A rapid detection method for food safety

Abstract

This invention discloses a rapid detection method for food safety, belonging to the field of food testing technology. It includes the following steps: sample pretreatment, capture of target analytes using functionalized magnetic nanoparticles, enrichment and reaction using a microfluidic chip, signal amplification and conversion, and data processing assisted by a deep learning model. A pre-trained deep learning model is used to adaptively filter noise and correct baseline drift on the acquired signal curve, and the concentration of the target analyte is calculated based on the corrected signal peak intensity. This invention employs Fe3O4@SiO2 core-shell structured magnetic nanoparticles with surface-modified specific recognition molecules, combined with a microfluidic chip platform, to achieve efficient capture and enrichment of target analytes. By immobilizing three-dimensional porous Prussian blue analog nanozymes within the chip channels, dual amplification of electrochemical or colorimetric signals is achieved. The acquired signal curve is input into a pre-trained one-dimensional convolutional neural network model for adaptive noise reduction and baseline correction, and quantitative analysis of the target analyte concentration is performed.

Description

Technical Field

[0001] This invention relates to the field of food safety testing technology, specifically to a rapid food safety testing method. Background Technology

[0002] Food safety is a major issue affecting people's livelihoods. Rapid screening for harmful contaminants such as pesticide residues, veterinary drug residues, heavy metals, and illegal additives in food is a crucial step in ensuring food safety. Currently, traditional food safety testing methods mainly include high-performance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS). Although these methods offer high accuracy, they suffer from drawbacks such as complex sample pretreatment, long testing cycles, expensive instruments, and the need for professional personnel to operate them, making them unsuitable for rapid on-site testing.

[0003] Existing rapid detection technologies, such as colloidal gold immunochromatography and enzyme inhibition methods, suffer from problems such as insufficient sensitivity, high false positive rates, and limited detectable indicators. Surface-enhanced Raman scattering (SERS) technology has advantages such as fast detection speed, strong fingerprint recognition specificity, and high sensitivity. However, existing SERS detection methods have technical bottlenecks such as poor substrate preparation repeatability, uneven mixing of samples and substrates, and low signal acquisition consistency, resulting in insufficient stability of detection results and difficulty in achieving standardized applications. At the same time, it is difficult to achieve both detection sensitivity and specificity, especially for the detection of trace (pM to fM level) hazards, which often leads to false positives or false negatives due to interference from complex food matrices. The signal amplification methods are also limited, mostly relying on chemical amplification, lacking effective physical or computational assistance, resulting in an unsatisfactory detection limit. The operation steps are still relatively cumbersome, requiring multiple steps of cleaning, centrifugation, and incubation, making it difficult to achieve true "sample in, result out" results. Summary of the Invention

[0004] The purpose of this invention is to provide a rapid detection method for food safety, addressing the problems of insufficient sensitivity, high false positive rate, and limited detectable indicators in existing rapid detection techniques such as colloidal gold immunochromatography and enzyme inhibition methods, as mentioned in the background. Surface-enhanced Raman scattering (SERS) technology has advantages such as fast detection speed, strong fingerprint recognition specificity, and high sensitivity. However, existing SERS detection methods suffer from technical bottlenecks such as poor substrate preparation repeatability, uneven mixing of sample and substrate, and low signal acquisition consistency, resulting in insufficient stability of detection results and difficulty in standardized application. Furthermore, it is difficult to balance detection sensitivity and specificity, especially for the detection of trace (pM to fM level) hazards, where interference from complex food matrices often leads to false positives or false negatives. Signal amplification methods are often limited, relying heavily on chemical amplification and lacking effective physical or computational assistance, resulting in an unsatisfactory detection limit. The operation steps are still relatively cumbersome, requiring multiple steps of cleaning, centrifugation, and incubation, making it difficult to achieve true "sample in, result out" results.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a rapid detection method for food safety, the detection method comprising the following steps: S1. First, sample pretreatment is required. Take the food sample to be tested, add 0.1~0.5mol / L hydrochloric acid solution or phosphate buffer, homogenize for 1~3 minutes, filter through a 0.22μm organic filter membrane to obtain the sample solution to be tested, and use functionalized magnetic nanoparticles with first recognition molecules on their surface to specifically capture the target analyte in the food sample solution, and separate and enrich it under an external magnetic field to obtain the magnetic particle-target analyte complex.

