Physical Surface Modification of Carbon-Nanotube/Polydimethylsiloxane Composite Electrodes for High-Sensitivity DNA Detection

The chemical modification of electrode surfaces has attracted significant attention for lowering the limit of detection or for improving the recognition of biomolecules; however, the chemical processes are complex, dangerous, and difficult to control. Therefore, instead of the chemical process, we physically modified the surface of carbon-nanotube/polydimethylsiloxane composite electrodes by dip coating them with functionalized multi-walled carbon nanotubes (F-MWCNTs). These electrodes are used as working electrodes in electrochemistry, where they act as a recognition layer for sequence-specific DNA sensing through π–π interactions. The F-MWCNT-modified electrodes showed a limit of detection of 19.9 fM, which was 1250 times lower than that of pristine carbon/polydimethylsiloxane electrodes in a previous study, with a broad linear range of 1–1000 pM. The physically modified electrode was very stable during the electrode regeneration process after DNA detection. Our method paves the way for utilizing physical modification to significantly lower the limit of detection of a biosensor system as an alternative to chemical processes.


Introduction
Highly sensitive and selective detection of specific DNA sequences is crucial for biotechnological applications such as clinical diagnosis [1,2] and environmental [3,4] and food monitoring [5]. Therefore, extensive efforts have been devoted to developing techniques for detecting highly sensitive sequence-specific DNA sensors based on optical [6,7], electrical [8,9] and electrochemical [1,3,10,11] methods. Among these techniques, electrochemical methods have received considerable attention because they can be used to fabricate fast, simple, highly sensitive, and miniature DNA sensors [12,13]. Electrochemical impedance spectroscopy (EIS) is a very accurate electrochemical method, which is sensitive to changes on the electrode surface and is suitable for label-free DNA detection [14,15].
In several studies, EIS-based sensors have been fabricated using a gold electrode as the working electrode [16][17][18]. However, gold is very expensive and a laborious and inefficient process for immobilizing the probe is usually required for gold electrodes, which is unfavorable for commercialization.
Additionally, as an alternative to glassy carbon electrodes (GCEs), Au electrodes are extensively used as working electrodes in electrochemical DNA detection. However, because GCEs themselves do not recognize DNA molecules, chemical processes for immobilizing probes, which are more complicated than those for Au electrodes, are required when GCEs are used for electrochemical DNA detection [19][20][21]. Moreover, to further reduce the limit of detection (LOD), chemical surface modification can involve other nanomaterials, such as carbon nanotubes [22] and graphene [23]. In these cases,

Materials and Reagents
The MWCNTs (diameter: 10-15 nm; length: 30-40 µm) were obtained from Hanwha Chemical (Daejeon, Korea) and the DNA samples, including probe DNA (P), complementary target DNA (T1), one base non-complementary target DNA (T2), and noncomplementary target DNA (T3), were obtained from Bionics (Seoul, Korea) ( Table 1). Nitric acid (HNO 3 , purity 69%) was obtained from Avantor (Radnor, PA, USA) and sulfuric acid (H 2 SO 4 , purity 98%) was obtained from Daejung (Siheung, Korea). The polydimethylsiloxane (PDMS) and potassium ferrocyanide (K 4 Fe(CN) 6 ) were obtained from Sigma-Aldrich (St. Louis, MO, USA). The potassium ferricyanide (K 3 Fe(CN) 6 ) was obtained from Junsei Chemical Co., Ltd. (Tokyo, Japan), membrane filters were obtained from Sigma-Aldrich (St. Louis, MO, USA) and 1 × phosphate-buffered saline (1 × PBS) solution (pH = 7.4) was obtained from Tech and Innovation Corporation (Chuncheon, Korea). Absolute ethanol was obtained from Fisher Scientific Inc. (Hampton, NH, USA) and isopropyl alcohol (IPA) solution was obtained from Sigma-Aldrich (St. Louis, MO, USA). The resistivity of deionized water (DI water) used throughout this study was 18.2 MΩ·cm. The Ag/AgCl reference electrode, with a potential of 0.197 V vs. SHE, and platinum wire, which was used as a counter electrode, was obtained from Princeton Applied Research (Oak Ridge, TN, USA). All other chemicals were of analytical reagent grade. To prepare a sample solution for the DNA detection experiments, target DNA with a specific concentration was added to the PBS solution containing 1000 nM probe DNA and 4 mM potassium ferrocyanide/potassium ferricyanide.

