Adsorption of BSA on passivated chromium studied by a flow-cell EQCM and XPS

Electrochimica Acta 52 (2007) 7660–7669

Adsorption of BSA on passivated chromium studied by a flow-cell EQCM and XPS I. Frateur a,∗,1 , J. Lecoeur a , S. Zanna a , C.-O.A. Olsson a,2 , D. Landolt b,1 , P. Marcus a,1 a

Laboratoire de Physico-Chimie des Surfaces, CNRS-ENSCP UMR7045, Ecole Nationale Sup´erieure de Chimie de Paris, 11 rue Pierre et Marie Curie, 75005 Paris, France b Ecole Polytechnique F´ ed´erale de Lausanne (EPFL), Laboratoire de M´etallurgie Chimique, CH-1015 Lausanne, Switzerland Received 25 July 2006; received in revised form 1 December 2006; accepted 17 December 2006 Available online 21 January 2007

Abstract The adsorption of bovine serum albumin (BSA) on a passivated chromium surface in deaerated pH 4 sulphate solution was studied in situ using a switch-flow cell in combination with an electrochemical quartz crystal microbalance (EQCM) and ex situ by X-ray photoelectron spectroscopy (XPS). EQCM measurements showed that the kinetics of BSA adsorption was fast, and that a steady-state was reached about 10 min after introducing the protein. They also showed that BSA adsorption was an irreversible process, or that the kinetics of desorption was very slow. The equivalent thicknesses of the adsorbed BSA layer estimated in situ by EQCM and ex situ by XPS are in excellent agreement, and are equal to 3.5 ± 0.7 nm, which corresponds to one horizontally orientated monolayer. © 2007 Elsevier Ltd. All rights reserved. Keywords: Chromium; BSA; EQCM; Flow cell; XPS

1. Introduction The initial stage involved in the adhesion of cells and the formation of biofilms on solid surfaces is the adsorption of proteins present in the medium. In particular, the interactions between cells and the surface of biomedical implants is mediated by pre-adsorbed proteins. Similarly, the growth of biofilms on construction materials used in marine environments and water distribution systems is initiated by the adsorption of biomolecules [1]. Stainless steels are often used in protein-containing environments, including marine, food, and biomedical applications. The bovine serum albumin (BSA) is a protein that is commonly used as a model protein for studying protein–surface interactions.

∗

Corresponding author. Tel.: +33 1 44 27 67 57; fax: +33 1 46 34 07 53. E-mail address: [email protected] (I. Frateur). 1 ISE member. 2 Permanent address: Outokumpu Stainless AB, Avesta Research Center, P.O. Box 74, SE-774 22 Avesta, Sweden. 0013-4686/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2006.12.060

The quartz crystal microbalance (QCM) technique is well suited for in situ studies of protein adsorption. However, in most cases, substrates are not electrochemically controlled and investigations are carried out in static conditions [2]; therefore, once the protein is adsorbed on the solid surface, the desorption of the protein cannot be followed in situ without removing the quartz crystal from the protein-containing solution i.e. without disturbing the microbalance signals. One way to perform QCM measurements under well-controlled hydrodynamic conditions is to use a flow cell. Flow cells have found many applications in the electrochemical literature since due to the small volume of the cell (0.1–1 mL), the inlet solution can be rapidly changed without interruption of the electrolyte flow. The coupling of flow cells with other measuring techniques, such as the electrochemical quartz crystal microbalance (EQCM), is easy to implement. Hamm et al. used an EQCM together with a flow cell described in detail by Ogle and Weber [3] to study the passivation of Fe–Cr alloys in acid sulphate electrolytes [4]. Garcia et al. investigated the formation of a scale deposit on gold and evaluated the efficiency of a well-known phosphonic acid (HEDP) inhibitor [5]. In combination with a switching system for changing the

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

electrolyte, the use of a flow-cell EQCM permits to follow the rate of adsorption as well as desorption. This offers the possibility for investigating the reversibility of an adsorption process by switching back and forth between a protein-free and a protein-containing solution. Thus, Galliano et al. designed a switch-flow cell that enabled an easy and rapid introduction and removal of adsorbates, without significantly disturbing the microbalance signals [6]. Their flow cell EQCM was a threeelectrode cell with two compartments separated by a membrane, permitting electrochemical control of the potential of the working electrode. The authors applied this device to the study of iodide adsorption on gold. To our knowledge, the switch-flow cell EQCM has never been used for the study of protein adsorption. The quartz crystal used as a sensor in the EQCM technique is not only sensitive to changes in mass of the working electrode. It is also highly sensitive to changes in temperature, pressure and viscous loading. Temperature and pressure can be controlled, but the principle of a switch-flow cell experiment is to change the electrolyte. Consequently, the changes in viscosity have to be compensated for. This can be done by simultaneous recording of the quartz crystal resonance frequency and resistance [6]. Another method is to switch off the driving circuit at regular intervals and to measure the attenuation rate of the oscillations as suggested by Rodahl and Kasemo [7]. In most protein adsorption studies using both QCM and Xray photoelectron spectroscopy (XPS), the latter was used only to characterize the substrate (polymeric films) before adsorption [8,9]. On the other hand, QCM and XPS have rarely been combined to study protein/substrate systems during and after adsorption, respectively [2]. In particular, the equivalent thickness of the adsorbed protein layer estimated from QCM measurements has never been compared to the one calculated from XPS data. The objective of this work was to study the adsorption of BSA on passivated chromium in a pH 4 deaerated sulphate solution. As Cr is a major constituent of stainless steels, it was chosen here as a model for this class of materials. The present study is intended as a first step to be followed by studies on ferritic stainless steels. The goal is to compare results of in situ EQCM and ex situ XPS. EQCM provides information on the kinetics of protein adsorption and on the amount of adsorbed protein. If combined with a switch-flow cell, it permits to study the reversibility of the adsorption process. In this work, a new switch-flow cell EQCM was designed similar to that developed by Galliano et al. [6], but allowing deaeration of both electrode compartments and use of commercially available quartz crystal electrodes. EQCM measurements were complemented by XPS analyses from which the chemical composition and the thickness of the surface layers can be obtained. The equivalent thicknesses of the adsorbed protein layer estimated by EQCM and by XPS were compared. After a detailed description of the experimental setup, the results obtained for one set of experimental conditions (−0.2 V versus SSE, 20 mg L−1 of BSA, pH 4) are presented.

