The drugs and instruments used in the experiment are detailed in the supporting information.

The Synthesis of probe was achieved through a hydrothermal method. The specific synthesis process is shown in Fig. 1. Firstly, dissolve 1 mmol acetylcysteine in deionized water via ultrasonication. Then, add 1 mL APTES and stir for 10 minutes to ensure uniform mixing. Subsequently, place the above solution in a polytetrafluoroethylene reactor and react at 180 degrees celsius for 10 hours. After the solution is cooled, separate the solid precipitate by centrifugation, take the upper transparent solution, and filter it through a 0.22 µm membrane to obtain a pure carbon quantum dot solution. Finally, the carbon quantum dot solution was freeze-dried to obtain a solid powder, which is reserved for further use.

The absolute quantum yield of the probe is calculated using the formula proposed by [29]:

In this formula, QY represents the absolute quantum yield, L_emission denotes the number of fluorescence emission photons from N,Si-CDs, and E_solvent and E_sample represent the number of photons absorbed by the solvent and sample, respectively, under the excitation light source. According to this equation, the absolute quantum yield of the probe was determined to be 16%.

Fluorescence titration experiments were carried out in probe solutions with a series of concentrations of CAP or TC. The procedure was as follows: firstly, prepare a solution of N, Si-CDs at a concentration of 0.5 mg/mL.Then, add differen concentrations of CAP and TC to separate probe solutions respectively to study their fluorescence responses. In the selectivity experiment, 15 µM of each interfering substance— neomycin, florfenicol, erythromycin, streptomycin sulfate, amoxicillin, curcumin, L-threonine, hcy (DL-homocysteine), D-2-phenylglycine, L-his, L-methionine, L-leucine, L-lys, Ca²⁺、Cd²⁺、Ni²⁺、Mn²⁺、Hg²⁺、K⁺、Mg²⁺、Zn²⁺、Al³⁺、Co²⁺ —was used to replace CAP and TC, and the fluorescence spectra were measured under identical conditions.

Honey (Beekeeping Mother, Yucheng Changpeng Yangfeng Co., Ltd), yili milk powder (Yili, Heilongjiang Yili Dairy Co., Ltd.), vitamin c effervescent drink (Compound Fruit Juice Drink brand, Nongfu Spring), plum juice (C'estbon, C'estbon Water Co., Ltd.), and lemon tea (C'estbon, C'estbon Water Co., Ltd.) were selected for real-sample detection. Honey was purchased from suguo supermarket, yili milk powder was purchased online, and the Vitamin C effervescent drink, plum juice, and lemon tea were purchased from the Education Supermarket of Nanjing Normal University. Dilute 1 mL of honey to 30 mL, centrifuge the solution, take the supernatant, and filter it through a 0.22 µm membrane. For Yili Milk Powder, weigh 0.1 gram and dissolve it in 10 mL of hot water, add an appropriate amount of acetonitrile to precipitate proteins, sonicate the mixture for 1 minute, centrifuge, and filter the supernatant through a 0.22 µm membrane. Dilute 1 mL of the Vitamin C effervescent drink, plum juice, and lemon tea to 20 mL, 20 mL, and 30 mL, respectively, for later use. To perform quantitative detection of CAP and TC in real samples, add the pretreated sample to the probe solution and record the fluorescence emission spectra. Specifically, add varying concentrations of CAP and TC standards to separate probe solutions, measure the fluorescence emission spectra, and repeat each measurement three times.

Carbon quantum dots were observed via HRTEM. As shown in Fig. 2 (a), they are spherical with an average size of 8.27 ± 0.58 nm (Fig. 2(b)). FT-IR and XPS analyses of N,Si-CDs were performed to identify functional groups and composition. From Figure S1, the FT-IR spectrum revealed that the surface of N,Si-CDs contained O-H (3430 cm^− 1) [30] (telescopic vibration), C-H (2933 cm^− 1) [31] (telescopic vibration), C = O (1631 cm^− 1) [32] (telescopic vibration), N-H (1567 cm^− 1) [33] (bending vibration), C-N (1409 cm^− 1) [34] (telescopic vibration), C-O (1124 cm^− 1) [19] (telescopic vibration), Si-O (1034 cm^− 1) [35] (telescopic vibration), Si-C (850 cm^− 1) [36] (telescopic vibration), Si-O (784 cm^− 1) [37] (telescopic vibration) and other groups. The XPS survey spectrum (Fig. 2(c)) exhibited five main peaks corresponding to O1s, N1s, C1s, Si2s, and Si2p, indicating successful doping of N and Si into the CDs. The elemental proportions were C: 53.24%, N: 8.55%, O: 21.19%, and Si: 17.02%. The high-resolution C1s spectrum (Fig. 2(d)) showed three peaks at 284.51 eV, 285.85 eV, and 288.05 eV, corresponding to C-C, C-O/C-N, and C = O, respectively. The N1s spectrum (Fig. 2(e)) revealed peaks at 399.1 eV and 401 eV, attributed to N-H and C-N bonds. The O1s spectrum (Fig. 2(f)) exhibited peaks at 530.9 eV and 532 eV, corresponding to Si-O and C-O groups. These results are consistent with the FT-IR analysis, further confirming successful doping of N and Si into the CDs and the presence of hydrophilic functional groups (e.g., -OH, -COOH and -NH₂) on the probe surface. XRD analysis (Figure S2) showed a distinct diffraction peak at 2θ = 21.4 °, a typical characteristic of carbon quantum dots [38], indicating good crystallinity of the synthesized N,Si-CDs. This result aligns with the reported diffraction behavior of carbon quantum dots in the literature.

