JPH0626178B2 - Pattern position detection method and device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern position detecting method and apparatus for detecting the center position of a two-dimensional pattern, and more particularly to a circuit pattern on a wafer by a reduction projection exposure apparatus. The present invention relates to a pattern position detecting method and apparatus suitable for detecting an alignment pattern during reduced projection exposure.

[Background of the Invention]

In the reduction projection exposure apparatus, as shown in FIG. 1, the circuit pattern 4 on the reticle 1 is repeatedly and successively reduced projection exposed onto the chip 5 on the wafer 3 via the reduction projection lens 2. At this time, the circuit pattern 4 needs to be accurately aligned with the chip 5, and this alignment is formed in advance on the reticle 1 and the wafer target patterns 7a and 7b previously formed on the wafer 3, for example. The reticle reference patterns 6a and 6b are used. If the wafer target pattern illumination light 8 illuminates the wafer target pattern 7a through the half mirror 10, the reticle reference pattern 6a, and the reduction projection lens 2,
The reflected light from the wafer target pattern 7a is magnified and imaged on the reticle reference pattern 6a via the reduction projection lens 2, and a composite image 9 of these patterns is formed by the half mirror 10,
Photoelectric conversion is performed on the two-dimensional image pickup device 12 via the magnifying lens 11. The combined image detection signal from the two-dimensional image pickup device 12 is then A / D converted by the preprocessing circuit 13 and then compressed in the direction orthogonal to the pattern position detection direction in order to improve the S / N ratio. The reticle reference pattern 6 is calculated by the computer 14 from the combined image detection signal
a, the center position of each of the wafer target patterns 7a is obtained. The amount of misalignment between the two patterns, and thus the amount of alignment, is known from the difference in the center positions, and the wafer table (XY table) 40 is driven and controlled in the x direction in accordance with this amount of alignment. . The same applies to the drive control of the XY table 40 in the y direction.
The reticle reference pattern 6b is provided for that purpose.

Here, how to obtain the center position of the wafer target pattern 7a will be described in detail as follows.

That is, the wafer terpolymers Getsuto pattern 7a shows a cross section in FIG. 2 (a), the SiO ₂ layer 21 to form a step on the line on the Si substrate (corresponding to the wafer 3) 20 is formed, further SiO ₂ layer twenty one
A photoresist 22 is applied to the exposed portion of the Si substrate 20 between steps as shown in the figure. The wafer target pattern 7a thus formed
On the other hand, when monochromatic light of the exposure wavelength is used as the wafer target pattern illumination light 8 in order to prevent the occurrence of chromatic aberration in the reduction projection lens 2 color-corrected to the exposure wavelength (g line), as shown in FIG. As shown, multiple interference between the incident light 23 and the reflected light 24 is caused in the photoresist 22.

By observing the wafer target pattern 7a from the upper direction due to this multiple interference, it can be seen that an equal-thickness interference fringe can be seen on the surface of the photoresist 22 as shown in FIG. 2 (b) (for example, D.R. (Die
by Trick W. Widmann) "Quantitative Everyue-Syon of
Photoresist patterns in the 1 micron range "(Quantitative Evalu
ation of Photoresist Patterns in the 1-μm Rang
e) (Applied Optics; April 1975, Vol.14, No.4 p931-9
34). In this case, assuming that the photoresist 22 is symmetrical with respect to the center between the pattern steps formed by the SiO ₂ layer 21 and is uniformly applied, the photoelectric conversion by the two-dimensional image pickup device 12, for example, along the line AA ′ is performed. The signal and the output of the preprocessing circuit 13 are also symmetrical with respect to the step center position x ₀ as shown in the second (c).

Therefore, the output of the preprocessing circuit 13 at the position I is newly set as X (I), and the symmetry evaluation function Z (I) shown in the equation (1) is obtained by calculation, and Z (I) is minimized. The symmetric center position, that is, the wafer target pattern center position can be obtained from the position I when the value is taken.

The symmetry evaluation function Z (I) will be described in detail with reference to FIGS. 3 (a) and 3 (b). This is nothing but determining the virtual symmetry center position Ia and obtaining Z (Ia). As shown in FIG. 3 (b), when the virtual symmetric center position Ia deviates from the true symmetric center position Iw, Z (Ia)> Z (Iw), and when Ia = Iw, Z
(Ia) is where the minimum value is reached. Then, each time Ia is updated, Z (Ia) is calculated, and then the minimum value of Z (Ia) is calculated, and Ia when the minimum value is obtained is taken as the true symmetric center position. .

