Dual‐Hydrogen Bond Donor‐Functionalized Carbon Nanotube Fibers: Enhancing Anion‐Sensing Performance Through Functionalization Approaches

Abstract

In this study, chemiresistive anion sensors are developed using carbon nanotube fibers (CNTFs) functionalized with squaramide‐based dual‐hydrogen bond donors (SQ1 and SQ2) and systematically compared the sensing properties attained by two different functionalization methods. Model structures of the selectors are synthesized based on a squaramide motif incorporating an electron‐withdrawing group. Anion‐binding studies of SQ1 and SQ2 are conducted using UV–vis titrations to elucidate the anion‐binding properties of the selectors. These studies revealed that the chemical interaction with acetate (AcO⁻) induced the deprotonation of both SQ1 and SQ2. Selectors are functionalized onto the CNTFs using either covalent or non‐covalent functionalization. For covalent functionalization, SQ1 is chemically formed on the surface of the CNTFs, whereas SQ2 is non‐covalently functionalized to the surface of the CNTFs assisted by poly(4‐vinylpyridine). The results showed that non‐covalently functionalized CNTFs exhibited a 3.6‐fold higher sensor response toward 33.33 mm AcO⁻ than covalently functionalized CNTFs. The selector library is expanded using diverse selectors, such as TU‐ and CA‐based selectors, which are non‐covalently functionalized on CNTFs and presented selective AcO⁻‐sensing properties. To demonstrate on‐site and real‐time anion detection, anion sensors are integrated into a sensor module that transferred the sensor resistance to a smartphone via wireless communication.

The effective electrical transduction property of chemiresisitve anion sensors is elucidated by comparing the functionalization techniques of selectors onto the carbon nanotube fiber (CNTF). Various dual‐hydrogen bond donor‐based selectors are developed and functionalized either covalently or non‐covalently to evaluate acetate sensing response. Much improved acetate sensing performance is achieved by non‐covalent selector functionalization with the demonstration of real‐time wireless detection of acetate using a mobile device.

1. Introduction

The significance of anion sensing is on the rise in diverse fields, including healthcare, environmental monitoring, and biotechnology.^{[ 1} , ² , ^{3 ]} For example, acetate (AcO⁻) is crucial in microbial metabolism, significantly affecting bacterial cell growth.^{[ 4} , ^{5 ]} In biotechnological and industrial microbiology, monitoring AcO⁻ concentrations is essential for optimizing fermentation processes and maintaining the health and productivity of microbial cultures.^{[ 6 ]} AcO⁻ can inhibit bacterial growth even at concentrations as low as 8 mm and reduce the yield and efficiency of bioproduction.^{[ 7 ]} Therefore, precise and reliable monitoring of AcO⁻ concentrations can provide valuable information about microbial behavior, environmental health, and metabolic conditions in clinical settings.

Anion recognition techniques employing proton nuclear magnetic resonance (¹H NMR) and UV–vis spectroscopic methods can provide precise and highly accurate results in molecular sensing and anion binding studies.^{[ 8} , ^{9 ]} However, the application of spectroscopic methods for the portable and on‐site detection of anions in environmental and biological samples is challenging because of the bulkiness of the equipment and the requirement of an expert for operation. Therefore, sensor technology is gaining considerable attention for its miniaturization and ease of use. To develop a portable and real‐time anion sensor, selector functionalization and efficient transduction upon chemical interaction between the selector and analyte are crucial.

To develop selectors, the structural design and characterization are important for elucidating the anion selectivity and binding mechanisms.^{[ 10} , ^{11 ]} Among selectors, dual‐hydrogen bond donor motifs exhibit effective binding properties toward Y‐shaped anions such as AcO⁻ owing to the geometric fitting with two N─H bonds providing a directional hydrogen bond.^{[ 12} , ^{13 ]} Various dual‐hydrogen bond donors such as thiourea, squaramide, and croconamide have been investigated to understand anion binding properties toward anions.^{[ 14} , ^{15 ]} We previously developed asymmetric N,N’‐substituted dual‐hydrogen bond donor motifs by incorporating an electron‐withdrawing group and a cationic moiety.^{[ 16} , ¹⁷ , ^{18 ]} The electron‐withdrawing group could enhance the hydrogen bonding capability of dual‐hydrogen bond donors by increasing the acidity of N─H protons.^{[ 19} , ²⁰ , ²¹ , ^{22 ]} In addition, cationic moieties such as imidazolium,^{[ 23} , ^{24 ]} triazolium,^{[ 25 ]} and pyridinium^{[ 26 ]} could enhance the binding affinity toward anions through electrostatic interactions. The decoration of a cationic moiety along with hydrogen bond donor motifs can further improve the anion binding affinity through additional electrostatic interactions.^{[ 27 ]} Indeed, a thiourea‐based dual‐hydrogen bond donor with a terminal pyridinium moiety exhibited a 2.8‐fold higher AcO⁻‐binding affinity than that with a hydroxyl terminal group.^{[ 17 ]}

In terms of transduction techniques, portable analysis methods such as optical,^{[ 28} , ^{29 ]} electrochemical,^{[ 30 ]} and chemiresistive^{[ 16} , ¹⁷ , ^{18 ]} techniques have been employed to interpret the chemical interactions between selectors and anions. Among these techniques, the chemiresistive method is gaining a great deal of attention owing to its simple operating principle and real‐time detection capability for point‐of‐care testing.^{[ 31 ]} To fabricate chemiresistive‐type sensors, an electrically conductive material with high functionality capable of incorporating selectors is essential. Among the various charge transduction layers, carbon nanotubes (CNTs) have been widely employed owing to their exceptional mechanical and electrical properties.^{[ 32 ]} Remarkable achievements and innovative applications have been made utilizing chemiresistive sensors by transducing chemical interactions using electrically conductive CNTs functionalized with selectors.^{[ 33} , ^{34 ]}

Despite the exceptional electrical properties of CNTs, their applications in the sensing field are frequently impeded by their insolubility in most organic solvents and the challenges associated with handling them.^{[ 35 ]}In addition, complicated chemical treatments on a substrate are necessary to obtain stable adhesion of CNT‐based composites in solutions which is particularly beneficial for sensing in a solution.^{[ 36} , ^{37 ]} To address these limitations, carbon nanotube fibers (CNTFs) have been fabricated by assembling multiple CNTs into a 1D continuous fiber structure. The CNTFs are synthesized by wet‐spinning a CNT dispersion and exploiting the lyotropic nematic liquid crystal properties of CNTs.^{[ 38 ]} This wet‐spinning technique results in the axial alignment and condensation of CNTs, leading to the high mechanical strength and electrical conductivity of CNTFs.^{[ 39 ]} The unique physical properties of CNTFs have enabled the successful development of wearable electronics such as energy storage and conversion devices.^{[ 40} , ⁴¹ , ^{42 ]}

To utilize CNTFs in chemiresistive anion sensors, the functionalization of selectors on the CNTFs is essential. In terms of selector functionalization on the surface of CNTs, both covalent and non‐covalent bonding techniques have generally been used to generate sensitivity toward specific analytes.^{[ 43} , ^{44 ]} For covalent bonding, selectors create strong bonds on the surface of CNTs by penetrating the C═C bonds of the CNT sidewalls, which can efficiently transduce electrical signals through the formation of selector‐anion interactions. However, rehybridization of the sp² carbon atoms to the sp³ character disrupts the intrinsic electrical properties of CNTs.^{[ 45 ]} In contrast, the non‐covalent functionalization of selectors causes minimal perturbation of the sp² character of carbon atoms, thereby preserving π‐conjugation and the intrinsic electrical properties.^{[ 46 ]} However, indirect chemical bonding between CNTs and selectors can lower the transduction efficiency upon the interaction with anions. Therefore, the transduction efficiency of covalently and non‐covalently functionalized selectors should be comparatively analyzed to improve the response of chemiresistive anion sensors.

