X-ray fitting analysis method and X-ray fitting analysis system based on a multi-physical-variable model

Abstract

To provide an X-ray fitting analysis method and an X-ray fitting analysis system based on a multi-physical variable model.SOLUTION: The method acquires an X-ray measuring signal of a detection object by using an X-ray measuring device, and constructs a plurality of fitting models based on a target frame. An initial fitting process is performed to generate a plurality of initial fitting results and generate a plurality of initial parameter ranges. Performing a first joint fitting process to obtain a plurality of first fitting results that meet a first fitting condition; A classification process is performed based on the data format of the first fitting result to generate a plurality of classification fitting results, and the plurality of classification fitting results are counted to obtain a plurality of classification parameter ranges. A second joint fitting process is performed to obtain a plurality of second fitting results respectively corresponding to the plurality of structural parameters satisfying a second fitting condition.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to an analysis method and an analysis system, and particularly to an X-ray fitting analysis method and an X-ray fitting analysis system based on a multi-physical variable model.

Background Art

[0002] Conventional three-dimensional NAND flash memories generally have a high aspect ratio structure that is stacked vertically and interconnected. For this reason, channel hole etching is the most important process step in its technology development and manufacturing, and any abnormality will affect the subsequent processes. For example, abnormalities in the deposition of the block / trap / tunnel layer and channel formation ultimately affect the function and reliability of the device. In addition, uniformity is extremely important in process control.

[0003] In order to accurately control process parameters, the critical dimension (CD) of each word line (WL) of the product is measured. However, most conventional measurement methods, such as a transmission electron microscope (TEM) or a critical dimension scanning electron microscope (CD-SEM) that detects cross-sectional images, can only be realized by destructive methods. Also, optical critical dimension (OCD) spectroscopic measurement is the main non-destructive method for measuring CD. However, with the increase in the number of layers, many problems occur.

Summary of the Invention

Problems to be Solved by the Invention

[0004] The technical problem that this invention aims to solve is to provide an X-ray fitting analysis method based on a multi-physical-variable model in order to overcome the shortcomings of the prior art. [Means for solving the problem]

[0005] To solve the above-mentioned technical problems, one of the technical solutions employed in the present invention is to provide an X-ray fitting analysis method based on a multi-physical-variable model. The X-ray fitting analysis method involves acquiring multiple X-ray measurement signals of an object to be detected by measuring using at least one X-ray measuring device, The processing unit Perform the following steps The aforementioned Processing unit arrangement The steps include constructing a plurality of fitting models based on the target frame of the object to be detected, defining a plurality of structural parameters in the target frame, and the plurality of fitting models each Using the above-mentioned plurality of X-ray measurement signals, First The process involves executing an initial fitting process to generate multiple initial fitting results for the multiple structural parameters, statistically analyzing the multiple initial fitting results to generate multiple initial parameter ranges, and then generating multiple fitting models based on the multiple initial parameter ranges. each Using the above-mentioned plurality of X-ray measurement signals, , fitting to a predetermined range of the target frame By performing the first co-fitting process, multiple first fitting results that satisfy the first fitting conditions are obtained. To gain Based on the data format of the plurality of first fitting results, the plurality of first fitting results The plurality of first fitting results are input into the trained classification model and the plurality of first fitting results are classified by a classification tree analysis process or a regression tree analysis process. Execute the classification process By The process involves generating multiple classification fitting results, statistically analyzing the multiple classification fitting results to obtain multiple classification parameter ranges, and then formulating the multiple fitting models based on the multiple classification parameter ranges. each Using the above-mentioned plurality of X-ray measurement signals, , perform fitting analysis over the entire range of the target frame.Execute the second joint fitting process. By , satisfying the second fitting condition su, katsu Multiple second fitting results corresponding to each of the aforementioned multiple structural parameters are obtained. To gain This includes the act of doing something.

[0006] To solve the above-mentioned technical problems, another technical solution employed in the present invention is to provide an X-ray fitting analysis method based on a multi-physical-variable model. The X-ray fitting analysis method involves acquiring multiple X-ray measurement signals of an object to be detected by measuring using at least one X-ray measuring device, The processing unit Perform the following steps The aforementioned Processing unit arrangement The step includes constructing a plurality of fitting models based on the target frame of the object to be detected, defining a plurality of structural parameters in the target frame, and the plurality of fitting models each Using the above-mentioned plurality of X-ray measurement signals, First The process involves executing a phase fitting process to generate multiple initial fitting results for the multiple structural parameters, statistically analyzing the multiple initial fitting results to generate multiple initial parameter ranges, generating a first set of parameters to be validated based on the multiple initial parameter ranges, inputting the first set of parameters to be validated into the multiple fitting models to verify their accuracy, and verifying the accuracy and the multiple initial parameter ranges. The first set of parameters to be verified is Optimization conditions of fulfill vinegar up to the above 1 Adjust the parameters being verified. The group of parameters to be verified that satisfies the optimization conditions is set as the optimization parameter group.This includes setting multiple weight values ​​for the optimization parameter group, applying the multiple weight values ​​to the optimization parameter group and inputting them into the multiple fitting models to verify the weight accuracy, adjusting the multiple weight values ​​based on the weight accuracy until the weight accuracy satisfies the weight optimization conditions, and applying the multiple weight values ​​that satisfy the weight optimization conditions to the optimization parameter group to generate a weight optimization parameter group.

