JPH10115932A - Exposure method using phase shift mask - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent decrease in exposure characteristics due to mask errors and to enable accurate transfer by using an iris filter to cut the zero-order light of exposure light which transmits a mask including a Levenson type phase mask pattern. SOLUTION: When a mask assembled in an exposure mask is formed with mask errors in the production process of the mask, the intensity distribution of transmitted light on an iris plane is represented by the sum of the intensity distribution of light transmitted through an ideal mask which is ideally produced without errors and the intensity distribution of light transmitted through a normal mask having no phase shifter and a pattern with a twice pitch as that of the ideal mask. Therefore, in the exposure process of a Levenson-type phase shift mask 3a, an iris filter 10 which cuts a part of the iris plane is used. Namely, a light-shielding member is arranged in the light-shielding part 5a in the center of the iris plane where the zero-order component appears, while the surrounding part of the light-shielding part 5a consists of a light-transmitting material. By applying the iris filter 10 for the mask 3a with errors, the zero- order light as the error component can be completely cut.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention suppresses a mask error generated in mask manufacturing in a process of manufacturing a semiconductor integrated circuit device, in exposure using a mask including a Levenson-type phase shift mask pattern for forming a fine pattern. And an exposure method for performing accurate exposure (transfer).

[0002]

2. Description of the Related Art As semiconductor devices become more highly integrated, circuit patterns formed on semiconductor substrates are rapidly becoming finer. In particular, one of the key technologies for high integration is photolithography technology, which is widely known as a basic technology in pattern formation.

[0003] The photolithography technique is a technique in which a pattern drawn on a mask is transferred to a photoresist applied on a semiconductor substrate, and the underlying film to be etched is patterned using the transferred photoresist. is there.

A development process is performed after the transfer to the photoresist. A photoresist in which exposed portions remain as a pattern after development is referred to as a negative type, and a photoresist in which an unexposed portion remains after development is referred to as a positive type. . Next, a conventional exposure method in the photolithography technique will be described.

FIG. 14 is a schematic view of a conventional exposure apparatus. In the drawing, reference numeral 101 denotes a light source (here, a light source refers to a secondary light source obtained by shaping light emitted from a light emitting source through a predetermined stop. ), 102 is a condenser lens, 103 is a mask on which a pattern of an arbitrary shape is drawn, 104 is a reduction projection lens, 105 is a pupil plane formed in the reduction projection lens 104, and 106 is a surface coated with a photoresist. Semiconductor substrates are shown.

An exposure method using the exposure apparatus shown in FIG. 14 will be described below. First, the exposure light emitted from the light source 101 is applied to the mask 1 via the condenser lens 102.
03 is irradiated. Next, the exposure light transmitted through the mask 103 enters the reduction projection lens 104, is shaped through a predetermined lens or the like, is reduced to a predetermined magnification through the pupil plane 105, and is exposed on the semiconductor substrate 106. The photoresist is exposed.

In the case of forming a pattern having a line width of about 0.5 μm, the light source 101 is generally an i-line (wavelength 365 nm) emitted from a high-pressure mercury lamp. Resolution limit dimension R and depth of focus DO of exposure optical system
F is generally represented by the following equation. R = k ₁ λ / (NA) DOF = k ₂ λ / (NA) ^{2 Further} , the symbol λ used in the above equation is the wavelength of the exposure light, N
A is the numerical aperture of the projection lens, and k ₁ and k ₂ are process-dependent constants. Generally, k ₁ = 0.6 and k ₂ = 0.5.

As can be seen from the above equation, if the resolution limit dimension R is to be improved, the values of K ₁ and λ are reduced and the NA
However, it is difficult to increase the wavelength of the light source 101 and increase the numerical aperture of the light source 101, whereby the depth of focus DOF of light becomes shallow, which may lead to a reduction in resolution. was there.

Therefore, instead of improving the resolution limit dimension R by improving the light source and the lens, it has been considered to improve the resolution by improving the mask structure. In recent years, a phase shift mask has attracted attention, and a Levenson-type phase shift mask, which is a type of phase shift mask, will be described below in comparison with a normal mask (binary mask).

FIGS. 15A, 15B, and 15C show the cross-sectional structure of the normal mask 103a, the electric field intensity on the mask 3a when the mask 3a is used, and the light intensity on the semiconductor substrate 1, respectively. ing. The mask 103 shown in FIG.
a is formed from a glass substrate 107 and a metal mask pattern 108 attached to the lower surface thereof. The electric field on the mask 103a is divided into a region where the metal mask pattern 108 is formed and a region where the metal mask pattern 108 is not formed. Are spatially pulse-modulated.

However, as shown in FIG. 15 (c), when the pattern becomes finer, the light intensity on the wafer 103a becomes unexposed in the non-exposed area where the metal mask pattern 108 is formed due to the light diffraction effect. The light wraps around, and the difference between the exposed area and the non-exposed area cannot be added. For this reason, the resolution is reduced, and it is difficult to transfer a fine pattern.

In the case of exposure using the above-described ordinary mask (binary mask) 103a, ± first-order diffracted light diffracted at a diffraction angle θ on the mask (hereinafter abbreviated as ± first-order light). And the 0th-order diffracted light (hereinafter, abbreviated as the 0th-order light), which is a non-diffracted light, is used. When defocusing is performed, an image is obtained as the focal point is shifted from the focal plane. It was blurred, making accurate exposure difficult.

