JP6750793B2 - Film thickness measuring device and film thickness measuring method - Google Patents

Description

The present invention relates to a film thickness measuring device and a film thickness measuring method, and more particularly to a film thickness measuring device and a film thickness measuring method for measuring a film thickness using reflected light.

Photoresist formed by spray-coating an uneven structure such as MEMS (MEMS, Micro Electro Mechanical Systems), transparent conductive resin formed by screen-printing or slit-coating a film for display, on the surface of ceramic parts, blades, etc. The formed thin film such as a lubricant coat is likely to have a film thickness distribution locally and locally, and a measurement technique for improving uniformity is required.

Generally, the film thickness of a thin film can be measured by SEM (Scanning Electron Microscope) observation of a cut surface of a sample. As the nondestructive method, the film thickness can be measured using a conventional optical film thickness measuring method (white interference method, reflection spectroscopy, ellipsometry, etc.). However, in the conventional optical film thickness measuring method, the spatial resolution in measuring the film thickness distribution is not so high. Therefore, when the film thickness continuously increases or decreases from several nm to several μm in the region of several μm□ to several hundred μm□, evaluate the quality of the thin film only by the average value of arbitrary measurement points. I can't. In such a case, statistical processing such as creating a histogram for the film thickness distribution as shown in FIG. 11 and analyzing the average film thickness from the created histogram is necessary.

JP, 2009-204313, A JP, 2007-248312, A

Examples of conventional methods for measuring the thickness of a thin film include (1) reflection spectroscopy, (2) ellipsometry, (3) white fringe scanning, and (4) confocal microscopy. Further, for example, the film thickness can be measured by the method described in Patent Document 1. However, each of these methods has the following problems.

(1) The reflection spectroscopy method measures the spectrum by dispersing the reflected light of a sample illuminated with white light. Among the methods using reflection spectroscopy, the peak valley method (PV method) is a method of analyzing the film thickness by obtaining the wavelength corresponding to the maximum and minimum of the reflectance from the spectrum. With this method, when using visible light, it is not possible to obtain the maximum and minimum spectral patterns with a film thickness of about a wavelength (about 0.5 μm) or less. Therefore, it cannot be applied to the measurement of the film thickness below the wavelength.

In addition, among the methods by reflection spectroscopy, the curve fit method is based on the condition that the spectral spectrum is determined by the wavelength of light, the film thickness of the thin film, the refractive index (n) between the thin film and the substrate, and the extinction coefficient (k). , Theoretically calculated in the form of wavelength dependence of absolute reflectance. Then, the film thickness is determined by parameter fitting the theoretically calculated absolute reflectance to the measured data. Generally, the film thickness can be obtained by the curve fitting method from the thickness of several nm to several μm. However, a sufficient number of samplings is required for each wavelength of illumination light used for measurement.

A general spectrometer cannot measure the spatial distribution of film thickness with high resolution, and only obtains an average value of the area illuminated by light. If there is a film thickness distribution (change) in the illumination area, the reflectance will be averaged, and an accurate absolute reflectance cannot be obtained. As described above, the curve fit method cannot measure the film thickness distribution with high resolution.

(2) In the method using ellipsometry (ellipsometry), the film thickness is obtained by analyzing the reflected light. However, since it is necessary to make the illumination light obliquely incident on the sample, it is not possible to measure the spatial distribution of the film thickness with high resolution, and the average area (1 mm□ to 100 um□) illuminated by the light cannot be measured. Only get the value. Furthermore, when the thin film itself has polarization characteristics, it becomes difficult to measure the film thickness with high resolution.

(3) In the method of scanning white interference fringes, the maximum intensity position of the interference fringes of white light is calculated by z-scanning the sample or the objective lens. Thereby, the film thickness of the thin film in which the interference fringes are generated can be obtained. In order to measure the film thickness of the thin film, it is necessary to separately measure the interference intensity signals from the base substrate and the thin film surface. However, when the film thickness is 1 μm or less, it becomes difficult to separate those intensity signals. Furthermore, when a transparent thin film with a low reflectance is present on a substrate with a high reflectance, the interference intensity from the transparent thin film cannot be measured accurately, making it even more difficult to measure the film thickness. Becomes It becomes more difficult to measure the film thickness of a multilayer film or the like.

(4) In the method using a confocal microscope, the distance between the objective lens and the sample is changed by z-scanning, and the film thickness is measured by detecting the in-focus position. When the film thickness is several μm or less, the reflection signals of the base substrate and the film surface cannot be separated, so that the film thickness cannot be measured. Since the signal separation limit largely depends on the numerical aperture of the objective lens, the film thickness measurement limit is several μm at high magnification and tens of μm at low magnification and wide field of view, which makes it difficult to measure the film thickness.

The method of patent document 1 measures a film thickness by switching the light of a several wavelength using a confocal optical system. Specifically, the measurement data of the reflectance for each of the first wavelength and the second wavelength is obtained, and the calculated data in which the relationship between the wavelength and the reflectance is shown for each film thickness is referred to. The thickness is approximated and calculated. By applying the curve fit method to each pixel from the images of a plurality of wavelengths, the film thickness can be calculated and the film thickness distribution can be obtained. By doing so, a film thickness of 1 μm or less can be accurately obtained. However, in the method of Patent Document 1, the accuracy of curve fitting is reduced for a film thickness of 1 μm or more.

In the method of Patent Document 1, optical interference must occur in the thin film due to the measurement principle, but at least two conditions must be satisfied for that purpose. (i) The distance between the focal positions of the film surface and the substrate surface must be within the depth of focus (depth of focus). (ii) The path difference due to the film thickness must be sufficiently smaller than the coherence length (coherence).

The problem of the depth of focus causes a deviation from theoretical calculation as the numerical aperture (NA) of the objective lens increases, even when the separation of focal positions does not completely occur.

