Luminescent Molecularly Imprinted Polymers Based on Covalent Organic Frameworks and Quantum Dots with Strong Optical Response to Quinoxaline-2-Carboxylicacid

Three-dimensional molecularly imprinted polymers (MIPs) based on quantum dots-grafted covalent organic frameworks (QDs-grafted COFs) are reported in this study. The compound 1,3,5-triformylphloroglucinol-P-phenylenediamine was used as COF material to react with the amino-modified CdSe/ZnS QDs by Schiff-base reactions. The amino-derived QDs reacted with quinoxaline-2-carboxylicacid (QCA) via a non-covalent interaction. The system combines the advantages of MIPs, COFs, and QDs for highly sensitive and selective QCA detection. The MIPs based on QDs-grafted COFs showed good chemical selectivity and thermal stability, as well as consistency in QCA optosensing. Under optimal conditions, the detection limit for QCA in meat and feed samples was 0.85 μmol L−1, over a linear concentration range of 1–50 μmol L−1. The current findings suggest a potential application of MIPs based on QDs-grafted COFs for the detection of trace levels of hazardous chemicals for food safety and environmental control.


Introduction
Luminescent covalent organic frameworks (COFs) with the porosity of COFs and intrinsic optical properties, which can transform the chemical signal produced by the host-guest interaction into detectable changes in luminescence, have attracted wide attention. Although various topological diagrams of building blocks with predesigned geometry and symmetry by virtue of dynamic covalent bonds have been developed for COFs synthesis [1][2][3][4][5], luminescent COFs remain a challenge. Most layered, two-dimensional COFs (2D-COFs) with uniform and perforated pores exhibit conjugated 2D layers and periodic columnar π-π arrays, which trigger the thermal decay of photoexcited states and result in less-or non-emissive luminescence [6]. The π-π accumulation-caused quenching mechanism has been explained by the quick dissipation of the COFs excitation energy [7,8]. Crystalline multiple-component COFs are constructed by both a heterogeneous and a homogeneous distribution of π-conjugated vertex units of dehydrobenzoannulene and exhibit unique solid-state optical properties [9]. Dalapati et al. have prepared a highly emissive COFs using an aggregation-induced emission-active chromophore to the π-frameworks. It uses an aggregation-induced emission mechanism to interact with the intralayer covalent bonding, and the interlayer non-covalent π-interactions work to reduce rotation-induced thermal decay of the photo-excited state [10]. The highly luminescent COFs are first achieved, which use a high sensitivity sensor to reduce the detection limit of ammonia at the

Synthesis of TpPa COFs
The typical Schiff-base reaction was used for the synthesis of TpPa COFs [22]. In a 500 mL capacity conical flask, 0.3 mmol of 1,3,5-triformylphloroglucinol (Tp) and 0.45 mmol of P-phenylenediamine (Pa) were mixed, then 1.5 mmol of acetic acid and 3 mL of mesitylene/dioxane (1:1) were added successively. Then, the mixture was sonicated for 10 min, after keeping it at 120 • C for three days with nitrogen sealing. After centrifugation, a red-colored precipitate was obtained, and anhydrous acetone was used to wash the precipitate six times. Finally, it was dried under a vacuum at 180 • C for 24 h.

Synthesis of MIPs Based on QDs-Grafted COFs
The MIPs based on QDs-grafted COFs were prepared by reverse microemulsion [23]. Triton X-100 (180.0 mL) and cyclohexane (750.0 mL) were added into a 25 mL round-bottom flask, and then stirred for 15 min at room temperature. CdSe/ZnS QDs (1.0 mL), silane reagent of TEOS (50.0 µL), and aqueous ammonia (25 wt%, 100.0 µL) were then added to the flask. After stirring for 2 h, APTES (20.0 µL) and TpPa (0.5 mg) were added to the mixture, which was stirred for another 2 h. QCA (3.0 mg) was dissolved in ethanol solution (10.0 mL) and added (100.0 µL) to the flask. The mixture was sealed and stirred for 24 h. After pre-polymerization, the product was purified by centrifugation and then washed twice with acetone and DDW. In all of the washing procedures, the mixture was centrifuged to remove the supernatant at 5000× g for 10 min, and then the precipitate was re-dispersed in the next solvent. After purification, the products were washed with methanol until no template molecules were found using UV-vis spectrophotometry. Lastly, the products were dried under vacuum at 40 • C for 10 h. The products without QCA were prepared by the same procedure.

