All the Chenpi used in this work was purchased from a local pharmacopoeia. Before use, the Chenpi was washed and crushed. Other chemicals were purchased from Sigma Aldrich and were utilized without purification (analytical grade). Ultrapure water (≥18 MΩ cm⁻¹; Millpore Co., USA) was used throughout the experiment.

To characterize the samples, a transmission electron microscope (TEM) was used to produce TEM images of C-dots that were further analyzed by JEOL-1230 TEM (JEOL, Japan). Fourier transform infrared (FT-IR) spectroscopy analysis was also used to characterize the C-dot samples and was performed using a USA spectrometer (Nicolet Nexus 670 FT-IR spectrometer) with a spectral area of 4000–400 cm⁻¹. The ultraviolet visible (UV-vis) spectrum was recorded via a UV-2450 ultraviolet visible spectrophotometer (Shimadzu Company, Japan). In addition, this paper used a Thermo Fisher Scientific K-Alpha 1063 X-ray photoelectron spectrometer (USA) to conduct X-ray photoelectron spectroscopy (XPS) analysis using Al and K excitation sources. Fluorescence spectrum analysis was conducted via a F-7000 fluorescence spectrophotometer (Hitachi Co., Ltd., Japan). Quinine sulfate was used to calculate the C-dots’ fluorescence quantum yield [11].

The carbon source used in this study was Chenpi, and the C-dots were made using a hydrothermal method. First, 2.0 g dried tangerine peel powder was poured into 10 mL ultrapure water. After a period of time, when the mixture was thoroughly stirred, it was transported to an autoclave (25 mL, Teflon lined) and then warmed to 180°C in an oven for 24 h. When the temperature gradually decreased to ambient conditions, the solution was taken out and centrifuged at 12000 rpm for 20 min. Next, the deionized water was dialyzed using a dialysis membrane with a molecular weight cutoff (MWCO) of 1 kD for a total of 3 days. Finally, the solution was dispersed in deionized water for subsequent experiments. Note: Solid C-dots were obtained via a freeze-drying process to calculate the concentration of C-dots.

This paper used 10 mM phosphate buffer solution (PB) to detect Hg²⁺ (pH 7.0, room temperature). First, 20 µL of C-dot solution, with a final concentration of 0.6 mg mL⁻¹, was poured into 880 µL PB buffer solution. Next, 100 µL of Hg²⁺ was added to the PB solution at specific concentrations. The solution was incubated at an ambient temperature of 20°C for 10 min. Then, the corresponding fluorescence emission spectrum was obtained at ambient temperature utilizing a 340 nm excitation wavelength.

In this experiment, tap water samples were obtained from different locations in the lab. The samples that were collected include river water and mountain spring water. All the water samples were subjected to a 24-h natural sedimentation treatment and filtered using filter paper with a maximum pore size of 15–20 µm to remove the suspended river and lake particles. Next, the water sample was centrifuged at 10000 rpm for 20 min. All of the collected supernatants were stored at 4°C. A 0.22 µm microporous membrane was used for sample filtration.

The size and morphology of the C-dots were examined using TEM. According to Fig. 1A, it is apparent that the C-dots have a quasi-spherical shape, and the nanoparticles are well dispersed. Approximately 100 particles were counted in order to obtain their size distribution histogram. These nanoparticles have a diameter of 8.6 ± 2.7 nm, obtained from Fig. 1B.

The functional groups of the C-dots were analyzed using FT-IR spectroscopy. According to Fig. 2A, an absorption band of 3413 cm⁻¹ corresponds to the stretching vibrations caused by different bonds, such as the N-H and O-H bonds [12]. Additionally, a strong band of 1604 cm⁻¹ relates to the vibrations resulting from the aromatic stretching behavior of C=C. Another peak at 2933 cm⁻¹ relates to the multiple vibrations caused by the bond stretching of C-H [13]. Moreover, another peak observed at 1760 cm⁻¹ is related to the uneven and asymmetric vibrations caused by COO⁻, with the peak at 1402 cm⁻¹ corresponding to the even and proportional stretch vibrations [14]. We also found a distinctive absorption band that can be used to classify C-N around 1209 cm⁻¹, meaning that an N atom was doped into the C point [15]. According to the analysis, a maximum of 1071 cm⁻¹ relates to the bending vibrations caused by the bonds in C-O [16].

