22nd International Symposium on Plasma Chemistry
July 5-10, 2015; Antwerp, Belgium
Analysis of defects in permeation barrier coatings for polymers
F. Mitschker1, S. Steves1, M. Gebhard2, M. Rudolph1, A. Devi2 and P. Awakowicz1
1
Institute for Electrical Engineering and Plasma Technology, Ruhr-University Bochum, DE-44780 Bochum, Germany
2
Inorganic Materials Chemistry, Ruhr-University Bochum, DE-44780 Bochum, Germany
Abstract: Barrier coatings like silicon oxide deposited by PECVD on polymers offer a
limited barrier performance. This is due to microscopic defects emerging in thin films.
Therefore, visualization of defects by reactive oxygen etching is performed to analyse
defect densities in barrier coatings. Defect densities are correlated with oxygen permeation
rates and deposition process parameters.
Keywords: barrier coatings, defects, PECVD, silicon oxide, permeation, polymer, PET
1. Introduction
A widely used plastic for packaging, polyethylene
terephthalate (PET) offer limited barrier properties against
gas permeation. For many applications of PET improved
barrier properties are essential. In the field of barrier
coatings silicon oxide (SiO x ) films deposited by plasmaenhanced chemical vapor deposition (PECVD) are of
great interest since they are transparent and offer a
significant increase in barrier performance.
The
deposition process is performed on PET foils by means of
a PECVD process in a pulsed microwave driven lowpressure plasma. While the adjustment of the microwave
power allows for a control of the ion production during
the plasma pulse, a substrate bias controls the energy of
ions impinging on the substrate. A barrier improvement
factor of up to 150 is achieved for barrier coatings with a
thickness of 25 nm. Still a residual oxygen transmission
rate (OTR) through the coating is present. This is led
back to the microscopic defects existing in PECVD
coatings that may arise from particles, surface roughness
or inhomogeneities on the surface of the polymer [1-3].
A detailed analysis of deposited films is performed by
means of oxygen permeation measurements and detection
of microscopic defects. The detection of defects is
arranged by an established technique based on the
increase of defect diameter by reactive oxygen etching of
the polymer and imaging with SEM [2]. The etching
process is performed in a capacitively coupled oxygen
plasma.
The defects are quantified by an image
processing software. Defects of various diameters and
distributions are observed leading to a classification of
defects. The defect density is correlated with film
thickness, OTR and various film compositions. The
effect of an additional substrate bias on the defect density
is also presented.
reactor chamber can be evacuated to a base pressure of
1 Pa. Microwave power is applied to the system by a
modified Plasmaline antenna [4]. For the deposition of
SiO x barrier coatings hexamethyldisiloxane (HMDSO) is
evaporated and fed into the chamber together with
oxygen.
A generator provides microwave power
(f = 2.45 GHz) with a maximum of P = 2 kW (continuous
wave). The microwave is applied in pulsed mode in order
to prevent the PET foils from overheating and to attain a
homogeneous coating. Oxygen and HMDSO gas fluxes
are chosen regarding the homogeneity criterion presented
by Deilmann et al. [5]. A feedback control allows for the
design of arbitrary voltage signals at the substrate holder.
This is implemented by a feedback of the substrate
voltage at the vacuum feed through using a digital
oscilloscope with a high voltage probe tip. The feedback
control applies a fast Fourier transform (FFT) of the
signal, Fourier components of the input signal are
iteratively adjusted to create the desired waveform at the
substrate [6, 7]. In this work the barrier properties of the
coatings are increased with a sinusoidal substrate bias
(13.56 MHz).
2. Experimental setup
The experimental setup (see Fig. 1) for the treatment of
PET foils (23 µm, Hostaphan, RD23, Mitsubishi) and
bottles is composed of a vacuum chamber with a volume
of 6 liter and is capable of treating various bottle sizes up
to 1.5 liter and PET foils by means of a foil carrier. The
Atomic oxygen etching of coated SiO x films is
performed in a capacitively coupled discharge with
optimized process parameters and solely fed with oxygen.
