Selective Separation and Quantification of Tetracycline Antibiotics Using Ionic Liquid‐Functionalized MOF‐Derived Diffusive Gradients in Thin Films

ABSTRACT

A selective phase for tetracycline antibiotics (TAs) uptake was designed by incorporating [BMIM][PF₆]‐functionalized ZIF‐67‐derived carbonaceous material into the diffusive gradients in a thin‐films (DGT) device. This material, integrated in the binding gel of DGT (ta‐DGT), exhibited specifically strong affinity toward tetracycline (TC), oxytetracycline (OTC), and chlortetracycline (CTC), with adsorption capacities of 12.22, 11.81, and 13.78 µg/disc, respectively. The ta‐DGT showed preferential uptake of TAs over coexisting antibiotics and organic contaminants, with adsorption ratios above 80% and stable performance over pH 4–9 and ionic strength of 0.001–0.1 M. Mechanistic analysis indicated that π–π electron donor–acceptor interactions and cation–π bonding facilitated the specific targeting of TAs onto ta‐DGT. During 7 days of field deployment in a domestic wastewater treatment plant in Beijing, ta‐DGT showed 5–40 times higher detectable concentrations than conventional HLB/XAD DGT devices. These results validated its sensitivity, reliability, and feasibility for the selective separation and quantification of TAs in complex aqueous matrices.

1. Introduction

Global emissions of antibiotics are ∼1 × 10⁵ to ∼2 × 10⁵ tons annually [1, 2]. Considerable studies have evidenced the pervasive presence of these pharmaceuticals over diverse environmental media, especially in both drinking [3] and wastewater systems [4]. The physicochemical properties, including high polarity, complex molecular structures, and zwitterionic nature of tetracyclines (TCs), cause their weak collection from conventional hydrophobic materials and strong susceptibility to competitive adsorption in mixtures of diverse organic compounds [5]. Hence, effective separation and precise quantification of TAs in aqueous matrices remain technically challenging.

Diffusive gradients in thin‐films (DGT) have been increasingly used as a sampling and analytical tool for measuring antibiotics due to the irrelevance of the water flow rate and diffusive boundary layer [6, 7]. The separation performance of a DGT device is triggered by the selective binding agent rather than its diffusion process [8]. Typical organic DGT (o‐DGT) devices often employ universal sorbents such as XAD18 or HLB resins to capture hydrophilic organic contaminants without molecular recognition [9, 10]. Recent studies have explored molecularly imprinted polymers and carbon‐based binding phases to improve antibiotic uptake, but their insufficient selectivity toward specific antibiotic classes may compromise the specific quantification of TAs in complex wastewater matrices [10, 11, 12]. Consequently, the binding phases of universal o‐DGT are prone to competitive occupation by organic compounds, resulting in suppressed uptake and insufficient mass for further instrumental measurements. The hydrophilicity and structural similarity to other pharmaceutical chemicals also complicate their enrichment in mixtures. Therefore, it is necessary to develop a binding phase that is specific to TAs via π–π electron donor–acceptor interactions and cation–π bonding for their selective uptake over diverse interferents.

Metal–organic frameworks (MOFs) are a class of crystalline porous coordination materials that have been widely explored as feasible sorbents for the adsorption of antibiotics [13], metals [14, 15], and persistent organic pollutants (POPs) [16, 17] in aqueous environments. MOFs such as Zn/Co/Fe‐ZIF have been investigated for antibiotic adsorption [18, 19, 20]. However, the limited water stability of pristine MOFs in aqueous environments, which may lead to framework degradation and metal ion leaching, restricts their practical application in water treatment [21, 22, 23]. In addition, conventional binding phases often lack sufficient molecular selectivity, making it challenging to selectively target tetracycline antibiotics (TAs) in complex aqueous matrices. To overcome this limitation, functionalized MOFs and MOF‐derived carbonaceous materials have been increasingly developed for water decontamination because their porous structures and active sites can be regulated [24]. In particular, ionic liquid (IL) functionalization can tailor the surface polarity and interaction sites of ZIF‐based materials [25], which may enhance the selective adsorption of TAs. Therefore, in our previous research, an IL was used to improve the water stability of ZIF‐based materials [21]. Furthermore, carbonization of the IL‐modified precursor can increase pore sizes and generate more active sites, thereby enabling fast transport and long‐term accumulation in binding phases for selective separation of TAs.

Despite current advances, the application of IL‐functionalized MOF‐derived carbonaceous materials in the assembly of selective DGT has not been systematically investigated, particularly for preferential separation and quantification of TAs under complex aqueous conditions. Regarding this issue, a binding agent was synthesized by carbonization of 1‐butyl‐3‐methylimidazolium hexafluorophosphate ([BMIM][PF₆])‐modified ZIF‐67 (BP/Z‐C) in this study for the development of TAs targeting DGT (ta‐DGT). The adsorption capacity, selectivity, and separation mechanisms of ta‐DGT toward TC, oxytetracycline (OTC), and chlortetracycline (CTC) were evaluated to ensure its performance. Selective measurements were then tested under different pH values, ionic strengths (ISs), temperatures, and deployment times. Field experiments were finally carried out in a WWTP located in Beijing to validate its effectiveness as a selective separation and quantification strategy for TAs in complex water environments.

