JP2003303768A - Pattern formation method and drawing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a pattern formation method for easily correcting a dimensional variation in an etching process. <P>SOLUTION: In the pattern formation method, a sample coated with a resist is irradiated with an energy beam, a pattern is formed by development, and the etching process is carried out with the pattern as a mask for forming a desired pattern on the sample. In this case, the energy beam is corrected for illumination so that the variation of pattern dimension in the etching process is corrected. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern forming method and a drawing method.

[0002]

2. Description of the Related Art A pattern forming method using an electron beam is capable of forming a much finer pattern as compared with a method using light, and is used for trial manufacture of semiconductor elements. Further, in the pattern forming method using an electron beam,
It is also used to make masks used in optical lithography.

FIG. 1 shows an example of the construction of a drawing apparatus using such an electron beam.

The electron beam emitted from the electron gun 1 is focused by the focusing lens 2 and irradiated on the first shaping aperture 3. The image of the first shaping aperture 3 is formed on the second shaping aperture 5 by the projection lens 4.

Here, a shaping deflector 6 is provided inside the projection lens 4, and by this shaping polarizer 6, the position of the image shaped by the first shaping aperture is positioned on the second shaping aperture 5. adjust. By doing so, a rectangular or triangular beam of arbitrary size is obtained.

The image formed by the second forming aperture 5 is reduced by the objective lens 7 and formed on a reticle mask 8 which is a sample.

The irradiation position of the beam on the reticle mask 8 is largely deflected by the high-precision main deflector 9 to the center of the sub-deflection region, and further, by the high-speed sub-deflector 10 in the sub-deflection region 11. adjust.

The reticle mask 8 alternately performs a stage step movement which moves between the frames 12 and a continuous stage movement which moves continuously within the frame 12.

In such an electron beam drawing apparatus, a pattern accuracy error called a proximity effect is observed. The proximity effect will be described below.

As shown in FIG. 2, the incident electrons 14 incident on the sample 13 are scattered inside the sample 13 to generate secondary electrons. A certain proportion of these secondary electrons and scattered incident electrons act as backscattered electrons 15 on the resist 16 formed on the surface of the sample 13. That is, the background irradiation amount is generated in addition to the irradiation amount by the incident electrons. The spread of this photosensitivity is about 10 μm in a device of 50 keV.

Further, the exposure amount of the resist by the backscattered electrons 15 varies depending on the density of the pattern to be formed. Since the pattern size after resist development is determined by the combined exposure of the original incident electron 14 exposure and the backscattered electron 15 exposure, the resist pattern size after development varies depending on the density of the pattern.

This is called the proximity effect. The proximity effect may also be caused by beam blurring or scattering of electrons in the resist.

Further, in the electron beam drawing apparatus, a pattern accuracy error called a long distance photosensitivity is also observed. The long-range photosensitivity will be described below.

As shown in FIG. 3, it is assumed that incident electrons 14 are incident on the surface of the sample 13. Some of the incident electrons 14 and secondary electrons are emitted from the sample surface and return to the lower surface 18 of the objective lens 17 of the device. The re-reflected electrons 19 that are further reflected on the lower surface 18 return to the resist 16 to be exposed. That is, the background dose is generated.

This phenomenon is called a long-range photosensitivity. The range covered by this long-range photosensitivity is several tens of meters from the beam irradiation position.
Therefore, when the average irradiation dose has a large distribution on the scale of several mm over the area of m, the pattern size of the resist after development has a large size distribution.

On the other hand, when a sample is etched by using the resist pattern as a mask to form a pattern, the progress of the etching varies mainly depending on the pattern density, so that the pattern dimension also varies. . This is called the loading effect.

The dimensional variation due to the loading effect largely depends on the pattern density of the formed resist pattern. The correction of the micro loading effect is performed after the correction of the proximity effect (for example, refer to Patent Document 1).

[0018]

[Patent Document 1] Japanese Patent No. 3074675

[0019]

In the conventional pattern forming method including energy beam writing, there is a problem that a pattern size variation occurs due to a loading effect during etching. Further, in Patent Document 1 described above, the microloading effect is corrected after the proximity effect is corrected, but the irradiation amount for correcting the microloading effect is experimentally obtained. However, as will be described later, since the loading effect interacts with the proximity effect and the long distance photosensitivity, accurate correction cannot be performed simply by determining the correction irradiation amount for correcting the loading effect. The present invention has been made in view of such a problem, and an object thereof is to provide a pattern forming method and a drawing method in which an accurate correction dose is determined, and the finished dimension variation is constant regardless of the pattern density. And

[0020]

A pattern drawing method according to a first aspect of the present invention is a pattern drawing method of irradiating an energy beam on a resist-coated sample to draw a pattern, wherein the pattern is drawn. Drawing data for performing the reference, the reference dose D ₀ for irradiating the energy beam, and the dimension variation Δ due to the influence of the loading effect.
Of the pattern-dependent distribution and the energy distribution s of the energy beam given to the resist, dividing the drawing area into a grid to form sub-drawing areas, and based on the drawing data , A step of obtaining a pattern area density distribution for each of the sub drawing areas, and a dose DC _f (x for correcting long-range photosensitivity in the sub drawing area based on the pattern area density and the reference dose D _0. ), And a dose DC _p for correcting the proximity effect for the pattern in the sub-drawing region based on the writing data and the reference dose D _0.
(X) calculating step, dose DC _f (x) for correcting the long-range photosensitivity, dose DC _p (x) for correcting the proximity effect, and pattern-dependent distribution of the dimensional variation Δ , And a step of obtaining an irradiation amount D (x) based on the energy distribution s of the energy beam, and the irradiation position and shape of the energy beam are determined based on the data on the position and shape of the pattern of the sub drawing area. Irradiating the energy beam only for a time period of the amount D (x).

