A method and system for detecting lattice thermal conductivity of semiconductor materials

By obtaining and analyzing the physical parameters of semiconductor materials, and combining with the graph neural network to predict Green Eisen parameters, a new lattice thermal conductivity formula is obtained, which solves the complexity and cost of traditional detection methods, and achieves fast and accurate Lattice thermal conductivity detection.

CN117849097BActive Publication Date: 2024-11-15XI AN JIAOTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410019181.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-01-05
Publication Date
2024-11-15
Estimated Expiration
2044-01-05

AI Technical Summary

Technical Problem

Traditional semiconductor material lattice thermal conductivity detection methods have the problem of complex theoretical calculations and relying on advanced instruments and high purity samples, with long experimental measurement cycles and high cost.

Method used

By obtaining multiple physical parameters related to semiconductor materials, performing correlation analysis, introducing the relationship between the material Vickers hardness and shear modulus, a new lattice thermal conductivity formula is obtained, and a graph neural network is used to predict the Green Eisen parameters , calculate the thermal conductivity of the lattice.

Benefits of technology

A new lattice thermal conductivity formula with concise, simple parameters and physical dimension alignment is realized, which reduces experimental costs, improves the universality and speed of detection, and avoids dependence on advanced instruments and high-purity samples.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117849097B_ABST
    Figure CN117849097B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for detecting lattice thermal conductivity of semiconductor materials, which relates to the field of thermal conductivity detection technology, and includes the following steps: obtaining multiple first physical parameters related to the lattice thermal conductivity of semiconductor materials and performing correlation analysis to obtain multiple second physical parameters; based on the second physical parameters, analyzing the traditional Slack model, and introducing the relationship between the material Vickers hardness and the shear modulus, to obtain a new lattice thermal conductivity formula; predicting the Greeneson parameters of semiconductor materials through graph neural networks to obtain predicted Greeneson parameters; substituting the predicted Greeneson parameters of semiconductor materials and the other related second physical parameters into the new lattice thermal conductivity formula to obtain lattice thermal conductivity. The present invention obtains some physical quantities as parameters of lattice thermal conductivity through correlation analysis, re-expresses lattice thermal conductivity in a more accurate and compact form, and finally obtains a new formula with simple parameters and aligned physical dimensions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of thermal conductivity detection, and in particular to a method and system for detecting lattice thermal conductivity of semiconductor materials. Background Art

[0002] Materials with ultra-high or ultra-low thermal conductivity have great potential in many technological applications. On the one hand, heat dissipation and on-chip cooling of devices are the main challenges for the future development of microelectronics, requiring high thermal conductivity materials. On the other hand, thermal insulators and the realization of direct thermoelectric conversion require lower thermal conductivity materials, where the thermoelectric conversion efficiency is determined by the thermoelectric figure of merit (ZT), defined as ZT = S 2 σ / (κ e +κ L )(where σ, S, T, κ e and κ L Represents electrical conductivity, Seebeck coefficient, absolute temperature, electronic thermal conductivity and lattice thermal conductivity respectively). Due to the trade-offs among carrier concentration, effective mass, σ and S and other electronic transport properties, reducing lattice thermal conductivity is an effective way to obtain high ZT materials and optimal thermal insulation materials.

[0003] For the detection of lattice thermal conductivity of semiconductor materials, the traditional trial and error method is mainly divided into two categories, theoretical calculation and experimental measurement. Among them, in terms of theoretical calculation: (1) Slack and some empirical formulas can be used to directly obtain lattice thermal conductivity by substituting parameters. (2) Phonon Boltzmann transport equation is used to calculate lattice thermal conductivity through various software packages such as ShengBTE. (3) Molecular dynamics calculations are performed using Green-Cooper or Fourier formulas. (4) And scattering matrix methods, etc. In terms of experimental measurement: (1) Micro-fabricated suspended thermal bridges, (2) Micro-Raman spectroscopy, (3) 3ω method, (4) Femtosecond laser time-domain thermoreflectance measurement technology (TDTR).

