Film thickness measurement method, film thickness measurement device, and program

Abstract

This invention provides a method for measuring film thickness that allows for simple and highly accurate measurement of the film thickness of a thin film on a non-mirror-finish substrate. [Solution] A method for measuring film thickness, comprising the steps of: irradiating a thin film on a non-mirrored substrate with light and measuring the reflected light to obtain an actual reflection spectrum S; a first substep of setting an assumed film thickness value for the thin film and calculating a theoretical reflection spectrum M when the substrate is mirrored; a second substep of calculating a theoretical baseline spectrum C to extract the periodic component related to the film thickness from the theoretical reflection spectrum M and the actual reflection spectrum S; a third substep of calculating the difference between the actual reflection spectrum S and the theoretical baseline spectrum C, and the theoretical fringe spectrum which is the difference between the theoretical reflection spectrum M and the theoretical baseline spectrum C, performing these steps with the assumed film thickness value as a parameter, and optimizing the assumed film thickness value so that the difference between the difference spectrum SC and the theoretical fringe spectrum MC is minimized; and a film thickness value determination step.

Description

[Technical Field]

[0001] This invention relates to a method for measuring film thickness, a device for measuring film thickness, and a program. [Background technology]

[0002] Spectroscopic reflection is a widely known method for non-contact, non-destructively measuring the thickness of thin films formed on substrates in semiconductor device manufacturing processes. This method utilizes the interference of light reflected from the thin film, and calculates the film thickness by analyzing the resulting reflection spectrum.

[0003] However, conventional spectral reflectance methods assumed that the surface of the object being measured was an optically flat, mirror-like surface. Therefore, there was a problem in that measurements were difficult on non-mirror surfaces where fine irregularities were formed by planarization treatment, such as the back surface of a silicon wafer. Here, non-mirror surfaces include, but are not limited to, pear-skin surfaces, matte surfaces, ground surfaces, and lapped surfaces. On non-mirror surfaces, the amount of reflected light decreases significantly due to light scattering at the surface, and the amplitude of interference fringes is attenuated, with this effect being particularly pronounced on the short wavelength side. For this reason, theoretical models that assume a mirror surface could not accurately fit the measured spectrum, making it difficult to obtain reliable film thickness values.

[0004] To address this problem, a technique has been proposed to remove gradual fluctuation components (trend components) other than interference fringes from the measured spectrum (see, for example, Patent Document 1). Patent Document 1 attempts to improve the accuracy of the analysis by applying signal processing such as a low-pass filter to the measured interference spectrum to remove the trend component, and then performing a Fourier transform.

[0005] Furthermore, techniques for separating the influence of substrate surface roughness on film thickness evaluation are also known (see, for example, Patent Document 2). In Patent Document 2, the influence of surface roughness on film thickness is evaluated by removing the oxide film from the substrate under evaluation, forming an oxide film again on both the substrate and a reference substrate, and comparing the results. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Patent No. 6912045 [Patent Document 2] Japanese Patent Publication No. 2023-069889 [Overview of the project] [Problems that the invention aims to solve]

[0007] However, the technology described in Patent Document 1 removes the trend component by signal processing and is not directly based on the physical phenomena that occur on non-mirror surfaces. Therefore, it is not possible to accurately model the complex optical behavior unique to non-mirror surfaces, particularly the wavelength-dependent amplitude attenuation, and high-precision film thickness measurement remains difficult.

[0008] Furthermore, the technology described in Patent Document 2 requires destructive processes such as oxide film removal and re-deposition in order to evaluate the effect of surface roughness. Therefore, it has the problem of not being applicable to rapid, non-destructive in-line measurement required for product quality control, etc.

[0009] This invention has been made in view of the problems of the prior art described above, and aims to provide a method for measuring film thickness that can easily and accurately measure the film thickness even of thin films on substrates with non-mirror surfaces. [Means for solving the problem]

[0010] The present invention has been made to achieve the above objective, and provides a method for measuring the film thickness of a thin film formed on a substrate having a non-mirror surface, comprising: a step of irradiating the thin film with light and measuring the spectrum of the reflected light to obtain an actual reflection spectrum S; a first substep of setting an assumed film thickness value for the thin film and calculating a theoretical reflection spectrum M assuming the substrate is a mirror surface; a second substep of calculating a theoretical baseline spectrum C to extract the periodic components related to the film thickness from the theoretical reflection spectrum M and the actual reflection spectrum S; a third substep of calculating a difference spectrum SC between the actual reflection spectrum S and the theoretical baseline spectrum C, and a theoretical fringe spectrum MC which is the difference between the theoretical reflection spectrum M and the theoretical baseline spectrum C, performing the above steps with the assumed film thickness value as a parameter to optimize the assumed film thickness value so that the difference between the difference spectrum SC and the theoretical fringe spectrum MC is minimized; and a film thickness value determination step of determining the optimized assumed film thickness value as the film thickness value of the thin film. Note that the theoretical baseline spectrum C is due to the wavelength dispersion (wavelength dependence) of the refractive indices of the substrate and the thin film.

