We prepared the N/P CDs via a simple hydrothermal reaction based on a previously reported method (Shangguan et al., 2016). Details on the synthesis of the N/P CDs are provided in the Supporting Information.

Different amounts of analytes (Fe³⁺ and MnO₄ ⁻) were gradually added into 10 μg·mL⁻¹ N/P CDs, and the fluorescence signals were monitored after incubation for 3 min. In order to further verify the recovery effect of F⁻, Fe³⁺ (40 μM) was added to 10 μg·mL⁻¹ N/P CDs, and finally, various concentrations of F⁻ (0–30 μM) were added. To study the selectivity of the N/P CDs for analytes, we introduced other relevant substances into the N/P CD solution and recorded the fluorescence spectrum. The fluorescence emission spectra of each sample were excited at 320 nm and emitted at 392 nm.

Escherichia coli ATCC 25922 (E. coli) was transferred to Luria-Bertani (LB) broth and incubated in a shaker (200 rpm) at 37°C overnight. Finally, the bacteria optical density at 600 nm (OD 600) was adjusted to reach 1.0.

For the detection of selective ions in bacteria, E. coli were cultured with 50 μg mL⁻¹ N/P CDs for 4 h. Next, the N/P CD-stained E. coli were further treated with FeCl₃ and KMnO₄ (200 μM each) solutions for another 2 h at 37°C, respectively. Images of the bacteria were immediately observed by fluorescence microscopy after washing three times with PBS.

Blue fluorescent N/P CDs were obtained through hydrothermal treatment using GMP as the only ingredient (Scheme 1). Because of the excellent solubility of GMP, the synthesis process can be carried out in water without the use of any harmful organic reagents, which was well consistent with green chemistry principles. The hydrothermal synthesis conditions were investigated to obtain the optimal optical properties of N/P CDs through a series of experiments (Supplementary Table S1). The highest QY of N/P CDs (53.72%), which was obtained under the optimized hydrothermal condition at 220°C for 6 h, was much higher than other reported CDs (Qu et al., 2012; Shi et al., 2016; Shangguan et al., 2017; Li et al., 2018; Khan et al., 2021). N, P co-doped effect may be the reason for the high QY of N/P CDs, which can modulate the electronic and chemical behaviors of the CDs. Thus, N/P CDs prepared from GMP had great potential as a promising fluorescent nanoprobe due to their high QY.

The microstructure of the N/P CDs was investigated through TEM (Figure 1A), and the N/P CDs presented well dispersed and spherical structure with an average size of ∼3.0 nm (Supplementary Figure S1). The XRD pattern (Supplementary Figure S2) revealed a wide diffraction peak at about 23.4°, indicating a disordered graphite-like structure (Tang et al., 2012). FT-IR spectra confirmed the surface chemistry of the N/P CDs. As illustrated in Figure 1B, the FT-IR spectra of GMP and N/P CDs were performed. Broad absorption peaks at 3,353 and 3,130 cm⁻¹ confirmed the stretching vibrations of O−H and N−H bonds present on the N/P CD surface. The peak at 1,688 cm⁻¹ indicated C=C and C=O groups. The peaks at 1,608 cm⁻¹ and 1,403 cm⁻¹ contributed to the bending vibrations of N-H and C-N, respectively (Ananthanarayanan et al., 2015). Furthermore, the stretching vibrations of P−O and P=O appeared at 887, 1,152, and 1,301 cm⁻¹, suggesting that the phosphorus element was perfectly adulterated into the C-dots (Du et al., 2014; Shangguan et al., 2017). XPS was performed to analyze the surface states of the N/P CDs. The XPS full survey spectrum (Figure 1C) clearly exhibited that the N/P CDs contained 46.98% carbon (C), 17.06% nitrogen (N), 24.85% oxygen (O), and 3.53% phosphorus (P), and their corresponding peaks were located at 286.1, 399.1, 531.1, and 133.1 eV, respectively. The high-resolution C 1s spectra revealed five major deconvoluted peaks at 284.0 eV (C=C), 284.6 eV (C−C), 285.1 eV (C−N), 286.0 eV (C−O), and 287.5 eV (C=O) (Figure 1D) (Gong X. et al., 2016; Huang et al., 2018). In the deconvoluted N 1s spectrum, three distinct peaks at 398.2, 399.0, and 400.1 eV indicated C=N−H, N−H, and C−N moieties, respectively (Figure 1E) (Huang et al., 2018). Additionally, the high-resolution P 2p spectrum showed two peaks at 132.6 and 133.5 eV, assigned to P−O and P=O, respectively (Figure 1F) (Shangguan et al., 2017). The results of FTIR and XPS spectra suggested that N/P CDs were perfectly adulterated with N and P atoms, and their surfaces contained related functional groups. In addition, the ζ−potential of the N/P CDs was −25.5 mV, suggesting that the surface of the N/P CDs contained the negative functional groups.

