DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0060] Embodiments of an optical proximity effect correcting method according to the present invention will be described below with reference to the attached drawings.

[0061] It should be noted that there is a copending U.S. patent application Ser. No. ______, entitled “METHOD FOR CORRECTING PHOTO-CONTIGUOUS EFFECT DURING MANUFACTURE OF SEMICONDUCTOR DEVICE ”, claiming a priority based on Japanese patent application No. Heisei 10-328080, invented by Keiichiro Tounai who is one of inventors of the present application, and assigned to an assignee who is an assignee of the present application. The content of the copending US application is incorporated herein by reference.

[0062] At first, a first embodiment is described.

[0063] This embodiment is a technique for modifying a wiring pattern in a layout data which is not yet modified, to generate a mask drawing data based on the modified wiring pattern, in accordance with the optical proximity effect correcting method of the semiconductor manufacturing process.

[0064] As shown in FIG. 1, a computer 66 stores therein the layout data which is not yet modified. The layout data which is not yet modified includes a wiring (design pattern) Pd which is a wiring layer pattern or an element thereof and a contact Ct which is a contact layer pattern or an element thereof.

[0065] The design pattern Pd is designed at an interval equal to or greater than a minimum space width sp 2 (for example, 0.18 μm or 0.24 μm) defined in accordance with a minimum design rule. A single correcting pattern 11 having a horizontally-reclined U-shaped style when it is viewed on a flat surface is added to an end of the design pattern Pd so as to surround three sides of the end. Accordingly, the proximity effect correction is carried out.

[0066] When a width of the correcting pattern 11 is assumed to be W 3 , a minimum space width sp 3 is narrower by 2*W 3 than the minimum space width sp 2 in the design, as the result of the addition of the correcting pattern 11 . Here, the space width implies a width in a left and right direction of FIG. 1 (an extending direction of the design pattern Pd), and the pattern width implies a width in an upper and lower direction of FIG. 1 .

[0067] In the above-mentioned case, a mask on which the proximity effect correction is performed has a problem that a resolution is deteriorated when the mask is transcribed. The original design pattern Pd at the beginning of the design is designed such that the space (sp 2 ) between the design patterns Pd is not smaller than a space corresponding to a practical resolution of an exposing apparatus, in accordance with the minimum design rule. Here, the practical resolution of the exposing apparatus implies a resolution when the exposing apparatus exposes a pattern. However, after the proximity effect correction, the minimum space width (sp 3 =sp 2 −2*W 3 ) between the correcting patterns 11 is not smaller than the space corresponding to the practical resolution.

[0068] At first, as shown in FIG. 2 , the proximity effect correcting mask in which the correcting pattern 11 is added to the design pattern Pd is detected for the space width between the correcting patterns 11 . As the detected result, a portion where the space width between the correcting patterns 11 is equal to or less than a predetermined space width sp 1 is extracted as an error pattern 1 (EP 1 ). As shown in FIG. 3 , the space width and the pattern width of the error pattern 1 are expanded by a certain width W 1 to generate an error pattern 2 .

[0069] As shown in FIG. 4, a portion overlapping with the error pattern 2 is removed from the correcting patterns 11 , 11 . Thus, the correcting patterns 11 , 11 are modified so as to enlarge the space width between the correcting patterns 11 (the modified correcting patterns are denoted by symbols 11 a , 11 b ). Accordingly, in the space between the correcting patterns 11 a , 11 b , the deterioration of the resolution in the mask transcription is protected without being smaller than the space corresponding to the practical resolution of the exposing apparatus.

[0070] At a next step, the computer 66 assumes these correcting patterns 11 a , 11 b to be modified wiring portions, and adds the modified wiring portions 11 a , 11 b to the design patterns Pd to generate a modified wiring pattern. The computer 66 stores the modified wiring pattern in a modified layout data memory section (not shown) in the computer 66 .

