JPH049704A - Optical interference film thickness measuring apparatus - Google Patents

Abstract

PURPOSE:To improve the positional resolving power and to measure the film thickness with high accuracy by arranging an end face and the other end face of a photoconductive member in tight contact with a projecting surface of an image intensifier and a plane of incidence of a photoelectric converting element, respectively. CONSTITUTION:A light projecting from a white light source 1 passes through an entrance window 8 opened at a light condenser 7 to an object 6 to be measured via a polarizing plate 5 or the like. The condenser 7 and entrance window 8 work as a shielding body and an aperture, respectively. The condenser 7 is comprised of a supporting body 9 and an optical fiber 10. A light image passing through an image intensifier 16 and a photoconductive member 17 via a spectroscope 14 etc. at the other end face of the fiber 10 is input to a photodetecting element 18 and detected as the intensity of the light for every wavelength. A data processor 19 takes the intensity of the light for every wavelength, thereby obtaining the wavelength of a bright part or a dark part resulting from the interference phenomenon and operating the film thickness of the object 6 from a predetermined formula. If one end face of the member 17 and the other end face thereof are held in tight contact with a projecting surface of the intensifier 16 and a plane of incidence of the element 18 respectively, the intensity ratio of the spectroscopic bright and dark parts can be made larger.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to an optical interference type film thickness measuring device that has excellent positional resolution and is suitable for measuring the film thickness of a measurement target in which local thickness unevenness exists.

(Prior art) For example, many devices that measure the thickness of polymer films utilize the absorption of β-rays or infrared rays by the film, but these methods are not suitable for highly accurate measurements. Not suitable. For more accurate measurements, for example, JP-A-56-1
The so-called optical interference method disclosed in JP-A No. 15905, JP-A-63-163], and JP-A-05 is employed.

Optical interference film thickness measurement equipment uses a method that detects changes in spectral intensity due to interference phenomena when parallel white light is reflected or transmitted by a film, depending on the incident angle of the white light, the film thickness, and the refractive index. By utilizing this dependence, changes in the spectral intensity of the reflected light or transmitted light are detected, and the film thickness is determined from the changes. However, when measuring the thickness of a film with local thickness unevenness using such a device,
There is a problem in that the SN ratio is greatly reduced.

That is, to detect a change in spectral intensity using the above-mentioned device is to know specifically whether it is a bright region or a dark region due to an interference phenomenon, and at what wavelength it occurs. Therefore, in order to detect the position of a bright area or a dark area with high precision, it is necessary that the intensity ratio and contrast between the bright area and the dark area be sufficiently large.

However, if there is local thickness unevenness such as a change in thickness within an optical image (hereinafter referred to as a spot) of white light irradiated onto the film, sufficient contrast cannot be obtained.

In other words, the interference light reflected or transmitted by a minute area within the spot has bright and dark areas at wavelengths corresponding to the thickness of that minute area.If there is a distribution of this thickness within the spot, then The spectral intensities with distributions in bright areas and dark areas overlap, resulting in a significant decrease in contrast, and as a result, accurate measurements cannot be made.

Further, the thickness measured in this manner is a typical thickness within the spot, and the fine thickness distribution within the spot is canceled out, making it possible to only measure with poor positional resolution.

In fact, in the polymer film manufactured by melting and discharging from J gold, there are deposits of A and 4 near the mouthpiece 1, and due to the influence of foreign substances, the length of the film is streaked in the direction of force, convex or There is a ``ko'' that causes uneven thickness of the concave. These lines have a great influence on the quality of the produced film, so it is important to quantitatively measure this shape in the width direction3. , ~3n+m, and the height of the convex portion (the height of the concave portion) is 0.15 μm or more, for example,
Not only is it impossible to measure this thickness unevenness with a spot diameter of 8 mm using the optical interference type film thickness measuring device disclosed in Japanese Patent Application Laid-Open No. 63-16311', )5, but also the contrast of the interference waveform is lost. The thickness itself cannot be measured.

