X-ray adaptation analysis method and X-ray adaptation analysis system based on a multiphysics variable model

Abstract

The present invention discloses an X-ray adaptation analysis method and an X-ray adaptation analysis system based on a multiphysics variable model.The procedure comprises the following: receiving an X-ray measurement signal (SX) of an inspection target (SP) by an X-ray measuring device (10); creating multiple fitting models based on a target architecture; performing an initial fitting process to generate multiple initial fitting results and generate multiple initial parameter ranges; performing a first co-fitting process to obtain multiple initial fitting results that satisfy a first fitting condition; performing a classification process based on the data form of the first fitting results to generate multiple classified fitting results and perform a count to obtain multiple classified parameter ranges; and performing a second co-fitting process to obtain multiple second fitting results that satisfy a second fitting condition and are mapped to the structural parameters.

Description

[0001] The present invention relates to an analysis method and an analysis system, in particular an X-ray adaptation analysis method and an X-ray adaptation analysis system based on a multiphysics variable model.

[0002] Existing 3D NAND flash memory typically features structures with a large depth-to-width ratio, repeatedly stacked and interconnected in a vertical direction. Therefore, etching the channel holes is the most critical process step in their technological development and manufacturing, where any anomaly impacts the subsequent process. For example, anomalies such as barrier / trap / tunneling layer deposits and channeling can impair the function and reliability of a component. Furthermore, uniformity is extremely important for process control.

[0003] To precisely control process parameters, the critical dimension (CD) of each wordline (WL) of the product is measured. However, previous measurements, such as transmission electron microscopy (TEM) or critical dimension scanning electron microscopy (CD-SEM) for examining cross-sections, can usually only be performed using destructive methods. Furthermore, spectroscopic optical critical dimension (OCD) measurement is the most important non-destructive method for measuring the critical dimension. However, the increasing number of layers also brings many challenges.

[0004] Starting from the disadvantages of the prior art, the present invention aims to offer an X-ray adaptation analysis method based on a multiphysics variable model.

[0005] To solve this problem, the present invention proposes an X-ray fitting analysis method based on a multiphysics variable model, comprising: performing a measurement by at least one X-ray measuring device to obtain multiple X-ray measurement signals of an inspection target; configuring a processing device to perform the following steps: creating multiple fitting models based on a target architecture of the inspection target, wherein the target architecture defines multiple structural parameters; performing an initial fitting process for the measurement signals with the fitting models to generate multiple initial fitting results of the structural parameters; counting the initial fitting results to generate multiple initial parameter ranges;Performing an initial co-fitting process for the measurement signals with the fitting models based on the initial parameter ranges to obtain several initial fitting results that satisfy an initial fitting condition, where the initial co-fitting process performs a fitting for a predefined area of ​​the target architecture; performing a classification process for the initial fitting results based on the data form of the initial fitting results to generate several classified fitting results; counting the classified fitting results to obtain several classified parameter ranges;and performing a second co-fitting process for the measurement signals with the fitting models based on the classified parameter ranges to obtain multiple second fitting results that satisfy a second fitting condition and are assigned to the structural parameters, wherein the second co-fitting process performs a fitting analysis for the entire range of the target architecture.

[0006] To solve this problem, the present invention further proposes an X-ray fitting analysis method based on a multiphysics variable model, comprising: performing a measurement by at least one X-ray measuring device to obtain multiple X-ray measurement signals of an inspection target; and configuring a processing device to perform the following steps: creating multiple fitting models based on a target architecture of the inspection target, wherein the target architecture defines multiple structural parameters; performing an initial fitting process for the X-ray measurement signals with the fitting models to generate multiple initial fitting results of the structural parameters; counting the initial fitting results to generate multiple initial parameter ranges; and generating a set of initial parameters to be checked based on the initial parameter ranges.Inputting the set of initial parameters to be checked into the fitting models to verify its accuracy, and adjusting the set of parameters to be checked based on the accuracy and the initial parameter ranges until an optimized parameter set is obtained when an optimization condition is met; setting several weight values ​​for the optimized parameter set and inputting the weight values ​​into the fitting models after applying them to the optimized parameter set to check the weighting accuracy; and adjusting the weight values ​​based on the weighting accuracy until the weighting accuracy satisfies a weight optimization condition, and applying the weight values ​​that satisfy the weight optimization condition to the optimized parameter set to produce a weight-optimized parameter set.

[0007] To solve this problem, the present invention further proposes an X-ray fitting analysis system based on a multiphysics variable model, comprising: performing a measurement by at least one X-ray measuring device to obtain multiple X-ray measurement signals of an inspection target; configuring a processing device to perform the following steps: creating multiple fitting models based on a target architecture of the inspection target, wherein the target architecture defines multiple structural parameters; performing an initial fitting process for the X-ray measurement signals with the fitting models to generate multiple initial fitting results of the structural parameters; counting the initial fitting results to generate multiple initial parameter ranges; generating a set of parameters to be checked based on the initial parameter ranges;Inputting the set of initial parameters to be checked into the fitting models to verify its accuracy, and adjusting the set of parameters to be checked based on the accuracy and the initial parameter ranges until an optimized parameter set is obtained when an optimization condition is met; setting several weight values ​​for the optimized parameter set and inputting the weight values ​​into the fitting models after applying them to the optimized parameter set to check the weighting accuracy; adjusting the weight values ​​based on the weighting accuracy until the weighting accuracy satisfies a weight optimization condition, and applying the weight values ​​that satisfy the weight optimization condition to the optimized parameter set to produce a weight-optimized parameter set.

