JPS5892805A - Detector for inclination of plane plate - Google Patents

Abstract

PURPOSE:To measure the inclination of a mask stably with a simple constitution, by irradiating an array of Fresnel zones, which is formed on the rear face of the mask, with rays and observing an optical image converged by the reflected light of a wafer. CONSTITUTION:Eight Fresnel zone plates similar to a zone plate 13-8 are arranged on the rear face of a mask 11 to form one set of a zone array 13. When the mask 11 is irradiated with a parallel light from above, the third-order focus of the zone array 13 is observed as a light spot by the reflected light from a wafer 12. If the mask 11 and the wafer 12 are parallel with each other, only a light spot 16 looks bright. If the wafer 11 is inclined, plural zone plates have bright centers. Thus, the inclination of the wafer 11 is measured stably with a simple constitution.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a precise inclination detection device for measuring the inclination of a flat plate. For example, among IC and L8I pattern printing methods using a photomask, measuring and controlling the relationship between the photomask and the wafer is one of the important issues in the proximity method and the contact method. That is, in the proximity method, 1 line between the mask and the sea urchin and 7 lines during exposure, and in the contact method, 7 lines are made.The gap between the mask and the wafer must be maintained at a set value of several S to several tens of microns. . If the mask and wafer are too close to each other than the set interval during alignment, there is an increased risk of contact between the mask and wafer, which may damage not only the wafer but also the mask assembly. The pattern on the wafer and the wafer may not be able to enter the focal point SJt of the 7-alignment microscope at the same time, or the position of the noodles may be uneven, that is, the pattern may be tilted, which tends to cause errors in alignment. In the trench proximity method, if the relationship between the mask and the wafer exceeds a predetermined value during exposure and printing, the diffraction phenomenon becomes noticeable and the pattern cannot be printed accurately.

Furthermore, if a tilt occurs between the mask and the wafer during printing using short-wavelength electromagnetic waves in which the diffraction effect is extremely small, a difference will occur between the printed pattern and the mask pattern.

Since the method of applying a contact to a wafer or mask may damage the contact portion, it is desirable to measure the gap and inclination without contact. Conventionally, such non-contact Ml determination of the gap has been carried out using a method using an air micrometer or a change in electrostatic capacity, or optically by making parallel light 41 incident obliquely on the mask and wafer. This has been done by adding the amount of positional deviation t611 of the nine rays reflected by the wafer.

Of these methods, the first and second methods require calibrating the reading of the measuring device and the gap amount.

In the third method, measurement is performed using the lateral shift of the light beam, which does not require calibration in principle, but if you try to increase the measurement sensitivity, the angle of incidence will increase and the measurement optical system will become larger. Inevitable. The diameter of the wafer currently used is 5 inches, and even in that case, at least three measurement points are required, so if s"xi = mini sensor, it becomes difficult to implement the device.

An object of the present invention is to provide a tilt detection device that is easy to calibrate, does not require a large measurement sensor, has a simple structure, and is capable of stable tilt measurement. .

Next, an embodiment of the present invention will be explained with reference to the drawings, but before that, the measurement using an Ht-Fresnel plate chamber between two flat plates will be explained based on FIG. 1. Fig. 1 (4) shows the fresnel band plate 3t attached to the mask 1 for the connection between the mask 1 and the wafer 2 [t-6111m] and the stripe 1 diagram (
B) ri is a cross-sectional view thereof.

In FIG. 1, the Fresnel zone plate a has a median value r rh of the n-th zone from the concentric circles.
“, 1==l, z, 3 ”・・””・(Z)
A customary area delimited by a circle as represented by [.

Light transmitting portions 3a and non-light transmitting portions 3b are arranged alternately from the center. There are two types, one in which the center is formed with a light-transmissive portion 31 and one in which it is not, but the one that is used in the present invention is the one in which the center is formed as a light-transmissive portion 3a.