[0011] This method of measuring film thickness allows for simple and highly accurate measurement of film thickness.

[0012] In this case, the method for measuring film thickness can include minimizing the difference, which means minimizing the sum of the squares of the differences between the difference spectrum SC and the theoretical fringe spectrum MC.

[0013] This method of measuring film thickness allows for simpler and more accurate measurement.

[0014] In this case, the film thickness measurement method can be such that the theoretical baseline spectrum C is a non-periodic spectrum obtained by removing periodic interference components from the theoretical reflection spectrum M.

[0015] Thus, the theoretical baseline spectrum C is preferably a reference spectrum that approximates the aperiodic optical response obtained by removing the periodic vibration components caused by interference in the combination of the film and the substrate to be measured.

[0016] At this time, the film thickness measurement method can be such that the theoretical baseline spectrum C is the average value of the upper envelope spectrum connecting the maximum values of interference and the lower envelope spectrum connecting the minimum values of interference in the theoretical reflection spectrum M.

[0017] Thereby, the film thickness can be measured with higher accuracy.

[0018] At this time, the film thickness measurement method can be such that the upper envelope spectrum is calculated as the reflection spectrum from only the substrate assuming that the thin film does not exist.

[0019] Thereby, the film thickness can be measured with higher accuracy. <0000%094> At this time, the film thickness measurement method can be such that the lower envelope spectrum is calculated as the reflectance under the condition where interference is weakest using the optical constants of the atmosphere, the thin film, and the substrate.

[0021] Thereby, the film thickness can be measured with higher accuracy.

[0022] Also, the film thickness measurement method can be such that the theoretical baseline spectrum C is a value based on the interface reflection spectrum between the substrate and the thin film.

[0023] Thereby, the film thickness can be measured with higher accuracy.

[0024] In this case, the third substep involves calculating the corrected measured fringe spectrum S-αC by multiplying the theoretical baseline spectrum C by a first correction coefficient α (hereinafter referred to as the first correction coefficient α), and calculating β(MC) by multiplying the theoretical fringe spectrum MC by a second correction coefficient β (hereinafter referred to as the second correction coefficient β). In the optimization step, the assumed film thickness value, the first correction coefficient α, and the second correction coefficient β are optimized so that the difference F between the corrected measured fringe spectrum S-αC and β(MC) is minimized.

[0025] This allows for more accurate measurement of film thickness.

[0026] In this case, the second correction coefficient β can be expressed as a function that changes monotonically with respect to wavelength, and the second correction coefficient β can be expressed as a linear function of wavelength, and the film thickness can be measured in this manner.

[0027] This makes it possible to measure film thickness more easily.

[0028] In this case, the second correction coefficient β can be expressed as a film thickness measurement method represented by an exponential function of wavelength.

[0029] This allows for more accurate measurement of film thickness.

[0030] The present invention also provides a film thickness measuring device comprising: a light source; an irradiation optical system for irradiating a thin film on a substrate having a non-mirror surface with light from the light source; a light receiving optical system for receiving reflected light from the thin film; a spectrometer for spectrally analyzing the received reflected light to obtain a reflection spectrum; and a calculation processing unit for controlling the execution of the above-described film thickness measurement method.

[0031] Such a film thickness measuring device allows for the measurement of film thickness in a simple and highly accurate manner.

[0032] Furthermore, the present invention can provide a program for a computer to function as the calculation processing unit of the above-mentioned film thickness measuring device.

[0033] This allows us to provide a program that can measure film thickness easily and with high accuracy. [Effects of the Invention]

[0034] As described above, the film thickness measurement method of the present invention makes it possible to measure film thickness simply and with high accuracy. The film thickness measurement device of the present invention makes it possible to measure film thickness simply and with high accuracy. The program of the present invention provides the calculation processing unit of the film thickness measurement device of the present invention with high-precision and high-speed processing functions. [Brief explanation of the drawing]