7661

2. Experimental 2.1. Switch-flow cell setup 2.1.1. Global experimental setup A schematic view of the global experimental setup that was used for the EQCM experiments is given in Fig. 1(a). Two solutions, a protein-free solution and a protein-containing one, were pumped continuously from two vessels put in a thermostated bath using a 12-roll peristaltic pump (Ismatec) with two pump channels placed upstream from the flow cell. The pump has an adjustable speed; a low pumping speed of 3.3 mL min−1 was chosen so as to minimize the noise of the frequency and resistance signals. Both solutions flowed through a low-pressure mobile-phase selection valve (Rheodyne 5011) that allowed for instant switching of the inlet solution to the cell while minimizing mixing, without the creation of a measurable pressure gradient. Downstream from the selection valve, one solution went through the flow cell while the other one flowed directly to a recycling tank. The electrolytes were transported in highly chemically resistive Tygon® tubes of 1.6 mm inside diameter. To perform gravimetric measurements under electrochemical control, the flow cell was connected to a microbalance control unit (RQCM) and to a potentiostat, both devices being computer controlled. 2.1.2. Flow cell A schematic view of the flow cell is given in Fig. 1(b). The flow cell was a two-compartment and a three-electrode cell that allowed electrochemical and mass change measurements in deaerated solution. The cell design developed for this study was based on the Maxtek’s FC-550 flow cell that can be used with Maxtek’s CHC-100 crystal holder (made from CPVC). The cell is made from Kynar® and has two stainless steel inlet and outlet tubes with a 0.047 in. inside diameter × 0.062 in. outside diameter. A Viton® O-ring provides sealing between the cell and the front face of the sensor crystal. Installation of the cell on the crystal holder creates a cylindrical flow chamber of approximately 0.1 mL. The electrolyte flows parallel to the quartz surface from the inlet to the outlet tubes, both of which are positioned perpendicularly to the quartz crystal. The commercially available flow cell was modified into a three-electrode and two-compartment cell in order to allow for QCM experiments under electrochemical control. A large hole was drilled so as to introduce the counter electrode and a small one to create a Luggin capillary. The airtightness of the counter electrode compartment was ensured by a manufactured Kel-F® cap and an O-ring. Holes were made inside the cap for solution inlet and outlet, and for electrical connection. The working and reference electrode compartment was separated from the counter electrode compartment by a porous glass membrane fixed on the flow cell with epoxy glue. This rigid glass membrane does not modify the initial volume of the flow cell (∼0.1 mL) and thus, keeps short the solution switch time. The electrolyte first flows through the working electrode compartment, then it flows through the counter electrode compartment before being stored in a recycling tank. This

7662

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

Fig. 1. Experimental setup for EQCM protein adsorption studies in deaerated medium: (a) global setup, (b) flow cell.

design allows one to work in deaerated solution in both compartments unlike the cell used by Galliano et al. [6] where only the working electrode compartment could be deaerated. The working electrode was the pure chromium deposited on the front face of the quartz crystal. The counter electrode was a rolled up gold sheet of about 6 cm2 surface area. The saturated sulphate reference electrode (SSE) immersed in saturated K2 SO4 was in electrical contact with the working electrode compartment by means of a salt bridge made up of a glass three-line stopcock and a Luggin capillary (small hole in the Kynar® piece extended by a stainless steel cylindrical tube) coming very near to the working electrode. The mixing between the studied solution and the K2 SO4 solution was achieved by turning the glass stopcock. To decrease the impedance of the

potential measurement circuit and to improve the performances of the potentiostatic system, the salt bridge was shunted by an electrical circuit that connected the SSE to the metallic tube of the Luggin capillary through a low capacity (0.1 ␮F) [10]. 2.1.3. Cleaning Before each experiment, all the Kynar® , CPVC and KelF pieces were cleaned 5 min in a ultrasonic bath in a 1/3 ethanol + 2/3 deionised water mixed solution, then 5 min in a ultrasonic bath in deionised water, and finally air dried. After cleaning in the 1/3 ethanol + 2/3 water solution, 0.5 L of deionised water flowed through the glass membrane by means of a suction pump, so as to remove the protein desorbed from the glass pores during ultrasonication.