Fluorescence Measurement
All of the measurements of fluorescence intensity were performed under the same conditions. The excitation and emission wavelength were set at 460 nm and 610 nm, respectively. The MIPs based on QDs-grafted COFs were mixed evenly for FL-4500 spectrofluorometer analysis.
During the detection, 3 mL of target standard solution at a specific concentration was added into a quartz cell before 1.0 mg of MIPs based on QDs-grafted TpPa COFs was added. After mixing, fluorescence intensity detection was performed using the FL-4500 spectrofluorometer.

Meat and Feed Samples
Recently, QCA was reported at trace levels in beef, pork, fish, pork feed, cattle feed, and fish feed samples. Therefore, meat and feed samples were collected from a local supermarket. We homogenized the samples by a high-speed food blender, and packed them (5.00 ± 0.01 g) into a 100 mL centrifuge tube. QCA was added at three concentrations: 0, 5, 10, and 20 µmol L −1 . Next, 5 mL of a solution containing 5% metaphosphoric acid (w/v) and 20% acetonitrile (v/v) was added. After the above mixture was mixed for 5 min, 10 mL of acetonitrile was added. Then, it was centrifuged at 5000× g for 10 min, and the supernatant was collected and evaporated to near dryness at 40 • C. Finally, the residue was re-dissolved in 1 mL of methanol and diluted to 100 mL with phosphate-buffered saline (pH 7.5).

Synthesis of MIPs Based on QDs-Grafted COFs for Detection of QCA
In the present study, novel MIPs based on QDs-grafted COFs were prepared for a sensitive and selective detection of QCA. The structure and synthesis procedures of the QDs-grafted COFs are depicted in Figure 1. The preparation of TpPa COFs was carried out by the Schiff-base reactions of Tp with Pa with the formation of a crystalline framework, in the presence of acetic acid using 1:1 mesitylene/dioxane as the solvent. The chemical stability of the COFs was obtained by an irreversible enol-to-keto tautomerization. A crystalline framework was formed that enhanced chemical stability. During the preparation, a one-pot reverse microemulsion strategy was used. With the surfactant interactions of reverse microemulsion, the QDs were encapsulated in aqueous domains by magnetic stirring. With the further hydrolysis and condensation reactions of the silane reagent of APTES and TEOS, silica nanospheres were fabricated. APTES was used as the functional monomer to react with the target QCA using H-bonding reactions. The APTES-modified QDs complex was further reacted with the TpPa COFs using the Schiff-base reaction. The cavity sites were left after the QCA was extracted from the QDs-grafted COFs using methanol to disrupt the H-bonding. Prior to the removal of the QCA, the fluorescence intensity of the QDs-grafted COFs was 37.4%; after removal, it was 99.2% ( Figure S1). The repeatability of QDs-grafted COFs was investigated for a QCA concentration of 1 µmol L −1 over five measurements. Relative standard deviations (RSD) of 1.81%, 1.36%, 2.15%, 1.42%, and 2.23% were obtained by spectrofluorometry, indicating good optosensing repeatability. In addition, the MIPs based on QDs-grafted COFs had high fluorescence quantum yields due to the hydrophobic QDs and the polymerization method of reverse microemulsion. The quantum yield was calculated by using Rhodamine B as a reference and was 33.5 %.

Characterization of MIPs Based on QDs-Grafted COFs
The TEM images in Figure 2a showed that the QDs were uniformly distributed and sized. QDS size was about 3-5 nm in diameter. In Figure 2b, the SEM images of TpPa COFs revealed a tree-like morphology. We also used SEM to observe MIPs based on QDs-grafted COFs (Figure 2c), which revealed a rough morphology of highly uniform spheres. It indicated that when QCA was removed from the surface of the QDs-grafted COFs, it left recognition sites. The tree-like morphology has been seen in QDs-grafted COFs. A lot of CdSe/ZnS QDs were aggregated in the leaves of the COFs. In addition, the MIPs based on QDs-grafted COFs had high fluorescence quantum yields due to the hydrophobic QDs and the polymerization method of reverse microemulsion. The quantum yield was calculated by using Rhodamine B as a reference and was 33.5%.

