JP5497503B2 - Repetitive pattern extraction method from pattern data - Google Patents

Description

  The present invention relates to a method, apparatus, and program for creating CP (Character Projection: partial batch exposure) drawing data used when manufacturing a semiconductor integrated circuit. In particular, even for drawing data for semiconductor integrated circuit design that includes a large number of patterns of various shapes and sizes having minute shape differences and dimensional differences, the number of shots at the time of drawing is reduced and the drawing time is reduced. The present invention relates to a technique for extracting a highly versatile repeating pattern that can be shortened.

  As a measure to shorten the time to create a semiconductor integrated circuit (improve throughput), for example, parallel processing of drawing by multi-beam, multi-column, etc. that repeats drawing and moving at the same time in multiple areas divided equally using multiple electron guns And partial batch exposure technology (CP drawing technology) that uses repetitive patterns that often appear in circuit patterns. An example of a conventionally known apparatus or technique using this CP drawing technique is, for example, the apparatus described in Patent Document 1 or Patent Document 2 shown below, or described in Non-Patent Document 1 or Non-Patent Document 2. There is technology.

  In general, a circuit pattern of a semiconductor integrated circuit is divided into a large number of rectangles (fracturing), and each rectangle is drawn separately according to one shot (VSB drawing). However, among the circuit patterns divided into a large number of rectangles, there are a plurality of patterns having a combination of a plurality of the same rectangles and having the same overall outer shape and size.

  Therefore, in the apparatuses shown in Patent Document 1 and Patent Document 2 described above, a repeated pattern is extracted for each pattern having the same shape and size, and the repeated pattern is represented by one figure (for example, “E” type, “H” type, Is prepared in advance as a mask that is regarded as an “I” type or “L” type character figure for convenience in this specification. Based on this, even if it is an overall pattern composed of a combination of the plurality of rectangles that would normally require a plurality of shots in order to individually draw each rectangle (for example, the entire pattern is an “E” type) In the case of (4), four shots are required consisting of a combination of four rectangles), and the entire pattern can be drawn together in one shot, thereby efficiently drawing the number of times (number of shots) ) To reduce drawing time. In addition, the more the entire patterns having the same shape and size exist in the circuit pattern, the more the number of drawing can be reduced.

  By the way, in recent years, with the miniaturization of semiconductor integrated circuits, the optical proximity effect (OPE: Optical Proximity Effect) has become apparent as the pattern dimensions greatly exceed the resolution limit at the time of exposure. It becomes difficult to transfer to. Therefore, when the pattern dimensions exceed the resolution limit at the time of exposure, the optical proximity effect correction generates a correction pattern that can cancel the OPE by correcting the dimensions and shape of the original circuit pattern based on the design data. (OPC) and super-resolution technology (RET) processing is required.

  However, if the above processing is performed, the circuit pattern and thus the entire pattern are complicated (diversified), so that the time required for drawing increases, and the cost for producing a semiconductor integrated circuit becomes very high. Inconvenient. Therefore, in the techniques shown in Non-Patent Document 1 and Non-Patent Document 2 described above, a partial pattern having the same shape and size is specified from a combination of the same rectangles constituting a part of the pattern from the entire pattern. Then, a combination of a plurality of rectangles is drawn at a time according to the specified partial pattern (that is, the partial pattern is extracted as a repeated pattern and drawn). In this way, even with drawing data after processing by optical proximity correction (OPC) and super-resolution technology (RET), the above-mentioned CP drawing technology can be applied to efficiently reduce the number of drawing operations. It has already been considered to reduce the cost for producing a semiconductor integrated circuit by reducing the time required for drawing.

JP-A-10-256113 Japanese Patent Laid-Open No. 2001-35770

Masaki Yamabe, "Optimization of MDP, mask writing, and maskinspection for mask manufacturing const reduction", Proc.SPIE 7028,70280V (2008) Masahiro Shoji, Tadao Inoue, and Masaki Yamabe, "Verification of extraction repeating pattern efficiency from many actual device data", Proc.SPIE 7275,7275Q (2009)

  However, in the conventional devices and techniques as described above, processing such as optical proximity effect correction (OPC) and super-resolution technique (RET), which is complicated at the time of circuit design, is performed. Even if a partial pattern is extracted as described above, for example, from the drawing data including a large number of overall patterns of various shapes and sizes having a minute shape difference or size difference complicated by Since only combinations of the same shape and size are extracted, only a large number of various partial patterns are extracted in the end, and there are too few partial pattern application points as repetitive patterns that can be used for CP drawing. There is a problem that the effect of reducing the number of times of drawing is small. Such a problem is that the entire pattern is irregularly arranged as compared with the memory circuit pattern. For this reason, the pattern is changed from one overall pattern to a plurality of different derived patterns by the above OPC and RET processing considering the adjacent pattern. This is particularly noticeable in logic circuit patterns that can be diversified.

  The present invention has been made in view of the above points, and efficiently extracts a versatile repetitive pattern from drawing data including a large number of fine and various overall patterns and sizes in a circuit pattern, An object of the present invention is to provide a CP drawing data creation method and apparatus, and a program for creating CP drawing data capable of reducing the number of shots at the time of drawing and shortening the drawing time.

  In the CP drawing data creation method according to the present invention, a repetitive pattern included in a circuit pattern is extracted from drawing data for designing a semiconductor integrated circuit, and a portion of the circuit pattern is shot by repeatedly shot the extracted repetitive pattern. A CP drawing data creation method for creating CP drawing data for efficiently drawing on a substrate, the step of dividing each pattern existing in a large number in the circuit pattern into a number of rectangles; Among the partial patterns constituting a part of each pattern obtained by combining arbitrary adjacent rectangles, one or more partial patterns having the same shape and size are extracted as one partial pattern group. The step and the plurality of extracted partial pattern groups are compared with each other, and the difference in shape and size is within a certain range. Each of the partial pattern groups includes a step of generating one representative pattern representing the partial pattern included in the set of partial pattern groups, and a step of generating CP drawing data based on the representative pattern.