Next, a description will be given of a Levenson type phase shift mask which is a kind of the phase shift mask. FIG.
(A), (b), and (c) show a cross-sectional view of the Levenson-type phase shift mask 103b and an electric field intensity distribution and a light intensity distribution on the phase shift mask 103b, respectively. The Levenson-type phase shift mask 103b shown in FIG. 16A is obtained by additionally forming a phase shifter 109 on the normal mask shown in FIG. 15A, and a pattern in which the phase shifter 109 is attached on the mask. And the unattached pattern are alternately arranged.

The phase shifter 109 includes the phase shifter 10
9 is to convert the phase of the light transmitted through 9 by 180 degrees.
In addition, the electric field on the Levenson-type phase shift mask 103b has an opposite phase in the adjacent exposure region, and the light cancels each other in a portion where light overlaps (between patterns) due to a light interference effect. Becomes Therefore, as shown in FIG. 16C, the light intensity distribution on the Levenson-type phase shift mask 103b accurately shows the light intensity peak corresponding to each pattern, and the light intensity difference between the exposed region and the non-exposed region. Can be sufficiently secured, resolution can be improved, and a fine pattern can be transferred.

In the case of the above-mentioned normal mask 103a, 0
The image was formed by three light beams, the next light and the ± 1st light,
In the case of this Levenson-type phase shift mask 103b,
The light transmitted through the mask is diffracted at an angle smaller than the diffraction angle θ of a normal mask, and an image is formed by two light beams of ± first-order light. The zero-order light, which is undiffracted light, does not appear theoretically and need not be considered here if the Levenson-type phase shift mask is accurately formed. as a result,
Even if the obtained image is out of focus in the height direction with respect to the focal plane, the image does not change, and there is an effect that accurate exposure is possible. For this reason, it can be understood that the importance of the Levenson-type phase shift mask is increasing with the miniaturization of the element.

However, in the case where exposure is performed using a Levenson-type phase shift mask, it is considered that ultra-fine processing can be performed in principle. There is a problem.

One of the mask errors of the Levenson-type phase shift mask is a phase error in which the phase difference of light transmitted through adjacent aperture patterns does not accurately become 180 degrees,
The other is a transmitted light intensity error that the transmitted light intensity of adjacent opening patterns is not exactly equal. Due to these mask errors, when transferred to the resist applied on the semiconductor substrate 106, the dimensions of the pattern passing through the phase shifter and the pattern not passing through the normal phase shifter on the mask are ideally There is a problem that accurate transfer cannot be performed, for example, where exactly the same should be achieved is not the same.

Such deterioration of pattern formation accuracy due to phase error and transmitted light intensity error becomes more serious as the elements become finer. In addition, JJAP Vol.33, PP6816
~ 6822, Proc, SPIE Vol.1674, P264, JJAPVol.34, PP
6578-6683, Proc, SPIE Vol.1927, P28- and the like.

[0019]

As described above, there has been a problem that accurate exposure may not be performed even in fine processing using a Levenson-type phase shift mask due to a mask error. According to the present invention, in the exposure using a Levenson-type phase shift mask, even if there is a mask error generated during the fabrication of the mask, the shape of the pattern finally formed on the semiconductor substrate can be obtained as an ideal one. The purpose is to do so.

[0020]

According to an exposure method of the present invention, a first step of irradiating an exposure light from a light source, a pattern of an arbitrary shape is drawn on the exposure light shaped through a predetermined optical system. A second step of irradiating the illuminated mask with the exposure light transmitted through the mask to a pupil filter on a pupil plane, and blocking the 0th-order light, which is undiffracted light, of the exposure light with the pupil filter A third step, including a fourth step of reducing the exposure light transmitted through the pupil filter to an arbitrary size and irradiating the object to be exposed, wherein the mask includes at least a Levenson-type phase shift mask pattern It is.

Further, in the exposure method according to the present invention, in the third step, the light-shielding portion of the pupil filter used is formed so as to completely shield the zero-order light.

Further, in the exposure method according to the present invention, the light-shielding portion of the pupil filter used in the third step has a size of 0
It is formed such that a part of the next light is transmitted and other light is shielded.

Further, in the exposure method according to the present invention, the light-shielding portion of the pupil filter used in the third step is made of a semi-transmissive substance to reduce the light intensity of the zero-order light to an arbitrary magnitude. It is.

Further, in the exposure method according to the present invention, in the first step of irradiating exposure light from a light source, the exposure light formed through a predetermined optical system is applied to a mask on which a pattern of an arbitrary shape is drawn. A second step of irradiating, a third step of irradiating the pupil filter on the pupil surface with the exposure light transmitted through the mask, and blocking the zero-order light, which is undiffracted light, of the exposure light with the pupil filter A fourth step of reducing the exposure light transmitted through the pupil filter to an arbitrary size and irradiating the exposure target with the object, wherein the mask is a Levenson-type phase shift mask formed only with a Levenson-type phase shift pattern It is characterized by being.

Further, in the exposure method according to the present invention, in the first step of irradiating exposure light from a light source, the exposure light formed through a predetermined optical system is applied to a mask on which a pattern of an arbitrary shape is drawn. A second step of irradiating, a third step of irradiating the pupil filter on the pupil surface with the exposure light transmitted through the mask, and blocking the zero-order light, which is undiffracted light, of the exposure light with the pupil filter A fourth step of reducing the exposure light transmitted through the pupil filter to an arbitrary size and irradiating the exposure object with the mask, wherein the mask includes a Levenson-type phase shift mask pattern and a normal binary mask pattern. It is a structure that includes.