Instruments
All electrochemical methods, including EIS, cyclic voltammetry (CV), and differential pulse voltammetry (DPV) were measured using a CHI622D (CH Instruments, Inc., Austin, TX, USA). A tip sonicator (HD 2070, Bandelin sonopuls, Berlin, Germany) was utilized to disperse the MWCNT solvent. A dip coater (PTL-UMB, MTI Co., Richmond, CA, USA) was used to modify the MWCNT/PDMS electrode surface with the F-MWCNT solution and a UV-visible spectrophotometer (Cary 50, Varian, Mulgrave, Australia) was used to measure ultraviolet (UV) absorption spectroscopy. A scanning electron microscope (SEM; S4800, Hitachi, Tokyo, Japan) was used to obtain cross-sectional images and surface morphology of the F-MWCNT/MWCNT/PDMS electrodes and Fourier-transform infrared spectroscopy (FTIR) was used to obtain an infrared spectrum of transmittance of the F-MWCNT solution (L160000A, PerkinElmer, Waltham, MI, USA).

Preparation of Functionalized CNTs
The preparation of the functionalized CNTs is summarized in Figure S1a. First, the MWCNTs (0.2 g) were sonicated in a 3:1 (v/v) mixture of nitric acid and sulfuric acid solution [33]. Subsequently, the MWCNTs were filtered and rinsed with DI water using vacuum filtration until the pH of the filtrate was neutral. Finally, the filtrate was dried in an oven (Figure S1b) and sonicated for 5 min with 60 mL of DI water to make the F-MWCNT solution. As shown in Figure S1c, our F-MWCNTs were uniformly dispersed in DI water because they have better dispersion than MWCNTs in water [34]. The FTIR spectra (Figure 1), showed that there were O-H, C=O, and C-O vibrations at 3372.11, 1642.21, and 1226.28 cm −1 , respectively, which clearly indicated that carboxylic and hydroxyl groups had been successfully attached to the surface of the MWCNTs.

Instruments
All electrochemical methods, including EIS, cyclic voltammetry (CV), and differential pulse voltammetry (DPV) were measured using a CHI622D (CH Instruments, Inc., Austin, TX, USA). A tip sonicator (HD 2070, Bandelin sonopuls, Berlin, Germany) was utilized to disperse the MWCNT solvent. A dip coater (PTL-UMB, MTI Co., Richmond, CA, USA) was used to modify the MWCNT/PDMS electrode surface with the F-MWCNT solution and a UV-visible spectrophotometer (Cary 50, Varian, Mulgrave, Australia) was used to measure ultraviolet (UV) absorption spectroscopy. A scanning electron microscope (SEM; S4800, Hitachi, Tokyo, Japan) was used to obtain cross-sectional images and surface morphology of the F-MWCNT/MWCNT/PDMS electrodes and Fourier-transform infrared spectroscopy (FTIR) was used to obtain an infrared spectrum of transmittance of the F-MWCNT solution (L160000A, PerkinElmer, Waltham, MI, USA).

