Method for generating a deep N-well pattern for an integrated circuit design

[0001] Embodiments of the present invention relate to circuit design and layout in complementary metal-oxide semiconductor (CMOS) circuits. In particular, embodiments of the present invention relate to the design of body-bias voltage connections in CMOS circuits.

[0002] As the operating voltages for CMOS transistor circuits have decreased, variations in the threshold voltages for the transistors have become more significant. Although low operating voltages offer the potential for reduced power consumption, gate threshold voltage variations due to process and environmental variables often prevent optimum efficiency and performance from being achieved due to increased leakage currents.

[0003] Threshold voltage variations may be compensated for by body-biasing. In typical CMOS transistors, the source of the transistor has a connection to the bulk region local to the transistor. This connection may be made to the substrate or to a well in the substrate containing the transistor. Body-biasing introduces a bias potential between the bulk and the source of the transistor that allows the threshold voltage of the transistor to be adjusted electrically.

[0004] Whereas the typical CMOS transistor is a three-terminal device, the body-biased CMOS transistor is a four-terminal device, and thus requires a more complex interconnect scheme. Connections for biasing may be made on the substrate surface using conventional metal/dielectric interconnects similar to those used for typical gate, drain, and source connections, or they may be made using buried complementary well structures.

[0005] For example, in a p-type substrate with a population of surface n-wells that contain p-channel field effect transistors (PFETs), a buried n-type layer may be formed in the substrate at a depth that allows for contact with the bottom of the N-wells, while providing sufficient clearance with respect to n-channel field effect transistors (NFETs) in the substrate.

[0006] Generally, prior art solutions assume that the resolvability of buried well features are of the same order as surface features. For example, some solutions teach forming apertures registered to transistor channel regions.

[0007] However, processes that try to form apertures in this way may end up with disconnected bits of buried N-well along with numerous width and spacing violations due to the size disparity between surface features and buried features.

[0008] The relative size of buried layer features is about an order of magnitude larger than surface features. With this built-in disparity, the mesh generating process and rules checking should accommodate a range of coverage and alignment issues not contemplated in the prior art. Thus, the prior art does not appreciate the disparity that exists in geometric sizes and spacing of buried substrate layers.

[0009] Thus, a need exists for a method for the design and layout of mask layers for patterned deep N-wells. There is also a need for a flexible and computationally efficient method that is adaptable to conventional computer aided design (CAD) tools.

[0010] Accordingly, embodiments of the present invention provide a method for generating and checking a deep N-well mask pattern that is computationally efficient and flexible. The method provides for the development of a pattern with a hierarchical structure that is subsequently edited and flattened while making efficient use of data storage and computational resources.

[0011] A method for the design and layout for a patterned deep N-well is disclosed. A tile is specified as a building block for the deep N-well pattern. The tile comprises a first element on a first layer and may also comprise a second element on a second layer. A two-dimensional region is covered with an array of contiguous tiles, with the elements on each layer connecting with elements of adjacent tiles to form extended shapes.

[0012] These and other objects and advantages of the present invention will no doubt become obvious to those of ordinary skill in the art after having read the following detailed description of the preferred embodiments which are illustrated in the various drawing figures.

[0023] In the following detailed description of the present invention, a method for generating a deep N-well pattern for an integrated circuit design, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be obvious to one skilled in the art that the present invention may be practiced without these specific details. In other instances well known methods involving photolithography, ion implantation, deposition and etch, etc., and well known structures such as ohmic contacts and barrier metallization, etc., have not been described in detail so as not to unnecessarily obscure aspects of the present invention.

[0024] FIG. 1 shows a plan view of a portion 100 of a p-type integrated circuit substrate comprising a deep N-well structure 110 with a plurality of openings 105. A plurality of surface N-wells 115 are coupled to the deep N-well structure 110. A plurality of NFETs 120 resides in the p-type substrate 100, and a plurality of complementary PFETs 125 reside in an N-well 115.

[0025] The design and layout of the surface N-wells 115, PFETs 125, and NFETs 120 is conventionally done with respect to a single rectilinear coordinate system. It is to be appreciated that in a preferred embodiment, the deep N-well 110 has linear features at least some of which are aligned with a rectilinear coordinate system that is rotated with respect to the coordinate system of the surface structures.

[0026] Details of the relative alignment between the surface N-wells 115 and the deep N-well structure 110 is disclosed in a U.S. Patent Application titled “Diagonal Deep Well Region for Routing Body-Bias Voltage for MOSFETs in Surface Well Regions,” Ser. No. 10/334,272, assigned to the assignee of the present application and filed on Dec. 31, 2002; the entire contents of which are incorporated herein by reference.

