**5. Choice of biomolecular probe**

**Figure 1.** Types of silicon surface chemical modifications for biosensors: (a) organosilane-based, (b) phosphonate-

**Figure 1** represents the main functionalization approaches employed to construct integrated optics (IO) biosensors. Before the biofunctionalization step, a previous chemical activation of the sensor surface is always needed. To this aim, our group employed the self-assembly of organofunctional alkoxysilanes (**Figure 1a**), an easy and versatile system for organic conju‐ gating [34]. However, silicon-based surfaces require a prior activation step to oxidize the surface and to expose the silanol groups for cross-linking with the silane. The formation of a thin silane self-assembled film allows applying a great number of chemical reactions. Immediately before silanization, surfaces are cleaned with oxidant media to remove organic pollutants and to increase the hydroxyl moieties on the surface [35]. The used oxidant is piranha solution [36–39], consisting of a concentrated sulfuric acid mixed with hydrogen peroxide at 3:1 ratio. This treatment is performed by heating for 30 min only. Hundreds of different organosilanes with different structures and functionalities are nowadays commer‐ cially available, although the most commonly employed are those with short alkyl chain that present an amino, thiol, epoxy, or carboxylic group at the terminus. Among this vast variety of compounds, 3-aminopropyltriethoxysilane (APTES) was chosen for its reactivity to

based, and (c) glutaraldehyde-based strategies.

184 Lab-on-a-Chip Fabrication and Application

aldehyde, carboxylic acid, and epoxy functionalities.

At this point, a wide variety of biomolecules (antibodies, nucleic acid sequences, peptides, enzymes, cell receptors) can be used as bioreceptors (**Figure 2**).

**Figure 2.** Types of bioconjugation methods on aminated surfaces: (a) *N*-hydroxysuccinimide–based, (b) succinic anhy‐ dride–based, (c) p-phenylenediisocyanate–based, and (d) glutaraldehyde-based strategies.

The choice of bioreceptor depends on the intended application of biosensor and it must meet two importantrequirements: high specificity forthe target molecule and high stability to retain its biological activity when immobilized on the support.

A first biofunctionalization approach, based on the covalent bind of a biomolecule on the activated silicon sensor surface, included the use of an IgG antibody as molecular probe directed against B-cell receptor. The chip was treated with the homobifunctional cross-linker glutaraldehyde (GA): This molecule, besides to be employed to form an aldehyde-terminat‐ ed surface, which allows the reaction of amine groups, by the formation ofimines (Schiff bases), acts as spacer in order to keep away from the surface the immobilized bioprobe that can react freely with target molecules [46, 47]. By this strategy, the antibody has been immobilized on

the surface via protein A in an oriented fashion [48]. The whole process is checked monitor‐ ing surface changes by ellipsometric measurements and FTIR spectromicroscopy. As report‐ ed in Table 1, using a random sampling of four different wells, it was observed for all of them after each functionalization step the surface layer thickness.


**Table 1.** Surface layer thickness on four random samples after each step of functionalization measured by ellipsometry technique.

The analyses of the FT-IR spectra led to the identification of several characteristic vibration bands that were coherent with the various functionalization steps. Table 2 reports a list of the major bands identified together to peak assignment.


**Table 2.** Major bands identified by FTIR spectromicroscopy and corresponding assigned peaks.

As experimental model, it was chosen a murine lymphoma cell line (A20) [49] that expresses high levels of membrane IgG. The most interesting point of this first approach is that the microfabricated biochip appears to be suitable to reveal specific bindings such as that between cell-surface proteins (receptor) and corresponding specific antibody. In addition, the num‐ ber of cells detected by the devices was 2.0 × 10−3 cells/μm2 .

the surface via protein A in an oriented fashion [48]. The whole process is checked monitor‐ ing surface changes by ellipsometric measurements and FTIR spectromicroscopy. As report‐ ed in Table 1, using a random sampling of four different wells, it was observed for all of them

**Thickness (nm)**

**Table 1.** Surface layer thickness on four random samples after each step of functionalization measured by ellipsometry

The analyses of the FT-IR spectra led to the identification of several characteristic vibration bands that were coherent with the various functionalization steps. Table 2 reports a list of the

Si–O str – 1127 – – – – Si–O–C *as* str – – 1250 1250 1258 1258 –(CH2)–str – – 1295–1305 – – – –O–CH2–str – – 1445–1475 – – – Amide II – – – – – 1531 C=O str – – – – 1635 1642 Saturated primary ammine (–NH2 def) – – 1650 – – –

Amide I C=O str – – – – 1650–1680 1638–1687 C=O str – – – – 1685–1705 1774 N–H str – – – – – 3121 Primary ammine –NH2 str – – 3250–3677 – – –

