JPS5973712A - Flatness measuring device - Google Patents

Abstract

PURPOSE:To peform highly accurate flatness measurement by a simple constitution, by arranging a plurality of light sources, light guiding means, and light detecting means on approximately the same plane, respectively. CONSTITUTION:Luminous flux, which is outputted from a light source 1 such as a semiconductor laser and an LED, is transmitted through a beam splitter 2, converged by a condenser lens 5 such as SELFOC lens and the like, and inputted into an optical fiber 6. The luminous flux, which is projected from the emitting end of the optical fiber 6, is converged on a material to be measured 17 by a condenser lens 16. Under the focussed state, the magnitude of the luminous flux on the material to be measured 17 becomes the minimum value. The luminous flux, which is reflected by the material to be measured, goes backward on the light path and inputted into the optical fiber 6 again. The reflected luminous flux is emitted from the optical fiber 6, transmitted through the condenser lens 5 and the beam splitter 2, and inputted into a light detector 3. Thus the signal, which is used for detecting the position of the material to be measured is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a flatness measuring device, and in particular to a flatness measuring device, which measures an object to be measured having an irregularly uneven shape by applying an external force to the flatness to an extent that does not cause any practical inconvenience. The present invention relates to a flatness measuring device having a function of performing correction.

A flatness measuring device having such a function is useful in many fields, but it is most effective in the field of semiconductor manufacturing. In recent years, with advances in semiconductor manufacturing technology, the size of silicon wafers has tended to increase.

In contrast, the required minimum pattern size is becoming smaller, and higher precision is required for the flatness of silicon. In other words, in projection printing, it is necessary to maintain the entire surface at the required depth of focus, and even when printing is performed through minute gaps, the printing is done in the form of uniform gaps over the entire surface. is inconclusive.

On the other hand, it is difficult to completely prevent warpage, deformation, etc. of the wafer itself through each process of semiconductor fabrication because the stress of the layer formed on the surface is large, and for reasons of economy and handling, the wafer thickness This is extremely difficult because it cannot be made that thick. Therefore, there is a need for a technology to improve the surface roughness of the wafer entrance surface to a level that does not cause any practical problems, by simultaneously applying an appropriate external force while holding the wafer to correct the unevenness of the wafer. Become.

Conventionally, as a method for measuring surface irregularities of sea urchins, etc.
Thousands of methods have been devised and implemented. One method is to obtain contour lines on the wafer surface using a traverse meter or a moiré method.

Since these methods essentially detect contour lines, converting the unevenness into numerical values requires sophisticated information processing to decipher the interference fringes, and a large number of two-dimensional This method had drawbacks such as the need for a photodetector. Another drawback is that when flatness is measured during a semiconductor process, the shapes of contour lines become unclear due to the influence of already formed patterns.

Another method for measuring the flatness of a wafer is to use a Fresnel zone plate disclosed in Patent Application Publication No. 143052/1985, which focuses a laser beam onto the surface to be measured and converts the diffracted light beam into a diffracted beam. There is a method to confirm that the gap is adjusted to a predetermined distance by forming intersecting fringes with the light beam that does not occur. This method allows measurement of multiple points at the same time, and also allows for the easy production of a large number of standard zone plates on a single substrate using highly productive methods such as step-and-repeat. It has advantages.

However, this method requires a large-diameter optical system, resulting in an expensive and large measuring device, and requires a large number of photodetectors to detect individual interference fringes. This method has the disadvantage that complex signal processing is required for signal discrimination.

As another method for detecting flatness, a patent application published in 1972
There is a method shown in No.-148147. This method is
This method uses two-dimensionally arranged fly-eye lenses to detect flatness from the intensity of reflected light, but silicon sea urchin...
It has been difficult to improve the measurement accuracy to the level required for flatness measurement.

The purpose of the present invention is to improve the drawbacks of conventional flatness measurement, such as the complicated structure and the influence of semiconductor patterns, and to provide a flatness measurement method that is thin and capable of high-density arrangement. The goal is to provide the following.

The present invention is also based on the patent application published by the same applicant on 1986-61.
This is related to the improvement of No. 62.

First, the content of the prior patent will be explained using FIG.

