2.3.1. Measurement Setup

Impedance measurements were performed with the IMPSPEC device from Meodat (Ilmenau, Germany), which was designed for fast and broadband impedance spectroscopy. The starting frequency was set to 5.895 Hz. The frequency was increased linearly by adding up frequency intervals of 5.895 Hz, until the final frequency of about 10 kHz was reached. The real part and the imaginary part of the impedance were displayed for each frequency. The impedance biosensor chip was connected to the IMPSPEC device cable by means of conductive silver and crocodile clamps. A peristaltic pump delivered the liquid samples through the PDMS channel and across the electrodes.

#### 2.3.2. HSA Adsorption for Testing SAMs

PBS redox was prepared by adding potassium hexacyanoferrate(II) and potassium hexacyanoferrate(III) to PBS to a final concentration of 15 mM each. HSA was dissolved in PBS redox at concentrations of 0/1/10/100/1000 ng/mL, resulting in five HSA samples. Starting with a zero sample containing PBS redox only and corresponding to HSA, 0 ng/mL, the four other HSA samples were subsequently applied with increasing HSA concentration to the same impedance biosensor chip. Each sample was applied for 2 min at a flow rate of 0.04 mL/min. After sample application, the pump was stopped, and after 30 s, the impedance measurement was performed before the next sample was applied. After measuring the last HSA sample (1000 ng/mL), the impedance biosensor chip was disposed, and a new biosensor chip was used.

#### 2.3.3. HSA Adsorption and Troponin I Assay

An impedance biosensor chip with freshly prepared antibody coating (see Section 2.2) and bonded PDMS microfluidic channel (see Section 2.1.2) was rinsed with PBS redox for 2 min at a flow rate of 0.04 mL/min. After switching off the pump and waiting for 30 s, the impedance of the antibody-coated biosensor chip was measured. After that, HSA dissolved at 1000 ng/mL in PBS was applied to the biosensor chip at a flow rate of

0.04 mL/min. Sampling time was 8 min if the antibody coating was performed without SAM (see Section 2.2.1) or 3 min if the antibody coating was performed via SAM (see Sections 2.2.2 and 2.2.3). This was followed by PBS redox rinsing for 2 min at the same flow rate. The pump was stopped, and after 30 s, the effect of HSA on the respective biosensor impedance was measured. Finally, troponin I (Fisher Scientific, Schwerte, Germany) dissolved at 1 ng/mL in PBS was applied to the biosensor chip for 1 min at 0.04 mL/min, followed by PBS redox for 2 min at the same flow rate. The pump was stopped, and after 30 s, the impedance resulting from troponin I binding on the respective biosensor chip was measured. After measuring the troponin I sample, the impedance biosensor chip was disposed, and a new biosensor chip was used.

#### **3. Results and Discussion**

#### *3.1. Basic Performance of the Microfluidic Impedance Biosensor Chip*

A preliminary experiment to test the performance of the microfluidic impedance bio-sensor chip was carried out by simply adsorbing the antibody on the working electrode, without the use of any SAM. While the working electrode was coated with anti-troponin I, the counter electrode remained uncoated. After application of HSA, 1000 ng/mL, for blocking remaining potentially accessible adsorption sites, troponin I, 1 ng/mL, was applied on the sensor surface. The results are shown in Figure 3.

**Figure 3.** Nyquist plots of an impedance biosensor chip with no intermediary self-assembled monolayer (SAM) for troponin I detection. Anti-troponin I was adsorbed on the working electrode, while the counter electrode remained uncoated. Samples containing human serum albumin (HSA), 1000 ng/mL, and troponin I, 1 ng/mL, were applied subsequently to the biosensor chip.

The resulting Nyquist plots show almost ideal semicircles, i.e., charge transfer kinetics prevails the diffusion to the layer. Changes in the double layer, such as affinity binding to the surface, can be observed. Hence, the electrode design of the microfluidic impedance biosensor chip together with the measurement protocol with the redox buffer allows basic biosensor measurements. The recording of the impedances requires only a few seconds, the sampling time a few minutes.

The initial impedance after antibody adsorption on the working electrode was 11 k Ω. It increased only slightly to 12 k Ω after HSA blocking, and the change in the complete Nyquist plot was negligible. After troponin I binding, however, the initial impedance increased to 18 k Ω, and a significant shift in the Nyquist plot was observed. This demonstrates that the chosen anti-troponin I is able to bind troponin I, while nonspecific HSA adsorption is blocked. Though this assay even worked with adsorbed antibody, antibody immobilization via SAM was to be performed in the following to show that the newly

developed microfluidic impedance biosensor chip is also able to handle standard layer setups generally recommended for impedance biosensors.