[0071] At a next step, the computer 66 generates a mask drawing data based on the modified wiring pattern. Here, a method of determining a width Lk of the correcting pattern 11 (refer to FIG. 10 ) and a method of calculating a value of the certain width W 1 are described with reference to FIG. 5 .

[0072] A graph of FIG. 5 shows a change of an edge deviation backward amount in a correcting width. In the graph of FIG. 5, a vertical axis indicates the edge deviation backward amount (nm). Here, as shown in FIG. 9 , the edge deviation backward amount implies an amount Lb reduced (deviated) from a design pattern P, when the design pattern P to which a correcting pattern Ph is added is reduced (deviated) by the optical proximity effect after the exposure transcription. A edge deviation backward amount Lb of 0 implies a situation that the correcting is optimally carried out and thereby a desired pattern is attained after the exposure transcription. In short, the edge deviation backward amount Lb of 0 implies a situation that only a portion of the correcting pattern Ph is reduced (deviated).

[0073] A horizontal axis in the graph of FIG. 5 shows the correcting width (μm). Here, the correcting width implies a width Lk of the correcting pattern Ph added to the design pattern P as shown in FIG. 10 . The correcting pattern Ph is generated in a single horizontally-reclined U-shaped style when it is viewed on a flat surface. The correcting pattern Ph has the same separation distances Lk from a short side Ps and a long side Pl of the design pattern P, respectively.

[0074] In FIG. 5 , the parameter includes a combination of an arrangement situation of the design pattern P and a wiring width Hk of the design pattern P. Namely, [Isolation 0.24] indicates that the design pattern P is singly arranged as shown in FIG. 10 and a wiring width Hk of the design pattern P is 0.24 μm. [Opposition 0.30] indicates that as shown in FIG. 11 , the design patterns P are arranged opposite to each other and each wiring width Hk of these design patterns P is 0.30 μm. In this opposite arrangement, a distance Hl between the short sides Ps of the design patterns P and a distance Hl between the long sides Pl thereof are both a value defined in accordance with the minimum design rule (for example, 0.24 μm)

[0075] The values of the width Lk of the correcting pattern 11 and the certain width W 1 are determined from the following viewpoints. The wiring width Hk and the arrangement state (the isolation or the opposition) of the design pattern P (Pd) are not uniform. Thus, the values are determined so as to give a desirable situation in which the edge deviation backward amount is small, to the design patterns P in all cases.

[0076] As shown in FIG. 5 , in the states of [Isolation 0.24], [Opposition 0.30] and [Isolation 0.30], when the width Lk of the correcting pattern 11 is [0.06 μm], the edge deviation backward amounts are all closer to [0]. For example, in a case of [0.24 μm Width Isolation Pattern], when the correcting width is [0.06 μm], the edge deviation backward amount is reduced to 50 nm. Thus, at first, the widths Lk of the correcting patterns 11 are uniformly set to [0.06 μm].

[0077] However, in the case of [0.30 μm Width Opposition Pattern], a 70 nm pattern is conversely expanded when the correcting width is [0.06 μm]. In short, this results in an over correcting. The wiring pattern is generated at a position exceeding a design (pattern) value in a direction opposite to a deviated direction to thereby drop the resolution. Thus, the pattern can not be resolved by a small variation of an exposure amount.

[0078] Thus, if the examples shown in FIGS. 2 to 4 are [0.30 μm Width Opposition Pattern], the certain width W 1 is set to [0.02 μm]. Accordingly, the width Lk of the correcting pattern 11 is reduced from the original 0.06 μm to 0.04 μm. The edge deviation backward amount becomes 0 (satisfying the pattern value). Hence, an erroneous connection to opposite other wiring patterns is avoided.

[0079] Here, the step for detecting the space width between the correcting patterns 11 shown in FIG. 2 is performed to check whether or not the design pattern Pd (P) is at the [opposite] state. As shown in FIG. 4 , after the correcting pattern 11 is modified, micro projections 15 a , 15 b remain in the ends of the modified correcting patterns 11 a , 11 b . These projections 15 a , 15 b may be detected, for example, as dust adhered to the mask, when the defect of the mask is detected (as pseudo defect). This results in a factor of disturbing the excellent detection of the mask defect.