In the figure, a luminous flux from a light source 11 is incident on an illumination fine nozzle W13a attached to a light shielding plate 12. Fiber Al 3 at + constituting illumination fiber J 13 a
13 at +... and optical detection fiber -
13b, 13b, . . . are placed in a substantially image forming relationship with the object to be measured 17 by the lens 16 in FIG. 1(b). The light flux guided by the optical detection fiber 13b is detected by the optical detector 15.

Now, when the object to be measured 17 is completely imaged with the exit surface end A14 of the optical fiber, the fiber 13
The light beam emitted from al + 13 at + . Furthermore, when the object to be measured 17 moves to an out-of-focus position, the fibers 13b, 13b, 1
A light flux is incident on 3b, . . . , and the amount of light is detected by the photodetector 15.

This can be realized by arranging a plurality of - end faces 14 or by arranging a plurality of such detection devices 10. However, the former method requires a lens that has good imaging performance over the entire wide wafer surface, which inevitably results in a large and expensive measuring device. Furthermore, implementing the latter method has disadvantages in that the number of parts increases and position adjustment among them is very complicated.

The present invention improves on the prior patent and provides a means for measuring flatness with high accuracy despite its simple configuration. The present invention will be explained in detail using a first embodiment of the present invention shown in FIG.

A light beam emitted from a light source 1 (FIG. 2b) such as a semiconductor laser or LED passes through a beam splitter 2, and then is condensed by a condensing lens 5 such as a SELFOC lens.
Optical fiber (-6) The optical fiber used here is not only a normal multi-mode fiber, but also a single-mode fiber, special fibers such as linearly polarized light, circularly polarized light, and other fibers with different polarization states. A polarization-maintaining optical fiber that transmits data without disturbance is used depending on the purpose.This optical fiber enters the gradient index condenser lens 16.The manufacturing method of this gradient index condenser lens is described in, for example, a patent application. As shown in Publication No. 135806 of 1982 and Publication No. 53702 of 1983, similar to the semiconductor manufacturing process, after buttering a suitable mask material, a refractive index distribution is formed by techniques such as ion exchange. The light beam emitted from the exit end of the optical fiber 6 is condensed onto the object to be measured 17 by the condenser lens 16. In the focused state, the size of the light beam on the object to be measured 17 Vi is minimum. The light beam reflected by the object to be measured travels the optical path in the opposite direction and enters the optical fiber 6 again.

This reflected light beam exits the optical fiber 6, passes through the condenser lens 5 and the beam splitter 2, enters the photodetector 3, and becomes a signal used to detect the position of the object to be measured. If necessary, a photodetector 4 monitors the output of the light source.
is used.

In Fig. 2, optical fibers 6, 6', 6'' and condensing lenses 16, 16', 16'' are arranged in a -dimensional manner, and the object to be measured is divided into small pieces and held by vacuum suction (not shown). Holder 18.18'.

18". Each holder 18°18'
, 18" is the actuator rotor through the ball bearing,
Rad l<'1.19', 19'' is connected.

As a result of such a structure, the local unevenness of the object to be measured 17 is caused by the actuator rod 19, 19'.

1? You can correct it as much as you want by taking it in and taking it out.

FIG. 3 is a diagram showing a form in which the flatness measuring device of the present invention is arranged three-dimensionally. Both the gradient index lens and the optical fiber mounting portion are processed on their respective flat plates by a single two-dimensional patterning process.

Therefore, by using the same method as mask alignment in a normal semiconductor manufacturing process, assembly with alignment accuracy of several micrometers can be achieved at once.

FIG. 4 is a diagram showing how the position of the object to be measured is detected according to the present invention. The horizontal axis represents the amount of defocus, and the vertical axis represents the amount of detected light. When the object to be measured 17 is in focus, the amount of detected light is at a maximum (Imax), and when out-of-focus occurs in the front and back, the amount of detected light decreases. This decrease in light intensity occurs because the reflected light beam is spread out because it is out of focus at one end of the fiber, but this detection sensitivity can be increased or decreased as necessary. It is possible. - for example, a semiconductor laser light source with λ = 0.83ttm;
Using a single mode fiber with a core diameter of 5 μm,
N, A, when condensing at a focal length of 0.6 h, the amount of defocusing that reduces the amount of detected light by half Z (]/2) 4'i approximately 0.
Focus detection is performed with extremely high precision of 3 μm.

Using the present invention, the object to be measured 17 is a desired plane.