The pattern area density distribution and the pattern area density distribution
Pattern of dimensional variation Δ due to loading effect
From the distribution of dependence,
And a step of obtaining a dimensional variation distribution Δ (x).
Dose DC to correct long range photosensitivity_f(X) is the above
Pattern area density distribution, the reference dose D₀, And before
Calculated based on the dimensional variation distribution Δ (x),
Dose DC to correct the effect_p(X) is the sub drawing area
Data on the position and shape of the pattern of the area, the standard
Dose D₀, The dimensional variation distribution Δ (x) and the energy
Calculated based on the energy distribution s of the rugie beam
It may be configured as follows. The dose D (x) is
Dose DC for correcting the proximity effect_p(X) and the distance
Dose DC to correct the photosensitivity effect_fIs the product of (x)
It may be configured as follows. In addition, the long-range photosensitivity
Correcting dose DC_f(X) is the reference dose D ₀Is
As an alternative, the dose D (x) may be obtained.
Yes. The dose D (x) is D (x) = DC
_p(X) x DC_f(X) / (1+ (2s (Δ) -1) ×
(DC_p(X) x DC_f(X) / D₀))
Is also good. In addition, the above-mentioned dimensional variation Δ is due to the loading effect.
In addition, the effect of non-uniformity of the etching rate of the resist
It may also include dimensional variation due to. The distance
Dose DC to correct the photosensitivity effect_f(X) is the reference light
Target D ₀The dose D (x) may be calculated as
Yes.

Further, in the pattern forming method according to the second aspect of the present invention, a step of forming a resist pattern by drawing a pattern on the resist by using the above pattern drawing method, and forming the resist pattern, and the resist Etching the sample using the pattern as a mask to form a pattern on the sample.

Here, the loading effect refers to an effect in which, when a sample is etched by using a resist pattern as a mask to form a pattern, the progress of etching mainly varies depending on the pattern density.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION Before describing the embodiments of the present invention, consideration by the present inventors regarding the exposure amount of an electron beam in consideration of the loading effect will be described.

FIG. 4 is a diagram showing an energy distribution given to a resist by an electron beam spot at a pattern end when writing is performed by an electron beam writing apparatus.

In FIG. 4, for simplification, the energy distribution at the pattern end is linear. Here, the sparse pattern area A and the dense pattern area B are corrected in a range where the influence of the proximity effect is considered without considering the loading effect. It is assumed that the two patterns are located at positions where the effects of long-range photosensitivity are nearly equal.

The exposure dose of the sparse pattern region A is Dr, the background exposure amount due to the long-range photosensing action is DB _f , and the dense pattern region B is
Df, and the background dose due to the proximity effect is DB _p . The sparse pattern area A and the dense pattern area B
, It is assumed that the influences of the long-range photosensitivity are close to the same degree, so that the background irradiation amount of the dense pattern region B due to the long-range photosensitivity is DB _f as in the case of the sparse pattern region A. . Further, in this example, for simplicity, it is assumed that the influence of the proximity effect can be ignored in the sparse pattern area A. In FIG. 4, w represents a half value of the spread of the energy distribution given to the resist by the incident electrons on the pattern end surface, which is about half of the blur of the electron beam spot, that is, the center of the spread, that is, The position of half of the reference dose D ₀ is the pattern edge of the resist pattern.

The irradiation amount Df of sparse pattern region A dose Dr and dense pattern region B of so as to satisfy the equations shown in each _{0.5Dr + DB f = 0.5D 0 0.5Df} + DB p + DB f = 0.5D 0 Decide Here, the background irradiation amount DB _p due to the proximity effect and the background irradiation amount DB _f due to the long-range photosensitivity are calculated by using the integral of the entire surface of the mask DB _p = η∫σ (x−x ′) D (x ′) dx ′. DB _f = θ∫p (x−x ′) D (x ′) dx ′. Here, D (x) represents the dose of incident electrons at the position x, and η and θ are parameters indicating the effects of proximity effect and long-range photosensitivity, respectively.
Further, the integrand σ (x) represents the spread of the proximity effect.
Also, the integrand function p (x) represents the spread of the long-range photosensitivity.

[0029] Since it is assumed that approximately equal the influence of the loading effect in the sparse pattern regions A and dense pattern region B where added dose DC _a for correcting the influence of the same loading effect in two patterns.

As shown in FIG. 5, the sparse pattern area A is now available.
In addition, a correction reference to correct the influence on the loading effect
Radiation DC_aThe dimension of the pattern is Δ
Suppose that you can correct it. That is, the corrected dose DC_aAdd
The irradiation dose in the sparse pattern area A is the reference irradiation dose.
D₀Is equal to half the value of _a
The irradiation dose in the sparse pattern area A before adding is the reference irradiation dose.
D₀To the right of the drawing than the position equal to half the value of
To move by Δ. To do this, (Dr + DC_a) (W-Δ) / 2w + (1 + DC_a/ D
C_mf) DB_f= 0.5D₀ Corrected dose DC to meet_aJust decide. Well
DC_mfIs the background dose DB due to long-range photosensitivity
_fIs the average irradiation amount in the integration area given by.

