JPH09280816A - Position detection system performance evaluation method and position detection apparatus using the same - Google Patents

Abstract

(57) Abstract: To accurately evaluate the adjustment state of a detection optical system of a position detection device to achieve highly accurate alignment. SOLUTION: The performance of the detection system is directly evaluated by creating alignment marks having two different step structures on the object, rotating the object integrally by 180 ° and measuring, and confirming the state of the detection system. Method and position detecting device adjusted by the method.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a position detection system performance evaluation method for detecting the position of an object and a position detection apparatus using the same, and more particularly to a semiconductor projection exposure apparatus for manufacturing a semiconductor IC or LSI. It is suitable for observing an image of an object, detecting its position with high accuracy, and performing alignment based on the detected information.

[0002]

2. Description of the Related Art The progress of semiconductor technology has been increasing at a rapid pace in recent years, and accordingly, the progress of fine processing technology has been remarkable. In particular, the optical processing technology using the semiconductor projection exposure apparatus at the center of this has entered the submicron region with 1M DRAM as the boundary.

As a means for improving the resolution, a semiconductor projection exposure apparatus has been used in the past in which the wavelength is fixed to increase the NA of the projection optical system or the exposure wavelength is set to g.
This is a method of shortening the wavelength from the line to the i-line and further to the oscillation wavelength of the excimer laser. Recently, attempts have been made to expand the limit of optical processing by light exposure using a phase shift mask, modified illumination, or the like.

On the other hand, with the improvement in resolution, it is necessary to improve the accuracy of alignment for relatively aligning the wafer and the reticle in the semiconductor projection exposure apparatus.
The semiconductor projection exposure apparatus is both an exposure apparatus and a position detection apparatus at the same time.

FIG. 7 shows the configuration of a conventional position detection optical system for alignment of a semiconductor projection exposure apparatus. The x and y axes are taken on the surface of the wafer 4 as shown in the figure, but the position detection system of the present exposure apparatus uses x and y axes.
Since the directions are the same, the measurement in the y direction will be described here. The position detection system is a generic name for all optical systems from the light source to the detection.

Light emitted from a light source such as a He-Ne laser (not shown) is guided to an illumination optical system 11 through a fiber-12. The light from the illumination optical system 11 is reflected by the polarization beam splitter 10 as an S-polarized component perpendicular to the plane of the drawing, and λ / 4
It is transmitted through the plate 7 and converted into circularly polarized light. After that, the light passes through the imaging optical systems 6 and 5, the mirror 13 and the projection exposure optical system 1,
The mark 31 formed on the wafer 4 placed on the stage 2 which can be driven in the xyz directions is illuminated with a marker. The reflected light or scattered light from the mark 31 passes through the projection exposure optical system 1, the mirror 13 and the image forming optical systems 5 and 6 again, and then passes through the λ / 4 plate 7 to be the component P on the paper surface. It is converted to polarized light. Since the light is converted into P-polarized light, the light passes through the polarization beam splitter 10 and the image forming lens 8 forms an image of the mark 31 on the photoelectric conversion element 9 such as a CCD camera. The signal detected by the photoelectric conversion element 9 is image-processed to detect the center position of the mark 31 with high accuracy, and the stage 2 is driven from the detected value to align the wafer 4.

[0007]

However, in the conventional position detection system, there is a so-called inter-process offset in which the detection of the center position of the mark, which should be the same, differs depending on each process. It wasn't messed up.

There are two main factors that cause inter-process offset.
There are two. The first is due to the asymmetry of the step structure of the alignment mark, the distortion of the detected waveform due to resist interference, etc., and the second is due to the adjustment state of the position detection system.

For example, when the position detection system has decentering coma, the detected waveform is asymmetrical according to the following principle. FIG.
Is a schematic diagram thereof. FIG. 8A shows a state in which a mark having a stepped shape in the cross section in the measurement direction is illuminated with the illumination light 41. The scattered lights 42a and 42b are asymmetrical as shown in the figure in consideration of the eccentric coma. . FIG. 8B is an image signal of the reference mark in the state of FIG. The detected light from the mark edge has a waveform asymmetric with respect to the center of the mark.

