JP4857963B2 - PATTERN EXTRACTION METHOD, PATTERN EXTRACTION DEVICE, AND SEMICONDUCTOR DEVICE MANUFACTURING METHOD - Google Patents

Description

  The present invention relates to a pattern extraction method, a pattern extraction device, and a semiconductor device manufacturing method.

  When manufacturing a semiconductor device, a fine pattern is processed using a lithography technique.

  In this lithography technology, in order to meet the demand for higher integration of semiconductor devices and higher operating speed, it is required to finely process patterns with higher resolution. Instead, an electron beam lithography technique using an electron beam is used.

  In this electron beam lithography technique, a point beam type exposure system is known. In this point beam exposure method, exposure is performed so that the entire pattern is filled with a finely focused electron beam. However, in this point beam type exposure method, since each of the plurality of patterns is exposed so as to be sequentially drawn, the drawing time becomes long. For this reason, since it is not easy to manufacture a semiconductor device with high throughput, this point beam type exposure method is not suitable for industrial mass production. For this reason, in the electron beam lithography technique, a variable shaped beam type exposure method has been proposed, and the throughput is improved by performing exposure while varying the size of the electron beam according to the shape of the pattern. . However, as the miniaturization progresses, the ratio of exposing a fine pattern by the point beam type exposure method is increasing, and thus the effect of improving the throughput by this variable shaped beam type exposure method is decreasing.

  Therefore, in order to further improve the throughput, a partial batch figure irradiation type exposure method has been proposed. This partial collective figure irradiation type exposure method is also referred to as a character projection (CP) method. In this CP method, a CP pattern repeatedly drawn in a layout pattern of an LSI (Large-Scale Integration) manufactured as a semiconductor device is extracted, and the extracted CP pattern is formed in advance on a CP mask as a mask pattern. The throughput is improved by transferring the CP pattern formed on the CP mask as necessary.

  By the way, the semiconductor business environment has changed from a memory to a system LSI as represented by SOC (System On Chip). For this reason, in an LSI layout pattern mainly composed of standard cells, a CP pattern that is repeatedly transferred is extracted, and a CP mask is produced using the extracted CP pattern (see, for example, Patent Document 1). Here, instead of forming a CP mask for each LSI layout pattern, a CP mask is created at the element figure level from the viewpoint of ensuring versatility, and the CP pattern at the element figure level is created from the layout pattern. Extraction is performed to form a CP mask. In this method, a complicated step of extracting graphic information is required. However, since the efficiency of replacement with a CP pattern is low, the throughput may not be sufficiently improved.

  On the other hand, in the photomask for optical lithography, the price increase has become apparent. For this reason, system LSIs, which are mainly used for a variety of products and small quantities, have become difficult to obtain a sufficient profit if they are manufactured by the conventional standard cell method. In order to improve this problem, the structured ASIC technology has attracted attention. In this structured ASIC, a general-purpose part including a lower wiring layer is formed in advance, and several other upper wiring layers and hole layers are customized so that it can be used for a system LSI for various kinds and small quantities. Yes. Here, it is sufficient to prepare a photomask for photolithography so as to correspond to several layers. For this reason, a significant increase in cost caused by the photomask can be avoided and manufacturing can be started from a pre-made general-purpose part, so that a short TAT can be realized. Further, as this structured ASIC technology, a technology for performing customization with only one hole layer has been proposed.

If a system LSI for various types and small quantities can be easily produced by applying electron beam exposure to a customized layer with only one hole as in this proposal, a photomask for optical lithography is no longer necessary, and photolithographic for optical lithography Since the cost and time required for manufacturing and inspecting the mask can be reduced, a low TAT and short TAT production environment can be realized. However, the CP pattern extraction and collation method based on the standard cell has a complicated procedure for obtaining exposure data, and the number of shots cannot be reduced, and a throughput suitable for the production level is achieved. Is difficult. For this reason, techniques focusing on CP generation and exposure of hole patterns have been disclosed (see Patent Documents 2 and 3).
Japanese Patent Laying-Open No. 2005-064404 JP 2004-259744 A JP 2002-252158 A

  However, in these methods, since the CP pattern is determined based on the frequency used in the design layout pattern of the target LSI, the method of extracting the CP pattern from the LSI design layout pattern is based on the standard cell. It is the same as the standard approach. For this reason, it is difficult to sufficiently solve the above-described problems.