Next, the dense pattern region B, and that the pattern dimension variations in accordance with the addition of corrected dose DC _a for correcting the influence of the loading effect delta 'When, (Df + DC a) ( w-Δ' ) / 2w + (1 + DC _a /
Df) DB _p + (1 + DC _a / DC _mf ) DB _f = 0.
It becomes 5D ₀ . In this _{case, DBar = (DC a / DC} mf) DB f, DBaf = (DC a / Df) DB p + (DC a / DC
_mf ) DB _f , as shown in FIG. 5, DBar and DBaf represent increments of the background irradiation amount, and the dimensional variation Δ ′ in the dense pattern area B is larger than that in the sparse pattern area A.

Now, assuming that DB _p = Df / 3, ignoring the background irradiation amount DB _f due to the long-distance photosensitivity, and calculating that Δ is very smaller than w, Δ′˜3Δ. That is, the dimension variation Δ ′ of the dense pattern area B is about 3 times the dimension variation Δ of the sparse pattern area A.

From the above, it can be seen that even if a uniform correction irradiation amount for simply correcting the loading effect is determined, accurate correction cannot be performed. Embodiments of the present invention will be described below with reference to the drawings. The present invention is not limited to the following embodiments and can be variously modified and used.

FIG. 6 shows a glass mask having a side of 150 mm, and consider the case of drawing on this glass mask by an electron beam drawing apparatus.

First, as shown in FIG. 6, the sample surface is divided into a square lattice (sub-drawing area) having δ _x and δ _y (δ _{x to} δ _{y to 1} mm) on both sides. Here, δ _x and δ _y may be a fraction of the typical magnitude (about 10 mm) of the dimensional variation due to the long-distance photosensitivity.

Further, in the electron beam drawing apparatus, when the size of the grating is larger than the size of the deflection area, making the size of the grating an integral multiple of the size of the deflection area improves the drawing efficiency. Is preferred. For example, when the size of the deflection area is 500 μm square, the size of the grating is 100
Take 0 μm = 1 mm square. If the drawing area of a glass mask of 150 mm on a side is 130 mm square, the size of the grid is 1 mm square, so the number of grids is 130 × 130 =
There are 16900. The deflection area is described as 1 mm square.

First, the dose D (x) in each shot, which is a combination of the correction of the proximity effect, the long-distance exposure effect, and the loading effect, is 0.5 D (x) (w-Δ (x)) / w + η∫σ (x −x ′) D (x ′) dx ′ + θ∫p (x−x ′) D (x ′) dx ′ = 0.5D ₀ (1) For the convenience of description, x and x ′ are two-dimensional vectors.

Here, η and θ are parameters showing the effects of the proximity effect and the long-range photosensitivity, respectively. Also,
σ (x) and p (x) are functions that give the spread of the proximity effect and the long-range photosensitivity, respectively. These are experimentally obtained in advance.

Δ (x) gives a dimensional variation due to the loading effect at the position x. It is considered to be constant within a region of about 1 mm square. Usually the proximity effect spread σ (x)
Is about 10 μm, and the spread of long range photosensitivity p
(X) is about several mm.

Here, in equation (1), the integration range on the left side is the pattern area on the entire surface of the mask. However, practically, the integration range in the first integration including σ (x−x ′) is limited to a region centered at the point x and having a radius of several tens of μm, and includes p (xx−x ′). The integration range in the second integration may be limited to a region centered at the point x and having a radius of about 30 mm.

As a method of correcting the proximity effect and the long distance photosensitivity, for example, the following method can be considered.

As D (x) = DC _p (x) × DC _f (x), the change in the dose DC _f (x) for correcting the long-range photosensing effect is the dose DC _p (x) for correcting the proximity effect. Equation (1) is 0.5DC _p (x) DC _f (x) (w−Δ (x)) / w + ηDC _f (x) ∫σ (x−x ′) DC _p (x ′) dx ′ + θ∫p (x−x ′) DC _p (x ′) DC _f (x ′) dx ′ = 0.5D ₀ (2)

Now, the dose DC for correcting the proximity effect
_p (x) satisfies 0.5DC _p (x) + ηw / (w−Δ (x)) ∫σ (x−x ′) DC _p (x ′) dx ′ = 0.5 ... (3) To decide. Under this condition, the remaining equation is 0.5DC _f (x) + θw / (w−Δ (x)) ∫p (x−x ′) DC _p (x ′) DC _f (x ′) dx ′ = 0. 5w / (w-Δ (x)) D ₀ (4). The integration included in the equation (4) does not cause a large error even if the area to be integrated is set to about 1 mm square and the dose DC _f (x) for correcting the long-range photosensitivity is constant within the area. . When limited to this 1 mm square region, the term including the integral of the equation (4) is θw / (w−Δ (x)) × Σp (x−x _j ) DC _f (x
_j ) ∫ _j DC _p (x ′) dx ′. Here, j is a subscript indicating a region, and ∫ _j means integration in the region j. The term including the integral of the equation (4) will be represented by the sum of the terms including the integral for each region.

