Highly sensitive and selective detection of naproxen via molecularly imprinted carbon dots as a fluorescent sensor

The overuse and inappropriate discharge of naproxen, a common anti-inflammatory medication and an emerging contaminant in water, is detrimental to human health and bodies of water. Here, we design a fluorescent sensor based on molecularly imprinted carbon dots (CDs) for highly selective detection of trace amounts of naproxen. The CDs were encapsulated into the pores of silica through a sol–gel based method and provide fluorescent signal. After removal of the template molecules, a molecularly imprinted polymer layer was formed and the fluorescence of the CDs sensor was selectively quenched by naproxen. A detection limit of as low as 0.03 μM and a linear range of 0.05–4 μM for detecting naproxen in aqueous solution were obtained. High recoveries of naproxen levels in waste water and urine samples for practical application were also achieved. In addition, the accurate detection performance of sensor was maintained during the UV degradation of naproxen.


Introduction
Naproxen, one of the four most commonly used nonprescription drugs, is oen used to relieve inammation, fever, and pain. 1 However, prolonged and excessive use of naproxen leads to side effects such as gastrointestinal bleeding, stomach ulcers, and increased risk of heart disease. 2 In addition, naproxen can be released into domestic sewage processing system via bodily excretions, and linger within natural water bodies due to its high chemical stability. The existence of naproxen impacts bivalve lipid peroxidation, microbial community structure, and cause antibiotic resistance. 3 The high polarity and low volatility of naproxen further promote expansion of pollution in water. Therefore, developing rapid, simple, and effective detection methods of naproxen is highly desirable. Traditional approaches for analysing the existence of naproxen include electrochemical techniques, 4,5 high performance liquid chromatography(HPLC), 6 ow injection analysis, 7 magnetic-solid phase extraction (M-SPE), 8 capillary electrophoresis (CE), 9 and absorption spectrophotometry. 10 These methods, however, demand specialized training, costly instruments, and complex sample preprocessing. Fast detection of naproxen in water is of great signicance. As an intriguing alternative to analytical methods, uorescence sensing has attracted increasing attentions in recent years, due to its high sensitivity and fast response. Fluorescent sensors have been widely developed in the detection of pollutants such as Hg 2+ , 11 Pb 2+ , 12 Cd 2+ , 13 Ag + , 14 peruorooctane sulfonate, 15 bisphenol A, 16 and phosphate. 17 Compared to organic uorophores, inorganic quantum dots possess desirable optical properties of broad excitation spectrum, narrow emission peak, and light bleaching resistance. 18 Traditional quantum dots, such as CdS 19 and CdTe, 20 are commonly based on rare metals. Recently, carbon dots have been widely exploited for applications since they are easy to prepare, costeffective and environmentally friendly. Fluorescent carbon dots in particular benet from their light stability, photoluminescence, and size-dependent emission wavelength, 21 and have been extensively utilized in the eld of photocatalysis, 22 cell imaging, 23 biosensors, 24-26 and more. Carbon dots have been developed rapidly for uorescent sensors, 27,28 however, they still largely serve as uorescence carrier with limited selectivity. 29,30 To improve the selectivity of carbon dots-based sensors, molecular imprinting technique has been introduced to provide recognition sites for the analyte. Molecularly imprinted polymers (MIP) are the functionalized polymers formed in the copolymerization process of functional monomers, crosslinking agents and target template molecules. 31 Aer removal of the template molecules, imprinted cavities of the exactly same shape and size complementary to target molecules are obtained, achieving specic recognition ability towards target molecules. [32][33][34][35] Given the high specicity, molecularly imprinted polymers have found great potentials in sensor applications. [36][37][38] Recent advances in combining uorescent quantum dots with molecular imprinting polymer have dramatically improved the selectivity of quantum dot sensors, 39,40 in the case of detecting metronidazole, 41 amoxicillin, 42 promethazine hydrochloride, 43 cocaine, 44 and doxorubicin. 45 However, to the best of our knowledge, detection of naproxen based on molecularly imprinted carbon dots has not been reported to date.
Herein, a highly selective uorescent sensor (MIP@CDs) for naproxen was developed, utilizing carbon dots as the uorophore and molecularly imprinted polymer to provide the recognition sites. As shown in Scheme 1, carbon dots were embedded in silica imprinted lm through one-step sol-gel polymerization. Aer the removal of template molecules, the recognition sites were then released. The uorescence of carbon dots was selectively quenched by naproxen, with a low detection limit of 0.03 mM and a linear range of 0.05-4 mM.