[0004] Traditional theoretical calculations are usually limited by complex formulas and input parameters that are difficult to obtain directly. Experimental measurements are usually highly dependent on advanced instruments and sample purity, and the experimental measurement cycle is long and the cost is high. Summary of the invention

[0005] The present invention provides a method and system for detecting the lattice thermal conductivity of semiconductor materials, which solves the problem that the existing traditional theoretical calculations are usually limited by complex formulas and input parameters that are difficult to obtain directly. Experimental measurements are usually highly dependent on advanced instruments and sample purity, and the experimental measurement cycle is long and the cost is high.

[0006] The present invention provides a method for detecting lattice thermal conductivity of semiconductor materials, comprising the following steps:

[0007] Acquiring a plurality of first physical parameters related to lattice thermal conductivity of semiconductor materials;

[0008] Performing lattice thermal conductivity correlation analysis on the plurality of first physical parameters to obtain a plurality of second physical parameters whose correlations are greater than set values;

[0009] Based on multiple second physical parameters, the traditional Slack model is analyzed, and the relationship between the Vickers hardness and shear modulus of the material is introduced to obtain a new lattice thermal conductivity formula;

[0010] Based on the new lattice thermal conductivity formula, the Greenesen parameters of semiconductor materials are predicted through graph neural network to obtain the predicted Greenesen parameters;

[0011] The predicted Grüneisen parameters of the semiconductor material and other related second physical parameters are substituted into the new lattice thermal conductivity formula to obtain the lattice thermal conductivity of the semiconductor material.

[0012] Preferably, a plurality of first physical parameters related to lattice thermal conductivity are obtained in the AFLOW database by Python code.

[0013] Preferably, the plurality of first physical parameters include thermal conductivity, electronic energy band, speed of sound, Poisson's ratio, bulk elastic modulus, shear modulus, Young's modulus, Debye temperature, Greeneisen parameter, isovolumetric heat capacity, lattice constant, atomic radius of different elements, atomic weight, electron affinity, ground state energy of each atom, number of valence electrons, electronegativity and atomic number.

[0014] Preferably, thermal conductivity correlation analysis is performed on the plurality of first physical parameters by Pearson correlation analysis, and the set value is 70%.

[0015] Preferably, the plurality of the second physical parameters include shear modulus, Young's modulus, isovolumetric heat capacity, Debye temperature, Grüneitz parameter, sonic velocity and bulk elastic modulus.

[0016] Preferably, the traditional Slack model is analyzed, and the relationship between the Vickers hardness and the shear modulus of the material is introduced to obtain a new lattice thermal conductivity formula, including the following steps:

[0017] The traditional Slack model is transformed to obtain the first positive correlation expression of physical dimension alignment;

[0018] The Debye temperature expression of the acoustic branch is combined with the first positive correlation expression to obtain the second positive correlation expression;

[0019] According to the positive correlation between lattice thermal conductivity and hardness, the second positive correlation expression is transformed into a third positive correlation expression;

[0020] According to the relationship between hardness and shear modulus, the third positive correlation expression is transformed into a fourth positive correlation expression;

[0021] According to the definition of Debye temperature and Grüneisen constant, the fourth positive correlation expression is transformed into a new lattice thermal conductivity formula.

[0022] Preferably, the first positive correlation expression is as follows:

[0023]

[0024] In the formula, κ L is the lattice thermal conductivity, M is the atomic mass, θ D,a is the Debye temperature of the acoustic branch, V is the unit cell volume, N is the number of atoms in the unit cell, and T is the temperature;

[0025] The second positive correlation expression is as follows:

[0026]

[0027] Where ρ is the density, v m is the speed of sound;

[0028] The third positive correlation expression is as follows:

[0029]

[0030] In the formula, H v is hardness;

[0031] The fourth positive correlation expression is as follows:

[0032]

[0033] Where G is the shear modulus, γ is the Greenesen constant, and B(γ) is the dimensionless term parameter related to the Greenesen parameter;

[0034] The new thermal conductivity formula is shown as follows:

[0035]

[0036] In the formula, κ L is the lattice thermal conductivity, G is the shear modulus, v m is the speed of sound, V is the volume of the primitive cell, N is the number of atoms in the primitive cell, T is the temperature, and γ is the Greeneson constant.