  According to the present invention, a plurality of patterns existing in a circuit pattern are divided into a large number of rectangles, and a part of each pattern obtained by combining arbitrary adjacent rectangles among the divided large numbers of rectangles is configured. Among the partial patterns to be processed, one representative pattern that represents not only partial patterns having the same shape and size but also similar partial patterns is determined. Then, CP drawing data is generated based on the determined representative pattern. In such a case, the shapes and dimensions in the circuit pattern in the original drawing data are similar to each other included in similar partial pattern groups (each including one or a plurality of the same partial patterns of different types) within a certain range. Since the CP drawing data is generated based on the representative pattern that can share all of the partial patterns in the same pattern, not only the same partial pattern having the same shape and size in the original drawing data but also the shape and A repeated pattern can be extracted from among similar partial patterns having similar sizes. The repeated pattern extracted in this way is highly versatile, and increases the number of places where the repeated pattern is applied compared to the conventional one. Therefore, it is complicated at the time of designing the original circuit pattern, or is complicated by processing such as optical proximity correction (OPC) and super-resolution technology (RET), Even drawing data that contains a large number of patterns (overall patterns) in a circuit pattern, a highly versatile repeating pattern that was difficult to extract in the past can be extracted by partial patterns that form part of each pattern. It becomes possible to create CP drawing data having a large effect of reducing the number of drawing times used.

  The invention can be constructed and implemented not only as a method invention but also as an apparatus invention. Further, the present invention can be implemented in the form of a program of a processor such as a computer or a DSP, or can be implemented in the form of a storage medium storing such a program.

  According to the present invention, by extracting a repeated pattern including not only the same partial pattern but also a similar partial pattern, a similar partial pattern can be drawn in addition to the same partial pattern by the extracted repeated pattern. Even if it is drawing data that contains a lot of patterns (total patterns) of various shapes and sizes in the circuit pattern, it is possible to extract many highly versatile repeated patterns from application points that were difficult to extract in the past. There is an effect that it is possible to create CP drawing data having a large effect of reducing the number of drawing times.

It is a flowchart which shows CP drawing data creation processing. It is a conceptual diagram which shows an example of the deformation | transformation range (diversity relaxation value) which accept | permits the difference of a shape and a dimension. It is a flowchart which shows one Example of a partial pattern repetition data extraction process. It is a flowchart which shows one Example of a representative pattern determination process. It is a flowchart which shows one Example of the comparison process in a group. It is a conceptual diagram which shows an example of the whole pattern before the diversity relaxation contained in the circuit pattern defined by the original drawing data. It is a conceptual diagram which shows an example of a partial pattern, Fig.7 (a) is a 1st C type figure, FIG.7 (b) is a 2nd C type figure, FIG.7 (c) is a 1st E type figure, FIG. 7D shows a partial pattern of the second E-shaped figure, and FIG. 7E shows a partial pattern of the third E-shaped figure. It is a conceptual diagram which shows an example of the partial pattern group extracted. It is a conceptual diagram for demonstrating concretely a representative pattern determination process. It is a conceptual diagram which shows the whole pattern after the diversity relaxation by which each partial pattern contained in the similar partial pattern group was replaced by the representative pattern, and was made common. It is a conceptual diagram which shows one Example of the whole structure of a mask drawing machine. It is a flowchart which shows the process which determines a diversity relaxation value based on the result of transferability simulation.

  Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

  FIG. 1 is a flowchart showing CP drawing data creation processing. This process is executed by a computer including a CPU, a ROM, a RAM, and the like. In FIG. 1, drawing data (original drawing data) for designing a semiconductor integrated circuit in which a circuit pattern is defined, CP drawing data finally created, and CP drawing data are created from the original drawing data. The storage devices such as a hard disk and a memory (or a predetermined area in a specific storage device) each storing intermediate data appropriately generated at the time are also shown for convenience in the drawing.

  In step S1, a “deformation range allowing a difference in shape and dimensions” (diversity relaxation value) to be referred to when extracting partial pattern repetition data described later is determined. Here, a method of determining a deformation range (diversity relaxation value) that allows the difference in shape and size will be described. FIG. 2 is a conceptual diagram illustrating an example of a deformation range (diversity relaxation value) that allows a difference in shape and size.

One method of determining the diversity mitigation value is to determine based on the design rule. In FIG. 2, the gate CD control value is used as a reference for each design rule (ITRS node) specified in the ITRS lithography requirements. As an example, a diversity alleviation value obtained according to the calculation formula shown in Equation 1 taking into account the MEEF (Mask Error Enhancement Factor) value at the time of mask drawing is shown.
(Equation 1)
Diversity relaxation value = {(gate part CD control value x magnification) / MEEF value} x 0.5 x 0.5

  More specifically, mask drawing is performed based on the gate CD control value (2.6 nm, 1.9 nm, 1.3 nm) for each design rule (65 nm, 45 nm, 32 nm) specified in the lithography requirements of the ITRS 2007 roadmap. Assuming that the MEEF value (2.2) is also taken into consideration, the maximum value of the dimensional difference used as a criterion for judging similar partial patterns when mitigating pattern diversity (to be described later) (that is, an allowable error for pattern similarity judgment to be described later) Dimension) is calculated. That is, the value calculated by “{(gate part CD control value × magnification) / MEEF value}” in the above equation 1 is the maximum allowable error between the design time of the data and the drawing result including the dimension error at the time of drawing. Represents a value. However, here, the value calculated by “{(gate part CD control value × magnification) / MEEF value}” is multiplied by half (0.5) as an allowable error value on the drawing data, and will be described later. In order to perform relaxation in both the enlargement and reduction directions based on the representative pattern, the diversity relaxation value is obtained by further multiplying half (0.5). Note that the “magnification” described in Equation 1 is “4” in the case of mask drawing.

