DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0069] Referring now to the figures of the drawing in detail and first, particularly, to FIGS. 1 a to 12 thereof, the first part of the invention—the detection of phase conflicts —will be explained.

[0070] In FIGS. 1 a - 1 c and 2 a - 2 d , the differences between known methods for detecting phase conflicts ( FIGS. 2 a - 2 d ) and the method according to the invention ( FIGS. 1 a - 1 c ) will be illustrated with reference to one and the same dark-field mask structure.

[0071] A dark-field mask 10 has transparent regions 1 which are to be imaged onto electrical circuit elements such as conductor tracks or the like.

[0072] In FIG. 2 b , the transparent regions are represented as a hatched polygon. Between the individual sections of the polygon there are critical regions 2 in which the distance between individual sections of the polygon is less than a predefined minimum value. The object is to assign to the individual sections of the polygon the phases which have a phase difference ΔØ=180°.

[0073] On the basis of FIG. 2 b , the method according to U.S. Pat. No. 5,923,566 leads to the definition of free spaces F ₁ , F ₂ and F ₃ , as illustrated in FIG. 2 c . The free spaces F ₁ and F ₃ indicate here the same elementary phase conflict determined by F ₁ . F ₃ is composed of F ₁ and F ₂ , in which case F ₂ does not represent a phase conflict, because of its even number of interactions (4). The same phase conflict is therefore unnecessarily indicated twice.

[0074] The method described in Moniwa et al., which was mentioned earlier, is illustrated in FIG. 2 d and supplies the three cycles (1251), (123451) and (23452). Of these cycles, only the two first mentioned ones have an uneven number of nodes and correspondingly represent two phase conflicts. The second cycle mentioned is composed of the two other elementary cycles. Because the third cycle has an even number of nodes, there is in reality only one phase conflict which is determined by the first elementary cycle and is unnecessarily indicated twice with the same cycle.

[0075] In contrast, in the method according to the invention, after the critical regions 2 have been determined, coherent regions 3 which lie outside the transparent and critical regions 1 and 2 and which are designated in FIG. 1 b as areas L 1 and L 2 are determined. These areas are subsequently also referred to as lands. The outer borders 4 of these lands are determined and their interactions or lines of contact with the critical regions 2 are determined. As is shown in FIG. 1 c , the outer border of the land L ₁ represents the single expected phase conflict in an efficient and unambiguous way. The localized phase conflict is indicated visually by means of the outer border 4 illustrated as a polygon in FIG. 1 c . On the other hand, the outer border around the land L ₂ indicates that there is no phase conflict because the number of lines of contact with critical regions 2 is even (4).

[0076] FIG. 3 a - 3 c show a more complex dark-field mask 100 with which the inadequacy of the method described in U.S. Pat. No. 5,923,566 in comparison with the present invention is illustrated. Firstly, according to FIG. 3 b the transparent regions 1 are illustrated in the form of polygons and critical regions 2 between them are determined. Then, the lands L ₁ to L ₄ are defined as illustrated. According to FIG. 3 c , those outer contours 4 a , 4 b and 4 c of the lands which have an uneven number of lines of contact with critical regions 2 indicate the localized phase conflicts. In the present case, the phase conflict, which is localized by the outer borders of L ₃ ,is not registered with the method in U.S. Pat. No. 5,923,566. The reason for this is that this complex phase conflict is restricted by two free spaces, namely the lands L ₃ and L ₄ with even numbers of interactions 8 and 6 , respectively. This example therefore shows the reliability of the method according to the invention in comparison with the method U.S. Pat. No. 5,923,566 in terms of dark-field masks.

