The invention relates to integrated device design. More particularly, the invention relates to caching of lithography and etch simulation results during optical and process correction (OPC).
As integrated circuits (ICs) become denser, the widths of lines and components, as well as the separation between lines becomes increasingly smaller. Currently, deep sub-micron (<0.25 μm) processes are being used. However, with deep sub-micron processes, silicon yield is affected by several factors including reticle/mask pattern fidelity, optical proximity effects, and diffusion and loading effects during resist and etch processing. Typical problems include line-width variations that depend on local pattern density and topology and line end pullback.
Optical and process correction (OPC) can be used to improve image fidelity. Optical proximity correction is a subset of optical and process correction. OPC techniques include, for example, introduction of systematic changes to the geometry of an IC reticle layout to compensate for nonlinear distortions introduced by optical diffraction and resist effects.
Two general categories of OPC are currently in use: rule-based OPC and model-based OPC. In rule-based OPC, a reticle layout is modified according to a set of fixed rules for geometric manipulation. However, rule-based OPC has limited capability and when more complex OPC is desired, model-based OPC is used.
Under model-based OPC, corrections to a reticle layout are calculated through computer simulations employing various models such as optical models. Model-based OPC can be very computationally intensive. Typically, for every edge in the reticle layout, an edge placement error is determined by simulation. Edge placement error identifies the deviation of the edge in a simulated layout (or actual layout) from the corresponding edge in the target layout. Based on an edge placement error, an edge may be pushed or pulled in an attempt to compensate for the error. The simulations and adjustments may need to be repeated (iterated) several times for each edge before the edge placement error is within acceptable limits.
The accuracy of the edge placement error can be increased if computations are done for fragments of edges. Fragmenting involves inserting additional vertices to create smaller sections of edges, or edge fragments. The granularity of the fragments defines the fineness of the OPC corrections that can be made. Fragmentation rules generally define where vertices should be added. The greater the number of vertices added, the more precise edge placement corrections can be made at the expense of increased OPC computations. In other words, increasing the granularity of edge fragments increases potential OPC accuracy but decreases speed.
Some existing OPC techniques attempt to reduce computational costs associated with model-based OPC by reusing OPC results for matching portions of reticle layouts. That is, the corrected state of a reticle layout portion is determined and stored with the initial state of the reticle layout portion in a database. Subsequently, when a reticle layout portion with a matching geometry is encountered, the OPC result is reused to avoid the necessity of having to re-perform the simulations. However, this reuse mechanism has proved to reduce OPC accuracy because it uses the results that are based solely on the matching geometry and fails to take into account other factors that contribute to computations performed during each iteration.
Thus, what is needed is an improved reuse mechanism that reduces computational costs associated with OPC without negatively affecting the accuracy of OPC.
The present invention relates to performing optical and process correction (OPC) on a reticle layout using previously calculated simulation results. According to one aspect of the present invention, one or more control points are identified within a reticle layout that is used in a simulation of a manufacturing process for an integrated device layer. Further, a current geometrical layout pattern is determined in the vicinity of the control points, and a cache is searched for a matching geometrical layout pattern. If the search is successful, a simulation result associated with the matching geometrical layout pattern is retrieved from the cache and reused for the current geometrical layout pattern. Alternatively, if the search is unsuccessful, a simulation result associated with the current geometrical design pattern is computed and stored in the cache for future reuse.
The invention is illustrated by way of example, and not by way of limitation in the figures of the accompanying drawings in which like reference numerals refer to similar elements.
Methods and apparatuses for performing optical and process correction (OPC) on a reticle layout using previously calculated simulation results are described. In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the invention. It will be apparent, however, to one skilled in the art that the invention can be practiced without these specific details. In other instances, structures and devices are shown in block diagram form in order to avoid obscuring the invention.
Reference in the specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.
Methods and apparatuses are described herein with respect to integrated circuit manufacturing; however, the techniques described can be applied to the manufacturing and/or design process of any integrated device. Integrated devices include integrated circuits, micromachines, thin film structures such as disk drive heads, gene chips, micro-electromechanical systems (MEMS), or any other article of manufacture that is manufactured using lithography techniques.
