PDMS-Parylene Adhesion Improvement via Ceramic Interlayers to Strengthen the Encapsulation of Active Neural Implants

Abstract

Parylene-C has been used as a substrate and encapsulation material for many implantable medical devices. However, to ensure the flexibility required in some applications, minimize tissue reaction, and protect parylene from degradation in vivo an additional outmost layer of polydimethylsiloxane (PDMS) is desired. In such a scenario, the adhesion of PDMS to parylene is of critical importance to prevent early failure caused by delamination in the harsh environment of the human body. Towards this goal, we propose a method based on creating chemical covalent bonds using intermediate ceramic layers as adhesion promoters between PDMS and parylene. To evaluate our concept, we prepared three different sets of samples with PDMS on parylene without and with oxygen plasma treatment (the most commonly employed method to increase adhesion), and samples with our proposed ceramic intermediate layers of silicon carbide (SiC) and silicon dioxide (SiO2). The samples were soaked in phosphate-buffered saline (PBS) solution at room temperature and were inspected under an optical microscope. To investigate the adhesion property, cross-cut tape tests and peel tests were performed. The results showed a significant improvement of the adhesion and in-soak long-term performance of our proposed encapsulation stack compared with PDMS on parylene and PDMS on plasma-treated parylene. We aim to use the proposed solution to package bare silicon chips on active implants.

Full text

Abstract—Parylene-C has been used as a substrate and
encapsulation material for many implantable medical devices.
However, to ensure the flexibility required in some
applications, minimize tissue reaction, and protect parylene
from degradation in vivo an additional outmost layer of
polydimethylsiloxane (PDMS) is desired. In such a scenario, the
adhesion of PDMS to parylene is of critical importance to
prevent early failure caused by delamination in the harsh
environment of the human body. Towards this goal, we propose
a method based on creating chemical covalent bonds using
intermediate ceramic layers as adhesion promoters between
PDMS and parylene.
To evaluate our concept, we prepared three different sets of
samples with PDMS on parylene without and with oxygen
plasma treatment (the most commonly employed method to
increase adhesion), and samples wit h our proposed ceramic
intermediate layers of silicon carbide (SiC) and silicon dioxide
(SiO2). The samples were soaked in phosphate-buffered saline
(PBS) solution at room temperature and were inspected under
an optical microscope. To investigate the adhesion property,
cross-cut tape tests and peel tests were performe d. The results
showed a significant improvement of the adhesion and in-soak
long-term performance of our proposed encapsulation stack
compared with PDMS on parylene and PDMS on plasma-
treated parylene. We aim to use the proposed solution to
package bare silicon chips on active implants.
I. INTRODUCTION
Neural interfaces, in general, are used to interact with the
nervous system, to record, stimulate or block electrical
activity. For this purpose, they are implanted close to the
targeted region. To reduce the foreign body response, prevent
water permeation into the implanted electronics, and avoid
diffusion of corrosion products to the tissue, it is necessary to
encapsulate the implantable device [1]. Polymers are
commonly used as an encapsulation material due to their
mechanical properties. In particular, parylene has the
advantage of easy deposition and etching processes, high
biocompatibility, and a good ionic barrier property, which
makes it a very promising material both for the substrate as
well as the encapsulation layer of the implant. However, its
high Young’s modulus makes it still relatively stiff compared
to the tissue [2]. This stiffness might be tolerated in some
applications, but might be completely unsuitable in others,
N. Bakhshaee, R. Dekker, W. A. Serdijn and V. Giagka are with the
Department of Microelectronics, Faculty of Electrical Engineering,
Mathematic s and Computer Science, Delft Un iversity of Technology, Delft,
The Netherlands. (N.bakhshaee@tudelft.nl, V.Giagka@tudelft.nl).
V. Giagka is also with the Fraunhofer Institute for Reliability and Micro-
integration IZM, Berlin, Germany.
where softer elastomers such as PDMS are necessary to avoid
tissue damage [3]. In addition, [4] suggests that strong tissue
reaction can be developed around parylene-C-based implants
after 8-month in vivo tests, while tissue reaction to PDMS is
minimal. [5] also shows oxidation and chlorine abstraction of
parylene after 3.25 years of implantation. As a consequence,
an additional outmost encapsulation layer with a similar
mechanical property to the tissue is desired.