2. Materials and Methods

2.1. Materials and Reagents

Unless otherwise stated, all chemicals in this study were analytical grade. Materials for the synthesis of selective adsorbents, including cobalt nitrate hexahydrate (Co(NO₃)₂·6H₂O), sodium chloride (NaCl), sodium hydroxide (NaOH), 37 wt% hydrochloric acid (HCl), 2‐methylimidazole and [BMIM][PF₆], were purchased from Macklin Biochemical Co. Ltd. Methanol was collected from Beijing Merida Technology Co. Ltd. Target standards TC, OTC, CTC, erythromycin, sulfamethoxazole, ofloxacin, and atrazine were obtained from Aladdin Biochemical Technology Co. Ltd. Acetone, methanol, and formic acid of HPLC grade were purchased from Beijing Yili Fine Chemicals Co. Ltd.

2.2. Synthesis and Characterization of the ta‐DGT Binding Agent (BP/Z‐C)

The BP/Z‐C material was synthesized by the carbonization of ZIF‐67 functionalized with [BMIM][PF₆]. Consequently, the preparation of ZIF‐67 was the key to producing BP/Co‐NPC. Detailed synthesis procedures are shown below.

2.2.1. Preparation of ZIF‐67

Briefly, 1.16 g Co(NO₃)₂·6H₂O dissolved in 30 mL methanol was ultrasonicated for 60 s before being mixed with 2.32 g 2‐methylimidazole in 30 mL methanol solution. After 3 min of 40 kHz ultra‐sonication, the mixing solution was stirred at 300 rpm and 30°C for at least 20 h until the formation of a purple suspension, followed by washing three times with methanol (30 mL each time). ZIF‐67 was finally collected by drying the purple cloudy solution at 60°C overnight. The synthesis yield of ZIF‐67 was 20.45%, calculated based on the mass of dried ZIF‐67 to the theoretical yield from the complete conversion of Co in Co(NO₃)₂·6H₂O.

2.2.2. Preparation and Carbonization of [BMIM][PF₆]/ZIF‐67

The wet impregnation protocol was applied for the synthesis of [BMIM][PF₆]/ZIF‐67 (BP/Z). ZIF‐67 was activated by drying at 105°C overnight to remove water and impurities. Different amounts of [BMIM][PF₆] (0.15 and 0.3 g) were dissolved in 30 mL of acetone and magnetically stirred for 1 h until a homogeneous solution was obtained. Then, 0.70 g of ZIF‐67 was added with 60 s water bath sonication to achieve the uniform dispersion of ZIF‐67. Subsequently, the suspension was added to a 50 mL round‐bottom flask with gentle evaporation at 30°C under continuous stirring (200 rpm) until acetone was fully removed. The purple BP _x /Z (x is the mass of [BMIM][PF₆] used for binding agent synthesis) composite powder was then collected by drying at 105°C overnight.

For the development of carbonized [BMIM][PF₆]/ZIF‐67 (BP _x /Z‐C), BP _x /Z was carbonized at 800°C for 5 h with a pure constant gas flow of N₂ and a heating rate of 10°C/min. The successfully carbonized BP _x /Z‐C powders were then dried at 60°C for subsequent fabrication of ta‐DGT binding gels and samplers.

2.2.3. Characterization

Scanning electron microscopy (SEM; ZEISS, GeminiSEM300, Germany) and transmission electron microscopy (TEM; FEI Talos F200X‐G2, Czech) analyses were performed to observe the morphological differences among ZIF‐67, BP_0.15/Z, and its carbonization derivative BP_0.15/Z‐C. A powder x‐ray diffractometer (XRD; Smartlab3) from Rigaku, Japan, equipped with a Cu sealed tube at 10°–60°, was employed to measure the crystalline structure of all materials. The infrared spectra of ZIF‐67, BP _x /Z and BP _x /Z‐C were obtained in wavenumbers ranging from 400 to 4000 cm⁻¹ using a Fourier transform infrared (FTIR) spectrometer (IRPrestige‐21, Japan). Raman microspectroscopy (Renishaw, UK) with an excitation wavelength of 532 nm was utilized to collect relevant Raman spectra ranging from 1000 to 2100 cm⁻¹.

2.3. Preparation and Performance of ta‐DGT Binding Gel

2.3.1. Preparation and Optimization of the Binding Gel

Based on the optimization results described in Section 3.2, BP_0.15/Z‐C₁₀ was selected as the optimized binding gel for ta‐DGT preparation. Briefly, 10 mg of BP_0.15/Z‐C and 1.5 g of agarose powder were dispersed in 100 mL of MQ water and microwaved until the agarose was completely dissolved and BP_0.15/Z‐C was uniformly dispersed. The mixture was stirred for 1 min and then pipetted into the middle of two preheated glass plates divided by a 0.8 mm polytetrafluoroethylene (PTFE) spacer. The gel sheet, cooled to room temperature for 15 min, was cut into several gel discs 2.2 cm in diameter and 0.8 mm in thickness and hydrated in MQ water with 0.1 M NaNO₃.

TC, OTC, and CTC were selected as representative TAs to test the performance of each type of ta‐DGT binding gel. All experiments were carried out in triplicate unless otherwise stated. Five hundred milligrams of each antibiotic was dissolved in 1 L of MQ water to prepare stock solutions and serially diluted to 1–500 mg/L for further use. Triplicate gel discs prepared from each amount of BP _x /Z‐C _y were separately treated with 2 mg/L solutions of TC, OTC, and CTC to test their performances. From the results of accumulation capacity and elution efficiency (see Section 3.2), BP_0.15/Z‐C with a mass of 10 mg was evidenced as the optimized material for the preparation of ta‐DGT binding gel and device. Hence, except for the experiments of capacity measurements and elution validation, the following treatments were carried out with only BP_0.15/Z‐C₁₀.