[0007] To solve the above-mentioned technical problems, another technical solution employed in the present invention is to provide an X-ray fitting analysis system based on a multi-physical-variable model. The X-ray fitting analysis system acquires multiple X-ray measurement signals of an object to be detected by measuring using at least one X-ray measuring device, The processing unit Perform the following steps The aforementioned Processing unit arrangement The steps include constructing a plurality of fitting models based on the target frame of the object to be detected, defining a plurality of structural parameters in the target frame, performing an initial fitting process on each of the plurality of X-ray measurement signals using the plurality of fitting models to generate a plurality of initial fitting results for the plurality of structural parameters, statistically analyzing the plurality of initial fitting results to generate a plurality of initial parameter ranges, generating a first set of parameters to be verified based on the plurality of initial parameter ranges, inputting the first set of parameters to be verified into the plurality of fitting models to verify accuracy, and based on the accuracy and the plurality of initial parameter ranges, The first set of parameters to be verified is Optimization conditions of fulfill vinegar up to the above 1 Adjust the set of parameters to be validated. The first set of parameters to be verified that satisfy the optimization conditions are set as the optimization parameter set.This includes setting multiple weight values ​​for the optimization parameter group, applying the multiple weight values ​​to the optimization parameter group and inputting them into the multiple fitting models to verify the weight accuracy, adjusting the multiple weight values ​​based on the weight accuracy until the weight accuracy satisfies the weight optimization conditions, and applying the multiple weight values ​​that satisfy the weight optimization conditions to the optimization parameter group to generate a weight optimization parameter group.

[0008] To solve the above-mentioned technical problems, another technical solution employed in the present invention is to provide an X-ray fitting analysis system based on a multi-physical-variable model. The X-ray fitting analysis system comprises at least one X-ray measuring device and a processing device. The at least one X-ray measuring device measures an object to be detected and acquires a plurality of X-ray measurement signals. The processing device performs the following steps: Placed The steps include: constructing a plurality of fitting models based on the target frame of the object to be detected, defining a plurality of structural parameters in the target frame; performing an initial fitting process on each of the plurality of X-ray measurement signals using the plurality of fitting models to generate a plurality of initial fitting results for the plurality of structural parameters; statistically analyzing the plurality of initial fitting results to generate a plurality of initial parameter ranges; generating a first set of parameters to be verified based on the plurality of initial parameter ranges; inputting the first set of parameters to be verified into the plurality of fitting models to verify accuracy; and based on the accuracy and the plurality of initial parameter ranges, The first set of parameters to be verified is Optimization conditions of fulfill vinegar up to the above 1 Adjust the set of parameters to be validated. The first set of parameters to be verified that satisfy the optimization conditions are set as the optimization parameter set.setting a plurality of weight values for the optimization parameter group, applying the plurality of weight values to the optimization parameter group, inputting them into the plurality of fitting models, and verifying the weight accuracy; adjusting the plurality of weight values based on the weight accuracy until the weight accuracy meets the weight optimization condition; and applying the plurality of weight values that meet the weight optimization condition to the optimization parameter group to generate a weight-optimized parameter group.

Advantages of the Invention

[0009] One of the beneficial effects of the present invention is that in the X-ray fitting analysis method and X-ray fitting analysis system based on the multi-physical variable model according to the present invention, first, by performing a rough first co-fitting process, the fitting results are classified to limit the range of structural parameters, and then by performing a precise second co-fitting process, when the detection object has a large number of material layers, the amount of calculation can be reduced while maintaining a certain level of fitting accuracy, thereby maintaining the stability of the fitting model.

[0010] In addition, in the X-ray fitting analysis method and X-ray fitting analysis system based on the multi-physical variable model according to the present invention, the mechanism of weight values is also applied to optimize the weight values, so the optimization speed is accelerated and the degree of variation of the fitting results can be reduced.

[0011] For a further understanding of the features and technical content of the present invention, reference should be made to the following detailed description and drawings of the present invention. However, the provided drawings are for reference and illustration purposes only and are not used to limit the present invention.

Brief Description of the Drawings

[0012] [Figure 1] It is a first schematic diagram of an X-ray fitting analysis system based on a multi-physical variable model according to a first embodiment of the present invention. [Figure 2]It is a second schematic diagram of an X-ray fitting analysis system based on a multi-physical variable model according to the first embodiment of the present invention. [Figure 3] It is a flowchart of an X-ray fitting analysis method based on a multi-physical variable model according to the first embodiment of the present invention. [Figure 4] It is a first schematic diagram showing a target frame of a detection object according to the first embodiment of the present invention. [Figure 5] It is a second schematic diagram showing a target frame of a detection object according to the first embodiment of the present invention. [Figure 6] It is a set of curve graphs of two initial fitting results according to the first embodiment of the present invention. [Figure 7] It is a set of curve graphs of two first fitting results obtained after executing the first joint fitting process according to the first embodiment of the present invention. [Figure 8] It is a classification fitting result obtained by a learned classification model according to the first embodiment of the present invention. [Figure 9] It is two second fitting results obtained after executing the second joint fitting process according to the first embodiment of the present invention. [Figure 10] It is a schematic diagram comparing the X-ray fitting analysis result according to the first embodiment of the present invention with the fitting analysis result of a single physical model. [Figure 11] It is a flowchart of an X-ray fitting analysis method based on a multi-physical variable model according to the second embodiment of the present invention. [Figure 12] It is a schematic diagram of the first iteration of step S26 according to the second embodiment of the present invention. [Figure 13] It is a schematic diagram of the second iteration of step S26 according to the second embodiment of the present invention.

