JPS6214935B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method of high-precision alignment exposure when forming multilayer fine patterns on a wafer in the process of manufacturing semiconductor integrated circuits.
In particular, the present invention relates to a close-up exposure method using a diverging light beam, and to an alignment exposure method that takes advantage of the fact that the pattern transfer magnification can be minutely changed by controlling the gap between the mask and the wafer.

In the lithography process of forming various fine patterns in multiple layers on a wafer in the manufacturing process of semiconductor integrated circuits, it is necessary to resolve the fine patterns and precisely align the positions of the patterns in the multiple layers. . In the manufacturing of semiconductor integrated circuits, in order to increase productivity and reduce manufacturing costs, on the one hand, the circuit patterns are made smaller to lower the degree of integration, and on the other hand, the wafer diameter is made larger to reduce the number of circuits per wafer. It has increased.

In response, lithography processes are using soft X-rays with shorter wavelengths instead of traditional ultraviolet light to increase resolution, and to extend mask life.
In order to increase the yield of wafers, a close exposure method has come to be used instead of a contact exposure method.
When soft X-rays are used, line widths of 1
It is possible to resolve fine circuit patterns of micrometers or less. However, with the miniaturization of circuit patterns, increasingly strict alignment accuracy between multilayer patterns is required.For example, for patterns with a line width of 1 .mu.m or less, alignment accuracy of approximately ±0.1 .mu.m is required.

On the other hand, Si wafers on which integrated circuits are formed inevitably expand and contract when metal evaporation films are formed, oxide films are attached, and high-temperature treatments are applied during the process of forming multilayer patterns. . Although the amount of expansion and contraction is minute, it becomes a non-negligible amount when the diameter of the wafer becomes large and the alignment is on the order of magnitude described above.

In conventional exposure methods, the expansion and contraction of the Si wafer directly causes alignment errors in circuit patterns, which cannot be removed. Also, X using divergent light
Japanese Patent Laid-Open No. 18970/1983 discloses a method of compensating for positional deviations due to expansion and contraction of a wafer by changing the distance between the X-ray source and the mask surface in line exposure. However, the positional deviation that can be compensated for by this method is a positional deviation that is proportional to the radius in the radial direction, as shown in FIG. As shown in the figure, the non-uniform positional deviation caused by the inclination of the gap could not be corrected over the entire surface of the wafer.
Note that in FIGS. 5 and 6, arrows indicate the direction and magnitude of positional deviation, and broken lines indicate projected patterns.

The present invention has been made to eliminate such defects, and involves placing a mask having a pattern A to be exposed in close opposition to a wafer on which a reference pattern B has already been printed, and placing a point above the mask on the wafer. Pattern A is printed using an X-ray exposure method in which pattern A is superimposed on pattern B using divergent light from a light source.
By performing the following procedure when aligning and exposing pattern B, pattern A and pattern This is for positioning B. In the following, symbols in the corresponding embodiments are attached for convenience of explanation.

At least three arbitrary points P ₁ on the top surface of the wafer,
Set P ₂ and P ₃ .

After performing trial exposure, at each point, the components (Δx ₁ , Δy ₁ ) (Δx ₂ ,
Δy ₂ ) (Δx ₃ , Δy ₃ ) is measured.

The component of the above measured deviation and the component of the position of each point (R ₁ , θ ₁ ) (R ₂ , θ ₂ ) (R ₃ , θ
₃ ), the component of the misalignment of the mask with respect to the wafer in the direction of the positioning axis, that is, the parallel misalignment components ΔX and ΔY with respect to the wafer top surface, the rotational misalignment component Δθ with respect to the wafer top surface, the distance deviation component ΔS with respect to the wafer top surface, and the wafer top surface. Calculate slope components Δα and Δβ for

These components can be represented by, for example, the formula in the description of the examples.
As shown in (3), it is uniquely determined by the position of each point and the component of deviation at each point. Therefore, by solving a geometric relational expression such as equation (3), the component of the shift of the mask with respect to the wafer is calculated from the position of each point and the component of shift at each point.

The position of the mask is corrected based on each calculated component, and the positional deviation of patterns A and B is corrected over the entire wafer.

Embodiments of the present invention will be described in detail below with reference to the drawings.

Figure 1 shows the principle of close-up exposure using a point light source.
1 is a light source, 2 is a mask, and 3 is a wafer. The pattern 4 on the mask 2 to be exposed is aligned with the pattern 5 already printed on the wafer 3, and then transferred to the position of the pattern 4'. That is, in this embodiment, a case is shown in which the pattern 4' is inserted into the center of the pattern 5.

Now, if the distance between the mask 2 and the wafer 3 is S, and the distance between the light source 1 and the wafer 3 is D, then the pattern 4 on the mask 2 at the radius R is radially oriented S, D by the radial light beam. , R is projected at the position of pattern 4' shifted by an amount determined by R. In other words, in close exposure using diverging light, the pattern transferred onto the wafer has a very small magnification (for example, D=120 mm, S= If it is 20μm, it is 1.00017
magnified by 2 times). On the other hand, the position of the pattern 5 already printed on the wafer 3 changes in the radial direction as the wafer 3 expands and contracts.

The present invention attempts to utilize the characteristic of the above-mentioned exposure method, which is transfer with a small magnification, for correcting pattern positional deviations caused by wafer expansion and contraction.