  In addition, when a CP pattern in which holes are formed at all points on the lattice is prepared and a part of the CP pattern is selected and exposed by the first shaping aperture, one-dimensional or two-dimensional pattern is formed on the lattice. Only applicable to holes that are continuous in dimension. Accordingly, the reduction in the number of shots is not as easy as the former, and it is difficult to improve the throughput.

  As described above, when the exposure is performed by the CP method, it may be difficult to improve the throughput.

  Accordingly, an object of the present invention is to provide a pattern extraction method, a pattern extraction apparatus, and a semiconductor device manufacturing method capable of improving throughput in the exposure by the CP method.

  In order to achieve the above object, the present invention provides a pattern extraction method for extracting a CP pattern from a layout pattern, the first step of generating a layout pattern matrix by dividing the layout pattern into a matrix shape, and A second step of performing a convolution operation on the layout pattern matrix generated in the first step by a kernel matrix, and a CP from the layout pattern based on a result of the convolution operation in the second step. And a third step of extracting a pattern.

In order to achieve the above object, the present invention provides a pattern extraction method for extracting a CP pattern from a layout pattern, the first step of generating a layout pattern matrix by dividing the layout pattern into a matrix shape, and A second step of performing a convolution operation on the layout pattern matrix generated in the first step with a first kernel matrix arranged so that weighting coefficients are radially symmetric through the center;
Based on the result of the convolution operation in the second step, a third step of extracting the candidate pattern of the CP pattern from the layout pattern, and the candidate pattern extracted in the third step in the layout pattern Based on the number of repeated repetitions and the number of holes of the candidate pattern, a fourth step of specifying the candidate pattern as the CP pattern, and the layout pattern matrix generated in the first step, A fifth step of performing a convolution operation with a second kernel matrix arranged so that a weighting coefficient rotates 180 degrees with respect to the center with respect to the CP pattern specified in the fourth step, and the fifth step And a sixth step of extracting the CP pattern from the layout pattern based on the result of the convolution operation at .

  In order to achieve the above object, the present invention provides a pattern extraction apparatus for extracting a CP pattern from a layout pattern, and generates a layout pattern matrix by generating a layout pattern matrix by dividing the layout pattern into a matrix. A layout pattern matrix data generated by the layout pattern matrix generation means, a convolution operation means for performing a convolution operation with a kernel matrix, and a result of the convolution operation performed by the convolution operation means And CP pattern extracting means for extracting the CP pattern from the layout pattern.

  In order to achieve the above object, the present invention provides a method for manufacturing a semiconductor device, comprising: a first step of generating a layout pattern matrix by dividing a layout pattern of the semiconductor device into a matrix; and A CP pattern is extracted from the layout pattern based on a second step of convolving the layout pattern matrix generated in one step with a kernel matrix and a result of the convolution operation in the second step. A third step.

  According to the present invention, it is possible to provide a pattern extraction method, a pattern extraction device, and a semiconductor device manufacturing method capable of improving throughput in the exposure by the character projection (CP) method.

  An embodiment of the present invention will be described.

(Configuration of exposure apparatus)
FIG. 1 is a block diagram showing a main part of an exposure apparatus 1 according to an embodiment of the present invention.

  As shown in FIG. 1, the exposure apparatus 1 includes an electron gun 101, a first shaping aperture AP1, a shaping deflector 104, a second shaping aperture AP2, a positioning deflector 107, a wafer table 108, and a control unit. 109, and performs exposure by the CP method. That is, the exposure apparatus 1 transfers the CP pattern repeated in the LSI layout pattern to the photoresist film PR formed on the wafer W.

  The electron gun 101 includes a cathode electrode, an anode electrode, and a grid electrode. The grid electrode guides the thermal electrons emitted from the cathode electrode to the anode electrode, and the anode electrode accelerates the electrons to generate an electron beam beam. Irradiate. Here, as shown in FIG. 1, the electron beam beam is irradiated to the first shaping aperture AP1 side.

