A Novel Membrane Charge Sensor - American Chemical Society

We report the design of a novel membrane charge sensor by the deposition of highly insulating polymer/lipid composite films on indium tin oxide (ITO) ...
1 downloads 0 Views 180KB Size
J. Phys. Chem. B 2002, 106, 477-486

477

A Novel Membrane Charge Sensor: Sensitive Detection of Surface Charge at Polymer/Lipid Composite Films on Indium Tin Oxide Electrodes Heiko Hillebrandt, Motomu Tanaka,* and Erich Sackmann Institut fu¨ r Biophysik, Technische UniVersita¨ t Mu¨ nchen, James-Franck-Strasse, D-85748 Garching, Germany ReceiVed: May 3, 2001; In Final Form: October 17, 2001

We report the design of a novel membrane charge sensor by the deposition of highly insulating polymer/lipid composite films on indium tin oxide (ITO) semiconductor electrodes. The lipid monolayers were deposited on soft Langmuir-Blodgett (LB) multilayers of cellulose derivatives (‘hairy-rod’ polymers) by continuous exchange of solvent. The optical transparency of ITO enables the parallel characterization of the polymersupported lipid monolayers by electrochemical impedance spectroscopy and fluorescence microscopy. The polymer/lipid composite system yielded an electric resistance of 2.5 × 106 Ω cm2 and a lateral diffusion constant for the lipids of 0.1 µm2/s. Such highly insulating and fluid composite films on ITO semiconductor electrodes can be utilized as membrane charge sensors to detect changes in surface charge by treating this electrolyte/(organic) insulator/semiconductor (EIS) system as an analogue of the metal/oxide/semiconductor (MOS) system. For this purpose, we incorporated 10 mol % of lipids with a chelator headgroup (nitrilotriacetic acid, NTA) to switch the membrane charge. A difference in surface charge density of ∆Q ) 2.2 × 10-6 C/cm2 changed the flat band potential of the EIS system by nearly 50%. This result suggests that the sensitivity limit for our setup is sufficient to detect the binding of charged proteins to a membrane surface.

Introduction Supported lipid membranes on solid surfaces are both scientifically and practically important as artificial mimics for cell membranes.1 To separate the lipid membranes from the solid support, the deposition of ultrathin and soft polymer films is a promising approach to the design of mechanically and thermodynamically stable membranes for biosensing applications.2 Such soft polymer cushions can prevent defects through a reduction of the van der Waals and electrostatic interactions, resulting in fluid and self-healing lipid membranes with high electrical resistances.3 Such defect-free and fluid model cell membranes should prevent leak currents and enable the design of sensitive electrochemical biosensors. Some electrochemical sensors based on thin polymeric films on electrodes have already been reported, such as polysiloxane films used as pH4,5 or cadmium sensors.6 Moreover, Mortimer and Beech7 studied the impedance spectroscopy of triblock copolymers on platinum electrodes to measure carbon monoxide concentrations. In the present study, we designed composite films of lipid monolayers and insulating ‘hairy-rod’ polymer cushions8,9 on indium tin oxide (ITO) semiconductor electrodes (Figure 1). The polycrystalline semiconductor ITO is polarizable in aqueous electrolytes without any insulating layers, and it has a high electron density (∼1021 cm-3). Because the Fermi energy EF of ITO with such a high electron density is close to the conduction band edge or even above, it has to be treated as a degenerated semiconductor. ITO is transparent to visible light (λ > 400 nm), allowing for combined characterizations by optical microscopy and electrochemical techniques for biotechnological studies. ITO has been used for cell experiments10-12 and for the characterization of lipid membranes containing proteins.13 As polymer cushions, we deposited Langmuir-Blodgett multilayers of the cellulose derivative isopentylcellulose cin-

Figure 1. Schematic illustration of the polymer-supported lipid monolayer system.

namate (IPCC) from the air/water interface. The soft and hydrophobic isopentyl side chains allow for the deposition of fluid lipid monolayers.14 In addition, the inter- and intralayer structure of the polymer films can be stabilized against organic solvents such as chloroform or ethanol by photo-cross-linking of the cinnamoyl side chains.15,16 The successive deposition of polymer monolayers on hydrophobized ITO electrodes results in an electrically insulating polymer cushion with resistances of up to 3.0 × 107 Ω cm2.3 The lipid monolayer was deposited on the cross-linked polymer cushions by the exchange of solvent.17,18 The monolayer self-assembles from the ethanol solution onto the surface when the solvent is diluted with an aqueous electrolyte. To change the surface charge density of the lipid membranes reversibly, we incorporated chelator lipids19,20 into the supported monolayer. Such lipids can form chelate complexes with divalent metal ions such as nickel and change their molecular charge. Furthermore, the complexation can be canceled by treatment with a competitor such as EDTA (see Figure 1). In the present paper, we used commercially available dioleoylglycero-(nitrilotriacetic acid)succinyl (DOGS-NTA) lipids to modify the membrane charge.