On the other hand, in an optical interference film measuring device, the area of the spot affects the S/N ratio. In other words, it is reflected by the film (
The amount of light transmitted (or transmitted) is proportional to the area of the spot, and the amount of light guided to the spectrometer changes accordingly.

Therefore, when the spot area is reduced, the amount of light guided to the spectroscope is also reduced, and the output of the spectrometer is proportionally reduced. However, since there is no change in the S/N ratio of the signal processing section,
The SN ratio of the entire device deteriorates. In order to compensate for this, it is possible to increase the intensity of the incident light.

However, in pinhole lens systems commonly used to create parallel light, increasing the pinhole diameter to increase the intensity reduces the parallelism of the light and deteriorates the interference contrast.

Regarding the lamp intensity of the light source, high-output lamps are generally made by making the filament part larger, and since the intensity per unit area of the filament is not high, this effect is lost due to pinholes.

Moreover, increasing the strength per unit area of the filament is not practical because it significantly shortens the life of the lamp.

On the other hand, even if the amplification factor of the output of the photoelectric conversion element in the spectrometer is increased, noise such as dark current and thermal noise will also be amplified, and the S/N ratio will not be improved as a result. Therefore, with the conventional optical interference type film thickness measuring device, it has not been possible to accurately measure and cover fine thickness unevenness.

(Problem to be solved by the invention) The second purpose of the invention is to solve the above drawbacks of the conventional device,
Optical beam thickness measurement device that can increase the contrast between bright and dark areas even if the object to be measured has local thickness unevenness, and can perform highly accurate measurements with good positional resolution. is to provide.

(Means for Solving the Problems) In order to achieve the arrangement, in this invention, parallel white light is made incident on the measurement object at a constant angle of incidence, and the reflected light or transmitted light from the measurement object is collected using a spectrometer. The device is configured to measure the film thickness of the measurement target by guiding the white light to the target and detecting its spectral intensity, the shielding body having a hole on the incident optical path of the white light for regulating the outer shape of the light beam irradiated onto the measurement target. and the spectrometer includes an image intensifier, a photoconductive member formed by bundling a plurality of optical fibers, and two photoelectric conversion elements, and one end surface of the photoconductive member is connected to the image intensifier. An optical interference type film thickness measuring device is provided, characterized in that the tensifier is disposed on the output surface of the tensifier with the other end surface in close contact with the entrance surface of the photoelectric conversion element.

(Function) The outer shape of the parallel white light is regulated by the hole in the shielding body, and the parallel white light is incident on the measurement target at a constant angle of incidence. The spot on the measurement target is a projection of the shape of the hole in the shield. The incident light is reflected (or transmitted) by the measurement target, and at this time, the intensity of the frequency component changes due to an interference phenomenon. This light is incident on a spectroscope and is separated into positions according to frequency.

The intensity change at each frequency of the interference phenomenon results in an intensity change at each position, that is, an image of light.

This light image is amplified by the image intensifier, enters one end face of the photoconductive member from the output surface of the image intensifier, propagates within each optical fiber without being dispersed, and reaches the other end face. The light image is output while being held and input to the light receiving element. During this time, the light is not dispersed by the lens system and is input to the light receiving element with only slight attenuation caused by the optical fiber. The light-receiving element has a position-wavelength correspondence material in advance, so that the intensity of each wavelength can be adjusted as follows:
Spectral intensity can be measured.

As is well known, the change in the intensity of the light thus separated depends on the film thickness of the object to be measured if the incident angle and the refractive index of the object to be measured are constant. Therefore, keeping the incident angle θ constant,
By measuring the refractive index n of the object to be measured in advance, the film thickness can be determined from the change in the spectral intensity of the reflected light using the following equation.

d = 1 / [[2(n''-5in''θ)1'
Town × (λ, ×λ□+/(λ, −λ,,I)l ]・・
・(1) Here, d is the film thickness to be measured, λ1, λ, +
1 (λ, >λ□1) represents the wavelength of adjacent bright or dark areas (peaks).

