Adsorption Capacity of Tetracycline in Solution by Cu-BTC@Carboxyl-Functionalized Carbon Nanotubes@Copper Alginate Composite Aerogel Beads

Abstract

In order to remove tetracycline (TC) from sewage more effectively, the adsorption performance of TC on alginate composite aerogel beads containing carbon nanomaterials was studied systematically. Carboxylated functionalized carbon nanotubes (F-CNTs)@Cu-based metal-organic framework (Cu-BTC) carbon nanomaterial composites (F-C) were prepared by a hydrothermal method, and the F-C powders were coated and fixed by macromolecular polymer copper alginate (CA). Then, F-CNTs@Cu-BTC@CA composite aerogel beads (F-C-CA) were prepared by a vacuum freeze-drying method. The new composite was characterized by BET, SEM, FTIR, and TGA, and its physical and chemical properties were analyzed. The results of batch adsorption experiments showed that F-C-CA aerogel beads had excellent adsorption capacity for TC. At 303 K, 10 mg F-C-CA aerogel beads adsorbed 20 mL 100 mg·L⁻¹ TC solution; the removal rate reached 94% after 48 h. After kinetic analysis, the adsorption process of F-C-CA on TC was found to be more coherent with the pseudo-second-order kinetic model (chemisorption process). The isotherm fitting analysis indicated that the adsorption behavior was more suitable to the Langmuir model (monolayer adsorption), and the fitted maximum adsorption was 297 mg·g⁻¹.

1. Introduction

Since ancient times, human beings have been constantly fighting against viruses and bacteria, and diseases caused by viruses and bacteria have brought great harm and challenges to human life and health [1]. The discovery and widespread use of antibiotics have provided humans with very effective ways in the process of fighting against viral and bacterial diseases [2]. Among the many antibiotics, tetracycline (TC) drugs are widely used because of their good therapeutic effect on bacterial infection [3]. However, the widespread use of TC can cause harm to humans and animals as well as the environment. In addition, TC has poor metabolism in human and animal bodies, which leads to the accumulation of TC in the natural environment [4]. On the one hand, TC in the natural environment can cause bacteria to develop drug resistance, making the therapeutic effect of TC reduced or even invalid [5]. On the other hand, long-term intake of TC will cause damage to the human body system [6]. Therefore, the question of how to efficiently and safely remove TC from the natural environment is being widely studied.

As an important physicochemical phenomenon, liquid–solid interfacial adsorption is of great importance in the field of environmental science research. In-depth study of liquid–solid interface adsorption can provide effective theoretical guidance for the development and engineering application of new adsorbents. Currently, the commonly used methods to remove TC in wastewater are mainly adsorption [7], chemical oxidation degradation [8], photocatalytic degradation [9], and membrane separation [10]. Among them, the adsorption method has the advantages of simple operation, no complex secondary products, good stability, and reusability, and it has been widely studied and used [11]. Common adsorbents include activated carbon, alginate, carbon nanotubes, chitosan, zeolite, graphene oxide, and metal-organic frameworks [12,13,14]. Metal-organic framework materials (MOFs) were widely studied because of their large specific surface area, inherent function, high porosity, and multi-purpose structure [15]. Zhou et al. used iron trimesic metal-organic frameworks (Fe-BTC) as adsorbents. They added 4 mg Fe-BTC to 20 mL TC solution (concentration range: 10–200 mg·L⁻¹), and the maximum adsorption amount (q_m) was 714 mg·g⁻¹ by Langmuir isotherm equation fitting at 298 K [16]. Chen et al. used UiO-66 to adsorb TC. They added 20 mg UiO-66 to a 40 mL conical flask containing TC solution (concentration range: 10–100 mg·L⁻¹), and at 298 K the q_m calculated by the Langmuir isotherm equation was 37.2 mg·g⁻¹ [17]. Although MOFs have a good adsorption effect as adsorbents, they are a carbon nanomaterial whose size is too small to be recycled. Therefore, in order to avoid secondary pollution, fixation of carbon nanomaterials by macromolecular polymers (such as chitosan and alginate) has attracted extensive attention. For example, Zhao et al. studied the us of chitosan to fix several different MOFs and measured their adsorption capacity for TC [6]. Ahmed M. Omer et al. reported on MIL-125 (Ti)/MIL-53 (Fe)/CNT@Alg microbeads, and the q_m was 294 mg·g⁻¹ by Langmuir isotherm fitting [18]. Therefore, it is an interesting research direction to find a reasonable and efficient macromolecular polymer composite for the fixation of carbon nanomaterials.

In this work, a new composite aerogel bead was developed for the adsorption of TC based on the theoretical basis of liquid–solid interfacial adsorption. It has been reported that TC can form strong coordination bonds with copper through the cation bonding bridge due to its electron-donor group [19]. Therefore, Cu-based MOF (Cu-BTC) and copper alginate (CA) were used in this study to form a double Cu-based composite adsorbent. The shape of Cu-BTC particles is a regular octahedral structure, and the particle size is 100 nm to 5 μt. The metal-organic frame ligand of Cu-BTC contains three dimeric Cu wheels in which the dimeric Cu wheel is formed by linking a Cu atom with the oxygen atoms of eight tetracarboxylate units. [20]. The sodium alginate (SA) molecule is a large chain polysaccharide substance consisting of two kinds of units, β-L-guluronic (G) units and α-D-annuronic (M) units. SA is cross-linked in CuSO₄ solution, and Cu atoms bind G and M units to form copper alginate (CA) molecules [21]. Each TC molecule contains four aromatic rings, which can form binding forces through electron donor–acceptor interactions when TC molecules are adsorbed onto the liquid–solid interface of the adsorbent, such as the cation bonding bridge (cation-n bond and cation-π bond), hydrogen bond, n-π electron donor–acceptor (n-π EDA), and π-π electron donor–acceptor (π-π EDA) interactions [22]. Accordingly, Cu-BTC@CA composites can be used as adsorbents to remove TC.