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

7663

2.2. Quartz crystals

2.5. X-ray photoelectron spectroscopy analyses

Polished AT-cut, 5 MHz, 1-in. diameter quartz crystals (Maxtek), with a nominal thickness of 333 ␮m, were sputter coated with (i) a 550 nm Cr layer on the front face, and (ii) a thin Cr adhesion layer and a 200 nm Au outer layer on the rear face. The chromium deposit on the front face was amorphous and metallic with a 99.99% purity. The exposed area of the front electrode was 1.37 cm2 , but the active oscillation region was limited to the overlapping area of the front and rear electrodes (0.34 cm2 ). As received quartz crystals were washed in absolute ethanol for 15 min in an ultrasonic bath, then air dried. Quartz crystals contaminated by the protein were cleaned in a 1 M NaOH solution during 2 h at 60 ◦ C, rinsed with deionised water for 15 min in an ultrasonic bath, then washed in absolute ethanol for 15 min in an ultrasonic bath, and finally air dried. It was checked by XPS that this cleaning procedure for contaminated crystals removed the protein from the surface and did not change significantly the thickness of the chromium oxide and hydroxide layers. Subsequently, the quartz crystal was clamped between the crystal holder (CHC-100, Maxtek) and the flow cell. Two gold pins inside the cavity of the crystal holder assured the electrical connection to the crystal front and rear electrodes, and two Orings prevented any contact between electrical connections and the solution. As the microbalance signals depend on the pressure applied to the crystal through the retainer cover, the same pressure was applied each time the cell was closed.

At the end of the EQCM measurements, the potential was switched off, and the cell was dismounted. After dismounting, the sample was rinsed in deionised water for a few minutes, then dried with clean compressed air, before introduction in the analysis chamber of the XPS spectrometer. XPS analyses were performed with a VG ESCALAB 250 spectrometer, using a monochromatized Al K␣ X-ray source. The calibration of the spectrometer was done with Cu 2p3/2 (932.9 eV), Au 4f7/2 (84.1 eV) and Ag 3d5/2 (368.6 eV). Base pressure during analysis was less than 10−9 Torr. The analysed area had a diameter of about 500 ␮m. The following XPS regions were recorded: Cr 2p, O 1s, N 1s, C 1s, S 2p, and Na 1s. To evaluate the degree of homogeneity of the surface layers, three different areas (at least, one situated at the centre and one at the edge of the Cr surface) were analysed for each sample. For quantification, Scofield cross sections [14] in combination with a measured transmission function were used. The data processing (background subtraction and peak fitting) was done with the AVANTAGE software from VG, using a Shirley type background and gaussian/lorentzian peak shapes. The binding energies were obtained from reference samples or from published data [15]. 3. Interpretation of experimental data 3.1. EQCM data

2.3. Solutions Two solutions were prepared in a 0.05 M Na2 SO4 supporting electrolyte: one without protein and another one with 20 mg L−1 of BSA. It was chosen to work with a comparatively high BSA concentration. As the protein changes conformation and behaviour with the pH, a mildly acidic pH of 4 was chosen. This pH is lower than the isoelectric point of the BSA in water at 25 ◦ C which is about 4.7–4.9 [11–13]. The pH of the solutions was adjusted to 4 with a diluted H2 SO4 solution. Both solutions were deaerated for at least 1 h by nitrogen bubbling in the vessels, and thermostated at 25 ◦ C. 2.4. EQCM measurements The oscillating circuit used was a Maxtek Research Quartz Crystal Microbalance (RQCM). This system allows for realtime and simultaneous recording of the quartz crystal frequency and resistance. The circuit also incorporates adjustable crystal capacitance cancellation, reducing error caused by parasitic capacitance of the crystal. This is essential for accurate measurements of lossy films, which is the case of adsorbed protein layer. The mass resolution is better than 0.4 ng cm−2 . Thanks to an optional data acquisition card, that can accommodate five scaleable analog inputs, the potential and current outputs of the potentiostat were registered simultaneously with the frequency and resistance data of the QCM. All data were logged and graphically displayed in real time using included Maxtek custom WindowsTM -based software.

The total frequency shift (f) that is measured experimentally is generated from the two contributions of mass loading (fm ) and viscous loading (fv ) [16]: f = fm + fv

(1)

The viscous loading is related to the density (ρ) and viscosity (η) of the solution, whereas the mass loading is related to gain or loss in weight, in our case to protein surface adsorption. The single measurement of the resonance frequency does not allow to separate the two contributions. This can be achieved by measuring the quartz crystal resistance, which depends linearly on viscous loading, in a similar way as fv [16–20]: fv = K1 (ρη)0.5

(2)

R = K2 (ρη)0.5

(3)

Thus, Martin et al. [16,18–20] found that, in the case of ideally smooth surfaces, rigidly attached films and Newtonian fluids, mass loading influences only the resonance frequency, whereas viscous loading affects both the frequency and the quartz crystal resistance near the resonance frequency. Therefore, the measurement of the resistance variation (R) can be used to obtain a real time correction for the viscous loading contribution, following: fm = f − K3 R

