Silicon-titanium oxide grating substrates (depth 20 nm, period 2,400 lines/mm) were supplied by MicroVacuum Ltd (Budapest, Hungary) and consist on a 170 nm thick SiO₂-TiO₂ oxide layer (n_f = 1.77) onto a glass substrate (n_s = 1.53). All refractive indices have been measured at 632.8 nm. The grating area dimensions were 2 mm in width per 12 mm in length on a total chip size of 12 mm in length, 8 mm in width and 0.5 mm thick. Previously to be coated with silicon nitride, the gratings were thoroughly cleaned with organic solvents and Milli-Q water.

Sodium cyanoborohydride, Sodium hydroxide, Ethanolamine, Glycine and HSA (Albumin solution from human serum 30% in 0.85% sodium chloride) were purchased from Sigma-Aldrich (USA). The triethoxysilane aldehyde (TEA), which was obtained from United Chemical Technologies (USA), was stored in the dark, under argon and at room temperature. Anti-HSA rabbit monoclonal antibodies (IgG) having a molecular weight of about 150 kDa were purchased from AntibodyBcn (Spain). Antibodies were prepared in phosphate buffer saline solution (0.01 M phosphate buffer, 0.0027 M potassium chloride and 0.137 M sodium chloride, PH 7.4). Hydrochloric acid was purchased from Merck (Germany).

The structure and measuring principle of the optical grating coupler sensor chip are depicted in Figure 1. A He–Ne laser (λ = 632.8 nm) is diffracted by the grating and, at a characteristic incident angle, there is a constructive interference (phase shift of the internal reflection is zero) that excites a guided mode, then the light propagates through the waveguide via total internal reflection and an evanescent field is generated onto the covering medium.

The change in the refraction index of the covering layer produced by the analyte recognition can be monitored on line by measuring the changes in the incoupling angle [10,25,26]. In this case, a commercial OWLS instrument from MicroVacuum Ltd. (Budapest, Hungary) was used. This system software measures the effective refractive index on the zero transverse electric (TE) and magnetic (TM) modes and converts them directly to adsorbed layer mass and thickness by assuming an optical uniform layer and a refractive index varying linearly with mass density [27] upon using Feijter’s Formula [28]. A full description of the technology and measuring principle is included available in [9,10].

Silicon nitride thin films were deposited by r.f. magnetron sputtering (Edwards Coating System ESM100) on titanium oxide gratings and silicon wafer substrates. Film deposition was carried out by sputtering of a pure Si₃N₄ target with an r.f. input power of 100 W and using argon as a sputtering gas. The cathode magnetron was 100 mm in diameter, the substrate was rotated in relation with the target and the substrate-target distance was kept at 5 cm. Before deposition, the chamber was evacuated to 1.5 × 10⁻⁵ mbar and then, backfilled to a working pressure of 5.0 × 10⁻³ mbar. Film thickness and refractive index of the film were determined by ellipsometry (monochrome He-Ne PLASMOS) on the substrates deposited on silicon. The optimal thickness selected for the silicon nitride layer is 10 nm, in order to get a layer thick enough to be uniform and at the same time, thin enough to be transparent and to avoid changing significantly the effective refractive index of the waveguide [11]. For the film required thickness of 10 nm, the sputtering deposition time was established at 3 min.

Silicon nitride coated optical grating coupler sensor chips were characterized by X-Ray Photoemission spectroscopy (XPS) after their activation to be used in immunosensing. XPS spectra were recorded in a Perkin-206 Elmer PHI 5500 Multitechnique System from Physical Electronics (Waltham, MA, USA) with a monochromatic X-ray source (Aluminum KR line of 1,486.6 eV energy and 350 W), placed perpendicular to the analyzer axis and calibrated using the 3d5/2 210 line of Ag with a full width at half-maximum (FWHM) of 0.8 eV. No surface sputtering was used previous to the surface spectra. The resolution selected for the spectra was 187.5 eV of pass energy and 0.8 eV/step. All measurements were taken in an ultra high vacuum (UHV) chamber pressure between 5 × 10⁻⁹ and 5 × 10⁻⁸ mbar. When necessary, a low energy electron flood gun (0–3 eV) was used to discharge the samples. Peak fitting was performed using MultiPak 220 V 6.0 A software from Physical Electronics Inc. (Chanhassen, MN, USA).

Atomic Force Microscopy (AFM) was used to characterize the topography of the optical gratings before and after the silicon nitride deposition. The measurements were performed in a Dimension 3100 AFM instrument (Veeco Instruments, USA) equipped with a rectangular silicon AFM tip (MikroMasch NSC18/AlBS) with a spring constant of 3.5 N/m, a radius of curvature about 10 nm and a resonance frequency around 75 kHz. The instrument was operated in Tapping mode, in air media and room temperature. The topographic images obtained were analyzed by using the free WSxM software (Nanotec Electrónica, Spain).