Preparation of Functionalized CNTs
The preparation of the functionalized CNTs is summarized in Figure S1a. First, the MWCNTs (0.2 g) were sonicated in a 3:1 (v/v) mixture of nitric acid and sulfuric acid solution [33]. Subsequently, the MWCNTs were filtered and rinsed with DI water using vacuum filtration until the pH of the filtrate was neutral. Finally, the filtrate was dried in an oven ( Figure S1b) and sonicated for 5 min with 60 mL of DI water to make the F-MWCNT solution. As shown in Figure S1c, our F-MWCNTs were uniformly dispersed in DI water because they have better dispersion than MWCNTs in water [34]. The FTIR spectra ( Figure  1), showed that there were O-H, C=O, and C-O vibrations at 3372.11, 1642.21, and 1226.28 cm −1 , respectively, which clearly indicated that carboxylic and hydroxyl groups had been successfully attached to the surface of the MWCNTs.  Figure 2. First, the MWCNT composite electrode layer was fabricated according to a previous study's methodology [28], which is explained in detail in Section 1 of the Supplementary Materials. Next, we  Figure 2. First, the MWCNT composite electrode layer was fabricated according to a previous study's methodology [28], which is explained in detail in Section 1 of the Supplementary Materials. Next, we coated the MWCNT/PDMS film with F-MWCNT solution according to the preparation method described above. Finally, the film was dip-coated at a dipping speed of 500 µm/s. In our experiments, the optimum repetition number for the same dip coating was found to be 60. coated the MWCNT/PDMS film with F-MWCNT solution according to the preparation method described above. Finally, the film was dip-coated at a dipping speed of 500 μm/s In our experiments, the optimum repetition number for the same dip coating was found to be 60.

Measurement Using EIS
Before the DNA detection experiment using EIS measurement, a sample solution was prepared by including one of the DNA targets (T1, T2, and T3) with a specific concentration of 1 nM DNA probe (P) and 4 mM [Fe(CN)6] 3−/4− in 1mL of PBS solution. The sample solvent was maintained at 25 °C for 7 min because the UV absorption of the sample solution was saturated after 7 min ( Figure S2), which indicated that the DNA reaction in the solution had stabilized. After starting the EIS measurements, 9 min and 0.26 V were required to obtain the stabilized value of the charge transfer resistance (Rct) ( Figure S3) Thus, a preparation time of 16 min was needed for EIS-based DNA detection in our experiments. Our three-electrode system, which consisted of an Ag/AgCl reference electrode, platinum counter electrode, and 5 mm × 5 mm F-MWCNT/MWCNT/PDMS working electrode, was placed in sample solutions, as shown in Figure S4. The EIS measurements were performed in the frequency range of 0.1 Hz to 10000 Hz, with an AC (alternating current) amplitude of 5mV and a DC (direct current) bias of 0.26 V for DNA detection.

SEM Measurments of the Composite Electrode
The cross-section and surface morphology of the F-MWCNT-modified and pristine MWCNT/PDMS electrodes were characterized using SEM. For the pristine MWCNT/PDMS electrode, most MWCNTs were buried in PDMS and only a few tubular structures of MWCNTs were visible near the electrode surface, as shown in Figure 3a After 60 dip-coating cycles with the F-MWCNT solution, many F-MWCNTs were observed on the surface of the F-MWCNT-modified MWCNT/PDMS electrode (see Figure  3b). These may act as active centers for DNA interactions. Figure 3c shows the cross-section of the F-MWCNT/MWCNT composite electrode. The results indicate that the mean thickness of the PDMS layers was approximately 430 μm. In addition, the mean thickness of the recognition layer was 15 μm, which increased by approximately 5 nm after the deposition of F-MWCNTs on the surface.

Measurement Using EIS
Before the DNA detection experiment using EIS measurement, a sample solution was prepared by including one of the DNA targets (T1, T2, and T3) with a specific concentration of 1 nM DNA probe (P) and 4 mM [Fe(CN) 6 ] 3−/4− in 1mL of PBS solution. The sample solvent was maintained at 25 • C for 7 min because the UV absorption of the sample solution was saturated after 7 min ( Figure S2), which indicated that the DNA reaction in the solution had stabilized. After starting the EIS measurements, 9 min and 0.26 V were required to obtain the stabilized value of the charge transfer resistance (R ct ) ( Figure S3). Thus, a preparation time of 16 min was needed for EIS-based DNA detection in our experiments. Our three-electrode system, which consisted of an Ag/AgCl reference electrode, platinum counter electrode, and 5 mm × 5 mm F-MWCNT/MWCNT/PDMS working electrode, was placed in sample solutions, as shown in Figure S4. The EIS measurements were performed in the frequency range of 0.1 Hz to 10000 Hz, with an AC (alternating current) amplitude of 5mV and a DC (direct current) bias of 0.26 V for DNA detection.