(4)

where K3 = K1 /K2 is a constant that is dependent on the crystal properties [6]. The theoretical value of K3 is K3theo = −2 Hz −1 , which correlates very well with the experimental

7664

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

value obtained from a two-point water/glycerol mix that gave exp K3 = −1.9 Hz −1 . This experimental value was used in the following. After correction of the total frequency variation for viscous loading, fm was converted to mass variation by unit area (m) using the theoretical Sauerbrey coefficient (C) of 0.0566 Hz cm2 ng−1 : fm = −C m

(5)

This mass variation is assumed to correspond only to BSA adsorption. Finally, the equivalent thickness of the adsorbed protein layer (dBSA (EQCM)) was estimated assuming a continuous layer and using the density of BSA in aqueous solution (ρBSA ) given as 1.36 g cm−3 [21]: dBSA (EQCM) =

m ρBSA

by the adsorbed BSA is given by: BSA sin θ INBSA = kA(θ)σN λBSA N TN DN    dBSA × 1 − exp − BSA λN sin θ

ox /I m The thickness of the oxide layer was calculated from the ICr Cr hy ox ratio, the thickness of the hydroxide layer from the ICr /ICr ratio, hy and the thickness of the BSA layer from the INBSA /ICr ratio:  ox Dm λm  ICr ox Cr (11) dox = λCr sin θ ln 1 + m Cr ox λox ICr DCr Cr

dhy =

hy λCr sin θ ln

3.2. XPS data For the analysis of XPS data, a three-layer model was assumed: an inner Cr2 O3 oxide layer (thickness: dox ) and an outer Cr(OH)3 hydroxide layer (thickness: dhy ) covered by an adsorbed BSA layer (thickness: dBSA ). In this work, these three layers were assumed to be continuous. The Cr 2p3/2 intensity emitted by the metallic chromium is expressed by:   dox m m m ICr = kA(θ)σCr λCr TCr DCr sin θ exp − ox λCr sin θ     dhy dBSA exp − BSA (7) × exp − hy λCr sin θ λCr sin θ The Cr 2p3/2 intensity emitted by the chromium oxide is given by:    dox ox ox = kA(θ)σCr λox T D sin θ 1 − exp − ICr Cr Cr Cr λox Cr sin θ     dhy dBSA exp − BSA (8) × exp − hy λCr sin θ λCr sin θ The Cr 2p3/2 intensity emitted by the chromium hydroxide is expressed by:    dhy hy hy hy ICr = kA(θ)σCr λCr TCr DCr sin θ 1 − exp − hy λCr sin θ   dBSA × exp − BSA (9) λCr sin θ The nitrogen signal comes only from the protein and is therefore a fingerprint of the adsorbed protein. The N 1s intensity emitted

  hy ox ox  ICr DCr λCr dox 1 + ox hy hy 1 −exp − ox ICr D λ λCr sin θ Cr

(6)

The value for ρBSA was shown to be almost independent of the temperature (between 20 and 30 ◦ C) [21,22] and the pH (for 2 < pH < 7) [23,24].

(10)

Cr

(12)    hy hy dhy INBSA DCr λCr TCr σCr 1 − exp − hy BSA BSA hy ICr DN λN TN σN λCr sin θ =

1 − exp(−dBSA /λBSA sin θ) N BSA exp(−dBSA /λCr sin θ)

(13)

In Eqs. (7)–(13), k is an instrumental constant, θ the take-off angle of the photoelectrons with respect to the sample surface, A(θ) the analysed area, σ X the photoionisation cross-section of element X, λM X the inelastic mean free path of the photoelectrons emitted by the X core level in the matrix M, TX the transmisM the bulk concentration of sion function for element X, and DX element X in the matrix M. The values of the different λM X were assessed by the TPP2M hy m ˚ , λox ˚ formula [25]. Thus, λCr = 16.1 A Cr = 18.5 A , λCr = BSA BSA ˚ , λN = 32.0 A ˚ , and λCr = 29.6 A ˚ . The values used 22.7 A m for the other constants were the following: θ = 90◦ , DCr hy ox = 0.06855 mol cm−3 , = 0.13827 mol cm−3 , DCr DCr = BSA −3 −3 0.02796 mol cm , DN = 0.01599 mol cm , TCr /TN = 1.083, σ Cr = 7.69, σ N = 1.80 [14]. 4. Results 4.1. EQCM measurements 4.1.1. Definition of the passive domain To define the passive domain of pure chromium in deaerated, pH 4, 0.05 M Na2 SO4 solution, polarization curves were plotted with the protein-free solution flowing through the cell, and with the protein-containing solution after the adsorption plateau was reached (Fig. 2). The start and end potential was the open circuit potential (OCP) that was stable from the very first minutes of immersion and reproducible (values comprised between −0.25 and −0.16 V versus SSE); the potential was first swept in the cathodic direction then in the anodic direction, with a sweep rate of 2 mV s−1 . It is shown that the adsorbed BSA has no influence on the passive domain, which corresponds to potentials ranking

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

Fig. 2. Polarization curves of pure Cr deposited on a quartz crystal, in deaerated, pH 4, 0.05 M Na2 SO4 solution (black curve) without and (grey curve) with adsorbed BSA. Sweep rate: 2 mV s−1 .

