A method for predicting the environmental transformation performance of pollutants
By utilizing the reactivity parameters of pollutants and environmental media through activation energy prediction models, the conversion performance of chemicals on the surface of environmental media can be quickly predicted, solving the problem of time-consuming and labor-intensive processes in existing technologies and achieving efficient risk assessment and safety evaluation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- RES CENT FOR ECO ENVIRONMENTAL SCI THE CHINESE ACAD OF SCI
- Filing Date
- 2023-03-31
- Publication Date
- 2026-06-30
AI Technical Summary
Existing technologies are insufficient for rapidly screening the conversion properties of tens of thousands of chemicals on environmental media surfaces. Experimental methods consume a large amount of manpower and resources, and high-precision computational chemistry methods lack sufficient data accumulation, making it difficult to achieve large-scale analysis.
By acquiring the reactivity parameters of pollutants and environmental media, and using a pre-trained activation energy prediction model, the reaction activation energy is output to predict the transformation performance of pollutants on the surface of environmental media.
It enables rapid screening of the transformation properties of pollutant molecules on the surface of environmental media, and is suitable for large-scale pollutant molecule transformation risk assessment and environmental safety assessment, simplifying the screening process.
Smart Images
Figure CN116403658B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of environmental pollution control technology, and in particular to a method for predicting the environmental transformation performance of pollutants. Background Technology
[0002] In recent years, the quantity and variety of chemicals produced by humans have been continuously increasing. These chemicals are widely used in industry, agriculture, medicine, and daily life, bringing convenience to people while inevitably entering the environmental media. Besides their inherent toxicity, the secondary transformation of these chemicals into highly toxic substances on the surface of environmental media has attracted widespread attention. Therefore, conducting assessments of the surface transformation performance of pollutants in environmental media is of great significance for environmental protection and pollution control.
[0003] Currently, research on the surface transformation properties of pollutants in environmental media generally relies on experimental methods or analysis at the atomic level using high-precision computational chemistry. However, experimental research or high-precision theoretical calculations are insufficient for rapidly screening the environmental transformation properties of tens of thousands of chemicals. Summary of the Invention
[0004] To address the aforementioned technical problems, this disclosure provides a method for predicting the environmental transformation performance of pollutants, thereby at least partially solving at least one of the aforementioned technical problems.
[0005] To address the aforementioned technical problems, this disclosure provides a method for predicting the environmental transformation performance of pollutants, comprising:
[0006] Obtain a first target reactivity parameter of the target pollutant and a second target reactivity parameter of the target environmental medium, wherein the first target reactivity parameter includes at least one of the following: a first ionization potential, a first affinity potential, molecular hardness, and molecular softness; and the second target reactivity parameter includes: redox potential.
[0007] The first target reactivity parameter and the second target reactivity parameter are input into a pre-trained activation energy prediction model, which outputs the target activation energy, wherein the target activation energy characterizes the activation energy of the transformation of the target pollutant on the surface of the target environmental medium; and
[0008] Based on the activation energy of the target reaction, the conversion performance of the target pollutant in the target environmental medium is predicted.
[0009] According to embodiments of this disclosure, the training method for the above-mentioned activation energy prediction model includes:
[0010] Obtain a training sample dataset, wherein the training sample dataset includes at least one set of sample data groups, and each set of sample data groups includes sample pollutant reactivity parameters, sample environmental medium reactivity parameters, and sample reaction activation energy of sample pollutant transformation on the surface of sample environmental medium;
[0011] Input the above-mentioned pollutant reactivity parameters and environmental medium reactivity parameters from the above-mentioned training sample dataset into the initial prediction model to obtain the predicted reaction activation energy;
[0012] The loss value is determined based on the predicted activation energy and the activation energy of the sample reaction; and
[0013] The network parameters of the initial prediction model are adjusted using the aforementioned loss value until the preset iteration conditions are met, thus obtaining the activation energy prediction model.