A pulsed radio frequency (rf) signal with a frequency of
13.56 MHz, a pulse frequency of 1 kHz and a duty-cycle
of 40% is used to generate the plasma at a pressure of
P-III-6-39
Fig. 1. Experimental setup for plasma treatment of plastic
foils and bottles.
1
20 Pa. Power dissipated in the discharge is 40 W.
Thereby, an etch rate of 1.4 µm h-1 is attained for PET
foil.
3. Method
Usually microscopic coating defects that limit the
barrier performance of barrier coatings are not visible by
optical or electron microscopy. In order to analyse defects
quantitatively, visualization has to be achieved. This is
arranged by an established technique based on the
increase of defect diameter by reactive oxygen etching of
the polymer surface and imaging with SEM [2]. In this
process the SiO x barrier film serves as a mask against
atomic oxygen etching. Hence, only the polymer surface
next to coating defects is etched underneath the SiO x
barrier coating, resulting in crater shaped structures as
shown in Fig. 2. Accordingly, defects are visualized and
quantified by means of SEM and an image processing
software. For each sample 40 SEM images are analyzed,
covering an overall measuring area of 0.56 mm² for defect
investigation.
OTR is measured with a Mocon
OX-TRAN 2/61 (Mocon Inc., Minneapolis, USA) using
the carrier gas method. Tests are performed at 23 °C and
0% relative humidity. For each sample the OTR is
analyzed before defect density is determined,
respectively.
Fig. 3. Type 1 and 2 defects in 25 nm SiO x barrier
coating deposited with an O 2 /HMDSO ratio of 50.
defect does not change for prolonged etch times even if
the diameter of a type 2 defect outgrows 1 µm. As
defects close-by each other may overlap after extended
etch times, following investigations are performed after
60 min of oxygen etching. The trend of defect diameters
is presented in Table 1. Besides that, the mean growth in
diameter of different defects per hour dd defect /dt and the
mean volume etch rate per hour dV defect /dt is shown. The
mean volume etch rate is calculated assuming a half
sphere for the crater like structure. With increasing initial
defect diameter a decrease of mean defect diameter
growth per hour dd defect /dt is clearly observed for defects
of type 1. Apart from that, a volume etch rate per hour of
dV defect /dt ≈ 1 µm³ h-1 is observed. This leads to the
conclusion that type 1 defects are initially big enough to
provide access for atomic oxygen during etching and
leaving of volatile etch products. However, the volume
etch rate of defect type 2 is lower by one order of
magnitude. This, in combination with the appearance as
dark spots in the images allows the assumption of an
unharmed SiO x film on top of the etched polymer surface
in this case. Atomic oxygen is able to pass through type 2
coating defects, but the path has to be small. Therefore,
the amount of atoms and molecules passing through is
constrained.
Fig. 2. Visualization of coating defects after 1 hour of
oxygen etching in rf CCP plasma.
4. Results
Two sorts of defects can be observed, that are classified
by its size after 60 min of oxygen etching. Defects of
type 1 have a diameter of ≥ 1 µm and type 2 defects a
diameter of ≤ 1 µm. In addition to the diameter the
appearance of sort 1 and 2 differ from each other. The
former shows a crater like structure visible as bright rings
surrounding the defect (Figs. 2 and 3). The latter appears
as a dark spot in SEM images, as shown in Fig. 3.
The growth of defects in 25 nm SiO x barrier coatings
on PET foil is shown in Fig. 4. The sample is etched for a
duration between 60 and 180 min. Fig. 4 shows
exemplarily that no additional defects appear for extended
etch times between 60 min and 180 min. However, the
diameter of observed defects increases. The type of
2
Table 1. Exemplarily trend of defect diameter d of etched
SiOx films on PET.
