Colorimetric and photothermal dual-mode lateral flow immunoassay based on Au-Fe₃O₄ multifunctional nanoparticles for detection of Salmonella typhimurium

Abstract

Au-Fe₃O₄ multifunctional nanoparticles (NPs) were synthesized and integrated with lateral flow immunoassay (LFIA) for dual-mode detection of Salmonella typhimurium. The Au-Fe₃O₄ NPs not only combined excellent local surface plasmon resonance characteristics and superparamagnetic properties, but also exhibited good photothermal effect. In the detection, antibody-conjugated Au-Fe₃O₄ NPs first captured S. typhimurium from complex matrix, which was then loaded on the LFIA strip and trapped by the T-line. By observing the color bands with the naked eyes, qualitative detection was performed free of instrument. By measuring the photothermal signal, quantification was achieved with a portable infrared thermal camera. The introduction of magnetic separation achieved the enrichment and purification of target bacteria, thus enhancing the detection sensitivity and reducing interference. This dual-mode LFIA achieved a visual detection limit of 5 × 10⁵ CFU/mL and a photothermal detection limit of 5 × 10⁴ CFU/mL. Compared with traditional Au-based LFIA, this dual-mode LFIA increased the detection sensitivity by 2 orders of magnitude and could be directly applied to unprocessed milk sample. Besides, this dual-mode LFIA showed good reproducibility and specificity. The intra-assay and inter-assay variation coefficients were 3.0% and 7.9%, and with this dual-mode LFIA, other bacteria hardly produced distinguishable signals. Thus, the Au-Fe₃O₄ NPs-based LFIA has potential to increase the efficiency of pandemic prevention and control.

Graphical Abstract

Au-Fe₃O₄ nanoparticle proved to be a promising alternative reporter for LFIA, achieving multifunctions: target purification, target enrichment, visual qualitation, and instrumental quantification, which improved the limitations of traditional LFIA.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00604-023-05645-x.

Introduction

Food safety problem continues to present a major threat to human health and social security. According to the World Health Organization, foodborne diseases cause ca. 600 million to fall ill, 420,000 to die, and hundreds of billions of US$ lost years, which severely threaten human health and impede socioeconomic development [1]. To effectively reduce the above losses, early and reliable detection plays a key role [2, 3]. Salmonella, as one of the most common foodborne pathogens, affects millions of people annually, sometimes with severe and fatal outcomes, and thus has always been an important monitoring target for food safety [1, 4, 5]. Currently, the main methods used in practice are culture-based assays, polymerase chain reaction (PCR) analyses, and enzyme-linked immunosorbent assay (ELISA). Culture-based assays are the gold standard method owing to their high accuracy but limited by time-consuming and labor-intensive manipulation. PCR has high sensitivity but requires sophisticated instruments and skilled technical staffs, restricting its applications. ELISA simplifies the detection procedure to a certain degree, yet it still suffers from the limitations like relatively low sensitivity and time consumption [6–9]. Thus, these methods cannot be applied to on-site detection or large-scale screening, which seriously restricts the efficiency of pandemic prevention and control.

Lateral flow immunoassay (LFIA) has shown great promise in on-site instant diagnosis, due to its enormous advantages including high speed, low cost, portability, convenience, free of instrument or skilled staff [10–12]. The most successful application of LFIA is the pregnancy test strips, which change hospital-dependent detection to rapid home self-testing, thereby highly improving diagnosis efficiency and reducing costs [13, 14]. However, applying traditional LFIA to bacteria detection faces three main bottleneck problems. One is the low sensitivity which may result in missed detection. The second is that it is a yes-or-no diagnosis which cannot provide accurate quantitative information for guiding therapy. The third is that it cannot be directly applied to complex matrix which must be diluted before flowing on the strips, and this seriously limits its practical application due to the fact that many pathogenic bacteria usually exist in complex biological samples at very low concentration [10, 15–17]. The key reason that induces the above drawbacks is that traditional LFIA mainly uses gold nanoparticles (NPs) as reporters.