SEM Measurments of the Composite Electrode
The cross-section and surface morphology of the F-MWCNT-modified and pristine MWCNT/PDMS electrodes were characterized using SEM. For the pristine MWCNT/PDMS electrode, most MWCNTs were buried in PDMS and only a few tubular structures of MWCNTs were visible near the electrode surface, as shown in Figure 3a. After 60 dipcoating cycles with the F-MWCNT solution, many F-MWCNTs were observed on the surface of the F-MWCNT-modified MWCNT/PDMS electrode (see Figure 3b). These may act as active centers for DNA interactions. Figure 3c shows the cross-section of the F-MWCNT/MWCNT composite electrode. The results indicate that the mean thickness of the PDMS layers was approximately 430 µm. In addition, the mean thickness of the recognition layer was 15 µm, which increased by approximately 5 nm after the deposition of F-MWCNTs on the surface.    Figure 4a. The circuit model is widely used for describing processes at the electrochemical interfaces [35,36], consisting of the active electrolyte resistance (R Ω ), double-layer capacitance (Cd), charge transfer resistance (Rct), and Warburg impedance (Zw). Rct and Zw describe the electron transfer and the mass transport of the electroactive species near the solution-electrode interface, respectively [37]. The Rct of the film decreased when increasing the number of dip coating cycles, as shown in Figure 4b; as compared to Rct for no dip coating (5500 Ω), that for 60 dip coatings (200 Ω) was about 27.5 times smaller. When the number of dip coatings was over 60, we found that the F-MWCNT/MWCNT layers were easily peeled off from the electrode during the electrochemical analysis. Thus, the number of dip coating cycles in the standard fabrication process was set to 60.

Characterization of the F-MWCNT-Modified Electrodes
We think these results were obtained for two reasons. First, the charge transfer resistance might be associated with the number of MWCNTs on the electrode surface. In the MWCNT electrode, only a small number of MWCNTs and active area were exposed because some of the MWCNTs were buried in the PDMS. However, after depositing F-MWCNTs, the number of MWCNTs and the active area of the electrode were increased. Second, we think the energy barrier of the redox species reaching the electrode was lowered due to Coulomb or steric interactions [38]. We speculate that the energy barrier of [Fe(CN)6] 3−/4− approaching the electrode surface is lowered by hydrogen bonding affinity to carboxylic and hydroxyl groups in F-MWCNTs.    Figure 4a. The circuit model is widely used for describing processes at the electrochemical interfaces [35,36], consisting of the active electrolyte resistance (R Ω ), double-layer capacitance (C d ), charge transfer resistance (R ct ), and Warburg impedance (Z w ). R ct and Z w describe the electron transfer and the mass transport of the electroactive species near the solution-electrode interface, respectively [37]. The R ct of the film decreased when increasing the number of dip coating cycles, as shown in Figure 4b; as compared to R ct for no dip coating (5500 Ω), that for 60 dip coatings (200 Ω) was about 27.5 times smaller. When the number of dip coatings was over 60, we found that the F-MWCNT/MWCNT layers were easily peeled off from the electrode during the electrochemical analysis. Thus, the number of dip coating cycles in the standard fabrication process was set to 60.   Figure 4a. The circuit model is widely used for describing processes at the electrochemical interfaces [35,36], consisting of the active electrolyte resistance (R Ω ), double-layer capacitance (Cd), charge transfer resistance (Rct), and Warburg impedance (Zw). Rct and Zw describe the electron transfer and the mass transport of the electroactive species near the solution-electrode interface, respectively [37]. The Rct of the film decreased when increasing the number of dip coating cycles, as shown in Figure 4b; as compared to Rct for no dip coating (5500 Ω), that for 60 dip coatings (200 Ω) was about 27.5 times smaller. When the number of dip coatings was over 60, we found that the F-MWCNT/MWCNT layers were easily peeled off from the electrode during the electrochemical analysis. Thus, the number of dip coating cycles in the standard fabrication process was set to 60.