[0080] The reason why the projections 15 a , 15 b may easily lead to the pseudo defect when the mask defect is detected is as follows. The sizes of the projections 15 a , 15 b on the mask pattern are micro and also equal to or smaller than the practical resolution when the mask is manufactured. Thus, a difference between a photographed image and a reference pattern before the mask transcription is detected as a large difference, because of the severe deterioration of the pattern form in the mask transcription and the limitation on the resolution in a photographing device, such as a CCD camera and the like, for the mask defect detection.

[0081] Then, in the first embodiment, a portion where a pattern width is equal to or less than a certain width W 2 is extracted from the modified correcting patterns 11 a , 11 b . The portion is removed from the modified correcting patterns 11 a , 11 b . Accordingly, the defect of the mask can be easily detected without erroneously detecting the pseudo defect.

[0082] That is, the first embodiment discloses the following technique. A portion where the space width between the correcting patterns 11 , 11 is equal to or less than a predetermined value sp 1 is extracted as an error pattern 1 (EP 1 ) from the proximity effect correcting mask in which the correcting patterns 11 , 11 are added to the design patterns Pd, Pd. Next, an error pattern 2 (EP 2 ) is generated in which the error pattern 1 (EP 1 ) is expanded by a certain width W 1 . Then, a portion overlapping with the error pattern 2 (EP 2 ) is removed from the correcting patterns 11 , 11 to thereby modify the space width between the correcting patterns 11 , 11 . Next, a portion where a pattern width between the modified correcting patterns 11 a , 11 b is equal to or less than another predetermined value w 2 is extracted and then removed from the modified correcting patterns 11 a , 11 b.

[0083] The reason why the projections 15 a , 15 b are generated is that as shown in FIG. 1 , axis lines Lc, Lc of the design patterns Pd, Pd opposite to each other are not in line with each other. If the axis lines Lc, Lc of the design patterns Pd, Pd are located on the same line as shown in FIG. 6 , the projections 15 a , 15 b never remain after the removal of the portion overlapping with the error pattern 2 (EP 2 ). In this case, only portions indicated by symbols K are removed, and the projections 15 a , 15 b do not remain.

[0084] In addition to the cases shown in FIGS. 1 to 4 in which the axis lines Lc, Lc of the design patterns Pd, Pd are not in line with each other, the case that the pseudo defect may be detected is considered as follows.

[0085] Firstly as shown in FIG. 7 , though the axis lines of the design patterns Pd, Pd are located on the same line, if the widths of added correcting patterns 11 c , 11 d are different from each other, minor projections 16 a , 16 b remain on both sides of the correcting pattern 11 d having a larger width after the correcting patterns 11 c , 11 d are removed partly to be modified. There may be a possibility that the projections 16 a , 16 b are detected as the pseudo defect. In the case of FIG. 7 , the method similar to that of removing the projections 15 a , 15 b in FIG. 4 may be employed in order to remove the projections 16 a , 16 b.

[0086] Secondly as shown in FIG. 8 , the axis lines Lc, Lc of the design patterns Pd, Pd are largely deviated from each other. Consequently, only minor areas indicated by symbols 17 a , 17 b are removed when the correcting patterns 11 e , 11 f are removed partly to be modified. At this time, since the removed portions 17 a , 17 b are minor, it may be considered that an occurrence of “break” (pseudo defect) in the mask is erroneously detected in the mask defect detection. Here, the following manner may be employed so as to avoid the occurrence of the minor removal portions 17 a , 17 b . In FIG. 2 , if a portion where the space width between the correcting patterns is equal to or less than the sp 1 is extracted as the error pattern 1 (EP 1 ), the error pattern 1 is removed when the width of the extracted error pattern 1 (pattern width) is equal to or less than the W 2 . The processes in FIGS. 3 and 4 after that are not carried out. In such a case, the minor removal portions 17 a , 17 b are never brought about since the removing the correcting pattern is not carried out.