To test the capability of the new structure to work as an immunosensor, standard antibody-antigen detection experiments were performed. The pair anti-HSA/HSA was selected for this purpose, as a model. In order to functionalize the silicon nitride coated OWLS chips for the antibody immobilization, the procedure described by Caballero et al. was followed [29] (Figure 2). Briefly, after cleaning with organic solvents and Milli-Q water, the silicon nitride coated chip was immersed in Piranha solution (1:3 v/v H₂O₂:H₂SO₄; Caution:Piranha is an extremely strong oxidant and should be handled very carefully) at 90 °C for 30 min. Afterwards, sequential immersion in aqueous solutions of NaOH (0.5 M) for 20 min., HCl (0.1 M) for 10 min. and a final immersion in NaOH (0.5 M) solution for 10 min. were performed to activate the surface. The samples were then rinsed thoroughly with HCl and Milli-Q water and dried in an oven at 100 °C for 20 min. Then, the chip was functionalized with triethoxysilane aldehyde (TEA) by the vapor-phase method for 1 hour and cured into the oven for 1 h at 100 °C. Afterwards, the samples were rinsed with absolute ethanol and dried under nitrogen. Once functionalized with the organosilane, the antibodies were immobilized on the chip surface by immersing it into a 10⁻⁷ M anti-HSA monoclonal antibody PBS solution (0.01 M phosphate buffer, 2.7 × 10⁻³ M potassium chloride and 0.137 M sodium chloride, pH 8.4), containing 4 mM sodium cyanoborohydride and allowed to react for 1 h at 37 °C. Afterwards, it was thoroughly rinsed with PBS to remove antibody excess. Subsequently, the aldehyde free surface groups were blocked by a solution of ethanolamine (100 mM ethanolamine in 10 mM PBS, pH 8.4) in the presence of 4 mM cyanoborohydride for 1–2 h at room temperature. Finally, the surface was thoroughly rinsed with PBS, pH 8.4. These antibodies will later recognize the HSA protein from the flowing solution.

The OWLS immunosensor system was operated in a continuous flow mode during sample measurement. For the sensing experiments, 500 μL of increasing concentrations (from 10⁻¹³ M to 10⁻³ M) of the HSA in PBS buffer were injected into the OWLS flow cell. The flow rate of the solution was kept at 30.4 μL/min and the on line sensor response was continuously collected as a surface mass change (ng/cm²). After the injection of each analyte concentration, the system was kept in stop-flow mode to favour the analyte incubation for 30 min. Then, the sample was rinsed by flowing PBS solution for 20 min to eliminate unspecific HSA adsorption. The changes of the surface mass during the measurement cycles were automatically calculated by the software implemented in the measurement system through the experimental values of the incoupling angles of the zero transverse electric and magnetic modes.

The deposition time and the applied bias of the sputtering process were optimized in order to obtain the best nitride layer in terms of high transmittance in the visible range and in terms of surface smoothness. After deposition, silicon nitride film thickness and refractive index were determined by ellipsometry measurements as 10.3 (±0.1) nm and 2.22 (±0.10), respectively. This refractive index value is higher that the corresponding to a stoichiometric silicon nitride film, which is 2.02 at 300 K [16].

The effects in the chemical composition of the outmost part of the silicon nitride layer produced by the chemical activation process with the NaOH were investigated by means of XPS. Figure 3(a) plots the surface full XPS spectrum of the activated layer, where it can be observed that there is a huge content of oxygen (up to 34% in atomic content). These results agree with the literature, where Si₃N₄ surfaces have been oxidized following the same procedure [30]. Moreover, it can be also observed that the ratio Si:N does greatly exceed the stoichiometric ratio of the silicon nitride, being in this case Si_6.8N₄. In order to investigate this in more detail, a high-resolution spectrum of the 2p Si peak was performed and analyzed. The peak could be deconvoluted into three contributions (Figure 3(b)), coming from three different oxidation states corresponding to silicon nitride, silicon oxide and silicon [30,31]. These results prove the presence of an oxygen-rich surface that will be then further functionalized by the TEA reagent.

