**3. Results**

*3.1. Scanning Electron Microscopy (SEM) Characterization of the TiO2- and Tobacco Mosaic Virus (TMV)-Modified Si3N4 Surface*

To characterize the surface morphology of the fabricated TiO2 and the enzymemodified Si3N4, scanning electron microscopy (SEM) images were taken with a Schottky field-emission microscope (JSM-7800F, JEOL GmbH, Freising, Germany). For higher conductivity, a ~5 nm platinum-palladium layer was sputtered onto the Si3N4 surface before SEM images were taken.

To achieve a high spatial resolution of the LAE, a low current in the absence of illumination is required. This can be achieved with a dense and non-porous TiO2 layer to avoid short circuits with the SnO2:F glass [29]. Additionally, it is important to exclude a direct contact of the analyte with the highly-doped SnO2:F layer, to circumvent unexpected surface reactions. A representative SEM image of the TiO2 surface is given in Figure 2a. The image shows a homogeneous and dense surface structure without visible cracks. The granularity (200–250 nm) is induced by the SnO2:F glass on which the ~200 nm TiO2 layer is deposited.

Exemplary SEM images of the TMV-modified (carrying the penicillinase molecules) LAPS-Si3N4 surface with part of the microfluidic channel are shown in Figure 2b (left and right). The channel boundary is cleanly cut with no visible fringes. On the Si3N4, the white cloud-like area is indicating the TMV-modified surface spot (left image). A zoom into this spot (right image) shows homogeneously distributed TMV particles with immobilized penicillinase. The TMV particles appear as typical 300 nm long nanotubes, as well as in shorter particle fractions (~50–200 nm) or elongated "end-to-end"—multimer structures (up to ~600 nm) as described in previous works as enzyme nanocarriers on a Ta2O5-sensor surface [24,26].

**Figure 2.** (**a**) Scanning electron microscope (SEM) image (magnification of 20,000× of the LAE showing the TiO2 film surface on the SnO2:F glass substrate. (**b**) SEM image depicting the part of the microfluidic channel where the enzyme-modified Si3N4 surface of the LAPS is located with a magnification of 60× (left) and with a zoom-in, showing the adsorbed TMV particles carrying the immobilized penicillinase with a magnification of 35,000× (right).

#### *3.2. Penicillin Detection with Penicillinase-Modified LAPS*

TMV particles were loaded inside the microchannel by drop-coating with subsequent penicillinase coupling by affinity binding of SA-penicillinase conjugates to the biotinylated TMV. Because modified TMV particles have been utilized for the first time inside a microfluidic channel for enzyme immobilization, chemical images and photocurrentvoltage curves were recorded by the LAPS to control the layout's functionality for penicillin detection. During the enzymatic conversion of penicillin to penicilloic acid, H+ ions are generated resulting in a local pH change in the solution. As a first experiment, the pH change resulting from varying penicillin concentrations was studied as chemical images inside the microchannel. The rear side of the LAPS is therefore scanned sequentially, with the resulting photocurrent depicted in Figure 3a. Each chemical image represents (from top to bottom) a different penicillin concentration (from 0.1 mM to 5.0 mM) as differential image. This differential image is obtained by subtracting the particular chemical image from the reference chemical image of the microfluidic structure recorded at an applied potential of −1.65 V in 0.33 mM PBS buffer at pH 7.0. For the reference, (Figure S1 (Supplementary Information)), a high flow rate of the analyte of 1.0 <sup>μ</sup>L·s<sup>−</sup><sup>1</sup> was chosen to suppress any pH changes inside the channel. During the enzymatic experiments, the flow was stopped, allowing an accumulation of enzymatically produced H+ ions. The resulting differential chemical images (Figure 3a) show a section of the microfluidic channel with 192 measurement points, visualizing a total area of 8.0 × 1.5 mm2. The applied potential of −1.65 V was selected to be close to the inflection point of the photocurrent–voltage curve (Figure 3b).

In Figure 3a, the bottom image shows the result for a penicillin concentration of 5 mM. Due to the H+ ion generation, the ΔPhotocurrent (ΔIphoto) decreased in the area with immobilized enzyme. The diameter of the pH-change region (~4.0 mm), corresponds to the drop-coated area of the TMV particles with immobilized penicillinase. In this area, ΔIphoto changed (decreased) by 7.3 ± 0.9 nA for a penicillin concentration of 5.0 mM. For lower penicillin concentrations, the ΔIphoto variations have been 6.0 ± 0.7 nA (1.0 mM) and 4.0 ± 0.5 nA (0.5 mM). For the lowest penicillin concentration (0.1 mM), a small photocurrent (ΔIphoto = 0.5 ± 0.0 nA) was detected. Here, due to the ΔIphoto scaling of the depicted chemical image, the change is hardly visible. Interestingly, the spatial pH change expansion for all concentrations was—more or less—in the same local area at 4.0 mm width (*x*-axis) and equally distributed indicating a rather low H+ ion diffusion away from the enzyme into the surrounding medium; only the photocurrent intensity changed by varying penicillin concentrations. Besides offering a mapping of the enzyme activity inside the microchannel, the chemical image mode can be used to determine the exact spots of immobilized enzymes. Moreover, possible enzyme detachment (e.g., due to shear stress induced by high flow rates) would be directly recognized.