[0008] To solve this problem, the present invention further proposes an X-ray fitting analysis system based on a multiphysics variable model, comprising at least one X-ray measuring device and a processing device. The at least one X-ray measuring device performs a measurement on an inspection target to obtain multiple X-ray measurement signals. The processing device is configured to perform the following steps: creating multiple fitting models based on a target architecture of the inspection target, wherein the target architecture defines multiple structural parameters; performing an initial fitting process for the X-ray measurement signals with the fitting models to generate multiple initial fitting results of the structural parameters; and counting the initial fitting results to generate multiple initial parameter ranges.Generating a set of parameters to be checked based on the initial parameter ranges; inputting the set of initial parameters to be checked into the fitting models to verify its accuracy, and adjusting the set of parameters to be checked based on the accuracy and the initial parameter ranges until an optimized parameter set is obtained when an optimization condition is met; setting several weighting values ​​for the optimized parameter set and inputting the weighting values ​​into the fitting models after they have been applied to the optimized parameter set to verify the weighting accuracy;Adjusting the weighting values ​​based on the weighting accuracy until the weighting accuracy satisfies a weight optimization condition, and applying the weighting values ​​that satisfy the weight optimization condition to the optimized parameter set to generate a weight-optimized parameter set.

[0009] One of the advantageous effects of the present invention is that, in the X-ray matching analysis method and the X-ray matching analysis system based on a multiphysics variable model according to the present invention, by performing a coarse first co-fitting process to classify the matching results in order to limit the range of structural parameters, and subsequently performing a fine second co-fitting process, the computational effort can be reduced in the case where the inspection target has a large number of material layers, while at the same time maintaining a certain matching accuracy in order to maintain the stability of the matching models.

[0010] Furthermore, in the X-ray adaptation analysis method and the X-ray adaptation analysis system based on a multiphysics variable model according to the present invention, a mechanism of weighting values ​​is applied and an optimization of the weighting values ​​is carried out, so that the optimization can be accelerated and the divergence of the adaptation results can be reduced.

[0011] For a better understanding of the features and technical content of the present invention, reference is made to the following detailed description and the drawings of the present invention, which, however, serve only for illustration and are not intended to limit the scope of protection of the present invention. Fig. Figure 1 shows a first schematic view of an X-ray adaptation analysis system based on a multiphysics variable model of a first embodiment of the present invention. Fig. Figure 2 shows a second schematic view of the X-ray adaptation analysis system based on a multiphysics variable model of the first embodiment of the present invention. Fig. Figure 3 shows a flowchart of an X-ray adaptation analysis method based on a multiphysics variable model of the first embodiment of the present invention. Fig. 4 and Fig. Figure 5 shows a first schematic view and a second schematic view of a target architecture of an inspection target of the first embodiment of the present invention. Fig. Figure 6 shows a series of curve diagrams of two initial fitting results of the first embodiment of the present invention. Fig. Figure 7 shows a series of curve diagrams of two first fitting results of the first embodiment of the present invention, which are obtained after a first co-fitting process. Fig. Figure 8 shows classified fitting results of the first embodiment of the present invention, which are obtained after passing through a trained classification model. Fig. Figure 9 shows two second adaptation results after a second co-adaptation process of the first embodiment of the present invention. Fig. Figure 10 shows a schematic representation of the X-ray adaptation analysis results of the first embodiment of the present invention in comparison to the adaptation analysis results of a single physical model. Fig. Figure 11 shows a flowchart of an X-ray adaptation analysis method based on a multiphysics variable model of a second embodiment of the present invention. Fig. Figure 12 shows a schematic representation of the first iteration of step S26 of the second embodiment of the present invention and Fig. Figure 13 shows a schematic representation of the second iteration of step S26 of the second embodiment of the present invention.

[0012] The following describes, with reference to specific embodiments, the embodiments of an X-ray adaptation analysis method and an X-ray adaptation analysis system based on a multiphysics variable model, which are disclosed within the scope of the present invention. Those skilled in the art will recognize that the advantages and technical effects of the present invention can be derived from the disclosures in this description. The present invention can be implemented or applied through other specific embodiments, whereby various modifications or alterations can be made to the details disclosed in this description as needed and depending on the application, without departing from the fundamental ideas of the present invention.Furthermore, it should be noted that the individual components of the present invention are not shown in their actual size, but only schematically. The embodiments described below serve to provide a more detailed description of the various configurations of the present invention, although the corresponding disclosures do not constitute a limitation of the scope of protection of the present invention. Furthermore, the term "or" used in this description may, if applicable, encompass any of the items listed or a combination of several such items. [First embodiment]

[0013] Fig. Figure 1 shows a first schematic view of an X-ray adaptation analysis system based on a multiphysics variable model of a first embodiment of the present invention. Fig. Figure 2 shows a second schematic view of the X-ray adaptation analysis system based on a multiphysics variable model of the first embodiment of the present invention. With reference to Fig. 1 and Fig. 2 provides the first embodiment of the present invention an X-ray adaptation analysis system 1 based on a multiphysics variable model, comprising several X-ray measuring devices 10 and a processing device 12.