(Example) Hereinafter, this invention will be explained in detail based on one embodiment.

FIG. 1 shows the configuration of an apparatus of the present invention in which a polarizer is disposed on the incident optical path and receives reflected light from the object to be measured.

In FIG. 1, eight lights emitted from a white light source 1 are transmitted through a condenser lens 2, a pinhole plate 3, a collimator lens 4,
The light passes through a polarizing plate (polarizer) 5 and further passes through an entrance window 8 formed in a condenser 7, which will be described later, to be guided to a measurement target (film) 6. This concentrator 7 and entrance window 8
serve as a shield and a hole, respectively. The size of the entrance window was a round hole with a diameter of 2 mm. Usually 3
A round hole with a diameter of mm or less.

As the white light source 1, a tungsten lamp, a xenon lamp, a halogen lamp, etc. can be used. Further, the shielding body does not need to be an independent one, and may be one with a hole formed on the incident optical path of another element.

A light condenser 7 is placed facing the measurement target 6 .

This condenser 7 consists of a hemispherical support 9 having a center of curvature at the point at which the film thickness on the measurement target 6 is to be measured and in which the entrance window 8 is bored, and a number of optical fibers 10. Each one end of the optical fiber 10 is integrally fitted and fixed into a hole bored in the support body 9, and each other end is bundled and fixed. One end of each optical fiber 10 is fixed such that the fiber axis faces the point at which the film thickness on the measurement object 6 is to be measured, that is, the center of curvature of the support 9.

Behind the condenser 7, a sorting lens 11 is arranged facing the other end surface of each optical fiber 10, and further behind the sorting lens 11, an extraction pinhole plate 12, a collimator lens 13, and a spectroscope are arranged. 14, imaging lens 15, image intensifier 16, fiber plate (light conductive member) 1
7. Light receiving elements, such as the linear image sensor 18, are arranged in this order.

As the spectrometer 14, a diffraction grating type one such as a plane diffraction grating, a prism type one, etc. can be used. For example, the fiber plate 17 has a diameter of 6μ.
A plate material made by bundling a plurality of optical fibers with a diameter of approximately 1.5 m is used. Further, the output surface of the image intensifier 16 and the fiber plate 17 are preferably brought into close contact with each other via transparent grease or the like. In addition, linear image sensor 1
8 and Fiberbrae River/17 are in close contact, I like it (-<
Fiber Brake!・17 and linear image sensor 18
Now, to explain the operation of the above-mentioned device, the white light emitted from the white light source 1 is focused by the condenser lens 2 onto the pinhole of the pinhole plate 3. After being made parallel by the collimator lens 4, P is
The vibration direction of the wave component or S-wave component becomes only linearly polarized light, which passes through the entrance window 8 of the condenser 7 and is incident on the point to be measured on the measurement object 6. At this time, the light beam is focused by the entrance window 8 into a spot of 2 m and a diameter of m.

The light incident on the measurement object 6-L is reflected by the measurement object 6, and at this time, due to an interference phenomenon, the spectral intensity can change to 1 as shown in FIG.

This change depends on the film thickness d of the measurement object 6 and is 7 if the incident angle θ of the light and the refractive index n of the measurement object 6 are constant as described above.

Incidentally, the light incident on the measurement object 6 is normally reflected specularly, but if the measurement object has wrinkles or has a movement or inclination of 1:-T:, the direction of reflection may vary.

However, in this embodiment, the fiber axes of each end of a large number of optical fibers IO are arranged so that the point to be measured on the measurement object 6 is directed. , extremely efficient light reception is possible because the condenser 7 is used.
Guided by the separation lens 11, the extraction pinhole plate 1
The sorting lens 11 guided by the second pinhole separates only the components in the same direction of the light emitted from each other end of each optical fiber 10 at the position of the extraction pinhole plate 12. Then, the extracted reflected light is made into parallel light by a cremeter lens 13, and then guided to a spectroscope 14 where it is separated into light beams of each wavelength forming the same plane.