Embodiments for Carrying Out the Invention

[0013] The embodiments of the "X-ray fitting analysis method and X-ray fitting analysis system based on a multi-physical-variable model" disclosed herein will be described below with reference to specific examples. Those skilled in the art will be able to understand the advantages and effects of the present invention from the disclosed content. The present invention can be carried out or applied through other different specific embodiments, and the various detailed descriptions herein can be modified and changed in various ways without departing from the spirit of the invention, based on different perspectives and applications. It should also be noted in advance that the drawings of the present invention are for illustrative purposes only and are not based on actual dimensions. The technical content relating to the present invention will be described in more detail using the following embodiments, but the disclosed content is not intended to limit the scope of protection of the present invention. Furthermore, the term "or" herein should be understood to include any one or more of the items listed in relation to the actual situation. [First Embodiment]

[0014] Figure 1 is a first schematic diagram of an X-ray fitting analysis system based on a multi-physical-variable model according to a first embodiment of the present invention. Figure 2 is a second schematic diagram of an X-ray fitting analysis system based on a multi-physical-variable model according to a first embodiment of the present invention. As shown in Figures 1 and 2, a first embodiment of the present invention provides an X-ray fitting analysis system 1 based on a multi-physical-variable model comprising a plurality of X-ray measuring devices 10 and a processing device 12.

[0015] In this embodiment, the X-ray measuring device 10 may include an X-ray generator, a group of X-ray optical elements, an X-ray detector, and a sample stage. The X-ray generator may be used to generate an X-ray beam having a predetermined energy. The group of X-ray optical elements may guide the X-ray beam and focus it onto the object SP to be detected. The X-ray detector may be used to receive a signal generated by the X-ray beam irradiating the object SP to be detected. However, the present invention is not limited thereto. The X-ray measuring device 10 may measure the object SP to be detected and acquire a plurality of X-ray measurement signals Sx. Although Figure 1 shows multiple X-ray measuring devices 10, in reality... , multiple Measure the target object SP under a number of different measurement conditions. One X-ray measuring device 10 is positioned Alternatively, as shown in Figure 2. , check Measure the object SP to be removed. Multiple different X-ray measuring devices 10 are arranged. This may also be done. In this embodiment, the X-ray measuring device 10 may be an X-ray reflectivity (XRR) measuring device, an X-ray fluorescence spectrometer (XRF) measuring device, a small-angle X-ray scattering (SAX) measuring device, an X-ray diffractometer (XRD) measuring device, or any other device capable of measuring using X-rays as a light source.

[0016] Each X-ray measurement signal Sx may include one or more of the transmission pattern, reflection pattern, diffraction pattern, and scattering pattern. For example, if the X-ray measurement signal Sx includes a reflection pattern, after the X-rays are incident on the object SP at a plurality of predetermined angles, the reflected X-rays are received using the receiver of the X-ray measurement device 10, and the X-ray measurement signal is fitted and analyzed to obtain structural parameters of the object SP, including one or more of the thickness, density, and roughness.

[0017] The processing unit 12 may be, for example, a computer system comprising a processor and memory, which executes a stored instruction set or code to acquire X-ray measurement signals and perform fitting analysis. arrangement This may be done. Furthermore, the processing device 12 may control the X-ray measuring device 10 to perform a corresponding measurement process for the object SP to be detected.

[0018] Refer to Figure 3, which is a flowchart of the X-ray fitting analysis method based on a multi-physical-variable model according to the first embodiment of the present invention. As shown in Figure 3, the X-ray fitting analysis method may include the following steps.

[0019] Step S10: Multiple X-ray measurement signals of the object to be detected are obtained by measuring using an X-ray measuring device. In this step, there may be one or more X-ray measuring devices 10.

[0020] Next, the processing unit 12 performs the following steps: arrangement It may also be used.

[0021] Step S11: Multiple fitting models are constructed based on the target frame of the object to be detected. Figures 4 and 5 are the first and second schematic diagrams showing the target frame of the object to be detected according to the first embodiment of the present invention, respectively. In step S11, as shown in Figure 4, the target frame of the object to be detected SP may be, for example, a sample in which the pattern on the carbon film has not been transferred to a multi-layer structure. Alternatively, as shown in Figure 5, the target frame of the object to be detected SP may be, for example, a sample in which the pattern on the carbon film has already been transferred to a multi-layer structure and the carbon film has been removed. Therefore, the target frame may comprise multiple material layers, and multiple structural parameters may be defined, for example, the thickness, density, or roughness of each material layer.

[0022] Step S12: An initial fitting process is performed on each of the measurement signals using those fitting models to generate multiple initial fitting results for their structural parameters.

[0023] The initial fitting process uses the processing unit 12 to execute multiple electromagnetic wave calculation engines corresponding to each of the fitting models, and based on the target frame, performs spectral fitting analysis on the corresponding X-ray measurement signal Sx to obtain the corresponding initial fitting result. The electromagnetic wave calculation engine may include, for example, one or more of the following: finite-difference time-domain (FDTD) algorithm, distorted wave Born approximation (DWBA) algorithm, rigorous coupled wave analysis (RCWA) algorithm, discrete dipole approximation (DDP) algorithm, and boundary element method (BEM).