[0014] In the present embodiment, the X-ray measuring device 10 can comprise an X-ray generator, X-ray optics, an X-ray detector, and a sample carrier. The X-ray generator can be used to generate an X-ray beam with a specific energy. The X-ray optics can guide the X-ray beam and focus it onto an inspection target SP. The X-ray detector can be used to receive a signal generated by irradiating the inspection target SP with the X-ray beam. However, the present invention is not limited to this. The X-ray measuring device 10 can be used to measure an inspection target SP in order to obtain multiple X-ray measurement signals Sx. Although in Fig. Since several X-ray measuring devices 10 are shown in Figure 1, in practical application it is possible to configure a single X-ray measuring device 10 such that the inspection target SP is measured under several different measurement conditions. Alternatively, several different X-ray measuring devices 10 can be provided for measuring the inspection target SP, as shown in Figure 1. Fig. 2 can be seen. In the present embodiment, the X-ray measuring devices 10 can include an X-ray reflectivity (XRR) meter, an X-ray fluorescence spectrometer (XRF) meter, a small-angle X-ray scattering (SAX) meter, an X-ray diffractometer (XRD) or another device that can use X-rays as a light source for measurement.

[0015] Each of the X-ray measurement signals Sx can include one or more transmission, reflection, diffraction, and scattering spectra. For example, if the X-ray measurement signal Sx includes a reflection spectrum, X-rays can be incident on the inspection target SP at several predefined angles. The intensity of the reflected X-rays is then received by a receiver in the X-ray measuring device 10, whereupon the X-ray measurement signal is adjusted and analyzed to obtain structural parameters of the inspection target SP, which may include, for example, one or more of its thickness, density, and roughness.

[0016] The processing device 12 can, for example, be a computer system with a processor and memory, which can be configured to execute a stored instruction set or program code to obtain an X-ray measurement signal and perform a matching analysis. Furthermore, the processing device 12 can control the X-ray measuring device 10 to perform a corresponding measurement operation on the inspection target SP.

[0017] It will be on Fig. Reference is made to Figure 3, which shows a flowchart of an X-ray adaptation analysis method based on a multiphysics variable model of the first embodiment of the present invention. As in Fig. As shown in section 3, the X-ray adaptation analysis procedure can include the following steps:

[0018] Step S10: Perform a measurement using the X-ray measuring device to obtain multiple X-ray measurement signals of the inspection target. In this step, the X-ray measuring device 10 may be present once or multiple times.

[0019] The processing device 12 can then be configured to perform the following steps:

[0020] Step S11: Create multiple adaptation models based on a target architecture of the inspection objective. Fig. 4 and Fig. Figure 5 shows a first schematic view and a second schematic view of the target architecture of the inspection target of the first embodiment of the present invention. In step S11, the target architecture of the inspection target SP can be shown as in Fig. Figure 4 illustrates, for example, a sample where the pattern on the carbon film has not yet been transferred to the multimaterial layer structure. Alternatively, the target architecture of the inspection target SP can be, as shown in Fig. Figure 5 shows, for example, a sample in which the pattern on the carbon film was transferred to the multi-material layer structure and the carbon film was removed. Accordingly, the target architecture can comprise multiple material layers and defines several structural parameters, which may include, for example, the thickness, density, or roughness of the individual material layers.

[0021] Step S12: Perform an initial fitting process for the measurement signals with the fitting models to generate several initial fitting results of the structural parameters.

[0022] The initial fitting process involves the processing device 12 executing several EMW calculation engines associated with the fitting models to perform a spectral fitting analysis of the corresponding X-ray measurement signal Sx based on the target architecture in order to obtain a corresponding initial fitting result. The EMW calculation engine may, for example, include one or more of the FDTD (Finite-Difference Time-Domain), DWBA (Distorted Wave Born Approximation), RCWA (Rigorous Coupled Wave Analysis), DDP (Discrete Dipole Approximation), and BEM (Boundary Element Method) algorithms.

[0023] Specifically, data measured after exposure to X-rays of varying energies can be fitted, for example, using the EMW computational engine described above, to obtain the structural parameters of the target architecture and the corresponding accuracy. Accuracy can encompass the variance and the magnitude of the error of a specific structural parameter, such as density. The variance can be, for example, the mean of the structural parameters resulting from multiple fittings, divided by the standard deviation, while the magnitude of the error can be described by a cost function and serves as a measure of the degree of agreement between the model and the fitted target data. The cost function can, for example, be the mean squared error (MSE) of the fitting results relative to the fitted target data.For a fitting model with N layers, several initial fittings using an EMW computational engine can result in a magnitude of error and a set of N variances of the structural parameters (mean density / density standard deviation of each of the N material layers). Therefore, if there are M fitting models, each fitted with M EMW computational engines, M magnitudes of error and M sets of N variances can be generated. It is referred to as... Fig. Reference is made to Figure 6, which shows a series of curve diagrams of two initial fitting results of the first embodiment of the present invention. The two upper curve diagrams in Fig. Figure 6 shows the X-ray measurement signal Sx and the adjusted reflectivity versus the angle, while the two lower curve diagrams in Fig. Figure 6 shows the determined density (divided into the real part rho and the imaginary part irho) versus the depth, with the lower data showing the accuracy, expressed as MSE (denoted as chisq), of approximately 452192 and 688786 respectively (decimal places omitted).