In Figure I/I41, r indicates the radius from the center of the transparent part 3a to the circumferential trench of the first non-transparent part.The formation of the non-transparent part 3b is done by uniformly applying 0% chromium on the mask 1. .3s
This is done by depositing it thickly and removing the transparent portion 3a1r. The lens action of the Fresnel zone plate 3 has been known for a long time.

There are an infinite number of focal points for one annular plate.

Its focal length is It is. It's a friend, and λ is the wavelength of light.

It has been confirmed that if the number of rings, that is, the maximum value of n, is tN, then if N is about 10 or more, there will be an explosive effect. -If you just observe it, it is possible even when N times 10. In addition, acid is clearly observed at the primary focus, and becomes blurred as it becomes a higher-order focus. The third focal point is clearly -mJ at 8 to 10 degrees. In Figure 1 (B), the circuit pattern of mask 1 ( Fresnel annular plate 3t formed on the same surface as (not shown)
-Illuminate. Transmissive portion 3at of Fresnel zone plate 3 - The transmitted light beam has a focal length 1111It of each order expressed by equations (2) (2) to (5) - for example, a primary focus (not shown),
The tertiary focus 4.5 converges on the hot spot 5, but if the focal length is longer than the gap between the mask 1 and the wafer 2, it will be reflected by the smooth surface of the wafer 2.

Upon reflection, the amount of light is somewhat reduced, but the light passes through the same location as when it is reflected on the m-plane, and is focused uniformly in space. Using a microscope 7 (not shown) equipped with an objective lens with a large numerical aperture, that is, a shallow depth of focus, the Fresnel zone plate 3? - It can be seen that when the focal point is in the same plane as the Fresnel zone plate 3, that is, the focal length is in the gap.
When it becomes I, a bright light spot 4 can be seen at the center of the Fresnel zone plate 3 to be struck as shown in FIG. 1 (4).

To explain in more detail, as in the case of pattern printing for IC and L8I, it is necessary to add a gap t-J of 10 μm, and the focal length of Fresnel's board 3 must be around 20 μm, which is twice the gap. do. As mentioned above, the first-order focus area is clear, so I would like to use this as much as possible, but the wavelength is 0.546 μm.
When observing with light of (mercury line), fl = 204 m
Then, from equation (2), rl==3.3mm,
If N=10, the outermost ring zone (f)rpQ is (1)
From the formula, rlO−rsW o, 53 pm,
This is almost the limit size for current mask pattern forming technology. Therefore, consider using a third-order focus, which has a wider annular zone than when using a first-order focus.

If the wavelength is α546 and t@=20s copper, (3)
From Equation 1 and Equation (1), r te −rs =0.91 am, which is still a technical limit value, but is easier to form than using one hot spot. The technology for forming such fine patterns is currently in the development stage, and the transparent part 3ai9 of the Fresnel zone plate 3 is similar to the band t- of the non-transparent part 3b.
When actually formed, the width may be slightly thicker than the designed value, i.e., bIlf8<n>. Such a slight increase or decrease in width will not affect the focal length if the edge of the band increases or decreases equally on both sides, but will only reduce the contrast of the nine converged light points. Of course, the number 1071111
When performing the above-mentioned light measurement, it is preferable to use a first-order focal point because it allows a brighter point of light to be seen. When using a tertiary focal point, the wafer 2t is approached from the mask 1, and the light point of the primary focal point increases at the beginning, and the light point of the tertiary focal point increases at the beginning. Dharma. It can be seen that if the focal length 1lII of the tertiary focus is twice the set gap, the gap will be at the set value when the tertiary focus is established.

Next, an embodiment of MX1 according to the present invention will be explained based on FIG. As shown in FIG. 1, when the Fresnel zone plate is 1115, at 9 o'clock, when one gap is exactly half the focal length, the gap at that point can be detected, but the inclination cannot be detected. Therefore, several Fresnel zone plates 1-m of different sizes, for example eight as shown in FIG. 2, are formed on the same mask 1llfi as one set of zone arrays 13 and used. FIG. 2(4) is a plan view of the Fresnel zone array 13, and the zone array 13 is formed on the lower surface of the mask 11 in FIG. 2(6). The largest annular plate 13-8L is not shown accurately, but the one indicated by the other circles is a single annular plate. Now, as 4 using 3rd order focus.