**Figure 3.** (**a**) Chemical images and (**b**) photocurrent-voltage curves for penicillin concentrations ranging from 0.1 to 5.0 mM after 5 min of enzymatic reaction in phosphate buffered saline (PBS) buffer, pH 7.0. (**c**) Mean calibration curve evaluated from the photocurrent-voltage curves (*n* = 4) with an average penicillin sensitivity of 42.3 mV/dec. The inlet represents the photocurrent change in dependence of the penicillin concentration, evaluated from Figure 3a.

In addition to the chemical images, for photocurrent–voltage curves, the photocurrent is recorded at a defined location inside the microchannel, while sweeping the applied bias potential from 0 to −3.0 V. The measurement spot with an illumination size of 250 × 250 μm<sup>2</sup> was located in the center of the determined pH-change area (x-axis 4.0 mm, y-axis 0.75 mm). Figure 3b displays the normalized I-V curves related to the previously discussed chemical images for different penicillin concentrations. The I-V curves exhibit the characteristic regions with inversion, depletion and accumulation. For example, the blue I-V curve represents the previously described reference measurement during vigorous flushing of the channel with PBS buffer solution (1.0 <sup>μ</sup>L·s<sup>−</sup>1). In the diagram, from −3.0 V to −2.4 V, the n-type semiconductor is in the inversion state. In the depletion region, the photocurrent decreases until an applied voltage of −1.0 V and reaches its minimum due to charge accumulation for further increasing bias potentials. During the enzymatic reaction, the H+ ion generation leads to a surface protonation of the Si3N4, followed by a shift of the photocurrent-voltage curve to more negative potentials. The potential changes were taken from the inflection point at the normalized photocurrent of 0.5. In contrast to the chemical image, a potential change (ΔU) of 8.0 ± 1.7 mV with respect to the reference I-V curve was detected even for 0.1 mM penicillin. For higher penicillin concentrations, the signal shift increased to 42.8 ± 6.2 mV, 62.3 ± 3.1 mV and 78.8 ± 0.8 mV for 0.5 mM, 1.0 mM and 5.0 mM, respectively. The evaluated calibration curve is depicted in Figure 3c. A mean penicillin sensitivity in the concentration range from 0.1 to 5.0 mM of 42.3 mV/dec was achieved.

The experiments highlight, that the combination of LAPS and penicillinase-functionalized TMV can be used for the detection of penicillin inside a microfluidic channel: here, a twodimensional mapping in x- and y-directions is possible.

#### *3.3. Impact of pH Changes on Penicillinase Activity*

A main aim of this study is to control the rate of enzymatic conversion by locally induced pH changes with the LAE. Therefore, the activity of penicillinase for pH values ranging from pH 4.0 to pH 7.0 in 0.33 mM PBS buffer with a constant penicillin concentration of 1.0 mM was characterized. The differential chemical images in Figure 4a visualize typical pH changes due to H+ ion accumulation after stopping the enhanced flow of 1.0 <sup>μ</sup>L·s<sup>−</sup>1. All results were obtained 5 min after the flow stopped. For pH 4.0, there is nearly no pH change, and thus no change in photocurrent ( ΔIphoto = 0.4 nA ± 0.0) detected, which can be attributed to the inhibition of the enzyme at a such low pH value (see also activity behavior of penicillinase [30]. For pH 5.0, a slight variation in photocurrent of 2.1 ± 0.1 nA occurs, indicating a low enzymatic activity. For higher pH values of pH 6.0 and pH 7.0, ΔIphoto is increased to 4.3 ± 0.4 nA and 5.9 ± 0.6 nA, respectively. Here, the pH values are closer to the penicillase's activity optimum of approximately pH 7.5 (for immobilized enzyme) [31], resulting in a higher catalytic conversion of penicillin.