[0024] Step S13: Counting the initial fitting results to generate multiple initial parameter ranges.

[0025] If multiple structural parameters are obtained after the initial fitting, a preliminary count can be performed to determine an initial parameter range for each of the structural parameters. It should be noted that with an increasing number of layers of the inspection target SP, too many flowing parameters can make the fitting model very complex, and the computation time also increases exponentially. In practice, simple modeling and fitting is not possible for three-dimensional NAND structures because the channel hole depth doubles at each node junction, making the sensitivity of the critical dimension in the lower region too weak. Simultaneously, the critical dimensions of adjacent word lines may give the same spectral response and thus be indistinguishable, leading to modeling problems.Although it is possible to use the critical dimension of a lower layer in combination with the critical dimension of a middle layer in a simplified manner to improve the stability of the model, this reduces the accuracy of the results for the critical dimension of the lower layer. Furthermore, if the ratio between the critical dimension of the middle layer and the critical dimension of the lower layer changes, the accuracy decreases even further. Therefore, the present invention employs an alternative, novel fitting method for the analysis to reduce the computational effort and increase the accuracy.

[0026] Step S14: Perform an initial co-fitting process for the measurement signals with the fitting models based on the initial parameter ranges to obtain several initial fitting results that satisfy an initial fitting condition.

[0027] In step S14, the first co-fitting process performs a rough adjustment for a predefined region of the target architecture. For example, if the target architecture is a NAND flash memory with 400 layers, the predefined region could represent the odd-numbered layers among those 400 layers. Therefore, the first co-fitting process in step S14 consists of adjusting the structural parameters, such as density, thickness, and roughness, of the odd-numbered layers.

[0028] In the first co-fitting process, a set of initial parameters to be checked can first be generated based on the initial parameter ranges obtained in step S13. This set of initial parameters corresponds to the structural parameters in the aforementioned predefined range (200 layers). The set of initial parameters to be checked is then fed into the aforementioned multiple fitting models to verify initial accuracy. For example, after the set of initial parameters to be checked has been fed into the fitting models and fitted using the appropriate EMW calculation engine, an error magnitude and N variances of the set of initial parameters to be checked are generated. It is then determined whether the error magnitude and variances are lower compared to the initial fitting result.

[0029] If this is not the case, the set of first parameters to be checked is set, for example, based on the random numbers of the initial parameter ranges and through nested combinations, and checked again until the first fitting condition is met. The set of first parameters to be checked that meets the first fitting condition is then used as the first fitting result. The first fitting condition can be considered met if, for example, the error magnitude and variances are smaller compared to the initial fitting result and the difference is greater than a predefined value.

[0030] For example, it will refer to the Fig. Reference is made to Figure 7, which shows a series of curve diagrams of two first matching results of the first embodiment of the present invention, obtained after the first co-matching process. In the two upper curve diagrams of Fig. Figure 7 shows the X-ray measurement signal Sx and the reflectivity obtained through the first co-adjustment process plotted against the angle, while in the two lower curve diagrams of Fig. Figure 7 shows the determined density (divided into the real part rho and the imaginary part irho) plotted against the depth. The lower data show the accuracy, expressed as MSE (denoted as chisq), of approximately 15374 and 349764 respectively (decimal places omitted), with both values ​​decreasing from the initial fitting result.

[0031] Step S15: Perform a classification process for the initial fitting results based on the data form of the initial fitting results to generate multiple classified fitting results.

[0032] In this step, classification accuracy and efficiency can be improved by using a decision tree algorithm in combination with machine learning. Specifically, the initial fitting result generated in step S14 can be fed into a trained classification model to generate classified fitting results. The trained classification model is then trained to classify the initial fitting results using either a classification tree analysis process or a regression tree analysis process, where the number of initial fitting results is greater than the number of classified results.For example, the classification tree analysis process or the regression tree analysis process can be trained to capture data features of the initial fitting results, such as heights, depths, curves, peaks, and valleys in a curve diagram, to perform similarity clustering of the initial fitting results, and also to rank the classifications in terms of accuracy (including variances and the cost function) and identify the classifications with a higher degree of accuracy. Therefore, the best classification can be determined by the decision tree algorithm, which in turn reduces the uncertainty of the architecture of a fitting model. It can be applied to the... Fig. Reference is made to Figure 8, which shows the classified fitting results of the first embodiment of the present invention, obtained after passing the data through a trained classification model. It should be noted that here, the features of changes in a structural parameter (e.g., density) with respect to depth are used as classification criteria for the data shapes, but the present invention is not limited to this. Other structural parameters, such as thickness and roughness, can also serve as a basis for classification in step S15.

