Method and system for measuring thickness of film

Abstract

The invention provides a film thickness measuring method and a film thickness measuring system. The measuring method comprises the following steps: acquiring estimated thicknesses of a plurality of film layers on the surface of a semiconductor device to be measured, and constructing a candidate thickness set for thick film layers reaching a preset thickness upper limit; constructing a first structure model according to the thick film layer and the estimated thickness of the standard film layer of which the estimated thickness is between the thickness upper limit and the preset thickness lower limit; providing detection light to the semiconductor device, and collecting actual spectral information of reflected light; according to the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness, calculating a local optimal solution of the first structure model about each candidate thickness, and determining the measurement thickness of the thick film layer according to the optimal solution; the estimated thickness, smaller than the thickness lower limit, of the multiple film layers is incorporated into the optimal first structure model to construct a second structure model; and according to the actual spectral information and second theoretical spectral information corresponding to the second structure model, calculating an overall optimal solution of the measured thicknesses of the plurality of film layers.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor manufacturing, and in particular to a film thickness measurement method, a film thickness measurement system, and a computer-readable storage medium. Background Art

[0002] In the complex process of semiconductor production, semiconductor measurement equipment is a key factor to ensure the improvement of chip yield. Optical measurement equipment has been widely used in semiconductor measurement processes because of its outstanding high precision and high stability, and can efficiently measure a variety of key process dimensional parameters at one time. The operating process of this type of equipment includes first collecting the spectral information of the sample to be tested, and then combining the pre-set sample model and the optical parameters of the system to accurately analyze the model properties of the sample to be tested by fitting the spectrum of the sample model, such as the thickness of the film and key dimensions. Therefore, in order to maintain or even improve the yield of chip manufacturing in large-scale mass production, it is necessary to fit the high-precision model properties of the sample to be tested based on the optical parameters of the system and the collected spectral information.

[0003] The accuracy of the fitting results of the properties of the sample model to be tested depends first on the initial structure and parameters of the sample model, such as the nonlinear refractive index (NK) of the material parameters, the film thickness parameter (T), etc. Figure 1 As shown, the initial structure of the sample model on the surface of the semiconductor device can usually be composed of several stacked thin films, the material parameters NK of two adjacent thin films are different, and the thickness T of each film is in nanometers (nm) or angstroms. Secondly, the accuracy of the fitting results also depends on the fitting method used in the specific application. The current mainstream fitting methods mostly use the strategy of regression analysis. This strategy calculates the theoretical spectrum of the model based on the initial structural parameters of the model to be tested, and compares it with the collected spectrum. If there is a large difference in the comparison results, the model parameters are modified according to certain rules based on the initial structural parameters, and calculated and compared again. Through iterative calculation and comparison, the model parameters whose comparison results meet the requirements are found as the high-precision properties of the sample to be tested obtained by fitting.

[0004] However, in the actual fitting process, the regression analysis fitting method essentially takes the initial structure of the stacked film as the calculation starting point and searches for the local optimal solution through iterative calculation. ), since the actual film thickness parameter T of these film layers may cover a relatively narrow value range, when fitting a thin film layer based on multiple collected spectra, a film thickness parameter T with a large degree of dispersion may be obtained, and thus it is impossible to fit the real film thickness value with high accuracy. For the thicker film layer in the sample to be tested (for example, the film thickness parameter ), since the actual film thickness parameters T of these film layers may cover a relatively wide range of values, when a film thickness parameter T is initially set as the starting point for fitting, it is easy to obtain a local optimal film thickness, and it is impossible to find the global optimal film thickness, so it is impossible to fit and obtain a high-precision true film thickness value.

[0005] In this regard, in order to improve the accuracy of the model fitting results of the sample to be tested, in the prior art, for the case where there are thinner film layers in the sample to be tested, a multiple fitting strategy can be adopted. First, the film thickness parameters of the thinner film layer are specified not to participate in the fitting, and the film thickness parameters of other film layers are fitted; then, the fitted model parameters are used as new parameters to be fitted, and the film thickness parameters that were not previously fitted are fitted again. This multiple fitting strategy can effectively improve the fitting accuracy of the thinner film layer, but this strategy cannot find the true film thickness for the thicker film layer. For the case where there are thicker film layers in the sample to be tested, a multi-group fitting strategy can be adopted. First, multiple initial film thickness parameters of the thicker film layer are given, and each initial film thickness parameter and other structural parameters constitute a group of initial structural parameters; then, these different groups of initial structural parameters are fitted separately to obtain different groups of local optimal solutions, and the best results among these solutions are taken as the final model attributes. This multi-group fitting strategy can effectively improve the fitting accuracy of the thicker film layer for the thicker film layer, but this strategy cannot be effectively used to improve the fitting accuracy of the thinner film layer.

[0006] In actual situations, the samples to be tested on the surface of semiconductor devices include both thicker film layers and thinner film layers. Therefore, neither the multiple fitting strategy nor the multiple group fitting strategy can obtain high-precision fitting results for all film layers of the actual sample model.

