3.1. PBS at Various Concentration Levels
In order to verify the discrimination capability of the μEoN-DP, electrical impedances of PBS at various concentration levels were measured. The electrical impedance of each concentration level was measured 10 times, and their average values were depicted in
Figure 3. The standard deviations of the magnitude and phase were less than 2.23% and 0.82%, respectively, for every concentration level and at all the investigated frequencies showing a high discrimination capability.
For the magnitude, differences among each concentration level of PBS were observed at higher frequencies, and those for the phase were observed at mid-range frequencies. The high impedance at low frequencies can be explained by the capacitive property of the double layer, which can be modeled as a CPE (constant phase element) [
39]. The CPE impedance is expressed by
Z = 1/{
Y0·(
jω)
n}, whose parameters will be extracted through curve fitting.
In the majority of frequency range, as the concentration levels of PBS decreased, the magnitude and the phase increased. The larger real (
Z′) part of the impedance was observed in the lower concentration levels of PBS, which indicate that the conductivity becomes lower as concentration levels of PBS decrease. The strong dependency of the real part of the impedance on the frequency can be attributed predominantly to the parallel connection of the resistance (R) and capacitance (C). From
Figure 3, as the concentration decreased, the absolute values of the real and imaginary parts of the impedance increased, and their ratio became larger. Thus, the phase values at low concentration levels of PBS were larger than those at high concentration levels. As the frequency increased, the real part of the impedance became more dominant. At low concentration levels of 0.125× and 0.0625×, the increasing tendency of the phase reversed after 400 kHz and 630 kHz, respectively, because the capacitance decrease influences the imaginary part of the impedance more than the frequency increase does.
From the experimental results of the sensor output, the μEoN-DP was verified to discriminate dissimilar materials having different electrical properties. In order to quantify the extent of the discrimination capability of the μEoN-DP, discrimination index (
DI), which will be employed in the next section for biotissue discrimination, can be defined as follows:
The variables (
C,
D) and subscript (i) represent the mean impedance value, standard deviation, and concentration level of PBS (1:0.0625×, 2:0.125×, 3:0.25×, 4:0.5×, and 5:1×), respectively. The
DI value, obtained at all investigated frequencies, becomes larger as the difference of the mean values in the numerator increases and/or the sum of the standard deviations in the denominator decreases. The difference of electrical impedance between two concentration levels of PBS was divided by the sum of their standard deviations to normalize the
DI value. The
DIs (
DI1,
DI2,
DI3, and
DI4) were then summed and depicted, as shown in
Figure 4a. The optimal frequencies at which the discrimination capability can be maximized for the magnitude and phase were found at 1 MHz and 150 kHz, respectively; these are the frequencies at which the experimental curves in
Figure 3 are best distinguishable.
Although we evaluated the discrimination capability of the μEoN-DP using PBS at various concentration levels, the experimental results still included the device resistance and sample-device interactions, such as charge transfer resistance and double layer capacitance. These interactions should be minimized from the sensor output to obtain only the sample resistance and the capacitance by curve fitting with the equivalent circuit shown in
Figure 4b. The subscripts i and s stand for internal and sample, respectively.
Cs (the sample capacitance) is connected in parallel with
Rs (the sample resistance), and they are in series with
Ri (the internal resistance) and the CPE. The internal resistance was experimentally confirmed as approximately 100 Ω. The CPE was placed in an equivalent circuit to reflect the double layer capacitance, charge transfer resistance, electrode polarization, and rough surface of the electrodes [
39]. The extracted sample resistance (
Rs), sample capacitance (
Cs), and CPE parameters (
Y0 and
n) are summarized in
Table 1.