[0035] [Figure 1] This is a diagram illustrating the processing procedure in the film thickness measurement method of the present invention. [Figure 2] This diagram illustrates the measured reflectance spectrum S and the theoretical reflectance spectrum M. [Figure 3] This diagram shows the change in the period of the fringe of the theoretical reflectance spectrum M when the assumed film thickness value is used as a parameter. [Figure 4] This is a diagram illustrating an example of a theoretical baseline spectrum C. [Figure 5] This is a diagram illustrating another example of the theoretical baseline spectrum C. [Figure 6] This diagram illustrates an example of calculating the corrected measured fringe spectrum S-αC. [Figure 7] This diagram illustrates an example of calculating β(MC) by multiplying the theoretical fringe spectrum by the coefficient β. [Figure 8] This is a diagram illustrating a film thickness measuring device according to the present invention. [Figure 9] This diagram shows the reflectance spectra of Example 1 and Comparative Example 1. [Figure 10] This diagram shows the reflectance spectra of Example 2 and Comparative Example 2. [Figure 11] This diagram shows the reflectance spectra of Example 3 and Comparative Example 3. [Modes for carrying out the invention]

[0036] The present invention will be described in detail below, but the present invention is not limited to these descriptions.

[0037] As described above, there was a need for a film thickness measurement method that could easily and accurately measure the film thickness even for thin films on substrates with non-mirror surfaces.

[0038] As a result of diligent study on the above problems, the present inventors have found that the film thickness of a thin film formed on a substrate having a non-mirror surface can be measured simply and with high accuracy by a film thickness measurement method that includes the steps of: irradiating the thin film with light and measuring the spectrum of the reflected light to obtain an actual reflection spectrum S; setting an assumed film thickness value for the thin film and calculating a theoretical reflection spectrum M assuming the substrate is a mirror surface; calculating a theoretical baseline spectrum C to extract the periodic component related to the film thickness from the theoretical reflection spectrum M and the actual reflection spectrum S; calculating a difference spectrum SC between the actual reflection spectrum S and the theoretical baseline spectrum C, and a theoretical fringe spectrum MC which is the difference between the theoretical reflection spectrum M and the theoretical baseline spectrum C, performing the above steps with the assumed film thickness value as a parameter to optimize the assumed film thickness value so that the difference between the difference spectrum SC and the theoretical fringe spectrum MC is minimized; and determining the film thickness value by determining the optimized assumed film thickness value as the film thickness value of the thin film.

[0039] The following explanation will use a specific example and refer to the diagram. The specific example is an example of measuring the film thickness of an SiO2 film formed on a non-mirror surface Si substrate.

[0040] (Method for measuring film thickness) First, the film thickness measurement method according to the present invention will be described. The object to be measured by the film thickness measurement method according to the present invention is a thin film formed on a substrate whose surface is not mirror-finish. The type of substrate is not particularly limited as long as the surface is not mirror-finish, for example, a matte surface, a ground surface, a lapped surface, etc. It can be any substrate that can support the film on its surface, such as a semiconductor substrate such as Si, SiC, or a compound semiconductor, a metal substrate, an insulating substrate, or a glass substrate. Furthermore, the type of thin film is not particularly limited as long as it is a film with uniform thickness and composition that can transmit measurement light. For example, it can be an insulating film such as SiO2 or SiN, a resist film, a semiconductor film, a coating protective film, etc. Furthermore, the roughness of the non-mirror-finish substrate surface is not particularly limited. For example, it can be applied to an anisotropic etched surface, a ground surface, a lapped surface, etc.

[0041] The film thickness measurement method according to the present invention is as shown in Figure 1, S1: Step to obtain the measured reflectance spectrum S. S2: Optimization step for assumed film thickness value S3: Film thickness value determination step It includes the following. Each step is explained in detail below.

[0042] (S1: Step to obtain the measured reflectance spectrum S) In this step (S1), the thin film to be measured is irradiated with light, and the spectrum of the reflected light is measured to obtain the measured reflection spectrum S. The light used for irradiation is not particularly limited. Examples of light used for irradiation include visible light, infrared light, and ultraviolet light. It is preferable to select light within an appropriate wavelength range depending on the type of thin film to be measured. Infrared light is particularly suitable when the film thickness is thick, and ultraviolet light is suitable when the film thickness is thin. The irradiation light may be incident perpendicularly or obliquely to the substrate. When measuring the entire surface of the substrate, oblique incidence is particularly preferable.

[0043] In step (S1) of obtaining such an experimentally measured reflection spectrum S, for example, the experimentally measured reflection spectrum S shown by the solid line in Figure 2 can be obtained. Figure 2 also shows the theoretical reflection spectrum M with synchronized interference periods, and the spectrum obtained by multiplying this theoretical reflection spectrum M by a correction factor of 0.5.

