JP3536362B2 - Drawing data generation method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a mask EB (Electorn Beam Lili) for verifying a ULSI mask image.
thography drawing) drawing to perform a display in accordance with the data at high speed
The present invention relates to an image data generation method . 2. Description of the Related Art Processing of mask data used for designing a semiconductor device or the like is becoming more and more complicated, and a demand for efficient verification of mask data is increasing. A particularly useful verification tool that meets these requirements is mask EB (Electorn Beamlithography drawing).
There is a data graphical image browser. This browser displays a pattern arranged in a predetermined area in a chip as a graphical image. Further, as EB data becomes more and more complicated, there is an increasing demand for browsing a final mask image. At the same time, if high browsing characteristics can be obtained, the overall time spent designing a semiconductor device can be reduced. In addition, portability of the EB browser is important for enabling the EB browser to be executed on hardware having different CAD systems. Further, there is a problem that a plurality of EB data formats cannot be supported because there is no effective EB data format in mask production. [0004] However, recent EB browsers do not perform appropriate data division and approximation of drawing in displaying a pattern, so that the browsing characteristics are low, and as a result, large-scale ULSI design is not possible. There is a problem that a large amount of data cannot be browsed. Further, this EB browser has a problem that it is highly dependent on the platform and cannot support a large number of EB data. The present invention has been made in view of the above-mentioned problems of the prior art, and is capable of displaying an image corresponding to EB data at a high speed.
It is an object to provide a drawing data generation method . In order to solve the above-mentioned problems of the prior art and to achieve the above object, a drawing data generating method according to the present invention uses a drawing data generating method executed by a computer.
The method comprises the steps of: providing a first partition in a matrix for dividing a chip region in each of the first regions;
Assuming that the pattern density is uniform, the first section
Among the pattern data for displaying the patterns arranged in the chip region, the number of the patterns that do not intersect with the line that defines the first section is set such that the number of the patterns included in between becomes constant. assignment pattern data, each of the chip areas by dividing the first second major matrix from compartment sections, said Ji
The density of the pattern is uniform over the entire
Assuming that the pattern of the pattern included in the second section
Among the patterns that cannot be assigned to the first section because they are located to intersect with the line so that the number is constant, the pattern that does not intersect with the line that defines the second section When the pattern data is allocated, and the result of the allocation is recorded in a recording unit provided in the computer, and when a part of the chip area is displayed in an enlarged manner, the pattern data of the pattern arranged in the part is displayed. Detecting the first section and the second section to which the first section and the second section are assigned, and reading the pattern data of the detected patterns assigned to the first section and the second section from the recording means, This is a method of generating drawing data based on the pattern data that has been drawn. The trapezoid drawn by the pixel approximate to a rectangle, Ripetishon region drawn by the pixels included in the predetermined region is approximated to a single rectangular box area, drawn by a predetermined number or less of pixels,
A collection of said patterns that can be referenced as a single entity
In this case , the drawing data is generated by approximating a single rectangular box area. According to the drawing data generating method of the present invention, first, among the pattern data for displaying the pattern arranged in the chip area, each of the first sections of the matrix for dividing the chip area is displayed. , The pattern data of the pattern not intersecting with the line defining the first section is allocated. Then, since the chip area is divided and located in each of the matrix-like second sections larger than the first section and intersects with the line, the chip area cannot be allocated to the first section. Among the patterns, the pattern data of the pattern not intersecting with the line defining the second section is allocated. Then, the result of the allocation is recorded in the recording means. Then, when displaying a part of the chip area in an enlarged manner,
The first section and the second section to which the pattern data of the pattern arranged in the portion are allocated are detected, and the detected pattern of the pattern allocated to the first section and the second section is detected. The pattern data is read from the recording unit, and drawing data is generated based on the read pattern data. In this case, the pattern drawn by the predetermined number of pixels or less is ignored, the trapezoid drawn by the predetermined number of pixels or less approximates a rectangle, and the repetition area drawn by the pixels included in the predetermined area is The pattern approximated to a single rectangular box area and drawn by a predetermined number or less of pixels, which can be referred to as a single entity, generates the drawing data by approximating to a single rectangular box area. Hereinafter, a graphical image browser according to an embodiment of the present invention will be described. FIG. 1 is a configuration diagram of a graphical image browser 1 according to the present embodiment. As shown in FIG. 1, a graphical image browser 1 according to the present embodiment includes a data conversion unit 2, a division processing unit 3, a mask image creation unit 4, a database 5, and an operation unit 9. The data converter 2 receives the EB pattern data and converts the EB pattern data into intermediate data 7. EB