In recent decades, researchers have been trying to develop various nanomaterials as new reporters to replace gold NPs, such as dyed beads, fluorescent NPs, surface-enhanced Raman scattering-active nanomaterials, and carbon nanomaterials, which have increased the sensitivity and realized quantitative detection [18–21]. Nevertheless, with these reporters, dilution is still necessary for complex samples, which may lower the sensitivity in practical detection. Magnetic NPs may be promising alternative reporters to solve this problem. Magnetic NPs can effectively capture, separate, and concentrate target bacteria from a large volume of sample into a small volume of buffer under an external magnetic field, thereby realizing purification and enrichment [22–25]. Some works have confirmed the superiority of magnetic reporters in the LFIA. For example, Hwang et al. coated Au NPs on Fe₃O₄ nanoclusters, with which sensitive detection of Salmonella in milk was achieved [26]. As other examples, Hu et al. and Wang et al., respectively, prepared fluorescent-magnetic nanocomposites by repeated assembling methods and developed multimodal LFIAs for Salmonella and S. pneumoniae detection in milk, serum, or whole blood samples [27, 28]. The above magnetic-colorimetric reporters-based LFIA exhibited obvious advantages over conventional LFIA, yet the preparation of the reporters generally involved more than two kinds of nanoparticles and their assembly, which were complicated and time/labor-consuming.

In our previous work, we proposed a universal seed-mediated growth method to directly synthesize hydrophilic coinage metal/iron oxide (M-Fe₃O₄) heterodimer NPs, such as Ag-Fe₃O₄, Pd-Fe₃O₄, and AgPt-Fe₃O₄ [29–32]. These M-Fe₃O₄ NPs not only combined multiple functions, but also showed improved bio-modifiability and water compatibility with good tunability. Herein, based on our previous work, we directly synthesized Au-Fe₃O₄ heterodimer NPs, avoiding the assembly of two kinds of NPs, which was much more convenient and time-saving. The Au-Fe₃O₄ NPs integrated the local surface plasmon resonance (LSPR) characteristics of Au and the superparamagnetic properties of Fe₃O₄, which were used as new reporters in LFIA for Salmonella typhimurium detection. Au part provided favorable colorimetric signal due to its excellent LSPR characteristics. Fe₃O₄ part enabled magnetic separation and enrichment of the targets, improving the detection sensitivity and applicability in complex matrix. Besides, the Au-Fe₃O₄ NPs exhibited good photothermal effect due to the combination of Au and Fe₃O₄ [33–35]. By measuring the photothermal signal with a portable infrared thermal camera, accurate quantification of S. typhimurium was achieved. Thus, as new reporters, Au-Fe₃O₄ possessed multifunctions: target purification, target enrichment, and dual-mode detection (visual qualitation and instrumental quantification), which improved the limitations of traditional LFIA and showed great application potential in practice.

Experimental section

Reagents and instruments

Tetramethylammonium hydroxide (TMAH), Tween 20, polyvinylpyrrolidone (PVP, MW = 10,000), trisodium citrate (TSC), oleic acid, chloroauric acid (HAuCl₄), 1-octadecene, and iron pentacarbonyl were purchased from Macklin Biochemical Co., Ltd. HS-PEG-COOH (MW 2000) was bought from Shanghai ToYong Bio-Tech. Co., Ltd. Mouse anti-S. typhimurium monoclonal antibody was obtained from Abcam. N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), morpholino ethanesulfonic acid (MES), goat anti-mouse antibody, N-hydroxysuccinimide (NHS), acridine orange hemi (AO), and bovine serum albumin (BSA) were bought from Sigma-Aldrich. Heat-inactive Escherichia coli, Listeria monocytogenes, S. typhimurium, and Staphylococcus aureus were supplied by Army Medical University. Sample pad, conjugate pad, absorbent pad, nitrocellulose (NC) membrane, and PVC substrate were purchased from Joey-biotech Co., Ltd. Milk was bought from Nei Monggol Yili Industrial Group Co., Ltd. Ultrapure water was produced by a Millipore Milli-Q system.

Transmission electron microscopy (TEM) images were acquired by a microscope (JEOL, JEM 1400). High-resolution (HR) TEM and energy-dispersive X-ray (EDX) elemental mapping were obtained using a microscope (FEI, Tecnai F30). Absorption spectra were recorded with a spectrophotometer (Shimadzu, UV-2450). The test strips were sprayed by HGS510 (AUTOKUN) and divided by HGS210 induction cutting machine (AUTOKUN). Hydrodynamic diameters and zeta potentials were measured by a Zetasizer Nano ZS instrument (Malvern). Photothermal signals were collected by an infrared thermal camera (FOTRIC 226 s) mounted with an 808 laser (FU808AD2000-F34). Magnetic separation was carried out with a magnet (Invitrogen, 12320D). The photographs were taken using iPhone11 (Apple).

