Facile synthesis of carbon nitride quantum dots as a highly selective and sensitive fluorescent sensor for the tetracycline detection

Enhanced blue fluorescent carbon nitride quantum dots (g-C3N4QDs) were synthesized by a simple solvothermal “tailoring” process from bulk g-C3N4 and analyzed by various characterization methods. The as-obtained g-C3N4QDs were successfully applied in the determination of tetracycline (TC) with a good linear relationship in the range of 0.23–202.70 μM. The proposed fluorescent sensor shows excellent stability, good repeatability, high selectivity and outstanding sensitivity to TC with a low detection limit of 0.19 μM. The fluorescence quenching mechanism of g-C3N4QDs with TC was mainly governed by static quenching and the inner filter effect. The method was successfully applied to monitor TC in tap water and milk powder samples.


Introduction
Since its re-discovery in 1990s, graphite carbon nitride (g-C 3 N 4 ) as a metal-free semiconductor material has been widely explored and used in sensing, 1-6 catalysis, 7,8 uorescence imaging, 9,10 and cancer treatment 11 due to its unique electronic structure, excellent chemical and thermal stability, and good biocompatibility. Usually, g-C 3 N 4 is prepared by the hightemperature pyrolysis of nitrogen-rich precursors such as melamine, cyanamide, and dicyandiamide. 12 However, the synthetic bulk g-C 3 N 4 materials generally have low specic surface area, poor luminescence performance, and insolubility in most solvents, which limit their practical applications. Therefore, it is rewarding to search for solutions to overcome the shortcomings and limitations of bulk g-C 3 N 4 materials. Quantum dots (QDs) as quasi-zero dimensional nanomaterials have attracted wide attention because of quantum effects, and possess exotic features, such as large specic surface area, good water solubility, and unique optical and electronic properties superior to those of large particles. Currently, researchers have developed various types of QDs including metallic QDs, 13 nonmetallic QDs, 14 and composite QDs 15 among others and applied them in electroluminescence devices, solar cells, photocatalysis, imaging and sensing based on their extraordinary optical properties: electrochemiluminescence, phosphorescence, and uorescence. 16 Thus, exploration of g-C 3 N 4 QDs may provide the promising properties and applications of g-C 3 N 4 materials. As a result, a series of uorescent g-C 3 N 4 QDs have been fabricated and used as promising uorescent sensor materials in recent years. [17][18][19] For instance, K. Patir et al. applied g-C 3 N 4 QDs as a photoluminescent sensor for Hg 2+ detection. 19 Up to now, several routes such as hydrothermal, solvothermal, microwave and chemical etching, which are mainly categorized into top-down or bottom-up approaches, have been employed to synthesize QDs. 16,20 However, the present methods of g-C 3 N 4 QDs whether top-down or bottom-up synthesis are usually complicated and difficult to control. Therefore, developing highly efficient and facile methods to synthesize g-C 3 N 4 QDs is of great importance. In this paper, we report a facile and environmentally friendly solvothermal "tailoring" method to synthesize uorescent g-C 3 N 4 QDs. Moreover, the application of g-C 3 N 4 QDs still needs further exploration.
Tetracycline (TC) is one of the major broad-spectrum antibiotics, which can inhibit a wide variety of bacteria, and has been extensively used in human therapy, animal disease control and agricultural feed additives because of its excellent therapeutic effect and low cost. 21 Nevertheless, the absorption and metabolism of TC in the body represent only a small proportion, and about 30 to 90% of TC is released into excreta in the form of parent compounds or metabolites. 22 Meanwhile, TC commonly has a long halife in natural environments. 21 Consequently, the abuses of TC results in residues widely present in animal products, soil, surface water, drinking water and groundwater, which would inhibit aquatic species growth and development. 23 In addition, TC residues would gradually accumulate in the food chain and nally affect the health of human beings. 23 Therefore, the rapid and accurate quantitative determination of TC concentration in natural environments is very necessary. Common methods for TC detection include the microbiological method, 24 high performance liquid chromatography, 25 enzyme immunoassay, 26 and capillary electrophoresis. 27 However, these detection methods have some disadvantages, such as poor detection sensitivity and selectivity, complex sample preparation, tedious operation, costly equipment, being time-consuming and using toxic reagents. Hence, developing a simple, inexpensive, eco-friendly, and rapid method with high selectivity and sensitivity is meaningful. The uorescent sensing method based on observation of direct emission quenching as the sensing signal is considered to have the advantages of high sensitivity and selectivity, simple operation and repeatability. Therefore, we aim to develop g-C 3 N 4 QDs as an efficient uorescent sensor to detect TC. Herein, we succeeded in synthesizing g-C 3 N 4 QDs with low costs, water solubility and bright blue uorescence via a facile and environmentally friendly solvothermal "tailoring" method. The prepared g-C 3 N 4 QD material displayed good stability, reproducibility, high selectivity and sensitivity for TC determination on the basis of the uorescence quenching method (Scheme 1), and was also successfully applied to detect TC in real water and milk powder samples. Moreover, the uorescence quenching mechanism of g-C 3 N 4 QDs was proposed.

