BRIEF DESCRIPTION OF THE DRAWINGS
These and other objects and advantages of the present invention will be more readily apparent from the following Detailed Description in which:
FIG. 1 depicts a series of steps in processing a layer of metallization in the prior art;
FIGS. 2A and 2B depict a series of steps in an illustrative embodiment of the invention;
FIGS. 3A and 3B depict first and second masks used in the practice of the invention;
FIGS. 4A and 4B depict a series of steps in a second illustrative embodiment of the invention; and
FIGS. 5A and 5B depict a series of steps in a third illustrative embodiment of the invention.
DETAILED DESCRIPTION
As is known in the art, several layers of metallization are formed one on top of the other on the surface of a semiconductor substrate. Patterns are formed in the layers of metallization using standard photolithographic steps so as to define conductive paths that interconnect the circuits formed in the underlying substrate. The general sequence for forming and processing one layer of aluminum metallization is shown in FIG. 1. Further details may be found in numerous texts on semiconductor processing such as S. A. Campbell, The Science and Engineering of Microelectronic Fabrication, Ch. 7 (Oxford, 2d ed. 2001) and J. D. Plummer et al., Silicon VLSI Technology, Ch. 5 (Prentice Hall, 2000).
As shown in FIG. 1, at step 10, a layer of metal is formed on the underlying surface. A uniform layer of photoresist is then formed on the metal layer at step 20. At step 30, the photoresist is exposed to actinic radiation in a pattern having features defined by a mask. Following the exposure step, portions of the photoresist are selectively removed at step 40 so as to expose portions of the underlying metal layer. As is known in the art, different types of photoresist are available, negative photoresists become less soluble in developer solution as a result of exposure to actinic radiation and positive photoresists become more soluble. Whichever type of photoresist is used, an exposure pattern is formed in the photoresist; and using well known methods the more soluble portions of the photoresist layer are removed. As a result, either the negative or the positive of this pattern is removed from the photoresist layer to expose the metal layer below. The exposed portions of the metal layer are then removed at step 50, thereby transferring the pattern from the photoresist to the metal layer. At step 60, the photoresist is removed leaving the pattern defined in the metal layer; and at step 70 an insulating layer is formed on the patterned metal layer. At step 80, vias are formed at selected places in the insulating layer to provide electrical connections to the patterned metal layer. At this point, another metal layer can be formed on top of the insulating layer using the steps just described.
In forming a structured ASIC, this process is repeated several times using standard (i.e., non-custom) masks to define the metallization layers that provide logic and hard functions such as memory, PLLs, clock and power bussing. The structured ASIC is completed using a few high resolution custom masks to define the critical metal layers.
As is known in the art, advanced technology nodes use copper metallization created by a damascene process, instead of aluminum metallization. The photolithographic process used in forming copper metallization is similar to that used in forming aluminum metallization; but in the damascene process, the work surface is a dielectric layer into which the mask pattern is transferred as a trench or a via that is subsequently filled with copper by electroplating.
The present invention alters the conventional sequence of processing steps in the formation of one or more layers of metallization. FIG. 2A is a flowchart depicting the steps of an illustrative embodiment. FIG. 2B is a series of sketches alongside the steps of FIG. 2A that depict the processing referred to in the steps. At step 120, a first layer of photoresist 210 is formed on a work surface 200 such as a layer of metallization or dielectric. At step 130, the photoresist is then exposed to actinic radiation in a pattern having features defined by a first mask such as mask 300 shown in FIG. 3A. As shown in FIG. 3A, the mask is an extremely high resolution mask and the features defined by the mask are in a regular array extending across the entire region of the photoresist where structures are to be formed. Thus, mask 300 is a standard or non-custom mask. Elements of the radiation pattern formed on the photoresist are represented as dashes 220 in FIG. 2B. It will be noted, however, that the radiation preferably is in a high frequency region, invisible to the naked eye; and the mask that forms the pattern of dashes 220 is transparent to such radiation in the region of the dashes and is opaque everywhere else.