In this study, we developed chemiresistive anion sensors that utilize CNTF functionalized with selectors. Various selectors composed of N,N’‐substituted dual‐hydrogen bond donors were synthesized and the anion binding affinity was investigated to understand the fundamental binding mechanism. Real‐time changes in the resistance of the CNTF were generated by transducing the chemical interactions between the selectors and anions. The anion‐sensing responses depending on the functionalization method, i.e., covalent or non‐covalent bonding, were evaluated to determine the efficiency of electrical transduction. Finally, the anion sensors were integrated with a wireless sensing module to demonstrate the real‐time and on‐site detection of anions. The structural properties of CNTFs combined with selectors have their unique potential and versatility for integration into various sensor configurations such as fiber‐based sensing platforms and e‐textiles.

2. Results and Discussion

CNTFs were functionalized with selectors comprising N,N’‐substituted dual‐hydrogen bond donors to develop chemiresistive anion sensors and real‐time resistance changes were measured upon injection of anions (Figure 1a). In particular, we characterized the anion‐sensing responses depending on the technique for functionalizing the selector on the CNTFs. Specifically, selectors can be covalently functionalized by forming direct chemical bonds with CNTFs, or they can be non‐covalently functionalized by attaching CNTFs to polymers containing selectors. To this end, asymmetric anion selectors were synthesized through N,N’‐substitution reactions of dual‐hydrogen bond donors, wherein an alkyl bromide chain was inserted into the dual‐hydrogen bond donor. Subsequently, a cationic pyridinium group was introduced via a quaternization reaction, particularly for the non‐covalent functionalization, to enhance the anion‐binding affinity. An electron‐withdrawing 3,5‐bis(trifluoromethyl)phenyl group was also inserted into all the selectors as a functional group to enhance hydrogen bond donation. Thus, various asymmetric selectors consisting of N,N’‐substituted dual‐hydrogen bond donors such as squaramide (SQ1 and SQ2), thiourea (TU), and croconamide (CA) were prepared to investigate the anion‐binding affinity and demonstrate the applicability of the selectors in chemiresistive anion sensors (Figure 1b and Schemes S1–S3, Supporting Information).

Figure 1.

a) Schematic illustration of the chemiresistive anion sensor comprising a CNTF functionalized with a selector (e.g., SQ2‐P4VP‐CNTF). b) Model structures of the anion selectors SQ1, SQ2, TU, and CA. c) Synthesis of CNTFs using a wet‐spinning process and subsequent covalent or non‐covalent functionalization of squaramide‐based selectors, i.e., SQ1‐CNTF and SQ2‐P4VP‐CNTF, respectively.

For use as electrical signal transducers, CNTFs were synthesized through a wet‐spinning process that facilitated the coagulation of single‐walled carbon nanotubes (SWCNTs) in a dispersion (Figure 1c). The surface chemistry of the CNTFs can be modified by decorating functional groups on the surface of the SWCNTs. For instance, CNTFs decorated with alkylamines (NH₂‐CNTF) were prepared by the amination of SWCNTs, followed by wet‐spinning the dispersion. To compare the anion‐sensing responses with respect to the selector functionalization method, squaramide‐based dual‐hydrogen bond donors (SQ1 and SQ2) were subsequently functionalized onto the surfaces of the CNTFs. For covalent functionalization, the NH₂‐CNTFs were immersed in a solution containing 1, which resulted in the formation of SQ1 on the surface of the CNTFs (SQ1‐CNTF) (Scheme S1, Supporting Information). For non‐covalent functionalization, polymers containing SQ2 were first prepared and physically attached to the CNTFs. To this end, poly(4‐vinylpyridine) (P4VP) was modified by decorating SQ2 (SQ2‐P4VP) through a quaternization reaction (Scheme S4, Supporting Information). Both SQ2‐P4VP and CNTFs were immersed in the solution, leading to the physical attachment of SQ2‐P4VP to the surface of the CNTFs (SQ2‐P4VP‐CNTF).

To compare the anion‐sensing responses with respect to the selector functionalization method, the anion‐binding affinities of the model structures were first investigated using SQ1 and SQ2 (Figure 2 ). The anion‐binding affinities of SQ1 and SQ2 were evaluated by UV–vis titration toward AcO⁻, chloride (Cl⁻), bromide (Br⁻), and nitrate (NO₃ ⁻) in dimethyl sulfoxide (DMSO). In the case of SQ1, the absorbance bands decreased at 280, 320, and 350 nm upon the addition of AcO⁻, whereas the absorbance intensity dramatically increased at 386 nm (Figure 2a). In addition, two isosbestic points were observed at 264 and 357 nm. These changes in absorbance upon adding AcO⁻ were attributed to the deprotonation of N─H protons in SQ1 resulting from the internal charge transfer.^{[ 47} , ^{48 ]} The deprotonation of N─H protons of dual‐hydrogen bond donors upon interacting with basic AcO⁻ occurs according to two equilibrium states.^{[ 49 ]} The first equilibrium state is associated with the formation of a hydrogen‐bond complex between SQ1 and AcO⁻, i.e., RH₂···AcO⁻, as expressed in Equation (1):

$$\begin{array}{ccc}\hfill \mathbf{R}{\mathrm{H}}_{\mathbf{2}}& +& {\mathrm{AcO}}^{-}\leftrightarrow \mathbf{R}{\mathrm{H}}_2···{\mathrm{AcO}}^{-},{\mathrm{K}}_1\hfill \\ & & =[\mathbf{R}{\mathrm{H}}_2···{\mathrm{AcO}}^{-}\left]/\right[\mathbf{R}{\mathrm{H}}_2]\left[{\mathrm{AcO}}^{-}\right]\hfill \end{array}$$ (1)

Figure 2.

a) UV–vis titration of SQ1 ([SQ1] = 4.1 × 10⁻² mm) upon the sequential additions of 0–1.04 equivalents of AcO⁻ in DMSO. b) Mole fractions of SQ1 and SQ1 complexes as functions of the equivalents of AcO⁻. c) ΔAbs of SQ1 toward AcO⁻, Cl⁻, Br⁻, and NO₃ ⁻. d) UV–vis titration of SQ2 ([SQ2] = 3.8 × 10⁻² mm) upon the additions of 0–1.00 equivalents of AcO⁻ in DMSO. e) Mole fractions of SQ2 and SQ2 complexes as functions of the equivalents of AcO⁻. f) ΔAbs of SQ2 toward AcO⁻, Cl⁻, Br⁻, and NO₃ ⁻.