[0006] S2. Next, the SERS substrate with magnetic nanoparticles modified with specific recognition molecules needs to be fixed in the detection channel of the microfluidic chip. The detection channel is rinsed with ethanol and deionized water 3-5 times in sequence, and then dried with nitrogen gas. The resulting sample solution is injected into the detection channel of the microfluidic chip at a flow rate of 5-20 μL / min. The reaction temperature is controlled at 25-37℃ and incubated for 3-5 minutes to allow the analyte in the sample to fully bind with the specific recognition molecules on the SERS substrate.

[0007] S3. Finally, signal conversion and amplification processing are required. The sample solution is introduced into the detection area of ​​the microfluidic chip. A signal amplification catalyst is fixed on the inner wall of the channel of the microfluidic chip. The catalyst enhances the original signal of the metal nanoclusters to generate a detectable physical signal. Data acquisition and processing are required. The physical signal is acquired, and a pre-trained deep learning model is used to adaptively filter noise and correct baseline drift on the acquired signal curve. The concentration of the target analyte is calculated based on the corrected signal peak intensity.

[0008] Preferably, the functionalized magnetic nanoparticles are Fe3O4@SiO2 core-shell structured nanoparticles, wherein the Fe3O4 core (diameter 15~30 nm) provides high saturation magnetization (≥60 emu / g) to ensure rapid magnetic response; the middle layer is a silica (SiO2) shell (thickness 5~15 nm), which protects the Fe3O4 core from oxidation and provides easily modified silanol groups (-Si-OH), which can be covalently linked to biorecognition molecules via silane coupling agents (such as APTES, MPS). The density of the first recognition molecule modified on the surface is controlled at 10^2~10^4 molecules / particle, which can be adjusted by changing the concentration of the antibody / aptamer and the reaction time. The first recognition molecule modified on the surface is a specific antibody or nucleic acid aptamer against the target analyte.

[0009] Preferably, the signal amplification catalyst is a three-dimensional porous Prussian blue analog nanozyme fixed on the inner wall of the microfluidic chip channel. Its preparation method is as follows: using transition metal ions such as Co, Fe, Ni, and Cu, and ferricyanide as precursors, PBA with nanocube, nanocubic octahedron, or nanoflower morphology is synthesized via a self-templating method or hydrothermal method (120~180℃, 6~24 hours) in the presence of surfactants (such as PVP or CTAB). The specific surface area can reach 200~500 m². 2 / g, with an average pore size of 5~20 nm, the detectable physical signal is a current signal or a visible light absorption signal. In the detection region, the signal amplification catalyst catalyzes the substrate solution to generate an electrochemical or colorimetric signal, thereby achieving dual signal amplification.

[0010] Preferably, the deep learning model is a one-dimensional convolutional neural network model. The training steps of the model include: acquiring multiple original signal samples containing noise and baseline drift, as well as the corresponding standard ideal signal; using the original signal samples as input and the standard ideal signal as output, training the one-dimensional convolutional neural network by minimizing the mean squared error loss function until convergence.

[0011] Preferably, the data acquisition and processing includes comparing the calculated concentration with a preset safety threshold. If the concentration exceeds the safety threshold, a positive alarm signal is output; otherwise, a negative result is output, and the detection data is uploaded to a cloud server for tracing.

[0012] Preferably, the homogenization process takes 1 to 3 minutes, the filtration uses a 0.22 μm organic filter membrane, the rinsing process involves rinsing the detection channel with ethanol and deionized water 3 to 5 times in sequence, followed by blowing with nitrogen gas to dry it, and the analytes include at least one of pesticide residues, veterinary drug residues, heavy metals, and illegal food additives.

[0013] Preferably, the microfluidic chip is made of polydimethylsiloxane (PDMS) or polymethyl methacrylate (PMMA) and is fabricated using soft photolithography or micro-injection molding. The chip includes the following functional areas: a sample inlet, a fluid delivery microchannel (width 100–300 μm, depth 50–100 μm, length 2–5 cm), a reaction / capture zone, a signal detection zone, and a waste outlet. The inner wall of the channel is treated with oxygen plasma to change it from hydrophobic to hydrophilic, facilitating subsequent chemical modification.