As experimental model, it was chosen a murine lymphoma cell line (A20) [49] that expresses high levels of membrane IgG. The most interesting point of this first approach is that the microfabricated biochip appears to be suitable to reveal specific bindings such as that between

**Table 2.** Major bands identified by FTIR spectromicroscopy and corresponding assigned peaks.

**Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6**

**Film Sample 1 Sample 2 Sample 3 Sample 4** Oxide 75.8 ± 0.4 72.4 ± 0.2 75.0 ± 0.3 73.4 ± 0.2 Aptes + GA 3.0 ± 0.4 3.1 ± 0.3 3.1 ± 0.3 2.1 ± 0.2 Protein A 0.68 ± 0.09 0.67 ± 0.08 0.75 ± 0.1 0.85 ± 0.1 χ<sup>2</sup> 0.54 0.43 0.47 0.45

after each functionalization step the surface layer thickness.

186 Lab-on-a-Chip Fabrication and Application

major bands identified together to peak assignment.

**Predicted peak Frequency cm−1**

technique.

Anyway, since this detection limit does not seem satisfactory and the idea that the contact probability between cells and antibodies on capture specific surface could be improved, we took advantage of a new functionalization strategy exploiting an Id-peptide as biomolecular probe. The choice of an Id-peptide was dictated by two main reasons: (i) Artificial peptides provide an opportunity to develop the desired molecular biosensor due to their desirable properties such as diversified structure and high affinity [50]. In addition, peptides with specific sequences can provide high affinity to particular ligands and be obtained by screen‐ ing and optimization of artificial peptide libraries; (ii) the used idiotype peptide is a small peptide ligand able to be recognized with high affinity and specificity from the B-cell recep‐ tors present on the lymphoma B cells [51–53]. The use of a small ligand as biorecognition element endowed with great specificity could highly enhance affinity and selectivity of the detection layer. In addition, it simplifies the functionalization procedure with respect to that employed for antibodies in which controlling protein orientation is still very challenging [54]. The peptide was immobilized on the silicon surface following the functionalization strategy schematized in **Figure 3**.

**Figure 3.** Functionalization approach utilized on silicon surface to conjugate an Id-peptide to detect lymphoma cells. After each passivation step, the new synthesized layer is reported in the figure with the same color of the molecule used in the chemical reaction (APTES is blue, BS<sup>3</sup> is red, Id-peptide is green).

This chemical procedure was developed on both crystalline flat and porous silicon samples; the nanostructured porous was chosen because its peculiar morphology allows the immobi‐ lization of a greater number of molecules with respect to a planar substrate and a number of functionalization investigation methods could be more easily exploited [55]. The aminosilan‐ ized surface has been activated by the homobifunctional cross-linkers bis[sulfosuccinimidyl] suberate (BS<sup>3</sup> ), which, acting as spacer, provide succinimidyl-activated carboxyl group that could react with amine-ended peptide to form an amide bond. Changes in chemical compo‐ sition of PSi surface were monitored by FTIR spectroscopy after each functionalization step until BS<sup>3</sup> (**Figure 4**).

**Figure 4.** FTIR spectra of silicon surface after each chemical modification step.

The analysis of the FTIR spectra in the range from 2500 to 500 cm−1 highlighted characteristic peaks of each molecule used in the different passivation steps, demonstrating the effective‐ ness of the functionalization procedure. Indeed, the characteristic peaks of Si–Hx bonds corresponding to the PSi after electrochemical etching (2100 and 680–630 cm−1) are no longer visible when the devices were thermally oxidized, whereas the appearance of the Si–O–Si characteristic band at 1100 cm−1 was detected.

The formation of the silane film was confirmed by the presence of peaks in the span 1440–1390 cm−1, relative to CH3 from APTES ethoxy moieties, and at 1655 cm−1 relative to an imine group from oxidation of an amine bicarbonate salt [56]. Moreover, the appearance of the peaks at 1640 and 1550 cm−1 that correspond to CO– and NH– groups of an amide bond, confirms the deposition of the BS3 .

The functionalization of porous silicon surface was also confirmed by spectroscopic reflec‐ tometry.

**Figure 5.** Reflectivity spectra on porous silicon surface before (solid line) and after APTES silanization (dashed line), and after BS3 functionalization (short dashed line).