The problem of coherence is that the coherence length is limited by the bandwidth and center wavelength even in the case of point light source illumination. The narrower the band width is to several nm, the longer the coherence length becomes, but the illumination intensity becomes weaker, so there is a demerit that the imaging time needs to be lengthened. If a bandwidth of about 10 to 20 nm is secured to secure the brightness of an image, the coherence length becomes about 10 to 20 μm.

Generally, depth of focus and coherence are not taken into consideration when theoretically calculating the absolute reflectance from the Fresnel coefficient. If the wavelength resolution of the spectroscope is several nm or if it is pure monochromatic light such as laser light, it is not necessary to consider coherence in the measurement reflectance. However, when using monochromatic light having a certain bandwidth, it must be taken into consideration that the coherence is attenuated as the bandwidth is expanded.

As described above, in the method of Patent Document 1, the coherence of reflected light is reduced when an objective lens having a high numerical aperture is used and when the film thickness is 1 μm or more. The divergence becomes large, and the optimal solution may not be obtained. In order to increase the coherence length, there is a method of narrowing the bandwidth of the illumination light, but the image becomes dark, which is not preferable from the viewpoint of image pickup time and S/N maintenance.

The present invention has been made to solve such a problem, and analyzes the film thickness distribution of a thin film with high resolution even when an objective lens having a high numerical aperture is used and the film thickness is 1 μm or more. An object of the present invention is to provide a film thickness measuring device and a film thickness measuring method capable of performing the same.

A film thickness measuring device according to the present invention is a film thickness measuring device for measuring a film thickness of a thin film of a sample including a substrate and a thin film provided on the substrate, and is an illumination light of at least a first wavelength. And a second wavelength illumination light, a photodetector for detecting reflected light reflected by the sample to obtain an image in a predetermined region of the sample, and illumination from the light source unit. A confocal optical system that guides light to the sample and guides the reflected light from the sample to the photodetector, and a first image by the illumination light of the first wavelength in order to calculate the film thickness of the thin film. And a processing unit that respectively obtains measurement data of reflectance with respect to the first wavelength and the second wavelength based on the second image by the illumination light of the second wavelength. The relationship with the reflectance is referred to the calculation data respectively shown for each film thickness of the thin film, the film thickness of the thin film is calculated by approximation from the measurement data, the reflectance in the calculation data, The term of the non-interference component is included. With such a configuration, the film thickness distribution in a thin film having a film thickness of 1 μm or more can be analyzed with high resolution.

Further, the term of the non-interference component includes the sum of reflectances in each reflection up to a predetermined number of times due to the interface on the substrate side. With such a configuration, multiple reflection in the thin film can be reflected, so that the film thickness can be analyzed with high accuracy even for a thin film having a film thickness of 1 μm or more.

The reflectance in the calculation data includes the product of the term of the interference component and the coherence factor indicating the ratio of the term of the interference component. With such a configuration, it is possible to analyze the film thickness of the thin film with high accuracy even when the reflectance is partially coherent including the non-interference component and the interference component.

The confocal optical system includes an objective lens, and the reflectance in the calculation data includes correction based on the numerical aperture of the objective lens. With such a configuration, it is possible to analyze the film thickness of the thin film with high accuracy even when using an objective lens having a high numerical aperture.

Further, the thin film is a multi-layered film in which a plurality of films are laminated, and the processing unit synthesizes the reflectances shown for each film thickness of each of the plurality of films to obtain the calculated data in the calculation data. The reflectance. With such a configuration, it is possible to analyze the film thickness of a thin film made of a multilayer film with high accuracy.

A film thickness measuring method according to the present invention is a film thickness measuring method for measuring a film thickness of the thin film of a sample including a substrate and a thin film provided on the substrate, and a confocal optical system is used. Illuminating the sample with illumination light of the first wavelength and illumination light of the second wavelength by switching, irradiating the sample with reflected light reflected by the sample through the confocal optical system, Based on the first image and the second image, the first image based on the light and the second image based on the light of the second wavelength are acquired, and the first image based on the first image and the second image to calculate the film thickness of the thin film. The measurement data of the reflectance with respect to the wavelength and the second wavelength are respectively obtained, and with reference to the calculation data in which the relationship between the wavelength and the reflectance is shown for each film thickness of the thin film, the film of the thin film is obtained from the measurement data. The thickness is approximated and calculated, and the reflectance in the calculation data includes a term of a non-interference component. With such a configuration, the film thickness distribution in a thin film having a film thickness of 1 μm or more can be analyzed with high resolution.

Further, the term of the non-interference component includes the sum of reflectances in each reflection up to a predetermined number of times due to the interface on the substrate side. With such a configuration, multiple reflections in the thin film can be reflected, so that the film thickness can be analyzed with high accuracy even for a thin film having a film thickness of 1 μm or more.

Further, the reflectance in the calculation data includes a product of an interference component term and a coherence factor indicating a ratio of the interference component term. With such a configuration, it is possible to analyze the film thickness of the thin film with high accuracy even when the reflectance is partially coherent including the non-interference component and the interference component.

The confocal optical system includes an objective lens, and the reflectance in the calculation data includes correction based on the numerical aperture of the objective lens. With such a configuration, it is possible to analyze the film thickness of the thin film with high accuracy even when using an objective lens having a high numerical aperture.

The thin film is a multilayer film in which a plurality of films are laminated, and the reflectances shown for the respective film thicknesses of the plurality of films are combined to obtain the reflectance in the calculation data. With such a configuration, it is possible to analyze the film thickness of the thin film including the multilayer film with high accuracy.

According to the present invention, there is provided a film thickness measuring device and a film thickness measuring method capable of analyzing the film thickness distribution of a thin film with high resolution even when an objective lens having a high numerical aperture is used and the film thickness is 1 μm or more. Can be provided.