To further enhance the adsorption capacity of the composites, carboxylated functionalized carbon nanotubes (F-CNTs) were loaded on the Cu-BTC surface. Thus, we first obtained F-CNTs@Cu-BTC (F-C) powder. The powder was compounded with SA, and, finally, F-CNTs@Cu-BTC@CA (F-C-CA) aerogel beads were obtained through cross-linking in copper sulfate solution. The purpose of adding F-CNTs is as follows. On the one hand, F-CNTs can improve the agglomeration phenomenon of Cu-BTC particles [18]. On the other hand, F-CNTs can act as a skeleton to connect Cu-BTC particles and CA molecules, which can improve the structure of composites. Interestingly, in this study, the introduction of F-CNTs was found to make the surface of composite aerogel beads rougher and wrinkled, and the specific surface area was significantly increased. Thus, the adsorption capacity is also improved.

In this study, F-C-CA aerogel beads were prepared by a hydrothermal method and freeze-drying method. The new adsorbents were characterized by Fourier transform infrared spectroscopy (FTIR), field emission scanning electron microscopy (SEM), thermogravimetric analysis (TGA), and BET analysis. In order to investigate the ability of F-C-CA beads to remove TC, batch adsorption experiments were carried out systematically, and the adsorption performance of CNTs, F-CNTs, Cu-BTC, F-C, CA, and C-CA on TC was tested as a control group. The results showed that F-C-CA beads had the best removal performance. Firstly, 10 mg adsorbent was added to 20 mL TC solution with an initial concentration of 100 mg·L⁻¹ at 303 K; the adsorption capacity at adsorption equilibrium was 189 mg·g⁻¹ (removal rate: 94%). In order to investigate the adsorption mechanism of the novel composite aerogel beads, kinetic model, isotherm model, and thermodynamic analyses were carried out on the experimental data. In order to make the choice of kinetic and isotherm models more convincing, five data model selection criteria were used to analyze experimental and fitted data [23,24]. After kinetic and isotherm fit analysis and model selection criteria determination, the experimental data of TC adsorption by F-C-CA fit well on the pseudo-second-order kinetic and Langmuir isotherm models, indicating that the adsorption process was mainly chemisorption and monolayer adsorption. The behavior of TC adsorption by F-C-CA was found to be spontaneous and heat-absorbing by thermodynamic analysis. The results of the regeneration experiments of the new composite aerogel adsorbent showed that the F-C-CA beads still had 76% adsorption effect after five regenerations.

2. Materials and Methods of the Experiment

2.1. Materials and Reagents

The materials and reagents used in the preparation of the new composite aerogel adsorbent were as follows: benzene-1,3,5-tricarboxylic acid (H₃BTC), multi-walled carbon nanotubes (MWCNTs, diameter 20–40 nm, length 5–15 μm), N-N-dimethylformamide (DMF) copper nitrate trihydrate (Cu(NO₃)₂·3H₂O), copper sulfate (CuSO₄), concentrated sulfuric acid (H₂SO₄, 98 wt%), nitric acid (HNO₃, 65 wt%), and sodium alginate (SA). The sources of all materials and reagents are shown in

Table 1. Source of materials and reagents.

Materials and Reagents	Source of Acquisition
H3BTC	Shanghai Hao Hong Biomedical Technology Co., Ltd. (Shanghai, China)
MWCNTs	Shenzhen Nanotech Port Co., Ltd. (Shenzhen, China)
DMF	Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China)
Cu(NO3)2·3H2O	Tianjin Ruijinte Chemical Co., Ltd. (Tianjin, China)
CuSO4	Tianjin Ruijinte Chemical Co., Ltd. (Tianjin, China)
H2SO4	Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China)
HNO3	Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China)
SA	Tianjin Guangfu Fine Chemical Research Institute (Tianjin, China)

.

2.2. Preparation of F-CNTs@Cu-BTC

F-CNTs were prepared according to the method reported by Li et al. [25]. Briefly, a certain amount of CNTs was oxidized by condensation reflux in an acid bath, followed by dilution, filtration, washing, and drying to obtain F-CNTs. The synthesis of F-C powders was based on the previous studies of V. Jabbari et al. [26] and Zhang et al. [27], and their methods were improved in this experiment. The specific preparation method is shown in Figure 1. Firstly, 0.75 g H₃BTC, 0.2 g F-CNTs, and 1.5 g Cu(NO₃)₂·3H₂O were weighed, and 25 mL deionized water, 25 mL DMF, and 25 mL ethanol were measured with a cylinder. The above materials and reagents were then placed in a beaker and stirred and sonicated to thoroughly mix. The temperature of the oven was set to 373 K, and the heating time was set to 48 h. The autoclave containing the mixed solution was then placed in the oven. At high temperatures, metal-organic ligands of Cu-BTC begin to form. At this moment, binding forces were formed between F-CNTs and Cu-BTC, mainly hydrogen bonds and π-π EDA. At the end of heating in the high-temperature oven, the autoclave was removed and placed in the room for natural cooling. When the autoclave finished cooling, the lid of the autoclave was opened and the Teflon liner removed. When the lid was opened, a dark blue solid (F-C) could be seen attached to the inner wall of the Teflon liner. The F-C solid was naturally dispersed into the remaining solution in the Teflon liner by repeatedly cutting it against the inner wall with a spoon. The solution was then placed in a centrifuge with the speed set at 6000 r, and the centrifugation time was set to 5 min. The supernatant was poured out of the centrifuge tube, and deionized water was added to it. It was cleaned in an ultrasonic cleaner for 10 min. The centrifugation–cleaning step was repeated three times to remove the excess DMF from the F-C solid. After centrifugation, the supernatant was poured off. The centrifuge tubes with the F-C solids were placed in a vacuum oven at a temperature of 393 K, and the heating time was set to 8 h. After heating and cooling, the F-C solids were removed and ground thoroughly to obtain the F-C composite powder.

