The Fluorescent Quenching Mechanism of N and S Co-Doped Graphene Quantum Dots with Fe3+ and Hg2+ Ions and Their Application as a Novel Fluorescent Sensor

The fluorescence intensity of N, S co-doped graphene quantum dots (N, S-GQDs) can be quenched by Fe3+ and Hg2+. Density functional theory (DFT) simulation and experimental studies indicate that the fluorescence quenching mechanisms for Fe3+ and Hg2+ detection are mainly attributed to the inner filter effect (IFE) and dynamic quenching process, respectively. The electronegativity difference between C and doped atoms (N, S) in favor to introduce negative charge sites on the surface of N, S-GQDs leads to charge redistribution. Those negative charge sites facilitate the adsorption of cations on the N, S-GQDs’ surface. Atomic population analysis results show that some charge transfer from Fe3+ and Hg2+ to N, S-GQDs, which relate to the fluorescent quenching of N, S-GQDs. In addition, negative adsorption energy indicates the adsorption of Hg2+ and Fe2+ is energetically favorable, which also contributes to the adsorption of quencher ions. Blue fluorescent N, S-GQDs were synthesized by a facile one-pot hydrothermal treatment. Fluorescent lifetime and UV-vis measurements further validate the fluorescent quenching mechanism is related to the electron transfer dynamic quenching and IFE quenching. The as-synthesized N, S-GQDs were applied as a fluorescent probe for Fe3+ and Hg2+ detection. Results indicate that N, S-GQDs have good sensitivity and selectivity on Fe3+ and Hg2+ with a detection limit as low as 2.88 and 0.27 nM, respectively.


Introduction
Graphene quantum dots (GQDs) have drawn research interest by reasons of their excellent photoluminescence, low toxicity, and good biocompatibility [1]. They have been applied in various fields, such as solar cells, light-emitting diodes, bioimaging, and fluorescent sensors [2]. Some reports have shown that the fluorescence intensity of GQDs can be quenched by Ag + and Hg 2+ ions [3][4][5]. According to the interaction between quantum dots and a quencher, the main quenching mechanism can be divided into Förster resonance energy transfer (FRET), the inner filter effect (IFE), and the dynamic and static quenching process [6,7]. The quenching process requires sufficient contact between quantum dots and a quencher. Diffusive encounters collision contact can be attributed to the dynamic quenching process, while new complex forms can be assigned to the static quenching process [8].

Synthesis of N-GQDs
The synthesis of N, S-GQDs was through a hydrothermal process using critic acid and thioacetamide as the initial carbon, nitrogen, and sulfur source, respectively. Briefly, 0.5 g critic acid and 0.0267 g thioacetamide were dissolved into 60 mL purified water by ultrasonic treatment for 5 min. Then the mixture was transferred into 100 mL Teflon-lined autoclave and heated at 180 • C for 10 h. After being cooled down to room temperature, the suspension was centrifuged at 10,000 rpms for 15 min. The obtained N, S-GQDs solution was collected after being further purified through dialysis (cutoff molecular weight: 300 Da) for 10 h.

Characterization
The morphology of as-synthesized N, S-GQDs was recorded on a JEOL-JEM 2100 transmission electron microscope (TEM) and Seiko SPA-400 SPM atomic force microscope (AFM). The optical properties of N, S-GQDs were carried out by UV-vis-1800 spectrophotometer (Jinghua Instrument, Shanghai, China). The Fourier transformed infrared spectra (FTIR) were recorded by an Avatar-360 spectrometer. K-Alpha + X-ray electron spectrometer was used to record the X-ray photoelectron spectroscopy (XPS) of N, S-GQDs. The photoluminescence spectra were conducted on a Horiba Fluorolog-3 fluorescence spectrophotometer. Fluorescence lifetime decays were acquired on a Quantaurus-Tau fluorescence lifetime spectrometer (Hamamatsu, Japan).

Detection of Metal Ions
The detection of Fe 3+ or Hg 2+ was performed by measuring the fluorescence spectra in the presence and absence of metal ions. In a typical analysis process, 200 µL N, S-GQDs (0.08 mg/mL) were dispersed into 1 mL PBS (phosphate buffered saline) buffer (0.1 M, pH 7), followed by the addition of a certain amount of Fe 3+ or Hg 2+ , respectively. The schematic diagram of detection process and device are shown in Figure S1 in Supplementary Materials. Then the solution was diluted to 5 mL with PBS buffer and incubated for 10 min at room temperature. The detection of Fe 3+ or Hg 2+ was assessed by the fluorescence quenching ratio (I/I 0 ) with various metal ion concentrations, where I and I 0 were corresponding to the fluorescence intensity in the presence and absence of metal ions, respectively. The selectivity of N, S-GQDs was investigated by adding various certain concentrations of interfering ions (50 µM).
nM, 1 μM, and 3 μM, respectively. Then, the appropriate volume of cation solutions were skipped into the fluorescence solution, and diluted to 5 mL with PBS buffer.