  As shown in FIG. 1, the first shaping aperture AP <b> 1 includes a light shielding plate in which an opening is formed, and the first shaping aperture AP <b> 1 is shaped by allowing a part of the electron beam irradiated from the electron gun 101 to pass through the opening. For example, it is formed into a rectangular shape. Then, as shown in FIG. 1, the electron beam formed by the first shaping aperture AP1 is emitted to the shaping deflector 104 side.

  As shown in FIG. 1, the shaping deflector 104 condenses the electron beam shaped by the first shaping aperture AP1, and deflects the collected electron beam. Here, the shaping deflector 104 uses the electron beam beam so that the electron beam beam selectively irradiates a plurality of partitioned areas in the second shaping aperture AP2 based on a control signal from the control unit 109. To deflect.

  The second shaping aperture AP2 is a so-called CP mask, which includes a light shielding plate having an opening formed so as to correspond to the CP pattern or the variable shaping pattern, and a part of the electron beam deflected by the shaping deflector is the opening. Is formed into a shape corresponding to the CP pattern. In the present embodiment, the second shaping aperture AP2 is formed with a CP pattern extracted by a pattern extraction device 201 described later. Then, as shown in FIG. 1, the electron beam formed by the second shaping aperture AP2 is emitted to the shaping deflector 104 side.

  FIG. 2 is a plan view showing the second shaping aperture AP2 according to the embodiment of the present invention.

  As shown in FIG. 2, in the second shaping aperture AP2, the beam portions H are formed in a lattice shape on the mask substrate MS. Here, the beam portion H has, for example, 100 thin film regions 11 in which the CP pattern is formed and four opening regions 13 in which the variable forming pattern is formed, as numbered from 0 to 99. Are formed on the mask substrate MS so as to be partitioned into rectangular shapes. In the second shaping aperture AP <b> 2, the electron beam is selectively irradiated by the shaping deflector 104 on the plurality of divided areas.

  As shown in FIG. 1, the positioning deflector 107 condenses the electron beam formed by the second shaping aperture AP2, and deflects the collected electron beam. Here, a plurality of regions in the photoresist film PR formed on the wafer W placed on the wafer table 108 are selectively irradiated based on a control signal from the control unit 109. The electron beam formed by the second shaping aperture AP2 is deflected.

  The wafer table 108 includes a mounting surface, and supports the wafer W on which the photoresist film PR is formed on the mounting surface. Here, the wafer W is coated with a photoresist solution by a resist coating device (not shown) and pre-baked by a heat processing device (not shown) to form a wafer W on which a photoresist film PR is formed. It is transferred from the outside by a wafer transfer system (not shown). Then, the wafer W on which the photoresist film PR is formed is fixed to the stage by, for example, vacuum suction. The wafer table 108 is provided with a reflecting mirror (not shown), and its position and inclination are detected by a laser interferometer (not shown) installed so as to correspond to the reflecting mirror. The wafer table 108 includes a drive motor (not shown), and the x direction along the surface of the wafer W supported by the wafer table 108, the y direction orthogonal to the x direction on the surface of the wafer W, and the wafer W It moves by a drive motor in each direction with z direction perpendicular | vertical to the surface of this. The surface of the wafer W is adjusted by rotating the wafer W in the rotational directions of the respective axes in the x, y, and z directions by the drive motor. Here, based on the position and inclination of the wafer table 108 detected by the laser interferometer, the control unit 109 controls the drive motor to adjust the position and inclination of the wafer table 108.

  The control unit 109 includes a computer and a program that causes the computer to function as predetermined means, and controls each unit. The operation of each unit is controlled so as to correspond to the exposure data set based on the input device design pattern.

(Configuration of pattern extraction device)
The pattern extraction apparatus 201 according to the embodiment of the present invention will be described below.

  FIG. 3 is a block diagram of the pattern extraction apparatus 201 according to the embodiment of the present invention.

  The pattern extraction apparatus 201 includes a computer and a memory that stores a program that causes the computer to function as a predetermined unit. As illustrated in FIG. 3, the layout pattern matrix generation unit 211 and the first tatami mat The computer functions as a program as each of the convolution calculation means 212, the CP pattern candidate extraction means 213, the CP pattern identification means 214, the second convolution calculation means 215, and the CP pattern extraction means 216. A CP pattern is extracted from the layout pattern. Each means will be described sequentially.

  The layout pattern matrix generation unit 211 generates a layout pattern matrix as data by dividing a device layout pattern into a matrix.