Further, in the exposure method according to the present invention, in the above-mentioned third step, the size of the pupil diameter (NA) that determines the non-transmissive area of the pupil plane is determined by the following equation: NA ≦ λ / (L (1 + σ)) (Λ is the wavelength of the exposure light, L is the pitch of the pattern to be exposed,
is adjusted so as to satisfy the coherence factor of the exposure light, and the pupil filter blocks the 0th-order light and simultaneously blocks the exposure light irradiated outside the pupil diameter.

[0027]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1 FIG. First, the principle of Embodiment 1 of the present invention will be described. FIG. 1 is a schematic diagram of a reduced projection exposure apparatus (hereinafter simply referred to as an exposure apparatus) used for manufacturing a semiconductor device. In the figure, reference numeral 1 denotes a wavelength λ = 248n.
m using a KrF excimer laser (where,
The light source is a secondary light source formed by shaping light emitted from a light emitting source through a predetermined stop. 2) a condenser lens which receives exposure light emitted from the light source 1 and shapes the light through a lens or the like; 3 denotes a fine pattern of 0.3 μm rule or less which is irradiated with light through the condenser lens 2. Is a mask, 4 is a reduction projection lens on which exposure light is incident via the mask 3, 5 is a pupil plane formed in the reduction projection lens 4, and 6 is a semiconductor substrate having a resist film formed on the surface. ing. Further, on the pupil plane 5, a light-shielding portion 5a that shields zero-order light, which is non-diffracted light transmitted through the mask 3, is formed.

The exposure light emitted from the light source 1 is formed at a predetermined magnification through a mask 3 and the like.
The resist is applied to one main surface of the semiconductor substrate 6 to expose the resist.

When the mask 3 incorporated in the exposure apparatus of FIG. 1 is formed with a mask error in the process of manufacturing the mask, the intensity distribution of the transmitted light appearing on the pupil plane 5 is ideally formed without error. It is expressed as the sum of the idealized "ideal mask" transmitted light intensity distribution and the transmitted light intensity distribution of a normal mask having a pattern of twice the pitch of the ideal mask and having no phase shifter. FIG. 2 is a conceptual diagram showing in principle that an image substantially similar to that of an ideal mask can be formed by applying a pupil filter that shields a part of the pupil plane during exposure of the mask.

The upper part of FIG. 2 shows the mask used for exposure, the middle part shows the light intensity on the pupil plane corresponding to each mask, the lower part on the left shows the light intensity on the pupil plane after error correction, and the lower part on the right shows the error. 3 shows a pupil filter 10 used for correction. The upper mask in FIG. 2 is not a Levenson-type phase shift mask 3a (hereinafter, abbreviated as error mask 3a) having a transmitted light intensity error, an ideal mask 3b having no error, and a Levenson-type phase shift mask from the left. A double pitch normal mask (binary mask) 3c without a shifter is shown.

These masks are formed by forming a metal mask pattern (chrome pattern) 8 on a glass substrate 7 as in the conventional case. Further, the Levenson-type phase shift mask pattern further has a phase shifter 9 formed at every other adjacent mask opening. The pitch of the opening pattern formed by the metal mask pattern 8 on the error mask 3a and the ideal mask 3b is the same,
The normal mask includes an error mask 3a and an ideal mask 3b.
The metal mask pattern 8 having openings formed at double pitches is formed.

The lower part of the upper broken line in FIG. 2 relatively shows the transmitted light intensity at the openings of the respective masks. Assuming that the transmitted light intensity of the opening of the ideal mask 3b is 1.0, the transmitted light intensity of the opening of the error mask 3a having the transmitted light intensity error of 10% is the transmitted light intensity of the opening via the phase shifter 9. Can be expressed as 1.0, and the transmitted light intensity of the opening not passing through the phase shifter 9 can be expressed as 1.1. The difference between the ideal mask 3b and the error mask 3a causes the error mask 3
It can be seen that the transmitted light intensity distribution a is obtained by adding the transmitted light intensity distribution of the normal mask 3c when the transmitted light intensity is 0.1 to the transmitted light intensity distribution of the ideal mask 3b.

Due to the linearity of the exposure system, the error mask 3
The image by a is given as the sum of the image by the ideal mask 3b and the image by the normal mask 3c having a double pitch. This is shown in the middle part of FIG. 2 for the light intensity distribution (Fourier spectrum) on the pupil plane. FIG.
As is well known, the light intensity distribution on the pupil plane by the ideal mask 3b shown in the center of the middle stage is composed of only ± 1st-order light without 0th-order light. Also, the light intensity distribution on the pupil plane by the normal mask 3c having a pattern of twice the pitch of the ideal mask 3b is shown in the middle right of FIG. 2, and as is well known, the zero-order light and ± It is composed of primary light. Since the light intensity distribution on the pupil plane of the error mask 3a is the sum of these,
As shown on the left of the middle part of FIG. 2, it is composed of ± 1st-order light which is an ideal mask component and has a relatively large intensity and 0th-order light which is an error component having a relatively small intensity. become.

In the light intensity distribution on the pupil plane of the normal mask 3c having a pattern of twice the pitch of the ideal mask 3b, if the influence of the relatively small ± first-order light is negligible, the pupil plane of the error mask 3a is ignored. The light intensity distribution is approximately represented as the sum of the light intensity distribution on the pupil plane by the ideal mask 3b and the zero-order light in the light intensity distribution by the normal mask 3c. From this, it is concluded that the deterioration of the imaging characteristics due to the error mask 3a is due to the 0th-order light in the light intensity distribution of the normal mask 3c having a pattern of twice the pitch of the ideal mask 3b. In the exposure of the error mask 3a, the light intensity on the pupil plane can be reduced by blocking the 0th-order light in the light intensity distribution on the pupil plane by the normal mask 3c having a pattern of twice the pitch of the ideal mask 3a. It is considered that the distribution can be the same light intensity distribution on the pupil plane as that obtained by the ideal mask 3b, and it is possible to avoid the deterioration of the imaging characteristics.