10.1021/jp011693o CCC: $22.00 © 2002 American Chemical Society Published on Web 12/19/2001

478 J. Phys. Chem. B, Vol. 106, No. 2, 2002 The electrical properties of the stratified composite films were studied by measuring the complex impedance over a large frequency range (10 mHz-50 kHz), and the measured spectra were analyzed in terms of equivalent circuits.21 In parallel, the homogeneity and fluidity of the polymer-supported lipid monolayers were monitored by fluorescence microscopy. The diffusion constants of the monolayers were measured by fluorescence recovery after photobleaching (FRAP).22,23 Here, we demonstrate that this electrolyte/(organic) insulator/ semiconductor (EIS) structure can be applied as a capacitive transducer for the highly sensitive detection of surface charge at the lipid membrane surface. Because this EIS structure was treated as an analogue of the metal/oxide/semiconductor (MOS) structure, changes in membrane surface charge density can be detected as changes in the space charge capacitance of the ITO semiconductor. To estimate the changes in membrane charge density, we measured the flat band potential of the composite system using the Mott-Schottky analysis as a semiquantitative approximation for a degenerated semiconductor. We propose that our setup, in combination with semiconductor technology, can serve as a reversible platform for screening applications in the future. Experimental Section Chemicals. Dimyristoylphosphatidylcholine (DMPC), cholesterol, and dioleoylglycero-(nitrilotriacetic acid)succinyl (DOGSNTA) were purchased from Avanti Polar Lipids Inc. (Alabaster, Al), and NBD-dipalmitoylphosphatidylethanolamine (NBDDPPE) from Molecular Probes (Leiden, The Netherlands). All other chemicals were purchased from Fluka (Neu Ulm, Germany) and were used without further purification. The standard electrolyte was degassed 10 mM HEPES [4-(2hydroxyethyl)piperazine-1-ethanesulfonic acid] titrated with NaOH to pH 7.5. For the reversible charging of chelator lipids, 10 mM HEPES buffers (pH 7.5) containing either 1 mM NiCl2 (nickel buffer) or 50 mM EDTA (ethylenediaminetetraacetic acid, EDTA buffer) were used. To minimize the effect of electrolyte on the impedance spectra, we matched the osmolarities of these two buffers by adding 80 mM NaCl to the nickel buffer. ITO Electrodes. Glass slides coated with 110 nm thick ITO were purchased from Balzers (Balzers, Lichtenstein). The lithographic process used in this study to prepare the electrode arrays (10 identical electrodes) is described elsewhere.3 Each electrode consisted of an array of circular disks (area 0.098 cm2) with thin contact lines (width 300 µm). The distance between the centers of the disks was about 4 mm, and the active area of each electrode was estimated with an accuracy of about (10%. First, the etched substrates were cleaned with acetone and ethanol. Then, they were immersed into a solution of 1:1:5 (v/ v) H2O2 (30%)/NH4OH (30%)/H2O for 5 min with ultrasonication and soaked for another 30 min at 60 °C. Finally, they were rinsed 10 times with water. Before silanization, the samples were thoroughly dried in a vacuum chamber. Silanization. The coupling reaction was accomplished by a 60 min sonication of the ITO substrates in a 5 vol % solution of octadecyltrimethoxysilane (ODTMS) in dry toluene (packaged with molecular shieves, water content < 0.005%) using n-butylamine (0.5 vol %) as the catalyst.24 Afterward, the samples were incubated for another 30 min in the same solution. To optimize the quality of the monolayers, the temperature was kept constant (T ) 293 K) throughout the silanization.25,26 The physisorbed silanes were removed from the surface by sonication for 2 min in pure toluene. The ODTMS monolayer deposition

Hillebrandt et al.

Figure 2. (a) Structure of the hairy-rod polymer, isopentylcellulose cinnamate (IPCC), used in this study. The degrees of substitution (DS) per glucose unit were DS(isopentyl) ) 2.9 and DS(cinnamoyl) ) 0.1. The average degree of polymerization was about 150. (b) Cross-linking reaction of cinnamoyl groups by photocycloaddition under illumination with UV light.

resulted in a hydrophobic surface, whose static contact angle with a water droplet was about 95°. Furthermore, the defect area ratio of the monolayer was quantitatively estimated by impedance spectroscopy in the presence of redox couples to be only 0.2%.27 Deposition of Hairy-Rod Polymers. Isopentylcellulose cinnamate (IPCC, Figure 2a) was synthesized as previously reported,15 resulting in derivatives with degrees of substitution of 2.9 for the isopentyl groups and 0.1 for the cinnamoyl groups. The Langmuir-Blodgett (LB) films of IPCC were deposited onto silanized substrates at a constant pressure of 18 mN/m and a temperature of T ) 20 °C. The successive deposition of LB films could be monitored by impedance spectroscopy in terms of a decrease in the interface capacitance.3 After deposition, the cinnamoyl side groups of the polymers were cross-linked (Figure 2b) by irradiation with a 500-W Hg(Xe) lamp (Oriel Instruments, Stratford, CT) for 4 min. The incident light from the lamp was reflected by an aluminum mirror to cut off deep-UV light (104 23 170

-560 -510 -730 -500 -730

a For comparison, some previous results from ref 27 are also included. b Measured at Ue ) 0 mV. c Subscript X denotes the different organic layers deposited on ITO. d Measured at -300 mV e Ue e 300 mV (linear region of the Mott-Schottky plots). e From ref 27. f Ten layers of IPCC. g Six layers of IPCC, cross-linked, treated with ethanol.