The simplest method is a method that uses the absolute light sensitivity of the detected luminous flux. By driving each actuator rod 19 back and forth, each photodetector 3
The output from increases and decreases according to FIG. Now, the reflection customization for the incident light at each point on the wafer is the same, and the amount of light detected by driving the output direction of the same size from all photodetectors when in focus is a specific value, for example (I max
/2). As is clear from FIG. 4, there are two out-of-focus positions t where the light amount is (Imax/2), but this determination is made based on the direction of increase/decrease in the detected light amount with respect to the insertion and removal of the actuator rod 19. I can do it.

In addition, as a simple method, the amount of detected light is the maximum value I ma
There is also a method that uses the fact that it is sufficiently close to x as a means of checking.

For example, this method requires that the amount of detected light be 95X or more of Imax, and it is easy to determine whether the flatness is good or bad.

As a second control method, there is a method of detecting the maximum amount of light detected by each photodetector [2], storing the value, and performing the above-mentioned control. When this kind of control is performed, even if the surface structure of each point on the wafer irradiated with each light beam is different, and as a result the reflectance varies locally, it is possible to accurately maintain a flat surface. Adjustments can be made.

As a third control method, there is a local maximum value detection method used in autofocus of optical discs and the like. Actuator - drives the rod 19 back and forth at a specific frequency;
Looking at the output of the photodetector 3 in response to this, the signal output applied to the out-of-focus position of zNO fluctuates at the same frequency as the frequency of the actuator rod 19, and its sign is reversed depending on the front focus and rear focus. On the other hand, Z=.

At the in-focus position, the output component of the same frequency as the driving frequency of the actuator rod 19 becomes zero, and a frequency component twice as high as that of the actuator rod 19 is generated.

The three signal detection methods described above can be used depending on the uniformity of the reflectance of the object to be measured 17, the degree of inclination of the surface, etc.

In the present invention, the ability to follow the movement of the object to be measured, which is a problem in cine cameras, autofocus of optical discs, etc., is not a problem at all.

The driving speed or driving frequency of the actuator rod 19 may be determined according to the allowable time for performing the necessary flatness correction.6 Also, once the flatness is controlled to a specified level, by fixing the actuator rod 19, Baking allows achieving flatness in the process. In addition to the surface of the wafer being flat, the semiconductor printing must also be placed in line with the correct imaging plane of the optical system. In the present invention, the arrangement surface of a plurality of arranged lenses is placed at a predetermined position with respect to the printing optical system.

FIG. 5 shows a second embodiment of the present invention in which a hologram lens is used as a condenser lens. The condensing lenses 16, 16' and 16'' used here are hologram lenses made of volume hologram lenses such as dichromate gelatin, and have the same light condensing ability as ordinary lenses. .

This hologram lens can be easily mass-produced by applying a photosensitive material onto the substrate 8 and then using an optical step-and-repeat method or an optical system in which multiple lenses are fired simultaneously. Further, like the above-mentioned gradient index lens, since the individual lens pitches are formed with a positional error of several μm or less during manufacturing, the optical axes of all lenses can be aligned at the same time using the mask alignment method.

Also shown in FIG. 5 is a method for realizing front-back direction discrimination of out-of-focus conditions. Now, two wavelengths λ3. Suppose that the light of λ is transmitted. Unless a normal lens is specially achromatized, there will be chromatic aberration on the optical axis, and the focal point of the same object will shift in the optical axis direction.

FIG. 5 shows a case where the surface of the object to be measured 17 is in perfect focus with respect to the light beam having the wavelength λ. At this time, for the light beam having the wavelength λ, the exit end face of the optical fiber 6 and the surface of the object to be measured 17 are located at a conjugate position in an imaging relationship through the lens 16. On the other hand, for light of a different wavelength λ2, the imaging position shield becomes the surface indicated by the broken line 120 in the figure, and for the light beam of this wavelength, the object to be measured
7 is out of focus. Letting this defocus amount be Δ, this value is proportional to the focal length of the lens 16. Furthermore, in a hologram lens, the focal length of the lens is inversely proportional to the wavelength used, so it is possible to create a large defocus with a very small change in wavelength.