from −0.9 to +0.1 V versus SSE. This passive potential domain is similar to the one obtained by Olsson et al. for Cr in deaerated 0.1 M H2 SO4 + 0.4 M Na2 SO4 solution [26]. They found that Cr was in the passive state for potentials comprised between −0.745 and +0.255 V versus SSE. A passive potential of −0.2 V versus SSE was chosen to perform the adsorption experiments. This potential is very close to the OCP and is situated in the upper part of the passive domain. 4.1.2. Adsorption experiments It was initially decided to apply a cathodic prepolarisation at −1.65 V versus SSE for 5 min to clean the Cr surface and to start with a surface state as reproducible as possible [4,26,27]. Although, this prepolarisation was found to modify the Cr topography and hence to increase the nanoscale roughness [28,29], the amount of BSA adsorbed on the Cr surface was similar with or without cathodic cleaning. It was therefore decided to suppress this initial step for the following experiments. Adsorption experiments were performed as follows: the potential was increased from the OCP to −0.2 V versus SSE using a scan rate of 20 mV s−1 in the protein-free solution; after a stabilisation period of 10 min at −0.2 V versus SSE, the protein-containing solution was switched to the cell and the microbalance signals were recorded for at least 20 min. It was verified that adsorption experiments performed in the same experimental conditions gave reproducible results. In particular, EQCM results obtained with an as received quartz crystal were similar to those obtained with a quartz crystal cleaned in 1 M NaOH. Therefore, the cleaning in basic solution does not alter the behaviour of the quartz crystal and the Cr surface. The EQCM signals for one switch experiment are presented in Fig. 3. When introducing the protein in the cell, the resonance frequency begins decreasing a few seconds after the switch and it reaches a plateau after about 10 min. No great resistance shift (maximum 2 ) is observed during the whole experiments. It can be deduced from these results that the kinetics of BSA adsorption is fast and that a steady-state is reached about 10 min after introduction of the protein. The amount of BSA

7665

Fig. 3. (Black thick curve) total frequency shift, (grey thick curve) mass frequency shift, and (black thin curve) resistance of a quartz crystal coated with pure chromium for an adsorption experiment. The vertical dotted line indicates the switch between deaerated, pH 4, 0.05 M Na2 SO4 solution and deaerated, pH 4, 0.05 M Na2 SO4 + 20 mg L−1 BSA solution. For the sake of clarity all signals were graphically offset to zero at BSA introduction.

adsorbed on the Cr surface was estimated by application of Eqs. (4)–(6). After correction for viscous loading, the mass frequency shift for Fig. 3 is equal to fm = −24 Hz which corresponds to a mass variation of m = 424 ng cm−2 and to an equivalent thickness of the adsorbed BSA layer of dBSA (EQCM) = 3.1 nm. All the adsorption experiments gave similar results, and the mean value calculated from five replicated measurements is dBSA (EQCM) = 3.5 ± 0.8 nm. 4.1.3. Adsorption–desorption experiments To study the reversibility of BSA adsorption, adsorption–desorption experiments were performed. The adsorption experiment was the same as described above. Once the adsorption plateau was reached, the protein-free solution was again switched in the cell and the microbalance signals were recorded for at least 20 min. The results of one adsorption–desorption experiment are shown in Fig. 4. It can be seen that once a steady-state is reached in the protein-containing solution, the introduction of the protein-free solution does not lead to an increase of the mass frequency which means that BSA adsorption is an irreversible process or that the kinetics of desorption is very slow. 4.2. XPS analyses All the XPS analyses performed in the same experimental conditions gave reproducible results. In particular, XPS results obtained with an as received quartz crystal can be reproduced with a crystal cleaned in 1 M NaOH. Moreover, for one given sample, results from the three different analysed areas (at least, one situated at the centre and one at the edge of the Cr surface) were similar. The nitrogen and the carbon signals, except contamination, come only from the protein and are therefore a fingerprint of the biomolecule.

7666

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

Fig. 6. C 1s XPS core level spectrum for passivated chromium after an adsorption experiment (10 min in the protein-free solution and 20 min in the protein-containing solution at −0.2 V vs. SSE). (—) Experimental spectrum, and (– – –) peak decomposition. Fig. 4. Mass frequency shift of a quartz crystal coated with pure chromium for an adsorption–desorption experiment. The vertical dotted lines indicate the switches between the protein-free and the protein-containing solutions. For the sake of clarity the signal was graphically offset to zero at BSA introduction.

Fig. 5. N 1s XPS core level spectrum for passivated chromium after an adsorption experiment (10 min in the protein-free solution and 20 min in the protein-containing solution at −0.2 V vs. SSE).