Although the center position of the wafer target pattern as a two-dimensional pattern can be obtained as described above, such a method is not without problems. This is because if X (I) has symmetry, it is good, but if X (I) has asymmetry, there is a great risk of mistaking the center position. That is, in the case of an Al pattern which has different inclinations in the pattern step portion on the left and right, or has a step difference larger than other patterns and has graininess, the resist coating state is asymmetrical and X ( As shown in FIGS. 4 (a) and 4 (b), the symmetry of the waveform of I) is greatly impaired. The situation is the same when random noise is mixed. By the way, referring to FIGS. 4 (a) and 4 (b), in the region from the virtual symmetric center positions Iw, Iv to the positions Iw + J, Iv + J. The asymmetrical portion is shown with diagonal lines. In this case, although the position Iw is the true center position, FIG.
As shown in (c), Z (Iv) <Z (Iw), and the position Iv is erroneously detected as the center position.

[Object of the Invention]

Therefore, it is an object of the present invention to provide a pattern position detecting method and apparatus capable of accurately obtaining the center position of a pattern even if the reflected light intensity distribution from the detected pattern is asymmetric.

[Outline of Invention]

To this end, the invention is such that the composite image detection signal is processed by a weighted symmetry evaluation function.
That is, when processing is performed by the symmetry evaluation function, the one-dimensional light intensity distribution signal (composite image detection signal) in the two fixed areas set as updatable becomes the one-dimensional light intensity distribution signal in the other area. After effectively strengthening by weighting in comparison,
The center of symmetry of the two-dimensional pattern is obtained by performing the symmetry evaluation process.

Example of Invention

 The present invention will be described below with reference to FIGS. 5 to 15.

First of all, the theoretical background will be described before starting the specific description of the present invention.

The TTL (Through The Lens) wafer target pattern detection system has already been described with reference to FIG. 1. In this detection system, light having the same wavelength as the exposure light is used as the wafer target pattern illumination light to prevent chromatic aberration in the reduction projection lens. Is usually used, or monochromatic light is used even if not of the same wavelength. by the way,
FIG. 5 (a) shows the photoresist 2 on the Si substrate 20 via the SiO ₂ layer 21.
2 shows the state where it is painted, but it will be S
Wafer target pattern illumination light is incident on the i substrate 20 25
When it is incident as, multiple reflection occurs and multiple interference occurs. When the incident light 25 is incident, the reflected light 26 from the surface of the photoresist 22 and the reflected light 27 from the surface of the SiO ₂ layer 21, the reflected light 28 from the surface of the Si substrate 20 and the photoresist 22.
Reflected light reflected on the surface of the SiO ₂ layer 21 after being reflected on the surface 2
9. Multiple reflections occur due to various reflected lights such as the reflected light 30 reflected on the surface of the Si substrate 20 after being reflected on the surface of the SiO ₂ layer 21, resulting in multiple interference. The multiple interference intensity in this case changes greatly periodically depending on the photoresist film thickness as shown in FIG. 5 (b). The period depends on the wavelength of incident light, the thickness of layers other than photoresist, and the refractive index. It depends on you. Note that FIG. 5 (b) shows the relative multiple interference intensity when the incident light intensity is standardized to 1.

Except when the uppermost layer has a high refractive index such as Al, other layers such as Poly-Si, HLD, Si ₃ N ₄ and PSG have a refractive index close to that of the photoresist. When the incident light intensity is 1, the multiple interference intensity is about 0.07-0.
In the range of 52, it greatly changes periodically according to the photoresist film thickness, and its outline is almost the same as that shown in FIG. 5 (b). That is, when the uppermost layer is other than Al, the intensity distribution of reflected light from the wafer target pattern depends on the photoresist film thickness.

Now, FIG. 6 (a) shows a state in which the photoresist 22 is applied to the upper part of the concave target pattern step formed by the SiO ₂ layer 21 on the Si substrate 20, together with the photoresist film thickness distribution. From this, it is clear that the photoresist film thickness changes abruptly at the pattern step. As a result, as estimated from the multiple interference intensity described in FIG. 5 (b), the reflected light intensity changes abruptly near the pattern step as shown in FIG. 6 (b).