[0014] According to embodiments of this disclosure, the method for determining the activation energy of the above-mentioned sample reaction includes:
[0015] Establish a structural model of the starting reactants, wherein the above structural model of the starting reactants is formed by combining the molecular structure model of the sample pollutants with the surface model of the sample environmental medium;
[0016] Establish a structural model of the reaction product, wherein the above-mentioned structural model of the reaction product is the structural model formed after the above-mentioned sample pollutant molecular structure model reacts with the above-mentioned sample environmental medium surface model;
[0017] Based on the above-mentioned initial reactant structure model and the above-mentioned reaction product structure model, a preset method is used to search for reaction stationary points and obtain multiple reaction pathways.
[0018] Based on the energy of the transition state structure corresponding to each of the above reaction pathways and the energy of the above starting reactant structure model, the activation energy of the above sample reaction of the above sample pollutants on the surface of the above sample environmental medium is determined.
[0019] According to embodiments of this disclosure, the above-mentioned establishment of the starting reactant structural model includes:
[0020] Based on the various adsorption modes of the above-mentioned sample pollutant molecules on the surface of the above-mentioned sample environmental medium, all complex models obtained by combining the above-mentioned sample pollutant molecule structure model with the above-mentioned sample environmental medium surface model are established.
[0021] Calculate the energy of each of the above complex models;
[0022] The target complex model is determined based on the energy of the aforementioned complex model, and the aforementioned target complex model is used as the structural model of the aforementioned starting reactant.
[0023] According to embodiments of this disclosure, based on the various adsorption modes of the above-mentioned sample pollutant molecules on the surface of the above-mentioned sample environmental medium, all complex models obtained by combining the above-mentioned sample pollutant molecule structure model with the above-mentioned sample environmental medium surface model include:
[0024] Based on the various adsorption modes of the above-mentioned sample pollutant molecules on the surface of the above-mentioned sample environmental medium, models of all initial adsorption complexes obtained by combining the above-mentioned sample pollutant molecule structure model with the above-mentioned sample environmental medium surface model are established.
[0025] Configuration optimization was performed on each of the above initial adsorption complex models to obtain the above complex models corresponding to each of the above initial adsorption complex models.
[0026] According to embodiments of this disclosure, the method for establishing the above-mentioned sample environmental medium surface model includes:
[0027] The cell structures of the sample environment medium components were selected from the inorganic crystal structure database, and the surface model of the sample environment medium was established.
[0028] According to embodiments of this disclosure, the method for establishing the molecular structure model of the above-mentioned sample pollutants includes:
[0029] The above-mentioned sample pollutants were placed in a unit cell of the same size as the surface model of the above-mentioned sample environmental medium, and the configuration was optimized to obtain the molecular structure model of the above-mentioned sample pollutants.
[0030] According to embodiments of this disclosure, the target environmental media include soil, sediment, or particulate matter.
[0031] According to an embodiment of this disclosure, the method for determining the first ionization potential includes: determining the difference between the single-point energy of the target pollutant cation fragment and the single-point energy of the target pollutant to obtain the first ionization potential, wherein the target pollutant cation fragment includes the structure obtained after the target pollutant loses one electron;
[0032] The method for determining the first affinity includes: determining the difference between the single-point energy of the target pollutant and the single-point energy of the target pollutant anionic fragment, and obtaining the first affinity, wherein the target pollutant anionic fragment includes the structure obtained after the target pollutant gains an electron;
[0033] The aforementioned molecular hardness and molecular softness are determined based on the aforementioned first ionization potential and the aforementioned first affinity.
[0034] According to embodiments of this disclosure, the first target reactivity parameter and the second target reactivity parameter are determined by searching a literature database; and / or
[0035] The aforementioned first target reactivity parameter and the aforementioned second target reactivity parameter were determined by density functional theory calculations.
[0036] According to embodiments of this disclosure, by obtaining a first target reactivity parameter of the target pollutant and a second target reactivity parameter of the target environmental medium, the first and second target reactivity parameters are input into a pre-trained activation energy prediction model, which outputs the activation energy characterizing the transformation of the target pollutant on the surface of the target environmental medium, i.e., the target activation energy. Then, the transformation performance of the target pollutant on the target environmental medium is predicted based on the target activation energy. Therefore, this disclosure utilizes an activation energy prediction model to rapidly screen the potential transformation risk of target pollutants based on the reactivity parameters of pollutant molecules and the environmental medium. The screening method is simple and efficient, suitable for large-scale screening and assessment of pollutant molecular transformation risk hazards, and has broad application prospects in the fields of pollutant molecular health risk assessment, environmental safety assessment, and prediction. Attached Figure Description
[0037] Figure 1 A flowchart illustrating a method for predicting the environmental transformation performance of pollutants according to embodiments of the present disclosure is shown schematically.