Preparation of bulk g-C 3 N 4
Firstly, the bulk g-C 3 N 4 was synthesized by a simple thermal polymerization of melamine according to our previous work. 28 Typically, 5 g of melamine was heated to 500 C in a tube furnace at a ramping rate of 10 C min À1 and kept at this temperature for 2 h and then continued to heat up to 520 C at the same heating rate and kept for another 2 h. Aer naturally cooling to room temperature, the obtained yellow product was ground into a homogeneous powder.

Preparation of g-C 3 N 4 QD S
The g-C 3 N 4 QDs were prepared by a solvothermal method according to ref. 29. The bulk g-C 3 N 4 (0.10 g) was mixed with ethylene glycol (15 mL) and ammonia (15 mL). The mixture was transferred into a Teon-sealed autoclave and kept at 180 C for 12 h. The resultant product was cooled to room temperature and ltered with a 0.22 mm membrane. Finally, the obtained ltrate containing highly dispersed g-C 3 N 4 QDs was stored at 4 C before use.

Characterization
The X-ray diffraction (XRD) patterns were obtained with an X-ray diffractometer (Bruker, Germany) using Cu K a radiation (40 kV, 30 mA). A transmission electron microscope JEM-2100F (Japan JEOL Ltd.) was used to record transmission electron micrographs (TEM) at an acceleration voltage of 200 kV. The XPS measurements were carried out on an ESCALAB 250Xi spectrometer (Thermo Scientic, USA) with a pass energy of 30 eV and 100 W. The absolute quantum yield is measured with a C11347-11 Quantaurus-QY absolute quantum yield measurement instrument (Hamamatsu, Japan). Fourier transform infrared spectra (FT-IR) were collected with a Thermo Scientic Nicolet 380 FT-IR spectrometer with a resolution of 4 cm À1 . The ultraviolet-visible (UV-Vis) absorption spectra were measured on a TU-1901 UV-Vis spectrophotometer (UV). The photoluminescence (PL) spectra were recorded on a Shimazu RF-5301 PC uorescence spectrophotometer.

Measurement of uorescence quantum yield
The photoluminescence absolute quantum yield (QY) of the prepared g-C 3 N 4 QDs was measured using an absolute PL quantum yield spectrometer with an integrating sphere (C11347-11, Hamamatsu, Japan) under excitation with a 150 W xenon light source at 325 nm. The test principle and method were referred to ref. 30 and 31. The absolute uorescence quantum yield ¼ the number of emitted photons/the number of absorbed photons. 30 2.6 Fluorescence detection of TC 0.5 mL of TC solution with various standard concentrations was mixed with 0.5 mL of g-C 3 N 4 QD solution and then diluted with deionized water to 5 mL (the nal concentration was 0.3 mg mL À1 ). Aer being stirred thoroughly and placed for 15 min at room temperature, the uorescence intensity of the mixture was collected under an excitation wavelength of 325 nm. To explore the selectivity, PL of the solution containing g-C 3 N 4 QDs and other substances (CTC, OTC, DTC, CIP, AMX, SDZ, CAM, Mg 2+ , Ba 2+ , K + , etc.) was detected using the same method. Moreover, PL intensities of the mixture of g-C 3 N 4 QDs and TC in the presence of other substances were also recorded under identical conditions. Scheme 1 Schematic illustration of the preparation of g-C 3 N 4 QDs and their fluorescence quenching property triggered by tetracycline.

