Silicon wafers were purchased from Siltronix, Archamps, France. All cleaning and etching reagents were of VLSI grade and supplied by Merck. Other chemicals were purchased from Sigma-Aldrich (Munich, Germany), Acros Organics (Geel, Belgium) or Fluka (Buchs, Switzerland) and were of the highest purity available. The 10 × PBS buffer (pH = 7.4) was obtained from Ambion (Darmstadt, Germany). Ultrapure water (MilliQ Billerica, MA, USA; 18.2 MΩcm) was used for solution preparation and rinses.

The silicon samples of 15 × 15 mm² size were cut from double-side polished (100) oriented p-type silicon wafers boron doped, 0.08-0.12-Ωcm resistivity and were cleaned in 3:1 96% H₂SO₄//30% H₂O₂ (piranha solution) for 15 min at 100°C and copiously rinsed with MilliQ water. The native oxide was removed by immersing the samples in 50% aqueous HF for 1 min. The hydrogen-terminated surfaces were electrochemically etched in a 1/1 50% HF/absolute ethanol mixture for 30 s at a current density of 80 mAcm^-2. The prepared PSi surface was rinsed with MilliQ water and dried under a nitrogen stream.

The freshly prepared PSi sample was transferred into a Schlenk tube containing neat undecylenic acid under argon bubbling and allowed to react at 150°C for 16 h. The PSi surface was subsequently rinsed twice for 30 min in an outgassed Schlenk tube containing acetic acid at 75°C and was blown dry under a nitrogen stream. The surface, now bearing acid terminations, was introduced into a Schlenk tube containing a solution mixture of 5 mM EDC and 5 mM NHS and allowed to react under continuous argon bubbling for 90 min in a water bath at 15°C. The resulting succinimidyl-ester-terminated surface (activated surface) was copiously rinsed with water and dried under a nitrogen stream. The activated surface was immersed in an outgassed Schlenk tube containing a solution of 10^-4 M GlyHisGlyHis peptide in 1 × PBS buffer at pH approximately 7, overnight. The resulting surface was thoroughly rinsed and dried.

The Fourier transform infrared (FT-IR) spectra were recorded using a Bruker (Equinox 55) spectrometer (Ettlingen, Germany) equipped with a deuterated triglycine sulphate detector. The samples were mounted in a purged sample chamber in transmission geometry at normal incidence. All FT-IR spectra were collected with 200 scans in the 900-4,000 cm^-1 spectral region at 4 cm^-1 resolution. Background spectra were obtained by using an untreated deoxidised flat silicon wafer mounted in the same geometry.

The X-ray photoelectron spectroscopy (XPS) spectra were obtained with a Thermo Electron VG ESCALAB 220i XL spectrometer (Thermo Electron Corporation, Waltham, MA, USA), using an Al Kα₁ monochromatic X-ray excitation, and providing an overall full width at half-maximum (fwhm) energy resolution of 0.31 eV.

All glassware was rinsed with 6 M HNO₃, then thoroughly with MilliQ water to avoid metal ion contamination.

The copper ions were accumulated at the GlyHisGlyHis-modified PSi electrode at open circuit potential by dipping the sample into 10 mL of a stirred aqueous solution of Cu(II) sulphate in acetate buffer (pH = 8) for 15 min. The sample was removed from the solution, thoroughly rinsed with MilliQ water, dried under a nitrogen stream and transferred to the electrochemical cell.

The electrochemical measurements were performed with an Autolab potentiostat using a three-electrode electrochemical cell comprising the modified PSi as working electrode, a platinum wire counter electrode and an Hg/Hg₂SO₄ reference electrode. The electrolyte was copper-free ammonium acetate at pH 4 adjusted with HCl. The solution was degassed with argon for 15 min prior to data acquisition. Cyclic voltammetry was performed at different sweep rates between -800 and 0 mV.

