JPS6298621A - Fine gap setting up device - Google Patents

Abstract

PURPOSE:To measure the gap on-line between a mask and a wafer subject to non-destruction and non-contact by a method wherein the gap between a pattern and a mirror is set up by measuring any positional deflection caused when a pattern and a vertical image of pattern on the mirror are observed in the oblique direction. CONSTITUTION:A wafer 2 below a mask 1 is vacuum-adsorbed on a wafer chuck 3 for setting up by vacuum 5. The wafer chuck 3 controls the wafer 2 set thereon by vertical inching mechanisms 4-1, 4-2 to make specified gap between the wafer 2 and the mask 1. The gap shall be set up with high precision since the gap between the mask 1 and the wafer 2 is radially irradiated with X-rays from a spot type exposure light source so that any transferred pattern width may change in response to the change in gap between the mask 1 and the wafer 2. Therefore, an illumination optical system 6, a detection optical system 7 and a pattern detector 8 are provided to measure the gap. In such a constitution, the gap between the mask 1 and the wafer 2 detected by the pattern detector 8 can be fedback to the vertical inching mechanisms 4-1, 4-2 to set up the gap with high precision.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to micron-order minute gap measurement technology, and in particular to measurement of the gap between a mask and a wafer during non-contact proximity exposure, such as an X-ray aligner in semiconductor manufacturing. suitable for

[Background of the invention]

Gap setting in conventional proximity aligners is performed by measuring the heights of the wafer and mask separately off-line and moving them to the exposure position to set the gap, as described in Japanese Patent Application Laid-open No. 13726/1983. Because of this, even if there is a shift during transportation, it cannot be detected, and it can only be measured off-line, so it is almost impossible to achieve submicron gap setting accuracy at the actual exposure position. [Object of the Invention] The object of the present invention is to measure the gap between the mask and the wafer in an X-ray mask aligner online in a non-destructive and non-contact manner. ,
The accuracy lies in obtaining submicron accuracy. [Summary of the Invention] The pattern on the mask is reflected on the mirrored wafer and observed as a virtual image. When viewed from an oblique direction, this corresponds to the size of the gap. As a result, the real image and virtual image are misaligned, and they appear as a double image. Therefore, the deviation of this double image is measured.
Conversely, you can calculate how much the gap is. The problem here is that
Since the real image and the virtual image are not on the same focal plane, a defocus occurs. Therefore, in the present invention, a plurality of separate patterns whose relative positional accuracy on a plane has been measured in advance are used. relative positional relationship. One real image and the other virtual image can be made to exist on the same focal plane. Therefore, if we measure the distance on the focal plane between the real image target pattern and the virtual image target pattern provided at this focal point position, we can can be converted into a gap.

If the absolute value of the gap is not required, by controlling the two to become a focused point, the gap can be set using the zero method, and the gap setting reproducibility accuracy can be greatly improved.

[Embodiments of the invention]

The present invention will be explained in detail below.

Figure 1 shows the main parts of the X-ray aligner, including mask and wafer gap detection.The wafer 2 under the mask 1 is set on the wafer chuck 3 by vacuum 5.The wafer chuck has a vertical fine movement mechanism. 4-1.4-
2, the wafer 2 set on it is massed.

As shown in the figure, the X-rays from the original source are radially irradiated from a point light source, so that the mask 1 and the wafer are controlled at constant intervals.

If the -2 gap changes, the width of the pattern to be transferred changes as the X-rays become wider. Therefore, it is necessary to set this gap with high precision. Therefore, in order to measure this gap, an illumination beam system 6 and a detection beam system 7. A pattern detector 8 is provided.

Mask 1 detected by pattern detector 8. The gap between the wafers 2 is fed back to the vertical fine movement mechanism 4-1 and 4-2, and the structure is such that this gap is set.
The principle of the 1tlT method will be explained.

Target pattern 9-1.9- formed on mask 1
2 blocks light, and the other parts are transparent and allow light to pass through. Normally,
This pattern is made of a material that has the same effect on X-rays.
A wafer 2 is placed under a mask 1 made of, for example, gold (Au), separated by a minute gap d. In this case, the patterns are arranged so that the boundary surfaces of the wafer 2 face each other. Mask 1 set like this
--2, the wafer 2 has a mirror surface 1, and the mask pattern 9-1.
, 9-2' Samuraitsu. Therefore, the pattern detector 8 is set in the direction of the angle θ with respect to 1/"v-2. Here, if the distance between the patterns 9-1 and 9-2 on the mask 1 is 1, then 1 If one design is carried out, the pattern 9- on the mask will be
1 and the imaginary 1?9-2' can both be placed on this focal plane 10.