Characterization and measurements
UV-vis absorption spectra were collected with a ultravioletvisible spectrophotometer (UV-2450, Shimadzu). Photoluminescence spectra were recorded using a uorescence spectrometer (Fluoro-Max-4, Horiba Scientic). Scanning electron microscopy (SEM) images were obtained using an emission scanning electron microscope (S 4800, Hitachi). Fourier transform infrared (FT-IR) measurement was carried out with a FT-IR spectrometer (Nicolet iS10, Thermo Scientic). Highperformance liquid chromatography (HPLC) was carried out using a UltiMate 3000 pump (Thermo Scientic). All separations were conducted on a Thermo Scientic ZORBAX SB-C18 (5 m, 4.6 mm Â 150 mm, Agilent) column. Methanol and 0.2% formic acid solution (80/20 volume ratio) comprised the mobile phase. The detection wavelength was 242 nm, the ow rate was 0.2 mL min À1 , and the column temperature was set at 40 C. The sample was ltered through a membrane lter with a diameter of 0.22 mm before testing. All optical measurements were carried out at ambient temperature.

Synthesis of CDs
Carbon dots were prepared by hydrothermal method according to the reported literature with slight modications. 46 Briey, 0.5 g citric acid anhydrous was dissolved in 10 mL deionized water using ultrasonator and degased in the atmosphere of nitrogen for 15 min. 2 mL AEAP was then added under vigorous stirring for 10 min. The solution mixture was transferred to a 50 mL Teonlined stainless steel autoclave and was kept at 200 C for 2 h. Aer the reaction was completed, the autoclave was cooled to room temperature. The resultant dark brown transparent liquid was ltered through ltration membrane lter (0.22 mm) and washed with petroleum ether three times. The obtained CDs were stored in the refrigerator at 4 C for further use.

Synthesis of the MIP@CDs
100 mg naproxen was dissolved in 30 mL ethanol by using ultrasonator, 0.4 mL APTES and 0.5 mL CDs were added to the solution respectively under stirring for 10 min. 1 mL TEOS was added dropwise, followed by the addition of 1 mL ammonia water. The solution mixture was then kept stirring for 24 h in the dark. The resultant product was then collected by centrifugation and washed thoroughly with methanol until no naproxen template molecules were detected from the washing solvent using ultraviolet-visible spectrophotometer. The obtained molecularly imprinted carbon dots (MIP@CDs) were then dried in vacuum oven for 24 h to get white powder product and stored in the refrigerator at 4 C. The non-imprinted CDs (NIP@CDs) were also prepared identically without adding any naproxen.

Detection of naproxen
The obtained white powder of MIP@CDs was dispersed into ultrapure water to make the stock solution (280 mg L À1 ). 500 mL of stock solution and 200 mL of phosphate buffer saline (pH ¼ 5.5, 0.1 M) were added to the cuvette, the solution volume was Scheme 1 Illustration of the preparation and fluorescence response of MIP@CDs to naproxen. adjusted to 3 mL using ultrapure water. Then various amount of naproxen solution (50 mM) was introduced to the cuvette gradually. Aer incubation for 10 min, the uorescence spectra was collected under the excitation wavelength of 360 nm.