Through the same optical fiber 6, two wavelengths λ8. λ,
The wavelength multiplexing method used in the field of optical communication can be used as a method for transmitting and independently detecting. FIG. 5 (bl) shows an example of this wavelength multiplexing technique. Light beams of two different wavelengths λ1.λ are transmitted through the optical fiber 6 and enter the glass block 20. This light beam is collimated. It passes through the lens 21, becomes a parallel light beam, passes through the prism 22, and enters the diffraction grating 2.
3. The traveling direction of the light beam diffracted by the diffraction grating varies greatly depending on the wavelength of the light, so the wavelength λ
After diffraction, the light flux of , is condensed by the collimation lens 21 and guided to the optical fiber 119, and the light flux of wavelength λ is guided to the optical fiber 119'.

Figure 6 1b+ shows how the luminous flux is split according to the wavelength, but if the traveling direction of the light is reversed, this can be used as it is to combine the luminous flux from light sources with two different wavelengths. can.

Figure 6 is a diagram showing the principle by which the direction of defocus can be detected by using two wavelengths, and the horizontal axis represents the amount of defocus;
The vertical axis indicates the amount of light detected. wavelength λ, and wavelength λ
The focusing position for the light beam of wavelength λ is shifted by a length Δ with respect to the amount of defocus 2, so the detected light intensity peak of the light beam of wavelength λ occurs at 2=Δ/2, while the peak of the detected light amount of the light beam of wavelength λ, The peak of the amount of detected light occurs at Z=-2Δ. 4. In the region where Z>0, the detected light amount I (c) of the luminous flux with wavelength λ is smaller than the wavelength λ.
The detected light quantity I(λ1) of the luminous flux of 1 is also larger, and the opposite is true in the region of z<0. Taking the difference between the two, Figure 6 1b
As shown in +, the direction of out-of-focus can be detected from the sign of the difference signal.

FIG. 7 is a diagram showing a third embodiment of the present invention.

On the substrate 24 are surface-emitting semiconductor lasers 1.1', 1.
' and photodetectors 3.3', 3' are created together with their power supply and amplification circuits (not shown). The light beam emitted from the laser 1 travels through a birefringent crystal plate 25 cut out with the crystal axis oriented in the approximately 45° direction. If the linearly polarized light emitted from the laser corresponds to the ordinary ray of this crystal, the light beam will proceed according to the normal law of refraction. The linearly polarized light that has passed through the crystal plate 25 passes through a quarter-wave plate 126 and becomes circularly polarized light 9, and is focused onto the object to be measured 17 by the aspherical molded lens 16. This aspherical molded lens is integrally formed as a -dimensional or two-dimensional fly's eye lens. Measured object 1
After the light beam reflected by 7 passes through the lens 16, it passes through the quarter-wave plate 126 again, and the linearly polarized light that is initially emitted from the semiconductor laser is linearly polarized light that has a plane of vibration in a direction perpendicular to the linearly polarized light emitted from the semiconductor laser. Therefore, it becomes an extraordinary ray with respect to the crystal plate 25. As a result, the reflected light beam enters the photodetector 3 through a different optical path from that of the light beam emitted from the laser. If the spatial distribution of the sensitive skin of the photodetector is appropriate, out-of-focus information can be detected from its output.

This configuration eliminates the need for components such as optical fibers and beam splitters. Furthermore, since the semiconductor laser and photodetector are two-dimensionally formed on the same plane, individual components can be eliminated. There is no need for alignment, and the advantage is that it can be formed ultra-thin.

A fourth embodiment of the present invention in which the illumination light flux and the detection light flux are separated is shown in FIG. The light beam from the surface-emitting semiconductor laser passes through the substrate 8 and passes through the off-axis hologram lens 1.
6. The off-axis hologram lens 16 is
It has a function of focusing a light beam onto the object to be measured 17 from an oblique direction. The reflected light flux enters a hologram lens 26 formed adjacent to the hologram lens 16, is diffracted, and enters photodetectors 33 and 34. In this embodiment, since the light beam obliquely enters the object to be measured 17, the light spot generated on the photodetector causes lateral movement according to the amount of defocus. Therefore, a two-part photodetector 33°3 as shown in FIG.
By using 4 and taking the difference, it is possible to determine the direction of out-of-focus.

As described above, the converging elements in various embodiments of the present invention include a gradient index lens, an in-line hologram lens, an off-axis hologram lens, and an aspherical molded lens. Any optical system can be used as long as it has imaging performance.