#### *3.2. Testing Thiol-SAMs Based on Aliphatic and Aromatic Hydrocarbon Spacer*

SAMs on working and counter electrode are recommended to minimize nonspecific protein binding on the electrode surface, as this would lead to false positive signals, particularly in real samples with high concentrations of nonanalyte proteins. Furthermore, a suitable SAM on the working electrode needs to provide functional groups for antibody immobilization. To test the performance of potential SAMs with the microfluidic impedance biosensor chip, both working and counter electrode were coated with the respective thiol compound. After the SAM coating, samples containing increasing concentrations of HSA were successively applied on the electrodes. The results are shown in Figure 4.

**Figure 4.** Nyquist plots of impedance biosensor chips for testing SAMs with HSA samples. Both working and counter electrode of such a chip were coated with (**a**) 4-mercaptobenzoic acid, (**b**) 1,4-benzenedithiol, or (**c**) 6-mercapto-1-hexanol. Samples containing HSA in increasing concentrations were applied subsequently to the biosensor chips.

Thiols with aromatic hydrocarbon spacers were chosen as they promise low initial impedance [12] and, hence, are beneficial for the charge transfer through the sensing layer. Both 4-mercaptobenzoic acid (Figure 4a) and 1,4-benzenedithiol (Figure 4b) resulted in SAMs yielding almost ideal semicircles except for a few frequencies at the beginning, confirming the possible charge transfer.

The initial impedance of the 4-mercaptobenzoic acid SAM was 4.9 kΩ, after HSA adsorption it was 6.6 kΩ. i.e., the shielding against nonspecific adsorption was not perfect. However, 4-mercaptobenzoic acid offers a carboxyl group allowing further protein coupling by simple carbodiimide chemistry. The additional immobilization steps could increase the density of the sensing layer in a way that nonspecific HSA adsorption would be reduced to a greater extent. For that reason, experiments with 4-mercaptobenzoic acid as SAM on the working electrode were continued (see Section 3.3).

The initial impedance of the 1,4-benzenedithiol SAM was 12 kΩ. This was higher than that obtained with 4-mercaptobenzoic acid, but it remained in this range after HSA sampling. Both 4-mercaptobenzoic acid and 1,4-benzenedithiol were applied in the same concentration (see Section 2.2.2), and their structure differs only in one functional group. However, 1,4-benzenedithiol has two thiol groups located *para* to each other which promotes the parallel or close to parallel orientation of the molecule on the gold surface [26,27], which is not the case for 4-mercaptobenzoic acid. Hence, a better surface coverage is obtained with 1,4-benzenedithiol, which makes it an ideal protection layer against unwanted protein adsorption.

Except for a few frequencies at the beginning, almost ideal semicircles in the Nyquist plot were also obtained for the 6-mercapto-1-hexanol SAM (Figure 4c), showing that the aliphatic hydrocarbon spacer is short enough to allow charge transfer through the sensing layer. The initial impedance was 5.2 kΩ, after HSA adsorption it was 7.3 kΩ. Hence, the results after HSA sampling were similar to those obtained with 4-mercaptobenzoic acid (Figure 4a) regarding both the initial impedances and the insufficient shielding against nonspecific protein adsorption. Better shielding might be achieved by using a thiol with a longer hydrocarbon chain, but this bears the risk of impeding the charge transfer. As 6-mercapto-1-hexanol in this study was not supposed to act as single SAM but as cocomponent in a DNA-based SAM (see Section 3.4), it is more important that the impedance values obtained with 6-mercapto-1-hexanol are not too high. This requirement is fulfilled, as shown above.

#### *3.3. Troponin I Assay Using Thiol-SAMs Based on Aromatic Hydrocarbon Spacer*

The working electrode was functionalized with 4-mercaptobenzoic acid acting as SAM with low impedance and providing functional groups for covalent antibody coupling. The counter electrode was coated with 1,4-benzenedithiol, as this showed the best shielding abilities against nonspecific protein adsorption compared with the other SAMs used here (see Figure 4). After antibody immobilization, samples containing HSA, 1000 ng/mL, and troponin I, 1 ng/mL, were applied successively to the microfluidic impedance biosensor chip. Figure 5 shows the results obtained with this assay.

**Figure 5.** Nyquist plots of an impedance biosensor chip with aromatic hydrocarbon-based SAM for troponin I detection. The working electrode was coated with anti-troponin I, which was immobilized on a SAM consisting of 4-mercaptobenzoic acid. The counter electrode was coated with 1,4-benzenedithiol. Samples containing HSA, 1000 ng/mL, and troponin I, 1 ng/mL, were applied subsequently to the biosensor chip.

The initial impedance after antibody coating was 15 kΩ. Antibody immobilization, however, did not reduce HSA adsorption as anticipated. Instead, after applying

1000 ng/mL HSA, the initial impedance almost doubled (27 k Ω) and remained in this range after troponin I, 1 ng/mL, was applied. Hence, troponin I detection was not possible with this impedance biosensor, at least not at that concentration. The functionality of the anti-troponin I used here was confirmed before (see Figure 3). It is unlikely that the antibody was harmed by the covalent coupling protocol, because this is a standard procedure for protein immobilization. Furthermore, the suitability of this approach is confirmed in the next section (Section 3.4). However, the accessibility of the antibody binding sites may be hindered by the comparatively large amount of nonspecifically adsorbed HSA.

