All the reagents were bought from Aladdin and the reagents were of analytical grade. Ultrapure water (18.2 MΩ cm) was obtained using the Millipore water purification system (Millipore, MA, United States).

Fluorescence data were recorded on a FluoroMax-4 Compact Spectrofluorometric (HORIBA, United States). Transmission electron microscopy (TEM) images were recorded using a JEOL 2020 transmission electron microscope. The UV-visible (UV-vis) absorption was obtained using T6 New Century spectrometer. Fourier transform infrared (FT-IR) spectroscopy measurements were recorded on a UERTEX-70 FT-IR spectrometer (Thermo-Fisher, United States). X-ray photoelectron spectroscopy (XPS) analysis was performed using an EscaLab 250xi spectrometer. Powder patterns of the samples were collected using a MiniFlex600 X-ray powder diffractometer. The RGB color intensity of the fluorescence photographs were captured using a smartphone under a WFH-204B portable UV lamp (365 nm).

All glassware was washed with aqua regia (V_HCl: V_HNO3 = 3:1) before the start of each experiment. The NCDs and Au NCs were prepared according to a literature procedure (Xie et al., 2018) and (Xie et al., 2009), respectively.

By introducing various Hg²⁺ concentrations (0.05–2.5 µM) into the above ratio fluorescent probe solution, the sensitivity of the response to Hg²⁺ was assessed. Under 400 nm excitation, the 410–780 nm fluorescence spectrum was recorded. Fluorescent probe solutions were formulated by mixing NCDs and Au NCs in 2 ml phosphate buffered saline (PBS) buffer (10 mM, pH = 7.4) at a fluorescence intensity ratio of 1:1. In a homemade closed box, color changes were observed under a 365 nm UV lamp. A variety of common metal cations, including Na⁺, K⁺, Ca²⁺, Ni²⁺, Co²⁺, Cd²⁺, Ba²⁺, Fe³⁺, Zn²⁺, Cr³⁺, Al³⁺, Ag⁺, Mg²⁺, Pb²⁺, and Cu²⁺ were selected to investigate their selectivity and anti-interference ability.

Different concentrations of Hg²⁺ were added into the 1:1 fluorescent probe solution, and under a 365 nm UV lamp, the corresponding fluorescent color images were captured using a homemade smartphone-based colorimetric device. The RGB values of the generated fluorescence color information were obtained using the free Android application Color Grab [loomatix (2021) Color Gr, 2021].

After the fluorescent paper sensor was developed according to the literature (Xie et al., 2019), the adjusted ratio fluorescent probe solution was added to the hydrophilic area to make it appear uniform and stable on the paper-based sensor. Different concentrations of the mercury ion solution were added to the hydrophilic area of the paper sensor. The RGB values were captured and recognized by the smartphone on Color Grab. The quantitative relationship between the color intensity and concentration was established for quantitative detection.

Tap water and Yellow River water (Lanzhou, China) were used for the recovery experiments in this study. All Yellow River water and tap water samples were carried out on the same day of collection. All water samples were filtered twice with a 0.22 μm microporous membrane before used for experiments. Various concentrations of Hg²⁺ were added to the water samples, while the probe solution was used for detection. Three independent experiments were performed to determine.

In the current design strategy, the mixing ratio of NCDs and Au NCs was adjusted to realize the visual mercury ion detection. The reaction mechanism is shown in Scheme 1. The NCDs play a background reference role in the sensing system, while the Au NCs were quenched at the Hg²⁺ reaction site. The working principle presents a special high affinity between Hg²⁺ and Au⁺ interactions; hence, the red fluorescence of Au NCs is quenched by Hg²⁺ (Yan et al., 2016; Sahu et al., 2020), while the blue fluorescence of NCDs is stable against Hg²⁺, leading to distinct ratiometric fluorescence changes when exposed to Hg²⁺(Scheme 1).

For rapid mercury ion detection, the paper-based sensor was designed according to reference (Xie et al., 2019). The preparation process is as follows: by using a hand-held puncher, 6 mm holes were punched into the fluorescent acetate filter paper to form hydrophilic reaction zones. Subsequently, 6 mm diameter holes were also punched into the sub-light black film paper, discarding the self-adhesive black paper and pasting together the two sub-optical black film paper sheets to form the hydrophobic barrier area. Finally, the filter paper containing the 6 mm diameter holes was pasted onto two sub-optical black film paper sheets to construct a fluorescent paper-based sensor with good affinity and water drainage. The construction process is illustrated in Figure 1.