[0077] The dark-field masks described above are shaped in such a way that the critical regions are rectangular, or more generally trapezoidal sections of different lengths which extend in one direction. However, it is also possible that two or more such trapezoidal, straight sections which extend in different directions will meet. In this case, not only are lands determined, but also overlapping regions between these trapezoidal, straight sections. An example of such a phase mask is illustrated in FIGS. 4 a - 4 c . On this phase mask 10 , three transparent regions 1 are arranged with respect to one another in such a way that critical regions 2 are produced within which the transparent regions 1 are less than a predefined minimum distance apart from one another. The critical regions 2 form a T-shaped structure in which, therefore, two rectangular, straight sections run toward one another and form an overlapping region 13 (See FIG. 4 c which shows an enlarged view of the overlapping region 13 ). The overlapping region 13 is determined by virtue of the fact that the intersecting, straight sections are drawn beyond the points at which they meet, the amount of intersection of the continuing lines which are drawn defining the overlapping region 13 . In order to determine the degenerated critical region, the overlapping region 13 is removed from the critical region 2 . The outer border 14 around the overlapping region 13 thus has three lines of contact with the three degenerated critical regions 2 a and thus signals a phase conflict on the basis of the uneven number of contact lines.

[0078] A further dark-field mask structure, which exhibits a dual T structure (2T structure) in comparison with the single T structure shown in FIGS. 4 a - 4 c is illustrated in FIG. 5 a . According to FIG. 5 b , two overlapping regions 13 are determined and these are removed from the critical, non-transparent regions 2 , resulting in the degenerated critical regions 2 a . The outer borders 14 around the overlapping regions 13 each have three lines of contact with end sections of straight, degenerated critical regions 2 a so that as a result two phase conflicts are indicated. This dual phase conflict cannot be registered with the method in the U.S. Pat. No. 5,923,566 already mentioned.

[0079] There are no such overlapping regions 13 in the dark-field masks in FIGS. 1 to 3 . As a result, in these cases the critical regions are identical to the degenerated critical regions.

[0080] In FIGS. 6 a and 6 b , a further dark-field mask is illustrated which contains a further type of region which includes a phase conflict. FIG. 6 a firstly illustrates the dark-field mask structure which has a coherent transparent region 1 which illustrates, for example, a conductor track structure which is to be imaged by the phase mask and which is surrounded by a non-transparent region. The transparent region 1 is shaped in such a way that it encloses a critical, non-transparent region 2 between two line sections, the non-transparent region 2 being less than a predefined structural size. This critical region 2 thus ends in the middle of a region composed of the transparent region 1 . In this case, it is referred to as an end region 13 a which is to be generated. The end region 13 a is generated by virtue of the fact that a boundary line or outer border 14 is drawn around the end region 13 a in such a way that it makes contact with the short side edge of the end section of the critical region 2 . Then, as mentioned, degenerated critical regions 2 a are again generated by removing possibly present overlapping regions 13 from the critical regions 2 . Because the critical region 2 does not contain an overlapping region 13 in the present case, a degenerated critical region 2 a (hatched) is generated from it without modification. The outer border 14 thus has a line of contact with the degenerated critical region 2 a so that because of the uneven number of contact lines it indicates a phase conflict.

[0081] In the text which follows, exemplary embodiments according to a second aspect of the method according to the invention relating to the application on bright-field masks are illustrated. FIG. 7 a illustrates a simple exemplary embodiment for a bright-field mask structure 20 , which contains non-transparent regions 21 against a transparent background. According to FIG. 7 b , phase-shifting elements 22 (illustrated by hatching) are first defined on each side of the non-transparent critical regions 21 . Critical regions 21 are considered to be regions which are less than a specific predefined minimum width or minimum structural size. The definition of these phase-shifting elements can be made, for example, as in U.S. Pat. No. 5,537,648 ( FIG. 6 and associated descriptive text). U.S. Pat. No. 5,537,648 is incorporated herein by reference. Then, according to FIG. 7 c , overlapping regions 23 are defined between straight sections of the critical regions 21 in the same way as already described in conjunction with the dark-field mask in FIG. 5 . By removing the overlapping regions 23 from the critical regions 21 , degenerated critical regions 21 a are generated. Large outer borders 24 with the overlapping regions 23 and lands are then generated, as shown in FIG. 7 e , which is an enlarged view of the circular detail in FIG. 7 d . A phase conflict is present precisely when an outer border 24 has an uneven number of interactions or lines of contact with degenerated critical regions 21 a . According to FIG. 7 e , the outer border 24 around the overlapping region 23 is in contact with the three degenerated critical regions 21 a and thus unambiguously signals a T phase conflict.