Etch time
d defect A
(type 1)
d defect B
(type 2)
d defect C
(type 1)
d defect D
(type 1)
60 min
2,28 µm
0,66 µm
1,31 µm
1,69 µm
80 min
2,35 µm
0,68 µm
1,44 µm
1,93 µm
100 min
2,46 µm
0,81 µm
1,60 mm
1,93 µm
120 min
2,49 µm
0,90 µm
1,70 µm
2,00 µm
180 min
2,70 µm
1,02 µm
2,13 µm
2,34 µm
A
B
C
D
dd defect /dt
0,21 µm h-1
0,21µm h-1
0,40 µm h-1
0,32 µm h-1
dV defect /dt
1,03 µm³ h-1
0,10 µm³ h-1
0,98 µm³ h-1
1,05 µm³ h-1
P-III-6-39
defect free areas enlarge with increasing film thickness
and agglomerates of defects can be found. It reveals that
defects arise probably from polymer areas with higher
local density of inhomogeneities on the surface and
therefore a higher amount of defects in these areas of the
barrier coating. A 25 nm SiOx film exhibits a noticeable
smaller defect density.
Fig. 4. Defects of a 25 nm SiO x film deposited with
O 2 /HMDSO = 50 on PET foil. After variation of etching
time in CCP defects are visualized by SEM.
The presented technique is handy for investigation of
coating uniformity, particularly. At local elevations of the
polymer surface barrier coatings are expected to fail [2].
The minimum coating thickness is determined by the
polymer surface roughness [8]. Thus, in order to achieve
a uniform coverage the coating thickness has to exceed
the surface roughness. Fig. 5 shows representative SEM
images for defects in silicon oxide coatings of various
film thicknesses. Fig. 6 shows the defect density per mm²
after an etch time of 60 min as a function of film
thickness. The SiO x barrier coatings in Figs. 5 and 6 are
deposited on PET foil with an oxygen to HMDSO ratio of
50 and a substrate bias of U bias = 36 V. As Fig. 5 shows,
a 2 nm thick film already covers the PET surface against
oxygen etching, but a high defect density is present (see
Fig. 6). In this case the coating is too thin to exhibit a
self-contained silicon oxide layer. This leads to cracking
and partial delamination of the silicon oxide film. Hence,
defects in a 2 nm barrier coating do not show a crater like
structure in SEM images. Coatings with a thickness of
3 nm exhibit a decreased defect density. The SiO x film is
thick enough to stay intact at the majority of etched
defects, but still cracking and delamination can occur,
locally. For small film thickness both types of defects are
randomly distributed. However, defects can also arise
from mechanical stress. In this case, defects are found to
be oriented in a straight line that originates from the
direction of mechanical stress, as can be seen in Fig. 4
(5 nm). Further increase of film thickness leads to a
decrease of defect density.
As a matter of fact,
distribution of defects gets more inhomogeneous. Thus,
P-III-6-39
Fig. 5. : SEM images of defects in SiOx coatings as a
function of film thickness. Coatings are deposited with an
oxygen to HMDSO ratio of 50 and a bias of U bias = 36 V.
Fig. 6. Overall defect density of SiO x films deposited
with an oxygen to HMDSO ratio of 50 and a substrate
bias of U bias = 36 V as a function of film thickness.
The process parameters have a big influence on the
barrier performance of the applied coatings. Hence,
analysis of OTR and defect density for variation of
oxygen to HMDSO ratio with and without bias is
performed. In Fig. 7a OTR is shown as a function of
O 2 /HMDSO ratio. In Fig. 7b only defects of type 2 are
3
taken into account. It indicates that no clear dependence
of O 2 /HMDSO ratio and influence on OTR is present.
However, the overall type 2 defect density for a
deposition with substrate bias is lower compared to a
deposition without bias, except of an O 2 /HMDSO ratio of
200. In this case a comparable defect density is observed.
In contrast, a decreasing defect density of type 1 is shown
in Fig. 7c for SiO x coatings deposited with an
O 2 /HMDSO ratio of larger than 100 and without bias.