If both sides of the equation (3) are integrated in the area where the pattern of the area j of about 1 mm square is present, the integration of the second term on the left side is a double two-dimensional spatial integration ∫ _j ∫σ (xx−x ′). ) DC _p (x ′) dx′dx. Here, among the above integrals, ∫ _j σ (xx−x ′)
Consider the dx part. The value of the integration changes depending on how many patterns exist in the area centered on the position x ′. However, if the pattern is dense, if x ′ is included in the region j, it may be approximated to 1 regardless of x ′. Therefore, this value is set to 1 as the first approximation.
In this case, the above integration becomes ∫ _j DC _p (x ′) dx ′ obtained by integrating DC _p (x) in the area j of 1 mm square. After all, (0.5 + ηw / (w−Δ (x _i ))) ∫ _j DC
_p (x ′) dx ′ = 0.5 × (pattern area in the region j) is obtained. That is, ∫ _j DC _p (x ′) dx ′ = 0.5 × (pattern area in the region j) / (0.5 + ηw / (w−Δ (x _i ))), and the equation (4) becomes the region i. About 0.5 DC _f (x _i ) + θw / (w−Δ (x _i )) / (0.5 + ηw / (w−Δ (x))) Σp (x _i −x _j ) DC _f (x _j ) × 0.5 × (pattern area in the region j) = 0.5 w / (w−Δ (x _i )) D ₀ ... (5) DC _f (x _i ) can be obtained by simultaneous equations (5) for each area on the entire surface of the mask. Still more improve the accuracy of the approximation is defined as the pattern density e _j within the area j e _{j =} (area of the region j) / (pattern area in the area j), ∫σ (x-x ') dx Can also be approximated as e _j . At this time, (0.5 + e _j ηw / (w−Δ (x _i ))) ∫DC
_{Since p} (x ′) dx ′ = 0.5 × (the pattern area in the region j), the equation (4) is slightly different from (5) in shape, and 0.5DC _f (x _i ) + θw / (w− Δ (x _i )) Σ1 / (0.5 + e _j ηw / (w−Δ (x _i ))) p (x _i −x _j ) DC _f (x _j ) × 0.5 × (pattern in region j Area) = 0.5 w / (w-Δ (x _i )) D ₀ ···· (5 ′). Further, it is naturally possible to improve the calculation accuracy by directly calculating DCp (x) from (3) and substituting it into (4) using an appropriate approximation.

Since the size of the mask is larger than the spread of p (x), it is necessary to calculate not only the entire region but also the region within a radius of 30 mm from x _i , for Σ in equation (5). The calculation becomes easier if this is done.

In the method described in equations (5) and (5 '), DC _f (x _i ) is a parameter η indicating the influence of the proximity effect, a parameter θ indicating the influence of the long-range photosensitivity, and If the pattern area in each region is known, the fine pattern distribution required when performing proximity effect correction is not necessary.

Dose DC for correcting long-range photosensitivity
_A dose DC for correcting the proximity effect obtained by previously calculating _f (x) and calculating the vicinity of the drawing area at the time of drawing
By calculating _p (x) and multiplying it by the value of the dose DC _f (x) that corrects the proximity effect given to that region, it is possible to correct all of the proximity effect, the long-range photosensitivity effect, and the loading effect.

The pattern density dependence of the dimensional variation Δ due to the loading effect can be obtained by etching resist patterns having different pattern densities.

Here, the simplest linear distribution is assumed as the effective dose distribution, but of course, other distribution may be used. Let u be the dimensional variation Δ due to the loading effect, and consider it one-dimensionally and u = 0 at the pattern end. Further, it is assumed that the dose distribution is given by D (x) × s (u). Here, s (u) represents the energy distribution given to the resist by the beam spot, and s (0)
= 0.5, s (∞) = 0, and s (−∞) = 1.

In the above example (in the case of linear distribution), -w≤u≤
s (u) = 0.5 × (w−u) / w in w, s in u <−w
(U) = 1, u> w and s (u) = 0.

When using the method described above, the equation (1) for giving the dose D (x) is as follows.

D (x) s (Δ) + η∫σ (x−x ′) D (x ′) dx ′ + θ∫p (xx−) D (x ′) dx ′ = 0.5D ₀ ... ··· (6) Here, as before, D (x) = DC _p (x) × DC
_{As f} (x), DC _p (x) + η (1 / s (Δ)) ∫σ (x−x ′) DC _p (x ′) dx ′ = 0. 5 ... (7) 0.5 DC _f (x) + θ (1 / s (Δ)) ∫p (x−x ′) DC _p (x ′) DC _f (x ′) dx ′ = 0.5 ( Dose DC _p (x) for correcting the proximity effect by solving 1 / s (Δ)) D ₀ (8),
A dose DC _f (x) that corrects the long-range photosensitivity can be obtained.

The energy distribution s (Δ) given to the resist by the spot beam can be experimentally obtained from the dose dependence of the resist pattern size variation.
For example, assuming a proper function form such as an error function, parameters may be determined by drawing a test pattern.

A rectangular pattern is arranged in an array at an equal pitch, and a region at the center is sufficiently larger than the distance affected by the proximity effect and smaller than the distance affected by the loading effect, for example, a size of 200 μm square. Take the area of. Linear patterns having different densities are drawn in the central portion in a region of about 100 μm. Here, in the drawing of the central pattern, the proximity effect correction and the long distance photosensitivity correction are performed. The representative value of the dimensional variation in the pattern after development is defined as the dimensional variation Δ due to the influence of the loading effect on the peripheral pattern density. As a typical value, for example, the dimensional variation Δ averaged around the most important density of the central pattern Δ
Is preferred.

FIG. 8 shows an example of a system configuration for realizing the above correction.

First, in the storage means 20 of the system,
A table of the dependence of the dimensional variation Δ due to the effect of the loading effect on the pattern area density and a function s (x) representing the energy distribution of the spot beam is stored.

Next, the drawing data and the reference dose D ₀ are input and stored in the storage means 21 of the data processing computer of the apparatus.