Fluorescence detection of TC in tap water and milk powder samples
The tap water samples were obtained from our lab, ltered through 0.22 mm membrane lters for further analysis and added with a series of different concentration levels of TC. Aer that, the solutions were analyzed with the same proposed method.
Milk powder samples were obtained from the local supermarket. The procedure for pretreating actual samples was performed on the basis of the reported works with little modication as follows. 32,33 Firstly, 2 g of milk powder was diluted to 20 mL with ultrapure water. Then 2 mL of 10% trichloroacetic acid (w : v, in water) was added into the sample solution and sonicated for 30 min to precipitate proteins and dissolve other organics in the matrix. Aer that, the mixture was centrifuged at 8000 rpm for 10 minutes to separate the precipitate. Thereaer, the supernatant was ltered with a 0.22 mm lter membrane to remove lipids, and the ltrate was taken for further analysis.

Physicochemical characterization of g-C 3 N 4 QDs
TEM and high resolution TEM were employed to observe the morphology of g-C 3 N 4 QDs. As presented in Fig. 1a, the g-C 3 N 4 QDs are well mono-dispersed and have a relatively uniform spherical shape. The particle sizes are mostly in the range of 2-6 nm with a narrow distribution, and the average particle size is approximately 3.5 nm. Fig. 1b shows the HRTEM images of g-C 3 N 4 QDs, and the spacing of the lattice fringe is 0.24 nm corresponding to the (100) plane of g-C 3 N 4 . 34 Fig. 1c displays the XRD patterns of bulk g-C 3 N 4 and g-C 3 N 4 QDs. As shown, the diffraction peaks of g-C 3 N 4 QDs are in good agreement with those of bulk g-C 3 N 4 , indicating that they have the same basic crystal structure. The strong peak at 27.5 is corresponding to the (002) plane, and a relatively weak diffraction peak at 13.1 is attributed to the (100) plane. These two characteristic diffraction peaks reect the inter-planar stacking of the aromatic ring structure and the in-planar tri-s-triazine unit packing motif, respectively. 35 Moreover, compared with bulk g-C 3 N 4 , the relative intensity of the (100) peak for g-C 3 N 4 QDs is much weaker, indicating that the in plane repeated unit structure may be damaged to a certain extent aer solvothermal treatment. 36 The FT-IR spectra of bulk g-C 3 N 4 and g-C 3 N 4 QDs are presented in Fig. 2. For both samples, they have an absorption peak located at 810 cm À1 belonging to the out of plane breathing vibration modes of heptazine heterocyclic rings. 37,38 Several intense bands in the range of 1200-1650 cm À1 are also detected for both bulk g-C 3 N 4 and g-C 3 N 4 QDs, which are attributed to typical stretching vibrations of tri-s-triazine units. The results indicated that the basic surface functional structures of g-C 3 N 4 are not changed in the process of solvothermal "tailoring". Moreover, the broad bands in the 3000-3500 cm À1 region assigned to stretching vibrations of N-H (À3100 cm À1 ) and O-H (À3300 cm À1 ) of adsorbed water are observed for the two samples. 11 Nevertheless, the intensity of the N-H and O-H peaks for g-C 3 N 4 QDs is much stronger than that of g-C 3 N 4 . And these hydrophilic groups are probably benecial to increase the water solubility of g-C 3 N 4 QDs. In addition, compared with bulk g-C 3 N 4 , two new peaks at ca. 1731 and 1032 cm À1 appeared, which are ascribed to C]O stretching vibration and C-O vibration in the C-O-C group, respectively. 39,40 It seems possible that oxygen atoms may be introduced into g-C 3 N 4 QDs.
XPS was performed to further investigate the chemical composition and structure information of g-C 3 N 4 QDs. The survey scan of the XPS spectrum displays three peaks located at 284.8, 398.7 and 532.8 eV (Fig. 3a), which are ascribed to C 1s, N 1s, and O 1s, correspondingly. In the high resolution N 1s spectrum (Fig. 3b), g-C 3 N 4 QDs possess two types of nitrogen species: N-(C) 3 (399.7 eV) and C-N]C (398.7 eV). 1,41 The deconvolution of the C 1s spectrum (Fig. 3c) presents four peaks at 284.8, 286.3, 288.1, and 289.5 eV, which could be ascribed to graphitic carbon (sp 2 C]C or sp 3 C-C), C-NH x (x ¼ 1, 2), N-C] N, and C-O. 12,[42][43][44] For the O 1s spectrum (Fig. 3d), it can be deconvoluted into two peaks at 531.8 eV and 532.8 eV, which are related to C-O-C and hydroxyls of adsorbed water, respectively. 45,46 The XPS results further conrm the presence of oxygen impurities in the g-C 3 N 4 QDs, which may promote the formation of intermolecular hydrogen bonds and thus enhance water solubility. Fig. 1 TEM (a), size distribution (a, inset) and HRTEM (b) images of g-C 3 N 4 QDs; XRD patterns (c) of the prepared bulk g-C 3 N 4 and g-C 3 N 4 QDs. Fig. 2 FT-IR spectra of g-C 3 N 4 and g-C 3 N 4 QDs. Fig. 4a shows the UV-Vis absorption and uorescence spectra of g-C 3 N 4 QD solution. The large peak from 200 to 270 nm with a shoulder centered at 262 nm is from p-p* electronic transition of the tri-s-triazine ring in graphitic carbon nitride. 47,48 Furthermore, another broad peak appears between 300 and 340 nm because of n-p* electronic transition of the C-N bond in the heptazine heterocycle structure. 45 The excitation spectrum exhibits a maximum peak at 325 nm, and an emission peak emerges at 394 nm when g-C 3 N 4 QDs are excited at this maximum excitation wavelength. Moreover, the g-C 3 N 4 QD solution shows bright blue uorescence under irradiation with 325 nm UV light (Fig. 4a inset). Compared with our previous assynthesized few-layer g-C 3 N 4 (ref. 28) and much of g-C 3 N 4 materials, 49,50 the absorption and uorescent behaviors of the as-prepared g-C 3 N 4 QDs show a clear blue shi, which could be due to the quantum connement effect. 43 The absolute quantum yield of the obtained g-C 3 N 4 QDs is 0.097, which is slightly higher than that of the reported carbon dots. [51][52][53] Fig. 4b displays the emission spectra of g-C 3 N 4 QDs at various excitation wavelengths. Commonly, surface defects can act as a capture center for excitons, resulting in surface defect state uorescence. Surface defect uorescence is caused by radiation relaxation from the excited state to the ground state, which can lead to multicolor emissions. When the sample is excited by light with different specic wavelengths, the photons whose energy satises the optical band gap will transit and accumulate in the adjacent surface defect centers, and then return to the ground state to emit different wavelengths of light, therefore showing excitation wavelength-dependent characteristics. 54,55 As shown in Fig. 4b, with excitation wavelength shiing from 310 to 330 nm, the emission peak positions are almost not changed. With continuously increasing the excitation wavelength from 330 to 350 nm, the emission peaks become broadened and appear slightly red-shied. The maximum uorescence emission peak is obtained when excited at 325 nm. This emission behavior suggests that there would be a slight inhomogeneity of particle size and a few surface defects involved in the surface functional groups of g-C 3 N 4 QDs, which therefore result in nonuniform surface states, which is consistent with the results of FT-IR and XPS.