2.3.2. Selectivity for TAs

Selective measurements of TAs onto ta‐DGT were investigated by applying triplicate binding gels in 10 mL of MQ solution with 2 mg/L TC, OTC, CTC, erythromycin, sulfamethoxazole, ofloxacin, and atrazine for 12 h. The 2 mL solution samples were collected pre‐ and post‐exposure to target chemicals for duplicate UV/Vis spectrometer (TU‐1900, Persee, Beijing) detection to test the selectivity of the BP_0.15/Z‐C₁₀ binding gel toward TAs.

2.3.3. Adsorption Capacity and Kinetics of ta‐DGT Binding Gel

Adsorption capacity evaluation is required for long‐term deployment of DGT devices or in highly polluted environments. The adsorption capacity of ta‐DGT for TC, OTC, and CTC was measured by performing triplicate BP _x /Z‐C _y binding gels in 10 mL solution with each TA at concentrations from 0.1 to 2 mg/L in conical glass flasks for 24 h with shaking at 200 rpm. Samples were collected at 0, 0.5, 2.5, 4, 6, 8, 10, 12, and 24 h for duplicate UV/Vis spectrometer detection for adsorption kinetics measurements. As the adsorption equilibrium was essentially reached within 12 h, samples collected at 12 h were applied for the test of adsorption capacity.

2.3.4. Effect of Temperature, pH, and IS

For the further application of ta‐DGT in real environments, it is necessary to test its performance under diverse temperature, pH, and IS conditions. All treatments in this section were carried out separately for TC, OTC, and CTC by mixing triplicate BP_0.15/Z‐C₁₀ binding gels with 10 mL of a 2 mg/L single‐analyte solution for 12 h at 200 rpm shaking under relevant conditions, followed by UV/Vis spectrometer detection. For temperature experiments, each treatment was performed on a temperature‐controlled stirrer at 25°C, 30°C, and 35°C. To investigate the influences of different pH conditions, each solution was adjusted by using 1 M HCl or NaOH to achieve a final pH ranging from 4 to 9. Experiments of IS were performed by mixing the solution with NaCl concentrations of 1 ∼100 mM.

2.3.5. Elution Efficiencies

Accurate elution efficiency is important for the measurement of the DGT concentration of TAs in real environments. Triplicate BP_0.15/Z‐C₁₀ gels were separately treated with 10 mL 0.5 mg/L TC, OTC, and CTC for 12 h. After adsorption, the gels were eluted with 10 mL of FA‐MeOH (5:5, v/v), which was selected as the optimum desorption solvent. Elution was performed for 30 min at 40 kHz in an ultrasonic bath.

2.4. ta‐DGT Assembly and Application

2.4.1. Diffusion Coefficient Determination

The diffusion coefficients of TC, OTC, and CTC in the agarose diffusive gel have not been previously reported, so they were measured using a previously reported assay with required modifications [26]. Normally, ta‐DGT devices were exposed to 5 L of 0.1 mg/L antibiotics, and triplicate samples were collected at 0.5, 6, 8, 10, 12, 36, and 48 h. For each treatment, 1 mL of solution was sampled at the same collection time as ta‐DGT deployment to confirm that the antibiotic concentration remained stable. The diffusion coefficient (D, cm²/s) was derived from the accumulated mass (ng) of antibiotics in ta‐DGT devices plotted as a function of time (s), which is shown as the equation below [27]:

$$D=\frac{\alpha \times \mathrm{\Delta }g}{A\times C_{\mathrm{sol}}}$$ (1)

where α is the slope for the linear regression of the antibiotic mass over deployment time, $\mathrm{\Delta }g$ (cm) is the thickness of the diffusive layer, A (cm²) is the area of the sampling window of the diffusion layer, and C _sol (ng/mL) is the antibiotic concentration in the solution. The D values at a given temperature were standardized into the diffusion coefficient of 25°C based on the equation in the study of Chen et al. [28].

2.4.2. ta‐DGT Assembly and Sampling Dynamics

A typical ta‐DGT device often involves plastic molding of a base and a cap as the protective facility, a 0.45‐µm polyether sulfone (PES) filter membrane of 0.14‐mm thickness, an agarose diffusive layer of 0.8‐mm thickness, and a 1.6‐mm BP_0.15/Z‐C₁₀ binding layer.

Sampling dynamics are an essential indicator to set up and correct the deployment time for practical application of DGT samplers, especially for water environments with significant variability in antibiotic concentrations. The uptake of TC, OTC, and CTC by the ta‐DGT device was measured by exposing the sampling devices to 5 L of solution containing 500 µg of three target antibiotic solutions for 48 h with constant stirring and sampling at 0, 12, 24, 36, and 48 h for FA‐MeOH (5:5 volume ratio) elution.

2.4.3. WWTP Wastewater Deployment and Retrieval

To evaluate the performance of ta‐DGT in field environments, a WWTP for domestic sewage in Chaoyang District, Beijing, was selected as the sampling site (Figure S1). Triplicate ta‐DGT devices were deployed in the influent and effluent of the WWTP for 7 days to measure TA uptake. Conventional HLB o‐DGTs containing 0.8‐mm agarose diffusive gels were deployed simultaneously with ta‐DGTs to estimate their practical performance. After deployment, both HLB and ta‐DGT were disassembled to transfer the binding gel into a 15 mL prebaked and washed amber glass vial. The internal standard ¹³C₈‐ofloxacin (100 ng) was spiked, followed by the addition of 10 mL 5:5 FA‐MeOH solution for 30 min in an ultrasonic bath. The extract was filtered with 0.22 µm to dryness under a gentle flow of N₂, redissolved in 10 mL of MQ water, and stored at −18°C.

