Method for exploring photochemical reactivity of microplastic-derived soluble organic matter

Abstract

The invention relates to the technical field of dissolvable organic matters, and discloses a method for exploring photochemical reactivity of microplastic derived dissolvable organic matters, which comprises the following steps: S1, sample preparation: selecting microplastic as a source of the dissolvable organic matters, and generating MP-DOM through a leaching experiment; different types of MP-DOM samples are subjected to illumination treatment, and the DOM samples before and after illumination are analyzed through Fourier transform ion cyclotron resonance mass spectrometry; s2, generating a molecular formula data set: calculating a plurality of molecular parameters for describing chemical characteristics of each molecular formula by using molecular formula data generated by Fourier transform ion cyclotron resonance mass spectrometry; according to the method for exploring the photochemical reactivity of the micro-plastic derived soluble organic matter, the FT-ICR-MS technology and the machine learning algorithm are combined, the RF model is superior to other models in predicting the MP-DOM photochemical reactivity, the complex DOM photochemical reactivity can be effectively analyzed, and a more accurate prediction result is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of dissolved organic matter, and specifically to a method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter. Background Art

[0002] Dissolved organic matter (DOM) is one of the largest active carbon reservoirs on Earth. Its chemical reactivity in aquatic environments is closely related to carbon and nutrient cycling, trace element transport, and responses to environmental pollutants. Photochemical reactions, such as photogeneration and photodegradation, directly affect the chemical properties of DOM and further influence its response to environmental pollutants. In recent years, with the increasing problem of plastic pollution, microplastic-derived DOM (MP-DOM) has become a research focus. Traditional DOM photochemical research relies on experimental data and molecular formula classification, but this method has problems such as "label conflict" and instability of experimental conditions. Therefore, a more efficient and accurate method is urgently needed to evaluate the photochemical reactivity of DOM. Summary of the Invention

[0003] (1) Technical problems solved In response to the shortcomings of the existing technology, the present invention provides a method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter, which has the advantage of being able to efficiently predict the reactivity of dissolved organic matter in different samples under light conditions.

[0004] (2) Technical solution To achieve the above-mentioned purpose of being able to efficiently predict the reactivity of dissolved organic matter in different samples under light conditions, the present invention provides the following technical solution: a method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter, comprising the following steps: S1. Sample preparation: Microplastics were selected as a source of dissolved organic matter and MP-DOM was generated through leaching experiments. Different types of MP-DOM samples were treated with light and analyzed by Fourier transform ion cyclotron resonance mass spectrometry before and after illumination. S2. Molecular formula data set generation: Using the molecular formula data generated by Fourier transform ion cyclotron resonance mass spectrometry, a variety of molecular parameters are calculated to describe the chemical characteristics of each molecular formula; S3. Machine Learning Model Establishment: Use a machine learning regression model to predict photochemical reactivity; assign three probabilistic labels of photochemical types to each molecular formula; S4. Label assignment and training: For each molecular formula, calculate the probability of its belonging to the three photochemical types in four illumination experiments; S5. Model training and evaluation: Use training and test datasets to train the machine learning regression model and evaluate the model performance; S6. Prediction and photochemical reactivity evaluation: For each molecular formula, a probability label of its photochemical reactivity is predicted through a machine learning regression model, and the relative intensities of different photochemical types are calculated based on the probability label, thereby achieving a semi-quantitative estimation of the sample's photochemical reactivity.

[0005] As a preferred technical solution of the present invention, in step S1, the microplastics are PE, PS, PET and PVC.

[0006] As a preferred technical solution of the present invention, in step S2, the molecular parameters include element ratio, modified aromatic index and equivalent double bond number.

[0007] As a preferred technical solution of the present invention, in step S3, the machine learning regression model includes random forest, linear regression, K nearest neighbor regression, XGBoost and support vector regression.

[0008] As a preferred technical solution of the present invention, the three photochemical types are photostable type, photosensitivity type and photoproduct type.