[0082] This phase conflict is not registered by the method according to U.S. Pat. No. 5,537,648, and this is because, when the non-directional conflict graph is applied, no cycle is produced because, specifically, only two phase-shifting elements and one coherent critical structure are present.

[0083] FIG. 7 c shows that overlapping regions 23 are also regions at which only two straight sections of the non-transparent regions 21 , which run in different directions, meet one another, in particular, therefore, buckling points of the circuit element, such as a conductor track, to be imaged. In order to determine the degenerated critical regions 21 a , they are also removed from the critical regions 21 . However, they generally do not represent phase conflicts because the outer borders associated with them have lines of contact with only two degenerated critical regions 21 a , that is to say an even number of degenerated critical regions 21 a . In general terms, there are phase conflicts at branching structures such as T branches, from which at least three straight sections start and extend in different directions.

[0084] FIGS. 8 to 11 show the mode of operation of the method according to the invention in accordance with its second aspect by means of a relatively complex bright-field mask structure.

[0085] FIG. 8 first illustrates a bright-field mask 200 with non-transparent regions 21 , which are to be imaged onto circuit elements such as printed circuit boards or the like. According to FIG. 9 , phase-shifting elements 22 (hatched polygons) are determined—as already explained—on each side of the straight sections of the non-transparent regions 21 because the non-transparent regions 21 can be classified as critical.

[0086] In this bright-field mask 200 , two further types of end region 23 a and 23 b occur in addition to the lands and the overlapping regions 23 , as in dark-field masks. Whenever straight sections of the non-transparent regions 21 end within a phase-shifting element 22 , an end region 23 a is to be generated. This is simply generated by drawing a boundary line of the end region 23 a —as shown in the circular section of FIG. 10 b, which is an enlarged view from FIG. 10 a ,—in such a way that it makes contact with the short side edge of the end section of the non-transparent region 21 . In addition, as is apparent from the circular detail in FIG. 10 c , which is an enlarged view from FIG. 10 a , an end region 23 b is generated at the point where a critical region 21 ends at a critical interaction region. Critical interaction regions are regions between phase-shifting elements 22 in which the phase-shifting elements 22 are less than a predefined distance apart. The end region 23 b is generated by virtue of the fact that a boundary line of the end region 23 b is drawn in such a way that it makes contact with the short side edge of the end section of the critical region 21 at the interaction zone.

[0087] In the center of the right-hand half of the diagram in FIG. 10 a , a further triangular branching structure which indicates a phase conflict and in which three straight non-transparent sections meet is illustrated. At the meeting point, a triangular overlapping region is formed whose outer border has lines of contact with the end edges of the three straight degenerated critical regions, and thus represents a phase conflict.

[0088] According to FIG. 11 , overlapping regions 23 are therefore determined between straight sections of the non-transparent regions 21 and end regions 23 a, b of straight sections which end in the middle of a phase-shifting element 22 or an interaction region. Degenerated critical regions are then defined by removing the overlapping regions 23 from the non-transparent regions 21 . Phase conflicts are indicated according to FIG. 11 by such large outer borders 24 of lands, overlapping regions 23 or end regions 23 a, b which enter into contact with an uneven number of end sections of the degenerated critical regions derived from the non-transparent mask fields 21 . End regions 23 a, b for bright-field masks and end regions 13 a for dark-field masks thus always signal a phase conflict because they enter into contact with precisely one end section of the non-transparent region 21 . FIG. 11 therefore shows the eight uniquely localized phase conflicts 24 relating to the bright-field mask structure in FIG. 8 because of the eight dark outer borders 24 . The second complex phase conflict on the left-hand side of the image cannot be registered with the prior art method because even numbers of the interactions 6 and 4 are assigned to the two free spaces adjoining it. There are thus two types of what are referred to as end region conflicts K 1 and K 2 . An end region conflict occurs precisely if the corresponding line ends within a phase-shifting element 22 (K 1 ) or a critical interaction region between two phase-shifting elements 22 (K 2 ). These two types of end region conflicts are shown in FIG. 11 .