9 and 10 show examples in which an erect equal-magnification imaging optical system is used as the imaging optical system in the present invention. In FIG. 9, the light flux emitted from the exit end of the optical fiber 6 is
The light is focused onto the object to be measured 17 by a gradient index imaging lens (product name: SELFOC lens), and the trench-reflected light beam is focused onto the optical fiber 6 as in the previous embodiment. . When such an erect equal-magnification imaging system is used, a plurality of imaging elements cooperate to form an image, and by appropriately arranging them, it is possible to improve the accuracy of out-of-focus detection. However, it still has the advantage that alignment of the optical axes of the imaging optical system and the optical fiber is basically unnecessary.

In the embodiment shown in FIG. 10, the optical fiber 6 is placed at the boundary between the gradient index lenses 27 and 27'.

While there is a loss in light quantity, the change in light quantity when out of focus becomes large.

As described above, the present invention has produced the following effects that cannot be achieved with conventional flatness measurement methods.

First of all, it is possible to realize a flatness measurement method that is not affected by the process even during the semiconductor manufacturing process. In the present invention, the size of the measurement point required to measure the flatness of the object to be measured can be set to a minimum diameter of about 1 μm, making it easy to avoid patterns on the wafer. In particular, stable flatness measurement can be achieved by setting a specific portion of each square 1p as a flatness measurement area or by using a scribe line used during cutting as a measurement point.

The second effect is 1. It is an object of the present invention to provide a thin and stable means for measuring flatness. Since the present invention has a configuration in which position measuring devices for minute objects to be measured are arranged in parallel,
Even if the size of the wafer increases from 5 inches to 6 inches, finches, etc., there is almost no need to increase the thickness of the measuring device.

Furthermore, since the present invention combines optical elements arranged at a predetermined pitch on a flat plate, it is difficult to align optical axes, which was a major factor in increasing the cost of conventional optical systems. The effect is great because the procedure only needs to be done once. 1

[Brief explanation of the drawing]

Figure 1 is a diagram showing the prior patent of the present invention, Figure 2 is a diagram showing the first embodiment of the present invention, and Figure 3 is the first embodiment of the present invention. 11kt as a perspective view, FIG. 4 is a diagram showing the out-of-focus signal detected by the present invention, FIG. 5 is a diagram depicting the second embodiment of the present invention, and FIG. 6 is a diagram depicting the second embodiment of the present invention. FIG. 7 is a diagram showing the out-of-focus signal detected in the second embodiment of the invention, FIG. 7 is a diagram showing the third embodiment of the invention, and FIG. 8 is a diagram showing the 40th embodiment of the invention. Figures 9 and 10 are diagrams showing modified examples of the embodiment of the present invention. In the figure, 11 is an incandescent light source, 12 is a light shielding plate, 13 is an optical fiber, 16 is a condensing lens, 15 is a photodetector, 6.6',
6" is an optical fiber, 7 is a light shielding plate, 8 and 9 are glass substrates, 1 is a light gamma such as a semicircular laser or LED, 2 is a beam sinter - 13°, 4 is a photodetector, 18.18', 18 “
19, 19', 19'' are actuator rods, 23 is a diffraction grating, and 26 is an off-axis hologram lens. Applicant: Canon Corporation F, Figure 7 (α) Sieve? (2) (α ) Su (b) Anger 11 ¥5 (α) (b)

Claims (1)

[Scope of Claims] (1) A non-contact position detection method comprising a light source, a means for guiding a light beam from the light source to an object to be measured, and a means for detecting a light beam reflected from the object to be measured, A flatness measuring device characterized in that a plurality of light sources, light guide means, and light detection means are arranged on substantially the same plane. (2. The flatness measuring device according to claim 1, wherein the light guide means is a hologram lens arranged on the same substrate. (8) The flatness measuring device according to claim 1) The flatness measuring device according to claim 1, wherein the light guide means is a gradient index condensing element having a parallel plate outer shape. (4) Scope of Claims The flatness measuring device according to claim 1, wherein the optical element is a multi-lens that does not integrally constitute a plurality of lenses. The flatness measuring device according to item 1, further comprising means for deforming the object to be measured, corresponding to the position detection means for the plurality of objects to be measured, and correcting the shape of the object to be approximately flat. A flatness measurement device with seven functions.