[0027] FIG. 2 shows a basic tile 200 comprising a first layer element 205, and a second layer element 210, in accordance with an embodiment of the present invention. Although in other embodiments, the tile may be limited to a single layer with one or more elements, tile 200 is presented in correspondence with the deep N-well 210 of FIG. 2. Each of the elements 205 and 210 represents a portion of the deep N-well 210, and thus represents a portion of a photomask used to fabricate the deep N-well 210.

[0028] Although tile 200 is shown as square; in general, rectangular tiles may be used. The minimum dimensions of the tile and the elements 205 and 210 are determined in part by the minimum attainable feature size for the deep N-well 210, and is thus determined in part by the depth at which the deep N-well is implanted. In one embodiment, the dimensions of the tile 200 and the elements 205 and 210 are selected to produce a deep N-well with the minimum attainable features. In other embodiments, the dimensions selected may be larger than the minimum attainable. For example, larger dimensions may be used to offset the impact of depletion regions on openings in the deep N-well 210.

[0029] In conventional usage, the deep N-well spacing only need be sufficient to ensure that two deep N-well structures at different potentials are electrically isolated. In the present invention, the spacing must also be large enough to ensure a sufficiently low resistance path between the surface P-well and the underlying p-substrate.

[0030] FIG. 3A shows a tile array 300 assembled from the tile 200 of FIG. 2, in accordance with an embodiment of the present invention. The use of a single building block that is stepped and repeated at regular intervals to cover a portion of the design area provides for a compact representation in the database associated with computer aided design (CAD) tools.

[0031] As can be seen in FIG. 3A, each of elements on the contiguous tiles mate to produce an extended pattern on each of the two layers. Although a single bar is used as an element on each of the two layers, one or more shapes of varying size and complexity may be used on a tile, and also on one or more layers.

[0032] The array of tiles 300 is an efficient tool for obtaining a uniform extended pattern that conforms to a coordinate system that is rotated with respect to the coordinate system used by the surface features and the tile edges. Since the tile edges conform to the surface coordinate system, the tile may be easily introduced to (CAD) tools that are used for integrated circuit layout and design. The tile array 300 may also be conveniently edited using conventional CAD tools.

[0033] FIG. 3B shows the tile array 300 of FIG. 3A after editing by the removal of one tile, in accordance with an embodiment of the present invention. The removal of one tile has effectively produced three sub-arrays having either (1,1,3) or (2,1,2) tiles. Removal of one or more tiles from the initial array is one technique that may be used in editing the pattern that is ultimately used to determine the shape of a fabricated deep N-well structure.

[0034] FIG. 4A shows the tile array 300 of FIG. 3A after merging of the tiles, in accordance with an embodiment of the present invention. Merging of the tiles results in a transformation of the tile array in the database of the CAD tool being used, and is typically done after any desired editing of the array through tile removal. The merging of the tiles produces extended shapes that reside on the two layers that were previously used by the individual tiles. Each extended shape (e.g., a stripe) is made up of a collection of segments.

[0035] With reference to FIG. 4A, the first layer now comprises a series of stripes 405 that comprise the elements 205 that were previously associated with individual tiles 200, and the second layer now comprises a series of stripes 410 that comprise the elements 210 that were previously associated with the individual tiles 200. The stripes on each layer may be further edited (e.g., trimmed or extended) by adding or removing segments. Extension of segments may be done to provide overlap to remove discontinuities that would otherwise be present at tile boundaries.

[0036] FIG. 4B shows the tile array 400 of FIG. 4A after editing of the second layer through the removal of a stripe 410. The removal of segments from a stripe within a layer provides another editing technique that is compatible with conventional CAD tools.

[0037] As indicated by the dashed lines at the ends of elements 205 and 210 of FIG. 2, the elements associated with a tile may extend beyond the tile boundary. Extension of elements produces an overlap that may be desired to avoid discontinuities when the elements are subsequently merged.

[0038] FIG. 5 shows a flattened, merged tile array 500 that results from flattening the merged tile array (two layers) 400 of FIG. 4A. Flattening is the process of combining the two layers into one, to produce a pattern 510 that exists on a single layer. The pattern 510 with openings 505 corresponds to a deep N-well structure that is produced from merging and flattening an unedited array of tiles as shown in FIG. 3A. At this point, editing and manipulation of segments is complete, and the segments cease to exist as independent entities.

