Chirality is one of the most universal and fundamental behavior in nature [1, 2]. Nowadays, chiral compounds are particularly widely used in medicinal chemistry. The biological behavior and toxicity of drug enantiomers often exhibits remarkable differences [3, 4]. As a typical example, R-thalidomide has the function of relieving nausea, whereas its mirror counterpart (S-thalidomide) could cause congenital disabilities [5]. Thus, efficient recognition and separation of chiral molecules is of great importance for medical and life sciences [6-8]. However, sensitive chiral recognition is always challenging due to the extreme similarity of two enantiomers. Up to now, some conventional techniques have been adopted for separation and detection of chiral enantiomers including high performance liquid chromatography (HPLC) [9], gas chromatography (GC) [10, 11], circular dichroism (CD) [12, 13], fluorescence spectroscopy [14-16] and surface-enhanced Raman spectroscopy (SERS) [17]. To achieve satisfactory results, complex sample pretreatment steps and high cost is usually required, which largely restrict their practical applications. Enantiomer sensing platforms with high sensitivity, low cost, and ease of use are highly desired to overcome these challenges.

Nanofluidic is a discipline that studies and applies fluid behavior and property in nanoconfinement [18, 19], which has been widely used in various fields including ultrafiltration and separation, bioanalysis, seawater desalination, and energy conversion [20-25]. Particularly, when nanofluidic channels are asymmetric, unique mass transfer behavior of ion current rectification (ICR) [26] will appear, which conducts ion current preferentially in one direction and exhibits a nonlinear electric response to the applied potential [27]. By measuring the changed ICR property before and after molecular recognition, sensitive and label-free detection of circulating tumor cell or biomolecules/ions could be achieved on asymmetric nanofluidic membranes [28]. It has been proved that ICR property could improve the detection sensitivity of bioanalysis considerately [28-33]. Obviously, nanofluidic ICR membranes display remarkable features of high sensitivity, low cost, and easy integratability with electrochemical/optical detection techniques. Despite many advantages descried above, there is rare work on constructing nanofluidic chiral sensors based on ICR. In our previous work, a series of nanofluidic ICR devices have been fabricated by integrating array nanochannels membrane with functional materials like nanoparticles based network [33], metal organic framework (MOF) [34], covalent organic framework (COF) [35] and others [28-32]. The asymmetric structure as well as the charge density endows the devices with remarkable ICR property and good bioanalysis performance. Among these functional materials, COF can be easily functionalized due to the advantages of periodic building block arrays, plenty of functional groups, regular shaped pores and high porosity. They are considered as ideal candidates for construction of superior chiral sensor owing to its higher stability in water than other porous materials and good biocompatibility caused by the absence of metals in the structure [36-39].

In this work, we constructed a nanofluidic chiral sensor based on ICR property for electrochemical enantioselective recognition (Scheme 1). First, an asymmetric nanochannel membrane was fabricated by in-situ growth of COF on anodic aluminum oxide (AAO) substrate via interfacial hydrothermal method [35]. Then, post-modification of l-penicillamine (l-Pen) on COF was performed, providing a chiral environment for enantioselective recognition and detection. Due to the asymmetric structure and charge density between COF and AAO, the as-prepared chiral COF/AAO (c-COF/AAO) membrane exhibits excellent ICR property. Binding of chiral target molecules will change the membrane surface charge density, leading to the variation of ion current as well as the ICR property. In combination with the electrochemical detection technique, the chiral recognition and in-situ detection can be performed by measuring the transmembrane ion current and ICR property before and after enantioselective recognition. As demonstration, S-/R-Naproxen (S-/R-Npx) enantiomers were used as the testing model. A super low detection limit of 3.88 pmol/L for nanofluidic ICR membranes was successfully achieved. The present work highlights the tremendous potential of nanofluidic ICR devices in analytical applications.