  The first convolution operation means 212 performs a convolution operation on the layout pattern matrix generated by the layout pattern matrix generation means 211 using the first kernel matrix. In the present embodiment, the convolution operation processing is executed using the first kernel matrix arranged so that the weighting coefficients are radially symmetrical via the center of the kernel matrix.

  The CP pattern candidate extraction unit 213 extracts candidate patterns that are CP pattern candidates to be extracted from the layout pattern, based on the result of the convolution operation unit 212 performing the convolution operation on the layout pattern matrix using the first kernel matrix.

  The CP pattern specifying unit 214 determines the candidate pattern as a CP pattern based on the number of repetitions of the candidate pattern extracted by the CP pattern candidate extraction unit 213 in the layout pattern and the number of holes included in the candidate pattern. As specified.

  The second convolution operation means 215 performs convolution operation processing on the layout pattern matrix generated by the layout pattern matrix generation means 211 using the second kernel matrix. In the present embodiment, the layout pattern matrix is used by using the second kernel matrix arranged so that the weighting coefficient is rotated 180 degrees with respect to the CP pattern specified by the CP pattern specifying means 214 with respect to the center. Is a convolution operation.

  The CP pattern extraction unit 216 extracts a CP pattern from the layout pattern based on the result of the convolution operation performed by the second convolution operation unit 215.

(Operation)
The operation of the pattern extraction apparatus 201 of this embodiment will be described.

  FIG. 4 is a flowchart showing the operation of the pattern extraction apparatus 201 in the embodiment according to the present invention.

  First, as shown in FIG. 4, a layout pattern matrix is generated by dividing a device layout pattern into a matrix (S11).

  Here, after the layout pattern matrix generation unit 211 receives data about a layout pattern in which holes formed in the manufacture of a semiconductor device are laid out, the layout pattern matrix generation unit 211 divides the layout pattern into a matrix. Then, a layout pattern matrix is generated as data.

  FIG. 5 is a diagram conceptually showing the layout pattern matrix generated in the embodiment according to the present invention.

  As shown in FIG. 5, a layout pattern in which holes formed in the manufacture of a semiconductor device are laid out in a two-dimensional plane is divided into a matrix so as to be a g × h unit region. For example, the layout pattern is divided so that the unit areas are arranged 6 × 6. Along with this, 1 is arranged when a hole is located in the divided unit area, and 0 is arranged when no hole is arranged. In this way, layout pattern matrix data is generated as bitmap data.

  Next, as shown in FIG. 4, the layout pattern matrix is convolved with the first kernel matrix (S21).

  Here, the layout pattern matrix generated by the layout pattern matrix data generation means 211 is used by using the first kernel matrix K1 arranged so that the weighting coefficients are radially symmetrical via the center of the kernel matrix. The first convolution operation means 212 executes a convolution operation process.

  Specifically, as shown in the following formula (1), the layout pattern matrix P is subjected to a convolution operation with the first kernel matrix K1, and the matrix W calculated by the convolution operation processing is used as data. Get as. In this equation, the layout pattern matrix P is obtained by dividing the layout pattern so as to be a unit area of g × h in the r × s matrix, and the first kernel matrix K1 is an electron in the layout pattern. The CP region irradiated with the line beam is divided into m × n with the same grid size as the unit region obtained by dividing the layout pattern, and the unit region is used in which the weighting factor is binarized and arranged. In the matrix W calculated by the convolution operation processing, data is arranged in a matrix form (g + m−1) × (h + n−1). In the formula, [] indicates that the decimal part is rounded down, and all the matrices are composed of one or more element indexes. When the matrix element index is zero or less in the formula, Consider element value 0.

  FIG. 6 is a diagram conceptually showing how the first convolution operation means 212 performs convolution operation processing on the layout pattern matrix in the embodiment according to the present invention.

  In the present embodiment, as shown in FIG. 6A, the first kernel matrix K1 is divided so that the unit areas are arranged in 3 × 3, and 1 is arranged as a weighting coefficient in all the unit areas. Is used. 6 (b), FIG. 6 (c), and FIG. 6 (d), the unit region located at the center in the first kernel matrix K1 is located in each unit region of the layout pattern matrix P. Thus, the first kernel matrix K1 is moved with respect to the layout pattern matrix P, and added as matrix data W as shown in FIGS. 6 (e), 6 (f), and 6 (g) sequentially.

