JP3414975B2 - Position shift amount measuring device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a positional deviation amount measuring device, and more particularly, to a positional deviation amount measuring device for measuring a positional deviation amount between various patterns formed on a wafer, that is, registration. The present invention relates to a registration measuring device capable of highly reliable measurement without being affected by irregularities (grains) on the surface of aluminum.

[0002]

2. Description of the Related Art In the manufacture of semiconductor ICs, various patterns are formed on a substrate wafer having a smooth surface. The positions of these patterns must be accurately formed, and the amount of misalignment between the already formed pattern and the pattern to be formed next, that is, registration, can be accurately measured. Has been done. For example, in a certain semiconductor manufacturing process, the amount of misalignment between a resist pattern formed by exposure through a mask or the like and an etched pattern that is already formed in the lower layer in the immediately preceding process. Is measured with high accuracy by a registration measuring device (positional shift amount measuring device). Then, when the deviation amount here exceeds the preset specified value, the formed resist pattern is cleaned and re-formed.

In recent years, 16M to 64M, 256M, D
With the dramatic increase in the storage capacity of RAM, the measurement and inspection of this displacement amount is becoming more and more important. In addition, the control value of the amount of positional deviation between the resist pattern formed by exposure and the etched pattern already formed in the preceding step is also small, and high-precision measurement is required.

FIG. 9A shows an example of a cross section of a wafer in an X direction or a Y direction in a certain process regarding one of the deviation amount measurement mark patterns provided for measuring the amount of positional deviation. As the alignment mark pattern, for example, first, a pattern 32 of silicon oxide is first formed on the surface of the wafer 31 with an appropriate gap L in a rectangular frame in the previous step, and the pattern 32 is covered to cover the wafer 1 A thin film 33 of aluminum is vapor-deposited on the entire surface of. Further, a rectangular pattern 34 of photoresist having a width d is formed near the center 31 of the gap L by exposure in the current process. The optical system of the measuring device irradiates these surfaces with epi-illumination light and reflects the reflected light into a 1-line CCD.
When detected by the sensor, the signal waveform shown in (b) is obtained. Since the aluminum thin film 33 is opaque, the pattern 32 is not detected, but the edge of the thin film 33 prevents the pattern 32 from being detected.
There therefore is curved, are detected peak p _K is for bend K, measured the distance L 'between the two peaks, the position of the center point m ₁ is obtained. Further, the peaks p _r for both edges of the pattern 34 are detected, the distance d ′ between the both peaks is measured, and the position of the center point m ₂ thereof is obtained. The distance δα between the center points m ₁ and m _{2 is} the amount of positional deviation of the pattern 34 with respect to the pattern 32. Although not shown in the drawing, the above-mentioned wafer 1 is subjected to reactive ion etching to leave a portion of the thin film 33 corresponding to the pattern 34 to form an aluminum wiring pattern and its position. The shift amount is measured by the same method as above.

The permissible values of the displacement amount δα become smaller and smaller as the integration degree of IC increases, and for example, 6
It is set to 80 to 100 nm at 4 mega, and the measurement accuracy for this is required to be 10 nm or less. It is 2
In the case of shifting to 56M, this allowable value becomes even tighter.

[0006]

Problems to be Solved by the Invention FIGS. 9 (a) and 9 (b)
Is a measurement waveform in an ideal measurement state where the deviation amount measurement mark pattern provided inside the frame has a large step. Actually, as the storage capacity of the DRAM increases, the step difference of the pattern inside the frame becomes lower as shown in FIG. 3 described later. Therefore, the edges of the pattern are rounded, various kinds of noise are added to the signal waveform, and the sample surface itself including the wafer has irregularities, which is affected. The irregularities are caused especially by irregular irregularities (hereinafter referred to as grains) appearing on a metal surface such as the aluminum thin film 33, and the irregularities appear as reflected light superimposed on the detection signal. for that reason,
The flat signal portion (reference level) of the detection signal waveform shown in FIG. 9B is undulated and noise is generated. As a result, a peak shift occurs or a pseudo peak occurs, and the position cannot be measured accurately. Moreover, the influence of the shift of the focus position of the objective lens significantly appears, and there is a problem that a stable measurement value cannot be obtained unless the focus is accurately performed.