Next, based on the drawing data stored in the storage means 21, the calculation means 24 divides the entire drawing area into 1 mm square grids, and the pattern area density for each 1 mm square grid (sub drawing area). Ask for.

Next, from the pattern area density distribution and the table of the pattern area density dependence of the dimension variation Δ due to the effect of the loading effect stored in the storage means 20, the dimension variation Δ in the computing means 25 is obtained. Dimensional variation distribution Δ
(X) is obtained, and the table of the size variation distribution Δ (x) is stored in the storage means 26.

Next, the pattern area density and the dimension variation distribution Δ (x) of the dimension variation Δ due to the influence of the loading effect,
The equation (8) is solved from the reference dose D ₀ by the long distance photosensitivity correction calculator 27 to obtain the dose DC _f (x) for correcting the long range photosensitivity, and the table is stored in the storage means 28.

On the other hand, in the drawing system 22, at the time of drawing, the data processing computer takes out the pattern data in the sub drawing area from all the pattern data stored in the storage means 21 and sends it to the figure dividing circuit 23.

Next, the figure dividing circuit 23 sends the data concerning the position and shape of the small figure in the sub drawing area obtained by the division to the proximity effect correction arithmetic circuit 29.

At this time, the proximity effect correction arithmetic circuit 29
The position information of the sub drawing area is sent to the storage unit 26 in which the size variation distribution Δ (x) due to the loading effect is stored, and the size variation distribution Δ (x) corresponding to the sub drawing area.
Retrieves the value of.

Further, the proximity effect correction arithmetic circuit 29 uses the value of the size variation distribution Δ (x) as a function s representing the energy distribution given to the resist by the energy beam.
Is sent to the storage means in which s (Δ
(X)) value is taken out.

Next, based on the equation (7), the dose DC _p (x) based on the proximity effect correction is calculated for each small figure in the sub drawing area, and is output to the arithmetic circuit 30.

Next, the arithmetic circuit 30 uses the table of the long-distance photosensitivity correction irradiation dose DC _f (x) stored in the storage means 28 to determine the long-distance photosensitivity correction irradiation dose DC corresponding to the sub drawing area. The value of _f (x) is taken out, and the final corrected dose D (x) = DC _p (x) × DC _f (x) is obtained, and is sent to the drawing circuit 31 together with the data on the position and shape of each small figure. Output.

The drawing circuit 31 determines the irradiation position and shape of the electron beam based on the data of the position and shape of the small figure in the sub drawing area, and applies the electron beam to the sample for the irradiation time of the irradiation amount D (x). Irradiate. This is repeated until the drawing is completed for all the small figures in the sub drawing area. When drawing in one sub drawing area is completed, the process moves to drawing in the next sub drawing area.

Depending on the system and required conditions, it may be possible to ignore the long-range photosensitivity. In that case, in the above process, the process of obtaining the dose DC _f (x) for correcting the long-range photosensitivity is omitted, and D (x)
= DC _p (x) × D ₀ .

In the above method, the parameter indicating the proximity effect in the proximity effect correction calculation is not η but ηw /
(W−Δ (x)), which takes a different value for each grid. However, in consideration of speeding up the operation or the configuration of the correction operation circuit, it is easier to configure the system if the parameter in the proximity effect correction is η and is constant over the entire mask. Therefore, it is possible to do the following.

First, without considering the loading effect, the dose D (x) in each shot is adjusted to 0.5 D (x) + η∫σ (x−x ′) D by correcting the proximity effect and the long-distance photosensitivity. (X ′) dx ′ + θ∫p (x−x ′) D _d (x ′) dx ′ = 0.5D ₀ ... (9) Descriptions of x and x ′ are two-dimensional vectors for convenience. Here, η and θ are parameters showing the influence of the proximity effect and the long-range photosensitivity, respectively, and σ (x) and p (x) are functions giving the spread of the proximity effect and the long-range photosensitivity, respectively. These are experimentally obtained in advance.

Normally, the spread σ (x) of the proximity effect is 10 μ
m, and the spread p (x) of long-range photosensitivity is several m
It is about m.

As a method for correcting the proximity effect and the long-range photosensitivity, for example, the following method can be considered. D
As (x) = DC _p (x) × DC _f (x), the change in the dose DC _f (x) that corrects the long-range photosensitivity is greater than the change in the dose DC _p (x) that corrects the proximity effect. If you're slow Te, the above equation is a good approximation _{0.5DC p (x) DC f (} x) + ηDC f (x) ∫σ (x-x ') DC p (x') dx '+ θ∫p (x-x It is possible to be ') DC _p (x') DC _f (x ') dx' = 0.5D ₀ (10).

Now, the dose DC for correcting the proximity effect
_p (x) and dose DC for correcting long-range photosensitivity
_f (x) is 0.5 DC _p (x) + η∫σ (x−x ′) DC _p (x ′)
dx ′ = 0.5 ... (11) 0.5DC _f (x) + θ∫p (x−x ′) DC _p (x ′)
DC _f (x ′) dx ′ = 0.5D ₀ (12) The integral included in the equation (12) is DC _f (x) within the area with the area being about 1 mm square.
There is no big error even if it is obtained with constant.