The adjustment of the position detection system has a problem of illumination.
In an alignment optical system using image processing, an alignment mark is often illuminated by a marker. Although the kerler illumination is a method of illuminating the detection surface uniformly, it does not guarantee the uniformity of the distribution of the pupil plane of the position detection system. Actually, the light source may be eccentric with respect to the pupil plane of the position detection system, and the incident angle distribution of the illumination light with respect to the detection plane may become asymmetric, causing a measurement error.

FIG. 9 (a) is a bird's-eye view of the measurement mark 31 in the y direction, FIG. 9 (b) is a sectional view of the mark 31 seen from the x direction, and FIG. 9 (c) is the observed signal. The waveform is shown. In the figure, 3
2a, light incident perpendicularly to the mark 2a, 32b and 32
Let c be incident light from directions that have the same angle with respect to the vertical direction but different directions, and these lights 32a, 32b,
Consider a case in which a mark 31 having a step structure is illuminated by 32c to detect a position. If the intensity of the illumination light 32b is weaker than the intensity of the illumination light 32c as shown in the figure, a difference occurs in the intensity of scattered light at the mark edge portion during mark detection, and accurate position detection cannot be performed. Assuming that the cross section of the mark is completely symmetrical, it is not necessary to consider the difference in the interference condition of the scattered light from the mark edge, but even in such an ideal case,
If the intensity of the light 32c is higher than that of the light 32b, the manner of light scattering at the mark edge portion is different, and the image signal of the obtained mark is asymmetric as shown in FIG. 9C.
That is, even if the mark itself is symmetrical, the waveform of the image signal detected when the illumination condition is asymmetric is distorted, and it is difficult to accurately detect the mark position. Of course, if the intensities of the light 32b and the light 32c are equal, the waveform becomes completely symmetrical due to the symmetry.

[0012] Several proposals have been made by the applicant of the present invention to solve the asymmetry of the waveform, and they are effective in actual products. Among them, there is a promising method that utilizes the fact that the degree of asymmetry of the signal waveform described above corresponds to the step amount of the silicon etching wafer. This step can be easily formed by etching or the like.

FIG. 6 is a graph showing the value of the signal asymmetry generated by the coma aberration of the position detection system and the nonuniformity of the illumination system with respect to the step difference of the silicon etching wafer. FIG.
If the strength of one edge of the signal is a, the strength of the other edge is b, and the strength of the entire mark is c, the evaluation value E is defined as E = (ab) / c (1) , The evaluation value E is a parameter indicating the waveform distortion.
Becomes FIG. 6 is a result of measuring the evaluation value E in a state in which the position detection system or the illumination system is incomplete by changing several heights d of the step of the Si alignment mark having the rectangular step structure. In Fig. 6, the horizontal axis is the HeNe laser wavelength λ of the detected light, and
The height d obtained by taking dulus is the evaluation value E on the vertical axis. As a result of the examination, it was confirmed from the experiment and the simulation that the evaluation value E changes like a periodic function as shown in FIG.

On the other hand, FIG. 6B shows a pair of steps when the end face (pupil surface) of the fiber 12 shown in FIG. 7 is decentered in a state where the position detection optical system including the projection exposure optical system has no aberration. The evaluation value E is shown. Due to the eccentricity, the distribution of pupils in the illumination system is non-uniform. The solid line is the characteristic when the eccentricity of the optical center of gravity on the pupil plane is shifted by 3% of the NA of the position detection system in the measurement direction, and the broken line is the characteristic when it is decentered by 1.5%. The vertical axis and the horizontal axis are the same as those in FIG. It can be seen that the evaluation value E changes with amplitude according to the amount of eccentricity of the optical center of gravity on the pupil plane. The present invention has focused on that the evaluation value E becomes 0 at a step of λ / 4 cycle regardless of the adjustment state of the scope such as the asymmetry of the position detection system.