The morphology of the prepared c-COF/AAO membrane was characterized by scanning electron microscopy (SEM) (Fig. 1A). The diameter of the fabricated AAO was about 50 nm (Fig. S1 in Supporting information). From the top and bottom view, c-COF covered the top surface of AAO and formed a dense membrane while no c-COF grew on the bottom of AAO. The corresponding EDS mapping (Fig. S2 in Supporting information) indicated the uniform presence of C, N and S in chiral COF, verifying the successful modification of l-Pen. From the cross-sectional view, a consistent layer of c-COF with a thickness of approximately 270 nm was obtained and no c-COF was found in AAO channels. Compared to the COF/AAO membrane (Fig. S3 in Supporting information), the chiral membrane still maintained integrity and homogeneity, which was consistent with the morphology of the original COF powder (Fig. S4 in Supporting information). In addition, the thickness of the COF membrane can be regulated by the growth time. As depicted in the SEM images in Fig. S5 (Supporting information), the COF membrane formed under 12 h exhibited some defects. Extending the growth time can enable the attainment of intact, defect-free COF membranes. The corresponding Ⅰ-Ⅴ curves showed the typical ICR properties (Fig. S6 in Supporting information). Considering that larger thickness would result in less ion permeability, the COF membrane with the growth time of 24 h was adopted in the following experiments.

According to the X-ray diffraction (XRD) pattern (Fig. S7 in Supporting information), l-Pen had no influence on the structure of COF. Nitrogen adsorption analysis results showed that both COF and c-COF possessed typical porous structures, and the Brunauer-Emmett-Teller (BET) surface area of COF and c-COF were achieved as 779.670 m²/g and 619.495 m²/g respectively (Fig. 1B and Fig. S8 in Supporting information). Compared with COF, the surface area of c-COF decreased slightly, mainly due to the introduction of l-Pen attached to each vinyl group (Fig. S9 in Supporting information), which was also confirmed by the reduction of the pore width from 1.64 nm for COF to 1.58 nm for c-COF (Fig. 1B and Fig. S8). Compared to the zeta potential of the unmodified COF (+11.3 mV, Fig. 1C), c-COF exhibited a significant reduction, reaching approximately −31.3 mV, which could be contributed to plenty of -COOH groups on the c-COF. Due to deprotonation of carboxyl group under neutral environment, c-COF is charged negatively. The -COOH groups also led to better hydrophilicity of the c-COF/AAO membrane than the COF/AAO membrane (Fig. S10 in Supporting information).

Analysis of the Fourier-transform infrared (FTIR) and X-ray photoelectron spectroscopy (XPS) further confirmed the incorporation of l-Pen into COF structure. Compared to COF, a new infrared absorption peak at 824 cm⁻¹ for c-COF was observed (Fig. 1D), which was attributed to the C-S bond, approving the successful grafting of l-Pen on vinyl-functionalized COF via the thiol-ene "click" reaction. The sulfur signal at a binding energy of 164.2 eV appeared in the XPS spectrum of c-COF (Fig. 1E), which arises from post-synthetic modification of l-Pen. CD spectroscopy was also used to demonstrate the chirality of c-COF (Fig. 1F). Compared with unmodified COF, there was an adsorption peak at 222 nm in the CD spectrum of l-Pen-COF and d-Pen-COF, which was not observed for pure COF. All above results demonstrate the successful construction of the c-COF/AAO membrane.

To investigate the ion transport properties of the prepared c-COF/AAO membranes, KCl was used as the testing electrolyte. Two Ag/AgCl electrodes were used for electrochemical detection when voltage bias was applied (Fig. 2A). Fig. 2B showed that the Ⅰ-Ⅴ curve of bare AAO was a straight line. After in-situ synthesis of COF on AAO surface, the ion current exhibited an obvious diode-like ICR behavior, which was caused by the asymmetric transmembrane transport of cations and anions [40]. The ICR ratio is calculated by I_-1/I₊₁, where I₊₁ represents the ion current recorded at +1 Ⅴ and I_-1 represents the ion current recorded at −1 Ⅴ. After the COF/AAO membrane surface was modified with l-Pen, the chiral membrane had superior cation-selectivity due to -COOH groups on the c-COF layer. The surface charge of the c-COF/AAO membrane was reversed, resulting in an inversion of the ion rectification direction (ICR = I₊₁/I_-1). The same ICR reversal phenomenon was also observed when the membrane surface was modified with d-Pen or l & d-Pen (l-Pen/d-Pen = 1/1) (Figs. S11 and S12 in Supporting information).