[0007] In order to solve the above-mentioned problems existing in the prior art, the art is in urgent need of a film thickness measurement technology that can simultaneously fit the thicker film layers and thinner film layers in the stacked thin films on the surface of semiconductor devices to obtain high-precision fitting results, thereby improving the fitting accuracy of the actual stacked film sample model. Summary of the invention

[0008] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or decisive elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0009] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for measuring film thickness, a system for measuring film thickness, and a computer-readable storage medium, which can simultaneously fit thicker film layers and thinner film layers in the stacked films on the surface of semiconductor devices to obtain high-precision fitting results, thereby improving the fitting accuracy of the actual stacked film sample model.

[0010] Specifically, the above-mentioned film thickness measurement method provided according to the first aspect of the present invention comprises the following steps: obtaining the estimated thickness of multiple film layers on the surface of the semiconductor device to be measured, and constructing a candidate thickness set for the thick film layer that reaches a preset upper thickness limit; constructing a first structural model of the semiconductor device according to the estimated thickness of the thick film layer and the standard film layer whose estimated thickness is between the upper thickness limit and the preset lower thickness limit; providing detection light to the semiconductor device and collecting the actual spectral information of the reflected light; calculating the local optimal solution of the first structural model for each candidate thickness according to the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness in the candidate thickness set, and determining the measured thickness of the thick film layer according to the optimal one; incorporating the estimated thickness of the thin film layer whose estimated thickness is less than the lower thickness limit among the multiple film layers into the optimal first structural model to construct a second structural model; and calculating the overall optimal solution of the measured thickness of the multiple film layers according to the actual spectral information and the second theoretical spectral information corresponding to the second structural model. …。

[0011] In addition, the film thickness measurement system provided according to the second aspect of the present invention includes a memory and a processor. The processor is connected to the memory and is configured to implement the film thickness measurement method provided by the first aspect of the present invention.

[0012] In addition, according to the third aspect of the present invention, there is also provided a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed by a processor, the above-mentioned film thickness measurement method provided by the first aspect of the present invention is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.

[0014] Figure 1 A schematic diagram showing an initial structure of stacked film layers on a semiconductor device surface;

[0015] Figure 2A structural block diagram of a film thickness measurement system provided according to some embodiments of the present invention is shown;

[0016] Figure 3 A flow chart of a method for measuring film thickness according to some embodiments of the present invention is shown;

[0017] Figure 4 A flow chart showing a method for measuring film thickness according to other embodiments of the present invention is shown;

[0018] Figure 5 A schematic diagram showing the goodness of fit results of different fitting methods provided according to an embodiment of the present invention; and

[0019] Figure 6 to Figure 9 Schematic diagrams of T1 results, T2 results, T3 results and T4 results respectively shown by different fitting methods provided according to an embodiment of the present invention.

[0020] Reference numerals:

[0021] 100 samples to be tested;

[0022] 110 semiconductor devices;

[0023] More than 120 membrane layers;

[0024] 200 Film thickness measurement system;

[0025] 210 Memory;

[0026] 220 processors;

[0027] Steps S310 to S360; and

[0028] Steps S410 to S470. DETAILED DESCRIPTION

[0029] The following specific embodiments illustrate the implementation of the present invention, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Although the description of the present invention will be introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to this implementation. On the contrary, the purpose of introducing the invention in conjunction with the implementation is to cover other options or modifications that may be extended based on the claims of the present invention. In order to provide a deep understanding of the present invention, the following description will include many specific details. The present invention can also be implemented without using these details. In addition, in order to avoid confusion or blurring the focus of the present invention, some specific details will be omitted in the description.

[0030] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0031] In addition, the terms "upper", "lower", "left", "right", "top", "bottom", "horizontal" and "vertical" used in the following description should be understood as the directions shown in the paragraph and the related drawings. Such relative terms are only used for the convenience of description and do not mean that the device described therein must be manufactured or operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0032] It is understood that although the terms "first", "second", "third", etc. may be used herein to describe various components, regions, layers and / or parts, these components, regions, layers and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers and / or parts. Therefore, the first component, region, layer and / or part discussed below may be referred to as a second component, region, layer and / or part without departing from some embodiments of the present invention.

[0033] As mentioned above, in actual situations, the samples to be tested on the surface of semiconductor devices include both thicker film layers and thinner film layers. Therefore, neither the multiple fitting strategy nor the multiple group fitting strategy can obtain high-precision fitting results for all film layers of the actual sample model.

[0034] In order to solve the above-mentioned problems existing in the prior art, the present invention provides a method for measuring film thickness, a system for measuring film thickness, and a computer-readable storage medium, which can simultaneously fit thicker film layers and thinner film layers in stacked films on the surface of semiconductor devices to obtain high-precision fitting results, thereby improving the fitting accuracy of the actual stacked film sample model.

[0035] Please see Figure 2 , Figure 2 FIG. 4 is a block diagram showing a structure of a film thickness measurement system according to some embodiments of the present invention. Figure 2 As shown, in some non-limiting embodiments, the film thickness measurement method provided by the first aspect of the present invention can be implemented in the film thickness measurement system 200 provided by the second aspect of the present invention.

[0036] Specifically, if Figure 2As shown, in some non-limiting embodiments, the film thickness measurement system 200 may include a memory 210 and a processor 220. The memory 210 stores computer instructions. The processor 220 may be connected to the memory 210 and configured to execute the computer instructions stored in the memory 210, so as to implement the above-mentioned film thickness measurement method provided by the first aspect of the present invention.

