JPH10268503A - Production of semiconductor device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the accuracy and uniformity of resist patterns by setting the phase shift quantity between phase and non-phase shift transmissible regions and/or the sizes of the respective transmissible regions. SOLUTION: The phase shift quantity between the phase shift transmissible regions and the non-phase shift transmissible regions and any one of the sizes of the respective transmissible regions are so set that the phases of the sum of the amplitude of the diffracted image of the light passed through the respective transmissible region and the amplitude of the diffracted image of the light transmitted through the light shielding films of the peripheries of the respective transmissible regions invert 180 deg. on the image plane from each other with respect to the phase shift transmissible regions and non-phase shift transmissible regions adjacent the each other for the fine patterns most requiring the phase shift effect. Namely, the quantity and size described above are so set that the amplitude C' of the diffracted image on the image plane of the light past the phase shift apertures attains C'=CO-2B=A-2B with respect to the amplitude A of the diffracted image on the image plane of the light past the non-phase shift apertures and the amplitude B of the diffracted image on the image plane of the light transmitted through the Cr regions. At this time, the phase shift region: C'+B=-A-B and the non-phase shift region: A+B are attained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a method for manufacturing a semiconductor device having a fine pattern.

[0002]

2. Description of the Related Art Higher performance / higher integration of semiconductor integrated circuits has been achieved by miniaturization of circuit patterns. The circuit pattern is formed by photolithography in which an original mask on which the pattern is drawn is printed on a photosensitive material (resist) film on a semiconductor substrate via a projection lens. In order to improve the resolution of an optical system to achieve miniaturization, the wavelength of exposure light has been shortened and the numerical aperture of a projection lens has been increased. However, it is considered that the exposure wavelength is an ArF excimer laser having a wavelength of 193 nm and the numerical aperture (NA) is about 0.7, and the resolution limit is about 0.18 μm. On the other hand, as a method for improving the resolution without depending on the performance of these optical systems, a phase shift method for controlling the phase of light transmitted through a mask is widely known. By making a dug part of the substrate in a specific area of the mask or by providing a transparent film having a refractive index different from that of air, light passing through the specific area (referred to as phase shifter) and passing through other areas An optical path difference is provided between the light beams, and a phase difference is realized between light beams transmitted through different regions on the mask. Thereby, the resolution limit can be further improved to about 0.1 μm.

In order to transfer a pattern accurately in the phase shift method, it is necessary to accurately control a phase difference of light transmitted through a mask caused by a phase shifter. This means that if the phase difference on the wafer surface is not exactly 180 degrees,
This is because line width variation occurs during defocusing.
Therefore, the substrate digging amount and the thickness of the transparent film have been made so that the phase difference is 180 degrees.

On the other hand, in order to manufacture a mask so that the phase difference is exactly 180 degrees, it is necessary to first accurately measure the phase of light transmitted through the mask and feed back the result to the manufacturing process. However, the property of light that we can directly measure is its intensity, and measuring the phase requires converting it into intensity using an interferometer. Therefore, the phase measurement of the phase shift mask has hitherto been performed by using the interference between light passing through a region where the phase shifter exists and a region where the phase shifter does not exist. In order to analyze the actual projected image of the mask pattern including the phase shift mask, the mask pattern is enlarged and observed using a microscope having the same optical conditions (NA, exposure wavelength, spatial coherence) as the projection exposure apparatus. , A device for monitoring the projected image (MSM
100, manufactured by Zeiss).

Further, a technique called a phase recovery method is known as one field of optics. In general, the phase recovery method is a general term for a method of obtaining a complex amplitude distribution itself of an image from two image intensity distributions on an image plane and a pupil plane, for the purpose of improving resolution in an electron microscope or an astronomical telescope having a large aberration. Has been considered. Further, as one of the algorithms of the phase recovery method, a method of obtaining a complex amplitude distribution from an image intensity distribution on an image plane and a defocus plane is also known. The phase retrieval method is described in, for example, Computer Technique for Image Processing in Electron Microscopy (Academic, New York, 197).
8 years) Pages 66 to 81 (Image Processing and Com
puter Aided Design in Electron Optics, Academic, N
ew York, 1973, pp.66-81).