The light emitted from the spectrometer 14 is focused on the image intensifier 16 by the imaging lens 15, and
Light having bright and dark areas generated by the interference phenomenon is amplified and appears as a light image on the output surface.

This light is transmitted to the -
・It is incident on the end face, but at this time the fiber break;・
Each light incident on one end face of each optical fiber constituting 17 is transmitted only through the corresponding optical fiber and output to the other end face, so the light image has almost the same intensity ratio. It is output to the other end face of the fiber play eye 7. When imaging using a lens instead of the fiber plate 17,
The intensity of the light is greatly attenuated from a fraction to several tenths.

However, when the fiber plate 17 is used, there is only slight attenuation at the end face or inside the optical fiber.

In this way, the light image transmitted through the fiber brake)-17 is
The light is directly input to the linear image sensor 18 and detected as the intensity of light for each wavelength. The data processing device 19 is
The intensity of the detected light for each wavelength is taken in, and the wavelength of the bright or dark area caused by the interference phenomenon is determined. The film thickness d of the measurement object 6 is calculated.

FIGS. 3 and 4 show the wavelength λ (nm
) and the spectral intensity ■. Figures 5 and 6 show measurement examples showing the relationship between the wavelength λ (
nrn) and the spectral intensity I. This is a measurement example. Figures 3 and 5 are measurements of the flat part of the film, and Figures 4 and 5 are measurements of the flat part of the film. This is a case in which a portion with local 19-dimensional unevenness called 2 was measured.The objects to be measured were all polyester films, and the measurements were made using a stylus-type thickness meter.

The thickness was 13.6 μn1.

In these examples, the actually detected spectral intensity is normalized by the spectral characteristics specific to the device, the Zunawaji light source 1, the spectrometer 14, the linear image sensor 18, etc., and the interference phenomenon is Only the changes in spectral intensity received by the light are extracted. This method will be described later.

Note that a 100 W halogen lamp was used as the white light source 1. A plastic optical fiber with a diameter of 1 mm is used as the optical fiber lO.
This book is also arranged. Furthermore, the spectrometer 14 used a plane diffraction grating with a blaze wavelength of 750 nm and a groove count of 1200/mm. Further, as the linear image sensor 18, a 1024-bit PCD (Pla, sma Co
<upledDevice> element was used.

In the conventional device, if there is local thickness unevenness, the intensity ratio between the bright and dark portions of the spectral intensity, that is, the contrast, almost disappears, making it impossible to calculate the thickness from this, but with the device of the present invention, The peaks and valleys can now be detected and the thickness can be measured. It can also be seen that the contrast is improved even in flat areas.

Next, the film thickness calculation method in the data processing device I9 will be described in detail.

FIG. 7 shows the internal configuration of the data processing device 19 in the present invention, including a buffer amplifier 20, an A/D converter 21, a microcomputer 22, and an image sensor drive circuit 23.
It also includes an input/output device, a storage device, etc. (not shown).

FIG. 8 is a flowchart of arithmetic processing in the data processing device 19.

The calculation processing can be broadly divided into - as shown by the dashed lines A, B, and C. ) input (B) (C)
It consists of three parts: film thickness measurement routine (C);

Here, blank data refers to the spectral characteristics of the entire optical system, which is measured by placing a suitable reflector in place of the measurement object 6 in the measurement system shown in FIG.

That is, in a normal optical system, the light source, spectroscope, image sensor, etc. each have their own spectral characteristics, so the optical system as a whole has a spectral pattern. Therefore, the measured data is measured as a superimposition of the spectral characteristics of the optical system (the spectral waveform caused by the interference of the thin film).
In order to obtain the waveform of the spectral intensity resulting from the interference of the measurement target 6, it is necessary to measure blank data in advance and remove the influence of the spectral characteristics of the entire optical system by dividing the measured data obtained by the sample. be.

The actual procedure will be explained below based on the flowchart shown in FIG.

First, the blank data (step 1) is measured.
Prepare a plate made of the same material as measurement object 6 and thick enough to be unaffected by interference phenomena, and place it as measurement object 6.
Place it in the position and measure. This blank measurement is performed before starting the film thickness measurement and is stored in the storage device.