PDMS has been selected for this purpose due to its low
Young’s modulus [2]. In such a scenario, the adhesion
between the parylene and PDMS layers will be critical for the
longevity of the neural interface [6]. Due to the different
polymeric backbones of parylene and PDMS, and their
hydrophobicity [7], adhesion between them is often based on
physical bonds, which are relatively weak, rendering the
layers prone to delamination, especially when water is
present at the interface [8]. A chemical bond on the other
hand is stronger by nature, hence is expected to lead to longer
encapsulation lifetimes. To create such chemical bonds a
number of different plasma treatments has been proposed to
enhance the bonding of PDMS to parylene for microfluidic
applications [7]. In another approach, thin ceramic
intermediate layers were used to make a chemical bond
between PDMS and polyimide [8].
Inspired by the above approach, we have investigated the
use of intermediate thin films, namely SiC and SiO2, to create
transient covalent bonds between our parylene and PDMS, as
illustrated in Fig. 1. Here, we assume that a carbon-carbon
chemical bond can be created between SiC and parylene. A
silicon-silicon covalent bond can be created between SiO2
and SiC [8]. PDMS is expected to have a strong adhesion to
SiO2 due to the presence of hydroxyl groups at the interface.
In the rest of this paper: Section II describes the
experiments, results are presented in Section III and
discussed in Section IV. Finally, Section V concludes the
paper.
PDMS-Parylene Adhesion Improvement via Ceramic Interlayers to
Strengthen the Encapsulation of Active Neural Implants
Nasim Bakhshaee Babaroud, Ronald Dekker, Wouter Serdijn, and Vasiliki Giagka
Figure 1. A smooth transition of chemical bonds between parylene and
PDMS to improve the adhesion.

II. METHODS
A. Sample Fabrication
1) Si-based samples
To evaluate the proposed concept, relevant test structures
were fabricated, as illustrated in Fig. 2.
A 400nm SiO2 layer was deposited at 400 ºC on a Si
wafer as an isolation layer by using a plasma-enhanced
chemical vapor deposition (PECVD) process. A 5µm
parylene layer was deposited using a SCS PDS 2010
parylene coater that employs a chemical vapor deposition
(CVD) technique at room temperature, followed by applying
a A-174 adhesion promoter. To create flavor 2(a), PDMS
(Dow Corning Sylgard 184) was mixed with a curing agent
at a 10:1 ratio, spin-coated at 1250 rpm directly on parylene,
and cured at 75°C for 3 hours. This resulted in a 50 µm thick
layer. For the samples of Fig. 2(b), an oxygen plasma
treatment (50 sccm of oxygen flow, 60 W, 0.25 mTorr,
Diener electronic GmbH Germany) was applied on the
parylene surface for 1 min before PDMS coating to increase
the surface activation energy in order to improve the
adhesion. The proposed SiC-SiO2 stack of Fig. 2(c) was
created as follows: 25 nm SiC is PECVD deposited on
parylene at 180ºC temperature. Then 25 nm SiO2 is
deposited in the same chamber. Finally, the PDMS layer is
spin-coated on top. For the SiC deposition, three different
recipes (R1, R2, and R3) have been developed, as in Table I.
Based on [9], a silane starving mode (R3) and a non-silane
starving mode (R1 and R2) were chosen. The silane starving
and non-silane starving modes refer to the precursor ratio
(SiH4/(CH4+SiH4)) of 0.17 and 0.29, respectively. Hydrogen
is also used as dilution gas. Based on this work, low silane
and high hydrogen flow result in higher Si-C bond density.
2) Free standing membrane
This set of test structures was developed for the needs of
the peel test described in Section II B.2. For these, the
Si/SiO2 substrates were removed to create free standing
parylene-PDMS stacks with all the flavors of Fig. 2. To see
the differences between the result for adhesion test for
samples with and without ceramic layers, these layers were
deposited only on half of the wafer (using some metal masks
to cover the other half during deposition) as in Fig.2(d).