Next, the pupil filter 10 that blocks the zero-order light component that appears on the pupil plane, which is a problem in the exposure using the error mask 3a, will be described. The pupil filter 10 that blocks the 0th-order light component is shown on the lower right side in FIG. The pupil filter 10 includes a light-shielding portion 5a at the center of the pupil plane where the zero-order light component appears.
A light-shielding member is disposed on the light-shielding portion 5a, and a portion around the light-shielding portion 5a is formed of a transparent material.

When the pupil filter 10 acts on the error mask 3a, all the zero-order light, which is an error component, can be blocked.
It can be seen that even when the error mask 3a is used, an ideal image having a light intensity distribution of only the ± first-order light components as shown in the lower left portion of FIG. 2 can be obtained.

The mask having the transmitted light intensity error has been described above. However, when a Levenson type phase shift mask (error mask) having a phase error due to the formation of the shifter is used, the image appearing on the pupil plane is also changed. As in the case of the light intensity error, it can be considered by analyzing the diffracted light flux, and the pitch L of the ideal mask (meaning the distance from the center of the pattern formed on the mask to the center of the adjacent pattern) is 1.
It is possible to consider that a normal mask of / 2 (however, the phase difference is 90 degrees) is added, and the same effect of reducing the mask error by the pupil filter 10 is expected.

Next, the result of obtaining the improvement of the optical image when the pupil filter 10 is operated by the optical image calculation will be described. The coefficients used for this calculation are numerical aperture NA = 0.55,
Illumination coherence factor σ = 0.2, illumination wavelength λ
= 248 nm.

FIG. 3 shows a transferred image by a mask having a phase error (10 deg) in a 0.2 μm 1: 1 L / S pattern. The upper, middle, and lower portions of FIG. Is ΔF (focus shift amount) =
When it is shifted to 1.0 μm, the focal position ΔF = 0 μm, and when there is no defocus, the focal position ΔF = −1.0 μm.
The left column in the figure shows the optical image without the pupil filter 10, and the right column shows the optical image with the pupil filter 10.

In the case where the focal position ΔF = 0 μm and there is no deviation and the ideal case (the middle part in FIG. 3), the obtained optical image is the same and ideal regardless of the presence or absence of the pupil filter 10. The focus shift amount ΔF = ± 1.0 μm
(Upper and lower parts of FIG. 3), an optical image passing through the pupil filter 10 can obtain an ideal optical image having a constant period and a constant amplitude, which is almost the same as the case of ΔF = 0 μm. Can be

However, without the pupil filter 10, ΔF
It can be seen that if there is a deviation of the focal position of = ± 1.0 μm, the amplitude of the optical image varies greatly due to the influence of the phase error, and the light intensity of the adjacent bright part greatly differs. That is, by applying the pupil filter 10, a great improvement in image quality is recognized.

FIG. 4 shows data in which the pattern dimension (CD) is a function of the focus shift amount (ΔF) obtained by the slice level method, that is, a method in which an optical image width having a certain light intensity or more is used as a pattern width. . The pattern dimension (CD) is 0.2 μm in a normal case as in the case described above. FIG.
4A shows the CD-Focus characteristic when the mask has a phase error (10 deg), and FIG. 4B shows the CD-Focus characteristic when the mask has a transmitted light intensity error (10%). .

In FIG. 4, ○ and □ indicate data when the pupil filter is actuated for exposure, and 、 and black □ indicate data when the pupil filter is not actuated. Among them, the large marks indicate the case where the phase shifter is formed in the pattern (opening) formed in the mask, and the small marks indicate the case where the phase shifter is not formed.
The marks indicate the dimensions of the bright part of the formed optical image, and the marks □ and black indicate the dimensions of the dark part, respectively.

In any of the characteristic diagrams shown in FIGS. 4A and 4B, the marks ●, □, and black □ (that is, the mark indicating the size of the filter and the size of the dark portion) have a defocus. 0 ± 1.
A CD difference appears in the range of 0 μm, a CD difference of about 0.04 μm at the maximum appears due to defocus, and a pattern in the case of ΔF = 0 μm without defocus (that is, exposure using an ideal mask). To get closer to the dimensions,
It can be seen that it is effective to operate the pupil filter 10 in any case where there is a light intensity error. FIG.
In (a) and (b), the □ mark and the black □ mark indicating the size of the dark portion are in an overlapping state, and indicate almost the same value.

FIG. 4 shows an example of 1: 1 L / S of 0.2 μm. Next, FIG. 5 shows a case where the L / S pattern dimension is changed from 0.16 to 0.24 μm. A plot (dependence of a dimensional difference on a focal position) of a mask error, a phase error (10 deg) and a transmitted light intensity error (10%) is shown. In FIG. 5, the left column shows the pattern dimensions without a mask error, the center column shows the pattern dimensions with a phase error, and the right column shows the pattern dimensions with a light intensity error.

L / S phase of 0.16-0.24 μm,
In the plot where there is a light intensity error, when the pupil filter is actuated, the plot approximates the ideal plot in the left column, the transfer error is reduced, and the pupil filter 10 FIG. 5 shows that there is no change in the effect.

When a pupil filter is applied to an ideal mask, it has no effect at all, and it goes without saying that normal transfer is possible. From the above, at the time of exposure using the error mask, the pupil filter that blocks the 0th-order light of the light transmitted through the mask is operated,
It is clear that the accuracy of the exposure can be greatly improved.