The reflecting plate at this time may be a mirror, but since the mirror and the sample have significantly different reflectances, the dynamic range when measuring the sample becomes narrow.

Therefore, it is preferable to use a plate made of the same material as the sample as the reflecting plate.

Although blank measurements are performed before starting measurement, it is preferable to periodically re-measure to compensate for changes in the light source lamp over time. FIG. 9 shows an example of measurement of blank data.

Next, enter the measurement conditions by inputting the approximate expected film thickness of the measurement object 6 (step 2), and calculating the dividing pitch for extreme value search based on this expected film thickness (step 3).
)do. This division pitch is a value determined so that the period of the interference waveform is divided into a constant number (approximately 10 divisions) according to the film thickness.

Further, from the expected film thickness, an expected wave number interval We given by the following equation is calculated (step 3).

Here, n represents the refractive index, θ represents the incident angle, and te represents the expected film thickness.

We predicts the wave number interval between adjacent maximum (minimum) values, and after detecting the extreme value of the spectral waveform of interference light, it is used to determine a normal part of the waveform where the film thickness can be measured with high accuracy.

Next, the film thickness measurement routine will be explained.

As shown in FIG. 8, the film thickness measurement routine is performed in the following steps.

(a) Spectral intensity measurement (b) Smoothing (C) Normalization (d) Create differential data (e) Extreme value interval detection ) Extreme value position interpolation (g) Extraction of normal parts fh,) Film thickness calculation The functions of each block will be explained in detail below.

In the spectral intensity measurement routine, the outputs of the image sensor 18 are sequentially read by the A/D converter 21 according to instructions from the microcomputer 22 (FIG. 7) of the data processing device 19 (step 4), and are used as raw measurement data.

Next, smoothing (step 5) is performed. There are various methods for smoothing, but it is preferable to use one that can be performed in a short time, such as a simple moving average. Also, set the number of points to take the moving average to 2.
” (n=1.2, . . .), it is preferable in terms of speeding up because the division can be performed by bit shifting. It is preferable to create a dark cell that covers the sensor so that it is not sensitive to light, and then subtracts the output of this dark cell from each output to compensate for the dark current of the linear image sensor 18.

Data smoothed in this way is called smoothed data. Next, the smoothed data is normalized by dividing it by blank data measured in advance by corresponding cell numbers (step 6). At this time, the smoothed data is multiplied by 2' (n-8, 9,
..., this operation is also performed by bit shifting) and then divides the integers. The obtained data is called normalized data.

FIGS. 10 and 11 show examples of smoothed data and normalized data when a polyester film having a thickness of 14.6 μm was measured using a stylus thickness meter.

Next, in the creation of difference data (step 7), the normalized data is scanned at predetermined division pitches to create difference data. If the number of the cell corresponding to the beginning of the valid portion of normalized data is 1 and the number of the cell corresponding to the end is N, then the fifth (I≦J≦(N-1)/h
) difference data D (J) is defined by the following equation.

D(1)=S (1+Jxh) -3(1+(J-1) xh) (3) Here, S (N) represents normalized data of cell number N, and h represents the division pitch.

Next, an extreme value existence section is detected based on the difference data (step 8). Since the local maximum value is the point where the differential coefficient changes from positive to negative, the area where the local value exists can be detected from the change in the sign of the difference data obtained by the above equation (3). FIG. 12 shows a flowchart for creating differential data.

In Figure 12, the cell number corresponding to the first valid portion of the normalized data is 1, and the cell number corresponding to the fifth valid portion is O.
(Step 1), the value obtained by adding the division pitch h determined by the expected film thickness to the cell number ■ is smaller than the cell number N corresponding to the end of the valid part of the normalized data (1+
h≦N) (step 2). When the discriminant answer in step 2 is affirmative (Yes), the number J is advanced by 1 (step 3), and the 5th difference data D(1) is divided between the 1st normalized data 5(1) and the (1+h)th Calculation is performed using the normalized data S (1+h) (step 4). Next, advance the cell number I by 1 (step 5
) Return to step 2. When such calculations are repeated and the discriminant answer in step 2 is negative (No), the number of differential data D
Set max to J (step 6) and end the calculation.