B. Characterization
1) Tape Test
To investigate the adhesion between SiC and parylene, a
tape test was performed based on ASTM 3359 [10]. To do
this, several samples with SiC only (deposited using the three
recipes of Table I) on parylene were created. Next, a tape test
was performed at a 180-degree on the grid pattern created
previously on the film. The grid pattern was optically
evaluated before and after test.
2) Peel Test
To do the peel test, one of the layers is clamped and
remains fixed in position while the other layer (which is
attached onto a moving arm of the testing tool, here a Zwick
1455 tensile testing machine, Fig. 3(b)) is pulled on at a
constant speed and at an 180-degree angle. For this test,
samples with the same encapsulation stacks as in Section III
A.1 were prepared. A Kapton tape was applied on the sample
before PDMS coating to create a clamping point for the tool
and initiate the peeling (see Fig. 3(a)). The peel test was
performed by pulling on the PDMS layer at 500µm/s and a
load of 10 N via the attached Kapton tape. The above test
structures allow for the evaluation of the adhesion of PDMS
to the layer underneath. Hence, for the samples in Fig. 2(c)
only the adhesion of PDMS to SiO2 can be evaluated. To
evaluate the adhesion among the remaining layers in Fig. 2(c)
the free standing membranes of Section II A.2 were created
to allow peeling from both the parylene and the PDMS sides.
3) PBS Soak Test
To evaluate the performance of our encapsulation stack in
a wet environment like the human body, samples (with grid
pattern created on them) were soaked in a PBS solution at
room temperature for 60 days. All samples were monitored
optically and a tape test was performed after 12 and 60 days.
4) Comparison of the three SiC recipes
In addition to adhesion improvement, ceramic layers also
act as a barrier layer against moisture. A layer with smaller
Figure 2. Encapsulation stacks: (a)PDMS spin coated on untreated
parylene, (b)PDMS spin coated on plasma-treated parylene, (c)PDMS spin
coated on parylene with ceramic interlayers, (d) free standing membrane.
Table I. Three different recipes used for SiC PECVD deposition at 180ºC.
Recipe
Deposition parameters
SiH4
(sccm)
CH4
(sccm)
H2
(sccm)
P
(W)
P
(mbar)
R1
20
45
0
4
0.7
R2
1.6
3.7
200
6
2.6
R3
2
10
90
30
2
Figure 3. Sample (a) before and (b) after installation in the peel
test machin e.

number of pinholes is expected to be a better moisture
barrier. To evaluate the number of pinholes of different
recipes for SiC, samples including only three different SiC
recipes on 400 nm PECVD SiO2 were placed inside Buffered
HydroFluoric acid (BHF) 1:7 for 10 minutes. BHF can
penetrate through the pinholes in SiC to reach SiO2 and etch
it, which can be detected optically.
Fourier Transform Infrared spectroscopy (FTIR) and
stress were measured for each recipe compared to a bare Si
wafer. A comparative study of the FTIR diagrams for the
three recipes could indicate the ones with the higher Si-C
bond density. Layers with a higher Si-C bond density are
expected to have less pinholes, but a higher amount of stress.
III. RESULTS
A. Tape Test
As shown in Fig. 4(a,b), some parts of the layer peeled off
from the substrate after the tape test. Detailed inspection
revealed that the delamination happened between parylene
and its SiO2 substrate (Fig. 4(b)), leaving the parylene-SiC
layers still firmly attached to each other. This result shows
that the adhesion of SiC to parylene is better than the
adhesion of parylene to SiO2 (which was confirmed by a
separate test).
B. Peel Test
The samples of Section II A.1 that had PDMS on parylene
with and without oxygen plasma were easily peeled. On the
other hand, no peeling was observed on samples that included
the whole stack. The adhesion was so strong that the peel test
caused the Si wafer to break and the PDMS layer to tear apart
before peeling started. Efforts to strengthen the PDMS layer
by making it thicker (80 µm) lead to the same result.
For the flexible test structures of Section II A.2, peeling
was performed by hand. As shown in Fig. 5, peeling of the
PDMS layer was very easy on the areas without ceramic
layers. But when the peeling reached to the region with
ceramic layer, it was not possible anymore and PDMS was
torn at the edge of this region, as can be seen from Fig. 5(b).