[0024] In detail, for example, data measured after X-ray beams of different energies are incident may be fitted using the electromagnetic wave calculation engine described above to obtain the structural parameters of the target frame and the corresponding precision. Precision may include, for example, the variance and error amount of a specific structural parameter, such as density. Variance may be, for example, the value obtained by dividing the average value of the structural parameters generated after performing multiple fittings by the standard deviation. The error amount may be described using a cost function to evaluate the goodness of fit of the model to the fitting target data. The cost function may be, for example, the mean squared error (MSE) of the fitting result to the fitting target data. In this way, when initial fitting is performed multiple times with one electromagnetic wave engine for one fitting model having N layers, one error amount and N variances for one group of structural parameters (the average density / standard deviation of density for each of the N material layers) can be generated. Therefore, when there are M fitting models, and each is fitted by M electromagnetic wave engines, it is possible to generate M error quantities and N variations in M ​​sets. Refer to Figure 6, which is a set of curve graphs of two initial fitting results according to the first embodiment of the present invention. The two upper curve graphs in Figure 6 plot the X-ray measurement signal Sx and the fitted reflectance against angle. The two lower curve graphs in Figure 6 plot the obtained density (divided into real part rho and imaginary part irho) against depth, with the lower data representing the precision expressed as MSE (indicated as chisq), which are approximately 452192 and 688786 (decimal places omitted), respectively.

[0025] Step S13: These initial fitting results are statistically analyzed to generate multiple initial parameter ranges.

[0026] Therefore, after performing an initial fitting to obtain a set of multiple structural parameters, preliminary statistics may be performed to obtain the initial parameter range for each structural parameter. It should be noted that as the number of layers of the object SP increases, if there are too many floating parameters, the fitting model becomes very complex, and the computation time increases exponentially. In practice, simply modeling and fitting is not feasible for three-dimensional NAND structures. Because the channel hole depth doubles at each node transition, the sensitivity of the critical dimension of the bottom layer becomes too weak, and at the same time, adjacent word line CDs may give the same spectral response and become indistinguishable, leading to correlation problems in the modeling. To improve the stability of the model, it is possible to simplify the use of the bottom layer CD in combination with the intermediate layer CD, but this reduces the accuracy of the result for the bottom layer CD. Furthermore, if the relationship between the intermediate layer CD and the bottom layer CD changes, the accuracy will decrease even further. For this reason, in this invention, a new fitting method is adopted to perform the analysis, thereby reducing the computational load and improving accuracy.

[0027] Step S14: Based on the initial parameter range, a first co-fitting process is performed on each measurement signal using the fitting model to obtain multiple first fitting results that satisfy the first fitting conditions.

[0028] In step S14, the first co-fitting process performs a rough fitting to a predetermined range of the target frame. For example, if the target frame is a NAND containing 400 material layers, the predetermined range may be an odd number of layers among the 400 material layers. Therefore, in step S14, the first co-fitting process performs a fitting to structural parameters such as density, thickness, and roughness of the odd number of layers.

[0029] In the first joint fitting process, a first set of parameters to be verified may be generated based on the initial parameter range obtained in step S13. This first set of parameters to be verified is a structural parameter corresponding to the predetermined range (200 layers within it) mentioned above. This first set of parameters to be verified is input into the aforementioned multiple fitting models to verify the first accuracy. For example, after inputting this first set of parameters to be verified into a fitting model and fitting it using the corresponding electromagnetic wave calculation engine, one error quantity and N variation numbers of this first set of parameters to be verified are generated, and it is determined whether the error quantity and variation numbers are lower than the initial fitting results.

[0030] If the error amount and variability are not lower than those of the initial fitting results, the first set of parameters to be validated is adjusted, for example, by using random numbers and cross-combinations based on the initial parameter range, and validated again until the first fitting condition is met. The first set of parameters to be validated that satisfies the first fitting condition is taken as the first fitting result. Satisfying the first fitting condition means, for example, that the error amount and variability are lower than those of the initial fitting result, and the difference is greater than a predetermined value.

[0031] For example, see Figure 7, which is a set of curve graphs of two first fitting results obtained after performing a first co-fitting process according to a first embodiment of the present invention. The two upper curve graphs in Figure 7 plot the X-ray measurement signal Sx and the reflectance obtained after performing the first co-fitting process against angle. The two lower curve graphs in Figure 7 plot the obtained density (divided into real part rho and imaginary part irho) against depth, with the lower data representing the precision expressed as MSE (indicated as chisq), which is approximately 15374 and 349764 (decimal places omitted), respectively. Both are lower compared to the initial fitting results.

[0032] Step S15: Based on the data format of the first fitting result, perform a classification process on the first fitting result to generate multiple classification fitting results.

[0033] In this step, the classification accuracy and efficiency may be improved by combining the decision tree algorithm with machine learning. Specifically, the first fitting results generated in step S14 may be input into a trained classification model to generate classification fitting results. The trained classification model is trained to classify these first fitting results using a classification tree analysis process or a regression tree analysis process, and the number of these first fitting results is greater than the number of classification results. For example, the classification tree analysis process or regression tree analysis process may be trained to extract data features of the first fitting results, such as highs and lows, bends, peaks and valleys on a curve graph, and similarity clustering may be performed on the first fitting results, while simultaneously ranking the accuracy (including variability and cost function) to extract classifications with high accuracy. Thus, the decision tree algorithm can determine the optimal classification and reduce uncertainty in the fitting model structure. Refer to Figure 8, which shows the classification fitting results obtained by the trained classification model according to the first embodiment of the present invention. It should be noted that, while the characteristic change in structural parameters (e.g., density) with respect to depth is used as the classification criterion for the data format here, the present invention is not limited to this, and other structural parameters such as thickness and roughness may also be used as classification criteria in step S15.