[0044] (S2: Optimization step for assumed film thickness value) In this step (S2), the following substeps are performed: a first substep to set an assumed film thickness value for the thin film and calculate the theoretical reflection spectrum M assuming the substrate is a mirror surface; a second substep to calculate the theoretical baseline spectrum C to extract the periodic component related to the film thickness from the theoretical reflection spectrum M and the measured reflection spectrum S; and a third substep to calculate the difference spectrum SC between the measured reflection spectrum S and the theoretical baseline spectrum C, and the theoretical fringe spectrum MC, which is the difference between the theoretical reflection spectrum M and the theoretical baseline spectrum C. These first to third substeps are performed using the assumed film thickness value as a parameter to optimize the assumed film thickness value so that the difference between the difference spectrum SC and the theoretical fringe spectrum MC is minimized. Each substep is described in detail below.

[0045] In the first substep, an assumed film thickness value for the thin film is set, and the theoretical reflection spectrum M is calculated assuming the substrate is a mirror surface, using known optical constants of the thin film and substrate (refractive index n(λ), extinction coefficient k(λ)). Specifically, in the case of normal incidence, the formula for calculating this theoretical reflection spectrum M is expressed as follows.

number

[0046] The symbols in the above formula are as follows: r1: Fresnel reflectance coefficient at the interface between the atmosphere and the thin film. r1 = (N0 - N1) / (N0 + N1) r2: Fresnel reflectance coefficient at the interface between the thin film and the substrate. r2=(N1-N S) / (N1+N S ) δ: Phase change of light in a thin film. δ = 2πN1d / λ N0: Complex refractive index of the atmosphere. Typically, N0 = 1 for air. N1: Complex refractive index of the thin film. N1 = n1 - ik1, where n1 is the refractive index and k1 is the extinction coefficient. N S : The complex refractive index of the substrate. N S =n S -ik S Here, n S is the refractive index, k S This represents the extinction coefficient. d: The assumed thickness of the thin film (assumed film thickness value). λ: Wavelength of light. i: Imaginary unit. By using the above formula to calculate the reflectance at each wavelength λ, the theoretical reflectance spectrum M(λ) can be obtained.

[0047] By changing the assumed film thickness value as a parameter, the period of the fringe in the theoretical reflection spectrum M changes, as shown in Figure 3. Using this, in the first substep, the assumed film thickness value can be changed to obtain an assumed film thickness value such that the period of the fringe in the theoretical reflection spectrum M is approximately equivalent to the period of the interference component (fringe) in the measured reflection spectrum S.

[0048] In the second substep, a theoretical baseline spectrum C is calculated to extract the periodic component related to the film thickness from the theoretical reflection spectrum M and the measured reflection spectrum S. This theoretical baseline spectrum C is a reference spectrum that approximates the non-periodic optical response for the combination of the film and substrate being measured, by removing the periodic vibrational component caused by interference. It is a reference spectrum used to separate the film thickness-dependent periodic interference component (fringe) included in the spectrum from the non-periodic component.

[0049] As shown in FIG. 4, the theoretical baseline spectrum C can be taken as the average value of the upper envelope spectrum connecting the maxima of the interference in the theoretical reflection spectrum M and the lower envelope spectrum connecting the minima of the interference. This makes it possible to accurately capture the physical vibration center of the interference fringes and enables fringe extraction with extremely high precision.

[0050] When it is assumed that the absorption of the thin film can be ignored, the upper envelope spectrum can be calculated as the reflection spectrum from only the substrate assuming that the thin film does not exist. Specifically, it can be calculated as follows.

[0051] The reflectance (upper envelope) R giving the maximum of the interference max (λ) corresponds to the reflectance of only the substrate assuming that the thin film does not exist and is given by the following equation.

Equation

[0052] Also, the lower envelope spectrum can be calculated as the reflectance under the condition where the interference is weakest using the optical constants of the atmosphere, thin film, and substrate. Specifically, it can be calculated as follows.

[0053] When it is assumed that the absorption of the thin film can be ignored, the reflectance (lower envelope) R giving the minimum of the interference min (λ) is given by the following equation considering the optical constants of the thin film.

Equation

[0054] Furthermore, the theoretical baseline spectrum C can also be a value based on the interface reflection spectrum between the substrate and the thin film. Figure 5 shows a comparison of the case where the theoretical baseline spectrum C is the average of the upper and lower envelope spectra described above (A), and the case where it is a value based on the interface reflection spectrum between the substrate and the thin film (B). As shown in Figure 5, the trends of both are similar, and it can be seen that both can be used as the theoretical baseline spectrum C. The lower spectrum in Figure 5 is obtained by subtracting the theoretical baseline spectrum C from (A) or (B) above from the upper spectral curve. The interfacial reflection spectrum R between the thin film (refractive index: n1) and the substrate (refractive index: n1) can be calculated using the following formula.