The properties of the NCDs and Au NCs were characterized using TEM, UV-vis absorption spectroscopy, FT-IR spectroscopy, X-ray diffraction (XRD), XPS, and fluorescence spectroscopy. The NCDs and Au NCs were approximately 3–5 nm in size (Supplementary Figure S1A,B), which proves that the synthesized nanomaterials have uniform particle size and dispersion distribution in aqueous solution. The characteristic peaks at wavenumbers 1171 cm⁻¹, 1310 cm⁻¹, and 1390 cm⁻¹ were assigned to the stretching vibrations of the C–O–C, C–N, and C–C bonds. The peak at 1650 cm⁻¹ was assigned to the stretching vibration of C=O, while 2950 cm⁻¹ and 3410 cm⁻¹ were assigned to the stretching vibrations of the C–H or N–H/O–H bonds (Supplementary Figure S1C). A broad peak at 2θ = 19.48° was observed owing to the disordered carbon atoms and graphite lattice spacing (Supplementary Figure S1D). In the excitation wavelength range of 290–400 nm, the fluorescence intensity of the NCDs (Supplementary Figure S2A) varied with different amplitudes and reached a maximum at 360 nm. Similarly, when the excitation wavelength was adjusted to 330–410 nm, the fluorescence intensity of the Au NCs showed the same trend and reached the maximum value of 350 nm (Supplementary Figure S2B). To balance the excitation wavelength between the NCDs and Au NCs, 400 nm was selected as the optimal excitation wavelength. This clearly shows that the NCDs contain C, H, N, and O elements. The appearance of peaks at 284.77, 399.87, and 531.29 eV are assigned to the characteristic peaks of C 1s, N 1s and O 1s, respectively, which correlate well with the above FT-IR results (Supplementary Figure S2C). The high-resolution C 1s spectrum consists of three peaks at 284.71, 286.03, and 287.62 eV, which belong to the C−C, C−O, and C=O/COOH bonds, respectively (Supplementary Figure S2D). The high-resolution spectrum of the carbon point N 1s can be divided into three peaks at 398.97, 399.87, and 401.46 eV, related to the C−N−C, N−H and N−C=C bonds (Supplementary Figure S2E). The high-resolution spectrum of O 1s contains two peaks at 531.28 and 533.01 eV (Supplementary Figure S2F), which are related to the C=O and C−OH/C−O−C bonds. XPS analysis showed that there were −OH, −COOH, and −NH₂ functional groups on the surface of the carbon point. As shown in the fluorescence spectra of the NCDs, the Au NCs and ratio fluorescent probe solutions are represented by letters a, b, and c, respectively (Supplementary Figure S3A). The ratio fluorescent probe solutions disperse well in water and exhibits dual-emission bands at 464 and 662 nm under excitation at 400 nm. Supplementary Figure S3B shows the UV−vis spectra. The black line represents the UV−vis absorption spectrum of NCDs. An acromion at 278 nm may be due to the n–π* excitation of the C=O group on the surface of the NCDs. The red line represents the UV−vis absorption spectrum of the Au NCs. No absorption was observed in the UV region. The blue line represents the UV−vis absorption spectrum of the ratio fluorescent probe solutions. The fluorescence intensity ratios (I₄₆₄/I₆₆₂) of the probe solutions remained steady over 1 h, proving their excellent photostability (Supplementary Figure S3C). Besides, the probe solutions were processed by centrifugation and found to have no significant increase or decrease in fluorescence intensity, and its fluorescence color was found to remain unchanged when photographed with a 365 nm UV lamp, indicating the good stability and usefulness of these ratiometric probe solutions.

pH is above 8 or below 6, the fluorescence intensity ratio of the probe solutions start to increase (Supplementary Figure S3E). The fluorescence intensity of the probe solutions tends to be stable in the pH range 7–8. Therefore, 7.4 was selected as the optimal pH. Kinetic experiments showed that the probe solutions can respond to Hg²⁺ in less than 5 min (Supplementary Figure S3F). To further illustrate the effect of these two fluorescent materials, a Hg²⁺ solution with a concentration ranging from 0.05 to 2.5 µM was added to a solution of NCDs and Au NCs The signal responses were then recorded. With an increase in Hg²⁺ concentration, NCDs did not response toward Hg²⁺, and its fluorescence intensity and color remained unchanged (Supplementary Figure S4A), while Au NCs continuously responded toward Hg²⁺, and its fluorescence intensity gradually decreased (Supplementary Figure S4B). With regards to the fluorescence color, the red became darker and difficult to distinguish with the naked eye. When the fluorescence intensity ratios of the probe solutions (NCDs: Au NCs) were set to 1:1, 1:2, and 1:3, respectively (Supplementary Figures S4C–E), the response toward Hg²⁺ was examined, and color changes were recorded.

