An Advanced, Silicon-Based Substrate for Sensitive Nucleic Acids Detection

Surface substrate and chemical functionalization are crucial aspects for the fabrication of the sensitive biosensor based on microarray technology. In this paper, an advanced, silicon-based substrate (A-MA) allowing enhancement of optical signal for microarray application is described. The substrate consists in a multilayer of Si/Al/SiO2 layers. The optical signal enhancement is reached by a combination of the mirror effect of Al film and the SiO2 thickness around 830 nm, which is able to reach the maximum of interference for the emission wavelength of the Cy5 fluorescent label. Moreover, SiO2 layer is suitable for the immobilization of single-strand DNA through standard silane chemistry, and probe densities of about 2000 F/µm2 are reached. The microarray is investigated in the detection of HBV (Hepatitis B Virus) pathogen with analytical samples, resulting in a dynamic linear range of 0.05–0.5 nM, a sensitivity of about 18000 a.u. nM−1, and a Limit of Detection in the range of 0.031–0.043 Nm as a function of the capture probe sequence.


Introduction
Microarrays (MAs) are one of the most powerful and widely used biotechnologies for nucleic acids detection and are particularly relevant in the field of genomics [1]. Currently, MAs are attracting great interest among biologists, biochemists, and biomedical researchers, mainly for gene expression evaluation and mutation detection [2,3]. Besides detecting nucleic acids, MAs are also widely used to analyze proteins [4,5], carbohydrates [6], and cells [7]. The strengths of this technology are high throughput, miniaturization, and multiplexing analysis [8,9]. The integration of microarray technology with microfluidics, microelectronics, MEMS, and optical technologies has led to the development of modern and advanced miniaturized Lab-on-chip (LoC) devices [10][11][12]. In this context, silicon material offers several advantages for creating technological advancements in smart LoC devices. Actually, it combines significant physical properties (such us good electrical, thermal, and photoconductivity) with important technological aspects including consolidated production technologies, industrialization processes, and the integration of "intelligence on board" through microelectronic circuitry [13][14][15].
MA consists of a large amount of arrayed capture probes anchored by a proper chemistry to the surface of a solid substrate. In case of DNA detection, the capture probes are single-strand oligonucleotides, which are generally 25 mer in length (ssDNA microarray). The fabrication processes can be summarised in two steps: (a) in situ synthesis of the capture probe and (b) the microspotting of pre-synthesized capture probes by contact or noncontact printing technologies [16]. In this context, the probe immobilization strategy plays a key role, strongly influencing probe density and, then,

Microarray Fabrication
The A-MA microarray substrate (7.5 × 2.5 cm) was manufactured by standard VLSI technology and consists of a silicon substrate (about 600 µm of thickness) featured by a multilayer of Al film placed by CVD method (900 nm) and a SiO 2 layer placed by TEOS precursor with PE-CVD technique, with thickness ranging from 780 nm to 925 nm.
(a) Chemical process. The A-MA substrates were chemically processed with the following steps: (a) cleaning (to remove organic contamination) and activation process (to increase the Hydroxyl group (-OH) density at SiO 2 surface) by plasma O 2 treatment was performed by Sentech plasma generator (300 s, 100 W); (b) vapor phase silanization with GOPS (glycidoxysilane) was carried out in under vacuum Oven with the follow conditions: 4 h, 0.1 Atm, and 125 • C; (c) spotting of capture probes by a Perkin Elmer piezo spotter dispensing drops of 300 pL of solution of oligonucleotides (10 µM) in Na 2 HPO 4 buffer (150 mM, pH = 9); (d) probe anchoring step by incubating the spotted substrate for 4 h at 30 • C and 90% humidity relative in a climatic cell;  Figure 1b reports the scheme of the layout used for this study. (d) Microarray for analytical performances characterization. Microarray layouts of size 6 × 18 spots (spot size 100 μm, pitch 250 μm) were printed by a Non-Contact PiezoArray (PERKIN ELMER) in clean room according to previously described procedures [28].

Hybridization Protocol
The hybridization reaction was carried out by a 250 µL solution containing P1* Cy5 and P2* Cy5 labelled perfect match targets at various concentrations of 0.0005 nM, 0.005 nM, 0.05 nM, 0.10 nM, 0.50 nM, 5.00 nM, 15.00 nM, and 50.00 nM in 20 mM of sodium phosphate buffer, 1 M NaCl, 5.2 mM fo KCl, 0.1% of Tween20, 2 × of Denardt's solution, and 20 µg/µL of ssDNA. After hybridization, a washing step consisting of 5 at 40 • C in 2X SSC, followed by a second wash of 5 at 40 • C in 0.2X SSC, was carried out.

