JP3535399B2 - Mask drawing data creation method - Google Patents

Description

Detailed Description of the Invention

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor manufacturing process and a photomask used therein, and more particularly to a mask drawing data creating method used in a semiconductor manufacturing process suitable for forming a fine pattern.

2. Description of the Related Art As LSIs are highly integrated, pattern sizes are being reduced and design data is increasing. In particular, when the optical proximity correction technology, which has been actively studied in recent years, is applied, the data of the photomask and the number of very small patterns included in the photomask are further increased. The optical proximity effect correction (OPC) is a mask pattern that is modified in advance in anticipation of the proximity effect in order to correct the optical proximity effect in which the pattern is not finished as desired due to the effects of light diffraction, resist development, etching, etc. It is a correction method to be set. This O
As a result of PC, the figure included in the mask is shown in FIG.
The shape shown in (a) may be included. Figure 1
(A) has minute projections and depressions, which cannot be faithfully reproduced by a laser beam mask drawing apparatus or an electron beam mask drawing apparatus. In other words, while the drawing data increases due to the minute protrusions and depressions, the minute protrusions and depressions are not reproduced in the shape of the light shielding pattern of the completed mask. Further, FIG. 1 shows the result of dividing FIG. 1A into shots of a variable shaped beam of an electron beam drawing apparatus.
It shows in (b). Although minute shots are generated around the minute protrusions and depressions, minute shots having a resolution lower than that of the drawing device cannot be drawn as desired because the dose control is not properly performed. That is, the conventional method has a problem that the accuracy of the pattern formed on the mask is deteriorated when the mask pattern includes a large number of minute patterns having the resolution of the mask drawing device. OP
As a problem in performing C, the amount of pattern data increases in addition to the generation of minute figures. The increase in the amount of data compared to before and after correction is the so-called 1
It is said that the dimensional correction is about twice, and the so-called two-dimensional correction with serifs for correcting the corner rounding is about 4 to 7 times. As the amount of correction data increases, the time for mask data processing increases, the mask drawing time increases (in the case of a vector scan mask drawing apparatus), and the correction processing time for the proximity effect in electron beam drawing increases. Dots occur. Of these, since the mask data processing can be performed off-line, the entire process is not rate-limited, but when the mask drawing time and the proximity effect correction processing are carried out in parallel with the drawing, these processing times increase. As it is, increase the turnaround time of the whole process. On the other hand, various methods have been proposed as a method for correcting the proximity effect caused by the backscattering of the electron beam, but the following methods are currently the mainstream methods. The first method is the Journal of Appli
There is a method called Ghost method described by G. Owen et al. in ed Physics vol. 54, p357 (1983). In this correction method, a target pattern is drawn with a normal beam narrowed down, and then the beam is blurred to the extent of backscattering, and a black-and-white inverted pattern is added with this beam. This additional exposure allows the background to remain constant at all locations. The second method is the Journal of Vacuum Science and Technology vol.
There is a method called dose correction described by M. Parikh et al. In 15, p931 (1978). In this correction method, the energy density at each exposed resist point is made constant by changing the dose of the element figure drawn by the electron beam.

In the prior art, in particular, when OPC is performed, minute protrusions or depressions are formed in the mask pattern, and when the protrusions or depressions are less than the resolution of the drawing apparatus, the protrusions or depressions are generated. The amount of data was increasing although it was not reproduced by the drawing device. In particular, when a variable shaping electron beam drawing apparatus is used as a mask drawing apparatus, dimensional accuracy in drawing is deteriorated due to minute shots generated around minute protrusions and depressions. Also,
When a mask subjected to OPC is subjected to proximity effect correction based on the Ghost method, data for additional exposure is created by inverting the pattern after OPC in black and white, but the inversion data of a complicated pattern after OPC is corrected by OPC. It is more complex than the previous inversion data and the amount of data is increasing. Therefore, the number of shots of additional exposure increases and the mask drawing time increases. In FIG. 2, (a) data before optical proximity correction,
(B) Optical proximity effect correction data, (c) Additional irradiation data (black and white inversion data) are shown. There are 723 shots only in the area in (c). Alternatively, when the mask data subjected to the optical proximity effect correction is created by performing the proximity effect correction based on the second method (irradiation amount correction), the optimum dose amount of each shot is calculated based on the mask data. When the mask data becomes complicated by the optical proximity correction, the amount of input data for the dose correction calculation increases and the calculation time also increases. The present invention has been made in consideration of the above circumstances,
It is an object of the invention to provide a data processing method for creating mask drawing data that does not adversely affect mask drawing accuracy even if a fine pattern is generated by optical proximity correction. Another object of the present invention is to provide a method for efficiently creating mask drawing data without increasing the calculation time of proximity effect correction in charged particle beam exposure, drawing time, and the memory requirement of hardware used for calculation.

