Professional Documents
Culture Documents
Brggemann Nanotechnology 2011
Brggemann Nanotechnology 2011
net/publication/51141032
CITATIONS READS
87 579
6 authors, including:
Some of the authors of this publication are also working on these related projects:
Controlled Engineering of Oxide Surfaces for Bioelectronics Applications Using Organic Mixed Monolayers View project
All content following this page was uploaded by Dorothea Brüggemann on 13 August 2014.
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
(http://iopscience.iop.org/0957-4484/22/26/265104)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 134.226.252.160
The article was downloaded on 16/09/2011 at 15:38
E-mail: a.offenhaeusser@fz-juelich.de
Abstract
We present a new biocompatible nanostructured microelectrode array for extracellular signal
recording from electrogenic cells. Microfabrication techniques were combined with a
template-assisted approach using nanoporous aluminum oxide to develop gold nanopillar
electrodes. The nanopillars were approximately 300–400 nm high and had a diameter of 60 nm.
Thus, they yielded a higher surface area of the electrodes resulting in a decreased impedance
compared to planar electrodes. The interaction between the large-scale gold nanopillar arrays
and cardiac muscle cells (HL-1) was investigated via focused ion beam milling. In the resulting
cross-sections we observed a tight coupling between the HL-1 cells and the gold nanostructures.
However, the cell membranes did not bend into the cleft between adjacent nanopillars due to the
high pillar density. We performed extracellular potential recordings from HL-1 cells with the
nanostructured microelectrode arrays. The maximal amplitudes recorded with the nanopillar
electrodes were up to 100% higher than those recorded with planar gold electrodes. Increasing
the aspect ratio of the gold nanopillars and changing the geometrical layout can further enhance
the signal quality in the future.
(Some figures in this article are in colour only in the electronic version)
0957-4484/11/265104+07$33.00 1 © 2011 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 22 (2011) 265104 D Brüggemann et al
materials have already been implemented into microelectrode production of metal nanowires (Sander and Tan 2003, Wolfrum
technology to improve the signal quality. For example, the et al 2006, Schröper et al 2008, Zhou et al 2009), polymer
enhanced surface areas of porous structures such as spongy nanostructures (Xiao et al 2007), semiconducting nanotubes
platinum black or Ti3 N4 lead to a reduced electrode impedance (Dresselhaus et al 2003, Mitchell et al 2002), carbon nanotubes
(Pine 1980, Regehr et al 1989). Despite their good electronic (Maschmann et al 2006), and composite nanostructures (Lee
properties porous electrode materials also show problems et al 2005, Shingubara 2003). The formation, and resulting
during signal recording. They are often blocked with proteins geometry, of the nanopores was thoroughly characterized for
used as surface modification in cell culture. Furthermore, Al foils (Masuda and Fukuda 1995, Masuda et al 1997, 1998,
porous structures show mechanical instabilities since they Nielsch et al 2002) and aluminum films on silicon substrates
detach during cell cultivation or cleaning (Taketani and Baudry (Sulka and Stepniowski 2009). In our approach, we use
2006). These disadvantages led to the development of different the nanoporous templates to structure the metal electrodes of
microstructured tip electrodes to avoid layers of dead cells and microelectronic circuits on silicon substrates. Subsequently,
to enhance the electrode surface (Thiébaud et al 1997, Isik et al we investigate the signal recordings with the nanopillar-
2005, Heuschkel et al 2002, Nam et al 2006). The electrode modified electrodes from cardiac-like muscle cells.
