JP2022045215A - Measurement device - Google Patents

Abstract

To provide a measurement device in which the influence of a surface state of a sample is suppressed and which can measure film thickness more easily.SOLUTION: The measurement device comprises: a light source which emits monochromatic light in 3 wavelengths or greater; a first sliding part which transmits the monochromatic light; a semipermeable membrane which reflects a portion of the monochromatic light passing through the first sliding part and transmits other portions of the monochromatic light; a second sliding part which reflects the monochromatic light passing through the semipermeable membrane; a load application part which relatively moves the first sliding part and the second sliding part while applying a load; an imaging part which acquires an interference image in which the monochromatic light reflected by the semipermeable membrane interferes with the monochromatic light reflected by the second sliding part; and a calculation part which converts a luminance distribution image for each wavelength region of the monochromatic light by the interference image into a hue distribution image, acquires an optical film thicknesses distribution image by converting a hue value for each pixel of the hue distribution image into an optical film thicknesses on the basis of a correspondence between a predetermined optical film thicknesses and the hue value, and calculates the optical film thickness distribution of a transparent film formed between the semipermeable membrane and the second sliding part.SELECTED DRAWING: Figure 1

Description

The present invention relates to a measuring device for measuring a film thickness.

As an example of the prior art of a measuring device for measuring a film thickness, for example, the film thickness measuring device disclosed in Patent Document 1 is known. In the film thickness measuring apparatus according to Patent Document 1, an interference image between a glass plate in a contact state and a sample is acquired by a white polarization interferometry method. A chromium thin film and a silica thin film are formed on the glass plate. The color information is converted into the HSV color space to obtain the hue value, and based on the calibration result with the clearance thickness between the two surfaces, the true contact portion corresponding to the clearance thickness or the film thickness zero is visualized.

Further, a measuring device for oil film thickness disclosed in Patent Document 2 is also known. The measuring device according to Patent Document 2 includes a transparent plate, a light source, a microscope, a camera, and an arithmetic unit. The light source and the camera are arranged on the upper surface side of the transparent plate. The light source instantaneously irradiates the contact portion between the transparent plate and the rolling element with light in a plurality of wavelength ranges. The camera captures the magnified contact area with a microscope when interference light is incident to generate an interference image. The arithmetic unit acquires the intensity of each wavelength range from the interference image and converts it into an oil film thickness. When the light emission time of the light source is L [μs], the rotation speed of the rolling element is V [m / s], the microscope magnification is M, the image pickup element size and the number of pixels of the camera are S [μm], and NP. , 0 <L × V + (S / NP) / M ≦ 5.

Further, a sliding device disclosed in Patent Document 3 is also known. In the sliding device according to Patent Document 3, the transparent sliding material and the reflective sliding material move relative to each other while receiving a load, and the liquid film exists between the transparent sliding material and the reflective sliding material and is transparent. The sliding material and the liquid film are made of a material that transmits light, and the reflective sliding material is made of a material that reflects light. The light from the white light source is converted into light composed of monochromatic light of three wavelengths using a bandpass filter, and the transparent sliding material and the liquid film are transmitted and irradiated to the reflective sliding material to cause light interference. The camera measures the brightness of each of two or more lights of different wavelengths in the resulting light interference. The arithmetic unit calculates the film thickness of the liquid film based on the brightness of each of the two or more lights measured by the camera.

Japanese Unexamined Patent Publication No. 2008-02318 JP-A-2017-207316 Japanese Unexamined Patent Publication No. 2017-053690

However, the measuring device according to the prior art has the following problems. In relation to Patent Document 1, since the white light source contains various wavelength components, the calibration curve of the optical film thickness and the hue is unique to the observation system (camera, lens, light source), and the calibration work is performed separately. There is a problem that is required (calibration curve specific correction problem). Further, since the hue is folded back in the range of 0 to 1, there is a problem that the film thickness convertible range is limited to one cycle (about 265 nm) (calibration curve cycle problem).

In addition, there are the following problems in relation to Patent Documents 2 and 3. That is, if there is a portion of the sample having a large surface roughness due to wear or the like, the intensity of the reflected light is attenuated by diffuse reflection in that portion, which deviates from the calibration curve, and it becomes difficult to convert the film thickness. 9A and 9B are sample interference images for explaining this problem, FIG. 9A is an interference image of a sample without wear, and FIG. 9B is an interference image of a sample including partial wear. Each is shown. As described above, in the sample containing wear, the intensity of the reflected light in the interference image is significantly reduced. This means that the occurrence of wear cannot be tolerated when observing the lubrication phenomenon over time, and it becomes a factor that significantly reduces the degree of freedom of measurement, such as handling only the completely non-contact lubrication phenomenon (diffuse reflection). problem).

Further, when oil is present between the glass and the sample, an absorption (absorption) phenomenon due to the oil occurs, and the reflected light is attenuated according to the absorption coefficient depending on the wavelength and the oil type. In order to estimate the film thickness accurately, it is necessary to obtain this absorption coefficient for each wavelength and oil type (absorption problem).

Further, when the interference images are collectively converted into the film thickness, if the RGB brightness is locally abnormal due to the above reason or other reasons, the film thickness that is significantly different from the surroundings can be obtained only in that portion. The reliability of the film thickness distribution is reduced (unique pixel problem).

Also, although calibration work is theoretically unnecessary, it is rare that the RGB brightness perfectly matches the calibration curve even without the above problems, and in practice, multiple reference points are selected in advance and they are selected. It is necessary to manually input the relationship between the RGB brightness and the film thickness value to correct the calibration curve (calibration curve manual correction problem).

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a measuring device capable of measuring the film thickness more easily by suppressing the influence of the surface condition of the sample.