[0089] FIG. 12 illustrates a portion of a gate plane that is to be manufactured with a bright-field phase mask. The conductor tracks illustrating a minimum width are critical structures and must therefore be implemented by means of phase elements arranged on each side, while the widened portions (contact pads or landing pads) do not have any critical widths.

[0090] On each side of the conductor tracks, generated phase-shifting elements are illustrated which have the two different phases 0° and 180° and are accordingly indicated by two different types of hatching. The parts of conductor tracks (gates in this case), which cannot be imaged directly with the given phase assignment because both sides of the gate are exposed with the same phase, are marked in the manner of a chessboard. In FIG. 12 , they are designated as phase conflicts 34 and 35 lying within the border 30 . A further phase conflict 36 is present outside the border 30 .

[0091] The cause of the left-hand one of the two non-localized phase conflicts is the fact that the respective gate adjoins the same phase shifter on both sides. The cause of the right-hand phase conflict is not so easily apparent. The question arises as to why it is not possible to find a phase assignment which avoids this phase conflict. The cause lies in the cycle which the border 30 shows. Along this path, five critical gates and five phase shifters are arranged in a row without gaps. The phase shifters affected by this path cannot be alternately assigned to both phases. This fact is easily apparent from the border 30 . However, only the method according to the invention makes it clear that there is actually only one, and not two phase conflicts present in the discussed part of the detail.

[0092] The second part of the method according to the invention—the elimination of the phase conflicts detected in the first part —will be explained below by means of exemplary embodiments in FIGS. 13 to 17 .

[0093] FIG. 13 a again shows the layout structure for the dark-field mask according to FIG. 2 b . The transparent regions 1 and critical regions 2 which coherently adjoin one another in this structure represent a closed layout group which is surrounded by the lands 3 , 5 and 6 . In the method according to the invention which is already described, a large outer border 4 which indicates a phase conflict is indicated around a land 5 lying further inward, by virtue of the fact that the large outer border has an uneven number of interaction lines. Then, a region border 40 is drawn around the layout group according to FIG. 13 b . The borders 4 and 40 represent polygons whose polygon edges lie partially parallel to one another and opposite one another. According to FIG. 13 c , parallel edge pairs lying opposite one another of the border 4 and of the border 40 , respectively, are determined first. In FIG. 13 c there are five such edge pairs. Connecting paths are generated between the latter and reduced to such a set of connecting paths in which each phase conflict is contained only once. According to FIG. 13 c , the connecting paths are thus reduced to an individual connecting path 45 . This connecting path 45 has a section which covers the transparent region 1 . This section is embodied as a phase limit between two regions of the transparent region 1 whose phase shifts have a phase difference of 180° from one another.

[0094] When such a phase mask is exposed, an undesired destructive interference between radiation beams, which have passed through the adjoining phase-shifting regions of the transparent region 1 and interfere with one another, occurs at the phase boundary 45 generated by the connecting path. The resulting non-exposed region must be subsequently exposed by means of a trim mask. This trim mask can advantageously be generated using the connecting path 45 which is obtained as above and which can be embodied, for example, as a transmitting region of a trim mask.

[0095] FIG. 14 illustrates a layout structure, with reference to which a bright-field mask structure is to be generated. As in FIG. 8, a bright-field mask structure with non-transparent regions 21 (dotted) which are to be imaged onto circuit elements such as conductor tracks or the like is also illustrated in FIG. 14 .