[0006]

However, when the above-described phase shift method is applied to actual semiconductor integrated circuit manufacturing, despite the fact that the phase difference is accurately set to 180 degrees,
There has been a problem that sufficient dimensional accuracy expected from optical theory cannot be obtained. As a result of the study by the inventor, it has become clear that the main causes are as follows.

According to the measurement by the inventors, a light-shielding film on a mask, which has conventionally been considered to have a light transmittance of 0 (Cr is usually used in many cases, so Cr is often used hereinafter).
The transmittance of the film (sometimes called a film or a Cr pattern) is not exactly zero. For example, the measurement result of the intensity transmittance of a Cr / Cr oxide laminated film having a thickness of 120 nm used in an i-line exposure mask was about 0.02%. With a conventional Cr mask that does not use the phase shift method, there is almost no problem even if the resist in the region corresponding to the mask light-shielding portion is exposed due to such a transmittance. However, in a phase shift mask that requires high phase accuracy, the Cr transmitted light greatly affects the phase of a projected image obtained on the wafer.

For example, as shown in FIGS. 8A and 8B, a simple repetitive phase shift mask having a period close to the resolution limit (a phase difference between adjacent apertures immediately after passing through the mask is 1).
80 degrees), there is a light-shielding portion (Cr region) that slightly transmits light between the two openings, and the transmitted light therefrom has a phase difference of 90 degrees with respect to the normal mask transmitted light. Suppose you had For each of the diffraction image of the light transmitted through each aperture and the diffraction image of the light transmitted through the Cr region, the amplitude distribution immediately after transmission through the mask, and the amplitude distribution of the projection image obtained on the imaging plane of the projection lens, These are shown by a solid line and a dotted line in FIG. 1, respectively. Note that since the period of the Cr pattern exceeds the resolution limit of the projection lens, the amplitude distribution B of the light passing through the Cr region becomes a uniform background on the image plane. FIG. 2A shows the amplitudes A and C of the diffraction images of the light passing through the phase shift aperture and the non-phase shift aperture at each aperture.
And the amplitude B at both openings of the diffraction image of the light passing through the Cr light shielding region. Even if the amplitude C of the phase shift area is set to be exactly opposite to the amplitude of the non-phase shift area A after passing through the mask (C = -A),
On the image plane, the background amplitude B is added to the complex amplitudes A and C in each of the phase shift portion and the non-phase shift portion, so that the amplitudes in each region are A + B and C + B, respectively, and the phase difference between the two is It deviates from 180 degrees. In general, the amplitude also changes to its absolute value. In fact, the error caused by this is that the phase difference of the Cr transmitted light is 90 degrees and the transmittance is 0.0
If 2%, 2 * at the center of the opening pattern
an (B / A) = reaches 2 degrees. Since the amplitude absolute value is extremely small between the openings, the error due to the background is further increased. Therefore, as shown in FIG.
The phase difference between the openings obtained on the actual wafer surface is 18
Therefore, the desired phase shift effect cannot be obtained.

Next, problems in phase measurement of such a phase shift mask will be described. As mentioned earlier, phase difference measurement has been performed using an interferometer.However, these interferometers are complicated and expensive, and the resolution becomes insufficient as the mask pattern becomes finer, making high-precision measurement difficult. There was a problem of becoming. On the other hand, the mask projection image monitoring apparatus using the above-mentioned microscope can measure only light intensity information, and it is difficult to measure the phase. By the way, using the mask projection image monitoring device, it is possible to measure the mask projection images at different positions in the focal direction. Therefore, it is considered that the phase distribution of the mask projection image can be obtained by applying the above-described phase recovery method to these measurement data. Here, if it is assumed that the phase distribution of the mask projection image represents the phase distribution of light immediately after transmission through the mask, it is expected that the phase accuracy of the phase shift mask can be estimated by this. However, as a result of the study by the inventor, when phase recovery is applied to a projection image obtained by an existing mask projection image monitoring apparatus, the above assumption is not necessarily satisfied for the same reason as described above for the phase shift mask. It has been found that accurate phase measurement becomes difficult. That is, in a normal mask in which the transmittance of the light-shielding region is not always completely zero, Cr
Transmitted light has a significant effect on measurement results. Since the mask-side numerical aperture of the optical system in the conventional projection image monitor is set to be equal to that of the exposure apparatus (the value obtained by dividing the wafer-side numerical aperture by the reduction magnification), for example, when measuring the phase shift mask shown in FIG. Is exactly equal to the distribution shown in FIG. 2b obtained on the image plane of the exposure apparatus. That is, the numerical aperture on the mask side of the projection image monitor is at most 0.2.
Cannot be exceeded, and the resolution is at the level where the smallest dimension on the mask can be resolved. Therefore, the complex amplitude distribution obtained on the detection plane (image plane) of the projection image monitor is such that the contribution of the light transmitted through the opening and the contribution of the light transmitted through the Cr region overlap each other, and it is often difficult to separate the two. . The problem is that we cannot determine from the measured data alone whether this is due to the influence of the Cr region or the phase error of the phase shifter itself. This problem has been clarified for the first time in an experiment conducted by the inventors in an attempt to apply the phase recovery method to the phase mask measurement.

