JPH10326743A - Method of exposure - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a simple method of exposure for forming a pattern with high precision of size control by lowering variation of pattern size generated near the limit of resolution. SOLUTION: In this method of exposure, each shading pattern of various size and density included in the exposure photo mask is formed in the predetermined size by uniformly providing negative bias, regardless of the size and form of the pattern, to the pattern size on the mask considering a projection magnification against the pattern formed on the substrate. The optimum bias and exposure are determined by obtaining the relation between the size of mask pattern and the exposure giving a designed size by means of experiment, simulation and the like for each shading pattern and convergence of the above relation for each shading pattern.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an exposure method applied to a projection exposure apparatus for manufacturing a semiconductor device.
The present invention relates to an exposure method suitable for forming a pattern having a size near a critical resolution on a semiconductor substrate in a projection exposure apparatus.

[0002]

2. Description of the Related Art In recent years, the degree of integration of a semiconductor integrated circuit represented by a DRAM (Dynamic Random Access Memory) has been steadily increasing, and as a result, the line width of a pattern has become extremely small. Accordingly, in a lithography process for forming a circuit pattern on a semiconductor substrate, transfer of a finer pattern is required.

In the current photolithography process, a circuit pattern on a photomask is printed on a photoresist applied on a semiconductor substrate by ultraviolet light to form a pattern by a reduction projection exposure apparatus (stepper or the like). .

Generally, the resolution R is given by the following equation (1) called Rayleigh's equation.

R = k ₁ λ / NA (1)

Here, λ is an exposure wavelength, NA is a lens numerical aperture, and k ₁ is a process coefficient.

From the above equation (1), it can be seen that the resolution limit is improved by shortening the wavelength λ, increasing the NA, and decreasing the process coefficient k ₁ .

Here, the reduction of the process coefficient k ₁ is achieved by increasing the resolution of the resist and by various improvements of an optical system collectively called a super-resolution technique.

In actuality, when various patterns having different dimensions are formed on a semiconductor substrate, dimensional controllability (in particular, controllability when only the dimensions are changed is called "mask linearity") is good. It is important that the contrast of the light diffraction image is sufficient.

The former (dimensional controllability) is achieved by shortening the wavelength and increasing the N
Due to the conversion to A, the latter (contrast of the optical diffraction image) is additionally achieved by reducing the process coefficient k ₁ .

[0011]

In the conventional pattern formation, the resolution has been steadily improved by shortening the wavelength, increasing the NA, and reducing the process coefficient.

However, shortening the wavelength and increasing the NA require a significant improvement in the exposure apparatus. As a result, the following problems have become apparent.

That is, when an exposure apparatus having a given performance is used, there is a limit only to the effect of improving the contrast, and the resolution is determined only by the mask linearity. This tendency is particularly remarkable when forming a light-shielding pattern (remaining pattern when a positive resist is used, and blanking pattern when a negative resist is used).

As an example of this case, using a KrF excimer laser (λ = 248 nm), NA = 0.60, σ
= 0.75 (σ: coherency factor)
Consider a case where a 1: 1 line / space (hereinafter referred to as L & S) and an isolated line pattern are exposed. Here, the resist is assumed to be a positive type.

FIG. 5 shows the relationship between the mask dimensions and the finished dimensions obtained by simulation (hereinafter, referred to as “mask dimensions”).
The values of the mask dimensions are converted on the wafer). In FIG. 5, the horizontal axis represents the mask pattern dimension, and the vertical axis represents the finished dimension of the resist pattern. In the figure, the black circles indicate the results of L & S (line and space), and the white circles indicate the results of the isolated lines.

Here, the exposure amount is adjusted to 0.18 μmL & S. The large-sized L & S pattern and the small-sized isolated line pattern are extremely thin.

The reason will be described below with reference to a light intensity distribution diagram.

FIG. 6A shows that 0.18 μm and 0.25 μm
The light intensity distribution when 1: 1 L & S is exposed is shown (solid line: 0.18 μm, broken line: 0.25 μm). FIG.
In (a), the horizontal axis indicates the position, and the vertical axis indicates the intensity. The solid line shows the light intensity distribution characteristic of 0.18 μm, and the broken line shows the light intensity distribution characteristic of 0.25 μm.

In the figure, the horizontal broken line (lth) is 0.1
This is a light intensity threshold value that gives dimensions as designed for 8 μmL & S. It can be seen that when the exposure amount is adjusted to the dimension of 0.18 μm, the finished dimension of the large dimension L & S is reduced. This is because the intensity of the light passing through the mask is greatly reduced when the size becomes smaller than the exposure wavelength.

FIG. 6B shows a similar light intensity distribution diagram for an isolated line pattern. However, the horizontal dashed line
This is a light intensity threshold value that gives dimensions as designed for 0.18 μmL & S. In this case, the size of the small-sized isolated line is reduced. This is because the intensity of the light passing through the mask increases as the wavelength becomes smaller than the dimensional exposure wavelength.