Next, the extreme value position is detected by interpolating the extreme value position using data before and after the extreme value existing section (step 9).

Normally, the sign of the difference data changes by 3 division pitch increments.
It is sufficient to perform quadratic interpolation using point data. By this interpolation, the position of the extreme value can be determined including between the cells of the image sensor 18.

In order to reread the position of this extreme value, it is necessary to know in advance the correspondence between the cell number of the image sensor I8 and the wavelength. This correspondence can be determined by placing an interference filter with a known wavelength on the reflection plate when measuring blank data, and determining the peak position.

Next, the normal region extraction method (step 10) will be described in detail.

There is a possibility of the following erroneous judgments in the extreme values obtained by the above procedure.

(a) If electrical and optical noise are superimposed, the noise will be incorrectly determined to be an extreme value.

(b) The spectral intensity waveform of the interference light is locally distorted due to factors such as local thickness unevenness and wrinkles of the measurement target, and a value different from the original extreme value is determined to be the extreme value.

(C) Extreme values in a portion where the waveform is distorted are not detected and are omitted, resulting in erroneous determination that extreme values that were originally separated are adjacent to each other.

In particular, when measuring an object with local unevenness using a spot with a small diameter, as with this device, there is a high possibility of these erroneous determinations, and if the calculation is performed as is, a large error will occur in the measured value. Therefore, it is desirable to use the method described below to determine the extreme values of the normal portion and the distorted portion of the waveform, and to calculate the film thickness using only the extreme values of the normal portion.

There are three methods for extracting normal parts:

The first method is to sequentially arrange the positions (wavelengths) of detected local maximum values and the positions (wavelengths) of local minimum values, and determine that a portion where local maximums and minimum values are not arranged alternately is an abnormal portion.

The detected maximum point and minimum point are pi (i=L2,...),
bi (i = 1.2, ...), and the position (wavelength) of the maximum value is λp + (t = 1.2, ..., λ□〉λPl
+1), the position (wavelength) of the minimum value is λb+(i=1.2
,...;λ1>λ1u+1).

Then, when these wavelengths are arranged in order of magnitude, there are missing or false detections in extreme value detection (non-extreme values are detected as extreme values).
If not, for example, λ, 1>λ, 1>λ, 2>λ, 2
〉・・・・・・, or λth〉λp+＞λ, 2〉λ,
2>..., λ, 1 and λ1 are arranged alternately.

If this order is out of order and, for example, λ, ; and λ, l+1 are lined up, it is assumed that there is a lack of extreme value detection between them, or that one or both of them is a false detection,
The area between pi and pi+1 (in terms of wavelength, λ, +1 to λpi) is defined as an abnormal section. Furthermore, in order to eliminate the possibility of abnormalities,
The minimum values on both sides adjacent to pi and pi+1 are bk and b
When k+1, between bk and bk+1 (in terms of wavelength,
λbk+I to λbk) may be defined as an abnormal section. If other methods described below are not combined, it is better to assume a wider abnormal interval. λ,1 and λ1. The same applies to cases where +1's are lined up or three or more maximum (minimum) values are lined up.

The second method is between the detected local maximum values or between the local minimum values,
Alternatively, the wave number interval between the maximum and minimum values is compared with the upper and lower limit values obtained from the above-mentioned expected wave number interval We, and a maximum (small) value outside this range is determined to be an abnormal part.

The case where the determination is made based on the wave number interval between local maximum values will be described in detail below.

The position (wavelength) of the detected maximum value is λ□(i=1.2,・
...;λ2. >λpl+1). The position of each local maximum (
(wavelength), the wave number which is the reciprocal of each is determined, and the wave number difference Wk between adjacent maximum values is calculated according to the following equation.