[0087] A second embodiment will be described below. In the first embodiment, the correcting pattern 11 having a horizontally-reclined U-shaped style when it is viewed on a flat surface is generated so as to surround the three sides of the ends of the design pattern Pd. Then, the above-mentioned modification is performed on the correcting pattern 11 .

[0088] On the contrary, a correcting figure described in the second embodiment is added in view of the following background.

[0089] As shown in FIG. 12, a pattern 41 in which the contact Ct and the wiring Pd overlap with each other and a pattern 42 having only the wiring Pd are used to design the pattern, in many cases. In order to connect the contact Ct in the pattern 41 to the pattern 42 of the wiring Pd, the pattern 41 and the pattern 42 may be in contact with each other. Thus, as denoted by a symbol 43 , a pattern arrangement may be considered in which only a part of the pattern 41 is projected which has a wiring width Hb different from a wiring width Ha of the wiring pattern 42 and the like. Actually, there may be a design based on such arrangement. However, this causes a pattern 43 in a form of small notch to be brought about.

[0090] The inhibition of the pattern in the form of small notch causes the number of CAD check items to be increased. Thus, this can not be inhibited since an operational speed is made slower. Hence, this causes the occurrence of the pattern arrangement 43 in which the contact Ct is not present in an inline of the pattern 42 at the end of the pattern Pd and the like.

[0091] It may be considered to limit the design rule so as not to bring about such a pattern 43 . However, the method of limiting the design rule and the method of validating the design rule are not still established. Such limitation and validation may be considered to drop a design efficiency and largely increase a time for the validation process. Thus, the subject is the manner of carrying out the proximity effect correction of the pattern 43 in which only a part of the pattern 41 is projected.

[0092] In the second embodiment, the correcting pattern (correcting figure) is added as follows, differently from that of the first embodiment.

[0093] As shown in FIGS. 13 and 14 , the layout data which is not yet modified is stored in the computer 66 . The correcting figure is added to the layout data which is not yet modified, by using the following procedure.

[0094] As shown in FIGS. 13 and 14 , the layout data which is not yet modified includes a wiring 21 which is a wiring layer pattern or an element thereof and a contact 22 which is a contact layer pattern or an element thereof.

[0095] (1) At first, at a first step, an area 23 formed by sides separated by a distance C from each of straight specific sides 22 a , 22 b adjacent to each other in the contact 22 is set as an angle margin area 23 .

[0096] (2) Next, at a second step, the computer 66 detects angles 24 , 24 . . . of 90 degrees of the wiring pattern 21 located in the angle margin are 23 .

[0097] As the result of the steps (1) and (2), the proximity effect correction is performed on only the angles 24 , 24 . . . belonging to a rectangular annular area 23 in which a width surrounding a square contact 22 is C. Thus, the proximity effect correction is not performed on a wiring portion in which the contact 22 is not present. Accordingly, correcting figures cannot be required in the portion where a contact 22 is not present. Hence, it is possible to reduce the increase of a data size of a mask data.

[0098] (3) At a third step, the computer 66 regards a side 25 between the detected angles 24 , 24 . . . as an end side 25 of the wiring pattern 21 .

[0099] (4) At a fourth step, as shown in FIGS. 15 and 16 , the computer 66 adds correcting FIGS. 31 g , 31 l , 31 q each of which has a predetermined width H and a predetermined length L, to each of the end side 25 and sides 26 , 26 . . . adjacent to the end side 25 .

[0100] (5) At a fifth step, the computer 66 assumes these correcting FIGS. 31 g , 31 l , 31 q to be modified wiring portions, and adds the modified wiring portions 31 g , 31 l , 31 q to the wiring pattern 21 to generate a modified wiring pattern. The computer 66 stores the modified wiring pattern in a modified layout data memory section (not shown) in the computer 66 .

[0101] (6) At a sixth step, the computer 66 generates a mask drawing data based on the modified wiring pattern.