  Returning to the description of FIG. 1, in step S <b> 2, original drawing data 1 (more specifically, circuit pattern data) created in advance is acquired. The drawing data 1 acquired here is design data (for example, CAD data) generated in advance by a user using a computer or the like, or optical proximity correction (OPC) and super-resolution technology (RET) for such design data. ), And defines a circuit pattern including a large number of complicated overall patterns of various shapes and sizes (see FIG. 6).

  In step S3, the acquired original drawing data 1 is fractured, and the result is output and stored as fractured drawing data 2. In this fracturing process, when the entire pattern is a polygon and cannot be represented by a single rectangle, the pattern is divided into a number of various rectangles based on the location where a step occurs (see FIG. 7).

  In step S4, the fractured drawing data 2 over the entire region is separated into a diversity alleviation target region and a diversity alleviation target region according to a predetermined predetermined pattern importance (see FIG. 6). Output and store as mitigation target data 3 or diversity non-division target data 4. The pattern importance degree defines an important part and an unimportant part of the entire pattern or a part of the pattern in the circuit. It is expected that important parts will be drawn with the shape, dimensions, and position as designed (high dimensional accuracy), but non-important parts will be drawn as a result of drawing compared to important parts. Even a slight difference may be allowed. As a method of determining such pattern importance, it is conceivable to determine at the time of circuit design, to determine according to the arrangement / regularity of drawing data, or to determine according to the user's range specification. Note that the process of separating the diversity relaxation target area and the diversity relaxation target area depending on the pattern importance may not be performed.

  In step S5, “partial pattern repetition data extraction processing” is executed. As will be described in detail later, in this process, only the separated diversity mitigation target data 3 is acquired, and a large number of the entire patterns divided by fracturing based on the acquired diversity mitigation target data 3 are formed. Of the rectangles, the same partial pattern is searched for whether there is a partial pattern (referred to as the same partial pattern to be distinguished from a similar partial pattern described later) having the same shape and size that is a combination of the same rectangles. Are extracted as one partial pattern group, and this is output and stored as partial pattern definition-added data 5. The partial pattern definition-added data 5 includes at least definition information (that is, information on different types of partial pattern groups) regarding the same partial pattern searched for at least the shape, size, or combination of rectangles. It is only necessary to include arrangement information indicating different arrangement positions in the same partial pattern, and the arrangement information may not be included (that is, the inclusion status of the arrangement information does not matter). In this way, as each identical pattern having the same shape and size that forms part of each overall pattern is searched, one or more identical partial parameters each having the same shape and size are included. A plurality of different types of partial parameter groups are extracted (see FIG. 8). At this time, the maximum number of types of partial parameter groups that can be stored is preferably about 10 million.

  In step S6, a “representative pattern determination process” is executed. As will be described in detail later, in this process, the partial pattern definition-added data 5 is acquired, and different types of partial pattern groups are sequentially compared based on the definition information, and the difference in shape and size is within a certain range. The partial pattern group is specified, and for each specified similar partial pattern group, one of the partial patterns included in the partial pattern group is determined as a representative pattern, and this is output as representative pattern definition data 6 and stored. (See FIGS. 8 and 9). The representative pattern definition data 6 only needs to include definition information of at least the partial pattern determined as the representative pattern.

  In step S7, the diversity mitigation target data 3 and the representative pattern definition data 6 are acquired, and the diversity mitigation target data 3 is replaced based on the determined representative pattern. That is, the representative pattern is sequentially read out from the representative pattern definition data 6, and the same partial pattern in which the difference in shape and size is within a certain range compared to the representative pattern read out from the diversity mitigation target data 3 and A similar partial pattern is identified, and all of the identified identical partial pattern and similar partial pattern are replaced with the compared representative pattern (see FIG. 10). In this way, by replacing the same partial pattern and similar partial pattern in which the difference in shape and size included in the circuit pattern in the original drawing data 1 is within a certain range by the respective representative patterns sequentially, the same and similar parts Drawing data (replaced circuit pattern data) including circuit patterns that have been replaced with the representative pattern in which all the patterns are shared by the same pattern is created. The drawing data replaced with the representative pattern (replaced circuit pattern data) is output and stored as representative pattern replaced data 7. Step S8 covers the entire region including not only the diversity mitigation target region but also the diversity mitigation target region by acquiring the representative pattern replaced data 7 and the diversity mitigation target data 4 and combining them. The synthesized data 8 to be generated is created, and this is output and stored.

  A step S9 executes a conversion process for removing rectangular overlaps and gaps (slits) in the entire synthesized area in accordance with the synthesized data 8, and outputs and stores the result as diversified data 9. That is, when the same partial pattern and the similar partial pattern are replaced with the representative pattern as described above, other rectangles and partial patterns near the boundary with the outer periphery of the replaced identical partial pattern and similar partial pattern. There may be overlaps or gaps (slits) between regions that did not exist before replacement (including those other than the same partial pattern and similar partial patterns). Therefore, processing for removing such overlapping areas and filling gaps is executed. Any method may be used for this processing, but a general method will be described here.