[0037] Preferably, the novel thermal conductivity formula is based on the effect of acoustic phonons on lattice thermal conductivity and the dominant effect of optical phonons on lattice thermal conductivity, and the novel thermal conductivity formula is expanded as follows:

[0038]

[0039] Where δ is the target parameter, which is between 1 and 2.

[0040] Preferably, the Greenesen parameters of semiconductor materials obtained in the AFLOW database are predicted by a graph neural network to obtain predicted Greenesen parameters; the error between the predicted Greenesen parameters and the database data is less than 50%.

[0041] A lattice thermal conductivity detection system, comprising:

[0042] A parameter acquisition module, used to acquire a plurality of first physical parameters related to the lattice thermal conductivity of the semiconductor material;

[0043] A correlation analysis module, used for performing lattice thermal conductivity correlation analysis on a plurality of first physical parameters to obtain a plurality of second physical parameters whose correlation is greater than a set value;

[0044] The formula derivation module is used to analyze the traditional Slack model based on multiple second physical parameters, and introduce the relationship between the Vickers hardness and shear modulus of the material to obtain a new lattice thermal conductivity formula;

[0045] The parameter prediction module is used to predict the Greenesen parameters of semiconductor materials through graph neural networks based on the new lattice thermal conductivity formula to obtain the predicted Greenesen parameters;

[0046] The thermal conductivity detection module is used to substitute the predicted Greenesen parameters and other related second physical parameters of the semiconductor material into the new lattice thermal conductivity formula to obtain the lattice thermal conductivity of the semiconductor material.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention first obtains a plurality of first physical parameters related to the lattice thermal conductivity of semiconductor materials, and then obtains some key physical quantities, namely a plurality of second physical parameters, as parameters of lattice thermal conductivity through correlation analysis. Through reasoning of the physical knowledge of these parameters, the lattice thermal conductivity is re-expressed in a more accurate and compact form, and finally a new lattice thermal conductivity formula with simple parameters and aligned physical dimensions is obtained. Since the new lattice thermal conductivity formula includes the Greenesen parameter, considering the shortcomings of the existing Greenesen parameter, the parameter is predicted by a graph convolutional neural network. The final formula is universal and a simple empirical formula, does not require experiments, has low cost, strong promotion ability, and clear physical meaning. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 It is a flow chart of a method for detecting lattice thermal conductivity of semiconductor materials of the present invention;

[0051] Figure 2 Schematic diagram of the Pearson correlation coefficient analysis process of an embodiment of the present invention;

[0052] Figure 3 This is a schematic diagram of quantitative analysis results of parameters with high correlation according to an embodiment of the present invention;

[0053] Figure 4 This is a schematic diagram of the result of predicting Greeneson parameters using a graph convolutional neural network in an embodiment of the present invention;

[0054] Figure 5 It is a schematic diagram comparing the predicted result of the lattice thermal conductivity of the embodiment of the present invention, the lattice thermal conductivity of the AFLOW database, and the experimentally measured lattice thermal conductivity. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0056] The present invention provides a method for detecting the lattice thermal conductivity of semiconductor materials. Some key physical quantities are detected in the AFLOW database as parameters of the lattice thermal conductivity through Pearson correlation analysis. Then, these parameters are linked to elastic mechanics, phonon-phonon scattering, phonon relaxation time approximation and domain knowledge, and the lattice thermal conductivity is re-expressed in a more accurate and compact form. Finally, a new formula with simple parameters and aligned physical dimensions is obtained. Taking into account the shortcomings of the existing Greeneson parameters (important parameters in the formula), the parameters are predicted by a graph convolutional neural network. The final formula has universality and a simple empirical formula, strong generalization ability, and clear physical meaning. Specifically, it includes the following steps:

[0057] The first step is to obtain a plurality of first physical parameters related to the lattice thermal conductivity of the semiconductor material.