Detection of naproxen in real water samples
Tap water, surface river water, and prepared human urine samples were selected to investigate the practical application for naproxen detection of MIP@CDs. Tap water sample was directly taken form the tap in the lab, and the surface river water sample was collected from local Caohejing River (Shanghai, China). Both of the water samples were ltered through microporous membrane (0.22 mm) to remove particulate matter before further use and monitored by HPLC to conrm the inexistence of naproxen. The urine sample was prepared according to the recipe in Table S1. † Different concentrations of naproxen were prepared using those water samples to study the uorescent sensing behaviour of MIP@CDs. For UV degradation experiments, 30 mL of naproxen solution (10 mg L À1 , 20 mg L À1 , and 10 mg L À1 , respectively) was placed in Petri dishes under UV light, and samples were taken every 15 min for a total of 3 h.

Characterization of the structure and spectra properties of MIP@CDs
The morphological structures of MIP@CDs and NIP@CDs were characterized by SEM and TEM techniques. Both MIP@CDs and NIP@CDs exhibit aggregated spherical structure with average diameters of about 50 nm (Fig. S1 †). In order to further conrm the formation of imprinted polymer layer, FTIR spectra of CDs, MIP@CDs and NIP@CDs were collected. As shown in Fig. 1, the characteristic peaks at 3357 cm À1 and 795 cm À1 for CDs can be assigned to the O-H stretching and bending vibration, accounting for the formation of hydrogen bond on the surface of CDs with AEAP. The vibration signals of C]O at 1482 cm À1 and 1651 cm À1 are from citric acid anhydrous. The peak at 1077 cm À1 is assigned to the C-O stretching vibration. All these characteristic peaks demonstrate that the obtained CDs are highly hydrophilic due to the rich carboxyl and hydroxyl groups on the surface. The FTIR spectra of MIP@CDs and NIP@CDs are of no signicant difference considering the similar preparation procedures and starting materials. The vibration signals of Si-O are found at 792 cm À1 and 452 cm À1 . The characteristic peaks at 2937 cm À1 and 1482 cm À1 can be assigned to the stretching vibration of CH 2 -N and C-H, respectively, conforming the existence of amide group. The peak at 1560 cm À1 is from the stretching vibration of N-H from aminopropyl group. 47 These experimental results reveal that the silica layer was formed on the surface of both MIP@CDs and NIP@CDs. XPS spectra of MIP@CDs and NIP@CDs both exhibit ve peaks at 102.1, 153.1, 284.4, 398.9 and 531.7 eV, which are attributed to Si 2p, Si 2s, C 1s, N 1s, and O 1 s, respectively, conrming the existence of the silica layer (Fig. S2 †).
The optical properties of CDs were examined by UV-vis absorption and uorescence spectra (Fig. S3 †). The as-prepared CDs exhibit an absorption peak around 360 nm and a maximum emission peak around 448 nm (Fig. S3a †). Furthermore, the maximum emission peak showed red-shi with increased excitation wavelength, demonstrating the unique characteristic of CDs. Aer being functionalized with silica layer, both MIP@CDs and NIP@CDs show the same absorption peak at 360 nm and emission peak at 448 nm as CDs (Fig. S3b and c †), thus the imprinted polymer layer doesn't alter the optical properties of CDs. The stability of MIP@CDs was also tested. As shown in Fig. S3d, † the emission intensity of MIP@CDs at 448 nm remained roughly unchanged within 14 days, illustrating an excellent stability.
Preliminary evaluation on the selective recognition of naproxen by MIP@CDs was also carried out. As shown in Fig. 2a, compared to NIP@CDs, the uorescence intensity of MIP@CDs before removing the template naproxen molecules was quite weak. When naproxen was washed away by methanol, the uorescence was almost completely recovered. The emission intensity dropped again when naproxen (14 mL, 50 mM) was added. However, the addition of naproxen doesn't change the uorescence intensity of NIP@CDs. These results demonstrate that MIP@CDs can detect naproxen through the change of uorescence intensity, and NIP@CDs without imprinted polymer show no response to naproxen.