Wafer 12ik when its focal point is observed as a light spot
The position of i is indicated by a broken line. That is, when the lower surface of the mask 11 is tilted with respect to the plane shown by the broken line, the tertiary focal point of the Fresnel ring array 13 is simultaneously observed as a light spot.

now.

If the mask 11 and the wafer 12 are parallel as shown in FIG. 8(8), the light spot 16 will appear bright at and near the intersection of this broken line and the surface of the wafer 12.

For example, in the case of Fig. 2 (1 m), the 64th ring plate 13-6
The center of the wax also looks bright. Next, since the center of the 5th #r ring plate is bright 9 degrees and the focal length of the ring plate, which makes this moth bright and vibrant, is known in advance, the gap can be determined from that. Therefore, by finding out that there is a plurality of brightest orbicular plates at the center in the orbicular series 13, a wall can be detected.

The second embodiment will be explained with reference to FIG. 3.4. In FIG. 3, the Fresnel strip has been modified from a circular shape to a straight one, and is made of a straight strip like the strip 2L in FIG. 3. If the distance from the center of symmetry to each line is -fr, then that distance is a combination of bands that can be determined by equation (1), and has the same distance t as a cylindrical lens. IC%L
Since the pattern IiI is formed by a group of straight lines extending in two directions perpendicular to each other, it is often easier to form the Fresnel strip 23-8 if it is also straight. This straight Fresnel strip 23-84 will be referred to as a one-dimensional Fresnel strip. Among the one-dimensional Fresnel band arrays 23, the largest one 23-81. Although not shown accurately, the other small pieces are also strips, and the only difference is the focal length. The orientation of this one-dimensional Fresnel strip does not necessarily have to be parallel to each center line as shown in FIG. 3, but may be as in the strip 23'-8 in the isA diagram. That is, as long as the strips in one strip are oriented in the same direction, the other strips in the strip row may be oriented in either direction.

Next, the Is3 embodiment will be explained based on FIG.

The third embodiment is based on the first embodiment. Unlike the second embodiment, the strip row consisting of a plurality of Fresnel strips having a discrete focal length Ill &t' is not used, but one as shown in FIG. 5(4).
A modified nine Fresnel strip plate 33t is used so that the focal length changes continuously depending on the location of the four strip rows.

In other words, this is considered to be the result of arranging one-dimensional Fresnel strip rows 23' as shown in FIG. 4 in a row with a size l1li, and connecting the adjacent strips. At this time, if we increase the number of strips between the ant-sized strip and the smallest strip to infinity, the distance between the focal length of the largest strip and the smallest strip will reach the limit. A one-dimensional Fresnel strip plate with a continuous focus having a focal length of 11 is created. For example, when measuring using a third-order focus.

If we focus only on the tertiary focal point, the position of the wafer 32 surface will be as shown by a broken line as shown in Figure 5 (B) when the light spot is on the pattern surface of the mask 31, and if the mask 31 and the wafer 32 are parallel, this broken line At and near the point where the light intersects with the surface of the wafer 32, a light spot 3@ is formed on one side at a position corresponding to FIG. 185 (4). This light spot 36 looks like a bunch of intersections intersecting in an X shape. Therefore, by observing the state of connection of this light spot 36, the inclination of the mask 31 and the wafer 32 can be measured.