[0058] The AFLOW database includes a large number of semiconductor materials, including a variety of binary inorganic materials and ternary materials, such as NaCl, KCl, AgCl, B2C2Mg, CoNb4Si, and LiPZn. A large number of first physical parameters related to lattice thermal conductivity in the AFLOW database are downloaded through self-written Python code. The number of first physical parameters is 151, as shown below: Thermal conductivity κ L , electron energy band gap, speed of sound v m , Poisson's ratio μ, bulk elastic modulus B, shear modulus G, Young's modulus E, Debye temperature θ D,a , Greeneson parameter γ, isovolumetric heat capacity C V , lattice constant a, atomic radius r of different elements a and r b , atomic weight m a and m b , electron affinity E ea and E eb , the ground state energy E of each atom a0 and E b0 , valence electron number VE a and VE b , electronegativity χ a and χ b and atomic number Z a and Z b .

[0059] Step 2: Perform lattice thermal conductivity correlation analysis on the multiple first physical parameters to obtain multiple second physical parameters whose correlation is greater than the set value.

[0060] Reference Figure 2 , multiple second physical parameters with linear correlation greater than 70% with lattice thermal conductivity were analyzed by Pearson correlation coefficient. Figure 3 Finally, the second physical parameters with the greatest correlation with lattice thermal conductivity are shear modulus G, Young's modulus E, and isovolumetric heat capacity C. V , Debye temperature θ D,a , Greeneson parameter γ and speed of sound v m and bulk elastic modulus B.

[0061] Step 3: Based on multiple second physical parameters, the traditional Slack model is analyzed, and the relationship between the Vickers hardness and shear modulus of the material is introduced to obtain a new lattice thermal conductivity formula.

[0062] Based on the Slack model, a series of analyses and approximations based on physical knowledge are carried out. The detailed analysis is as follows:

[0063] The original Slack model is of the form:

[0064]

[0065] Where M is the atomic mass, V is the volume of the unit cell, and θ D,a is the Debye temperature of the acoustic branch, γ is the Grüneisen constant, N is the number of atoms in the unit cell, and T is the absolute temperature. A and γ are dimensionless parameters to be determined, and then the lattice thermal conductivity formula is written as a first positive correlation expression aligned with physical dimensions (2):

[0066]

[0067] Combined with the definition of Debye temperature and Finally, the Debye temperature of the acoustic branch is:

[0068]

[0069] Where h is Planck's constant, v m is the speed of sound, k B is the Boltzmann constant.

[0070] Combining formula (2) and formula (3), the lattice thermal conductivity formula can be expressed as (4):

[0071]

[0072] After removing the constant term, we get formula (5), which is the second positive correlation expression:

[0073]

[0074] Where ρ is the density, since the hardness H v and ρv 2 The physical dimensions are similar to the physical image, and the lattice thermal conductivity and hardness H v There is a positive correlation, so ρv in formula (5) 2 You can use H v By making approximate substitution, we get the third positive correlation expression:

[0075]

[0076] According to previous studies by Teter et al., the hardness H v Strictly proportional to the shear modulus G, therefore, the hardness H v Can be made by H v =0.151G is accurately estimated, and then formula (7) is obtained, that is, the fourth positive correlation expression:

[0077]

[0078] Among them, B(γ) is a dimensionless parameter related to the Greeneisen parameter, and then according to previous studies, the relaxation time τ and It is a positive correlation, as shown in formula (8):

[0079]

[0080] Formula (8) shows that B(γ) is in the form of a negative exponential of e. The relaxation time of phonon scattering (a parameter proportional to the lattice thermal conductivity) can be written as form.

[0081] According to the Debye temperature θ D The definition of θ D =ω max / k B and the definition of the Greenesen constant γ γ = -dlnω / dlnV, we find that θ D and γ are negatively correlated, so we can approximately get B(γ)=e -γ .