Optimization of detection conditions
In order to achieve high sensitivity for naproxen recognition, detection conditions, such as pH value, incubation time and concentration of MIP@CDs solution, were optimized. The effect of pH on the uorescence quenching behaviour of naproxen for MIP@CDs was investigated by using PBS buffer to adjust pH. As shown in Fig. 2b, the uorescence quenching ratio at 448 nm of MIP@CDs by naproxen rst increased with increasing pH values from 4.4 to 5.5, then decreased from 5.5 to 9.5. The reduced sensitivity at lower or higher pH is ascribed to the proton protonation in acid environment and ionization in alkaline environment, both of which will cause structural damage of the recognition cavities in the silica shell, thus leading to a decreased response sensitivity to naproxen. 48 The effect of incubation time was also explored, Fig. 2c shows that the uorescence quenching efficiency by naproxen to MIP@CDs reached a maximum aer 10 min and remained relatively stable with further prolonging time to 30 min, which means a relatively completed binding between naproxen and the imprinted cavities of MIP@CDs occurred within 10 min. The amount of MIP@CDs used for the quenching experiment was also optimized (Fig. 2d). The highest uorescence quenching ratio was observed at the concentration of 200 mg L À1 within the range from 50 mg L À1 to 600 mg L À1 . On the basis of the above experiments for optimization of detection conditions, pH of 5.5, the incubation time of 10 min and 200 mg L À1 of MIP@CDs were selected for the subsequent studies.

Analytical performance for the detection of naproxen
The analytical performance of MIP@CDs for the detection of naproxen was investigated under the optimized conditions. As shown in Fig. 3a, the uorescence spectra of MIP@CDs were collected with different amounts of naproxen solution (50 mM) to investigate its quenching behaviour. The uorescence intensity of MIP@CDs decreased distinctly with increasing the amount of naproxen, whereas no obvious uorescence intensity change was observed for NIP@CDs upon adding naproxen. The better response sensitivity of MIP@CDs attributes to the existence of imprinted cavities in the silica shell, which provides selective recognition and binding affinity to naproxen molecules. 49 The quenching efficiency of naproxen molecules was then quantied by the following Stern-Volmer equation: where F 0 and F are the uorescent intensities at 448 nm of MIP@CDs in the absence and presence of naproxen, respectively, C is the concentration of naproxen, and K SV is the Stern-Volmer quenching constant. A good linear relationship between  F 0 /F (quenching ratio) and different concentration of naproxen was observed, and the linear regression equation was obtained as F 0 /F ¼ 1 + 0.815C [naproxen] , where K SV-MIP was determined as 8.15 Â 10 5 M À1 . As a comparison, the Stern-Volmer equation for NIP@CDs was also determined as F 0 /F ¼ 1 + 0.162C[ naproxen ] with K SV-NIP of 1.62 Â 10 5 M À1 (Fig. 3b). The imprinting factor (IF), dened as IF ¼ K SV-MIP /K SV-NIP , is an important parameter to evaluate the selectivity of imprinted polymer. Here IF was calculated to be 5, which suggests an excellent selectivity towards to naproxen by MIP@CDs. Limit of detection for MIP@CDs was also calculated to be 0.03 mM based on the ratio of signal-to-noise of 3, which is close to or lower than that of the existing sensors for naproxen, as shown in Table S2. † 50,51 These detailed studies demonstrate that the developed MIP@CDs sensor shows a wide linear range response (0.05 $ 4 mM) with a low limit of detection (0.03 mM) to naproxen. Given the strong hydrogen bonding reaction between the carboxyl group of naproxen and amino group of the imprinted cavities, the uorescence quenching mechanism may possibly be speculated as energy transfer or charge transfer between naproxen and MIP@CDs. Since there was no obvious overlap between the absorption spectra of naproxen and the emission spectrum of MIP@QDs ( Fig. S4a †), energy transfer cannot happen to induce uorescence quenching. As previously reported, 52 charges in the conductive bands of QDs could transfer to a lowest unoccupied orbital of analyte molecules, thus charge transfer is considered as the main reason for uorescence quenching. Lifetimes of MIP@CDs in the absence and presence of naproxen were also measured (Fig. S4b †). The lifetime of MIP@CDs decreased from 17.17 ns to 14.34 ns aer adding naproxen, which means dynamic quenching occurred. The collisional encounters and weak bonding reaction between MIP@CDs and naproxen depopulates the excited state of MIP@CDs, causing the decreasing of lifetime.