FIG. 5A is an enlarged view of one mark portion. As scattered light from the mark edge, the light from the upper left edge is 51, 52, the light from the upper right edge is 54, 55, the light from the lower left edge is 53, and the light from the lower right edge is 56. . Even light emitted from the same edge, the light 52 and the light 55 have an asymmetric relationship with the light 51 and the light 54 at the time of image formation due to the influence of aberrations such as coma of the position detection system.

Here, the wavefront of each of the light beams 51 to 56 is defined with θ as a phase.

[0017]

[Equation 1] Expressed as In the equation (3) expressed as the evaluation amount E of the final waveform, a is a combination of the lights 51, 52 and 53, and b is a combination of the lights 54, 55 and 56. Therefore, the difference a between the detected waveform signals
−b is represented by a−b = ∫ ₀ (U ₅₁ + U ₅₂ + U ₅₃ ) ² dθ−∫ ₀ (U ₅₄ + U ₅₅ + U ₅₆ ) ² dθ (3).

The result of the numerical calculation is shown in FIG. 5B, and the two curves correspond to coma aberrations of λ / 10 and λ / 20. The experimental result corresponding to this is shown in FIG. 6A, and it can be seen that the two show a good agreement. From both the theoretical simulation and the experiment, it was confirmed that the signal waveform was changed depending on the step amount of the alignment mark.

Therefore, the applicant of the present invention uses a wafer having a sensitive step amount (λ / 8) for adjustment of the position detection optical system.
The level difference (λ / 4) that is insensitive to the level difference of the alignment mark
* Proposals have been made for a vertical structure that is an (integer). This method makes it possible to adjust the detection system sensitively, and enables highly accurate alignment with an actual device.

However, the evaluation amount E in the equation (1) is an indirect thing called distortion of the signal waveform and does not directly correspond quantitatively to accuracy. The position measurement value changes even if the processing method is changed, for example, the window width of the pattern matching of the signal processing is changed. Correspondingly, it is necessary to correspond to the standard value of this distortion amount, which makes the turn complicated.

Further, a dedicated evaluation software for measuring the amount of distortion is also required for automation, which increases the load of error examination.

[0022]

In view of the above problems, the present invention proposes a direct evaluation using a position measurement amount. Therefore, in the present invention, an object (for example, a wafer)
It is characterized in that alignment marks having two different step structures (for example, the amount of step difference) are formed on the upper part, and the object is rotated integrally and measured in a plurality of states to confirm the state of the detection system. For example, the distance between a plurality of alignment marks having steps different from each other is measured by the movement of a highly accurate movable stage with a laser interferometer, and further the posture of the object is set to 1
The state of the detection system is confirmed by comparing the results of measuring the same inter-mark distance by rotating the same by 80 degrees.

In the position detecting system performance evaluation method of the present invention, the adjustment state of the detection optical system constituting one element of the position detecting device for measuring the position of the object is adjusted by a plurality of alignment marks having different step structures in advance. It is characterized by measuring and confirming the created object in a plurality of states.

[0024]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is an explanatory diagram according to a first embodiment of the present invention when two alignment marks are formed in a wafer by a distance L and have different step amounts. The distance L can be created with an error on the order of submicrons in the current technology, but it is not necessary to know the absolute value of the length within a measurement error (for example, on the order of nanometers). Even if you do not know the absolute value of the distance L exactly,
It is an effective point of the present invention that the state of the alignment detection system can be confirmed.

In this embodiment, when the wavelength used for the alignment detection system is λ, two different step structures (vertical structures) are formed in one Si wafer, for example, λ / 8 and 3λ /.
Alignment marks are created so that the amount becomes 8. This quantity is such that the signal distortions have opposite signs and take the maximum value as shown in FIG. The value of λ is the center wavelength when the detection wavelength of alignment has a width. The step difference of λ / 8 is (1 + 8N) * λ / 8, 3λ /
The step difference of 8 may be the one obtained by taking the modulus with λ of (3 + 8M) * λ / 8. Here, N and M are parameters having integer values of 0, 1, 2, ....