[0037] Furthermore, in some optional embodiments, the film thickness measurement system 200 may also include: a detection light source and a spectrometer (not shown in the drawings). The detection light source may be used to provide polarized detection light to the surface of the semiconductor device 100 to be tested to form reflected light. The spectrometer may be used to collect reflected light emitted from the surface of the semiconductor device 100 to be tested and determine its actual spectral information.

[0038] Specifically, during the thin film measurement process, the thin film thickness measurement system 200 can first collect actual spectral data from the sample to be measured, and then calculate the corresponding theoretical spectral data based on the optical system parameters and the preset model parameters. Finally, by comparing and fitting the actual spectral data with the theoretical spectral data, the measurement device can accurately obtain the model parameters of the sample to be measured. Here, the specific measurement process does not involve the technical improvement of the present invention. Those skilled in the art can refer to the prior art Solis-Tinoco, V. A. Cevedo-Barrera, A. Vazquez-Estrada, O. Munguia-Cervantes, J. Hernandez-Como, N. Olguin, Luis F. Garcia-Valenzuela, A. Fast and accurate optical determination of gold-nanofilms thickness [J]. Optics & Laser Technology, 2021, 134 (1). DOI: 10.1016 / j. optlastec.2020.106604. to understand the specific implementation process, so it will not be repeated. This prior art shows that in the process of measuring the gold film, the corresponding film layer information can be obtained by collecting spectral data and fitting the theoretical spectrum with the collected spectrum.

[0039] In addition, optionally, the film thickness measurement system 200 of the present invention can also support model fitting of the sample to be measured by different types of spectral ellipsometric measurement machines.

[0040] The following will describe the steps of the above-mentioned film thickness measurement method in conjunction with some embodiments of the film thickness measurement system 200. Those skilled in the art will understand that these embodiments of the film thickness measurement system 200 are only some non-limiting implementation methods provided by the present invention, which are intended to clearly demonstrate the main concept of the present invention and provide some specific solutions that are convenient for the public to implement, and do not constitute a restriction on the implementation subject of each step in these film thickness measurement methods. Similarly, the film thickness measurement method also only provides some specific solutions that are convenient for the public to implement, and is not used to limit the entire working mode or all functions of the film thickness measurement system 200.

[0041] Please see Figure 3 , Figure 3 A flow chart of a method for measuring film thickness provided according to some embodiments of the present invention is shown.

[0042] like Figure 3 As shown, in some embodiments of the present invention, the film thickness measurement method may include the following steps: First, step S310 may be performed: obtaining the estimated thickness of multiple film layers on the surface of the semiconductor device to be measured, and constructing a candidate thickness set for the thick film layers that reach a preset upper thickness limit.

[0043] Specifically, it can be combined Figure 1 It is commonly understood that the sample 100 to be tested can be placed on a spectroscopic ellipsometric measurement platform, and the sample 100 to be tested can include a semiconductor device 110 as a substrate, and a plurality of film layers 120. Figure 1 As shown, the sample 100 to be tested may include a substrate and n thin films thereon, wherein T i represents the thickness of the i-th film from top to bottom, N i and K i represents the material parameters of the i-th film from top to bottom, for example, it may include the refractive index and extinction coefficient, and 1≤i≤n. Furthermore, the materials of adjacent films may be different, that is, N i and K i N of the adjacent layers may not be i-1 and K i-1 , or N i+1 and K i+1 equal.

[0044] For the above Figure 1 The sample 100 to be tested, which includes both a relatively thick film layer and a relatively thin film layer, can be set with multiple groups of initial structural parameters.

[0045] Specifically, we can look at Figure 4 Common understanding, Figure 4A flow chart of a method for measuring film thickness according to some other embodiments of the present invention is shown.

[0046] like Figure 4 As shown, in some embodiments, step S410 and step S420 may be performed, wherein step S410 is to specify that the thinner film layer material does not participate in the fitting. Step S420 includes setting M initial film thickness parameters for the thicker film layer.

[0047] Specifically, the estimated thickness of the plurality of film layers 120 on the surface of the semiconductor device 110 to be tested may be obtained first. Optionally, the estimated thickness may be a target film thickness parameter {T 01 ,T 02 ,…,T 0n Alternatively, the estimated thickness may also be the film thickness parameters {T1′, T2′, . . . , T n ′}.

[0048] Furthermore, in the sample 100 to be tested, the i-th film layer can be preset as a thinner film layer, and the j-th film layer can be preset as a thicker film layer, that is, the film thickness parameter T of the j-th film layer is j It is understandable that the "i-th layer" and "j-th layer" herein do not refer to only one layer, but may include several thinner layers and thicker layers. A thickness upper limit is preset, for example, the preset thickness upper limit may be Since the thinner layers in the plurality of film layers 120 cannot reach this upper limit of thickness, the i-th film layer may not be involved in the fitting, that is, the film thickness parameter T i It does not need to be included in the fitting. The thicker film layer can reach this upper thickness limit. Therefore, based on these thick film layers that reach the preset upper thickness limit, the film thickness parameter T j Set multiple initial values, let T j ={T j1 ,T j2 ,…,T jM}, and {T j1 ,T j2 ,…,T jM} can be used as a candidate thickness set, where M is the film thickness parameter T j The initial value of .