Characterization of the F-MWCNT-Modified Electrodes
We think these results were obtained for two reasons. First, the charge transfer resistance might be associated with the number of MWCNTs on the electrode surface. In the MWCNT electrode, only a small number of MWCNTs and active area were exposed because some of the MWCNTs were buried in the PDMS. However, after depositing F-MWCNTs, the number of MWCNTs and the active area of the electrode were increased. Second, we think the energy barrier of the redox species reaching the electrode was lowered due to Coulomb or steric interactions [38]. We speculate that the energy barrier of [Fe(CN)6] 3−/4− approaching the electrode surface is lowered by hydrogen bonding affinity to carboxylic and hydroxyl groups in F-MWCNTs.  We think these results were obtained for two reasons. First, the charge transfer resistance might be associated with the number of MWCNTs on the electrode surface. In the MWCNT electrode, only a small number of MWCNTs and active area were exposed because some of the MWCNTs were buried in the PDMS. However, after depositing F-MWCNTs, the number of MWCNTs and the active area of the electrode were increased. Second, we think the energy barrier of the redox species reaching the electrode was lowered due to Coulomb or steric interactions [38]. We speculate that the energy barrier of [Fe(CN) 6 ] 3−/4− approaching the electrode surface is lowered by hydrogen bonding affinity to carboxylic and hydroxyl groups in F-MWCNTs.
Cyclic voltammetry (CV) is one voltammetry technique for measuring the current response of a redox active solution to a linearly cycled potential sweep. To investigate whether the reactions near the electrodes are diffusion-controlled, we performed CV measurements with  Figure 5a). Figure 5b indicates that the anodic and cathodic peak currents (I pa and I pc , respectively) increased linearly with the square root of the scan rate (ν). The linear equations can be described as; I pa (A) = 8.598 × 10 −4 × ν 1 2 + 1.311 × 10 −4 with an R 2 = 0.9953 and I pc (A) = −0.001 × ν 1 2 − 1.006 × 10 −4 with an R 2 = 0.9989. These results indicate that the oxidation-reduction reactions were diffusion controlled, indicating that the composite electrode is appropriate for quantitative electrochemical analysis [39,40].  Figure 5a). Figure 5b indicates that the anodic and cathodic peak currents (Ipa and Ipc, respectively) increased linearly with the square root of the scan rate (ν). The linear equations can be described as; ( ) = (8.598 × 10 ) × + 1.311 × 10 with an = 0.9953 and ( ) = −0.001 × − 1.006 × 10 with an = 0.9989. These results indicate that the oxidation-reduction reactions were diffusion controlled, indicating that the composite electrode is appropriate for quantitative electrochemical analysis [39,40].

DNA Detection Using F-MWCNT-Modified Electrodes
As discussed above, a F-MWCNT/MWCNT/PDMS electrode could be used as a DNA sensor. The schematic of the DNA detection mechanism is described in Figure 6. It is well known that single-stranded DNA (s-DNA) is adsorbed on the surfaces of F-MWCNT or MWCNT through π-π interactions, whereas double-stranded DNA (d-DNA) is not [4,[41][42][43][44][45]. For target T3 and probe DNA, the sequences of the probe and target are all mismatched, so that the single-stranded target and probe DNA might be adsorbed onto the recognition layer (F-MWCNT/MWCNT) of the electrode through π-π interactions [4,43,44]. The adsorbed s-DNA can act as the barrier of charge transfer between the oxidation-reduction couple ([Fe(CN)6] 3−/4− ) and the MWCNTs, increasing Rct, whereas for target T1 and probe DNA, d-DNA can form through hybridization between T1 and probe DNA because their sequences match completely. The vast majority of d-DNA remains far from the electrode rather than being adsorbed onto the recognition layer [45]. Thus, because d-DNA does not behave as a charge transfer barrier, Rct is expected to be lower than that for the former case with s-DNA adsorbed on the electrode. In the case of T2, the probe and target DNA have a one-base mismatch. Because the level of sequence mismatch is between those for T1 and T3, resistance might also be between those of T1 and T3.