Characterization of MIPs Based on QDs-Grafted COFs
The TEM images in Figure 2a showed that the QDs were uniformly distributed and sized. QDS size was about 3-5 nm in diameter. In Figure 2b, the SEM images of TpPa COFs revealed a tree-like morphology. We also used SEM to observe MIPs based on QDs-grafted COFs (Figure 2c), which revealed a rough morphology of highly uniform spheres. It indicated that when QCA was removed from the surface of the QDs-grafted COFs, it left recognition sites. The tree-like morphology has been seen in QDs-grafted COFs. A lot of CdSe/ZnS QDs were aggregated in the leaves of the COFs. The FTIR spectra of COFs ( Figure S2a) and MIPs based on QDs-grafted COFs ( Figure S2b) was used to judge the occurrence of the polymerization reactions. It can be seen from Figure S2a that the peaks at 1690 cm −1 and 1250 cm −1 were the stretching vibration of C=O and C-N, respectively, for the TpPa, and a strong peak at 1577 cm −1 was attributed to the C=C stretch in the keto form. Characteristic peaks at 1616 cm −1 were attributed to the decreased C=O stretching frequencies due to the covalent interactions of TpPa COFs and QDs. Therefore, the QDs were attached to the COF surface successfully.
An XRD pattern from MIPs based on QDs-grafted COFs is shown in Figure S2c. It revealed a cubic structure of the peaks at 26.46°, 44.10°, and 51.91°, which were similar to QD patterns. The XRD pattern of QDs was reported in previous work [24]. This confirmed that QDs were grafted on the COFs.
The TGA analysis of MIPs based on QDs-grafted COFs is shown in Figure S2d. The TGA data exhibited only 3.57% weight loss under 100 °C, while the loss was 11.75 % over the range 100-300 °C. There was no significant weight loss until up to 300 °C. Thus, the TGA data demonstrated excellent thermal stability of the QDs-grafted COFs.

Kinetic Adsorption
The kinetic adsorption of 1 μmol L -1 of QCA was determined (Figure 3). After 60 min of shaking, 87.9% of QCA was adsorbed, and the equilibrium was achieved after 90 min. Therefore, QCA readily found the imprinted cavities on the MIPs based on QDs-grafted COFs surfaces, and the QDs-grafted COFs exhibited good mass transport. The FTIR spectra of COFs ( Figure S2a) and MIPs based on QDs-grafted COFs ( Figure S2b) was used to judge the occurrence of the polymerization reactions. It can be seen from Figure S2a that the peaks at 1690 cm −1 and 1250 cm −1 were the stretching vibration of C=O and C-N, respectively, for the TpPa, and a strong peak at 1577 cm −1 was attributed to the C=C stretch in the keto form. Characteristic peaks at 1616 cm −1 were attributed to the decreased C=O stretching frequencies due to the covalent interactions of TpPa COFs and QDs. Therefore, the QDs were attached to the COF surface successfully.
An XRD pattern from MIPs based on QDs-grafted COFs is shown in Figure S2c. It revealed a cubic structure of the peaks at 26.46 • , 44.10 • , and 51.91 • , which were similar to QD patterns. The XRD pattern of QDs was reported in previous work [24]. This confirmed that QDs were grafted on the COFs.
The TGA analysis of MIPs based on QDs-grafted COFs is shown in Figure S2d. The TGA data exhibited only 3.57% weight loss under 100 • C, while the loss was 11.75% over the range 100-300 • C. There was no significant weight loss until up to 300 • C. Thus, the TGA data demonstrated excellent thermal stability of the QDs-grafted COFs.

Kinetic Adsorption
The kinetic adsorption of 1 µmol L −1 of QCA was determined (Figure 3). After 60 min of shaking, 87.9% of QCA was adsorbed, and the equilibrium was achieved after 90 min. Therefore, QCA readily found the imprinted cavities on the MIPs based on QDs-grafted COFs surfaces, and the QDs-grafted COFs exhibited good mass transport.