[0033] Step S16: Counting the classified fitting results to obtain multiple classified parameter ranges.

[0034] If one takes Fig. As an example, the classified fitting results show that the classification in step S15 yields five types of curves, each with five types of parameter ranges. This allows the range of each structural parameter to be limited, significantly reducing the computational effort required for the fitting analysis. Since the classified parameter ranges obtained in this step are based on the results of the first co-fitting process, their accuracy has already been optimized.

[0035] Step S17: Perform a second co-fitting process on the measurement signals with the fitting models based on the classified parameter ranges to obtain multiple second fitting results that satisfy a second fitting condition and are associated with the structural parameters. In this step, the second co-fitting process performs a fitting analysis for the entire target architecture. Although the number of layers to be computed is larger compared to the first co-fitting process, the range of structural parameters to be calculated has been restricted, thus significantly reducing the overall computational effort while maintaining a certain level of accuracy.

[0036] Similarly, in the second co-fitting process, a set of second parameters to be checked can first be generated based on the classified parameter ranges obtained in step S16. This set of second parameters is then fed into the aforementioned multiple fitting models to perform a second accuracy check. For example, after the set of second parameters to be checked has been fed into the fitting models and fitted using the appropriate EMW calculation engine, an error magnitude and N variances of the set of second parameters to be checked are generated, and it is then determined whether the error magnitude and variances are lower compared to the first fitting result.

[0037] If this is not the case, the set of second parameters to be checked is adjusted, for example, based on random numbers from the classified parameter ranges and through nested combinations, and checked again until the second fitting condition is met. The set of second parameters to be checked that meets the second fitting condition is then used as the second fitting result, which is the final fitting result. The second fitting condition can be considered met if, for example, the error magnitude and variances are smaller compared to the first fitting result and the difference is greater than another predefined value. [The last sentence appears to be incomplete and requires context. It has been omitted.] Fig. 9 and Fig. 10 referred, whereby Fig. 9 two second adaptation results after a second co-adaptation process of the first embodiment of the present invention and Fig. Figure 10 shows a schematic representation of the X-ray adaptation analysis results of the first embodiment of the present invention in comparison to the adaptation analysis results of a single physical model.

[0038] In the two upper curve diagrams of Fig. Figure 9 shows the X-ray measurement signal Sx (thick line) and the reflectivity obtained through the second co-matching process (thin line) plotted against the angle, while in the two lower curve diagrams of Fig. Figure 9 shows the determined density (divided into the real part rho and the imaginary part irho) plotted against the depth. The lower data show the accuracy, expressed as MSE (denoted as chisq), of approximately 6104 and 5740 respectively (decimal places omitted), with both values ​​decreasing compared to the initial fit result and the first fit result.

[0039] The comparison results for structural parameters such as thickness, density (real part), density (imaginary part) and roughness are in Fig. Figure 10 shows the upper part representing a single physical model and the lower part a multiphysics model. The histograms represent the means and the thin lines represent the variances. Fig. As can be seen in section 10, the variances obtained when calculating with a multiphysics model are relatively small. [Second embodiment]

[0040] The second embodiment of the present invention provides a further X-ray adaptation analysis system based on a multiphysics variable model. With respect to the system architecture, the X-ray adaptation analysis system of the present embodiment is essentially the same as that described in Fig. 2 and Fig. Three examples are shown, with the difference that the details of the fit analysis are different, which is why a repetition of the description is omitted here.

[0041] Fig. Figure 11 shows a flowchart of an X-ray adaptation analysis method based on a multiphysics variable model of the second embodiment of the present invention. As in Fig. As shown in 11, the second embodiment of the present invention provides an X-ray adaptation analysis method based on a multiphysics variable model, comprising the following steps:

[0042] Step S20: Perform a measurement using the X-ray measuring device to obtain multiple X-ray measurement signals of the inspection target.

[0043] The processing device is then configured to perform the following steps:

[0044] Step S21: Creating multiple adaptation models based on a target architecture of the inspection objective. As described in the first example, the target architecture defines several structural parameters.

[0045] Step S22: Perform an initial fitting process for the X-ray measurement signals with the fitting models to generate several initial fitting results of the structural parameters.

[0046] Step S23: Counting the initial fitting results to generate multiple initial parameter ranges.

[0047] Step 24: Generating a set of parameters to be checked based on the initial parameter ranges.

[0048] Step S25: Inputting the set of initial parameters to be checked into the fitting models to verify its accuracy, and adjusting the set of parameters to be checked based on the accuracy and the initial parameter ranges until an optimized parameter set is obtained if an optimization condition is met.

[0049] In steps S23 to S25, several sets of parameters to be checked can first be generated based on the initial parameter ranges obtained in step S23. Each of these sets is then fed into the aforementioned multiple fitting models to verify their accuracy. For example, if each set of parameters to be checked is fed into the fitting models and fitted using the appropriate EMW calculation engine, an error amount and several corresponding variances are generated. It is then determined whether the error amount and the variances satisfy the optimization condition, i.e., whether the error amount is less than a predefined error amount and the variances are less than a predefined variance. If this is not the case, the set of parameters to be checked is discontinued, e.g.,based on the random numbers of the initial parameter ranges as well as by nested combinations, and checked again until the optimization condition is met, with the multiple sets of parameters to be checked that meet the optimization condition being used as the optimized parameter set.

