JPH07306523A - Method for designing mask pattern - Google Patents

Abstract

PURPOSE:To make it possible to easily design mask patterns having excellent resolution and depth of focus without generating parasitic peaks around the main peak of an image plane intensity distribution. CONSTITUTION:After mutual intensity distribution is calculated, the intensity distribution of the projection images on a wafer to be exposed which is the imaging plane is calculated with the real part of the mutual intersity distribution as the temporally mask patterns. The calculated projection images and the desired patterns are then compared and the temporally patterns are adopted as the designed mask pattern when the desired patterns are obtd. The patterns formed by shifting the origins of the real parts of the mutual intensity distribution by adding a specified value to the real parts of the mutual intensity at all the points on the imaging plane are freshly adopted as the temporally patterns if the calculated projection images are not the desired patterns.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of designing a mask pattern of an original drawing substrate used in a projection exposure method used in photolithography for forming a fine pattern, and more particularly to a hole pattern formation at the time of manufacturing a large scale integrated circuit. The present invention relates to a method of designing a mask pattern of an original drawing substrate used for projection exposure.

[0002]

2. Description of the Related Art When manufacturing a large scale integrated circuit (LSI), a fine pattern is formed on a wafer by projecting and exposing a mask pattern of a photomask (original substrate) on which a required pattern is drawn. This is performed, for example, by a projection exposure apparatus using ultraviolet light as a light source, and the mask pattern on the photomask is printed on the photoresist on the wafer.

FIG. 9 is a perspective view showing a schematic structure of such a projection exposure apparatus. In the figure, 91 is a light source, 92 is an irradiation lens for collimating the light from the light source 91 into parallel light, 93 is a photomask, 94 is a first projection lens, and 95 is
Is a second projection lens, 96 is a pupil in a projection optical system formed by the first projection lens 94 and the second projection lens 95, 9
Reference numeral 7 is a wafer onto which the mask pattern of the photomask 93 is transferred. In recent years, projection exposure optical systems have reached a resolution close to a theoretical limit determined by the wavelength of the light source used within a predetermined region on the wafer surface. Here, in order to meet the demand for higher resolution, various measures as described below have been made.

One is a so-called grazing incidence illumination exposure method, which is a method in which the shape of the light source is four points or annular zones separated from each other. In addition to this, there is a method of providing a transmittance adjustment filter (pupil plane filter) on the pupil of the projection exposure optical system. This method is described in Reference 1 (“High-resolution optical lithography using a new illumination system-projection exposure method using oblique incidence illumination and pupil plane filter (SSBL method)-” Seitaro Matsuo, Katsuyuki Harada, Applied Physics Vol. 61, No. 11 ( 1992) pp.1139-1142).
The effect is that, when the contrast of the formed line and space pattern, the depth of focus, and the resolution are comprehensively evaluated, a performance improvement of about 1.6 to 1.8 times is achieved.

Another method is to use a phase shift mask instead of the pupil filter. This is a method in which projection exposure is performed using a photomask having mask objects composed of phase objects (shifters) so that the phases of light passing between adjacent mask patterns are different from each other. Then, the contrast of the images of the adjacent patterns is improved by the interference of the light having the phase difference. At the same time, the depth of focus is also improved, and an overall improvement effect of about 1.8 times can be obtained.

However, even in the above two methods,
In the case where the target mask pattern is an isolated hole pattern, the improvement effect is small and the effect cannot be obtained very much. This is because the distance to the adjacent square hole pattern is about several times the wavelength of the light used and no interference occurs. Therefore, in the case of the hole pattern, as shown in FIG.
As shown in (b), the periphery is covered with the light shielding portion 101.
Auxiliary patterns 103, 103 on the individual hole patterns 102
By providing a, the transmittance and the phase are adjusted to improve the resolution.

Alternatively, as shown in FIG. 10 (c), there is also a method in which the background 101a of the hole pattern 102 is made to have a weak transparency and further has a phase difference. However,
In 0 (a) and (b), since the auxiliary patterns 103 and 103a cannot be set large, the improvement effect is small. Also,
In FIG. 10C as well, since the transmittance of the background 101a makes the vibrating intensity distribution parasitic around the hole pattern 102, the improvement effect is limited.