[0038] Figure 2 A flowchart illustrating a training method for an activation energy prediction model according to an embodiment of the present disclosure is shown schematically.
[0039] Figure 3 This illustration schematically shows a diagram of the partitioning of a training sample dataset according to an embodiment of the present disclosure;
[0040] Figure 4 This illustration schematically shows a comparison of the performance of different machine learning algorithms according to embodiments of the present disclosure;
[0041] Figure 5 The diagram illustrates a performance analysis of an activation energy prediction model according to an embodiment of the present disclosure. Detailed Implementation
[0042] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.
[0043] Since the Industrial Revolution, the quantity and variety of chemicals produced by humankind have continuously increased. These chemicals are widely used in industry, agriculture, medicine, and daily life, bringing us convenience while inevitably entering the environmental media. Besides their inherent toxicity, the secondary transformation of these chemicals into highly toxic substances on the surface of environmental media has attracted widespread attention. Therefore, conducting performance assessments of pollutant surface transformation is of great significance for environmental protection and pollution control.
[0044] In related technologies, research on the surface transformation performance of pollutants in environmental media generally relies on experimental methods or high-precision computational chemistry methods for analysis at the atomic level. While experimental methods can provide relatively intuitive information on the surface reactivity of chemicals, observing the experimental reaction process of each compound requires significant human, material, and financial resources. Although high-precision theoretical calculations can provide microscopic explanations of the electronic structure, ground-state energy, and potential energy surface of reaction processes, the limited data accumulation in computational chemistry, coupled with the need to establish numerous structures and perform optimization and computational work, makes conducting high-precision theoretical calculations for tens of thousands of chemicals a challenging task.
[0045] To address the aforementioned technical problems, this disclosure provides a method for predicting the environmental transformation performance of pollutants. This method is based on an activation energy prediction model, which uses the reactivity parameters of pollutants and the reactivity parameters of the environmental medium as inputs to output the activation energy of the pollutant transformation on the surface of the environmental medium. Thus, the transformation performance of pollutants on the environmental medium is predicted based on the activation energy, achieving rapid prediction of the transformation performance of pollutant molecules on the surface of the environmental medium.
[0046] Figure 1 A flowchart illustrating a method for predicting the environmental transformation performance of pollutants according to embodiments of the present disclosure is shown.
[0047] like Figure 1 As shown, according to an embodiment of the present disclosure, the method for predicting the environmental conversion performance of pollutants includes operations S110 to S130.
[0048] In operation S110, a first target reactivity parameter of the target pollutant and a second target reactivity parameter of the target environmental medium are obtained, wherein the first target reactivity parameter includes at least one of the following: a first ionization potential, a first affinity potential, molecular hardness, and molecular softness; the second target reactivity parameter includes: redox potential.
[0049] According to embodiments of this disclosure, the target pollutant can be any chemical pollutant. For example, the pollutant can include phenolic compounds. Specifically, the target pollutant can include phenol and phenol molecules para-substituted with different substituent groups (F, Cl, Br, NO2, NH2, CH3, OCH3, OH).
[0050] According to embodiments of this disclosure, the target environmental medium can be any environmental medium, such as soil, sediment, particulate matter, etc. Specifically, for example, the target environmental medium can include Fe(III)-montmorillonite, K(I)-montmorillonite, Zn(II)-montmorillonite, Ti(III)-montmorillonite, and Ni(II)-montmorillonite.
[0051] According to embodiments of this disclosure, the first target reactivity parameter and the second target reactivity parameter are determined by searching a literature database; and / or
[0052] The first and second target reactivity parameters were determined by density functional theory calculations.
[0053] According to embodiments of this disclosure, for example, the first target reaction parameters of the target pollutant can be calculated using high-precision quantum chemical calculation methods with the Gaussian 16 quantum chemistry software.