[0014] Preferably, the method for immobilizing functionalized magnetic nanoparticles with specific recognition molecules on their surface within a microfluidic detection channel is as follows: A suspension containing magnetic nanoparticles (0.5~2 mg / mL) is pumped into the channel at a flow rate of 5~20 μL / min. Simultaneously, a programmable neodymium magnet array is placed below the chip to generate a local high-gradient magnetic field (0.3~0.8 T), precisely adsorbing the magnetic nanoparticles onto a specific area (detection window) on the inner wall of the channel, forming a stable and uniform single-layer / multilayer magnetic particle film. After immobilization, the detection channel is rinsed 3~5 times sequentially with ethanol (70%~100%) and deionized water at a flow rate of 10~30 μL / min, for 2 minutes each time, to remove unimmobilized particles and residual reagents. Finally, high-purity nitrogen gas (flow rate 0.5~1 L / min) is introduced to dry the channel.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention combines the self-signal of the functionalized magnetic nanoparticles of the first recognition molecule with the catalytic signal of the three-dimensional porous Prussian blue analog nanozyme, achieving a detection limit at the picomolar level. This improves sensitivity by 2-3 orders of magnitude compared to traditional colorimetric / electrochemical methods. It also enables intelligent data processing with high accuracy. Furthermore, it introduces a one-dimensional convolutional neural network deep learning model for adaptive denoising and baseline correction of the original signal, overcoming the problems of strong subjectivity and poor adaptability in traditional methods that involve manually setting parameters. This significantly improves the accuracy, reproducibility, and anti-interference ability of quantification.

[0016] This invention is simple, fast, and integrated: it integrates capture, labeling, reaction, detection, and data processing into a microfluidic chip platform, simplifying the steps to "sample injection-reaction-result output," with a total time of less than 40 minutes. It is ideal for rapid on-site screening and emergency monitoring. For the first time, it combines the precise fluid control technology of microfluidic chips with a functionalized SERS substrate, achieving controllable and uniform bonding between the sample and the substrate. This solves the technical problem of poor repeatability of traditional SERS detection signals and has outstanding substantive features compared with existing publicly available detection methods. This method does not require large instruments and can be applied to the rapid on-site detection of pesticide residues, veterinary drug residues, illegal additives, and other harmful substances in various food matrices such as vegetables, fruits, aquatic products, and meat products. The detection results are consistent with national standard methods by more than 95%, demonstrating good industrial application value. Attached Figure Description

[0017] Figure 1 This is a comparison table of the technical effects of the present invention; Figure 2 This is a schematic diagram of the microfluidic chip structure of the present invention; Figure 3 This is a sensitivity comparison experiment for the present invention; Figure 4This serves to verify the applicability of the present invention to various substrates. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1-2 A rapid detection method for food safety, comprising the following steps: S1. First, sample pretreatment is required. Take the food sample to be tested, add 0.1~0.5 mol / L hydrochloric acid solution or phosphate buffer, homogenize for 1~3 minutes, and filter through a 0.22 μm organic filter membrane to obtain the sample solution. Then, use functionalized magnetic nanoparticles with a first recognition molecule modified on the surface to specifically capture the target analyte in the food sample solution, and separate and enrich it under an external magnetic field to obtain a magnetic particle-target analyte complex. The functionalized magnetic nanoparticles are Fe3O4@SiO2 core-shell structured nanoparticles, in which the Fe3O4 core (diameter 15~30 nm) provides high saturation magnetization (≥60 emu / g) to ensure rapid magnetic response; the middle layer is a silicon dioxide (SiO2) shell (thickness 5~15 nm). The nanoparticles (nm) protect the Fe3O4 core from oxidation and provide easily modifiable silanol groups (-Si-OH). Biorecognition molecules can be covalently linked via silane coupling agents (such as APTES, MPS). The density of the first recognition molecule on the surface is controlled at 10^2~10^4 molecules / particle, which can be adjusted by changing the concentration of the antibody / aptamer and the reaction time. The first recognition molecule on the surface is a specific antibody or nucleic acid aptamer targeting the analyte. The signal amplification catalyst is a three-dimensional porous Prussian blue analog nanozyme immobilized on the inner wall of the microfluidic chip channel. The preparation method is as follows: Using transition metal ions such as Co, Fe, Ni, and Cu and ferricyanide as precursors, PBA with nanocube, nanocubic octahedron, or nanoflower morphology is synthesized in the presence of surfactants (such as PVP, CTAB) via a self-templating method or hydrothermal method (120~180℃, 6~24 hours). The specific surface area can reach 200~500 nm. 2 / g, with an average pore size of 5~20 nm, the detectable physical signals are current signals or visible light absorption signals. In the detection region, the signal amplification catalyst catalyzes the substrate solution to generate electrochemical or colorimetric signals, achieving dual signal amplification.