Reflectivity spectra of porous silicon devices during functionalization steps are reported in **Figure 5**. The deposition on pores walls of a thin layer, constituted by the different organic chemical compounds, produces red shifts of spectra due to the increase of the average refractive index of porous silicon surfaces [57]. After silanization and cross-linker modifica‐ tion, a red shift of 21 and 15 nm was recorded, respectively. The same chemical modifica‐ tions were performed also on flat silicon devices; in the latter, the whole functionalization procedure was followed by spectroscopy ellipsometry, in order to quantify layer thickness variations after thermal oxidation (SiO2), silanization (APTES), and cross-linker functionali‐ zation (BS3 ). As showed in Table 3, the thickness of oxidized silicon devices was 74 ± 1 nm; this value increased of 3 and about 2 nm after treatment with aminosilane and BS3 , respectively.


The values reported are the average of five determinations on each sample.

**Table 3.** Surface layer thickness on four random samples after each step of functionalization measured by ellipsometry technique.

### **6. Biosensing**

**Figure 4.** FTIR spectra of silicon surface after each chemical modification step.

characteristic band at 1100 cm−1 was detected.

.

functionalization (short dashed line).

deposition of the BS3

188 Lab-on-a-Chip Fabrication and Application

tometry.

and after BS3

The analysis of the FTIR spectra in the range from 2500 to 500 cm−1 highlighted characteristic peaks of each molecule used in the different passivation steps, demonstrating the effective‐ ness of the functionalization procedure. Indeed, the characteristic peaks of Si–Hx bonds corresponding to the PSi after electrochemical etching (2100 and 680–630 cm−1) are no longer visible when the devices were thermally oxidized, whereas the appearance of the Si–O–Si

The formation of the silane film was confirmed by the presence of peaks in the span 1440–1390 cm−1, relative to CH3 from APTES ethoxy moieties, and at 1655 cm−1 relative to an imine group from oxidation of an amine bicarbonate salt [56]. Moreover, the appearance of the peaks at 1640 and 1550 cm−1 that correspond to CO– and NH– groups of an amide bond, confirms the

The functionalization of porous silicon surface was also confirmed by spectroscopic reflec‐

**Figure 5.** Reflectivity spectra on porous silicon surface before (solid line) and after APTES silanization (dashed line),

Once the chemical modified silicon chips have been obtained, a procedure to immobilize a small peptide forlabel-free detection of cancer cells was settled. The used experimental system takes advantage of the properties of an idiotype peptide isolated from peptide libraries able to bind the variable region of the B-cell receptor on A20 lymphoma cells [51]. The selected peptide, named A20-36 (pA20-36), whose sequence is EYVNCDNLVGNCVI, was linked on silicon-modified surfaces and used as molecular probe. A random peptide (RND), SSAYGSCKGPCSSGVHSI, was used as negative control. To determine the optimal peptide concentration to obtain a uniform coverage of planar and porous surfaces, a titration was carried out. Based on the obtained results [58], 150 μM concentration was used for both peptides.

The detection of lymphoma cancer cells fulfilled on both planar and porous peptide-modi‐ fied silicon surfaces is showed in **Figure 6**. The panels *a* and *b* report microscope light images of the planar device surfaces after incubation with a low (100 cells) or high number (50,000 cells) of A20 cells. The choice of the high number of cells was made in order to have satura‐ tion binding conditions. The same number of cells (50,000) was incubated on A20-36-peptidemodified porous silicon surface, but a lower number of detected cells were observed on light microscope (**Figure 6**, panel *c*). The chip was not able to bind lymphoma cells when function‐ alized with RND peptide (**Figure 6**, panel *d*), whereas no myeloma cells (5T33MM), a surface IgG-positive B-cell line unable to bind to pA20-36 peptide [51], were detected when incubat‐ ed on the device functionalized with pA20-36 (**Figure 6**, panel *e*).

The number of A20 cells detected on functionalized planar surface device was about 8500 and, taking in account an average area of 80 μm2 for a single cell, filled ~680,000 μm2 , a value concordant with the available functionalized area (~1.0 × 106 μm2 ); when the detection was performed on porous silicon device, the number of cells that effectively bind the chip was lower (400), filling an area of about 32,000 μm2 . The exiguous number of A20 cells on the porous silicon surface was probably caused by the peculiar morphology of the support; being highly porous, with pore diameter of about 50 nm, and pore upper edges lowerthan 1 nm in thickness, its inner surface is many order of magnitude greater than the top active one. Hence, just a very low number of peptides are really available on the pore upper edges to bind the cells (that cannot enter into the pores). Therefore, the consequence of this condition is the decrease in the number of cells detected on porous silicon biochip resulting lowerrespect to that on the planar surface.

**Figure 6.** Optical images of A20 cell detection on both planar and porous silicon devices. Planar silicon pA20-36 modi‐ fied sensor after incubation with 1 × 104 A20 cells/mL (*a*) and 5 × 106 A20 cells/mL (*b*). Porous silicon pA20-36 modified surface after incubation with 5 × 106 A20 cells/mL (*c*). Planar silicon RND-modified-sensor after incubation with 5 × 106 A20 cells/mL (*d*). Planar silicon pA20-36 modified sensor after incubation with 5 × 106 5T33MM cells/mL (*e*).