[0034] Step S16: Statistically analyze these classification fitting results to obtain multiple classification parameter ranges.

[0035] Using Figure 8 as an example, after the classification in step S15, the classification fitting results show five curve types, and five corresponding parameter ranges can be obtained. In this way, the range of each structural parameter can be limited, and the computational amount required for fitting analysis can be significantly reduced. Furthermore, since the classification parameter range obtained in this step is based on the results of the first co-fitting process, its accuracy is optimized.

[0036] Step S17: Based on these classification parameter ranges, a second joint fitting process is performed on each of these measurement signals using these fitting models to satisfy the second fitting conditions. su, katsu Multiple second-fitting results corresponding to each of these structural parameters are obtained. In this step, the second co-fitting process performs fitting analysis over the entire range of the target frame. Although the number of layers to be computed is greater than in the first co-fitting process, the range of structural parameters that need to be computed is already limited, so the overall computational load can be significantly reduced while maintaining a certain level of accuracy.

[0037] Similarly, in the second joint fitting process, a second set of parameters to be validated may be generated based on the initial parameter range obtained in step S16. This second set of parameters to be validated is then input into the aforementioned fitting models to verify the second accuracy. For example, after inputting this second set of parameters to be validated into a fitting model and fitting it using the corresponding electromagnetic wave calculation engine, one error quantity and N variations in this second set of parameters to be validated are generated, and it is determined whether the error quantity and variations are lower than the first fitting results.

[0038] If the error amount and variability are not lower than those of the first fitting result, this second set of parameters to be validated is adjusted, for example, by random numbers and cross-combinations based on the classification parameter range, and validated again until the second fitting condition is met. This second set of parameters to be validated that satisfies the second fitting condition is taken as the final fitting result, the second fitting result. Satisfying the second fitting condition means, for example, that the error amount and variability are lower than those of the first fitting result, and the difference is greater than another predetermined value. Refer to Figures 9 and 10. Figure 9 shows two second fitting results obtained after performing the second co-fitting process according to the first embodiment of the present invention. Figure 10 is a schematic diagram comparing the X-ray fitting analysis result according to the first embodiment of the present invention with the fitting analysis result of a single physical model.

[0039] The two upper curve graphs in Figure 9 plot the X-ray measurement signal Sx (thick line) and the reflectance obtained after performing the second co-fitting process (thin line) against angle. The two lower curve graphs in Figure 9 plot the obtained density (divided into real part rho and imaginary part irho) against depth, with the lower data representing the precision expressed as MSE (indicated as chisq), which is approximately 6104 and 5740 (decimal places omitted), respectively. Both decrease compared to the initial fitting results and the first fitting results.

[0040] In the comparison results in Figure 10, structural parameters such as thickness, density (real part), density (imaginary part), and roughness are compared. The upper panel shows a single physical model, and the lower panel shows multiple physical models. Here, the histogram represents the mean value, and the thin lines represent the variability. As can be seen from Figure 10, the calculation of the multiple physical model results in a smaller amount of variability. [Second Example]

[0041] A second embodiment of the present invention provides another X-ray fitting analysis system based on a multi-physical-variable model. Although the details of the fitting analysis differ in this embodiment, the system structure is basically the same as the system structure shown in Figures 2 and 3, so a repeated explanation is omitted here.

[0042] Figure 11 is a flowchart of an X-ray fitting analysis method based on a multi-physical-variable model according to a second embodiment of the present invention. As shown in Figure 11, the second embodiment of the present invention provides an X-ray fitting analysis method based on a multi-physical-variable model. The X-ray fitting analysis method includes the following steps.

[0043] Step S20: Multiple X-ray measurement signals of the object to be detected are obtained by measuring using an X-ray measuring device.

[0044] Next, the processing unit performs the following steps: arrangement It will be done.

[0045] Step S21: Multiple fitting models are constructed based on the target frame of the object to be detected. As described in the first embodiment, multiple structural parameters are defined in the target frame.

[0046] Step S22: An initial fitting process is performed on each of the X-ray measurement signals using those fitting models to generate multiple initial fitting results for their structural parameters.

[0047] Step S23: These initial fitting results are statistically analyzed to generate multiple initial parameter ranges.

[0048] Step S24: Based on these initial parameter ranges, generate the first set of parameters to be validated.

[0049] Step S25: Input this first set of parameters to be validated into these fitting models to verify their accuracy. Based on this accuracy and the initial parameter ranges, adjust this set of parameters to be validated until the optimized set of parameters is obtained when the optimization conditions are met.

[0050] Here, in steps S23 to S25, first, based on the initial parameter range obtained in step S23, multiple groups of parameters to be verified may be generated, and each group of parameters to be verified may be input into the aforementioned multiple fitting models to perform accuracy verification. For example, each group of parameters to be verified may be input into a fitting model, and fitting may be performed with the corresponding electromagnetic wave calculation engine. After this, one error amount and multiple corresponding variability numbers are generated, and then it is determined whether the error amount and variability numbers satisfy the optimization conditions. For example, the optimization conditions are that the error amount is lower than a predetermined error amount and the variability numbers are lower than a predetermined variability number. If they are not lower, these groups of parameters to be verified are adjusted, for example, by adjusting with random numbers and cross-combinations based on the initial parameter range, and verification is performed again until the optimization conditions are satisfied. Multiple groups of parameters to be verified that satisfy the optimization conditions are designated as the optimization parameter group.