[0009] As a preferred technical solution of the present invention, in step S5, the model performance is evaluated by cross-validation, root mean square error and determination coefficient.

[0010] As a preferred technical solution of the present invention, in step S1, a xenon lamp is used as a light source for the illumination treatment.

[0011] As a preferred technical solution of the present invention, in step S2, the molecular parameters also include molecular mass, molecular structure complexity and photochemical degradation rate.

[0012] (3) Beneficial effects Compared with the prior art, the present invention provides a method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter, which has the following beneficial effects: 1. This method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter, by combining FT-ICR-MS technology and machine learning algorithms, shows that the RF model is superior to other models in predicting the photochemical reactivity of MP-DOM. It can effectively analyze the complex photochemical reactivity of DOM and provide more accurate prediction results.

[0013] 2. This method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter solves problems such as "label conflict" and sample source differences in traditional molecular matching methods, and provides a solution based on probabilistic labels.

[0014] 3. This method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter can provide a new photochemical reactivity research tool for microplastic-derived DOM (MP-DOM), providing important data support for environmental pollution research. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flow chart of the method of the present invention; Figure 2 Performance comparison table of five selected machine learning algorithms; Figure 3 Performance of random forest models for three photochemical reaction types; Figure 4 VK diagram and Shapley value for molecules with well-defined photochemical reactivity. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] Example: See also Figures 1-4 The present invention proposes a method for exploring the photochemical reactivity of microplastic-derived dissolved organic matter, which specifically includes the following steps: Step 1: Sample Preparation: Common microplastics (in this example, PE, PS, PET, and PVC) were selected as a source of dissolved organic matter (DOM) and microplastic-derived dissolved organic matter (MP-DOM) was generated through leaching experiments. In the leaching experiments, microplastic particles were mixed with water and exposed to a xenon lamp at a constant temperature to simulate illumination, generating MP-DOM samples at different illumination times. During illumination, the chemical components of the microplastics interacted with the dissolved organic matter, generating dissolved organic matter with varying photochemical reactivity. The DOM samples before and after illumination were then analyzed using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) to obtain detailed molecular formula data.

[0018] Step 2: Molecular formula data set generation: Using the molecular formula data generated by Fourier transform ion cyclotron resonance mass spectrometry, multiple molecular parameters are calculated to describe the chemical properties of each molecular formula (MF). In this embodiment, the multiple molecular parameters specifically include: Element ratios (such as O / C, H / C, etc.) reflect the molecular composition of dissolved organic matter and its possible photochemical reactivity by calculating the element ratios; The modified aromatic index (AImod) reflects the aromatic structure of dissolved organic matter molecules and thus affects their photochemical stability; The number of double bonds equivalent (DBE), which indicates the possible ring structure and degree of unsaturation in the molecule, all of which may affect the molecule's response to light; In this embodiment, the molecular parameters may also include parameters such as molecular mass, molecular structure complexity, and photochemical degradation rate, so as to further optimize the prediction of photochemical reactivity.

[0019] Step 3. Machine learning model establishment: Five machine learning regression models, including random forest (RF), linear regression (LR), K-nearest neighbor regression (KNN), XGBoost (XGB) and support vector regression (SVR), are used to predict photochemical reactivity; for each molecular formula, the probability labels of three photochemical types (photostable type, photosensitivity type, photoproduct type) are assigned. Through these labels, the model can predict the photochemical reactivity of each molecular formula.

[0020] Step 4: Label Assignment and Training: For each molecular formula, calculate the probability of its attribution to the three photochemical types across the four illumination experiments. For example, if a MF was classified as photostable twice, photosensitized once, and photoproductive once in the four classification experiments, the label for the MF would be [0.5, 0.25, 0.25]. This process is used to train the machine learning model, enabling it to automatically learn and optimize the attribution rules for each photochemical type.