Both ion concentration and valence have considerable influence on nanofluidic ion transport behavior. Fig. 2C showed the Ⅰ-Ⅴ curves of the prepared c-COF/AAO membrane in various concentrations of KCl solution. All the curves displayed typical ICR properties, and the ion current increased with ion concentration owing to the raised ion strength. The ICR ratio was calculated in Fig. 2D. It was observed that the ICR ratio increased first and then decreased with the increasing KCl concentration. The phenomenon was caused by the formation of the electric double layer (EDL) formed by absorption of counterions on nanochannels surface [41, 42]. In low concentration electrolytes, the ICR ratio firstly increased with the increase of electrolyte concentration. The maximum value was reached in 1 mmol/L KCl electrolyte solution. Continuing increase of KCl electrolyte concentration resulted in the decreased ICR ratio. This is because the EDL is attenuated in higher electrolyte concentrations, and the ion transport controlled by the surface charge is weakened. In order to explore the effect of ion valence and type on the ion transport, divalent salt BaCl₂ and K₂SO₄ were used as the testing electrolyte instead of KCl. Fig. S13 (Supporting information) showed the Ⅰ-Ⅴ curves of the c-COF/AAO membrane measured in different electrolyte concentrations. Compared to the Ⅰ-Ⅴ curves of KCl, the ion current of BaCl₂ and K₂SO₄ were obviously larger than KCl for the same concentration and nearly the same change trends occurred for ICR properties in three different electrolytes solutions. However, the maximum ICR ratio was achieved at 1 mmol/L for monovalent salts and 0.1 mmol/L for divalent salt. It had been proved that when the negative charges on nanochannel surface is balanced by cations, the ICR ratio could reach the maximum value [35]. Since Ba²⁺ (SO₄^2-) carries more amount of charge than K⁺ (Cl^-), it has a stronger shielding effect. Therefore, divalent salt at lower concentrations reached maximum ICR ratio. As shown in Fig. S14 (Supporting information), the ion transport properties of the COF membrane were also tested in electrolyte solutions of different concentrations and different valence states. Since the COF/AAO membrane was positively charged, the Ⅰ-Ⅴ curves were reversed compared with that of the c-COF/AAO membrane, but its change trend was basically the same.

As known, pH value is also the important factor affecting ion transport behavior in nanochannel. Fig. 2E showed the Ⅰ-Ⅴ curves of the c-COF/AAO membrane were measured in KCl solution under different pH values, and the ICR ratio was shown in Fig. 2F. Obviously, the ICR properties could be easily influenced by pH, and the ICR reversed at pH 5. This pH-dependent ion transport properties of the nanofluidic membrane could be explained by the enhanced asymmetry of charge density. Due to the large pore size of AAO, it is difficult for AAO to generate effective ICR. Thus, the ICR property of the c-COF/AAO membrane is actually caused by c-COF. The surface of c-COF is endowed with carboxyl functional groups. When the membrane is placed in a solution with pH higher than 6, the c-COF is charged negatively due to deprotonation of carboxyl group. The higher the pH, the more negative charges the membrane has. Therefore, increased pH value can lead to higher ICR ratio. And the c-COF membrane exhibits a more prominent negative potential at pH 11 compared to that in other pH conditions (Fig. S15 in Supporting information). Under this condition, the device will attract more cations, resulting in superior cation selectivity and permeability. When pH is lower than 6, the carboxyl groups combine with protons, and the surface charge changes from negative to positive, resulting in reversed ICR property. The effect of pH on the surface charges was verified by measuring the zeta potential of the membrane at different pH environments (Fig. S15). The above results indicate that the c-COF/AAO membrane exhibits good Ⅰ-Ⅴ properties and high ion rectification in electrolyte solutions at different concentrations and pH, which paves the way for subsequent selective identification.