Sensing for TC
To evaluate the potential application of g-C 3 N 4 QDs as an optical sensor, we investigated the g-C 3 N 4 QDs as a uorescent probe to detect TC. As shown in Fig. 5a, the uorescence intensity decreases sharply when TC is added into the g-C 3 N 4 QD solution, indicating that its uorescence is sensitive to TC. Moreover, the effect of pH on the uorescence intensity was explored.   F 0 and F represent the PL intensities of g-C 3 N 4 QDs without and with TC. As depicted in Fig. 5b, with the pH value changing from 4.0 to 9.0, the PL intensity of g-C 3 N 4 QDs gradually increases rst and reaches a maximum under neutral conditions (pH 7-8), and then decreases slowly. In the presence of TC, the quenching efficiency [(F 0 À F)/F 0 ] displays almost the same change trend. In view of that the neutral ultrapure water system is environmentally friendly and readily available, using the ultrapure water (pH 7-8) as the solvent is also researched, and the results showed that the uorescence intensity and quenching efficiency can reach the maximum in the ultrapure water system. Further studies reveal that the PL intensity can remain unaffected by ion strength when adding NaCl into the ultrapure water system with the salt concentration increasing from 0 to 500 mM (Fig. 5c). Therefore, ultrapure water is selected as a solvent in the following research. In addition, the PL intensity drops quickly with the addition of TC in the rst 10 min and then reaches balance within 15 minutes (Fig. 5d), implying that the uorescence quenching of g-C 3 N 4 QDs is quite rapid and can be applied in fast sensing of TC. What's more, the PL intensity can remain stable during the following continuous irradiation for 60 minutes, which veries the good light stability of the as-prepared g-C 3 N 4 QDs. Hence, the reaction time between TC and g-C 3 N 4 QDs is determined to be 15 min. Under the optimized experimental conditions, the analytical performance of g-C 3 N 4 QDs towards different concentrations of TC was evaluated. As shown in Fig. 6a, the uorescence intensity of g-C 3 N 4 QDs is gradually declined with the increase of TC concentration. As a note, the signals exhibit dri which oen occurs in optical chemical sensors because of scattering, photobleaching and so on. Fluorescence quenching efficiency vs. TC concentration exhibits good linear relationships in the range of 0.23 to 11.26 mM and 11.26 to 202.70 mM (Fig. 6b). The corresponding calibration equations are y ¼ 0.00778x + 0.0172 (R 2 ¼ 0.9997) and y ¼ 0.00224x + 0.0930 (R 2 ¼ 0.9988), where y and x are the quenching efficiency [(F 0 À F)/F 0 ] and the concentration of TC, respectively. The limit of detection (LOD) of 0.19 mM is calculated according to 3 times the standard deviation of the blank measurements divided by the slope of the standard curve (LOD ¼ 3SD/s). For comparison, some of the reported uorescence sensors based on other materials for TC detection are displayed in Table 1, indicating that the proposed sensing system has superior sensitivity, a wider detection range and a relatively lower detection limit.
The selectivity of the present method was further investigated in the presence of potentially interfering representative metal ions (Na + , Mg 2+ , Ba 2+ , K + , Cu 2+ , and Fe 3+ ) and common antibiotics (AMX, CIP, SDZ, CAM, TC, CTC, OTC, and DTC). As described in Fig. 7a, the uorescence of g-C 3 N 4 QDs is distinctly quenched only by TC, OTC and DTC, which belong to tetracycline antibiotics (TCs). Among them, the quenched efficiency of TC is the highest. However, another tetracycline antibiotic CTC produces a uorescence enhancement phenomenon. What's more, except TCs, other above mentioned components show negligible uorescent quenching. Moreover, the experiments of adding the mixtures of 200 mM TC and 400 mM of the above mentioned components were performed. The interference effects of other coexisting metal ions and antibiotics except TCs on the uorescence quenching of TC can be ignored (Fig. 7b). Therefore, it can be concluded that the uorescent sensor based on the as-synthesized g-C 3 N 4 QDs has excellent selectivity to TCs.
The stability of g-C 3 N 4 QDs was also studied. The relative uorescence intensities of g-C 3 N 4 QDs in the absence and presence of TC are basically unchanged aer storage for 15 days at room temperature (Fig. 7c), which demonstrates that the asobtained uorescent sensor of g-C 3 N 4 QDs has excellent stability. Additionally, to evaluate the reproducibility of g-C 3 N 4 QDs, we synthesized g-C 3 N 4 QDs repeatedly 7 times. The relative uorescence intensities of g-C 3 N 4 QDs with and without TC remain almost constant (Fig. 7d). The relative standard deviation (RSD) of TC detected for the seven parallel tests is 1.10%, and the average recovery calculated as average measured value/spiked value Â 100% is 99.63% (Table S1 †), indicating that the presented g-C 3 N 4 QD uorescent sensor is quite satisfactory in reproducibility.