Construction of anti-S. typhimurium antibody-modified Au-Fe₃O₄ NPs

Au-Fe₃O₄ heterodimer NPs were first prepared by seed-mediated approach referring to our previous work with slight modification [29]. Au³⁺ was reduced by Fe₃O₄, producing a gold seed on each Fe₃O₄ NP, which further grew by repeatedly adding TSC and Au precursor. The detail procedure was presented in the Supporting Information (S.1). Then, the biomodification of Au-Fe₃O₄ NPs was performed as follows.

Ten milliliters of Au-Fe₃O₄ NPs was added with 0.01 g of HS-PEG-COOH and shook for 1 h at 37 °C. Then, the above Au-Fe₃O₄ NPs were washed 3 times with 100 mM MES buffer (pH = 5.0) containing 0.05% Tween 20 and dispersed in 10 mL of MES buffer containing 2.5 mM NHS and 5 mM EDC. After 0.5 h reaction, the Au-Fe₃O₄ NPs were washed with PBS and dispersed in pH 7.2 PBS (0.01 M, 10 mL) to react with anti-S. typhimurium antibody (10 µg) for about 4 h. Finally, the obtained antibody-functionalized Au-Fe₃O₄ (Au-Fe₃O₄-Ab) NPs were washed 3 times by PBS, which were then blocked by 2% BSA, dispersed in 1 mL of PBS, and stored at 4 ℃. The complete preparation and modification process were illustrated in Scheme 1A.

Scheme 1.

A Schematic diagram for the preparation and modification of Au-Fe₃O₄ NPs. B Schematic diagram for the fabrication of LFIA strip

Fabrication of LFIA strips

As illustrated in Scheme 1B, the test strip was constituted with sample pad, conjugate pad, NC membrane, and absorbent pad pasted onto the PVC plate. The overlap widths were about 2 mm between NC membrane and conjugate pad, 5 mm between NC membrane and absorbent pad, and 5 mm between conjugate pad and sample pad. Mouse anti-S. typhimurium antibody (1.0 mg/mL) was sprayed at the T-line at a rate of 1.0 µL/cm. Goat anti-mouse antibody (1.0 mg/mL) was sprayed at the C-line at a rate of 1.0 µL/cm. Then, the strips were dried for 2 h at 37 °C, which were then cut into 3 mm width and kept in sealed plastic bag.

Detection of S. typhimurium with the dual-mode LFIA

In a typical test, 40 µL of Au-Fe₃O₄-Ab was added to 600 µL of test samples and incubated at 37 °C with gentle shaking. Then, the mixture was separated by a magnet and resuspended with a solution containing 70 µL of pH 7.2 PBS, 10 µL of Au-Fe₃O₄-Ab, and 20 µL of running solution (2.0% BSA and 0.25% Tween 20 in pH 7.2 PBS). The above solution was transferred to a microtiter plate and the sample pad of the LFIA strip was immersed in it. All the liquid was absorbed and migrated along the strip. After 15 min, by observing T-line color with the naked eyes, qualitative detection was achieved. Meanwhile, the photothermal signal of the T-line was measured for quantitative detection, which was defined as follows.

$$\mathrm{\Delta }T={\mathrm{\Delta }T}_1-{\mathrm{\Delta }T}_0$$

where ΔT₁ was the temperature variation of the tested strips before and after irradiation and ΔT₀ was the temperature variation of the unused strips.

Detection of S. typhimurium in simulated complex samples

The whole milk was first tested by PCR methods to certify that they were not contaminated by S. typhimurium as our previous work [36]. Then, the milk was mixed with S. typhimurium suspension in 0.1 M pH 7.2 PBS (9:1 by volume) to simulate complex samples. For detection, Au-Fe₃O₄-Ab was added to incubate with the simulated samples. After the reaction, separation, and immunochromatography with the dual-mode strip as described in the previous section, naked-eye observing and photothermal signal measuring were performed. Blank milk samples were treated the same except no bacterium was added. Meanwhile, PBS samples with the same bacteria concentrations were detected for control.