modifications enabled signal recording and cell stimulation
from various tissue slices. For the detection of signals from sin- 2. Experimental details
gle cells in networks, a miniaturization of the microelectrodes
toward nanoelectrodes is now of great importance. Recently, 2.1. Fabrication of metal electrode arrays
a novel microelectrode modification method was presented for
P-doped silicon wafers with a (100) orientation (Si-Mat,
neural electrode engineering with gold nanoflakes (Kim et al
Landsberg, Germany) served as substrates for the production
2010). With these nanoflake-modified MEAs an impedance
of planar gold MEAs. They were oxidized in a diffusion
reduction was observed, and in vitro recording and stimulation
oven of type TS-6304 (Tempress Systems Ing, Heerde, NL)
of cultured hippocampal neurons was demonstrated. Besides
to obtain a top layer of 1500 nm of SiO2 . An adhesion layer of
the lowering of the electrode impedance, the nanoflakes also
10 nm of titanium was deposited onto the oxide in a sputter
improved the cell–chip coupling. Other approaches to modify
system ZH 620 (Emerald Technology Ventures AG, Zürich
MEAs for improved cell recordings include patterning with
Switzerland), followed by 200 nm of gold and a second layer
nanostructures such as vertically aligned carbon nanotubes
of 10 nm of titanium. For photolithographic structuring of
(CNTs) or CNT meshes (Wang et al 2006, Nguyen-Vu et al
the metal layers, the positive resist AZ 6612 (MicroChemicals
2006, Pancrazio 2008, Gabay et al 2007, Galvan-Garcia et al
GmbH Ulm, Germany) and a mask-aligner MA 6 (Süss
2007, Yu et al 2007, de Asis et al 2010), silicon nanowires
MicroTec AG, Garching, Germany) were used. Afterward,
(SiNWs) (Shalek et al 2010, Kim et al 2007) or metallic
the metal layer was etched via ion beam etching with argon
nanorods (Zhou et al 2009). Some techniques even aim at
in an Ionfab 300 Plus system (Oxford Instruments GmbH,
sticking the nanostructures through the membrane without
Wiesbaden, Germany) to obtain prestructured conductive
damaging the cell to achieve optimal coupling conditions at the
paths. A passivation layer of 300 nm of SiO2 , 200 nm of Si3 N4
interface (Shalek et al 2010, Kim et al 2007, McKnight et al
and 300 nm of SiO2 (ONO stack) was then deposited onto the
2004, Verma and Melosh 2010, Almquist and Melosh 2010).
wafer via plasma enhanced chemical vapor deposition (Surface
A tight coupling between the cell membrane and the
Technology Systems, STS, Newport, UK). In a second
recording device leads to an increase of the voltage at the
photolithographic step the ONO stack was structured using
cell–substrate interface so that small amplitude signals from the photoresist AZ 1518 (MicroChemicals). The electrode
biological systems can be detected (Fromherz et al 1991, Egert areas of the conductive paths were opened via reactive ion
et al 1998, Heuschkel et al 2002, Taketani and Baudry 2006, etching. SiO2 was removed using CF4 and CHF3 . For opening
Joye et al 2009). An interesting approach for improving the the Si3 N4 we employed CF4 , O2 and CHF3 . The top Ti
cell–chip coupling and thus enhancing the signal-to-noise ratio layer was also removed in this process. Finally, the gold
of neuro–electronic hybrid systems was recently presented by MEA wafers were coated with 800 nm of aluminum from an
Hai et al (2009a, 2009b). They reported on the fabrication of electron beam evaporation system BAK640 (Unaxis GmbH,
spine-shaped gold protrusions to enhance the electrode surface. Hanau, Germany) and were transferred out of the clean room
The adherence and electrical coupling of the neurons were to perform the nanostructuring process (see figure 1).
improved when growing on the gold spines. The cleft between
the neuronal cell membrane and the gold microstructures
2.2. Nanostructuring process
was reduced, and the neurons even engulfed the protrusions,
leading to field potentials that were significantly larger than A custom-built cell made out of polymethylmethacrylate was
the signals from flat electrodes. used for the anodization of the top aluminum layer. The wafers
A new approach for chip-based electrode modification and with the gold MEAs were cut into single chips with a size of
miniaturization is presented in this work. We report on the 11 × 11 mm2 to fit into this cell. The anodization bath consisted
fabrication of gold nanopillar arrays on MEAs using a top-layer of 0.3 M oxalic acid (Merck KGaA, Darmstadt, Germany). The
template of nanoporous aluminum oxide. Nanoporous alumina chip with the top aluminum film was fixed in the cell and served
templates can be produced via self-assembly (Masuda and as the anode. A platinum sheet of larger surface area than the
Fukuda 1995). This template material is widely used for the exposed alumina surface was used as a cathode. A potential
2
Nanotechnology 22 (2011) 265104 D Brüggemann et al
Figure 1. Fabrication of gold nanopillars on microelectrode arrays. 2.4. Impedance spectroscopy and recording of extracellular
1. Aluminum was deposited onto the MEAs via e-beam evaporation. signals
2. Anodization in 0.3 M oxalic acid yielded self-assembled
nanopores of Al2 O3 . Afterward, the barrier layer of the pore bottom Electrochemical analysis of the large-scale gold nanopillar
was removed in 5% H3 PO4 . 3. Electrochemical filling of the pores arrays was carried out via impedance spectroscopy in 0.15 M
was performed in a K[Au(CN)2 ] bath at 60 ◦ C for 120–130 s. 4. NaCl (Sigma). In a three-electrode arrangement with a
Finally, the Al2 O3 template was removed in 20% KOH to obtain platinum wire and a Ag/AgCl (3 M KCl) electrode we recorded
free-standing gold nanopillars. Their height was between 300 and
impedance spectra using the impedance analyzer Solartron
400 nm, and they were approximately 60 nm in diameter.