Following the exposure step, portions of the photoresist are selectively removed at step 140 so as to expose portions 202 of the underlying work surface 200. As is known in the art, both positive and negative types of photoresist are available although FIG. 2B illustrates the use of positive photoresists and positive masks. Whichever type of photoresist is used, an exposure pattern is formed in the photoresist, and using well known methods the more soluble portions of the photoresist layer are removed. As a result, either the positive or negative of this pattern is removed from the photoresist layer to expose a first, high resolution pattern on the work surface below. The remaining portions of the photoresist layer 210 are then hard baked so that they will not be affected by subsequent processing steps.
A second layer of photoresist 230 is then formed at step 150 on the first layer of photoresist 210 and on the exposed pattern 202 on the work surface. At step 160 the second layer of photoresist is then exposed to actinic radiation in a second pattern having features defined by a second mask such as mask 310 shown in FIG. 3B. Preferably, the second mask has a lower resolution than the first mask and as a result is considerably less expensive than the first mask even though mask 310 is a custom mask. Advantageously, the lower resolution exposure of step 160 is also made at a lower frequency than the high radiation exposure of step 130 using less expensive exposure equipment. The features defined by the second mask are aligned with the features defined by the first mask. Elements of the radiation pattern formed on the photoresist are represented by dashes 240 in FIG. 2B. Again, the radiation is typically invisible; and the mask is transparent to such radiation in the region of the dashes 240 and is opaque everywhere else.
Following the exposure step, portions of the second layer of photoresist are selectively removed at step 170 so as to expose portions 204 of the underlying work surface. Again, a positive or a negative photoresist can be used, although FIG. 2B illustrates the use of positive photoresists and positive masks. As shown in the bottom sketch of FIG. 2B, the portions of photoresist removed from the second layer are aligned with the regions of the first photoresist layer from which photoresist was removed in step 140 so that the removal of portions of the second photoresist layer exposes a third pattern 204 on the work surface that is a subset of the first pattern 202 previously exposed on the work surface. In logical terms, the third pattern is the logical AND of the first and second radiation patterns.
Further, the process used for removing portions of the second photoresist layer preferably removes those portions of the second photoresist layer while leaving the first photoresist layer in place. As a result, the features of the third pattern exposed on the work surface have the high resolution of the features of the first pattern even though the third pattern was determined, in part, by the lower resolution second mask. The exposed portions of the work surface may then be processed using standard lithographic processing techniques. For example, if the work surface is a layer of metallization, portions of the metallization may be removed to define connection patterns; or if the work surface is a dielectric, portions of the dielectric may be removed prior to electroplating copper in the removed portions.
FIGS. 3A and 3B illustrate masks 300 and 310 and their relationship to the pattern being formed on work surface 200. Please note that features defined by the masks in the die periphery region are not shown for simplicity's sake. Mask 300 illustratively is a high-grade optical proximity correction (OPC) mask and/or phase shift mask (PSM) that exposes on photoresist layer 210 a regular array of circular regions through an array of transparent circular apertures 302. All other regions of mask 300 are opaque at the frequency of radiation used during the exposure step. Photoresist layer 210 is then removed in these circular regions to expose circular regions 202 on work surface 200. Mask 310 illustratively is a low-grade binary mask having opaque regions 312 and transparent regions 314 and exposes on photoresist layer 230 regions that are images of transparent regions 314. The exposed regions on photoresist layer 230 are aligned with some 304 of the previously exposed circular regions as represented in FIG. 3B. As a result, when the exposed regions on photoresist layer 230 are removed at step 170 only some 204 of the previously exposed circular regions on the work surface are again exposed. These regions may then be subject to further processing, for example, to form vias.
For example, the masks of FIGS. 3A and 3B can be used to form the interconnection and vias in Altera Corporation's Hardcopy™ structured ASICs. In such an application, mask 300 which illustratively is a high-grade optical proximity correction (OPC) mask and/or phase shift mask (PSM) is used to form a pattern on the work surface that can be used to make every connection that might be made in that layer of work surface in the structured ASIC. Mask 310 which illustratively is a low-grade binary mask is then used to form a pattern on the work surface that makes only those connections that are required in that layer of work surface in the specific structured ASIC that is desired.