It is a block diagram which illustrated the film thickness measuring device which concerns on embodiment. 2 is an optical model illustrating a substrate on which a thin film according to the embodiment is formed. It is a figure which illustrated the optical path difference by refraction of the reflected light in the board|substrate with which the thin film which concerns on embodiment was formed. It is a figure which illustrated the multiple reflection without interference in the thin film which concerns on embodiment. It is a graph which illustrated the relationship between the wavelength and reflectance concerning an embodiment, and a horizontal axis shows wavelength and a vertical axis shows absolute reflectance. It is a graph which illustrated the relationship between the wavelength and reflectance concerning an embodiment, and a horizontal axis shows wavelength and a vertical axis shows absolute reflectance. It is a flowchart figure which illustrated the film thickness measuring method which concerns on embodiment. It is a graph which illustrated the relationship between the wavelength and reflectance concerning an embodiment, and a horizontal axis shows wavelength and a vertical axis shows absolute reflectance. 6 is a graph illustrating the relationship between the film thickness and the residual error of the thin film according to the embodiment, where the horizontal axis represents the thin film thickness and the vertical axis represents the residual error. It is a figure which illustrated the handling of the reflectance calculation with a zero film thickness in the case of perfect non-coherent according to the embodiment. It is a schematic diagram which illustrated the film thickness distribution of the thin film on a board|substrate.

(Embodiment)
The present invention utilizes reflection spectral film thickness measurement using the following four points for the purpose of measuring and evaluating the film thickness distribution.
(1) By using a confocal optical system, it is expected that the back surface reflected light of the transparent substrate can be removed and the net front surface reflected light can be accurately measured.
(2) Instead of splitting the reflected light as in the conventional spectroscopic film thickness meter, the reflection intensity of the sample is measured using monochromatic light as the illumination. Then, each time the illumination wavelength is changed, the reflection intensity is measured for the same visual field. Therefore, the wavelength dependence of reflectance, that is, the reflection spectrum can be obtained. At this time, the reflectance can be two-dimensionally imaged by imaging for each wavelength.
(3) By performing z-scanning using a confocal optical system, the surface shape can be measured three-dimensionally even though the surface shape includes an error of about film thickness. It is possible to estimate how much the measurement point is inclined with respect to the illumination optical axis. Therefore, it can be used also when correcting the reflectance change due to the inclination.
(4) By using a confocal optical system and an objective lens having a large numerical aperture NA in order to measure the film thickness distribution of a fine pattern, a reflection image can be captured even with a pattern width of several μm.

Hereinafter, a specific configuration of this embodiment will be described with reference to the drawings. The following description shows preferred embodiments of the present invention, and the scope of the present invention is not limited to the following embodiments. In the following description, the same reference numerals indicate substantially the same contents.

The configuration of the film thickness measuring device 100 according to this embodiment will be described. FIG. 1 is a diagram illustrating a configuration of a film thickness measuring device 100 according to the embodiment. As shown in FIG. 1, the film thickness measuring device 100 according to the present embodiment includes a light source 11, an interference filter 12, lenses 13a, 13b, 13c, a slit 14, a beam splitter 15, a vibrating mirror 16, an objective lens 17, and a stage. 18, a photodetector 19, and a processing device 20. The light source 11 and the interference filter 12 form the light source unit 10. The lenses 13a, 13b, 13c, the slit 14, the beam splitter 15, the oscillating mirror 16 and the photodetector 19 form a confocal optical system 101. A sample 30 is placed on the stage 18. The film thickness measuring device 100 measures the film thickness of the thin film of the sample 30 including the substrate and the thin film provided on the substrate.

As the light source 11, a white light source such as a mercury-xenon lamp containing a plurality of bright lines in a continuous spectrum is used. Note that, for example, a xenon lamp having a wide continuous spectrum from the ultraviolet region to the infrared region (185 nm to 2000 nm) may be used. Of course, the light source 11 is not limited to the xenon lamp, and a white diode, a white laser, or the like may be used. As will be described later, any light source may be used as long as the wavelength can be selected.

An optical system for observing the sample 30 with the light from the light source 11 will be described. The light emitted from the light source 11 passes through the interference filter 12 and is converted into light of a specific wavelength. As the interference filter 12, for example, a plurality of bandpass filters that selectively transmit light having a specific wavelength can be used. As a result, the plurality of single-wavelength illumination lights are selectively transmitted. For example, as the wavelength of the illumination light, it is possible to select wavelengths of 405 nm, 436 nm, 488 nm, 514 nm, 546 nm, 578 nm corresponding to the bright line of the mercury-xenon lamp, and 633 nm which is not the bright line of the mercury-xenon lamp. When using a mercury xenon lamp, it is also possible to select light having a wavelength other than the wavelength corresponding to the bright line with a filter. Since the intensity of light other than the wavelength of the bright line is small, it can be balanced by widening the half-value width of the interference filter. The wavelength switching may be continuous or intermittent, and for example, 5 to 7 wavelengths may be selected from 400 nm to 650 nm. As described above, the film thickness measurement device 100 includes the light source unit 10 that can switch at least the illumination light of the first wavelength and the illumination light of the second wavelength.

A laser light source that emits laser light of a single wavelength may be used as the light source 11, and a wavelength conversion element may be provided. For example, the generation of the second harmonic allows wavelength conversion of single-wavelength light that enters the wavelength conversion element. A variable wavelength laser can also be used as the light source 11. Furthermore, a plurality of laser light sources that emit laser lights of different wavelengths may be provided to select light having a desired wavelength from the plurality of laser light sources.

The confocal optical system 101 guides the illumination light from the light source unit 10 to the sample 30, and guides the reflected light from the sample 30 to the photodetector 19. The illumination light of a single wavelength that has passed through the interference filter 12 passes through the lens 13a and enters the slit 14. The illumination light is shaped into a line shape in the X direction through the slit 14. Then, the linear illumination light enters the beam splitter 15. The beam splitter 15 splits the light so that the amount of reflected light and the amount of transmitted light are approximately 1:1 regardless of the polarization state. Therefore, almost half of the illumination light passes through the beam splitter 15.