To investigate the performance of the prepared nanofluidic membrane for chiral recognition, we placed the c-COF/AAO membrane between two half-cells, and applied two Ag/AgCl electrodes to provide transmembrane potential (Fig. 3A). The Ⅰ-Ⅴ curves under a voltage bias were recorded by filling the two halves of the cell with 100 mmol/L KCl solution. Then, chiral molecular solutions of 0.1 mmol/L S-Npx/R-Npx were added into the cells. It was found that no obvious change occurred for the unmodified COF/AAO membrane (Fig. 3B). However, the c-COF/AAO membrane displayed obvious different ion current in S-Npx and R-Npx solutions (Fig. 3D). The ion current at +1.0 Ⅴ of the c-COF/AAO membrane decreased markedly with the addition of 0.1 mmol/L S-Npx solution, while the ion current remained almost constant after adding 0.1 mmol/L R-Npx solution. This was also evidenced by the changes in zeta potential (Fig. 3, Fig. 3). We found that the zeta potential of COF remained basically unchanged after the addition of S-Npx/R-Npx. However, the zeta potential changed from −31.3 mV to −19.7 mV when c-COF combined with S-Npx, while nearly no change was observed when R-Npx was added. We also studied the chiral recognition performance of the c-COF/AAO membrane in lower concentrations of KCl solutions (1 mmol/L and 10 mmol/L). The similar recognition capability for Npx enantiomer occurs (Fig. S16 in Supporting information). However, the largest change of ion current values was achieved in 100 mmol/L KCl. Larger current response is helpful to increase detection sensitivity. Therefore, 100 mmol/L KCl was used as the electrolyte during the following chiral recognition experiment. These results demonstrate the perfect enantioselective recognition ability of the prepared c-COF/AAO membrane. Notably, the d-Pen-COF/AAO membrane produced the opposite result with regards to recognition (Figs. S17 and S19 in Supporting information), and the L & D-Pen-COF/AAO membrane did not show chiral recognition ability (Figs. S18 and S19 in Supporting information).

The concentration of l-Pen was optimized in order to achieve efficient enantiomers recognition. The results of optimizing the concentration of l-Pen showed that the Ⅰ-Ⅴ curve was reversed at 1 mmol/L, which indicated that the positively charged COF/AAO membrane was changed to the negatively charged c-COF/AAO membrane (Fig. 4A). Subsequently, the Ⅰ-Ⅴ properties increased continuously as the l-Pen concentration further increased until the ICR ratio reached its maximum at the concentration of 10 mmol/L (Fig. 4B). Therefore, 10 mmol/ l-Pen solution was selected as a suitable candidate to modify the COF/AAO membrane in order to obtain the maximum Ⅰ-Ⅴ curves changes.

To achieve the optimum enantiomer reaction time, 20 µL of 0.1 mmol/L S-/R-Npx solutions were added into the two half cells for different times (5, 10, 15, 20, 25, 30 min) at room temperature. The Ⅰ-Ⅴ properties of the c-COF/AAO membrane after reaction with the enantiomers were investigated. The measured Ⅰ-Ⅴ curves were shown in Fig. 4C, and the corresponding current values at +1.0 Ⅴ was displayed in Fig. 4D. When S-Npx interacted with the c-COF/AAO membrane, the ion current at +1.0 Ⅴ decreased continuously with reaction time and then reached a stable state after 15 min. However, the Ⅰ-Ⅴ curves had no significant changes after adding 0.1 mmol/L R-Npx with the increase of reaction time (Fig. S20 in Supporting information). Therefore, the reaction time of the c-COF/AAO membrane and Npx with 15 min was adopted in the following experiments.

The chiral recognition performance of the chiral nanochannel membrane was investigated by the ion current response to different concentrations of S-Npx or R-Npx (Fig. 5A). The concentrations ranged from 1 × 10⁻⁴ mol/L to 1 × 10⁻¹¹ mol/L. The Ⅰ-Ⅴ curves of Npx at different concentrations after chiral recognition were shown in Fig. 5B and Fig. S21 (Supporting information). We found that upon addition of S-Npx, the Ⅰ-Ⅴ curves of the c-COF/AAO membrane changed dramatically with the increase of S-Npx concentration, while the Ⅰ-Ⅴ curves changed less with the addition of R-Npx. The ion current at +1.0 Ⅴ obtained from the Ⅰ-Ⅴ curves also gave an account of the selectivity of chiral membranes for Npx (Fig. 5C). The ion current gradually decreased with the increase of S-Npx concentrations and remained almost unchanged when R-Npx was added. Fig. 5C indicated that the current change is linear proportional to the logarithmic value of the S-Npx concentrations in the range from 10⁻¹¹ mol/L to 10⁻⁴ mol/L. The detection limits were estimated to be 3.88 pmol/L of S-Npx using a 3 SD/L method. The COF/AAO membrane was modified with d-Pen in the same way as l-Pen. The results are shown in Figs. S22 and S23 (Supporting information). We found that the ion current at +1.0 Ⅴ decreased with increasing concentrations of R-Npx, which showed that the d-Pen-COF/AAO membrane had a higher selectivity for R-Npx and almost no response to S-Npx. Hydrogen bonds can be formed between the amino groups of l-Pen and the carboxyl of Npx [43]. This difference in selectivity was mainly caused by the difference in the interaction between the chiral membrane and S-Npx/R-Npx.