Selectivity and anti-interference
The selectivity and anti-interference performances of MIP@CDs were also investigated. Functional and structural analogs (ibuprofen ketoprofen, and fenoprofen) were selected for the selectivity experiments. As shown in Fig. 4a, MIP@CDs exhibited a much higher uorescence quenching efficiency towards naproxen than that of its analogs, due to the existence of imprinted cavities with the exact shape and size complementary to template molecules. Meanwhile, NIP@CDs without imprinted cavities displayed similar inefficient quenching behaviours for naproxen and analogs, which highlights the necessity of imprinted cavities for the selectivity performance. Ibuprofen was chosen for competitive binding experiment to evaluate the anti-interference ability of MIP@CDs. With increasing the C IBU /C NAPROXEN ratio from 0 to 4, the uorescence quenching ratio of MIP@CDs remained unchanged by naproxen, illustrating no competitive binding from ibuprofen and high affinity of naproxen to MIP@CDs.
Considering the coexistence of interfering substance in real water samples, the uorescence quenching performances of naproxen to MIP@CDs in the presence of common metal ions, inorganic salt ions, and potentially existing organics (such as Cu 2+ , Ca 2+ , K + , Na + , Mg 2+ , Cl À , SO 4 2À , NO 3 À , NO 2 À ), were investigated. As illustrated in Fig. 4b, the quenching ratios of naproxen displayed no obvious change while coexisting with interfering substances. These results further demonstrated that MIP@CDs exhibited excellent anti-interference property and could be applied to selectively detect naproxen molecule in practical applications.

Practical applications
To assess the practical application of MIP@CDs in real samples, recovery studies were performed by spiking the real samples (tap water, river water, and simulative human urine) with various amounts of naproxen. The averaged recoveries with  relative standard deviation (RSD, n ¼ 3) for each concentration are listed in Table 1. Satisfactory recoveries with low RSDs were obtained for all three real samples, demonstrating the prepared MIP@CQD sensor can be utilized for accurate determination of trace naproxen in complicated real samples. Naproxen may also generate intermediates when exposed to sunlight irradiation (UV degradation) in natural water, which complicates the analysis of its existence. Thus, we also investigated the detection performance of MIP@CDs for naproxen under UV light. As shown in Fig. 5, the concentration of naproxen solution decreased under UV irradiation for all three samples, and the degradation curve is accorded with the rstorder degradation curve. There is no signicant difference between HPLC and MIP@CDs test results for detection of naproxen concentration, indicating that the intermediates produced during the UV degradation have no effects on the uorescence response of MIP@CDs to naproxen.

Conclusions
In summary, we have successfully developed a uorescent CDs sensor for specic recognition and sensitive detection of naproxen molecules based on molecularly imprinted polymer. The uorescence of the obtained MIP@CDs was selectively quenched by naproxen, and a low LOD of 0.03 mM with a linear range of 0.05-4 mM was achieved. Anti-interference performance in the presence of coexisting substances and recovery studies of the developed MIP@CDs sensor in real samples were also investigated to reveal a potential application for practical naproxen monitoring and evaluation in the environment.

Conflicts of interest
There are no conicts to declare.