Statistical evaluation
Statistical evaluation was employed to check the feasibility of the proposed uorescence sensor method for TC detection. TC standard solutions of 9.01 and 157.66 mM were prepared and determined by the presented method in six parallel tests. The measured values, mean values and standard deviations (S) are summarized in Table S2. † A 95% condence level was selected in the statistical evaluation. Considering that the outliers may affect the accuracy and precision of the results, the Grubbs test was used to determine whether outliers should be discarded. 61 The G value was calculated by formula (1).
where x q is the questionable value and xis the average value. As shown in Table S2, † the outliers of the two sets of data are 9.31 and 158.22 mM, respectively. By calculation, the G values of the two sets are 1.28 and 1.24, which are both lower than the critical value G 0.05,6 of 1.89. 62 Therefore, the outliers of 9.31 and 158.22 mM should be retained. Aer checking the outliers, the t test was used to determine the systematic error of the test data. Formula (2) was used to calculate the value of t. If t $ t a,f , where a ¼ 1 À 0.95 ¼ 0.05 and f ¼ n À 1, it means that there is a signicant difference between the average measured value and true value. Otherwise, there is no signicant difference. The calculated t values of the two sets of data are 0.59 and 2.24 respectively, which are lower than the t 0.05,5 of 2.571. 62 Hence, there is no signicant difference between the average measured value and true value. Moreover, the RSDs of the two sets of data are 3.30% and 1.10%, respectively, indicating that the precision is satisfactory. Above all, the described uorescence sensor method is feasible for TC detection within the range of the two calibration curves with high accuracy and precision.