All barrier coatings that offer a reasonable barrier
performance with an OTR of < 10 cm³ m-2 day-1 exhibit
less than 200 type 1 defects per mm2. Nevertheless, no
universal relation is found. SiO x coatings deposited with
substrate bias and low O 2 /HMDSO ratio of 25 and 50,
offer below 200 defects/mm2, but no oxygen barrier is
observed for these ratios. Thus, this method is not up to
investigate the film composition or porosity since
SiO x C y H z coatings in case of a low O 2 /HMDSO ratio
protect the polymer surface as well as a SiO x film.
Furthermore, the porous structure with a high amount of
nanopores assumed for coatings without oxygen
permeation barrier appear to be too small for atomic
oxygen to enter. Hence, the discussed technique is
especially appropriate when the chemical composition of
barrier coatings is known. In this regard, determined
defect densities give insight in inhomogeneities and
morphology of coated polymer surfaces and homogeneity
of deposited barrier films.
MW
MW + Bias
(a)
type 2 defects
(b)
type 1 defects
(c)
Fig. 7. (a) Oxygen transmission rate (OTR) and defect
densities of (b) type 1 and (c) type 2 defects for 25 nm
SiO x coatings as a function of O 2 /HMDSO ratio with a
substrate bias of U bias = 36 V (MW + bias) and grounded
substrate holder (MW). PET foil OTR reference is
indicated as dashed line.
4
5. Conclusion
Two different kinds of coating defects in SiO x barrier
coatings can be observed. After 60 min of oxygen
etching the diameter of defect type 1 is ≥ 1 µm and the
diameter of type 2 is ≤ 1 µm. Two etch mechanisms seem
to be present, respectively. Entering of atomic oxygen
and leaving of volatile etch products through defects of
type 1 is possible without serious hindrance. On the
contrary, type 2 defects exhibit an unharmed SiO x film
that constrains atoms and molecules from passing
through.
The overall defect density decreases with increasing
film thickness. With decreasing defect density the
distribution of defects gets more inhomogeneous and
agglomerates appear. This is led back to inhomogeneities
on the underlying polymer surface in certain areas.
For defects of type 2 no dependence of oxygen
transmission rate or O 2 /HMDSO ratio is observed.
However, coatings deposited with bias show less defects
of type 2 than coatings without bias. Type 1 defects show
a decreased density for an O 2 /HMDSO ratio above 100.
Barrier coatings with an OTR of < 10 cm³ m-2 day-1
exhibit less than 200 type 1 defects/mm². Depositions
with substrate bias and with an O 2 /HMDSO ratio of 25
and 50 show no barrier performance, but also a defect
density below 200 mm-2. Hence, the presented technique
is suitable for investigation of homogeneity and
morphology of barrier coatings and the underlying
polymer surface.
6. Acknowledgement
The authors gratefully acknowledge the support
provided by the German Research Foundation (DFG)
within the framework of the Transregional Collaborative
Research Center TRR 87/1 (SFB-TR 87) “Pulsed high
power plasmas for the synthesis of nanostructured
functional layers”.
7. References
[1] A.S. da Silva Sobrinho, et al. J. Vac. Sci. Technol.
A, 18, 149 (2000)
[2] A.S. da Silva Sobrinho, et al. Plasma Polymers, 3,
4 (1998)
[3] A.S. da Silva Sobrinho, et al. Surf. Coatings,
116-119, 1204-1210 (1999)
[4] M. Deilmann, et al. Plasma Process. Polymers, 6,
695-699 (2009)
[5] M. Deilmann, et al. Plasma Process. Polymers, 6,
362-699 (2009)
[6] M.M. Patterson, et al.
Plasma Sources Sci.
Technol., 16, 257 (2007)
[7] S. Steves, et al. J. Phys. D: Appl. Phys., 46, 084013
(2013)
[8] H. Chatham. Surf. Coatings, 78, 1-9 (1996)
P-III-6-39