As described above, in the vicinity of the limit resolution, it is difficult to form patterns having different dimensions and different densities with desired dimensions.

As a method for avoiding this problem, a method has been proposed in which the dimensions of the pattern on the mask are corrected in advance so that the finished dimensions of the pattern are as designed. For example, in the literature (E. Kawamura, et. Al., “S
simple Method of Correcting
Optical Proximity Effect
for 0.35μm Logic LSI Cir
kits ", Jpn.J.Apl.Phys.Vo.
l. 34 (1995) pp. 6547-6551).

However, since the optimum correction value (mask bias value) differs depending on the pattern size and the sparse density, it takes much time and effort to design, manufacture or inspect a photomask.

SUMMARY OF THE INVENTION Accordingly, the present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide a pattern having a desired size in various sizes and sparse densities with high dimensional control accuracy. An object of the present invention is to provide an exposure method capable of forming a pattern.

[0025]

In order to achieve the above object, an exposure method according to the present invention is an exposure method for transferring a pattern on a mask onto a semiconductor substrate by a projection exposure apparatus. Including a pattern group of at least one of a dimension and a different shape, the pattern dimension on the mask in consideration of the projection magnification is larger than the pattern dimension formed on the semiconductor substrate. It is characterized by being composed of small patterns regardless of the type.

Further, in the present invention, the uniform bias amount of the mask pattern is set to a mask bias amount and an appropriate exposure amount for providing a desired size on the semiconductor substrate for the patterns having the different shapes and different pattern dimensions. Is determined in advance by experiments or simulations, and a region where the relationship converges is determined and determined.

[0027]

Embodiments of the present invention will be described below. In a preferred embodiment of the present invention, in a method for exposing a pattern having various dimensions and low density included in an exposure photomask, a projection magnification is considered with respect to a dimension of a pattern formed on a substrate. By applying a uniform negative bias amount to the pattern size on the mask irrespective of the pattern size and pattern shape, each pattern can be formed almost as desired.

Further, in the embodiment of the present invention, as a method of determining the bias amount, the relationship between the mask pattern size and the exposure amount giving the size as designed is determined by means of experiments and simulations for various patterns. , A convergence range of the above relationship of various patterns is obtained, and an optimum bias amount and an optimum exposure amount are determined from this range.

The operation and effect of the embodiment of the present invention will be described. For a light-shielding pattern of various sizes and sparse density included in an exposure photomask, a uniform negative bias amount appropriate to the size is given. Each pattern can be formed to a desired size. For this reason, it is not necessary to obtain an appropriate bias amount for each of various patterns.
Man-hours related to manufacturing and inspection can be significantly reduced.

[0030]

EXAMPLES Examples of the present invention will be described below in order to further describe the above-described embodiments of the present invention with specific examples.

[Embodiment 1] FIG. 1 shows an embodiment of the present invention in which λ = 248 nm, NA = 0.60, σ = 0.75.
In the case of exposing 0.18 and 0.25 μmL & S (black circles and black triangles) and 0.18 and 0.25 μm isolated line patterns (white circles and white triangles), the mask bias amount (horizontal axis) and the design The relationship with the exposure amount (hereinafter, abbreviated as “Eop”) giving the dimension as the value (vertical axis) is shown. Here, the mask bias amount is based on the dimension of the line portion, and the pitch is not changed. The resist is assumed to be a positive type. L & S of 0.18 and 0.25 μm are indicated by black circles and triangles, and isolated line patterns of 0.18 and 0.25 μm are indicated by white circles and open triangles.

When the bias amount is 0, the E
op is very different. In particular, since Eop of 0.18 μmL & S is larger than others, if this is adopted, other patterns will be very thin.

However, the larger the Eop at a bias amount of 0, the greater the bias amount dependence of Eop.

For this reason, for an appropriate negative bias amount (finished size−mask size), the E
op does not intersect at one point, but tends to converge.

In FIG. 1, for patterns having different dimensions and different sparse densities, an overlap region between a region where the dimensional difference on the substrate is within 5% and a region where the dimensional difference is within -5%, for example, under a certain exposure amount. , For all of the above-described pattern groups, the conditions are such that the dimensional variation is within ± 5%.

In FIG. 1, the convergence range that satisfies the condition that the amount of dimensional variation falls within the range of ± 5% is indicated by the hatched area.

By selecting a bias amount within this range, Eop can be made uniform for patterns having different dimensions and different densities.

In this embodiment, the optimum mask bias amount is about -0.05 μm, and the optimum exposure amount in this case is about 3 μm.
6.5 mJ / cm ² .

When the bias amount is 0 μm (the exposure amount is 53 mJ /
cm ² ) and -0.05 μm (exposure amount 36.5 mJ / c)
FIG. 2 shows the relationship between the mask size and the finished size in the case of m ² ).