( ) zFUd 2RT

(3)

where  is the dielectric constant of the electrolyte ( ) 80 for water) and c is its concentration. The second term, ψd - ψP, is the potential across the Stern layer of adsorbed ions, which can be neglected because it is very small compared to the other contributions.33 In our experimental system, the contribution from the electrochemical double layer capacitance (∼1 × 10-4 F cm-2) can be neglected, because it is more than a factor of 10 above the semiconductor capacitance (see Table 1). Finally, ψP - ψS ) UP is the potential drop within the organic insulator (polymer/lipid composite) film. Assuming an ideal condensator, UP can be written as

ψ P - ψS ) UP )

σS CP

(4)

where CP is the capacitance of the organic insulator. Thus, the potential at the semiconductor surface ψS is given by

ψ S ) Ue -

σS 2RT sinh-1 CP zF

(

σd

)

x8okTc

(5)

Equation 5 predicts that ψS is dependent on the external potential Ue, the capacitance of the organic insulator CP, and the electrolyte concentration c. If additional fixed charges σ0 are present at the surface of the organic insulator (Figure 4b), then charge neutrality requires Figure 5. Simulation of the normalized capacitance C/CP as a function of semiconductor surface potential ΨS (low-frequency case of n-type semiconductor). The space charge capacitance of the ITO electrode was calculated using no ) 1021 cm-3, p0 ) 105 cm-3, T ) 300 K, and (ITO) ) 3.6. The capacitance of the organic insulator was set to CP ) 10-8 F/cm2.

capacitance of the semiconductor space charge region across the organic insulator by a field effect. Our semiquantitative description follows the works of Siu et al.33 and Bootsma et al.,34 which assume a completely blocked interface without any direct interactions between the electrolyte and the semiconductor surface. In Figure 4, the charge distributions and corresponding potential profiles of the EIS system are presented for two different situations. Figure 4a represents the uncharged EIS system under the external bias potential Ue. Considering only electrostatic effects, charge neutrality demands

σd + σS ) 0

(6)

The modified semiconductor potential ψS′ with additional surface charges can be calculated in a similar manner, and the semiconductor surface potential ψS′ is then also affected by the surface charges σ0, as previously reported by Siu and Cobbold.33 Sensitivity of the EIS. As outlined above, an additional charge σo on the surface of the organic insulator, as well as an applied bias Ue, will affect the surface potential ψS of the semiconductor and its space charge capacitance CSC. Because surface charges induce changes in the potential profile similar to those induced by a certain bias potential, we address the sensitivity of the EIS system in terms of an external potential Ue. The total capacitance C of the EIS can be represented as a serial connection of the space charge capacitance CSC and the organic insulator capacitance CP

(1)

where σS represents the mobile charge density in the space charge region and σd represents the charges in the diffusive layer at the insulator/electrolyte interface. The external potential can be separated as follows (see Figure 4)

Ue ) (Ue - ψd) + (ψd - ψP) + (ψP - ψS) + ψS

σd + σ0 + σS ) 0

(2)

The first term in eq 2 corresponds to the potential across the diffusive layer, Ue - ψd ) Ud. This potential drop can be related to the charge density in the diffusive layer, σd, according to the Poisson-Boltzmann equation for symmetric electrolytes with valence z35

C)

CPCSC CP + CSC

(7)

As the capacitance of the diffuse double Cd layer is more than 1 order of magnitude larger than this combined capacitance, it can be neglected in the following considerations. Figure 5 shows a simulation of the normalized capacitance C/CP as a function of the semiconductor surface potential ψS for low-frequency perturbations, which is linked to the external potential Ue (see eq 5). For a given insulator thickness, the value of CP is constant and corresponds to the maximum value of C/CP. The differential capacitance of an n-type semiconductor space charge region can be given by31

Novel Membrane Charge Sensor

CSC(ψS) )

J. Phys. Chem. B, Vol. 106, No. 2, 2002 481

∂σS ) ∂ψS po (1 - e-βψS) + eβψS no

o

x2dn

x

(8)

po -βψS (e + βψS - 1) + eβψS - βψS - 1 no

where dn denotes the Debye length dn ) xo/eβno and β ) e/kT. Note that eq 8 is based on the charge carrier density described by a Boltzmann distribution. For the simulations, we used the majority charge carrier density no ) 1021 cm-3 (verified by Hall measurements at the Walter Schottky Institute, Technical University Munich), the minority density po ) 105 cm-3, the insulator capacitance CP ) 10-8 F/cm2, the dielectric constant (ITO) ) 3.6,36 and the temperature T ) 300 K. The low-frequency case was simulated in Figure 5, because CSC dominates the impedance spectra of the system at frequencies below 10 Hz. The inversion of the ITO electrodes could not be observed experimentally, because the ITO decomposed in this potential region (Ue < -800 mV). The simulation curve in Figure 5 shows that the EIS system exhibits a very narrow potential region where its capacitance is sensitive to changes in the external potential or surface charge. Therefore, the charge carrier densities no and po, as well as the potential drop across the organic insulator UP, of the EIS must be tuned carefully for sensitivity according to the target system. Results and Discussion IPCC Polymer Cushions. After silanization, the IPCC monolayers were transferred onto the hydrophobized ITO surface by successive LB deposition. As the thickness of the IPCC monolayers is d ≈ 0.9 nm,14,37 the total thickness of the polymer cushion can be adjusted to an accuracy on the order of nanometers. For this study, we used IPCC polymer cushions with 6, 10, and 20 monolayers. As schematically depicted in Figure 2b, the cinnamoyl groups of the IPCC polymers can establish intra- and interlayer cross-links through photocycloaddition under UV illumination (λ ≈ 250 nm).15 After photocross-linkage, the polymer network is stable against treatment with organic solvents such as chloroform or ethanol. Furthermore, cinnamoyl side groups even enable the microstructuring of hairy-rod films by photolithography.16 Figure 6 shows the impedance spectra of a cross-linked polymer cushion with 10 layers of IPCC before (2) and after (4) the treatment with ethanol. The impedance spectra were analyzed with the equivalent circuit from the inset of Figure 6. The resistance R0 ) 4.0 × 103 Ω corresponds to the ohmic contributions from electrolyte and contacts, whereas the dielectric properties of the polymer film can be represented by a parallel combination of the resistance RP and the capacitance CP. The reference spectrum of the polymers before ethanol treatment yielded a cushion resistance of RP g 1.0 × 107 Ω cm2 and a capacitance of CP ) 2.3 × 10-7 F cm-2. Although the resistance of the polymer cushion RP was too high to be measured accurately with our setup, the lower limit given above indicates that the IPCC cushion behaves as an organic insulator,3 separating the semiconductor surface from the electrolyte. Judging from the increase in the capacitance CP compared to our previous study,3 photo-cross-linkage under UV illumination seems to reduce the cushion thickness. This can be explained either by an increased packing density after cross-linkage or by the photocleaving of the cellulose backbone by UV light.