As one measure to avoid such a problem, a measure for increasing the contrast is adopted. As a simple method for improving contrast, it is possible to reduce the ratio σ between the pupil diameter of the objective lens and the diameter of the light source image at the pupil position, that is, the value σ of (diameter of the light source image at the pupil position / pupil diameter). Done. The light source usually reaches the objective lens through a circular pinhole, and is illuminated in a telecentric manner onto the wafer by epi-illumination (Kohler illumination). Therefore, the detection optical system is arranged in the direction perpendicular to the wafer. In this case, empirically, when σ is σ = 0.5 to 0.6, the most stable detected image can be obtained. However, if a measure is taken to increase the contrast by further reducing σ, the influence of the grain becomes large at the same time, which causes a problem that the original peak becomes difficult to detect. Of course, if .sigma. Is large, the effect of grain can be reduced, but the peak level is correspondingly decreased and the effect of noise is increased. An object of the present invention is to solve such a problem of the conventional technique, and to provide a position deviation amount measuring device capable of measuring the position deviation amount with high accuracy while suppressing the influence of grains. is there.

[0008]

The feature of the position shift amount measuring apparatus of the present invention for achieving such an object is that the irradiation light from a light source is introduced into an objective lens through an aperture stop having a cross slit. Has an epi-illumination optical system that performs epi-illumination on the sample, and at the pupil position of the objective lens, one slit of the cross forms an image in the direction along the pattern,
The other slit of the cross is a cross slit which forms an image in a direction perpendicular to the pattern, and the width of the slit and σ = 0 set to σ = 0.3 or less in the ratio σ of the image to the pupil diameter of the objective lens. The detection signal is obtained by a light receiving element selected to have a slit length set to 6 or more and arranged in a direction perpendicular to a plurality of patterns.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION As described above, with respect to the ratio σ, the width of the slit in which the cross slit is set to σ = 0.3 or less in the ratio σ of the image to the pupil diameter of the objective lens and σ = 0.
Since the slit length is selected to be 6 or more, the direction perpendicular to the pattern, in other words, in the relationship between the pattern and the light receiving element arranged at a right angle, in other words,
In the direction over a plurality of formation patterns, σ = 0.6, so that a large horizontally long slit is formed along the slit when viewed from the light receiving element,
In the direction along the pattern, in other words, in the detection direction of the pattern corresponding to the direction of each pattern, it becomes a vertically long slit which is at a right angle to the light receiving element where σ = 0.3 or less. Thereby, the contrast of the detection signal from the pattern (edge) with respect to the pattern detection direction is improved, and at the same time, σ is larger than σ = 0.6 along the detection direction, so that the contrast of the unevenness of the grain is lowered. be able to. As a result, it is possible to suppress the noise level of the detection signal due to the grain and obtain a detection signal having a good peak value.

First, the principle of the present invention will be described with reference to FIGS. Shown in (a) is the relationship between the pupil diameter of the objective lens and the slit image of the cross slit at the pupil position of the objective lens. 53a is a slit image at the pupil position of the cross slit, and 1a is the pupil diameter (field of view) in the objective lens. However, the cross slit image 53a has a slit width (σ = 0.
006) and a slit length (σ = 0.9) set to σ = 0.6 or more. The CCD is an example of a light receiving element arranged in a direction perpendicular to a plurality of patterns, and Ds is a detection signal of this CCD. (B) is a CC in the slit image 53b in the case where a vertically long slit is provided only in the direction along the pattern.
Detection signal Ds by D and registration mark M
And (c) is the relationship between the detection signal Ds in the slit image 53c and the registration mark M in the case where a horizontally long slit is provided only in the direction perpendicular to the pattern. As described above in the case of the circular pinhole, in the slit having a small σ in the direction along the pattern shown in (b), the light receiving element (C
The contrast of the detection signal Ds detected by (CD) is improved. However, at the same time, a large number of peaks are also generated due to the influence of grains, which becomes noise.

On the other hand, in the slit having a large σ in the direction perpendicular to the pattern shown in (c), the contrast of the detection signal Ds detected by the light receiving element (CCD) obtained in the field of view is lowered and the detection signal is decreased. Is also averaged from the grains, resulting in a distorted waveform. Then, when the cross slit shown in (a) is used, the detection signal Ds becomes the sum of (b) and (c), the characteristic of (b) appears preferentially, and the signal from the grain has a horizontal slit side. Appears and is integrated and averaged to reduce the contrast. The reason can be explained as follows.