SiO 2 Thickness and Optical Measurements
The measurements of SiO 2 thickness and the reflectance spectra were performed by Ellipsometer SENTCH850.
The microarray fluorescent images were acquired by Optical Reader SN 038 [3], set in multisot 5 acquisition mode. Images were analyzed by a customized software InCheck Platform H-MAT version 2.5.1 (developed by STMicroelectronics) with the following analysis protocol. For each spotted slide, the average and standard deviation of signal intensity (σ) of the 42 replicas of each Cy5-labled oligo DNA concentration of layout in Figure 1 was calculated. These values were correlated with the specific SiO 2 thickness.

Substrate Optical Characterization
The A-MA substrate structures were composed of a multilayer structure comprising Silicon substrate, on top of which was placed a layer of Al and SiO 2 film with thickness of about 830 nm ( Figure 2). This structure has been designed to allow enhancement of light occurring at the substrate surface, since SiO 2 film thickness was modulated as function of Cy5-emitted light, and Al acts as mirror.

Hybridization Protocol
The hybridization reaction was carried out by a 250 μL solution containing P1* Cy5 and P2* Cy5 labelled perfect match targets at various concentrations of 0.0005 nM, 0.005 nM, 0.05 nM, 0.10 nM, 0.50 nM, 5.00 nM, 15.00 nM, and 50.00 nM in 20 mM of sodium phosphate buffer, 1 M Nacl, 5.2 mM fo KCl, 0.1% of Tween20, 2 × of Denardt's solution, and 20 μg/μL of ssDNA. After hybridization, a washing step consisting of 5′ at 40 °C in 2X SSC, followed by a second wash of 5′ at 40 °C in 0.2X SSC, was carried out.

SiO2 Thickness and Optical Measurements
The measurements of SiO2 thickness and the reflectance spectra were performed by Ellipsometer SENTCH850.
The microarray fluorescent images were acquired by Optical Reader SN 038 [3], set in multisot 5 acquisition mode. Images were analyzed by a customized software InCheck Platform H-MAT version 2.5.1 (developed by STMicroelectronics) with the following analysis protocol. For each spotted slide, the average and standard deviation of signal intensity (σ) of the 42 replicas of each Cy5labled oligo DNA concentration of layout in Figure 1 was calculated. These values were correlated with the specific SiO2 thickness.

Substrate Optical Characterization
The A-MA substrate structures were composed of a multilayer structure comprising Silicon substrate, on top of which was placed a layer of Al and SiO2 film with thickness of about 830 nm ( Figure 2). This structure has been designed to allow enhancement of light occurring at the substrate surface, since SiO2 film thickness was modulated as function of Cy5-emitted light, and Al acts as mirror.

Reflectance Measurements
In order to evaluate the mirror effect of the Al layer in the multilayer structure of the A-MA substrate, reflectance measurements were carried out on substrates featured by SiO2 layers with thicknesses ranging between 780 nm and 925 nm. The obtained reflectance spectra were expressed by percentage of reflectance variation (R%) compared to the signal of the reference sample (i.e., pure Al layer) (Figure 2a). The R% values at the first interference maximum and minimum are plotted in Figure 3b.
The R% variation over the explored wavelength range in the case of both 1st max (from 0.6 to 0.55) and 1st min (from 0.44 to 0.41) returns in the range of variability of the signal of Al reference sample within the explored wavelength range.
This finding indicates that the variation of the Al metal reflectivity is negligible over the explored SiO2 thickness range with a consequent negligible variation of the fluorescence signal.