#11~#! If the Fresnel ring array shown in Example 3 or a one-dimensional Fresnel belt plate modified from the same is formed at at least three locations on the pattern surface of the mask, the [6111] distance between the mask and the wafer can be established. When using this method, it is desirable that the wafer surface onto which the Fresnel ring array or one-dimensional Fresnel zone array is projected is a smooth surface on which no printing pattern is formed. Normally, in the production of IC%LET, several types of masks are used for the pattern printing, one of which is
There is a pattern of Fresnel band plates or one-dimensional Fresnel band plates, ie, a pattern of Fresnel band plates, which has been previously printed before printing the sheet of mask. Therefore, if the position of the Fresnel band array in each mask is made different for each mask, it is convenient to perform printing on each mask, since it is possible to detect a light spot using a smooth wafer surface.

In addition, in the above embodiment, the focal length of the Fresnel strip is 1lIi
1t-distance, and the light spot reflected once by the wafer was converged on the mask to detect it, but the focal length and 'h
It is also possible to detect the five gaps equally, that is, by collapsing a light spot on the wafer.

In the above embodiment, IIJIs detection of a mask and a wafer was explained as an example. However, when detecting the inclination of two adjacent objects with 2,000 parallel planes, if the inner plane is smooth and at least one The present invention can be used to detect the inclination of an object made of any material as long as the object is transparent.

(9) Since this embodiment has been described with reference to a chrome mask, a Fresnel zone plate or a one-dimensional Fresnel zone plate, that is, a Fresnel zone plate, is formed of a transparent part and a completely opaque part, but the opaque part is not necessarily It is not necessary to completely block the light t, and it may be formed of a material such as a see-through mask. Or to make it opaque.

``It is known that if the phase of light is made to be different from that of the transparent part, the intensity of the focal point will be four times greater than if the part is simply made non-transparent, so even if a thin film to be used as a T-plate is formed, good.

Similarly, in the above embodiment, the light that was once reflected by the wafer and imaged on the lower surface of the nine-light point t-mask was detected from the upper dIJ side of the mask using an optical telescopic mirror.

Generally, a printing pattern is formed on the bottom surface of the mask, and in normal 7-line operations, it is convenient to detect the slope of the darkness between the bottom surface of the mask and the top surface of the wafer.

Therefore, in the above embodiment, the light spot is imaged on the lower surface of the mask, but in general inclination detection of two flat plates, it is not necessary to image the light spot km on the opposing i# plane of the flat plate! ! There isn't.

When detecting the inclination of two flat plates, both of which are made of transparent materials,
Nine light particles formed on the surface of a flat plate corresponding to the wafer may be observed using a light beam transmitted through the flat plate.

As described above, according to the present invention, except for forming a Fresnel strip on the mask, it is only necessary to use a nine optical system using an objective lens Xt with a large numerical aperture as a detection optical system.
It is very easy to install. Furthermore, a plurality of Fresnel strips having a focal length of 1 times or r12 times the gap before and after the focal length of 1et-4 are formed on the pattern surface of the mask, and the illumination incident on the Fresnel strips and the focal point of the microscope I! Using the shallow 1WIL, monochromatic light is used to illuminate the pattern surface of the mask or the wafer tr.
The tilt can be detected by observing rt.

[Brief explanation of the drawing]

Figure 1 shows the plane and section rMt of the Fresnel strip used in the present invention.
2 is a plan view and a cross-sectional view of a first embodiment of the present invention, FIG. 3 is a plan view of a second embodiment of the present invention, and FIG. 4 is a plan view of a first embodiment of the present invention. 1112 shown in Figure 4113
FIG. 5 is a plan view of a modification of the embodiment of the present invention, and FIG. 5 is a plan view and a cross-sectional view of the third embodiment of the present invention. [Explanation of symbols of main parts] Fresnel strip...-, 13, 23, 23', 3
3 Figure 2 CB) Ward 4... 23cm Figure 5 (B):

Claims (1)

[Claims] Light 4 on a leather Fresnel board set on a flat plate! irradiate with iI,
A device for detecting the inclination of a flat plate, characterized in that the inclination tp of the cough flat plate with respect to a predetermined plane is determined from the convergence shape of the light converged on the unique focal point of the Fresnel band itself.