Pore-suspending lipid bilayers on porous alumina investigated by electrical impedance spectroscopy

Article abstract of DOI:10.1021/jp030762r

Nonordered and ordered porous alumina substrates with pore diameters of 20 and 50 nm, respectively, were utilized to immobilize lipid membranes spanning the pores of the porous material. The substrates were characterized by means of interferometry and electrical impedance spectroscopy. For impedance data reduction, an equivalent circuit representing the electrical behavior of porous alumina was developed on the basis of the parallel layer model. It turned out that the electrical parameters of the as prepared alumina substrates prevent a sensitive monitoring of the formation of immobilized lipid membranes. Thus, we established a technique to modify the substrates with respect to their electrical properties, leading to a significantly increased capacitance of porous alumina, which allowed for a sensitive detection of pore-spanning lipid bilayers by impedance spectroscopy. Two different membrane preparation techniques based on vesicle spreading were investigated. First, negatively charged giant liposomes were spread onto the porous alumina surface under an applied dc voltage of +100 mV. Second, large unilamellar vesicles containing lipids bearing a thiol anchor were used to chemisorb on gold functionalized porous alumina substrates and subsequently rupture to form planar pore-spanning membranes. For both techniques, impedance spectra were obtained, which indicate the formation of lipid bilayers on top of the porous alumina substrates.

Full text of DOI:10.1021/jp030762r

J. Phys. Chem. B 2003, 107, 11245-11254
11245
Pore-Suspending Lipid Bilayers on Porous Alumina Investigated by Electrical Impedance
Spectroscopy
Janine Drexler and Claudia Steinem*
Institut fu¨r Analytische Chemie, Chemo- und Biosensorik, UniVersita¨t Regensburg, 93040 Regensburg, Germany
ReceiVed: June 17, 2003
Nonordered and ordered porous alumina substrates with pore diameters of 20 and 50 nm, respectively, were
utilized to immobilize lipid membranes spanning the pores of the porous material. The substrates were
characterized by means of interferometry and electrical impedance spectroscopy. For impedance data reduction,
an equivalent circuit representing the electrical behavior of porous alumina was developed on the basis of the
parallel layer model. It turned out that the electrical parameters of the as prepared alumina substrates prevent
a sensitive monitoring of the formation of immobilized lipid membranes. Thus, we established a technique
to modify the substrates with respect to their electrical properties, leading to a significantly increased capacitance
of porous alumina, which allowed for a sensitive detection of pore-spanning lipid bilayers by impedance
spectroscopy. Two different membrane preparation techniques based on vesicle spreading were investigated.
First, negatively charged giant liposomes were spread onto the porous alumina surface under an applied dc
voltage of +100 mV. Second, large unilamellar vesicles containing lipids bearing a thiol anchor were used
to chemisorb on gold functionalized porous alumina substrates and subsequently rupture to form planar pore-
spanning membranes. For both techniques, impedance spectra were obtained, which indicate the formation
of lipid bilayers on top of the porous alumina substrates.
Introduction
To investigate ion transport across suspended membranes,
electrically insulating lipid bilayers are required. By means of
electrical measurements, the insulating properties of such
membranes can be obtained. In this paper, we describe the
investigation and modification of porous alumina substrates and
pore-suspending lipid membranes by means of electrical imped-
ance spectroscopy. Impedance spectroscopy provides compre-
hensive information about the dominant electrical processes of
the system under investigation. Impedance data are generally
discussed in terms of equivalent circuits mostly consisting of
parallel and series combinations of ohmic resistances and
capacitances. For porous alumina, various models have been
proposed to represent its electrical impedance behavior. Early
studies of Hoar and Wood12 described an equivalent circuit
consisting of a parallel circuit of a capacitance and an ohmic
resistance resembling the Al2O3 pore columns, which is in
parallel with a parallel circuit of a capacitance and a resistance
and in series with an ohmic resistance representing the Al2O3
barrier layer at the pore bottoms and the pores filled with
electrolyte, respectively. Hoar and Wood argued that the
capacitance and ohmic conductance of the Al2O3 pore columns
are so small that they cannot be detected in the observed
frequency range, and hence, the equivalent circuit is reduced
by these two elements. For calculating area-related specific
parameters, they used the geometric area of the porous alumina
exposed to the aqueous phase. The model of Hoar and Wood
was adopted by Mansfeld and Kendig13 for the investigation of
anodizing layers. Hitzig et al.14 proposed a model considering
two separate oxide phases, an Al2O3 barrier layer at the bottom
of the pores and a pore-containing Al2O3 layer. Each oxide layer
is characterized by a parallel connection of an ohmic resistance
and a capacitance. They anticipated that the electrical properties
of the two oxide layers can be distinguished and that both can
be treated as separate quasi-homogeneous oxide phases. Vereeck-
en and co-workers15,16 described an equivalent circuit composed
Solid supported membranes (SSMs) as well as black lipid
membranes (BLMs) are established model systems to investigate
basic physical properties of lipid bilayers and membrane-
confined proteins. Particularly, SSMs have been studied over
recent years as a promising tool for constructing biosensors.
Owing to their high long term stability and their amenability to
sensitive surface analysis tools such as scanning probe micros-
copy, ATR infrared spectroscopy, the quartz crystal microbal-
ance technique, and surface plasmon resonance spectroscopy,
SSMs enable one to rationally design biosensor devices based
on sophisticated micromachined chip technology.1-8 However,
it turned out that the suitability of SSMs is restricted by their
close surface proximity (typically 0.2-2 nm) that limits or even
prevents incorporation of large membrane-spanning proteins and
the establishment of chemical and electrochemical transmem-
brane gradients. These problems can be circumvented using
black lipid membranes first established by Mu¨ller and Rudin.9
BLMs allow in principle the incorporation of functional
transmembrane proteins; they lack, however, long term stability
and are thus not suited for biosensor applications.
To combine the merits of both membrane systems, SSMs and
BLMs, we followed a new strategy based on lipid membranes
suspending pores of a porous matrix. In this system, membranes
anchored to the upper surface of the porous material resemble
SSMs, while pore suspending regions can be viewed as nano-
BLMs. Recently, we have described a procedure that allows in
principle the preparation of pore-suspending bilayers on porous
alumina surfaces by a vesicle spreading technique. By means
of scanning force microscopy, we were, for the first time, able
to visualize these membranes and characterize their mechanical
stability.10,11
* To whom correspondence should be addressed. Telephone: +49 941
943 4547/4548. Fax: +49 941 943 4491. E-mail: claudia.steinem@
chemie.uni-regensburg.de.