  That is, as shown in FIG. 6B, in the layout pattern matrix P, the portion corresponding to the coordinates (1, 1) is arranged as data because holes are formed. As shown in e), 1 is added to the portion corresponding to the coordinates (1 to 3, 1 to 3) in the matrix W after the convolution operation. Then, as shown in FIG. 6C, in the layout pattern matrix P, the portion corresponding to the coordinates (2, 1) is arranged as 0 because no hole is formed. In the matrix W after the convolution calculation, 0 is added to the portion corresponding to the coordinates (2-4, 1-3). Then, as shown in FIG. 6D, in the layout pattern matrix P, the portion corresponding to the coordinates (3, 1) is arranged as data because holes are formed. As shown in g), 1 is added to the portion corresponding to the coordinates (3-5, 1-3) in the matrix W after the convolution operation. For this reason, in the matrix W after the convolution operation, the element value of the coordinates (3, 1 to 3) is 2.

  Then, by repeating this operation for all the unit areas in the layout pattern matrix P, a matrix W after the convolution operation is obtained as shown in FIG.

  In this way, the process of convolving the layout pattern matrix data with the kernel matrix is completed.

  Next, as shown in FIG. 4, candidate patterns that are candidates for the CP pattern extracted from the layout pattern are extracted (S31).

  Here, based on the result of the convolution operation unit 212 convolving the layout pattern matrix data with the first kernel matrix K1, the candidate CP pattern candidate extracted from the layout pattern is used as the CP pattern candidate extraction unit. 213 extracts. That is, a candidate pattern that is a candidate for the CP pattern is extracted using the matrix W after the convolution operation as described above.

  FIG. 7 is a diagram conceptually showing how candidate patterns that become CP pattern candidates are extracted in the embodiment according to the present invention. In FIG. 7, the layout pattern matrix P generated as shown in FIG. 5 and the matrix W after the convolution operation acquired as shown in FIG. In FIG. Then, here, as shown in FIG. 6 (h), a region where holes are formed in the layout pattern matrix P as shown in FIG. 5 (region where 1 is arranged in P in FIG. 5) is obtained. The unit area of the matrix W after the convolution operation is shown by hatching.

  In the present embodiment, as shown in FIG. 7, the center coordinates (2, 2) of the first kernel matrix K1 are moved to the portion where the element value is 2 which is the maximum value in the matrix W after the convolution operation. Then, the hole pattern included in the first kernel matrix K1 is extracted as a candidate pattern that is a candidate for the CP pattern. Specifically, as shown in FIGS. 7A, 7B, and 7C, coordinates (3, 1), (3) having an element value of 2 in the matrix W after the convolution operation are performed. , 2), (3, 3), the center coordinate (2, 2) of the kernel matrix K1 is moved to the coordinates (1, 1) and the coordinates (3, 1) in the layout pattern matrix P. A pattern of two holes to be arranged is included in the kernel matrix K1. Then, as shown in FIGS. 7D and 7E, coordinates (6, 4) and (7, 4) having an element value of 2 in the matrix W after the convolution operation are represented by the kernel matrix K1. When the center coordinates (2, 2) are moved, the pattern of two holes arranged at the coordinates (6, 2) and the coordinates (5, 4) in the layout pattern matrix P is included in the kernel matrix K1. . Therefore, in the layout pattern matrix P, a pattern composed of two holes arranged at coordinates (1, 1) and coordinates (3, 1), and coordinates (6, 2) and coordinates (5, 5 in the layout pattern matrix P). 4) is extracted as a candidate pattern.

  In the present embodiment, the maximum value of the element values in the matrix W after the convolution operation is continuously calculated between adjacent unit areas, and the pattern extracted as the CP pattern in the layout pattern matrix P is the kernel matrix. Duplicated extraction. For this reason, the process is executed so that the pattern is extracted only once according to a predetermined rule. Specifically, in the layout pattern matrix P, the pattern composed of two holes arranged at the coordinates (1, 1) and the coordinates (3, 1) has the first kernel matrix K1 at a plurality of positions as described above. In this case, the pattern extracted first when each unit area of the layout pattern matrix P is sequentially moved is set as a candidate pattern. That is, as shown in FIG. 7F, the pattern extracted in FIG. 6A is used as the first candidate pattern, and is generated as matrix data so that the portion corresponding to the hole is 1. Similarly, in the layout pattern matrix P, a pattern composed of two holes arranged at coordinates (6, 2) and coordinates (5, 4) is shown in FIG. The pattern extracted in step 2 is used as the second candidate pattern, and is generated as matrix data so that the portion corresponding to the hole is 1.