[0054] Gaussian 16 is a chemical analysis tool whose executable can run on various types of mainframes, supercomputers, workstations, and personal computers, with corresponding versions available. Gaussian functions include: transition state energy and structure, bond and reaction energies, molecular orbitals, atomic charges and potentials, vibrational frequencies, infrared and Raman spectra, nuclear magnetic properties, polarizability and hyperpolarizability, thermodynamic properties, and reaction pathways. Calculations can be performed on the ground or excited states of the system. It can predict the energy, structure, and molecular orbitals of periodic systems, and is used to study many topics in chemistry, such as the effects of substituents, chemical reaction mechanisms, potential energy surfaces, and excitation energies.
[0055] In operation S120, the first target reactivity parameter and the second target reactivity parameter are input into the pre-trained activation energy prediction model, and the target reaction activation energy is output. The target reaction activation energy characterizes the activation energy of the target pollutant transformation on the surface of the target environmental medium.
[0056] In operation S130, the conversion performance of the target pollutant in the target environmental medium is predicted based on the target reaction activation energy.
[0057] According to embodiments of this disclosure, by obtaining a first target reactivity parameter of the target pollutant and a second target reactivity parameter of the target environmental medium, the first and second target reactivity parameters are input into a pre-trained activation energy prediction model, which outputs the activation energy characterizing the transformation of the target pollutant on the surface of the target environmental medium, i.e., the target activation energy. Then, the transformation performance of the target pollutant on the target environmental medium is predicted based on the target activation energy. Therefore, this disclosure utilizes an activation energy prediction model to rapidly screen the potential transformation risk of target pollutants based on molecular and environmental medium reactivity parameters. The screening method is simple and efficient, suitable for large-scale screening and assessment of pollutant molecular transformation risk hazards, and has broad application prospects in the fields of pollutant molecular health risk assessment, environmental safety assessment, and prediction.
[0058] According to an embodiment of this disclosure, the method for determining the first ionization potential includes: determining the difference between the single-point energy of the target pollutant cationic fragment and the single-point energy of the target pollutant to obtain the first ionization potential, wherein the target pollutant cationic fragment includes the structure obtained by the target pollutant after losing one electron;
[0059] The method for determining the first affinity includes: determining the difference between the single-point energy of the target pollutant and the single-point energy of the target pollutant anionic fragment, and obtaining the first affinity, wherein the target pollutant anionic fragment includes the structure obtained after the target pollutant gains an electron;
[0060] Molecular hardness and molecular softness are determined based on the first ionization potential and the first affinity potential.
[0061] According to embodiments of this disclosure, the single-point energy of the target pollutant, the single-point energy of the target pollutant cationic fragment, and the single-point energy of the target pollutant anionic fragment can be determined using the following method: The high-precision double-hybrid functional MO6-2X method, combined with the 6-311G(d,p) basis set, is used to perform structural optimization and frequency calculation on the neutral molecule of the target pollutant. Based on the structural optimization, the MO6-2X / 6-311++G(3df,2p) method is used to calculate the single-point energy of the target pollutant, the single-point energy of the target pollutant cationic fragment, and the single-point energy of the target pollutant anionic fragment. The single-point energy of the target pollutant is denoted as E. M The single-point energy of the target pollutant cation fragment, i.e., the target pollutant molecule losing one electron and becoming a +1 valence ion M-1, is denoted as E. M-1 The single-point energy of the target pollutant anionic fragment, i.e., the target pollutant molecule gaining an electron and becoming a -1 valence M+1 fragment, is denoted as E. M+1 .
[0062] According to embodiments of this disclosure, the first ionization potential, first affinity potential, molecular hardness, and molecular softness of the target pollutant can be calculated using the following formula:
[0063] IP = E M-1 –E M (one)
[0064] EA = E M –E M+1 (two)
[0065]
[0066]
[0067] in,
[0068] IP, EA, η and S are the first ionization potential, the first affinity, molecular hardness and molecular softness, respectively.
[0069] E M The single-point energy of the target pollutant molecule;
[0070] E M-1 The single-point energy of the target pollutant's cationic fragment;
[0071] E M+1 This refers to the single-point energy of the anionic fragment of the target pollutant.