number

[0055] In the third substep, the difference spectrum SC (Figure 6) between the measured reflectance spectrum S and the theoretical baseline spectrum C, and the theoretical fringe spectrum MC, which is the difference between the theoretical reflectance spectrum M and the theoretical baseline spectrum C, are calculated (Figure 7). Figure 6 shows an example of the calculation of the difference spectrum SC, where C is the theoretical base spectrum, S is the measured spectrum, αC is the theoretical spectrum multiplied by an appropriate coefficient α, and S-αC is the corrected measured fringe spectrum obtained by taking the difference between them. Figure 7 shows an example of the calculation of the theoretical fringe spectrum MC, where M is the theoretical spectrum, C is the theoretical base spectrum, MC is the difference between them, and β(MC) is the difference multiplied by an appropriate coefficient β.

[0056] Thus, in order to correct the influence of a non-mirrored substrate with higher precision, it is preferable to calculate the corrected measured fringe spectrum S-αC by multiplying the theoretical baseline spectrum C by a first correction coefficient α, and to calculate β(MC) by multiplying the theoretical fringe spectrum MC by a second correction coefficient β. This corrected measured fringe spectrum S-αC is a calculation that extracts the fringe from the measured spectrum. The coefficient α plays a role in correcting the decrease in light intensity at non-mirrored surfaces. The first correction coefficient α mainly plays a role in correcting the decrease in the overall amount of reflected light due to light scattering at the non-mirrored substrate surface. This coefficient allows the baseline of the corrected measured fringe spectrum (S-αC) extracted from the measured reflectance spectrum S to be aligned with the theoretical standard.

[0057] Light reflection on non-mirrored surfaces is characterized by two main physical phenomena. First, there is light scattering due to surface irregularities, which reduces the overall intensity of reflected light. The first correction coefficient α compensates for this effect. Second, there is a phenomenon of attenuation of the amplitude of interference fringes. In particular, when the period of surface roughness is shorter than the wavelength, the scattering effect is greater for shorter wavelength light, and the amplitude attenuation becomes more pronounced. The second correction coefficient β, defined as a function of wavelength, models this wavelength-dependent amplitude attenuation. By separating and modeling these two physical phenomena, this algorithm eliminates the effects of surface irregularities and roughness, making it possible to accurately extract only the periodic information that depends on the film thickness.

[0058] The above β(MC) is a calculation that extracts the fringe from the theoretical reflection spectrum and compresses its amplitude by the coefficient β (MC is approximately centered at 0). The coefficient β serves to compensate for the attenuation of interference amplitude on non-mirrored substrates.

[0059] As described above, it is preferable to obtain the corrected measured fringe spectra S-αC and β(MC), and then optimize the assumed film thickness value, the first correction coefficient α, and the second correction coefficient β so that the difference F, i.e., (S-αC)-β(MC), is minimized. By optimizing the assumed film thickness value in this way, the film thickness can be determined with higher accuracy. The second correction coefficient β mainly plays a role in modeling the attenuation of the amplitude of interference fringes due to the influence of non-mirrored surfaces. In particular, since this attenuation tends to depend on wavelength, by making β a function of wavelength (for example, a linear function or exponential function as described in claims 10 and 11), it becomes possible to reproduce the complex behavior of the measured reflection spectrum with extremely high accuracy.

[0060] Here, by rearranging the equation and highlighting the model equation implemented in the program, it can be transformed as follows: (M' is the model equation.) F=(S-αC)-β(MC) =S-αC-β(MC) =S-(αC-β(MC)) =S-M'

[0061] This model equation may have difficulty handling situations where the amplitude is smaller at shorter wavelengths, the extracted fringe slopes upward, or the center of the extracted fringe is not zero. To address these issues, it is possible to change β from a coefficient that takes a constant value across all wavelengths to a function that takes a larger value at longer wavelengths, such as a linear function or an exponential function. Thus, it is preferable that the second correction coefficient β be a value expressed by a function that changes monotonically with respect to wavelength, or a value expressed by a linear function of wavelength or an exponential function of wavelength.

[0062] As explained above, in the optimization step of the assumed film thickness value (S2), the first to third substeps described above are performed using the assumed film thickness value as a parameter, and the value is calculated based on the difference between the difference spectrum SC and the theoretical fringe spectrum MC. For example, the assumed film thickness value is optimized so that the difference is minimized. Minimizing the difference includes, but is not limited to, the case where the sum of squares of the differences is used as the evaluation function and that evaluation function takes a local minimum value. For example, minimizing the difference includes the case where the sum of squares of the differences between the difference spectrum SC and the theoretical fringe spectrum MC is minimized. This makes it possible to measure the film thickness more simply and with higher accuracy.