In order to achieve the above object, the measuring device according to claim 1 has a light source that emits monochromatic light having three or more wavelengths, a first sliding portion that transmits the monochromatic light, and the first sliding portion. The semitransparent film that reflects a part of the monochromatic light transmitted through the portion and transmits the other portion, and the second sliding portion that reflects the monochromatic light transmitted through the semitransparent film, and the above while applying a load. The load-applying portion that causes the first sliding portion and the second sliding portion to move relative to each other interferes with the monochromatic light reflected by the semitransparent film and the monochromatic light reflected by the second sliding portion. The imaging unit that acquires the interference image and the brightness distribution image for each wavelength range of the monochromatic light generated by the interference image are converted into a hue distribution image, and based on a predetermined correspondence between the optical film thickness and the hue value. The optical film thickness distribution image is obtained by converting the hue value for each pixel of the hue distribution image into an optical film thickness, and the optical of the transparent film formed between the semitransparent film and the second sliding portion is optical. Includes a calculation unit that calculates the film thickness distribution.

Further, in the invention according to claim 2, in the invention according to claim 1, the arithmetic unit has a plurality of optical film thicknesses corresponding to one hue value due to the existence of a plurality of cycles in the corresponding relationship. In this case, the reference pixel in the hue distribution image is selected based on a predetermined method, the reference period to which the hue value of the reference pixel belongs is determined, the reference film thickness is determined, and then the optical film of the adjacent pixel is determined. The optical film thickness distribution is calculated by determining the thickness within the range of the reference period.

Further, the invention according to claim 3 is the invention according to claim 2, wherein the calculation unit has a plurality of optical film thicknesses corresponding to the hue values of each pixel for each pixel included in the hue distribution image. Candidates are selected based on the correspondence relationship, and the minimum error norm, which is the minimum value among the error norms between the RGB brightness of each pixel and the RGB brightness corresponding to each of the plurality of optical film thickness candidates, is calculated. The pixel having the smallest minimum error norm is selected as the reference pixel, and the optical film thickness corresponding to the minimum error norm of the reference pixel is determined as the reference film.

Further, the invention according to claim 4 is the invention according to claim 2 or 3, wherein the calculation unit has the optical film thickness distribution due to the peculiar pixels present in the color distribution image. When the loss of the optical film thickness value occurs in the calculation, the calculation of the optical film thickness distribution is repeatedly executed by changing the reference pixel until the loss of the optical film thickness value disappears.

Further, in the invention according to claim 5, in the invention according to any one of claims 1 to 4, the arithmetic unit converts the hue value of each pixel of the hue distribution image into a sine component and a cosine component. Hue filtering processing for removing singular pixels in the hue distribution image is performed by separating, performing hue filtering processing on each of the sine component and the cosine component, and then taking the inverse tangent and returning to the original state. Do more.

Further, in the invention according to claim 6, the filter used in the hue filtering process in the invention according to claim 5 is a Gaussian filter.

Further, the invention according to claim 7 is the invention according to claim 5 or 6, wherein the calculation unit has a hue from among optical film thickness value candidates corresponding to the hue distribution before the hue filtering process. Further processing is performed to restore the sharpness of the hue distribution image by selecting the value of the optical film thickness closest to the optical film thickness obtained from the hue after the filtering process.

According to the measuring device according to the present invention, the influence of the surface state of the sample is suppressed, and it is possible to provide a measuring device capable of measuring the film thickness more easily.

It is a block diagram which shows an example of the measuring apparatus which concerns on embodiment. In the embodiment, (a) is a diagram showing the relationship between the optical film thickness and RGB luminance, and (b) is a diagram showing the relationship between the optical film thickness and the hue value. In the embodiment, (a) is a diagram for explaining an optical film thickness conversion process, and (b) is a diagram for explaining a reference film thickness determining process. (A) shows an image obtained by converting an interference image including partial wear into a hue distribution, and (b) shows an image obtained by subjecting the image of (a) to the optical film thickness conversion process according to the present embodiment. There is. (A) is a diagram showing the relationship between an interference image in the presence of singular pixels and an abnormal distribution of optical film thickness generated when the interference image is converted into an optical film thickness distribution, and (b) is a diagram showing a large number of structures. The hue distribution image before the hue filtering process and the hue distribution image after the hue filtering process are shown for the interference image including defects. (A) is a diagram for explaining the principle of the hue filtering process according to the present embodiment, and (b) is an image of the optical film thickness distribution when the hue filtering process according to the present embodiment is applied. It is a figure for demonstrating the film thickness defect complementation process which concerns on embodiment. It is a flowchart which shows the process flow of the measurement processing program which concerns on embodiment. (A) is a diagram showing an interference image of a sample without wear, and (b) is a diagram showing an interference image of a sample with partial wear.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the following description, the italic characters represented in the mathematical formula are described as non-italic characters.

<Measuring device configuration>
An example of the configuration of the measuring device 10 according to the present embodiment will be described with reference to FIG. 1. The measuring device 10 is a device for measuring the thickness (optical film thickness) of the transparent film on the sliding surface, for example, the thickness of the oil film, and is sometimes called a sliding device. The basic principle of optical film thickness measurement by the measuring device 10 is the optical interferometry. The measuring device 10 is roughly divided into an optical system 30, a sliding system 40, and an arithmetic unit 21.

(Optical system)
The optical system 30 includes a light source 15, a bandpass filter 16, a passage 17, a microscope 18, a half mirror 19, and a camera 20. The light source 15 is a portion that generates measurement light used in the measuring device 10, and in the present embodiment, a white light source is used as an example. The bandpass filter 16 is an element that extracts the measurement light of monochromatic light from the light of the light source 15 which is white light. In the present embodiment, as will be described later, light having three wavelengths is used as the measurement light, so that the bandpass filter 16 transmits the monochromatic light having the three wavelengths. The measurement light transmitted through the bandpass filter 16 passes through the passage 17 and is reflected by the half mirror 19 and is incident on the sliding system as irradiation light Pi. The microscope 18 includes a lens and the like (not shown), and has a function of enlarging a contact portion (described later).