In addition, grab samples were collected on the first and final days of deployment with 300 mL PP bottles at approximately 13:00. The water temperature was 24.6°C, and the pH was 7.21 for influents and 6.97 for effluents. The grab sampling water samples were transferred to the lab as soon as possible for 0.22‐µm filtration. TAs in the filtered water samples were measured using solid‐phase extraction (SPE) of HLB cartridges (200 mg, 6 mL, Sigma‐Aldrich, UK) with ¹³C₈‐ofloxacin (100 ng) as the surrogate standard [29].

2.5. Instrumental Analysis

2.5.1. UV/Vis Spectrometer Measurements

For all binding gel performance experiments, the concentrations of TC, OTC, and CTC were measured at 350, 276.25, and 275 nm, respectively, by a UV‒vis spectrophotometer (TU‐1900, Persee, Beijing). In this study, UV‒vis measurements were only applied in single‐analyte systems for binding gel optimization and adsorption experiments, where no spectral interference occurred. For each detection, the sample was filtered via a 0.45‐µm membrane. The accumulation q _t (µg/disc) of each TA is defined below:

$$q_{\mathrm{t}}=\frac{\left(c_0-c_{\mathrm{t}}\right)\times \left(V_{\mathrm{sol}}+V_{\mathrm{disc}}\right)}{n}$$ (2)

where $c_0$ (µg/mL) is the antibiotic concentration at the starting time, $c_{\mathrm{t}}$ (µg/mL) is the antibiotic concentration at time t, $V_{\mathrm{sol}}$ (mL) is the volume of antibiotic solution, $V_{\mathrm{disc}}$ (mL) is the volume of binding gel, and n is the number of gel discs in the treatment solution.

2.5.2. High‐Performance Liquid Chromatography‒Mass Spectrometry Measurements

For all DGT device performance and application experiments, the concentrations of TC, OTC, and CTC were measured by high‐performance liquid chromatography‒mass spectrometry (HPLC–MS, Agilent 1290 Infinity/6460 LC/QQQ MS). This method was selected to ensure accurate separation and quantification of target antibiotics in multianalyte systems and environmental samples. The target TAs from the extracts of samples were analyzed by HPLC–MS following the procedures in Table S2 [30]. Blanks (MQ water) and ta‐DGT blanks were tested to assess the limit of quantification (LOD) by extrapolating the measured concentration to S/N values of 10, as provided in Table S3. The concentrations of TC, OTC, or CTC in the water (C _DGT) were calculated using the equation below:

$$C_{\mathrm{DGT}}=\frac{M\mathrm{\Delta }g}{DAt}$$ (3)

where M is the mass of TAs accumulated on the binding gel, $\mathrm{\Delta }g$ is the thickness of the agarose diffusion layer and filter membrane, D is the diffusion coefficient of TC, OTC, or CTC through the diffusive gel, A is the sampling area exposed to the target analyte of the DGT device, and t is the deployment time of the DGT device.

The mass of TAs accumulated in the binding gel (M) was calculated from the analyte concentration in the extract solution according to:

$$M=\frac{C_{\mathrm{e}}\left(V_{\mathrm{e}}+V_{\mathrm{g}}\right)}{f_{\mathrm{e}}}$$ (4)

where C _e is the analyte concentration in the extract solution, V _e is the volume of the extract solution, V _g is the volume of the binding gel, and f _e is the extraction efficiency of the analyte from the binding gel.

3. Results and Discussion

3.1. Characterization of the Binding Material BP/Z‐C

To investigate the synthesis and properties of [BMIM][PF₆]‐functionalized ZIF‐67 and carbonized BP/ZIF‐67, several characterization techniques, including SEM, TEM, XRD, FTIR, and Raman spectroscopy, were employed. SEM measurements confirmed the typical dodecahedral morphology of raw ZIF‐67 in Figure 1A. The similar morphology in Figure 1A,B showed that [BMIM][PF₆] has no significant effect on the material structure of ZIF‐67 after synthesis. This is also confirmed by the XRD spectra with intensive peaks at 2θ degrees of 7.22°, 10.32°, 12.64°, 14.60°, 16.38°, 17.94°, 22.04°, 24.40°, and 26.58° for both ZIF‐67 and BP_0.15/Z. These results are consistent with the simulated XRD standard pattern of ZIF‐67 (CCDC No. 671073, Cambridge Crystallographic Data Centre), showing that the crystal structure remained unchanged pre‐/post‐incorporation with [BMIM][PF₆] (Figure 2A). The XRD spectra of BP_0.15/Z‐C had no featured peaks of ZIF‐67 but showed peaks at 44.2° and 51.9° that are consistent with the XRD standard pattern of metallic Co (PDF No. 04‐003‐6596, ICDD) and are assigned to the crystal lattice planes of Co (111) and Co (200) [31, 32].

FIGURE 1.

SEM images of ZIF‐67 (A), BP_0.15/Z (B), BP_0.15/Z‐C (D); SEM element mapping of BP_0.15/Z (C), BP_0.15/Z‐C (E); TEM images of BP_0.15/Z‐C (F); TEM element mapping of BP_0.15/Z (G).

FIGURE 2.