That is, the distance between the imaging plane 11 of the pattern detection optical system 7 and the focal plane 10? : When J-0 and 1-, a bar appears on the imaging plane 11.

When turn detector 8 is installed, pattern 9-1゜9-2
Patterns 9-1' and 9-2 become the rear focus and rear bit, respectively, and the contrast is lowered by 1.

Now, let's consider the target gap d between the mask 1 and the wafer 2. d is between the pattern 9-1 and its virtual image 9-1'
It becomes 1/2 of the distance. If you observe this from an oblique direction by θ, it will look like d', 12.

Real image 9-1 and imaginary residue 9 when viewed from the true gap d and the oblique direction by -4
-1′ inter-pattern distance dlO is 2danθz
Since the relationship d1 holds true, if we find d+, we will get the required d
is found.

However, as mentioned above, the pattern 9-1 and the virtual image 9-1' cannot exist in the same focal plane, so the pattern 9-1 and the virtual image 9-1' are detected at a high WI [i] because they cannot exist in the same focal plane. Although it is difficult to measure, since the trap pattern 9-1 and the virtual image 9-2' are on the same focal plane as mentioned earlier, they are both in the focused position with good contrast to the detector 8. Even if the detector 8 is located away from the in-focus point, the distance between the two cameras will be explained later in order to be evenly affected by the defocus. Distance change when finding from image signal, f
That is, there is an advantage that the gap detection n degree change is small. Find the distances @d2 and 13 between the patterns 9-1 and 9-2' on the focal plane 10, where n is CtL, and dzcoa.
After d is determined from the relational expression θ = 2d, a method for determining the distance d2 between the patterns 9-1 and 9-2' will be explained.

FIG. 3 shows a target pattern 79- provided on the mask 1.
An example of 1.9-2 is shown.

For example, the length of the pattern is 7100 μm, the pattern line width is 5 μm, and the distance of 9-1.9-2 is 1. If this is the angle θ = 3 cf when illumination is detected using the optical system shown in Fig. 1, then if the set gap d = 20 μm, then the pattern spacing of the marks 9-1 and 9-2 for gap setting. If ヱ is designed to this value of 25 μm, pattern 9-
1.9-2' can both be detected as a focused pattern. At this time, therefore, the patterns are perceived as being approximately 47 μm apart on the detector.

As a detection optical system for pattern distance 1, for example, the fourth
Use what is shown in the figure.

When the objective lens 14 magnifies the image by, for example, 20 times and forms it on the imager 8, the distance between the patterns is 47 x 2.
It becomes 0-960 μm. As the imager 8, one pixel is 13μ
When a linear sensor is used, a cylindrical lens 15 for image compression is provided in front of the linear sensor 8 corresponding to 738 pixels to improve the S/N ratio.

The light intensity signal detected by the linear sensor is the signal 16-1
．． The signal will be as shown by signal 16-2, so measuring this electrically and reading the position between the signals is as follows:
By feeding back the values detected by the method described above, which can be realized with sufficient accuracy using the conventional technology, to the fine movement up and down mechanism shown in Fig. 1, 4-1 and 4-2,
Pattern transfer can be performed with high precision because the designed gap can be precisely set.

〔Effect of the invention〕

When the present invention is applied to an X-ray aligner, it is possible to set the gap between the mask and the substrate 1 without contact at the exposure position, and by feeding this back online, the gap can be set with high precision. can.

Since it can be performed at a different exposure position, the conventional technology, JP-A-57-
The method disclosed in No. 7931 eliminates the need for stage evacuation for necessary measurements. Let's make a huge improvement in throughput.

[Brief explanation of drawings]

Fig. 1 is an overall explanatory diagram of an embodiment of the present invention when used for setting the gap of Explanatory diagram of the pattern, 4th is an explanatory diagram of the optical system for detecting the distance of the pattern. 1... X-ray mask 2... Sea urchin/X
-3... Fine movement stage 5... Vacuum 6... Illumination optical system 7... Detection optical system 8... Imager
9...Mask pattern 10...Focal plane
11... Image forming surface 14... Objective lens 15... Cylindrical lens 16... Detection signal

Claims (1)

[Claims] 1. A minute gap setting device that sets the gap between the pattern and the mirror surface by measuring the amount of positional deviation that occurs when a virtual image of the pattern reflected on the mirror surface is observed from an oblique direction.