DNA Detection Using F-MWCNT-Modified Electrodes
As discussed above, a F-MWCNT/MWCNT/PDMS electrode could be used as a DNA sensor. The schematic of the DNA detection mechanism is described in Figure 6. It is well known that single-stranded DNA (s-DNA) is adsorbed on the surfaces of F-MWCNT or MWCNT through π-π interactions, whereas double-stranded DNA (d-DNA) is not [4,[41][42][43][44][45]. For target T3 and probe DNA, the sequences of the probe and target are all mismatched, so that the single-stranded target and probe DNA might be adsorbed onto the recognition layer (F-MWCNT/MWCNT) of the electrode through π-π interactions [4,43,44]. The adsorbed s-DNA can act as the barrier of charge transfer between the oxidation-reduction couple ([Fe(CN) 6 ] 3−/4− ) and the MWCNTs, increasing R ct , whereas for target T1 and probe DNA, d-DNA can form through hybridization between T1 and probe DNA because their sequences match completely. The vast majority of d-DNA remains far from the electrode rather than being adsorbed onto the recognition layer [45]. Thus, because d-DNA does not behave as a charge transfer barrier, R ct is expected to be lower than that for the former case with s-DNA adsorbed on the electrode. In the case of T2, the probe and target DNA have a one-base mismatch. Because the level of sequence mismatch is between those for T1 and T3, resistance might also be between those of T1 and T3.
We performed EIS measurements for the sample solutions described in the experimental section, which include probe DNA (P) and one target DNA (T1, T2, or T3), with the target as shown in Figure 7a. R ct was calculated from the fitting of the Nyquist plots in Figure 7a, the values of which were about 940, 1190, and 2000 Ω for the T1, T2, and T3 targets, respectively, as shown in Figure 7b. The highest R ct values were obtained for probe and T3, whereas the lowest were for probe and T1, as expected due to the mechanism discussed above. We defined the DNA detection sensitivity, г, by the equation г= ∆R/R • , where R • is the R ct when hybridized with T3 and ∆R is the difference between the R ct for the present target and R • . According to this equation, the sensitivities for perfectly matched and one-base mismatched targets (T1 and T2) were 52.5% and 40.4%, respectively; the sensitivity differs by 12.1% even for a one-base mismatch, indicating the good sequence selectivity of our system. We performed EIS measurements for the sample solutions described in the experimental section, which include probe DNA (P) and one target DNA (T1, T2, or T3), with the target as shown in Figure 7a. Rct was calculated from the fitting of the Nyquist plots in Figure 7a, the values of which were about 940, 1190, and 2000 Ω for the T1, T2, and T3 targets, respectively, as shown in Figure 7b. The highest Rct values were obtained for probe and T3, whereas the lowest were for probe and T1, as expected due to the mechanism discussed above. We defined the DNA detection sensitivity, г, by the equation г = ∆R/R°, where R° is the Rct when hybridized with T3 and ∆R is the difference between the Rct for the present target and R°. According to this equation, the sensitivities for perfectly matched and one-base mismatched targets (T1 and T2) were 52.5% and 40.4%, respectively; the sensitivity differs by 12.1% even for a one-base mismatch, indicating the good sequence selectivity of our system.