Hereinafter, how to obtain the correction amount will be described in geometric detail. As shown in FIG. 2, in a coordinate system with the origin at the center of the wafer 3, at a point P represented by a polar coordinate position (R, θ), a projected image pattern 4' of a pattern 5 and a pattern 4 on the wafer 3 FIG. 3 depicts in detail the positional deviation between them. Δx and Δy at point P in FIG.
y, the relative rotation error Δθ of the mask and wafer and the change ΔS of the gap between the mask and wafer, or the light source,
It includes the distance change ΔD between the wafers and the components in each linear position direction of the radial position shift based on the expansion and contraction of the wafers. Here, the mask-wafer gap S
When the mask surface is tilted from parallel to the wafer surface, the gap changes depending on the position on the wafer.
The mask surface with respect to the wafer surface can be expressed by vertical translation ΔS and tilt errors Δα and Δβ around the X and Y orthogonal axes.

To summarize the above error components, Δx and Δy can be expressed geometrically by the following equations.

Δx=ΔX−Rsinθ・Δθ+R/Dcosθ・ΔS+R ² /Dsinθcosθ・Δα+R ² /Dcos ²・Δβ ...(1) Δy=ΔY+Rcosθ・Δθ+R/Dsinθ・ΔS+R ² /Dsin ² θ・Δα+R ² /Dsinθcosθ・Δβ ... (2) In the above equation, the error components after the third term on the right side are unique to close exposure using a point light source, and the direction of the positional shift is the same radial direction as the expansion and contraction of the wafer. Therefore, by effectively utilizing this, even if the wafer expands or contracts, it is possible to align the pattern on the wafer with the projected image of the mask pattern over the entire wafer. The right sides of equations (1) and (2) contain six error components ΔX, ΔY, Δθ, ΔS,
Δα and Δβ are included, and Δx and Δy are 6
If it is possible to measure individual data, that is, the relative positional deviation of the transfer pattern in the X and Y directions at three positions,
Six erroneous components can be found by solving the six-element simultaneous equations.

Therefore, for example, after accurately aligning marks separately provided in a certain mask-wafer gap S, trial exposure and development are performed, and positional deviations in the X and Y directions are measured at at least three locations on the wafer. For example, in a pattern like Figure 3, a, b, c,
By measuring d, the positional deviation of pattern 4' with respect to pattern 5 can be calculated as Δx=a-b/2, Δy=d-c/2
It can be found as As shown in FIG. 4, three points on the wafer 3, P ₁ (R ₁ , θ ₁ ), P ₂ (R ₂ , θ
₂ ), positional deviation in x and y directions in P ₃ (R ₃ , θ ₃ ) (Δx ₁ , Δy ₁ ) (Δx ₂ , Δy ₂ ) (Δx ₃ , Δy ₃ )
, and solve the following equation (3) to obtain the six error components ΔX,
Find ΔY, Δθ, Δα, and Δβ. In addition, formula (3)
is simply a matrix representation of equations (1) and (2).

Based on this result, during the main exposure, the gap between the mask and wafer was set to S-ΔS, and -Δα, - from the parallel plane.
After tilting by Δβ and precisely aligning with a separately provided mark, exposure is performed with corrections of -ΔX, -ΔY, and -Δθ. Correction of ΔX, ΔY, Δθ is as follows:
It is not relevant to the essence of the invention. Correction of expansion and contraction due to the wafer process depends on ΔS, Δα, and Δβ. That is, as is clear from the above description, positional deviation errors due to expansion and contraction of the wafer can be eliminated over the entire wafer by changing the gap between the mask and the wafer.

Although it is generally thought that the amount by which a wafer expands and contracts uniformly is greater than the amount by which it expands and contracts non-uniformly, it is certain that non-uniformity also exists as a factor. Furthermore, if the preceding exposure is performed with the gap tilted, the positional shift will be unevenly distributed as shown in FIG.

As explained in detail above, according to the method of the present invention, in the exposure method using the radiation beam from a point light source, for example, the expansion and contraction of the wafer and the mask. Even if the wafer gap is tilted, by correcting the gap between the mask and the wafer by a calculated amount, alignment can be achieved over the entire wafer surface, which is extremely effective.

[Brief explanation of the drawing]

Fig. 1 is a principle diagram of the proximity exposure method using a point light source; Figs. 2, 3, and 4 are explanatory diagrams of the calculation method of the correction amount showing an embodiment of the alignment exposure method of the present invention; Fig. 5
FIG. 6 is an explanatory diagram showing an example of positional deviation of a pattern on a wafer. 1...Light source, 2...Mask, 3...Wafer, 4
...pattern to be exposed on mask 2, 4'...projection pattern of pattern 4 onto wafer 5, 5...pattern already printed on wafer 3, 6...mask pattern, position of wafer pattern Displacement measurement element.

Claims (1)

[Claims] 1. Place a mask with pattern A to be exposed in close opposition to a wafer on which pattern B, which serves as a reference, has already been printed, and overlap pattern A on pattern B using a diverging light beam from a point light source above the mask. When aligning and exposing pattern A and pattern B using the printing X-ray exposure method, set at least three arbitrary points on the top surface of the wafer, perform trial exposure, and then Measure the component of the deviation between pattern A and pattern B according to the orthogonal coordinate system, and use the measured deviation component and the position component of each point on the top surface of the wafer,
The components of the deviation in the direction of the positioning axis of the mask with respect to the wafer are uniquely determined by these components, that is, the parallel deviation component with respect to the top surface of the wafer, the rotational deviation component with respect to the top surface of the wafer, the deviation component of the distance with respect to the top surface of the wafer, and the tilt component with respect to the top surface of the wafer. The mask position is corrected based on each calculated component, and the positional deviation between the patterns is corrected over the entire wafer.
An alignment exposure method characterized by aligning.