Reflectance Measurements
In order to evaluate the mirror effect of the Al layer in the multilayer structure of the A-MA substrate, reflectance measurements were carried out on substrates featured by SiO 2 layers with thicknesses ranging between 780 nm and 925 nm. The obtained reflectance spectra were expressed by percentage of reflectance variation (R%) compared to the signal of the reference sample (i.e., pure Al layer) (Figure 2a). The R% values at the first interference maximum and minimum are plotted in Figure 3b.
The R% variation over the explored wavelength range in the case of both 1st max (from 0.6 to 0.55) and 1st min (from 0.44 to 0.41) returns in the range of variability of the signal of Al reference sample within the explored wavelength range.
This finding indicates that the variation of the Al metal reflectivity is negligible over the explored SiO 2 thickness range with a consequent negligible variation of the fluorescence signal.  Figure 4 illustrates the obtained microarray fluorescence image, highlighting also the relative capture probes layout.  Figure 4 illustrates the obtained microarray fluorescence image, highlighting also the relative capture probes layout.  (1): in which A is the difference of signal between max and min, xc (725 ± 6 nm) is the x position of min, w is the full width at half maximum (210 ± 10 nm), Y is the signal recorded (52650 ± 1700 a.u.) and x is the SiO2 thickness (830 ± 8 nm). The fitting shows a maximun of fluorescence peak at 830 ± 8 nm and a bandwidth value of 105 ± 20 nm. These data confirm the experimental finding above reported showing a maximum of interference for the Cy5 emission wavelength at a SiO2 thickness value of about 830 nm with a width of 90 nm at half maximum.
in which A is the difference of signal between max and min, xc (725 ± 6 nm) is the x position of min, w is the full width at half maximum (210 ± 10 nm), Y is the signal recorded (52650 ± 1700 a.u.) and x is the SiO 2 thickness (830 ± 8 nm). The fitting shows a maximun of fluorescence peak at 830 ± 8 nm and a bandwidth value of 105 ± 20 nm. These data confirm the experimental finding above reported showing a maximum of interference for the Cy5 emission wavelength at a SiO 2 thickness value of about 830 nm with a width of 90 nm at half maximum.
in which A is the difference of signal between max and min, xc (725 ± 6 nm) is the x position of min, w is the full width at half maximum (210 ± 10 nm), Y is the signal recorded (52650 ± 1700 a.u.) and x is the SiO2 thickness (830 ± 8 nm). The fitting shows a maximun of fluorescence peak at 830 ± 8 nm and a bandwidth value of 105 ± 20 nm. These data confirm the experimental finding above reported showing a maximum of interference for the Cy5 emission wavelength at a SiO2 thickness value of about 830 nm with a width of 90 nm at half maximum.   Figure 6 reports the scheme of the chemical steps used in the present study for the grafting of the ssDNA probes that comprises (a) plasma-O 2 cleaning and activation, (b) epoxy-silanization, (c) capture probe spotting and anchoring, and (d) BSA passivation. In order to follow the effectiveness of the chemical process, each step was characterized by XPS analysis. After plasma-O 2 process (step (a)), the XPS spectra shows the diagnostic peaks for Si2p (30%) and O1s (56.5%), proving the goodness of the cleaning. Upon silanization (step (b)), the XPS spectrum in the range of 0-1000 eV exhibits the specific peaks for Si2p (27%), O2p (51.6%), and C1s (10%). A Gaussian-Lorentz deconvolution for the C1s peak reports two main components: a first centered at about 285.2 eV attributed to C-C and C-H carbon species and a second centered at about 286.7 eV attribute to C-O species. The ratio C1s/Si2p value for silanized surface is 0.37. This is diagnostic of the presence of organic layer covering the substrate surface in line with epoxy-silane derivatization. Finally, after passivation with BSA protein (step (d)), the XPS spectra show the following peaks: C1s at 285 eV (17.31%), O1s (53.70), N1s at 400 eV (2.46%), and Si2p (24.77%). Both the peak for nitrogen (N1s) and the ratio C1s/Si2p ratio of about 0.7 (versus C1s/Si2p ratio of 0.37 for the silanized surface) confirm the presence of a protein layer.  Figure 6 reports the scheme of the chemical steps used in the present study for the grafting of the ssDNA probes that comprises (a) plasma-O2 cleaning and activation, (b) epoxy-silanization, (c) capture probe spotting and anchoring, and (d) BSA passivation. In order to follow the effectiveness of the chemical process, each step was characterized by XPS analysis. After plasma-O2 process (step (a)), the XPS spectra shows the diagnostic peaks for Si2p (30%) and O1s (56.5%), proving the goodness of the cleaning. Upon silanization (step (b)), the XPS spectrum in the range of 0-1000 eV exhibits the specific peaks for Si2p (27%), O2p (51.6%), and C1s (10%). A Gaussian-Lorentz deconvolution for the C1s peak reports two main components: a first centered at about 285.2 eV attributed to C-C and C-H carbon species and a second centered at about 286.7 eV attribute to C-O species. The ratio C1s/Si2p value for silanized surface is 0.37. This is diagnostic of the presence of organic layer covering the substrate surface in line with epoxy-silane derivatization. Finally, after passivation with BSA protein (step (d)), the XPS spectra show the following peaks: C1s at 285 eV (17.31%), O1s (53.70), N1s at 400 eV (2.46%), and Si2p (24.77%). Both the peak for nitrogen (N1s) and the ratio C1s/Si2p ratio of about 0.7 (versus C1s/Si2p ratio of 0.37 for the silanized surface) confirm the presence of a protein layer.