The layout of the auxiliary pattern or the background phase shifter shown in FIG. 10 can be intuitively understood, but recently, the transmittance and the phase distribution of the hole pattern mask on the photomask are theoretically obtained. Efforts are being made. One of them is FLEX, which improves the formation characteristics of a hole pattern, which is a fine mask pattern that exists independently, by raising and lowering the image plane position without making any modifications to the mask pattern or the projection optical system. There is a law.
Here, as a method of vertically vibrating the image forming surface, there is a method of mechanically vertically moving a wafer table on which a wafer is placed by using a piezo element or the like.

Furthermore, instead of effectively increasing the depth of focus by vertically vibrating the image forming surface on which the resist is applied and forming a pattern, the image plane position is moved up and down by a pupil filter method that modulates the lens pupil. There is a super-flex method in which the image plane position is moved up and down by a modulation mask method in which the mask is modulated (reference 2: "Axial Image Super").
posing (Super-Flex) Effect by a Mask Modulation Meth
od for Optical Lithography "by H. Fukuda, Digest of M
icroProcess '91, No.B-9-1, p.190).

An outline of the superflex method will be described below. Each point on the image plane is a point where the wave front of the arriving light is in phase, and shifting the wave front of the light arriving at the image plane is equivalent to shifting the image plane position in the optical axis direction. Is. Therefore, the light coming to one point on the image plane has a phase and an amplitude according to the position of the transmission part of the pupil aperture.

Therefore, when the phase of the transmitted light is changed concentrically from the center of the pupil to the periphery, the image forming position in the optical axis direction of the light transmitted from each concentric ring zone of the pupil is Each will be different. High-order diffracted light is mainly incident on the part far from the center of the pupil. Therefore, the pupil filter method in the Superflex method divides the diffracted light by the order of diffraction and forms an image on each image plane. Is a way to contribute.

On the other hand, the modulation mask method uses the fact that the pupil plane is the Fourier transform plane of the photomask plane (strictly speaking, when plane wave irradiation is performed by a point light source). That is,
By multiplying the Fourier transform image of the mask pattern by the above-described transmittance distribution and phase distribution of the pupil filter and performing Fourier transform on the image, the transmittance distribution and phase distribution of the mask pattern of the photomask to be modulated can be obtained.

Here, a method of designing a pupil filter in the superflex method will be briefly described. The pupil filter distribution is designed as follows. In the case of a normal pupil, the defocus function of the projection optical system is exp {i2πz
It is given by (f ² + g ² )}. Here, z is the defocus amount, and f and g are two-dimensional coordinates on the pupil plane. When synthesizing two defocus planes of z−D and z + D above and below the ideal image plane, the defocus function of the projection optical system is exp.
{I2πz (f ² + g ² )} cos {2πD (f ² + g
² ) -H} and cos {2πD (f ² + g ² )-
The term H} is the modulation amount of the pupil filter. H is an amount related to the phase change associated with the two defocuses.

The distribution of the pupil filter is obtained by performing optimization using D and H as fitting parameters. The effect is superior to that of the mask pattern shown in FIG. 10, and an improvement effect of about 3 times is obtained in terms of the depth of focus. FIG. 11 shows a mask pattern obtained by further converting the pupil filter distribution in the above-mentioned super flex method into mask modulation, and FIG. 11A is a plan view showing the amplitude transmittance of the obtained mask pattern by contour lines. , (B) are distribution charts showing the transmittance distribution.

The drawback of this method is that small peaks are parasitic around the main peak of the image intensity. The level at which the resist film coated on the wafer is exposed is 33% of the peak value.
%, The presence of this parasitic peak needs attention. This also applies to the masks shown in FIGS. 10B and 10C.

Another theoretical hole pattern mask design method is a method called Bessel beam mask (Reference 3: "Real and Imaginary Phase-Shitting Mask").
s ", FMSchellenberg and MD Levenson, Engineering T
echnology (1992-11)). FIG. 12 is a plan view showing this mask pattern, and (a) shows the mask pattern,
(B) shows the state after Fourier transform.

In FIG. 12, reference numeral 111 denotes a light shielding portion, and 112
Is a shifter having a transmittance of 1.0 and a phase of 180 °, and 113 is a circular transmitting portion. Here, if the width of the shifter 112 is optimally selected, the Fourier transform of this mask pattern will be as shown in FIG.
As shown in (b), it becomes a Bessel function of the 0th order of the first kind, and the transmitted light propagates without spreading. However, since this Fourier transform image is formed at the position of the projection optical pupil, the light is cut off at the portion outside the pupil diameter, and the original effect cannot be sufficiently exerted. Moreover, the pattern shape is limited to a circle.