When limited to this 1 mm square area, the term including the integral of the equation (12) is θΣp (x−x _j ) DC _f (x _j ) ∫ _j DC _p (x ′).
It becomes dx '. Here, j is a subscript indicating the region, ∫ _j means the integral within the region j, and the term including the integral of the equation (12) is represented by the sum of the terms including the integral for each region. Becomes

If both sides of equation (11) are integrated in the area where the pattern of this area of about 1 mm square exists, the integration of the second term is a double two-dimensional space integral ∫ _j ∫σ (xx ') DC. It becomes _p (x ') dx'dx. This is similar to the above and becomes ∫ _j DC _p (x ′) dx ′. If the integration area at x ′ is this 1 mm square area, then (0.5 + η) ∫ _j DC _p (x ′) dx ′ = 0.5 × (pattern area of area j) is obtained. That is, ∫ _j DC _p (x ′) dx ′ = (pattern area of region j) /
(0.5 + η), and the equation (12) becomes 0.5DC _f (x _i ) + θ / (0.5 + η) Σp (x _i −x _j ) DC _f (x _j ) × (the pattern of the region j) for the region i. Area) = 0.5D ₀ ... (13) DC _f (x _i ) can be obtained by simultaneous equations (13) for each region on the entire surface of the mask.

Since the size of the mask is larger than the spread of q (x), Σ in equation (13) should be calculated not for the entire region but only for the region within a radius of 30 mm from the region containing x _i. This will make the calculation easier.

In the method described here, DC
_{If f} (x _i ) is θ, η, and the pattern area in each region is known, the fine pattern distribution required when performing proximity effect correction is not required.

Dose DC for correcting long-range photosensitivity
_f (x) is obtained in advance, and the dose DC _p (x) for correcting the proximity effect obtained by calculating the vicinity of the sub drawing area at the time of drawing is calculated, and the long-range exposure given to the sub drawing area is calculated. By multiplying the value of the dose DC _f (x) that corrects the action, both the proximity effect and the long-range photosensitivity can be corrected.

Next, the correction dose of the loading effect is obtained. Here, it is approximated that the dimensional variation depends on the pattern density and does not depend on the pattern shape. In order to simplify the system, it is effective to use the same grid as the pattern density distribution used for the long-range photosensitivity correction.

In the example of FIG. 4, it is assumed that the dimensions change by Δ. In order to correct this dimensional variation, the dose is corrected as shown in FIG.

In the example of the dense pattern area B in FIG. 4, it is assumed that the irradiation amount before the loading effect correction is D (x) and the effective background irradiation amount due to the proximity effect and the long distance photosensing at this time is Db. When the corrected irradiation amount is Da (x), the background irradiation amount is approximated to Db × Da (x) / D (x). Here, it is given by _{Db = 0.5 (D 0 -D (} x)) ··· (15). Therefore, if Da (x) is selected so that Da (x) × (w−Δ) / 2w + Db × Da (x) / D (x) = 0.5D ₀ The resist size varies by -Δ. Since the dimensional variation due to etching is Δ, the two dimensional variations cancel each other out to obtain the desired dimension after etching. Since 0.5D (x) + Db (x) = 0.5D ₀ , the equation (16) is solved and Da (x) = D (x) / (1− (Δ / w) × D (x) / D ₀ ) ... (17) is obtained. By drawing with the dose Da (x),
The desired dimensions are obtained after the etching process.

The pattern density dependence of Δ is obtained by etching resist patterns having different pattern densities.

In the example of FIG. 7, as an effective dose distribution,
The simplest linear distribution was assumed, but of course other distributions may be used. Now, one-dimensionally consider, the position of the pattern end is u = 0. It is assumed that the dose distribution is D (x) s (u). However, s (0) = 0.5, s (∞) = 0,
Let s (-∞) = 1. If the method described above is used, then:

(Da (x) / D (x)) × Db (x) + Da (x) s (Δ) = 0.5D ₀ ... (18) Therefore, Da (x) = D (x) / (1+ (2s (Δ) −1) × (D (x) / D ₀ )) ... (19) The energy distribution s (Δ) given to the resist by the spot beam can be experimentally obtained from the dose dependency of the resist pattern size variation.
For example, assuming a suitable function form such as an error function, the parameters may be determined by drawing the test pattern.

A rectangular pattern is arranged in an array at an equal pitch, and the area is sufficiently larger than the distance affected by the proximity effect and smaller than the distance affected by the loading effect, for example, a size of 200 μm square. Take the area of Sasa. Linear patterns having different densities are drawn in the central portion in a region of about 100 μm. Here, in the drawing of the central pattern, the proximity effect correction and the long distance photosensitivity correction are performed. The representative value of the dimensional variation of the pattern after development is defined as the dimensional variation Δ due to the variation of the loading effect with respect to the peripheral pattern density. As the representative value, for example, it is desirable to use the dimensional variation Δ averaged around the most important density of the central pattern.

The correction method described so far can be realized, for example, by using a system configuration as shown in FIG.

First, by experiment or simulation, a function s representing the pattern density dependence of the size variation Δ due to the variation of the loading effect and the energy beam distribution.
(X) is obtained and the table is stored in the storage means 20.

Next, the drawing data and the reference dose D ₀ are input and stored in the storage means 21 of the data processing computer of the apparatus.

Next, the drawing data stored in the storage means 21 is input to the calculating means 24, and based on this, the entire area is divided into 1 mm square sub drawing areas, and the pattern area density of each sub drawing area is calculated. Ask.

Next, the calculating means 32 solves the equation (13) on the basis of the pattern area density of each sub-drawing region to determine the long-range photosensitivity of each sub-drawing region when the reference irradiation amount is D _0. A table of the dose DC _f (x) to be corrected is created and stored in the storage means 33.