The surface of each silicon chip presents a functionalized available area of about 1.0 × 106 μm2 so the maximum number of cells that can be bound on the device was 10,000 (covering an area of about 800,000 μm2 ). Since the number of cells detected on planar and porous silicon surfaces was by count 8500 and 400 (evaluated by optical microscopy), the efficiency of detection is 85 and 4%, respectively. Moreover, comparing the efficiency of detection of the flat silicon device based on Id-peptide-BCR recognition with an analogous silicon-based bioanalytical system in which an anti-IgG-BCR was used as molecular probe [48], it is clear that the first biochip resulted more efficient in detecting A20 cells (8.5 × 10−3 vs. 2.0 × 10−3 cells/μm2 , respectively).

alized with RND peptide (**Figure 6**, panel *d*), whereas no myeloma cells (5T33MM), a surface IgG-positive B-cell line unable to bind to pA20-36 peptide [51], were detected when incubat‐

The number of A20 cells detected on functionalized planar surface device was about 8500 and,

performed on porous silicon device, the number of cells that effectively bind the chip was

silicon surface was probably caused by the peculiar morphology of the support; being highly porous, with pore diameter of about 50 nm, and pore upper edges lowerthan 1 nm in thickness, its inner surface is many order of magnitude greater than the top active one. Hence, just a very low number of peptides are really available on the pore upper edges to bind the cells (that cannot enter into the pores). Therefore, the consequence of this condition is the decrease in the number of cells detected on porous silicon biochip resulting lowerrespect to that on the planar

**Figure 6.** Optical images of A20 cell detection on both planar and porous silicon devices. Planar silicon pA20-36 modi‐

A20 cells/mL (*a*) and 5 × 106

The surface of each silicon chip presents a functionalized available area of about 1.0 × 106

so the maximum number of cells that can be bound on the device was 10,000 (covering an area

was by count 8500 and 400 (evaluated by optical microscopy), the efficiency of detection is 85

A20 cells/mL (*d*). Planar silicon pA20-36 modified sensor after incubation with 5 × 106

, a value

); when the detection was

. The exiguous number of A20 cells on the porous

A20 cells/mL (*b*). Porous silicon pA20-36 modified

5T33MM cells/mL (*e*).

μm2

A20 cells/mL (*c*). Planar silicon RND-modified-sensor after incubation with 5 × 106

). Since the number of cells detected on planar and porous silicon surfaces

taking in account an average area of 80 μm2 for a single cell, filled ~680,000 μm2

ed on the device functionalized with pA20-36 (**Figure 6**, panel *e*).

concordant with the available functionalized area (~1.0 × 106 μm2

lower (400), filling an area of about 32,000 μm2

190 Lab-on-a-Chip Fabrication and Application

fied sensor after incubation with 1 × 104

surface after incubation with 5 × 106

of about 800,000 μm2

surface.

This difference is likely due to the better accessibility of the A20-36 Id-peptide on the BCR with respect to the anti-IgG. In fact, the binding of the idiotype peptide should occur with the more exposed variable region of the receptor in contrast with the interaction between IgG and BCR in which the variable regions of the immunoglobulin bind the less exposed constant region of the receptor. Furthermore, also the difference in affinity constants between the two ligandreceptor systems coupled to diverse functionalization approaches might have had a decisive role in the detection efficiency.

Cell detection was also investigated by atomic force microscopy analysis (AFM) (**Figure 7**).

**Figure 7.** Representative AFM image in the trace direction of live A20 cells detected on silicon surface.

Both light microscopy and AFM analysis showed a good biocompatibility of substrate since viability and cell morphology were not affected.

At this point, since cancerous cells are coexisted with other cell types in the body and it is very important to selectively differentiate cancer cells from other ones, in order to assess the real performance of the biochip, lymphoma cells detection was carried out on devices incubated with mixed samples of A20 and 5T33MM cells (3.5 × 10<sup>5</sup> /mL). The detection of lymphoma cells in system mixed is reported in **Figure 8**.

**Figure 8.** Detection of A20 cells (green) in system mixed with 5T33MM cells (red) by fluorescence macroscopy after incubation on planar silicon pA20-36 modified sensor.

The mixed system has been prepared in three different ratios (A20:5T33MM = 1:1, 1:10, 1:100) of the two labeled live cell lines [59]. The efficiency of detection also in a complex system demonstrated the high selectivity of the device, confirming that the use of an Id-peptide immobilized on a silicon-based chip could be a good proof-of-concept for future researches.