  First, the entire area of each rectangle and each partial pattern is slightly enlarged. At this time, as the enlargement width of the region, the width of the overlap or slit (that is, the diversity relaxation value) is used as a guide. The slit is filled by enlarging the area. In the enlargement process of the area, the entire pattern in which each rectangle is connected is regarded as one process unit, not each rectangle unit after fracturing, and only the outer frame of the entire pattern is overlapped. Since the part is ignored, this eliminates the overlap. Naturally, overlapping of regions that can be newly generated in accordance with the filling of the slit is also eliminated.

  Next, the entire area of the enlarged rectangle and each partial pattern is reduced (returned) to the original size. Here, the reduction width is the same as the enlargement width used during the enlargement process. In the reduction process, as in the enlargement process, the entire pattern is regarded as one processing unit, and therefore, the reduction process is reduced as one large pattern in which the slit is filled. In this way, an overall pattern with the same dimensions as the original but without overlap or slits is obtained. Therefore, the entire pattern is obtained by executing the above-described conversion process regardless of whether or not there is an overlap of areas or a gap.

  In step S10, the CP data is generated by acquiring the diversity-reduced data 9 and executing the partial pattern repetition data extraction process again in all areas based on this data. That is, all the same partial patterns and similar partial patterns are replaced with the representative pattern and shared by the execution of the above replacement process (step S7), so that the original drawing data 1 is composed of the same rectangular combination. A part having a different shape and size that could not be extracted as a repeated part of a partial pattern having the same shape and size is newly re-extracted as a repeated part of the same partial pattern. Therefore, the number of shots at the time of drawing can be reduced by using the partial pattern re-extracted in this way as a repeated pattern. The number of types of partial patterns to be extracted at this time is preferably about 1000. In this way, CP drawing data that reduces the number of shots during drawing and shortens the drawing time is created.

  The “partial pattern repetition data extraction process” (step S5 in FIG. 1) will be described with reference to FIG. FIG. 3 is a flowchart showing an example of “partial pattern repetition data extraction processing”.

  In step S21, the diversity mitigation target data 3 is acquired and the graphic (rectangular) list 11 is generated. In the figure (rectangle) list 11, a large number of rectangles having different sizes obtained by dividing each overall pattern are registered. A step S22 extracts a rectangle having a size that does not fit within a predetermined partial pattern size determined in advance from the graphic list 11, and cuts the rectangle into a plurality of rectangles of the maximum size that can fit within the partial pattern size.

  That is, as described above, in the fracturing, when the entire pattern is a polygon and cannot be represented by a single rectangle, the rectangle is divided on the basis of the location where the step is generated. The rectangle may be cut into rectangles that do not fit within a predetermined partial pattern size. Therefore, such a largely cut rectangle is subdivided into a maximum size rectangle that can fit in the partial pattern size. The rectangle of the maximum size that can be accommodated in the partial pattern size after subdivision is output as the partial pattern candidate list 12, while the remaining rectangles (rectangles smaller than the maximum size) among the rectangles divided by subdivision are displayed in the graphic list 11. Return to.

  In step S23, rectangles adjacent to each other are extracted from the graphic list 11, and a plurality of rectangles that are within the partial pattern size (as large as possible) are collectively output to the partial pattern candidate list 12. In step S24, isolated rectangles (rectangles not adjacent to other rectangles) dispersed within a range that fits in the partial pattern size are extracted from the graphic list 11, and these are collectively output to the partial pattern candidate list 12. In step S25, the respective CP candidate graphics output to the partial pattern candidate list 12 are checked, the same graphics are combined into one, and these are output to the partial pattern list 13 as a partial pattern group. At this time, the partial pattern group may be determined by a predetermined number in order from the one with the best CP drawing efficiency (this can be calculated from the number of appearances of each partial pattern and the number of rectangles in the partial pattern). Then, the remaining partial pattern candidate graphics are returned to the graphics list 11 by returning them to individual rectangles.

  A step S26 outputs (generates) the partial pattern group of the partial pattern list 13 and the rectangle left in the graphic list 11 as the partial pattern definition-added data 5. In this way, the same partial pattern that is the combination of the same rectangles and has the same shape and size is output as one partial pattern group, and the other partial patterns are individually output as one partial pattern. Output individually as a group. The partial pattern definition-added data 5 may store definition information common to at least one or more partial patterns included in the partial pattern group definition information.

  Next, the “representative pattern determination process” (step S6 in FIG. 1) will be described with reference to FIG. FIG. 4 is a flowchart showing an example of the “representative pattern determination process”.

  In step S31, the partial pattern definition-added data 5 output (generated) with the execution of the “partial pattern repetition data extraction process” is acquired, and each partial pattern group is set to be vertically long or horizontally long according to the partial pattern definition information. For example, the data is roughly classified according to a predetermined classification standard and grouped, and the result is output to the grouping list 14. The step S32 acquires the grouping list 14 and compares the partial pattern groups included in the classified groups to determine one or a plurality of representative patterns for each group and specify similar partial patterns. The “in-group comparison process” is executed, and the result is output to another group list 15. This process will be described later (see FIG. 5 described later). In step S33, with respect to the determined representative pattern and the specified similar partial pattern, the definition information of each representative pattern and the definition information of the partial pattern replaced by the set are output as representative pattern definition data 6.

  The “in-group comparison process” (step S32 in FIG. 4) will be described with reference to FIG. FIG. 5 is a flowchart showing an example of the “in-group comparison process”.