[0050] Step S26: Define several weighting values ​​for the optimized parameter set and input the weighting values ​​into the fitting models after applying them to the optimized parameter set to verify the weighting accuracy. For example, each set of parameters to be verified that meets the optimization condition includes structural parameters such as thickness, density, and roughness of each material layer of the inspection target. It is applied to Fig. Reference is made to Figure 12, which shows a schematic representation of the first iteration of step S26 of the second embodiment of the present invention. In the first iterative calculation, each set of parameters to be checked in the optimized parameter set can 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 sets of parameters to be checked, they are each multiplied by a weight value of 0.5 to obtain the optimized parameter set after applying the weight values.

[0051] The set of parameters to be checked can then be entered into the fitting models to generate multiple results to be checked and to determine multiple parameter variances for each of the results to be checked, as well as an error value with respect to the X-ray measurement signals, in order to check the weighting accuracy. As can be seen from the middle of Fig. As can be seen from the 12 distribution diagrams shown for the thickness, density, and roughness spaces, the results for the structural parameters are rather divergent. In the Fig. Figure 12 shows two fitting results and two density diagrams on the right, with the resulting weighting accuracies, expressed as MSE (denoted as chisq), at 9339 and 8474 respectively (decimal places omitted). It appears there is still room for optimization of the weighting values. Furthermore, the first iteration generates two new sets of parameters to be checked, which can be used in the second iteration.

[0052] Step S27: Adjusting the weighting values ​​based on the weighting accuracy until the weighting accuracy meets a weight optimization condition, and applying the weighting values ​​that meet the weight optimization condition to the optimized parameter set to create a weight-optimized parameter set.

[0053] It will be directed to the Fig. Reference is made to Figure 13, which shows a schematic representation of the second iteration of step S26 of the second embodiment of the present invention. In the second iterative calculation, the two further sets of parameters to be checked, which result from the first iteration, can each be multiplied by a weighting value of 0.7 and 0.3, respectively, to obtain the optimized parameter set after application of the weighting values.

[0054] Similarly, the two new sets of parameters to be checked can be fed into the fitting models to generate multiple results to be checked and to determine multiple parameter variances for each of the results to be checked, as well as an error value with respect to the X-ray measurement signals, to verify the weighting accuracy. As can be seen from the distribution diagrams for the thickness, density, and roughness space in the middle of Fig. As can be seen in Figure 13, the results for the structural parameters are more convergent than in the first iteration, suggesting that the weight distribution of 0.7 and 0.3 may be better. In the Fig. Figure 13 shows two fitting results and two density plots on the right, with the resulting weighting accuracies, expressed as MSE (denoted as chisq), at 8453 and 5505 respectively (decimal places omitted). It appears that the weighting accuracy is higher than in the first iteration.

[0055] If the weighting accuracy reaches a predetermined range of values ​​(e.g., MSE is less than a certain value), meaning that the weight optimization condition is met, the weighting values ​​that meet the weight optimization condition can be applied to the optimized parameter set to generate a weight-optimized parameter set that defines several structural parameters resulting from the fit analysis. [Advantageous effects of the exemplary embodiments]

[0056] One of the advantageous effects of the present invention is that, in the X-ray matching analysis method and the X-ray matching analysis system based on a multiphysics variable model according to the present invention, by performing a coarse first co-fitting process to classify the matching results in order to limit the range of structural parameters, and subsequently performing a fine second co-fitting process, the computational effort can be reduced in the case where the inspection target has a large number of material layers, while at the same time maintaining a certain matching accuracy in order to maintain the stability of the matching models.

[0057] Furthermore, in the X-ray adaptation analysis method and the X-ray adaptation analysis system based on a multiphysics variable model according to the present invention, a mechanism of weighting values ​​is applied and an optimization of the weighting values ​​is carried out, so that the optimization can be accelerated and the divergence of the adaptation results can be reduced.

[0058] The above disclosure merely presents preferred possible embodiments of the present invention and is not intended to limit the claims of the present invention, so that all equivalent technical modifications made by applying the content of the description and the drawings of the present invention are included in the claims of the present invention. Reference symbol list 1 X-ray adaptation analysis system 10 X-ray measuring device 12 Processing device SP Inspection Target Sx X-ray measurement signal S10, S11, S12, S13, S14, S15, S16, S17, S20, S21, S22, S23, S24, S25, S26, S27 Step rho Real part irho Imaginary part chisq MSE