The reflected light Pr reflected by the sliding system 40 passes through the half mirror 19 and is incident on the camera 20. The camera 20 measures the brightness of each of the three wavelengths of light in the light interference generated by the sliding system 40. As will be described later, in the present embodiment, the wavelengths of 600 nm, 560 nm, and 470 nm 3 wavelengths are selected as the monochromatic light. This is because a general color camera is used as a photodetector in consideration of the spectral sensitivity of RGB of a general color camera. The full width at half maximum (FWMH) of each monochromatic light is about 10 nm. In the present embodiment, the smaller the full width at half maximum is, the better, but reasonable results are obtained even when monochromatic light having a full width at half maximum of 10 nm is used. In the present embodiment, a bandpass filter is used to configure the irradiation light Pi from only monochromatic light, but it may be realized by combining a light source that emits monochromatic light, for example, an LED or a laser.

In the present embodiment, the case where these visible lights are used as the wavelength of the irradiated light will be described as an example, but the measurement principle of the measuring device 10 is not limited to visible light and has an arbitrary wavelength. This measurement principle also holds for light (electromagnetic waves). Further, in the present embodiment, the number of monochromatic light is 3, but this is the minimum number of wavelengths, and may be 4 or more wavelengths.

(Sliding system)
The sliding system 40 of the measuring device 10 includes a transparent member 11, a translucent film 12, a reflective member 13, and a load mechanism 14. The measurement target of the measuring device 10 is the thickness of a transparent film, for example, an oil film (not shown) formed between the translucent film 12 and the reflective member 13. The oil film is, for example, a lubricant supplied to a portion (“contact portion”) where the reflective member 13 and the translucent film 12 are in contact with each other. The oil film exists between the translucent film 12 and the reflective member 13, and is arranged so that the translucent film 12 provided on the side opposite to the optical system 30 side of the transparent member 11 and the oil film are in contact with each other. 13 is in contact with the oil film.

The transparent member 11 and the oil film transmit light having three wavelengths, and the reflecting member 13 reflects light having three wavelengths. The irradiation light Pi incident from the optical system is transmitted through the transparent member 11 and the oil film and irradiates the reflective member 13, causing optical interference.

The load mechanism 14 rotates the transparent member 11 and causes the transparent member 11 and the reflective member 13 to move relative to each other while applying a load.

The arithmetic unit 21 is a device that comprehensively controls the entire measuring device 10, and is composed of, for example, a PC (Personal Computer) or the like. The arithmetic unit 21 includes a CPU, ROM, RAM, etc. (not shown), and executes a measurement processing program described later. That is, the luminance of the interference light is acquired for each wavelength range from the interference image, the luminance is converted into the hue, and the hue is converted into the optical film thickness.

(Operation of measuring device)
First, when the transparent member 11 is rotated by the load mechanism 14 and the transparent member 11 and the reflective member 13 are moving relative to each other while receiving a load, white light from the light source 15 is incident on the bandpass filter 16. Then, the irradiation light Pi composed of monochromatic light having three wavelengths transmitted through the bandpass filter 16 is reflected by the half mirror 19 and is incident on the sliding system 40. A part of the irradiation light Pi made of monochromatic light of three wavelengths is reflected at the interface between the translucent film 12 and the oil film, and the other part (other part) of the irradiation light Pi made of monochromatic light of three wavelengths is reflected. Reflects at the interface between the member 13 and the oil film. When light interference is caused by both reflected lights, the light interference distribution is imaged by the camera 20, and the image is input to the arithmetic unit 21 each time the image is taken by the camera 20. Then, for each of the input images, the arithmetic unit 21 executes a measurement processing program described later.

<Background technology of this embodiment>
(Definition of transparent film)
In this embodiment, a measuring device capable of measuring the thickness distribution of the transparent film existing between the sliding materials will be described, but the definition of the transparent film will be described before that. In the present embodiment, the transparent film does not mean only a film that causes almost no light attenuation due to transmission such as a water film, but also a film that causes light attenuation due to transmission such as an engine oil film. That is, it also means a colored transparent film. Further, in the sense of a region for transmitting light existing between the sliding surfaces, it also means a gap between the sliding surfaces when no liquid or solid is contained. A specific example is a valley portion of surface roughness when surfaces having surface roughness slide with each other in air or vacuum. In the present embodiment, unless otherwise specified, the transparent film means the above.

(Importance of measuring transparent film thickness distribution)
In the sliding portion of the mechanical element, a transparent film is often present between the sliding surfaces. For example, many sliding parts are lubricated with a liquid such as oil, and these liquids are often transparent, so that a transparent film exists between the sliding surfaces. In addition, a transparent film may be formed on the surface of the sliding material due to a chemical reaction between the sliding material and surrounding substances. Specific examples include a ZnDTP reaction film when the steel is rubbed in engine oil and an oxide film when the steel is rubbed in the atmosphere. All of these are transparent. Further, although the transparent film as an object does not exist between the sliding surfaces, there may be a region through which light is transmitted. As a specific example, there is a valley portion of the surface roughness when the surfaces having the surface roughness slide with each other in the air or vacuum. Even in such a case, it is useful to measure the thickness distribution of the region. This is because the distribution of zero film thickness points means the distribution of true contact points.

Since these transparent film thickness distributions affect the friction characteristics, it is important to measure the transparent film thickness distribution in the design of sliding parts.

(Optical interferometry)
The "optical interferometry" is a widely used method for measuring the transparent film thickness distribution, especially the oil film thickness distribution. This is a method of measuring the transparent film thickness using the interference of light.