Application in actual samples
Based on the above results, the standard addition method was conducted to examine the practical application of the proposed uorescent sensor in tap water and milk powder samples. The concentration monitored in all samples was derived from the standard curves and regression equations. As described in Table S3, † recoveries of TC at the four fortication levels in assay in tap water and milk powder were determined to be 97.19-102.76% with RSDs of 0.23-3.75%, and 95.08-104.85% with RSDs of 0.41-3.14%, correspondingly. The results indicated that the proposed uorescent probe is reliable and can be applied to detect TC in real samples.

Proposed quenching mechanism
We further researched the uorescence quenching mechanism of g-C 3 N 4 QDs caused by TC. Fluorescence quenching is divided into static and dynamic quenching processes. 63 The static quenching is due to the formation of a non-luminescent complex between the ground state uorescent substance and quencher. 63 The collisional interaction between the quencher and excited state uorescent substance will result in dynamic quenching. 63 The dynamic quenching can be described by the Stern-Volmer equation [Eqn (3)], and the static quenching process has a similar equation form [Eqn (4)].
where F 0 and F are the uorescence intensities of g-C 3 N 4 QDs in the absence and presence of TC, respectively; K sv represents the quenching constant; K s denotes the association constant of the complex; [Q] is the concentration of the quencher. The curves of F 0 /F vs.
[Q] at various experimental temperatures are shown in Fig. 8a, and the parameters of plots are displayed in Table S4. † The slopes of linear regression equations are 4.56 Â 10 3 , 4.15 Â 10 3 and 3.71 Â 10 3 L mol À1 at the temperature of 303, 313 and 333 K, correspondingly. Obviously, the quenching efficiency is reduced with the increase of temperature, indicating that there is a static quenching process, 64 because of that increasing the temperature will decrease the association constant of the complex formed between g-C 3 N 4 QDs and TC.
In addition, we also measured the uorescence lifetime of g-C 3 N 4 QDs without and with TC (Fig. 8b), and the uorescence decay curves are well-tted by a three-exponential function [Eqn (5)]. Among them, s 1 , s 2 and s 3 are the time constants of the three radiative decay channels; A 1 , A 2 and A 3 are the corresponding amplitudes. 65 The relevant parameters are  summarized in Table S5, † and the calculated average lifetimes are 5.66 and 5.56 ns for g-C 3 N 4 QDs and g-C 3 N 4 QDs/TC based on eqn (6), respectively. Nearly unchanged uorescence lifetime further indicates that the uorescence quenching of g-C 3 N 4 QDs with TC addition is mainly governed by a static quenching process.
Furthermore, in order to get more information about the mechanism of uorescence quenching, we analyze the uorescence spectra of g-C 3 N 4 QDs and the UV-vis absorption spectrum of TC carefully (Fig. 8c). The excitation and emission spectra of g-C 3 N 4 QDs overlap with the absorption peaks of TC, which suggests that the uorescence of g-C 3 N 4 QDs may be quenched either by the inner lter effect (IFE) or uorescence resonance energy transfer (FRET). FRET is a non-radiative energy transfer caused by vibration collision within a limited distance of 10 nm between the donor and acceptor, which will shorten the uorescence lifetime of the donor, 66 whereas the IFE is a radiation energy transfer and not limited by distance, which is induced by the formation of the ground-state complex. Therefore, for the IFE, the absorption spectra change and the uorescence lifetime remains constant of the uorescent substance with the quencher. 65 As shown in Fig. 8d, the absorption peaks of TC shi from 275 and 357 nm to 263 and 381 nm accordingly, and the relative intensity changes aer reaction with g-C 3 N 4 QDs. Moreover, combined with the front results of uorescence lifetime analysis, we deduced that the uorescence quenching of g-C 3 N 4 QDs with TC is attributed to the IFE, not FRET.

Conclusion
In summary, we have successfully prepared a g-C 3 N 4 QD solution with excellent uorescence properties and water-solubility by a simple, low cost and environmental-friendly solvothermal "tailoring" method. The resultant g-C 3 N 4 QDs have a "strong quenching" behavior in the presence of TC, which is mainly controlled by a static quenching process because of the IFE mechanism. Based on this, we developed a rapid uorescent sensor to detect TC in a highly selective and sensitive manner with a wide detection range and low detection limit. The proposed g-C 3 N 4 QDs have promising application prospects in the monitoring of TC in real samples.

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