According to the present invention, there is provided a resist formed using the above mask by using a phase shift mask capable of correctly obtaining a phase difference of 180 degrees on a wafer surface without being affected by the effect of the light transmitted through the Cr region. It is a first object of the present invention to provide a method for manufacturing a semiconductor device having higher performance by greatly improving the accuracy and uniformity of a pattern.

Further, the present invention uses the mask manufactured based on the data obtained by accurately measuring the phase of the light passing through the opening pattern without being affected by the effect of the light transmitted through the Cr region or the like. It is a second object of the present invention to provide a method for manufacturing a semiconductor device with higher performance by greatly improving the accuracy and uniformity of a resist pattern formed using a mask.

[0012]

SUMMARY OF THE INVENTION The first object of the present invention is to provide, for a fine pattern requiring the phase shift effect the most, the above described method for each of a phase shift transmission region and a non-phase shift transmission region adjacent to each other. The phase shift is performed so that the phase of the sum of the amplitude of the diffraction image of the light transmitted through each transmission region and the amplitude of the diffraction image of the light transmitted through the light-shielding film around each transmission region is inverted by 180 degrees on the image plane. This is achieved by setting at least one of the amount of phase shift between the transmission region and the non-phase-shift transmission region and the size of each transmission region.
In particular, when the phase difference of the light transmitted through the light shielding film with respect to the transmitted light in the non-phase shift region is about 180 degrees, almost sufficient effect can be obtained only by adjusting the size of the transmission region in accordance with the light shielding film transmittance. Can be When the phase difference is about 90 degrees,
An almost sufficient effect can be obtained only by adjusting the amount of phase shift between the phase shift region and the non-phase shift transmission region according to the light-shielding film transmittance. The phase of the light transmitted through the light shielding film can be adjusted by the thickness and composition of the light shielding film.

A first object of the present invention is to transmit light through the light-shielding region at a position substantially equidistant from the boundary between each of the openings and the light-shielding region in a light-shielding region sandwiched between a pair of transmission regions having different phases. This can also be achieved by providing a phase shifter in the light-shielded region so that the phases of the light beams are inverted. Further, the amplitude absolute value transmittance of the light shielding area may be set to 0.5% or less.

The second object is to form an enlarged image of the light transmitted through the mask through a projection lens having a resolution sufficiently smaller than the characteristic dimension of the structure on the mask, and to form a plurality of positions (de-images) near the image plane. The optical image intensity distribution of the mask projection image is measured on the focus plane), and the complex amplitude distribution of the mask projection image near the imaging plane is obtained from the plurality of optical image intensity distributions by using a phase recovery technique. This is achieved by measuring the phase accuracy of the shift mask and adjusting the phase of the phase shift mask using information on the obtained phase accuracy. The characteristic size may be, for example, half the minimum pattern size on the mask or a typical phase defect size on the mask. To this end, specifically, it is desirable that the numerical aperture on the mask side of the projection optical system be 0.25 or more, more preferably 0.5 or more. At this time, the measurement area on the mask is illuminated with a spatial coherence degree of 0.1 or less, and the measurement area on the mask is not changed for different focal positions. It is desirable to remove noise caused by the reflection of light between each optical element and the mask.