Figure 6. Absolute impedance of ITO electrodes covered by 10 layers of photo-cross-linked IPCC before (2) and after (4) the treatment with ethanol. The measured spectra were analyzed with the equivalent circuit from the inset. The symbols represent the measured data, and the solid lines are the corresponding fits. After treatment with ethanol, the capacitance CP increased from 2.3 × 10-7 to 2.9 × 10-7 F cm-2. On the other hand, the frequency exponent of the constant phase element (CPE) remained almost constant (R ) 0.98-0.96), suggesting that the homogeneity of the polymer film was not disturbed.

The effect of the photo-cross-linkage of cinnamoyl side groups on the structure of polyglutamate films was previously studied using X-ray scattering.38 It was reported that the structural homogeneity and layer-by-layer architecture of the film was preserved, even though the thickness decreased. After ethanol treatment (∆ in Figure 6), the resistance decreased to RP ) 1.0 × 106 Ω cm2, whereas the capacitance increased to CP ) 2.9 × 10-7 F cm-2. If one assumes that the dielectric constant of the polymer cushion is not altered by the ethanol treatment, this change in capacitance corresponds to a decrease in thickness of nearly 30%, which can be attributed to the removal of non-cross-linked IPCC. On the other hand, if ethanol remains inside the polymer film, this might increase the apparent capacitance as well, as alcohol has a high dielectric constant in the region of  ) 24.39 Nevertheless, the IPCC films were stable against further treatment with ethanol after they were exposed to the alcohol treatment. Supported Lipid Monolayers. A lipid monolayer was deposited onto an IPCC cushion pretreated with ethanol through exchange of solvent by an aqueous buffer. After the exchange of solvent, the homogeneity of the lipid film was first checked by fluorescence microscopy. We observed homogeneous fluorescence intensity for a lipid monolayer on 10 layers of crosslinked IPCC over the whole electrode array (Figure 7a). The lateral diffusion constant of the polymer-supported lipid monolayer was measured in a FRAP experiment. After the fluorescence dyes were irreversibly bleached by a flash of highintensity laser light, the intensity inside the bleaching spot (diameter of 15 µm) was recorded as a function of time (Figure 7b). From this intensity vs time curve, one can extract the fluorescence recovery R and the lateral diffusion coefficient D.28 We obtained a fluorescence recovery of R ) 90% and a diffusion constant of D ) 0.1 µm2 s-1. This diffusivity is comparable to the value reported by Sigl et al.14 for a DMPC monolayer on 6 layers of IPCC, D ) 0.4 µm2/s. The smaller diffusion constant in our experiment is attributed to the 34 mol % of cholesterol in the membrane, which is known to reduce the lateral fluidity.40

482 J. Phys. Chem. B, Vol. 106, No. 2, 2002

Hillebrandt et al.

Figure 7. (a) Fluorescence image of a polymer-supported lipid monolayer on 10 layers of cross-linked IPCC. The fluorescence signal was very homogeneous over the whole sample. (b) The recovery of fluorescence intensity vs time of the same sample, yielding a diffusion constant of D ) 0.1 µm2/s and a recovery of R ) 90%. I0 is the intensity before the bleaching flash, I(0) is the intensity at the beginning of recovery (t ) 0 s), and I(∞) is the intensity after recovery.

Figure 8. Absolute impedance of ITO electrodes covered by 10 layers of photo-cross-linked IPCC before (4) and after ([) deposition of the lipid monolayer. The electrochemical parameters were obtained with the equivalent circuit in Figure 4. The deposition of the lipid monolayer increased the resistance to RP ) 2.5 × 106 Ω cm2. The thickness of the monolayer could be calculated from the decrease in the capacitance, ∆CP, as d ) 1.5 nm.

The changes in dielectric properties of the system due to monolayer deposition were also studied by impedance spectroscopy (Figure 8). After monolayer formation ([), the resistance of the dielectric film RP increased to 2.5 × 106 Ω cm2, indicating the highly insulating behavior of the lipid monolayer. In addition, the decrease in parallel capacitance CP corresponds to a dielectric thickness of d/ ) 0.68 nm. Using the dielectric constant of the hydrocarbon region of the lipid membrane, M ) 2.2,41,42 the thickness of the monolayer can be calculated as d ) 1.5 nm. Flatband Potential of the EIS. In Figure 9, the impedance spectra of a 10-layer IPCC film under various bias potentials (500 mV g Ue g -600 mV) are presented. Impedance spectroscopy over a large frequency range (90 mHz-50 kHz) is a powerful technique for separating the contributions of the semiconductor and the organic insulator, which dominate the spectra in different frequency regimes. The equivalent circuit (see inset of Figure 6) was extended by another set of resistance and capacitance (RSC and CSC) to account for the electrical characteristics of the ITO electrode at small frequencies (inset Figure 9). The external bias potential affects the current response of the EIS system in two frequency regimes. In the frequency regime