Fluorescent Quenching Mechanisms of N, S-GQDs with Fe 3+ or Hg 2+
DFT simulations were carried out to investigate the quenching mechanism of N, S-GQDs on Fe 3+ or Hg 2+ . The adsorption energy (Ead) and Mulliken populations were calculated to investigate the interaction between N, S-GQDs and a quencher. For the simulations, 18 atoms' hexagonal configurations were applied, as shown in Figure 1. Atomic population (Mulliken) analysis indicates that carbon atoms in pure graphene are neutral, as shown in Table 1. As set out in Table 1, the S atoms have a positive charge, while the negative charges are located on N and O atoms. This can be ascribed to the electronegativity difference of N, S, and O atoms. N and O atoms have larger electronegativity and stronger charge-accepting ability than S. The positively charged S atom breaks the electro-neutral situation of surrounding C atoms, and that is conductive to the generation of charge favorable sites for the interaction of cations and N, S-GQDs [31]. The atomic populations of Fe 3+ @N, S-GQDs and Hg 2+ @N, S-GQDs are listed in Supplementary Materials, Table S1. The adsorption energy can be defined as the energy difference between N, S-GQDs and the adsorption system. The Ead of Fe 3+ and Hg 2+ were calculated as −2.43 and −3.27 eV, respectively, which indicate the adsorption of Hg 2+ and Fe 2+ are energetically favorable. According to the DFT calculations, the adsorption of Fe 3+ or Hg 2+ change the charge distribution on N, S-GQDs as shown in Table S1. Atomic population calculations reveal that the amount of charge transfer between N, S-GQDs and adsorbed Fe 3+ is 0.66 e − and the charge transfer for Hg 2+ is 0.13 e − , respectively, implying the interaction existence between N, S-GQDs and a quencher (Fe 3+ or Hg 2+ ). Based on the DFT simulations, the quenching mechanism of N, S-GQDs by Fe 3+ and Hg 2+ might be attributed to the dynamic quenching.   In order to further investigate the fluorescent quenching mechanism of N, S-GQDs by Fe 3+ or Hg 2+ , N, S-GQDs were synthesized by a hydrothermal method. In the following, the fluorescence lifetime and UV-vis spectra were measured to determine the fluorescence quenching mechanism of N, S-GQDs, as shown in Figure 2. As shown in Figure 2, the average fluorescence lifetime of N, S-GQDs without metal ions (Fe 3+ or Hg 2+ ) is 8.15 ns. While the average fluorescence lifetime of N, S-GQDs in the presence of Fe 3+ and Hg 2+ can be determined as 8.02 and 7.72 ns, respectively. The average fluorescence lifetime of N, S-GQDs decay after the addition of Hg 2+ , indicating the fluorescence quenching is probably attributed to the non-radiative electron transfer between N, S-GQDs and a quencher [6,32]. However, the average fluorescence lifetime in the presence of Fe 3+ only changes slightly, which indicates the quenching process is probably related to other quenching mechanisms [33][34][35]. Figure 2d depicts the UV-vis spectra and fluorescence spectra of N, S-GQDs in the absence and presence of Fe 3+ or Hg 2+ , respectively. The absorption threshold and peak position change cannot be observed, which indicates that no stable metal complexes form [36]. It is noticeable that the absorption spectra of N, S-GQDs with Fe 3+ have overlaps with the emission and excitation spectra of N, S-GQDs, which is related to the characteristic of IFE quenching [7,33]. The fluorescence lifetime of N, S-GQDs changed in the presence of Hg 2+ but no changes in the absorption spectra of N, S-GQDs can be observed. Therefore, based on the DFT and experimental analyses, the fluorescence quenching mechanism of N, S-GQDs by Fe 3+ is mainly attributed to the IFE, while the fluorescence quenching by Hg 2+ is related to the dynamic quenching process, as shown in Figure 3. In order to further investigate the fluorescent quenching mechanism of N, S-GQDs by Fe 3+ or Hg 2+ , N, S-GQDs were synthesized by a hydrothermal method. In the following, the fluorescence lifetime and UV-vis spectra were measured to determine the fluorescence quenching mechanism of N, S-GQDs, as shown in Figure 2. As shown in Figure 2, the average fluorescence lifetime of N, S-GQDs without metal ions (Fe 3+ or Hg 2+ ) is 8.15 ns. While the average fluorescence lifetime of N, S-GQDs in the presence of Fe 3+ and Hg 2+ can be determined as 8.02 and 7.72 ns, respectively. The average fluorescence lifetime of N, S-GQDs decay after the addition of Hg 2+ , indicating the fluorescence quenching is probably attributed to the non-radiative electron transfer between N, S-GQDs and a quencher [6,32]. However, the average fluorescence lifetime in the presence of Fe 3+ only changes slightly, which indicates the quenching process is probably related to other quenching mechanisms [33][34][35]. Figure 2d depicts the UV-vis spectra and fluorescence spectra of N, S-GQDs in the absence and presence of Fe 3+ or Hg 2+ , respectively. The absorption threshold and peak position change cannot be observed, which indicates that no stable metal complexes form [36]. It is noticeable that the absorption spectra of N, S-GQDs with Fe 3+ have overlaps with the emission and excitation spectra of N, S-GQDs, which is related to the characteristic of IFE quenching [7,33]. The fluorescence lifetime of N, S-GQDs changed in the presence of Hg 2+ but no changes in the absorption spectra of N, S-GQDs can be observed. Therefore, based on the DFT and experimental analyses, the fluorescence quenching mechanism of N, S-GQDs by Fe 3+ is mainly attributed to the IFE, while the fluorescence quenching by Hg 2+ is related to the dynamic quenching process, as shown in Figure 3.