As a format of the pattern data, for example, a drawing pattern format JEOL52 for an EB drawing system manufactured by JEOL Ltd., or Etec Systems I, USA
nc. For example, a drawing pattern format MEBES for an EB drawing system manufactured by the company is used. The EB pattern data consists of a series of EB lithography drawing commands based on geometric patterns such as rectangles and trapezoids. That is, concepts such as the position of the gate and the position of the contact hole in the chip are uniformly grasped as a graphical mask image when verifying the ULSI mask image. Therefore, in the EB pattern data, it is not necessary to use the concept such as the position of the gate. Note that the EB pattern data does not include information on colors. The intermediate data 7 is generally in a human-readable format, which makes it easier to verify the pattern data. Further, the intermediate data 7 can represent various types of patterns and a hierarchical structure based on a general format of EB pattern data, and can support a plurality of EB pattern data. That is, when the EB pattern data is converted into the intermediate data 7 to convert the EB pattern data corresponding to each of the plurality of EB devices into display data unique to various display devices, the number of conversion programs is significantly reduced. can do. As described above, the EB pattern data does not include information on colors, but the intermediate data 7 has information on colors and types (styles). The intermediate data 7 is preferably in a format that can incorporate information about color and type. These colors and types are used to more clearly distinguish pattern data of a graphical image by assigning different colors to different mask layers, for example. In order to hierarchically represent the pattern represented by the intermediate data 7 , a single pattern shown in FIG.
Collection of objects 21 that can be referenced as entities
In addition, the concept of repetition is used. In here, the object 21 is a pattern or library. As shown in FIG. 3A, the repetition 22 describes a repetition of the object 21 according to a two-dimensional matrix. The division processing unit 3 reads the intermediate data 7 from the data conversion unit 2, divides the read intermediate data 7 into smaller areas, and temporarily stores the result in the database 5. As described above, the technique of dividing data into smaller areas is generally used and effective as a technique for solving a problem of handling a very large amount of pattern data. In the division processing section 3, computational complexity is using a split method is a (K logN). Here, K indicates the number of sections, and N indicates the total number of patterns. Further, in the division processing unit 3, it is assumed that the density of the rectangle is uniform in the entire chip area.
Under such an assumption, a very high-speed search performance can be obtained by selecting sections of equal size including a predetermined number of patterns. For example, as shown in FIG. 2, the chip pattern data 10 represented by the intermediate data 7 is divided by the matrix 11 like the partition pattern data 12. An example in which the chip area is divided into 3 × 3 matrix sections will be described below. FIG. 3 (B)
FIG. 3 is a diagram for explaining a case where a chip area is divided into 3 × 3 matrix-shaped section groups. As shown in FIG. 3B, objects 31a to 31
f is arranged. First, the chip area 30 is divided into sections 41 to 49 by a matrix-shaped section group 32, and an object is allocated to each section. At this time, the objects 31b, 31c, 31e, 31f of the chip area 30
Intersects with the line of the section group 32, so that the assignment to the section is not performed at this time. Here, the objects 31a and 31d are allocated to the sections 41 and 46, respectively. Next, the chip area 30 is divided into sections 50 to 53 by a section group 33, and objects not allocated by the section group 32 are assigned to each section. At this time, since the object 31c of the chip area 30 intersects with the line of the section group 33, the assignment to the section is not performed. At this time, the objects 31b, 31e,
31f is allocated to the sections 51, 52, 53, respectively. Next, the object 31c remaining until the end is assigned to a section 54 of the section group 34 having the same shape as the chip area 30. The number of patterns included in a single section is constant (limited). Therefore, if the size of the chip is specified, the total number of divisions can be calculated from the number of given patterns. That is, the division parameter is given by the following equation (1). Hereinafter, a method for roughly calculating the total number of divisions will be described. When dividing the chip region, m is the level of division, n is the number of final divisions, Rm is the number of patterns remaining after performing the m-th division, B is the number of patterns included in each section, When Km indicates the number of partitions when the m-th division is performed, the following expression is established. [0018] In equations (2) and (5), m = n,
By further transforming the equations, the following equations (6) and (7) hold. Here, the final value of R _n is ideally B
Becomes equal to Therefore, R _n can be replaced as in the following equation (8). When equation (8) is further modified, equation (9) is obtained. From equation (9), it can be seen that the final number of divisions n is determined by a function having (logN) as a variable. When the sum of the number of relatively small patterns is large, the distribution of patterns to each section is practically substantially constant. When dividing the format of the intermediate data 7 having a hierarchical structure, it is important to keep the format structure of the intermediate data 7 unchanged so as to make the data as compact as possible. In order to divide the above-mentioned two structural elements, those elements are treated as a rectangle by the following method. That is, FIG.