between 1 Hz and 1 kHz, the impedance spectra are dominated by the dielectric properties of the IPCC cushion. For 500 mV g Ue g -100 mV, the resistance of the polymer cushion remains almost constant at RP ≈ 1.2 × 104 Ω cm2. For Ue < -100 mV, the value decreases to RP(-600 mV) ) 7.0 × 102 Ω cm2, indicating the loss of ohmic behavior above a certain threshold potential. Because this significant change in resistance RP is reversible, its low value at Ue ) -600 mV cannot be interpreted as a breakdown of the dielectric film.43 It has been reported that the current-voltage characteristics of cellulose acetate films exhibit ohmic conduction for low fields and nonohmic conductance for high fields.44,45 Electronic charge migration in polymers can be described as a hopping process, as the charges are mostly located in traps.46 In the nonohmic region, the I-V curve has a slope of ∼2, which can be explained in terms of space-charge-limited currents (SCLCs).47 However, we did not observe such behavior in the nonohmic region of the IPCC conductance (Ue < -100 mV). Tentatively, the reduction of RP is attributed to reversible conformational changes of the polymers in high electric fields, opening pores for ionic currents. In fact, Albery et al.48 proposed a dual transmission line for charge transfer through polymer films consisting of a resistance for electronic motion in the conducting polymer and a resistance for ionic motion through pores. On the other hand, the change in the polymer cushion capacitance CP is within (10%, which is comparable to the accuracy of our measurements. Hence, we concluded that the thickness of the IPCC film was not altered by the applied bias potential. In the low-frequency regime (ω < 1 Hz), the impedance spectra (Figure 9) are dominated by the capacitance of the space charge region, CSC. For Ue e -500 mV, even the resistance of the semiconductor RSC becomes visible. A Mott-Schottky plot of the CSC values measured at different bias potentials is shown in Figure 10. The linear part of this plot was interpolated to the x axis, yielding a flat band potential for the EIS system of UFB(IPCC) ) -730 mV. At |Ue| > (300 mV, this plot shows a clear deviation from the linear relationship expected from the Mott-Schottky equation

CSC-2 )

2 kT ∆Ψ SCoen e

(

)

(9)

where ∆Ψ represents ∆Ψ ) Ue - UFB. It has been reported that such nonlinear behaviors can be observed if doping distributions are nonhomogeneous as a result of surface layers49-51 or if additional donor states are present in the band gap, leading to a potential-dependent space charge density.32

Novel Membrane Charge Sensor

J. Phys. Chem. B, Vol. 106, No. 2, 2002 483

Figure 9. Impedance spectra of an IPCC cushion (10 layers) under different bias potentials. To take the electrical characteristics of the ITO electrodes into account (dominating the impedance spectra below 1 Hz), another set of resistance and capacitance (CSC and RSC) was introduced to the equivalent circuit.

Figure 11. Chemical structure of the nitrilotriacetic acid (NTA) headgroup of the chelator lipid in the Ni-loaded and unloaded states. The reversible complexation of nickel ions changes the molecular net charge of the NTA-lipid by 1e-.

Figure 10. Mott-Schottky plot for the EIS system (10 layers of IPCC). The values of the capacitance CSC were obtained from the impedance spectra of Figure 9. The intercept of the extrapolated linear part with the x axis is the flat band potential UFB of the system, UFB ) -730 mV.

Dean et al.52 proposed a theoretical model of disordered semiconductors with different donor states in the band gap. Depending on the distribution of the additional band gap states, the resulting Mott-Schottky plots exhibited pronounced nonlinearity. However, if the distribution of localized states decays exponentially from the band edges and a constant donor band is present, the global shape of the plot should be similar to our results in Figure 10, showing a linear part between two almostconstant regions. They claimed that the interpolation of the linear region to 1/CSC2 ) 0 can provide only a rough estimate for the flat band potential of the semiconductor and that the error of this approximation is dependent on the actual donor states in the band gap. Because a precise treatment of a degenerated and polycrystalline semiconductor in contact with electrolyte is still very complicated, the Mott-Schottky analysis was applied

throughout this paper to obtain a semiquantitative value of the flat band potential of the composite system. Table 1 summarizes the calculated impedance data in this study, as well as some values obtained previously.27 The flat band potentials of the ITO with and without the silane monolayer coating were calculated with the same approximation described above to be UFB(ITO) ) -560 mV and UFB(silane) ) -510 mV, respectively. The difference between the UFB values before and after the deposition of the polymer film, ∆UFB ) UFB(IPCC) - UFB(silane) ) 220 mV, can be related to the potential drop across the organic insulator. The main aim of this study is to measure changes in surface charge of the polymer-supported lipid monolayer as a function of changes in the flat band potential UFB of the EIS system. It should be noted here that all changes in the impedance spectra presented in Figure 9 are completely reversible, verifying that the applied potential damaged neither the ITO electrodes nor the IPCC film. Indeed, previous studies reported that the degradation of ITO occurs only at |Ue| > 800 mV.13,53,54 Detection of Membrane Surface Charge. Figure 11 shows the chemical structure of the nitrilotriacetic acid (NTA) chelator group of the NTA-DOGS lipid in the complexed (Ni-loaded) and decomplexed (unloaded) states. The NTA group can form an octahedral complex in the presence of divalent ions such as nickel and change its molecular net charge from 2e- to 1e-.55

484 J. Phys. Chem. B, Vol. 106, No. 2, 2002

Figure 12. Absolute impedance of a lipid monolayer containing 10 mol % of NTA-DOGS on 6 layers of IPCC at Ue ) 0 mV. In the presence of nickel ions (Ni-loaded, 4), the lipid monolayer bears 1e-/ NTA. After treatment with EDTA (+), the NTA headgroups are unloaded, and the monolayer is charged with 2e-/NTA.