These results are confirmed by the associated photocurrent-voltage (I-V) curves, recorded in the center of the area with immobilized enzyme (x-axis 4.0 mm, y-axis 0.75 mm). The corresponding I-V curve for pH 4.0 PBS buffer solution is shown in Figure 4b. Between the reference (blue) and 1.0 mM penicillin I-V curve (orange), there is only a marginal shift of 6.8 ± 3.6 mV. Based on the original pH sensitivity of the sensor of 40 mV·pH−<sup>1</sup> (data not shown), this refers to an additional pH drop of 0.17. For pH 7.0 (Figure 4c), the I-V curve shifted by 61.0 ± 0.7 mV, which is equivalent to a pH change of 1.5 from pH 7.0 to pH 5.5. Lowering the pH values in relation to pH 7.0, the photocurrent–voltage curves shifted to more negative voltages by 49.8 ± 2.8 mV at pH 6.0 and 27.0 ± 3.0 mV for pH 5.0 (not shown). It should be mentioned that, beside the inhibited enzyme's activity at lower pH values, the PBS buffer capacity is also reduced at lower pH values. By that, the enzymatically produced H+ ions at pH 5 and pH 4 have a higher impact on the resulting pH change, than at higher pH values of pH 7 and pH 6.

**Figure 4.** (**a**) Chemical images for 1.0 mM penicillin in pH 4 to pH 8 PBS buffer after 5 min of enzymatic reaction. Photocurrent-voltage curve for (**b**) pH 4 and (**c**) pH 8 PBS buffer with and without 1.0 mM penicillin.

Both, the chemical images and the I-V curve reveal the possibility to inhibit the enzymatic reaction by lowering the pH in a microfluidic setup, where the chemical imaging allows determination of 2-dimensional distribution of pH-triggered enzyme activity.

#### *3.4. pH Manipulation with LAE*

By utilizing a LAE, a direct charge transfer at the semiconductor/electrolyte interface between generated holes and species in the solution is possible. Depending on the applied LAE potential, e.g., photoelectrocatalytic water splitting takes place, allowing a flexible pH-value adjustment inside the channel. First, a potential of 0.3 V is applied to the LAE against the Pt-counter electrode. A typical transient current response is rendered in Figure 5a. Without illumination, the current equilibrates at 18 nA (3 s). During this condition, no surface reactions are triggered. When illuminating the rear side of the LAE with an area of 0.25 × 1.0 mm2, a current peak occurs (4–5 s), which can be assigned to accumulated holes perturbing the surface, resulting in a capacitive discharge [32]. Afterwards, the current equilibrates at 1.04 μA after 50–60 s. Most of the current occurring during illumination can be assigned to the photoelectrocatalytic oxygen-evolution reaction of water where, besides oxygen, H+ ions are produced, resulting in a pH change. After switching-off the illumination, the current decreases again to its dark current value.

**Figure 5.** (**a**) Transient photocurrent signal for 60 s of illumination. (**b**) Chemical images of static (top) and dynamic (bottom) pH changes inside the microfluidic channel induced by the LAE.

Similarly to the H+ ions generated by the enzymatic reaction, it is possible to visualize the photoelectrocatalytically produced protons with differential chemical images by the LAPS. In Figure 5b, differential chemical images of a 2.0 × 10.0 mm<sup>2</sup> area and an applied potential of −1.45 V were recorded in PBS buffer, pH 7.1. The top image shows a pH change induced by the LAE without flow. It can be seen that in the illuminated area, the LAPS photocurrent changes by 10.7 ± 1.9 nA. The corresponding I-V curve reveals a shift of 134.4 mV to more negative voltages, which is equal to a pH decrease by ~3.5.

To exclude that the LAE illumination wavelength of 405 nm affected the functionality of the enzyme, the LAE illumination spot was positioned 3–4 mm downwards (in flow direction) from the enzyme. After 10 s of photoelectrocatalysis, a steady flow of 0.05 <sup>μ</sup>L·s<sup>−</sup><sup>1</sup> was applied, moving the generated protons upwards the channel to the location of the immobilized enzyme. Hereby, an equilibrium between the proton generation and transport is formed, resulting in a consistent pH change inside the channel. The result is depicted in the bottom image in Figure 5b. The differential chemical image shows an equally distributed pH variation with an average change of the LAPS photocurrent of 3.4 ± 0.6 nA. From the related I-V curve, a pH decrease of 1.0 (ΔU = 41.2 mV) was obtained. Additionally, the start of the flow can be seen in the transient current measurement in Figure 5a. After 10 s

of illumination, there is a small current increase. Since the volume above the illuminated area of 1.0 × 0.25 mm<sup>2</sup> is only 0.02 μL, due to the fluidic dimensions and no-flow conditions during the first 10 s, a reduced mass transfer can lead to a depletion of reaction partners, decreasing the current [33]. Hence, providing fresh solution with starting the flow after 10 s, the current increases to a nearly constant value of 1.04 μA, as the reaction rate between generated electron-hole pairs and reaction partners in solution equilibrate.