[0096] According to FIG. 15 , phase-shifting elements 22 (hatched) are determined—as already explained for FIG. 9 —on each side of the straight sections of the non-transparent regions 21 because the non-transparent regions 21 are less than a critical structural width. The phase-shifting elements 22 have two types of hatching in FIG. 14 , indicating that they have two different phase shifts with a phase difference of 180° with respect to one another. Phase conflicts occur where phase-shifting elements 22 occur with the same phase shift on each side of a non-transparent region. By way of example, three phase conflicts are designated by circles.

[0097] In FIG. 16 , for the sake of clarity the phase-shifting elements are firstly illustrated with the same type of hatching. In addition, the complete and minimal set of 12 phase conflicts, as determined by means of the method according to the invention, is designated by means of the large outer borders 4 and the outer borders 24 around overlapping regions and end regions.

[0098] Furthermore, the layout groups which are coherent in the fashion described are determined and drawn around these region borders 40 . Five layout groups or regions are produced which can be identified by means of the five region borders 40 which are illustrated. These layout groups are formed with the inclusion of any interaction regions present. In FIG. 16 , there is, for example, an interaction region bottom right, which has already been explained in detail in FIG. 10 in order to explain the end region type 23 b , and has been illustrated in enlarged form in FIG. 10 c . The end region 23 b which is contained in the interaction region, the other end regions 23 a , the overlapping regions 23 containing the phase conflicts and lands which form phase conflicts are not part of the layout group.

[0099] Then, pairs of edges which lie opposite one another between the phase conflict polygons are determined in the form of their outer borders 4 or 24 and in each case at the next adjacent region borders 40 , lying further out, and between the phase conflict polygons 4 and 24 , and a connecting path is formed in each case between the edges of each of these pairs.

[0100] The set of connecting paths obtained is then reduced to the set in which each phase conflict, i.e. each outer border 4 , 24 , by means of which a phase conflict is indicated, occurs only once.

[0101] FIG. 17 illustrates the two types of phase-shifting elements 22 whose phase shifts have a phase difference of 180° from one another once more with two different types of hatching. In addition, the regions of the connecting paths which run via phase-shifting elements 22 are illustrated as black strokes 45 . The connecting paths have already been reduced to the aforesaid set of connecting paths in which each phase conflict, i.e. each outer border 4 , 24 occurs only once. The sections 45 of the connecting paths which are obtained in this way and which cover the phase-shifting elements 22 are embodied as phase boundaries within a phase-shifting element 22 by means of which a phase-shifting element 22 is divided into two phase-shifting regions whose phase shifts have a phase difference of 180° from one another.

[0102] As is apparent from the triangular branching contained in the center of the right-hand half of FIGS. 14 to 17 , and from the associated triangular overlapping region 23 and the triangular outer border 24 , it is not always the case that the connecting paths are generated between parallel edges. In the case of such phase conflicts, a connecting path to the layout border 40 lying further out (to a buckling point between two polygon edges in the case shown) is drawn so that it starts merely from the outer border 24 .

[0103] At the interaction region already mentioned which is bottom right in the figure, a connecting path is generated at the edge of the interaction region and thus at the edge of a phase-shifting region. In this case, this connecting path indicates that the phase-shifting region which is present on the opposite side is to be given a contrasting coloring, or an opposed phase shift—as indicated by the types of hatching.

[0104] As in the case of the dark-field mask, destructive interference between radiation beams which pass through the adjoining phase-shifting elements 22 and interfere with one another also occurs here when the phase mask is exposed at the phase boundaries 45 . The resulting non-exposed region must be subsequently exposed by means of a trim mask. This trim mask can in turn advantageously be generated using the connecting paths 45 which are obtained as above and can be embodied, for example as transmitting regions of the trim mask.

[0105] The precise position of the selected connecting paths and of the phase boundaries 45 defined by them can be selected here within relatively wide limits because in many cases when selecting the minimum set of connecting paths it is also possible to obtain other connecting paths by, for example, selecting a different pair of edges and/or changing the position of the connecting path along the length of the edges which are opposite one another.