The binding stoichiometry was confirmed by a Job plot analysis based on absolute absorbance changes (ΔAbs) at 386 nm (Figure S1a, Supporting Information). The results revealed that SQ1 exhibited a 1:1 binding stoichiometry with AcO⁻ up to 1 equivalent, which implies the hydrogen bond interaction between SQ1 and AcO⁻ with a maximum mole fraction [SQ1···AcO⁻] of 63% (Figure 2b). Further addition of AcO⁻ decreased the mole fraction [SQ1···AcO⁻] as a result of the deprotonation of SQ1 and the formation of a hydrogen‐bond self‐complex, i.e., H(AcO)₂ ⁻,^{[ 50 ]} which corresponds to the second equilibrium state as presented in Equation (2):

$$\begin{array}{ccc}\hfill \mathbf{R}{\mathrm{H}}_2& & ···{\mathrm{AcO}}^{-}+{\mathrm{AcO}}^{-}\leftrightarrow \mathbf{R}{\mathrm{H}}^{-}+\mathrm{H}{{\left(\mathrm{AcO}\right)}_2}^{-},{\mathrm{K}}_2\hfill \\ & & =[\mathbf{R}{\mathrm{H}}^{-}\left]\right[\mathrm{H}{{\left(\mathrm{AcO}\right)}_2}^{-}\left]/\right[\mathbf{R}{\mathrm{H}}_2···{\mathrm{AcO}}^{-}\left]\right[{\mathrm{AcO}}^{-}]\hfill \end{array}$$ (2)

Additional UV–vis titrations toward Cl⁻, Br⁻, and NO₃ ⁻ exhibited minor changes in absorbance, indicating hydrogen‐bonding interactions with SQ1 without deprotonation (Figures S2a–c, Supporting Information). The ΔAbs at 386 nm with respect to the equivalents of anions confirmed the characteristic binding of SQ1 with AcO⁻, which caused hydrogen‐bond interactions and subsequent deprotonation, thus presenting AcO⁻ selectivity (Figure 2c). The association constants (K₁) of SQ1 toward anions were calculated based on the 1:1 stoichiometry, wherein SQ1 exhibited a K₁ value for AcO⁻ (1.95 × 10⁶ m ⁻¹) that was three orders of magnitude higher than its K₁ values for Cl⁻, Br⁻, and NO₃ ⁻ (Table S1, Supporting Information).

UV–vis titration was also conducted for SQ2 toward AcO⁻ in DMSO (Figure 2d). Similarly, an additional absorbance band emerged at 386 nm, whereas the absorbance bands at 280, 320, and 350 nm diminished upon adding AcO⁻. Additionally, isosbestic points were identified at 264 and 354 nm. These absorbance changes were attributed to the deprotonation of SQ2 identical to SQ1. Specifically, the hydrogen‐bond complex, i.e., SQ2···AcO⁻, was formed with 1:1 binding stoichiometry in the first Equation 1 with a maximum mole fraction [SQ2···AcO⁻] of 48% up to the addition of one equivalent (Figures 2e; Figure S1b, Supporting Information). Subsequent additions of AcO⁻ decreased the mole fraction [SQ2···AcO⁻], which resulted from the deprotonation of SQ2 and the formation of the hydrogen‐bond self‐complexes, as indicated by the Equation 2.^{[ 50 ]} UV–vis titration of SQ2 toward Cl⁻, Br⁻, and NO₃ ⁻ confirmed minor absorbance changes (Figures S2d–f, Supporting Information). The ΔAbs with respect to the equivalents of anions at 386 nm confirmed the characteristic binding interactions between SQ2 with AcO⁻, whereas minor changes in the absorbance of SQ2 toward Cl⁻, Br⁻, and NO₃ ⁻ were attributed to hydrogen‐bond interactions (Figure 2f). SQ2 exhibited a K₁ value of 2.16 × 10⁶ m ⁻¹ for AcO⁻, which was substantially higher than those for Cl⁻, Br⁻, and NO₃ ⁻ (Table S1, Supporting Information).

Based on the titration results of SQ1 and SQ2, the influence of the cationic moiety (i.e., pyridinium) on the binding affinity toward AcO⁻ was investigated by comparing the association constants K₁ and K₂. The results revealed that the values of both K₁ (2.16 × 10⁶ m ⁻¹) and K₂ (1.04 × 10⁶) of SQ2 were higher than K₁ (1.95 × 10⁶ m ⁻¹) and K₂ (1.61 × 10⁵) of SQ1. Thus, a 6.4‐fold increase in the K₂ value was attained with SQ2 with respect to that of SQ1, indicating that the cationic pyridinium moiety in SQ2 promotes stronger binding affinity toward AcO⁻, inducing deprotonation upon interaction with AcO⁻.

To prepare the electrical transducer, electrically conductive and mechanically robust CNTFs were synthesized by a wet‐spinning process utilizing the nematic liquid‐crystal properties of SWCNTs. These properties were observed using polarized optical microscopy (POM), wherein the dark region in the POM image indicates SWCNTs aligned along the polarizer direction (Figure 3a). The aligned SWCNTs coagulated to form continuous fibers through wet‐spinning, followed by winding onto a bobbin (Figure S3, Supporting Information). To utilize the CNTFs for covalent functionalization, NH₂‐CNTFs were also prepared by dispersing aminated SWCNTs in chlorosulfonic acid (CSA), followed by an identical wet‐spinning process. The surface morphologies of the CNTFs and NH₂‐CNTFs were investigated using scanning electron microscopy (SEM), and both the CNTFs and NH₂‐CNTFs showed a 1D structure and rough surface morphology (Figure 3b,c). For the case of NH₂‐CNTF, high‐resolution X‐ray photoelectron spectroscopy (XPS) was performed for N 1s, and the primary amine‐related peak was observed at 400.8 eV (Figure S4, Supporting Information).^{[ 51} , ^{52 ]} Magnified SEM images revealed that the SWCNT bundles were aligned along the fiber direction (Figures S5a,b, Supporting Information). The directional alignment of multiple CNTF bundles led to anisotropic charge carrier mobility in the axial direction, resulting in high electrical conductivity.^{[ 53 ]}

Figure 3.

a) POM image of the SWCNT dispersion in a CSA solution. SEM images of b) NH₂‐CNTF, c) CNTF, d) SQ1‐CNTF, and e) SQ2‐P4VP‐CNTF. High‐resolution XPS for N 1s spectra of f) SQ1‐CNTF and g) SQ2‐P4VP‐CNTF.