  A step S41 extracts a partial pattern group at the head of the list output first from the partial pattern group included in the grouping list 14 and determines any of the partial patterns included in the partial pattern group as a representative pattern. . At this time, the partial pattern group determined as the representative pattern is assumed to be checked data. The representative pattern determination method is not limited to extracting partial pattern groups in the order of the list. For example, any determination method may be used, such as selectively extracting a partial pattern group according to the complexity of the shape or the size of the area and determining it as a representative pattern. In step S42, it is determined whether or not there is a partial pattern group in the grouping list 14 that has not been checked yet. When it is determined that there is a partial pattern group that has not been checked yet in the grouping list 14 (with step S42), a partial pattern group that has not been checked is extracted from the grouping list 14 and determined as a comparison pattern (step S42). S43).

  A step S44 determines whether or not the patterns are similar by superimposing the representative pattern and the comparison pattern. In this similarity determination, if the difference between the representative pattern and the comparison pattern is within the allowable error dimension based on the diversity relaxation value, it is determined that the patterns are similar to each other, and if the difference is outside the allowable error dimension, the mutual pattern is determined. Are not similar (see FIG. 9 for details). If it is determined that the representative pattern and the comparison pattern are similar (similarity in step S44), the partial pattern included in the comparison pattern group is specified as a similar partial pattern similar to the representative pattern and output to the group list 15 (step S45). . Then, the process returns to step S42. At this time, the partial pattern group including the partial patterns that are similar partial patterns is deleted from the grouping list 14. On the other hand, when it is determined that the representative pattern and the comparison pattern are not similar (dissimilarity in step S44), the partial pattern group set as the comparison pattern is set as checked data, and the process returns to step S42.

  In the process of step S42, if it is determined that there is no unchecked partial pattern group in the grouping list 14 (no step S42), the partial pattern group other than the representative pattern and used as checked data The check result is returned to unchecked (step S46). A step S47 decides whether or not there is a partial pattern group in the grouping list 14 that has not been checked yet. If it is determined that there is a partial pattern group that has not been checked in the grouping list 14 (the presence of step S47), the process returns to the process of step S41, and the processes of steps S41 to S47 are repeated. On the other hand, when it is determined that there is no unchecked partial pattern group in the grouping list 14 (No in step S47), the process ends.

  Next, the pattern diversification mitigation procedure according to the CP drawing data creation method according to the present invention will be described taking the case of mask drawing as an example. FIG. 6 is a conceptual diagram illustrating an example of an overall pattern before diversification included in a circuit pattern defined by original drawing data. The circuit pattern defined by the original drawing data 1 has various shapes and sizes having minute shape differences and dimensional differences complicated by optical proximity effect correction (OPC) and super-resolution technology (RET) processing. In this case, there are two types of overall patterns A and B (for convenience, the first or the second pattern). (Referred to as the second C-shaped figure), two, two, and one total pattern C to E (total number of five total shapes C-E (first to third E for convenience)) This is illustrated as an example including a pattern graphic).

  Comparing the two types of C-shaped figures A and B shown here, the first C-shaped figure A and the second C-shaped figure B are approximately the same in size, but the shape of the details is different. The first C-shaped graphic A has a finer and more complex uneven shape than the second C-shaped graphic B. That is, it can be said that the first C-shaped graphic A is a pattern having a fine shape difference and a dimensional difference generated by, for example, OPC processing, as compared with the second C-shaped graphic B.

  Next, when the three types of E-type figures C to E are compared, the first E-type figure C and the second E-type figure D are approximately the same in size but different in shape. The E-shaped figure C has a finer and more complex uneven shape than the second E-shaped figure D. That is, it can be said that the first E-shaped graphic C and the second E-shaped graphic D are patterns having minute shape differences and dimensional differences. On the other hand, the first E-shaped figure C and the third E-shaped figure E have the same shape but different sizes, and the second E-shaped figure D and the third E-shaped figure E also have the same size. The shape is also different. In the case of such drawing data 1, in the conventional apparatus, all of the C-type figures A and B and the E-type figures C to E are regarded as whole patterns having different shapes and sizes, and any combination thereof Since it is not possible to extract a common repetitive pattern from each other and each individual pattern A to E must be drawn individually and for each divided rectangle, drawing takes time.

  In FIG. 6, a region Y indicated by a bold frame indicates a non-diversification target region in the entire region X forming the circuit pattern. In this example, the first and second C-shaped figures A and B and the first and second E-shaped figures C and D arranged in the area Y each include four figures in total. ing. Therefore, when fracturing is performed based on the drawing data having the pattern arrangement as shown in FIG. 6, the fractured drawing data 2 of each of the four figures included in the region Y is used as the diversity-reduction target data 4. Then, the fractured drawing data 2 of each of the five other figures is output and stored as the diversity alleviation target data 3 (see step S4).

  FIG. 7 is a conceptual diagram showing an example of a partial pattern. FIG. 7A is a first C-shaped graphic A, FIG. 7B is a second C-shaped graphic B, and FIG. E-shaped figure C, FIG. 7 (d) shows a second E-shaped figure D, and FIG. 7 (e) shows a partial pattern of the third E-shaped figure E, respectively.

  Each of the entire patterns A to E is drawn by being divided into a large number of the same or different rectangles corresponding to the beam size of the drawing machine. For example, in the first C-shaped graphic A shown in FIG. 7A, there are a total of 23 pieces, that is, 5 pieces (× 2) in the range A1, 5 pieces in the range A2, 5 pieces in the range A3, and 3 pieces in the range A4. Divided into rectangles. Similarly, a total of 11 second C-shaped figures B, 37 total 1st E-shaped figures C, 22 total 2nd E-shaped figures D, 37 37th E-shaped figures E It is divided into rectangles. Usually, since the entire pattern divided in this way is drawn individually for each rectangle, it takes time to finish drawing. For example, when drawing the first C-shaped graphic A, since 23 rectangles are shot individually, it takes time corresponding to the number of shots.