In addition, by using a simple square hole pattern as a starting point for initial design, the area near the hole is divided into minute mosaics, and the transmittance and phase change of each mosaic piece are obtained by an optimization method by a computer and finally determined. There is also a method of obtaining design results (Reference 4: "Systematic Design of Ph
ase-Shifting Masks with Extended Depth of Focusand
/ or Shifted Focus Plane "by Y.Lin, A.Pfan and A.Zakh
or; IEEE Trans.Semicon.Manufact. Vol6, No.1 (1993) 1-
twenty one). However, this method imposes a heavy burden on the computer and requires a large-scale calculation for a long time.

[0019]

As described above,
Conventionally, in order to meet the demand for high resolution, a grazing incidence illumination exposure method, a method of providing a pupil plane filter on a projection exposure optical pupil, a method of using a phase shift mask in combination, and a flex (FLE) method.
X) method, a method called Bessel beam mask, a method using a minute mosaic, etc. were used. However, when the grazing incidence illumination exposure method is used, if the target mask pattern is an isolated hole pattern, the improvement effect is small, and there is a problem that the effect is not very obtained.

Further, the flex method has a problem that a small mountain is parasitic around the main peak of the intensity of the obtained image, and in the method called the Bessel beam mask, light is cut off in a portion outside the pupil diameter. Therefore, the original effect cannot be sufficiently exhibited, and in addition, there is a problem that the pattern shape is limited to a circular shape. The method using the small mosaic has a problem that a large-scale calculation for a long time is required. That is, the mask pattern of the hole pattern is required to have excellent resolution and depth of focus without causing a parasitic peak around the main peak of the image plane intensity distribution. There was no simple way to design a pattern.

The present invention has been made in order to solve the above problems, and is excellent in resolution and depth of focus without causing a parasitic peak around the main peak of the image plane intensity distribution. It is an object of the present invention to make it possible to easily design a mask pattern.

[0022]

A mask pattern designing method of the present invention calculates a mutual intensity distribution on a substrate to be exposed, and a real part of the mutual intensity distribution is used as a temporary mask pattern on the substrate to be exposed. The feature is that a projected image is calculated.

[0023]

The use of the real part of the mutual intensity of the light intensity on the image plane as an image means that it is obtained as a transmittance distribution having positive and negative values, and the real part of the mutual intensity distribution is the mask. It becomes the design value of the pattern.

[0024]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a flow chart showing a design procedure of a hole pattern forming mask according to the present invention. First, an initial mask pattern is set (step S1). The initial mask pattern may be the same as the target hole pattern, but the mask pattern is slightly smaller than the target hole pattern because the 30% level of the image intensity is adjusted to the desired design dimension. preferable.

Next, the optical conditions are set (step S
2). What is important here is that the condition is such that the pupil diameter is large and close to perfect image formation, and that the condition is such that the outer periphery of the pupil has a phase inversion effect. Under the simulation condition for obtaining the projection optical image, a large pupil diameter calculated means that an image close to perfect image formation can be obtained. An image in which the edge of the mask is emphasized can be obtained by installing a phase inversion pupil filter in a portion near the outer circumference of the pupil. Next, the mutual intensity distribution is calculated based on the optical condition set in step S2 (step S3).

Here, the calculation of the mutual strength will be briefly described. Assuming that two points in the free space are P ₁ and P ₂ , a light source is a group of point light sources, an electric field at P ₁ is V ₁ and an electric field at P ₂ is V ₂ , mutual strength is defined by the following mathematical formula 1. It

[0027]

[Equation 1]

Since each point light source contributes to V ₁ and V _2, it can be rewritten in the form of the sum of the point light source points, as shown in the following formula 2.

[0029]

[Equation 2]

Since the light source is usually a heat source (such as a mercury lamp), the time average from different point light sources, that is, Equation 2
The second term on the right side of is zero. The first term on the right side of the equation 2 indicates the time average from the same point light source, and if the spectrum of the light source is sufficiently narrow (quasi-monochromatic light), the time average can be expressed as an effective value. . Therefore, the mutual strength is expressed by the following Expression 3.