[0082] Finally, we obtain an explicit formula (9) that is aligned with physical dimensions and can accurately and quickly detect lattice thermal conductivity:

[0083]

[0084] where v m It can be replaced by the mechanical modulus of the material, according to previous research:

[0085]

[0086]

[0087]

[0088] Among them, v t is the shear wave velocity, v l is the longitudinal wave velocity.

[0089] Therefore, the target parameter is the bulk elastic modulus B (in order to obtain the sound velocity v m ), shear modulus (G), Greeneisen constant (γ), number of atoms in the unit cell (N), unit cell volume (V), temperature (T).

[0090] Furthermore, formula (9) is expanded. Since formula (9) only considers the effect of acoustic phonons on lattice thermal conductivity, which is dominant in most materials, but optical phonons in a few materials play a dominant role in lattice thermal conductivity, such as BaO materials, the formula can be further expanded to (10):

[0091]

[0092] Among them, the undetermined parameter δ is between 1 and 2, but due to different material systems, the contribution of the four-phonon scattering process is different, so the above formula cannot give an accurate value. When δ = 1, the heat transport in the material has only three-phonon scattering processes, and when 1 < δ < 2, the heat transport in the material has both three-phonon and four-phonon scattering processes. Therefore, in the present invention, only the test results of formula (9) are shown.

[0093] Step 4: Based on the new lattice thermal conductivity formula, predict the Greenesen parameters through graph neural network.

[0094] Reference Figure 4 Due to the large number of materials, there are few reference data for the existing Greenisen parameters. Therefore, the present invention predicts the data obtained from the AFLOW database through a graph convolutional neural network. With less input data, machine learning is performed to obtain a large number of Greenisen parameters that can be substituted into the formula to predict the lattice thermal conductivity. Figure 4 The horizontal axis is the Greenessen parameter in the AFLOW database, and the vertical axis is the result of the graph convolutional neural network on the parameter prediction. The shaded area is the error between the database data and the prediction result within 50%. MAE is the mean absolute error value of 0.043, R 2 The determination coefficient is 0.869. The black dotted line represents the functional relationship of y=x, which indicates the closeness between the predicted results and the database results.

[0095] Step 5: Based on the formula (9) for rapid detection of lattice thermal conductivity, a large number of parameters related to the formula are extracted from the AFLOW and MaterialProjects databases, and finally directly substituted into formula (9) to rapidly detect the required lattice thermal conductivity. The test results are as follows: Figure 5 As shown. The horizontal axis represents the predicted value of the new lattice thermal conductivity formula, the vertical axis represents the thermal conductivity result of the AFLOW database, and the hollow points are the comparison between the predicted value and the database value. The solid points of different forms represent the experimental results of different symmetries, and the shadows represent that the predicted results are within the 100% error range. The illustration in the upper left corner divides the predicted results into different crystal symmetry group numbers, and it is found that the predicted results are applicable to different crystal symmetries and symmetry group numbers.

[0096] Based on the same inventive concept, the present invention also provides a semiconductor material lattice thermal conductivity detection system, including a parameter acquisition module, a correlation analysis module, a formula derivation module, a parameter prediction module and a thermal conductivity detection module. The parameter acquisition module is used to obtain multiple first physical parameters related to the lattice thermal conductivity of the semiconductor material. The correlation analysis module is used to perform lattice thermal conductivity correlation analysis on multiple first physical parameters to obtain multiple second physical parameters whose correlation is greater than a set value. The formula derivation module is used to analyze the traditional Slack model based on multiple second physical parameters, and introduce the relationship between the material Vickers hardness and shear modulus to obtain a new lattice thermal conductivity formula. The parameter prediction module is used to predict the Greeneson parameters of the semiconductor material through a graph neural network based on the new lattice thermal conductivity formula to obtain the predicted Greeneson parameters. The thermal conductivity detection module is used to substitute the predicted Greeneson parameters of the semiconductor material and the other related second physical parameters into the new lattice thermal conductivity formula to obtain the lattice thermal conductivity of the semiconductor material.