Results and discussion

Principle of the dual-mode LFIA for detecting S. typhimurium

As illustrated by Fig. 1A, Au-Fe₃O₄-Ab was first added to the samples to capture and separate S. typhimurium. This step achieved enrichment and purification of target bacteria, which would greatly facilitate enhancing detection signal and reducing interference. Then, the obtained Au-Fe₃O₄-Ab-S. typhimurium complexes were loaded on the LFIA strip, which were captured by the anti-S. typhimurium antibody on the T-line, inducing a colored band. Free Au-Fe₃O₄-Ab migrated further, which were captured by the anti-mouse antibody on the C-line to produce a quality control band. For negative samples without S. typhimurium, Au-Fe₃O₄-Ab would flow past the T-line without interaction and only react with the C-line. Thus, as illustrated in Fig. 1B, positive samples produced both colored C-line and T-line, and negative samples only produced colored C-line. While for the ones whose C-line did not show color, indicating the antibody had lost bioactivity, the results were invalid. By observing the color bands with the naked eyes, qualitative detection was performed free of instrument. Meanwhile, by measuring the photothermal signal of the T-line with a portable infrared thermal camera, quantification was achieved. Figure 1B also shows the typical positive and negative results from visual and thermal camera observation. It could be seen that the positive one showed two clear colored bands, and its T-line produced an obvious high temperature zone. While for the negative one, only the C-line produced color, and the temperature of its T-line hardly rose.

Fig. 1.

A Schematic diagram for S. typhimurium detection with the dual-mode LFIA. B Interpretation of the detection results and the typical positive and negative results from visual and infrared thermal camera observation

Characterization of Au-Fe₃O₄ and Au-Fe₃O₄-Ab NPs

As shown in Fig. 2A-B, Au-Fe₃O₄ NPs possessed good dispersion with an average size of 35.6 ± 2.3 nm (200 particles were counted for calculation) and mainly exhibited one-to-one dumbbell structure. From the EDX elemental mapping of one single NP (Fig. 2C–E), it could be seen that Fe element and Au element were respectively distributed in the two sections of the heterodimer, which further confirmed its structure. Besides, the Au-Fe₃O₄ solution showed a wine-red color with a maximum absorbance peak at 522 nm (Fig. 2F).

Fig. 2.

Characterization of Au-Fe₃O₄ and Au-Fe₃O₄-Ab NPs. A TEM image and size distribution of Au-Fe₃O₄ NPs. B HR-TEM image of a single Au-Fe₃O₄ NP. C–E EDX element mappings of Fe, Au, and the overlay. F UV–Vis spectra of Au-Fe₃O₄ and Au-Fe₃O₄-Ab solution and the corresponding photos (inset). G TEM image and size distribution of Au-Fe₃O₄-Ab NPs. H Hydrodynamic diameters and zeta potentials of Au-Fe₃O₄ and Au-Fe₃O₄-Ab. I Photographs of the test strips loaded with Au-Fe₃O₄ and Au-Fe₃O₄-Ab NPs

For construction of Au-Fe₃O₄-Ab, carboxyl groups were first introduced on Au-Fe₃O₄ NP surface with HS-PEG-COOH based on coordination interaction and then conjugated with the amino groups of anti-S. typhimurium antibody through carbodiimide chemistry. The Au-Fe₃O₄-Ab remained the optical and morphology properties of Au-Fe₃O₄ NPs. As shown in Fig. 2F–G, Au-Fe₃O₄-Ab NPs had good dispersibility with an average size of 37.0 ± 2.3 nm (200 particles were counted for calculation) and exhibited a wine-red color with a maximum absorbance peak at 528 nm, which only slightly changed compared with Au-Fe₃O₄ NPs. To confirm the conjugation, the hydrodynamic diameter and zeta potential of the NPs before and after functionalization were measured, and the results are shown in Fig. 2H. It could be seen that the hydrodynamic diameter produced an obvious increase from 76 to 150 nm, and the zeta potential changed from − 66 to − 16 mV, which demonstrated the antibody was successfully coupled with the NPs [37–39]. Furthermore, Au-Fe₃O₄ and Au-Fe₃O₄-Ab were, respectively, loaded on the LFIA strips, and as shown in Fig. 2I, the strip with Au-Fe₃O₄-Ab produced a colored C-line, while the strip with Au-Fe₃O₄ produced no colored line. This demonstrated that the anti-S. typhimurium antibody on the NPs maintained good bioactivity, which could react with the goat anti-mouse antibody on the C-line.