C. PBS Soak Test
As it was expected, for those samples that had PDMS on
parylene with and without oxygen plasma, delamination of
PDMS happened after creating the grid patterns on the film
and even before a tape test. For those samples that included
the whole stack of encapsulation, tape tests performed after
12 and 60 days revealed no delamination.
D. Comparison of the three SiC recipes
As illustrated by the optical image in Fig. 6, there are
more pinholes in the R1 (without hydrogen dilution gas)
compared to R2 and R3. The presence of the pinholes
depends on the chemistry used for deposition and the
cleanliness of the surface. The characterizations for three
different SiC layers show less pinholes in the recipes with
higher amount of hydrogen dilution gas (R2 and R3), as it can
compensate the effect of low deposition temperature (180 ºC,
due to the low glass transition temperature of parylene) and
lead to higher Si-C bond density [9]. As expected, more
pinholes and less stress are observed in the recipe without
any hydrogen gas (R1).
To gain an insight regarding the Si-C bond density of each
of the developed recipes, we look at their FTIR diagram in
Fig. 7 (here all waveforms are normalized to the maximum
peak). In the FTIR spectrum for a SiC layer, based on [9] we
expect the following: a peak at a wavenumber (1) between
720 to 780 cm-1 corresponding to Si-C stretching bonds, (2)
around 2000 cm-1 related to Si-Hn bonds, and (3) between
2800 to 3000 cm-1 for C-Hn bonds. From these results we
observe that R3 indeed has a high Si-C peak, as expected.
However, it also exhibits a higher C-Hn peak, which could
explain its relatively higher number of pinholes compared to
R2 (Fig. 6).
R1, R2 and R3, resulted in 205, 585 and 530 MPa
compressive stress, respectively. Stress could negatively
affect the adhesion and lead to delamination. To investigate
this, especially in the humid environment in the long term,
samples with different recipes for SiC were placed in soak
and investigated after 12 and 60 days. The result showed that
the adhesion was so strong that there was no delamination
even for those recipes with high amount of compressive
stress (R2 and R3).
Figure 6. Optical image of pinholes in SiC using recipe (a) R1, (b) R2, (c) R3.
Figure 5. Flexible sample including p arylene, ceramic layers (on only half
of the sample) and PDMS (a) during peeling, (b) torn PDMS after peeling.
Figure 4. Optical image of a sample with SiC on parylene, (a) before the
tape test, (b) after the tape test with zoomed-in picture of the removed
area.

IV. DISCUSSION
To improve the adhesion of PDMS to parylene, SiC and
SiO2 are used as intermediate adhesion layers. The adhesion
of SiC to parylene was initially evaluated with a tape test.
The result of this test revealed strong adhesion between the
two layers under test, however, it must be noted that such a
test is not ideal for investigating thin ceramic layers, as
peeling them at 180-degree angle may induce damage in the
thin layer. In addition, the tape test can only give a
qualitative evaluation of the adhesion strength. The peel test
can give a more quantitative evaluation, as the force at
which each layer is peeled from its substrate can be
recorded. Here, such a test showed that the adhesion for the
stack that includes ceramic layers was much stronger than
the other two variations. However, peeling of PDMS was not
possible for the sample with ceramic layers since the PDMS
layer was torn before peeling. Therefore, no quantitative
result was reported. The peel test also revealed no difference
in the adhesion between plasma treated and non-treated
samples. For a better and more complete comparison
between these samples, the effect of different power,
pressure, and oxygen flow parameters during the plasma
treatment process should also be investigated.
As mentioned before, it is expected that ceramic layers
can also act as a barrier layer against moisture, which is
important especially when the adhesion is not achieved. Our
so far soak tests revealed no difference in the performance of
the three different recipes of SiC, despite their different
characteristics in terms of pinhole density and stress, as
adhesion was never compromised throughout the soaking
period. To better understand the resulting barrier properties,
water vapor transmission rate tests could be employed for a
more thorough investigation.
V. CONCLUSION
In conclusion, in this paper we investigate the effect of
using thin ceramic layers to improve the adhesion of PDMS
to parylene for the encapsulation of implantable devices.
Results show that the adhesion of PDMS to parylene after
using intermediate SiO2 and SiC is so strong that no
delamination was seen after 60 days soak test at room
temperature. In comparison, PDMS-on-parylene and PDMS-
on-plasma treated parylene delaminated easily by hand
during the same test. This improvement was also confirmed
by tape and peel tests on dedicated test structures.