The anion selectors were functionalized onto CNTFs using covalent and non‐covalent methods. Covalent functionalization was achieved by direct chemical bonding through the amine groups on the surface of NH₂ ⁻CNTFs, which formed SQ1‐CNTF. For non‐covalent functionalization, the SQ2 model structure was linked along the polymer to form a polymeric selector. Specifically, P4VP was employed to synthesize the polymer selector through the quaternization of its pyridyl groups. The chemical modification of P4VP with dual‐hydrogen bond donors was confirmed by FT–IR analysis, wherein SQ2‐P4VP exhibited a characteristic FT–IR vibration peak at 1793 cm⁻¹ associated with the ring breathing mode of the squaramide unit (Figures S6a,b, Supporting Information).^{[ 54 ]} Additionally, the absence of vibration of the pyridyl groups at 1594 cm⁻¹ indicated the transformation of pyridyl groups into pyridinium, thereby confirming the chemical modification of P4VP.^{[ 55} , ^{56 ]} Subsequently, the polymeric selector was physically attached to the surface of the CNTFs using a dip‐coating technique.

The microstructure and surface chemical properties of the functionalized CNTFs were investigated using SEM and XPS. In the case of CNTFs with covalent functionalization, minor changes in the surface morphology were observed with respect to that of the pristine CNTFs (Figure 3d). In contrast, for the non‐covalently functionalized CNTFs, the polymer chains containing the SQ2 selector were homogeneously attached to the surfaces of the CNTFs in the form of a gel‐like surface coating (Figure 3e). The functionalized CNTFs were examined by XPS to confirm the surface functionalization (Figure 3f,g; Figure S7, Supporting Information). The high‐resolution N 1s XPS spectrum of SQ1‐CNTF exhibited a sharp single peak at 400.6 eV, corresponding to nitrogen from the amide C─NH─C (Figure 3f). On the other hand, two characteristic N 1s peaks at 400.1 and 402.2 eV were observed for SQ2‐P4VP‐CNTF, which originated from chemical bonds of the amide C─NH─C and pyridinium N, respectively (Figure 3g).^{[ 16} , ^{57 ]} The difference in C─NH─C binding energy between SQ1‐CNTF and SQ2‐P4VP‐CNTF is mainly due to the low degree of functionalization of SQ1 and the difference in molecular structure. Both SQ1‐CNTF and SQ2‐P4VP‐CNTF demonstrated a characteristic F 1s peak at 688.9 eV, indicating fluorine atoms from the ─CF₃ groups (Figure S7b,d, Supporting Information).^{[ 18 ]} These results confirmed the functionalization of squaramide‐based dual‐hydrogen bond donor motifs onto CNTFs through two different functionalization methods.

Raman spectroscopy analysis was performed to investigate the characteristic D/G band intensity ratio of CNTF, NH₂‐CNTF, SQ1‐CNTF, and SQ2‐P4VP‐CNTF (Figure S8, Supporting Information). While the D/G band intensity ratio was 0.018 for the pristine CNTF, noticeably increased ratios were observed for NH₂‐CNTF (0.063) and SQ1‐CNTF (0.041). This result is mainly attributed to the cleavage of C═C bonds and defect formation on the SWCNT sidewalls, which indicates successful covalent functionalization of the CNTFs. In contrast, SQ2‐P4VP‐CNTF exhibited a relatively low D/G band intensity ratio of 0.011, comparable to the pristine CNTF. This suggests that the non‐covalent functionalization method effectively preserved the intrinsic electronic structure of SWCNT sidewalls.

The anion‐sensing properties were characterized by evaluating the resistance changes of the sensors upon injecting anions such as AcO⁻, Cl⁻, Br⁻, and NO₃ ⁻ in the concentration range of 5–33.33 mm in acetonitrile, which aligns with the AcO⁻ concentrations that regulate bacterial growth, such as E. coli.^{[ 58} , ⁵⁹ , ^{60 ]} To generate a stable electrical signal, an anion sensor was fabricated using a functionalized CNTF on a glass substrate with electrical contacts (Figure 4a). A poly(dimethylsiloxane) (PDMS) mold with a cylindrical hole with a diameter of 3 mm was attached to the top of the functional CNTF, exposing the sensing layer to the external environment. Before injecting an anion analyte, 10 µL of pure acetonitrile solvent was injected into the solution chamber and the sensor resistance decreased by ≈0.74 ± 0.23% (n = 3) and gradually saturated within 3–5 min to stabilize a baseline resistance (Figure S9, Supporting Information). Subsequently, 2 µL of an analyte anion solution was injected into the solution chamber, and the change in resistance was measured in real time. The sensor response was evaluated using the normalized resistance, i.e., (R–R₀)/R₀ × 100 (%), where R₀ and R are the resistances after injecting the baseline and analyte solutions, respectively.

Figure 4.

a) Schematic (top) and photograph (bottom) of a CNTF‐based anion sensor attached to a PDMS mold with a cylindrical solution chamber. b) Dynamic response transitions and c) average responses of the SQ2‐P4VP‐CNTF sensor upon the injection of 33.33 mm AcO⁻, Cl⁻, Br⁻, and NO₃ ⁻ (n = 3). Dynamic response transitions of d) SQ1‐CNTF and e) SQ2‐P4VP‐CNTF for AcO⁻ concentrations of 5–33.33 mm. f) Average sensor responses of SQ2‐P4VP‐CNTF toward 5–33.33 mm of AcO⁻ (n = 3).

Dynamic response transitions were evaluated toward various anions to understand the selective sensing property of the sensor. The sensor with non‐covalent functionalization, SQ2‐P4VP‐CNTF, exhibited a large response change with a rapid increase in resistance after injection of the AcO⁻ solution, whereas the change in resistance was negligible with Br⁻, Cl⁻, and NO₃ ⁻ (Figure 4b). Thus, high selectivity was attained toward AcO⁻ with an average response of 2.23 ± 0.12% (n = 3) at 33.33 mm while maintaining negligible responses (< 0.1%) toward Cl⁻, Br⁻, and NO₃ ⁻ (Figure 4c). The AcO⁻‐selective sensing response is attributed to the deprotonation of SQ2, which can interfere with the charge carriers in the CNTF as a result of the internal charge transfer in SQ2. On the other hand, minor chemiresistive signals were generated by hydrogen bond interactions between SQ2 and anions such as Cl⁻, Br⁻, and NO₃ ⁻. A pristine CNTF with no selector was also prepared to compare its anion‐sensing properties, and the corresponding sensor showed no response toward 33.33 mm AcO⁻, Cl⁻, Br⁻, or NO₃ ⁻ (Figure S10, Supporting Information).