2.3. Synthesis of F-CNTs@Cu-BTC@CA Aerogel Beads

The preparation process of F-C-CA aerogel beads is displayed in Figure 1. Firstly, 1 g sodium alginate (SA) was placed in 49 mL deionized water and stirred on a magnetic stirrer for 5 h to fully dissolve it. Secondly, the prepared F-C powder (0.15, 0.20, 0.25, 0.30, 0.35, 0.40 g) was placed in the above-mentioned SA solution and stirred for 3 h so that the F-C powder could be fully dispersed in the solution. F-C is bound to SA mainly by hydrogen bonds between the hydroxyl oxygen (-O-) of the carboxyl group (-COO-) and the carbonyl oxygen (=O). Thirdly, the mixture of F-C and SA was dripped uniformly through a syringe into a 10% CuSO₄ solution and left for 12 h to fully cross-link. During the formation of copper alginate (CA), Cu²⁺ replaces Na⁺ in SA and binds the β-L-guluronic (G) unit and the α-D-annuronic (M) unit [28]. Thus, the cross-linked CA structure was very dense and could fix the F-C on the surface and the inside of the composite well. Fourthly, the cross-linked F-C-CA aerogel beads were placed into a beaker containing deionized water, and the beaker was placed onto a magnetic stirrer and stirred for 10 min. The remaining solution was poured off after cleaning, and the appropriate amount of deionized water was added. This step was repeated three times to fully clean off the excess CuSO₄. Finally, F-C-CA aerogel beads were obtained by vacuum freeze-drying for 12 h.

2.4. Characterization

The microstructure of the materials in this experiment was observed by field emission scanning electron microscopy (SEM, Hitachi, SU8020, Tokyo, Japan). In the present work, the results of specific surface area and pore size structure of all materials were obtained by the nitrogen adsorption–desorption method (also known as BET) (Micromeritics, ASAP2460 Version 3.00, Atlanta, GA, USA). Before the BET test, the material needed to be vacuum-dried at 353 K in advance in order to ensure that the moisture inside the material was fully removed and to avoid affecting the test results. Thermal stability was analyzed by a thermogravimetric analyzer (TGA, Q5000, TA Instruments, New Castle, DE, USA). In this study, samples were heated from 303 K to 873 K in a N₂ atmosphere at a rate of 10 K·min⁻¹. The functional groups in the materials were characterized by a Fourier transform infrared spectrometer (FTIR, Tensor 27, Bruker, Karlsruhe, Germany), and the wavelength range was 4000–400 cm⁻¹.

2.5. Adsorption Experiments

All adsorption experiments were carried out in a thermostatic water bath oscillator at a preset temperature at 190 rpm. The concentration of TC was measured by a UV–visible spectrophotometer (TU-1810, Beijing, China), and the wavelength for detecting TC concentration was 360 nm. In order to explore the most appropriate addition amount of F-C powder in the F-C-CA aerogel beads, the mass ratio between F-C and SA ((F-C:SA) × 100%) was set at 15%, 20%, 25%, 30%, 35%, and 40%. In order to demonstrate the good performance of the composite aerogel beads, the difference in adsorption capacity between the composite and its components was investigated. In order to explore the time when the adsorption behavior reached equilibrium and to analyze the adsorption kinetics, 125 mg F-C-CA aerogel beads was placed into 250 mL TC solution with a concentration of 100 mg·L⁻¹. The concentration of TC was measured at intervals of 2, 2, 4, 4, 8, 8, 16, 16, 32, 32, 64, 64, 128, 128, 256, 256, 600, and 600 min. In order to explore the influence of different temperatures and initial concentrations on adsorption behavior and to conduct adsorption isotherm analysis, different temperatures were set at 303, 313, and 323 K, and different initial concentrations were set at 40, 70, 100, 130, and 160 mg·L⁻¹. The initial pH of the TC solution can have a more pronounced effect on the adsorption behavior, and, in this study, the initial pH = 2, 3, 4, 5, 6, 7. In order to explore the influence of the amount of adsorbent on the adsorption effect, different quantities of adsorbent (2, 6, 10, 15, 20 mg) were placed into TC solution. In order to make sufficient contact between the TC molecules and the adsorption sites on the liquid–solid interface of the adsorbent, the time of oscillation in a constant temperature water bath oscillator was set to 48 h (refer to Section 3.2.3). The formula for the removal rate R_e (%) is as follows:

$$R_e=\frac{c_0-c_e}{c_0}\times 100\%$$

(1)

where c₀ (mg·L⁻¹) represents the initial concentration of the TC solution, and c_e (mg·L⁻¹) represents the concentration of the TC solution after the adsorption process has reached equilibrium.

The equation for equilibrium adsorption capacity q_e (mgum⁻¹) is:

$$q_e=\frac{c_0-c_e}{m}\times V$$

(2)

where m (g) represents the mass of the F-C-CA aerogel beads, and V (L) indicates the volume of the TC solution in the adsorption experiment.

In order to reduce the error in the experimental data, each set of adsorption experiments was repeated three times. The error in the experimental results was expressed by standard deviation (σ).

2.6. Model Selection Criteria

In order to make the results of the kinetic and isothermal fits more reliable, data model selection criteria were used to determine the credibility of several fitting models in this study. Among them, the Akaike Information Criterion (AIC) was created and developed by the Japanese statistician Hiroji Akaike. It is based on the concept of entropy and is able to measure the complexity of the estimated model and the goodness of this model’s fit to the data [23]. The AIC equation is as follows:

$$AIC=2p+N\ln \left(\frac{SSE}{N}\right)$$

(3)

The p in the equation represents the number of indeterminate parameters in the data fit model; the SSE stands for the sum of squares of the mean errors in the overall data sample, and N indicates the number of all samples in the overall sample.

When the number of samples is small (N/P < 40), then the corrected Akaike Information Criterion (AIC_C) can be used. The equation for the AIC_C is as follows.

$$AI{C}_C=AIC+\left[\frac{2p\left(p+1\right)}{N-p-1}\right]$$

(4)

In determining the reliability of a data model, the smaller the value of AIC and AIC_C, the more realistic the data fitting model will be in terms of the overall mechanics and nature of the data. [23]. The Akaike weight (w_i) is also an essential way of determining the reliability of a data fitting model and is often used in conjunction with AIC and AIC_C. The w_i equation is as follows [29]:

$$w_i=\frac{\exp \left(-0.5\Delta AI{C}_{c\left(i\right)}\right)}{{{\displaystyle \sum }}_{i=1}^R\exp \left(-0.5\Delta AI{C}_{c\left(i\right)}\right)}$$

(5)

$$\Delta AI{C}_{c\left(i\right)}=AI{C}_{c\left(i\right)}-AI{C}_{c, min}$$

(6)

R in Equation (5) indicates the number of data fitting models attempting to fit the data to an overall sample; AIC_{C, min} represents the smallest calculated AIC_C value among all models attempting to fit the data, and AIC_C(i) means the value of AIC_C for any one data fitting model. In determining which of a series of models is the most reliable, the sum of the w_i values of all models is 1, and, of all the models fitted to the data, the model which has the highest w_i value is judged to be the most appropriate model.