The present invention for solving the above problems adopts the following configuration. (1) In the method for creating drawing data of a mask used in lithography, when the difference between X coordinates or Y coordinates of two or more vertices defined in the pattern data is larger than 0 and falls within a predetermined range , It is characterized in that a plurality of coordinates are replaced with one representative coordinate. (2) The representative coordinate is the average value of the plurality of coordinates or a point on the data grid closest to the average value.

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 3 is a flow chart for explaining the mask drawing data creating method of the present invention. After the layout design is completed (STEP0), pattern data processing 1 (STEP1), which is a graphic calculation processing, is performed, optical proximity correction (STEP2) is further performed, and then a graphic simplification processing is performed (STEP3). After that, pattern data processing 2 considering the mask process is performed (STEP
4) Format the drawing data (STEP 5).
The simplification processing of STEP3 may be performed anywhere after the layout design and before the mask drawing data is generated, but it is preferable to perform the simplification processing in the upstream as much as possible because the amount of processing data in the subsequent processing can be reduced. . Then, STEP3 is demonstrated in detail. Simplification processing is performed on the figure group shown in FIG. FIG. 4B shows the result of decomposing this figure group into directed line segments counterclockwise.
Non-axis parallel line segments 3, 7, 10, 14, 20, 24, 2
Since 6, 28 and 30 are sufficiently short, they are replaced with patterns parallel to the axis for simplification. Focusing on the line segment 20, the line segments 19 and 21 connected to this line segment form an angle of 90 ° with each other.
Is. In this case, the intersection of the line segments 19 and 21 is set as a new vertex. Attention is paid to the line segment 14. The coordinates of the start point and the end point of the line segment 14 are (x1 y1) and (x2 y2), respectively. Since the two line segments connected to this line segment are both parallel to the X axis, the vertices of (x1 y1) and (x2 y2) are deleted and a new ((x1 + x
Add 2) / 2 y1), ((x1 + x2) / 2 y2) vertices. Focusing on line segment 3, since the two line segments connected to this line segment are parallel to the Y axis, the vertices of (x1 y1) and (x2 y2) are deleted and a new (x1 (y1 + y2) Add vertices of / 2), (x2 (y1 + y2) / 2). Similar processing is performed for other shaded segments. Figure 4
A pattern after the above simplification is shown in (c).
FIG. 4D shows the resist pattern shape formed on the wafer when the pattern before simplification is exposed. This is the result obtained by the exposure simulation. FIG. 4 (e) shows the result of simulation performed on the simplified pattern. In the simplification according to this embodiment, there is almost no difference in the resist size, and the difference in the resist size before and after the simplification is within the dimensional error allowed in the resist pattern. FIG. 4F shows the result of dividing the pattern before the simplification into shots of the variable shaped beam, and FIG. 4G shows the result of dividing the pattern after the simplification into shots. In FIG. 4F, a minute shot is generated around the minute diagonal line.
In (g), a minute shot does not occur. In simplification
May make the following judgment on whether or not simplification is possible.
Yes. As shown in the example of FIG. 5, pattern data including line segments that are not axially parallel but have various angles and lengths are prepared in advance, a mask is created, and the pattern is transferred onto a wafer. The length of a pattern on a mask or wafer is measured to find the length and angle of a line segment that is not parallel to the axis when the dimensional error falls within the allowable range. As in the example of FIG. 5, both the patterns before and after the simplification may be prepared, and the finishes on the mask and the wafer may be compared before and after the simplification. A simulator may be used instead of actually transferring a pattern to a mask or a wafer. At the time of simplification, it is determined whether or not the simplification is performed by referring to the lengths and angles of the line segments that are not axis-parallel obtained in advance. Another criterion for determining whether or not simplification is possible will be described. Non-axis-parallel line segments 3, 7 included in FIG.
Out of 10, 14, 20, 24, 26, 28, 30
With respect to 3, 7, 10, and 14, since there is no change in the area of the pattern before and after simplification, it is determined that simplification is possible. The pattern areas of the remaining 20, 24, 26, 28 and 30 change due to simplification, but the change amount is +