λjk+l λpk Here, (k-1, 2, 3,...) When the error tolerance rate is ±a'+(at >o) for the wave number interval We determined by equation (2), the value of Wk is The allowable range is determined by the following formula.

We(l at)≦Wk≦We(1+a+)
-(5) For a value Wk that does not satisfy this formula (5), it is assumed that there is an abnormality in either or both of the maximum parts pk and pk+1 on both sides, or that the maximum part between them is missing, and both The area between the maximum parts pk and pk+1 (in terms of wavelength, λ, +1 to λ1) is determined to be an abnormal section, and is not used for film thickness calculation.

Note that when determining based on the wave number interval between the maximum value and the minimum value, it is preferable to combine the first method described above and arrange the positions of the maximum values in advance. Note that in determining the wave number difference Wk' between the maximum and minimum values, a value We' obtained by multiplying this by 172 is used instead of We.

The third method is to find the difference between adjacent maximum and minimum values (corresponding to the amplitude of the interference waveform), and determine that a value that is significantly different from the others is an abnormal portion.

The detected local maximum value (intensity) is I++ (t=1.2,...
), and the minimum value is L+(i=1.2, . . . ).

In addition, the order is that each position (wavelength) is λ, l〉λp
It is assumed that they are arranged so that l+++ λ1>λbl+I.

At this time, the intensity difference 11 between the peaks and valleys is determined according to the following equation (6).

1 to 1 □ - 1b & ... (6
) (k=1.2,...,) Then, among the obtained ■, the one that is significantly different from the others is that there is an abnormality in either or both of the maximum points pk and bk, and the difference between both pk and bk is (In terms of wavelength, λ-2
．． ~λpk or λ□~λpk) is determined to be an abnormal section and is not used for film thickness calculation.

To determine Ik, the average value of all I's obtained is 1. For the error tolerance rate a2, L(lax)≦I,≦r
−(1+ a 2) −(71 Either consider ik that does not fit in the above equation (7) as abnormal, or consider I
Find the variance σ2 of , 1, (1-2σ)≦I, ≦I,
(1+2σ) ...(8) I that does not fit into the above formula (8)
There are methods such as making h an abnormality.

Alternatively, in combination with the first method described above, maximum points and minimum points may be arranged alternately, and the peak-to-valley difference may be determined only for the points in the correct order.

The three types of determination methods described above may be used alone or in combination. Among the detected extreme values, a portion excluding the section determined as an abnormal section by these determination methods is set as a normal section, and the film thickness is calculated using only this normal section. Alternatively, the film thickness may be calculated by extracting only the longest continuous normal section and using only that part.

Table 1 shows the local maximum detected from the normalized data in Figure 11,
This is the wavelength of the minimum value. When applying the second method of stating the wave number interval between maximum and minimum values, n=1. ．． 6, θ=
yr/6 (rad), expected film thickness 14.6 μ
For m, the expected wave number interval is We'= -We =11266 (1/m)
-(9) If the allowable wave number interval is 20%, the upper and lower limits of the wave number interval Wk are 9013≦Wk≦1.351.9, and the waveform in Fig. 11 continuously satisfies this condition and is the longest continuous one. This is five cycles from the long wavelength side.

Table 1 or the wavelengths of two adjacent maximum points and minimum points are λ□,
When λ4゛(λ,〉λ□′), film thickness calculation is performed using the waveform of two adjacent maximum points (or minimum points) for the extreme values included in the same continuous normal region (including both ends). When λ4 and λff1a1 are used, the calculation is performed according to the above equation (1).

...Calculate according to (10). This method is suitable for cases where the membrane thickness is thin, the area is normal or small.

When a normal region contains multiple poles and multiple pairs of adjacent poles can be selected, there are multiple film thicknesses that can be obtained in this way, but these are processed using methods such as averaging to determine the film thickness. .

The film thickness calculated from the above measurement example was 14.57 μm.

All of the above procedures are controlled by the microcomputer 22 and are performed automatically. If the sample to be measured is different, it is sufficient to start again by inputting the expected film thickness.