The principle of the optical interferometry is as follows. That is, one of the sliding materials is made of a material that transmits light, and the material of the other sliding material is made of a material that reflects light. These sliding materials may be referred to as "transparent sliding material" and "reflective sliding material", respectively. The transparent sliding material is transmitted from the light source, and the other sliding material is irradiated with light through the transparent film. At this time, the difference in refractive index between the transparent sliding material and the transparent film is appropriately set so that light is partially reflected and partially transmitted at the interface. When the difference in refractive index is small and reflection is not sufficiently generated, a partially reflective film may be formed on the surface of the transparent sliding material in contact with the transparent film.

The light transmitted through the transparent film is reflected by the reflective sliding material, passes through the transparent film and the transparent sliding material again, and returns to the light source direction. In the present embodiment, the light reflected at the transparent sliding material-transparent film interface is referred to as "front surface reflected light", and the light reflected by the reflective sliding material is referred to as "back surface reflected light". Light interference occurs between the front surface reflected light and the back surface reflected light due to an optical path difference, and when the surface of the reflected sliding material is observed from the transparent sliding material side through the transparent film with a light detector (camera, etc.), light interference is observed. Ru. Since the optical path difference, that is, the transparent film thickness affects the color (wavelength and brightness) of this optical interference, information on the transparent film thickness can be obtained from the interference color. This is the principle of optical interferometry.

<Measurement principle of this embodiment>
The measuring device 10 according to the present embodiment is intended to solve the above-mentioned problem with respect to the above-mentioned method for measuring the film thickness by optical interference. Hereinafter, the measurement principle of the measuring device 10 according to the present embodiment will be described in more detail. In this embodiment, an embodiment in which the number of wavelengths of monochromatic light used for measurement is three wavelengths will be illustrated and described.

The reflected light brightness I _j of each monochromatic light (j = 1, 2, 3) is the intensity of the reflected light ( _front surface reflected light) on the front surface and the reflected light (back surface reflected light) on the back surface of I ₁ j and I 2 j, respectively. It is expressed by the following (Equation 1), where λ _j is the intensity and t is the optical film thickness to be measured. FIG. 2A is a plot of (Equation 1) in the case where each of the three wavelengths is λ ₁ = 600 nm, λ ₂ = 560 nm, and λ ₃ = 470 nm. In the following description, the case where the measurement target of the measuring device 10 is an oil film as an example will be described.

Here, the luminance is often treated as an integer value from 0 to 255 in data processing, but in the present embodiment, it is defined by a numerical value in the range of 0 to 1. Further, the reflected light brightness _Ij of each monochromatic light corresponds to the RGB brightness of the captured image, and the wavelength combination is selected within the range where the crosstalk phenomenon does not occur with respect to the spectral sensitivity of the camera. In the following description, each of the three wavelengths is λ ₁ = 600 nm, λ ₂ = 560 nm, and λ ₃ = 470 nm.

ここで、Ｒは裏面における乱反射による減衰率、ｋは測定対象である油膜の吸収係数である。 The above (Equation 1) is an equation when there is no decrease in brightness due to the diffuse reflection problem and the absorption problem among the problems in the above-mentioned prior art, and the reflected light intensity when the diffuse reflection problem and the absorption problem are taken into consideration is as follows. It is changed as in (Equation 2) of.

Here, R is the attenuation rate due to diffused reflection on the back surface, and k is the absorption coefficient of the oil film to be measured.

ただし、（０≦Ｈ≦２π） On the other hand, the hue H is defined by the following (Equation 3) using RGB luminance ( _Ir , _Ig , _Ib ).

However, (0 ≦ H ≦ 2π)

In the prior art, there are cases where a formula is used in which the hue is classified according to the magnitude of each luminance, but in the present embodiment, the hue is defined by an inverse trigonometric function and a radian in consideration of ease of handling. Substituting the reflected light from the above monochromatic light into the hue equation shown in (Equation 3) as _Ir = I ₁ , _Ig = I ₂ , and I _b = I ₃ , (Equation 3) is shown below (Equation 3). It becomes like the formula 4).

Here, it is assumed that R and k are equal regardless of the wavelength, and the amount of light and the light receiving sensitivity are determined by I ₁₁ = I ₁₂ = I ₁₃ for the front surface reflected light intensity and I ₂₁ = I ₂₂ = I ₂₃ for the back surface reflected light intensity. When is adjusted so that is satisfied, the following (Equation 5) can be obtained by substituting (Equation 2) into (Equation 4) and rearranging.

That is, if the amount of light and the light receiving sensitivity are adjusted (calibrated) in advance, the hue H depends only on the wavelength and the optical film thickness of each monochromatic light, as shown in (Equation 5). Therefore, according to the measuring device 10 according to the present embodiment, the influences of the diffused reflection problem and the absorption problem described above can be suppressed. This is determined by I _1j and I _2j even if the amplitude (ie, the magnitude of the RGB brightness) in the calibration curve is reduced by R and k because the hue H is independent of the magnitude of the RGB brightness. This is because the hue H does not change if the center of the amplitude and the attenuation rate (R, k) of the amplitude are equal for each monochromatic light.

FIG. 2B shows a calibration curve of the optical film thickness t and the hue H created in the range of the optical film thickness of 0 nm to 1500 nm based on (Equation 5). Here, as described above, in the present embodiment, the wavelengths are λ ₁ = 600 nm, λ ₂ = 560 nm, and λ ₃ = 470 nm. This calibration curve is analytically calculated using (Equation 5) from the wavelength combination of the selected monochromatic light regardless of the combination of the camera 20 and the lens (microscope 18). Therefore, pre-calibration becomes unnecessary, and the above-mentioned calibration curve-specific correction problem that the optical film thickness and hue calibration curves are unique to the observation system (camera, lens, light source) can be solved.