The Bayesian Information Criterion (BIC) and the Hannan Information Criterion (HIC) are improvements on the AIC in terms of the information criteria used to make reliability judgments on data fitting models, and their equations are shown below [30]:

$$BIC=plnN+N\ln \left(\frac{SSE}{N}\right)$$

(7)

$$HIC=2pln\left(lnN\right)+N\ln \left(\frac{SSE}{N}\right)$$

(8)

Smaller values of BIC and HIC indicate a higher degree of confidence in the model fitted to the data.

The results of the above data fit model selection criteria are presented in Section 3.3 and Section 3.4.

2.7. Regeneration Experiments

In practical applications, the reusability of materials is an important factor in the evaluation of adsorbents. When the adsorption process had reached equilibrium, TC was separated from the liquid–solid interface of the adsorbent by soaking in NaOH solution with 0.001 M concentration for 12 h. Then, it was rinsed several times with deionized water to remove the excess NaOH. F-C-CA aerogel beads were then pre-frozen and vacuum freeze-dried to complete regeneration. The above steps were repeated five times, and the renewable performance of the F-C-CA aerogel beads was assessed by the ratio of the regenerated aerogel (q_e, n) to the adsorption capacity of the original aerogel (q_e, 0).

3. Results and Discussions

3.1. Characterization Results

3.1.1. SEM

The morphology of several different materials is illustrated in Figure 2a–f. Figure 2a shows the microscopic shape of Cu-BTC. The shape of the Cu-BTC particles is a regular octahedral structure [31]. Figure 2b illustrates the morphology of F-CNTs, whose elongated structure can act as a connecting skeleton in the F-C-CA composite. Figure 2c demonstrates the morphology of F-C, with a tight connection between F-CNTs and Cu-BTC. Figure 2d shows the surface topography of CA. The surface of CA was dense and wrinkled, which is the same as previous research results [21]. Figure 2e illustrates the morphology of the C-CA. The surface of the material became significantly rougher after Cu-BTC was added to CA. Figure 2f shows the surface morphology of F-C-CA aerogel beads. Compared with CA and C-CA, the surface of F-C-CA was rougher and had a large number of folds. According to the surface morphology of the three, a preliminary prediction can be made: the specific surface area of F-C-CA with the roughest surface and gully structure was larger than that of CA and C-CA. Thus, the adsorption capacity of F-C-CA, as inferred from the surface morphology results, should be better than both of them.

3.1.2. BET

Figure 3a–d shows the BET results and pore size distributions (see insets in Figure 3a–d) of F-CNTs (Figure 3a), CA (Figure 3b), C-CA (Figure 3c), and F-C-CA (Figure 3d). The N₂ adsorption–desorption isotherms of F-CNTs, CA, and C-CA were type II isotherm, which mainly reflects the adsorption process on mesoporous, non-porous, or macroporous adsorbents [32]. The F-C-CA is type IV isotherm; adsorption hysteresis loops appear in the middle segment [33]. By BET test and calculation, the specific surface areas of F-CNTs (131 ± 9.2 m²·g⁻¹), CA (4.2 ± 0.3 m²·g⁻¹), C-CA (62.4 ± 2.1 m²·g⁻¹), and F-C-CA (97.7 ± 2.9 m²·g⁻¹) were obtained. It can be seen that the specific surface areas of C-CA and F-C-CA with carbon nanomaterials added were greatly improved compared with that of CA. Meanwhile, the results of specific surface area also corresponded to the surface roughness of the materials in Section 3.1.1; that is, the rougher the surface was, the larger the specific surface area was. The results of BET show that the specific surface area of C-CA was smaller than that of F-C-CA. This is because the F-CNTs act as a bonding skeleton in the composite, exposing more carbon nanomaterials to the surface of CA, and so the surface was rougher. By analyzing the results of the pore size distribution, it was found that the pore structure of C-CA and F-C-CA was much more than CA. The pore structure of F-C-CA was higher than C-CA, and the proportion in the mesoporous range (2–50 nm) was higher than that of C-CA. The distribution and content of pore size play a decisive role in the specific surface area of materials and also reflect the adsorption capacity of materials to a certain extent.

3.1.3. TGA Analysis and DTG Analysis

Thermogravimetric analysis (TGA) reflects the thermal stability of the material and allows some inference to be made about the structure and functional groups contained in the material, the results of which are shown in Figure 4a. When heated to around 473 K, each material lost about 5%–10% of its weight, mainly due to water loss from the material. For F-CNTs, only about 7% of their weight was lost from 473 K until heating to 873 K, which was due to the excellent thermal stability of CNTs themselves, and the lost mass was caused by the decomposition of carboxyl groups on F-CNTs by heating. The weight loss of Cu-BTC and F-C was about 30% between 473 and 593 K, which was related to the loss of free binding DMF [34]. The weight loss of Cu-BTC between 593 K and 630 K was about 30%, which was caused by the decomposition of the ligand structure [35]; this indicates that the carboxyl and ester functional groups in the material have been decomposed and volatilized at high temperatures [36]. As can be seen from the TGA curves, there was almost no thermal weight loss of the sample after 630 K, indicating that only the carbon structure, with strong thermal stability, remained in the sample after high-temperature calcination. At this point, F-C was about 5% more than the remaining mass of Cu-BTC, which was the CNTs in F-C. Between 473 K and 543 K, the hydroxyl functional group and the carboxyl functional group in CA were decomposed, which resulted in a deficit of about 25% of the mass. CA lost about 25% of its mass between 543 K and 673 K due to thermal decomposition of the molecular chain of alginate [33]. C-CA and F-C-CA lost about 45% of their mass between 473–673 K, and the reasons for the mass reduction were similar to the above analysis results.

