Content uploaded by Vasiliki Giagka
Author content
All content in this area was uploaded by Vasiliki Giagka on Jan 23, 2020
Content may be subject to copyright.
WAFER-SCALE GRAPHENE-BASED SOFT ELECTRODE ARRAY WITH
OPTOGENETIC COMPATIBILITY
Andrada I. Velea,
1,2
, Sten Vollebregt
2
, Gandhika K. Wardhana
1,2
and Vasiliki Giagka
1,3
1
Bioelectronics Section, Department of Microelectronics, TU Delft, Delft, THE NETHERLANDS,
2
Electronic Components, Technology and Materials Section, Department of Microelectronics, TU
Delft, Delft, THE NETHERLANDS and
3
Technologies for Bioelectronics Group, Department of System Integration and Interconnection
Technologies, Fraunhofer IZM, Berlin, GERMANY
ABSTRACT
This paper reports on the characterization of a
microfabricated wafer-scale, graphene-based, soft implant
for spinal cord applications. Graphene is used because of
its high transparency and good conductivity, making it
suitable for optogenetic applications. Moreover it has a
high mechanical strength and is potentially biocompatible.
The implant consists of multi-layered chemical vapor
deposited graphene, in the form of electrodes and tracks,
encapsulated between 2 layers of silicone. Methods such
as Raman spectroscopy, optical transmittance, and
electrical measurements combined with bending tests and
in-vitro experiments, using phosphate-buffered saline
(PBS) solution, were employed to characterize the device.
The results have shown high bendability and no critical
damage of the graphene after immersing the device in
PBS solution up to 7 days. To the authors’ best
knowledge, this is the first work that presents a soft and
fully scalable optogenetics-compatible graphene-based
spinal cord electrode array.
KEYWORDS
Wafer-scale implant, graphene, silicone, optogenetic
compatibility.
INTRODUCTION
Spinal cord stimulation studies are an important
research topic, mostly for pain relief but also for restoring
locomotion after spinal cord injuries. Apart from clinical
trials, more exploratory research is mostly conducted in
rodents. However, the availability of suitable
neurotechnologies tailored to small animals is limited. For
instance, the spinal cord shows significant mechanical
mismatch with rigid neural implants. Even flexible
implants encapsulated in stiffer polymers such as
polyimide can cause significant damage and compression
to the spinal cord [1].
Therefore, effort has been put into the design and
fabrication of reliable soft spinal cord implants, based on
silicone elastomers.
Existing spinal cord microelectrode arrays (MEAs)
feature opaque electrodes [1, 2] which are used to
electrically activate the tissue or record evoked electrical
activity. Alternatively, tissue activation by means of light,
i.e. optogenetics, is promising to increase the resolution of
activation. This is due to the fact that it activates the tissue
in a more type-specific manner. Only neurons that are
genetically modified to respond to light will be activated,
as long as they are in the vicinity of the light source and
the intensity of it is enough to trigger action potentials [3].
To quantify the effect of this new technology there is
great need to monitor the electrical response of cells at the
site of optogenetic stimulation. To this end, MEAs with
transparent electrodes are a necessary tool. In such a
system, the transparency will guarantee that light, coming
from an external source, can pass through the electrode
site, activating the target location. Moreover, the
conductivity of MEAs will allow for in situ recording of
the evoked electrical response. This temporally and
spatially concurrent recording during stimulation is not
possible with conventional electrodes, as these systems
require both the activating and recorded signals to be in
the electrical domain. Transparent parylene-based
graphene electrode arrays, with metal tracks, have
previously been developed to allow for electrophysiology,
in-vivo imaging and optogenetics in the brain [4].
However, parylene, with a Young’s modulus of ~2 GPa
[5] is still very stiff for the spinal cord.
This work proposes a wafer-scale fabrication process
for a silicone-sandwiched graphene-based implant. The
materials used ensure the required transparency and
conductivity but also provide the necessary softness for
interacting with the spinal cord. The design of the
proposed implants ensures that any metal layers, if
present, only appear at areas which are not under
mechanical strain as they have been proven to have poor
adhesion to graphene, thus causing delamination [6].