As shown in (A), the library 20 is handled as a rectangular box area 23, and the repetition 22 is handled as a rectangular box area 24. When the library contains a large sum of pattern data, it can be referred to as a single entity.
In order to subdivide the set of objects themselves, a new level of division is introduced and a hierarchy of divisions is created. The database 5 stores the results of the division processing by the division processing unit 3 described above. In this database 5, pattern data belonging to (assigned to) each section is stored in a high-speed accessible state. Further, the database 5 stores a look-up table used to search for pattern data belonging to each section. By the operation of the operation unit 9 by the user,
When the browser instructs to zoom (enlarge) and display the chip area, only the pattern belonging to the section corresponding to the zoom area is required to be displayed on the display 8. Database 5 because of the large amount of pattern data
Is generally implemented as an external file. for that reason,
It is not necessary to load all the pattern data into the memory. As a result, a memory having a relatively small storage capacity may be used for allocating the database lookup table, and there is almost no need to limit the number of patterns. The mask image creating section 4 creates drawing data according to the user's request in response to the operation of the operation section 9 by the user, and outputs this drawing data S4 to the display 8. At this time, the mask image creation unit 4 reads out pattern data necessary for the creation from the database 5. For example, if the user requests to enlarge and draw an area corresponding to the section 41 shown in FIG.
The pattern data belonging to 1, 50, 54 is read from the database 5. In order to enable a high-speed search for a pattern in the chip area, the mask image creation unit 4
The drawing data S4 is created using the following drawing approximation rules to reduce the drawing time. (1) Small patterns that are hardly visible are ignored. (2) A small unequal quadrilateral approximates a rectangle. This is because in most graphical packages, the drawing speed of the inequilateral quadrilateral is slower than the drawing speed of the rectangle. (3) A small repetition area approximates a single rectangular box area. (4) Objects that can be referenced as a single entity
A small set of objects approximates a single rectangular box area. Since the approximation in drawing distorts the mask image, the user can determine the degree of approximation by specifying (inputting) the degree of approximation by operating the operation unit 9. Next, a method for creating the drawing data S4 will be described. For example, patterns 70, 71, 72 as shown in FIG. 5 are formed on a semiconductor chip. Here, the pattern 70 is formed on the pattern 72,
The pattern 71 is formed on the same layer as the pattern 70. At this time, the mask image creation unit 4 creates the drawing data S4 so that the patterns 70 and 71 are displayed in red and the pattern 72 is displayed in blue. Further, the mask image creating section 4 includes a pattern 70 and a pattern 72.
And the intersection 74 between the pattern 71 and the pattern 72 are displayed by mixing, for example, blue, dots, and red dots. Thus, the user looks at the display screen, and the intersection 73 becomes the pattern 70, 71.
And the pattern 72 can be easily recognized as a crossing portion, and the efficiency of pattern verification can be increased. The processing in the graphical image browser 1 shown in FIG. 1 will be described with reference to the flowchart shown in FIG. Step S1: The EB pattern data is input to the data converter 2, and the intermediate data 7 is generated.
The intermediate data 7 is output from the data conversion unit 2 to the division processing unit 3. Step S2: In the division processing section 3, the intermediate data 7 is divided as described above, and the result of the division processing is stored in the database 5. At the same time,
A look-up table 6 is generated and stored in the database 5. Step S3: The lookup table 6 is read out and stored in a memory which can be accessed at high speed. Step S4: An operation signal S9 corresponding to the operation of the operation unit 9 by the user is input to the mask image creation unit 4, and if this is not a cancellation request, the process of step S5 is executed. To cancel the processing. Step S5: As described above, the drawing data corresponding to the user's request in step S4 is created in the mask image creating section 4. At this time, pattern data necessary for the drawing data is read from the database 5. The evaluation of the above-described graphical image browser 1 will be described below. FIGS. 6 and 7 are images when browsing a chip composed of 2.73 million patterns using a Unix work station with a processing speed of 25 mips. Here, FIG. 6 is an image when the drawing approximation rule by the graphical image browser 1 is used, and FIG. 7 is an image when the drawing approximation rule is not used. At this time, the image shown in FIG.