10.1021/jp030762r CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/16/2003

11246 J. Phys. Chem. B, Vol. 107, No. 40, 2003
Drexler and Steinem
of a constant phase element (cpe) instead of a capacitance and
an ohmic resistance connected in series. In the case of ideally
polarizable electrodes, the presence of a cpe is discussed in
terms of surface roughness,17,18 specific anion adsorption, effects
of the solvent, and electrode potential.19,20 De Levie21 and
Scheider22 addressed the dispersion phenomenon in terms of
infinitely long, branched ladder-networks of capacitances and
resistances. Different degrees of branching lead to different
values of the index n.15 Vereecken and co-workers15,16 concluded
that information about pore size, volume, and thickness cannot
be obtained directly from electrical impedance spectroscopy;
only the barrier layer at the bottom of the pores is measured.
The main problem is that the active area of the porous alumina
surface that was used for calculating area-related specific
parameters was not explicitly given.
Figure 1. Schematic drawing of the experimental setup used for
impedance analysis of porous alumina.
To obtain lipid membrane specific parameters that can be
compared with literature values, the actual measured area of
the lipid bilayer is, however, absolutely required. Our objective
was to elucidate whether only the area of the pore suspending
part of the lipid bilayer is measured or whether the entire lipid
membrane covering the top of the porous alumina substrate
needs to be taken into account. In this study, we developed an
electrical equivalent circuit based on a simple parallel layer
model well describing our impedance and interferometric results
of porous alumina and those of others. Moreover, the model
addresses the question of the electrically active area, enabling
us to calculate specific parameters of the immobilized lipid
bilayers. The obtained electrical parameters of the porous
alumina layers indicated that a sensitive detection of lipid
bilayers immobilized on the porous alumina surface is not
feasible. Thus, we established a technique first described by
Nielsch et al.23 to thin out the barrier layer at the bottom of the
pores, allowing us to sensitively detect lipid membranes covering
the porous substrate. Insulating lipid membranes on the porous
alumina substrates were achieved by two different strategies
based on vesicle fusion, and electrical parameters of the resultant
lipid membranes were obtained by means of equivalent circuit
modeling.
Impedance spectra were recorded with the impedance analyzer
from Solartron Instruments (SI 1260, Farnborough, U.K.). The
frequency was varied between 0.1 and 106 Hz, while the applied
voltage amplitude was set to 30 mV (rms). A two electrode
configuration was used, in which the porous alumina substrate,
placed in a Teflon chamber sealed by an O-ring, served as
working electrode, while a platinized platinum wire in the
electrolyte solution served as counter electrode. Electrical contact
of the aluminum was accomplished by a slightly larger steel
disk contacting the aluminum from the back-side (Figure 1).
The geometric, electrolyte-exposed area of the porous alumina
sample was 0.3 and 0.8 cm2, respectively. All measurements
were performed in 0.1 M Na2SO4 or 10 mM Bis-Tris, 10 mM
Na2SO4, pH 5.5. For data representation and evaluation, the
absolute value of the impedance |Z(ω)| and the phase difference
between voltage and current æ(ω) were used. Data analysis was
performed using different equivalent circuits as outlined below
using the program ZView2.
Interference Reflectance Spectra. Interference reflectance
spectra of porous alumina were recorded by using an Ocean
Optics S 1000 spectrometer (Dunedin, FL) fitted with a
bifurcated fiber optic probe. A tungsten light source was focused
onto the center of the porous alumina surface with a spot size
of approximately 1-2 mm. Spectra were recorded in the
wavelength range of 400-900 nm. Illumination and detection
of light were performed along the surface normal. Data
acquisition and evaluation were achieved with a self-written
program in LabView 6.0. Extraction of the effective optical
thickness from the interference pattern was accomplished by
Fourier analysis and curve fitting of the reflectance spectra
considering eq 2:
Materials and Methods
Materials. Aluminum substrates (thickness 0.5 mm, purity
99.999%) and platinum wires (thickness 1 mm, purity 99.95%)
were purchased from Goodfellow (Goodfellow Cambridge
Limited, U.K.). 1-Palmitoyl-2-oleoyl-sn-glycero-3-[phospho-
rac-(1-glycerol)] (POPG), 1,2-dipalmitoyl-sn-glycero-3-phos-
phothioethanol (DPPTE), and 1,2-dioleoyl-sn-glycero-3-phos-
phocholine (DOPC) were purchased from Avanti Polar Lipids
(Alabaster, AL). Cholesterol was obtained from Sigma, and
soybean asolectin, from Fluka. Bis-Tris was purchased from
Sigma. Orthophosphoric acid (85%), sulfuric acid (95-97%),
oxalic acid, and chromium(VI) oxide dihydrate were from
Merck. The water used was ion-exchanged and Millipore-filtered
(Millipore, France).
mλ ) 2neffd
(2)
where m is the fringe order, λ is the wavelength, neff is the
effective refractive index, and d is the thickness of the layer.
Preparation of Porous Alumina Substrates. High purity
aluminum foils were first annealed for 3 h at 500 °C under a
nitrogen atmosphere to remove mechanical stresses and recrys-
tallize the sample to increase grain boundaries. To smooth the
surface, the aluminum substrates (2 × 2 cm2) were electro-
polished four times in a sulfuric/phosphoric acid/water (2:2:1)
mixture at 70 °C for 40 s and thoroughly rinsed with water
directly after removal of the acidic solution. Nonordered pores
were obtained by a one-step anodization process in 0.3 M oxalic
acid at 2 °C and 40 V for several hours. To control the
anodization process, the current was continuously recorded. The
Electrical Impedance Spectroscopy (EIS). Porous alumina
substrates and lipid membrane covered porous alumina were
investigated using electrical impedance spectroscopy. The
complex electrical impedance Z(ω) is determined by the ratio
between the applied voltage U(ω) and the measured current
response I(ω) of the sample as a function of frequency (eq 1):
U(ω)
I(ω)
Z(ω) )
) |Z(ω)| exp[iæ(ω)]
(1)

Pore-Suspending Lipid Bilayers on Porous Alumina
J. Phys. Chem. B, Vol. 107, No. 40, 2003 11247
porous substrates display pores with pore diameters of (20 (
5) nm with a surface porosity of (16 ( 2)% obtained from
scanning electron microscopy after gold coating. Hexagonally
ordered pores were obtained by a two-step anodization process
according to Go¨sele and co-workers.23-25 First, aluminum was
anodized in 0.3 M oxalic acid at 2 °C and 40 V for several
hours. Second, the anodically etched porous alumina layer was
removed by incubating the sample in a stirred acidic chromium-
(VI) oxide solution at 70 °C for several hours followed by a
second anodization step conducted under the same conditions
as those for the first one. Due to prestructuring of the surface
during the first anodization process, pores that are formed in
the second step are hexagonally ordered with pore sizes of
(50 ( 5) nm and a surface porosity of (50 ( 10)%. Pore
depth depends on etching time and is reported to be about
1-2 µm/h.23-26
To decrease the thickness of the barrier layer at the bottom
of the pores, an exponentially decreasing dc voltage starting at
40 V and stopping at 50 mV with a time constant of τ ) 40
min was applied to the porous alumina substrate in 0.3 M oxalic
acid at 2 °C. Afterward, the porous alumina samples were
immediately rinsed with water.