After that, the light traveling in the right direction in FIG. 1 enters the vibrating mirror 16. The vibrating mirror 16 scans the sample 30 in the Y direction with linear illumination light in the X direction. As a result, the surface of the sample 30 can be scanned in XY. As the vibrating mirror 16, for example, a galvano mirror, a polygon mirror, or the like can be used.

The illumination light reflected downward by the vibrating mirror 16 is condensed by the objective lens 17 and irradiated on the sample 30. The sample 30 is placed on the stage 18. Then, the reflected light from the sample 30 passes through the objective lens 17 again, is reflected again by the vibrating mirror 16, and enters the beam splitter 15. Then, about half of the incident light is reflected by the beam splitter 15 and enters the lens 13c. The lens 13c forms an image of the combined light on the light receiving surface of the photodetector 19. The light transmitted through the lens 13c is received by the photodetector 19.

In the present embodiment, the photodetector 19 is a CCD line sensor that captures a confocal image of the sample 30. Illumination light transmitted from the light source 11 through the slit 14 is reflected by the sample 30 and detected by the CCD line sensor. By scanning the sample 30 with the vibrating mirror 16, a slit confocal image is captured. In this way, the photodetector 19 detects the reflected light reflected by the sample 30 and acquires an image in a predetermined region of the sample 30. Note that the scanning method and the like may be different as long as the confocal optical system method is used, and slits and photodetectors that are adapted to the method can be appropriately used. For example, a vibrating mirror for scanning in the X and Y directions may be used, and it is also possible to use an AOD which is an acousto-optic device in the X direction.

The stage 18 has a Z-axis drive motor (not shown), and can move the sample 30 in the vertical direction of FIG. The stage 18 is controlled so that the sample surface comes to the focal position by moving in the Z-axis direction. The focus position can be adjusted by moving the objective lens 17 instead of moving the stage 18 in the Z direction.

In the confocal optical system, it is considered that the focus position changes when the observation wavelength is changed, and the change in brightness due to this is expected. This is to measure the deviation of the focus position of each wavelength in advance and store it in the PC, and to automatically finely adjust the Z position of the sample 30 or the objective lens 17 by the deviation when switching the wavelength. You can cancel at. Alternatively, an omnifocal image may be produced by Z scanning at each wavelength. The wavelength dependence of the observation optical system itself can be corrected by calculation by previously measuring a sample such as silicon or quartz glass having a known reflection spectrum.

In the present invention, the confocal microscope is used to measure the reflectance while switching the illumination wavelength. The processing device 20 measures the reflectance from the images obtained by the respective photodetectors 19 when the illumination light having a plurality of different wavelengths is irradiated. Then, the film thickness of the thin film formed on the substrate is calculated. That is, the processing device 20 calculates the film thickness of the thin film by illuminating light of a certain wavelength (first wavelength) (first image) and illuminating light of a different wavelength (second wavelength) from the image (first image). (2nd image) and the measurement data of the reflectance for each wavelength is obtained. Then, the processing device 20 refers to the calculation data in which the relationship between the wavelength and the reflectance is shown for each film thickness of the thin film, and approximates and calculates the film thickness of the thin film from the measured data.

Next, a film thickness measuring method using the film thickness measuring device 100 described above will be described. As described above, the film thickness measuring device 100 according to the present embodiment uses the confocal microscope capable of selecting the wavelength. In this embodiment, a method of analyzing the thickness of the resist film having a thickness of 1 μm or more will be described. First, the calculation data referred to in the analysis of the film thickness will be described. Then, the method of measuring the reflectance and calculating the film thickness from the measured reflectance measurement data will be described with reference to the calculation data.

[Introduction of coherence factor]
Even if the coherence of the reflected light is reduced, such as when using a high numerical aperture objective lens or when the film thickness is 1 μm or more, the reflectance can be calculated so that the reflectance can be calculated. Introduce a coherence factor. The reflectance R due to the interference of the thin film in the sample 30 is divided into the term of the interference component and the term of the non-interference component. Therefore, the reflectance R due to the interference of the thin film in the sample 30 can be written as Expression (1) using the term R _ic of the non-interference component, the term R _{if of the} interference component, and the coherence factor Γ.

If the state in which the reflected light in the thin film completely interferes is completely coherent (Γ=1), and the state in which there is no interference is completely incoherent (Γ=0), the general state is considered to be in the middle of them. You can When theoretically calculating the reflectance, it can be calculated from the Fresnel coefficient for Γ=1 (R=R _ch ) and Γ=0 (R=R _ic ), respectively. Therefore, by taking the difference between them, the interference component R _if can be calculated by the equation (2).

Therefore, the reflectance R can be theoretically calculated by using the formula (3) with the coherence factor Γ.

The coherence includes temporal coherence and spatial coherence, but the former can be estimated in the form of coherence distance Lc. The wavelength band Δλ of the illumination light and the central wavelength λ can be used to calculate as in Expression (4). For example, when the central wavelength λ=546 nm and the wavelength band Δλ=16 nm, the coherence length Lc is about 18 μm .

The coherence factor Γ is 1 in the range where the film thickness t is sufficiently smaller than the coherence distance L _c , and the coherence factor Γ<<1 in the range where the film thickness t is large. It is generally thought to change. It may be distorted from the Gaussian type due to the influence of multiple reflections, but it is approximate. The correction coefficient is β. When the refractive index of the thin film is n ₁ , the path difference is 2tn ₁ . (Tn ₁ is replaced with ΔL later by correcting the effect of the numerical aperture NA.)