<Batch conversion from hue to optical film thickness: optical film thickness conversion process>
From the above, the calibration curve of the optical film thickness and the hue is analytically derived, and the hue can be converted into the optical film thickness by using (Equation 5). However, as described above, since the hue is folded back at a cycle of about 250 nm (in FIG. 2B, the cycles P1, P2, and P3 are shown), it is said that there are a plurality of film thickness candidate values for one hue. , The same problem as the calibration curve period problem described above has not been solved yet. That is, even if (Equation 5) is used, the problem that the convertible range is limited within the folding cycle remains. Therefore, in the present embodiment, the hue is monotonically decreased in the range from the optical film thickness of 0 nm to 1040 nm shown by the dotted line R in FIG. 2 (b), and the film thicknesses having the same value are separated by about 250 nm. A new method is adopted to uniquely determine the film thickness. Here, to add to the monotonous decrease in hue, as shown in FIG. 2 (b), the hue is 0 in the period of the optical film thickness of about 250 nm (the period of P1, P2, P3 shown in FIG. 2 (b)). It takes a value in the range of ~ 2π, but it decreases monotonically within each cycle, so when the curves of each cycle are continued, the entire optical film thickness decreases monotonically. However, the range in which the hue is monotonically reduced and the period of the repeated optical film thickness change depending on the combination of the three wavelengths. Hereinafter, a method for converting a hue to an optical film thickness according to this concept (hereinafter, may be referred to as “optical film thickness conversion process”) will be described in detail.

The optical film thickness distribution is determined by the unevenness, inclination, etc. of the surface of the sample, but if the resolution is sufficiently large, the optical film film between adjacent pixels can be measured except for extreme steps such as laser texture and defects such as scratches. It is considered that the change stays in the range of several nm to several ten and several nm at most. In that case, if the optical film thickness of a reference pixel (hereinafter, "reference pixel") and the hue of the adjacent pixel are known, the optical film thickness of the adjacent pixel is among a plurality of optical film thickness value candidates obtained from the hue. , The one closest to the optical film thickness of the reference pixel can be uniquely determined. However, this determination method is based on the precondition that the hue changes monotonically in the vicinity of the optical film thickness of the reference pixel (hereinafter, “reference film thickness”). That is, as long as the optical film thickness of an arbitrary reference pixel is first determined by a predetermined method, the optical film thickness can be determined in order from the reference pixel under the above assumption. That is, finally, all the hues around the reference pixel can be uniquely converted into the optical film thickness. Hereinafter, the present optical film thickness conversion process will be described more specifically.

[Procedure 1]: An arbitrary reference pixel is selected, and an optical film thickness corresponding to the hue of the reference pixel (hereinafter, may be referred to as “reference film thickness”) is determined. Here, the determination of the reference film thickness means to determine one optical film thickness from a plurality of optical film thickness value candidates based on a predetermined procedure. One of the predetermined procedures is manual, for example, a human visually recognizes the hue of the reference pixel, or by a rule of thumb, determines which hue cycle to enter in comparison with the hue curve of FIG. 2 (b). , The optical film thickness of that period is adopted as the optical film thickness of the reference pixel. As a method for determining the reference film thickness, there is a method for automatically determining the standard film thickness other than the manual method, and a method for automatically determining the standard film thickness (reference film thickness determination process) will be described later.

[Procedure 2]: Move to an adjacent pixel and select an optical film thickness close to the moving source (reference pixel) from the optical film thickness value candidates corresponding to the hue. Of course, this selection can be made automatically.
[Procedure 3]: [Procedure 2] is executed linearly and repeated until the end of the interference image is reached.
[Procedure 4]: Returning to the reference pixel, [Procedure 2] and [Procedure 3] are repeated in all directions to convert the entire interference image into an optical film thickness distribution.

The procedure of the optical film thickness conversion process will be specifically described with reference to FIG. 3A. In FIG. 3A, four pixels PX1, PX2, PX3, and PX4 are arranged in the horizontal direction.

First, in [Procedure 1], the reference pixel is selected as the pixel PX1, and the hue value of the reference pixel is determined.
In the example of FIG. 3A, the case where the hue value of the reference pixel is determined to be 3.14 is illustrated. When the hue value is determined, the corresponding optical film thickness value candidate is determined. Therefore, for example, a person determines one of the optical film thickness value candidates by visual inspection, and the hue value is shown in FIG. 2 (b). Determine which period of the hue curve it is in. As a result, the optical film thickness corresponding to the hue value of the reference pixel is determined to be 511 nm from the optical film thickness candidates (767 nm, 511 nm, 256 nm).

Hereinafter, according to [Procedure 2], the hue value of the pixel PX2 adjacent to the reference pixel PX1 is automatically determined to be 2.09, and the optical film thickness is automatically determined to be 527 nm in the range of the period P2. Will be done. Hereinafter, the same applies to the pixels PX3 and PX4.

The above-mentioned method is similar to the phase shift method for a white light source which is conventionally known, but the phase shift method requires unwrapping processing for hue, which complicates the filtering processing and the filtering removal processing described later. Therefore, in the present embodiment, the hue is not converted and is handled as it is.

<Example>
The effect of the measuring device 10 according to the present embodiment will be described with reference to FIG. 4 (a) shows the hue distribution obtained by converting the interference image including the partial wear shown in FIG. 9 (b), and FIG. 4 (b) shows the present embodiment with respect to FIG. 4 (a). The optical film thickness distribution obtained by performing the optical film thickness conversion process is shown. In FIG. 4B, parts other than the sliding portion are masked for easy viewing. As shown in FIG. 9B, although the RGB brightness of the original interference image is greatly reduced due to wear, the use of hue provides a level of optical film thickness that does not cause any problem as data. You can see that there is.

<Reference film thickness determination process>
Here, a reference film thickness determination process for selecting a reference pixel and automatically executing a process for determining the reference film thickness will be described. FIG. 2A is a diagram showing calibration curves of optical film thickness and RGB brightness when each of the three measurement wavelengths is λ ₁ = 600 nm, λ ₂ = 560 nm, and λ ₃ = 470 nm. rice field. If there is no decrease in RGB brightness due to diffuse reflection or absorption problems and the optical film thickness is within this range, the RGB brightness of each pixel of the interference image matches any combination of RGB brightness on this calibration curve. Should be. Based on this, the error norm v between the RGB luminance I _pj of the pixel of interest and the RGB luminance value I _j (t) of the reflected light corresponding to the arbitrary optical film thickness t on the RGB calibration curve shown in FIG. 2 (a). (T) is defined by the following (Equation 6).