[0031]

[Equation 3]

That is, the mutual intensity is an electric field having a phase difference due to a difference in optical path length of a light beam propagating from the light source, which is formed at points P ₁ and P _2, and a complex conjugate product of them is used as a light source. It is the one that was integrated over the span. Therefore, when the light source is close to the point light source, the correlation is strong and the mutual intensity has a finite value (complex number). However, when the light source is widely spread, the mutual intensity is averaged and becomes close to zero. In the above description, for simplicity, the case where there is no mask or lens between the light source and P ₁ and P ₂ has been described. Here, when a mask or lens is inserted between the light source and P ₁ or P ₂ , it is necessary to consider the transmission function of the mask and the point response function of the lens, but the basic idea of mutual intensity is common. Is.

As described above, the mutual intensity is a quantity representing how much the two points (P ₁ , P ₂ ) on the two-dimensional pattern are imaged while maintaining the coherence. It is known that the mutual intensity when two points (P ₁ , P ₂ ) coincide with each other becomes a real number and becomes equal to the light intensity at that point. The mutual intensity distribution J (P) sets the reference point P ₁ on the image plane and the mutual intensity J with the point P on the coupling plane.
(P ₁ , P) is calculated, and P is obtained by scanning the image plane.

The mutual intensity distribution is generally a complex two-dimensional distribution. The reference point depends on the pattern, but the center of the area where the patterns are dense or close to each other may be selected. For details of the calculation of the mutual intensity distribution, see Japanese Patent Laid-Open No. 6-291.
No. 80, and Document 5 ("Analitical Method fo
r Image Characteristivs of Auuular Illumination wi
th a Spatial Filter in Optical Projection Lithogra
phy '' Y. Takeuchi et al., Jpn.J.Appl.Phys. valo.31 (1992) N
o.12B p.4120).

The procedure for designing a mask pattern will be described below. After calculating the mutual intensity distribution (step S
3) The real part of the mutual intensity distribution is used as a temporary mask pattern (step S4), and the intensity distribution of the projected image on the exposed wafer, which is the image plane, is calculated (step S5). Then, the calculated projected image is compared with the desired pattern (step S
6) If the calculated projected image has a desired pattern, the temporary mask pattern is set as the designed mask pattern (step S7).

On the other hand, when the calculated projected image does not have the desired pattern, the constant value is added to the real parts of the mutual intensities at all points on the image plane to obtain the mutual intensity distribution. The original part (0 point) of the real part is shifted to be a temporary mask pattern again (step S8). Then, the intensity distribution of the projected image on the image plane, that is, the wafer to be exposed is calculated again (step S5). Here, when the shift amount of the mutual intensity distribution has a negative value, it means that the negative value is changed to the real part of the mutual intensity of all points on the image plane.

By the way, to determine whether or not the projected image has a desired pattern, a slice level is set according to the resist used in the assumed exposure process, and contour lines of light intensity corresponding to the slice level are set. Is extracted and the contour is compared with the desired pattern shape.

For example, a case of a 0.8 μm square hole pattern will be specifically described. As the initial mask pattern, a square pattern of 0.64 μm square, which is 80% smaller than the desired pattern, was adopted. Next, the mutual intensity was calculated under the optical condition 1 shown in Table 1 below. The optical condition 1 is merely an optical condition for calculating the mutual intensity, and may not be an optical condition that can be actually realized.

[0039]

Under optical condition 1, the wavelength of the light source is 0.
436 μm, the ratio of the size of the light source to the aperture of the projection optical system (coherence degree σ) was set to 0.01, and the aperture tangent was set to 0.7. The aperture tangent here is an amount equal to the numerical aperture (NA) when the aperture is small, and is an amount obtained by dividing the aperture radius by the image-side lens focal length of the telecentric projection optical system. When the aperture tangent is between 0.0 and 0.5, the amplitude transmittance of light is 1.0 and the aperture tangent is 0.5.
From 7 to 7, the amplitude transmissivity was set to -1.0 to give the effect of phase inversion when the aperture tangent was from 0.5 to 7.0.

Next, using the real part of the obtained mutual intensity distribution as a temporary mask pattern, the projected image on the image plane was calculated under the optical condition 2 in Table 1. Since the obtained projected image was an image in which the center of the hole was conspicuous and was different from the desired pattern, the one obtained by shifting the real part of the mutual intensity distribution by -0.66 level was again used as a temporary mask pattern and shown in Table 1. The projected image was calculated under the optical condition 2. This time, as shown in FIG. 8, the obtained projected image had a desired pattern, so the temporary mask pattern was used as the mask pattern. FIG. 6 shows the mask pattern determined as described above.
Details of FIG. 8 will be described later.