The derivative thermogravimetric (DTG) curves are demonstrated in Figure 4b. All materials had a peak at 373 K due to evaporation of unbound water from the sample [37]. The peak near 473 K represented the loss of bound water from the material. Briefly speaking, peaks near 523 K corresponded to decomposition of functional groups such as hydroxyl and carboxyl, and peaks near 623 K corresponded to decomposition of material molecular structure and molecular chains. From the DTG figure, we can see that the peaks on the curves of C-CA and F-C-CA are closer to the right than those on the curves of CA, which indicates that the thermal stability of the composite material obtained by adding carbon nanomaterials to CA was improved.

3.1.4. FTIR

FTIR spectral characterization results are shown in Figure 4c. For Cu-BTC, the peak at 1715 cm⁻¹ represents C=O on the carboxyl functional group. The peak at 1050 cm⁻¹ represents R–CH₂–OH, and the peaks appearing at 1586 cm⁻¹ and 724 cm⁻¹ indicate C=C bonds on benzene rings and C–H on substituted benzene rings. The peaks of the FTIR spectral profile of F-C are similar to those of Cu-BTC, which demonstrates that the addition of CNT to the preparation of F-C powder did not affect the structure and functional groups of Cu-BTC. The curves of C-CA and F-C-CA are very similar. The curves of both of them have a broad peak at 3300 cm⁻¹, which indicates the binding –OH in samples; the peak at 1715 cm⁻¹ represents the C=O on the carboxyl functional group, which can determine that the composite material contained the carboxyl functional group; the peak at 1050 cm⁻¹ represents R–CH₂–OH. On the curves of C-CA and F-C-CA, there are also peaks at 1587 cm⁻¹ and 725 cm⁻¹, which indicate the presence of benzene ring structure in the material [38].

3.2. Results of Adsorption Experiments

3.2.1. The Effect of F-C Content on Adsorption Capacity

Figure 5a indicates the influence of different F-C contents in F-C-CA on adsorption performance. When the content of F-C was 0—that is, CA aerogel beads—the adsorption capacity (q_e) was 156 mg·g⁻¹ (σ = 2.8). After the addition of F-C, the q_e of F-C-CA was obviously improved, and with the increase in F-C content, the q_e of the F-C-CA aerogel beads was better. This is because the addition of F-C increased the specific surface area of the material, thus increasing more effective adsorption sites. When the ratio of F-C to SA was 35%, the adsorption performance of F-C-CA reached the best level (q_e = 189 mg·g⁻¹, σ = 1.7). When the ratio of F-C to SA was 40%, the adsorption effect of the F-C-CA aerogel beads was no longer improved. Although more F-C increases the specific surface area of the F-C-CA aerogel beads, the effective adsorption sites of adsorbents may not increase once F-C content reaches a certain level. Meanwhile, the mechanical strength of the aerogel beads deteriorated, and the aerogel beads were damaged after 48 h oscillation in the constant temperature water bath oscillator. Therefore, the optimal ratio of F-C to SA in the preparation process was 35%.

3.2.2. Results of Adsorption Properties of Different Materials

The results for the adsorption capacity of the seven different materials are presented in Figure 5b. Ten milligrams of adsorbent was placed into 20 mL 100 mg·L⁻¹ TC solution. The equilibrium adsorption quantities (q_e) of CNTs, F-CNTs, Cu-BTC, F-C, CA, C-CA, and F-C-CA were 97, 129, 175, 179, 156, 180, and 189 mg·g⁻¹, respectively (σ = 2.2, 1.7, 2.2, 2, 2, 3, 1.7). The removal rates were 48.5%, 64.5%, 87.5%, 89.5%, 78%, 90%, and 94.5%, respectively. In order to make the results more intuitive, the above results can also be seen in

Table 2. Comparison of adsorption properties of different materials.

Materials	qe (mg·g−1)	σ	Re (%)
CNTs	97	2.2	48.5
F-CNTs	129	1.7	64.5
Cu-BTC	175	2.2	87.5
F-C	179	2	89.5
CA	156	2	78
C-CA	180	3	90
F-C-CA	189	1.7	94.5

. CNTs without oxidation treatment are hydrophobic, and so they are easy to agglomerate in aqueous solution under the action of van der Waals forces, which reduces the effective contact area and leads to poor adsorption capacity. Interestingly, the adsorption capacity of F-C was superior to both the Cu-BTC and F-CNTs parent materials. The reason for this result may be that the effective adsorption sites in the F-C composite were increased after the addition of F-CNTs. The results for adsorption capacity (q_e (CA) < q_e (C-CA) < q_e (F-C-CA)) corresponded to the results for specific surface area in Section 3.1.2. Despite the fact that the specific surface areas of the four carbon nanomaterials powders were higher than the three polymer materials containing alginate, the adsorption capacity of C-CA and F-C-CA exceeded that of the three carbon nanomaterials. This was because copper alginate itself contains abundant functional groups, resulting in a high proportion of effective adsorption sites. Moreover, the high specific surface areas of the three carbon nanomaterials are mainly determined by a large number of microporous structures (<2 nm); a small microporous structure makes it difficult for TC molecules to enter and leads to a small effective contact area. In addition, the effective adsorption functional groups on the surface of the three carbon nanomaterials are also fewer than those in CA, resulting in a smaller proportion of effective adsorption sites.

3.2.3. Influence of Different Conditions on Adsorption Capacity

Figure 5c illustrates the results of the effect of the contact time between the adsorbent and the TC solution on the adsorption performance of the F-C-CA beads. The removal efficiency of the F-C-CA aerogel beads for TC was fast in the early stage, and about 60% TC was removed within 200 min. This was due to the large number of unoccupied adsorption sites in the early stages of the adsorption process; TC molecules rapidly bound to the bare adsorption sites [39], and, at this time, the concentration of TC was high, which accelerated the binding of TC molecules to adsorption sites on the liquid–solid interface of the adsorbent under the driving force of concentration gradient. As the contact time increased, the rate of adsorption slowed down, reaching equilibrium at around 2200 min. This was due to the rapid occupation of a large number of adsorption sites in the early stages of the adsorption process, and the concentration gradient driving force of TC became weaker [40]. When all of the adsorption sites were occupied, the adsorption process reached equilibrium.