Based on the above, when actually performing exposure using the exposure apparatus shown in FIG. 1, a pupil filter 10 is provided on the pupil plane 5, and an area for blocking the 0th-order light transmitted through the mask 3. By forming the light-shielding portion 5a in advance, it is possible to perform almost normal transfer without using a mask 3 (error mask) having a phase and light intensity error in the mask manufacturing stage. is there.

As described above, the pupil filter 10 is operated,
By removing the error component, the effect as a Levenson-type phase shift mask, that is, it is possible to accurately perform fine processing with an ideal rule of 0.3 μm or less. Also, when the exposure method using the slice level method is adopted, there is no problem as long as the light intensity is equal to or less than a predetermined value, even if the zero-order light, which is an error component, is not completely blocked. In contrast to 10, it is possible to shield most of the area where the zero-order light can be irradiated and to transmit part of the area, and to make the light-shielding portion 5a half-tone (semi-transmission). By processing the pupil filter 10 so that the influence of the zero-order light component can be reduced, more ideal exposure can be performed than in the case where the transfer is performed without passing through the pupil filter 10 at all. .

Embodiment 2 The first embodiment is effective when transferring a very fine pattern using a Levenson-type phase shift mask, such as a pattern having a design rule of about 0.2 μm, that is, all or zero-order light transmitted through the mask is transferred. The method of performing ideal exposure by removing most of the light that is an error component by shielding most of the light is described.

Next, in the second embodiment, an ultra-fine 0.2 μm 1: 1 L / S Levenson type pattern as described in the first embodiment and a contact or pad are used. Exposure method in which a large pitch pattern formed by opening chromium etc. is mixedly mounted on a single mask, such as a normal mask type, etc. explain.

When a normal binary mask 3c (having a pattern formed of metal openings such as chromium) which is not of the Levenson type is used, as shown in the upper right side of FIG. 2, the light passes through the mask 3c. The light is formed with a luminous flux of 0-order light and ± 1st-order light. Therefore, Embodiment 1
Pupil filter 1 that blocks 0-order light as shown in
When 0 is applied, it is obvious that the 0th-order light is removed and normal imaging becomes impossible.

Originally, if the zero-order light is excluded from the optical image formed by the three-beam interference of the light beams of the zero-order light and the ± first-order light, the optical image will be formed at twice the pitch of the mask pitch. For this reason, when the Levenson-type phase shift mask pattern and the normal mask pattern (binary mask pattern having no phase) are mixedly mounted on one mask, a pupil filter that blocks the zero-order light is used. It is understood that adverse effects are inevitable.

However, as a compromise, it is possible to form an image of each pattern in a state close to normal by using a mask on which the ultra-fine pattern and the large pitch pattern are mixedly mounted. That is, the 0th-order light of the light flux transmitted through the mask is not completely blocked,
By using a method of blocking a part of the next light, a large pitch pattern 0 formed in the Levenson type phase shift mask pattern is used.
The near-normal exposure is performed by reducing the next light component and securing the light intensity required for normal imaging of the zero-order light component of the large pitch pattern formed in the normal mask pattern.

As a method of blocking a part of the zero-order light component, a method of reducing the filter diameter of the pupil filter 10 will be described, which shows that exposure in a state close to normal can be performed. FIG. 6 shows a calculated optical image of a 0.60 μm L / S normal mask (binary mask). The optical conditions are numerical aperture NA = 0.55, light source coherence factor σ =
0.2, and the wavelength of illumination λ = 248 nm.

FIGS. 6A to 6F show optical images obtained by changing the filter diameter (wave number).
In FIG. 7, the ratio between the filter diameter and the non-diffraction light source image diameter corresponding to FIGS. 6A to 6F is 1.0, and since the 0th-order light is completely shielded, it is normal. For example, a peak of an optical image appearing every 1.2 μm appears every 0.6 μm which is a half cycle of the mask pattern, and it can be seen that normal exposure cannot be performed.

6 (b), 6 (c)... (F),
Filter diameters of 0.08 / λ, 0.06 / λ, 0.04 /
λ, 0.02 / λ, 0, and the ratio of the non-diffractive optical image diameter to 0.73, 0.55, 0.36, 0.18, 0,
By gradually reducing the light shielding ratio of the zero-order light, an image having an original normal cycle, that is, an image close to an ideal optical image that can be formed by a normal mask pattern is obtained.

In particular, when the filter diameter is set to 0.08 / λ or less and the ratio to the non-diffractive optical image is set to 0.73 or less, the slice level can be sufficiently increased when the zero-order light is not completely blocked. It can be said that imaging by the method is possible. on the other hand,
For an image formed by a Levenson-type phase shift mask pattern having a mask manufacturing error, a pupil filter 10 is used.
And normal image formation is possible by removing the zero-order light component. In the case where there is no filtering effect, for example, as shown in FIG. It can be said that normal imaging cannot be expected.

FIGS. 8A to 8F show L of 0.16 μm.
Is the result of calculating an optical image of a Levenson-type phase shift mask on which an ultrafine pattern of / S is formed, and the mask error is a transmittance difference at the opening of the mask (a transmittance between 0 and 180 degrees). The difference is set to 10%, and the filter diameter is calculated as a parameter under the same conditions as in FIG. The table in FIG. 7 shows the ratio between the filter diameter and the non-diffractive optical image diameter in FIGS. 8 (a) to 8 (f).