XRD patterns of ZIF‐67, BP_0.15/Z, and BP_0.15/Z‐C (A); FTIR spectra of ZIF‐67, BP_0.15/Z, and BP_0.15/Z‐C (B); Raman spectra of BP_0.15/Z‐C (C); N₂ adsorption–desorption isotherms of ZIF‐67, BP_0.15/Z, and BP_0.15/Z‐C (D).

From the chemical formula of [BMIM][PF₆] and ZIF‐67, despite the common elements C and N for both of these materials, Co was specific for ZIF‐67, while F and P were exclusive for [BMIM][PF₆]. Hence, from the images of Figure 1C,G, the uniform co‐distribution of F and P provided by the IL, along with Co from ZIF, confirmed the successful functionalization of [BMIM][PF₆] onto ZIF‐67. After calcination at 800°C, the rough surface and collapsed crystal structure of BP_0.15/Z‐C are shown in Figure 1D, and its elements, including Co, F, P, N, and C, remained uniformly distributed throughout the particles after carbonization (Figure 1E). It is worth noting that, from the results of TEM in Figure 1F, carbon nanomaterials generated from BP_0.15/Z‐C post‐calcination at 800°C can increase the amount of activated Co in the binding agent to be the key potential player for capturing more TC molecules.

FTIR spectral measurements of ZIF‐67, BP_0.15/Z and BP_0.15/Z‐C provided information about chemical bonds and molecular interactions (Figure 2B). For pure ZIF‐67, the spectral peak at 434.5 cm⁻¹ was assigned to the Co─N bond, while peaks at 1139.8, 1407.8, and 2926.4 cm⁻¹ presented C═C, C─N, and C─H, respectively [28]. With the incorporation of [BMIM][PF₆], identical spectral structures were found for both ZIF‐67 and BP_0.15/Z, generating exclusive peaks at 557 and 852.9 cm⁻¹ from the [PF₆]⁻ anion. Calcination at 800°C forced the shifts of Co─N, C═C, and C─N peaks toward higher frequencies [21], suggesting strengthened Co─C/N interactions within the carbon matrix. In addition, Co facilitated the graphitization of carbon, which is confirmed by the Raman spectral results in Figure 2C. The carbon structure of BP_0.15/Z‐C₁₀ showed one peak at ∼1355 cm⁻¹ for the D band in carbon that was assigned to the degree of material disorder and defects, and the other peak at ∼1590 cm⁻¹ for the G band that represented the degree of graphitization of the carbon structure. The I _D/I _G ratio of 0.93 for BP_0.15/Z‐C indicated stronger graphitization of this carbon material [33]. The pore structure and surface properties of ZIF‐67, BP_0.15/Z, and BP_0.15/Z‐C were further investigated by N₂ adsorption–desorption measurements (Figure 2D). According to the IUPAC classification, all exhibit typical Type I isotherms, indicating a predominantly microporous structure. A sharp increase in N₂ uptake at low relative pressure (P/P ₀ < 0.1) confirms the presence of abundant micropores. In the intermediate pressure range (P/P ₀ = 0.1–0.8), the relatively flat adsorption curve suggests that adsorption mainly occurs on the external surface. For the carbonized composite BP_0.15/Z‐C, the additional adsorption uptake at high relative pressure (P/P ₀ ≈ 1.0), along with the presence of a distinct hysteresis loop, indicates the formation of mesopores after carbonization.

3.2. Adsorption Kinetics and Elution Efficiency of Binding Gel

To optimize the weight ratio of [BMIM][PF₆] and ZIF‐67 during the synthesis of BP _x /Z‐C _y , different amounts of BP _x /Z‐C (0, 2, 5, 10, 15, and 20 mg) were incorporated into the binding gel, and their performances were evaluated using adsorption capacity toward TC, OTC, and CTC. From the results of Figure 3A, the highest adsorption capacity of the three tested TAs was found for the binding material with [BMIM][PF₆]:ZIF‐67 at a weight ratio of 1.5:7 (BP_0.15/Z). Slight mass decreases were found in accumulated antibiotics on the BP_0.3/Z binding gel, resulting from the reduced surface area [34]. Therefore, although [BMIM][PF₆] at a certain weight ratio can improve the number of active sites for Co to adsorb antibiotics, its increasing addition possibly limits the surface area for further improvement. In addition, 10 mg of BP_0.15/Z‐C was mixed with 10 mL gel solution containing 0.15 g agarose to achieve the maximum adsorption capacity toward TCs (Figure 3B). The corresponding uptake values were 48.10% for TC, 50.85% for OTC, and 57.85% for CTC. Therefore, the optimized weight ratio of [BMIM][PF₆] to ZIF‐67 was 0.15:0.7, BP_0.15/Z‐C, to prepare 10 mg binding agents in 10 mL 1.5% (w:v) agarose gel solution (BP_0.15/Z‐C₁₀) for the assembly and application of ta‐DGT in laboratory conditions or real environments. The sorption dynamics of TAs on ta‐DGT were tested by using a BP_0.15/Z‐C₁₀ binding gel in 2 mg/L TC, OTC, and CTC solutions. After exposure to three different TA solutions, their accumulation initially increased to ∼70% of their maximum adsorption linearly within 2 h (Figure S2). From the time‐dependent analysis, the adsorption capacity was perfectly fitted in pseudo‐second‐order with R ² > 0.99 (Table S1), indicating that the binding mechanisms were mainly driven by chemical adsorption. After 12 h of exposure, the accumulated mass remained similar to that of the 24 h measurements, showing the adsorption capacity of 12.22 µg for TC, 11.81 µg for OTC, and 13.78 µg for CTC per gel disc. This proved that the adsorption capacity of BP_0.15/Z‐C₁₀ binding gels for TAs can prevent the risk of insufficient uptake in real environments, contributing to the comprehensive investigation of their behaviors.