To compare the AcO⁻ sensitivity with respect to the functionalization method, the dynamic response transitions were investigated in the concentration range of 5–33.33 mm (Figure 4d,e). For covalently functionalized SQ1‐CNTF, the sensor showed a response of 0.61% toward 33.33 mm AcO⁻ with an experimental detection limit of 16.67 mm (Figure 4d). The negative drift of SQ1‐CNTF in the AcO⁻ concentration range of 5–8.33 mm is mainly attributed to the low selector density on the surface of CNTFs. For this reason, the contribution of surface‐solvent interaction on resistance transition exceeded the chemical interaction between SQ1 and AcO⁻. On the other hand, non‐covalently functionalized SQ2‐P4VP‐CNTF exhibited a relatively high response of 2.23 ± 0.13% (n = 3) at 33.33 mm, and the detection limit was 0.63 ± 0.07% (n = 3) at 5 mm (Figure 4e). The average sensor responses of SQ2‐P4VP‐CNTFs exhibited high linearity (R² = 0.9884) in the AcO⁻ concentration range of 5–33.33 mm with negligible device‐to‐device response variation less than ± 0.16% (Figure 4f). A 3.6‐fold increase in the AcO⁻ response was achieved at 33.33 mm through non‐covalent selector functionalization (SQ2‐P4VP‐CNTF) compared to that with covalent functionalization (SQ1‐CNTF). This result implies that the chemical interactions between the selector and anion were efficiently transduced by the non‐covalent selector functionalization on the surface of the CNTF. In other words, the changes in the charge density on the surface of the CNTF after selector deprotonation can lower hole conduction in the CNTF, resulting in a large increase in resistance with a high sensor response. In addition, the selector density on the surface of the CNTFs can be increased by non‐covalent functionalization as a result of the physical attachment of SQ2‐P4VP. In contrast, surface‐specific chemical reactions are involved in the covalent functionalization of SQ1, which limits the number of selectors on the CNTF surface. To evaluate the respective degrees of functionalization, high‐resolution C 1s XPS analysis was performed for SQ1‐CNTF and SQ2‐P4VP‐CNTF (Figure S11, Supporting Information). The estimated selector densities of SQ1‐CNTF and SQ2‐P4VP‐CNTF were obtained by examining the area ratio of CF₃ bonds, which originated from the electron‐withdrawing group of selector molecules, to sp² C═C bonds.^{[ 61 ]} SQ2‐P4VP‐CNTF exhibited a 2.8‐fold higher CF₃ area ratio of 0.14 compared to 0.05 for SQ1‐CNTF. This result indicates that higher selector densities on the CNTF surface were achieved through non‐covalent functionalization methods. The higher selector density of SQ2‐P4VP‐CNTF contributes to the increased sensor responses toward AcO⁻ upon deprotonation. Furthermore, the cationic pyridinium moiety lowers the acidity (pK _a) of SQ2, leading to effective deprotonation upon adding AcO⁻, which improves the sensing response. Therefore, we concluded that the non‐covalent functionalization of selectors on CNTFs could maximize the sensing response toward AcO⁻ by effective charge transduction.

To extend our selector library and validate the application of CNTFs in anion sensors, various dual‐hydrogen bond donors, such as TU and CA were additionally synthesized, and their anion‐binding affinities were characterized by UV–vis titrations (Figures 5 and  6 ). An increase in absorbance at ≈330 nm was observed for TU upon the injection of AcO⁻ in DMSO (Figure 5a). The formation of a new absorbance band upon interacting with AcO⁻ was attributed to the deprotonation of TU inducing internal charge transfer.^{[ 62 ]} Deprotonation of TU proceeds with two equilibrium steps identical to those of SQ1 and SQ2 shown in Equations 1 and 2.^{[ 49 ]} A Job plot analysis indicated a 1:1 binding stoichiometry of TU with AcO⁻ (Figure S12a, Supporting Information). Variations in the mole fractions with respect to equivalents of AcO⁻ revealed that TU formed a hydrogen bond complex with a maximum mole fraction of 58% (Figure S12b, Supporting Information). The mole fraction of the hydrogen bond complex decreased after a further increase in the AcO⁻ concentration, resulting in the deprotonation of TU, leaving the hydrogen‐bond self‐complex of AcO⁻, i.e., H(AcO)₂ ⁻. The selective AcO⁻ binding property of TU was confirmed by UV–vis titrations, indicating a minor ΔAbs toward Cl⁻, Br⁻, and NO₃ ⁻ (Figures 5b and S13a–c, Supporting Information). Association constants (K₁) were calculated using the absorbance changes at 330 nm upon the addition of anions, and the results revealed that the value for AcO⁻ (K₁ = 2.12 × 10⁵ m ⁻¹) was 10.9‐fold higher than those for Cl⁻, Br⁻, and NO₃ ⁻ (Table S1, Supporting Information).

Figure 5.

a) UV–vis titration of TU ([TU] = 4.1 × 10⁻² mm) upon the addition of 0–1.43 equivalents of AcO⁻ in DMSO. b) ΔAbs of TU at 330 nm upon interacting with AcO⁻, Br⁻, Cl⁻, and NO₃ ⁻ in the range of 0–4 equivalents. c) Schematic illustration and d) SEM image of TU‐P4VP‐CNTF. e) Dynamic response transitions of TU‐P4VP‐CNTF toward AcO⁻, Br⁻, Cl⁻, and NO₃ ⁻ at 33.33 mm. f) Concentration‐dependent response transitions of TU‐P4VP‐CNTF toward 5–33.33 mm AcO⁻.

Figure 6.

a) UV–vis titration of CA ([CA] = 2.0 × 10⁻² mm) upon addition of 0–1.65 equivalents of AcO⁻ in dry DCM. b) ΔAbs of CA at 420 nm upon interaction with AcO⁻, Br⁻, Cl⁻, and NO₃ ⁻ in the range of 0–5 equivalents c) Schematic illustration and d) SEM image of CA‐P4VP‐CNTF. e) Dynamic response transitions of CA‐P4VP‐CNTF toward AcO⁻, Br⁻, Cl⁻, and NO₃ ⁻ at 33.33 mm. f) Concentration‐dependent response transitions of CA‐P4VP‐CNTF toward 5–33.33 mm AcO⁻.

To maximize the anion‐sensing response, TU was non‐covalently functionalized onto the surface of the CNTFs with the assistance of P4VP. Specifically, TU‐P4VP was prepared by the quaternization of the pyridyl N in P4VP, which converted it into ammonium‐alkylated pyridinium, and a subsequent chemical reaction with 3,5‐bis(trifluoromethyl)phenyl isothiocyanate to form TU (Scheme S4, Supporting Information). The presence of pyridinium groups was confirmed by FT–IR spectroscopy by assigning a vibration peak at 1639 cm⁻¹ (Figure S6c, Supporting Information). TU‐P4VP was physically attached to the CNTF through dip‐coating to form TU‐P4VP‐CNTF (Figure 5c). To validate the non‐covalent functionalization of the selector on the CNTF, the surface morphology of TU‐P4VP‐CNTF was investigated using SEM (Figure 5d). The physical attachment of the polymeric film to the surface of the CNTFs indicated dense coatings of TU‐P4VP through van der Waals interactions.^{[ 63 ]}

The anion‐sensing properties were also investigated by measuring the resistance changes upon injecting the analyte solutions. Notably, TU‐P4VP‐CNTF exhibited a high selectivity for AcO⁻ with a response of 0.85% at 33.33 mm. In contrast, negligible responses occurred after exposure to 33.33 mm Cl⁻, Br⁻, and NO₃ ⁻ (Figure 5e). The concentration‐dependent responses of TU‐P4VP‐CNTF were investigated upon the injection of AcO⁻ in the range of 5–33.33 mm (Figure 5f). The sensor responses decreased with the decreasing AcO⁻ concentration, and the experimental limit of detection was identified as 5 mm with a response of 0.17%.