[0020] S2. Next, the SERS substrate with magnetic nanoparticles modified with specific recognition molecules needs to be immobilized within the detection channel of the microfluidic chip. The microfluidic chip is made of polydimethylsiloxane (PDMS) or polymethyl methacrylate (PMMA) and is fabricated using soft photolithography or micro-injection molding. The chip includes the following functional areas: a sample inlet and a fluid delivery microchannel (width 100~300 μm, depth 50~100 μm, length 2~5 μm). The microfluidic chip consists of a reaction / capture zone, a signal detection zone, and a waste outlet. The inner wall of the channel is treated with oxygen plasma to change it from hydrophobic to hydrophilic, facilitating subsequent chemical modification. The detection channel is rinsed with ethanol and deionized water 3-5 times sequentially, and then dried with nitrogen gas. The resulting sample solution is injected into the detection channel of the microfluidic chip at a flow rate of 5-20 μL / min. The reaction temperature is controlled at 25-37℃, and the chip is incubated for 3-5 minutes to allow the analyte in the sample to fully bind with the specific recognition molecules on the SERS substrate. The deep learning model is a one-dimensional convolutional neural network model. The training steps of this model include: acquiring multiple raw signal samples containing noise and baseline drift, as well as the corresponding standard ideal signal; using the raw signal samples as input and the standard ideal signal as output; and training the one-dimensional convolutional neural network by minimizing the mean square error loss function until convergence.

[0021] S3. Finally, signal conversion and amplification are required. The sample solution is introduced into the detection area of ​​the microfluidic chip. A signal amplification catalyst is fixed on the inner wall of the channel of the microfluidic chip. The catalyst enhances the original signal of the metal nanoclusters, generating a detectable physical signal. Data acquisition and processing are then required. The physical signal is acquired, and a pre-trained deep learning model is used to adaptively filter noise and correct baseline drift on the acquired signal curve. The concentration of the target analyte is calculated based on the corrected signal peak intensity. Data acquisition and processing includes comparing the calculated concentration with a preset safety threshold. The comparison is performed, and if the result exceeds the safety threshold, a positive alarm signal is output; otherwise, a negative result is output, and the detection data is uploaded to the cloud server for traceability. The homogenization time is 1-3 minutes, and a 0.22μm organic filter membrane is used for filtration. The rinsing process involves rinsing the detection channel with ethanol and deionized water 3-5 times in sequence, followed by blowing with nitrogen to dry it. The analytes include at least one of pesticide residues, veterinary drug residues, heavy metals, and illegal food additives. The method for fixing functionalized magnetic nanoparticles with specific recognition molecules on the surface into the microfluidic detection channel is as follows: A suspension containing magnetic nanoparticles (0.5-2 mg / mL) is pumped into the channel at a flow rate of 5-20 μL / min. At the same time, a neodymium magnet array with a programmable switch is placed below the chip to generate a local high gradient magnetic field (0.3-0.8 T), which precisely adsorbs the magnetic nanoparticles onto a specific area (detection window) on the inner wall of the channel, forming a stable and uniform magnetic particle monolayer / multilayer film. After immobilization, the detection channel was rinsed 3-5 times with ethanol (70%~100%) and deionized water at a rate of 10~30 μL / min, for 2 minutes each time, to remove unfixed particles and residual reagents. Finally, high-purity nitrogen gas (flow rate 0.5~1 L / min) was introduced to dry the channel.