[0097] The final formula was substituted into the data of AFLOW and Materials Project, and finally the lattice thermal conductivity of 4601 binary inorganic materials and 6995 ternary materials in MP were detected, and the lattice thermal conductivity of 4363 materials was obtained from AFLOW. It can speed up the material discovery process targeting lattice thermal conductivity and reduce the cost of material design and screening in the fields of thermal management, on-chip cooling and thermoelectrics.

[0098] The present invention combines machine learning models with reasoning about physical knowledge to derive an accurate and rapid method for detecting lattice thermal conductivity. This method solves the limitations of theoretical calculations of lattice thermal conductivity, which are insufficient in precision, time-consuming, and difficult to accurately calculate high-order force constants; solves the dependence of experiments on advanced instruments and sample quality, and solves the problem of long experimental cycles; solves the problem that previous machine learning models cannot give clear physical formula explanations or the formulas are complex, the physical parameter dimensions are confusing, and it is difficult to make rapid predictions.

[0099] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0100] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for detecting lattice thermal conductivity of semiconductor materials, It is characterized in that The following steps are involved: Acquiring a plurality of first physical parameters related to lattice thermal conductivity of semiconductor materials; Performing lattice thermal conductivity correlation analysis on the plurality of first physical parameters to obtain a plurality of second physical parameters whose correlations are greater than set values; Based on multiple second physical parameters, the traditional Slack model is analyzed, and the relationship between the Vickers hardness and shear modulus of the material is introduced to obtain a new lattice thermal conductivity formula; Based on the new lattice thermal conductivity formula, the Greenesen parameters of semiconductor materials are predicted through graph neural network to obtain the predicted Greenesen parameters; Substituting the predicted Grüneisen parameter and other relevant second physical parameters of the semiconductor material into the new lattice thermal conductivity formula, the lattice thermal conductivity of the semiconductor material is obtained; The traditional Slack model is analyzed, and the relationship between the Vickers hardness and the shear modulus of the material is introduced to obtain a new lattice thermal conductivity formula, which includes the following steps: The traditional Slack model is transformed to obtain the first positive correlation expression of physical dimension alignment; The Debye temperature expression of the acoustic branch is combined with the first positive correlation expression to obtain the second positive correlation expression; According to the positive correlation between lattice thermal conductivity and hardness, the second positive correlation expression is transformed into a third positive correlation expression; According to the relationship between hardness and shear modulus, the third positive correlation expression is transformed into a fourth positive correlation expression; According to the definition of Debye temperature and Greeneisen constant, the fourth positive correlation expression is transformed into a new lattice thermal conductivity formula; The first positive correlation expression is as follows: In the formula, κ L is the lattice thermal conductivity, M is the atomic mass, θ D,a is the Debye temperature of the acoustic branch, V is the unit cell volume, N is the number of atoms in the unit cell, and T is the temperature; The second positive correlation expression is as follows: Where ρ is the density, v m is the speed of sound; The third positive correlation expression is as follows: In the formula, H v is hardness; The fourth positive correlation expression is as follows: Where G is the shear modulus, γ is the Greenesen constant, and B(γ) is the dimensionless term parameter related to the Greenesen parameter; The novel lattice thermal conductivity formula is shown below: In the formula, κ L is the lattice thermal conductivity, G is the shear modulus, v m is the speed of sound, V is the volume of the primitive cell, N is the number of atoms in the primitive cell, T is the temperature, and γ is the Greeneson constant.

2. A method for detecting lattice thermal conductivity of semiconductor materials according to claim 1, It is characterized in that A number of first physical parameters related to lattice thermal conductivity were obtained in the AFLOW database using Python code.

3. A method for detecting lattice thermal conductivity of semiconductor materials as claimed in claim 2, It is characterized in that The multiple first physical parameters include thermal conductivity, electronic energy band, speed of sound, Poisson's ratio, bulk elastic modulus, shear modulus, Young's modulus, Debye temperature, Greeneisen parameter, isovolumetric heat capacity, lattice constant, atomic radius of different elements, atomic weight, electron affinity, ground state energy of each atom, number of valence electrons, electronegativity and atomic number.