Further studies on the optical properties of the probe were conducted. As shown in Fig. 3(a), a clear UV absorption peak at 290 nm was observed, which is attributed to the characteristic π-π*electron transitions in the carbon nuclei. Meanwhile, the probe showed a strong fluorescence emission peak at 403 nm, emitting bright blue fluorescence. To further verify its luminescent properties, the CIE color coordinates of probe were measured, yielding results of (0.1522, 0.0710) (Fig. 3(b)). This confirms that the probe emits blue fluorescence under UV excitation. From Figure S3, the fluorescence intensity of the probe varied with the excitation wavelength. When the excitation wavelength was between 250–270 nm, the fluorescence intensity gradually decreased; between 290–320 nm, it increased significantly; and between 330–340 nm, it decreased again. Notably, as the excitation wavelength increased, no significant blue or red shift was observed in the emission peak, indicating its independence from the excitation wavelength.

We systematically studied the stability of the probe in terms of chemical stability, photostability, and thermal performance. First, the fluorescence changes of probe in different concentrations of sodium chloride solution were studied, as shown in Fig. 4(a), the results showed that the fluorescence intensity slightly decreased in 0.1-1.0 mol/L NaCl solutions, indicating that the probe maintained good chemical stability in high-ionic-strength environments. The fluorescence response of N,Si-CDs to varying pH values was further studied as shown in Fig. 4(b). The fluorescence intensity gradually increased when the pH ranged from 3 to 7 but decreased when the pH increased from 7 to 12. The probe exhibited optimal fluorescence stability under neutral conditions, whereas acidic or alkaline environments may damage surface functional groups, leading to fluorescence quenching [39]. We also investigated the fluorescence stability of N,Si-CDs after adding CAP and TC. As shown in Figure S4(a, b), although the fluorescence intensity decreased upon the addition of CAP or TC, it remained stable within 1 hour, indicating good temporal stability during the detection process. Photostability test revealed that continuous irradiation at 365 nm for 1 hour caused only minor changes in fluorescence intensity (Figure S5), demonstrating excellent photostability. At the same time, the long-term stability test shows that after being stored for 30 days, from Figure S6, although the fluorescence intensity has decreased, it still has good fluorescence performance.

Finally, the thermal performance of the probe was analyzed via thermogravimetric analysis (TGA) (Figure S7), the results indicate that when the temperature increased to 200°C, the probe exhibited approximately 22% weight loss, likely due to the evaporation of adsorbed water or the disruption of weak interactions. At 700°C, the weight loss reached 46%, attributed to the decomposition of functional groups. Above 700°C, the weight remained nearly constant, indicating that N,Si-CDs possess certain structural stability at high temperatures. These results collectively demonstrate that N,Si-CDs show good stability under chemical, optical, and thermal aspects, supporting their reliability in practical applications.

Due to the complexity of environmental and biological systems, the probe need to possess excellent selectivity for practical applications. Therefore, we selected a series of common interfering substances for selectivity testing, including CAP, TC, neomycin, florfenicol, erythromycin, streptomycin sulfate, amoxicillin, curcumin, l-threonine, dl-homocysteine (Hcy), d-2-phenylglycine, l-his, l-methionine, l-leucine, l-lys, and various metal ions (Ca²⁺, Cd²⁺, Ni²⁺, Mn²⁺, Hg²⁺, K⁺, Mg²⁺, Zn²⁺, Al³⁺, Co²⁺). After adding these substances to the probe solution, their fluorescence spectra were measured. As shown in Fig. 5(a). The results indicated that only the addition of CAP and TC significantly decreased the fluorescence intensity, while the influence of other substances was negligible. This demonstrates that N,Si-CDs exhibit outstanding selectivity for CAP and TC. To evaluate the anti-interference ability of the probe, we investigated its specific recognition of CAP and TC in the presence of interfering substances. From Fig. 5(b, c), when adding other interfering substances to the probe solution, the fluorescence spectrum of the probe remained largely unchanged when other substances were added. This result indicates that the probe can specifically recognize CAP and TC in complex environments, exhibits good anti-interference performance.