[0022] Example 1: Detection of organophosphorus pesticide residues in vegetables This embodiment is used to detect residues of two organophosphorus pesticides, dichlorvos and dimethoate, in leafy green vegetables. The specific steps are as follows: S1. Sample pretreatment: Take 1g of homogenized green vegetable sample, add 3mL of 0.2mol / L hydrochloric acid solution, vortex for 2 minutes, and filter through a 0.22μm organic filter membrane to obtain the sample solution to be tested. S2. Microfluidic chip pretreatment: The gold nanorod SERS substrate with organophosphorus pesticide aptamers modified on the surface is fixed onto the detection channel of the microfluidic chip, rinsed three times each with anhydrous ethanol and deionized water, and then dried by blowing with nitrogen gas. S3. Sample introduction and reaction: Inject the sample solution to be tested into the detection channel at a flow rate of 10 μL / min, control the temperature at 30℃, and incubate for 4 minutes; S4. Raman signal acquisition: A 785nm portable Raman spectrometer was used with an incident power of 10mW and an integration time of 2 seconds. The spectra of three uniformly distributed detection points in the channel were acquired and the average value was taken. S5. Analysis Results: The sample spectrum is within 1090 cm⁻¹. -1 and 1235cmcm -1 A distinct characteristic peak appeared at the point, matching the standard spectrum of dichlorvos. Substituting the values ​​into the standard curve, the dichlorvos content was calculated to be 0.82 mg / kg, exceeding the national limit. The test result deviated from the value determined by gas chromatography by 3.2%.

[0023] Example 2: Detection of Malachite Green Residue in Aquatic Products This embodiment is used to detect malachite green residue in grass carp. The specific steps are as follows: S1. Sample pretreatment: Take 2g of grass carp muscle homogenized sample, add 5mL of pH7.0 phosphate buffer, homogenize for 3 minutes, filter through a 0.22μm organic filter membrane to obtain the sample solution to be tested. S2. Microfluidic chip pretreatment: The gold nanorod SERS substrate with malachite green antibody modified on the surface is fixed onto the detection channel of the microfluidic chip, rinsed 5 times each with anhydrous ethanol and deionized water, and then dried by blowing with nitrogen gas. S3. Sample introduction and reaction: Inject the sample solution to be tested into the detection channel at a flow rate of 15 μL / min, control the temperature at 37℃, and incubate for 5 minutes; S4. Raman signal acquisition: A 785nm portable Raman spectrometer was used with an incident power of 15mW and an integration time of 3 seconds. The spectra of the four detection points in the channel were acquired and the average value was taken. S5. Analysis Results: The sample spectrum is at 1618 cm⁻¹ -1 1395cm -1 and 1175cm -1 The characteristic peak of malachite green was observed at the point, and the calculated content was 2.3 μg / kg. The deviation of the detection result from the value determined by liquid chromatography-tandem mass spectrometry was 4.7%.

[0024] Example 3: Traditional SERS detection method This embodiment uses the traditional centrifuge tube method for SERS detection of the same vegetable sample as in Example 1. The gold nanorod sol is directly mixed with the sample solution and then dropped onto a silicon wafer for detection. The detection time is 30 minutes. The coefficient of variation of the detection result is 18.7%, which is much higher than the 3.6% of the method of the present invention. Moreover, the limit of detection is only 5 μg / L, and the sensitivity is much lower than that of the present invention.

[0025] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A rapid detection method for food safety, characterized in that: Its detection method includes the following steps: S1. First, sample pretreatment is required. Take the food sample to be tested, add 0.1~0.5mol / L hydrochloric acid solution or phosphate buffer, homogenize for 1~3 minutes, filter through a 0.22μm organic filter membrane to obtain the sample solution to be tested, and use functionalized magnetic nanoparticles with first recognition molecules on their surface to specifically capture the target analyte in the food sample solution, and separate and enrich it under an external magnetic field to obtain the magnetic particle-target analyte complex. S2. Next, the magnetic nanoparticle SERS substrate with specific recognition molecules on its surface needs to be fixed in the detection channel of the microfluidic chip. The detection channel is rinsed with ethanol and deionized water 3-5 times in sequence, and then dried with nitrogen gas. The resulting sample solution is injected into the detection channel of the microfluidic chip at a flow rate of 5-20 μL / min. The reaction temperature is controlled at 25-37℃ and incubated for 3-5 minutes to allow the analyte in the sample to fully bind with the specific recognition molecules on the SERS substrate. S3. Finally, signal conversion and amplification processing are required. The sample solution is introduced into the detection area of ​​the microfluidic chip. A signal amplification catalyst is fixed on the inner wall of the channel of the microfluidic chip. The catalyst enhances the original signal of the metal nanoclusters to generate a detectable physical signal. Data acquisition and processing are required. The physical signal is acquired, and a pre-trained deep learning model is used to adaptively filter noise and correct baseline drift on the acquired signal curve. The concentration of the target analyte is calculated based on the corrected signal peak intensity.