Characterization of N, S-GQDS
After the quenching mechanism analysis of N, S-GQDs, the as-synthesized N, S-GQDs were applied to detect Fe 3+ and Hg 2+ , respectively. The morphological properties of as-synthesized N, S-GQDs were carried out by TEM and AFM images as shown in Figure 4. The high-resolution transmission electron microscopy (HRTEM) image, as shown in the inset of Figure 4a, displays the crystal lattice of N, S-GQDs with a lattice spacing distance of 2.23 Å , which corresponds to the (1120) plane of graphite [5]. The TEM image shows that the as-synthesized N, S-GQDs are well-dispersed and have a narrow sized distribution ranging from 1.5 to 4 nm with an average size of 2.5 nm, as shown in Figure 4b. AFM images reveal the thickness of N, S-GQDs, as shown in Figure 4c. The topographic height of 1.0-2.5 nm can be seen in Figure 4d with an average height of 1.5 nm. The thickness of single layered graphene is 0.34 nm, which suggests that most N, S-GQDs are single or bilayered. Therefore, it can be concluded that the as-synthesized quantum dots are graphene quantum dots.

Characterization of N, S-GQDS
After the quenching mechanism analysis of N, S-GQDs, the as-synthesized N, S-GQDs were applied to detect Fe 3+ and Hg 2+ , respectively. The morphological properties of as-synthesized N, S-GQDs were carried out by TEM and AFM images as shown in Figure 4. The high-resolution transmission electron microscopy (HRTEM) image, as shown in the inset of Figure 4a, displays the crystal lattice of N, S-GQDs with a lattice spacing distance of 2.23 Å, which corresponds to the (1120) plane of graphite [5]. The TEM image shows that the as-synthesized N, S-GQDs are well-dispersed and have a narrow sized distribution ranging from 1.5 to 4 nm with an average size of 2.5 nm, as shown in Figure 4b. AFM images reveal the thickness of N, S-GQDs, as shown in Figure 4c. The topographic height of 1.0-2.5 nm can be seen in Figure 4d with an average height of 1.5 nm. The thickness of single layered graphene is 0.34 nm, which suggests that most N, S-GQDs are single or bilayered. Therefore, it can be concluded that the as-synthesized quantum dots are graphene quantum dots.