Factors other than the coherence length have an influence on the depth of focus. The focal point shift between the film surface and the substrate surface acts in the direction of reducing the coherence. The effect on coherence is negligible unless the half-width when the relationship between the shift from the focus position and the received light intensity (I-Z curve) is approximated by a Gaussian function or Lorentz function is sufficiently small with respect to the film thickness t become unable. It is considered that the correction of this effect can be absorbed by the correction coefficient β.

The film thickness can be analyzed by the curve fitting method by including the correction coefficient β in the parameter with respect to the measured value based on the theoretical value. It is also possible to analyze the film thickness distribution of the entire image using the value of the correction coefficient β obtained by the analysis of the representative points.

[Theoretical calculation of absolute reflectance]
FIG. 2 is an optical model illustrating the substrate 31 on which the thin film 32 according to the embodiment is formed. As shown in FIG. 2, the optical model shows a cross section of the structure of the sample 30 to be measured. The sample 30 includes a substrate 31 and a thin film 32 provided on the substrate 31. The substrate 31 is, for example, a silicon substrate Si, and the thin film 32 is, for example, a photoresist film (Photo Resist) PR.

From the top, the structure of the sample 30 is air air/photoresist film PR/silicon substrate Si, and the complex refractive index of each wavelength λ is N ₀ , N ₁ , N ₂ , and the refractive index is n ₀ =1 and n _1. , N ₂ , and extinction coefficients are k ₀ =0, k ₁ , k ₂ . The complex index of refraction is defined as in equations (7) to (9). The film thickness of the photoresist film PR to be obtained is defined as a film thickness t.

[1. In case of complete interference]
Next, the thin film interference strength due to the photoresist film PR on the silicon substrate Si will be considered. It is assumed that light of wavelength λ is incident perpendicularly to the thin film (in FIG. 2, it is obliquely incident considering the numerical aperture NA of the objective lens). Part of the incident light is reflected at the interface 0 of the air air/photoresist film PR. Further, part of the light is transmitted and reflected at the interface 1 of the photoresist film PR/silicon substrate Si. A part of the reflected light from the interface 1 is transmitted through the interface 0, and a part of the light is reflected by the interface 0 and then reflected by the interface 1 again. Generally, in a thin film, such multiple reflection occurs. The reflectance R is obtained from the total of these reflected lights that interfere with each other. If the complex refractive index is known, the reflectance R at normal incidence is determined only by the wavelength λ and the film thickness t.

The amplitude reflectances at the interface 0 (air air/photoresist film PR interface) and the interface 1 (photoresist film PR/silicon substrate Si interface) are expressed by equations (10) and (11), respectively. At this time, the phase change and the amplitude change of the light transmitted once through the thin film are represented by δ and γ as in equations (12) and (13).

The amplitude reflectance of the entire film structure considering multiple reflection is given by the equation (14).

Since the reflectance R is the square of the absolute value of the amplitude reflectance, the equation (15) is obtained.

Therefore, if the complex refractive index is known, the reflectance at normal incidence can be calculated from only the wavelength λ and the film thickness t.

Here, the correction of the oblique incidence effect by the numerical aperture NA of the objective lens will be considered. FIG. 3 is a diagram exemplifying an optical path difference due to refraction of reflected light in a substrate on which the thin film according to the embodiment is formed. As shown in FIG. 3, light incident at an angle θ ₀ with respect to the optical axis is refracted at the interface 0 by θ ₁ according to Snell's law of the equation (16), is reflected at the interface 1, and is reflected again. The reflected light E ₁ is refracted at the interface 0. Interference due to multiple reflection occurs between the reflected light E ₁ and the reflected light E ₀ reflected at the interface 0. The difference from the case of vertical incidence is that the optical path difference between the reflected light E ₀ and the reflected light E ₁ is as in Expression (17). The apparent film thickness t decreases as the incident angle increases.

Since the incident light includes various lights from the angle θ _NA depending on the numerical aperture NA of the objective lens to 0 degrees, strictly speaking, it is necessary to calculate the interference of the reflected light at all angles. However, since it takes a large calculation load to handle all the angle dependences of the amplitude reflectance and the optical path difference, the approximation is performed by using the amplitude reflectance and the average optical path difference of vertical incidence. Since incident light is non-polarized light, it is considered that the angle dependence of the optical path difference is dominant rather than the angle dependence of the reflectance.

Since the relationship between the numerical aperture NA and the refraction angle is as shown in Expression (18), the average optical path difference is expressed by Expression (17) in the range of −θ _max to +θ _max , and the average of the angle distribution is expressed by Expression (19). Just ask. Therefore, by replacing the formula (17) with the formula (19) and calculating the formula (15), the reflectance for an arbitrary numerical aperture NA can be easily calculated.

The calculation of the single-layer thin film 32 described so far can be performed by calculating from the substrate 31 side using the synthetic Fresnel coefficient synthesis rule even in the case of a multilayer film.

[2. In case of complete non-coherence]
Next, the reflectance R _ic when there is no thin film interference due to the photoresist film PR on the silicon substrate Si will be examined. FIG. 4 is a diagram exemplifying multiple reflection without interference in the thin film 32 according to the embodiment. As in the case of perfect coherence, multiple reflection occurs in the thin film 32, but the reflected lights of the interface 0 and the interface 1 do not interfere with each other. Therefore, the reflectance R _ic is calculated from a simple sum of these reflected lights. can get.

Assume that the reflectance of the interface 0 is R ₀ , the reflectance of the interface 1 is R ₁ , and the internal transmittance of the thin film 32 is T _i . From the Fresnel coefficient, the reflectance of each interface is given by equation (20) and equation (21). The internal transmittance T _i is expressed by equation (22) using the relationship between the absorption coefficient α and the extinction coefficient k ₁ according to Lambert's law. When the reflectance of m reflections by the substrate 31 is R ^m , the sum of m from 0 to ∞ is the reflectance R _ic , which can be written as equation (23).