Optosensing QCA Based on MIPs
QCA adsorption quenched the MIPs based on QDs-grafted COFs fluorescence in accordance with the Stern-Volmer equation: where F0 and F are the fluorescence intensity of MIPs based on QDs-grafted COFs in the absence and presence of QCA, respectively, KSV is the Stern-Volmer constant, and [C] is the concentration of quencher QCA. The fluorescence was examined for various concentrations ranging over 0-0.1 mmol L −1 . In Figure 4, the fluorescence intensity was reduced in the presence of QCA increasing concentrations. The fluorescence was quenched because of the H-bonding between the QDs-grafted COFs and QCA at the imprinted recognition sites. Analysis with the Stern-Volmer equation was performed in the linear range 1-50 μmol L −1 , and the correlation was 0.9847. The detection limit was calculated as 0.85 μmol L −1 , which was the concentration needed to quench the MIPs based on QDs-grafted COFs at a level three times the standard deviation of the blank signal to divide the slope of the standard curve.

Optosensing QCA Based on MIPs
QCA adsorption quenched the MIPs based on QDs-grafted COFs fluorescence in accordance with the Stern-Volmer equation: where F 0 and F are the fluorescence intensity of MIPs based on QDs-grafted COFs in the absence and presence of QCA, respectively, K SV is the Stern-Volmer constant, and [C] is the concentration of quencher QCA. The fluorescence was examined for various concentrations ranging over 0-0.1 mmol L −1 . In Figure 4, the fluorescence intensity was reduced in the presence of QCA increasing concentrations. The fluorescence was quenched because of the H-bonding between the QDs-grafted COFs and QCA at the imprinted recognition sites. Analysis with the Stern-Volmer equation was performed in the linear range 1-50 µmol L −1 , and the correlation was 0.9847. The detection limit was calculated as 0.85 µmol L −1 , which was the concentration needed to quench the MIPs based on QDs-grafted COFs at a level three times the standard deviation of the blank signal to divide the slope of the standard curve.

Optosensing QCA Based on MIPs
QCA adsorption quenched the MIPs based on QDs-grafted COFs fluorescence in accordance with the Stern-Volmer equation: where F0 and F are the fluorescence intensity of MIPs based on QDs-grafted COFs in the absence and presence of QCA, respectively, KSV is the Stern-Volmer constant, and [C] is the concentration of quencher QCA. The fluorescence was examined for various concentrations ranging over 0-0.1 mmol L −1 . In Figure 4, the fluorescence intensity was reduced in the presence of QCA increasing concentrations. The fluorescence was quenched because of the H-bonding between the QDs-grafted COFs and QCA at the imprinted recognition sites. Analysis with the Stern-Volmer equation was performed in the linear range 1-50 μmol L −1 , and the correlation was 0.9847. The detection limit was calculated as 0.85 μmol L −1 , which was the concentration needed to quench the MIPs based on QDs-grafted COFs at a level three times the standard deviation of the blank signal to divide the slope of the standard curve.

Interference Experiments for QCA Selectivity
The performance of MIPs based on QDs-grafted COFs was further evaluated by interference experiments, as shown in Figure 5. Several kinds of structural analogs, CBX, MEQ, MQCA, and QLA were used to demonstrate the selectivity of QDs-grafted COFs. Equilibrium binding experiments were performed for initial concentrations of 1-50 µmol L −1 . The fluorescence intensity changes of MIPs based on QDs-grafted COFs with QCA were much greater than those without QCA. Because there were no tailor-made recognition sites on the MIPs based on QDs-grafted COFs formed QCA, less QCA was adsorbed. The coefficient ratio between MIPs based on QDs-grafted COFs with QCA and without QCA was 1.74. For the MIPs based on QDs-grafted COFs with QCA, it was mainly specific adsorption due to the selective recognition capability of tailor-made cavities, while it was mainly non-specific adsorption for the MIPs based on QDs-grafted COFs without QCA. Obviously, the MIPs based on QDs-grafted COFs had the highest binding affinity for QCA; however, there were different degrees of recognition for CBX, MEQ, MQCA, and QLA. The absorptions of CBX and QLA indicated no significant changes in fluorescence intensity, even though similar H-bonding can occur between the structural analogs and the APTES-modified QDs. The size and shape of MEQ and MQCA were more similar to that of QCA, and therefore it is possible that the molecule of MEQ and MQCA fitted neatly into the cavity of the MIPs and formed a hydrogen bond with the amino group of APTES on the MIPs, causing the MIPs to exhibit cross-binding ability. However, because of the lack of spatial complementarity, the degree of rebinding of the MIPs with MEQ and MQCA was smaller than that with QCA.