In an alternative process for practicing the invention, a hard mask may be used instead of a layer of photoresist. The hard mask is a layer of material such as silicon nitride or silicon carbide. FIG. 4A is a flowchart depicting the steps of one such alternative embodiment. FIG. 4B is a series of sketches alongside the steps of FIG. 4A that depict the processing referred to in the steps. At step 510, a hard mask layer 410 is formed on a work surface 400 such as a layer of metallization or dielectric. At step 520, a first layer of photoresist 420 is formed on hard mask layer 410. At step 530, the photoresist is then exposed to actinic radiation in a pattern having features defined by a first mask such as mask 300 shown in FIG. 3A. As shown in FIG. 3A, the mask is an extremely high resolution mask and the features defined by the mask are in a regular array extending across the entire region of the photoresist where structures are to be formed. Thus, mask 300 is a standard or non-custom mask. Elements of the radiation pattern formed on the photoresist are represented as dashes 430 in FIG. 4B. It will be noted, however, that the radiation preferably is in a high frequency region, invisible to the naked eye; and the mask that forms the pattern of dashes 430 is transparent to such radiation in the region of the dashes and is opaque everywhere else.
Following the exposure step, portions of the photoresist are selectively removed at step 540 so as to expose portions 412 of the underlying hard mask layer 410. As is known in the art, both positive and negative photoresists are available, although FIG. 4B illustrates the use of positive photoresists and positive masks. Whichever type of photoresist is used, an exposure pattern is formed in the photoresist, and using well known methods the more soluble portions of the photoresist layer are removed. As a result, either the positive or negative of this pattern is removed from the photoresist layer to expose a first, high resolution pattern on the work surface below. At step 550, the exposed portions of hard mask layer are removed so as to expose portions 402 of the underlying work surface 400. Illustratively, this removal is accomplished by an etching process.
A second layer of photoresist 450 is then formed at step 560 on the exposed portions of the work surface 400 and the remaining portions of hard mask layer 410. At step 570 the second layer of photoresist is then exposed to actinic radiation in a second pattern having features defined by a second mask such as mask 310 shown in FIG. 3B. Preferably, the second mask has a lower resolution than the first mask and as a result is considerably less expensive than the first mask even though mask 310 is a custom mask. Advantageously, the lower resolution exposure of step 570 is also made at a lower frequency than the high resolution exposure of step 530 using less expensive exposure equipment. The features defined by the second mask are aligned with the features defined by the first mask. Elements of the radiation pattern formed on the photoresist are represented by dashes 460 in FIG. 4B. Again, the radiation is typically invisible; and the mask is transparent to the radiation of the region of the dashes 460 and is opaque everywhere else.
Following the exposure step, portions of the second layer of photoresist are selectively removed at step 580 so as to expose portions 404 of the work surface. Again, a positive or a negative photoresist can be used, although FIG. 4B illustrates the use of positive photoresists and positive masks. As shown in FIG. 4B, the portions of photoresist removed from the second layer are aligned with the regions of the first hard mask layer that were removed in step 550 so that the exposed portions 404 form a third pattern on the work surface that is a subset of the first pattern 402 previously exposed on the work surface.
As a result, the features of the third pattern exposed on the work surface have the high resolution of the features of the first pattern even though the third pattern was determined, in part, by the lower resolution second mask. The third portions of the work surface may then be processed using standard lithographic processing techniques.
Alternatively, a dual set of hard masks may be used. FIG. 5A is a flowchart depicting the steps of this alternative embodiment. FIG. 5B is a series of sketches alongside the steps of FIG. 5A that depict the processing referred to in the steps. At step 710, a first hard mask layer 610 is formed on a work surface 600 such as a layer of metallization or dielectric. At step 720, a first layer of photoresist 620 is formed on hard mask layer 610. At step 730, the photoresist is then exposed to actinic radiation in a pattern having features defined by a first mask such as mask 300 shown in FIG. 3A. As shown in FIG. 3A, the mask is an extremely high resolution mask and the features defined by the mask are in a regular array extending across the entire region of the photoresist where structures are to be formed. Thus, mask 300 is a standard or non-custom mask. Elements of the radiation pattern formed on the photoresist are represented as dashes 630 in FIG. 5B. It will be noted, however, that the radiation preferably is in a high frequency region, invisible to the naked eye; and the mask that forms the pattern of dashes 630 is transparent to such radiation in the region of the dashes and opaque everywhere else.