  By the way, the whole pattern divided into a large number of rectangles can be regarded as a combination of a plurality of partial patterns formed by arbitrarily combining a plurality of rectangles. In some cases, there are a plurality of partial patterns (identical partial patterns) having the same value. As described above, by performing the above-described “partial pattern repetition data extraction process” (see FIG. 3), the same partial pattern is searched and extracted as one partial pattern group.

  Specifically, for example, the first figure C type A shown in FIG. 7A is a combination of four partial patterns A1 to A4, A1 is two, and the others are partial pattern groups having one appearance location. Extracted as In the second C-shaped figure B shown in FIG. 7B, the partial patterns B1 and B2 are extracted as a partial pattern group having two appearance locations and B3. In the first E-shaped graphic C shown in FIG. 7C, the partial pattern C1 is extracted as a partial pattern group having three appearances, C4, C5 are two, and C2, C3, and C6 are one appearance location. . In the second E-shaped graphic D shown in FIG. 7D, partial patterns D1 are extracted as a partial pattern group having four appearances, D2 and D5 are two, and D3 and D4 are one appearance. In the third E-shaped figure E shown in FIG. 7E, partial patterns E1 are extracted as a partial pattern group having three appearances, E4, E5 two, and E2, E3, E6 having one appearance location. .

  Further, the same partial pattern described above is not limited to being present only in individual whole patterns, but may be present in other whole patterns. For example, the first C-shaped figure A shown in FIG. 7A and the first E-shaped figure E shown in FIG. 7C are greatly different from each other in the overall shape. It can be understood that there are partial patterns that have the same shape and size as a combination of the same rectangles (specifically, partial patterns A1, C1, A2, C2, A3 and the like). C3, A4 and C4 are the same partial pattern). In such a case, even partial pattern groups that form a part of different overall data are grouped together as partial pattern groups of the same type.

  In the “partial pattern repetition data extraction process” described above (see FIG. 3), instead of comparing the whole pattern in this way, a partial pattern that is a partial range of the pattern is extracted for each whole pattern and compared, A partial pattern group is extracted by collecting patterns having the same shape and size as partial patterns. FIG. 8 shows a partial pattern group extracted from each overall pattern divided into rectangles as shown in FIG. FIG. 8 is a conceptual diagram illustrating an example of the extracted partial pattern group. In the drawing, the numbers described below the partial pattern groups represented by graphics correspond to the same partial patterns included in each partial pattern group shown in FIG.

  As shown in FIG. 8, 16 different partial pattern groups are obtained from the drawing data having the entire pattern as shown in FIG. That is, A1 and C1, A2 and C2, A3 and C3, A4 and B3 and C4, B1 and D1, B2 and D2, C5 and D5, C6, D3, D4, E1, E2, E3, E4, E5, E6 Each is extracted as a different type of partial pattern group. For example, since the partial patterns A1 and C1 have the same shape and size, they are extracted as a partial pattern group of the same type. On the other hand, these partial patterns A1 (C1) and partial pattern E1 are extracted as separate types of partial pattern groups because they have the same shape but different sizes. Further, regarding this partial pattern E1, there is no other partial pattern having the same shape and size, and therefore the partial pattern E1 is extracted as a single partial pattern group. In this way, among the partial patterns that constitute a part of the overall pattern obtained by combining arbitrary rectangles adjacent to each other among a large number of divided rectangles, the shape and size are the same and combinations of rectangles are also possible. One or more partial patterns that are the same are extracted together as a partial pattern group of the same type.

  FIG. 9 is a conceptual diagram for specifically explaining the above-described “representative pattern determination process” (see FIG. 3). First, a plurality of partial pattern groups extracted as described above are roughly classified according to a predetermined classification criterion into groups such as portrait, landscape, or others. In the example of FIG. 8, they are classified into two groups, which are vertically long and horizontally long (in the figure, those classified into vertically long groups are shown in a frame indicated by a dotted line). The partial pattern groups are compared in each group. First, the comparison in the horizontally long group will be described.

  FIG. 9A shows a case where, for example, a partial pattern group of the type indicated by the graphic C5 (hereinafter the same) is determined as the representative pattern, and another graphic (partial pattern group) classified into the same group as the graphic C5. Are sequentially compared to determine whether or not the partial patterns included in each partial pattern group are similar. Here, only the case where the graphic A1 and the graphic E1 are compared as comparison targets will be shown and described. The figure C5 and the figure A1 (or the figure E1) are overlapped with each other according to a predetermined reference position (for example, an axial center, a side surface, or a center position), and the horizontal and vertical differences that can occur due to the overlapping state at this time are determined. Whether or not the graphic A1 (or the graphic E1) is a partial pattern similar to the graphic C5 is determined based on whether or not both are within the allowable error dimension (diversity relaxation value). If the difference between the horizontal direction and the vertical direction is within the allowable error dimension, the pattern is specified as a similar partial pattern. In this example, looking at the degree of overlap between the figure A1 and the figure C5 of the figure E1, the figure A1 is similar because the figure A1 has a difference within an allowable error dimension in both the horizontal direction and the vertical direction. Specific to partial pattern. On the other hand, the figure E1 is not specified as a similar partial pattern because the difference exceeds the allowable error dimension in both the horizontal direction and the vertical direction.

  FIG. 9B shows a case where, for example, the figure B2 is next determined as the representative pattern from the figures not specified in the similar partial pattern group in the comparison with the figure C5. In this example, looking at the degree of overlap between the figure A2 and the figure B2 of the figure E2, the figure A2 is similar because the figure A2 has a difference within an allowable error dimension in both the horizontal direction and the vertical direction. Specific to partial pattern. On the other hand, since the figure E2 has a difference exceeding the allowable error dimension in the vertical direction, the figure E2 is not specified as a similar partial pattern. In this way, the representative pattern and the similar partial pattern are specified from the partial pattern group classified into the horizontally long group.