In the spectrum of the freshly prepared PSi layer (Figure 1a), one observes peaks characteristic of the SiH stretching modes, namely, the bands at 2,085 cm^-1, 2,115 cm^-1and 2,140 cm^-1 ascribed to monohydride, dihydride and trihydride contributions, respectively 16. The absence of any sizeable contribution in the 1,000-1,200 cm^-1 range demonstrates that the PSi surface is oxide free 19. The peak around 910 cm^-1 corresponds to the deformation vibrations mode of Si-H₂ (scissor deformation) 20. After reaction with undecylenic acid (Figure 1b), the signature of acid chains grafted at the surface appears clearly. It consists of the contribution of the methylene backbone (symmetric and antisymmetric CH₂ stretching mode at 2,855 and 2,925 cm^-1, respectively, and CH₂ scissor deformation mode at 1,465 cm^-1), and that of the carboxyl groups (C=O stretching mode at 1,715 cm^-1 and the C-O-H modes at 1,280 and 1,415 cm^-1) 121. After treatment of the obtained acid surface in EDC/NHS solution 22, a prominent triplet appears (Figure 1c) attesting the formation of a succimimidyl-ester termination. The main peak of this triplet at 1,740 cm^-1 is ascribed to the antisymmetric stretching mode of the carbonyl groups of the succinimide cycle. The smaller peak at 1,785 cm^-1 is ascribed to the corresponding symmetric mode, and that at 1,820 cm^-1 is attributed to the ester C=O stretch 2223. Other characteristic bands of the terminal succinimidyl ester group include that with corresponding to the antisymmetric and symmetric stretching of the C-N-C group at 1,205 and 1,370 cm^-1, and the C-O(-N) stretching vibration at 1,065 cm^-1 2223. After amidation (Figure 1d), the bands corresponding to the terminal succinimidyl ester group disappear and two broad characteristic bands are observed at 1,650 and 1,550 cm^-1, commonly labelled amide I (νC=O) and amide II (δNH) 123. The PSi surface remains essentially oxide free and the SiH stretching band intensities have decreased due to the partial substitution of the surface SiH species by the grafted chains.

Figure 2 shows the C1s high-resolution XPS spectrum of the GlyHisGlyHis-modified PSi surface. This spectrum shows a peak centred at 285.4 eV with a fwhm of 1.4 eV and a weaker peak at 289 eV. The shoulders observed on either side of the main peak are indicative of the presence of carbon atoms in different environments. The signal can be deconvoluted into six peaks attributed to the different contributions of the carbon atoms. A tentative depiction of the types of carbon atoms that are distinguishable by XPS is shown in Figure 2. The weak peak at 284.5 eV is attributed to the carbon (shown as i) bonded to silicon. The CH₂ moieties (shown as ii) in the alkyl chain are represented by two peaks. The first one at 285.2 eV is for the nearest carbon atoms from the PSi substrate and the second peak at 285.7 eV is for the carbon atoms close to the attached peptide 24. The peak at 286.2 eV consists of the carbon atom of the alkyl chain (shown as iii) directly bonded to the peptide and the carbon atoms C=CH-N in the imidazole cycles 2425. The peak at 287 eV is ascribed to the carbon atoms (shown as iv) adjacent to amide functions and the carbon atoms (N=CH-N) in the imidazole cycles. Finally, the contribution at high binding energy (289 eV) is assigned to the acid and amide carbons (shown as v) 2426.

Figure 3 shows the reaction scheme of copper complexation on the GlyHisGlyHis-modified PSi during the accumulation step. The GlyHisGlyHis-modified PSi electrode is electrochemically inactive (Figure 4a) in the absence of copper (II). After copper accumulation for 15 min in a 0.1 mM Cu²⁺ solution and washing, the voltammogram recorded in a buffer solution that was free of copper exhibits cathodic and anodic peaks attributed to the quasi-reversible process of the Cu(I)/Cu(II) couple of copper chelated by the GlyHisGlyHis peptide immobilised on the PSi surface (Figure 4b).