As the concentration decreased, the resistances increased, whereas the capacitance decreased. As shown in
Figure 4b, the sample resistances (
Rs) were inversely proportional and roughly linear to the decrease in concentration levels because the conductivity of deionized water can be assumed to be extremely low; therefore, the conductivity of the 1:1 mixture of 1× PBS and deionized water became half the conductivity of 1× PBS. However, the contribution of deionized water will not be negligible, as concentration significantly decreases. The extracted resistance and capacitance were in sufficient agreement with those theoretically estimated based on the mixing ratio of PBS and deionized water; discrepancies were less than 8.02% and 1.85%, respectively. As the concentration decreased, the influence of the CPE on the sensor output also decreased. At a concentration of 1×, the CPE impedance accounted for 99.9% at 100 Hz and 29.6% at 1 MHz in the capacitive effect on the sensor output, which decreased to 99.1% and 5.7% at the same frequencies at a concentration of 0.0625×.
In order to discuss the advantage of micro EIS over macro EIS, their specific electrical properties were investigated in terms of sensitivity, based on the equivalent circuit and extracted parameters from PBS experiments. "Microsized" is derived from the macro by miniaturizing the dimensions of the electrodes, whereas their effective sensing areas are kept constant. In a microscale, each magnitude contributed by the sample resistance (Rs) and capacitance (Cs) is smaller than that in a the macroscale, which was verified using a COMSOL simulation (COMSOL, Inc., USA). Thus, the sensitivity in a microscale tends to be larger than that in a macroscale, assuming that the same amount of perturbation in the magnitude of electrical impedance is given to the sample. This tendency becomes more prominent at high frequencies, which can explain the reason why the highest DI for the magnitude was found at 1 MHz.
3.2. Depth Profiling Into Four-Layered Porcine Tissue
In order to evaluate the effectiveness of the discrimination capability of the μEoN-DP for biotissues, depth profiling was conducted using four-layered (fat
1–muscle
1–fat
2–muscle
2) porcine tissue. Depth profiling was conducted up to 25 mm with a penetration interval of 1 mm. As shown in
Figure 2c, the fat
1 (8 mm), muscle
1 (7 mm), fat
2 (3 mm), and muscle
2 (7 mm) tissues were arranged in order within 25 mm of the porcine tissue.
The impedances measured in the same layer of porcine tissue were averaged, as shown in
Figure 5a,b. For the magnitude and phase, the maximum standard deviations of fat
1 tissue were 17.53% at 1.57 kHz and 6.55% at 6.33 kHz, respectively. The values for the muscle
1 tissue were 6.26% and 3.87%, both at 998 Hz, respectively. Fat
2 and muscle
2 showed similar extents of standard deviations. From the experimental impedances, it was confirmed that the magnitudes of fat tissue were higher than those of muscle tissues at all the investigated frequencies. This implies that more electrical current can flow through muscle tissues with less resistance, which was in good agreement with the reported study [
40].
To suggest the best discrimination results of depth profiling considering real-time applications for practical use, it is necessary to find the optimal frequency at which the discrimination capability is maximized. In order to find the optimal frequency, the
DI was evaluated for the porcine tissues. The
DIij for two of the dissimilar tissues is defined as the ratio of the absolute value of mean difference and the sum of the standard deviations, which is shown in Equation (2) as a generalized expression. The variables (
C,
D, and
N) represent the mean impedance value, standard deviation and the total number of tissues, respectively. The subscripts (f, m, i, and j) represent fat tissue, muscle tissue, i
th tissue, and j
th tissue, respectively.
The DI basically includes three values of DIij for three physical boundaries within 25 mm in the porcine tissue. Still, there is the fourth possible combination of dissimilar tissues, i.e., fat1 and muscle2, which is necessary to completely reflect the symmetric properties of DI. Thus, further comparison between fat1 and muscle2 tissues was also included in the definition of DI, even if there is no physical boundary between them. The optimal frequency was determined as the frequency at which sum of the four DIij was maximized. The magnitude differences between dissimilar tissues seem to be larger at low frequencies around 5 kHz compared to those at higher frequencies. However, the largest DIs for the magnitude and phase (271.51 and 98.03, respectively) were found both at 1 MHz because the mean difference values were divided into sum of the standard deviations in the definition of DI. It should be noted that the small value of the denominator implies small deviation, indicating a high repeatability of the experimental results.