First, the operation of the means for achieving the first object will be described with reference to FIG. FIG. 3a is FIG.
Similarly to the above, when the phase shift method is applied to the simple repetitive pattern of FIG. The amplitudes at the two openings are shown independently of each other. In the present invention, instead of setting the amplitude of the phase shift opening immediately after passing through the mask to be exactly opposite to the amplitude of the non-phase shift opening as in the related art, the phase shift opening is Is the diffraction image amplitude C ｀ on the image plane of the light that has passed through the non-phase shift opening, and the amplitude A of the diffraction image on the image plane of the light that has passed through the non-phase shift opening and the amplitude B on the image plane of the Cr-region transmitted light , C ｀ = C0−2B = −A−2B. At this time, the amplitude obtained on the image plane is as follows because the amplitude B of the diffraction light on the image plane of the light transmitted through the Cr region is added to the phase shift region and the non-phase shift region.

Phase shift area: C ｀ + B = -AB, non-phase shift area A + B Therefore, it can be seen that the absolute values of the two are equal and correctly opposite in phase, and a desired phase shift effect can be obtained on the image plane. . At this time, C ｀ and A are generally not in opposite phases, and their absolute values are generally not equal. This can be realized by appropriately shifting the amount of phase shift between the phase shift aperture and the non-phase shift aperture from 180 degrees and changing the size of both apertures. The amount of phase shift and the amount of correction of the aperture size are determined by the relationship between the amplitudes A, B, and C, and thus depend on the pattern design layout.

The above-described method is based on the concept that the influence of the light transmitted through the Cr region is canceled by the phase and the size of the phase shift opening. However, it is also possible to cancel the influence inside the Cr region. That is, when the phase of the transmitted light from the half region is inverted with respect to the phase of the transmitted light from the other half region in the Cr region sandwiched between the pair of openings whose phases are inverted, both cancel each other out. , Or both have the same effect on each of the openings, thereby canceling out the effect of the Cr transmitted light. Further, when the amplitude absolute value transmittance of the Cr film is 0.5% or less, the effect can be substantially ignored.

Next, the operation of the means for achieving the second object will be described. As described above, in the conventional projection image monitor, the mask-side numerical aperture of the measurement optical system is set equal to the mask-side numerical aperture of the exposure apparatus. That is, since the numerical aperture NAw on the wafer side of the exposure apparatus is about 0.7 at the maximum and the reduction magnification M is 4 or 5, the numerical aperture NAm on the mask side of the measuring optical system is at most NAm = NAw / M = 0.7 / 4. =
It is about 0.175. (Although it is possible to mount a high NA objective lens in a conventional projection image monitoring apparatus, since the actual aperture is defined by another part of the optical system, it is impossible to set the NAm to 0.2 or more. Therefore, the numerical aperture on the mask side of the measuring optical system has room for further increasing by the magnification of the exposure apparatus using the mask. That is, by making the resolution of the measurement optical system sufficiently higher than the minimum dimension of the mask structure, high-precision measurement becomes possible. For example, the information of the transmitted light in the Cr region is separated from that of the aperture, and the phase difference introduced by the phase mask (a phase distribution corresponding immediately after passing through the mask) can be obtained. Further, a minute phase defect or the like on a mask which cannot be detected by the conventional method can be detected.

Next, in the inspection of the phase mask using the present invention, the complex amplitude distribution of the image plane is obtained from a plurality of optical image distributions measured in a plane perpendicular to different optical axes near the image plane. The procedure will be described. The complex amplitude distributions a and O of the mask pattern diffraction image on the image plane and the pupil plane have a Fourier transform F relationship with each other (that is, a = F (O)). Also, the operation of multiplying the complex amplitude distribution of the diffraction image by the defocus wavefront aberration distribution on the pupil plane is a defocus aberration operator D, and the pupil function is P
Defines the image complex amplitude distribution on the defocus plane
b is obtained as b = F (P (D (O))). Therefore, b = F (P (D (Finv (a)))). (Equation 1) where Finv is an inverse Fourier transform.