  Next, as shown in FIG. 4, a candidate pattern is specified as a CP pattern (S41).

  Here, based on the number of repetitions in which the candidate pattern extracted by the CP pattern candidate extraction unit 213 is repeatedly extracted as a candidate pattern in the layout pattern and the number of holes included in the candidate pattern, the CP pattern specifying unit 214 identifies the candidate pattern as a CP pattern.

  This is performed because the number of CP patterns formed on the CP mask is finite, and it is necessary to specify a pattern suitable for improving the throughput among the candidate patterns extracted in the above.

  Specifically, for example, when it is necessary to specify only one pattern as the CP pattern, the number of repetitions of the first candidate pattern extracted as described above is larger than that of the second candidate pattern in the layout pattern. If there are many, the first candidate pattern is identified as the CP pattern instead of the second candidate pattern. Note that the first candidate pattern extracted as a candidate pattern in the above includes two holes, but if a candidate pattern with many holes is extracted, the candidate pattern with many holes is designated as CP. Specify as a pattern.

  Next, as shown in FIG. 4, the layout pattern matrix data is convolved with the second kernel matrix (S51).

  Here, the second convolution operation means 215 performs convolution operation processing on the layout pattern matrix data generated by the layout pattern matrix generation means 211 using the second kernel matrix K2. In the present embodiment, the layout pattern matrix is used by using the second kernel matrix arranged so that the weighting coefficient is rotated 180 degrees with respect to the CP pattern specified by the CP pattern specifying means 214 with respect to the center. Convolution data. That is, as the second kernel matrix K2, the holes included in the CP pattern specified by the CP pattern specifying means 214 correspond to the positions rotated by 180 ° with the center of the CP pattern as the rotation axis. The one having a weighting factor of 1 and a weighting factor of 0 other than that position are used.

  FIG. 8 is a diagram showing a second kernel matrix used when the layout pattern matrix is subjected to a convolution operation process in the embodiment according to the present invention.

  In this step, for convenience of explanation, not the layout pattern matrix shown in the above-mentioned figure but the layout pattern matrix P shown in FIG. 8A, the CP pattern shown in FIG. It will be described as being specified. That is, as shown by 1 as an element value in FIG. 8B, a CP pattern in which holes are arranged at coordinates (2, 1), (3, 2), (2, 3) in a 3 × 3 matrix. It shall be specified.

  For this reason, in the present embodiment, as shown in FIG. 8C, the second kernel matrix K2 is divided so that the unit area is a 3 × 3 matrix, and the weighting coefficient is centered in the unit area. Is used so that the CP pattern is rotated 180 degrees. That is, in a 3 × 3 matrix, a weighting factor of 1 is arranged as an element value at coordinates (2,1), (1,2), (2,3), and the others are those in which 0 is arranged. Use.

  Then, in the same manner as the above process, the layout pattern matrix P shown in FIG. 8A is subjected to a convolution operation using the second kernel matrix K2 shown in FIG. As shown in (d), the matrix W after the convolution operation is acquired as data. In FIG. 8D, the layout pattern matrix P shown in FIG. 8A and the matrix W after the convolution operation obtained here are overlapped so that their unit areas correspond to each other. They are also shown. Here, the region where holes are formed in the layout pattern matrix P is illustrated by hatching the unit region of the matrix W after the convolution operation obtained as shown in FIG.

  Next, as shown in FIG. 4, a CP pattern is extracted from the layout pattern (S61).

  Here, based on the result of the convolution operation performed by the second convolution operation unit 215 as described above, the CP pattern extraction unit 216 extracts the CP pattern from the layout pattern.