FIG. 2 is a correlation diagram showing the relationship between the level shift amount of the real part of the mutual intensity and the obtained projected image. In the figure, the horizontal axis is the level shift amount, and when the shift amount is a negative value, it means that this negative value is added to the real part of the mutual intensities of all points on the image plane. ing. The vertical axis is a qualitative coordinate axis that represents the intensity distribution of the projected image.

From FIG. 2, the optimum projected image is obtained when the level shift amount is −0.66, and the level shift amount is −0.
It can be seen that when the value is smaller than 66, the projected image of the center of the hole becomes conspicuous, and when it is larger than -0.66, the projected image of the periphery of the hole becomes conspicuous. Further, there is a point that the optimum shape is obtained when the level shift amount is on the positive side of -0.66, but in this case, the goodness of image formation is unstable with respect to the change of the shift amount. Do not adopt here.

The image intensity distribution 6 obtained under the optimum level shift amount of -0.66 is a hole pattern image obtained by the mask pattern designing method of the present invention. The simulation load required in the design method of the mask pattern according to the present invention is much smaller than that in the method of optimizing this mosaic piece by using the above-mentioned minute mosaic.

Incidentally, step S in the flow chart of FIG.
In 1, the hole pattern (mask pattern) for 0.8 μm square is designed by using a hole pattern of 0.8 μm square with 80% reduction, but there is an optimum value for this reduction ratio. However, the change of the reduction ratio near the optimum value does not have a great influence on the optimum mask pattern design. And
The mask pattern used in the initial stage is a square hole pattern that has not been devised except for adjusting the reduction ratio.

The simulation images of the three hole patterns will be compared below. One is a conventional image with a square hole pattern that has not been devised for the mask pattern.
The other is an image by the trimming phase shift mask shown in FIG. 10A, and the last is an image created by the mask pattern according to the present invention. FIG. 3 shows 0.8 μm using a conventional mask.
FIG. 3A shows a square hole pattern image, and FIG.
It is a figure which shows distribution of horizontal intensity of AA 'of (b). Optical conditions are as follows: light source wavelength 0.436 μm, coherence σ =
0.2, numerical aperture NA 0.5, defocus 0.0. As is clear from the figure, the projected image obtained by the square mask pattern is circular.

FIG. 4 shows the intensity distribution (a) of the amplitude transmittance in the trimming phase shift mask and the contour distribution (b) of the planar intensity distribution showing the shape of the mask pattern. FIG. 4A shows the distribution of horizontal intensity at AA ′ in FIG. Further, FIG. 5 is an intensity distribution (a) showing a projected image by the trimming phase shift mask of FIG. 4 and a contour distribution (b) of a planar intensity distribution. FIG. 5A is FIG. 5B.
7 shows the distribution of horizontal intensity at AA ′ in FIG.

FIG. 4 shows that when the mask pattern is 0.8 μm square, 0.15 μm inside the four sides of 0.8 μm square.
An auxiliary pattern area having a width is provided, and the amplitude transmittance of light passing through this area is set to 1.0, while the amplitude transmittance of the central square portion is set to -0.2. That is, the central square portion forms a phase inversion (180 °) phase shifter for transmittance adjustment (0.2).

The projected image obtained by this mask pattern is close to a square, as shown in FIG. In FIG. 5A, the width of 30 levels of the peak of the intensity distribution is as thick as about 1.12 μm. FIG. 3 shows a 0.8 μm square hole pattern image by a conventional mask.
As shown in (a), the width of the 30% level is 0.71
It is as thin as μm.

FIG. 6 shows intensity distributions (a) and (b) of the amplitude transmittance of the mask pattern according to the present invention, and a contour distribution (c) of the planar intensity distribution showing the shape of the mask pattern. FIG. 6A shows a horizontal intensity distribution in AA ′ of FIG. 6C, and FIG. 6B shows a horizontal intensity distribution in BB ′ of FIG. 6C. As shown in FIG. 6, the phases are inverted between the central portion and the peripheral portion of the mask pattern, and the transmittances at the four corners are large. FIG. 7 is a perspective view showing the amplitude transmittance by this mask pattern.