[0049] Afterwards, please continue to return to Figure 3 As shown, step S320 may be performed: constructing a first structural model of the semiconductor device according to the estimated thickness of the thick film layer and the standard film layer whose estimated thickness is between an upper thickness limit and a preset lower thickness limit.

[0050] Specifically, if Figure 4As shown, step S430 may be performed: other parameters of the initial structure involved in fitting may be combined with the obtained M initial film thickness parameters to form M groups of different initial structure parameters.

[0051] In some embodiments, a lower thickness limit may be preset. For example, the lower thickness limit may be According to the above thick film layer and the estimated thickness of the standard film layer with an estimated thickness between the above upper and lower thickness limits, M groups of initial structural parameters can be obtained to construct a first structural model of the semiconductor device. The first structural model can be represented as {T1, T2, …, T i-1 ,T i+1 ,…,T jm ,…,T n}, where T is the film thickness, i is the serial number of the thin film layer, j is the serial number of the thick film layer, n is the total number of film layers on the surface of the semiconductor device 110, T jm Can be selected from the candidate thickness set {T j1 ,T j2 ,…,T jM}, m is the serial number of the candidate thickness, and M is the total number of candidate thicknesses in the candidate thickness set, so 1≤m≤M.

[0052] Then, if Figure 3 As shown, step S330 may be performed: providing detection light to the semiconductor device, and collecting actual spectrum information of the reflected light.

[0053] Specifically, in some embodiments, the detection light source in the film thickness measurement system 200 can provide polarized detection light to the surface of the semiconductor device 110 to be tested to form reflected light. Then, the reflected light emitted from the surface of the semiconductor device 110 to be tested can be collected by a spectrometer to determine its actual spectrum information.

[0054] Continue as Figure 3 As shown, step S340 can be executed afterwards: according to the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness in the candidate thickness set, the local optimal solution of the first structural model for each candidate thickness is calculated respectively, and the measured thickness of the thick film layer is determined according to the optimal one.

[0055] Specifically, in some preferred embodiments, when using optical measurement equipment to analyze the properties of thin film samples on the semiconductor device 110 to be tested, two major types of parameters need to be comprehensively considered: the first type is the parameters of the optical system itself, including the incident light cone angle (AOI), numerical aperture (NA), polarization angle (P) and analyzer angle (A), etc.; the second type is the model parameters based on the sample to be tested, including film thickness parameters (T) and model material parameters (NK), etc.

[0056] In this regard, before respectively calculating the local optimal solution of the first structural model for each candidate thickness based on the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness in the candidate thickness set, the thin film thickness measurement method may further include the following steps.

[0057] You can first prepare a number of multi-layer calibration samples with known film thickness. Then, you can provide detection light to each multi-layer calibration sample to obtain the calibration spectrum information of its reflected light. After that, you can build and train a regression analysis model of theoretical spectrum information-multi-layer thickness based on the calibration spectrum information of each multi-layer calibration sample and the thickness of each film layer:

[0058] Spec Theory (T1,T2,…,T n )=F1(T1,T2,…,T n )

[0059] Among them, {T1,T2,…,T n} is the film thickness of multiple film layers of the multi-film layer calibration sample, n is the total number of film layers of the multi-film layer calibration sample, Spec Theory (T1,T2,…,T n ) is the theoretical spectral information corresponding to the multi-layer calibration sample, and F1(·) is the transfer function of the regression analysis model.

[0060] Furthermore, in some optional embodiments, a regression analysis model of theoretical spectral information-multi-film layer thickness can be constructed and trained based on the calibration spectral information of each multi-film layer calibration sample, as well as the thickness and material parameters of each film layer:

[0061] Spec Theory (T1,T2,…,T n )=F2(T1,T2,…,T n ,NK1,NK2,…,NK n )

[0062] Among them, F2(·) is the transfer function of the regression analysis model, {NK1,NK2,…,NK n} are material parameters of each film layer, for example, may include refractive index N and / or extinction coefficient K.

[0063] Furthermore, in some other optional embodiments, a regression analysis model of theoretical spectral information-multi-film layer thickness can be constructed and trained based on the optical parameters of the measurement system that collects the actual spectral information, the calibration spectral information of each multi-film layer calibration sample, and the thickness and material parameters of each film layer:

[0064] Spec Theory (T1,T2,…,Tn )=F3(T1,T2,…,T n ,NK1,NK2,…,NK n ,SysParam)

[0065] Wherein, F3(·) is the transfer function of the regression analysis model, and SysParam is the optical parameter of the measurement system, for example, it may include at least one of the incident light cone angle, numerical aperture, polarization angle, and analysis angle.

[0066] Then, if Figure 4 As shown, step S440 may be executed to perform fitting based on these M groups of initial structural parameters to obtain a group of locally optimal model parameters.

[0067] Specifically, for each set of initial structural parameters in the first structural model of the above-mentioned semiconductor device, a first objective function that minimizes the difference between actual spectral information and first theoretical spectral information can be constructed, and the Levenberg-Marquardt (LM) optimization algorithm can be used to solve the first objective function, thereby obtaining the local optimal solution of each corresponding candidate thickness.