Claims

[1] X-ray adaptation analysis method based on a multiphysics variable model, comprising: - Performing a measurement using at least one X-ray measuring device (10) to obtain multiple X-ray measurement signals (SX) of an inspection target (SP); - Configure a processing device (12) to perform the following steps: - Creating multiple adaptation models based on a target architecture of the inspection target (SP), where the target architecture defines multiple structural parameters; - Performing an initial fitting process for the measurement signals with the fitting models to generate multiple initial fitting results of the structural parameters; - Counting the initial fitting results to generate multiple initial parameter ranges; - Performing an initial co-fitting process for the measurement signals with the fitting models based on the initial parameter ranges to obtain several initial fitting results that satisfy an initial fitting condition, wherein the initial co-fitting process performs a fitting for a predetermined range of the target architecture; - Performing a classification process for the initial fitting results based on the data format of the initial fitting results to generate multiple classified fitting results; - Counting the classified fitting results to obtain multiple classified parameter ranges; and - Performing a second co-fitting process for the measurement signals with the fitting models based on the classified parameter ranges to obtain multiple second fitting results that satisfy a second fitting condition and are assigned to the structural parameters, with the second co-fitting process performing a fitting analysis for the entire range of the target architecture. [2] X-ray adaptation analysis method according to claim 1, wherein the step of performing a measurement by the at least one X-ray measuring device (10) to obtain multiple X-ray measurement signals (SX) of the inspection target (SP) comprises configuring a single X-ray measuring device (10) to measure the inspection target (SP) under several different measurement conditions or configuring several different X-ray measuring devices (10) to measure the inspection target (SP). [3] X-ray matching analysis method according to claim 1 or 2, wherein each of the X-ray measurement signals (SX) comprises one or more of a transmission spectrum, a reflection spectrum, a diffraction spectrum and a scattering spectrum and each of the structural parameters comprises one or more of the thickness, density and roughness. [4] X-ray matching analysis method according to any one of claims 1 to 3, wherein the target architecture comprises multiple material layers, and wherein the initial matching process comprises executing an EMF calculation machine associated with each of the matching models with the processing device (12) to perform a spectral matching analysis of the corresponding X-ray measurement signal (SX) based on the target architecture in order to obtain the corresponding initial matching result. [5] X-ray matching analysis method according to any one of claims 1 to 4, wherein the first co-matching process comprises: - Generating a set of initial parameters to be checked based on the initial parameter ranges, corresponding to the structural parameters in the given ranges; - Inputting the set of first parameters to be checked into the fitting models to verify an initial accuracy, and adjusting the set of first parameters to be checked based on the initial accuracy and the initial parameter ranges until the first fitting condition is met, and using the set of first parameters to be checked that satisfies the first fitting condition as the first fitting results. [6] X-ray fitting analysis method according to claim 5, wherein the classification process comprises inputting the first fitting results into a trained classification model to generate the classified results, wherein the trained classification model is trained to classify the first fitting results by a classification tree analysis process or a regression tree analysis process, wherein the number of first fitting results is greater than the number of classified results. [7] X-ray matching analysis method according to any one of claims 1 to 6, wherein the second co-matching process comprises: - Generating a set of second parameters to be checked based on the classified parameter ranges, corresponding to the structural parameters in the entire range; - Inputting the set of second parameters to be checked into the fitting models to check for a second accuracy, and adjusting the set of second parameters to be checked based on the second accuracy and the classified parameter ranges until the second fitting condition is met, and using the set of second parameters to be checked that meets the second fitting condition as the second fitting results. [8] X-ray adaptation analysis method based on a multiphysics variable model, comprising: - Performing a measurement using at least one X-ray measuring device (10) to obtain multiple X-ray measurement signals (SX) of an inspection target (SP); and - Configure a processing device (12) to perform the following steps: - Creating multiple adaptation models based on a target architecture of the inspection target (SP), where the target architecture defines multiple structural parameters; - Performing an initial fitting process for the X-ray measurement signals (SX) with the fitting models to generate multiple initial fitting results of the structural parameters; - Counting the initial fitting results to generate multiple initial parameter ranges; - Generating a set of initial parameters to be checked based on the initial parameter ranges; - Inputting the set of initial parameters to be checked into the fitting models to verify its accuracy, and adjusting the set of parameters to be checked based on the accuracy and the initial parameter ranges until an optimized parameter set is obtained if an optimization condition is met; - Defining multiple weighting values ​​for the optimized parameter set and entering the weighting values ​​into the fitting models after they have been applied to the optimized parameter set to verify the weighting accuracy; and - Adjusting the weighting values ​​based on the weighting accuracy until the weighting accuracy meets a weight optimization condition, and applying the weighting values ​​that meet the weight optimization condition to the optimized parameter set to create a weight-optimized parameter set. [9] X-ray fitting analysis method according to claim 8, wherein the step of entering the set of first parameters to be checked into the fitting models in order to check its accuracy comprises: Inputting the set of parameters to be checked into the fitting models to generate multiple results to be checked, and determining multiple parameter variances for each of the results to be checked and an error value with respect to the X-ray measurement signals (SX) to check the weighting accuracy. [10] X-ray matching analysis method according to claim 9, wherein each of the X-ray measurement signals (SX) comprises one or more of a transmission spectrum, a reflection spectrum, a diffraction spectrum and a scattering spectrum and each of the structural parameters comprises one or more of the thickness, density and roughness. [11] X-ray adaptation analysis system (1) based on a multiphysics variable model, comprising: - at least one X-ray measuring device (10) used to perform a measurement on an inspection target (SP) in order to obtain multiple X-ray measurement signals (SX); and - a processing device (12) configured to perform the following steps: - Creating multiple adaptation models based on a target architecture of the inspection target (SP), where the target architecture defines multiple structural parameters; - Performing an initial fitting process for the measurement signals with the fitting models to generate multiple initial fitting results of the structural parameters; - Counting the initial fitting results to generate multiple initial parameter ranges; - Performing an initial co-fitting process for the measurement signals with the fitting models based on the initial parameter ranges to obtain several initial fitting results that satisfy an initial fitting condition, wherein the initial co-fitting process performs a fitting for a predetermined range of the target architecture; - Performing a classification process for the initial fitting results based on the data format of the initial fitting results to generate multiple classified fitting results; - Counting the classified fitting results to obtain multiple classified parameter ranges; and - Performing a second co-fitting process for the measurement signals with the fitting models based on the classified parameter ranges to obtain multiple second fitting results that satisfy a second fitting condition and are assigned to the structural parameters, with the second co-fitting process performing a fitting analysis for the entire range of the target architecture. [12] X-ray adaptation analysis system (1) according to claim 11, wherein the step of performing a measurement by the at least one X-ray measuring device (10) to obtain multiple X-ray measurement signals (SX) of the inspection target (SP) comprises configuring a single X-ray measuring device (10) to measure the inspection target (SP) under several different measurement conditions or configuring several different X-ray measuring devices (10) to measure the inspection target (SP). [13] X-ray adaptation analysis system (1) according to claim 11 or 12, wherein each of the X-ray measurement signals (SX) comprises one or more of a transmission spectrum, a reflection spectrum, a diffraction spectrum and a scattering spectrum and each of the structural parameters comprises one or more of the thickness, density and roughness. [14] X-ray matching analysis system (1) according to one of claims 11 to 13, wherein the target architecture comprises multiple material layers, and wherein the initial matching process comprises executing an EMW calculation machine associated with each of the matching models with the processing device (12) to perform a spectral matching analysis of the corresponding X-ray measurement signal (SX) based on the target architecture in order to obtain the corresponding initial matching result. [15] X-ray adaptation analysis system (1) according to any one of claims 11 to 14, wherein the first co-adaptation process comprises: - Generating a set of initial parameters to be checked based on the initial parameter ranges, corresponding to the structural parameters in the given ranges; - Inputting the set of first parameters to be checked into the fitting models to verify an initial accuracy, and adjusting the set of first parameters to be checked based on the initial accuracy and the initial parameter ranges until the first fitting condition is met, and using the set of first parameters to be checked that satisfies the first fitting condition as the first fitting results. [16] X-ray adaptation analysis system (1) according to claim 15, wherein the classification process comprises inputting the first adaptation results into a trained classification model to generate the classified results, wherein the trained classification model is trained to classify the first adaptation results by a classification tree analysis process or a regression tree analysis process, wherein the number of first adaptation results is greater than the number of classified results. [17] X-ray adaptation analysis system (1) according to any one of claims 11 to 16, wherein the second co-adaptation process comprises: - Generating a set of second parameters to be checked based on the classified parameter ranges, corresponding to the structural parameters in the entire range; - Inputting the set of second parameters to be checked into the fitting models to check for a second accuracy, and adjusting the set of second parameters to be checked based on the second accuracy and the classified parameter ranges until the second fitting condition is met, and using the set of second parameters to be checked that meets the second fitting condition as the second fitting results. [18] X-ray adaptation analysis system (1) based on a multiphysics variable model, comprising: - at least one X-ray measuring device (10) that performs a measurement to obtain multiple X-ray measurement signals (SX) of an inspection target (SP); - Configure a processing device (12) to perform the following steps: - Creating multiple adaptation models based on a target architecture of the inspection target (SP), where the target architecture defines multiple structural parameters; - Performing an initial fitting process for the X-ray measurement signals (SX) with the fitting models to generate multiple initial fitting results of the structural parameters; - Counting the initial fitting results to generate multiple initial parameter ranges; - Generating a set of parameters to be checked based on the initial parameter ranges; - Inputting the set of parameters to be checked into the fitting models to verify its accuracy, and adjusting the set of parameters to be checked based on the accuracy and the initial parameter ranges until an optimized parameter set is obtained when an optimization condition is met; - Defining multiple weighting values ​​for the optimized parameter set and entering the weighting values ​​into the fitting models after they have been applied to the optimized parameter set to verify the weighting accuracy; and - Adjusting the weighting values ​​based on the weighting accuracy until the weighting accuracy meets a weight optimization condition, and applying the weighting values ​​that meet the weight optimization condition to the optimized parameter set to create a weight-optimized parameter set. [19] X-ray adaptation analysis system (1) according to claim 18, wherein the step of entering the set of parameters to be checked into the adaptation models in order to check its accuracy comprises: Inputting the set of parameters to be checked into the fitting models to generate multiple results to be checked, and determining multiple parameter variances for each of the results to be checked and an error value with respect to the X-ray measurement signals (SX) to check the weighting accuracy. [20] X-ray adaptation analysis system (1) according to claim 18 or 19, wherein each of the X-ray measurement signals (SX) comprises one or more of a transmission spectrum, a reflection spectrum, a diffraction spectrum and a scattering spectrum and each of the structural parameters comprises one or more of the thickness, density and roughness.

Images