FIG. 7 is an explanatory diagram thereof. Reference numeral 1 is an objective lens, 9 is a wafer, and a grain portion G is provided on the surface. In the telecentric relationship, the light should originally be parallel to the sample, but in the case of a horizontally long slit as shown in the figure, the illumination light is irradiated to the sample from multiple directions, and the parallelism of the irradiation light is Reduced to low coherent light. Moreover, the grain is the registration mark M.
Unlike FIG. 6 (c), since there are many irregularities, signals from more grains are received than in the case of a vertically long slit. This is because the lateral direction is perpendicular to the pattern and is not along the pattern, and thus is more susceptible to the grain. In this respect, the vertical direction is
Since the direction is along the pattern, the effect of the vertical grain is reduced as compared with the effect of the horizontal grain due to the existence of the pattern. It is considered that an integrated signal is obtained due to the influence of grain. On the other hand, since the vertically long slit is along the pattern, the pattern is irradiated with high coherent light. Therefore, the detection signal for the pattern edge maintains high contrast.

Since the signal from the grain is suppressed and the pattern edge is detected by such high and low coherent light, it is important that the center of gravity of the cross slit is located at the center of the pupil, which is preferable. The reason is that, when the center of gravity of the slit does not coincide with the center of the pupil, the average irradiation angle of the illumination light is not perpendicular to the wafer surface. As a result, the symmetry of the waveform is broken and it becomes impossible to obtain an accurate measurement value.

FIG. 8 is an explanatory diagram of the CCD detection signal in the relation between the width and the length of the cross slit,
(A) shows the relationship between the width and length of the cross slit,
(B) to (d) are C when the width and the length are changed.
The state of the detection signal of CD is shown. As shown in FIG. 8A, the vertical slit width perpendicular to the CCD is σ = 0.3, and the vertical slit length is σ =
Fixing to 0.6, the lateral slit width parallel to the CCD is σ = A, and the slit length is σ = B.
When the states of the CCD detection signals when these A and B are changed are confirmed by experiments, the results shown in (b) to (d) are obtained. Here, the vertical width is fixed while the horizontal slit width is σ = A, and the slit length is σ = B. The reason for this is that the relationship between the horizontal width and length improves the contrast of the detection signal. This is because it is related to the (width in the horizontal direction) and the decrease in contrast of the grain (length in the horizontal direction).

As shown in FIG. 8B, the length B is σ =
When fixed at 0.6, σ = 0.4 for width A
The noise level increases and the contrast of the detection signal deteriorates. On the other hand, when the width A is σ = 0.3 or more, a good detection signal can be obtained. Further, when the width A is fixed to σ = 0.3 as shown in FIG. 8C, the noise level increases at σ = 0.5 for the length B and the contrast of the detection signal is poor. Become. On the other hand, the length B is σ = 0.
When it is 6 or more, a good detection signal can be obtained.

FIG. 8 (d) shows the state of the detection signal and noise when the width A and the length B are changed at the same time. As you can see, for width A and length B, width A
For σ = 0.4 and for length B σ = 0.5,
It is difficult to obtain a good detection signal because the noise becomes large. As described above, it can be seen that a good detection signal can be obtained when the width A is σ = 0.3 or less and the length B is 0.6 or more. In this way, the condition of the cross slit is the ratio σ, and the ratio σ with the image to the pupil diameter of the objective lens
= 0.3 or less and σ = 0.6 or more, the cross slit can detect about 3 nm. In this respect, the conventional round pinhole has two steps.
The limit is about 0 nm.

Incidentally, the condition includes zero when the width is as follows, but since this is a slit,
The width is not zero, and the minimum value should be selected according to the design in relation to the selected pupil diameter.
It is not specified. Similarly, in the case of the length above, infinity is included, but since this is a cross slit, there is an upper limit naturally from the design,
This also means that the maximum value may be selected by design in relation to the selected pupil diameter.

