These are subjects
to further studies.
5.2. On chemical properties and morphological structure
Regarding the carbon content, the increasing trends of energy per
molecule from 20 to 50 kJ cm−3
(RF) and from 200 to 900 kJ cm−3
(MW) correlate with the decreasing trends of the bulk carbon concentration
from 5 to 4% (RF) and from 3 to 1% (MW), respectively. The effect
demonstrates the higher efﬁciency of precursor decomposition and increased
oxidation of the SiOx matrix at higher speciﬁc energy. Moreover,
this energy dependent trend is related to an enhanced stoichiometric
ratio from x = 1.8 to 2.1 for the RF deposition and 1.7 to 2.2 for the
MW deposition. This observation is in accordance with the expectation
based on the state of the art. Nevertheless, the difference in speciﬁc energies
between the RF and MW experiments is about one order of magnitude.
Therefore, a comparison between RF and MW series can be
performed qualitatively at best, by discussing general trends. On the
other hand it is a surprising result, that, given the largely diversiﬁed
ﬁlm structure produced by heavily differing speciﬁc energies between
RF and MW experiments allows to distill similarities and parallels related
to the chemical properties, which effectuate only small variations of
the SiO2 chemistry in the coatings produced by PECVD.
The dendritic nanostructures deposited with the high power MW jet
could be used in applications, where a large speciﬁc surface is needed,
e.g. heterogeneous chemistry, catalytic surfaces, sensors, hydrogen storage
or battery separators. Further advantage of the ﬁlms is their SiO2
composition with low residual carbon content, which promises good
chemical stability. The compact and smooth ﬁlms produced with the
low power RF jet could be used as a functional interlayer with low surface
roughness for multilayer coatings. The presence of silanol groups in
the ﬁlms together with low carbon concentration makes the coating relevant
for biochemical tests, too. The ﬁlm properties could be even enhanced
by selective functionalization. The results suggest that using a
complex precursor molecule and tuning of input power can produce
both nanostructures and smooth ﬁlms (and their combinations) by a
single process.
6. Conclusion
In this study we investigated atmospheric pressure PECVD based on
tetrakis(trimethylsilyloxy)silane (TTMS). The usage of TTMS for thin
ﬁlm deposition by means of a radiofrequency non-thermal atmospheric
pressure plasma jet (RF jet) is reported herewith for the ﬁrst time. We
Table 4
Characteristic infra-red absorption bands of SiOx network.
Wavenumber [cm−1
] Vibration mode Appearance in samples
RF1 RF2 MW1 MW2
1279 Si–CH3 w n w n
1130–1200 Si–O–Si transv. optical (TO) w w s s
1050–1085 Si–O–Si asym. stretching (AS) s s s s
920–940 Si–OH stretching w m m m
800–810 Si–O–Si bending w w m m
450 Si–O–Si rocking m m h h
Notation: s — strong, m — medium, w — weak, n — no peak, and h — hypothetical. Indicators:
Si–CH3 — carbon content, TO — porosity or roughness, AS — stoichiometry, and Si–OH
— oxidation grade.
117J. Schäfer et al. / Surface & Coatings Technology 295 (2016) 112–118
compared results from deposition experiments involving TTMS in two
different plasma sources: an RF jet and a microwave plasma jet (MW
jet). Surprisingly, it could be found, that – despite the dramatic speciﬁc
energy differences between both jets – the chemical composition of the
deposited ﬁlms is very similar, close to SiO2. However, largely differing
ﬁlm morphologies could be obtained covering the broad interval of deposition
rates from 0.1 nm/s to 17 nm/s.
Nine different analytical approaches regarding ﬁlm and surface analysis
have been applied during this study in order to present a consistent
picture of the morphological structure (SEM, cross-section polishing for
SEM, AFM, imaging spectroscopic reﬂectometry) and chemical composition
(transmission FTIR, ATR-FTIR, EDX, XPS, and sputtering XPS) of
deposited ﬁlms.
By deposition from TTMS with the MW jet intact dendritical nanostructures
have formed. Their chemical composition resembles that of
SiO2, with only 1% contamination by residual carbon and can promote
interesting applications that require chemically stable yet large effective
surface areas. These properties qualify the coating for example for adsorption
tests regarding surface sensor applications or as matrix surface
for heterogeneous catalysis. On the other hand, the deposition by the RF
jet demonstrates, that a comparable chemical composition of the ﬁlm
can be achieved with an energy input 18-times smaller than in the
MW jet. Here, the resulting surface exhibits an extremely low surface
roughness. The technological potential of the studied ﬁlms demonstrates
the relevance of investigations that are aimed to introduce precursor
molecules of large and complex molecular structure into gas
discharges at atmospheric pressure accenting the role of chemical
functionalization and surface nanostructuring.
Acknowledgments
The authors thank David Nečas for help with ISR data analysis and
Petr Klapetek for AFM measurements. The work has been funded in
part by the following sources: 7th European Framework Program, project
no. 316216, “PlasmaShape”; DFG, Transregio 24 “Fundamentals of
Complex Plasmas”, project CZ.1.05/2.1.00/03.0086, funded by the European
Regional Development Fund; project LO1411 (NPU I) funded by
the Ministry of Education Youth and Sports of Czech Republic; “CEITEC
— Central European Institute of Technology” (CZ.1.05/1.1.00/02.0068)
from the European Regional Development Fund; and by project
TE02000011 of the Czech Technology Agency.
References
[1] H. Nagel, A.G. Aberle, R. Hezel, Prog. Photovolt. Res. Appl. 7 (1999) 245–260.
[2] S.E. Alexandrov, N. McSporran, M.L. Hitchman, Chem. Vap. Depos. 11 (2005)
481–490.