Since it is as very small as 0.0009 μm 2, it is not affected by simplification. The change amount of the area that does not affect the finished size may be obtained in advance by experiments or simulations as described in claim 3. Still another simplification determination method will be described with reference to FIG. The simplification candidate pattern or a part of the pattern is extracted, and the pattern and its peripheral pattern are input to the simulator to perform mask exposure and development simulations. If the pattern obtained by the simulation is within the allowable dimensional error required for the mask, it is determined that simplification may be performed (STEP 2), and simplification candidates are simplified (STEP).
5). If the mask simulation result is not within the allowable dimensional error required for the mask, further wafer exposure / development simulation is performed (STE).
P3). If the dimension of the predicted pattern on the wafer obtained by the wafer exposure / development simulation falls within the allowable dimension error required for the wafer pattern, it is determined that simplification may be performed (STEP 4), and simplification is performed (STEP 4). STE
P5). If it does not fall within the allowable dimensional error, simplification cannot be performed. Here, it is determined whether the simplified mask pattern is within the allowable dimensional error (STEP1, STEP
Although both 2) and the simplification of the wafer pattern within the permissible dimensional error are determined (STEP 3, STEP 4), only one of them may be performed. Next, at the stage where the proximity effect correction in step 2 of FIG. 3 is completed, FIG.
The case in which the layout is obtained will be described. In general proximity effect correction software, the width and length of the minimum jog can often be specified, which can prevent generation of a minute pattern that cannot be drawn by the mask drawing apparatus to some extent. However, this function is not sufficient, and when the jogs on the opposite sides are deviated by a minute distance as shown in FIG. 7, as a result of shot division of the variable shaped beam, a minute shot is generated as shown in FIG. 7C. Will end up.
Therefore, in the layout of FIG.
Search for coordinates that have similar coordinates and Y coordinates. Whether or not they are close to each other is determined in consideration of the minimum shot size of the variable shaped beam that can be stably shaped. The minimum shot size that can be stably shaped with the variable shaped beam is 0.2 μm, and this is 50 nm when converted on the wafer (in the case of a tetraploid). If a shot smaller than this size occurs, the accuracy may be rather deteriorated. Therefore, a plurality of coordinates having a difference within 50 nm are replaced with the same representative coordinates. At this time, the average value of the plurality of coordinates may be used as the representative coordinate. For example, X coordinates x1 of vertices 1 and 2 in FIG.
Since the X coordinate x2 of the vertices 3 and 4 has a difference of 10 nm, it is replaced with the representative coordinate. The result of using the average value of the X coordinates of these vertices as the representative coordinates is shown in FIG. When dividing the shots of the variable shaped beam at the coordinate position of either the X coordinate or the Y coordinate, the above processing may be performed only for the coordinates (either X or Y) to be used. Further, in the example of FIG. 7, the coordinates in which the coordinates are close to each other are replaced with the representative coordinates. However, when a pattern in which the graphic is inverted in black and white is drawn, the vertices of the adjacent graphics have a sufficient X coordinate or Y coordinate. For the close vertices, the operation of replacing the X coordinate or the Y coordinate of the vertices with the representative coordinate may be performed. FIG. 7D shows the result of shot division of the simplified result of FIG. 7B in parallel with the Y axis. When the simplification is performed, the number of shots is reduced from 3 to 2, and the minute shots smaller than the minimum shot size are not included. Next, let us consider the error due to simplification. With simplification (Fig. 7 (e)) and without simplification (Fig. 7 (f)),