Different ambient temperatures will affect the adsorption process. Results, as shown in Figure 5d, showed that the higher the temperature, the better was the adsorption effect of F-C-CA aerogel beads on TC (q_{e, 323 K} > q_{e, 313 K} > q_{e, 303 K}). On the one hand, higher temperature makes the molecular movement more active, making TC molecules more fully bound to the adsorption sites on the liquid–solid interface of the adsorbent. On the other hand, the adsorption process of TC on F-C-CA aerogel beads was endothermic [41].

Figure 6a shows the effect of initial pH of the adsorbed matter. Because TC is unstable in alkaline conditions [42], when pH < 6.5, the molecular structure of TC changes [24], and the main products are isotetracycline and 4-epi-isotetracycline [43]. We only conducted experiments under acidic conditions (2 ≤ pH ≤ 7). As illustrated in Figure 5a, when pH = 3, the adsorption effect of the F-C-CA aerogel beads was the best (q_e = 195 mg·g⁻¹, R_e = 97.5%, σ = 2.4). The adsorption effect was the worst at pH = 2 (q_e = 105 mg·g⁻¹, R_e = 52.5%, σ = 2.8). This is because, in solutions containing a high concentration of H⁺, TC mainly exists in the form of TCH₃⁺ and TCH₂⁺, which are positively charged, and the dissociation of H⁺ in the –COOH and –OH on the surface of F-C-CA is inhibited, resulting in the material surface approaching non-electronegativity (see Figure 6b Zeta potential diagram). Therefore, in the environment with low pH, electrostatic adsorption was inhibited, resulting in a poor adsorption effect. When 2 ≤ pH ≤ 3, the adsorption effect of F-C-CA on TC was improved with the decrease in H⁺ concentration. This is because H⁺ in solution had a weakened inhibition on the H⁺ dissociation of –COOH and –OH on the surface of F-C-CA [44]. As can be seen from the Zeta potential diagram in Figure 5b, the surface electronegativity of the material gradually increased. Meanwhile, the electrostatic adsorption between the TC molecule and F-C-CA surface was enhanced, such that the adsorption performance was better. When 3 ≤ pH ≤ 7, the adsorption effect of F-C-CA on TC decreased with the increase in pH. This is because TC exists in many forms in weak acid environments: cation (TCH₃⁺), amphoteric (TCH₂^±), and anion (TCH⁻, TC²⁻). In addition, with the decrease in H⁺ concentration, the proportion of anions (TCH⁻ and TC²⁻) increased, and the electrostatic adsorption was inhibited, leading to a poorer adsorption effect [4,45].

Figure 6c illustrates the results of the effect of the initial concentration of the TC solution on the adsorption process. When the initial concentration (c₀) was low, the equilibrium adsorption capacity (q_e) was small. As can be seen from the results in Figure 5c, q_e increased with increasing c₀. The reason for this phenomenon is that the higher concentration TC solution has a greater concentration gradient driving force, resulting in more adsorption sites on the F-C-CA beads. As c₀ increased, the adsorption sites were gradually occupied, resulting in a decrease in the rate of increase in q_e [46].

Figure 6d illustrates the results for the influence of the adsorbent dose on the behavior of the adsorption. At 303 K, different masses of F-C-CA aerogel beads were added to 20 mL 100 mg·L⁻¹ TC solution. The removal rate (R_e) of TC by F-C-CA increased from 45% to 97%, and this was because the addition of aerogel beads provided a greater number of adsorption sites for TC molecules, thereby improving the R_e of TC. However, the equilibrium adsorption capacity (q_e) of F-C-CA on TC declined from 449 mg·g⁻¹ to 97 mg·g⁻¹ (σ = 1.7, 1.4, 1.7, 2.4, 3.3). This phenomenon is due to the decrease in the utilization rate of effective adsorption sites on the aerogel beads with the increase in adsorbent, resulting in a decrease in the q_e of F-C-CA [33].

3.3. Adsorption Kinetics

In order to explore the type and mechanism of the adsorption process of F-C-CA on TC, three kinetic models were used to fit the experimental data in this study (see Figure 7a). The three kinetic models are the pseudo-first-order (PFO) model, pseudo-second-order (PSO) model, and Elovich model. The PFO model can predict that the adsorption process is a physical adsorption process [14]. The PSO model is a model that indicates that the adsorption process is dominated by chemical adsorption [22]. The Elovich model can demonstrate that the adsorbent is a non-homogeneous solid [47]. The above kinetic models are formulated as follows:

$$\mathrm{PFO} \mathrm{model}: q_t=q_e\left(1-e^{-k_{1}t}\right)$$

(9)

$$\mathrm{PSO} \mathrm{model}: q_t=\frac{k_2{q}_e^{2}t}{1+k_2{q}_{e}t}$$

(10)

$$\mathrm{Elovich} \mathrm{kinetic} \mathrm{model}: q_t=\frac{\ln \left(\alpha \beta \right)}{\beta }+\frac{\ln \left(t\right)}{\beta }$$

(11)

where q_t (mg·g⁻¹) indicates the adsorption quantity at a certain moment, which changes with the independent variable t (min) of time and is the dependent variable of t; q_e (mg·g⁻¹) is the equilibrium adsorption capacity, which is an unknown parameter in the model formula, and its fitting value can be obtained through fitting calculation; k₁ (min⁻¹) and k₂ (min⁻¹) are the rate constants of the PFO and PSO models. They are also unknown parameters in their respective model formulas, and their values can be obtained by fitting calculation; α (mg·g⁻¹) represents the initial adsorption rate; β (g·mg⁻¹) indicates the relationship between activation energy for chemisorption and coverage of the surface.

The parameter values and model reliability evaluation results of the three adsorption kinetics model formulas are shown in

Table 3. Results for parameters and the information criterion determination of kinetic models.