The chemical configuration and surface elements of the C-dots were examined using XPS. According to Fig. S1, the primary strong peaks of the C-dots are located at 284.8, 399.7, and 531.9 eV, and are related to C1s, N1s, and O1s peaks, respectively [12]. According to the calculation and analysis results, the C, N, and O contents were 63.76%, 4.37%, and 31.86%, respectively. We deconvoluted the high-resolution C1 spectrums at different locations (284.8, 286.1, 286.6, and 287.9 eV) to obtain the corresponding peaks and assigned them to the C-C/C=C, C-OH, C-N, and C-O/C=O groups (Fig. 2B) [13]. The high-resolution N1s spectrum results were interpreted and shown in Fig. 2C. A peak at 399.7 eV was allocated to the C-N group, while 401.8 eV was allocated to N-H group [14]. The O1s spectrum in Fig. 2D shows two fitting peaks eV, of which the peak at 531.2 had been allocated to group C=O, whereas 532.4 was assigned to the group C-O [13]. The results strongly suggest that the C-dots that were synthesized were functionalized with several groups, such as amino, carboxyl, and hydroxyl groups. These functional groups can effectively enhance the C-dots water solubility needed for sensing applications in aqueous solutions [15].

The fluorescence and UV visible absorption spectrum were further examined to characterize the optical properties of the C-dots. According to Fig. 3A, the C-dots have a relatively wide UV absorption band. The characteristic absorption near 290 nm is a result of π-π∗ conversion in the C=C group. The C-dot’s maximum fluorescence excitation/emission wavelengths are 340 and 429 nm. In Fig. 3A the blue fluorescence at a wavelength of 365 nm is shown. Utilizing quinine sulfate as a reference, we were able to determine the fluorescence quantum yield and obtained a value of 10.8% [11]. Fig. S2 shows the fluorescence spectra of the C-dots with different excitation wavelengths in water. Furthermore, as the excitation wavelength was progressively increased from 260 to 440 nm the maximum emission gradually shifted from 394 to 500 nm. During this process, the emission intensity gradually decreased. The reason for this is due to the surface energy trap or the incomplete size inhomogeneity of some of the functional classes on the surface of the C-dots [14].

In our analysis, we analyzed the C-dots fluorescence stability under different experimental conditions. The research methods used include fluorescence spectrum analysis and the change of fluorescence intensity as a result of photobleaching, ionic strength, and pH value. It is clear from Fig. S3A that the fluorescence intensity reduces under alkaline and acidic conditions, with the highest value occurring at pH 7. Therefore, further sensing experiments were carried out in neutral media. We exposed the C-dots to a NaCl mixture containing various proportions to analyze the fluorescence intensity variations resulting from ionic strength. Fig. S3B shows that when the NaCl concentration gradually reaches 4.0 M, the fluorescence intensity does not significantly change. A Xe lamp measured photostability via constant UV (360 nm) irradiation exposure. The change in fluorescence intensity under continuous UV irradiation is shown in Fig. S3C. The fluorescence intensity changed minimally over 1 h, revealing that the C-dots can resist photobleaching. Moreover, the C-dots exhibit strong stability in the air when stored for a month at room temperature (Fig. S3D). According to the above data, the C-dots synthesized in this paper have excellent stability, thus, increasing their potential use in numerous applications.

There are different types of oxygen-containing functional groups (that include hydroxyl and carboxyl groups) on the surface of the C-dots, making them an excellent candidate for detecting metal ions [13,15]. In this study, Hg²⁺ was added to the C-dots aqueous mixture, and a thorough analysis of the fluorescence response of Hg²⁺ to the C-dots was conducted. According to Fig. 3B, when there is no Hg²⁺, the C-dots have a relatively strong fluorescence emission. In the presence of 200 nM Hg²⁺, the fluorescence of the C-dots is severely weakened. This analysis revealed that the efficient quenching of the C-dots fluorescence might result from the relatively sturdy chelating properties of the carboxyl group for mercury on the surface of C-dots, which gradually forms non-fluorescent complexes. Additionally, a non-radiative electron/hole recombination process occurs, quenching the static fluorescence caused by the annihilation of the electron transfer process [17,18]. This quenching mechanism has been demonstrated in previous studies, where the coupling agents such as cysteine can restore the quenched fluorescence of the C-dots [19]. Moreover, after adding cysteine to the solution, the fluorescence recovers to some extent as shown in Fig. 3B, and is consistent with our expected results. The fluorescence quenching of the C-dots occurs primarily due to the chelation between the C-surface groups and Hg²⁺.