Since the directions in which the signal distortion and the offset are generated correspond to each other, if the aberration remains in the detection system, the arrow of 1 is shown for the mark of 3λ / 8 on the left side in FIG.
A measurement error offset of Δ1 occurs in the direction indicated by (1) and a measurement error offset of Δ2 occurs in the direction indicated by arrow of2 for the right λ / 8 mark (Δ1 and Δ2 are positive values). At this time, the distance between the two alignment marks is measured as L0 = L + Δ1 + Δ2 (4) because of the error that occurs when the true value is L.

Subsequently, this wafer is rotated 180 degrees as shown in FIG. 2 and measurement is performed again. Because of the rotation, the positions of the two alignment marks are interchanged with those in FIG. Since the direction of the offset generated corresponding to the amount of step is unchanged, each marker
In the same manner, offsets of Δ1 and Δ2 occur in the directions of arrows of1 and of2 in FIG. Therefore, the distance between the two alignment marks is now measured as L180 = L-Δ1-Δ2 (5).

From the half of the difference between the two measured values L0 and L180, the sum of the generated offsets of Δ1 + Δ2 is obtained as (L0-L180) / 2 = Δ1 + Δ2 (6). Therefore, by adjusting the position detection system so that the sum of the generated offsets is less than or equal to the standard value, it is possible to make adjustments by a direct evaluation amount.

If the alignment marks having two different step amounts can be arranged in the detection range of the position detection system, the two alignment marks can be simultaneously observed within the detection range. However, when there is a measurement error due to a position within the detection range due to, for example, nonuniformity of illumination, the wafer is mounted on a movable stage with a laser interferometer while monitoring the position with high accuracy, and moved. It is necessary to make the position where the alignment mark is detected by the position detection system the same on the order of submicron and perform measurement.

When the value of L is large and the alignment marks having two different step amounts cannot be arranged within the detection range of the position detection system, the stage with the laser interferometer is moved to move the mark position to the position detection system. Are measured in the same way. In the measurement that moves the stage with the laser interferometer, the movement error becomes the measurement error, so the same measurement is repeated,
High-accuracy measurement can be performed by averaging multiple measurements.

When the distance L is used as the evaluation amount as in this embodiment, it is necessary to consider the rotation error ε at the time of setting. For example, the distance L is 100 μm and the rotation error ε is 100P.
In terms of PM, the rotation error is 10 nm. If this error cannot be ignored, it is necessary to reduce the rotation error ε or employ a measurement method that is not affected by the rotation error ε. With current hardware, it is possible to achieve a rotation error of 10 PPM or less. Therefore, by using two types and two marks instead of the above two types and two marks, measurement that is not affected by the rotation error ε can be performed. It can be carried out.

FIG. 3 and FIG. 4 are explanatory views of a mark according to the second embodiment of the present invention in which either one of the two marks is prepared and another mark is formed in the middle. FIG. 3 shows a state in which measurement is first performed, and FIG. 4 shows a state in which the wafer of FIG. In FIG. 3, a mark A having a step of 3λ / 8 is placed at a distance L1 on the left side across a mark C of λ / 8.
A mark B having a step of 3λ / 8 is formed at a distance L2 on the right side. In FIG. 4, mark A and mark B are rotated by 180 °.
The relationship of is exchanged. Embodiment 1 using two marks
In the second embodiment, the distance L is used as the evaluation amount, but in the second embodiment, the distance L1
And the distance L2, in other words, the position of the mark C with respect to the marks A and B is the evaluation amount. In the measurement of Figure 3

[0033]

[Equation 2] In the measurement of Figure 4

[0034]

(Equation 3) Since the rotation errors ε 0 and ε 180 are 10 PPM or less and the difference between L1 and L2 is a submicron order amount,

[0035]

(Equation 4) it is conceivable that. Here, take the difference between the measured values of A-C and B-C.