[0068] The objective equation of the LM optimization algorithm, that is, the functional form of the first objective function, can be shown as follows:

[0069] min{Func object1}=|Spec Meas -Spec Theory (T 1,Fit ,T 2,Fit ,…,T i-1,Fit ,T i+1,Fit ,…,T jm,Fit ,…,T n,Fit )|

[0070] In the functional formula of the first objective function above, Func object1 represents the value of the first objective function, Spec Meas represents the actual spectral information, {T 1,Fit ,T 2,Fit ,…,T i-1,Fit ,T i+1,Fit ,…,T jm,Fit ,…,T n,Fit} represents the first structural model with respect to the candidate thickness T jm The local optimal solution of is the film thickness parameter after fitting. Theory (T 1,Fit ,T 2,Fit ,…,T i-1,Fit ,T i+1,Fit ,…,T jm,Fit ,…,Tn,Fit ) represents the first theoretical spectrum information obtained under the local optimal solution, that is, the theoretical spectrum of the sample to be tested obtained under the fitted film thickness parameters, and |·| represents the absolute value operation.

[0071] Then, the candidate thicknesses T can be compared jm The objective function value corresponding to the local optimal solution of , and take the minimum value as the optimal one. Then, the candidate thickness T corresponding to the optimal one j,Fit , determine the measured thickness of the thick film layer.

[0072] Specifically, in some optional embodiments, through the functional formula of the first objective function, the LM optimization algorithm can make the first objective function Func object1 The value of is as close to 0 as possible, so as to ensure that the fitting spectrum of the sample to be tested is consistent with the collected spectrum to the greatest extent.

[0073] In addition, by applying the LM optimization algorithm to each set of initial structural parameters in the above step S320, M sets of corresponding fitted film thickness parameters T can be obtained. Then, in these M sets of fitting results, the first objective function Func is selected object1 The set of fitting results with the smallest value is used as the film thickness parameter T i A set of local optimal solutions obtained when not participating in fitting.

[0074] Continue as Figure 3 As shown, step S350 may be performed afterwards: the estimated thickness of the thin film layer whose estimated thickness is less than the lower thickness limit among the multiple film layers is incorporated into the optimal first structural model to construct the second structural model.

[0075] Specifically, in order to overcome the problem of insufficient fitting accuracy of the film thickness parameter of the i-th thinner layer, the film thickness parameter T is specified in step S310. i It does not participate in the fitting, and after the fitting of steps S320 to S340, the film thickness parameter T is obtained. i When not involved in fitting, the obtained set of local optimal solutions {T 1,Fit ,T 2,Fit ,…,T i-1,Fit ,T i+1,Fit ,…,T j,Fit ,…,T n,Fit}.

[0076] like Figure 4 As shown, in order to further optimize the film thickness parameters and improve the spectrum fitting results, step S450 can be performed to convert the film thickness parameters T of the thinner layer that was not previously included in the fitting into iIn other words, the estimated thickness of the thin film layer whose estimated thickness is less than the lower limit of the thickness among the multiple film layers can be included in the optimal first structural model, so that step S460 can be executed to obtain a new set of structural parameters to be fitted {T 1,Fit ,T 2,Fit ,…,T i-1,Fit ,T i ,T i+1,Fit ,…,T j,Fit ,…,T n,Fit}, as the second structural model. In the second structural model, the film thickness parameter T i The parameters in the initial structure may be kept unchanged, and the other parameters may be the results obtained by fitting in steps S320 to S340.

[0077] Finally, continue back to Figure 3 As shown, step S360 may be executed: calculating an overall optimal solution for the measured thicknesses of the plurality of film layers according to the actual spectral information and the second theoretical spectral information corresponding to the second structural model.

[0078] Specifically, in some embodiments, for the structural parameters to be fitted in the second structural model reset in the above step S350, a second objective function that minimizes the difference between the actual spectral information and the second theoretical spectral information can be constructed and continued to be used with the LM optimization algorithm, so that step S470 can be performed to fit the final local optimal solution as the final model attribute of the sample to be tested. At this time, the objective equation of the LM optimization algorithm can be transformed from the functional formula of the first objective function to the functional formula of the second objective function, and the functional formula of the second objective function can be as follows:

[0079] min{Func object2}=|Spec Meas -Spec Theory (T′1, Fit ,T′2, Fit ,…,T′ i-1,Fit ,T i,Fit ,T′ i+1,Fit ,…,T′ j,Fit ,…,T′ n,Fit )|

[0080] In the functional formula of the second objective function above, Func object2 represents the value of the second objective function, {T′ 1,Fit ,T′ 2,Fit ,…,T′ i-1,Fit ,T i,Fit ,T′ i+1,Fit ,…,T′ j,Fit ,…,T′ n,Fit} represents the overall optimal solution of the second structural model, Spec Theory (T′ 1,Fit ,T′ 2,Fit ,…,T′ i-1,Fit ,T i,Fit ,T′ i+1,Fit ,…,T′ j,Fit ,…,T′ n,Fit ) represents the second theoretical spectrum information obtained under the overall optimal solution. {T′ 1,Fit ,T′ 2,Fit ,…,T′ i-1,Fit ,T i,Fit ,T′ i+1,Fit ,…,T′ j,Fit ,…,T′ n,Fit This set of film thickness parameters is the model properties finally fitted for the sample to be tested.