MATERIALS AND METHODS
The proposed solution was implemented by
employing a scalable, 2-mask microfabrication process,
inspired by the “flex-to-rigid” (F2R) approach [7],
together with a transfer-free CVD process for graphene
growth [8] as shown in Fig.1.
Figure 1: Microfabrication process flow. SiO2 deposition
(I). Graphene growth on the molybdenum (Mo) catalyst
(II). First layer of silicone spin-coated and cured on top
of the structure (III). Release of the complete area using
DRIE for Si and wet etching steps to remove the oxide
and Mo layers (IV, V).
As the developed prototypes are intended to be used
for in-vivo experiments in rats, the MEAs had to be
designed accordingly. A total of 12 electrodes were
distributed as in Fig. 2, each of them being connected to
individual test pads. This number allows for a large
coverage of the spinal cord but also fits into the limited
space available for implantation in rats [9].
For the microfabrication process, 6 µm of plasma-
enhanced chemical vapor deposited (PECVD) oxide were
deposited on both sides of a double-side polished (DSP)
silicon (Si) wafer. Then, on the frontside, 50 nm of
molybdenum (Mo) were sputtered and patterned. This
serves as a catalyst for graphene growth. Next, graphene
was grown using a CVD process [8] to ensure uniformity
of the layer, reproducibility and scalability of the
manufacturing.
Before encapsulating the structures in silicone,
Raman spectroscopy and two-point measurements were
employed to evaluate the graphene layer. Next, 50 µ m of
Sylgard 184 silicone were spin coated on top of the
structures and cured for 1 h at 90 °C. Finally, for the
backside of the wafer, a deep reactive ion etching (DRIE)
process, was employed only over the areas containing the
implants, thus suspending them completely (Fig. 3). This
was followed by wet etching of the oxide and Mo layers.
Before spin-coating the final encapsulation layer on the
backside of the wafer, the structure was again evaluated
by means of Raman spectroscopy and 2-point
measurements to ensure that no critical damage was
present after the microfabrication process. Moreover,
optical transmittance measurements were performed to
approximate the number of graphene layers and the
transparency of the released graphene-on-silicone
structure. In addition, in-vitro experiments were
conducted using PBS saline solution for 24 h and 7 days
to quantify how much graphene degrades over time.
The structures were then cut out from the wafers and
bending tests were performed by placing the implant
around rods with different diameters to emulate handling
and deformation expected at the site of implantation. For
the actual implant, the final layer of silicone encapsulation
can be applied and openings for the electrodes and test
pads can be created before removing the samples from the
Si wafer. The robust process achieves high yield and the
authors are employing a similar microfabrication process
as a basis for the first graphene-based active implants
(details can be found in [10]).
RESULTS AND DISCUSSION
Raman Spectroscopy
After developing the graphene MEAs on the Si wafer,
Raman spectroscopy was employed to determine the
presence of graphene and to qualitatively evaluate it. The
results in Fig. 3 demonstrate that a multi-layer graphene is
present, judging from I
2D
/I
G
< 1 [11]. To monitor
graphene, Raman spectroscopy was employed after
suspending the implants and removing the oxide and Mo
layers (the structures are shown in Fig. 4). Also, after
conducting the in-vitro experiments, in saline solution, to
emulate the biological environment, the graphene layer
had to be investigated to evaluate if the graphene is
damaged by the PBS solution.
The results in Fig. 5, illustrate that there was no
critical damage present on the graphene layer after the
microfabrication process. After immersing the structures
in PBS, the D peak (indicating the number of defects
present) slightly increases. This could be caused by
surface contamination of the graphene layer with different
particles from the solution. The additional peaks seen on
the Raman spectra represent the influence of the silicone
Raman signal over the measurement.
Figure 4: Photograph of the 8 implants fabricated on a
single 10 cm wafer.
Figure 2: Masks used to develop and suspend the
graphene-based passive implants.