Characterization of N, S-GQDS
After the quenching mechanism analysis of N, S-GQDs, the as-synthesized N, S-GQDs were applied to detect Fe 3+ and Hg 2+ , respectively. The morphological properties of as-synthesized N, S-GQDs were carried out by TEM and AFM images as shown in Figure 4. The high-resolution transmission electron microscopy (HRTEM) image, as shown in the inset of Figure 4a, displays the crystal lattice of N, S-GQDs with a lattice spacing distance of 2.23 Å , which corresponds to the (1120) plane of graphite [5]. The TEM image shows that the as-synthesized N, S-GQDs are well-dispersed and have a narrow sized distribution ranging from 1.5 to 4 nm with an average size of 2.5 nm, as shown in Figure 4b. AFM images reveal the thickness of N, S-GQDs, as shown in Figure 4c. The topographic height of 1.0-2.5 nm can be seen in Figure 4d with an average height of 1.5 nm. The thickness of single layered graphene is 0.34 nm, which suggests that most N, S-GQDs are single or bilayered. Therefore, it can be concluded that the as-synthesized quantum dots are graphene quantum dots.  XPS measurements were performed to characterize the elemental composition properties of as-synthesized N, S-GQDs, as shown in Figure 5. Figure 5a shows the survey spectra of N, S-GQDs. Four dominant peaks at 168.8, 285.2, 401.7, and 533.4 eV can be identified as S 2p, C 1s, N1s, and O1s, respectively. The C, S, O, and N configuration in N, S-GQDs were investigated by the deconvolution of S 2p, C 1s, N1s, and O1s peaks, as shown in Figure 5b-e. As shown in Figure 5b, the high-resolution spectrum of C 1S can be deconvoluted into four peaks at 283.9, 284.8, 285.7, and 288.4 eV, which can be assigned to C-S-C, C-C, C-O/C-H, and O-C=O, respectively [37,38]. The covalent bond of C-S-C at 283.9 eV confirms the presence of S doping in N, S-GQDs. Figure 5c represents the high-resolution spectrum of N 1s. The binding energy peaks located at 399.4 and 401.2 eV are the contributions of pyrrolic N (C-N-C) and O=C-N, respectively [39,40]. The S 2p spectrum, as shown in Figure 5d XPS measurements were performed to characterize the elemental composition properties of as-synthesized N, S-GQDs, as shown in Figure 5. Figure 5a shows the survey spectra of N, S-GQDs. Four dominant peaks at 168.8, 285.2, 401.7, and 533.4 eV can be identified as S 2p, C 1s, N1s, and O1s, respectively. The C, S, O, and N configuration in N, S-GQDs were investigated by the deconvolution of S 2p, C 1s, N1s, and O1s peaks, as shown in Figure 5b-e. As shown in Figure 5b, the high-resolution spectrum of C 1S can be deconvoluted into four peaks at 283.9, 284.8, 285.7, and 288.4 eV, which can be assigned to C-S-C, C-C, C-O/C-H, and O-C=O, respectively [37,38]. The covalent bond of C-S-C at 283.9 eV confirms the presence of S doping in N, S-GQDs. Figure 5c represents the high-resolution spectrum of N 1s. The binding energy peaks located at 399.4 and 401.2 eV are the contributions of pyrrolic N (C-N-C) and O=C-N, respectively [39,40]. The S 2p spectrum, as shown in Figure 5d [40,44]. It is worthy to notice that the absorption bands at 1195 cm −1 are assigned to the vibration of C-S or C-N bonds, indicating that S and N atoms have been doped into the graphene configuration. Peaks at 1573 and 1673 cm −1 are corresponding to C=N and C=C stretching. Nanomaterials 2019, 9, x FOR PEER REVIEW 8 of 18 Figure 6 shows the FTIR spectrum of as-synthesized N, S-GQDs. The bands appearing around 3417 cm −1 are related to the stretching vibration of O-H or N-H bonds. The broad absorption band at 2528 cm −1 is assigned to the vibration of S-H. The absorption peaks at 1709 and 1403 cm −1 , which are ascribed to the contribution of carboxyl groups (C=O) and the amide linkage (O=C-NH) bending vibration, indicate that there are abundant oxygen-contained groups on the surface of as-synthesized N, S-GQDs [40,44]. It is worthy to notice that the absorption bands at 1195 cm −1 are assigned to the vibration of C-S or C-N bonds, indicating that S and N atoms have been doped into the graphene configuration. Peaks at 1573 and 1673 cm −1 are corresponding to C=N and C=C stretching.

Optical Analyses of N, S-GQDS
UV-vis and fluorescence spectrum were applied to investigate the optical properties of N, S-GQDs, as shown in Figure 7. As shown in Figure 7a (blue curve), two clear characteristic absorption peaks at 230 and 350 nm can be observed, respectively. The absorption peak at 230 nm is contributed from the π-π* transition [40,45]. Typical n-π* transition absorption bands of C=O and C=N are plotted at 350 nm [5]. The fluorescence excitation wavelength of N, S-GQDs is 350 nm, as given by the photoluminescence excitation spectrum of Figure 7a (black curve), indicating that the n-π* transition is the domain excitation mode in the photoluminescence progress of N, S-GQDs [46]. The maximum wavelength of photoluminescence is 438 nm (red curve), which further proves N, S-GQDs aqueous solution emits strong blue fluorescence under UV lamp irradiation. It can be seen in Figure 7a that the fluorescence intensities of N, S-GQDs are quenched by Hg 2+ and Fe 3+ ions, respectively. Furthermore, the photoluminescence wavelength reveals nearly excitation-dependence properties, as shown in Figure 7b. Excitation dependent properties of N, S-GQDs are the result of surface defect and particle size variation. The emission wavelength changes slightly and is attributed to the narrow particle distribution of as-synthesized N, S-GQDs. The stabilities of as-synthesized N, S-GQDs were evaluated by recording the fluorescence intensity of N, S-GQDs for 60 days, as shown in Figure S2. The results indicate that N, S-GQDs have good fluorescent stability. Besides that, heteroatom N and S doping into graphene quantum dots' lattice also contributes the excitation-dependence properties of N, S-GQDs [37].