[0102] Also, as shown in FIG. 15 , if the notch is not tentatively present, a width from a side 29 a of the wiring pattern 21 is, for example, about 0.06 μm (refer to a symbol Hn), in the correcting FIG. 31 l . However, in the example of FIG. 15 , the correcting FIG. 31 l is set on the basis of a width Hm, of the end side 25 , shorter than a width Hp of the wiring pattern 21 . Thus, the added correcting FIG. 31 l has a width He, which is about 0.01 and 0.02 μm from the side 29 a.

[0103] There may be a case that a portion M, of the correcting FIG. 31 l , which is about 0.01 to 0.02 μm from the side 29 a of the wiring pattern 21 is not recognized as a part of the correcting FIG. 31 l when the mask defect is detected and that it is erroneously detected as dust. In this case, a projection width He (0.01 to 0.02 μm in the example of FIG. 15 ) from the side 29 a in the correcting FIG. 31 l is not constant, since it is varied in accordance with a difference between the wiring width Hp and the width Hm of the end side 25 (a notch Kr of the end of the wiring pattern 21 ). This fact causes the detection of the mask defect to be further difficult.

[0104] Also, as shown in FIG. 16 , there may be a case that an area which is not covered by the correcting FIGS. 31 g , 31 q as indicated by a symbol E occurs in a portion where the correcting FIG. 31 q and the correcting FIG. 31 g are adjacent to each other, because of the relation between the two end sides 25 , 25 . Also, this area E may be erroneously recognized as “break” in an inspection of the mask. Similarly, the above-mentioned area D may be erroneously recognized as “break” in the inspection of the mask.

[0105] So, the second embodiment discloses the following technique. If the end of the wiring pattern 21 is corrected, a correcting FIG. 31 is generated on the basis of a side of a pattern generated without consideration of the minor irregularities (a part of a contact pattern 22 and the like) located at the end of the wiring pattern 21 .

[0106] In short, as shown in FIG. 17, a correcting FIG. 33 having a length L 1 is added to a wiring pattern Pd having a wiring width W 11 originally. On the contrary, as shown in FIG. 18 , if only a part of a wiring 22 and a contact are projected from an end of the pattern wiring Pd, a size of a correcting FIG. 34 is determined on the basis of a width W 22 of the projected contact 22 (as an end side 25 of the wiring pattern Pd). Thus, only the correcting FIG. 34 having a length L 2 (L 2 <L 1 ) is added. Hence, in the case shown in FIG. 18 , the size of the correcting figure is determined on the basis of the width W 11 of the wiring pattern, and it is not determined on the basis of the width W 22 of the projected contact 22 .

[0107] The second embodiment will be described below in detail with reference to FIGS. 19 to 21 .

[0108] (1) As shown in FIGS. 19 to 21 , a conventional method is firstly used to specify an angle margin 23 on the basis of a contact pattern 22 .

[0109] (2) Angles 24 , 24 . . . of a wiring pattern 21 in this specified angle margin 23 are detected.

[0110] (3) A side 25 between the detected angles 24 , 24 . . . is assumed to be an end side 25 . In FIGS. 19 and 20 , a single end side 25 is detected. And, two end sides 25 ( 25 a , 25 b ) are detected in FIG. 21 . The items (1) to (3) are identical to those of the conventional method.

[0111] (4) As shown in FIGS. 19 to 21 , a side 28 parallel to a single noted end side 25 ( 25 a ) is searched in the angle margin 23 .

[0112] (5) An inner angle α of 90 degrees on the noted end side 25 ( 25 a ) or the parallel side 28 is searched. Here, the inner angle implies that a vertex of the angle (a convex portion of the angle) looks toward an upper side (a left upper side or a right upper side) in FIGS. 19 to 21 and that a lower side (a left lower side or a right lower side) is excluded.