Hereinafter, in the present invention, the conversion between the image plane and the defocus plane is repeated using (Equation 1) in accordance with the algorithm schematically shown in FIG. That is, the phase part of the complex amplitude distribution obtained by the conversion is left as it is, only the absolute value thereof is replaced with the measured value, and this is used as a new complex amplitude distribution of the conversion destination. This conversion / inversion is repeated until the change in the phase distribution becomes sufficiently small. Assuming that the change in the complex amplitude distribution is sufficiently small, the obtained phase distribution, and thus the complex amplitude distribution, can be considered to satisfy the measurement results.

Specifically, first, the light intensity distribution is measured on the image plane and the defocus plane, and the square root thereof is calculated as the amplitude absolute value distribution.
am and bm. Next, assuming that the phase distribution on the image plane is f0,
Let the initial value of the image plane amplitude distribution a be a0 = am * exp (i * f0) (i
Is the imaginary unit). By substituting this into (Equation 1), the defocus surface complex amplitude distribution becomes b0 = F (P (D (Finv (a0))))
= b0 '* exp (i * g0). Here, b0 'and g0 are the amplitude absolute value and the phase of b0, respectively. Here, a value obtained by replacing the amplitude absolute value b0 ′ in the above equation with the measured value bm is newly set as a complex amplitude distribution b0 on the defocus surface, and the image surface complex amplitude distribution is inversely calculated from b0 using (Equation 1). get a1. That is, a1 = Finv (P
(Dinv (F (b0)))) = a1 '* exp (i * f1). Where a
1 'and f1 are the absolute amplitude and phase of a1, and Dinv is the inverse transform of D. Next, the amplitude absolute value a1 ′ in the above equation is replaced with the measured value of the amplitude absolute value am on the image plane, and the complex amplitude distribution on the defocus plane is calculated again using (Equation 1) to be b1. This process is repeated i times, and the i-th defocus plane complex amplitude distribution bi is determined from the i-th image plane complex amplitude distribution, and the phase distribution fi of the defocus plane and the measured value bm of the absolute amplitude value on the defocus plane Having the (i + 1) th image plane complex amplitude distribution, the image plane phase distribution obtained from the i-th defocus plane complex amplitude distribution obtained above, and the measured value of the amplitude absolute value on the image plane. And The calculation is terminated when the difference between the (i + 1) th image plane phase distribution and the ith image phase distribution is sufficiently small. As a convergence condition of the calculation, an appropriate condition generally used in numerical calculation by an iterative method can be used. In addition, as the initial phase distribution,
The phase distribution that would be expected as naturally as possible (for example,
It is desirable to assume a design value in the case of phase shift mask measurement. In the above description, the calculation is repeated between the image plane and the defocus plane. However, one is not necessarily the image plane, and the calculation is repeated between two or more arbitrary defocus planes. You may. Also, the above discussion assumes that the aberrations of the optical system do not depend on defocus.
If this does not hold, it is desirable to make appropriate corrections.

Next, points to be noted in optical image measurement and data processing for performing phase recovery using the above algorithm will be described. Phase recovery requires coherent illumination in principle. Therefore, it is preferable to make the illumination stop of the mask illumination system as small as possible, and it is desirable that the spatial coherence degree s be 0.1 or less.
In the phase recovery process, Fourier transform is used. Therefore, the measurement pattern is periodic or, if an actual mask pattern that is not necessarily periodic is used for measurement, a large light-shielding area should be provided around the measurement area so that the influence from the surroundings can be ignored sufficiently. Must be provided. In order to make this possible for an arbitrary mask pattern, a field stop for defining a mask illumination area may be provided on a conjugate plane of the mask of the mask illumination optical system. When the mask position is moved in order to measure the optical image by changing the focal position, it is preferable to move the measurement area on the mask together with the condenser lens or the mask stage in the optical axis direction so that the measurement area does not change with the movement. Further, in order to maintain uniformity of illumination, it is ideal to move the entire illumination system together with the mask. It should be noted that if the field stop is too small, light diffraction at the stop will occur and the illumination conditions of the mask will change. As a general noise countermeasure, a light-shielding cover is installed to block external light noise, and surface reflection of each optical element is prevented to prevent reflection of light between the photodetector, optical element in the optical system, and mask. It is preferable to take measures such as shifting the normal direction of the detector surface slightly from the optical axis. In addition, non-uniformity of the light source, speckles due to lighting system dust,
It is desirable to improve mechanical accuracy such as camera noise and focus.