For nanofluidic sensors, the reusability and stability are of critical importance for sample detection. The Ⅰ-Ⅴ curves of the present c-COF/AAO membrane were achieved by adding and removing S-Npx (Fig. 5, Fig. 5). The ion current decreased after addition of S-Npx solution and recovered to the original state after water washing. This is due to the reversibility of hydrogen bonds between the probe and the chiral molecules, and the S-Npx trapped by c-COF/AAO can be released after 1 h of immersion in water. To investigate the long-term stability, the chiral membrane was immersed in 100 mmol/L KCl for 15 days. As displayed in Fig. S24 (Supporting information), it still exhibited good ability to recognize naproxen. This device shows good repeatability and reusability, demonstrating high potential for practical applications.

To illustrate the enantioselectivity of the chiral nanochannel membrane for other enantiomers, the ability of the c-COF/AAO membrane to recognize other enantiomers (l-/d-glutamate, l-/d-histidine, l-/d-phenylalanine and l-/d-tyrosine) was further investigated under the same conditions and concentration of electrolyte (0.1 mmol/L). The Ⅰ-Ⅴ curves after chiral recognition were shown in Fig. S25 (Supporting information), and the ion current at +1.0 Ⅴ also changed differently after recognition of the glutamate, histidine, phenylalanine, and tyrosine enantiomers. Owing to the hydrogen-bond interaction, the present device also exhibits some recognition ability for other chiral molecules. But the interaction between probe and naproxen is relatively stronger. This is due to the size-matching effect play an important role in chiral recognition [44]. The Npx molecule, measuring ~13 Å, is larger than the other four chiral molecules, each under 10 Å, making naproxen more suitably sized for c-COF pores (15.8 Å). Next, the applicability of the c-COF/AAO membrane for Npx recognition in different electrolyte solutions were investigated. The Ⅰ-Ⅴ curves also changed differently of the c-COF/AAO membrane in BaCl₂ and K₂SO₄ solutions when adding 0.1 mmol/L S-Npx/R-Npx (Fig. S26 in Supporting information). With the addition of S-Npx, we found that the ion current at response +1.0 Ⅴ fell notably. This strategy has good selectivity and stable performance for the recognition of Npx, which clearly indicates that S-Npx has strong binding with the chiral nanochannel surface.

In summary, a highly selective nanofluidic chiral sensor was successfully constructed by post-modification of l-Pen on the COF/AAO membrane. The prepared c-COF/AAO membrane exhibited good chiral selectivity for Naproxen. Differences in interactions between enantiomers and chiral nanofluidic ICR sensor were responsible for the different levels of enantiomeric recognition. With the addition of enantiomers, the changes in the surface charge density of the c-COF/AAO membrane led to different ion transport behaviors, and its outstanding selective ion transport enables good chiral discrimination of S-/R-Npx enantiomers. In combination with the electrochemical detector, sensitive enantioselective recognition and detection can be achieved. The results show that the prepared c-COF/AAO ICR membrane indicates high specificity and sensitivity for S-/R-Npx with a detection limit as low as 3.88 pmol/L. This work provides a novel and simple strategy for the preparation of chiral nanofluidic ICR device which has the ability of highly selective and sensitive recognition and detection for chiral molecules.

Declaration of competing interest

I confirm that all the results in the manuscript are original and have not been plagiarized. There are no conflicts of interest in relation to this present manuscript.

CRediT authorship contribution statement

Chong Wang: Writing – original draft, Investigation, Formal analysis, Data curation. Hao Xie: Software, Methodology, Investigation. Rulan Xia: Software, Investigation. Xuewei Liao: Writing – review & editing, Supervision, Investigation, Conceptualization. Jin Wang: Visualization, Software. Huajun Yang: Formal analysis, Data curation. Chen Wang: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Acknowledgments

This work was supported by grants from the National Natural Science Foundation of China (Nos. 22274076, 22304084), the Primary Research & Development Plan of Jiangsu Province (No. BE2022793), the Natural Science Foundation of Jiangsu Province of China (No. BK20230377), and Jiangsu Provincial Department of Education (No. 211090B52303).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110642.