The result of simulating the finished shape of the mask is shown. When the dimensions are measured at several points in the simulation result, the error is about 1 nm, and it can be seen that the dimensions are well within the dimensional accuracy required for the mask. That is, it can be seen that the effect of simplification does not matter. Next, another example of simplification of step 3 in FIG. 3 will be described. As described above, when the correction is performed using the general proximity effect correction tool, the edges are divided so that the jog having the specified length or less does not occur as much as possible. However, it is often the case that the remainder obtained by dividing the edge at equal intervals becomes a minute protrusion, or the correction portion of the slant line becomes a minute recess. Such minute protrusions and recesses cannot be reproduced by a mask drawing apparatus, and in the case of a variable shaped beam drawing apparatus, they may become minute shots, resulting in a decrease in drawing accuracy. Therefore, in the present invention, fine irregularities are removed by the following method. Figure 8 shows the procedure for erasing minute irregularities.
, And will be described according to this procedure. Here, FIG. 9A shows an example of a layout in which minute irregularities are present. First, a Δd1 thickening process is performed to erase the minute recesses (STEP 0). Here, thick sizing of 10 nm was performed. Δd1 may be set to half the width of the recess to be erased. The result of the thick sizing process is shown in FIG. Then, if this is narrowed by Δd1 and sizing,
As shown in (c), the figure has the same size as the original pattern, and the minute concave portions disappear (STEP 1). Next, in order to remove the minute protrusions, a Δd2 narrowing sizing process is performed (STEP 2). Here, fine sizing of 10 nm was performed. Δd2 may be set to half the width of the minute protrusion to be erased. The result of the narrowing sizing process is shown in FIG. Then, if this is thickened by Δd2, FIG.
As shown in (e), the figure has the same size as the original pattern and the fine protrusions disappear (STEP 3). The procedure to remove the protrusions and depressions this time is STEP0 → STEP1 → ST
I performed in the order of EP2 → STEP3, but STEP2 → ST
You may perform in order of EP3->STEP0-> STEP1.
This time, we removed both the small depressions and the minute protrusions. If only the minute depressions need to be removed, STE
P0 → STEP1 may be performed, and STEP2 → STEP3 may be performed when only the minute protrusions need to be removed. FIG. 9F shows the result of simulating the resist shape when transferred to the wafer using the mask before simplification, and FIG. 9G shows the result of simulating the resist shape after simplification. There is no difference between the two, and even if the minute protrusions or dents are removed to simplify the figure, it does not affect the accuracy of the pattern formed on the wafer. 10 and 11 show the present invention.
FIG. 6 is a diagram for explaining an example of FIG. That is, this fruit
In the example, a new vertex is generated in the pattern data.
Other pattern data when performing the processing
Is shown. An optical proximity effect correction process is a typical process for generating a new vertex in pattern data. FIG. 10 shows a flow for generating vertices so that minute figures are not generated in the optical proximity effect correction. An example of pattern data is shown in FIG.
Input the pattern data for the proximity effect correction process, and set the division points where the new vertices are generated (ST.
EP0). There are various methods for setting the division points, such as a method of dividing at a constant interval and a method of dividing at a portion facing the apex. In the example of FIG. 11B, the division point is set at the portion facing the apex. Then, for each division point, it is checked whether or not the difference in X coordinate (or Y coordinate) from other vertices or other division points is smaller than a threshold value. The threshold value used here is determined in consideration of the minimum shot size of the variable shaped beam described in the second embodiment. In the example of FIG. 11B, the end point of the boundary line segment 7 (start point of the line segment 8) (X1 Y1)
, And the difference in the X coordinate at the start point of the boundary line segment 11 (end point of the line segment 10) (X2 Y2) is sufficiently small, the X coordinate of these two dividing points is replaced with the representative coordinate, and ((X1 + X
2) / 2Y1) and ((X1 + X2) / 2Y2) (FIG. 11 (c), STEPs 1 and 2). Subsequently, each boundary line segment is moved by a correction amount in the direction perpendicular to the direction of the line segment (STEP 3). The correction result is shown in FIG. Further, FIG. 11E shows a result of dividing the result of FIG. 11D into shots of a variable shaped beam. For comparison, FIG. 11F shows the result of shot division performed by performing the optical proximity correction without applying the present invention. When the present invention is not applied, minute shots are generated, but by applying the present invention, the generation of minute shots can be avoided.