Now, among the plurality of error norms obtained by substituting the plurality of film thickness candidate values obtained from the hue values of each pixel into (Equation 6), the smallest one is the minimum error norm of the pixel, and the film giving it. The thickness candidate value is defined as the provisional film thickness value. This minimum error norm should be 0 for pixels in which the RGB luminance is not reduced at all, and the provisional film thickness value giving this minimum error norm should be the actual optical film thickness value.

However, in reality, the minimum error norm is not 0 for most pixels because the RGB brightness is lowered due to the diffused reflection problem and the absorption problem. Further, if the decrease in brightness is large, it is possible that the wrong provisional film thickness value is selected, and in that case, the minimum error norm is also considered to be large. Conversely, it can be said that a pixel having a small minimum error norm has a small decrease in luminance and high reliability. Therefore, such a pixel may be selected as a reference pixel. Specifically, the most reliable reference pixel is automatically selected by the following procedure, and the reference film thickness is automatically determined.
[Procedure 1]: A plurality of film thickness candidate values are calculated from the hue values of each pixel. Of the plurality of error norms obtained by substituting these into (Equation 6), the smallest one is stored as the minimum error norm of the pixel, and the film film candidate value giving it is stored as a provisional film film value.
[Procedure 2] The processing of [Procedure 1] is performed on all the pixels, the pixel having the smallest minimum error norm is used as the reference pixel, and the provisional film thickness value thereof is used as the reference film thickness value.

By the above-mentioned reference film thickness determination process, it is possible to automatically determine a highly reliable reference pixel and a reference film thickness value. However, in rare cases, singular pixels may be erroneously detected. In that case, a pixel whose brightness has not been reduced may be manually selected as a reference pixel, the pixel may be subjected to the process of the above [Procedure 1], and the optical film thickness value may be automatically determined.

The reference film thickness determination process will be described more specifically with reference to FIG. 3 (b). In the example shown in FIG. 3B, the interference color (RGB luminance: R = 0.356, G = 0.727, B = 0.728) of the pixel PX1 is converted into a hue, and the hue value of the pixel PX1 is 3. It is .14. Candidates for the optical film thickness in this case are 256 nm, 511 nm, and 767 nm, respectively, corresponding to the periods P1, P2, and P3 shown in FIG. 2 (b). At this time, the error norms for these optical film thicknesses are 0.521, 0.001, and 0.497, respectively. Therefore, the minimum error norm of the pixel PX1 is 0.001 ([Procedure 1]). The provisional film thickness value in this case is 511 nm. As a result of calculating the minimum error norm for all the pixels, assuming that the minimum error norm of the pixel PX1 is the smallest, the pixel PX1 is selected as the reference pixel, and the reference film thickness is determined to be 511 nm.

<Hue filtering process>
By the above-mentioned procedure, it is possible to accurately convert the interference image into the optical film thickness while avoiding the problem in the prior art. Here, in the above procedure, it is assumed that there is no extreme step and the change in the optical film thickness between pixels remains at several nm to several ten and several nm. However, in reality, there may be pixels whose hue is greatly disturbed by camera noise, local scratches, foreign matter, etc. (hereinafter referred to as "unique pixels"). For example, although there is no wear in the interference image of FIG. 9A, there are many portions where the optical film thickness suddenly changes due to structural defects. In such a case, since the information of the adjacent pixel is dragged in the above procedure, when a certain point is a singular pixel, an inaccurate value is obtained not only in the optical film thickness of the pixel but also in the optical film thickness conversion of the subsequent pixels. Is obtained.

The influence of the singular pixel will be described with reference to FIG. 5 (a). FIG. 5A <1> is an example of an interference image including singular pixels, and the singular pixels are shown together with the reference pixels. 5 (a) and <2> show the conversion process from the interference image shown in FIG. 5 (a) and <1> to the optical film thickness distribution. As shown in FIG. 5 (a) <2>. When the optical film thickness is linearly converted from the position B of the reference pixel to the position S of the singular pixel, the abnormal distribution of the optical film thickness spreads from the position S to the side opposite to the position B. Therefore, in practice, it is necessary to remove these peculiar pixels.

In general image processing, filtering processing is often performed to remove peculiar pixels. Typical filters include a Gaussian filter and a median filter, both of which are to be performed individually for each of the RGB luminance of the interference image. If it is normal image processing, there is no problem as it is, but the hue handled here may not work well with normal filtering processing. This is because it can be said that the hue is obtained by excluding the saturation and brightness information from the RGB luminance and extracting the relative relationship between R, G, and B, but in the filtering process for the RGB luminance of the interference image, the saturation and brightness information is also included. Include and filter. Therefore, the hue in the interference image after filtering drags their influence, and unnatural color wrapping may occur, which may be unsuitable for use in terms of optical film thickness conversion. This is particularly remarkable when the brightness decrease due to the diffused reflection problem, the absorption problem, or the like is large.

Ｈｓ、Ｈｃは、値が－１～１で定義され、かつ折り返しではないため、Ｈｓ、Ｈｃに色相フィルタリング処理を行っても不連続は発生しない。そして、色相フィルタリング処理後の値をＨｓ’、Ｈｃ’とすると、Ｈｓ’、Ｈｃ’を以下に示す（式８）で戻すことにより不連続を回避した色相フィルタリング処理後の色相Ｈ’が得られる。

Therefore, in the present embodiment, a new method of applying a Gaussian filter to the hue instead of the interference image is adopted. Hue H is defined by 0 ≦ H ≦ 2π, so if a Gaussian filter is applied directly to the hue, the folded parts of 0 and 2π will be separated even though they are originally continuous, which is not possible. The hue distribution will be appropriate. In order to avoid this problem, in the present embodiment, the hue H is separated into a sine and cosine component Hs and a cosine component Hc as shown in the following (Equation 7).

