**3. Results**

#### *3.1. PSi-Based Biosensor Characterization*

The effective optical thickness of Porous Silicon is strongly dependent on its refractive index. When an analyte penetrates the pores, the refractive index of the porous layer increases, therefore inducing a shift in the EOT. In the case of a membrane, to make sure that the analyte remains inside the porous matrix, two approaches are possible: size exclusion or binding to the pore wall. For this project, size exclusion was chosen by accordingly selecting different pore size for each layer of the membrane. The first layer, also called the sensing layer, has an average pore size 41.05 nm, as described in Table 1 and illustrated in Figure 3. Larger pores could be etched, but the resulting layer was too easily damaged, as an increase in pore size induced a decrease of the pore wall thickness. The selected pore size is a trade-off between pore opening and mechanical integrity. The bacterial lysates are composed, among others, of cell wall fragments, DNA and RNA molecules, and cytoplasmic liquid and ribosomes, which are assumed to penetrate the sensing layer. In order to keep the bulkier ones in the top layer, the contrast layer was designed to have a smaller pore size, as indicated in Table 1. These choices in pore size also have two other motivations: (1) they allow the PlyB221 endolysin to flow through the membrane and not be retained in the sensing layer and (2) they prevent the non-lysed

bacteria from penetrating the membrane, therefore enabling a selective detection. Figure 4 depicts a PSi membrane, with an up-close view of the transition between the top sensing layer and the contrast layer.

**Table 1.** Pore diameter, thickness and porosity measured for each of the three layers of the porous silicone membrane (PSiM).


\* The thickness and porosity could not be accurately measured, as part of the layer was etched away during the DRIE step.

**Figure 3.** Scanning electron microscopy (SEM) image of the top layer porous silicon membrane. The average pore size is ~41 nm, with a standard deviation of ~20 nm.

**Figure 4.** Scanning electron microscopy (SEM) cross-section image of a multilayered porous silicon membrane. The inset shows a close-up of the top of the porous membrane, in which the sensing and the contrast layers can be identified.

The theoretical sensitivity of the porous membrane sensor was also approximated by calculating the EOT in both air and in ethanol, applying a Fourier transform to the Fabry Perot fringes of the optical spectra, as illustrated in Figure 5. By plotting the EOT versus refractive index variation (Figure 5c), an approximation of the sensitivity can be made, which amounts to 5745.8 ± 847.7 nm·RIU−<sup>1</sup> or, expressed in relative changes of EOT, as 54.4 ± 8.3 %·RIU−1.

**Figure 5.** Porous silicon membrane characterization. (**a**) Reflection spectra of a porous silicon membrane in air (blue) and ethanol (orange). (**b**) Representative Fourier transform of the reflectance spectra in panel. (**c**). Calibration curve of the EOT versus the refractive index variation.

#### *3.2. B. cereus Lysate Observation and Characterization*

Endolysins are encoded and used by bacteriophages at the end of their replication cycle. They degrade the peptidoglycan of the targeted bacteria, creating an opening in the cell wall for the phage. When endolysins are recombinantly produced and added exogenously to bacteria, they lead to cell lysis by breaking down the exposed peptidoglycan layer which make them promising antimicrobial agents [42]. To observe the effect of the PlyB221 endolysin, *B. cereus* was observed using SEM before and after the lysis. Close up images of an intact versus a lysed bacterium are depicted in Figure 6.

**Figure 6.** Scanning electron microscopy (SEM) image of (**a**) an intact *B. cereus* bacterium and (**b**) a *B. cereus* lysed bacterium.

To further characterize the bacterial lysate, the average number and size of bacterial clusters were compared before and after lysis, using images of the same magnification. Due to a lack of contrast and sharpness, only qualitative observations could be made. Over the same area of inspection, there are nearly 20 times more clusters of bacterial lysate than of bacteria; the average size of each cluster is also decrease by more than a 10-fold. As illustrated in Figure 6, some larger clusters of bacterial lysate remain, but these are surrounded by much smaller clusters, whose area goes down to the nanometer range (<50 nm). These smaller clusters are assumed to penetrate the first porous layer, but are unable to diffuse to the second porous layer because of a size exclusion effect.

#### *3.3. B. cereus Lysate Detection PSi Layers and PSi Membranes*

In order to establish the added-value of a flow-through approach to the optical detection with respect to the traditional flow-over approach, the performances of PSi membranes were compared to those measured on PSi layers.