[0072] The M06-2X functional belongs to the Minnesota functional series and is a method in density functional theory. It can be used for thermodynamic, kinetic and non-covalent interaction systems of main group elements, and provides reliable prediction results, providing strong theoretical support for understanding the electronic properties of chemical substances from the microscopic world.
[0073] A basis set is a set of mathematical functions used in quantum chemical calculations to describe the wavefunction of a system. To balance computational accuracy and efficiency, valence electron orbitals are split into more basis functions. The inner electron orbitals of the 6-311G(d,p) and 6-311++G(3df,2p) basis sets are composed of combinations of six Gaussian functions, while each valence electron is split into two basis functions, consisting of linear combinations of three and one Gaussian functions, respectively. Furthermore, to better describe the properties of electron cloud deformation and nonbonding interactions, polarization and dispersion functions for d and p are introduced into the split valence basis sets.
[0074] Figure 2 A flowchart illustrating a training method for an activation energy prediction model according to an embodiment of the present disclosure is shown.
[0075] like Figure 2 As shown, the training method for the activation energy prediction model in this embodiment includes operations S210 to S240.
[0076] In operation S210, a training sample dataset is obtained, wherein the training sample dataset includes at least one set of sample data groups, and each set of sample data groups includes sample pollutant reactivity parameters, sample environmental medium reactivity parameters, and sample reaction activation energies of sample pollutant transformation on the sample environmental medium surface.
[0077] In operation S220, the sample pollutant reactivity parameters and sample environmental medium reactivity parameters from the training sample dataset are input into the initial prediction model to obtain the predicted reaction activation energy.
[0078] In operation S230, the loss value is determined based on the predicted reaction activation energy and the sample reaction activation energy.
[0079] In operation S240, the network parameters of the initial prediction model are adjusted using the loss value until the preset iteration conditions are met, thus obtaining the activation energy prediction model.
[0080] According to embodiments of this disclosure, the sample contaminant may be a phenolic compound.
[0081] According to embodiments of this disclosure, the reactivity parameters of the sample pollutant may include a first ionization potential, a first affinity potential, molecular hardness, and molecular softness.
[0082] According to embodiments of this disclosure, the sample environment medium can be an Fe(III)-montmorillonite surface, a K(I)-montmorillonite surface, a Zn(II)-montmorillonite surface, a Ti(III)-montmorillonite surface, and a Ni(II)-montmorillonite surface.
[0083] According to embodiments of this disclosure, the reactivity parameters of the sample environment medium can be determined by consulting literature. These parameters may include five metal cations from Fe(III)-, K(I)-, Zn(II)-, Ti(III)-, and Ni(II)-montmorillonite. 3+ K + Zn 2+ Ti 3+ , and Ni 2+ The redox potential (RP).
[0084] According to embodiments of this disclosure, the initial model can be trained using N-fold cross-validation to improve the generalization ability of the activation energy prediction model. Alternatively, a grid search method can be used to adjust the network parameters of the initial model to obtain the optimal model.
[0085] According to embodiments of this disclosure, the training method for the activation energy prediction model includes: dividing the training sample dataset into a training set and a test set according to a certain ratio, training the initial model using the training set, and testing the trained activation energy prediction model using the test set.
[0086] According to embodiments of this disclosure, the initial model may include an initial model built using six machine learning algorithms, such as support vector machine, random forest, and decision tree.
[0087] Figure 3 A schematic diagram illustrating the partitioning of a training sample dataset according to an embodiment of the present disclosure is shown.
[0088] like Figure 3 As shown in the embodiments of this disclosure, the training sample dataset is randomly divided into a training set and a test set in a ratio of 8:2, and the division of the training set and the test set data is reasonable.
[0089] Figure 4 The diagram illustrates a comparison of the performance of different machine learning algorithms according to embodiments of the present disclosure.