The thinning of the large-sized L & S pattern and the small-sized isolated line pattern observed at the bias amount 0 is reduced to within ± 5% of the dimensional fluctuation. In FIG.
As a comparative example, regarding the relationship between the mask size and the finished size with a bias amount of 0 (no bias), L & S is shown by a small black circle and a broken line connecting it, and an isolated line is shown by a small white circle and a broken line connecting it.

As described above, the large dependence of Eop on the bias amount of a small-sized dense pattern is due to the fact that the ratio of the amount of light passing through the mask largely depends on the line size.

FIG. 3A shows a bias value of -0.05 μm.
2 shows the light intensity distribution when 0.18 and 0.25 μmL & S patterns are exposed by applying the above formula. The horizontal dashed line
This is a light intensity threshold value that gives dimensions as designed for 0.18 μmL & S. As compared with the case where the bias value is 0 (see FIG. 6A), the finished size does not largely depend on the mask size.

FIG. 3B shows a similar light intensity distribution diagram for an isolated line pattern. Like L & S, the finished dimensions do not depend much on the mask dimensions.

Embodiment 2 FIG. 4 shows that λ = 248 nm, N
In the case of exposing a 0.25 μm 1: 1 and isolated dot pattern (square of 0.25 μm on a side) at A = 0.60 and σ = 0.75, a mask bias amount and an exposure amount giving dimensions as designed values (Hereinafter, Eop). Here, the bias amount is based on the size of the dot portion, and the pitch is not changed. The resist is assumed to be a positive type.

When the bias amount is 0, the E
Although op greatly differs, each Eop can be made uniform by selecting an appropriate bias amount.

In FIG. 4, the convergence range that satisfies the condition that the dimensional variation is within ± 3% is indicated by hatched areas. For a 0.25 μm dot pattern, the optimum mask bias amount is about −0.07 μm, and the optimum exposure amount in this case is about 27 mJ / cm ² .

In the embodiment described above, λ = 248n
m, NA = 0.60, σ = 0.75, two types of patterns, line and dot, and two types of pattern pitch: 1: 1 dense pattern or isolated pattern, but these parameters are as far as listed here. Instead, the same effect can be obtained by optimizing the bias amount and the exposure amount through experiments and simulations.

[0048]

As described above, according to the present invention,
For light-shielding patterns of various dimensions and low density included in the photomask for exposure, it is possible to form each pattern as desired by applying a uniform and uniform negative bias amount to the dimensions. It works.

Therefore, according to the present invention, the labor involved in the design, manufacture, and inspection of a photomask can be greatly reduced as compared with the conventional method of obtaining appropriate bias amounts for various patterns.

[Brief description of the drawings]

FIG. 1 is a diagram showing a relationship between a mask bias amount and an exposure amount that gives a finished dimension as designed according to an embodiment of the present invention.

FIG. 2 is a diagram showing a relationship between a mask dimension and a finished dimension in one embodiment of the present invention.

3A and 3B are diagrams illustrating light intensity distributions according to an embodiment of the present invention, in which FIG. 3A is a diagram illustrating an L & S and FIG. 3B is a diagram illustrating a light intensity distribution of an isolated line pattern, respectively.

FIG. 4 is a diagram showing a relationship between a mask bias amount and an exposure amount that gives a finished dimension as designed according to the second embodiment of the present invention.

FIG. 5 is a diagram showing a relationship between a mask dimension and a finished dimension in the related art.

FIG. 6 is a diagram showing a light intensity distribution in the related art;
(A) is a figure which shows L & S, (b) is a figure which shows the light intensity distribution of an isolated line pattern, respectively.

Claims (3)

[Claims] An exposure method for transferring a pattern on a mask onto a semiconductor substrate by a projection exposure apparatus, wherein the mask includes at least one group of patterns having different pattern dimensions and different shapes. The pattern size on the mask in consideration of the projection magnification is made up of a uniformly small pattern regardless of the pattern size and the pattern shape with respect to the size of the pattern formed thereon. Exposure method. 2. A uniform bias amount of the mask pattern is obtained by previously experimenting a relationship between a mask bias amount and an appropriate exposure amount giving a desired size on the semiconductor substrate for the pattern having the different shape and different pattern size. And / or by simulation, etc., and this relationship is
An exposure method, wherein an area to be converged is determined. 3. An exposure method in which a mask includes a pattern group in which patterns having various sizes and sparse densities are mixed, and the pattern of the mask is transferred onto a semiconductor substrate by a projection exposure apparatus. Or, from the relationship between the mask bias amount obtained in advance by simulation and the exposure amount at which the resist pattern dimensions are as designed, the relationship between the mask bias amount and the optimal exposure amount for the various patterns becomes a desired dimensional variation amount. A convergence range that can be accommodated is obtained, and within this range, a mask bias amount and an optimum exposure amount are determined, and a uniform negative mask bias amount allows each resist pattern to have a desired size irrespective of pattern size and pattern shape. An exposure method, characterized in that the exposure method can be formed.