The proposed ceramic layers have a dual function, acting
at the same time as a barrier layers against water permeation.
In this paper three different SiC recipes were evaluated with
respect to the number of pinholes present in each layer. The
result shows that the non-silane starving mode with high
amount of hydrogen leads to less pinholes in the layer.
Future work will be focusing on further investigations of
the adhesion under the effect of applying bias voltage at
elevated temperatures, as we expect that such experiments
may reveal failure mechanisms which cannot be observed in
passive tests, as in [11]. Eventually it is our intension to use
the proposed process for the packaging of bare silicon chips
on active implants [12].
ACKNOWLEDGMENT
This research was supported by POSITION-II project
funded by the ECSEL JU under grant number Ecsel-
783132-Position-II-2017-IA.
REFERENCES
[1] V. Giagka and W. A. Serdijn, “Realizing flexible bioelectronic
medicines f or ac cessin g the peripheral nerves – technology
considerations,” Bioelectron. Med., vol. 4, no. 8, pp. 1–10, 2018.
[2] C. Hassler, T. Boretius, and T. Stieglitz, “Polymers for neural
implants,” J. Polymer Sci.-Part B: Polymer Phys., 49, pp. 18-33,
2011.
[3] I. R. Minev et al. "Electronic dura mater for long-term multimodal
neural interfaces," Science 347, no. 6218, pp. 159-163, 2015.
[4] N. de la Oliva, M. Mueller, T. Stieglitz, X. Navarro, and J. del Valle,
"On the use of Pa rylene C polymer as substrate for periph eral nerve
electrodes," Scientific reports 8, no. 1:5965, 2018.
[5] R. Caldwell, M. G. Street, R. Sharma, P. Takmakov, B. Baker, and L.
Rieth, “Characterization of Parylene-C degradation mechanisms: In
vitro reactive accelerated aging model compared to multiyear in vivo
implantation,” Biomaterials, 119731, 2019.
[6] A. Vanhoestenberghe, N. Donaldson, “Corrosion of s ilicon integrated
circuits and lifetime predictions in implantable electronic devices,” J.
Neural Eng. 10(3), 031002, 2013.
[7] P. Rezai, P. R. Selvaganap athy, G. R. Wohl. “Plas ma enhanced
bonding of polydimethylsiloxane with parylene and its
optimization,” J. Micromechanics and Microengineering 21(6),
065024, 2011.
[8] J. Ordonez, S. C. Boehler, M. Schuettler, and T. Stieglitz. “Silicone
rubber and thin-film polyimide for hybrid neural interfaces—A
MEMS-based adhesion promotion technique,” In 2013 6th
International IEEE/EMBS Conference on Neural Engineering (NER),
pp. 872-875. IEEE, 2013.
[9] J. M. Hsu, P. Tathireddy, L. Reith, A. R. Normann, F. Solzbacher,
“Characterization of a-SiCx: H thin films as an encapsulation silicon
based neural interf ace device s,” Thin solid films 516, no. 1, pp. 34-41,
2007.
[10] ASTM, D. 3359, Standard Test Methods for Measuring Adhesion by
Tape Test, (Method B–Cross-Cut Tape Test), Vol. 06.01,
2002. American Society for Testing Materials, West Conshohocken,
PA, 19428.
[11] K. Nanbakhsh, M. Kluba, B. Pahl, F. Bourgeois, R. Dekker, W.
Serdijn, and V. Giagka, “Effect of Sig nals on the Encapsulation
Performan ce of Parylene Coated Platinum Tracks for Active Medical
Implants,” in Proc. 41st Int. Conf. of the IEEE Engineering in
Medicine and Biology (EMBC) 2019, Berlin, Germany, pp. 3840-
3844, IEEE, Jul. 2019.
[12] V. Giagka, A. Demosthenous, and N. Donaldson, “Flexible active
electrode arrays with ASICs that fit inside the rat’s spinal canal,”
Biomed. Microdev., vol. 17, no. 6, pp. 106 – 118, Dec. 2015.
Figure 7. FTIR spectra for three different SiC deposition recipes.