First, we investigated the optical characteristics of the obtained N/P CDs by optical spectroscopy. The UV/vis spectra of N/P CDs contained three dominant absorption bands at 240 nm, 280, and 320 nm, ascribing to π −π^∗ transition of C=C, C=N or N=P bonds, and n−π^∗ transition of C=O bonds, respectively (Figure 2A) (Dong et al., 2013). The N/P CD solution was light brown, transparent, and clear under normal light, while it emitted a bright blue color under a 365 nm UV lamp (inserted in Figure 2A). Under 320 nm excitation, the N/P CDs showed a maximum signal at 392 nm. In addition, by varying excitation wavelength from 260 to 380 nm, the N/P CDs showed a typical excitation-dependent fluorescence feature (Figure 2B). The aforementioned result could be attributed to the impact of the surface states of CDs (Zhou et al., 2019). The QY of N/P CDs excited by 320 nm was as high as 53.72%, which was higher than other reported N/P CDs (Zheng et al., 2013; Gong Y. et al., 2016; Khan et al., 2021; Nandi et al., 2021). The fluorescence lifetime of N/P CDs was fitted by a multi-exponential function, giving two decay times, τ₁ = 3.93 ns (40.1%) and τ₂ = 10.55 ns (59.9%), with an average lifetime of 7.89 ns (Figure 2C).

Next, we evaluated the optical stability of the N/P CDs under various conditions, such as pH, salt medium, temperature, and UV-light treatment. The fluorescent signals of N/P CDs at different pH values were investigated, and the results are shown in Figure 2D. At lower and higher pH, N/P CDs showed low fluorescence intensity. However, the fluorescent intensity was maintained constant at a pH range of 4–10. This result may be caused by the protonation−deprotonation of the phosphoric acid group and the amino group on the surface of the N/P CDs (Zhang et al., 2017). Supplementary Figure S3 showed that the N/P CDs remained stable even in a high ionic strength (600 mM NaCl), which was attributed to no ionization of surface functional groups of the N/P CDs (Zhang et al., 2014). At the same time, the fluorescence intensity remained unchanged at the temperatures between 25 and 80°C and continuous illumination for 2 h (Supplementary Figure S4 and Supplementary Figure S5). The excellent thermal stability of the N/P CDs was attributed to the synergic effect of stable composition on the surface of N, P-doped CDs. The oxygen- and nitrogen-rich groups can effectively inhibit the aggregation at a higher temperature and protect the N/P CDs from degradation induced by thermal oxidation (Khan et al., 2018). In addition, the photobleaching-resistant property was possibly ascribed to the electrostatic repulsions between the negatively charged nanoparticles (Huang et al., 2013). The prepared N/P CDs possessed outstanding fluorescence properties and abundant functional groups, which inspired us to investigate their sensing performance.