[0051] Step S26: Multiple weight values ​​are set for the optimization parameter group, these weight values ​​are applied to this optimization parameter group and input into these fitting models, and the weight accuracy is verified. For example, each group of parameters to be verified that satisfies the optimization conditions includes structural parameters such as the thickness, density, and roughness of each material layer of the object to be detected. Refer to Figure 12, a schematic diagram of the first iteration of step S26 according to the second embodiment of the present invention. In the first iteration calculation, each group of parameters to be verified in the optimization parameter group may be multiplied by a weight value between 0 and 1 (note that the sum of all weight values ​​is 1). For example, if there are two groups of parameters to be verified, the optimized parameter group to which the weight value has been applied is obtained by multiplying each by a weight value of 0.5.

[0052] Next, these parameters to be validated are input into these fitting models to generate multiple validation results, and the weighting accuracy is verified by determining the amount of variation in the multiple parameters of each validation result and the error values ​​for these X-ray measurement signals. As can be seen from the distribution plots of thickness space, density space, and roughness space shown in the center of Figure 12, the obtained structural parameter results are relatively varied. On the right side of Figure 12, two fitting results and two density curve graphs are shown, and the obtained weighting accuracy is shown in MSE (indicated as chisq), which is approximately 9339 and 8474 (decimal places omitted), respectively, suggesting that there is still room for optimization of the weight values. Furthermore, two new sets of parameters to be validated may be generated in the first iteration and used in the second iteration.

[0053] Step S27: Adjust these weight values ​​according to the weight accuracy until the weight accuracy satisfies the weight optimization conditions, and then apply these weight values ​​that satisfy the weight optimization conditions to the optimization parameter set to generate the weight optimization parameter set.

[0054] Refer to Figure 13, a schematic diagram of the second iteration of step S26 according to the second embodiment of the present invention. In the second iteration, the other two groups of parameters to be verified generated from the first iteration can be multiplied by weight values ​​of 0.7 and 0.3, respectively, to obtain a group of optimized parameters to which the weight values ​​have been applied.

[0055] Similarly, two new sets of parameters to be validated are input into these fitting models to generate multiple validation results, and the weighting accuracy is verified by determining the amount of variation of multiple parameters in each validation result and the error values ​​for these X-ray measurement signals. As can be seen from the distribution plots of thickness space, density space, and roughness space shown in the center of Figure 13, the obtained structural parameter results converge more than in the first iteration, indicating that weightings such as 0.7 and 0.3 may be better. On the right side of Figure 13, two fitting results and two density curve graphs are shown, and the obtained weighting accuracy is shown in MSE (indicated as chisq), which is approximately 8453 and 5505 (decimal places omitted), respectively, which appears to be more accurate than in the first iteration.

[0056] Therefore, when the weight accuracy reaches a predetermined range (for example, when the MSE is less than a specific value), it means that the weight optimization condition is met, and these weight values ​​that satisfy the weight optimization condition can be applied to the optimization parameter group to generate a weight optimization parameter group. That is, this weight optimization parameter group defines a plurality of structural parameters obtained by fitting analysis. [Beneficial effects from the examples]

[0057] One of the beneficial effects of the present invention is that, in the X-ray fitting analysis method and X-ray fitting analysis system based on the multi-physical-variable model according to the present invention, by first performing a rough first co-fitting process to classify the fitting results and limit the range of structural parameters, and then performing a precise second co-fitting process, it is possible to maintain a certain level of fitting accuracy while reducing the amount of computation when the object to be detected has many material layers, thereby maintaining the stability of the fitting model.

[0058] Furthermore, in the X-ray fitting analysis method and X-ray fitting analysis system based on the multi-physical-variable model according to the present invention, a weighting mechanism is also applied, and optimization is performed with respect to the weighting values, thereby accelerating the optimization speed and reducing the degree of variation in the fitting results.

[0059] The information disclosed herein represents only preferred embodiments of the present invention and does not limit the scope of the claims. Accordingly, all equivalent technical modifications made using the specification and drawings of the present invention are included within the scope of the claims. [Explanation of symbols]

[0060] 1: X-ray fitting analysis system 10: X-ray measuring device 12: Processing Unit SP: Detected object Sx: X-ray measurement signal S10, S11, S12, S13, S14, S15, S16, S17, S20, S21, S22, S23, S24, S25, S26, S27: Step rho: Practical Department irho: imaginary part chisq:MSE

Claims

1. An X-ray fitting analysis method based on a multi-physical-variable model, The aforementioned X-ray fitting analysis method is By using at least one X-ray measuring device, multiple X-ray measurement signals of the object to be detected are obtained, The processing device is arranged such that it performs the following steps, Includes, The aforementioned step is, Based on the target frame of the object to be detected, multiple fitting models are constructed, and multiple structural parameters are defined in the target frame. Using each of the aforementioned fitting models, an initial fitting process is performed on the aforementioned X-ray measurement signals to generate a plurality of initial fitting results for the aforementioned structural parameters. The process involves statistically analyzing the multiple initial fitting results to generate multiple initial parameter ranges, Based on the aforementioned multiple initial parameter ranges, a first co-fitting process is performed to fit the multiple X-ray measurement signals to a predetermined range of the target frame using the respective multiple fitting models, thereby obtaining multiple first fitting results that satisfy the first fitting conditions. Based on the data format of the plurality of first fitting results, a classification process is executed on the plurality of first fitting results by inputting the plurality of first fitting results into a trained classification model and classifying the plurality of first fitting results using a classification tree analysis process or a regression tree analysis process, thereby generating a plurality of classification fitting results. The process involves statistically analyzing the multiple classification fitting results to obtain multiple classification parameter ranges, Based on the aforementioned range of classification parameters, a second collaborative fitting process is performed to execute a fitting analysis on the multiple X-ray measurement signals for the entire range of the target frame using each of the multiple fitting models, thereby obtaining multiple second fitting results that satisfy the second fitting conditions and correspond to each of the multiple structural parameters. including, An X-ray fitting analysis method characterized by the following:

2. The step of obtaining multiple X-ray measurement signals of the object to be detected by measuring using the at least one X-ray measuring device is: This includes arranging one X-ray measuring device to measure the object to be detected under multiple different measurement conditions, or arranging multiple different X-ray measuring devices to measure the object to be detected. The X-ray fitting analysis method according to claim 1.

3. Each of the aforementioned X-ray measurement signals includes one or more of the transmission pattern, reflection pattern, diffraction pattern, and scattering pattern, and each of the aforementioned structural parameters includes one or more of the thickness, density, and roughness. The X-ray fitting analysis method according to claim 1.

4. The target frame comprises a plurality of material layers, and the initial fitting process includes using the processing apparatus to execute an electromagnetic wave calculation engine corresponding to each fitting model, performing spectral fitting analysis on the corresponding X-ray measurement signal based on the target frame, and obtaining the corresponding initial fitting result. The X-ray fitting analysis method according to claim 1.

5. The first joint fitting process described above is: Based on the aforementioned ranges of initial parameters, a first set of parameters to be verified is generated that corresponds to the aforementioned ranges of structural parameters within a predetermined range. The first set of parameters to be verified is input into the plurality of fitting models to verify the first accuracy, and based on the first accuracy and the plurality of initial parameter ranges, the first set of parameters to be verified is adjusted until it satisfies the first fitting condition, and the first set of parameters to be verified that satisfies the first fitting condition is taken as the plurality of first fitting results. The X-ray fitting analysis method according to claim 1.

6. The number of the plurality of first fitting results is greater than the number of the plurality of classification results, The X-ray fitting analysis method according to claim 5.

7. The second joint fitting process is as follows: Based on the aforementioned ranges of classification parameters, a second set of parameters to be verified is generated, corresponding to the aforementioned ranges of structural parameters within the entire range. The second set of parameters to be validated is input into the multiple fitting models to validate the second accuracy, and based on the second accuracy and the range of the multiple classification parameters, the second set of parameters to be validated is adjusted until it satisfies the second fitting condition, and the second set of parameters to be validated that satisfies the second fitting condition is taken as the multiple second fitting results. The X-ray fitting analysis method according to claim 1.

8. An X-ray fitting analysis method based on a multi-physical-variable model, The aforementioned X-ray fitting analysis method is By using at least one X-ray measuring device, multiple X-ray measurement signals of the object to be detected are obtained, The processing device is arranged such that it performs the following steps, Includes, The aforementioned step is, Based on the target frame of the object to be detected, multiple fitting models are constructed, and multiple structural parameters are defined in the target frame. Using each of the aforementioned fitting models, an initial fitting process is performed on the aforementioned X-ray measurement signals to generate a plurality of initial fitting results for the aforementioned structural parameters. The process involves statistically analyzing the multiple initial fitting results to generate multiple initial parameter ranges, The first set of parameters to be verified is generated based on the aforementioned multiple initial parameter ranges, The first set of parameters to be verified is input into the multiple fitting models to verify their accuracy, and based on the accuracy and the range of the multiple initial parameters, the first set of parameters to be verified is adjusted until it satisfies the optimization conditions, and the first set of parameters to be verified that satisfies the optimization conditions is designated as the optimized parameter set. Setting multiple weight values ​​for the aforementioned optimization parameter group, applying the multiple weight values ​​to the aforementioned optimization parameter group and inputting them into the aforementioned fitting models, and verifying the weight accuracy, The process involves adjusting the multiple weight values ​​based on the weight accuracy until the weight accuracy satisfies the weight optimization conditions, applying the multiple weight values ​​that satisfy the weight optimization conditions to the optimization parameter group, and generating the weight optimization parameter group. including, An X-ray fitting analysis method characterized by the following:

9. The step of inputting the first set of parameters to be verified into the multiple fitting models and verifying the accuracy is: The first set of parameters to be verified is input into the multiple fitting models to generate multiple verification results, and the weighting accuracy is verified by determining the amount of variation of the multiple parameters in each verification result and the error value with respect to the multiple X-ray measurement signals. The X-ray fitting analysis method according to claim 8.

10. Each of the aforementioned X-ray measurement signals includes one or more of the transmission pattern, reflection pattern, diffraction pattern, and scattering pattern, and each of the aforementioned structural parameters includes one or more of the thickness, density, and roughness. The X-ray fitting analysis method according to claim 9.