[0090] Initial models were built using extreme gradient boosting (XGBoost), decision trees (DT), random forests (RF), linear regression (LR), support vector machines (SVM), and multilayer perceptrons (MLP), respectively. Then, a 5-fold cross-validation method was used with the training dataset. This involved dividing the training dataset into five parts, using four parts as the training set and the remaining part as the validation set. The training set was used to train the initial model, and the validation set was used to evaluate the model's performance. This process was repeated five times, each time using one part as the validation set, resulting in five trained models and five evaluation results. (Using R...) 2 Both RMSE and other methods are used for model evaluation, such as Figure 4 As shown, among the six machine learning models, the decision tree model has the highest R-value on the training set data. 2 =0.899 is the largest, and RMSE=0.236 is the smallest, therefore the optimal model is the decision tree model.
[0091] Figure 5 The diagram illustrates a performance analysis of an activation energy prediction model according to an embodiment of the present disclosure.
[0092] The performance of the activation energy prediction model was evaluated using a test set, and the predicted values were compared with the actual values. The performance metrics of the activation energy prediction model included the goodness of fit R0. 2 And RMSE, the calculation formula is as follows:
[0093]
[0094]
[0095] Where n represents the total number of samples in the dataset, and y i The distribution represents the best model prediction and the true value for the i-th data point. All predicted values The average value.
[0096] The robustness of the machine learning model is verified based on the above results.
[0097] like Figure 5 As shown, after inputting the independent variables from the test set—namely, the sample pollutant reactivity parameters and the sample environmental medium reactivity parameters—into the activation energy prediction model, based on R... 2Model evaluation was performed using RMSE, where R0 of the test set data was used. 2 =0.928, RMSE=0.222, indicating that the method described in this disclosure can rapidly predict and screen the environmental transformation performance of pollutants based solely on the calculation of environmental media and pollutant molecular reactivity parameters, and is applicable to pollutant environmental transformation risk assessment.
[0098] According to embodiments of this disclosure, the method for determining the activation energy of a sample reaction includes:
[0099] Establish a structural model of the starting reactants, which is formed by combining the molecular structure model of the sample pollutants with the surface model of the sample environmental medium;
[0100] Establish a structural model of the reaction product, wherein the structural model of the reaction product is the structural model formed after the reaction between the molecular structural model of the sample pollutant and the surface model of the sample environmental medium;
[0101] Based on the structural models of the starting reactants and the reaction products, a preset method is used to search for reaction stationary points, resulting in multiple reaction pathways.
[0102] Based on the energy of the transition state structure corresponding to each reaction pathway and the energy of the initial reactant structure model, the activation energy of the sample reaction for the transformation of sample pollutants on the surface of the sample environmental medium is determined.
[0103] According to embodiments of this disclosure, the method for establishing a sample environment medium surface model includes: selecting the cell structure of the sample environment medium components from an inorganic crystal structure database and establishing a sample environment medium surface model.
[0104] According to embodiments of this disclosure, the method for establishing a sample environmental medium surface model specifically includes: selecting the corresponding component's unit cell structure from an Inorganic Crystal Structure Data (ICSD) database based on the known surface structure of adsorbed environmental soil components and their binding mode with pollutant molecules and their mechanism of action on pollutant transformation; constructing the selected unit cell structure using methods such as isomorphic substitution; cutting the unit cell structure according to the Miller index (hkl) of the required environmental medium surface; and adding a vacuum layer of x nm in the z-axis direction of the unit cell structure to eliminate interlayer interactions, thus establishing an initial environmental medium surface model. Then, based on the size of the pollutant molecules under study, the established initial environmental medium surface model is expanded, i.e., expanded by x1 and y1 times in the x-axis and y-axis directions of the unit cell structure to obtain a supercell surface model suitable for the pollutant molecule size. The unit cell size of this supercell surface structure is denoted as l. a *l b *l cFinally, the supercell surface model was optimized to obtain the sample environment medium surface model, and the energy of the sample environment medium surface model was calculated, denoted as E. sur .
[0105] According to embodiments of this disclosure, the method for establishing a molecular structure model of a sample pollutant includes: placing the sample pollutant within a unit cell of the same size as the surface model of the sample environmental medium, and performing configuration optimization to obtain a molecular structure model of the sample pollutant.
[0106] According to embodiments of this disclosure, the method for establishing a molecular structure model of a sample pollutant further includes: calculating the energy of the molecular structure model of the sample pollutant, denoted as E. mol .