2. The method for rapid detection of food safety according to claim 1, characterized in that: The functionalized magnetic nanoparticles are Fe3O4@SiO2 core-shell structured nanoparticles. The Fe3O4 core (15-30 nm in diameter) provides high saturation magnetization (≥60 emu / g), ensuring a rapid magnetic response. The intermediate layer is a silica (SiO2) shell (5-15 nm thick), which protects the Fe3O4 core from oxidation and provides easily modifiable silanol groups (-Si-OH), allowing for covalent linking of biorecognition molecules via silane coupling agents (such as APTES, MPS). The density of the first surface-modified recognition molecule is controlled at 10^2-10^4 molecules / particle, adjustable by varying the concentration of the antibody / aptamer and the reaction time. The first surface-modified recognition molecule is a specific antibody or nucleic acid aptamer targeting the analyte.

3. The method for rapid detection of food safety according to claim 1, characterized in that: The signal amplification catalyst is a three-dimensional porous Prussian blue analog nanozyme fixed on the inner wall of the microfluidic chip channel. Its preparation method is as follows: using transition metal ions such as Co, Fe, Ni, and Cu, and ferricyanide as precursors, PBA with nanocube, nanocubic octahedron, or nanoflower morphologies is synthesized via a self-templating method or hydrothermal method (120~180℃, 6~24 hours) in the presence of surfactants (such as PVP or CTAB). The specific surface area can reach 200~500 m². 2 / g, with an average pore size of 5~20 nm, the detectable physical signal is a current signal or a visible light absorption signal. In the detection region, the signal amplification catalyst catalyzes the substrate solution to generate an electrochemical or colorimetric signal, thereby achieving dual signal amplification.

4. The rapid detection method for food safety according to claim 1, characterized in that: The deep learning model is a one-dimensional convolutional neural network model. The training steps of the model include: acquiring multiple original signal samples containing noise and baseline drift, as well as the corresponding standard ideal signal; using the original signal samples as input and the standard ideal signal as output, training the one-dimensional convolutional neural network by minimizing the mean square error loss function until convergence.

5. The rapid detection method for food safety according to claim 1, characterized in that: The data acquisition and processing work includes comparing the calculated concentration with a preset safety threshold. If the concentration exceeds the safety threshold, a positive alarm signal is output; otherwise, a negative result is output, and the detection data is uploaded to a cloud server for traceability.

6. The rapid detection method for food safety according to claim 1, characterized in that: The homogenization process takes 1 to 3 minutes, and a 0.22 μm organic filter membrane is used for filtration. The rinsing process involves rinsing the detection channel with ethanol and deionized water 3 to 5 times in sequence, followed by blowing with nitrogen gas to dry it. The analytes include at least one of pesticide residues, veterinary drug residues, heavy metals, and illegal food additives.

7. The method for rapid detection of food safety according to claim 1, characterized in that: The microfluidic chip is made of polydimethylsiloxane (PDMS) or polymethyl methacrylate (PMMA) and is fabricated using soft photolithography or micro-injection molding. The chip includes the following functional areas: a sample inlet, a fluid delivery microchannel (width 100–300 μm, depth 50–100 μm, length 2–5 cm), a reaction / capture zone, a signal detection zone, and a waste outlet. The inner wall of the channel is treated with oxygen plasma to change it from hydrophobic to hydrophilic, facilitating subsequent chemical modification.

8. The rapid detection method for food safety according to claim 1, characterized in that: The method for immobilizing functionalized magnetic nanoparticles with specific recognition molecules on their surface within a microfluidic detection channel is as follows: A suspension containing magnetic nanoparticles (0.5~2 mg / mL) is pumped into the channel at a flow rate of 5~20 μL / min. Simultaneously, a programmable neodymium magnet array is placed below the chip to generate a local high-gradient magnetic field (0.3~0.8 T), precisely adsorbing the magnetic nanoparticles onto a specific area (detection window) on the inner wall of the channel, forming a stable and uniform single-layer / multilayer magnetic particle film. After immobilization, the detection channel is rinsed 3~5 times sequentially with ethanol (70%~100%) and deionized water at a flow rate of 10~30 μL / min, for 2 minutes each time, to remove unimmobilized particles and residual reagents. Finally, high-purity nitrogen gas (flow rate 0.5~1 L / min) is introduced to dry the channel.

Images