The N/P CDs were utilized as a promising fluorescence sensor because of their excellent fluorescence properties and water stability. First, we studied the fluorescence response of the prepared CDs toward various metal ions. As presented in Figure 3A, pure N/P CD solution emitted strong blue luminescence, and the fluorescence of the N/P CDs varied from blue to colorless by adding Fe³⁺ ions under UV light. However, the addition of other tested metal ions including Cu²⁺, Hg²⁺, Fe²⁺, Ag⁺, La³⁺, Co²⁺, Mn²⁺, Ba²⁺, Ca²⁺, Mg²⁺, Na⁺, K⁺, and Zn²⁺ showed very little effect on the N/P CD fluorescence. These initial results indicated that N/P CDs can specifically detect Fe³⁺ compared to other tested metal ions. We subsequently evaluated the sensitivity of the prepared N/P CDs toward Fe³⁺ ions by the fluorescence titration experiments. Figure 3B showed a gradual decrease in the fluorescence of N/P CDs with increasing Fe³⁺ concentration and showed 98% fluorescence quenching with 40 μM Fe³⁺. Figure 3C showed a good linear correlation (R ² = 0.998) between the intensity at 392 nm and Fe³⁺ concentration (0.01–16 μM). The limit of detection (LOD) was calculated as 12 nM Fe³⁺ based on the formula, LOD = 3σ/slope, where σ was the standard deviation of blank samples. The LOD was well below the LODs of the previously reported CDs (Supplementary Table S2). The N/P CDs/Fe³⁺ system could be further applied for F⁻ determination. The fluorescence signal at 392 nm recovered progressively with increasing F⁻ concentration (Figure 3D). The recovered fluorescent intensity had a good linear correlation (R ² = 0.996) with F⁻ concentration (0–20 μM) (Figure 3E). The LOD of F⁻ was calculated as 8.5 nM (3σ/slope).

To evaluate the specificity of N/P CDs in testing Fe³⁺, a series of anti-jamming experiments were carried out (Supplementary Figure S6A). Most metal ions showed neglected influence on Fe³⁺ detection, indicating that N/P CDs had excellent selectivity for Fe³⁺. On the other hand, the anti-interference sensing ability of the N/P CDs/Fe³⁺ system toward F⁻ was further ascertained by the competing experiments. Only F⁻ recovered the fluorescence of the N/P CDs/Fe³⁺ system; other anions did not cause obvious fluorescence enhancements (Supplementary Figure S6B).

In order to further confirm that the system we developed had potential biological applications, the fluorescence response of N/P CDs toward Fe³⁺ and F⁻ was explored in HEPES buffer at different pH ranges. N/P CDs can maintain stable fluorescence from pH 4–8 (Supplementary Figure S7), indicating their outstanding photostability in a wide pH range. A notable fluorescence quenching was observed after adding Fe³⁺ with pH from 1 to 12, suggesting that N/P CDs were effectively quenched by Fe³⁺ from pH 1–12. In addition, a stable fluorescence-recovered phenomenon of N/P CDs/Fe³⁺ system to F⁻ was observed at pH 4–10. Thus, N/P CDs can sequentially detect Fe³⁺ and F⁻ at biological pH values.

Time-dependent fluorescence variation of N/P CDs to Fe³⁺/F⁻ sequential detection was investigated (Supplementary Figure S8). N/P CDs maintained stable fluorescence under aqueous media, indicating good water solubility and stability abilities. However, with the addition of Fe³⁺, the fluorescence of N/P CDs was rapidly quenched within 30 s and then remained stable thereafter (Supplementary Figure S8A). Subsequent addition of F⁻ to the CDs/Fe³⁺ system resulted in fluorescence recovery within 30 s, which indicated that Fe³⁺ can be released from N/P CD surfaces (Supplementary Figure S8B). The aforementioned results demonstrated that N/P CDs could be employed for the real-time Fe³⁺ and F⁻ determination.