#### *3.5. Regulation of Enzyme Activity by the LAE*

In the final experiment, the enzymatic activity of penicillinase was directly regulated by the LAE inside the microfluidic channel. Here, photocurrent-time measurements (called as constant potential LAPS measurements) were performed to study the temporal photocurrent change during H+ ion generation of the enzymatic reaction. The applied potential was −1.45 V.

Figure 6a shows the photocurrent change for 1.0 mM penicillin in PBS buffer, pH 7.1, without manipulation of the pH with the LAE. The microchannel was rinsed with a pump rate of 1.0 <sup>μ</sup>L·s<sup>−</sup><sup>1</sup> for the first 60 s. After stopping the dosage, the photocurrent starts to decrease due to the accumulation of enzymatically produced H+ ions and reaches an equilibrium after around 300 s. The measured photocurrent drop of 7.0 nA corresponds to a pH change of 1.75. After 300 s, the flow was started again and the channel was rinsed with fresh solution, whereby the photocurrent increased again to its initial value.

**Figure 6.** Constant potential LAPS measurements. (**a**) Photocurrent response for 1.0 mM penicillin in PBS buffer, pH 7.1. In (**b**–**d**) the blue curves depict the transient photocurrent decrease due to pH changes induced by the LAE with an illumination width of 250 μm, 500 μm and 1500 μm, respectively. The orange curves show the concatenated, additional change in photocurrent when 1.0 mM penicillin is added to the PBS buffer.

In the following measurements, the pH was regulated by the LAE and the enzymatic response was determined by the LAPS. First, the LAE induced a pH change by photoelectrocatalytic water oxidation in PBS buffer without penicillin. Since a simultaneous operation between LAPS and LAE is not possible, due to the influence of the 405 nm light beam of the DLP projector on the LAPS chip during measurements, the illustrated curves in Figure 6b–d are concatenated and normalized: The photocurrent was measured during the 60 s of rinsing, stopped while the LAE pH changed, and directly started after the LAE illumination was switched off. During the LAE illumination, the output flow of the pump was reduced to 0.05 <sup>μ</sup>L·s<sup>−</sup>1. In Figure S2 (Supplementary Information), the effect of the lower flow velocity on the enzyme activity without LAE is analyzed. After changing the flow rate from 1.0 <sup>μ</sup>L·s<sup>−</sup><sup>1</sup> to 0.05 <sup>μ</sup>L·s<sup>−</sup>1, a small drop of ΔIphoto = 0.5 nA occurred. Nevertheless, the delta photocurrent after stopping the flow reached again 7.0 nA, which is identical to the photocurrent change without the decreased flow rate.

As the change of the pH is defined by the number of generated H+ ions during the photoelectrocatalytical water oxidation, this can be influenced by varying the reactive area of the LAE with differently sized illumination spots. These have been changed with illumination lengths inside the microchannel between 250 μm and 1500 μm with the help of the DLP projector.

The pH change for an illuminated area of 1.0 × 0.25 mm<sup>2</sup> is depicted in Figure 6b. The blue curve shows the constant photocurrent shift of 3.8 nA that corresponds to a pH shift from pH 7.1 to pH 6.3. Subsequently, the measurement was repeated with 1.0 mM penicillin in the PBS buffer (orange curve). Here, after the LAE-induced pH drop, the photocurrent further decreased until it reached an equilibrium after approximately 300 s. The total photocurrent change is 8.6 nA. This is equivalent to a pH change of 2.1. As the LAE altered the pH value to 6.3, the additional pH drop by the enzymatic reaction was 1.3. In Figure 6c, the LAE was illuminated with a beam width of 500 μm, leading to a photocurrent drop of 7.2 nA. With 1.0 mM penicillin, it changed by 9.4 nA. The pH therefore decreases from pH 7.1 to pH 5.4 and pH 4.8 after the LAE and penicillin reaction, respectively. The largest illumination width of 1500 μm resulted in a ΔIphoto of 10.3 nA, which corresponds to a pH change of 2.6 (Figure 6d). Since this change already results in pH 4.4 inside the microchannel, no further change in the photocurrent was observed while adding 1.0 mM penicillin in the solution. This indicates that the enzymatic catalysis of penicillin was inhibited. These results underline the high potential of the proposed combination of LAPS and LAE. The flexible generation of pH gradients, using a LAE by changing the illumination spot, offers the spatially resolved control of the enzyme activity inside the microfluidic channel. Furthermore, the triggered enzymatic reaction rate can be label-free monitored by the enzyme-LAPS to validate the resulting impact on the enzymatic inhibition.