[0036]

(Equation 5) Furthermore, the value of the measurement error Δ1 + Δ2 can be obtained from the difference σ0−σ180 = 4 (Δ1 + Δ2) (11) without being affected by the rotation error. Therefore, by adjusting the position detection system so that the sum of the generated offsets becomes equal to or less than the standard value, it is possible to make adjustments by a direct evaluation amount.

The Si wafer having the alignment marks of two kinds of steps shown in this embodiment can be easily manufactured by the current lithography technique. In manufacturing, first apply photoresist to a Si wafer, transfer the reticle pattern containing the desired pattern to the wafer using a semiconductor exposure device and develop it, and then use an etching device to etch the reticle to the desired level difference. I do. The desired step amount is, for example, a step amount that is λ / 8 with respect to the used wavelength λ of the alignment detection system. At this time, the pattern formed on the wafer is the desired pattern and an alignment mark used in the next exposure. It is also possible to share a desired pattern with the alignment mark.

A photoresist is again applied to the etched wafer, alignment is performed with the alignment mark created by etching, and then a second exposure is performed with a reticle containing a desired pattern. If the desired pattern transferred during the second exposure is different from the pattern used during the first exposure, a new reticle is needed. Also, if the reticle pattern used during the previous exposure was devised and the desired pattern could be formed on the wafer by shift exposure, the same reticle was used to shift the exposure by a predetermined shift amount before exposure. Good. The exposed wafer is again developed and then etched by an etching apparatus so that the desired step difference from the previous one is obtained. The desired amount of step difference this time is, for example, 3λ / 8.

By this procedure, one Si wafer having two kinds of alignment marks with different step amounts can be produced. Further, when it is desired to create several kinds of marks having different step differences, the above-mentioned procedure may be repeated.

In this embodiment, the marker used for the adjustment measurement is used.
And the alignment mark have the same shape. Therefore, if the X-direction mark and the Y-direction mark of the alignment mark are used, the mark for adjustment is created at an arbitrary position on the wafer by using the masking mechanism of the exposure device and the above-mentioned wafer shift together. be able to.

Taking the reticle shown in FIG. 10 as an example, patterns 111X and 1X are formed on the wafer by the first exposure as shown in FIG.
11Y is formed. This is similar to the reticle pattern.

During the second exposure, after alignment, as shown in FIG.
The masking blade in the semiconductor exposure apparatus is set so that only the X-direction mark 100X of 0 is transferred, and the wafer is shifted for exposure. By doing so, only the X-direction mark 100X is transferred, and the pattern 121X as shown in FIG. 12 is formed on the wafer. Next, when the masking blade setting is changed so that only the Y-direction mark in FIG. 10 is transferred and the wafer shift amount is changed and shift exposure is performed, a pattern 131Y as shown in FIG. 13 is formed on the wafer. It

By the procedure described above, it is possible to form a desired pattern at an arbitrary position.

Factors that cause an error in creating the adjustment pattern include a reticle manufacturing error, an alignment error, and a laser interferometer stage error. These are all submicron-level values. Since the present invention is a relative measurement, in principle less susceptible to these errors,
Further, as shown in the equation (9), due to the submicron setting accuracy, the error factor can be suppressed to a negligible size.

Although an Si etching wafer is used for adjustment in the above description, the present invention is not limited to this, and various substrates can be used. For example, it is possible to determine by preparing a plurality of types of wafers having a step structure, which is a problem in detecting offset. By doing so, it becomes possible to directly judge the performance of the detection system with respect to a clear target by the evaluation quantities such as the expressions (6) and (11).

It is generally difficult to actually form different semiconductor processes on one wafer, for example, to form alignment marks in a gate process and a metal wiring process in the same process as the actual process. Therefore, in the application of the present invention, for example, only the alignment mark portions of the two semiconductor process wafers may be separated from the wafer and adhesively supported on the quartz substrate. The object of the present invention can be achieved by rotating the substrate 180 degrees and measuring with a detection system.