To investigate the relationship between CAP/TC concentrations and changes in the probe's fluorescence, we performed fluorescence titration experiments and recorded the corresponding spectral changes. As shown in Figs. 6(a, b), when CAP increases to 40 µM, the fluorescence weakens at 403 nm. When CAP is between 10–40 µM, there is a linear relationship between fluorescence intensity and CAP concentration. The equation is Y = -113.6x + 6559.9 (Y and x represent fluorescence intensity and CAP concentration, respectively), and the correlation coefficient R² = 0.9943. Based on the calculation formula for the detection limit LOD = 3σ/k, the detection limit of CAP was determined to be 0.54 µM, σ and k are the standard deviation of the blank sample and the slope of the equation, respectively.

From Figs. 6(c, d), when the amount of TC in the probe solution increases, the fluorescence decreases at 403 nm and there is a red shift. There is a linear relationship between the fluorescence intensity and TC concentrations from 10–30 µM. The equation is Y = -134.24x + 4929.8 (Y and x represent fluorescence intensity and TC concentration, respectively), and the correlation coefficient R² = 0.9843. Based on the calculation formula for the detection limit LOD = 3σ/k (0.45 µM), σ and k are the standard deviation of the blank sample and the slope of the equation, respectively. Compared with other sensors for detecting CAP and TC (Table S1), N,Si-CDs showed excellent detection performance in this study.

We conducted the following tests on the sensing and detection mechanisms of CAP and TC respectively. For TC detection, first, we conducted Zeta potential detection on N, Si-CDs and N, Si -CDs + TC, as shown in Figure S9. The results were − 19.9 mV and − 24.7 mV, respectively, indicating that there is no electrostatic attraction between the probe and TC. We also measured the UV absorption spectra of the probe before and after adding TC, as shown in Figure S8 (a). After adding TC, there is an absorption peak around 370 nm, prove a non-fluorescent ground-state complex might be formed between TC and N,Si-CDs. We speculated that it was a static quenching mechanism [19]. In order to avoid the absorption peak of TC itself, we separately tested the UV absorption spectrum of TC and compared it with N, Si-CDs and N, Si-CDs + TC. TC showed an absorption peak around 350 nm (as shown in Figure S8 (a)), and the absorption peak shifted after adding TC to the probe, proving the formation of non fluorescent ground state complexes. In addition, we tested the fluorescence lifetime of the probe and probe + TC, as shown in Fig. 7 (a, b). The fluorescence lifetime changed from 7.35 ns to 7.39 ns. The fluorescence lifetime can be calculated according to the following formula: τ_avg = τ₁B₁ + τ₂B₂ + τ₃B₃, τ is the lifespan, B is the weight (in Table S2). This subtle change further confirms that the mechanism for detecting TC is static quenching [40].

For CAP detection, first, we conducted Zeta potential detection on N, Si-CDs and N, Si-CDs + CAP, as shown in Figure S9. The results were − 19.9 mV and − 25.8 mV, respectively, indicating that there is no electrostatic attraction between the probe and CAP. We also measured the UV absorption spectra of the probe before and after adding CAP, as shown in Figure S8 (b). There was no significant change after adding CAP. In addition, we tested the fluorescence lifetime of the probe and probe + CAP, as shown in Fig. 7 (a, c). The fluorescence lifetime changed from 7.35 ns to 7.50 ns. The fluorescence lifetime can be calculated according to the following formula: τ_avg = τ₁B₁ + τ₂B₂ + τ₃B₃, τ is the lifespan, B is the weight (in Table S2). We further investigate the UV absorption spectra, excitation spectra, and emission spectra between CAP and the probe. In Figure S10, it can be seen that there is a clear overlap between the three. In summary, it can be concluded that the quenching mechanism of the probe on CAP is the internal filtration effect (IEF) [41].

To establish a convenient on-site detection method for CAP and TC, a smartphone-based analysis platform was developed. Visual detection of target substances was achieved through the ColorPicker application, as shown in Fig. 8. The specific operation process is as follows: gradient-concentration CAP/TC solutions (0–40 µM) were mixed with N,Si-CD probes respectively, and the solution colors were captured under ultraviolet light. After obtaining the color information through the smartphone camera, the clorPicker application automatically converted the chromaticity signal into RGB parameters. The intensity of the blue channel was selected as the quantitative analysis index, and there is a linear relationship between the antibiotic concentration and the B value. Within the range of 5–30 µM, the relationship between CAP and B values is shown in the equation: B = -1.8x + 160 (R²=0.9972), and LOD = 3σ/k (33.7 µM). Within the range of 10–40 µM, the relationship between TC and B values is shown in the equation: B = -2.4357x + 181.36 (R²=0.9964), with a LOD of 24.9 µM (calculated as 3σ/k). This sensing mode exhibited excellent concentration-dependent response characteristics.

To verify the practicality and accuracy of N,Si-CDs in real-world environments, we used the standard addition method to detect CAP and TC in samples of honey, Yili milk powder, Vitamin C Effervescent Drink, Plum juice, and Lemon tea. Tables 1 and S3 show that the recovery rates of CAP in real samples ranged from 93% to 110.9%, and those of TC ranged from 95.3% to 118.4%. The results indicate that N,Si-CDs have broad application prospects for the detection of CAP and TC in actual samples.