In an actual projected image, much of important information is often included in an area having a low light intensity around the main pattern. In order to avoid these being buried in measurement noise, after capturing the main pattern image with normal exposure, increase the integrated exposure, amplify and capture the low intensity area around it, and scale and connect both. It is desirable to use such data as optical image data. It is also desirable to correct the lateral shift of the optical image due to optical axis deviation / stage error or the like by data processing after measurement. In addition, the background noise observed in the light-shielded area on the mask and the illuminance unevenness in the measurement area observed in the transparent area on the mask should be measured in advance and excluded from the data. Is desirable. Further, processing such as processing the background caused by the mask structure itself is useful.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION

(Embodiment 1) This embodiment shows an example in which the present invention is applied to the formation of a DRAM gate pattern. First, measurement of the phase difference of the phase shift mask according to the present invention will be described. FIG. 5 schematically shows a mask measuring apparatus used in this embodiment. The mask 3 on the mask stage 2 was illuminated with the illumination light 1 which was substantially coherent in space, and the pattern on the mask 3 was enlarged using the enlargement projection optical system 4 to form an image on the CCD sensor 5. The mask-side numerical aperture NAm of the magnifying projection optical system 4 is variable up to 0.9 at the maximum. After a signal from the CCD sensor was input to the computer 6, the signal was processed to obtain a mask projection image intensity distribution. By repeating the above measurement while moving the mask stage in the direction of the optical axis 7, the projected image intensity distribution of the mask pattern at the focal point position 8, the +50 μm defocus plane 9 and the −50 μm defocus plane 10 was measured. The illumination range on the mask is defined by the field stop 11 on a plane conjugate with the mask plane. The field stop 11 moves together with the condenser lens 12 when moving the mask stage 2 in the optical axis direction. The spatial coherence degree of the mask illumination system 13 is 0.0
Set to 5. In the computer, the obtained measured data of the mask projection image intensity distribution was input to a phase recovery program using the algorithm described above, and the phase distribution at the focal point 8 was output.

Next, using the above device, a 1 Gbit DR
In the AM gate pattern phase shift mask, a plurality of masks in which the amount of reentrant of the phase shifter was variously changed were measured. The results obtained by setting the mask-side numerical aperture to 0.5 and 0.15 (mask-side numerical aperture of an exposure apparatus in which the mask is used) were measured, and the phase distribution of the gate pattern portion was obtained. 6 is shown. Assuming that the numerical aperture is 0.5 at any of the digging amounts, the phase of light passing through the openings on the mask and the phase of weak light transmitted through the Cr region between the openings are clearly separated and observed. You can see that it is done.
When the engraving amount is 250 nm (FIG. 6a), a phase difference of 180 degrees is obtained between the adjacent openings in the measurement result of the numerical aperture 0.5, whereas 1 is obtained in the measurement result of the numerical aperture 0.15.
83 degrees. Actually, when a DRAM was experimentally manufactured using this mask, a line width fluctuation which was considered to be caused by a focus fluctuation occurred. On the other hand, when the engraving amount is 246 nm (FIG. 6B), the phase difference due to the numerical aperture of 0.5 is 177 degrees, but at the numerical aperture of 0.15, 180 degrees is obtained. From this, it was found that a phase difference of 180 degrees can be obtained on an actual wafer by setting the phase difference of the phase shifter to 177 degrees. This is considered to be due to the influence of light transmitted through the Cr region between the openings. Therefore, the above device was manufactured using a mask in which the amount of shifter depression of the phase shift mask was set to an amount corresponding to a phase difference of 177 degrees. As a result, a dimensional change due to a focus change that occurs when the phase deviates from 180 degrees was suppressed, and a gate pattern could be formed accurately. Although the optimum engraving amount of the phase shifter depends on the pattern as described above, since the dimensions of the pattern in the region other than the gate pattern are relatively large, even if the phase difference is optimized and set to the gate pattern, it is not so large. There was no problem. The above DRAM
In the manufacturing process, a phase shift mask was used for layers other than the gate pattern, but the optimum phase shift amount in each layer was set individually by performing the above measurement for each layer. Note that the configuration of the measuring apparatus, the type of device to be applied, and the like are not limited to those shown in the present embodiment, and various changes can be made without departing from the gist of the present invention.