Similarly, the anion‐binding properties of CA were investigated through UV–vis titrations in dry DCM to maintain the central carbonyl groups of croconamide without hydration.^{[ 64 ]} The CA exhibited an increased absorbance peak at 420 nm and a decreased absorbance peak at 372 nm after the addition of 1.65 equivalent AcO⁻, presenting isobestic points at 376 and 500 nm (Figure 6a). In contrast, only minor absorbance changes were observed for Cl⁻, Br⁻, and NO₃ ⁻, indicating weak chemical interactions with CA (Figure 6b; Figure S14a–c, Supporting Information). The distinct changes in the absorbance upon the injection of AcO⁻ were attributed to the deprotonation of CA. To further investigate the anion‐binding properties of CA, a Job plot analysis was conducted using the absorbance changes at 420 nm. The results revealed a 1:1 binding stoichiometry between CA and AcO⁻ (Figure S15a, Supporting Information). The concentration of deprotonated CA (i.e., CA─H⁺) species reached a maximum mole fraction of 58% upon adding two equivalents of AcO⁻. This is because highly acidic CA is prone to deprotonation upon adding basic AcO⁻ by producing a hydrogen‐bond self‐complex (i.e., H(AcO)₂ ⁻). A further increase in the AcO⁻ concentration resulted in the additional deprotonation of N─H protons, leading to the doubly deprotonated CA (CA─2H⁺) with a decreased mole fraction of CA─H⁺ (Figure S15b, Supporting Information).^{[ 65 ]} These deprotonation processes of CA can be expressed by the following Equations (3) and (4):^{[ 66 ]}

$$\begin{array}{ccc}\hfill \mathbf{R}{\mathrm{H}}_2& +& 2{\mathrm{AcO}}^{-}\leftrightarrow \mathbf{R}{\mathrm{H}}^{-}+\mathrm{H}{{\left(\mathrm{AcO}\right)}_2}^{-},{\mathrm{K}}_3\hfill \\ & & =[\mathbf{R}{\mathrm{H}}^{-}][\mathrm{H}{{\left(\mathrm{AcO}\right)}_2}^{-}]/[\mathbf{R}{\mathrm{H}}_2]{[{\mathrm{AcO}}^{-}]}^2\hfill \end{array}$$ (3)

$$\begin{array}{ccc}\hfill \mathbf{R}{\mathrm{H}}^{-}& +& 2{\mathrm{AcO}}^{-}\leftrightarrow {\mathbf{R}}^{2-}+\mathrm{H}{{\left(\mathrm{AcO}\right)}_2}^{-},{\mathrm{K}}_4\hfill \\ & =& [{\mathbf{R}}^{2-}][\mathrm{H}{{\left(\mathrm{AcO}\right)}_2}^{-}]/[\mathbf{R}{\mathrm{H}}^{-}]{[{\mathrm{AcO}}^{-}]}^2\hfill \end{array}$$ (4)

CA was also non‐covalently functionalized onto CNTFs with the assistance of modified P4VP with CA. Similarly, the pyridyl groups of P4VP were quaternized to form croconamide‐based selectors, resulting in CA‐P4VP (Scheme S4, Supporting Information). In the FT–IR spectrum of CA‐P4VP, a vibration peak related to pyridinium was observed at 1639 cm⁻¹ (Figure S6d, Supporting Information). Subsequently, CA‐P4VP was physically attached to the CNTF through dip‐coating to form CA‐P4VP‐CNTF (Figure 6c). An SEM image of the CA‐P4VP‐CNTF revealed the successful attachment of polymeric films to the CNTFs (Figure 6d).

The anion‐sensing characteristics of CA‐P4VP‐CNTF were evaluated by monitoring the changes in the resistance upon introducing anion solutions. CA‐P4VP‐CNTF showed a sensor response of 0.42% at 33.33 mm AcO⁻. In contrast, negligible anion‐sensing responses were observed toward 33.33 mm Cl⁻, Br⁻, and NO₃ ⁻, thereby indicating high selectivity for AcO⁻ (Figure 6e). The concentration dependence of the response was investigated upon the injection of 5–33.33 mm AcO⁻ (Figure 6f). CA‐P4VP‐CNTF exhibited an increased sensor response at high AcO⁻ concentrations. The experimental detection limit of CA‐P4VP‐CNTF was 5 mm, with a sensor response of 0.06%.

The sensitive and selective detection of AcO⁻ was mainly attributed to the deprotonation of TU and CA, which results in large changes in the resistance of the functionalized CNTFs. TU‐P4VP‐CNTF exhibited a higher sensitivity to basic AcO⁻ than CA‐P4VP‐CNTF, despite the higher acidity of croconamides compared to that of thioureas.^{[ 14} , ^{15 ]} This result may be attributed to the partial deprotonation of highly acidic croconamides (pK _a = 6–8) under neutral pH conditions in acetonitrile before adding anions, which consequently diminishes their capacity to interact with anions effectively.^{[ 15 ]}

To summarize the sensor responses with respect to the functionalized selectors, dynamic resistance transitions toward 33.33 mm of AcO⁻ and average sensor responses were presented in Figure S16 (Supporting Information). From the result, SQ2‐P4VP‐CNTF exhibited the highest AcO⁻ response of 2.23 ± 0.13% (n = 3) followed by the responses of TU‐P4VP‐CNTF (0.84 ± 0.08%, n = 3) > SQ1‐CNTF (0.51 ± 0.08%, n = 3) > CA‐P4VP‐CNTF (0.38 ± 0.12%, n = 3).

The pH effect on AcO⁻‐sensing was investigated using SQ1‐CNTF and SQ2‐P4VP‐CNTF in aqueous solutions with pH 2, 7, and 12 (Figure S17, Supporting Information). The sensors exhibited negligible response toward 33.33 mm of AcO⁻, primarily due to competing hydrogen‐bonding interaction between selectors and water molecules. Further refinement of selector structures should be conducted to increase binding affinity toward AcO⁻ in aqueous solution with sub‐mm concentration detection capability. Nevertheless, our work provides novel chemiresistive AcO⁻‐sensing properties (Table S2, Supporting Information) with competitive achievements regarding the detection of AcO⁻ in mm concentration ranges as compared to the electrochemical sensing technique.^{[ 67} , ⁶⁸ , ⁶⁹ , ^{70 ]}

To validate the AcO⁻‐sensing in real‐world applications, we simulated AcO⁻ detection in a Luria‐Bertani (LB) broth, which is a common cell culture medium for E. coli. Pre‐processing, such as centrifugation, filtration, and dilution in an organic solvent of real‐world samples, can improve sensing reliability by removing biocomponents, including cell fraction.^{[ 60} , ⁷¹ , ^{72 ]} In this regard, the analyte samples were prepared by pre‐processing the LB broth by diluting it with an organic solvent (e.g., acetonitrile). Subsequently, 2 µL of the analyte sample was injected into SQ2‐P4VP‐CNTF sensors, and resistance transition was measured (Figure S18a, Supporting Information). From the results, the sensor exhibited a response of 0.45 ± 0.23% (n = 3) upon injecting the analyte sample containing 40 mm AcO⁻, while a negligible response was observed for the analyte sample without AcO⁻ (Figure S18b, Supporting Information). The negligible response in the absence of AcO⁻ indicates the high selectivity of the sensors toward AcO⁻ considering that LB broth contains a high concentration of interfering Cl⁻.