[0081] Through the above-mentioned embodiments of the present invention, the present invention adopts a new method to improve the fitting accuracy of the sample model. By deeply optimizing and upgrading the independent multiple fitting and multi-group fitting strategies in the calculation process, and adopting a distributed global fitting strategy, the model fitting accuracy of the samples to be tested including both thicker film layers and thinner film layers can be effectively improved, thereby improving the overall model fitting accuracy of the samples to be tested.

[0082] Next, a specific embodiment will be used in combination with a film thickness measurement system 200 to fully introduce the above film thickness measurement method and illustrate the beneficial effects.

[0083] Specifically, in one embodiment of the present invention, the plurality of film layers 120 on the surface of the semiconductor device 110 in the sample 100 to be tested includes 4 layers (n=4) of thin films. The expected film thickness parameters of the 4 layers are respectively The preset upper thickness limit can be The preset lower thickness limit can be In this embodiment, the first and second film layers are relatively thin, that is, T1 and T2 are relatively small and smaller than a preset lower limit of thickness, that is, and The third layer of film is thicker, that is, T3 is larger and greater than the preset upper limit of thickness, that is,

[0084] Then, these given preset film thickness parameters can be used as initial structural parameters, and actual spectral information can be collected at multiple locations (for example, 13 locations) on the surface of the semiconductor device 110 of the sample 100 to be tested, and used to independently apply the multiple fitting strategies in the prior art, the film thickness measurement method of multiple groups of fitting strategies, and the film thickness measurement method based on the step-by-step global fitting strategy provided by the present invention to the structural model of the sample 100 to be tested, so that the model properties of the sample 100 to be tested measured by the three different methods can be fitted. The fitting results are as follows:

[0085] Please see Figure 5 , Figure 5 A schematic diagram showing the goodness of fit results of different fitting methods provided according to an embodiment of the present invention is shown.

[0086] like Figure 5 As shown, in the test example of the thin film thickness measurement method using the multiple fitting strategy, for different actual spectral information, the goodness of fit (GOF) obtained by the multiple fitting strategy is generally too low, and the average GOF is about 0.6767.

[0087] GOF indicates the degree of fit between the theoretical spectrum and the actual spectrum. The closer the GOF value is to 1, the more consistent the two are. The GOF calculation method can be as follows:

[0088]

[0089] Among them, x i represents the data points in the actual spectrum, represents a data point in the theoretical spectrum, represents the mean of the data points in the theoretical spectrum, and n represents the number of data points in the actual spectrum.

[0090] Further, please refer to Figure 6 to Figure 9 , Figure 6 to Figure 9 Schematic diagrams of T1 results, T2 results, T3 results and T4 results respectively shown by different fitting methods provided according to an embodiment of the present invention.

[0091] like Figures 6 to 9As shown in the figure, in the test case of the thin film thickness measurement method using the multiple fitting strategy, the thickness of each film layer obtained by fitting can be found to have significant differences between the fitted parameters and the expected parameters of each film layer thickness. This is because T3 is thicker and its true thickness may cover a wider range of values. Therefore, in the measurement method of the multiple fitting strategy, starting from the expected parameters of T3, the local optimal parameters found for T3 are quite different from the true thickness of T3, resulting in insufficient accuracy of the fitted T3 parameters, and further resulting in significant differences between the fitted parameters and the expected parameters of other film layers.

[0092] Therefore, according to the test results of the above-mentioned film thickness measurement method using the multiple fitting strategy, when there is a thicker film layer in the sample 100 to be tested, the multiple fitting strategy may not be able to obtain a high-precision model fitting result for the sample 100 to be tested.

[0093] As shown in Figure 5, in the test example of the thin film thickness measurement method using multiple fitting strategies, for different actual spectral information, the GOF average value obtained by the multiple fitting strategies is relatively normal, and the GOF average value is about 0.9663.

[0094] And, continue as Figures 6 to 9 As shown in the test example of the thin film thickness measurement method using the multi-group fitting strategy, the thickness of each film layer is fitted, and it can be found that the fitted parameters of each film layer thickness are close to the expected parameters. However, for the thinner layers of the first and second layers, in the measurement method using the multi-group fitting strategy, the fitted parameters T1 and T2 have a high degree of dispersion, where the standard deviation σ of T1 is about 4.1563, and the standard deviation σ of T2 is about 4.5394.

[0095] The standard deviation σ can represent the degree of data dispersion. The closer σ is to 0, the smaller the degree of data dispersion is, and the higher the data stability and consistency is. The calculation method of σ can be as follows:

[0096]

[0097] Among them, y i represents a data point in a set of data, represents the mean of this set of data, and n represents the number of data points in this set of data.

[0098] Therefore, according to the results of the test example of the film thickness measurement method using the multi-group fitting strategy, although the multi-group fitting strategy can effectively obtain high-precision results for the thicker film layer T3, and on this basis effectively improve the accuracy of the fitting results of other film layers to obtain a normal GOF. However, for thinner film layers, the parameters after fitting by the multi-group fitting strategy are still relatively unstable and inconsistent, and the standard deviation of such thinner film layers is generally higher.