Fig. 5 shows the N 1s XPS core level spectrum for passivated chromium after an adsorption experiment (10 min in the protein-free solution and 20 min in the protein-containing solution at −0.2 V versus SSE). The N 1s peak, centered at 400.3 eV,

is symmetric as expected for the amine or amide groups of the BSA [30]. From the XPS analysis of the chromium surface passivated during 30 min at −0.2 V versus SSE in the protein-free solution (not shown here), it was checked that nitrogen contamination at 400.3 eV was negligible in the absence of protein. The small peak observed at 397 eV does not come from the adsorbed protein since it is already present for the as-received quartz crystals. It would correspond to chromium nitride arising from the sputtering process. The C 1s XPS core level spectrum obtained under the same experimental conditions as the N 1s spectrum of Fig. 5 is presented in Fig. 6. The C 1s signal was fitted with three contributions corresponding to well identified carbon bonds present in the BSA molecule: the first peak named C1 , at the lowest binding energy of 285.0 eV, is attributed to C–C, C C, and C–H; the second peak named C2 , at 286.4 eV, is assigned to C–N and C–O single bonds; the third peak named C3 , at 288.4 eV, is attributed to O C–O and O C–N (peptide bonds) bonds [2,31,32]. The curve fit parameters that were used for the C 1s XPS spectrum are given in Table 1. From the nitrogen and carbon XPS intensities, different N/C atomic ratios can be estimated, in particular N/Ctotal where Ctotal = C1 + C2 + C3 , and N/(C2 + C3 ) that does not take into account the carbon contamination included in C1 . The values estimated for Cr put into contact with the protein are

Table 1 Parameters used for the decomposition of N 1s, C 1s, O 1s, and Cr 2p3/2 XPS spectra Position (eV) N 1s

FWHM

Tail mix (%)

Tail height (%)

Tail exponent

G/L (%)

1.41

C 1s

C1 C2 C3

285.0 286.4 288.4

1.36 1.36 1.36

100 100 100

0 0 0

0 0 0

50 50 50

O 1s

O2− OH− H2 O

530.0 531.8 533.5

1.30 1.90 1.90

100 100 100

0 0 0

0 0 0

50 50 50

Cr 2p3/2

Cr(met) Cr3+ (ox) Cr3+ (hy)

574.0 576.4 577.8

0.89 2.20 2.20

0 100 100

0 0 0

0.15 0 0

50 50 50

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

Fig. 7. O 1s XPS core level spectrum for passivated chromium after an adsorption experiment (10 min in the protein-free solution and 20 min in the protein-containing solution at −0.2 V vs. SSE). (—) Experimental spectrum, and (– – –) peak decomposition.

N/Ctotal = 0.20 and N/(C2 + C3 ) = 0.49. XPS analysis was also performed with BSA powder, and the following atomic ratios were obtained: N/Ctotal = 0.23 and N/(C2 + C3 ) = 0.50. The very good agreement between the values calculated for BSA powder and those assessed for Cr put into contact with the protein shows that BSA is adsorbed on the Cr surface and that it is chemically intact. The O 1s XPS core level spectrum obtained under the same experimental conditions as the N 1s and C 1s spectra of Figs. 5 and 6 (Fig. 7) shows that oxygen is present on the surface as oxide (O2− at 530.0 eV), hydroxide (OH− at 531.8 eV) and water (H2 O at 533.5 eV). The curve fit parameters are given in Table 1. The oxide peak can be attributed only to chromium oxide whereas the hydroxide peak corresponds to both chromium hydroxide and oxygen present in the BSA molecule (the O 1s XPS core level spectrum obtained with BSA powder (not shown here) exhibits one single contribution centered at 531.8 eV). The decomposition of the oxygen spectrum justifies the three-layer model that was assumed for the analysis of XPS data (an inner Cr2 O3 oxide layer, and an outer Cr(OH)3 hydroxide layer covered by an adsorbed BSA layer). The Cr 2p3/2 XPS core level spectrum for passivated chromium after an adsorption experiment is presented in Fig. 8. It was decomposed into three peaks: one at 574.0 eV relative to metallic chromium, one at 576.4 eV relative to Cr2 O3 , and one at 577.8 eV relative to Cr(OH)3 . This decomposition into three contributions (metal, oxide, hydroxide) was also used by other authors [33–35]. The curve fit parameters are given in Table 1. From the O2− and Cr3+ (ox) XPS intensities, the O/Cr atomic ratio in the oxide was estimated at about 1.54 which corresponds indeed to Cr2 O3 . This result justifies the distinction between chromium oxide and chromium hydroxide that was made in the decomposition of Cr 2p3/2 XPS spectrum. From the Cr 2p3/2 and N 1s signals, the equivalent thicknesses of the oxide, hydroxide, and adsorbed BSA layers were estimated by application of Eqs. (11)–(13). For a given sample, the values of these thicknesses obtained for the three different analysed areas were similar; therefore, the thicknesses of the three surface layers are homogeneous on the Cr surface. The

7667

Fig. 8. Cr 2p3/2 XPS core level spectrum for passivated chromium after an adsorption experiment (10 min in the protein-free solution and 20 min in the protein-containing solution at −0.2 V vs. SSE). (—) Experimental spectrum, and (– – –) peak decomposition.