Optical Analyses of N, S-GQDS
UV-vis and fluorescence spectrum were applied to investigate the optical properties of N, S-GQDs, as shown in Figure 7. As shown in Figure 7a (blue curve), two clear characteristic absorption peaks at 230 and 350 nm can be observed, respectively. The absorption peak at 230 nm is contributed from the π-π* transition [40,45]. Typical n-π* transition absorption bands of C=O and C=N are plotted at 350 nm [5]. The fluorescence excitation wavelength of N, S-GQDs is 350 nm, as given by the photoluminescence excitation spectrum of Figure 7a (black curve), indicating that the n-π* transition is the domain excitation mode in the photoluminescence progress of N, S-GQDs [46]. The maximum wavelength of photoluminescence is 438 nm (red curve), which further proves N, S-GQDs aqueous solution emits strong blue fluorescence under UV lamp irradiation. It can be seen in Figure 7a that the fluorescence intensities of N, S-GQDs are quenched by Hg 2+ and Fe 3+ ions, respectively. Furthermore, the photoluminescence wavelength reveals nearly excitation-dependence properties, as shown in Figure 7b. Excitation dependent properties of N, S-GQDs are the result of surface defect and particle size variation. The emission wavelength changes slightly and is attributed to the narrow particle distribution of as-synthesized N, S-GQDs. The stabilities of as-synthesized N, S-GQDs were evaluated by recording the fluorescence intensity of N, S-GQDs for 60 days, as shown in Figure S2. The results indicate that N, S-GQDs have good fluorescent stability. Besides that, heteroatom N and S doping into graphene quantum dots' lattice also contributes the excitation-dependence properties of N, S-GQDs [37].

Selectivity of N, S-GQDs for Fe 3+ and Hg 2+ Detection
The selectivity of N, S-GQDs were evaluated by measuring the fluorescence quenching ratio as a result of the addition of various cations, including Ag + , K + , Ba 2+ , Mg 2+ , Ca 2+ , Al 3+ , Co 2+ , Cd 2+ , Zn 2+ , Cu 2+ , Fe 2+ , Na + , Ni 2+ , and Pb 2+ . The fluorescence quenching ratios were recorded in the presence of 50 µ M various cations, as shown in Figure 8a. Of all the cations tested, only Hg 2+ and Fe 3+ led to efficient quenching of fluorescence intensity. Moreover, the interfering tests of N, S-GQDs, as shown in Figure 8b, were investigated through recording the fluorescence quenching ratio on Fe 3+ or Hg 2+ with the addition of various interfering cations (50 µ M). As displayed in Figure 8c, the presence of interfering ions does not have an obvious impact on Fe 3+ or Hg 2+ detection. When Fe 3+ or Hg 2+ coexist in the detection sample, the interfering of another cation can be eliminated by adding ascorbic acid (AA) and cysteine (Cys), respectively.

Selectivity of N, S-GQDs for Fe 3+ and Hg 2+ Detection
The selectivity of N, S-GQDs were evaluated by measuring the fluorescence quenching ratio as a result of the addition of various cations, including Ag + , K + , Ba 2+ , Mg 2+ , Ca 2+ , Al 3+ , Co 2+ , Cd 2+ , Zn 2+ , Cu 2+ , Fe 2+ , Na + , Ni 2+ , and Pb 2+ . The fluorescence quenching ratios were recorded in the presence of 50 µM various cations, as shown in Figure 8a. Of all the cations tested, only Hg 2+ and Fe 3+ led to efficient quenching of fluorescence intensity. Moreover, the interfering tests of N, S-GQDs, as shown in Figure 8b, were investigated through recording the fluorescence quenching ratio on Fe 3+ or Hg 2+ with the addition of various interfering cations (50 µM). As displayed in Figure 8c, the presence of interfering ions does not have an obvious impact on Fe 3+ or Hg 2+ detection. When Fe 3+ or Hg 2+ coexist in the detection sample, the interfering of another cation can be eliminated by adding ascorbic acid (AA) and cysteine (Cys), respectively.