This reaction can be reversed by treatment with competitors such as EDTA. These chelator lipids are often applied for the reversible and specific immobilization of recombinant histidinetagged proteins to lipid membranes.19,20,56 In this study, lipid monolayers with 10 mol % of NTA-DOGS were deposited on IPCC cushions by solvent exchange from ethanol to nickel buffer to form a monolayer with Ni-loaded NTA groups (1e-/NTA). Afterward, the NTA groups were unloaded by treatment with EDTA buffer, resulting in a monolayer with 2e-/NTA. By assuming an area per lipid molecule in the fluid phase of about 0.65 nm2,57 the change in surface charge density of the supported lipid monolayer induced by the loading and unloading of the NTA groups can be estimated as ∆Q ) 2.2 × 10-6 C/cm2. We performed charging experiments with lipid monolayers deposited directly on silanized electrodes and on electrodes coated with 6, 10, and 20 monolayers of IPCC. As predicted in the theory section, the sensitivity of our EIS system was strongly dependent on the potential drop across the insulator, i.e., the thickness of the polymer cushion. If the monolayer was deposited onto thick IPCC cushions (10 or 20 layers) or directly onto the silane monolayer, the changes in surface charge density induced by the loading or unloading of the NTA lipids (10 mol %) did not alter the measured impedance in the frequency range between 10 mHz and 50 kHz. On the other hand, the lipid monolayer on 6 layers of IPCC changed the impedance spectra significantly, in coincidence with the loaded or unloaded states of NTA lipids (Figure 12). The switching between these two states was reversible and reproducible for more than 2 weeks with the same sample. Figure 12 shows the impedance spectra of a lipid monolayer with 10 mol % of NTA-DOGS on 6 layers of IPCC measured at Ue ) 0 mV. According to the reversible loading and unloading of the NTA headgroups by treatment with nickel and EDTA buffer, the monolayer surface charge can be switched from 1e-/NTA to 2e-/NTA, respectively. The charging of the supported lipid monolayer resulted in a decrease of the resistance RP from 1.7 × 105 to 2.3 × 104 Ω cm2 and an increase of the capacitance CP from 5.6 × 10-7 to 7.0 × 10-7 F cm-2. These

Hillebrandt et al.

Figure 13. Impedance spectra of the lipid monolayer with Ni-loaded NTA-DOGS under various bias potentials. The flat band potential of the EIS system with 1e-/NTA was calculated by a Mott-Schottky analysis to be UFB(loaded) ) -500 mV.

values were obtained using the equivalent circuit from the inset of Figure 9. The change in RP can be interpreted in terms of the modified potential profile induced by the additional charges σ0, as observed for the IPCC films under external bias potential (see Figure 9). On the other hand, the interpretation of changes in the capacitance CP is more difficult. One possibility is a decrease in chain packing due to charging of the NTA groups,19 resulting in a decrease in the dielectric thickness. However, this effect seems not to be sufficient to explain the relatively large change that we observed. Although the space charge capacitance CSC could be observed in the frequency regime from 10 to 100 mHz, any quantitative changes between the two different states of the NTA lipids could be gained at Ue ) 0 mV. Thus, impedance spectra were measured under various bias potentials (200 mV g Ue g -600 mV) to obtain the flat band potential UFB in the loaded (Figure 13) and unloaded (Figure 14) states of the NTA groups. Similarly to the current response of the IPCC films under various bias potentials (Figure 9), changes in the organic insulator resistance RP, as well as in the space charge capacitance of the semiconductor CSC, could be observed. The capacitance CSC and the resistance RP were extracted from these impedance spectra using the equivalent circuit given in the inset of Figure 9. Figure 13 presents impedance spectra of the lipid monolayer with Ni-loaded NTA-DOGS under various bias potentials. According to the decrease in bias potential, the resistance of the organic insulator RP decreased reversibly from RP(200 mV) ) 5.0 × 105 Ω cm2 to RP(-600 mV) ) 3.7 × 104 Ω cm2. The estimated values of CSC yielded a flat band potential for the EIS system with a lipid monolayer bearing 1e-/NTA of UFB(loaded) ) -500 mV. Figure 14 contains impedance spectra of the EIS system with unloaded NTA lipids after treatment with EDTA under various bias voltages. The decrease in resistance of the insulator from RP(200 mV) ) 4.3 × 104 Ω cm2 to RP(-600 mV) ) 2.9 × 103 Ω cm2 was similar to that for the loaded monolayer. From the Mott-Schottky analysis of the capacitance CSC, we obtained a flat band potential for the EIS system with a lipid monolayer bearing 2e-/NTA of UFB(unloaded) ) -730 mV.

Novel Membrane Charge Sensor

J. Phys. Chem. B, Vol. 106, No. 2, 2002 485 the NTA groups are loaded with nickel, the resulting octahedral complex is known to serve as a selective binding side for recombinant proteins exhibiting 6 histidine residues (Histag).20,59,60 Therefore, an EIS system based on a polymersupported monolayer with NTA lipids can be used as a recyclable platform for the selective detection of charged protein binding for noninvasive applications in pharmaceutical screening tests. Furthermore, our introduced membrane voltmeter can also be extended to other natural ligands and synthetic epitopes such as biotin tethers56 and glycan-phosphatidyl inositol (GPI) linkages61 for the docking of the extracellular domains of various receptors and cell adhesion molecules (CAMs).