[0113] (6) It is judged whether or not lengths Ls 1 , Ls 2 of vertical sides in contact with the noted end side 25 ( 25 a ) or the parallel side 28 are equal to or greater than a predetermined value “height”. Here, the vertical side points out a side extending in an upper and lower direction in FIGS. 19 to 21 , and implies that a side extending in a left and right direction is excluded. Here, the predetermined value “height” is, for example, 0.04 μm.

[0114] (7) If the lengths Ls (Ls 1 , Ls 2 ) of the vertical sides are equal to or greater than the predetermined value “height”, the inner angle α in contact with the vertical side is recognized as a reference angle θ. Here, the length Ls 1 in FIGS. 19 and 21 and the length Ls in FIG. 20 fall below the predetermined value “height”. Thus, the inner angle α in contact with the sides having these lengths Ls 1 , Ls is not recognized as the reference angle θ. The length Ls 2 in FIGS. 19 and 21 is equal to or greater than the predetermined value “height”. Thus, the inner angle α in contact with the side having the length Ls 2 is recognized as the reference angle θ.

[0115] (8) Next, as shown in FIGS. 19 and 21 , the end side 25 ( 25 a , 25 b ) is extended to a position corresponding to the reference angle θ in an extending direction of the end side 25 ( 25 a , 25 b ). Then, the extended side is defined as a correcting reference side 29 .

[0116] (9) A correcting FIG. 36 is added by using the conventional method, on the basis of this correcting reference side 29 . The correcting FIG. 36 generated on the basis of the correcting reference side 29 is shown in FIGS. 22 and 23 . FIG. 22 in which FIG. 15 is modified corresponds to FIG. 19 . And, FIG. 23 in which FIG. 16 is modified corresponds to FIG. 21 .

[0117] (10) At a next step, the computer 66 assumes the correcting FIG. 36 to be a modified wiring portion, and then adds the modified wiring portion to the wiring pattern 21 to generate a modified wiring pattern. The computer 66 stores the modified wiring pattern in the modified layout data memory section (not shown) in the computer 66 .

[0118] (11) At a next step, the computer 66 generates a mask drawing data based on the modified wiring pattern.

[0119] The above-mentioned steps enable the influence resulting from the minor irregularities at the end of the pattern to be removed and also enable the adequate proximity effect correction.

[0120] Here, as shown in FIG. 20 , if the reference angle θ is not present as the results of the steps (6) and (7), the correcting figure is not added. The reason why the correction is not carried out if the length Ls is equal to or less than the predetermined value “height” as shown in FIG. 20 is described.

[0121] As shown in FIG. 24 , if the wiring patterns 21 in which projection amounts of the end sides 25 as shown in FIG. 20 are equal to or less than the predetermined value “height” are opposite to each other, the addition of the correcting FIG. 36 brings about the excessive correction to thereby result in the connection in which it is embedded in a part of a gap S between the wiring patterns 21 (refer to a symbol 51 ).

[0122] This is described in comparison with the case in which a distance between correcting FIGS. 36, 36 opposite to each other as shown in FIG. 25 is Q identical to that of FIG. 24 . In FIG. 24 , the wiring patterns 21 , 21 are opposite to each other in a wide range as compared with the case of FIG. 25 . So, the gap S is formed in the portion where they are opposite to each other in this wide range. Accordingly, they may be easily connected so as to embed a part of the gap S.

[0123] The optical proximity effect correcting method in the semiconductor manufacturing process according to the present invention is provided with: adding a first correcting region ( 11 ) around a portion of a first design pattern (Pd), the portion facing a second design pattern (Pd), and a first corrected design pattern including the first correcting region ( 11 ) and the first design pattern (Pd); detecting a space between the first corrected design pattern and the second design pattern (Pd); judging whether the space is smaller than or equal to a predetermined value (sp 1 ); and deleting at least a portion of the first correcting region ( 11 ) such that the space is larger than the predetermined value (sp 1 ), when the space is smaller than or equal to the predetermined value (sp 1 ). Thus, there is no case that the interval falls below the practical resolution of the exposing apparatus. Hence, the deterioration of the resolution can be avoided in the mask transcription.