In FIG. 8A, since the pupil filter completely blocks the zero-order light, which is the main error component, there is almost no change in the maximum value of the adjacent peaks, and the ideal image formation is achieved. Becomes possible. In addition, as shown in FIGS. 8B to 8F, the influence of the 0th-order light component gradually increases, and the difference between adjacent peak values tends to increase.

Considering all the above, for example,
L / S pattern (normal pattern) of about 0.6 μm which is a large pitch pattern and 0.16 which is an ultra fine pattern
At the time of exposure using a mask in which an L / S pattern (Levenson type phase shift pattern) of about μm is mounted on one mask, a light source is placed at the position of the pupil plane 5 of the exposure apparatus shown in FIG. Optimal exposure can be performed by attaching a pupil filter having a light-shielding portion whose area ratio to the non-diffracted image diameter of 1 (the ratio to the diameter of the image by the 0th-order light) is 0.73 times, and performing exposure. In other words, optimal exposure is actually possible.

The setting of the light-shielding portion of the pupil filter so that the area ratio with respect to the specific diffraction image diameter of the light source is 0.73 times is merely an example, and the pattern to be exposed is not limited. It goes without saying that the optimum value differs depending on the size and various exposure conditions.

That is, the light-shielding portion of the pupil filter 10 is set so as to shield a part of the zero-order light component of the transmitted light through the mask, and the pupil filter 10 is operated to perform exposure. Even when a mask in which a mask pattern and a Levenson-type phase shift mask pattern coexist is used, an image in a state close to normal can be obtained. By using this method, a plurality of different dimensions can be obtained in one exposure step. There is an effect that the pattern can be exposed with high accuracy.

Embodiment 3 In the second embodiment, a Levenson-type phase shift mask pattern and a normal binary mask pattern are mixedly mounted on a single mask used for exposure. A technique was described in which part of the light was blocked by adjusting the size of the filter diameter of the pupil filter, the error component of the Levenson-type phase shift mask pattern was reduced, and the normal mask pattern could be transferred with accurate dimensions. .

In the third embodiment, as in the case of the second embodiment, any one of the Levenson-type phase shift mask patterns and the ordinary binary mask pattern is mixed and mounted on one mask. A technique capable of transferring the image to a size close to normal will also be described.

In the second embodiment, the diameter of the light shielding portion 5 a of the pupil filter 10, that is, the filter diameter is adjusted so as to block a part of the zero-order light, and a predetermined value is determined from the center of the zero-order light component appearing on the pupil plane 5. In the third embodiment, the exposure light in the circle within the range of the distance is shielded and the 0th-order light component located on the outer periphery of the exposure light is transmitted. ) Type, and by reducing the transmittance of the zero-order light component as a whole, the error component of the Levenson-type phase shift mask pattern is reduced, and the normal mask pattern is accurately transferred.

Next, FIGS. 9 (a) to 9 (f) show a normal mask pattern having a shape of 1: 1 L / S at 0.60 μm and an optical image obtained by exposing the light while changing the transmittance of the pupil filter.
This will be described with reference to FIG. The table showing the filter transmittance of FIG.
An optical image of the Levenson-type phase shift mask pattern described next is shown in the table of FIG. 10 together with the filter transmittances of FIGS. 11 (a) to 11 (f).

In addition, the coefficients used for calculating the optical image are NA = 0.55, the coherence factor σ = 0.2 of the light source, and the wavelength λ of the illumination = 248 nm.
In the case of FIG. 9A, the transmittance of the filter is 0%. In this case, since the 0th-order light is completely blocked, a peak which should appear every 1.2 μm appears every 0.6 μm. And the error is too large to allow normal transfer. 9B to 9F, the transmittance of the light shielding portion of the filter is 20%, 40%, 60%, 80%, 100%.
%, The image gradually becomes an optical image capable of normal transfer, and its peak also appears every 1.2 μm, and it can be seen that the peak serving as an error component becomes smaller.

FIG. 11 shows the effect of the transmittance of the pupil filter for blocking the zero-order light on the Levenson-type phase shift mask pattern. As in the case of the second embodiment, this is a case where a 0.16 μm 1: 1 L / S pattern is formed as the mask pattern, and the coefficients used for the optical image calculation are the same as those of the above-described normal mask pattern. Note that FIG.
The transmittances corresponding to (a) to (f) are as shown in FIG. 7, and 0%, 20 corresponding to FIGS. 11 (a) to (f).
%, 40%, 60%, 80%, and 100%.

As described in Embodiments 1 and 2, in the case of exposure using a Levenson-type phase shift mask, ideal exposure is possible when the zero-order light component is completely shielded. As can be seen, as the transmittance is closer to 0%, there is no difference in the height of the adjacent peaks, and the optical image is closer to a normal optical image. Conversely, as the transmittance approaches 100%, the effect of a mask error such as a phase error and a transmittance error of the Levenson-type phase shift mask increases, and the optical image is disturbed.

To summarize the above, using a mask in which a Levenson-type phase shift mask pattern and a normal binary mask pattern are mixed by the slice level method, the zero-order light component of the transmitted light of the mask and When light is shielded by a semi-transmissive pupil filter, for example, by setting the transmittance of the pupil filter to 60%, it can be said that any pattern can be transferred normally, and actually good transfer is possible.

Therefore, the pupil plane 5 of the exposure apparatus shown in FIG.
Then, a pupil filter of 60% transmitted light is formed in a region irradiated with the 0th-order light, and the same exposure is performed by using the above-mentioned mixed type mask. Normal transfer can be performed.

The transmittance of the light-shielding portion 5a of the pupil filter 10 is set to 60% because the pattern formed on the mask is a normal mask pattern of 0.6 μm 1: 1 L / S and a 0.16 μm 1: 1 L / S pattern. However, it is needless to say that it is necessary to change the value of the effective transmittance when, for example, the dimension of the pattern is changed.