1260 (Solartron Analytical, UK). The amplitude of the applied
AC voltage was 10 mV.
At day in vitro 4 (DIV 4) signal recording from the HL-1
difference of 40 V was applied between the chip and the
cells was performed with an MEA amplifier system designed
counter electrode resulting in the formation of a nanoporous
at our institute (Krause et al 2000) and operated using the
aluminum oxide film. During the anodization the electrodes in
software MED 64 conductor™ (KF Technology srl, Rome,
the middle of the MEA chip as well as the outer bond pads
Italy). A chlorinated silver wire was used as a reference
were in direct contact with the aluminum layer. When the electrode. On average three different cultures of HL-1 cells
nanopores reached the gold electrodes an increasing current with one or two nanostructured and planar chips each were
indicated complete anodization. A low-resistance contact of analyzed for the respective electrode types. Thereby, the
the microelectrodes to the power supply was maintained even electrode diameters ranged from 8 to 20 μm. Action potentials
after aluminum oxide formation via the feed lines and the were recorded from 64 electrodes per chip and averaged
outer bond pads. Subsequently, the pores were widened in 5% afterward to obtain an average action potential per chip.
phosphoric acid (Merck KGaA) for 45 min, and the remaining
bottom barrier layer was removed to achieve better contact with
2.5. Microscopic techniques
the underlying gold electrodes.
After formation of the alumina template, gold was Live–dead staining was performed using an Axio Imager
electrochemically deposited in the nanopores connecting to the Z.1 Apotome (Carl Zeiss GmbH, Oberkochen, Germany).
3
Nanotechnology 22 (2011) 265104 D Brüggemann et al
Figure 3. FIB cross-section of an HL-1 cell at DIV 4 grown on a macroscopic gold nanopillar array. (a) Cross-section through a whole HL-1
cell on gold nanopillars. (b) Magnification of the cell–nanopillar interface.
4
Nanotechnology 22 (2011) 265104 D Brüggemann et al
Figure 4. (a) SEM image of a microelectrode modified with gold nanopillars. The nanopillars had an average height of 300–400 nm and were
approximately 60 nm in diameter. (b) Live–dead staining of HL-1 cells on nanostructured gold MEAs with calcein and EthD-1. The red
arrows indicate where the nanostructured electrodes are situated beneath the cell layer. (c), (d) Optical light microscopy images of cultured
HL-1 cells on gold microelectrode arrays with a diameter of 15 μm at DIV 4. The electrodes with gold nanopillars in figure (c) appear black,
whereas the planar electrodes in (d) are yellow.
enhance the surface area of the microscopic metal electrodes. presented process yielded stable nanostructures that could be
The resulting nanostructured MEAs can be seen in figure 4(a). used for repetitive measurements, although mechanical forces
In the center of the electrode a dense growth of nanopillars was due to intensive cleaning can induce bending of the nanopillars.
observed. However, at the outer edge of the planar electrode no In preparation for the signal recording, the nanostructured
pillars were grown. This process-related effect occurred for all and planar microelectrodes were coated with fibronectin which
electrodes but it is worth noting that for the smallest electrodes yielded confluent layers of HL-1 cells, as can be seen in
(8 μm diameter) the result was that only about 20–30% figures 4(c) and (d). In the light microscopy image in
of the total electrode surface was covered with nanopillars. figure 4(c) the electrodes appear brown/black due to the gold
We believe that this effect is caused by the inhomogeneous nanopillars at the surface whereas the planar gold electrodes
coverage of the passivated microelectrodes. The Al film only shown in figure 4(d) are yellow. Live–dead staining with
adheres tightly to the inner part of the electrodes because of calcein and EthD-1 of the on-chip HL-1 cultures showed
the height difference near the surrounding passivation layer. vital cell growth throughout all measurements, as can be seen
Hence, during the anodization, the nanopores can only reach exemplarily in figure 4(b).