Figure 2. (A) Interference reflectance spectrum of nonordered porous
alumina etched in 0.3 M oxalic acid at T ) 2 °C and U ) 40 V for
3 h. (B) Effective optical thickness vs anodization time. The solid line
is the result of a linear regression with a slope of 6.4 µm/h.
Gold coating of the upper surface of the porous alumina
substrates was achieved by gold sputtering (25 nm) using a
sputter coater with a thickness control unit (Cressington sputter
coater 108 auto, Cressington mtm20, Elektronen-Optik-Service
GmbH, Dortmund, Germany). Surfaces were cleaned before
lipid membrane preparation using a plasma cleaner (Harrick
Scientific Corporation, New York, NY).
observable vesicles exhibited diameters between 1 and 30 µm.
Vesicle spreading was achieved by applying a voltage of +100
mV to the alumina substrate relative to a platinum wire for
several hours. Afterward, the surface was rinsed several times
with water to remove intact vesicles from the solution and
surface.
Preparation of Planar Barrier Layers. Planar barrier layers
were prepared according to two different procedures. For both
methods, aluminum was first electropolished as described above.
In the first method, the anodization protocol was started as
described above but was stopped immediately after the current
minimum had been reached. The obtained planar barrier layer
is supposed to have the same thickness as that of the barrier
layer at the pore bottoms of porous alumina. In the second
method, aluminum was anodized at room temperature in 0.3 M
boric acid, while the current was limited to 0.1 A. The current
decreases exponentially to zero in about 60-120 s, denoting
the end of anodization. The resulting thickness of the planar
barrier layer is 1.2-1.7 nm/V.27 After anodization, the samples
were immediately rinsed with water.
(ii) 1,2-Dipalmitoyl-sn-glycero-3-phosphothioethanol (DPPTE)
and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) were
mixed in appropriate proportions from stock solutions in
chloroform, dried on the bottom of glass test tubes with a stream
of nitrogen, desiccated under vacuum for 3 h, and rehydrated
by the addition of outgassed 10 mM Bis-Tris, 10 mM Na2SO4,
pH 5.5 at 50 °C for 30 min. The lipid suspension was vortexed
several times, and large unilamellar vesicles were prepared by
the extrusion method (Miniextruder Liposofast, Avestin, Ottawa,
Ontario, Canada) using a polycarbonate membrane of 400 nm
pore size. Bilayer preparation was achieved by incubating the
vesicle suspension (1 mg/mL) with the gold coated porous
alumina substrate overnight at 50 °C. Afterward, the sample
was rinsed with buffer solution several times in order to remove
vesicles attached to the lipid bilayer and from solution.
Preparation of Lipid Bilayers on Porous Alumina Sub-
strates. Lipid bilayers were prepared by spreading and fusing
unilamellar vesicles onto porous alumina. We followed two
different strategies: (i) adhesion and spreading of negatively
charged giant unilamellar vesicles, applying a positive bias
potential, and (ii) self-assembly of large unilamellar vesicles
containing lipids with a thiol anchor onto gold covered porous
alumina substrates.
Results
Interference Reflectance Analysis of Porous Alumina.
Porous alumina exhibits a Fabry-Perot fringe pattern when
illuminated with white light (Figure 2). A Fabry-Perot fringe
pattern is created by multiple reflections of illuminated white
light on the air-porous alumina and porous alumina-aluminum
interfaces. Interference reflectance spectra were monitored for
porous alumina substrates in air obtained after different anod-
ization times, and the effective optical thicknesses 2neffd were
obtained by Fourier analysis and curve fitting of the reflectance
spectra (Figure 2). The effective optical thickness increases
linearly with increasing anodization time. It is assumed that the
effective refractive index neff remains constant during the etching
procedure, while the thickness d of the porous layer increases
continuously. Using the effective medium approximation ac-
cording to Bruggeman developed for cylindrical pores (eq 3),
(i) Giant unilamellar vesicles were prepared from a lipid
mixture composed of 70 wt % asolectin, 25 wt % POPG,
and 5 wt % cholesterol. Lipids were mixed in appropriate
proportions from stock solutions in chloroform and dried at the
bottom of glass test tubes under a stream of nitrogen and in a
vacuum for 3 h. The lipid film was rehydrated by the addition
of water at room temperature for several hours, resulting in a
lipid concentration of 0.1-0.2 mg/mL. Giant unilamellar
vesicles are formed due to electrostatic repulsion between the
negatively charged lipid bilayers, and they were stable for
several weeks.28 To control the formation of giant unilamellar
vesicles, 1 mol % â-Bodipy was added to the lipid films, and
vesicles were observed by fluorescence microscopy. Resulting

11248 J. Phys. Chem. B, Vol. 107, No. 40, 2003
Drexler and Steinem
SCHEME 1: Schematic Drawing of the Alumina Substrates and Terminology: (A) Porous Alumina Substrate; (B)
Planar Barrier Layer
an effective refractive index neff of the porous alumina layer
can be calculated:29
n2air - n2
eff p +
n2air + n2eff
n2Alox - n2
eff(1 - p) ) 0
n2Alox + n2eff
(3)
The refractive index of air is nair ) 1.003, and that of aluminum
oxide is nAlox ) 1.77 (λ ) 550 nm). The porosity p for
nonordered, as-etched pores is p ) 0.16, which was determined
from scanning electron microscopy images assuming ideal
cylindrical pores. Taking these values into account, the effective
refractive index of the porous alumina layer is calculated to be
neff ) 1.63. From the obtained slope (Figure 2), a linear pore
growth rate of 2.0 µm/h is then calculated, which agrees very
well with those reported by others.23-26
Impedance Analysis of Porous Alumina. A prerequisite for
the evaluation of impedance spectra of lipid membrane covered
porous alumina is a comprehensive understanding of the
impedance behavior of porous alumina itself. Various electrical
models have been discussed in the literature to model the
electrical properties of porous alumina.12-16 To clear up whether
the channel array (Scheme 1A) of the porous alumina contributes
to the overall impedance, porous aluminas with different layer
thicknesses were prepared and impedance spectra were taken.
Independent of the anodization time, which results in an
increased thickness of the channel array (see Figure 2B), the
same impedance spectra (Figure 3) were obtained, providing
clear evidence that the thickness of the channel array does not
contribute to the overall impedance of porous alumina, particu-
larly the frequency dependent part given by the capacitance.
To model this electrical behavior of porous alumina by an
equivalent circuit, the two different layers, the barrier layer at
the bottom of the pores and the channel array composed of
Al2O3 and electrolyte, are treated separately. The channel array
can be ascribed by a parallel layer model,30 in which Al2O3
and electrolyte are stacked across the electrode, which is
composed of aluminum and the barrier layer. For this parallel
layer model, the complex conductivity ψj (eq 4)
Figure 3. (A and B) Impedance spectra of nonordered porous alumina
in 100 mM Na2SO4. The result of fitting the parameters of equivalent
circuit 1a is given as a solid line. Cb ) 0.20 µF/cm2; Rb > 107 Ω cm2.