11. An X-ray fitting analysis system based on a multi-physical-variable model, The aforementioned X-ray fitting analysis system is An X-ray measuring device that measures an object to be detected and acquires multiple X-ray measurement signals, A processing unit is configured to perform the following steps, Equipped with, The aforementioned step is, Based on the target frame of the object to be detected, multiple fitting models are constructed, and multiple structural parameters are defined in the target frame. Using each of the aforementioned fitting models, an initial fitting process is performed on the aforementioned X-ray measurement signals to generate a plurality of initial fitting results for the aforementioned structural parameters. The process involves statistically analyzing the multiple initial fitting results to generate multiple initial parameter ranges, Based on the aforementioned multiple initial parameter ranges, a first co-fitting process is performed to fit the multiple X-ray measurement signals to a predetermined range of the target frame using the respective multiple fitting models, thereby obtaining multiple first fitting results that satisfy the first fitting conditions. Based on the data format of the plurality of first fitting results, a classification process is performed on the plurality of first fitting results by inputting the plurality of first fitting results into a trained classification model and classifying the plurality of first fitting results using a classification tree analysis process or a regression tree analysis process, thereby generating a plurality of classification fitting results. The process involves statistically analyzing the multiple classification fitting results to obtain multiple classification parameter ranges, Based on the aforementioned range of classification parameters, a second collaborative fitting process is performed to execute a fitting analysis on the multiple X-ray measurement signals for the entire range of the target frame using each of the multiple fitting models, thereby obtaining multiple second fitting results that satisfy the second fitting conditions and correspond to each of the multiple structural parameters. including, An X-ray fitting analysis system characterized by the following features.

12. The step of obtaining multiple X-ray measurement signals of the object to be detected by measuring using the at least one X-ray measuring device is: This includes arranging one X-ray measuring device to measure the object to be detected under multiple different measurement conditions, or arranging multiple different X-ray measuring devices to measure the object to be detected. The X-ray fitting analysis system according to claim 11.

13. Each of the aforementioned X-ray measurement signals includes one or more of the transmission pattern, reflection pattern, diffraction pattern, and scattering pattern, and each of the aforementioned structural parameters includes one or more of the thickness, density, and roughness. The X-ray fitting analysis system according to claim 11.

14. The target frame comprises a plurality of material layers, and the initial fitting process includes using the processing apparatus to execute an electromagnetic wave calculation engine corresponding to each fitting model, performing spectral fitting analysis on the corresponding X-ray measurement signal based on the target frame, and obtaining the corresponding initial fitting result. The X-ray fitting analysis system according to claim 11.

15. The first joint fitting process described above is: Based on the aforementioned ranges of initial parameters, a first set of parameters to be verified is generated that corresponds to the aforementioned ranges of structural parameters within a predetermined range. The first set of parameters to be verified is input into the plurality of fitting models to verify the first accuracy, and based on the first accuracy and the plurality of initial parameter ranges, the first set of parameters to be verified is adjusted until it satisfies the first fitting condition, and the first set of parameters to be verified that satisfies the first fitting condition is taken as the plurality of first fitting results. The X-ray fitting analysis system according to claim 11.

16. The number of the plurality of first fitting results is greater than the number of the plurality of classification results. The X-ray fitting analysis system according to claim 15.

17. The second joint fitting process is as follows: Based on the aforementioned ranges of classification parameters, a second set of parameters to be verified is generated, corresponding to the aforementioned ranges of structural parameters within the entire range. The second set of parameters to be validated is input into the multiple fitting models to validate the second accuracy, and based on the second accuracy and the range of the multiple classification parameters, the second set of parameters to be validated is adjusted until it satisfies the second fitting condition, and the second set of parameters to be validated that satisfies the second fitting condition is taken as the multiple second fitting results. The X-ray fitting analysis system according to claim 11.

18. An X-ray fitting analysis system based on a multi-physical-variable model, The aforementioned X-ray fitting analysis system is An X-ray measuring device that measures an object to be detected and acquires multiple X-ray measurement signals, A processing unit is configured to perform the following steps, Equipped with, The aforementioned step is, Based on the target frame of the object to be detected, multiple fitting models are constructed, and multiple structural parameters are defined in the target frame. Using each of the aforementioned fitting models, an initial fitting process is performed on the aforementioned X-ray measurement signals to generate a plurality of initial fitting results for the aforementioned structural parameters. The process involves statistically analyzing the multiple initial fitting results to generate multiple initial parameter ranges, The first set of parameters to be verified is generated based on the aforementioned multiple initial parameter ranges, The first set of parameters to be verified is input into the multiple fitting models to verify their accuracy, and based on the accuracy and the range of the multiple initial parameters, the first set of parameters to be verified is adjusted until it satisfies the optimization conditions, and the first set of parameters to be verified that satisfies the optimization conditions is designated as the optimized parameter set. Setting multiple weight values ​​for the aforementioned optimization parameter group, applying the multiple weight values ​​to the aforementioned optimization parameter group and inputting them into the aforementioned fitting models, and verifying the weight accuracy, The process involves adjusting the multiple weight values ​​based on the weight accuracy until the weight accuracy satisfies the weight optimization conditions, applying the multiple weight values ​​that satisfy the weight optimization conditions to the optimization parameter group, and generating the weight optimization parameter group. including, An X-ray fitting analysis system characterized by the following features.

19. The step of inputting the first set of parameters to be verified into the multiple fitting models and verifying the accuracy is: The first set of parameters to be verified is input into the multiple fitting models to generate multiple verification results, and the weighting accuracy is verified by determining the amount of variation of the multiple parameters in each verification result and the error value with respect to the multiple X-ray measurement signals. The X-ray fitting analysis system according to claim 18.

20. Each of the aforementioned X-ray measurement signals includes one or more of the transmission pattern, reflection pattern, diffraction pattern, and scattering pattern, and each of the aforementioned structural parameters includes one or more of the thickness, density, and roughness. The X-ray fitting analysis system according to claim 19.

Images