the bottom of the electrode in the center of the circle so that Signal recordings from HL-1 cells were performed at DIV
only in this area can the pores be filled with gold. 4 with planar and nanopillar-modified MEAs. To reduce the
In general, the nanostructuring process was more effect of cell culture quality on the comparison of the different
challenging on microelectrodes than on planar gold surfaces electrode types, we always plated cells on planar and pillar
due to the topography of the passivation layer, the electrode electrodes in the same cell culture run. Action potentials were
areas, and the feed lines. The different materials and the recorded repeatedly from the different HL-1 cultures for all
topography of the chips also led to unequal etch rates, e.g. at the different electrode arrays. Figure 5 shows an exemplary
corners of the passivation. Furthermore, the anodization sequence of action potentials and a single action potential
parameters had to be adjusted according to the underlying recorded with a nanostructured gold electrode. The number
microelectronic circuits to avoid a delamination of the of channels showing action potentials depended on the density
aluminum film due to inappropriately high voltages. The of the HL-1 cell culture. For very dense cell cultures all 64
5
Nanotechnology 22 (2011) 265104 D Brüggemann et al
Figure 5. Extracellular recordings of HL-1 cells at DIV 4 using a nanostructured gold electrode (20 μm diameter). The signal has been
inverted after recording. (a) Train of action potentials. (b) Single action potential starting with a slight dip due to the initial depolarization of
the cell and followed by the spike caused by the inward sodium current into the cell.
Table 1. The table summarizes the comparison of the maximal cell coupling, i.e. the seal resistance between the cell
peak-to-peak amplitudes of action potentials and the noise levels membrane and the electrode, determines the extracellular
recorded with nanostructaured and planar MEAs. About 10
potential at the junction and the interface impedance of the
successive action potentials were averaged per electrode.
electrode material can strongly influence the signal quality.
Maximal Furthermore, the input impedance of the amplifier system and
Electrode Electrode peak-to-peak Rms noise
the leak resistance of the feed lines on the chip affect signal
diameter (μm) type amplitude (mV) (μV)
recordings. In this work, we addressed the improvement
10 Planar 0.40 ± 0.01 7. 0 of the interface impedance by enhancing the surface area
10 Pillar 0.83 ± 0.02 6. 6
20 Planar 0.92 ± 0.03 6. 5 due to the nanostructuring process. Although the impedance
20 Pillar 1.50 ± 0.02 6. 7 reduction was evident, as shown in figure 2, this effect is
less pronounced compared to other electrode modification
methods. For example, impedance reduction factors between
15 and 23 were reported using various CNT coated electrodes
electrodes exhibited cell activity. The resulting signals varied (Chen et al 2010, Nguyen-Vu et al 2006, Keefer et al 2008). In
strongly with regard to their shape, amplitude, and frequency. principle, similar or even higher impedance reduction values
Variability in the chip fabrication as well as differences in cell can also be obtained using nanostructured metal electrodes
fitness, adhesion, and the exact location of a cell with respect (Zhou et al 2009, Kim et al 2010). Nevertheless, the data in
to the recording electrode influenced the signal transfer to the table 1 show that even for electrodes as small as 10 μm the
chip. We assume that the variability in the recorded signals is impedance reduction has almost no effect on the noise level of
dominated by these features. our recorded traces. This suggests that in our setup we are not
To characterize the amplitude response for each electrode operating at the Johnson noise limit of the impedance at the
type we used the highest average peak-to-peak action potential, electrode–electrolyte interface.
which should represent signals under the best cell culture
The observed improvement in maximal amplitudes of the
conditions (see table 1).
cell signals is either due to an increased junction resistance or
The corresponding rms noise of these traces was
calculated in the intervals from 300 to 200 ms before the peak a better ratio of leak to electrode impedance. In principle,
of each action potential. Similar noise levels were recorded in the aspect of impedance reduction is advantageous for very
the literature involving nanostructured electrode systems such small electrodes, where the signal quality is mainly limited
as microelectrodes modified with gold nanoflakes (3.5 μV, by the high impedance. However, to efficiently transfer our
Kim et al 2010), CNT electrodes (7 μV, Gabay et al 2007) process to the nanostructuring of sub-10 μm electrodes it has
or carbon nanofiber electrodes (17 μV noise, Yu et al 2007). to be adjusted in such a way that the pillar coverage extends
However, a direct comparison is difficult since the noise is also to the outer rim of the electrodes. For larger electrode
influenced by the electrode size and the electronic setup for diameters, the impedance of the planar surfaces is already
different experiments. In our experiments the structure of quite low. In the case where the impedance of the electrode
the electrode had only a minor effect on the recorded noise. is already significantly lower than the leak impedance of the
Nevertheless, the maximal amplitudes were increased for the feed lines and the load impedance of the electronic system, a
nanostructured MEAs by up to 100%. further reduction of the electrode impedance due to electrode
The quality of the signals recorded via extracellular modifications will have no significant effect on the signal
on-chip measurements depends on different parameters. The amplitude.
6
Nanotechnology 22 (2011) 265104 D Brüggemann et al