FIG. 13 is a diagram showing the relationship between film position and thickness measured with this apparatus.

The thickness measured with a stylus thickness gauge is 4.371 m.
The thickness of the polyester film was measured at intervals of one cylinder in the width direction of the film, and the results obtained using the above film thickness calculation method are plotted. From the figure, it can be seen that local thickness changes are captured and there is "C".

In this example, the spot diameter was set to 2 mm, but the spot diameter may be made smaller in order to measure even finer changes. In this case, it is preferable to make the diameter of the optical fiber 10 of the condenser 7 thinner and denser, with the spot diameter being 172 mm or less, depending on the spot diameter.

(Effects of the Invention) As explained above, according to the present invention, parallel white light is made incident on a measurement object at a constant angle of incidence, and the reflected light or transmitted light from the measurement object is guided to a spectrometer, and its spectral intensity is A device configured to measure the film thickness of a measurement target by detecting the white light, wherein a shielding body having a hole for regulating the outer shape of the light beam irradiated onto the measurement target is disposed on the incident optical path of the white light, and The spectrometer includes an image intensifier, a photoconductive member formed by bundling a plurality of optical fibers, and a photoelectric conversion element, and one end surface of the photoconductive member is connected to the output surface of the image intensifier. By arranging the other end face in close contact with the incident surface of the photoelectric conversion element, even if there is local thickness unevenness in the measurement target, the intensity ratio of the bright part and the dark part of the spectral intensity, that is, the contrast, can be maintained. Can be made larger. As a result, it becomes possible to accurately detect the wavelengths of the bright and dark areas that contain film thickness information, and the excellent effect of improving positional resolution and making it possible to perform highly accurate film thickness measurements is achieved. be.

[Brief explanation of drawings]

Fig. 1 is a schematic diagram showing an embodiment of the optical interference type film thickness measuring device according to the present invention, Fig. 2 is a schematic diagram showing the interference waveform of spectral intensity due to a thin film, and Figs. Graphs showing measurement examples using the invented device; Figures 5 and 6 are graphs showing measurement examples using the conventional device; Figure 7 is a schematic configuration diagram of the data processing section; Figure 8 is the calculation processing necessary for film thickness measurement. Figure 9 is blank data, Figures 10 and 11 are stylus-type film thickness measurement results of 14.6 μm.
Figure 1 shows smoothed data and normalized data when the film thickness of a polyester film of
FIG. 2 is a flowchart for creating differential data, and FIG. 13 is a diagram showing thickness unevenness when a polyester film having a stylus-type film thickness measurement result of 14.3 μm is measured using the apparatus of the present invention. 1... White light source, 2... Condenser lens, 3...
・Pinhole plate, 4... Collimator lens, 5...
Polarizing plate, 6...Measurement object, 7...Concentrator (shielding body)
, 8... Entrance window (hole), 9... Support, lO...
Optical fiber, 11... Lens for classification, 12... Pinhole plate, 13... Collimator lens, 14... Spectroscope, 15... Imaging lens, 16... Image intensifier, 17 ... Fiber plate (photoconductive member), 18... Linear image sensor, 19... Data processing device. Applicant Higashi Co., Ltd. Company Representative Patent Attorney Kan Nagado Figure L-Nico/11! l Comparison of light of light

Claims (1)

[Claims] Parallel white light is incident on the measurement target at a constant angle of incidence, and the reflected or transmitted light from the measurement target is guided to a spectrometer.
The apparatus measures the film thickness of the measurement target by detecting the spectral intensity, and the apparatus includes a shielding body having a hole for regulating the outer shape of the light beam irradiated onto the measurement target on the incident optical path of the white light. and the spectrometer includes an image intensifier, a photoconductive member formed by bundling a plurality of optical fibers, and a photoelectric conversion element, and one end surface of the photoconductive member is connected to the image intensifier. An optical interference type film thickness measuring device, characterized in that an output surface is disposed with the other end surface in close contact with the entrance surface of the photoelectric conversion element.