[0063] In particular, in the third substep, it is preferable to calculate the corrected measured fringe spectra S-αC and β(MC) using the theoretical baseline spectrum C obtained in the second substep, as shown in Figures 6 and 7. Then, it is preferable to optimize the assumed film thickness value, the first correction coefficient α, and the second correction coefficient β so that F=(S-αC)-β(MC) is minimized. This allows for more accurate determination of the film thickness.

[0064] (S3: Film thickness determination step) In this step (S3), the optimized assumed film thickness value described above is determined as the actual film thickness value. In this way, the film thickness can be measured simply and with high accuracy.

[0065] With the film thickness measurement method according to the present invention described above, even if the shape and intensity of the entire spectrum change significantly due to the non-mirrored surface of the substrate, it is possible to accurately fit the film thickness by focusing only on the periodicity of the fringe. As a result, the optimized assumed film thickness value can be determined as the film thickness value of the thin film to be obtained.

[0066] (Other embodiments) In the above embodiment, the film thickness was calculated by fitting the fringe component (S-αC) extracted from the measured spectrum with β(MC), which is the fringe component calculated from the theoretical model and adjusted by the coefficient β. However, the present invention is not limited to this. For example, the method of removing the trend component using the theoretical baseline spectrum C in the present invention can also be applied to film thickness measurement by Fourier analysis. Specifically, the spectrum αC, obtained by multiplying the theoretical baseline spectrum C by the coefficient α, is subtracted from the measured reflectance spectrum S obtained from the object to be measured to obtain a spectrum (S-αC) from which the trend component has been removed. Then, this spectrum from which the trend component has been removed may be subjected to Fourier analysis, and the film thickness may be calculated based on the position of the peak appearing in the obtained power spectrum. With this method, the trend component can be removed with higher precision than conventional methods, thus improving the accuracy of film thickness measurement by Fourier analysis.

[0067] (Film thickness measuring device) Next, the film thickness measuring device according to the present invention will be described. As shown in Figure 8, the film thickness measuring device 10 according to the present invention comprises a light source 1, an irradiation optical system 2 that irradiates a thin film 21 on a substrate 20 whose surface is not mirrored with incident light L1 from the light source 1, a light receiving optical system 3 that receives reflected light L2 from the thin film 21, a spectrometer 4 that spectrally analyzes the received reflected light L2 to obtain a reflection spectrum, and a calculation processing unit 5 that controls the device to execute the film thickness measuring method according to the present invention described above. It may also include a mounting table 6 on which the substrate 20 is placed, and a control unit 7 that controls each of the above components. The calculation processing unit 5 is composed of, for example, a personal computer, and can implement a program for executing the film thickness measuring method according to the present invention. Since such a film thickness measuring device is capable of executing the film thickness measuring method according to the present invention described above, it can measure film thickness simply and with high accuracy.

[0068] (program) Furthermore, the program according to the present invention is designed to make a computer function as the calculation processing unit of the aforementioned film thickness measuring device. This makes it possible to provide a program that can measure film thickness simply and with high accuracy. [Examples]

[0069] The present invention will be described in detail below with reference to examples, but this is not intended to limit the present invention.

[0070] The film thickness of a silicon oxide (SiO2) film with a target thickness of 540 nm, formed on the back surface (pearlescent surface) of a silicon wafer using a thermal oxidation method, was measured using the method of the present invention (Example 1) and a method based on a mirror model (Comparative Example 1). Figure 9 shows the reflection spectra. In the examples, the measured spectrum and the theoretical spectrum show good agreement and are therefore represented as a single curve. As shown in Example 1, it became possible to fit even small amplitude fringes. As a result, in Example 1, the measured film thickness value was 540 nm, and the value was obtained that matched the actual film thickness (target film thickness).

[0071] On the other hand, in Comparative Example 1, the theoretical spectrum was calculated using a method based on a mirror model. Specifically, this method involved determining the film thickness using the least squares method with the wavelength-wise difference between the theoretical spectrum R(λ) and the measured spectrum S(λ). Due to the small amplitude of the fringe and low reflectivity, the fitting was based on the reflectivity of a thin oxide film, resulting in a poor fit between the measured spectrum and the theoretical spectrum, as shown in Figure 9. As a result, the measured film thickness in Comparative Example 1 was 86 nm, which deviated significantly from the actual film thickness (target film thickness). In the example of a film thickness of 540 nm shown in Figure 9, the measured values ​​of α and β are as follows. α:0.52 β:-0.168235+0.000592λ

[0072] Similar to Example 1 and Comparative Example 1, the film thickness of silicon oxide (SiO2) films with target thicknesses of 900 nm and 2600 nm, formed on the back surface (pearlescent surface) of a silicon wafer using a thermal oxidation method, was measured using the method of the present invention (Examples 2 and 3) and the method based on a mirror model (Comparative Examples 2 and 3). Figure 10 shows the reflection spectra of Example 2 and Comparative Example 2 with a film thickness of 900 nm, and Figure 11 shows the reflection spectra of Example 3 and Comparative Example 3 with a film thickness of 2600 nm. As shown in Figures 10 and 11, it can be seen that with the method of the present invention, the fitting between the measured spectrum and the theoretical spectrum is good even for different film thicknesses, and values ​​that match the actual film thickness (target film thickness) can be obtained.