FIGURE 3.

Adsorption capacity of BP/Z‐C gel with different weight ratios between [BMIM][PF₆] and ZIF‐67 (A), with different amounts in 10 mL gel solution toward TC, OTC, and CTC (B); elution efficiencies of TC, OTC, and CTC from BP_0.15/Z‐C₁₀ binding gel by using FA‐MeOH of different volume ratios (C).

Suitable desorption solvents and methods are important to acquire stable elution efficiencies of TAs from binding gels for the precise calculation of their concentrations in the environment. In previous studies, methanol or methanol–formic acid (0.1%) was used to extract sulfonamides, fluoroquinolones, and macrolides from HLB or XAD18 binding gels [12, 35], but not for TCs. The NaOH of 0.1 M pure solution or of 0.5 M mixture with oxalic acid and methanol showed 44%–60% or 68%–94% elution efficiencies for TC removal from graphene or porous carbon material binding gels [36, 37]. It is obvious that using extraction solvents containing a certain amount of NaOH cannot obtain fixed elution efficiencies due to the strong impacts of OH⁻ on both antibiotics and gel structures [38]. The organic solvent, acetonitrile, prepared with 0.025 M H₃PO₄, was then employed to elute TAs, but no elution efficiencies were reported in a relevant study [39].

Consequently, to avoid uneven damage from inorganic bases or acids on the chemical properties of target antibiotics, we selected acetonitrile and a mixture of methanol with formic acid (FA‐MeOH) as two extraction solvents to elute antibiotics from the BP_0.15/Z‐C₁₀ binding gel. Based on previous studies of conventional o‐DGT sampling antibiotics [39], acetonitrile/water mixtures containing 3%, 10%, and 50% acetonitrile (v/v), as well as 100% acetonitrile, were tested as eluents for the ta‐DGT binding gel. Because the binding agent developed in this study was a carbonaceous material that is distinct from the resinous materials (e.g., HLB and XAD) of o‐DGT, FA‐MeOH mixed solvents with different volume ratios (FA:MeOH = 1:9, 4:6, 5:5, 6:4, and 9:1) were also employed to extract antibiotics from the binding gels. From the results in Figure S3, 10 mL of 3%, 10%, 50%, and 100% (v:v) acetonitrile solutions showed weak removal efficiencies, all lower than 15%. To improve the recovery of antibiotics from the binding layer, FA‐MeOH mixtures with varying ratios were tested as elution solvents. The highest elution efficiencies of ∼63.17% for CTC, ∼37.33% for OTC, and ∼54.15% for TC were found by using the FA‐MeOH solvent with a 5:5 volume ratio (Figure 3C). As increasing or decreasing the concentration of formic acid did not further improve the recovery, 10 mL of 1:1 FA‐MeOH was used to elute the three TAs from the BP_0.15/Z‐C₁₀ binding layers.

3.3. Selective Separation Mechanisms of TAs in ta‐DGT

Achieving preferential uptake of TAs by the ta‐DGT sampler is essential to guarantee its practical applications, especially in environments contaminated with diverse organic pollutants. Different types of antibiotics (e.g., macrolides, sulfonamides, quinolones) and hydrophilic POPs are potential competitors affecting the uptake of TAs in DGT samplers. Thus, representative substances, such as 2 mg/L, were mixed with TC, OTC, and CTC in 10 mL of solution to test the adsorption preference of ta‐DGT. As shown in Figure 4A, after exposure to the mixture of typical organic compounds for 12 h, the selectivity of each binding gel disc from ta‐DGT toward TC, OTC, and CTC was 11.06, 9.63, and 13.22 µg, respectively, leading to 5–10 times higher adsorption efficiencies compared to erythromycin, sulfamethoxazole, ofloxacin, and atrazine. In Figure 4B, the adsorption ratios of the TAs (TC, OTC, and CTC) were all above 80%, whereas that of ofloxacin was only ∼20% and < 10% for other organic compounds, demonstrating the preferential adsorption of ta‐DGT toward TAs. Therefore, ta‐DGT can effectively measure TAs in aquatic environments with multiple types of organic pollutants, thereby improving the measurement accuracy of the DGT device for the detection of the bioavailable concentration of antibiotics.

FIGURE 4.

Selective adsorption capacity (A) and adsorption ratio (B) of BP_0.15/Z‐C₁₀ binding gel for the mixture of different organic pollutants.

The adsorption of organic compounds onto carbon‐based materials is normally driven by multiple interactions, including surface complexation, van der Waals forces, π–π donor–acceptor effects, and cation–π bonding. Surface complexation tends to occur between zwitterionic TAs molecules and adsorbents with abundant ionic/polar groups on the surface. However, erythromycin, sulfamethoxazole, and atrazine, with higher contents of polar sites than the three TAs, had significantly lower adsorption affinities, indicating that surface complexation was not the primary mechanism for preferential TAs adsorption on BP_0.15/Z‐C₁₀. Increasing van der Waals forces were caused by efficient contact surface area, but the results of Table S4 and Figure S5 show that the smallest surface area of BP_0.15/Z‐C₁₀ contributed to the highest adsorption capacity compared to ZIF‐67 and BP/ZIF‐67. Therefore, van der Waals forces were not the main driver for the adsorption of TAs. In addition, compared to the binding agent specific to quinolone antibiotics developed in our previous study, the large amount of F⁻ in BP_0.15/Z‐C₁₀ repelled the fluorine in ofloxacin, resulting in its low accumulation (Figure 4) [12]. The strong electron‐withdrawing from the ketone group of the TAs molecule enabled the enone structure as a π–electron acceptor to interact with the graphene surface of BP_0.15/Z‐C₁₀ via π–π EDA interactions. Moreover, the Co of BP_0.15/Z‐C₁₀ and the protonated amino group on the carbon ring of TAs can attach to ketone and graphene π‐electrons, respectively, suggesting that π–π EDA interactions and cation–π bonding were two effective mechanisms for the preferential accumulation of TAs.