It consisted of multiple patterns, and it took less than 1 second to display the entire pattern. At this time, patterns smaller than one pixel can be ignored, and reference can be made to smaller trapezoids and repetitions of 5 × 5 pixels, and a single entity.
A set of various patterns is drawn as a rectangle in an area smaller than 44 × 44 pixels. On the other hand, the image shown in FIG. 7 is composed of 2.73 million patterns, and it took about 2 minutes and 40 seconds to display the entire image. Similarly, FIG. 8A is an image when the drawing approximation rule by the graphical image browser 1 is used, and FIG. 8B is an image when the drawing approximation rule is not used. In the case where the display shown in FIG.
A trapezoid smaller than × 10 pixels was drawn as a rectangle. At this time, displaying the image shown in FIG. 8A was instantaneous, but it took a few seconds to display the image shown in FIG. 8B. However, as shown in FIG.
When the drawing approximation rule is used, a slightly jagged portion 60 occurs. If the above-described graphical image browser of this embodiment is created using Xwindows and the C language, Xwindows will become widely used as a graphical interface for CAD systems. Can be increased. According to the drawing data generating method of the present invention,
An image corresponding to the electron beam drawing data can be displayed at high speed. Further, according to the drawing data generation method of the present invention, by using the intermediate data, the number of conversion programs for converting the electron beam drawing data unique to the EB device into the display data unique to the display device can be reduced. Also, the drawing data of the present invention
According to the generation method , the user can easily recognize the intersection of the patterns formed in the plurality of layers.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a configuration diagram of a graphical image browser according to an embodiment of the present invention. FIG. 2 is a diagram illustrating a case where intermediate data is divided. FIG. 3 (A) can be referred to as a single entity
FIG. 4B is a diagram for explaining the concept of a set of objects and repetition, and FIG. 4B is a diagram for describing a case where the object is divided into 3 × 3 matrix sections. FIG. 4 is a diagram for explaining processing in the graphical image browser shown in FIG. 1; FIG. 5 is a diagram for explaining a case where an intersection is displayed for a pattern of a plurality of layers. FIG. 6 is an image when browsing a chip composed of 2.73 million patterns using a Unix work station with a processing speed of 25 mips, and is an image when a drawing approximation rule is used. FIG. 7 is an image when browsing a chip composed of 2.73 million patterns using a Unix work station with a processing speed of 25 mips, and is an image when a drawing approximation rule is not used. 8 (A) in the case of using the drawing approximation rules, (B)
Is an image when the drawing approximation rule is not used. [Description of Signs] 1 ... Graphical image browser 2 ... Data conversion unit 3 ... Division processing unit 4 ... Mask image creation unit 5 ... Database 6 ... Lookup table 7 ... Intermediate data 8 Display 9 Operation units 41 to 54 Partition

────────────────────────────────────────────────── ─── Continuation of front page (56) References JP-A-2-201577 (JP, A) JP-A-63-7632 (JP, A) JP-A-5-324766 (JP, A) (58) Field (Int.Cl. ⁷ , DB name) G06F 17/50 658

Claims (1)

(57) [Claim 1] A drawing data generating method executed by a computer.
Therefore, the density of the pattern in the entire chip region is increased in each of the first matrix-shaped sections for dividing the chip region.
Assuming that it is uniform, the first section
Allocate the pattern data of the pattern that does not intersect with the line defining the first section, among the pattern data for displaying the pattern arranged in the chip area, so that the number of patterns is constant , in each of the chip areas by dividing the first large matrix of the second zone than compartments, the chip area total
Assuming that the density of the pattern in the body is uniform
The number of the patterns included in the second section is constant
As will be, among the patterns that can not be performed the allocation of the to the first compartment in order to position intersects with the line, the pattern of the pattern which does not intersect the line which defines the second section Allocating data, recording the result of the allocation in a recording means provided in the computer, and displaying the pattern data of the pattern arranged in the portion when the portion of the chip area is enlarged and displayed. Detecting the first section and the second section, and reading the pattern data of the pattern allocated to the detected first section and the second section from the recording means, and reading the read pattern data. Is a method of generating drawing data based on the above, ignoring the pattern drawn by a predetermined number of pixels or less, and A trapezoid drawn by an element approximates a rectangle, and a repetition area drawn by pixels included in a predetermined area approximates a single rectangular box area, and is drawn by a predetermined number or less of pixels. A drawing data generation method for generating the drawing data by approximating a set of the patterns that can be referred to as a single entity to a single rectangular box area.