Note that c indicates the pattern width.

Similarly, FIGS. 7 (a) and 7 (b) show the convex target pattern step difference. Also in this case, it can be seen that the intensity of the reflected light changes rapidly near the pattern step portion due to the rapid change in the photoresist film thickness.

The case where no Al layer is present has been described above.
(a) shows an Al layer 31 in which a SiO ₂ layer 21 and an Al layer 31 are formed on a Si substrate 20.
It shows a state of multiple interference in the case where a photoresist 22 is further applied on the above. When the incident light 32 is incident, the reflected light 33 from the surface of the photoresist 22 and the reflected light 34 from the surface of the Al layer 31 are reflected by the surface of the photoresist 22 and then the Al layer.
31 Multiple reflections occur due to the reflected lights 35 and 36 reflected on the surface, and multiple interference occurs. However, the complex index of refraction of Al is 0.46-i4.24, which is much larger than the complex index of refraction of 1.67-i0 of photoresist (AZ1350J). Therefore, the reflectivity on the surface of Al layer 31 is very large. Will be things. As a result, the balance of the reflected light intensities is impaired, and the periodic change of the multiple interference intensity with respect to the photoresist film thickness is as shown in FIG. 8 (b) when the uppermost layer (the layer immediately below the photoresist) is other than Al. It is much smaller than That is, when the uppermost layer is Al, the intensity distribution of the reflected light from the wafer target pattern depends not on the photoresist film thickness but on the diffraction effect at the pattern step portion.

FIG. 9 (a) shows a state of diffracted light 32 at the pattern step portion when the concave target pattern step is formed on the Si substrate 20 by the SiO ₂ layer 21 and the Al layer 31.
Some Al particles have large graininess, and as a result, Fig. 9 (b)
As shown in (1), speckle noise is generated in the reflected light intensity, but it can be seen that the reflected light intensity corresponding to the step portion drops sharply due to the diffraction effect at the step portion. Figure 10 (a), (b)
Shows the convex shape, but in this case as well, the reflected light intensity sharply drops at the pattern step portion.

Therefore, as is clear from the above description, the intensity of the reflected light from the wafer target pattern always changes greatly in the vicinity of the pattern step, regardless of the material of the uppermost layer forming the pattern. In particular, when the pattern material is Al, the reflected light intensity always sharply drops in the vicinity of the pattern step portion as compared with the other materials. Therefore, if the reflected light intensity signal in the area near the pattern step portion is weighted and is emphasized more effectively than that in other areas, even if the reflected light intensity distribution from the detected pattern is asymmetric, The center position of the pattern can be obtained with high accuracy by stabilizing the pattern without being affected by random noise.

The wafer target pattern is shown in FIG. 1, for example.
The present invention will be specifically described as being detected by a TTL pattern detection system.

First, referring to FIGS. 11 to 13, the first embodiment of the present invention will be described. As in the case of FIG. 1, the reflected light from the wafer target pattern is detected by the two-dimensional image pickup device. It is sent to the computer in the form of A / D conversion by the preprocessing circuit, and further converted into a one-dimensional detection signal. On the computer, the pattern width in the pattern position detection direction
The value of the weighted symmetry evaluation function Y (I) for each virtual symmetric center position is obtained by the equation (2) with reference to the information (c).

However, f ₂ (J) is a weighting function and its outline is as shown in the eleventh. Therefore, the equation (2) is nothing but the equation (1) to which the weighting function f ₁ (J) for enhancing the one-dimensional detection signal in the two specific pattern regions is added. By the way first
The weighting function f ₁ (J) shown in Fig. ₁ is as follows.

However, the value of b depends on the pattern width c and the resolution a of the pattern detection system on the wafer. Is required as. If the pattern width c is 6 μm and the resolution a of the pattern detection system on the wafer is 0.04 μm, the value of b can be obtained as 75. Equation (3) or first
According to Fig. 1, for the one-dimensional detection signal near the pattern step, the maximum W ₂ / W ₁ is higher than that for other areas.
Double weighting is performed, but the values of W ₁ and W ₂ may be set arbitrarily and appropriately according to the pattern material. Further, the values of P and Q are appropriately determined in consideration of the pattern material, the size of the pattern step portion and the inclination state thereof.