Since the values of Hs and Hc are defined by -1 to 1 and are not folded, discontinuity does not occur even if the hue filtering process is performed on Hs and Hc. Then, assuming that the values after the hue filtering process are Hs'and Hc', the hue H'after the hue filtering process that avoids the discontinuity can be obtained by returning Hs'and Hc'by the following (Equation 8). ..

FIG. 5 (b) <1> shows the hue distribution (that is, the hue distribution according to (Equation 5)) before the hue filtering process obtained by converting the interference image containing a large number of structural defects shown in FIG. 9 (a). 5 (b) <2> shows a hue distribution (that is, a hue distribution shown in (Equation 8)) obtained by subjecting the hue distribution of FIG. 5 (b) <1> to a hue filtering process according to the above procedure. In this example, a Gaussian filter is used as a filter. As shown in FIG. 5B <1>, a large number of color skips occur in the hue distribution before the hue filtering process. This color skipping itself is significant information on one side because it indicates the optical film thickness of the defective portion, but it causes a conversion error. On the other hand, referring to FIG. 5 (b) <2>, these color skips can be removed to a level where there is no problem in the hue distribution by the hue filtering process (the crushing is not noticeable). I understand.

<Hue filtering restoration process: sharpness restoration process>
The hue filtering process enables automatic film thickness conversion from hue even if the hue image contains singular pixels. However, the image of the optical film thickness distribution after the hue filtering process has reduced sharpness due to filtering. Therefore, it is preferable to restore the original sharpness. Hereinafter, this sharpness restoration process will be specifically described with reference to FIG. 6A.

First, the optical film thickness is determined from the hue after hue filtering. This determination is made in accordance with the above-mentioned [Procedure 1] to [Procedure 4]. In the example shown in FIG. 6A, the pixel PX1 is determined to be 520 nm, the pixel PX2 is determined to be 530 nm, and the pixel PX3 is determined to be 550 nm. Next, for each pixel, the value of the optical film thickness closest to the optical film thickness obtained from the hue after the hue filtering process is selected from the optical film film value candidates corresponding to the hue distribution before the hue filtering process. In the example shown in FIG. 6A, the values of 511 nm for the pixel PX1, 527 nm for the pixel PX2, and 579 nm for the pixel PX3 are selected. As described above, it is possible to acquire a distribution image of the optical film thickness from which peculiar pixels are removed while maintaining the sharpness before applying the hue filtering process.

FIG. 6B shows an image of the optical film thickness distribution obtained by applying the above-mentioned hue filtering process to the interference image shown in FIG. 9A. In FIG. 6B, for the sake of visibility, the portion of the non-sliding surface that is not directly related to the measurement is removed by masking. As is clear with reference to FIG. 6B, even when a large number of defects are included, an accurate optical film thickness distribution can be obtained by applying the above-mentioned hue filtering process.

<Film thickness defect complement processing>
By using the above-mentioned hue filtering process, it is possible to convert the optical film thickness on most of the sliding surfaces, but there are some surface conditions that are difficult to handle even with the hue filtering process. For example, it is a case where a defect such as an extreme step or a scratch due to a laser texture is included.

With reference to FIG. 7, a method for converting an optical film thickness, that is, a film thickness defect complementing process, which is useful in the above cases, will be described. First, as shown in FIG. 7 <1>, an arbitrary reference pixel 1 is selected, and optical film thickness conversion is performed along a straight line. The optical film thickness conversion at this time is performed according to the methods of [Procedure 1] to [Procedure 4] of the above optical film thickness conversion process. Then, when the pixel reaches a pixel whose optical film thickness is discontinuous, the conversion is terminated at that portion. By the above processing, an optical film thickness distribution including linear data defects starting from a singular pixel can be obtained.

Next, as shown in FIG. 7 <2>, a reference pixel 2 other than the reference pixel 1 that can avoid the singular pixel and reach the data loss is selected, and conversion is performed again along the straight line from the reference pixel 2. At this time, if the point where the optical film thickness value is known by the optical film thickness conversion with respect to the reference pixel 1 is selected as the reference point, it is not necessary to manually select the optical film thickness value. By this processing, the optical film thickness value can be obtained even in the data missing portion in terms of the optical film thickness with respect to the reference pixel 1. Instead, a new data missing part occurs, but in the new data missing part, the optical film thickness value determined by the optical film thickness conversion by the first reference pixel is input as it is, and the above processing is performed on the peculiar pixel. As shown in FIG. 7 <3>, the optical film thickness distribution in the region excluding the singular pixel can be obtained by repeating until there is no data missing portion other than the above.

<Measurement processing>
With reference to FIG. 8, the measurement process executed by the measuring device 10 according to the present embodiment will be described. FIG. 8 is a flowchart showing a processing flow of a measurement processing program executed by the measuring device 10. The measurement processing program is stored in, for example, a storage means such as a ROM (not shown) included in the arithmetic unit 21, and a CPU (not shown) reads the measurement processing program from the ROM or the like and stores the measurement processing program in a RAM or the like (not shown). Deploy and run.

With reference to FIG. 8, in step S10, an interference image is acquired from the camera 20.

In step S11, the interference image acquired in step S10 is converted into a hue image.

In step S12, a hue filtering process is performed on the hue image converted in step S11. In this hue filtering process, after separating the hue into a sine component and a cosine component, for example, a Gaussian filter is used as a filter.

In step S13, the reference pixel is selected by the reference film thickness determination process described above, and the reference film thickness is determined.

In step S14, the above-mentioned optical film thickness conversion process ([Procedure 1] to [Procedure 4]) is executed to convert the hue to the optical film thickness.