Reusability was an important characteristic of a probe used for practical detection. A switchable change in the fluorescence signal at 392 nm could be repeated, and no obvious signal loss was detected by the alternate introduction of Fe³⁺ and F⁻ (Supplementary Figure S9), indicating that N/P CDs can be reused for Fe³⁺ and F⁻ detection.

The interaction between Fe³⁺ and the multifunctional groups (amino, phosphoric acid, and carboxyl groups) of N/P CDs could modulate the fluorescence properties. Fe³⁺ quenched the fluorescence of N/P CDs because of the strong complexation interaction between them (Mohammed and Omer, 2020). The average fluorescence lifetime of N/P CDs was 7.89 ns However, with Fe³⁺ and F⁻, the average lifetimes of N/P CDs were 7.85 [τ₁ = 4.48 ns (40.7%) and τ₂ = 10.16 ns (59.3%)] and 7.87 [τ₁ = 5.16 ns (35.3%) and τ₂ = 9.35 ns (64.7%)] ns (Figure 4A), respectively, suggesting Fe³⁺−induced fluorescence quenching was probably a static quenching process (Lu et al., 2017). To further clarify this deduction, UV/vis absorption spectra, FT-IR, and zeta-potential analysis were implemented. Absorption spectra of N/P CDs were notably affected and changed after the addition of Fe³⁺ (Figure 4B), which was attributable to the strong chemical interaction between Fe³⁺ and N/P CDs followed by stable metal complex formation (Rajendran et al., 2021). Absorption peaks of N/P CDs shifted to their former position again with the addition of F⁻. Comparing the N/P CDs before and after treating with Fe³⁺, the characteristic vibration peaks of C=O, P−O, and P=O groups (1,688, 887, and 1,301 cm⁻¹) demonstrated obvious changes (Figure 4C) (Shangguan et al., 2017). Meanwhile, the zeta potential increased from −25.5 to +32.4 mV when Fe³⁺ was present and then reversed back to −6.7 mV for the N/P CDs/Fe³⁺ system after treating with F⁻ (Figure 4D), which proved that Fe³⁺ was indeed selectively complexed with the negatively charged groups of the CDs.

Simultaneously, various anions were used to demonstrate the anion-sensing abilities of the N/P CDs based on similar research methods for metal cations. In order to evaluate the sensing behaviors of the N/P CDs toward anions, we separately added 19 anionic salt aqueous solutions (50 μM) to the aqueous suspension of N/P CDs (10 μg·mL⁻¹). The fluorescence of the N/P CDs changed from blue to colorless in the case of MnO₄ ⁻ treatment under UV light excitation (365 nm) (Figure 5A). However, adding other tested anions did not induce conspicuous fluorescence changes in the N/P CD solution.

The sensitivity of the N/P CDs toward MnO₄ ⁻ in water was explored by performing the concentration−dependent titration experiments. Upon introducing various concentrations of the MnO₄ ⁻ (0–50 μM) into the N/P CD solution (10 μg·mL⁻¹), the fluorescence signals of the N/P CDs gradually decreased and showed a complete quenching response by MnO₄ ⁻ (Figure 5B) at the maximum concentration. Furthermore, Figure 5C showed a good linear relationship of the fluorescence values at 392 nm versus the concentrations of MnO₄ ⁻ (0–20 μM). The LOD of MnO₄ ⁻ was 18.2 nM. The detection performance of the N/P CDs for MnO₄ ⁻ in this work was better than that of the previously reported literature presented in Supplementary Table S3.

The competition tests should be conducted to confirm the muscular anti-interference ability of N/P CDs toward MnO₄ ⁻ in the presence of other 17 inorganic anions (Supplementary Figure S10A). The results indicated that other interfering anions did not induce significant changes in fluorescence, suggesting that MnO₄ ⁻ anions can be selectively distinguished by N/P CDs in the coexistence of other inorganic anions.