[0099] Next, please continue to return to Figure 5 In the test example of the thin film thickness measurement method using the above-mentioned step-by-step global fitting strategy provided by the present invention, for different actual spectral information, the GOF mean value obtained by the step-by-step global fitting strategy is relatively normal, and the GOF mean value is about 0.9663.

[0100] And, continue as Figures 6 to 9 As shown in the test example of the thin film thickness measurement method using the step-by-step global fitting strategy, the thickness of each film layer obtained by fitting can be found to be similar to the expected parameters. However, what is better than the above-mentioned multiple groups of fitting strategies is that for the first and second thin film layers, the discrete degree of the parameters T1 and T2 obtained by the step-by-step global fitting strategy is significantly reduced, among which the standard deviation σ of T1 is about 2.1513, and the standard deviation σ of T2 is about 2.7062.

[0101] According to Figures 5 to 9 From the fitting results obtained by the three different methods shown in , it can be found that, compared with the multiple fitting strategies, the step-by-step global fitting strategy can effectively obtain high-precision results for the thicker film layer T3, and effectively improve the accuracy of the fitting results of other film layers, thereby obtaining a normal GOF. Moreover, compared with the multi-group fitting strategy, for the thinner film layers T1 and T2, the step-by-step global fitting strategy can significantly reduce the discreteness of the parameters after fitting, reducing the standard deviation σ of T1 from about 4.1563 to about 2.1513, and reducing the standard deviation σ of T2 from about 4.5394 to about 2.7062, which means that the stability and consistency of the parameters after fitting are improved by 48.24% (for T1) and 40.38% (for T2), respectively, thereby effectively improving the accuracy of the model fitting results. It can be concluded that the above-mentioned film thickness measurement method based on the distributed global fitting strategy provided by the present invention can achieve the simultaneous response to the problem of insufficient model fitting accuracy in multiple scenarios (i.e., both thicker film layers and thinner film layers), thereby improving the model fitting accuracy in these specific cases.

[0102] Therefore, the above-mentioned film thickness measurement method provided by the present invention is to combine two independent multiple fitting strategies for different scenarios (for scenarios with smaller film thickness, the film thickness parameter ) and multiple fitting strategies (for scenes with thicker film layers, film thickness parameters ) is deeply optimized and upgraded to simultaneously address the problem of insufficient model fitting accuracy in these two scenarios, thereby improving the model fitting accuracy in these specific cases, thereby improving the overall fitting accuracy of the model of the samples to be tested.

[0103] In summary, the present invention provides a method for measuring film thickness, a system for measuring film thickness, and a computer-readable storage medium, which can simultaneously fit thicker film layers and thinner film layers in stacked films on the surface of semiconductor devices to obtain high-precision fitting results, thereby improving the fitting accuracy of the actual stacked film sample model.

[0104] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.

[0105] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for measuring film thickness, characterized in that: The following steps are involved: Obtaining estimated thicknesses of multiple film layers on the surface of the semiconductor device to be tested, and constructing a candidate thickness set for the thick film layers that reach a preset upper thickness limit; constructing a first structural model of the semiconductor device according to the estimated thickness of the thick film layer and a standard film layer whose estimated thickness is between the upper thickness limit and a preset lower thickness limit; Providing detection light to the semiconductor device and collecting actual spectrum information of the reflected light; According to the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness in the candidate thickness set, respectively calculating the local optimal solution of the first structural model with respect to each candidate thickness, and determining the measured thickness of the thick film layer according to the optimal one thereof; Incorporating the estimated thickness of the thin film layer whose estimated thickness is less than the lower thickness limit among the multiple film layers into the optimal first structural model to construct a second structural model; as well as An overall optimal solution for the measured thicknesses of the plurality of film layers is calculated based on the actual spectral information and the second theoretical spectral information corresponding to the second structural model.

2. The measuring method according to claim 1, characterized in that: The first structural model is represented as {T1, T2, ..., T i-1 , T i+1 , ..., T jm , ..., T n }, where T is the film thickness, i is the serial number of the thin film layer, j is the serial number of the thick film layer, n is the total number of film layers on the surface of the semiconductor device, T jm Selected from the candidate thickness set {T j1 , T j2 , ..., T jM }, m is the serial number of the candidate thickness, and M is the total number of the candidate thicknesses in the candidate thickness set.

3. The measuring method according to claim 2, characterized in that: The step of respectively calculating the local optimal solution of the first structural model with respect to each candidate thickness according to the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness in the candidate thickness set comprises: Construct and solve a first objective function that minimizes the difference between the actual spectral information and the first theoretical spectral information: min{Func object1 }=|Spec Meas -Spec Theory (T 1，Fit ,T 2，Fit ,...,T i-1，Fit ,T i+1，Fit ,...,T jm，Fit ,...,T n，Fit )| Among them, Func object1 represents the value of the first objective function, Spec Meαs represents the actual spectrum information, {T 1，Fit , T 2，Fit , ..., T i-1，Fit , T i+1，Fit , ..., T jm,Fit , ..., T n，Fit } represents the first structural model with respect to the candidate thickness T jm The local optimal solution of Spec Theory (T 1，Fit , T 2，Fit , ..., T i-1,Fit , T i+1,Fit , ..., T jm,Fit , ..., T n，Fit ) represents the first theoretical spectrum information obtained under the local optimal solution, and |.| represents the absolute value operation.