Models	Parameters
Pseudo-first-order model	k1 (min−1)	qe (mg·g−1)	R2	AIC	AICC	wi	BIC	HIC
	5.67 × 10−3	165 ± 7	0.92	110	111	3.66 × 10−5	112	110
Pseudo-second-order model	k2 (min−1)	qe (mg·g−1)	R2	AIC	AICC	wi	BIC	HIC
	4.11 × 10−5	184 ± 6	0.97	91.8	92.6	0.34	93.7	92.1
Elovich kinetic model	α (mg·g−1·min−1)	β (g·mg−1)	R2	AIC	AICC	wi	BIC	HIC
	7.16 ± 0.9	0.04	0.97	90.5	91.3	0.66	92.4	90.8

. Through the R² values of the three kinetic models (PFO: R² = 0.92, PSO: R² = 0.97, Elovich: R² = 0.97), we can preliminarily judge that the adsorption behavior of F-C-CA on TC was more consistent with the PSO and Elovich models. Since the R² values of the PSO and Elovich models were the same, model selection information criteria should be further used to determine which model is most appropriate. The results in Table 3 show that the Elovich model had the smallest values for AIC, AIC_C, BIC, and HIC, whereas w_i was the largest. Thus, the Elovich model has the highest reliability, and, therefore, the Elovich model better reflects the adsorption mechanism and behavior of F-C-CA aerogel beads on TC, which indicates that the new composite aerogel beads have a heterogeneous solid surface.

3.4. Adsorption Isotherm

Figure 7b shows the fitting results of Langmuir, Freundlich, Sips, Temkin, and Dubinin-Radushkevich (D-R) isotherm models to the experimental data at 303 K. The Langmuir model usually proves that the adsorbent surface is homogeneous [47]. The Freundlich model is often used to indicate that the surface of the adsorbent is multilayered and non-homogeneous [48]. The Sips isothermal model is an improved empirical formulation of the Langmuir and Freundlich models. The closer the value of b in the Sips model equation is to 1, the more homogeneous the adsorbent surface is. When the value of n is equal to 1, the Sips model equation is the Langmuir equation [49]. The Temkin model can describe the relationship between the adsorbent and the heat of adsorption. The data from the adsorption experiments can be fitted to the D-R model to determine whether the adsorption behavior is physical, ion exchange, or chemisorption [47]. The formulas of the five isothermal models are as follows:

$$\mathrm{Langmuir}: q_e=\frac{q_m{K}_L{c}_e}{1+K_L{c}_e}$$

(12)

$$\mathrm{Freundlich}: q_e=K_F{c}_e^{1/n}$$

(13)

$$\mathrm{Sips}: q_e=\frac{q_m{(K_S{c}_e)}^b}{1+{(K_S{c}_e)}^b}$$

(14)

$$\mathrm{Temkin}: q_e=B_T\ln \left(A_T{C}_e\right)$$

(15)

$$\mathrm{D}\text{-}\mathrm{R}: q_e=q_m\exp [-A_D(\ln (1+\frac{1}{C_e}){)}^2]$$

(16)

where q_m (mg·g⁻¹) is the maximum sorption amount calculated by fitting the isotherm model; it is an unknown parameter in the model and can be obtained through fitting calculation. K_L (L·mg⁻¹) is related to the energy of adsorption, and its value can be obtained by fitting calculation. K_F (mg^1−1/n·L^1/n·g⁻¹) is an unknown parameter, and its value can be calculated by fitting; n represents the parameter related to adsorption strength. The value of n was greater than 1 (n = 2.91), indicating that TC had good heterogeneous adsorption performance on F-C-CA [50]. K_S [(L/mg)^1/b] is the affinity constant for adsorption; b is the index of heterogeneity. In the Temkin isothermal model, B_T (J·mol⁻¹) represents the energy constant, and A_T (L·mg⁻¹) is the Temkin isotherm equilibrium binding constant. A_D = K_DR²T², where K_D (mol²·J⁻²) is associated with adsorption energy; R is the universal gas constant; and E is the average free energy. The value of E can be calculated by E = (2K_D)⁻⁰.⁵. When E < 8 kJ·mol⁻¹, this indicates that physical adsorption is dominant. When 8 < E < 16 kJ·mol⁻¹, this means that the adsorption behavior is based on ion exchange as the main mode of adsorption. When E > 16 kJ·mol⁻¹, this indicates that, in this adsorption process, the adsorbent and the adsorbed material are mainly bound to each other by chemisorption [47].

Table 4. Results for parameters and the information criterion determination of isotherm models.

Models	Parameters
Langmuir	qm (mg·g−1)	KL (L·mg−1)		R2	AIC	AICc	wi	BIC	HIC
	297 ± 14	0.25 ± 0.04		0.98	25.1	31.1	0.64	24.3	23.0
Freundlich	Kf (mg1−1/n·L1/n·g−1)	n		R2	AIC	AICc	wi	BIC	HIC
	87.6 ± 13.5	2.91 ± 0.48		0.94	31.9	37.9	0.02	31.2	32.4
Sips	qm (mg·g−1)	KS [(L·mg−1)1/b]	b	R2	AIC	AICc	wi	BIC	HIC
	288 ± 33	0.27 ± 0.08	1.08 ± 0.29	0.99	26.9	50.9	3.14 × 10−5	25.8	23.8
Temkin	BT (J·mol−1)	AT (L·mg−1)		R2	AIC	AICc	wi	BIC	HIC
	62.9 ± 5.1	2.57 ± 0.64		0.98	26.4	32.4	0.34	25.6	24.3
D-R	qm (mg·g−1)	AD	E (kJ·mol−1)	R2	AIC	AICc	wi	BIC	HIC
	236 ± 20	5.76 ± 1.96	742	0.88	35.6	41.6	3.2 × 10−3	34.9	33.5

shows the values of the parameters derived from the isotherm fit and the results of the determination of the model selection criteria. We found that the R² of the Sips, Langmuir, and Temkin isotherm models were higher; they were 0.99, 0.98, and 0.98, respectively. However, after judging the five isotherm models by several information criteria, it could be seen that the AIC, AIC_C, BIC, and HIC values of the Langmuir isotherm model were the smallest, and w_i was the largest. Thus, the Langmuir model was the most appropriate to describe the adsorption behavior of F-C-CA on TC, and the adsorption behavior of F-C-CA on TC was more consistent with the monolayer adsorption process. The maximum adsorption capacity (q_m) of F-C-CA on TC obtained by the Langmuir isotherm model was 297 ± 14 mg·g⁻¹.