According to the present invention, by simplifying a pattern having a fine texture exceeding the resolution limit of a mask drawing apparatus, the amount of data can be reduced without degrading the accuracy of mask making. Can be done. In particular, when a variable shaping electron beam exposure apparatus is used, it is possible to reduce minute shots and prevent deterioration of dimensional accuracy due to minute shots.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a conventional technique.

FIG. 2 is a diagram illustrating a conventional technique.

FIG. 3 is a flowchart for explaining a mask drawing data creation method of the present invention.

FIG. 4 is a pattern diagram for explaining the mask drawing data creation method shown in FIG.

5 is a diagram for explaining the mask drawing data creation method shown in FIG. 3, and is an example of a mask pattern used in an experiment for obtaining a rule of simplification propriety.

FIG. 6 is a diagram for explaining the mask drawing data creation method shown in FIG. 3, and is a flowchart in the case of determining whether or not simplification is possible by simulation.

FIG. 7 is a diagram for explaining another example of the mask drawing data creation method of the present invention.

FIG. 8 is a diagram for explaining another example of the mask drawing data creation method of the present invention, and is a data processing flowchart when removing minute protrusions and minute depressions.

FIG. 9 is a diagram for explaining another example of the mask drawing data creation method of the present invention.

FIG. 10 is an embodiment of a mask drawing data creation method of the present invention .
It is a diagram for explaining the
It is a data processing flowchart when generating a new vertex.
It

FIG. 11 is an embodiment of a mask drawing data creation method of the present invention .
FIG. 3 is a pattern diagram for explaining

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Sachiko Kobayashi 1 Komukai Toshiba Town, Saiwai-ku, Kawasaki City, Kanagawa Prefecture Toshiba Research & Development Center (72) Inventor Shigehiro Hara Komukai Toshiba, Kawasaki City, Kanagawa Prefecture No. 1 in the Town, Toshiba Research & Development Center (72) Inventor Higurashi, etc. Komukai, Komukai, Kawasaki-shi, Kanagawa No. 1 in the Toshiba R & D Center, Toshiba (72) Inventor Uno Ko, Sachi-ku, Kawasaki, Kanagawa Muko Toshiba-cho No. 1 In Toshiba Research and Development Center Co., Ltd. (56) Reference JP-A-7-153676 (JP, A) JP-A-8-306608 (JP, A) JP-A-20-202013 (JP, A) Tokushuhei 11-509006 (JP, A) International Publication 96/005224 (WO, A1) (58) Fields investigated (Int.Cl. ⁷ , DB name) G03F 1/08 H01L 21/027

Claims (2)

(57) [Claims] 1. Drawing of a mask used in lithography
In the data creation method, set it on the line segment of the pattern data.
Of the two or more defined vertices, it is set on the line segment parallel to the x- axis.
If the difference in the x- coordinates of the defined vertices is greater than 0,
A set of x-coordinates at the vertices if they fall within the range
With one representative coordinate, or a line parallel to the y- axis
The y- coordinate difference of the apex defined above is greater than 0.
If it falls within a predetermined range, one pair at the apex
Characterized in that the y coordinate of is replaced with one representative coordinate
Mask drawing data creation method. 2. The representative coordinate is an average value of the set of coordinates.
Or at the point on the data grid closest to this mean
The mask drawing data according to claim 1, wherein
How to make.