On the other hand, in the drawing system 22, when drawing is started, drawing data is input from the storage means 21 to the figure dividing circuit 23 to cut out each drawn figure.

Next, using a proximity effect correction circuit 34, based on the drawing data of the sub-drawing region near, the dose for proximity effect correction when the base dose and D ₀ DC p _(x)
Is determined using equation (11).

Next, in the loading effect correction circuit 35, the value of the dose DC _f (x) for correcting the long-distance photosensitivity corresponding to the sub drawing area is fetched from the storage means 33, and D (x) = DC _p ( x) × DC _f (x) is obtained, and then the dimensional variation Δ due to the loading effect corresponding to the pattern density is retrieved from the storage means 20. After that, an energy distribution s (Δ) regarding the spot beam is stored in a storage means, and a function s representing the distribution of the energy beam
Obtained from the table of (x). Finally Da (x) = D (x) /
(1+ (2s (Δ) −1) × (D (x) / D ₀ ))
The irradiation amount Da (x) is sent to the drawing circuit 31.

The drawing circuit 31 receives the applied dose Da.
The sub drawing area is drawn by (x). The above process is repeated in all sub drawing areas to draw in all drawing areas. Since the values in the table are discrete, it is effective to use interpolation when obtaining the size variation Δ due to the loading effect and the energy distribution s (Δ) regarding the spot beam based on the table.

Further, instead of obtaining the table of the dose DC _f (x) for correcting the long-range photosensitivity in advance, the pattern density is calculated by obtaining the drawing data slightly ahead of the current drawing during the drawing. By doing so, since the calculation of the pattern density is performed almost in parallel with the drawing, it is possible to eliminate the calculation time required to create the table of the dose DC _f (x) for correcting the long-range photosensitivity. .

So far, the correction has been limited to the loading effect. However, this method is also effective for correcting dimensional variations other than the loading effect.

In the etching apparatus, it is ideal that the etching rate is uniform on the sample surface, but in reality it has a slight distribution. Assuming that the pattern dimension variation due to the non-uniformity of etching is given by the dimension variation Δ regardless of the pattern density, the correction equation (10) in the above discussion is used.
It is possible to correct the dimension by using.

For this, the distribution data of the pattern dimension variation is used in advance, and this is stored in the storage means in the form of a table or a mathematical expression obtained by fitting, and the energy distribution s given to the resist by the spot beam.
When calculating (Δ), the dimensional variation Δ is obtained and the correction amount is calculated.

Further, when both the loading effect and the non-uniformity of etching exist, as shown in FIG. 10, the influence of the two is obtained in advance, and the distribution ΔL of the dimensional variation due to the loading is non-uniform. Distribution of dimensional variation due to sex
The sum ΔL + ΔU with U is again set as Δ, and the correction equation (10)
To obtain the desired dimensions.

Differences from FIG. 9 are as follows. In this example, there is provided a means 36 for storing the size variation distribution ΔU (x) of process non-uniformity obtained by experiments or the like.

Further, the dimension variation Δ due to loading in the example of FIG. 9 is replaced with the distribution ΔL of dimension variation due to loading.

The loading correction circuit 37 is used for long-distance exposure.
Dose DC to correct the application_fTable of (x), for pattern density
Cloth table storage means 33 and non-uniformity dimensional variation distribution ΔU
From (x) the storage means 38, it is possible to sense the long distance of the sub drawing area.
Dose DC to correct light effect _f(X), pattern density,
And the distribution of non-uniformity dimensional variation ΔU (x)
put out.

The loading correction circuit 37 has Δ (x) = Δ
It has a function of calculating L (x) + ΔU (x). Based on the obtained Δ (x), the corrected dose Da (x) is obtained from s (Δ (x)), D (x) and the reference dose D ₀ .

Although not described so far, as is well known, the sign of Δ differs depending on whether the resist is positive or negative. It also depends on the conditions of the etching process. As shown in FIG. 11A, a resist pattern 16 is formed by drawing a pattern on the resist film 16 coated on the sample 13 and developing it by using the above-described pattern drawing method. The resist pattern 16 is a pattern in which the proximity effect, the long distance photosensitivity, and the loading effect are corrected. Subsequently, as shown in FIG. 11B, the resist pattern 16 is used as a mask to etch the sample 13,
Thereafter, by removing the resist pattern 16, a pattern in which the dimensional variation due to etching is corrected with high accuracy can be formed on the sample 13 (FIG. 11).
(See (c)).

Furthermore, this method can be applied to a transfer method using an electron beam. In the transfer method, a pattern is formed by irradiating a sample with a patterned electron beam obtained by scanning the patterned mask continuously or with the electron beam turned on and stepwise moving while turning on and off. Form. Also here, by adjusting the current density of the scanning electron beam or the beam irradiation time, it is possible to give a distribution to the irradiation amount density on the sample and correct the dimensional variation. Even when scanning is performed by moving in steps, it is effective to move the beam positions so as to overlap each other in order to maintain the continuity of the pattern.

In the transfer method, in many cases, the transfer pattern is limited to a small area on the wafer, and the long-range photosensitivity or loading effect and process nonuniformity are caused in a wider range than the area formed by one mask pattern. May appear. In this case, when calculating the correction amount described above, the influence of the adjacent pattern around the pattern data is also calculated.