mean values calculated from 12 points are: dox = 2.6 ± 0.4 nm, dhy = 1.4 ± 0.3 nm, and dBSA (XPS) = 3.5 ± 0.7 nm. 5. Discussion From the EQCM measurements, it can be concluded that the kinetics of BSA adsorption is fast, that a steady-state is reached about 10 min after introducing the protein, and that the adsorption process is irreversible. The amount of adsorbed protein estimated by EQCM is equal to m = 424 ng cm−2 which corresponds to an equivalent thickness of the BSA layer of dBSA (EQCM) = 3.5 ± 0.7 nm. Our results can be compared to those obtained by several authors with different substrates. Azioune et al. studied by QCM the adsorption of human serum albumin (HSA) on polypyrrole films at pH 7.4, without any electrochemical control [2]. They found a slower kinetics of adsorption and a lower mass variation (176 ng cm−2 ) than for the BSA/Cr system. However, they worked with a static cell. As part of surface induced immune complement and plasma coagulation studies, Berglin et al. showed by QCM that the adsorption of human serum proteins on poly(alkyl methacrylates) was fast and that a steady-state was reached after a few minutes; they also found that washing the protein-contaminated substrates with a protein-free solution led to partial desorption of the proteins [9]. Giacomelli et al. who studied the adsorption of BSA onto hydrophobic TiO2 surfaces by ellipsometry found that, for a BSA solution of low concentration (40 mg L−1 ) and pH 5.2, the adsorbed protein layer was thin and compact; thus, the observed thickness was about 2.5 nm and the calculated amount of adsorbed BSA was about 200 ng cm−2 [36]. H¨oo¨ k et al. determined by ellipsometry and optical waveguide lightmode spectroscopy an experimental mass of adsorbed HSA on TiO2 -coated substrates of about 188 ± 22 ng cm−2 , for 80 mg L−1 of HSA and pH 7.4 [37]. They also calculated an average thickness of the protein film equal to 2.8 nm. At last, Hughes Wassell and Embery estimated for BSA adsorption onto titanium powder an amount of adsorbed protein of about 65 ng cm−2 , for 20 mg L−1 of BSA and pH 5.15 [38]. Therefore, our estimated values of the amount and equivalent thickness of adsorbed BSA on passivated Cr are higher than those deduced

7668

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669

by other authors by QCM and other techniques for albumin adsorption on polymers or titanium oxide. The equivalent thicknesses of the adsorbed BSA layer estimated in situ by EQCM and ex situ by XPS are in excellent agreement, and are equal to 3.5 ± 0.7 nm. However, much water will leave the surface during transfer from electrolyte to ultra high vacuum (UHV). For the XPS estimate, nitrogen was used as a marker element for BSA. Thus, it is only necessary to assume that there is no desorption of nitrogen during transfer for the comparison to be correct. The in situ adsorption experiments were carried out under conditions where a stable passive film was formed. The oxide layer does not make substantial changes during transfer, whereas the hydroxide layer may thin considerably. The effects of transfer environment on passive films have been studied in detail by Courty et al. [39] and Bardwell et al. [40]. The dimensions of BSA molecules in aqueous solution are 4 nm × 13 nm for pH 4 [41]. If it is assumed that the protein adsorbs without major conformational changes, forming a layer of adsorbed molecules that maintained the original size and shape they had in solution, one monolayer “side-on” i.e. horizontally orientated [42] of BSA would be adsorbed on the passivated Cr surface. However, it is well known from kinetic and spectroscopic arguments that protein molecules can undergo structural and orientational changes following attachment to the surface. Moreover, the protein molecules may also change conformation after transfer from electrolyte to UHV. Giacomelli et al. deduced from their results, obtained for a low BSA concentration in solution of 40 mg L−1 and pH 5.2, that no significant conformational changes took place during the adsorption process onto hydrophobic TiO2 surfaces, and that the protein layer was formed by adsorption of the molecules in a “side-on” orientation [36]. The average thickness of the albumin film on TiO2 , calculated by H¨oo¨ k et al. for 80 mg L−1 of HSA and pH 7.4, was lower than that of one monolayer; they explained their result by surface induced denaturation of the protein and spreading on the surface [37]. The results of Giacomelli et al. and H¨oo¨ k et al. seem to be contradictory (no conformational changes and conformational changes, respectively) since they estimated similar thicknesses of adsorbed protein (2.5 and 2.8 nm, respectively) at pH corresponding to the same structure of the protein. However, the conclusions depend strongly on the dimensions of the protein that are taken into account at a given pH, and these dimensions can vary significantly from one author to another. 6. Conclusions A new design of a switch-flow cell in combination with a quartz crystal microbalance was developed for protein adsorption studies under electrochemical control and in deaerated solutions. It was used to study the adsorption of BSA on chromium passivated at −0.2 V versus SSE in pH 4 sodium sulphate solution. One original aspect of this work was the coupling between electrochemical quartz crystal microbalance (EQCM) and X-ray photoelectron spectroscopy (XPS) to study the BSA/Cr system in situ during adsorption and ex situ after adsorption.