[0018]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is an explanatory view of a registration measuring apparatus of an embodiment to which the position shift amount measuring apparatus of the present invention is applied, and FIG. 2 is an explanatory view of the shape of its slit, and FIG.
4A and 4B are explanatory views of the deviation amount measurement mark pattern and the detection waveform, FIG. 4 is an explanatory view of a chip position on the wafer on which the deviation amount measurement mark pattern is formed, and FIG. 5 is a cross slit and a circular pinhole in the present invention. It is explanatory drawing of the variation amount with. Reference numeral 100 denotes a registration measuring device, 1 denotes epi-illumination, and reflects light reflected from the wafer 9 by C
It is an objective lens to be sent to the CD linear sensor 10 and the CCD linear sensor 11. A half mirror 2, a relay lens 3, a half mirror 4, a cylindrical lens 5 (X-axis direction), and a cylindrical lens 6 (Y-axis direction) are provided in alignment with the center of the objective lens 1. The CCD linear sensor 10 and the CCD linear sensor 11 are detectors for receiving the reflected light in the X direction and the Y direction, respectively, which are provided for detecting the positional deviation amount in the registration measurement, and are controlled by the control circuit 20. It is driven via the CCD drive circuit 14 of the CCD linear sensors 10 and 11.

Here, the half mirror 2 is an illumination optical system 5.
The irradiation light from 0 is received and directed toward the objective lens 1 to perform epi-illumination on the wafer 9. Half mirror 4 is X
The reflected light from the wafer 9 is separated into the detection system of the direction and the Y direction. The illumination optical system 50 controls the light emitted from the light source 51 by a condenser lens system 52 provided in the middle of the optical path, an aperture stop 53,
The half mirror 2 is irradiated through the relay lens 54 and supplied to the objective lens 1. 7 is a wafer chuck, 8
Is an XYZ for moving the wafer chuck 7 in the XYZ directions.
The moving stage 9 is a wafer chucked by the wafer chuck 7.

On the wafer 9, the deviation amount measurement mark pattern shown in FIG. 3 has four corners A of each chip A = NO.1 to NO.4 ,.
E = chip A formed in NO.17 to NO.20 (see FIG. 4),
It has B, C, D and E chips. As shown in FIG. 3A, the deviation amount measurement mark pattern is
It consists of a double rectangular frame. The inner frame is the reference layer mark pattern (reference layer pattern 24) already formed in the preceding process (silicon substrate layer 27), and the outer frame is the alignment layer mark pattern (generated in the current process). It is a laminated layer pattern 25). (B) is a sectional view in the X direction, and (c) is a detection signal waveform thereof.

In the figure, 23 is an interlayer insulating film, and 26 is an aluminum wiring layer. 28 is a portion of the aluminum wiring layer which is raised due to the bending. Of course, the alignment layer pattern 25 is formed of a resist. The waveform shown in (c) shows the CCD linear sensor 10 (or CCD linear sensor 1 in the Y direction).
1), which is input to the control device 20 via the A / D conversion circuit (A / D) 15. In the signal waveform, 25a is a peak signal generated by the matching layer pattern 25, 24a is a peak signal generated by the reference layer pattern 24, and 28a is a noise component generated by unevenness of the aluminum wiring layer. Note that FIG.
The level of the detection signal here is lower than that of the signal of (b) because DR of higher density such as 64M and 256M.
This is because the mark has a low step difference corresponding to AM.

Now, referring to FIG. 1, the A / D 15 is controlled by the control unit 20 to control the CCD linear sensors 10, 1.
The detection signal of 1 is digitized and sent to the control device 20. The control device 20 includes an image memory 16, a digital signal processor (DSP) 17, a focus controller 18, an MPU 19, a memory 21 and the like, and the MPU 19 and the image memory 16 and the DSP 1 via a bus 22.
7, the focus controller 18, the memory 21 and the like are connected to each other. The A / D 15 receives the detection signals from the CCD linear sensors 10 and 11 and sends out the A / D converted data to the image memory 16 at a predetermined sampling cycle. The image memory 16 sequentially stores the data from the A / D 15.