Reliability of the dual-mode LFIA

Fluorescence microscopy and TEM analyses were performed to investigate the capture of S. typhimurium with Au-Fe₃O₄-Ab. From Fig. 3A, it can be seen that many green fluorescence dots, indicating AO-stained bacteria, could be found after the capture of S. typhimurium by Au-Fe₃O₄-Ab. While using Au-Fe₃O₄ to treat S. typhimurium, no bacterium was observed (Fig. S1A). This demonstrated that Au-Fe₃O₄-Ab could efficiently capture S. typhimurium based on antibody-antigen interaction. TEM image more intuitively showed the combination between S. typhimurium and Au-Fe₃O₄-Ab. From Fig. 3B, it can be seen that many Au-Fe₃O₄-Ab NPs bound with the bacterium surfaces. While using Au-Fe₃O₄ to treat S. typhimurium, few NPs combined with S. typhimurium, and after separation, only Au-Fe₃O₄ NPs were observed rather than bacteria (Fig. S1B-C). This matched well with the microscopy analyses.

Fig. 3.

A–B Fluorescence microscopic image and TEM image of S. typhimurium captured by Au-Fe₃O₄-Ab. C Photothermal signals obtained from positive samples (1 × 10⁷ CFU/mL S. typhimurium) and blank samples enriched by different magnetic attraction times. The lower is the histogram of photothermal signals, the upper are the corresponding photothermal images. The numbers above the columns are the signal-to-noise ratios. D Results from detection of different bacteria (5 × 10⁷ CFU/mL). The lower is the photothermal signal histogram and images of the test strips, the upper are the corresponding photographs by phones. Error bars are calculated from three experiments

Then, we explored the effect of magnetic enrichment on detection signal. As shown in Fig. 3C, compared with the samples without enrichment, the enriched ones achieved higher photothermal signals, facilitating enhanced detection sensitivity. Besides, the magnetic attraction time in separation also influenced the detection signal, due to the fact that the attraction time affected the capture efficiency. With the increase of attraction time, the signals from positive samples and negative ones both increased, and 15 min induced the highest signal-to-noise ratio (the ratio of the positive sample signal to the negative sample signal), which was chosen as the optimal attraction time. Other conditions, including irradiation power density, irradiation time, amount of Au-Fe₃O₄-Ab NPs, and incubation time, were also optimized. Finally, 4 W/cm², 100 s irradiation, 40 µL of Au-Fe₃O₄-Ab NPs, and 20-min incubation were selected (Fig. S2-5).

Under optimal conditions, the specificity and reproducibility of the dual-mode LFIA were evaluated. S. typhimurium, L. monocytogenes, S. aureus, and E. coli at the same concentration (5 × 10⁷ CFU/mL) and the blank samples without bacteria were detected by the strips. As shown in Fig. 3D, S. typhimurium samples produced obvious colored T-line and strong photothermal signal, while other bacteria showed similar phenomenon as that of the blank, in which no colored T-line or obvious photothermal signal was found. Meanwhile, we calculated the coefficient of variations of the intra-assay and inter-assay as 3.0% and 7.9%, confirming the good reproducibility.

All the above results demonstrated that this Au-Fe₃O₄ multifunctional NPs-based LFIA enabled efficient capture of S. typhimurium, and the pre-enrichment enhanced the detection signal, definitely improving the sensitivity. Besides, this LFIA had good specificity and reproducibility, assuring the detection reliability.

Analytical performance of the LFIA

To evaluate the sensitivity and dynamic range, the dual-mode LFIA was used to detect different concentrations of S. typhimurium. As shown in Fig. 4A, for the negative samples, only the C-line generated a colored band, and when the S. typhimurium concentration reached 5 × 10⁵ CFU/mL, the T-line began to display a visible color. As the bacteria concentration increased further, the T-line color became darker, due to the fact that more Au-Fe₃O₄-Ab-S. typhimurium complexes were accumulated on the test zone. Therefore, by observing the color change of the T-line with the naked eyes, qualitative detection of S. typhimurium was achieved with a limit of 5 × 10⁵ CFU/mL. Meanwhile, the photothermal signals of the strips were measured, which are shown in Fig. 4B. Similar with the visual detection, the photothermal signal increased with the bacteria concentration. According to the lowest distinguishable signal (S_blank + 3SD, where S_blank is the average signal of blank groups, and SD is their standard deviation), the limit of the photothermal detection could achieve 5 × 10⁴ CFU/mL, which was one order of magnitude lower than that of the visual observing. Moreover, a good linear relationship was exhibited at the concentration from 5 × 10⁶ to 1 × 10⁸ CFU/mL with R² = 0.9989 (Fig. 4C), facilitating accurate quantification of S. typhimurium. Compared with the commercial LFIA [27], this dual-mode LFIA increased the sensitivity by 1–2 orders of magnitude, which might be attributed to the pre-enrichment of the target bacteria.