4. The measuring method according to claim 3, characterized in that: The step of determining the measured thickness of the thick film layer according to the best one thereof comprises: Compare the candidate thicknesses T jm The objective function value corresponding to the local optimal solution of , and take the minimum value as the optimal one; and The candidate thickness T corresponding to the optimal one j，Fit , determine the measured thickness of the thick film layer.

5. The measuring method according to claim 3, characterized in that: The second structural model is represented as {T 1，Fit , T 2，Fit , ..., T i-1，Fit , T i , T i+1，Fit , ..., T j，Fit , ..., T n,Fit }, the step of calculating the overall optimal solution of the measured thickness of the multiple film layers according to the actual spectral information and the second theoretical spectral information corresponding to the second structural model comprises: Construct and solve a second objective function that minimizes the difference between the actual spectral information and the second theoretical spectral information: min{Func object2 }=|Spec Meas -Spec Theory (T′ 1,Fit ,T′ 2,Fit ,...,T′ i-1，Fit ,T i,Fit ,T′ i+1,Fit ,...,T′ j，Fit ,...,T′ n,Fit )| Among them, Func object2 represents the value of the second objective function, {T′ 1，Fit , T′ 2，Fit , ..., T′ i-1,Fit , T i,Fit ,T′ i+1,Fit , ..., T′ j，Fit , ..., T′ n，Fit } represents the optimal solution of the second structural model, Spec Theory (T′ 1，Fit , T′ 2，Fit , ..., T′ i-1，Fit , T i，Fit , T′ i+1，Fit , ..., T′ j，Fit , ..., T′ n，Fit ) represents the second theoretical spectrum information obtained under the overall optimal solution.

6. The measuring method according to claim 1, wherein: Before respectively calculating the local optimal solution of the first structural model with respect to each candidate thickness according to the actual spectral information and the first theoretical spectral information corresponding to each candidate thickness in the candidate thickness set, the measurement method further includes the following steps: Prepare multiple multi-layer calibration samples with known film thickness; Providing the detection light to each of the multi-layer calibration samples respectively to obtain calibration spectrum information of the reflected light; and According to the calibration spectrum information of each of the multi-film layer calibration samples and the thickness of each of the film layers, a regression analysis model of theoretical spectrum information-multi-film layer thickness is constructed and trained: Spec Theory (T1,T2,...,T n )=F1(T1,T2,...,T n ) Among them, {T1, T2, ..., T n } is the film thickness of multiple film layers of the multi-film layer calibration sample, n is the total number of film layers of the multi-film layer calibration sample, Spec Theory (T1, T2, ..., T n ) is the theoretical spectral information corresponding to the multi-film layer calibration sample, and F1(·) is the transfer function of the regression analysis model.

7. The measuring method according to claim 6, characterized in that: The step of constructing and training a regression analysis model of theoretical spectral information-multi-film layer thickness based on the calibration spectral information of each of the multi-film layer calibration samples and the thickness of each of the film layers comprises: According to the calibration spectrum information of each of the multi-film layer calibration samples, as well as the thickness and material parameters of each of the film layers, a regression analysis model of theoretical spectrum information-multi-film layer thickness is constructed and trained: Spec Theory (T1,T2,...,T n )=F2(T1,T2,...,T n ,NK1,NK2,...,NK n ) Wherein, F2(·) is the transfer function of the regression analysis model, {NK1, NK2, ..., NK n } are the material parameters of each of the film layers.

8. The measuring method according to claim 7, characterized in that: The step of constructing and training a regression analysis model of theoretical spectral information-multi-film layer thickness based on the calibration spectral information of each of the multi-film layer calibration samples, as well as the thickness and material parameters of each of the film layers, comprises: According to the optical parameters of the measurement system for collecting the actual spectral information, the calibrated spectral information of each of the multi-layer calibration samples, and the thickness and material parameters of each of the film layers, a regression analysis model of theoretical spectral information-multi-layer thickness is constructed and trained: Spec Theory (T1,T2,...,T n )=F3(T1,T2,...,T n ,NK1,NK2,...,NK n ,SysParam) Wherein, F3(·) is the transfer function of the regression analysis model, and SysParam is the optical parameter of the measurement system, including at least one of the incident light cone angle, numerical aperture, polarization angle, and analysis angle.

9. A film thickness measurement system, characterized in that: include: a memory having computer instructions stored thereon; as well as A processor is connected to the memory and is configured to execute computer instructions stored in the memory to implement the thin film thickness measurement method according to any one of claims 1 to 8.

10. The measuring system according to claim 9, characterized in that: Also includes: A detection light source, used for providing polarized detection light to the surface of the semiconductor device to be tested to form reflected light; as well as The spectrometer is used to collect the reflected light emitted from the surface of the semiconductor device to be tested and determine its actual spectrum information.

11. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the thin film thickness measuring method according to any one of claims 1 to 8 is implemented.

Images