To understand the relationship between the anion‐binding affinity of the selectors and the sensor response, the association constants of the selectors and their respective sensor responses of the functionalized CNTFs toward AcO⁻ are summarized in Table 1 . The correlations between the association constants and sensor responses of SQ1, SQ2, and TU were first investigated. Among these selectors, SQ2 exhibited the highest K₁ value, followed by SQ1 and TU, indicating strong hydrogen bonds between SQ2 and AcO⁻. In terms of K₂, the highest value was attained with SQ2, indicating the strongest deprotonation tendency among the selectors. However, the K₂ value of TU was slightly higher than that of SQ1 because of the presence of the pyridinium functional group, which implies that TU is more acidic and easier to deprotonate than SQ1. In terms of the sensing response, the CNTF sensor functionalized with the SQ2 selector presented an exceptionally high response (2.23 ± 0.13%) followed by TU (0.84 ± 0.08%) and SQ1 (0.51 ± 0.08%) toward 33.33 mm AcO⁻. These response trends correspond to the magnitude of the K₂ values of the selectors, i.e., SQ2 > TU > SQ1. This result explains the fact that the sensor response is primarily influenced by the tendency of deprotonation of selectors. Therefore, the sensor responses were strongly correlated with the binding constant K₂, which is associated with the formation of deprotonated selectors (Figure S19, Supporting Information). This can be further proven by the selective AcO⁻‐sensing properties of the sensors, as Cl⁻, Br⁻, and NO₃ ⁻ formed hydrogen bonds with the selectors and exhibited negligible sensing responses.

Table 1.

Association constants (K₁ and K₂) of selectors toward AcO⁻ as determined by UV–vis titrations in DMSO and the corresponding AcO⁻‐sensing responses at 33.33 mm after non‐covalently functionalizing each selector onto CNTFs.

Selector	SQ1	SQ2	TU
K1 (m −1)	1.95 × 106	2.16 × 106	2.12 × 105
K2 (‐)	1.61 × 105	1.04 × 106	2.26 × 105
Response (%)	0.51 ± 0.08	2.23 ± 0.13	0.84 ± 0.08

The association constants (K₃ and K₄) for CA and the corresponding sensor response of CA‐P4VP‐CNTF are listed in Table 2 , considering the use of different solvents in the UV–vis titration and equilibrium states of CA. The results revealed that relatively low association constants were attained, with the lowest sensing response toward AcO⁻ (0.38 ± 0.12% at 33.33 mm), which mainly resulted from the partial deprotonation of highly acidic croconamides in the baseline solution before the injection of AcO⁻ solution.

Table 2.

Association constants (K₃ and K₄) of CA toward AcO⁻ as determined by UV–vis titration in dry DCM and the sensing responses of CA‐P4VP‐CNTF toward 33.33 mm AcO⁻.

Selector	CA
K3 (m −1)	9.90 × 104
K4 (m −1)	1.21 × 104
Response (%)	0.38 ± 0.12

Finally, we demonstrated the wireless detection of AcO⁻ using a sensor module that can transfer a sensor resistance signal to a smartphone in real time (Figure 7 ). The sensor module was designed to be portable, with dimensions of 50 mm × 80 mm to enable the on‐site detection of anionic analytes (Figure 7a). The sensor module had electrical contacts that could be connected to an anion sensor, and the sensing data were transferred to a smartphone via Bluetooth communication. An anion sensor fabricated with a single line (1‐wire) of SQ2‐P4VP‐CNTF within the sensing chamber was integrated into the sensor module. The sensor exhibited responses of 1.41% and 0.97% for 33.33 and 16.67 mm AcO⁻, respectively (Figure 7b). To further increase the resolution of the resistance changes, another anion sensor was fabricated by assembling a continuous SQ2‐P4VP‐CNTF with three lines (3‐wire) located within the sensing chamber (Figure S20a,b, Supporting Information). The resistance changes were evaluated by calculating the absolute resistance change (ΔR), i.e., R–R₀ and the performance of 1‐wire SQ2‐P4VP‐CNTF and 3‐wire SQ2‐P4VP‐CNTF was compared upon exposure to AcO⁻ at 33.33 mm. The absolute resistance change was ≈2.73‐fold higher with the 3‐wire SQ2‐P4VP‐CNTF compared with that of the corresponding 1‐wire (Figure S20c, Supporting Information). As a result, a high‐resolution AcO⁻‐sensing response was achieved using 3‐wire SQ2‐P4VP‐CNTF, with concentration‐dependent responses of 1.64% and 0.91% at 33.33 and 16.67 mm, respectively (Figure 7c). Regarding the reusability of the sensors, the recovery property was characterized by adding 2 µL AcO⁻ followed by 10 µL of HBr (0.1 m) (Figure S21, Supporting Information). Upon injecting the HBr solution, the sensor response recovered to its initial state, which is attributed to the protonation of deprotonated selectors. In addition, two sensing chambers were formed in series on a single sensor chip to amplify the response signals or for serial analysis of multiple analytes. For example, continuous measurement using two sensing chambers was demonstrated for the amplified AcO⁻ response by sequentially injecting the analyte solution (33.33 mm) into sensing chambers (1,2) (Figure 7d). Upon separately injecting the AcO⁻ solution into each chamber, step‐like response changes of 1.83 and 1.84% were observed, resulting in an overall response change of 3.67% (Figure 7e). These results demonstrate the potential application of our sensor platform for the real‐time wireless screening of multiple anionic species using various anion selectors.

Figure 7.

a) Real‐time wireless anion sensing by integrating an anion sensor into a sensing module. Dynamic response transitions of b) 1‐wire SQ2‐P4VP‐CNTF and c) 3‐wire SQ2‐P4VP‐CNTF toward 16.67 and 33.33 mm AcO⁻. d) Schematic illustration of continuous sensing measurement by forming two chambers (1,2) for the sequential injection of anion solutions. e) Dynamic response transitions of 3‐wire SQ2‐P4VP‐CNTF upon sequentially adding the 33.33 mm AcO⁻ solution into chambers (1,2).