Interference Experiments for QCA Selectivity
The performance of MIPs based on QDs-grafted COFs was further evaluated by interference experiments, as shown in Figure 5. Several kinds of structural analogs, CBX, MEQ, MQCA, and QLA were used to demonstrate the selectivity of QDs-grafted COFs. Equilibrium binding experiments were performed for initial concentrations of 1-50 μmol L −1 . The fluorescence intensity changes of MIPs based on QDs-grafted COFs with QCA were much greater than those without QCA. Because there were no tailor-made recognition sites on the MIPs based on QDs-grafted COFs formed QCA, less QCA was adsorbed. The coefficient ratio between MIPs based on QDs-grafted COFs with QCA and without QCA was 1.74. For the MIPs based on QDs-grafted COFs with QCA, it was mainly specific adsorption due to the selective recognition capability of tailor-made cavities, while it was mainly non-specific adsorption for the MIPs based on QDs-grafted COFs without QCA. Obviously, the MIPs based on QDs-grafted COFs had the highest binding affinity for QCA; however, there were different degrees of recognition for CBX, MEQ, MQCA, and QLA. The absorptions of CBX and QLA indicated no significant changes in fluorescence intensity, even though similar H-bonding can occur between the structural analogs and the APTES-modified QDs. The size and shape of MEQ and MQCA were more similar to that of QCA, and therefore it is possible that the molecule of MEQ and MQCA fitted neatly into the cavity of the MIPs and formed a hydrogen bond with the amino group of APTES on the MIPs, causing the MIPs to exhibit cross-binding ability. However, because of the lack of spatial complementarity, the degree of rebinding of the MIPs with MEQ and MQCA was smaller than that with QCA. The mechanism may involve the interactions with the shape, size, and functionality of QCA. In addition, charge transfer from the luminescence QDs to QCA is caused by fluorescence quenching, as suggested by Tu et al. [25]. Furthermore, from the UV absorption spectrogram, the UV absorption band of QCA is near to the band-gap of the luminescence QDs, as revealed by UV-absorption spectra in Figure S3. Charges in the CdSe/ZnS QDs conduction bands can be transferred to the lowest The mechanism may involve the interactions with the shape, size, and functionality of QCA. In addition, charge transfer from the luminescence QDs to QCA is caused by fluorescence quenching, as suggested by Tu et al. [25]. Furthermore, from the UV absorption spectrogram, the UV absorption band of QCA is near to the band-gap of the luminescence QDs, as revealed by UV-absorption spectra in Figure S3. Charges in the CdSe/ZnS QDs conduction bands can be transferred to the lowest unoccupied molecular orbital of QCA. Because there was no spectral overlap between QCA absorption and QD emission, energy transfer is unlikely.

Stability Analysis
To further analyze the analytical performance of the MIPs based on QDs-grafted COFs, stability was investigated ( Figure S4). At room temperature, the MIPs based on QDs-grafted COFs were stable for nearly 60 days. This indicated that the MIPs based on QDs-grafted COFs possessed sufficient stability for QCA analysis.

Detection of QCA in Meat and Feed Samples
QCA detection in meat and feed samples was performed in a recovery experiment using 5, 10, and 20 µmol L −1 QCA. The recovery ranged from 93.8 to 101.2%, as shown in Table 1, indicating that the MIPs based on QDs-grafted COFs can be used to detect QCA accurately in complex samples.

Conclusions
This study reports the clear advantage of MIPs coated on QDs-based COFs for simple and rapid optosensing of QCA. The MIPs were prepared by a one-pot water-in-oil reverse microemulsion strategy. Here, QDs were used as an optical response element, and the COFs were used as an optosensing platform to combine the MIPs to provide high selectivity. The high sensitivity derived from the dual-signal amplification by high QD fluorescence quantum yields and the large COF-specific surface area. The results demonstrate that the MIPs based on QD-grafted COFs arrays detected QCA with high sensitivity and selectivity. The simple preparation, good specificity, low cost, good repeatability, and rapid detection make this approach attractive for hazard analysis.

Conflicts of Interest:
The authors declare no competing financial interest.