(C) Specific capacitances Cb of porous alumina obtained after different
anodization times.
with σj being the conductivity and ꢀj being the permittivity of
phase j, follows a linear mixing rule of the two layers (eq 5):
ψeff ) øAloxψAlox + øelψel
(5)
ø
Alox and øel are the volume fractions of Al2O3 and electrolyte,
ψj ) σj + iωꢀj
(4)
respectively.

Pore-Suspending Lipid Bilayers on Porous Alumina
J. Phys. Chem. B, Vol. 107, No. 40, 2003 11249
SCHEME 2: Equivalent Circuits Used throughout This Study To Model the Impedance Data
This treatment results in an equivalent circuit composed of
two parallel RC-elements, which transforms into one RC-element
with a conductivity σeff (eq 6),
The result that only the barrier layer is measured by
impedance analysis and that the channel array does not
contribute to the overall impedance was corroborated by
investigating planar barrier layers (Scheme 1B) with the same
thickness as that of the barrier layer at the bottom of the pores
of a porous substrate and with the same geometric area.16 The
obtained impedance spectra were almost identical with the same
electrical parameters obtained for planar barrier layers and for
porous layers (data not shown). This result confirms the
hypothesis of solely measuring the barrier layer of the porous
alumina.
σeff ) øAloxσAlox + øelσel
and a permittivity ꢀeff (eq 7),
ꢀeff ) øAloxꢀAlox + øelꢀel
(6)
(7)
As σel . σAlox and ꢀel . ꢀAlox, σeff, ꢀeff and thus Geff and Ceff
are determined by the properties of the electrolyte. Hence, the
channel array can be represented by a small ohmic resistance
Rel that merges with the bulk electrolyte resistance. This is
consistent with the observation that, independent of the thickness
of the channel array, the same impedance spectrum was
obtained. The barrier layer can be modeled by a capacitance
Cb representing the dielectric behavior of Al2O3 in parallel with
an ohmic resistance Rb representing the ion conductance across
the barrier layer. Equivalent circuit 1a depicted in Scheme 2
combines the electrical elements of the channel array, the barrier
layer, and the electrolyte. Fitting the parameters of the equivalent
circuit to the impedance spectra results in very good agreement
(Figure 3A and B).
As the electrical properties of the porous alumina remain
constant during the anodization process and as the electrical
properties are solely determined by the barrier layer, it can be
concluded that the thickness of the barrier layer is constant and
independent of the anodization time. Even more important is
the result that the measured active area of porous alumina is
only determined by the geometric area of the barrier layer.
Taking this area into account, which is determined by the area
surrounded by the O-ring, a specific capacitance of Cb ) (0.20
( 0.03) µF/cm2 and a resistance Rb > 107 Ω cm2 can be
calculated independent of anodization time (Figure 3C).
As the specific capacitance of the barrier layer Cb is much
smaller than that of a solid supported lipid bilayer, which is
typically in the range 0.6-1.0 µF/cm2,2,7,31-36 and the barrier
layer resistance Rb is larger than 107 Ω cm2, a lipid bilayer on
top of the porous alumina substrate could hardly be detected.
To overcome this problem, we have established a method to
modify the electrical behavior of the barrier layer by decreasing
the thickness of the barrier layer and, as a consequence,
increasing its specific capacitance (Scheme 1A). An exponen-
tially decaying voltage is applied to the as etched porous
alumina, starting from 40 V and eventually stopping at 50 mV
with a time constant of τ ) 40 min. The process was followed
by impedance spectroscopy. As expected for an effective
thinning of the barrier layer, the impedance of the porous
alumina substrate changes considerably dependent on the applied
final voltage (Figure 4). Fitting the parameters of equivalent
circuit 1a depicted in Scheme 2 to the data indicates that the
capacitance of the barrier layer increases, while its resistance
decreases. However, it turned out that with decreasing voltage
the agreement between impedance data and fit became less
perfect. Deviations from an ideal impedance behavior are often
treated by using the constant phase element (cpe) instead of an
ideal capacitance. The origin of a cpe-behavior is discussed
controversially. It is generally believed that it originates from

11250 J. Phys. Chem. B, Vol. 107, No. 40, 2003
Drexler and Steinem
Figure 4. Impedance spectra before (b) and after (O) the application
of an exponentially decaying voltage down to 1 V during the etching
procedure. The parameters of equivalent circuit 1 were fitted to the
data with the following results: (b) Ab ) 0.20 µF/cm2, n ) 0.99, Rb
> 107 Ω cm2; (O) Ab ) 4.6 µF/cm2, n ) 0.97, Rb ) 13 kΩ cm2. The
solid lines are the results of the fitting procedure.
-1
Figure 5. (A) Plot of Ab vs the applied voltage U. The slope was
determined to be m ) 0.126 cm2 µF-1 V-1. (B) Plot of n vs the applied
voltage.
electrical parameters always reach their equilibrium values, all
porous alumina substrates were soaked in aqueous solution
overnight. During the incubation in aqueous solution, a slight
increase in barrier layer thickness occurs, resulting in a slightly
decreased Ab-value and an increased Rb-value. The final values
for Ab were between 4 and 6 µF/cm2, and they were always
constant for one prepared alumina sample. The final values for
Rb varied between 3 × 104 and 5 × 106 Ω cm2. The thinning
process not only results in a decreased barrier layer thickness
but also widens the pores up to 2-5 nm, leading to an increased
surface porosity by ∼2-5%.
a distribution in the current density along the electrode surfaces
as a result of surface inhomogeneity. This is inferred from the
analogy with the behavior of porous electrodes, which has been
extensively discussed by de Levie.21 To account for the observed
deviation of the thinned porous alumina substrates in our case,
we introduced the cpe-element instead of the capacitance Cb.
In contrast to the fitting results obtained by using equivalent
circuit 1a, those achieved by using equivalent circuit 1 exhibit
good agreement between data and fit. The result of the fitting
procedure is shown in Figure 4.
Fusion of Giant Liposomes on Nonordered Porous Sub-
strates. The preparation of lipid bilayers on porous surfaces
starting from giant unilamellar vesicles allows, in principle, the
formation of single lipid bilayers suspending the pores of the
porous substrate, as these vesicles are excluded from the pores
due to their size, which exceeds the pore size by at least a factor
of 50. In a first set of experiments, porous alumina was coated
with a thin gold layer providing the substrate for chemisorption
of 3-mercaptoethylamine. At pH 5.5, this chemisorbed mono-
layer is positively charged so that negatively charged giant
liposomes adhere to the surface and might spread at high
adhesion forces to form planar bilayers.35 Impedance spectra
were taken before and after incubation of the surface with giant
liposomes. After incubation, no change in the impedance of the
system was obtained, indicating that the vesicles did not form
an insulating surface-attached planar bilayer. In a different set
of experiments, noncoated porous alumina was used and
incubated with negatively charged giant liposomes while
applying a voltage of +100 mV for 16 h. Again, impedance
spectra were taken before and after the incubation process. After
incubation, the impedance spectrum was significantly altered,
indicating the formation of an insulating lipid layer (Figure 6).