The incubation time, C-dot concentration, solution pH, and other parameters were optimized to improve the detection sensitivity. The highest fluorescence intensity value occurred when the pH was 7.0 as displayed in Fig. S3A. Based on this, we used a pH 7.0 buffer to achieve the highest fluorescence intensity. According to Fig. S4A, the fluorescence quenching rate [(F₀-F)/F₀] is highest when the C-dot concentration was 0.6 mg mL⁻¹ and therefore was the best detection concentration. Furthermore, we discussed the dependence between the fluorescence quenching efficiency of Hg²⁺ and the reaction time (Fig. S4B). After adding 200 nM Hg²⁺, the fluorescence quenching gradually reaches a constant value within 10 min and, according to the above results, has rapid quenching kinetics.

We adopt the best determination conditions to comprehensively analyze and evaluate the Hg²⁺ detection method proposed in this study as well as discuss its analytical performance. The fluorescence spectrum analysis results of the C-dots synthesized under various Hg²⁺ concentrations are shown in Fig. 4A. The fluorescence intensity of the C-dots is significantly reduced after gradually increasing the concentration of Hg²⁺. According to Fig. 4B, there exists a relation among the fluorescence intensities and the mercury’s mass between 10–300 nM, which appears to follow a linear behavior. The calibration equation is as follows: y = 0.0018x + 0.00567 (R² = 0.9798), where x represents the concentration of Hg²⁺ and y represents the fluorescence quenching rate [(F₀-F)/F₀]. In addition, this study calculated the limit of detection (LOD) according to a three-times standard deviation, and the result (7.0 nM) exceeded the safety level (10 nM) that is set as a standard requirement by the US Environmental Protection Agency for drinking water. These results indicate that it is an ideal candidate for detecting mercury in water specimens. Table S1 lists a comparative analysis between the results obtained in this paper and the Hg²⁺ detection methods previously reported [9,17,20,21–26]. The results show that this detection method is analogous or even superior to most C-dot detection methods.

Furthermore, careful selection can be regarded as a critical factor when evaluating the efficiency of the sensor systems. When there are representative metal ions, the fluorescence intensity will change. The influences of some of the potentially competing substances in natural water specimens include Hg²⁺, Al³⁺, Cu²⁺, Ca²⁺, Fe³⁺, Cd²⁺, Ba²⁺, Na⁺, Mg²⁺, Zn²⁺, Co²⁺, Pb²⁺, Ni²⁺, Mn²⁺, and K⁺ were investigated. Fig. S5 (black column) shows that the Hg²⁺ causes greater fluorescence quenching than the other ions. Furthermore, Hg²⁺ and each potentially competing substance were mixed for interference testing. As shown in Fig. S5 (red column), the other substances show negligible interference in the detection of Hg²⁺, showing superior selectivity for Hg²⁺. The results show that this detection method has good selectivity and specificity towards Ho2+, which is consistent with that reported in the iterature [27–31].

To verify the practical application of this method, we use the basic addition method to analyze and determine Hg²⁺ concentrations in the stream, pond, and tap water. All the water samples were subjected to a 24-h natural sedimentation treatment, and a filter paper with a maximum pore size of 15–20 µm was used to filter the suspended particles in the river and lake water. Table 1 shows that no Hg²⁺ in the original water samples was detected. All samples were spiked with a known amount of Hg^2+, and the recovery and relative standard deviation (RSD) were calculated. According to Table 1, the recovery of all the water samples ranged from 92.9%–111.5%, and the RSD was less than 7.2%, verifying that this method can be applied to detect Hg²⁺ in water samples.