  In the present embodiment, the maximum element value is 3 in the matrix W after the convolution calculation, as shown in FIG. For this reason, the center coordinate (2, 2) of the second kernel matrix K2 is moved to the portion where the maximum value is 3, and the pattern included in the portion corresponding to the second kernel matrix K2 is changed to the layout pattern matrix. As a CP pattern. Then, data about the position where the CP pattern is arranged in the layout pattern matrix is acquired.

  Thereafter, using the extracted data relating to the CP pattern, the corresponding portion in the layout pattern is replaced with the CP pattern to generate exposure data. Then, the exposure apparatus 1 performs exposure under the conditions corresponding to the generated exposure data.

  As described above, in this embodiment, after the layout pattern matrix P is generated by dividing the layout pattern into a matrix, the generated layout pattern matrix P is radially symmetric with the weighting coefficient at the center. A convolution operation is performed with the first kernel matrix K1 arranged as follows. Then, based on the result of the convolution operation, a CP pattern candidate pattern is extracted from the layout pattern. For this reason, in the present embodiment, candidate patterns for the CP pattern can be extracted by a simple process, so that the processing time can be reduced. That is, since processing is performed by integer arithmetic, high-speed processing can be realized. Therefore, this embodiment can improve the throughput.

  In addition to this, after specifying the candidate pattern as a CP pattern based on the number of repetitions of the extracted candidate pattern in the layout pattern and the number of holes of the candidate pattern, the layout pattern matrix P Is convolved with the second kernel matrix K2 arranged so that the weighting coefficient rotates 180 degrees with respect to the center of the identified CP pattern. Then, based on the result of the convolution operation, a CP pattern is extracted from the layout pattern. For this reason, in the present embodiment, the CP pattern can be extracted by a simple process, as described above, and the processing time can be reduced. Thus, throughput can be improved.

  In the above embodiment, the pattern extraction device 201 corresponds to the pattern extraction device of the present invention. In the above embodiment, the layout pattern matrix generation unit 211 corresponds to the layout pattern matrix generation unit of the present invention. Moreover, in said embodiment, the 1st convolution operation means 212 is corresponded to the convolution operation means of this invention. In the above embodiment, the CP pattern candidate extraction unit 213 corresponds to the CP pattern extraction unit of the present invention. In the above embodiment, the second convolution operation means 215 corresponds to the convolution operation means of the present invention. In the above embodiment, the CP pattern extraction unit 216 corresponds to the CP pattern extraction unit of the present invention.

  Moreover, when implementing this invention, it is not limited to said embodiment, A various deformation | transformation form is employable.

  For example, the kernel matrix is not limited to one having a unit area of 3 × 3.

  FIG. 9 and FIG. 10 are diagrams showing modifications of the embodiment according to the present invention. FIG. 9 shows a case where the first kernel matrix K1 performs a convolution operation, and FIG. 10 shows the second kernel matrix K2. This shows a case of performing a convolution operation in each of the above. In FIG. 9, (a) shows the first kernel matrix K1, (b) shows the matrix W obtained by the convolution operation, and (c) shows the extracted candidate patterns. In FIG. 10, (a) shows the second kernel matrix K2, and (b) shows the matrix W obtained by the convolution operation. As shown in FIGS. 9 and 10, a kernel matrix having various unit areas can be used, such as a 10 × 10 kernel matrix.

  Further, the first kernel matrix K1 is not limited to the case where all weighting factor values are 1.

  FIG. 11 is a diagram showing a first kernel matrix K1 as a modification of the embodiment according to the present invention.

  As the first kernel matrix K1, one in which the weighting coefficient values are arranged so as to be radially symmetric through the center of the kernel matrix can be applied. For example, as shown in FIG. In other words, 0 may be arranged as a weighting factor between portions where 1 is arranged.

  In addition to the above, in the above-described embodiment, the case where the process of performing the convolution operation with each of the first kernel matrix K1 and the second kernel matrix K2 has been described, but the present invention is not limited thereto. For example, when it is not necessary to extract a candidate pattern of a CP pattern as in the case of using an existing CP mask, it is not necessary to perform a convolution operation with the first kernel matrix K1.

  Further, since it can be applied at the time of pattern matching, it can also be applied to pattern inspection and interlayer overlay processing.