3.4. Effects of Environmental Conditions

The implications of pH, IS, and temperature on ta‐DGT performance are illustrated in Figure 5. In the pH range from 4 to 9, the accumulation of 3 TAs on the BP_0.15/Z‐C₁₀ gel remained relatively stable and higher than ∼9 µg/disc, corresponding to an uptake of higher than ∼45%, indicating stable performance at different aquatic acidities (Figure 5A). The increasing adsorption of these three antibiotics found at pH values of 4 and 5 was probably due to metal release from the adsorbent materials. Hence, the H⁺ replacement of Co²⁺ in BP_0.15/Z‐C₁₀ can provide hydrogen bonding and electrostatic attraction toward antibiotics to slightly enhance the adsorption of ta‐DGT binding gel in low pH environments. With changing ISs from 0.001 to 0.1 M, the uptake of three TAs on the ta‐DGT binding gel exhibited relatively insignificant variations, with 9.62–11.75 µg/disc for TC, 9.67–10.37 µg/disc for OTC, and 9.51–11.89 µg/disc for CTC, corresponding uptake percentages of 48.10%–58.75%, 48.35%–51.85%, and 47.55%–59.45%, respectively (Figure 5B). The minor increase in the accumulated TC mass at 0.1 M IS was possibly attributed to its changing diffusion coefficient through agarose gel [40]. In addition, the high IS can influence the hydrophobicity of antibiotics in aqueous environments to further disturb their accumulation in DGT [41].

FIGURE 5.

Effect of pH (A), ionic strength (B), and temperature (C) on TAs adsorption on the BP_0.15/Z‐C₁₀ binding gel.

Due to the requirement for long‐term deployment in water, the changing temperature can influence the collection of antibiotics on the DGT binding gel. In spite of the well‐studied equation derived for the diffusion coefficient within a certain temperature range, the effects of this environmental factor on the adsorption of new binding agents toward antibiotics should be taken seriously into consideration for future practical applications. Therefore, the measurement of TAs on ta‐DGT binding gel at temperatures of 25°C, 30°C, and 35°C indicated their feasible performance in real environments (Figure 5C). The adsorption of TAs on porous carbon is reported as an endothermic process, suggesting that their increasing mass accumulated on the binding agent with growing temperature [42, 43].

3.5. Diffusion Coefficient

Antibiotic accumulation in ta‐DGT increased linearly with deployment time (Figure 6). By using Equation (1), the D values corrected at 25°C are 1.34 × 10⁻⁶, 1.23 × 10⁻⁶, and 1.04 × 10⁻⁶ for TC, OTC, and CTC, respectively (Table 1). Given the certain thickness and area of the diffusive layer in the DGT device, the diffusion of antibiotics through the agarose gel was proven to be strictly related to the surrounding temperature and molecular weight [30]. Normally, at the same temperature, compounds with larger molecular weights tend to have lower D values and vice versa. Therefore, among the three typical TAs tested in this study, the highest diffusion coefficient was found for TC, with a molecular weight of 444.44 g/mol, whereas the lowest diffusion coefficient was measured for CTC, with a value of 478.88 g/mol. The diffusion coefficients of three target TAs acquired from our study were in the consistent range measured by You et al. [36].

FIGURE 6.

Adsorption kinetics of chlortetracycline (A), oxytetracycline (B), and tetracycline (C) in ta‐DGT.

TABLE 1.

Diffusion coefficient (D) values of TC, OTC, and CTC in agarose diffusion gel.

Tetracycline antibiotics	D (× 10−6 cm2/s)	R 2	Molecular Weight (g/mol)
CTC	1.04576 ± 0.12	0.95	478.88
OTC	1.22994 ± 0.06	0.99	460.43
TC	1.33588 ± 0.15	0.94	444.44

3.6. Performances of ta‐DGT in WWTP

The deployment kinetics of ta‐DGT in fresh water with three TAs at a concentration of 0.5 mg/L under laboratory conditions showed a linear relationship within 36 h, and plateauing of plots demonstrated that the stable deployment time should be longer than 2 days (Figure S4). During the plateau period of deployment, the masses of TC, OTC, and CTC collected by ta‐DGT at 2.06, 2.18, and 4.73 µg, respectively, were attributed to their different adsorption capacities on BP_0.15/Z‐C₁₀ binding gel, as shown in Figure 4A. Both of these results showed identical accumulation patterns with the order of CTC > OTC > TC, confirming the strong competitive adsorption of CTC among the three target TAs in ta‐DGT.

To compare the performance of ta‐DGT with conventional o‐DGT (HLB and XAD‐18), the 7‐day deployment of these three devices was applied in both the influent and effluent of a WWTP in Beijing to acquire the time‐averaged concentrations of TC, OTC, and CTC. Available binding sites of o‐DGT devices are likely to be occupied by organic pollutants with high concentrations (e.g., pesticides, herbicides, organic dyes, polychlorinated benzenes), thereby eliminating the accumulation of antibiotics with weak diffusion efficiencies and binding affinity on either HLB or XAD binding gels. Consequently, measurements of TAs by highly selective DGT were of great importance to investigate their bioavailability and potential ecological risks.