[3] J. Schäfer, S. Horn, R. Foest, R. Brandenburg, P. Vašina, K.-D. Weltmann, Surf. Coat.
Technol. 205 (2011) S330–S334.
[4] D. Gnanasekaran, K. Madhavan, B.S.R. Reddy, J. Sci. Ind. Res. 68 (2009) 437.
[5] B. Barwe, F. Riedel, O.E. Cibulka, I. Pelant, J. Benedikt, J. Phys. D Appl. Phys. 48 (2015)
314001.
[6] T. Maschmeyer, M.C. Klunduk, C.M. Martin, D.S. Shephard, J.M. Thomas, B.F.J.
Johnson, Chem. Commun. (1997) 1847–1848.
[7] R. Duchateau, Chem. Rev. 102 (2002) 3525.
[8] G. Sievers, S. Mueller, A. Quade, F. Steffen, S. Jakubith, A. Krutha, V. Brueser, J. Power,
Sources 268 (2014) 255–260.
[9] D.A. Loy, K.J. Shea, Chem. Rev. 95 (1995) 1431.
[10] I.L. Dubchak, V.E. Shklover, M.Y. Antipin, Y.T. Struchkov, V.M. Kopylov, A.M.
Muzafarov, P.L. Prikhod'ko, A.A. Zhdanov, J. Struct. Chem. 23 (1982) 219.
[11] M. Bordeau, J. Dedier, E. Frainnet, J.-P. Fayet, P.J. Mauret, Organomet. Chem. 61
(1973) 91.
[12] A.I. Lazarev, The Vibrational Spectra and Structure of Silicates [in Russian], Nauka,
Leningrad, 1968.
[13] J. Hnilica, J. Schäfer, R. Foest, L. Zajčková, V. Kudrle, J. Phys. D Appl. Phys. 46 (2013)
335202.
[14] H. Kim, H. Oh, C. Lee, D. Jung, J.H. Boo, Bull. Kor. Chem. Soc. 35 (2014) 2941.
[15] C. Tendero, C. Tixier, P. Tristant, J. Desmaison, P. Leprince, Spectrochim. Acta B 61
(2006) 2–30.
[16] L. Bárdos, H. Baránková, Thin Solid Films 518 (2010) 6705–6713.
[17] D. Trunec, Z. Navrátil, P. St'ahel, L. Zajíčková, V. Buršíková, J. Čech, J. Phys. D Appl.
Phys. 37 (15) (2004) 2112–2120.
[18] D. Trunec, L. Zajíčková, V. Buršíková, F. Studnička, P. St'ahel, V. Prysiazhnyi, V. Peřina,
J. Houdková, Z. Navrátil, D. Franta, J. Phys. D Appl. Phys. 43 (22) (2010) 225403.
[19] J. Schäfer, R. Foest, A. Quade, A. Ohl, K.-D. Weltmann, J. Phys. D Appl. Phys. 41 (2008)
194010.
[20] J. Schäfer, R. Foest, A. Quade, A. Ohl, J. Meichsner, K.-D. Weltmann, Eur. Phys. J. D 54
(2009) 211.
[21] J. Hnilica, V. Kudrle, P. Vašina, J. Schäfer, V. Aubrecht, J. Phys. D Appl. Phys. 45 (2012)
055201.
[22] M. Moisan, C. Beaudry, P. Leprince, Phys. Lett. 50 (1974) 125.
[23] J. Schäfer, R. Foest, A. Ohl, K.-D. Weltmann, Plasma Phys. Control. Fusion 51 (2009)
124045.
[24] D. Nečas, V. Čudek, J. Vodák, M. Ohlídal, P. Klapetek, J. Benedikt, K. Rügner, L.
Zajíčková, Meas. Sci. Technol. 25 (2014) 115201.
[25] D. Franta, D. Nečas, L. Zajíčková, Opt. Express 15 (2007) 16230–16244.
[26] R. Foest, M. Schmidt, H. Gargouri, Eur. Phys. J. D 68 (2014) 23.
[27] C. Martinet, R.A.B. Devine, J. Appl. Phys. 77 (9) (1995) 4343–4348.
[28] D.R. Anderson, Infrared, Raman and Ultraviolet Spectroscopy, in: A.L. Smith (Ed.),
Analysis of Silicones, Wiley-Interscience, New York 1974, p. 247 (Chapter 10).
[29] C. Rau, W. Kulisch, Thin Solid Films 249 (1994) 28–37.
[30] K. Aumaille, C. Vallée, A. Granier, A. Goullet, F. Gaboriau, G. Turban, Thin Solid Films
359 (2000) 188.
[31] L. He, T. Inokuma, Y. Kurata, S. Hasegawa, J. Non-Cryst. Solids 185 (1995) 249.
[32] A. Kučírková, K. Navrátil, L. Pajasová, V. Vorlíček, Appl. Phys. A Mater. Sci. Process. 63
(1996) 495.
[33] R.A.B. Devine, Trans. Mater. Res. Soc. Jpn. 8 (1992) 165.
[34] W. Kaiser, P.H. Keck, C.F. Lange, Phys. Rev. 101 (1956) 1264.
[35] P.G. Pai, S.S. Chao, Y. Takagi, G. Lucovsky, J. Vac. Sci. Technol. A 4 (1986) 690.
[36] J. Hnilica, V. Kudrle, L. Potočňáková, IEEE Trans. Plasma Sci. 40 (2012) 2925.
[37] J. Schäfer, Z. Bonaventura, R. Foest, Eur. Phys. J. Appl. Phys. 71 (2015) 20804.
118 J. Schäfer et al. / Surface & Coatings Technology 295 (2016) 112–118