The DSP 17 is controlled by the MPU 19 to receive the digital data of the image memory 16 and calculate a deviation amount ΔX (ΔY) from the digital data at high speed.
Send to. This is a processor dedicated to calculating the amount of deviation. Note that ΔX corresponds to the interval δα in the X direction described in FIG. 9, and ΔY corresponds to the interval δα in the Y direction described in FIG. 9. As the processing of the deviation amount ΔX (ΔY) of the DSP 17, the pixel position of each of the detected peaks is detected, and as described earlier with reference to FIG. 5, the CCD linear sensor in the X direction is detected from the peak pixel position. The signal from 10 is received from the A / D 15 and the distance L'between both peaks of the reference layer pattern 24 is measured to obtain the position of the center point m ₁ . Further, the peaks p _r for both edges of the matching layer pattern 25 are detected, the distance d ′ between both peaks is measured, the position of the center point m ₂ thereof is determined, and both center points m ₁ and m ₂ are obtained.
The amount of deviation ΔX in the X direction of the position of the alignment layer mark pattern 25 is calculated as the interval δα. Next, the DSP 17
A / D1 signals from the CCD linear sensor 11 in the Y direction
5, the positional deviation amount ΔY of the similar alignment layer mark pattern 25 is calculated. The shift amount ΔX (ΔY) calculated in this way is sent to the MPU 19 via the bus 22.

The focus controller 18 is the MPU 1
CCD linear sensor 10, 11, A / D controlled by 9
15. The image memory 16 is controlled to receive data from the image memory 16 and move the XYZ moving stage 8 in the Z direction via the MPU 19 to perform focusing. Memory 21
A registration measurement program 21a, a positional deviation amount determination program 21b, and the like are provided in. 23
Is a drive signal for moving the XYZ moving stage 8 in the X, Y, Z directions in accordance with a control signal from the control device 20.
It is a stage drive circuit for sending to the YZ moving stage 8.

Now, the registration measurement program 21a is executed by the MPU 19 to execute C
The detection data (voltage value of the detection signal) regarding the reflected light from the A / D converted pattern, which is stored in the image memory 16 by performing the processing of fetching the detection signal from the CD sensor, is read and fetched. On the other hand, the focus controller 18 is controlled by differentiating the peak corresponding to the signal of the edge portion of the pattern formed on the wafer 9 (here, the reflected light reception signal from one edge). Then, the distance of the objective lens 1 is set to the optimum focus state. Next, the detection signals of the reference layer pattern 24 and the alignment layer mark pattern 25 are taken into the image memory 16, and the data of the reflected light is taken in by driving the DSP 17, and the deviation amount ΔX (ΔY) is received and a predetermined area of the memory 21 is received. The displacement amount measurement mark patterns at the four corners of each chip are stored in correspondence with A = NO.1 to NO.4, ... E = NO.17 to NO.20 (see FIG. 4). Then, the position shift amount determination program 21b is called. The misregistration amount determination program 21b compares the misregistration amount ΔX (ΔY) with a predetermined reference value to determine the misregistration amount for each matching layer mark pattern 25 in each of the misalignment amount measurement mark patterns NO.1 to NO.20. A process of determining whether or not it is within the range is performed, and a process of displaying the shift amount and the determination result on the display is performed.

FIG. 2A is an explanatory view of the shape of the cross slit of the aperture stop 53. Here, as described above with reference to FIGS. 6 and 7, the slit width is σ = 0.3 or less with respect to the ratio σ between the pupil diameter 1a of the objective lens and the image width of the aperture stop 53 at the pupil position. Is set and the slit length is σ
= 0.6 or more. For example, when the peak values of the two detection signals at both ends are obtained along the X direction (X axis deviation amount measurement), σ in the X direction is σ = 0.6 or more, and σ in the Y direction is σ = 0. It becomes less than or equal to 3. This makes X
It is possible to improve the contrast of the detection signal of the axis and suppress the fluctuation of the reference level due to the grain. Further, when the peak values of the two detection signals at both ends are obtained along the Y direction (measurement of the deviation amount of the Y axis), σ in the Y direction is σ = 0.6.
As described above, σ in the X direction is σ = 0.3 or less. As a result, the contrast of the Y-axis detection signal can be improved and the fluctuation of the reference level due to the grain can be suppressed.
As a result, a peak signal that is less likely to be affected by grain irregularities is obtained as a detection signal in each of the X and Y directions.