Control experiments, in which giant liposomes were incubated
with porous alumina without applying a potential, and those in
which porous alumina was incubated in an aqueous solution
while applying a voltage of +100 mV did not lead to an altered
impedance response after the incubation process. Hence, we
conclude that an insulating lipid bilayer has been formed on
the porous alumina surface under these conditions. To evaluate
the measured impedance spectra, a model providing a math-
ematical expression for the electrical response of the system is
required. In addition to the porous alumina, whose impedance
As dendritelike pore structures at the barrier layer emerge
during the thinning process, only the remaining thin barrier layer
at the bottom of the small channel area (Scheme 1A) contributes
to the impedance behavior of the porous alumina substrate, in
accordance with the theoretical treatment of the porous alumina
based on the parallel layer model described above. Again, the
channel array of the porous alumina and the small channel array
emerging during the thinning process at the barrier layer are
not resolved by impedance spectroscopy, as they are indistin-
guishable from the bulk electrolyte. Thus, the active area of
the thinned porous alumina substrate is again the geometric area
of the barrier layer. The area-specific values for Ab together
with the exponent n as a function of the applied voltage are
given in Figure 5. n decreases nonlinearly from 0.99 in the range
10-40 V to 0.96-0.90 at 100-50 mV, while Ab-1 depends in
a linear fashion on the applied voltage. As a first approximation,
the values for Ab can be treated as a capacitance Cb, as n is still
close to one, indicating a capacitive behavior of the barrier layer.
Assuming a plate condenser for the barrier layer with a dielectric
constant of ꢀAlox ) 9.0, one can calculate the thickness of the
barrier layer at each applied voltage. The slope of the plot in
Figure 5A then translates into a barrier layer thinning rate of 1
nm/V. This agrees very well with the reported value for barrier
layer thicknesses of 1.3 nm/V and corroborates our model and
approach.23
When the as prepared porous alumina substrates are stored
in an aqueous solution, the electrical parameters are altered. Ab
only slightly decreases, while Rb increases considerably with
time. After 3 h of storing the porous alumina substrates in 0.1
M Na2SO4, no further changes in the electrical parameters were
detected and equilibrium was reached. To ensure that the

Pore-Suspending Lipid Bilayers on Porous Alumina
J. Phys. Chem. B, Vol. 107, No. 40, 2003 11251
Figure 6. Impedance spectra of a nonordered porous alumina substrate
before (]) and after (b) incubation with giant liposomes while applying
a voltage of +100 mV for 16 h. (]) Equivalent circuit 1 was used to
model the impedance spectra of the porous alumina substrate. The solid
line is the result of the fitting procedure with the following values: Ab
) 4.2 µF/cm2, n ) 0.94, Rb ) 71 kΩ cm2. (b) Equivalent circuit 2
was used for data evaluation. The parameters of the porous alumina
were kept constant during the fitting procedure. The result of the fitting
routine is given as dotted line with Cm ) 2.4 µF/cm2. Rm was not defined
in the impedance spectra.
Figure 7. Impedance spectra of a nonordered gold covered porous
alumina substrate before (]) and after (b) incubation with DPPTE/
DOPC (72:28) unilamellar vesicles (400 nm nominal diameter)
overnight at 50 °C. (]) Equivalent circuit 1 was used to model the
impedance spectra of the porous alumina substrate. The result of the
fitting procedure is given as a solid line with the following parameters:
Ab ) 5.9 µF/cm2, n ) 0.89, Rb ) 207 kΩ cm2. (b) Equivalent circuit
2 was used for data evaluation. The parameters of the porous substrate
were kept constant during the fitting procedure. The result of the fitting
routine is shown as dotted line leading to a membrane capacitance of
Cm ) 0.99 µF/cm2 and a resistance of Rm ) 4 × 105 Ω cm2.
response is modeled by equivalent circuit 1 depicted in Scheme
2, additional electrical elements have to be added to the
equivalent circuit to account for the insulating lipid layer. The
standard and most simple equivalent circuit of a lipid layer is
composed of a parallel connection of a resistance Rm and
capacitance Cm. Combining the equivalent circuit of the lipid
bilayer with that representing porous alumina results in equiva-
lent circuit 2 in Scheme 2. Fitting this simple equivalent circuit
to the data reveals only limited agreement between the data and
the model (Figure 6) with a specific capacitance value of 2.4
µF/cm2. The membrane resistance was not defined in the
impedance spectrum. During the fitting routine, the parameters
of the porous alumina substrate were kept constant, as it is
assumed that they do not change during the incubation process,
which was experimentally verified by the control experiments.
Fusion of DPPTE-Containing Vesicles on Nonordered
Porous Substrates. In a second set of experiments, we used
vesicles that are known to chemisorb onto gold surfaces,
resulting in almost defect free planar lipid bilayers. In general,
vesicles were composed of 1,2-dipalmitoyl-sn-glycero-3-phos-
phothioethanol (DPPTE) and 1,2-dioleoyl-sn-glycero-3-phos-
phocholine (DOPC). To find the lipid mixture providing the
best insulating planar lipid bilayers by vesicle spreading and
fusion, vesicles composed of different mole fractions of DPPTE/
DOPC were incubated with planar gold surfaces and investigated
by impedance spectroscopy. It turned out that at least 50 mol %
DPPTE is required for the formation of insulating planar lipid
bilayers on gold, indicated by specific capacitances ) (1.0 (
0.2) µF/cm2 and resistances > 107 Ω cm2. This optimized
system was transferred to porous alumina. Nonordered porous
alumina substrates were coated with a thin gold layer on top of
the surface, ensuring a selective functionalization of the upper
surface. These alumina substrates were incubated with a vesicle
suspension composed of DPPTE/DOPC (72:28). To exclude
vesicles from the pores, the nominal vesicle size was chosen to
be 400 nm, which exceeds the pore size by a factor of 20. The
incubation process resulted in a significant change in the
impedance spectrum compared to the spectra of gold covered
neat porous alumina taken prior the incubation process (Figure
7). As a control, impedance spectra of porous substrates
incubated in pure buffer and in DOPC vesicle suspensions at
50 °C were taken. Only very slight changes in the impedance
spectra were observed after incubation of the substrates in pure
buffer and DOPC vesicle suspensions, respectively. These results
indicate that the obtained changes in the impedance spectra can
be attributed to the formation of an insulating lipid bilayer on
the porous matrix. Fitting the parameters of equivalent circuit
2 to the obtained data, while keeping the parameters of the
porous alumina substrate constant results in moderate agreement
between the data and the fit. In this case, the obtained specific
capacitance of 0.99 µF/cm2 agrees with the expected value of
(1.0 ( 0.2) µF/cm2 for a solid supported membrane obtained
from DPPTE/DOPC vesicle spreading and fusion on gold. This
capacitance value is therefore indicative of a high surface
coverage with bilayers with respect to planar gold electrodes
as a reference.