Time-dependent fluorescence fluctuation to detect MnO₄ ⁻ with N/P CDs was explored. When mixing various concentrations of MnO₄ ⁻ with N/P CD (10 μg·mL⁻¹) suspension, the fluorescence signal progressively decreased (Supplementary Figure S10B). In addition, the N/P CDs showed a significant quenching effect on MnO₄ ⁻ at various pH ranges (1–13) (Supplementary Figure S10C), which can be applied to detect and image MnO₄ ⁻ in biological systems.

To clarify the detection mechanism, we recorded UV−vis spectra of the MnO₄ ⁻ in water (Supplementary Figure S11A). The excitation spectra of N/P CDs (260–380 nm) displayed broad overlaps with UV−vis spectra of MnO₄ ⁻ (270–600 nm), suggesting that the internal filtration effect (IFE) was responsible for fluorescence quenching (Li et al., 2018; Ji et al., 2020). Additionally, the emission spectra of N/P CDs overlapped with the UV−vis spectrum of MnO₄ ⁻, indicating that resonance energy transfer (RET) was possibly produced from N/P CDs to analytes (Li et al., 2018). To further confirm whether RET happened during the MnO₄ ⁻ sensing process, the fluorescence lifetimes of N/P CDs were measured with or without MnO₄ ⁻. Lifetimes were as follows: 7.89 ns (before the addition of MnO₄ ⁻) and 7.87 ns (after the addition of MnO₄ ⁻; Supplementary Figure S11B). The lifetimes had almost no change in the presence of MnO₄ ⁻, which indicated that the static quenching effect (SQE) was formed between N/P CDs and MnO₄ ⁻. In the RET process, target molecules can significantly change the fluorescence lifetime of the fluorophore (Joseph and Anappara, 2016; Liu et al., 2016; Sun et al., 2016; Li et al., 2018). Thus, RET was not the main reason to induce the fluorescence quenching. In conclusion, IFE and SQE played a main role in inducing the fluorescence quenching process.

To demonstrate the potential analytical application in a real sample, quantitative analytes’ (Fe³⁺, F⁻, and MnO₄ ⁻) detection tests were carried out in lake water by the standard spiking method. The results are listed in Supplementary Table S4, and the recoveries changed between 97.6 and 102.9% with relative standard deviations (RSDs) lower than 1.1%. Therefore, N/P CDs had excellent potential for these analytes’ detection in actual water samples.

To study the potential biomedical applications of the N/P CDs, the cytotoxicity assay was first examined using HeLa cells by the traditional MTT method. Approximately 90% of cells still survived even with 200 μg·mL⁻¹ N/P CDs for 24 h (Figure 6), suggesting that N/P CDs were nontoxic in nature and displayed good biocompatibility. In order to investigate the potential biological imaging functions of N/P CDs, fluorescence imaging was performed in E. coli to monitor Fe³⁺ and F⁻. As displayed in Figure 7, E. coli stained by 50 μg·mL⁻¹ N/P CDs emitted bright green and red emissions when stimulated by 488 and 552 nm lasers, respectively, suggesting that N/P CDs were internalized by E. coli. After treatment with 200 μM Fe³⁺ for 30 min, the fluorescence signals of E. coli with green and red emissions disappeared. Subsequently, the fluorescence recovered when 150 μM of F⁻ was treated with the aforementioned bacteria, and an evident “on−off−on” fluorescence response appeared. Subsequently, the N/P CDs also were employed to detect MnO₄ ⁻ in E. coli by fluorescence microscopy. As shown in Supplementary Figure S12, no emission was observed for MnO₄ ⁻ (200 μM)-treated E. coli under 488, and 552 nm laser excitation, respectively. But without MnO₄ ⁻, a bright green and red emission was observed for only N/P CDs (50 μg·mL⁻¹) treated with E. coli. These results suggested that the N/P CDs could be applied in the visual detection of Fe³⁺, F⁻, and MnO₄ ⁻ in bacteria and possessed great promise in bioimaging and biosensing applications.