[0047]

According to the present invention, the alignment marks having two different step structures are formed on the object, and the object is rotated integrally and measured in a plurality of states to confirm the state of the detection system. , It became possible to directly evaluate the performance of the position detection system. Since the state of the position detection system in the semiconductor exposure apparatus can be accurately evaluated, the detection system can be adjusted so that the value of the generated offset is equal to or less than the standard value, and high-precision alignment can be achieved.

Further, according to the present invention, the performance of the detection system can be quantitatively evaluated in advance for the most critical process in actual use, so that the adjustment state of the detection system can be improved and the alignment can be improved. It is possible to reduce the offset between the steps.

[Brief description of drawings]

FIG. 1 is a layout diagram of a mark according to a first embodiment of the present invention.

FIG. 2 is an explanatory view showing a state in which the mark according to the first embodiment is rotated by 180 °.

FIG. 3 is an explanatory diagram of marks according to the second embodiment of the present invention.

FIG. 4 is an explanatory view showing a state in which the mark of the second embodiment is rotated by 180 °.

FIG. 5 is a diagram showing a state of scattered light from an alignment mark.

FIG. 6 is a diagram showing a relationship between an adjustment state of a position detection system and a step amount.

FIG. 7 is a layout diagram of the optical system of the position detection system.

FIG. 8 is a detection signal when the position detection system has an aberration.

FIG. 9: Bird's-eye view of alignment mark and detection signal

FIG. 10 is an example of a reticle for producing the wafer according to the first embodiment.

11 is a mark formed by the first exposure using the reticle shown in FIG.

12 is a mark formed by the second exposure using the reticle shown in FIG.

13 is a mark formed by the third exposure using the reticle shown in FIG.

[Explanation of symbols]

1 projection exposure optical system, 2 XYZ drive stage, 3 wafer chuck, 4 wafers, 5, 6, detection optical system, 7 λ / 4 plate, 8
Imaging optical system, 9 photoelectric conversion elements, 1
0 polarization beam splitter, 11 illumination optical system, 1
2 fiber, 13 mirror, 31
Alignment mark, 32 illumination light, 34 mark
Center, 41 illumination light, 42 edge scattered light, 51-56 edge scattered light

Claims (7)

[Claims] 1. An adjustment state of a detection optical system that constitutes one element of a position detection device for measuring the position of an object is adjusted to a plurality of objects in which a plurality of alignment marks having a predetermined different step structure are prepared in advance. Position detection system performance evaluation method characterized by measuring and confirming in the state. 2. The distance between a plurality of alignment marks of the object having the predetermined step structure is measured in a first state, and then the object having the predetermined step structure is measured by the measurement method.
The position detection system performance evaluation method according to claim 1, wherein the distance between the plurality of alignment marks is measured again by rotating 0 degree, and a difference between the two measurement values is used as an evaluation amount. 3. The detection system performance evaluation method according to claim 1, wherein the object having the predetermined step structure is a Si wafer. 4. When the use wavelength or the use center wavelength of the detection optical system is λ, the plurality of step amounts are (1 + 8).
N) * λ / 8 and (3 + 8M) * λ / 8 (N, M = 0,
The position detection system performance evaluation method according to claim 1, wherein the position detection system performance evaluation method is in the vicinity of 1, 2 ,. 5. When measuring the distances of the plurality of alignment marks, the stage with the laser interferometer is moved while detecting the position of the object having the predetermined step structure, and the detection is performed. 2. The position detecting system performance evaluation method according to claim 1, wherein the position detected by the optical system is always constant. 6. A position detecting device, wherein the detection optical system is adjusted based on the evaluation amount according to claim 2. 7. The step amount is (1 + 8N) * λ, where λ is the use wavelength or the use center wavelength of the object position detection system.
/ 8 and (3 + 8M) * λ / 8 (N, M = 0,1,2,
..) The position detecting device according to claim 6, wherein the adjacent marks are arranged on the same substrate.