The 7‐day C _DGT of these three target antibiotics in ta‐DGT developed in our study were 4.93 and 2.40 µg/L CTC, 4.40 and 4.11 µg/L OTC, and 15.16 and 5.98 µg/L TC in the influent or effluent of the selected WWTP, showing 5–40 times higher than those from HLB or XAD‐18 DGT (Figure 7). The distinct selectivity for the three target TAs enabled their detectable amounts after elution from the ta‐DGT binding gel, while their accumulated mass on the HLB or XAD‐18 binding gel was hardly measurable by the LC‒MS instrument. From the C _DGT results obtained by ta‐DGT, the concentrations of TC and CTC were significantly reduced with ∼50% removal efficiency from the influent to effluent in a WWTP, but the decrease in OTC was weak.

FIGURE 7.

Comparison of average concentrations of TC, OTC, and CTC by ta‐DGT with conventional o‐DGT (HLB and XAD) for 7 days in the influent and effluent of WWTP in August.

3.7. ta‐DGT vs. Grab Sampling

The daily concentrations of TC, OTC, and CTC in the influent or effluent from grab sampling were collected on the first and final day of the week. The concentrations of the three target TAs collected from the grab sampling method were in the range of 15.52–43.83 µg/L in the influent and 12.12–29.73 µg/L in the effluent (Figure 8). Since no special equipment or complicated collection work is needed, grab sampling is still the most commonly used technique. However, significant differences in the concentrations of the three antibiotics between the two sampling days for each sampling site can be attributed to the fluctuating discharge of wastewater containing TC, OTC, and CTC.

FIGURE 8.

Comparison of average concentrations of TC, OTC, and CTC collected by ta‐DGT and grab sampling for 7 days in the influent and effluent of WWTP in June.

The ta‐DGT enabled the on‐site and continuous uptake of TAs from WWTP, allowing the accurate calculation of their time‐average concentrations. As an in situ tool, ta‐DGT does not affect water properties (e.g., pH, IS, and flow rate), so the sampling process remains under natural conditions. Differences in antibiotic concentrations measured by the two methods are possibly caused by the storage and transportation of wastewater samples before analysis for grab sampling. Moreover, using SPE to concentrate the amount of TAs after grab sampling can change their available fractions. Moreover, antibiotics in wastewater can be adsorbed onto sludge particles mixed with dissolved organic matter and metals that impact the bioavailability, adsorption, and desorption of antibiotics, thereby affecting their concentrations in the effluent.

4. Conclusion

This study reports an antibiotic‐selective DGT originating from an IL‐modified MOF. Herein, the IL [BMIM][PF₆] of 0.15 or 0.3 g was incorporated into 0.7 g MOF ZIF‐67 for carbonization at 800°C to produce carbonaceous materials (BP_0.15/Z‐C, BP_0.3/Z‐C) for the selective accumulation of TAs. This design establishes a direct link between functional material modification and DGT application, as the binding phase was engineered to meet the requirements of rapid uptake, stability, and selective affinity toward target antibiotics. SEM and TEM images confirmed the even and complete incorporation of [BMIM][PF₆] into ZIF‐67, and carbonization enhanced the binding capabilities via increased Co content and graphitization, as evident in the Raman spectroscopy and FTIR spectra. With an adsorption capacity higher than 11 µg per gel for TC, OTC, and CTC, the carbonaceous material produced from 0.15 g IL was verified as the optimized ratio for the synthesis of the binding agent for ta‐DGT. Selectivity tests demonstrated the strong preference of the binding gel for TAs over other organic pollutants, such as pesticides, azo dyes, quinolones, and macrolides, allowing accurate measurements in complex environments. This selective recognition was mainly attributed to the specific interactions introduced by IL functionalization and carbonization, including π–π electron donor–acceptor interactions and cation–π bonding. Under the conditions of varied pH, IS, and temperature, the reliable performance of ta‐DGT enables its further suitability for practical applications. For 1 week of deployment in a WWTP, ta‐DGT outperformed conventional o‐DGT devices by specific uptake of TC, OTC, and CTC in both the influent and effluent of the WWTP. Unlike grab sampling, ta‐DGT can achieve constant accumulation of antibiotics in complex environmental fluctuations to provide stable time‐averaged antibiotic concentrations. These findings provide a feasible strategy for integrating IL‐functionalized MOF‐derived carbonaceous materials with DGT devices, offering a promising tool for the selective monitoring of emerging contaminants in complex aqueous matrices.

Author Contributions

Jie Wang: methodology, investigation, resources. Yaming Li: resources, investigation. Hanbing Li: conceptualization, methodology, formal analysis, validation, writing – original draft, writing – review and editing, funding acquisition. Yuyang Li: resources, writing – original draft. Sumei Li: writing – review and editing. Lei Yan: supervision, investigation. Yixuan Liang: resources. Dayi Zhang: writing – review and editing. Xiaowei Wang: validation, formal analysis, writing – review and editing. Sha Chen: conceptualization, supervision, writing – review and editing. Danyang Li: resources.

Funding

This study was supported by the General Project of Science and Technology Plan of Beijing Municipal Education Commission (No. KM202310005018) and the National Natural Science Fund of China for Young Scholars (No. 52200223).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: jssc70470‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data supporting this study's findings are available from the corresponding author upon reasonable request.