However, in any case, when exposure is performed using a mask on which a normal mask pattern and a Levenson-type phase shift mask pattern are mixed, zero-order light, which is transmitted light of the mask, is transmitted at a predetermined transmittance. A semi-transmissive filter is used to shield light, and the influence of the zero-order light on the transfer image is suppressed to a level that does not adversely affect the transfer of the normal mask pattern. It becomes possible.

Embodiment 4 In the above-described first to third embodiments, all or a part of the zero-order light transmitted through the mask is shielded by attaching a pupil filter to the pupil plane of the exposure apparatus, thereby reducing a mask error. .
In the fourth embodiment, the zero-order light (error component)
A description will be given of a technique for eliminating the influence of ± second order light, which is an error component during transfer using an Levenson-type phase shift mask pattern having an error, together with the reduction of the error.

FIG. 12A shows a pupil plane diffraction image obtained by ideal exposure using a normal mask, and FIG. 12B shows an ideal pupil diffraction image using a Levenson-type phase shift mask. It is a pupil plane diffraction image at the time of exposure. As shown in FIG. 12A, when a normal mask is used, the pupil plane diffraction image has ± 0 order light and ± 1 order light and ± 2 order diffracted light (hereinafter abbreviated as ± 2 order light) around the 0 order light. ... And the light intensity is the largest for the zero-order light, and the higher the order, the smaller the effect on the image formation.

In the first embodiment, the exposure using the Levenson type phase shift mask having an error is 0.16
When the pitch of the pattern to be transferred is as small as 1 μm1: 1 L / S, the numerical aperture NA, that is, the pupil diameter is adjusted so as to transmit ± first order light as shown by the broken line 11 in FIG. It is not necessary to consider a light beam of a higher order than the ± second order light, and no error was caused by the ± second order light and the like.

However, when a large pitch pattern using a Levenson-type phase shift mask is transferred, or when the pupil diameter is within a range including ± secondary light, for example, the dashed line 1 in FIG.
In the case of the range indicated by 2, ideally, it is desirable that only the ± first-order light appear as shown in FIG. 12B, but the zero-order light, ± Secondary light may enter the pupil.

The shielding of the zero-order light is performed by the pupil filter 10.
However, it is difficult to transmit the ± first-order light and to shield the ± second-order light and the higher-order light flux only by operating the pupil filter 10. When ± second order light enters the pupil, ± 1 order light and ± 2 order
A half-pitch image is formed due to interference with the next light, which causes a problem that the effect of reducing the influence of the mask error is reduced.

As a method of reducing the influence of the ± second-order light due to the error component, the size of the pupil is adjusted, and the position shown by the two-dot chain line 13 in FIG. Including
There is a method of blocking the ± second-order light by setting a range that does not include the ± second-order light and is in contact with a part of the boundary of the ± second-order light or a pupil diameter smaller than the range.

The range indicated by the two-dot chain line 13 in FIG. NA ≦ λ / (L (1 + σ)) As shown in FIG. 12A, the center position of ± primary light is ±
1 / 2L, and the center position of ± secondary light is ± 1
/ L. The size (radius) of each light beam component can be expressed as NA · σ. Also, the limit NA at which the center of the ± 1st order light enters the pupil is NA = λ /
The limit NA at which the center of the 2L, ± 2 order light enters the pupil is NA =
It is represented as λ / L. Accordingly, by subtracting the radius NA · σ of this light flux from the limit NA at which the center of the ± second order light enters the pupil, the NA in the range indicated by the two-dot chain line 13 in the figure is obtained, and the relationship represented by the above equation is obtained. Can be shown.

FIG. 13 shows an example of an optical image of a large pitch pattern of 0.40 μm 1: 1 L / S formed on a Levenson-type phase shift mask when the size of the pupil diameter (numerical aperture NA) is changed. (A) to (d). FIG. 13 (a)
Is set so that its pupil diameter NA is 0.25.
3 (b), (c) and (d) each have an NA of 0.30,
The coefficients are set to 0.35 and 0.40, and other coefficients and the like are σ = 0.2, pupil filter size = light source size, and pupil filter (light-shielding portion) transmittance is 0%.

In FIGS. 13A to 13D, in the case of FIG. 13A having the smallest pupil diameter, there is no variation in the size of the peak appearing in the optical image, and the most ideal image is obtained. It can be seen that it can be obtained. On the other hand, when the pupil diameter is as shown in FIGS.
It can be seen that as the size increases, the variation in the size of the adjacent peaks of the optical image increases.

In the above equation NA ≦ λ / (L (1 + σ)), λ =
Substituting 0.248, L = 0.8 μm, and σ = 0.2 gives a solution of NA ≦ 0.258. When compared with the calculation result of NA = 0.25 in FIG. 13A that satisfies this value, the transfer in the best condition without the adverse effect of ± secondary light is adjusted to the pupil diameter as shown in the above equation. It can be said that what is possible is supported by doing so.

Therefore, the pupil plane of the exposure apparatus shown in FIG.
A pupil filter that blocks ± 1st-order light of the mask diffracted light is formed. Further, the size of the pupil diameter is set to NA ≦ λ / (L ( 1+
σ)) by adjusting the size to satisfy FIG.
As shown in the figure, the luminous flux appearing on the pupil plane can be made only ± 1 order light,
By this image formation, it is possible to perform exposure of an ideal large pitch pattern using a Levenson type phase shift mask.

[0086]

The present invention has the following effects by using the above-described exposure method.