Prior to the silicon nitride layer deposition, the grating substrate was characterized by means of atomic force microscopy (AFM). Figure 4(a) shows 2D topography AFM images of the grating and a line profile. As it can be seen, the grating has a period of 450 nm, a grafting apparent depth of around 7 nm and a RMS value on the flat top surfaces of 0.7 nm. Due to the AFM tip convolution, it is likely that the measured depth of the grating is underestimated. After the deposition of the silicon nitride layer, Figure 4(b), it can be noticed that surface roughness has increased up to 2.6 nm (RMS value), presenting a grainier morphology, while the grating structure remained barely altered (period of 450 nm and apparent depth of 7 nm). Once the chip was functionalized with the anti-HSA antibodies, the morphology of the grating was again characterized by the AFM technique. Figure 4(c) shows the changes produced by the organic layer: the grating structure period remains unaltered (450 nm) but the structure depth decreases slightly (5–6 nm) and the surface becomes again smoother, with RMS values around 1.8 nm.

Once fabricated, the functionalized silicon nitride chip performance in the OWLS instrument was checked. Figure 5 shows the plot of OWLS incoupling angle spectra after the silicon nitride layer deposition. It can be seen that both TE and TM peaks are perfectly visible after the silicon nitride deposition and the signal to noise ratio increases from 41 to 56 in the OWLS spectra when the chip is functionalized with antibodies with respect to the bare Si₃N₄ coated chip. This is related to the decrease in the roughness as a flatter surface avoids scattering during the incoupling of the laser. The same figure shows the OWLS incoupling angle peaks after the chip functionalization with the anti-HSA antibody layer. Worth noticing, the functionalization of the chip did not affect the chip structure neither covered the whole chip surface, allowing the coupling of the laser. The chips may be used a minimum of three times, maintaining their properties before the nitride layer gets damaged.

Human serum albumin detection experiments were performed in order to establish the biosensing viability of the whole developed system. The albumin concentration in the blood serum of an average adult human is 42.0 ± 3.5 mg/mL, (~0.6 mM) with an interval of 35–50 mg/mL [32,33]. Hyperalbuminemia and hypoalbuminemia are pathogenesis related with high or low HSA concentrations in body fluids, respectively. The immunosensing experiments have been specifically design to show high sensitivity and dynamic range performance in the range of interest. HSA was recognized by the anti-HSA antibody attached to the silicon nitride coated OWLS waveguide chip. The system proved to be stable; responses did not decrease significantly during the measurement. A typical curve of the measurements performed with the system is shown in Figure 6. Measurements started by stabilizing the system by flowing PBS for two hours approximately in order to obtain a steady baseline. Then, solutions with increasing HSA protein concentrations within the range of 10⁻¹³ M–10⁻³ M were injected in the cell. After each injection, the system was kept in stop-flow mode in order to stabilize the signal, and then it was thoroughly rinsed with PBS in order to remove all the protein non-specifically adhered to the chip surface. The system response was sigmoidal with the protein concentration, as predicted by the Nerst equation and the Langmuir adsorption theory [34]. The system proved to be very stable, achieving repeatable results with an error of less than 10%. Considering the amount of antibody captured probes used on the immobilization procedure (equivalent to 104 pmol/cm²) and taking into account that an antibody perfect monolayer would have a density of is 1.7 pmol/cm² (antibodies are around 10 nm in diameter [35]) the antibody amount used in the immobilization process is well in excess. If the recognition ratio between the protein and the antibody is 0.37:1 [36], the maximum saturation value of protein to be recognised for an ideal antibody monolayer would be 336 ng/cm². Thus, the experimental data measured for the adsorbed protein in the detection range assayed, shown in Figure 6, is far from saturation, therefore confirming that the biosensing system is well designed to work in the desired range.

The relationship between the adsorbed mass, attributed to the HSA protein and the protein concentration plotted in a semi-logarithmic scale is shown in Figure 7. The experimental mass values were fitted using a sigmoidal fit with a correlation factor of R² = 0.9969. The graphic also suggests a linear fit within the range of 10⁻³ M to 10⁻⁶ M with a R² value of 0.99689. The sensitivity of the system within the dynamic range is 34 ng/cm² for the HSA detection.

These results show a biosensing system with a dynamic working range particularly well fitted and designed for the specific application, with a LOD that, despite not reaching the values reported by similar studies [37], is three orders of magnitude better than needed for diagnosis of hyper/hypoalbuminemia and far more stable. Furthermore, our system presents a very good sensitivity and the possibility of extracting values of adsorbed mass, in contrast with previous studies [37] where was only possible to measure changes in refractive index units (RIU). This model study of the detection and quantification of HSA, demonstrate the potential of the Si₃N₄ covered grating sensor chips. The ability of the silicon nitride to be the functional layer in a real-time biosensor, renders it a promising material for integration of optical waveguides with microelectronics; for instance, the integration of optical waveguides into printed-circuit boards, or into field effect transistors.