The effect of the numerical aperture NA on the internal transmittance T _i is corrected. The internal transmittance T _{i in} the case of the angle θ ₁ may be obtained by replacing the film thickness t in the formula (22) with the angle average <t>. When the incident angle is 0 to θ _max , the formula (24) is obtained, and the average angle <t> can be written as the formula (25). Here, the internal transmittance T _i is set again as shown in Expression (26).

The above calculation can be executed even in the case of a multilayer film by synthetically calculating the interface reflectance from the substrate 31 side.

[3. In case of partial interference]
Using the perfect coherent reflectance R _ch , the perfect non-coherent reflectance R _ic , and the coherence factor Γ, the intermediate state (partial coherence) reflectance R is calculated as equation (27). can do.

In the coherence factor Γ of the equation (30), the film thickness tn ₁ of the equation (5) is replaced with ΔL by using the correction of the numerical aperture NA of the equation (19). In the case of a multilayer film, the total ΔL of each layer is used as the final ΔL.

5 and 6 are graphs illustrating the relationship between the wavelength and the reflectance according to the embodiment, where the horizontal axis indicates the wavelength and the vertical axis indicates the absolute reflectance. As shown in FIG. 5, in the case of the numerical aperture NA=0.3, the dependency of the reflectance R on the correction coefficient β at the wavelength of 400 to 700 nm was calculated under the condition of the film thickness t=1000 nm. When the correction coefficient β=0, Γ=1, when the correction coefficient β=1, Γ<0.8, when the correction coefficient β=2, Γ<0.65, and when the correction coefficient β=10, Γ< It is 0.1, and the interference component attenuates as Γ decreases. Further, since Γ has wavelength dependence, the interference component is attenuated from the shorter wavelength side. As shown in FIG. 6, when the numerical aperture NA=0.3 and the film thickness t=2000 nm, the interference component is attenuated as the film thickness increases.

[Film thickness analysis]
Next, a method of measuring the reflectance to obtain the measurement data of the reflectance, and referring to the above-mentioned calculation data, the method of calculating the film thickness will be described.

FIG. 7 is a flowchart illustrating the film thickness measuring method according to the embodiment. First, as shown in step S1 of FIG. 7, the illumination light of different wavelengths is switched to illuminate the thin film with the illumination light. The sample 30 is irradiated with the illumination light of the first wavelength and the illumination light of the second wavelength by switching through the confocal optical system 101. For example, seven wavelengths of 436 nm, 486 nm, 514 nm, 546 nm, 578 nm, and 633 nm are selected as the wavelength of the illumination light.

Next, as shown in step S2 of FIG. 7, images by illumination lights of different wavelengths are acquired. The reflected light reflected by the sample 30 is detected via the confocal optical system 101 to acquire a first image of light of the first wavelength and a second image of light of the second wavelength.

Next, as shown in step S3 of FIG. 7, the reflectance is measured and the measurement data is obtained. That is, in order to calculate the film thickness of the thin film 32, the measurement data of the reflectance with respect to the first wavelength and the second wavelength are obtained based on the first image and the second image, respectively. The reflectance is obtained from the brightness value I ^sample of the captured image. The film thickness t can be obtained by fitting the measured value of the reflectance R of each wavelength to the equation (27) by the method of least squares using the film thickness t and the correction coefficient β as parameters.

When measuring the reflectance R, the reflectance of a reference sample whose reflectance R is known is measured in order to correct the characteristics of the optical system and the characteristics of the light source used for the measurement. For example, when silicon (Si) is used as a reference (quartz glass or the like may be used), a reflected image of silicon (Si) is captured at each wavelength (color balance, gain control, etc. are kept constant). The reflectance R of the sample is calculated as a relative value with respect to the luminance value I ^Si of this silicon (Si), and further correction is performed using the known wavelength dependency data R ^Si of the reflectance of silicon (Si). Therefore, the reflectance of the sample is obtained by the equation (32). I ⁰ is a dark luminance value received when the shutter of the illumination light is closed.

With the gain adjustment of the confocal microscope fixed, images of the same field of view were acquired for the reference silicon (Si) and the sample 30 by changing the wavelength. The absolute reflectance is calculated from the equation (32) by using the average value of the luminance values regarding the area designated on the image or the luminance value of each pixel as I ^sample and I ^Si .

A photoresist film PR (about 1586 nm: cross-section measurement by white light interferometry) on a silicon wafer was used to verify the calculation method of the coherence factor correction. The optical demodulation is the same as in FIG. The objective lens of the film thickness measuring device 100 has a numerical aperture NA=0.3.

Next, as shown in step S4 of FIG. 7, the film thickness is calculated from the reflectance measurement data with reference to the reflectance calculation data. That is, the film thickness of the thin film 32 is approximated and calculated from the measurement data with reference to the calculation data in which the relationship between the wavelength and the reflectance is shown for each film thickness of the thin film 32.

FIG. 8 is a graph illustrating the relationship between the wavelength and the reflectance according to the embodiment, where the horizontal axis represents the wavelength and the vertical axis represents the absolute reflectance. FIG. 8 shows a plot of the measured reflectance obtained from the equation (32) and the calculated reflectance obtained from the equation (27). The correction coefficient β is calculated as 0 and 0.5. The residual Σ between the reflectance Rmes of the measured value and the reflectance Rcal of the calculated value is obtained by the equation (33).

FIG. 9 is a graph illustrating the relationship between the film thickness of the thin film and the residual error according to the embodiment, where the horizontal axis represents the thin film thickness and the vertical axis represents the residual error.