Capture Probe Density Evaluation
In order to evaluate the A-MA surface grating ability, intra-spot probe densities average values were assessed by means of a calibration curve. This curve was obtained by spotting onto A-MA substrate 10 different concentrations (5000-10 F/um 2 ) of P1 Cy5-labeled probes and by recording fluorescence images (Figure 7a). Signal fluorescence intensities extracted by the images were plotted versus Cy5-labeled oligoprobes concentration (F/um 2 ) to obtain the calibration curve illustrated in Figure 7b. A good liner correlation was found (Y = 755.12 + 17.55X; R 2 = 0.9884). The curve was then used to evaluate the probe density values as described below.

Capture Probe Density Evaluation
In order to evaluate the A-MA surface grating ability, intra-spot probe densities average values were assessed by means of a calibration curve. This curve was obtained by spotting onto A-MA substrate 10 different concentrations (5000-10 F/µm 2 ) of P1 Cy5-labeled probes and by recording fluorescence images (Figure 7a). Signal fluorescence intensities extracted by the images were plotted versus Cy5-labeled oligoprobes concentration (F/µm 2 ) to obtain the calibration curve illustrated in Figure 7b. A good liner correlation was found (Y = 755.12 + 17.55X; R 2 = 0.9884). The curve was then used to evaluate the probe density values as described below. To evaluate the probe density values, A-MAs were prepared by spotting Cy5-labeled P1 and P2 probes according to the chemical process described in the previous section and the layout reported in Figure 1b. The obtained fluorescence images were analysed, and the related signals used to read the corresponding probe density values (F/µm 2 ) using the calibration curve above described. Table 1 reports the obtained results for both probes Cy5-labeled P1 and P2. Values around 2000 fluorophores per µm 2 were found. This finding confirms previously published studies on probe grafting on silicon oxide surfaces [17,19].

Analytical Performances Characterization
The A-MA analitycal performance in terms of linearity, sensitivity LoD, and LoB was assessed performing hybridation with P1* and P1* Cy5 labelled perfect match targets at various concentrations of 0.00 nM, 0.0005 nM, 0.005 nM, 0.05 nM, 0.10 nM, 0.50 nM, 5.00 nM, 15.00 nM, and 50.00 nM, To evaluate the probe density values, A-MAs were prepared by spotting Cy5-labeled P1 and P2 probes according to the chemical process described in the previous section and the layout reported in Figure 1b. The obtained fluorescence images were analysed, and the related signals used to read the corresponding probe density values (F/µm 2 ) using the calibration curve above described. Table 1 reports the obtained results for both probes Cy5-labeled P1 and P2. Values around 2000 fluorophores per µm 2 were found. This finding confirms previously published studies on probe grafting on silicon oxide surfaces [17,19].

Analytical Performances Characterization
The A-MA analitycal performance in terms of linearity, sensitivity LoD, and LoB was assessed performing hybridation with P1* and P1* Cy5 labelled perfect match targets at various concentrations of 0.00 nM, 0.0005 nM, 0.005 nM, 0.05 nM, 0.10 nM, 0.50 nM, 5.00 nM, 15.00 nM, and 50.00 nM, respectively. The Foreground Median (F Med ) as the median of pixel intensity of fluorescence signal and the Background Mean (B M ) as the mean of pixel intensity of local background were measured from each hybridized spots. Figure 8 illustates the average F Med values obtained for both P1 and P2 perfect match over the explored concentration range. Good linearity can be observed for the range between 0.05 nM and 0.5 nM for both specific targets. In order to calculate both the LoD and the LoB for each probes, a methodology based on linear regression has been applied. More in details, the LoB value for each probe has been found by appliying Equation (2), in which NB is number of background replica and estimating the 95% percentile p = (100 − α) = 95.
From this procedure was obtained for P1 and P2 probes LoB values of about 204 and 257, respectively.
The LoD was estimated set to 4 times the LoB signal intensity value (Cut-off) in the regression linear equation above reported. The finding LoD are 0.031 nM and 0.043 nM for specifc P1 and P2, respectively. Regarding the sensitivity, it was defined as the slope of the regression linear equation, and a value was found of 17702.2 a.u. nM −1 for the specific P1 probe and 17882.7 a.u. nM −1 for P2, respectively.