FIG. 12 shows a processing flow by the computer. According to this, in processing, first, b, P, and
The values of Q, W ₁ , W ₂ and N are set, and the data acquisition of the one-dimensional detection signal is performed. The one-dimensional detection signal data fetched after this has the noise data component removed by smoothing processing, and the edge thereof is further emphasized by high-pass processing. As a result, the one-dimensional detection signal data is obtained as waveform data in which the noise component is small and the step portion is clarified. By appropriately processing the one-dimensional detection signal data processed in this way, the coarse symmetric center position Ig of the pattern can be obtained as the center position between edges or by area calculation. Once the coarse symmetry center position Ig is determined, it is then determined whether this is within the set value. If it is not within the set value, the wafer stage is moved to move the wafer target pattern toward the center of the detection area, and the above process is repeated again. On the other hand, if it is within the set value, the formula
The process shown in (2) is supposed to be performed every time the position I is updated. As a result, the value of Y (I) corresponding to the position I is obtained. If the minimum value is obtained from these values of Y (I), the position I when the minimum value is obtained can be obtained as the true symmetric center position Iw. It is a thing.

To find the center of symmetry in this way, refer to Fig. 13.
Even if the symmetry of the waveform of the one-dimensional detection signal is significantly impaired as shown in (a) and (b), if the position I is at the true center position of symmetry Iw, the weighting function (f ₁ (J)) 33 Since the detection signal portion corresponding to the vicinity of the pattern step portion is emphasized by, the value of Y (Iw) corresponding to the true symmetric center position Iw can be seen from the difference in the double diagonal display area in the weighting function 33. As shown in FIG. 13 (c), it becomes a minimum. Therefore, the position of the wafer target pattern can be detected with high accuracy.

Next, the second embodiment will be described. FIG. 14 (a) shows a weighting function f ₂ (J) according to the embodiment. The weighting function f ₂ (J) in this case is as follows.

As can be seen from FIG. 14 (b), the weighting function (f ₂ (J)) 34 is a wafer target pattern made of a pattern material such that the detection signal waveform near the pattern step changes abruptly in a narrow region, For example, Al with a steep slope at the pattern step
It is extremely effective for patterns and the like. Therefore, the weighting function as used in the previous embodiment is used for the pattern in which the detection signal waveform depends on the multiple interference and the waveform change in the pattern step is relatively gradual as compared with the Al pattern. On the other hand, for the Al pattern, the weighting function f ₂ (J)
When such a material is used, if the weighting function is selectively used in consideration of the pattern step shape and the pattern material, the pattern position can always be detected stably and with high accuracy. The values of W ₁ and W ₂ in FIG. 14 (a) are set in the same manner as in the first embodiment, and the value of R is set in consideration of the pattern material and the inclination at the pattern step portion. It has become a thing. The computer processing in the second embodiment is performed as shown in FIG.

Finally, a third embodiment of the present invention will be described.

The weighted symmetry evaluation function in the third embodiment is not of the form as shown in formula (2) but as shown in formula (5).

However, X '(I) is a derivative of X (I).

As shown in FIGS. 15 (a) to 15 (c), when the true symmetric center position is obtained from the detection signal that occupies a large proportion of the entire flat area, the position can be obtained without error. That is, when using the weighted symmetry evaluation function as shown in equation (2), as shown in FIG.
Since the value of the symmetry evaluation function at is weighted with respect to the substantially equal flat regions on both sides of the position Iv, it becomes almost equal to that of the symmetry evaluation function at the true symmetric center position Iw. Therefore, as shown in FIG. 15 (c), there are two or more minimum points with almost the same value, and there is a risk that the virtual symmetric center position Iv will be erroneously detected as a true one. is there. However, in the case of using the one shown in the formula (5), the waveform is effectively divided by dividing the value of the weighted symmetry evaluation function as shown in the formula (2) by the differential value of the detection signal. As a result of weighting the area having a large change, the true symmetric center position Iw can be reliably obtained even if the proportion of the flat area is large.
When the virtual symmetry center position Iv matches the true symmetry center position Iw, the value of the denominator in the equation (5) becomes extremely large. Therefore, the weighted symmetry evaluation function Y ^* (Iw) has a minimum value. It will be taken. The weighting function f in equation (5) is
₃ (J) may be the weighting functions f ₁ (J) and f ₂ (J) described above, and the value of n is supposed to be appropriately set according to the detected signal waveform. . Further, although the denominator in the equation (5) is a differential function, any function other than the differential value, such as an autocorrelation function, that represents a change in signal intensity can be used.