The film thickness t at which the residual Σ is minimized can be taken as the film thickness at the measurement point of this sample. Therefore, here, when the correction coefficient β=0 (complete coherence), the film thickness t=44 nm is determined, and when the correction coefficient β=0.5 (partial coherence), the film thickness t=1580 nm. Has been determined. The analysis result with the coherence factor Γ is consistent with the measurement result of white interference. Even when measuring six wavelengths, the film thickness t of 1 μm or more can be measured with high accuracy by introducing the coherence factor Γ into the analysis.

Since the film thickness t can be obtained for each pixel of the image by the same method, the film thickness distribution can be displayed. Therefore, the film thickness distribution can be easily histogram-analyzed. Even if an abnormal point of the measured value occurs due to the presence of dust or foreign matter on the thin film, it is easy to perform interpolation processing (noise reduction) with a normal value around. A bird's-eye view can also be used as the display method.

[Supplement of reflectance calculation]
In the case where the thin film 32 in the sample 30 is a multilayer film, the supplementary description will be made on the handling of the reflectance calculation when there are some zero-thickness layers in the multilayer film. In the case of complete coherence, the formula of the Fresnel coefficient of each interface can be applied as it is even if the thickness is 0, and therefore no problem occurs. However, in the case of completely non-coherent, when the layer thickness is 0, the equations (20), (21), and (23) cannot be applied as they are, and therefore a special calculation law described below is used. Need to be introduced.

FIG. 10 is a diagram exemplifying how to handle the reflectance calculation when the film thickness is zero in the case of complete non-coherence according to the embodiment. As shown in FIG. 10, consider the overlapping of three consecutive layers (m-1), m, and (m+1) in the N-layer film. Focusing on m layer, the optical constants is _n m, _{k m} and thickness _{t m.} The interface between the (m-1) layer and the m layer is a (m-1, m) interface, and the interface between the m layer and the (m+1) layer is a (m, m+1) interface. The Fresnel coefficients of the interface are set to r _{m-1, m} and r _{m, m+1,} respectively, and can be expressed in the same manner as in Expression (10) and Expression (11). This relationship can be considered for the (m-1) layer and the (m+1) layer.

When the composite reflectances R′ _{m+2 and m+3} up to the (m+2) layer can be calculated and the film thickness t _m+1 >0 of the (m+1) layer, the composite reflectance R′ _{m+1 of the} (m+1, m+2) interface _{, m+2} can be calculated in the same manner as the equation (23). Here, when the thickness m of the m layer is t _m =0, the Fresnel coefficients of the upper and lower interfaces, r _{m−1, m} and r _{m, m+1} , are given by special equations (34) and (35). The replacement process is performed according to the rules. Using this revised Fresnel coefficient, the reflectivity of each interface is redefined as equation (36) and equation (37). The internal transmittance T _i at this time can be expressed by Equation (38) and Equation (39), respectively.

The combined reflectance R'm of the (m, m+1) interface _{. m+1} is given by equation (40) according to the composition rule.

Furthermore, the combined reflectance R′ _{m−1,m of} the upper m layer also becomes the expression (41) according to the combining rule.

If we look closely at equation (41), it is exactly the equation of the reflectance when there is no m layer. By applying such a Fresnel coefficient conversion rule, it is possible to calculate the reflectance of perfect incoherence even when a layer having a film thickness of 0 exists in the multilayer film. The same calculation can be performed in the case where J number of 0-thickness layers are distributed in the N-layer film.

Next, effects of the film thickness measuring device 100 and the film thickness measuring method according to the present embodiment will be described. In the film thickness measuring device 100 of the present embodiment, the reflectance in the calculation data in which the relationship between the wavelength and the reflectance is shown for each film thickness of the thin film includes the term of the non-interference component. Therefore, it is possible to analyze the film thickness with high accuracy even for a thin film having a film thickness of 1 μm or more that reduces coherence. Thereby, the film thickness distribution in a thin film having a film thickness of 1 μm or more can be analyzed with high resolution.

In addition, the term of the non-interference component R _ic of the reflectance in the calculation data includes the sum of the reflectance in each reflection up to a predetermined number of times due to the interface on the substrate 31 side. Thereby, multiple reflections in the thin film 32 can be reflected, so that the film thickness can be analyzed with high accuracy even for a thin film having a film thickness of 1 μm or more.

The reflectance in the calculation data includes the product of the term of the interference component R _if and the coherence factor Γ indicating the ratio of the term of the interference component. Thereby, even when the reflectance is a partial coherence including the non-interference component R _ic and the interference component R _if , the film thickness of the thin film can be analyzed with high accuracy.

Furthermore, the reflectance in the calculation data includes correction by the numerical aperture NA of the objective lens. As a result, the film thickness of the thin film can be analyzed with high accuracy even when an objective lens having a high numerical aperture is used. Further, the film thickness distribution can be analyzed with high resolution.

Even when the thin film is a multilayer film, the film thickness measuring device 100 synthesizes the reflectances shown for the respective film thicknesses of the plurality of laminated films to obtain the reflectance in the calculation data. Therefore, it is possible to analyze the film thickness of a thin film including a multilayer film with high accuracy. Further, the film thickness distribution in the multilayer film can be analyzed with high resolution. Even when some film thicknesses in the multilayer film are 0, the film thickness can be analyzed with high accuracy by applying the Fresnel coefficient conversion rule without changing the configuration of the optical model.

In this way, the film thickness measuring device 100 can obtain a spectrum for each pixel of the observed image. Then, the spectroscopic spectrum is converted into an absolute reflectance, and fitting is performed using the film thickness in consideration of the coherence factor as a parameter. As a result, not only a film thickness of 0 to 1 μm but also a film thickness of 1 μm or more can be accurately obtained.

In addition, the film thickness measuring device 100 can measure the film thickness distribution on the substrate by measuring the reflection intensity while switching the illumination wavelength using a confocal microscope. Therefore, even if the unevenness of the film thickness distribution is large, non-contact, non-destructive measurement can be performed in a short time. Furthermore, it is possible to measure the film thickness of a surface having a step greater than the depth of focus of the optical microscope.