Detection Limit of the DNA Sensor
As shown in Figure 8b, Rct in the EIS analysis with target T1 was linearly dependent on the logarithm of the T1 concentration in the range of 1-1000 pM in the sample solutions. In Figure 8a, a decrease in the Rct is clearly observed alongside an increase in the concentration of T1. The reason might be that as the concentration of T1 increases, the amount of DNA attached to the active site increases, which increases the current and reduces Rct. The linear regression equation was determined to be = −2.17 × 10 • + 1.58 × 10 with an R 2 = 0.992, where C1 is the concentration of T1 ranging from 1 to 1000 pM, as  We performed EIS measurements for the sample solutions described in the exper mental section, which include probe DNA (P) and one target DNA (T1, T2, or T3), wit the target as shown in Figure 7a. Rct was calculated from the fitting of the Nyquist plots i Figure 7a, the values of which were about 940, 1190, and 2000 Ω for the T1, T2, and T targets, respectively, as shown in Figure 7b. The highest Rct values were obtained for prob and T3, whereas the lowest were for probe and T1, as expected due to the mechanism discussed above. We defined the DNA detection sensitivity, г, by the equation г = ∆R/R where R° is the Rct when hybridized with T3 and ∆R is the difference between the Rct fo the present target and R°. According to this equation, the sensitivities for perfectl matched and one-base mismatched targets (T1 and T2) were 52.5% and 40.4%, respec tively; the sensitivity differs by 12.1% even for a one-base mismatch, indicating the goo sequence selectivity of our system.

Detection Limit of the DNA Sensor
As shown in Figure 8b, Rct in the EIS analysis with target T1 was linearly dependen on the logarithm of the T1 concentration in the range of 1-1000 pM in the sample solution In Figure 8a, a decrease in the Rct is clearly observed alongside an increase in the concen tration of T1. The reason might be that as the concentration of T1 increases, the amount o DNA attached to the active site increases, which increases the current and reduces Rct. Th linear regression equation was determined to be = −2.17 × 10 • + 1.58 10 with an R 2 = 0.992, where C1 is the concentration of T1 ranging from 1 to 1000 pM, a

Detection Limit of the DNA Sensor
As shown in Figure 8b, R ct in the EIS analysis with target T1 was linearly dependent on the logarithm of the T1 concentration in the range of 1-1000 pM in the sample solutions. In Figure 8a, a decrease in the R ct is clearly observed alongside an increase in the concentration of T1. The reason might be that as the concentration of T1 increases, the amount of DNA attached to the active site increases, which increases the current and reduces R ct . The linear regression equation was determined to be R ct = −2.17 × 10 2 ·logC 1 + 1.58 × 10 3 with an R 2 = 0.992, where C 1 is the concentration of T1 ranging from 1 to 1000 pM, as shown in Figure 8b. The LOD was extrapolated to be 19.9 fM using a signal-to-noise ratio of 3:1. Our LOD decreased by 1250 times compared to that obtained in a previous study, wherein a MWCNT electrode was used as a working electrode for DNA detection in EIS [27]. The 27.5 times reduction in R ct by F-MWCNT modification to the MWCNT/PDMS electrode surface might be an important reason for the drastic decrease in LOD. The change in the adsorption/desorption characteristics of the s-DNA and d-DNA on the electrode surface by F-MWCNT modification may also have contributed to lowering the LOD. Furthermore, F-MWCNTs can interact with DNA through hydrogen bonding and electrostatic interactions, as well as π-π stacking interactions, and these interactions are more complicated than the interactions in MWCNTs. Although a comprehensive study of DNA adsorption characteristics on F-MWCNTs was beyond the scope of this study, it is known that graphene oxide, which has a chemical structure similar to that of F-MWCNT, absorbs s-DNA well, but does not efficiently absorb d-DNA [46]. Graphene-oxide-based DNA sensing can be performed in less than 30 min, whereas it takes several hours to achieve a similar detection rate with carbon nanotubes, which do not have functional groups [47]. Based on these results, we speculate that F-MWCNTs are more efficient for discriminating s-DNA and d-DNA than MWCNTs owing to their additional interactions, such as hydrogen bonding.
trode surface by F-MWCNT modification may also have contributed to lowering th Furthermore, F-MWCNTs can interact with DNA through hydrogen bonding and e static interactions, as well as π-π stacking interactions, and these interactions ar complicated than the interactions in MWCNTs. Although a comprehensive study o adsorption characteristics on F-MWCNTs was beyond the scope of this study, it is that graphene oxide, which has a chemical structure similar to that of F-MWCNT, a s-DNA well, but does not efficiently absorb d-DNA [46]. Graphene-oxide-based sensing can be performed in less than 30 min, whereas it takes several hours to ac similar detection rate with carbon nanotubes, which do not have functional group Based on these results, we speculate that F-MWCNTs are more efficient for discrim s-DNA and d-DNA than MWCNTs owing to their additional interactions, such as gen bonding.