The present invention has been described above by taking the TTL pattern detection system of the reduction projection exposure apparatus as an example and using monochromatic light as the illumination light.
Obviously, it can also be applied to a pattern detection system.

〔The invention's effect〕

As described above, according to the present invention, even if the reflected light intensity distribution from the detected pattern is asymmetric, the center position of the pattern can be obtained. Therefore, for example, when the present invention is applied to the detection of a wafer target pattern when performing reduction projection exposure, it can contribute to the improvement of wafer alignment accuracy and hence to the improvement of semiconductor productivity.

[Brief description of drawings]

FIG. 1 is a diagram for explaining a method of aligning a chip with a circuit pattern by a conventional wafer target pattern when performing reduction projection exposure on a wafer, FIG. 2
FIGS. 4 (a) to 4 (c) and FIGS. 3 (a) and 3 (b) are views for explaining a method for obtaining the center position of the wafer target pattern, and FIGS. FIG. 5 (a) to (b), FIG. 6 (a), (b), FIG. 7 (a), (b), FIG. 8 (a) ) To (b), FIGS. 9 (a) and (b) and FIGS. 10 (a) and (b) are diagrams for explaining the theoretical background of the present invention, and FIG. 11 is a diagram for explaining the present invention. FIG. 12 is a diagram showing a weighting function according to the first embodiment, FIG. 12 is a diagram showing a flow of processing by a computer according to the embodiment, and FIGS. 13 (a) to 13 (c) are diagrams showing the symmetry of the detection signal waveform. Figures 14 (a) and 14 (b) for explaining that the pattern symmetry center position can be obtained even if it is damaged.
FIG. 15A is a diagram showing a weighting function according to the second embodiment of the present invention and a detection signal waveform to which the weighting function is applied.
FIG. 7C is a diagram for explaining the third embodiment of the present invention when the detection signal waveform is special. 1 ... Reticle, 2 ... Reduction projection lens, 3 ... Wafer, 4 ... Circuit pattern, 5 ... Chip, 6a, 6b ...
... reticle reference pattern, 7a, 7b ... wafer target pattern, 12 ... two-dimensional image sensor, 14 ... calculator.

Claims (4)

[Claims] 1. An optical image from a pattern detected by a detection optical system is received by a photoelectric conversion means to obtain at least a one-dimensional video signal, and a folding point as a virtual center position is updated every predetermined time. In addition, the obtained at least one-dimensional video signal is symmetrically folded back at the folding point to check the matching degree between the two signals, the folding point having the highest matching degree is obtained, and the folding point having the highest matching degree is obtained. In the pattern position detecting method in which is detected as the center position of the pattern, when checking the degree of coincidence between both signals, by weighting only the area distant by a desired distance from the folding point, A pattern position detecting method characterized by checking the degree of coincidence of signals. 2. The desired distance is approximately 1 of the width of the pattern.
The pattern position detecting method according to claim 1, wherein / 2 is set. 3. A detection optical system for detecting an optical image from a pattern, a photoelectric conversion means for receiving an optical image detected by the detection optical system to obtain at least a one-dimensional video signal, and the photoelectric conversion means. The obtained at least one-dimensional video signal includes means for folding back at the folding point as the virtual center position to check the degree of coincidence between the two signals, and means for obtaining the folding point having the highest degree of coincidence. In a pattern position detection device that detects a high turnaround point as the center position of a pattern, the means for checking the degree of coincidence between both signals, when checking the degree of coincidence between the two signals, only in an area distant by a desired distance from the turnaround point. A pattern position detecting device, characterized in that the degree of coincidence between the two signals is checked by weighting. 4. The desired distance is approximately 1 of the width of the pattern.
The pattern position detecting device according to claim 3, wherein / 2 is set.