[0107] According to embodiments of this disclosure, establishing the initial reactant structure model includes: based on various adsorption modes of sample pollutant molecules on the sample environmental medium surface, establishing all complex models obtained by combining the sample pollutant molecule structure model with the sample environmental medium surface model; calculating the energy of each complex model, denoted as E. com The target complex model is determined based on the energy of the complex model, and the target complex model is used as the starting reactant structure model.
[0108] According to embodiments of this disclosure, the target complex model can be, for example, the most stable structure among all complex models. Specifically, the target complex model can be the most stable structure among all complex structures. com The lowest level of structure.
[0109] According to embodiments of this disclosure, based on the various adsorption modes of sample pollutant molecules on the surface of the sample environmental medium, all complex models obtained by combining the sample pollutant molecule structure model and the sample environmental medium surface model include: establishing all initial adsorption complex models by combining the sample pollutant molecule structure model and the sample environmental medium surface model based on the various adsorption modes of sample pollutant molecules on the surface of the sample environmental medium; and optimizing the configuration of each initial adsorption complex model to obtain a complex model corresponding to each initial adsorption complex model.
[0110] According to embodiments of this disclosure, the CASTEP module of Materials Studio simulation software can be used to optimize the configuration of the initial adsorption complex model, obtaining the complex corresponding to each initial adsorption complex model, and calculating the energy of each complex structure, denoted as E. com .
[0111] Materials Studio is a materials computational simulation software developed specifically for researchers in the field of materials science, running on a PC platform. The software employs a flexible client-server architecture, supports multiple operating platforms, and helps users more easily build three-dimensional structural models and conduct in-depth research on the properties and related processes of various crystalline, amorphous, and polymeric materials.
[0112] CASTEP (short for Cambridge Sequential Total Energy Package) is an ab initio quantum mechanics program based on density functional methods. Typical applications include research on surface chemistry, bond structure, density of states, and optical properties. CASTEP can also be used to study the charge density and 3D form of the wave function of a system.
[0113] According to embodiments of this disclosure, when performing configuration optimization, the generalized gradient approximation method is used to handle the exchange correlation energy; the cutoff energy is set to a, and the k-point is set to b; the interaction between the nucleus and valence electrons is handled using an ultrasoft pseudopotential; long-range and dispersion corrections are performed using the Grimme method; the convergence parameters for geometry optimization are set as follows: the interatomic interaction force is no greater than c; the maximum atomic displacement is no greater than d; and the convergence threshold for the change in the total energy of the system is e.
[0114] In one embodiment, the generalized gradient approximation method is used to handle the exchange correlation energy; the cutoff energy is set to 400 eV, and the k-point is 2×2×1; the interaction between the nucleus and valence electrons is handled using an ultrasoft pseudopotential; long-range and dispersion corrections are performed using the Grimme method; the convergence parameters for geometry optimization are set to an interatomic interaction force of no more than 0.05 eV / atom; and the maximum atomic displacement is no greater than... The convergence threshold for the change in the total energy of the system is 2.0 × 10⁻⁶. -6 eV / atom.
[0115] The activation energy includes the difference between the energy of the transition state structure and the energy of the initial reactant structure.
[0116] The formulas for calculating activation energy include:
[0117] E barrier =E TS –E com
[0118] in,
[0119] E barrier Activation energy;
[0120] E TS The energy of the transition state structure;
[0121] E comThe energy represents the structure of the complex.
[0122] According to embodiments of this disclosure, the preset method may include, for example, a linear synchronization transition / secondary synchronization transition method (LST / QST).
[0123] The specific embodiments described above further illustrate the purpose, technical solutions, and beneficial effects of this disclosure. It should be understood that the above descriptions are merely specific embodiments of this disclosure and are not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.