Although the embodiments of the present invention have been described above, the present invention includes appropriate modifications that do not impair the object and advantages thereof, and is not limited by the above embodiments.

0, 1 Interface 10 Light source part 11 Light source 12 Interference filter 13a, 13b, 13c Lens 14 Slit 15 Beam splitter 16 Vibrating mirror 17 Objective lens 18 Stage 19 Photodetector 20 Processing device 30 Sample 31 Substrate 32 Thin film 100 Film thickness measuring device 101 Confocal optics

Claims (10)

A film thickness measuring device for measuring the film thickness of the thin film of a sample including a substrate and a thin film provided on the substrate,
A light source unit capable of switching between at least first wavelength illumination light and second wavelength illumination light;
A photodetector that detects reflected light reflected by the sample to obtain an image in a predetermined region of the sample,
A confocal optical system that guides the illumination light from the light source unit to the sample, and guides the reflected light from the sample to the photodetector,
In order to calculate the film thickness of the thin film, based on the first image by the illumination light of the first wavelength and the second image by the illumination light of the second wavelength, for the first wavelength and the second wavelength And a processing unit for respectively obtaining reflectance measurement data,
The processing unit refers to calculation data in which the relationship between the wavelength and the reflectance is shown for each film thickness of the thin film , and when approximating and calculating the film thickness of the thin film from the measurement data ,
The film thickness t is approximated and calculated by fitting the measured data of the reflectance at each wavelength to the following formula (A) using the film thickness t and the correction coefficient β as parameters.

here,
R _ch is the reflectance of perfect coherence and is calculated by the equation (B),
R _ic is the reflectance of perfect incoherence, and is calculated by the equation (C),
Γ is a coherence factor, which is obtained by the equation (D),
ΔL is obtained by the equation (E),
a _c is obtained by the equations (F) and (G),
n ₀ =1 and n ₁ and n ₂ are the refractive indices of the thin film and the substrate,
k ₁ and k ₂ are extinction coefficients of the thin film and the substrate,
R ₀ is the reflectance of the interface between air and the thin film,
R ₁ is the reflectance of the interface between the thin film and the substrate,
T _i is the internal transmittance of the thin film,
NA is the numerical aperture of the objective lens of the confocal optical system,
δ and γ are the phase change and the amplitude change of the light transmitted through the thin film once, respectively,
λ and Δλ are the central wavelength and wavelength band of the illumination light,
The film thickness measurement apparatus, wherein the reflectance in the calculation data includes a term of a non-interference component. The film thickness measuring device according to claim 1, wherein the term of the non-interference component includes a sum of reflectances in each reflection up to a predetermined number of times by the interface on the substrate side. The film thickness measurement apparatus according to claim 1, wherein the reflectance in the calculation data includes a product of an interference component term and a coherence factor indicating a ratio of the interference component term. The confocal optical system includes an objective lens,
The film thickness measuring device according to claim 1, wherein the reflectance in the calculation data includes correction based on the numerical aperture of the objective lens. The thin film is a multilayer film in which a plurality of films are laminated,
The film according to any one of claims 1 to 4, wherein the processing unit combines the reflectances shown for the respective film thicknesses of the plurality of films to obtain the reflectance in the calculation data. Thickness measuring device. A film thickness measuring method for measuring a film thickness of a thin film of a sample including a substrate and a thin film provided on the substrate,
Through the confocal optical system, the sample is irradiated by switching between illumination light of the first wavelength and illumination light of the second wavelength,
The reflected light reflected by the sample is detected via the confocal optical system to obtain a first image of the light of the first wavelength and a second image of the light of the second wavelength,
In order to calculate the film thickness of the thin film, the measurement data of the reflectance for the first wavelength and the second wavelength is obtained based on the first image and the second image,
The relationship between the wavelength and the reflectance, with reference to the calculation data respectively shown for each film thickness of the thin film , when approximating the film thickness of the thin film from the measurement data ,
The film thickness t is approximated and calculated by fitting the measured data of the reflectance at each wavelength to the following formula (A) using the film thickness t and the correction coefficient β as parameters.

here,
R _ch is the reflectance of perfect coherence and is calculated by the equation (B),
R _ic is the reflectance of perfect incoherence, and is calculated by the equation (C),
Γ is a coherence factor, which is obtained by the equation (D),
ΔL is obtained by the equation (E),
a _c is obtained by the equations (F) and (G),
n ₀ =1 and n ₁ and n ₂ are the refractive indices of the thin film and the substrate,
k ₁ and k ₂ are extinction coefficients of the thin film and the substrate,
R ₀ is the reflectance of the interface between air and the thin film,
R ₁ is the reflectance of the interface between the thin film and the substrate,
T _i is the internal transmittance of the thin film,
NA is the numerical aperture of the objective lens of the confocal optical system,
δ and γ are the phase change and the amplitude change of the light transmitted through the thin film once, respectively,
λ and Δλ are the central wavelength and wavelength band of the illumination light,
The film thickness measuring method, wherein the reflectance in the calculation data includes a term of a non-interference component. The film thickness measuring method according to claim 6, wherein the term of the non-interference component includes a sum of reflectances in each reflection up to a predetermined number of times by the interface on the substrate side. 8. The film thickness measuring method according to claim 6, wherein the reflectance in the calculation data includes a product of a term of an interference component and a coherence factor indicating a ratio of the term of the interference component. The confocal optical system includes an objective lens,
The film thickness measuring method according to claim 6, wherein the reflectance in the calculation data includes correction based on the numerical aperture of the objective lens. The thin film is a multilayer film in which a plurality of films are laminated,
The film thickness measuring method according to any one of claims 6 to 9, wherein the reflectances shown for the respective film thicknesses of each of the plurality of films are combined to obtain the reflectance in the calculation data.

Images