The average relative EOT shift measured on PSi layers, in a flow-over approach, is presented in Figure 7a. No distinction could be made between the bacterial lysate detection and the control tests using either the buffer or the endolysin suspension. Both control tests induced minute decreases of the relative EOT: on average −0.24% and −0.01% after 1 h for PBS and the PlyB221 endolysin, respectively. A small increase of relative EOT was measured in the presence of bacteria lysate, namely 0.05%, but this decrease was not significant when compared to the noise level. This noise level was calculated based on the overall standard deviation of the signal in PBS and was expressed as 3σ = 1.08%.

**Figure 7.** Characteristic relative effective optical thickness (EOT) shift measured on (**a**) a porous silicone (Psi) layer or (**b**) a PSi membrane for 1 h in phosphate buffered saline (PBS), in a PlyB221 endolysin suspension and in a *B. cereus* lysate (*n* ≥ 3). In (**a**), the inset illustrates the flow-over approach; there was no significant shift visible during the detection of either the PlyB221 endolysin or bacterial lysate. In (**b**), the inset illustrates the flow-through approach; there was a significant shift in the relative EOT during the detection of bacterial lysate.

The performances of PSi membranes for the same three tests in flow-through operation are illustrated in Figure 7b. The buffer control test induced a decrease in relative EOT of −0.32% after 1 h. The control test with only the PlyB221 endolysin gave rise to a 0.28% increase of relative EOT. This increase was however not significant and remained below the noise level, which was calculated in the same manner described previously and was equal to 0.84% in the case of PSi membranes. Upon the penetration of bacterial lysate inside the membrane, a significant increase of relative EOT was measured, exceeding the noise level after 6 min and reaching an average +2.43% after 1 h.

#### *3.4. Determination of the Limit of Detection of B. cereus*

The limit of detection was determined using decreasing concentration of *B. cereus*. Results are displayed in Figure 8. As presented above, the detection of 10<sup>6</sup> CFU/mL is significant compared to the negative control tests in PBS and in endolysin only. For a concentration of 10<sup>5</sup> CFU/mL, the relative EOT shift amounted 0.96%, which is just above the noise level but still significantly different to both control tests. For 10<sup>4</sup> CFU/mL, the relative EOT shift decreased below the noise level to 0.64%. While this difference remains

significant with respect to the control test in PBS, it is not with respect to the endolysin control test.

**Figure 8.** Characteristic relative effective optical thickness (EOT) shift measured on a PSi membrane after 1 h in PBS, in a PlyB221 endolysin suspension and in increasing concentrations of *B. cereus* lysate (*n* ≥ 3). The dashed red line represents the noise level, fixed as 3σ of the signal measured in PBS. The detection limit is 10<sup>5</sup> CFU/ml of *B. cereus*.

#### *3.5. Specificity Testing with S. epidermidis*

To illustrate the specificity of the sensing platform, a negative control test was performed using a *S. epidermidis* and a positive control test was carried out using a mixture of *B. cereus* and *S. epidermidis*. The relative EOT shifts observed for both tests are depicted in Figure 9, where it is compared to controls in PBS and with the PlyB221 endolysin only, as well as to the detection of *B. cereus* lysate only. The detection using *S. epidermidis* induced no significant shift in relative EOT. After 1 h, the relative EOT was increased by 0.29%, which is comparable to the control test with only the PlyB221 endolysin. The positive control test showed a significant relative EOT increase of 1.59%.

#### *3.6. Versatility of the Platform: Detection of S. epidermidis in PBS with Lysostaphin*

To illustrate that the biosensor can be used for the detection of other bacteria/lytic enzyme pairs, the same detection protocol was repeated using *S. epidermidis* as targeted strain and lysostaphin as selective lytic agent. The relative EOT shift was measured overtime and compared to control tests in PBS and lysostaphin only. As depicted on Figure 10, the relative EOT shift induced after 1 h by the penetration of lysostaphin into the porous membrane is comparable to the one measure for the PlyB221 endolysin and amounts 0.26%. The penetration of *S. epidermidis* lysate causes a significant shift in relative EOT after 30 min, reaching a value of 1.83% after 1 h.

**Figure 9.** Characteristic relative EOT shift measured on a PSi membrane after 1 h in PBS, in a PlyB221 endolysin suspension, in a *S. epidermidis* suspension, in a mixture of *B. cereus* lysate and *S. epidermidis* and in *B. cereus* lysate only (*n* ≥ 3). The dashed red line represents the noise level, fixed as 3σ of the signal measured in PBS. Only the detection of mixture of *B. cereus* lysate and *S. epidermidis* induced a significant shift.

**Figure 10.** Relative effective optical thickness (EOT) shift measured on a PSi membrane for 1 h in PBS (grey), in a lysostaphin suspension (blue) and in a *S. epidermidis* lysate (red). The bacterial lysate is detected after 30 min.