Conclusions
In this work, an advanced, silicon-based substrate (A-MA) able to sensitively enhance fluorescent signals for microarray applications has been presented. The substrate contains a multilayer structure consisting of an Si substrate with Al film acting as a light mirror, and an SiO2 layer properly modulates the thickness (835 nm) to achieve the fluorescence enhancement by light constructive interference for the emission wavelength of the Cy5 fluorescent label. The immobilization of the ssDNA probe was achieved by means of standard silane chemistry, and probe densities of about 2000 F/um 2 were found. The A-MA investigated in the detection of HBV (Hepatitis B Virus) pathogen showed a dynamic linear range of 0.05-0.5 nM, a sensitivity of about 18000 a.u. nM −1 , and a Limit of Detection in the range of 0.031-0.043 nM as a function of the capture probe sequence.
In conclusion, the main strength of the A-MA substrate can be summarized as it follows: (i) the total reflection of fluorescence-signal; (ii) the good optical transmission of SiO2 layer; (iii) the ability to modulate the SiO2 thickness to gain the maximum interference in correlation to the fluorescencelabel emission wavelength; and (iv) the high chemically versatility of SiO2 surface, being easily modified by silane-strategy to obtain covalent capture probes anchoring. All these features make this substrate very promising in the development of new microarray platforms for the next generation of molecular diagnostic devices. In order to calculate both the LoD and the LoB for each probes, a methodology based on linear regression has been applied. More in details, the LoB value for each probe has been found by appliying Equation (2), in which NB is number of background replica and estimating the 95% percentile p = (100 − α) = 95. LoB = [NB(p/100) + 0.5] From this procedure was obtained for P1 and P2 probes LoB values of about 204 and 257, respectively.
The LoD was estimated set to 4 times the LoB signal intensity value (Cut-off) in the regression linear equation above reported. The finding LoD are 0.031 nM and 0.043 nM for specifc P1 and P2, respectively. Regarding the sensitivity, it was defined as the slope of the regression linear equation, and a value was found of 17702.2 a.u. nM −1 for the specific P1 probe and 17882.7 a.u. nM −1 for P2, respectively.

Conclusions
In this work, an advanced, silicon-based substrate (A-MA) able to sensitively enhance fluorescent signals for microarray applications has been presented. The substrate contains a multilayer structure consisting of an Si substrate with Al film acting as a light mirror, and an SiO 2 layer properly modulates the thickness (835 nm) to achieve the fluorescence enhancement by light constructive interference for the emission wavelength of the Cy5 fluorescent label. The immobilization of the ssDNA probe was achieved by means of standard silane chemistry, and probe densities of about 2000 F/µm 2 were found. The A-MA investigated in the detection of HBV (Hepatitis B Virus) pathogen showed a dynamic linear range of 0.05-0.5 nM, a sensitivity of about 18000 a.u. nM −1 , and a Limit of Detection in the range of 0.031-0.043 nM as a function of the capture probe sequence.
In conclusion, the main strength of the A-MA substrate can be summarized as it follows: (i) the total reflection of fluorescence-signal; (ii) the good optical transmission of SiO 2 layer; (iii) the ability to modulate the SiO 2 thickness to gain the maximum interference in correlation to the fluorescence-label emission wavelength; and (iv) the high chemically versatility of SiO 2 surface, being easily modified by silane-strategy to obtain covalent capture probes anchoring. All these features make this substrate very promising in the development of new microarray platforms for the next generation of molecular diagnostic devices.
Author Contributions: S.C. conceived the study and wrote the manuscript; S.P. helped conceive the study, prepared the microarray, and contributed to writing the manuscript; S.C. and N.V. performed the hybridization test and analyzed the data; R.P. discussed the results and contributed to the final manuscript. All authors have read and approved the final version of this manuscript and agreed to be accountable for all aspects of the work.

Conflicts of Interest:
The authors declare no conflict of interest.