The cross slit shown in FIG. 2 (a) realizes this and has a slit width of 86 μm and a slit length in the X and Y directions of about 1300 μm.
m. Then, the pupil diameter 1a of the objective lens 1 (see FIGS.
(See) is 1.4 mm. The calculated σ can be calculated from the width, the length, and the pupil diameter, and the σ value in the direction along the pattern with respect to the peak value signals in the X direction and the Y direction is σ = 0.06. The σ value along the direction perpendicular to the pattern in which two peak values (X direction corresponds to the peak signals 25a, 25a in FIG. 3C) is σ = 0.93.
Is. Therefore, in this example, in the direction in which the detection signal is sampled, the ratio σ in the direction parallel to the ratio σ in the direction orthogonal thereto is about 15 times.

Various experiments were conducted in order to explain the effect under the above conditions. As an example, the effect of the cross slit shown in FIG. 2A will be described in the table of FIG. FIG. 5 shows a cross slit having a slit width of 86 μm and a length in the X and Y directions of 1300 μm when the pupil diameter of the objective lens is 1.4 mm, and the conventional ratio σ =
The statistical variation 3σ is calculated for a pinhole of 0.5 and the variation in measurement is compared. In the case of a pinhole, it is calculated from the measured values when the focus position on the wafer is optimized. In this respect,
In the case of the cross slit, it has been confirmed that the depth of focus becomes deep, and even if the focus position is slightly shifted, the variation in the measurement result is not significantly affected.

As shown in this table, 3σ at the same point has less variation as compared with the case of a pinhole,
Stable measured values are obtained. In addition, 3 of the same point method
σ is a value obtained by repeatedly measuring the points A, B, C, D, and E shown in FIG. 4 n times for each point, and expressing the variation of the deviation amount by 3σ, which is 3σ in the dynamic method.
Makes two measurements for every 4 points on each chip,
For NO.1 to NO.20, the first measurement value-the second measurement value is calculated, and 3 [sigma] is calculated for all points.

The above limit value ratio σ = 0.3 or less and the ratio σ
= 0.6 or more is due to the combination of the slit width and the length, and there is a correlation between them. Therefore,
As a result better than the pinhole, the results obtained by the experiment are ratio σ = 0.3 or less and ratio σ = 0.6 or more. That is, the condition that the σ value along the direction of obtaining the two peak value signals is the ratio σ = 0.3 or less and the σ value in the direction orthogonal to the direction of obtaining the two peak values is the ratio σ = 0.6 or more. It is what However, according to the experimental results, as a preferred embodiment, it is preferable that the σ value in the direction perpendicular to the direction in which the two peak signals are obtained has a ratio σ = 0.1 or less.

In FIG. 2B, instead of the cross slit, point light sources having a ratio σ = 0.3 or less are arranged at right angles to the directions in which two peak values occur, and the σ value in this direction is substantially 0. It is an example of the aperture stop 52 set to 0.6 or more. In any case, it is important that the position of the center of gravity of the cross slit is aligned with the center of the pupil. As described above, when the center of gravity of the slit does not coincide with the center of the pupil, the average irradiation angle of the illumination light is not perpendicular to the wafer surface, so the symmetry of the waveform is broken. This makes it impossible to obtain accurate measured values.

As described above, in the embodiments, the slit widths in the X and Y directions are the same, but as long as the condition of the ratio σ = 0.3 or less is satisfied, the widths of both slits are set to the same value. It may be different. By the way, in this invention,
With respect to the pattern edge, the smaller the value of σ, the higher the contrast is obtained, and the influence from the grain is that σ can be set large. Therefore, the slit width can be set small, and
If the slit length is increased, a good detection signal can be obtained for the pattern edge in the defocused state.
At this time, the detection signal from the grain is further suppressed, and its influence can be further suppressed.

[0033]

According to the present invention as described above,
Regarding the ratio σ, the cross slit is selected so as to have the slit width set to σ = 0.3 or less and the slit length set to σ = 0.6 or more in the ratio σ of the image to the pupil diameter of the objective lens. Therefore, in the direction orthogonal to the pattern, in other words, in the direction straddling a plurality of formation patterns, a horizontally long slit larger than σ = 0.6 is formed, and the pattern corresponds to the direction along the pattern, in other words, the direction of each pattern. In the detection direction of σ = 0.3 or less, the slit becomes a vertically long slit, which improves the contrast of the detection signal with respect to the pattern detection direction, and at the same time, along the detection direction over the distance at which signals of a plurality of patterns perpendicular to the pattern are simultaneously obtained. When σ is larger than σ = 0.6, it is possible to reduce the contrast of the unevenness of the grains.
Thereby, the noise level of the detection signal due to the grain can be suppressed. As a result, a detection signal having a good peak value can be obtained.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a registration measuring device according to an embodiment to which a position shift amount measuring device of the present invention is applied.