Figure 3: Raman spectroscopy evaluation (using a 633
nm laser) after the CVD process. The I2D/IG ratio, indicate
that a multi-layer graphene has been grown [11].
Two-Point Measurements
Apart from Raman spectroscopy, the electrical
properties of the graphene were investigated. To this end,
two-point measurements were employed after graphene
growth (Fig. 6), after suspension of the passive implants
(Fig. 6 and Table 1), as well as after conducting the in-
vitro experiments (Table 1). All measurements were
performed over a graphene line of 70 µ m in width and 1
mm in length.
A significant difference in resistance can be seen
from Fig. 6. This originates from the fact that after
graphene growth, the measured value is the resistance of
both Mo and graphene, whereas for the suspended
implants, only the graphene layer was probed as the Mo
catalyst had been removed. From this it can be concluded
that the resistance of Mo is significantly lower than that of
graphene. Moreover, for the suspended implants, the
probes were landed directly on graphene which was
resting on a 50 µm soft silicone substrate, without any
metal interface between the measured layer and the two-
point probes. Therefore, the accuracy of the measurement
was significantly reduced due to the less controllable
contact resistance. Ideally, the graphene track should be
measured using a four-point measurement setup to reduce
the contact resistance and, on top of that, a metal layer is
needed for landing the probes and not damaging the
graphene layer. Such designs were not included in the
current layout but will be included in future work.
Having the implants suspended, in-vitro experiments
using PBS solution were conducted for periods of 24 h
and 7 days respectively. After each period of time, the
samples were electrically evaluated, and the results
synthesized in Table 1 demonstrate that graphene does not
deteriorate in-vitro. The observed differences likely
originate from the inaccuracy of the two-point
measurements.
Bending Tests
As the implantation site for the described structures is
subjected to different types and degrees of movements,
mechanical evaluation is also needed. To this end,
bending tests using metal rods with diameters ranging
from 8 mm down to 3 mm, followed by resistive
measurements were employed. In Fig. 7, a passive
implant on a 3 mm metal rod is shown while Table 2
synthesizes the electrical measurements performed after
bending the samples on different rods. The measured
resistance is ~2 MΩ, which originates from the large
length-to-width (L/W) ratio of the implant and the high
contact resistance. Since the probing setup was not
tailored for such small prototypes, longer tracks had to be
measured such that the probes could properly be landed
on the sample under test. However, the results in Table 2,
all in the same range, demonstrate that the passive
structures can be bent down to 3 mm without damage.
Figure 5: Raman Spectroscopy results DRIE process and
after immersing the structure in PBS for 24 h and 7 days,
respectively. The grey area represents the silicone Raman
signal.
Table 1. Two-point electrical measurement results
Sample type Voltage
range
Measured
resistance
After DRIE -5 V to 5 V 200 kΩ
After DRIE and
24 h in PBS -5 V to 5 V 160 kΩ
After DRIE and
7 days in PBS -5 V to 5 V 130 kΩ
Figure 7: Implant released from the wafer and placed on
top of a 3 mm bending rod.
Figure 6: Two-point electrical evaluation of graphene
after the CVD process. The results show a resistance
value of ~250 Ω for graphene on Mo, and ~200 kΩ for
graphene on silicone. The contact resistances from the
probes also contribute to the overall resistance.
Optical Transmittance
As these MEAs are intended to be used in
optogenetics, optical transmittance measurements had to
be conducted to evaluate the degree of transparency for
the final implant. From the optical transmittance result in
Fig. 8 it was determined that the implant consists of ~11
graphene layers and has a transmittance of 72-77% [12].
CONCLUSIONS
This paper describes the methods used to evaluate
graphene-based passive implants for spinal cord
applications. Graphene quality (using Raman), electrical
and mechanical characterizations of the final prototypes
were performed. Although there were no dedicated
structures to evaluate the electrical conductivity of the
implants, it has been proven that graphene does not
deteriorate and moreover, still conducts after immersing
the implants in PBS up to 7 days, as well as after bending
them over different rod diameters, down to 3 mm.