Stability of the Modified Electrode
The stability of the fabricated DNA sensor over 7 days was investigated by m ing the average Rct daily (Figure 9 and Figure S5). For the experiment, 2 mL of solvent containing 1 nM P and T1 DNA was poured into the modified electrode. first stored at room temperature for 1 week and then examined via its EIS response ( 9a) after hybridization in 4 mM [Fe(CN)6] 3−/4− solution with 1 × PBS buffer. Between experiment, the electrode was cleaned by ultrasonication in absolute ethanol for and DI water for 3 min. Figure 9b shows the charge transfer resistance (Rct) and the r standard deviation of Rct is 2.31%, indicating the good stability of the fabricated ele The CV curves of the fabricated electrode are shown in Figure S5 (one cycle and cles). Even after 100 cycles, the results show that there was no significant chang indicates that our physical surface modification using F-MWCNTs is very stable the regeneration process. It is important to note that our electrode method is not a cal but a physical process; therefore, it is easy to control, thereby reducing LOD, applicable to other carbon sensors.

Stability of the Modified Electrode
The stability of the fabricated DNA sensor over 7 days was investigated by measuring the average R ct daily (Figure 9 and Figure S5). For the experiment, 2 mL of sample solvent containing 1 nM P and T1 DNA was poured into the modified electrode. It was first stored at room temperature for 1 week and then examined via its EIS response (Figure 9a) after hybridization in 4 mM [Fe(CN) 6 ] 3−/4− solution with 1× PBS buffer. Between every experiment, the electrode was cleaned by ultrasonication in absolute ethanol for 3 min and DI water for 3 min. Figure 9b shows the charge transfer resistance (R ct ) and the relative standard deviation of R ct is 2.31%, indicating the good stability of the fabricated electrode. The CV curves of the fabricated electrode are shown in Figure S5 (one cycle and 100 cycles). Even after 100 cycles, the results show that there was no significant change. This indicates that our physical surface modification using F-MWCNTs is very stable during the regeneration process. It is important to note that our electrode method is not a chemical but a physical process; therefore, it is easy to control, thereby reducing LOD, and is applicable to other carbon sensors.

Summary
We investigated the physical surface modification of multi-walled carbon-nanotube/polymer electrodes treated using F-MWCNTs and a dip-coating process. The F-MWCNTs could be successfully deposited using solution-based physical processes, such as dip coating, because of their high dispersibility in water. The electrode surface can be used as a recognition layer, enabling the simple fabrication of DNA sensors. The vital

Summary
We investigated the physical surface modification of multi-walled carbon-nanotube/polymer electrodes treated using F-MWCNTs and a dip-coating process. The F-MWCNTs could be successfully deposited using solution-based physical processes, such as dip coating, because of their high dispersibility in water. The electrode surface can be used as a recognition layer, enabling the simple fabrication of DNA sensors. The vital properties of our structure are that by modifying it with F-MWCNTs, the active area of the electrode was increased, which resulted in a low charge transfer resistance. Compared to EIS-based DNA sensors that use electrodes chemically modified with nanomaterials, our electrode has a simpler and more controllable fabrication process with a very low LOD of 19.9 fM. The physically modified electrodes were stable during the regeneration process. Our results indicate that the physical surface modification of electrodes is a promising alternative approach to chemical surface.  Table S1: Fitted values of the equivalent circuit parameters from dark Nyquist plots of devices without and with 5% Cs.