Figure 14. Impedance spectra of the lipid monolayer with unloaded NTA-DOGS under various bias potentials. A Mott-Schottky analysis yielded a flat band potential for the EIS system with 2e-/NTA of UFB(unloaded) ) -730 mV.

The obtained results demonstrate that the change in membrane surface charge density of ∆Q ) 2.2 × 10-6 C/cm2 (corresponding to about one charge per 8 nm2) can be detected using the flat band potential of the EIS system, which was altered by almost 50%. As our measurement accuracy is in the range of 10%, we can roughly estimate our sensitivity limit to be around 0.03 charges per square nanometer. Summary We demonstrated that polymer-supported lipid monolayers on ITO semiconductor electrodes provide a promising potential as biomimetic voltage sensors for detecting the binding of charged proteins to model membrane surfaces. The electrical response of the designed electrolyte/(organic) insulator/semiconductor (EIS) system was interpreted as that of the conventional MOS system. Because we can control the potential drop across the organic insulator through the number of polymer layers, the sensitivity of the EIS system can be tuned flexibly according to the target. Impedance spectroscopy over a large frequency range enables the contributions of the organic insulator to be separated from those of the semiconductor electrodes, as they dominate the spectra in different frequency regimes. The changes in membrane charge density could be detected by measuring the flat band potential UFB of the EIS system utilizing the Mott-Schottky approach as a semiquantitative measure for a degenerated semiconductor. It was demonstrated that the change in membrane charge density of ∆Q ) 2.2 × 10-6 C/cm2 alters the flat band potential of the system (lipid monolayer with 10 mol % of NTA-DOGS on 6 layers of IPCC) by almost 50%. This suggests roughly that the sensitivity of our membrane voltmeter should be in the range of 0.03 charges per square nanometer. Assuming a size extension for fusion proteins such as green fluorescence protein (GFP) of about 10 nm2,58 the membrane voltmeter should be able to detect the binding of various charged proteins to membrane surfaces. Furthermore, the polymer-supported lipid monolayers were mechanically and thermodynamically stable enough to reproduce the results for more than 2 weeks. The NTA lipids used in this study can be switched reversibly between two different states (Ni-loaded and unloaded). When

Acknowledgment. One of the authors (M.T) is deeply indebted to Prof. G. Wegner for many suggestions on cellulose chemistry. The authors are grateful to Prof. R. Tampe´ and Drs. L. Schmitt and I.T. Dorn for helpful discussions, as well as Prof. M. Stutzmann and Prof. U. Stimming for their constructive suggestions. Additionally, we thank Z. Guttenberg for the FRAP experiments and J. Nissen for the fluorescence microscopy studies. This work was supported by the Deutsche Forschungs Gemeinschaft (SFB 563) and by the Fond der Chemischen Industrie. M.T. acknowledges the Alexander von Humboldt Foundation for a postdoctoral fellowship and DFG for a Habilitation fellowship (Emmy Noether-Programm, Ta 259/11). References and Notes (1) Sackmann, E. Science 1996, 271, 43-48. (2) Sackmann, E.; Tanaka, M. Trends Biotechnol. 2000, 18, 58-64. (3) Hillebrandt, H.; Wiegand, G.; Tanaka, M.; Sackmann, E. Langmuir 1999, 15, 8451-8459. (4) Vogel, A.; Hoffmann, B.; Sauer, T.; Wegner, G. Sens. Actuators B 1990, 1, 408-411. (5) Sauer, T.; Caseri, W.; Wegner, G.; Vogel, A.; Hoffmann, B. J. Phys. D 1990, 23, 79-84. (6) Ben Ali, M.; Kalfat, R.; Sfihi, H.; Chovelon, J. M.; Ben Ouada, H.; Jaffrezic-Renault, N. Sens. Actuators B 2000, 62, 233-237. (7) Mortimer, R. J.; Beech, A. J. Electrochem. Soc. 2000, 147, 780786. (8) Wegner, W. Thin Solid Films 1992, 216, 105-116. (9) Wegner, G. Mol. Cryst. Liq. Cryst. 1993, 235, 1-34. (10) Gross, G. W.; Wen, W. Y.; Lin, J. W. J. Neurosci. Methods 1985, 15, 243-252. (11) Gross, G. W.; Rhoades, B. K.; Reust, D. L.; Schwalm, F. U. J. Neurosci. Methods 1993, 50, 131-143. (12) Hillebrandt, H.; Abdelghani, A.; Abdelghani-Jacquin, C.; Aepfelbacher, M.; Sackmann, E. Appl. Phys. A 2001, 73, 539-546. (13) Gritsch, S.; Nollert, P.; Ja¨hnig, F.; Sackmann, E. Langmuir 1998, 14, 3118-3125. (14) Sigl, H.; Brink, G.; Schulze, M.; Wegner, G.; Sackmann, E. Eur. Biophys. J. 1997, 25, 249-259. (15) Seufert, M.; Christo, F.; Wegner, G. AdV. Mater. 1995, 7, 52-55. (16) Wiegand, G.; Jaworek, T.; Wegner, G.; Sackmann, E. Langmuir 1997, 13, 3563-3569. (17) Miller, C.; Cuendet, P.; Gra¨tzel, M. J. Electroanal. Chem. 1990, 278, 175-192. (18) Raguse, B.; Braach-Maksvytis; Cornell, B. A.; King, L. G.; Osman, P. D. J.; Pace, R. J.; Wieczorek, L. Langmuir 1998, 14, 648-659. (19) Schmitt, L.; Dietrich, C.; Tampe´, R. J. Am. Chem. Soc. 1994, 116, 8485-8491. (20) Shnek, D. R.; Pack, D. W.; Sasaki, D. Y.; Arnold, F. H. Langmuir 1994, 10, 2382-2388. (21) Macdonald, J. R. Impedance Spectroscopy; John Wiley & Sons: New York, 1987. (22) Axelrod, D.; Koppel, D. E.; Schlessinger, E.; Elson, E.; Webb, W. W. Biophys. J. 1976, 16, 1055-1069. (23) Merkel, R.; Sackmann, E.; Evans, E. J. Phys. Fr. 1989, 50. (24) Mooney, J. F.; Hunt, A. J.; McIntosh, J. R.; Liberko, C. A.; Walba, D. M.; Rogers, C. T. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 12287-12291. (25) Brzoska, J. B.; Shahidzadeh, N.; Rondelez, F. Nature 1992, 360, 719-721. (26) Parikh, A. N.; Allara, D. L. J. Phys. Chem. 1994, 98, 7577-7590. (27) Hillebrandt, H.; Tanaka, M. J. Phys. Chem. B 2001, 105, 42704276.