Selectivity of N, S-GQDs for Fe 3+ and Hg 2+ Detection
The selectivity of N, S-GQDs were evaluated by measuring the fluorescence quenching ratio as a result of the addition of various cations, including Ag + , K + , Ba 2+ , Mg 2+ , Ca 2+ , Al 3+ , Co 2+ , Cd 2+ , Zn 2+ , Cu 2+ , Fe 2+ , Na + , Ni 2+ , and Pb 2+ . The fluorescence quenching ratios were recorded in the presence of 50 µ M various cations, as shown in Figure 8a. Of all the cations tested, only Hg 2+ and Fe 3+ led to efficient quenching of fluorescence intensity. Moreover, the interfering tests of N, S-GQDs, as shown in Figure 8b, were investigated through recording the fluorescence quenching ratio on Fe 3+ or Hg 2+ with the addition of various interfering cations (50 µ M). As displayed in Figure 8c, the presence of interfering ions does not have an obvious impact on Fe 3+ or Hg 2+ detection. When Fe 3+ or Hg 2+ coexist in the detection sample, the interfering of another cation can be eliminated by adding ascorbic acid (AA) and cysteine (Cys), respectively.  Some reports have shown that the good affinity between quencher cations and functional groups on the surface of quantum dots relates to the good selectivity of GQDs' fluorescence probe [47][48][49]. XPS and FTIR analysis show that there are many carboxyl, hydroxyl, amino, and hydrosulfonyl groups on the surface of N, S-GQDs. Fe 3+ and Hg 2+ have good affinity with them. Hence, N, S-GQDs have better selectivity to Fe 3+ or Hg 2+ than other cations. Bond populations and electron density difference also were calculated to investigate the interaction between N, S-GQDs and a quencher (Fe 3+ or Hg 2+ ). As listed in Table S1, the bond population of Fe-S was calculated as 0.17, indicating Fe 3+ can covalently bond with S. The electron density difference of Fe 3+ @N, S-GQDs and Hg 2+ @N, S-GQDs are plotted in Figure 9. The red areas represent the electron enrichment and the blue areas indicate electron withdrawal. The electron density difference distributed between Fe and S atoms, as shown in Figure 9a, shows the characteristic of covalent bond distribution, which agrees with bond population analysis. According to the electron density difference and bond population of Hg 2+ @N, S-GQDs, as shown in Figure 9c,d, there are no obvious covalent or chemical bonds formed between Hg and doped atoms (N, S). The atomic population of doped S shows that a doped S atom has a positive charge population. This situation could reduce the ability for S atoms forming chemical bonds with interfering cations. Meanwhile, energetically favorable adsorption and the charge favorable sites on the surface of N, S-GQDs also avail the electron transfer between Fe 3+ , Hg 2+ , and N, S-GQDs [9]. Therefore, N, S-GQDs have better selectivity to Fe 3+ and Hg 2+ than other interfering cations. Some reports have shown that the good affinity between quencher cations and functional groups on the surface of quantum dots relates to the good selectivity of GQDs' fluorescence probe [47][48][49]. XPS and FTIR analysis show that there are many carboxyl, hydroxyl, amino, and hydrosulfonyl groups on the surface of N, S-GQDs. Fe 3+ and Hg 2+ have good affinity with them. Hence, N, S-GQDs have better selectivity to Fe 3+ or Hg 2+ than other cations. Bond populations and electron density difference also were calculated to investigate the interaction between N, S-GQDs and a quencher (Fe 3+ or Hg 2+ ). As listed in Table S1, the bond population of Fe-S was calculated as 0.17, indicating Fe 3+ can covalently bond with S. The electron density difference of Fe 3+ @N, S-GQDs and Hg 2+ @N, S-GQDs are plotted in Figure 9. The red areas represent the electron enrichment and the blue areas indicate electron withdrawal. The electron density difference distributed between Fe and S atoms, as shown in Figure 9a, shows the characteristic of covalent bond distribution, which agrees with bond population analysis. According to the electron density difference and bond population of Hg 2+ @N, S-GQDs, as shown in Figure 9c,d, there are no obvious covalent or chemical bonds formed between Hg and doped atoms (N, S). The atomic population of doped S shows that a doped S atom has a positive charge population. This situation could reduce the ability for S atoms forming chemical bonds with interfering cations. Meanwhile, energetically favorable adsorption and the charge favorable sites on the surface of N, S-GQDs also avail the electron transfer between Fe 3+ , Hg 2+ , and N, S-GQDs [9]. Therefore, N, S-GQDs have better selectivity to Fe 3+ and Hg 2+ than other interfering cations.

The Fluorescence Properties of N, S-GQDs under Acidity and Alkalinity Situations
The fluorescence stability and fluorescence quenching ratio were tested by measuring fluorescence intensity change on various pH values from 2.0 to 12.0, as shown in Figure 10