  FIG. 9C shows a comparison of a vertically long group different from the above horizontally long group. In this example, the figure A4 is determined as the representative pattern. Looking at the degree of overlap between figure D3 and figure A4, figure D3 has a difference within the allowable error dimension in both the horizontal and vertical directions, so figure D3 is identified as a similar partial pattern. Is done. Similarly, the figure D4 is specified as a similar partial pattern. On the other hand, the figure E4 is not specified as a similar partial pattern because the difference exceeds the allowable error dimension in both the horizontal direction and the vertical direction.

  In this way, by comparing the partial pattern groups within each group, in the example shown in FIG. 8, the graphic C5 as the representative pattern (indicated by hatching in the drawing) and the graphic A1, as its similar partial pattern C1, figure B2 as a representative pattern and figures A2, C2, A3, C3 as a representative pattern, figure B1 as a representative pattern and figure E5 as a similar pattern, figure A4 as a representative pattern, and figure D3 as a similar pattern D4, graphics C6, E1, E2, E3, E4, and E6 that do not have similar partial patterns as representative patterns are specified.

  In accordance with the representative pattern and the similar partial pattern specified as described above, the similar partial pattern is replaced with the representative pattern from the drawing data including the entire pattern as shown in FIG. Diversified data with similar partial patterns in common is created (see steps S7 to S9). FIG. 10 is a conceptual diagram showing an overall pattern after diversification in which each partial pattern included in a similar partial pattern group is replaced with a representative pattern and shared.

As can be understood by comparing FIG. 10 and FIG. 6, the C-type graphic A in the upper left of FIG. 6 is replaced with a C-type graphic A ′. That is, the partial pattern A2 and A3 are replaced with the representative pattern B2 to be shared with the partial pattern B2, and the partial pattern A4 and the partial pattern B3 are the same. The figure A ′ is close to the overall shape of B.
Further, in the E-type graphic C at the lower left of FIG. 6 and the E-shaped graphic D at the upper right of FIG. 6, the partial pattern C1 is the partial pattern C5, the partial patterns C2 and C3 are the partial pattern B2 (D2), and the partial pattern D4 is the partial pattern A4. 10 are transformed into E-shaped figures C ′ and D ′ in FIG. 10, respectively. That is, instead of simply replacing one overall pattern with the other overall pattern, there are similarities in the partial patterns that form a part of each of them even though they are figures having different overall shapes (overall patterns). It is possible to increase the number of occurrences of the repeated pattern by replacing the similar partial pattern with the representative pattern and sharing it. Note that, as a result of the common use, the entire pattern may have the same shape.

  In this way, in the example shown here, the number of partial patterns having the same shape and size obtained by simplifying the combination of rectangles increases, so that the number of drawing operations can be efficiently reduced. Specifically, for the pattern existing in the diversity alleviation target region shown in FIG. 10, the number of drawing can be greatly reduced by drawing using a CP mask in which each partial pattern is formed in advance. For example, when the partial pattern C5 (D5) is included in the repetitive pattern extracted from the diversified data, not only the graphic C ′ and the graphic D ′ including the partial pattern C5 (D5) but also the partial pattern C5 ( A pattern that can be drawn with the partial pattern C5 (D5) is also included in the figure A ′ that includes the partial pattern A1 similar to D5). A CP mask created using a repeated pattern extracted from diversity-reduced data in this way has more patterns that can be drawn than when diversity reduction (that is, sharing of partial patterns with representative patterns) is not performed. It will be. It should be noted that portions where the partial pattern is not defined on the CP mask are drawn individually on the drawing target without using the CP mask.

  Here, FIG. 11 shows an example of a mask drawing machine that draws a circuit pattern on a mask dry plate using a CP mask created according to the CP drawing data created in accordance with the execution of the CP drawing data creation process. FIG. 11 is a conceptual diagram showing an embodiment of the overall configuration of the mask drawing machine.

  The electron beam irradiated from the electron gun 101 is adjusted in the size of the rectangular window M by the rectangular aperture 102 and the variable shaper 103 (or mask selection changer) or the drawing pattern of the CP mask 104 for mask drawing (hereinafter simply referred to as CP mask). Is selected. When the mask is selected, a partial pattern is formed on the mask dry plate 105 via the CP mask 104. A CP mask 104 in which a figure N (re-extracted repetitive pattern, corresponding to A1, C1, E1, etc. in FIG. 8) is formed in advance according to the CP drawing data created by the above-described CP drawing data creation processing (see FIG. 1). Even if the original drawing data is drawing data including fine and diverse patterns complicated by processing such as optical proximity correction (OPC) and super-resolution technology (RET), Pattern drawing can be performed with a smaller number of drawing times than in the past.

  The drawing machine is not limited to the mask drawing machine described above, and may be a direct drawing type drawing machine that does not use a mask. In the case of a direct drawing type drawing machine, a repetitive pattern (A1 in FIG. 8) that is re-extracted after being replaced with a representative pattern according to the CP drawing data created by the above-described CP drawing data creation processing (see FIG. 1). (Or equivalent to C1, E1, etc.), the same and similar partial patterns are repeatedly drawn directly on the substrate. Then, since the number of patterns that can be drawn with fewer shots by CP drawing increases, the number of drawing times can be reduced as compared with the conventional case.