[0021] S5. Model training and evaluation: The machine learning regression model is trained using a training dataset (80%) and a test dataset (20%). During the training process, cross-validation is used to avoid overfitting, and the model performance is evaluated using indicators such as root mean square error (RMSE) and coefficient of determination (R²).

[0022] In this example, five different regression models were compared to determine the most suitable model. Experiments demonstrated that the random forest (RF) model outperformed other models in multiple evaluation metrics, particularly in predicting photochemical reactivity.

[0023] S6. Prediction and Photochemical Reactivity Assessment: For each molecular formula, a machine learning regression model predicts a probabilistic label for its photochemical reactivity and calculates the relative intensities of different photochemical types based on the probabilistic labels. This probabilistic label calculation enables the model to provide a semi-quantitative assessment of the photochemical reactivity of each sample, helping researchers more accurately predict the likely photochemical behavior of different samples in real-world environments.

[0024] This study used four common microplastics (PE, PS, PET, and PVC) as research subjects, analyzing DOM samples before and after illumination using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS). Through illumination experiments, molecular formula data for the different samples was obtained. Combined with a machine learning model, the reactivity of different photochemical types within each sample was successfully predicted. The experimental results demonstrate that this method provides a more accurate and comprehensive assessment of photochemical reactivity than traditional molecular matching methods.

[0025] like Figure 2 The table shows the performance of the five selected machine learning algorithms. Among them, the RF model has a higher accuracy in the prediction of "photostable" molecules, and the model can also provide good discrimination in the prediction of "photosensitive" and "photoproduct" molecules.

[0026] Figure 3 The classification results of photochemical reaction types based on the random forest model are presented, indicating that the RF model has good classification performance in identifying photostable types, photosensitivity types and photoproduct types. Figure 4 The VK diagram and Shapley value of molecules with clear photochemical reactivity before and after illumination are displayed to further analyze the relationship between molecular formula and photochemical reactivity.

[0027] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter, characterized in that: The following steps are involved: S1. Sample preparation: Microplastics were selected as the source of dissolved organic matter and MP-DOM was generated through leaching experiments. Different types of MP-DOM samples were treated with light, and the DOM samples before and after light treatment were analyzed by Fourier transform ion cyclotron resonance mass spectrometry; S2. Molecular formula data set generation: Using the molecular formula data generated by Fourier transform ion cyclotron resonance mass spectrometry, a variety of molecular parameters are calculated to describe the chemical characteristics of each molecular formula; S3. Machine Learning Model Establishment: Use a machine learning regression model to predict photochemical reactivity; assign three probabilistic labels of photochemical types to each molecular formula; S4. Label assignment and training: For each molecular formula, calculate the probability of its belonging to the three photochemical types in four illumination experiments; S5. Model training and evaluation: Use training and test datasets to train the machine learning regression model and evaluate the model performance; S6. Prediction and photochemical reactivity evaluation: For each molecular formula, a probability label of its photochemical reactivity is predicted through a machine learning regression model, and the relative intensities of different photochemical types are calculated based on the probability label, thereby achieving a semi-quantitative estimation of the sample's photochemical reactivity.

2. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1, characterized in that: In step S1, the microplastics are PE, PS, PET and PVC.

3. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1, characterized in that: In step S2, the molecular parameters include element ratio, modified aromatic index and equivalent double bond number.

4. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1, characterized in that: In step S3, the machine learning regression model includes random forest, linear regression, K nearest neighbor regression, XGBoost and support vector regression.

5. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1 or 4, characterized in that: The three photochemical types are photostable, photosensitivity, and photoproduct.

6. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1, characterized in that: In step S5, the model performance is evaluated by cross-validation, root mean square error and determination coefficient.

7. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1, characterized in that: In step S1 , a xenon lamp is used as a light source for the illumination treatment.

8. The method for investigating the photochemical reactivity of microplastic-derived dissolved organic matter according to claim 1 or 3, characterized in that: In step S2, the molecular parameters also include molecular mass, molecular structure complexity and photochemical degradation rate.

Images