This paper demonstrates the potential of soft,
graphene-based passive implants, to enable unique,
currently not available, spinal cord tissue monitoring
during optogenetic activation, which is paramount to map
the neuronal activation at this level.
ACKNOWLEDGMENTS
We acknowledge the staff of Else Kooi Laboratory
(EKL) and the Bioelectronics group from Delft University
of Technology for their support throughout the project.
REFERENCES
[1] I. R. Minev et al., “Electronic dura mater for long-
term multimodal neural interfaces”, Science, vol. 347, no.
6218, pp. 159-163, 2015.
[2] 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, 2015.
[3] K. L. Montgomery et al., “Beyond the brain:
Optogenetic control in the spinal cord and peripheral
nervous system”, Science Transl. Med., vol. 8, no.
337rv5, pp. 1-12, 2016.
[4] D. W. Park et al., “Fabrication and utility of a
transparent graphene neural electrode array for
electrophysiology, in vivo imaging and optogenetics”,
Nat. Protoc., vol. 11, no. 11, pp. 2201-2222, 2016.
[5] K. Scholten and E. Meng, “Materials for
microfabricated implantable devices: a review”, Lab on a
Chip, vol. 15, no. 22, pp, 4256-4272, 2015.
[6] J. Robinson et al., “Contacting graphene”, Appl. Phys.
Lett., vol 98, no. 5, 2011.
[7] B. Mimoun et al., “Flex-to-rigid (F2R): A generic
platform for the fabrication and assembly of flexible
sensors for minimally invasive instruments”, IEEE
Sensors, vol. 13, no. 10, pp. 3873-3882, 2013.
[8] S.Vollebregt et al., “A transfer-free wafer-scale CVD
graphene fabrication process for MEMS/NEMS sensors”,
in Proc. IEEE MEMS 2016, Sanghai, China, 2016, pp. 17-
20.
[9] V. Giagka et al., “Flexible platinum electrode arrays
for epidural spinal cord stimulation in paralyzed rats: An
in vivo and in vitro evaluation”, in Proc. 3
rd
Annual Conf.
IFESSUKI 2012, Birmingham, UK, 2012, pp. 52-53.
[10] A. I. Velea, S. Vollebregt, T. Hosman, A. Pak and V.
Giagka “Towards a microfabricated flexible graphene-
based active implant for tissue monitoring during
optogenetic spinal cord stimulation”, in Proc. IEEE
NMDC 2019, Stockholm, Sweden, 2019.
[11] Z. Jian et al., “Irradiation effects of graphene and thin
layer graphite induced by swift heavy ions”, Chinese
Phys. B, vol. 24, no. 8, 2015.
[12] K. F. Mak et al., “Measurement of the optical
conductivity of graphene”, Phys. Rev. Lett, vol. 101, no.
19, pp. 2-5, 2008.
CONTACT
*A. I. Velea; A.I.Velea@student.tudelft.nl;
*S. Vollebregt; S.Vollebregt@tudelft.nl;
*G. K. Wardhana; G.K.Wardhana@tudelft.nl;
*V. Giagka; V.Giagka@tudelft.nl.
Table 2. Bending test results (using two-point
electrical measurements)
Sample type Voltage
range
Measured
resistance
No bending -5 V to 5 V 1.2 MΩ
Bending (8 mm rod) -5 V to 5 V 1.8 MΩ
Bending (5 mm rod) -5 V to 5 V 2.5 MΩ
Bending (3 mm rod) -5 V to 5 V 2.4 MΩ
Note: In case the application requires, this resistance
could be lowered (Fig. 5) by leaving the Mo on some
parts of the graphene tracks. Using Mo instead of
another metal has the advantage that the graphene-Mo
bonds are strong, originating from the CVD process.
Figure 8: Optical transmittance results for the graphene-
based implant. On the X-axis, the light wavelength range
(visible spectrum). On the Y-axis, the percentage of the
light transmitted through the sample.