[0073] In the example of a film thickness of 900 nm shown in Figure 10, the measured values ​​of α and β are as follows. α:0.47 β: 0.012103 + 0.000475λ

[0074] Furthermore, in the example of a film thickness of 2600 nm shown in Figure 11, the measured values ​​of α and β are as follows. α:0.62 β:-0.235087+0.000677λ

[0075] Furthermore, in Example 2, it was confirmed that even when the second correction coefficient β was optimized as an exponential function of wavelength, the film thickness value could be calculated with substantially the same accuracy as when a linear function was used.

[0076] As described above, according to the embodiment of the present invention, the film thickness of a thin film formed on a substrate with a non-mirror surface could be measured accurately using a simple method.

[0077] This specification includes the following embodiments: [1]: A method for measuring the thickness of a thin film formed on a substrate whose surface is not mirror-finish, The steps include irradiating the thin film with light, measuring the spectrum of the reflected light to obtain an actual reflection spectrum S, A first substep involves setting an assumed film thickness value for the thin film and calculating the theoretical reflection spectrum M assuming the substrate is a mirror surface, A second substep involves calculating a theoretical baseline spectrum C to extract the periodic component related to the film thickness from the theoretical reflection spectrum M and the measured reflection spectrum S, A third substep involves calculating the difference spectrum SC between the measured reflection spectrum S and the theoretical baseline spectrum C, and the theoretical fringe spectrum MC, which is the difference between the theoretical reflection spectrum M and the theoretical baseline spectrum C. An optimization step of the assumed film thickness value, in which the assumed film thickness value is optimized so that the difference between the difference spectrum SC and the theoretical fringe spectrum MC is minimized, by performing the optimization using the assumed film thickness value as a parameter, A film thickness value determination step in which the optimized assumed film thickness value is determined as the film thickness value of the thin film, A method for measuring film thickness, characterized by including the following: [2]: The method for measuring film thickness according to [1] above, wherein the difference being minimized means that the sum of the squares of the differences between the difference spectrum SC and the theoretical fringe spectrum MC is minimized. [3]: The method for measuring film thickness according to [1], wherein the theoretical baseline spectrum C is a non-periodic spectrum obtained by removing periodic interference components from the theoretical reflection spectrum M. [4]: The method for measuring film thickness according to [1], wherein the theoretical baseline spectrum C is the average value of the upper envelope spectrum formed by connecting the maximum interference values ​​in the theoretical reflection spectrum M and the lower envelope spectrum formed by connecting the minimum interference values. [5]: The method for measuring film thickness according to [3] or [4] above, wherein the upper envelope spectrum is calculated as the reflection spectrum from the substrate alone, assuming that the thin film is not present. [6]: The method for measuring film thickness according to [3] or [4] above, wherein the lower envelope spectrum is calculated as the reflectance under conditions where interference is most destructive, using the optical constants of the atmosphere, the thin film, and the substrate. [7]: The method for measuring film thickness according to [1] or [2] above, wherein the theoretical baseline spectrum C is a value based on the interfacial reflection spectrum of the substrate and the thin film. [8]: In the third substep, the corrected measured fringe spectrum S-αC is calculated by multiplying the theoretical baseline spectrum C by a first correction coefficient α, and β(MC) is calculated by multiplying the theoretical fringe spectrum MC by a second correction coefficient β. The film thickness measurement method according to [1], [2], [3], [4], [5], [6], or [7] above, wherein the optimization step involves optimizing the assumed film thickness value, the first correction coefficient α, and the second correction coefficient β so that the difference F between the corrected measured fringe spectrum S-αC and β(MC) is minimized. [9]: The film thickness measurement method of [8] above, wherein the second correction coefficient β is a function that changes monotonically with respect to wavelength.

[10] The second correction coefficient β is expressed as a linear function of wavelength in the film thickness measurement method of [8] described above.

[11] The second correction coefficient β is expressed as an exponential function of wavelength in the film thickness measurement method of [8] described above.