Fig. 4.

A–B Photographs of the test strips and the corresponding photothermal signals obtained from detection of different concentrations of S. typhimurium. The asterisk represents the detection limit. The red line represents the lowest distinguishable signal by photothermal measuring. C Calibration plot of photothermal signal versus S. typhimurium concentration. The insets are the corresponding photothermal images. Error bars are calculated from three experiments

Application to simulated complex samples

To test the feasibility of the dual-mode LFIA in complex biological matrixes, unprocessed milk spiked with S. typhimurium was used to mimic real samples for detection. As shown in Fig. 5, all the positive milk samples produced two obvious colored bands, and the negative one only produced a colored C-line, which were almost the same with those of the buffer samples. By measuring the photothermal signal, the milk samples induced slightly lower signal than the buffer samples, and the recovery was 72.70 ± 8.28% (Table S1), which was tolerable, due to the complexity of the unprocessed milk, while the commercial LFIA can be only applied to diluted milk, because the unprocessed milk cannot migrate on strips. This dual-mode LFIA’s better analytical performance in complex matrixes was mainly attributed to the pre-enrichment and pre-purification by magnetic separation, which made it possess great potential application in practice.

Fig. 5.

Photothermal signal histograms of the test strips and their corresponding photographs obtained from detecting different concentrations of S. typhimurium in buffer and milk samples. Error bars are calculated from three experiments

Conclusions

In this work, we directly synthesized Au-Fe₃O₄ multifunctional NPs and integrated them with LFIA as promising alternative reporters for dual-mode detection of S. typhimurium. Table 1 shows the analytical performance comparison between this dual-mode LFIA with other recently reported LFIA. It can be seen that this dual-mode LFIA indeed improved the limitations of traditional LFIA (Au-based LFIA) [40], and it increased the detection sensitivity by 2 orders of magnitude. Moreover, compared with carbon nanoparticles, quantum dots, up-converting materials, etc. [39, 41–44], Au-Fe₃O₄ NPs enabled enrichment and purification of target bacteria from complex matrix, thus achieving direct application to unprocessed whole milk. Besides, Au-Fe₃O₄ NPs could provide photothermal signal, facilitating quantitative detection with a portable infrared thermal camera, which was more sensitive than color observing and more cost-effective than fluorescence or magnetic signal determination [27, 45]. However, there is still room for improvement in this method. In the future, we will try to optimize the Fe₃O₄ size and the Au absorption to obtain faster enrichment and better photothermal signal and then make the manipulation more convenient and achieve more sensitive detection.

Table 1.

An overview on recently reported LFIA for detection of Salmonella

Materials used	Signal	Detection limit (CFU/mL)	Assay time	Linearity range (CFU/mL)	Complex samples	Ref
Gold nanoparticles	Color	107	30 min	Qualitative		[40]
Gold-coated gold nanoparticles	Color	104	20 min	Qualitative	Diluted fat-reduced milk	[41]
Nitrogen-rich carbon nanoparticles	Color	102	20 min	102–108	Diluted juice and vegetable salad	[42]
Au@Ru nanocomposites	Color	9.8 × 104	10 min	2.9 × 106–2.9 × 1011	Diluted drinking water, juice, and milk	[39]
Quantum dot nanobeads	Fluorescence	5 × 103	12 min	5 × 104–107	Broth processed by centrifugation	[43]
Up-converting phosphor	Fluorescence	104	20 min	104–107	Food samples cultured for about 14 h	[44]
Magnetic nanoparticles	Color	1.95 × 105	30 min	Qualitative	Low fat milk	[45]
Fluorescent-magnetic nanospheres	Fluorescence and magnetic signal	3.5 × 103	35 min	1.88 × 104–1.88 × 107	Milk, serum, and whole blood	[27]
Au-Fe3O4 nanoparticles	Color and photothermal signal	5 × 104	50 min	5 × 106–1 × 108	Whole milk	This work

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

This work was financially supported by the Natural Scientific Foundation of Shandong (ZR2020MB064, ZR2020ZD13, ZR2022JQ07), the Science and Technology Projects of Qingdao (21–1-4-sf-7-nsh), the National Natural Science Foundation of China (No. 21876206), and the Fundamental Research Funds for the Central Universities (No. 22CX03033A).

Data availability

Date will be made available on request.

Declarations

Conflict of interest

The authors declare no competing interests.