(Embodiment 2) Projection image monitor (MSM)
100, manufactured by Zeiss Co., Ltd.), the hole pattern projection image on the conventional Cr mask was measured, and the phase distribution was obtained by applying the phase recovery method as in Example 1. According to optical theory, the phase converges to 90 degrees away from the center, although the phase should alternate in the sidelobes around the main pattern (FIG. 7a). At this time, when the image amplitude distribution on the pupil plane was output, a peak appeared at the origin (FIG. 7b), and the phase of a portion corresponding to this peak was shifted by about 90 degrees from the phase of another region in the pupil plane. Then, in the final stage of the data processing, this peak was removed from the pupil plane, and this was subjected to Fourier transform. As a result, a theoretical phase distribution in which the phase was alternately inverted for each side lobe was obtained (FIG. 7C).
This is because background light that is out of phase by about 90 degrees with respect to light that has passed through a hole is superimposed from a wide area of the mask, and a portion corresponding to this background is selectively removed in a frequency space. This has eliminated the effect. This was confirmed from the fact that the attenuation of FIG. 7A was not observed in the case where a pinhole formed of a sufficiently thick stainless thin film was used instead of the Cr mask. As described above, in this embodiment, despite the use of measurement data obtained by a conventional projection image monitoring device that does not always have sufficient resolution, it is desired to know by selectively removing the influence of Cr transmitted light and the like by appropriate data processing. Phase information could be obtained.

Next, an example in which this method is applied to the manufacture of a 256 MDRAM will be described. First, as a mask for forming a contact hole pattern of the device, a so-called halftone phase shift mask designed to have a light transmittance of 3% in the light-shielding region is formed by changing the deposition conditions of the light-shielding region film, Was measured by the method described above. That is, after measuring masks having different deposition conditions using the MSM 100, when the complex amplitude distribution on the pupil plane was calculated, a peak of the amplitude absolute value appeared at the origin in any of the masks. The complex amplitude at the pupil plane origin is considered to be a superposition of the background from the halftone region and the diffraction image of the hole pattern itself. Therefore, the amplitude of the diffraction image of the hole pattern itself was obtained by interpolation from the absolute value of the image amplitude at a position deviated from the origin, and the complex amplitude of the background was obtained by subtracting this from the measured value of the complex amplitude at the origin. Further, the phase difference between the background complex amplitude obtained in this way and the complex amplitude of the hole pattern itself was calculated, thereby obtaining the phase shift amount of the halftone portion. The above device was manufactured using a mask formed under the deposition conditions where the phase shift amount was 180 degrees as a result of the measurement.

This method is applicable not only to the case of the hole pattern but also to the case where the magnitude of the influence of the light transmitted through the light shielding area can be quantitatively predicted in advance.

[0029]

As described above, according to the method of manufacturing a semiconductor device according to the present invention, a semiconductor device is formed by projecting and exposing a phase shift mask onto a semiconductor device substrate through a projection lens to form the circuit pattern on the substrate. When manufacturing the device, for the fine pattern requiring the phase shift effect most on the mask, for each of the phase shift transmission region and the non-phase shift transmission region adjacent to each other, on each of the transmission region peripheral masks The phase shift transmission region and the phase shift transmission region are inverted such that the phase of the sum of the amplitude of the diffraction image of the light transmitted through the light shielding film and the amplitude of the diffraction image of the light transmitted through each transmission region is inverted by 180 degrees on the image plane. By setting at least one of the phase shift amount between the phase shift transmission regions and the dimensions of each transmission region, the precision of the resist pattern formed using the mask can be improved. , It is possible to greatly improve the uniformity, to produce a higher performance of the semiconductor device.

Further, by using the method of manufacturing a semiconductor device according to the present invention, when manufacturing the above mask, the light transmitted through the mask is magnified through a projection lens having a resolution sufficiently smaller than the characteristic dimension of the structure on the mask. An image is formed, an optical image intensity distribution of the mask projection image is measured at a plurality of positions (defocus plane) near the image plane, and a phase recovery method is used from the plurality of optical image intensity distributions using a phase recovery technique. Determine the complex amplitude distribution of the mask projection image, by using a mask manufactured based on data obtained by estimating the phase of light that has passed through the phase shift mask from this information, the accuracy of the resist pattern formed using the mask, The uniformity can be greatly improved, and a higher-performance semiconductor device can be manufactured.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a problem of a conventional method.