486 J. Phys. Chem. B, Vol. 106, No. 2, 2002 (28) Soumpasis, D. M. Biophys. J. 1983, 16, 95-97. (29) Ku¨hner, M.; Tampe´, R.; Sackmann, E. Biophys. J. 1994, 67, 217226. (30) Stelzle, M.; Weissmu¨ller, G.; Sackmann, E. J. Phys. Chem. 1993, 97, 2974-2981. (31) Sze, S. M. Semiconductor DeVices: Physics and Technology; John Wiley & Sons: New York, 1985. (32) Dean, M. H.; Stimming, U. J. Electroanal. Chem. 1987, 228, 135151. (33) Siu, W. M.; Cobbold, R. S. C. IEEE Trans. Electron DeVices 1979, 26, 1805-1815. (34) Bootsma, G. A.; De Rooij, N. F.; Silfhout, A. V. Sens. Actuators 1981, 1, 111-136. (35) Brockris, J. O. Modern Electrochemistry; Plenum Press: New York, 1970. (36) Mu¨ller, H. K. Phys. Status Solidi 1968, 27, 723-731. (37) Schaub, M.; Fakirov, C.; Schmidt, A.; Lieser, G.; Wenz, G.; Wegner, G.; Albouy, P. A.; Wu, H.; Foster, M. D.; Majrkzak, C.; Satija, S. Macromolecules 1995, 28, 1221-1228. (38) Iida, S.; Schaub, M.; Schulze, M.; Wegner, G. AdV. Mater. 1993, 5, 564-568. (39) Atkins, P. W. Physical Chemistry, 3rd ed.; Oxford University Press: Oxford, U.K., 1986. (40) Almeida, P. F. F.; Vaz, W. L.; Thompson, T. E. Biochemistry 1992, 31, 6739-6747. (41) Montal, M.; Mueller, P. Proc. Natl. Acad. Sci. U.S.A. 1972, 69, 3561-3566. (42) Dilger, J. P.; McLaughlin, S. G.; McIntosh, T. J.; Simon, S. A. Science 1979, 206, 1196-1198. (43) Blok, J.; LeGrand, D. G. J. Appl. Phys. 1969, 40, 288-293. (44) Kumar, A.; Nath, R. Phys. Status Solidi 1980, 61, 301-305.

Hillebrandt et al. (45) Khare, P. K. Indian J. Pure Appl. Phys. 1994, 32, 160-165. (46) Pai, D. M. J. Chem. Phys. 1970, 52, 2285. (47) Lambert, M. A.; Mark, P. Current Injection in Solids; Academic Press: New York, 1970. (48) Albery, W. J.; Mount, A. R. J. Chem. Soc., Faraday Trans. 1994, 90, 1115-1119. (49) Boddy, P. J. J. Electrochem. Soc. 1968, 115, 199. (50) Nogami, G. J. Electrochem. Soc. 1986, 133, 525-531. (51) Schoonman, J.; Vos, K.; Blasse, G. J. Electrochem. Soc. 1981, 128, 1154. (52) Dean, M. H.; Stimming, U. J. Phys. Chem. 1989, 93, 8053-8059. (53) Chyan, O. M.-R.; Rajeshwar, K. J. Electrochem. Soc. 1985, 132, 2109-2115. (54) Asanov, A. N.; Wilson, W. W.; Oldham, P. B. Anal. Chem. 1998, 70, 1156-1163. (55) Beauchamp, A. L.; Israeli, J.; Saulnier, H. Can. J. Chem 1969, 47, 1269-1273. (56) Savage, D.; Mattson, G.; Desai, S.; Nielander, G.; Morgensen, S.; Conklin, E. AVidin-Biotin Chemistry: A Handbook; Pierce Chemical Company: Rockford, IL, 1992. (57) Sackmann, E. Biological membranes architecture and function. In Handbook of Biological Physics; Lipowsky, R., Sackmann, E., Eds.; Elsevier Science: New York, 1995; Vol. 1. (58) Yang, F.; Moss, L. G.; Phillips, G. N. Jr. Nat. Biotechnol. 1996, 14, 1246-1251. (59) Gritsch, S.; Neumeier, K.; Schmitt, L.; Tampe´, R. Biosens. Bioelectron. 1995, 10, 805-812. (60) Dorn, I. T.; Pawlitschko, K.; Pettinger, S. C.; Tampe´, R. Biol. Chem. 1998, 379, 1151-1159. (61) Cross, G. A. M. Annu. ReV. Cell Biol. 1990, 6, 1-39.