Fusion of DPPTE-Containing Vesicles on Ordered Porous
Substrates. The experiments described above clearly demon-
strate that insulating lipid bilayers on porous alumina can be
obtained. To gain a larger porosity and a less rough surface,
we next used ordered pores instead of nonordered pores. The
surface porosity of the ordered porous material is about 50%
compared to that of nonordered porous alumina, which only
exhibits a porosity of around 20%. An impedance spectrum of
an ordered porous alumina substrate after the thinning process
is depicted in Figure 8. The parameters of equivalent circuit 1
shown in Scheme 2 are fitted to the data, which results in an
Ab-value of 7.7 µF/cm2, slightly larger than those obtained for
nonordered pores after the thinning process. Also the value for
n, which was determined to be 0.86, is smaller than that of
nonordered pores. Lipid vesicles composed of DPPTE/DOPC
(72:28) obtained according to the extrusion method with
polycarbonate membranes of 400 nm were incubated with the

11252 J. Phys. Chem. B, Vol. 107, No. 40, 2003
Drexler and Steinem
area needs to be known to calculate area independent specific
electrical parameters of a lipid bilayer. From the impedance
spectra of planar barrier and porous alumina layers recorded at
different anodization times, it is clearly evident that only the
barrier layer at the bottom of the pores is detected by impedance
analysis within the observed frequency regime. The same
conclusion was drawn by De Laet et al.16 To model our
impedance data, we used an equivalent circuit, which combines
the electrical behavior of the channel array, based on a parallel
layer model,30 and the barrier layer of the porous alumina.
Moreover, in this model, the active measured area, which is
the geometric area of the porous alumina surface, is unambigu-
ously defined. It should be mentioned that our results and those
of De Laet et al.16 are in contrast to those of Hitzig et al.14
They proposed that both the channel array and the barrier layer
are detected by electrical impedance spectroscopy and that the
two oxide layers are distinguishable. Hitzig et al.14 investigated
porous alumina substrates with different layer thicknesses and
found that the capacitance of the barrier layer remains constant,
while the resistance of the channel array increases and its
capacitance decreases with increasing layer thickness.
Figure 8. Impedance spectra of a gold covered ordered porous alumina
substrate before (]) and after (b) incubation with DPPTE/DOPC (72:
28) unilamellar vesicles (400 nm nominal diameter) overnight at 50
°C. (]) Equivalent circuit 1 was used to model the impedance spectra
of the porous alumina substrate. The solid line is the result of the fitting
routine with the following values: Ab ) 7.7 µF/cm2, n ) 0.86, Rb )
146 kΩ cm2. (b) Equivalent circuit 2 was used for data evaluation.
The values for the porous alumina substrate were kept constant during
the fitting routine. The dotted line is the result of the fitting procedure
with Cm ) 0.9 µF/cm2. Rm was only poorly defined in the impedance
spectra.
Fitting the parameters of equivalent circuit 1a (Scheme 2)
results in a very good agreement between data and fit and leads
to specific barrier layer capacitance values of Cb ) (0.20 (
0.03) µF/cm2 and specific resistances Rb > 107 Ω cm2 for the
porous alumina substrates. These values are in good agreement
with those reported by Hoar and Wood,12 who found barrier
layer capacitances ) 0.47 µF/cm2 and resistances > 107 Ω cm2.
They also used the solution exposed geometric area given by
the sealing ring, so that their calculated specific parameters
can be compared to our values. In general, in most publications
the area that is used to calculate specific parameters is not
explicitly defined, so that a comparison is difficult. Assuming
that the solution exposed geometric area is used, Mansfeld and
Kendig13 reported a capacitance value for the barrier layer of
Cb ) 0.5 µF/cm2. De Laet et al.16 applied a constant phase
element instead of a capacitance to represent the barrier layer,
obtaining Ab ) 0.94-1.03 µF/cm2 and n ) 0.94-0.95.
To sensitively detect a lipid bilayer with a specific capacitance
of roughly 1 µF/cm2 on top of the porous substrate, capacitance
values of the porous alumina substrates must be considerably
increased. A thinning procedure forming dendrite like structures
at the barrier layer, first described by Nielsch et al.,23 was
applied, and for the first time impedance spectra were taken at
different stages of this process. It turned out that with decreasing
voltage the agreement between data and fit became less perfect
by using equivalent circuit 1a. This might be a result of the
formation of dendritelike structures at the barrier layer. To
account for this deviation from an ideal capacitive behavior,
we replaced the capacitance Cb in the equivalent circuit by a
constant phase element (cpe).37 Fitting the parameters of this
modified equivalent circuit to the data led to a very good
agreement. Interestingly, the value for n decreases with decreas-
ing voltage. Though the origin of the cpe-behavior has not been
well understood and is discussed controversially in the literature,
there is some evidence that electrode roughness contributes to
this behavior.17-20 The decrease in n might reflect the micro-
structure of the barrier layer and could be used as a semiquan-
titative indicator for its increasing inhomogeneity with decreas-
ing voltage.
gold covered porous alumina substrate, and impedance spectra
were again taken (Figure 8). Even without data evaluation it is
evident that incubation of porous alumina with the vesicle
suspension results in a significant alteration of the impedance
spectrum compared to that obtained prior to the incubation
process. Control experiments done at ordered porous substrates
incubated at 50 °C in buffer solution and in DOPC vesicle
suspensions, respectively, again lead to only minor changes in
the impedance spectra. Equivalent circuit 2 was fitted to the
obtained impedance data, leading to a specific capacitance of
0.9 µF/cm2. However, as can be seen from the fitting results,
the agreement between data and fit is only limited. Impedance
spectra were taken over a period of more than 30 h without
any detectable alteration.
Discussion
To sensitively investigate the electrical properties of lipid
membranes immobilized onto porous alumina substrates, first
the material itself needs to be thoroughly characterized and its
electrical behavior modified. Porous alumina substrates were
characterized by means of interferometry and electrical imped-
ance spectroscopy. Interferometric measurements of porous
alumina samples prepared at different anodization times revealed
a linear dependence of the effective optical thickness on
anodization time. This result is consistent with that observed
by Vereecken and co-workers.16 By means of ellipsometry, they
observed a linear increase in porous alumina film thickness
dependent on anodization time. This result validates our setup
of preparing porous alumina substrates. To garner information
about the electrical behavior of porous alumina, electrical
impedance spectroscopy proved to be very useful.12-16,37 Though
several equivalent circuits are available in the literature to model
the impedance response of porous alumina, none of them
addresses the question of the active electrically measured area.