FIG. 8 shows an intensity distribution (a) of amplitude transmittance showing a projected image formed by the mask pattern of FIG. 6 and a contour distribution (b) of a planar intensity distribution showing the shape of the mask pattern. FIG. 8A shows the distribution of horizontal intensity at AA ′ in FIG. 8B. Compared with FIG. 5B, an image having a more square shape is obtained in FIG. 8B. This is because the mask pattern according to the present invention is a more optimized version of the trimming phase shift mask. Moreover, this can be realized by a simple simulation procedure as described above.

As for the defocus characteristic, the phase shift mask as shown in FIG. 10 and the mask as shown in FIG. 12 (a) are characterized in that the depth of focus is increased. It is considered that these are optimized, and the feature of increasing the depth of focus is maintained at the same time.
As shown in FIG. 8 (a), small sub-peaks appear on both sides of the main peak, but this is not a problem at all.

[0053]

As described above, according to the present invention, the projected optical image of a hole figure having a simple shape is calculated under special optical conditions, and the level of the real part of the mutual intensity distribution obtained thereby is calculated. The design value of the optimum mask pattern is obtained by performing the adjustment. Therefore, the calculation cost required for the calculation can be remarkably reduced, and the projected image formed by the designed mask pattern has an excellent feature that the specified peak is small and the hole shape of the mask pattern is retained. .

[Brief description of drawings]

FIG. 1 is a flowchart showing a design procedure of a hole pattern forming mask according to the present invention.

FIG. 2 is a correlation diagram showing the relationship between the level shift amount of the real part of mutual intensity and the obtained projection image.

FIG. 3 is a 0.8 μm square hole pattern image by a conventional mask, (a) is a distribution diagram showing an intensity distribution,
(B) is a contour map showing the intensity distribution as seen in a plane.

FIG. 4 shows an intensity distribution (a) of amplitude transmittance in a trimming phase shift mask and a contour distribution (b) of a planar intensity distribution showing a shape of a mask pattern.

5 is an intensity distribution (a) showing a projected image by the trimming phase shift mask of FIG. 4 and a contour distribution (b) of a planar intensity distribution.

6A and 6B are intensity distributions (a) and (b) of the amplitude transmittance of the mask pattern according to the present invention, and a contour distribution (c) of a planar intensity distribution showing the shape of the mask pattern.

7 is a perspective view showing an amplitude transmittance by the mask pattern of FIG.

8A and 8B are an intensity distribution (a) of amplitude transmittance showing a projected image formed by the mask pattern of FIG. 6 and a contour distribution (b) of a planar intensity distribution showing the shape of the mask pattern.

FIG. 9 is a perspective view showing a schematic configuration of a projection exposure apparatus.

FIG. 10 is a plan view showing a phase shift mask for forming a hole pattern.

FIG. 11 is a plan view (a) showing a mask pattern by a super flex method with contour lines of amplitude transmittance and a distribution chart (b) showing its transmittance distribution.

12A and 12B are plan views for explaining a Bessel beam mask, in which FIG. 12A shows the mask and FIG. 12B shows a state after Fourier transform.

[Explanation of symbols]

91 ... Light source, 92 ... Irradiation lens, 93 ... Photomask,
94 ... First projection lens, 95 ... Second projection lens, 96 ...
Pupil, 97 ... wafer.

Claims (3)

[Claims] 1. A method of designing a mask pattern for designing a mask pattern on an original substrate for use in a projection exposure apparatus, comprising: a step of calculating a mutual intensity distribution on a substrate to be exposed; and a real part of the mutual intensity distribution. Calculating a projected image on the substrate to be exposed as a temporary mask pattern. 2. The method for designing a mask pattern according to claim 1, wherein the step of comparing the projected image with the pattern to be formed, and the step of comparing the mutual intensity distribution when the projected image is different from the pattern to be formed. And a step of calculating a projected image on the substrate to be exposed by newly using a real mask level-shifted as a temporary mask pattern. 3. The method of designing a mask pattern according to claim 1 or 2, wherein a pupil diameter is calculated to be larger than that when the mask pattern is actually used, and phase inversion occurs at a portion near the outer circumference of the pupil. A method of designing a mask pattern, which comprises setting optical conditions for the calculation and calculating the mutual intensity distribution.