A closer look at the Nyquist plots shows significantly increased linear ranges than obtained before. This is characteristic for the Warburg impedance and indicates that the diffusion of the charge carriers prevails over the charge transfer kinetics. Hence, in order to promote the charge transfer, another SAM was to be applied. SAMs with hydrocarbon spacers of larger chain lengths are favorable regarding prevention of nonspecific protein adsorption, but may result in high initial impedances hindering the charge transfer even more. As wet chemistry methods are preferred due to the less complex coating procedures, conductive polymers are not tested here. Instead, oligonucleotide spacers were tested, as DNA strands represent large chains with negative backbone, promising a lower initial impedance.

#### *3.4. Troponin I Assay Using Thiol-SAMs Based on DNA Spacer*

The SAM on the working electrode was formed by ssDNA carrying a thiol group and 6-mercapto-1-hexanol serving as coimmobilization agent. The latter was added to support the alignment of the ssDNA strands perpendicular to the surface so that the subsequent hybridization is not hindered [13,14]. After hybridization of the thiolated ssDNA with the complementary ssDNA carrying an amino group, functional groups for antibody immobilization were available (for details see Section 2.2.3). The counter electrode was coated with 1,4-benzenedithiol as before, because of the excellent shielding abilities. After antibody immobilization on the working electrode, HSA and troponin I samples were applied successively to the microfluidic impedance biosensor chip. The results are shown in Figure 6.

**Figure 6.** Nyquist plots of an impedance biosensor chip with DNA-based SAM for troponin I detection. The working electrode was coated with anti-troponin I, which was immobilized on a SAM consisting of thiolated DNA and 6-mercapto-1-hexanol. The counter electrode was coated with 1,4-benzenedithiol. Samples containing HSA, 1000 ng/mL, and troponin I, 1 ng/mL, were applied subsequently to the biosensor chip.

In contrast to the sensing layer based on a SAM with aromatic hydrocarbon spacer (Figure 5), the Nyquist plots obtained with the sensing layer using a DNA-based SAM showed ideal semicircles (Figure 6), indicating an enhanced charge transfer. After antibody coupling, the initial impedance was 21 k Ω. Application of the HSA sample, 1000 ng/mL, did not have a significant effect on the impedance values. Sampling with troponin I, 1 ng/mL, however, led to an initial impedance of 33 k Ω and a significant shift in the Nyquist plot. Hence, aside from kinetically controlled charge transfer, the DNA-based SAM allowed both shielding against nonspecific protein adsorption and detection of the cardiac marker corresponding to the antibody in a clinically relevant concentration. Furthermore, it is confirmed that the carbodiimide coupling procedure used here did not affect the integrity of the antibody.

Finally, 1,4-benzenedithiol was used to replace 6-mercapto-1-hexanol as coimmobilization agent. This allowed the addition of potential benefits arising from the aromatic nature of the hydrocarbon spacer. Furthermore, this would simplify the coating procedure with regard to the use of fewer chemicals. The further immobilization and assay procedures remained the same. The results are shown in Figure 7.

**Figure 7.** Nyquist plots of an impedance biosensor chip with DNA-based SAM for troponin I detection. The working electrode was coated with anti-troponin I, which was immobilized on a SAM consisting of thiolated DNA and 1,4-benzenedithiol. The counter electrode was coated with 1,4-benzenedithiol. Samples containing HSA, 1000 ng/mL, and troponin I, 1 ng/mL, were applied subsequently to the biosensor chip.

Again, ideal semicircles were obtained in the Nyquist plot, confirming the improved charge-transfer in DNA-based sensing layers. The initial impedances decreased to 14 k Ω and remained in this range after application of the HSA sample, which led only to a negligible shift in the Nyquist plot. An initial impedance of 20 k Ω and a significant shift of the Nyquist plot was obtained by applying the troponin I sample. Though 1,4- benzenedithiol does not have the linear chain structure like 6-mercapto-1-hexanol, it still stabilizes the thiolated ssDNA strands in a way that the accessibility of the binding sites in the sensing layer is not hindered. As a result, the troponin I assay was also performed successfully with this DNA-based sensing layer.

The next step would include tests in real sample media, such as blood and blood derivatives (serum, plasma) or saliva. Those media contain enzymes that may degrade the DNA-based sensing layer. It has to be tested whether this effect is notable already at a sampling time of 3 min (see Section 2.3.3). If so, the sampling time could easily be reduced. Furthermore, the addition of ethylenediaminetetraacetic acid (EDTA) to the sample or the use of EDTA plasma instead of serum could be considered. EDTA complexes Ca2+ and Mg2+ which are required as cofactors for the DNA-degrading enzymatic reaction. Hence, this complexation would reduce the interfering effects of the DNA-degrading enzymes, while troponin detection is still possible as is the detection of other proteins [28–30].