  In the above-described embodiment, the method of determining the diversity relaxation value based on the data design rule is shown. However, the present invention is not limited to this, and a transferability simulation is executed on a computer and the user is displayed on a display or the like. A method of arbitrarily determining the diversity relaxation value while confirming the simulation result. A method in which the user arbitrarily determines the diversity relaxation value according to the result of such transferability simulation will be described with reference to FIG. FIG. 12 is a flowchart showing processing for determining the diversity relaxation value based on the result of the transferability simulation.

  In step S61, initial parameters are set. This initial parameter may be a diversity relaxation value determined according to the design rule (see FIG. 2), or may be arbitrarily specified by the user based on past performance values. In addition to setting a uniform diversity relaxation value for all regions, a different diversity relaxation value may be set for each region designated as a diversity relaxation target. In step S62, the CP drawing data creation process shown in FIG. 1 is executed. Since this process has already been described, a description thereof is omitted here.

  In step S63, a transferability simulation is executed based on the CP drawing data 10 generated along with the execution of the process. In step S64, it is determined whether or not the simulation result check by the user is “OK”. If it is determined that the simulation result check is not “OK” (NO in step S64), for example, the user is prompted to input a parameter (diversity relaxation value), and the parameter is set according to the parameter input by the user. Is reset (step S65), the process returns to step S62, and the processes of steps S62 to S64 are repeated. If it is determined that the simulation result check is “OK” (YES in step S64), the set parameter is set to the diversity alleviation value (step S66).

DESCRIPTION OF SYMBOLS 1 ... Drawing data 2 ... Fractured drawing data 3 ... Diversity relaxation object data 4 ... Diversity relaxation object data 5 ... Partial pattern definition data 6 ... Representative pattern definition data 7 ... Representative pattern replacement data 8 ... Synthesized data 9 ... Diversified data 10 ... CP drawing data 11 ... Graphic list 12 ... Partial pattern candidate list 13 ... Partial pattern list 14 ... Grouping list 15 ... Group list 101 ... Electron gun 102 ... Rectangular aperture 103 ... Variable shaper 104 ... CP mask 105 for mask drawing ... Mask dry plate M ... Rectangular window N ... Whole pattern

Claims (10)

CP drawing data for extracting a repetitive pattern included in a circuit pattern from drawing data for designing a semiconductor integrated circuit, and drawing the extracted repetitive pattern efficiently on a substrate by repeatedly shooting the extracted repetitive pattern CP drawing data creation method for creating
Dividing each pattern present in the circuit pattern into a plurality of rectangles;
Among partial patterns obtained by combining arbitrary adjacent rectangles among the plurality of divided rectangles, one to a plurality of partial patterns having the same shape and size are defined as one partial pattern. Extracting as a partial pattern group,
The extracted plurality of partial pattern groups are compared with each other, and a partial pattern included in the set of partial pattern groups is represented for each set of similar partial pattern groups having a difference in shape and size within a certain range. Generating two representative patterns;
Generating CP drawing data based on the representative pattern. The step of generating CP drawing data based on the representative pattern includes:
Creating replaced circuit pattern data obtained by replacing a partial pattern in the set in the original circuit pattern with the representative pattern;
In the replaced circuit pattern data, creating diversity-reduced circuit pattern data in which at least gaps between the patterns generated when replacing with the representative pattern are removed;
Of the partial patterns obtained by combining arbitrary adjacent rectangles out of a large number of rectangles from the circuit pattern data with reduced diversity, one or more partial patterns having the same shape and size are taken as one part The CP drawing data creation method according to claim 1, further comprising: creating CP drawing data extracted as a pattern group.   Each pattern present in the circuit pattern is a pattern having various shapes and sizes having fine shape differences and dimensional differences complicated by optical proximity correction, and the patterns are drawn on a mask. The CP drawing data creation method according to claim 1 or 2, wherein the created CP drawing data is used.   Each pattern present in the circuit pattern has a complicated shape at the time of circuit design, and the created CP drawing data is used when the pattern is drawn directly on the substrate. The CP drawing data creation method according to claim 1 or 2.   Each of the circuit patterns that exist in the circuit pattern based on the pattern importance set for each pattern that exists in the circuit pattern when the circuit pattern is designed in the original drawing data creation or after the original drawing data is created 5. The CP drawing data creation according to claim 1, further comprising a step of classifying each of the patterns into a target pattern to be replaced with a representative pattern and a non-target pattern not to be replaced with a representative pattern. Method.   6. The CP drawing data creation method according to claim 1, wherein the difference between the shape and the dimension is determined based on a predetermined semiconductor design rule.   The CP drawing data creation method according to claim 1, wherein the difference between the shape and the dimension is input in accordance with a user operation corresponding to a transfer result of the transferability simulation. CP drawing data for extracting a repetitive pattern included in a circuit pattern from drawing data for designing a semiconductor integrated circuit, and drawing the extracted repetitive pattern efficiently on a substrate by repeatedly shooting the extracted repetitive pattern CP drawing data creation device for creating
A dividing means for dividing each pattern present in the circuit pattern into a plurality of rectangles;
Among partial patterns obtained by combining arbitrary adjacent rectangles among the plurality of divided rectangles, one to a plurality of partial patterns having the same shape and size are defined as one partial pattern. Extraction means for extracting as a partial pattern group,
The extracted plurality of partial pattern groups are compared with each other, and a partial pattern included in the set of partial pattern groups is represented for each set of similar partial pattern groups having a difference in shape and size within a certain range. Pattern generation means for generating two representative patterns;
A CP drawing data creation device comprising drawing data generation means for generating CP drawing data based on the representative pattern.   A computer-executable program for realizing the CP drawing data creation method according to any one of claims 1 to 7.   A CP mask on which the representative pattern is formed based on CP drawing data created by the apparatus according to claim 8 or according to the program according to claim 9.

Images