The experimental results at the optimal frequency of 1 MHz were plotted as shown in
Figure 5c,d with respect to the penetration depth. For both the magnitude and the phase, there were significant differences between dissimilar tissues at the optimal frequency. However, slight discrepancies were observed between the similar tissues of fat
1 and fat
2, which will be investigated using a 3D COMSOL simulation in the next section.
3.3. Simulation Verification of Selective Passivation Layer
In order to verify the advantage of selective passivation over complete passivation in terms of the sensor output, the following two 3D COMSOL simulations were conducted: (a) only the sensing IDEs with SPL were exposed to biotissues; and (b) the entire electrode (IDEs and connection lines) with CPL were exposed to biotissues. This will also reveal the reason for the slight differences in the electrical impedance between fat
1 and fat
2 tissues. In the simulation, the material properties of fat
1 and fat
2 tissues were assumed to be identical in order to investigate only the influence of connection lines. As shown in
Figure 5c,d, the μEoN-DP with SPL showed clear discrimination results compared to the results with CPL.
For the case with the CPL, the magnitude decreased and the phase increased along with the penetration depth of the electrodes, which implies that the immersed connection lines in the biotissues considerably influenced the sensor output. The total output impedance becomes smaller as the penetration proceeds into the biotissues. This can be explained by the fact that the influence of the connection lines on the sensor output increases because the impedance of the connection lines and sensing IDEs are electrically connected in parallel. This decreasing tendency in total output impedance was more prominent in fat2 tissues compared to that in other tissues. The magnitude and phase of the fat2 tissues are similar to those of the muscle tissues, which makes it difficult to discriminate even between dissimilar tissues. Thus, a post-compensation process that can minimize the unwanted error induced from the connection lines is required to obtain more accurate results.
For the case with SPL, although the fat
1 and fat
2 tissues were assumed to have the identical electrical properties, a slight decrease in magnitude and increase in phase were observed in the simulation results as the penetration depth of the electrodes increased in the fat tissues. The slight discrepancies between the fat
1 and fat
2 tissues were also observed through simulations. The simulation results were in sufficient agreement with the experimental ones; the discrepancies were as small as 6.02% for both magnitude and phase. It is evident from the simulation and experimental results that the impedance variation in each tissue and discrepancies between fat
1 and fat
2 tissues can be accounted for the influence of connection lines on the sensor output; the SPL cannot perfectly block the alternating current through the connection lines. In the muscle tissues, the influence of the connection lines on the sensor output appeared to be negligible compared to that in the fat tissues. This can be explained by the fact that electric fields induced from the connection lines faces greater difficulty in passing through the muscle tissues than in passing through fat tissues. This is because the conductivity of muscle tissues is several hundred times higher than that of fat tissues, which implies that the muscle tissues can confine more electric fields within the SPL than fat tissues can. The dependencies of the magnitude on the type of passivation layer are listed in
Table 2.
With the help of the SPL, the relative influence of the connection lines on the sensor output can be effectively minimized without any post-compensation process. The magnitude decreased by 38.59% with the CPL as the penetration depth increased in the fat1 tissue, while the magnitude decreased only slightly by 8.70% with the SPL. In addition, the averaged magnitude with CPL is estimated through simulation to decrease by 90.14% between the fat1 and fat2 fat tissues, whereas the averaged magnitude with the SPL was experimentally confirmed to decrease by only 4.25%.
These results suggest that the μEoN-DP could clearly discriminate between dissimilar tissues so that the maximum depth of buried tumors during partial nephrectomy can be located to reduce the positive surgical margin ratio. In the case of laparoscopic surgery, an iatrogenic organ injury caused by Veress needle insertion can be promptly detectable, enabling immediate management to reduce perioperative complications. In addition, the μEoN-DP can be utilized as a spinal needle capable of locating the exact position for drug delivery during spinal or epidural anesthetic procedures.