FIG. 1 is a block diagram showing a main part of an exposure apparatus 1 according to an embodiment of the present invention. FIG. 2 is a plan view showing the second shaping aperture AP2 according to the embodiment of the present invention. FIG. 3 is a block diagram of the pattern extraction apparatus 201 according to the embodiment of the present invention. FIG. 4 is a flowchart showing the operation of the pattern extraction apparatus 201 in the embodiment according to the present invention. FIG. 5 is a diagram conceptually showing the layout pattern matrix generated in the embodiment according to the present invention. FIG. 6 is a diagram conceptually showing how the first convolution operation means 212 performs convolution operation processing on the layout pattern matrix in the embodiment according to the present invention. FIG. 7 is a diagram conceptually showing how candidate patterns that become CP pattern candidates are extracted in the embodiment according to the present invention. FIG. 8 is a diagram showing a first kernel matrix used when the layout pattern matrix is subjected to a convolution operation process in the embodiment according to the present invention. FIG. 9 shows a matrix in the case of performing a convolution operation with the first kernel matrix K1 as a modification of the embodiment according to the present invention. FIG. 10 shows a matrix in the case of performing a convolution operation with the second kernel matrix K2 as a modification of the embodiment according to the invention. FIG. 11 is a diagram showing a first kernel matrix K1 as a modification of the embodiment according to the present invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Exposure apparatus, 101 ... Electron gun, AP1 ... 1st shaping | molding aperture, 104 ... Molding deflector, AP2 ... 2nd shaping | molding aperture, 107 ... Positioning deflector, 108 ... Wafer table, 109 ... Control part 201 ... Pattern extraction apparatus (Pattern extraction device), 211 ... Layout pattern matrix generation means (layout pattern matrix generation means), 212 ... First convolution calculation means (convolution calculation means), 213 ... CP pattern candidate extraction means (CP pattern extraction means) , 214 ... CP pattern specifying means, 215 ... Second convolution operation means (convolution operation means), 216 ... CP pattern extraction means (CP pattern extraction means)

Claims (6)

A pattern extraction method for extracting a CP pattern from a layout pattern,
A first step of generating a layout pattern matrix by dividing the layout pattern into a matrix;
A second step of convolving the layout pattern matrix generated in the first step with a kernel matrix;
And a third step of extracting the CP pattern from the layout pattern based on the result of the convolution operation in the second step. The pattern extraction method according to claim 1, wherein in the second step, the kernel matrix is arranged such that weighting coefficients are radially symmetric through the center of the kernel matrix. The said kernel matrix is arrange | positioned so that a weighting coefficient may rotate 180 degree | times on the basis of the center of the said kernel matrix with respect to the specific pattern specified as said CP pattern in the said 2nd process. Pattern extraction method. A pattern extraction method for extracting a CP pattern from a layout pattern,
A first step of generating a layout pattern matrix by dividing the layout pattern into a matrix;
A second step of performing a convolution operation on the layout pattern matrix generated in the first step with a first kernel matrix arranged so that weighting coefficients are radially symmetric through the center;
A third step of extracting a candidate pattern of the CP pattern from the layout pattern based on the result of the convolution operation in the second step;
A fourth step of identifying the candidate pattern as the CP pattern based on the number of repetitions of the candidate pattern extracted in the third step and the number of holes of the candidate pattern;
The layout pattern matrix generated in the first step is arranged such that a weighting coefficient rotates 180 degrees with respect to the CP pattern specified in the fourth step with reference to the center. A fifth step of performing a convolution operation with a kernel matrix;
And a sixth step of extracting the CP pattern from the layout pattern based on the result of the convolution operation in the fifth step. A pattern extraction apparatus for extracting a CP pattern from a layout pattern,
A layout pattern matrix generating means for generating a layout pattern matrix by dividing the layout pattern into a matrix;
A convolution operation means for convolving the layout pattern matrix data generated by the layout pattern matrix generation means with a kernel matrix;
A pattern extraction apparatus comprising: CP pattern extraction means for extracting the CP pattern from the layout pattern based on the result of the convolution operation performed by the convolution operation means. A method for manufacturing a semiconductor device, comprising:
A first step of generating a layout pattern matrix by dividing a layout pattern of the semiconductor device into a matrix;
A second step of convolving the layout pattern matrix generated in the first step with a kernel matrix;
And a third step of extracting a CP pattern from the layout pattern based on the result of the convolution operation in the second step.

Images