In the reduced projection exposure method, of the exposure light transmitted through the mask including the Levenson-type phase shift mask pattern, the 0th-order light is blocked by a pupil filter. This has the effect of suppressing deterioration in characteristics and enabling accurate transfer.

In a reduction projection exposure method using a mask on which a Levenson-type phase shift mask pattern is formed,
By completely shielding the zero-order light out of the exposure light transmitted through the mask, the adverse effect of the mask error is eliminated, and there is an effect that accurate exposure can be performed.

In reduced projection exposure using a mask on which a Levenson-type phase shift mask pattern and a normal binary mask pattern are formed, of the exposure light transmitted through the mask, the 0th-order light is partially shielded, There is an effect that any of the mask patterns can be accurately exposed.

In reduced projection exposure using a mask on which a Levenson-type phase shift mask pattern and a normal binary mask pattern are formed, of the exposure light transmitted through the mask, zero-order light is transmitted through a semi-transmissive light-shielding portion. The light intensity of the zero-order light is reduced by passing through the pupil filter having the above-mentioned structure, whereby the mask pattern of both the Levenson-type phase shift mask pattern and the normal mask pattern can be accurately exposed. There is.

In reduced projection exposure using a mask on which only an Levenson-type phase shift pattern having an error is formed, of the transmitted light of the mask, the 0th-order light is shielded to eliminate the adverse effect of the mask error, thereby achieving accurate There is an effect that it is possible to perform a proper exposure.

In reduction projection exposure using a mask having an error, on which a Levenson-type phase shift pattern and a normal binary mask pattern are formed, the 0th-order light of the transmitted light of this mask is partially blocked. This has the effect of eliminating the adverse effects of mask errors and enabling accurate exposure.

In a reduced projection exposure using a Levenson-type phase shift mask, even when a mask having an error is used, the size of the pupil is set to ± 2 of the mask diffraction light.
By performing exposure while adjusting so as to block the next light beam and a higher-order light beam, an adverse effect of a mask error can be suppressed, and accurate exposure can be performed.

[Brief description of the drawings]

FIG. 1 is a view of an exposure apparatus showing a first embodiment of the present invention.

FIG. 2 is a view of a mask according to the first embodiment of the present invention.

FIG. 3 is a diagram showing characteristics of the first embodiment of the present invention.

FIG. 4 is a diagram showing characteristics of the first embodiment of the present invention.

FIG. 5 is a diagram showing characteristics of the first embodiment of the present invention.

FIG. 6 is a diagram showing characteristics of the second embodiment of the present invention.

FIG. 7 is a diagram showing variables according to the second embodiment of the present invention.

FIG. 8 is a diagram showing characteristics of the second embodiment of the present invention.

FIG. 9 is a diagram showing characteristics according to the third embodiment of the present invention.

FIG. 10 is a diagram showing variables according to Embodiment 3 of the present invention.

FIG. 11 is a diagram showing characteristics according to the third embodiment of the present invention.

FIG. 12 is a diagram showing characteristics of the fourth embodiment of the present invention.

FIG. 13 is a diagram showing characteristics of the fourth embodiment of the present invention.

FIG. 14 is a diagram showing a conventional technique.

FIG. 15 is a diagram showing a conventional technique.

FIG. 16 is a diagram showing a conventional technique.

[Explanation of symbols]

 1. Light source 2. 2. Condenser lens Mask 3a. Error mask 3b. Ideal mask 3c. 3. Normal mask 4. Reduction projection lens Pupil plane 6. Semiconductor substrate 7. Glass substrate 8. Metal mask pattern 9. Phase shifter 10. Pupil filter 11. Broken line 12. Dash-dot line 13. Two-dot chain line

Claims (7)

[Claims] In an exposure method using an exposure apparatus, a first step of irradiating exposure light from a light source, a pattern of an arbitrary shape is drawn on the exposure light shaped through a predetermined optical system. A second step of irradiating the mask with the exposure light transmitted through the mask, and irradiating the pupil filter on the pupil surface with the 0th-order diffracted light, which is undiffracted light, of the exposure light by the pupil filter Step, a fourth step of reducing the exposure light transmitted through the pupil filter to an arbitrary size and irradiating the object to be exposed, wherein the mask includes at least a Levenson-type phase shift mask pattern. Exposure method using a phase shift mask. 2. The exposure method using a phase shift mask according to claim 1, wherein, in the third step, the light shielding portion of the pupil filter used is formed so as to completely shield the zero-order diffracted light. 3. The phase shift mask according to claim 1, wherein the light-shielding portion of the pupil filter used in the third step is formed so as to transmit a part of the zero-order diffracted light and shield the other light. Exposure method using 4. The light-shielding portion of the pupil filter used in the third step is made of a semi-transmissive substance, and reduces the light intensity of the 0th-order diffracted light to an arbitrary magnitude.
An exposure method using the phase shift mask described in the above. 5. The exposure method using a phase shift mask according to claim 1, wherein the mask is a Levenson type phase shift mask formed only with a Levenson type phase shift pattern. . 6. The phase shift mask according to claim 1, wherein the mask has a structure including a Levenson type phase shift mask pattern and a normal binary mask pattern. Exposure method using 7. In a third step, the size of a pupil diameter (NA) that determines a non-transmissive area of a pupil plane is set to NA ≦ λ / (L (1
+ Σ)), where λ is the wavelength of the exposure light, L is the pitch of the pattern to be exposed, and σ is the coherence factor of the exposure light). An exposure method using a phase shift mask according to any one of claims 1 to 4, wherein the exposure light irradiated outside is shielded.