FIG. 2 is an explanatory view of the shape of the slit,
(A) is an explanatory view of the cross slit, (b) is a modification of other slits.

3A and 3B are explanatory diagrams of a deviation amount measurement mark pattern and a detection waveform, FIG. 3A is an explanatory diagram of a form of a rectangular frame, FIG. 3B is a sectional view thereof, and FIG. ) Is the detection signal waveform.

FIG. 4 is an explanatory diagram of a chip position on a wafer on which a deviation amount measurement mark pattern is formed.

FIG. 5 is an explanatory diagram of a variation amount between a cross slit and a pinhole in the present invention.

6A and 6B are explanatory views of a relationship between a cross slit and a registration mark pattern, FIG. 6A is an explanatory view of a relationship between the cross slit and a detection signal, and FIG. FIG. 5 is an explanatory diagram of a relationship between a vertically elongated slit and a detection signal in the direction along the line, and FIG. 7C is an explanatory diagram of a relationship between a horizontally elongated slit and a detection signal in a direction perpendicular to the pattern.

FIG. 7 is an explanatory diagram of a relationship between a reflection state from a grain and a cross slit.

FIG. 8 is an explanatory diagram of a CCD detection signal in the relationship between the width and the length of a cross slit,
(A) is a figure which shows the width of a cross slit, and the relationship of length,
(B) to (d) are C when the width and the length are changed.
It is a figure which shows the state of the detection signal of CD.

FIG. 9 is an explanatory diagram of an example of a cross section of the wafer in the X direction or the Y direction in a process regarding one of the deviation amount measurement mark patterns provided for measuring the positional deviation amount; a) is a cross-sectional view of the mark pattern,
(B) is a waveform diagram of the detection signal.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Objective lens, 2 ... Half mirror, 1a ... Pupil of objective lens 3 ... Relay lens, 4 ... Half mirror, 5,6 ... Cylindrical lens, 50 ... Illumination optical system, 7 ... Wafer chuck, 8 ... XYZ moving stage, 9 ... Wafer, 10, 11
... CCD linear sensor, 14 ... CCD drive circuit, 15 ...
A / D conversion circuit (A / D), 16 ... Image memory, 17 ...
High-speed numerical processor (DSP), 18 ... Focus controller, 19 ... MPU, 20 ... Control device, 21 ...
Memory, 22 ... Bus, 21a ... Registration measurement program, 21b ... Position deviation amount determination program, 23
... Stage drive circuit, 50 ... Illumination optical system, 51 ... Light source,
52 ... Condensing lens system, 53 ... Aperture stop, 54 ... Relay lens.

Claims (3)

(57) [Claims] 1. A sample is obtained by performing epi-illumination through an objective lens of an epi-illumination optical system and receiving reflected light from a plurality of patterns formed on the sample at a predetermined distance through the objective lens. A position deviation amount measuring device for measuring a position deviation amount based on a detected signal, wherein the irradiation light from a light source is introduced into the objective lens through an aperture stop having a cross slit to perform epi-illumination on the sample. The cross slit has an illumination optical system, and at the pupil position of the objective lens, one slit of the cross forms an image in a direction along the pattern, and the other slit of the cross forms a direction perpendicular to the pattern. In the ratio σ of the image to the pupil diameter of the objective lens, the slit width is set to σ = 0.3 or less and the slit width is set to σ = 0.6 or more. A positional deviation amount measuring device which obtains the detection signal by a light receiving element selected to have a lit length and arranged in a direction perpendicular to the plurality of patterns. 2. The sample is a wafer, the slit width is σ = 0.1 or less, and the center of gravity of the cross slit is set to be the center of the pupil position. The position shift amount measuring device according to claim 1, wherein the mark pattern is a mark pattern for registration measurement formed on the wafer, and the light receiving element is a linear image sensor. 3. The plurality of patterns are rectangular patterns each having a side parallel to a direction perpendicular to the arrangement direction of the light receiving elements,
The positional deviation amount measuring device according to claim 1, wherein the positional deviation amount is measured in two directions, that is, the direction of the right side and the direction of the parallel side.