EQCM measurements showed that the kinetics of BSA adsorption was fast, and that a steady-state was reached about 10 min after introducing the protein. They also showed that BSA adsorption was an irreversible process, or that the kinetics of desorption was very slow. The equivalent thicknesses of the adsorbed BSA layer estimated in situ by EQCM and ex situ by XPS are in excellent agreement, and are equal to 3.5 ± 0.7 nm, which corresponds to one horizontally orientated monolayer. Acknowledgements The authors wish to thank the students Sabrina Dubois, Olivier Bredy, Shujing Cheng, and Aur´elie Ithurbide for the development of the experimental procedure and for the carrying out of some experiments. References [1] D. Haras, Mat´eriaux et techniques 93 (2005) 27. [2] A. Azioune, F. Siroti, J. Tanguy, M. Jouini, M.M. Chehimi, B. Miksa, S. Slomkowski, Electrochim. Acta 50 (2005) 1661. [3] K. Ogle, S. Weber, J. Electrochem. Soc. 147 (5) (2000) 1770. [4] D. Hamm, K. Ogle, C.-O.A. Olsson, S. Weber, D. Landolt, Corros. Sci. 44 (2002) 1443. [5] C. Garcia, G. Courbin, F. Ropital, C. Fiaud, Electrochim. Acta 46 (2001) 973. [6] F. Galliano, C.-O.A. Olsson, D. Landolt, J. Electrochem. Soc. 150 (11) (2003) B504. [7] M. Rodahl, B. Kasemo, Sens. Actuators A 54 (1996) 448. [8] P. Rossini, P. Colpo, G. Ceccone, K.D. Jandt, F. Rossi, Mater. Sci. Eng. C 23 (2003) 353. [9] M. Berglin, M. Andersson, A. Sellborn, H. Elwing, Biomaterials 25 (2004) 4581. [10] C.C. Hermann, G.C. Perrault, A.A. Pilla, Anal. Chem. 40 (1968) 1173. [11] R.M.C. Dawson, et al., Data for Biochemical Research, 3rd ed., Clarendon Press, Oxford, 1993, p. 381. [12] D. Malamud, J.W. Drysdale, Anal. Biochem. 86 (1978) 620. [13] P.G. Righetti, T. Caravaggio, J. Chromatogr. 127 (1976) 1. [14] J.H. Scofield, J. Electron Spectrosc. Relat. Phenom. 8 (1976) 129. [15] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, Physical Electronics Inc., Eden Prairie, MN, 1995. [16] S.J. Martin, V.E. Granstaff, G.C. Frye, Anal. Chem. 63 (1991) 2272. [17] K.K. Kanazawa, J.G. Gordon, Anal. Chim. Acta 175 (1985) 99. [18] S.J. Martin, G.C. Frye, A.J. Ricco, S.D. Senturia, Anal. Chem. 65 (1993) 2910. [19] H.L. Bandey, S.J. Martin, R.W. Cernosek, A.R. Hillman, Anal. Chem. 71 (1999) 2205. [20] S.J. Martin, H.L. Bandey, R.W. Cernosek, A.R. Hillman, M.J. Brown, Anal. Chem. 72 (2000) 141. [21] M.J. Hunter, J. Phys. Chem. 70 (10) (1966) 3285. [22] M.J. Hunter, J. Phys. Chem. 71 (12) (1967) 3717. [23] R. Jaenicke, M.A. Lauffer, Biochemistry 8 (7) (1969) 3077. [24] K. Gekko, A. Kimoto, T. Kamiyama, Biochemistry 42 (2003) 13746. [25] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interface Anal. 21 (1993) 165. [26] C.-O.A. Olsson, D. Hamm, D. Landolt, J. Electrochem. Soc. 147 (7) (2000) 2563. [27] D. Hamm, C.-O.A. Olsson, D. Landolt, Corros. Sci. 44 (2002) 1009. [28] V. Maurice, W.-P. Yang, P. Marcus, J. Electrochem. Soc. 141 (1994) 3016. [29] D. Zuili, V. Maurice, P. Marcus, J. Phys. Chem. B 103 (1999) 7896. [30] Y.F. Dufrˆene, T.G. Marchal, P.G. Rouxhet, Appl. Surf. Sci. 144–145 (1999) 638.

I. Frateur et al. / Electrochimica Acta 52 (2007) 7660–7669 [31] C.D. Tidwell, D.G. Castner, S.L. Golledge, B.D. Ratner, K. Meyer, B. Hagenhoff, A. Benninghoven, Surf. Interface Anal. 31 (2001) 724. [32] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers—The Scienta ESCA 300 Database, Wiley, Chichester, UK, 1992. [33] I. Olefjord, B. Brox, U. Jelvestam, J. Electrochem. Soc. 132 (1985) 2854. [34] I. Olefjord, L. Wegrelius, Corros. Sci. 31 (1990) 89. [35] S. Haupt, H.-H. Strehblow, J. Electroanal. Chem. 228 (1987) 365. [36] C.E. Giacomelli, M.J. Esplandi´u, P.I. Ortiz, M.J. Avena, C.P. De Pauli, J. Colloid Interface Sci. 218 (1999) 404.

7669

[37] F. H¨oo¨ k, J. V¨or¨os, M. Rodahl, R. Kurrat, P. B¨oni, J.J. Ramsden, M. Textor, N.D. Spencer, P. Tengvall, J. Gold, B. Kasemo, Colloids Surf. B 24 (2002) 155. [38] D.T. Hughes Wassell, G. Embery, Biomaterials 17 (1996) 859. [39] C. Courty, H.J. Mathieu, D. Landolt, Electrochim. Acta 36 (1991) 1623. [40] J.A. Bardwell, G.I. Sproule, D.F. Mitchell, B.R. Macdougall, M.J. Graham, J. Chem. Soc. Faraday Trans. 87 (1991) 1011. [41] D.C. Carter, J.X. Ho, Adv. Protein Chem. 45 (1994) 153. [42] A. Baszkin, D.J. Lyman, J. Biomed. Mater. Res. 14 (1980) 393.