Cyclic voltammetry is an efficient method to extract kinetic parameters such as heterogeneous rate constants k° and charge transfer coefficients α for surface immobilised redox species by examining the variation of peak potential versus experimental time scale (i.e. scan rate). Data analysis relies on a theoretical methodology developed by Laviron et al. 27.

The degree of kinetic reversibility displayed by a surface redox reaction depends on the scan rate. It is expected 28 that a surface redox reaction will exhibit a reversible behaviour (manifested by a peak potential variation quasi-constant with logarithm of scan rate (ln v) when the scan rate is small, and an irreversible behaviour (indicated by a linear variation of peak potential with ln v) when the scan rate is large. This general prospect was confirmed in our experiments. When the scan rate is higher than 0.02 Vs^-1, the cathodic peak potential E_pc shifts negatively and the anodic peak potential shifts positively with increasing scan rate. Figure 5 shows plots of the anodic peak potential as a function of the logarithm of the scan rate for a GlyHisGlyHis-modified PSi surface after copper accumulation. This figure shows that the peak potentials are practically invariant when the scan rate is low and in contrast for high scan rate, the peak potentials vary linearly as a function of ln v.

The heterogeneous scan rate k° and the charge transfer coefficient α can be determined using the following equations for the cathodic and anodic peak potentials:

Where E° is the standard potential of the surface redox species, v is the scan rate (volt/second), n is the number of transferred electrons, R is the ideal gas constant, T is the temperature and F is the Faraday constant.

These equations have been established by Laviron considering the limiting conditions where the reaction is totally irreversible. He considered that this case corresponds to the experimental condition where δE_p > 200 mV 27, where δE_p denotes the peak potential separation. In our case, this condition is fulfilled for scan rates above 0.2 Vs^-1. The data E_p,a = f(ln v) of Figure 5a are replotted as Figure 5b by considering the highest scan rates. The plot yields a straight line with a slope equal to RT/(1 - α)nF deduced from Eq. 2 and using the anodic potential peak. The value determined for α is 0.77.

On the basis of Eqs. 1 and 2, the heterogeneous rate constants k° can be calculated with the help of the following expression:

which is valid for E_p > 200 mV. The calculated k° value is 1.56 s^-1.

The dependence of the cyclic voltammetry current density at the GlyHisGlyHis-modified PSi on Cu²⁺ concentration in the accumulation solution was calibrated (Figure 6). Copper ions were accumulated at the GlyHisGlyHis-modified SiP electrodes at open circuit potential by immersing the electrodes into 10 mL of stirred aqueous solutions of copper (II) sulphate of different concentrations (10^-7, 10^-6, 10^-5, 10^-4 and 10^-3 M) in acetate buffer (pH = 8) for 20 min. The electrode was then removed, rinsed with copper-free ammonium acetate solution and transferred to a cell with ammonium acetate electrolyte (pH = 4) for cyclic voltammetry.

Figure 6 shows that the relation between current and concentration is clearly non-linear but does follow a "Langmuir" relation. Experimental data for the different Cu²⁺ concentrations and peak currents were fitted using the following Langmuir equation:

Where I_∞ is the limiting current density corresponding to the saturation of the surface by copper ions, K is the pseudo-adsorption coefficient which represents the apparent stability constant of the complex formed on the PSi surface by binding of copper to the GlyHisGlyHis peptide and C is the Cu²⁺ concentration in the accumulation solution.

The Langmuir curve (solid line in Figure 6) gives a good fit of the experimental data. The value of the limiting current density obtained is 13.44 μA cm^-2 and the apparent stability constant of the complex Cu-GlyHisGlyHis formed on the PSi surface is K = 3 × 10⁵ M^-1.

The value of I_∞ gives an indication on the sensitivity of the sensor, which has implications for the detection limit whilst the value of K is indicative of the affinity of the peptide for the metal ion and hence determines the usable concentration range of the sensor. As a consequence of the high affinity constant for Cu-GlyHisGlyHis, the final sensor is expected to operate in a low concentration range with a low detection limit.