Claims
1. A method for predicting the environmental transformation performance of pollutants, comprising: The method involves obtaining a first target reactivity parameter for the target pollutant and a second target reactivity parameter for the target environmental medium. The first target reactivity parameter includes at least one of the following: a first ionization potential, a first affinity potential, molecular hardness, and molecular softness. The second target reactivity parameter includes: redox potential. The first target reactivity parameter and the second target reactivity parameter are input into a pre-trained activation energy prediction model to output the target activation energy, wherein the target activation energy characterizes the activation energy of the transformation of the target pollutant on the surface of the target environmental medium; and Predict the conversion performance of the target pollutant in the target environmental medium based on the target reaction activation energy; The training method for the activation energy prediction model includes: Obtain a training sample dataset, wherein the training sample dataset includes at least one set of sample data groups, and each set of sample data groups includes sample pollutant reactivity parameters, sample environmental medium reactivity parameters, and sample reaction activation energy of sample pollutant transformation on the surface of sample environmental medium; The sample pollutant reactivity parameters and sample environmental medium reactivity parameters from the training sample dataset are input into the initial prediction model to obtain the predicted reaction activation energy; The loss value is determined based on the predicted activation energy of the reaction and the activation energy of the sample reaction; and The network parameters of the initial prediction model are adjusted using the loss value until the preset iteration conditions are met, thus obtaining the activation energy prediction model.
2. The method according to claim 1, wherein, The method for determining the activation energy of the sample reaction includes: Establish a structural model of the starting reactant, wherein the structural model of the starting reactant is formed by combining the molecular structure model of the sample pollutant with the surface model of the sample environmental medium; Establish a structural model of the reaction product, wherein the structural model of the reaction product is the structural model formed after the reaction between the molecular structural model of the sample pollutant and the surface model of the sample environmental medium; Based on the structure models of the starting reactants and the reaction products, a preset method is used to search for reaction stationary points, resulting in multiple reaction pathways. The activation energy of the sample reaction, which involves the transformation of the sample pollutant on the surface of the sample environmental medium, is determined based on the energy of the transition state structure corresponding to each reaction path and the energy of the initial reactant structure model.
3. The method according to claim 2, wherein, The establishment of the structural model of the starting reactant includes: Based on the various adsorption modes of the sample pollutant molecules on the surface of the sample environmental medium, establish all complex models obtained by combining the sample pollutant molecule structure model with the sample environmental medium surface model; Calculate the energy of each of the aforementioned complex models; The target complex model is determined based on the energy of the complex model, and the target complex model is used as the structural model of the starting reactant.
4. The method according to claim 3, wherein, Based on the various adsorption modes of the sample pollutant molecules on the surface of the sample environmental medium, all complex models obtained by combining the sample pollutant molecule structure model with the sample environmental medium surface model include: Based on the various adsorption modes of the sample pollutant molecules on the surface of the sample environmental medium, establish all initial adsorption complex models obtained by combining the sample pollutant molecule structure model with the sample environmental medium surface model; Configuration optimization is performed on each of the initial adsorption complex models to obtain the complex model corresponding to each of the initial adsorption complex models.
5. The method according to claim 2, wherein, The method for establishing the surface model of the sample environment medium includes: The cell structures of the sample environment medium components are selected from the inorganic crystal structure database, and the surface model of the sample environment medium is established.
6. The method according to claim 2, wherein, The methods for establishing the molecular structure model of the sample pollutants include: The sample pollutant was placed within a unit cell of the same size as the surface model of the sample environmental medium, and its configuration was optimized to obtain the molecular structure model of the sample pollutant.
7. The method according to claim 1, wherein, The target environmental media include soil, sediment, or particulate matter.
8. The method according to claim 1, wherein, The method for determining the first ionization potential includes: determining the difference between the single-point energy of the target pollutant cationic fragment and the single-point energy of the target pollutant to obtain the first ionization potential, wherein the target pollutant cationic fragment includes the structure obtained by the target pollutant after losing one electron; The method for determining the first affinity potential includes: determining the difference between the single-point energy of the target pollutant and the single-point energy of the target pollutant anionic fragment, and obtaining the first affinity potential, wherein the target pollutant anionic fragment includes the structure obtained after the target pollutant gains an electron; The molecular hardness and the molecular softness are determined based on the first ionization potential and the first affinity potential.
9. The method according to claim 1, wherein: The first target reactivity parameter and the second target reactivity parameter were determined by searching literature databases; and / or The first target reactivity parameter and the second target reactivity parameter are determined by density functional theory calculation.