However, to evaluate the electrical parameters of a lipid bilayer
attached to the top of the porous alumina substrate, the active
However, as n typically ranges between 0.99 and 0.90 and
is, thus, still close to one, the parameter Ab can be regarded as
a capacitance as a first approximation. The successful thinning
of the barrier layer can be deduced from the fact that the

Pore-Suspending Lipid Bilayers on Porous Alumina
J. Phys. Chem. B, Vol. 107, No. 40, 2003 11253
parameter Ab increases considerably with decreasing voltage.
A large Ab-value is a prerequisite for the sensitive electrochemi-
cal detection of a lipid bilayer on top of the porous alumina
substrate.
the cases of lipid bilayers obtained from spreading and fusion
of charged vesicles on a charged surface with typical specific
capacitances of 0.9-1.1 µF/cm2. Capacitances obtained from
lipid bilayers achieved by fusing giant liposomes onto porous
alumina are 2.4 µF/cm2 and are thus at least a factor of 2 larger
than those obtained by vesicle fusion onto charged surfaces on
gold surfaces. The larger capacitance indicates an incomplete
surface coverage. The overall measured capacitance is the sum
of the capacitances arising from the lipid bilayer with a dielectric
constant of the hydrocarbon chains of 2.0-2.2 and the defects
occupied by water with a dielectric constant of 80. According
to this picture, one might find it useful to calculate the surface
coverage Θ with bilayers by assuming a parallel combination
of two capacitances reflecting the defect free bilayers, Cbi (0.7
µF/cm2), and the defects, Cd (4 µF/cm2), represented mainly
by the barrier layer (eq 8):
For preparation of a lipid bilayer on the porous alumina
substrate, vesicle spreading is a rather useful approach, as size
exclusion together with a functionalization of the upper surface
of the porous alumina substrate prevents the formation of lipid
bilayers within the pores, as observed by Bourdillon and co-
workers.38,39 Moreover, vesicles allow for the insertion of
membrane proteins, and the resultant planar lipid bilayers are
solvent free. To induce vesicle adhesion and spreading via
electrostatic interactions, the surface needs to be charged. This
can be achieved by adsorbing charged molecules on the surface
or by applying a voltage between a reference electrode and the
surface, resulting in an accumulation of charges at the inter-
face.28,32 In the case of negatively charged giant liposomes, the
formation of insulating planar bilayers was not observed on
porous alumina surfaces, which were functionalized with a
positively charged mercaptoethylamine. However, applying a
voltage of +100 mV led to a significant change in the
impedance spectra compared to those of bare porous alumina
substrates. For data analysis, an equivalent circuit has been
applied representing the porous alumina substrate and the lipid
bilayer. The electrical parameters of the porous alumina substrate
were determined before incubation with the vesicle suspension
and kept constant during the fitting routine after bilayer
formation. The assumption that the electrical parameters of
porous alumina do not change during the incubation process
seems to be justified, as no significant changes in the impedance
behavior were observed in any of the control experiments.
Cm - Cd
Cbi - Cd
Θ )
(8)
Cm is the measured capacitance. The surface coverage for giant
liposomes then amounts to roughly 50%.
Feder et al.42 found that giant vesicles retain their original
topology while spreading, but they flatten and exhibit fractal-
like structures for moderately strong adhesion. If the attraction
is too strong, vesicles burst; if it is too weak, no adhesion is
observed due to the Helfrich steric repulsion. It is conceivable
that not all giant liposomes adhere that strongly to the surface
on which they spread and fuse to form planar lipid bilayers,
which would lead to structural inhomogeneities and thus to
defects within the lipid bilayer structure.
The simplest and most well-known equivalent circuit repre-
senting a lipid bilayer is a parallel connection of a capacitance
and a resistance.7 Using this model, Wiegand et al.7 reported
membrane capacitances of 0.74-1.12 µF/cm2 and resistances
between 103 and 105 Ω cm2 for lipid bilayers immobilized on
semiconductor and gold surfaces. In particular, the membrane
resistance strongly depends on pH, ionic strength, temperature,
and the concentration of charged lipids and cholesterol, respec-
tively. In general, lipid bilayers prepared from vesicle spreading
and fusion on a hydrophilic surface exhibit slightly larger
capacitances than those obtained from organic solvent31 or
vesicle fusion on a hydrophobic monolayer.35,40 For example,
lipid bilayers composed of N,N-dimethyl-N,N-dioctadecyl-
ammonium bromide immobilized on 3-mercaptopropionic acid
modified gold electrodes typically exhibit membrane capaci-
tances of about 1.0-1.1 µF/cm2,2,36 while fusion of a DMPG/
DMPE mixture on a positively charged mercaptoethylamine
monolayer results in a specific capacitance of 0.9 µF/cm2.35 The
larger capacitances are discussed in terms of incomplete vesicle
fusion, leading to defect structures within the bilayer. The
increased number of defects and thus water within the bilayer
leads to a larger overall dielectric constant. Fromherz et al.41
had overcome this problem by touching one single giant vesicle
to the open gate of a field-effect transistor. The lipid bilayer
kept its continuous shape without rupturing, and it adhered onto
the cationic polylysine film due to electrostatic interaction. They
obtained a capacitance of 0.7 µF/cm2, since fusion of several
vesicles is not required for bilayer formation. Purrucker et al.33
were able to obtain lipid bilayers from vesicle fusion onto
smooth silicon-silicon dioxide surfaces with low membrane
capacitances of 0.7 µF/cm2. Here, the low roughness of the
surface presumably favors fewer defects within the bilayer. In
our case, the obtained capacitances can be best compared with
In a different approach starting from large unilamellar vesicles
with a nominal diameter of 400 nm, the upper surface was
coated with a thin gold layer, while lipid vesicles were used
carrying thiol anchors to allow for chemisorption of the vesicles
on the gold surface. By using 400 nm vesicles, the tendency of
the vesicles to diffuse into the pores is low due to size exclusion,
while there is a strong affinity toward the upper gold surface.
To find an optimized system, vesicles composed of DPPTE
and DOPC varying in the amount of DPPTE were incubated
with planar gold electrodes and impedance spectra were taken.
It turned out that at least 50 mol % DPPTE is required to
obtain almost defect free planar lipid bilayers with capacitance
values of (1.0 ( 0.2) µF/cm2. The optimized system was trans-
ferred to nonordered porous alumina substrates with a surface
porosity of around 20%. Data analysis revealed a capacitance
of 0.99 µF/cm2 using an equivalent circuit consisting of one
parallel RC-circuit representing the electrical behavior of a lipid
membrane. According to eq 8, the capacitance value translates
into a surface coverage of Θ > 90%. The obtained capacitance
on the porous surface is the same as that achieved on planar
gold electrodes as a reference for high coverage. To increase
the surface porosity and to obtain a more defined and less rough
surface, we also used hexagonally ordered porous alumina
substrates. Even though the porosity of ordered porous alumina
substrates is significantly higher (50%) than that of nonordered
pores (20%), a similar value of 0.9 µF/cm2 for the membrane
capacitance was obtained assuming equivalent circuit 2 shown
in Scheme 2. The fact that the same capacitance was obtained
for a lipid bilayer on nonordered and ordered poressthough
the porosity of these two substrates is about 30% differents
verifies further that not only are the pore columns of the channel
array covered by a lipid bilayer but also the pores are membrane
suspended.