[0039] FIG. 6 shows a representative deep N-well 610 in a CMOS circuit substrate 600 fabricated with a mask derived from the merged and flattened pattern of FIG. 3B. In this example, a hole in the deep N-well has been produce through the removal of a tile.

[0040] During the editing of the tile arrays and layers, the removal of tiles and stripe segments may be predicated on design rules so that the editing may be performed automatically by a CAD tool. For example Cadence Assura™ may be used. Editing and checking of the extended shapes (stripes) on the layers is easier to handle than attempting to manipulate a single complex shape existing on a single layer.

[0041] There are many conditions that may drive the removal of a tile or the editing of a layer. For example, an N-well 620 may comprise a PFET 625 that does not require body-biasing. In another example, the deep N-well 610 may be trimmed to provide decoupling or isolation for an analog NFET 630.

[0042] In general, during tile array editing, tiles that are not connected to surface N-wells may be removed. Tiles may also be removed to disconnect surface N-wells from the deep N-well structure. Thus, the tiles may be removed as unwanted, or as unused, although a particular tile that has no direct connections to a surface N-well may be kept for purposes of body-bias distribution.

[0043] Design checks are made to insure that that isolation and connection of the surface N-wells is robust. A tenuous connection will negatively affect the body-bias potential, and poor isolation will lead to leakage. The robustness of deep N-well connections may be checked through perturbing the design by incrementally increasing the size of the opening in the deep N-well database structure. If the incremental increase causes an N-well to become disconnected, an edit may be made to remove the ambiguity.

[0044] Proximity checks may also be made for unconnected N-wells and NFET devices to make sure that they are sufficiently removed from the deep N-well. Also, it is desirable that the circuit path for body-biased NFETs not be unduly constricted by the deep N-well.

[0045] FIG. 7 shows a process flow diagram 700 for generating a deep N-well pattern, in accordance with an embodiment of the present invention. In step 705, the individual tile geometry is specified. The tile size, number of layers, and elements on each layer are determined. This may be done on the basis of established rules, and may be done with or without user input.

[0046] In step 710, an array of tiles is created over an area of the circuit in which a deep N-well structure is desired. The use of a tile allows the initial generation of the structure to be done efficiently and automatically. In step 715, the array of tiles may be automatically edited by removing tiles on the basis of predefined requirements for voltage supply connections. The removal of tiles may result in a single larger array of tiles being converted to a collection of smaller arrays.

[0047] In step 720, the tiles are merged to produce extended shapes on each of the layers, with each of the extended shapes comprising the elements from the layers on each of the tiles. Each layer may contain a number of discrete extended shapes comprising segments.

[0048] In step 725, the layers may be edited manually to introduce features into the N-well structure that are not addressed by the automatic process. These features may be associated with special requirements for connection and isolation. The editing of layers may also include providing a waiver for an edited feature that has produced, or may produce, a design rule violation.

[0049] The geometry of the shapes at boundaries may also be checked for acute angles that may be incompatible with the photolithographic processes used for fabrication. One or more instances of an extended shape may be removed from a layer, or have segments added or deleted.

[0050] For example, in an integrated circuit with a high degree of homogeneity (e.g., memory), one or more extended shapes may be removed entirely. In such cases, the opening size and pitch may be made larger in the deep N-well structure.

[0051] In step 730, design rule checks are made to insure the proper relationship between the deep N-well structure as represented on the layers, and the connected and unconnected N-wells, as well as the proper relationship to circuit components such as analog devices.

[0052] Perturbation of the shapes resident on the layers may be performed to detect connection ambiguities and circuit paths. For example, an opening may be expanded by altering the size and/or location of the extended shapes on the layers that form the opening perimeter. The alteration of one or more extended shapes on one or more layers is computationally more efficient than modifying a complex unified structure residing on a single layer.

[0053] In step 735, a check is made for design rule violations. If a design rule violation is found, step 725 is repeated to deal with the violation. If a design rule violation is not found, step 740 is performed.

[0054] In step 740, the layers are flattened to produce a single layer. Flattening produces the final pattern that will correspond to the deep N-well structure. At this point the bulk of the computationally intensive modifications to the deep N-well layout have been performed by working with tiles and layers. In the finished flat pattern, proximity correction or other global modifications may be made.

[0055] The foregoing descriptions of specific embodiments of the present invention have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed, and obviously many modifications and variations are possible in light of the above teaching. For example, there are many combinations of the parameters for the implant and anneal process steps, and their sequencing, that may be used to produce the structures described herein. The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the Claims appended hereto and their equivalents.