In step S15, it is determined whether or not there is a film thickness defect due to the peculiar pixel, and if the determination is negative, the process proceeds to step S16. On the other hand, if the determination becomes affirmative and a defect portion due to a peculiar pixel occurs, the film thickness defect portion complement processing is executed, the reference pixel is changed until the defect portion is filled, and the optical film thickness conversion is repeated. .. It should be noted that the film thickness defect complementing process may be performed as needed, and is not necessarily an essential process.

In step S16, the hue filtering restoration process is performed to restore the sharpness of the image of the optical film thickness, and the present measurement processing program is terminated. The hue filtering / restoring process may be performed as needed, and is not necessarily an essential process.

As described in detail above, according to the measuring device according to the present embodiment, it is possible to perform high-speed conversion of the optical film thickness of the entire interference image even when the hue of the interference image is used. Further, the range of the convertible optical film thickness in the measuring device 10 is a range in which the hue monotonically decreases with respect to the optical film thickness, for example, the combination of measurement wavelengths is (λ ₁ = 600 nm, λ ₂ = 560 nm, λ ₃ = 470 nm). ), It is 0 nm to 1040 nm, which solves the calibration curve cycle problem that the optical film thickness convertible optical film thickness range is limited to one cycle (about 265 nm) of the optical film thickness. Is planned.

Here, referring to FIG. 2A, in the range of the optical film thickness of 0 nm to 200 nm, there is a section where the gradient (slope) of the monotonous decrease is small, and the conversion accuracy may deteriorate in this section. In that case, a transparent film called a spacer layer may be arranged on the translucent film 12 on the surface of the transparent member 11 with a thickness of, for example, 150 nm to 200 nm. As a result, the optical film thickness becomes larger than the actual thickness by the amount obtained by multiplying the thickness of the spacer layer by the refractive index, so that it is possible to avoid the range of the optical film thickness in which the accuracy deteriorates.

Further, in the present embodiment, since the information of the adjacent pixels is referred to and the film thickness defect portion is complemented as necessary, the reliability is lowered due to the occurrence of local optical film thickness deviation, which is the above-mentioned peculiar pixel problem. The outbreak is also suppressed. Further, in the measuring device 10 according to the present embodiment, if the reference film thickness determination process is adopted, the entire process is automated, so that manual calibration work is actually required. The problem has also been improved. As described above, the present invention is capable of solving the above-mentioned problems of the prior art and improving the applicable range, convenience, and accuracy of real-time observation of the optical film thickness on the sliding surface. ..

10 Measuring device 11 Transparent member 12 Translucent film 13 Reflective member 14 Load mechanism 15 Light source 16 Band pass filter 17 Passage 18 Microscope 19 Half mirror 20 Camera 21 Computing device 30 Optical system 40 Sliding system P1, P2, P3 Period PX1, PX2 , PX3, PX4 pixel Pi irradiation light Pr reflected light

Claims (7)

A light source that emits monochromatic light of three or more wavelengths,
The first sliding portion that transmits the monochromatic light and
A semipermeable membrane that reflects a part of the monochromatic light that has passed through the first sliding portion and transmits the other portion.
A second sliding portion that reflects the monochromatic light that has passed through the semipermeable membrane, and
A load-applying portion that causes the first sliding portion and the second sliding portion to move relative to each other while applying a load.
An image pickup unit that acquires an interference image in which the monochromatic light reflected by the semipermeable membrane and the monochromatic light reflected by the second sliding portion interfere with each other.
The brightness distribution image for each wavelength range of the monochromatic light based on the interference image is converted into a hue distribution image, and the hue value for each pixel of the hue distribution image is calculated based on a predetermined correspondence between the optical film thickness and the hue value. An arithmetic unit that acquires an optical film thickness distribution image by converting to an optical film thickness and calculates the optical film thickness distribution of the transparent film formed between the semitransparent film and the second sliding portion. Measuring equipment including. The calculation unit selects a reference pixel in the hue distribution image based on a predetermined method when a plurality of optical film thicknesses correspond to one hue value due to the existence of a plurality of cycles in the correspondence. Then, the reference period to which the hue value of the reference pixel belongs is determined to determine the reference film thickness, and thereafter, the optical film thickness of the adjacent pixel is determined within the range of the reference period to calculate the optical film thickness distribution. The measuring device according to claim 1. The calculation unit selects a plurality of optical film thickness candidates corresponding to the hue values of each pixel for each pixel included in the hue distribution image based on the correspondence relationship, and the RGB brightness of each pixel and the plurality of pixels. The minimum error norm, which is the minimum value among the error norms with the RGB brightness corresponding to each of the optical film thickness candidates of, is calculated, the pixel with the smallest minimum error norm is selected as the reference pixel, and the minimum of the reference pixel is selected. The measuring device according to claim 2, wherein the optical film thickness corresponding to the error norm is determined as a reference film thickness. The calculation unit eliminates the loss of the optical film thickness value when the loss of the optical film thickness value occurs in the calculation of the optical film thickness distribution due to the peculiar pixels existing in the hue distribution image. The measuring device according to claim 2 or 3, wherein the calculation of the optical film thickness distribution is repeatedly executed by changing the reference pixel up to. The calculation unit separates the hue value of each pixel of the hue distribution image into a sine and cosine component, performs hue filtering processing on each of the sine and cosine component, and then takes an inverse tangent to the original. The measuring apparatus according to any one of claims 1 to 4, further performing a hue filtering process for removing singular pixels in the hue distribution image by returning to. The measuring device according to claim 5, wherein the filter used in the hue filtering process is a Gaussian filter. The calculation unit selects an optical film thickness value closest to the optical film thickness obtained from the hue after the hue filtering process from the optical film thickness value candidates corresponding to the hue distribution before the hue filtering process. The measuring device according to claim 5 or 6, further performing a process of restoring the sharpness of the hue distribution image.

Images