The Fluorescence Properties of N, S-GQDs under Acidity and Alkalinity Situations
The fluorescence stability and fluorescence quenching ratio were tested by measuring fluorescence intensity change on various pH values from 2.0 to 12.0, as shown in Figure 10. The fluorescence intensities of N, S-GQDs remain stable in alkaline conditions, while there is a decrease of fluorescence intensity in the pH range from 2.0 to 7.0. The decrease of fluorescence intensity of N, S-GQDs in acidic solution can be attributed to the protonation of amino and carboxylic groups in N, S-GQDs. As the carboxyl and amino groups on the surface of N, S-GQDs tend to bond with protons, resulting in the redistribution of surface electrons in acidic conditions. The fluorescence intensity decrease is assigned to the aggregation of N, S-GQDs after the protonation process of amino and carboxylic groups take place [50]. However, the carboxyl groups on the surface of N, S-GQDs are deprotonated in alkaline medium and form a negative charge shell, which makes the fluorescence properties of N, S-GQDs retain a stable performance [51]. As shown in Figure 10, the fluorescence quenching performance of N, S-GQDs show pH-dependent behavior in the presence of 1 µM Hg 2+ and 1µM Fe 3+ , respectively. Carboxyl groups on the surface of N, S-GQDs are well protonated in acidic medium, which can weaken the affinity between metal ions (Hg 2+ or Fe 3+ ) and carboxyl groups, resulting in less fluorescence quenching. Hence, the fluorescence quenching performance of as-synthesized N, S-GQDs has pH-dependence behavior. Considering the fluorescence intensity and best fluorescence quenching of N, S-GQDs, the following photoluminescence spectra were measured under pH 7. decrease is assigned to the aggregation of N, S-GQDs after the protonation process of amino and carboxylic groups take place [50]. However, the carboxyl groups on the surface of N, S-GQDs are deprotonated in alkaline medium and form a negative charge shell, which makes the fluorescence properties of N, S-GQDs retain a stable performance [51]. As shown in Figure 10, the fluorescence quenching performance of N, S-GQDs show pH-dependent behavior in the presence of 1 μM Hg 2+ and 1μM Fe 3+ , respectively. Carboxyl groups on the surface of N, S-GQDs are well protonated in acidic medium, which can weaken the affinity between metal ions (Hg 2+ or Fe 3+ ) and carboxyl groups, resulting in less fluorescence quenching. Hence, the fluorescence quenching performance of as-synthesized N, S-GQDs has pH-dependence behavior. Considering the fluorescence intensity and best fluorescence quenching of N, S-GQDs, the following photoluminescence spectra were measured under pH 7.

Detection of Fe 3+ and Hg 2+ Using N, S-GQDs as a Sensor
The fluorescence quenching performance after the addition of various concentrations of Fe 3+ is plotted in Figure 11. As shown in Figure 11a, the fluorescence intensities of N, S-GQDs are quenched by Fe 3+ . The fluorescence quenching ratios are performed to determine the sensitivity of as-synthesized N, S-GQDs for Fe 3+ detection, as shown in Figure 11b-d. Figure 11b shows the fluorescence quenching ratio as a function of Fe 3+ concentration in the range 0-100 μM. Two individual linear relationships between the fluorescence quenching ratio and the concentration of Fe 3+ can be observed in the dynamic ranges 1-90 nM and 0.1-30 μM, as shown in Figure 11c,d, respectively. As shown in Figure 11c, the linear regression equation (I and I0 refer to the fluorescence intensity of N, S-GQDs at 438 nm in the presence and absence of Fe 3+ , respectively) is I/I0 = 0.98951-0.0005194 [Fe 3+ ]. The corresponding regression coefficient (R 2 ) is 0.985. A good detection limit (LOD) of 2.88 nM can be obtained by the standard curve of the fluorescence quenching ratio and Fe 3+ concentration (1-90 nM). The detection limit is defined by a signal-to-noise ratio of 3. In addition, the LOD can be calculated as 55.49 nM in the dynamic range 0.1-30 μM. According to the Standards for Drinking Water Quality of China National Standers, the threshold limit of Fe 3+ ion content in water is 0.3 mg/L (5.37 μM) [52]. Therefore, the dynamic range and detection limit of as-synthesized N, S-GQDs have comparable application potential for Fe 3+ detection in drinking water.
Besides that, the as-synthesized N, S-GQDs were applied to determine Hg 2+ , as shown in Figure  12. As described in Figure 12a