Following the exposure step, portions of the photoresist are selectively removed at step 740 so as to expose portions 612 of the underlying hard mask layer 610. As is known in the art, both positive and negative photoresists are available, although FIG. 5B illustrates the use of positive photoresists and positive masks. Whichever type of photoresist is used, an exposure pattern is formed in the photoresist, and using well known methods the more soluble portions of the photoresist layer are removed. As a result, either the positive or negative of this pattern is removed from the photoresist layer to expose a first, high resolution pattern on the work surface below. At step 750, the exposed portions of hard mask layer are removed so as to expose portions 602 of the underlying work surface 600. Illustratively, this removal is accomplished by an etching process.
At step 760 a second hard mask layer 640 is formed on the exposed portions of work surface 600 and the remaining portions of the first hard mask layer. The second hard mask layer is sufficiently different from the first hard mask layer that portions of the second hard mask layer can be removed by a process applied to both layers without significant removal of the first layer. Typically, the two hard mask layers are different materials.
A second layer of photoresist 650 is then formed at step 770 on the second hard mask layer 640. At step 780 the second layer of photoresist is then exposed to actinic radiation in a second pattern having features defined by a second mask such as mask 310 shown in FIG. 3B. Preferably, the second mask has a lower resolution than the first mask and as a result is considerably less expensive than the first mask even though mask 310 is a custom mask. The features defined by the second mask are aligned with the features defined by the first mask. Elements of the radiation pattern formed on the photoresist are represented by dashes 660 in FIG. 5B. Again, the radiation is typically invisible; and the mask is transparent to such radiation in the region of dashes 660 and is opaque everywhere else.
Following the exposure step, portions of the second layer of photoresist are selectively removed at step 790 so as to expose portions 644 of the underlying second hard mask layer 640. Again, a positive or a negative photoresist can be used. As shown in FIG. 5B, the portions of photoresist removed from the second layer are aligned with the regions of the first hard mask layer that were removed in step 750.
At step 800, the exposed portions of the second hard mask layer are removed as to expose a third pattern 604 on the work surface that is a subset of the first pattern 602 previously exposed on the work surface. Illustratively, this removal is accomplished by an etching process that removes those portions of the second hard mask layer while leaving the first hard mask layer in place. As a result, the features of the third pattern exposed on the work surface have the high resolution of the features of the first pattern even though the third pattern was determined, in part, by the lower resolution second mask. The third portions of the work surface may then be processed using standard lithographic processing techniques.
In still another embodiment of the invention, a single layer of negative photoresist is used. The practice of the invention is similar to that described in conjunction with FIGS. 2A and 2B except that the steps are carried out on a single layer of photoresist. In this process, a layer of photoresist is first formed on a work surface such as a layer of metallization or dielectric. The photoresist is then exposed to actinic radiation at a first wavelength to which the photoresist is sensitive in a pattern having features defined by a first mask such as the complement of mask 300 shown in FIG. 3A. As shown in FIG. 3A, the mask is an extremely high resolution mask and the features defined by the mask are in a regular array extending across the entire region of the photoresist where structures are to be formed. Thus, the mask is a standard or non-custom mask.
The photoresist is then exposed to actinic radiation in a second pattern having features defined by a second mask such as the complement of mask 310 shown in FIG. 3B. Preferably, the second mask has a lower resolution than the first mask and as a result is considerably less expensive than the first mask even though mask 310 is a custom mask. Advantageously, the lower resolution exposure is also made at a lower frequency than the high resolution exposure using less expensive exposure equipment. The features defined by the second mask are aligned with the features defined by the first mask.
Following the two exposure steps, the portions of the photoresist that were not exposed during either exposure step are selectively removed so as to expose portions of the underlying work surface. As a result, the areas of photoresist that were not exposed in either or both exposure steps are removed from the photoresist layer to expose a third, high resolution pattern on the work surface below. In logical terms, the third pattern is the complement of the logical OR of the first and second radiation patterns; and in the case where the first and second radiation patterns are the complements of masks 300 and 310, respectively, the third pattern is the logical AND of the first and second radiation patterns of FIG. 2B.
While the invention has been described in terms of specific embodiments, numerous variations of the invention may be practiced. For example, a wide variety of photoresists and a wide variety of hard masks (such as different hard mask materials or different numbers of masks) may be used in the practice of the invention. Where two layers of photoresist are used, care must be taken in selecting the materials to ensure that the process for removing portions of the upper layer does not affect the portion of the lower layer that remains in place.