[12] : A film thickness measuring device comprising: a light source; an irradiation optical system for irradiating a thin film on a substrate having a non-mirror surface with light from the light source; a light receiving optical system for receiving reflected light from the thin film; a spectrometer for spectrally analyzing the received reflected light to obtain a reflection spectrum; and a calculation processing unit for controlling the execution of the film thickness measuring method described in any one of [1] to

[11] above.

[13] : A program to cause the computer to function as the calculation processing unit for the film thickness measuring device described in

[12] above.

[0078] It should be noted that the present invention is not limited to the embodiments described above. The embodiments described above are illustrative, and any configuration that is substantially identical to the technical idea described in the claims of the present invention and achieves similar effects is included within the technical scope of the present invention. [Explanation of Symbols]

[0079] 1...Light source, 2...Irradiation optical system, 3...Reception optical system, 4...Spectroscope, 5...Calculation processing unit, 6...Mounting platform, 7...Control unit, 10...Film thickness measuring device, 20... Substrate, 21... Thin film. L1...incident light, L2...reflected light.

Claims

1. A method for measuring the film thickness of a thin film formed on a substrate whose surface is not mirror-finish, The steps include irradiating the thin film with light, measuring the spectrum of the reflected light to obtain an actual reflection spectrum S, A first substep involves setting an assumed film thickness value for the thin film and calculating the theoretical reflection spectrum M assuming the substrate is a mirror surface, A second substep involves calculating a theoretical baseline spectrum C to extract the periodic component related to the film thickness from the theoretical reflection spectrum M and the measured reflection spectrum S, A third substep involves calculating the difference spectrum S-C between the measured reflection spectrum S and the theoretical baseline spectrum C, and the theoretical fringe spectrum M-C, which is the difference between the theoretical reflection spectrum M and the theoretical baseline spectrum C. An optimization step of the assumed film thickness value, in which the assumed film thickness value is optimized so that the difference between the difference spectrum S-C and the theoretical fringe spectrum M-C is minimized, by performing the optimization using the assumed film thickness value as a parameter, A film thickness value determination step in which the optimized assumed film thickness value is determined as the film thickness value of the thin film, A method for measuring film thickness, characterized by including the following:

2. The method for measuring film thickness according to claim 1, wherein the minimum difference means that the sum of the squares of the differences between the difference spectrum S-C and the theoretical fringe spectrum M-C is minimized.

3. The method for measuring film thickness according to claim 1, characterized in that the theoretical baseline spectrum C is a non-periodic spectrum obtained by removing periodic interference components from the theoretical reflection spectrum M.

4. The method for measuring film thickness according to claim 1, characterized in that the theoretical baseline spectrum C is the average value of the upper envelope spectrum, which connects the maximum interference values ​​in the theoretical reflection spectrum M, and the lower envelope spectrum, which connects the minimum interference values.

5. The method for measuring film thickness according to claim 4, characterized in that the upper envelope spectrum is calculated as the reflection spectrum from the substrate alone, assuming that the thin film does not exist.

6. The method for measuring film thickness according to claim 4, characterized in that the lower envelope spectrum is calculated as the reflectance under conditions where interference is most destructive, using the optical constants of the atmosphere, the thin film, and the substrate.

7. The method for measuring film thickness according to claim 1, characterized in that the theoretical baseline spectrum C is a value based on the interface reflection spectrum of the substrate and the thin film.

8. In the third substep described above, the corrected measured fringe spectrum S-αC is calculated by multiplying the theoretical baseline spectrum C by a first correction coefficient α, and β(M-C) is calculated by multiplying the theoretical fringe spectrum M-C by a second correction coefficient β. The method for measuring film thickness according to claim 1, characterized in that, in the step of performing the optimization, the assumed film thickness value, the first correction coefficient α, and the second correction coefficient β are optimized so that the difference F between the corrected measured fringe spectrum S-αC and β(M-C) is minimized.

9. The method for measuring film thickness according to claim 8, characterized in that the second correction coefficient β is a function that changes monotonically with respect to wavelength.

10. The method for measuring film thickness according to claim 8, characterized in that the second correction coefficient β is expressed as a linear function of wavelength.

11. The method for measuring film thickness according to claim 8, characterized in that the second correction coefficient β is expressed as an exponential function of wavelength.

12. Light source and An illumination optical system that irradiates a thin film on a substrate whose surface is not mirror-like with light from the aforementioned light source, A light-receiving optical system that receives reflected light from the thin film, A spectrometer that spectrally analyzes the received reflected light to obtain a reflectance spectrum, A calculation processing unit that controls the execution of the film thickness measurement method described in any one of claims 1 to 11, A film thickness measuring device characterized by comprising the following features.

13. A program for causing a computer to function as the calculation processing unit of the film thickness measuring apparatus described in claim 12.

Images