3. Conclusion

In summary, we performed a comparative study to understand the effective charge transduction properties of CNTF‐based chemiresistive anion sensors depending on the covalent and non‐covalent selector functionalization techniques. Electrically conductive and mechanically strong CNTFs were prepared using a wet‐spinning process and subsequently functionalized with squaramide‐based dual‐hydrogen bond donors (SQ1 and SQ2). The anion‐binding properties were first investigated by UV–vis titrations, which revealed the deprotonation of both SQ1 and SQ2 upon chemically interacting with AcO⁻, whereas hydrogen bonding was observed with Cl⁻, Br⁻, and NO₃ ⁻. To fabricate chemiresistive anion sensors, SQ1 was covalently functionalized via a chemical reaction with alkyl amines on the CNTF surface. In contrast, SQ2‐P4VP was non‐covalently functionalized onto the CNTF through physical attachment via van der Waals interactions. The sensor responses were then evaluated in terms of the normalized resistance change upon injecting anion solutions into the sensor. As a result, SQ2‐P4VP‐CNTF exhibited a 3.6‐fold increased sensor response (2.23%) toward 33.33 mm AcO⁻ compared with the response (0.61%) of SQ1‐CNTF. This enhanced anion sensor response through the non‐covalent functionalization of SQ2 is attributed to the increased selector density and the formation of cationic pyridinium on the surface of the CNTF.

To expand the anion selector library, other selectors comprising different dual‐hydrogen bond donors, TU and CA, were non‐covalently functionalized onto CNTFs with the assistance of P4VP. As a result, TU‐P4VP‐CNTF and CA‐P4VP‐CNTF exhibited responses of 0.85 and 0.42% toward 33.33 mm AcO⁻, respectively, which was achieved through the deprotonation of selectors and charge transduction. The correlation between the AcO⁻‐sensing responses and selector types was explained by calculating the equilibrium constants (K₂ and K₄) for selector deprotonation, wherein a strong correlation was attained, decreasing in the order SQ2‐P4VP‐CNTF > TU‐P4VP‐CNTF > CA‐P4VP‐CNTF. All the anion sensors showed negligible responses toward Cl⁻, Br⁻, and NO₃ ⁻, which confirms that the anion‐sensing mechanism can be attributed to the internal charge transfer of selectors upon deprotonation. Finally, practical AcO⁻ detection was demonstrated by integrating a 3‐wire SQ2‐P4VP‐CNTF anion sensor and two sensing chambers into a sensing module, which could transmit amplified resistance data via wireless communication in real time.

4. Experimental Section

Materials

All chemicals and reagents were used without additional purification unless otherwise mentioned. 3‐Bromopropylamine hydrobromide, triethylamine, 6‐amino‐1‐hexanol, iodomethane, chlorosulfonic acid (CSA), 2‐aminoethanethiol hydrochloride, dicumyl peroxide, tetrabutylammonium bromide, tetrabutylammonium chloride, and tetrabutylammonium nitrate were purchased from Sigma–Aldrich. 3,5‐Bis(trifluoromethyl)phenyl isothiocyanate, 3,4‐dimethoxycyclobut‐3‐ene‐1,2‐dione, 3,5‐bis(trifluoromethyl)aniline, and tetrabutylammonium acetate were purchased from TCI Chemicals. Croconic acid, AgNO₃, and hydrobromic acid (48% aqueous) were purchased from Alfa Aesar. Poly(4‐vinylpyridine) (P4VP) (M_w = 200000) was purchased from Scientific Polymer Product, INC. Deuterated dimethyl sulfoxide (DMSO‐d₆) for NMR spectroscopy was purchased from Cambridge Isotope Laboratories, Inc. The SWCNTs were obtained from OCSiAl.

Synthesis

The model structures of the selector molecules SQ1, SQ2, TU, and CA were synthesized based on our previous works with slight modifications (Supporting Information).^{[ 17} , ¹⁸ , ^{19 ]} To synthesize the CNTFs, a CNT solution was prepared at a concentration of 20 mg mL⁻¹ in CSA and dispersed for a week. The mixture was then transferred into a glass syringe (25G needle diameter) and injected into an acetone bath at a rate of 0.1 mL min⁻¹ using a syringe pump. The resulting CNTF was drawn out and wound onto a Teflon bobbin at a speed of 2 m min⁻¹. Subsequently, the fiber was washed by soaking in acetone for 30 min and dried in a vacuum oven at 170 °C overnight. To synthesize the NH₂‐CNTFs, CNTs were first decorated with alkyl amines. To this end, 2‐aminoethanethiol hydrochloride and dicumyl peroxide were added to the CNT dispersion and refluxed under a N₂ atmosphere. After 48 h, the NaOH solution was added and diluted with DI water. The resulting products were filtered and washed with DI water and acetone. After drying in a vacuum oven, amine‐grafted CNTs were obtained. The NH₂‐CNTFs were synthesized by wet‐spinning amine‐functionalized CNTs in a manner identical to that described above.

Non‐Covalent Functionalization of Selectors on CNTFs

Dual‐hydrogen‐bond donor‐functionalized P4VP was first prepared by a quaternization reaction using pyridyl groups, as shown in Scheme S4 (Supporting Information) (SQ2‐P4VP, TU‐P4VP, and CA‐P4VP). Subsequently, 4 mg of modified P4VP was dissolved in 100 µL of MeOH. The CNTFs were immersed in the solution for 2 h and then dried in air to obtain the functionalized CNTFs.

Sensor Fabrication and Measurement

The anion sensor was fabricated by attaching a PDMS mold containing a cylindrical hole with a diameter of 3 mm on top of the functionalized CNTFs to create the solution chamber. The sensing properties of the functionalized CNTFs were characterized toward AcO⁻, Cl⁻, Br⁻, and NO₃ ⁻ in acetonitrile. First, 10 µL of acetonitrile was injected into the solution chamber to reach a baseline resistance. After stabilizing the resistance, 2 µL of anion solution was injected in the concentration range of 5–33.33 mm. The sensor response was measured as the normalized resistance change, i.e., (R−R₀)/R₀ × 100 (%), where R and R₀ are the resistances after injecting the anion solution and the baseline solvent, respectively.

To investigate the AcO⁻sensing properties using a cell culture medium, Luria–Bertani (LB) broth was prepared by dissolving 10 g of tryptone, 5 g of yeast extract, and 10 g of NaCl in 1 L of distilled water. AcO⁻ was then added to LB broth using tetrabutylammonium‐based salts to simulate AcO⁻ detection within the cell culture medium. The analyte samples were prepared by diluting the LB broth containing AcO⁻ to a volume ratio of 5% in acetonitrile. For comparison, a control sample was prepared by diluting LB broth in acetonitrile without the addition of AcO⁻. To characterize the sensing properties, 10 µL of acetonitrile was injected into the SQ2‐P4VP‐CNTF sensor to form baseline resistance. Subsequently, 2 µL of analyte samples were injected into the SQ2‐P4VP‐CNTF sensor, and the sensor responses, i.e., (R−R₀)/R₀ × 100 (%), were evaluated as described above.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Choi S.‐H., Lee J.‐S., Lee S., Jeong H. S., Choi S.‐J., Dual‐Hydrogen Bond Donor‐Functionalized Carbon Nanotube Fibers: Enhancing Anion‐Sensing Performance Through Functionalization Approaches. Small 2025, 21, 2405070. 10.1002/smll.202405070

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.