3.5. Adsorption Thermodynamic

Due to the effect of temperature on the movement of TC molecules, an energy-dependent mechanism inevitably occurs during the adsorption process. From Section 3.2.3, it is clear that an increase in temperature increased the adsorption capacity of F-C-CA on TC. Based on this result, we can tentatively determine that the adsorption behavior of F-C-CA on TC is endothermic. The above results are further illustrated by thermodynamic analysis. The formulas of enthalpy change (ΔH⁰), entropy change (ΔS⁰), and Gibbs free energy change (ΔG⁰) are as follows [22].

$$\Delta G^0=-RTLn{K}_0$$

(17)

$$ln{K}_0=-\frac{\Delta H^0}{RT}+\frac{\Delta S^0}{R}$$

(18)

$$K_0=\frac{q_e}{c_e}$$

(19)

where R is the universal gas constant (8.314 J·mol⁻¹·K⁻¹); ΔG⁰ (kJ·mol⁻¹) is the Gibbs free energy change. ΔH⁰ (kJ·mol⁻¹) is the enthalpy change, and ΔS⁰ (J·mol⁻¹·K⁻¹) is the entropy change. T (K) represents the Kelvin temperature; q_e (mg·g⁻¹) stands for the amount of adsorption at equilibrium; c_e (mg·g⁻¹) means the concentration of the solution at which the adsorption process reaches equilibrium [51].

The calculated results of K₀ were 32.3 (303 K), 40.0 (313 K), and 54.4 (323 K). The results of ΔH⁰, ΔS⁰, and ΔG⁰ obtained by linear fitting of lnK₀ and 1/T (see Figure 7c) are listed in

Table 5. Results of thermodynamic analysis.

ΔH0 (kJ·mol−1)	ΔS0 (J·mol−1·K−1)	ΔG0 (kJ·mol−1)
		303 K	313 K	323 K
21.2	98.6	−8.75	−9.60	−10.7

. As ΔG⁰ < 0, the adsorption process of F-C-CA aerogel beads on TC was spontaneous. ΔH⁰ > 0 indicates that the adsorption process of F-C-CA on TC was an endothermic process. ΔS⁰ > 0 indicates that the disorder and randomness of the solid–liquid interface increase after F-C-CA adsorbs TC [38].

3.6. Results of Regeneration Experiment

Figure 7d shows that, with the increase in regeneration times, the adsorption effect of F-C-CA on TC gradually weakened, which was attributed to the decrease in effective adsorption sites of the adsorbent. Firstly, the effective functional groups on the F-C-CA aerogel beads changed irreversibly with the repeated use of the adsorbent. Secondly, the pore structure and specific surface area of the adsorbent changed during the process of application and regeneration. Finally, some TC molecules bound to the adsorption site could not be fully removed during the desorption process. After five sorption–desorption cycles, the adsorption effect of F-C-CA on TC was 76% higher than that of the adsorbent without regeneration treatment.

3.7. Analysis of Adsorption Mechanism

The adsorption mechanism of the F-C-CA aerogel beads on TC is shown in Figure 8. The adsorption of TC molecules on F-C-CA was accomplished by various forces. First of all, TC is a nano-sized small molecule which can form a weak adsorption force with the adsorbent under the action of van der Waals forces [52]. Secondly, the difference in charge between the TC molecule and F-C-CA can form electrostatic adsorption forces. Thirdly, hydroxyl (-OH) and carboxyl (-COOH) in TC and F-C-CA can be connected by hydrogen bonding [53]. Fourthly, the complex between the TC molecule and Cu abundant in F-C-CA can form a TC–metal complex through a cation bonding bridge. Fifthly, π-π electron donor–acceptor (π-π EDA) interactions exist between the benzene ring structure of the TC molecule and F-C-CA (mainly Cu-BTC and F-CNTs). Finally, the adsorption of TC on F-C-CA has n-π EDA interactions. In this case, n is the electron donor, which can be an oxygen atom or a nitrogen atom. The aromatic ring on the molecular structure of TC acts as the π electron acceptor, and the carbonyl oxygen (=O) on F-C-CA acts as the electron donor, thus forming the n-π EDA interaction between the adsorbent and TC [54].

3.8. Comparative Study with Other Adsorbents

Table 6. The comparison results of adsorption properties of TC on various adsorbents.

Adsorbent	Experimental Conditions			Ref.
	T (K)	c0 (mg·L−1)	qm (mg·g−1)
Fe-BTC powder	298	10–200	714	[16]
UiO-66 powder	298	10–100	37.2	[17]
MIL-125/MIL-53/CNT@Alg beads	298	50–300	294	[18]
MOF-525@GO powder	303	100–200	436	[49]
CA-BT microspheres	318	20–140	154	[22]
SA-Cu bead	318	15–90	53	[21]
N-doped MWCNTs	N/A	8–111	147	[30]
CNT/β-CD/MnFe2O4	298	10–100	90	[7]
CA-CNTs membrane	318	40–140	230	[33]
F-C-CA aerogel beads	303	40–160	297	This work

shows the comparison results of adsorption properties of TC between various adsorbents previously studied and the new composite material studied in this paper.

4. Conclusions

In summary, a new composite aerogel with excellent adsorption effect on TC was prepared through a green and safe way in this study. The adsorption capacity of aerogels was improved by adding carbon nanomaterials to alginate. According to the characteristics of complexation between Cu and TC, two Cu-containing materials (copper alginate and Cu-BTC) were used for the composite. In order to further improve the adsorption effect of the composites, CNTs were added to the surface of Cu-BTC particles. Finally, F-C-CA aerogel beads were obtained. When the ratio of F-C to SA was 35%, the adsorption capacity of the new material was the best. The optimal initial pH of F-C-CA for TC adsorption was 3. According to the fitting results of adsorption kinetics and adsorption isotherms, the adsorption behavior of F-C-CA on TC was more consistent with the Elovich kinetic model and Langmuir isotherm model. Therefore, the surface of the F-C-CA composite was a heterogeneous solid surface, and the adsorption behavior of F-C-CA on TC was more consistent with the monolayer adsorption process. The maximum adsorption capacity (q_m) obtained by fitting the Langmuir isotherm model was 297 mg·g⁻¹. In conclusion, the F-C-CA aerogel beads prepared in this study have good application prospects.