FIG. 2 is a schematic diagram showing a problem of a conventional method.

FIG. 3 is a schematic view illustrating the principle of the present invention.

FIG. 4 is a diagram schematically showing a phase recovery algorithm according to the present invention.

FIG. 5 is a diagram schematically showing a mask measuring apparatus used in one embodiment of the present invention.

FIG. 6 is a diagram schematically showing an optical image measurement result in one example of the present invention.

FIG. 7 is a diagram schematically showing an optical image measurement result in another example of the present invention.

FIGS. 8A and 8B are an example of a main part cross-sectional view and a main part plan view of a line and space part of a phase shift mask. FIGS.

[Explanation of symbols]

1: illumination light, 2: mask stage, 3: mask, 4: magnifying projection optical system, 5: CCD sensor, 6: computer, 7: optical axis, 8: focal point position, 9, 10: defocus plane, 11 : Field stop, 12: condenser lens, 1
3: Mask illumination system.

Claims (6)

[Claims] A mask having a circuit pattern is projected and exposed on a semiconductor device substrate through a projection lens;
A method of manufacturing a semiconductor device comprising a step of forming the circuit pattern on the substrate, wherein the mask has a phase shift transmission region for substantially inverting phases of light transmitted on both sides of a light shielding region on the mask, and A phase shift mask including a non-phase-shifting transmission region, for each of the phase-shifting transmission region and the non-phase-shifting transmission region adjacent to each other for a fine pattern requiring the most phase-shift effect; The phase of the sum of the amplitude of the diffraction image of the light transmitted through the surrounding light-shielding region and the amplitude of the diffraction image of the light transmitted through each of the transmission regions is inverted in the range of 180 ± 1 degrees from each other on the image plane. A method of manufacturing a semiconductor device, comprising setting at least one of a phase shift amount between a phase shift transmission region and a non-phase shift transmission region and each of the transmission region dimensions. 2. The thickness and composition of a light-shielding film constituting the light-shielding region are set so that a phase difference of light transmitted through the light-shielding region with respect to transmitted light in a non-phase shift region is approximately 180 degrees.
And setting at least one of the phase shift transmission region size and the non-phase shift transmission region size according to the transmittance of the light shielding film, or the light shielding film so that the phase difference is about 90 degrees. 2. The semiconductor device according to claim 1, wherein a thickness and a composition of the light-shielding film are set, and a phase shift amount between the phase-shift transmission region and the non-phase-shift transmission region is set according to the transmittance of the light-shielding film. Manufacturing method. 3. The light transmitted through the light-shielding region at a position substantially equidistant from the boundary between each of the transmission regions and the light-shielding region in the light-shielding region sandwiched between the phase-shifting transmission region and the non-phase-shifting transmission region. A method for manufacturing a semiconductor device, wherein a phase shifter for inverting phases is provided in a light shielding region. 4. A step of inspecting a mask having a light-shielding pattern on a transparent substrate; a step of judging the quality of the mask based on the inspection; A method of manufacturing a semiconductor device comprising a step of forming the circuit pattern on the substrate by projecting and exposing the substrate on the substrate, wherein the step of inspecting the mask comprises: Illuminate and form an image of the pattern on the mask on an image plane through an enlargement optical system having a resolution sufficiently smaller than the characteristic dimension of the structure on the mask, and perpendicular to the optical axis at different positions near the image plane. The two-dimensional light intensity distribution of the projected image of the mask is measured on a plurality of planes, and the light in the vicinity of the image plane is measured from the projected image light intensity distribution on the plurality of planes by using a phase recovery technique. Obtains an image complex amplitude distribution, a method of manufacturing a semiconductor device characterized by further comprising the step of obtaining the optical image from the complex amplitude distribution of the light transmitted through the mask phase. 5. The magnifying optical system having a mask-side numerical aperture of 0.5.
6. The method according to claim 5, wherein
The manufacturing method of the semiconductor device described in the above. 6. The method according to claim 5, wherein said mask is a phase shift mask.