4. A method for detecting lattice thermal conductivity of semiconductor materials as claimed in claim 3, It is characterized in that Thermal conductivity correlation analysis is performed on the plurality of first physical parameters by Pearson correlation analysis, and the set value is 70%.

5. A method for detecting lattice thermal conductivity of semiconductor materials as claimed in claim 4, It is characterized in that The plurality of the second physical parameters include shear modulus, Young's modulus, isovolumetric heat capacity, Debye temperature, Grüneisen parameter, speed of sound and bulk elastic modulus.

6. A method for detecting lattice thermal conductivity of semiconductor materials according to claim 1, It is characterized in that The new lattice thermal conductivity formula is based on the effect of acoustic phonons on lattice thermal conductivity and the dominant effect of optical phonons on lattice thermal conductivity. The new lattice thermal conductivity formula is expanded as follows: Where δ is the target parameter, which is between 1 and 2.

7. A method for detecting lattice thermal conductivity of semiconductor materials according to claim 1, It is characterized in that The Greenesen parameters of semiconductor materials obtained in the AFLOW database are predicted by graph neural network to obtain predicted Greenesen parameters; the error between the predicted Greenesen parameters and the database data is less than 50%.

8. A semiconductor material lattice thermal conductivity detection system, It is characterized in that include: A parameter acquisition module, used to acquire a plurality of first physical parameters related to the lattice thermal conductivity of the semiconductor material; A correlation analysis module, used for performing lattice thermal conductivity correlation analysis on a plurality of first physical parameters to obtain a plurality of second physical parameters whose correlation is greater than a set value; The formula derivation module is used to analyze the traditional Slack model based on multiple second physical parameters, and introduce the relationship between the Vickers hardness and shear modulus of the material to obtain a new lattice thermal conductivity formula; The parameter prediction module is used to predict the Greenesen parameters of semiconductor materials through graph neural networks based on the new lattice thermal conductivity formula to obtain the predicted Greenesen parameters; A thermal conductivity detection module, which is used to substitute the predicted Grüneisen parameter of the semiconductor material and other related second physical parameters into the new lattice thermal conductivity formula to obtain the lattice thermal conductivity of the semiconductor material; The traditional Slack model is analyzed, and the relationship between the Vickers hardness and the shear modulus of the material is introduced to obtain a new lattice thermal conductivity formula, which includes the following steps: The traditional Slack model is transformed to obtain the first positive correlation expression of physical dimension alignment; The Debye temperature expression of the acoustic branch is combined with the first positive correlation expression to obtain the second positive correlation expression; According to the positive correlation between lattice thermal conductivity and hardness, the second positive correlation expression is transformed into a third positive correlation expression; According to the relationship between hardness and shear modulus, the third positive correlation expression is transformed into a fourth positive correlation expression; According to the definition of Debye temperature and Greeneisen constant, the fourth positive correlation expression is transformed into a new lattice thermal conductivity formula; The first positive correlation expression is as follows: In the formula, κ L is the lattice thermal conductivity, M is the atomic mass, θ D,a is the Debye temperature of the acoustic branch, V is the unit cell volume, N is the number of atoms in the unit cell, and T is the temperature; The second positive correlation expression is as follows: Where ρ is the density, v m is the speed of sound; The third positive correlation expression is as follows: In the formula, H v is hardness; The fourth positive correlation expression is as follows: Where G is the shear modulus, γ is the Greenesen constant, and B(γ) is the dimensionless term parameter related to the Greenesen parameter; The novel lattice thermal conductivity formula is shown below: In the formula, κ L is the lattice thermal conductivity, G is the shear modulus, v m is the speed of sound, V is the volume of the primitive cell, N is the number of atoms in the primitive cell, T is the temperature, and γ is the Greeneson constant.