11254 J. Phys. Chem. B, Vol. 107, No. 40, 2003
Drexler and Steinem
(16) De Laet, J.; Terryn, H.; Vereecken, J. Electrochim. Acta 1996, 41,
1155.
Conclusion
We established a procedure to modify porous alumina, so
that insulating lipid bilayers immobilized on the porous matrix
can be investigated by means of impedance spectroscopy. By
vesicle spreading techniques, insulating lipid membranes were
formed on porous alumina, and the membranes were character-
ized by their electrical parameters. These membranes spanning
the pores of the porous substrate will open up the avenue to
nanocompartments separated by a semipermeable lipid mem-
brane serving as containers for biochemical reactions. Experi-
ments are underway to demonstrate that these model membranes
are suited for the insertion of fully functional transmembrane
proteins and the establishment of electrochemical and chemical
gradients.
(17) Diao, P.; Jiang, D.; Cui, X.; Gu, D.; Tong, R.; Zhong, B. J.
Electroanal. Chem. 1999, 464, 61.
(18) Kerner, Z.; Pajkossy, T. J. Electroanal. Chem. 1998, 448, 139.
(19) Sadkowsky, A. J. Electroanal. Chem. 2000, 481, 222.
(20) Sadkowsky, A.; Motheo, A. J.; Neves, R. S. J. Electroanal. Chem.
1998, 455, 107.
(21) de Levie, R. Electrochim. Acta 1964, 9, 1231.
(22) Scheider, W. J. Phys. Chem. B 1975, 79, 127.
(23) Nielsch, K.; Mu¨ller, F.; Li, A.-P.; Go¨sele, U. AdV. Mater. 2000,
12, 582.
(24) Li, A.-P.; Mu¨ller, F.; Birner, A.; Nielsch, K.; Go¨sele, U. J. Appl.
Phys. 1998, 84, 6023.
(25) Li, A. P.; Mu¨ller, F.; Birner, A.; Nielsch, K.; Go¨sele, U. J. Vac.
Sci. Technol. A 1999, 17, 1428.
(26) Jessensky, O.; Mu¨ller, F.; Go¨sele, U. Appl. Phys. Lett. 1998, 72,
1173.
(27) Serebrennikova, I.; Vanysek, P.; Birss, V. I. Electrochim. Acta 1997,
42, 145.
(28) Schmidt, C.; Mayer, M.; Vogel, H. Angew. Chem., Int. Ed. 2000,
39, 3137.
Acknowledgment. Funding by the DFG is gratefully
acknowledged (STE 884/2-1 and 2-2). We thank K. Nielsch
and A. Janshoff for fruitful discussions.
(29) Bruggeman, D. A. G. Ann. Phys. (Leipzig) 1935, 24, 636.
(30) Macdonald, J. R. Impedance Spectroscopy; John Wiley & Sons:
New York, 1987.
References and Notes
(31) Asaka, K.; Ottova, A.; Tien, H. T. Thin Solid Films 1999, 354,
201.
(32) Michalke, A.; Galla, H.-J.; Steinem, C. Eur. Biophys. J. 2001, 30,
421.
(33) Purrucker, O.; Hillebrandt, H.; Adlkofer, K.; Tanaka, M. Electro-
chim. Acta 2001, 47, 791.
(1) Sackmann, E. Science 1996, 271, 43.
(2) Michalke, A.; Schu¨rholz, T.; Galla, H.-J.; Steinem, C. Langmuir
2001, 17, 2251.
(3) Knoll, W.; Frank, C. W.; Heibel, C.; Naumann, R.; Offenha¨usser,
A.; Ru¨he, J.; Schmidt, E. K.; Shen, W. W.; Sinner, A. ReV. Mol. Biotechnol.
2000, 74, 137.
(34) Raguse, B.; Braach-Maksvytis, V.; Cornell, B. A.; King, L. G.;
Osman, P. D. J.; Pace, R. J.; Wieczorek, L. Langmuir 1998, 14, 648.
(35) Steinem, C.; Janshoff, A.; Ulrich, W.-P.; Sieber, M.; Galla, H.-J.
Biochim. Biophys. Acta 1996, 1279, 169.
(36) Steinem, C.; Janshoff, A.; Galla, H.-J.; Sieber, M. Bioelectrochem.
Bioenerg. 1997, 42, 213.
(4) Salamon, Z.; Tollin, G. Biophys. J. 1996, 71, 858.
(5) Majewski, J.; Wong, J. Y.; Park, C. K.; Seitz, M.; Israelachvili, J.
N.; Smith, G. S. Biophys. J. 1998, 75, 2363.
(6) Sackmann, E.; Tanaka, M. Trends Biotechnol. 2000, 18, 58.
(7) Wiegand, G.; Arribas-Layton, N.; Hillebrandt, H.; Sackmann, E.;
Wagner, P. J. Phys. Chem. B 2002, 106, 4245.
(8) Hillebrandt, H.; Wiegand, G.; Tanaka, M.; Sackmann, E. Langmuir
1999, 15, 8451.
(37) Goeminne, G. T. H.; Vereecken, J. Electrochim. Acta 1995, 40,
479.
(9) Mu¨ller, P.; Rudin, D. O. Nature 1967, 213, 603.
(10) Hennesthal, C.; Steinem, C. J. Am. Chem. Soc. 2000, 122, 8085.
(11) Hennesthal, C.; Drexler, J.; Steinem, C. Chem. Phys. Chem. 2002,
10, 885.
(38) Delrouyre, V.; Laval, J.-M.; Bourdillon, C. J. Am. Chem. Soc. 2001,
123, 9176.
(39) Marchal, D.; Boireau, W.; Laval, J. M.; Moiroux, J.; Bourdillon,
C. Biophys. J. 1998, 74, 1937.
(12) Hoar, T. P.; Wood, G. C. Electrochim. Acta 1962, 7, 333.
(13) Mansfeld, F.; Kendig, M. W. J. Electrochem. Soc. 1988, 135, 828.
(14) Hitzig, J.; Ju¨ttner, K.; Lorenz, W. J.; Paatsch, W. J. Electrochem.
Soc. 1986, 133, 887.
(40) Plant, A. L.; Gueguetchkeri, M.; Yap, W. Biophys. J. 1994, 67,
1126.
(41) Fromherz, P.; Kiessling, V.; Kottig, K.; Zeck, G. Appl. Phys. A
1999, 69, 571.
(15) Van der Linden, B.; Terryn, H.; Vereecken, J. J. Appl. Electrochem.
1990, 20, 798.
(42) Feder, T. J.; Weissmu¨ller, G.; Zeks, B.; Sackmann, E. Phys. ReV.
E. 1995, 51, 3427.