A clear fluorescence color change from red to blue was observed (Supplementary Figure S4F) at a 1:1 ratio. Figure 2A illustrates the close relationship between the changes in fluorescence intensity and Hg²⁺ concentration. A good linear correlation (R² = 0.9935) was obtained from the standard curve for Hg²⁺ concentrations ranging from 0.05 to 2.5 µM (Figure 2B). The limit of detection (LOD) for Hg²⁺ was 2.7 nM based on the 3δ/slope. The linear range and LOD results were compared with those obtained using other methods (Table 1).

As the Hg²⁺ concentration increase from 0.05 to 2.5 µM, fluorescence images show a shift from red to blue under 365 nm UV light. Figure 3A shows a schematic of Hg²⁺ detection using a smartphone. Coupled with a 365 nm UV light, the RGB values of the solution were detected and recorded immediately, as shown in Figure 3B.

Based on the analysis of the response results, the ratio of the blue and red channels (B/R) were selected to calculate the signal response values. The linear relationship between B/R and the Hg²⁺ concentration (0.05–1.8 µM) was obtained as shown in Figure 3C, with a linear equation of B/R = 0.948C + 0.693. Therefore, it seems that there is a good linear relationship between the signal response values and the Hg²⁺ concentration. The LOD was calculated to be 25 nM based on the 3δ/slope.

To facilitate the visualization of Hg²⁺, the probe solution was repeatedly and uniformly added dropwise to the paper-based sensor. Under a 365 nm UV lamp, the paper-based sensor exhibited a highly uniform red luminescence. Figure 4A shows a schematic diagram of mercury ion detection by the paper-based sensor. The paper color appears as a gradient of red to pink violet and finally to dark blue with an increase in Hg²⁺ concentration from 0.05 to 2.5 µM (Figure 4B).

The RGB values of the test paper were identified using the application Color Grab. The results are shown in Figure 4B, and good linearity was observed through the fitted linear equation (B/R = 1.424C + 0.777, R² = 0.9885) when the Hg²⁺ concentration was in the range of 0.05–1.2 µM. The LOD was 32 nM based on the 3δ/slope (Figure 4C). As shown in Supplementary Figure S5C, the paper-based sensor was in the fluorescence color maintenance stage within 20 days at room temperature, and a distinct color variance still reemerged in the presence of 2.5 µM Hg²⁺.

In contrast, when different concentrations of mercury ion solutions were added to the paper-based sensor with the NCDs single probe solution, the blue fluorescence did not change under a 365 nm UV lamp (Supplementary Figure S5A), while the red fluorescence of Au NCs gradually weakened until it completely faded (Supplementary Figure S5B). It is difficult for the naked eye to separate color variations; however, the dual-emission ratio fluorescent paper sensor has a wide color range and is more suitable to observe for naked eye (Supplementary Figure S4F). These data suggest that the fluorescent paper chip sensor can simplify the detection process and achieve quantitative analysis of mercury ions using smartphones.

As shown in Figures 5A,C, the fluorescence intensity at 662 nm increased up to 80% when the Hg²⁺ concentration was at 2.5 µM. The value of I₄₆₄/I₆₆₂ did not change significantly with the addition of 25 μM of other metal cations (Na⁺, K⁺, Ca²⁺, Ni²⁺, Co²⁺, Cd²⁺, Ba²⁺, Fe³⁺, Zn²⁺, Cr³⁺, Al³⁺, Ag⁺, Mg²⁺, Pb²⁺, and Cu²⁺). Therefore, the fluorescent probe solution has a high selectivity toward Hg²⁺ because of the specific binding between the mercury ions and the functional groups on the Au NCs.

With the addition of 1 μM of the metal cations, the I₄₆₄/I₆₆₂ ratio increased by approximately 15% (Figure 5D). However, this phenomenon was significantly different after the addition of mercury ions. The corresponding fluorescence images (Figure 5B) were also photographed using a 365 nm UV lamp. The data suggests that the ratio fluorescent probe solution exhibits excellent selectivity and anti-interference capability for Hg²⁺. Cu²⁺ ion shows slight suppression effect on BSA-AuNCs fluorescence mainly results from its low concentration in the selectivity and anti-interference experiment. The pre-separation method should be adopted if the matrixes contain large amount of Cu²⁺ ion higher than 10 times.

The ratio fluorescence probe solution was applied to determine the Hg²⁺ in tap water (from our laboratory) and Yellow River water samples (from An Ning, Lanzhou.). All the water samples were treated with ordinary qualitative filter paper and 0.22 μm Supor filters. Hg²⁺ at concentrations of 0.5, 1.2, 1.6, and 2.0 μM were spiked to test the rate of recovery. The results are shown in Table 2. The recovery rates obtained ranged from 90.5 to 107.6%. The results indicate that the designed probe solution has practical applicability for the detection of Hg²⁺ in water samples.