Although the above description has been limited to the writing using the electron beam, this can also be used in the writing using the laser beam, for example. In this case, it is considered that the proximity effect and the long-range photosensitivity are mainly caused by the beam blur or the leaked beam in the optical system, but the dimensional variation due to the process is the same as that by the electron beam. The laser beam is turned on and off for drawing by using, for example, an acousto-optic device. By adjusting the on time, the irradiation density of the laser beam is adjusted to correct the dimensional variation.

[0108]

As described above in detail, according to the present invention, the dimensional variation due to etching can be corrected with high accuracy.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of an electron beam drawing apparatus.

FIG. 2 is a cross-sectional view of a substrate illustrating a proximity effect.

FIG. 3 is a schematic diagram illustrating a long-range photosensing action.

FIG. 4 is a diagram illustrating a loading effect.

FIG. 5 is a diagram illustrating a conventional method of correcting a loading effect.

FIG. 6 is a diagram showing an example of sample division according to the embodiment of the present invention.

FIG. 7 is a diagram illustrating a loading effect correction method according to an embodiment of the present invention.

FIG. 8 is a flowchart showing a work flow according to the embodiment of the present invention.

FIG. 9 is a flowchart showing a work flow according to the embodiment of the present invention.

FIG. 10 is a flowchart illustrating a work flow according to the embodiment of the invention.

FIG. 11 is a process sectional view of the pattern forming method according to the embodiment of the present invention.

[Explanation of symbols]

1 electron gun 2 Converging lens 3 Molding aperture 4 Projection lens 5 Molding aperture 6 Forming deflector 7 Objective lens 8 reticle mask 9 High-precision main deflector 10 High-speed sub-deflector 11 subfields 12 frames 13 samples 14 incident electrons 15 Backscattered electrons 16 Resist 17 Objective lens 18 Lower surface of objective lens 20 storage means 21 storage means 22 Drawing system 23 Figure division circuit 24 Computing means 25 Computing means 26 storage means 27 Remote Sensitivity Correction Calculator 28 storage means 29 Proximity effect correction calculation circuit 30 arithmetic circuit 31 Drawing circuit 32 arithmetic circuits 33 storage means 34 Proximity effect correction circuit 36 storage means 37 Loading correction means 38 Memory circuit

─────────────────────────────────────────────────── ─── Continued Front Page (51) Int.Cl. ⁷ Identification Code FI Theme Coat (Reference) H01J 37/04 H01J 37/305 B 37/305 H01L 21/30 541M 541S

Claims (8)

[Claims] 1. A pattern drawing method for drawing a pattern by irradiating a resist-coated sample with an energy beam, comprising: drawing data for drawing the pattern; and a reference irradiation amount for irradiating the energy beam. D ₀ , a step of storing the pattern-dependent distribution of the dimensional variation Δ due to the influence of the loading effect, and the energy distribution s of the energy beam given to the resist, dividing the drawing area into a grid pattern, A step of forming, a step of obtaining a pattern area density distribution for each of the sub drawing areas based on the drawing data, and a long distance in the sub drawing area based on the pattern area density and the reference dose D _0. Calculating a dose DC _f (x) for correcting the photosensitivity, and using the drawing data and the reference dose D ₀ . On the basis of,
Calculating a dose DC _p (x) for correcting the proximity effect for the pattern in the sub drawing area; a dose DC _f (x) for correcting the long-range photosensitivity; and a dose for correcting the proximity effect. DC _p (x), determining the dose D (x) based on the pattern-dependent distribution of the dimensional variation Δ and the energy distribution s of the energy beam, and the position and shape of the pattern in the sub drawing area. Determining the irradiation position and shape of the energy beam on the basis of the data on the energy beam, and irradiating the energy beam for a period of time corresponding to the irradiation amount D (x). 2. A step of drawing a pattern on the resist using the pattern drawing method according to claim 1, followed by developing the resist to form a resist pattern, and etching the sample using the resist pattern as a mask. A pattern forming method comprising: a step of forming a pattern on the sample. 3. A step of obtaining a dimensional variation distribution Δ (x) in each of the sub drawing regions from the pattern area density distribution and a pattern dependence distribution of the dimensional variation Δ due to the influence of the loading effect. The dose DC _f (x) for correcting the long-range photosensitivity is
The dose DC _p (x) that is calculated based on the pattern area density distribution, the reference dose D ₀ , and the dimensional variation distribution Δ (x) and corrects the proximity effect is the pattern of the pattern of the sub drawing area. Position and shape data,
The pattern drawing method according to claim 1, wherein the pattern irradiation method is calculated based on the reference dose D ₀ , the dimensional variation distribution Δ (x), and the energy distribution s of the energy beam. 4. The dose D (x) is a product of a dose DC _p (x) for correcting the proximity effect and a dose DC _f (x) for correcting the long-range photosensitivity. The pattern drawing method according to any one of claims 1 to 3, which is characterized. 5. A dose DC for correcting the long-range photosensitivity.
5. The pattern drawing method according to claim 4, wherein the dose D (x) is obtained by assuming that _f (x) is the reference dose D ₀ . 6. The dose D (x) is D (x) = DC _p
(X) × DC _f (x) / (1+ (2s (Δ) −1) ×
The pattern drawing method according to claim 1, wherein the pattern drawing method is obtained by (DC _p (x) × DC _f (x) / D ₀ ). 7. The dimensional variation Δ includes a dimensional variation due to the non-uniformity of the etching rate of the resist in addition to the loading effect.
The described pattern drawing method. 8. A dose DC for correcting the long-range photosensitivity.
7. The pattern drawing method according to claim 6, wherein the dose D (x) is obtained by assuming that _f (x) is the reference dose D ₀ .