Detection of Fe 3+ and Hg 2+ Using N, S-GQDs as a Sensor
The fluorescence quenching performance after the addition of various concentrations of Fe 3+ is plotted in Figure 11. As shown in Figure 11a, the fluorescence intensities of N, S-GQDs are quenched by Fe 3+ . The fluorescence quenching ratios are performed to determine the sensitivity of as-synthesized N, S-GQDs for Fe 3+ detection, as shown in Figure 11b-d. Figure 11b shows the fluorescence quenching ratio as a function of Fe 3+ concentration in the range 0-100 µM. Two individual linear relationships between the fluorescence quenching ratio and the concentration of Fe 3+ can be observed in the dynamic ranges 1-90 nM and 0.1-30 µM, as shown in Figure 11c,d, respectively. As shown in Figure 11c . Therefore, the dynamic range and detection limit of as-synthesized N, S-GQDs have comparable application potential for Fe 3+ detection in drinking water.
Environmental Protection Agency (0.002 mg L , 9.96 nM) for the safety demands of Hg in drinking water [52,53]. In addition, another linear relationship can be observed in the range 100-1000 nM with the detection limit of 36.85 nM, which also meets the requirements of the China National Standards for the upper safety limit of total mercury content in industrial effluent (0.05 mg L −1 , 249 nM) [52]. The doped atom N and S not only has contributed to the fluorescence quenching process of N, S-GQDs, but also makes the as-synthesized N, S-GQDs have a comparable detection performance on Fe 3+ or Hg 2+ to other fluorescence sensing systems. The detection performances of N, S-GQDs for Fe 3+ or Hg 2+ are compared with various reported fluorescence sensors, as listed in Tables 2 and 3.  Besides that, the as-synthesized N, S-GQDs were applied to determine Hg 2+ , as shown in Figure 12. As described in Figure 12a, the as-synthesized N, S-GQDs exhibit Hg 2+ concentration-dependent fluorescence properties. The fluorescence intensity of N, S-GQDs can be quenched after the addition of various amounts of Hg 2+ . There is a good linear relationship (I/I 0 = 0.99383−0.00181 [Hg 2+ ], R 2 = 0.991) between the fluorescence quenching ratio and Hg 2+ concentration in the range 1-30 nM, as shown in Figure 12c. The detection limit can be calculated as 0.27 nM, which satisfies the requirements of the Chinese National Standards (0.001 mgL −1 , 4.98 nM) and the United States Environmental Protection Agency (0.002 mg L −1 , 9.96 nM) for the safety demands of Hg 2+ in drinking water [52,53]. In addition, another linear relationship can be observed in the range 100-1000 nM with the detection limit of 36.85 nM, which also meets the requirements of the China National Standards for the upper safety limit of total mercury content in industrial effluent (0.05 mg L −1 , 249 nM) [52]. The doped atom N and S not only has contributed to the fluorescence quenching process of N, S-GQDs, but also makes the as-synthesized N, S-GQDs have a comparable detection performance on Fe 3+ or Hg 2+ to other fluorescence sensing systems. The detection performances of N, S-GQDs for Fe 3+ or Hg 2+ are compared with various reported fluorescence sensors, as listed in Tables 2 and 3.

Real Sample Detection Analyses
To evaluate the applicability of an as-synthesized N, S-GQDs fluorescence probe, the performance of N, S-GQDs in real drinking water samples was investigated. Figure S3 displays the fluorescence intensity change spectra of N, S-GQDs with different concentrations of Fe 3+ (70 nM, 1 µM, and 3 µM) and Hg 2+ (50, 100, and 300 nM), respectively. The recoveries of Fe 3+ or Hg 2+ detection are given in Table S2. Recoveries of drinking water sample detection are close to 100%. The obtained results imply that the as-synthesized N, S-GQDs have potential as a fluorescent probe for Fe 3+ or Hg 2+ detection in drinking water

Conclusions
In this work, DFT and experimental studies indicate that non-radiative electron transfer between Hg 2+ cations and N, S-GQDs cause dynamic fluorescence quenching. The good sensitivity of N, S-GQDs for Fe 3+ is mainly attributed to the IFE quenching. Doped atoms can produce charge favorable sites, which can interact with Fe 3+ or Hg 2+ , leading to the fluorescence quenching of N, S-GQDs. E ad calculations reveal that the adsorption of Hg 2+ or Fe 3+ are energetically favorable, which also contribute to the non-fluorescence electron transfer. The as-synthesized N, S-GQDs were applied to detect Fe 3+ and Hg 2+ in aqueous solution. The results show that as-synthesized N, S-GQDs have good selectivity and sensitivity to Fe 3+ and Hg 2+ , respectively. The analysis of the fluorescence quenching ratio displays a good linear relationship in both Fe 3+ and Hg 2+ detection. For Fe 3+ detection, a good detection limit of 55.49 nM can be found in the dynamic response range of 0.1 µM to 30 µM, which is lower than the recommended content of Fe 3+ in drinking water by China National Standers. Furthermore, a good detection limit of 0.27 nM with a wide dynamic range (1-30 nM) was obtained for Hg 2+ detection, implying that this method has potential for nanomolar level detection of Hg 2+ in aqueous solution. Therefore, the facile, inexpensive, and sensitive N, S-GQDs present a promising candidate as a fluorescence probe for fluorescence detection.
Supplementary Materials: The following are available online at http://www.mdpi.com/2079-4991/9/5/738/s1, Figure S1: The schematic diagram of detection device geometry and testing process, Figure S2: The stability of fluorescence intensity of as-synthesized N, S-GQDs solutions, Figure S3: The fluorescence intensity of N, S-GQDs in real sample detection, Table S1: The atomic populations of Fe 3+ @N, S-GQDs and Hg 2+ @N, S-GQDs, Table S2: Recovery of Fe 3+ and Hg 2+ detection in drinking water samples.
Author Contributions: Y.Y. synthesized the samples, performed PL experiments, and wrote the main manuscript. X.X. and Z.W. analyzed the data. R.Z., T.Z., S.P., and X.Z. performed XPS experiments. Y.W. revised and edited the paper.