Parts of the description will be presented using terminology commonly employed by those skilled in the art to convey the substance of their work to others skilled in the art, such as reticle/mask layouts, geometries, model-based simulations and so forth. Also, parts of the description will also be presented in terms of operations performed by a computer system, using terms such as determining, generating and so forth. As well understood by those skilled in the art, these operations and quantities operated on, take the form of electrical, magnetic, or optical signals capable of being stored, transferred, combined, and otherwise manipulated through electrical/optical components of a digital system; and the term digital system includes general purpose as well as special purpose data processing machines, systems, and the like, that are standalone, adjunct or embedded.
Various operations will be described as multiple discrete steps performed in a manner that is most helpful in understanding the present invention. However, the order of description should not be construed as to imply that these operations are necessarily performed in the order they are presented, or even order dependent. Lastly, repeated usage of the phrase “in one embodiment” does not necessarily refer to the same embodiment, although it may.
The present invention relates to performing optical and process correction (OPC) on a reticle layout using previously calculated simulation results. The simulation results may include various values that are computed when iteratively employing model-based simulations. These values may include edge placement errors, intensity values and other lithography and etch simulation results associated with different geometrical layout patterns of the reticle layout. Once a simulation result is computed for a certain geometrical layout pattern of the reticle layout, it is saved in a cache. Subsequently, when a matching geometrical layout pattern is encountered, the simulation result saved in the cache is reused, thus avoiding the necessity of having to repeat the expensive computation for the matching geometrical layout pattern.
Corrections depend on certain parameters computed during the simulation. Some of these parameters are determined using expensive computations. For example, the determination of convolution and intensity involves very computationally intensive operations. In particular, to compute intensity at a specific point within the reticle layout, a contribution of each reticle layout segment positioned in the vicinity of this point need to be determined and multiplied by a transmission value, and then the multiplication results need to be added together. Because the number of reticle layout segments may be significant (e.g., there may be more than 100 segments), these computations can be very costly. The OPC module 100 reduces the amount of expensive computations required to perform the simulation by reusing simulation results of the expensive computations for matching geometrical patterns of the reticle layout.
For the illustrated embodiment, OPC module 100 includes a geometry identifier 102, a simulator 104, a cache 110, and a layout correction analyzer 116. Geometry identifier 102 is responsible for defining a reticle layout area for which simulation is to be performed. In one embodiment, geometry identifier 102 defines a reticle layout area by identifying, in the reticle layout, one or more points (referred to herein as control points) at which computations (e.g., intensity computations) are to be made and determining a geometrical layout pattern in the vicinity of the control points.
Simulator 104 is responsible for performing model-based analyses for each reticle layout area to determine the appropriate corrections for the particular reticle layout area using optical and/or resist models. In one embodiment, simulator 104 includes a matching geometry finder 106 and a simulation result calculator 108. Matching geometry finder 106 is responsible for facilitating a search of cache 110 for a geometrical layout pattern matching the pattern defined by geometry identifier 102. Cache 110 may be in the form of any data structure including, for example, a lookup table, a tree data structure, a sparse matrix data structure, or a hash table. Cache 110 stores data identifying geometrical layout patterns and associated simulation results. A simulation result may include intensity, EPE, or any other lithography and etch simulation result associated with a specific geometrical layout pattern. For each layout pattern, cache 110 may store more than one parameter. For example, cache 110 may store multiple intensity values for a layout pattern, with each intensity value corresponding to a specific control point in a group of control points. In another example, cache 110 may store both intensity and EPE for a layout pattern, etc.
If a matching layout pattern is not found in cache 110, simulator 104 performs the computations and stores the appropriate result(s) 112, together with data identifying the layout pattern, in cache 110 for future use. If cache 110 contains an entry for a matching layout pattern, a simulation result 114 associated with the matching layout pattern is retrieved from cache 110 and used by simulator 104 for further computations if the simulation result 114 is not the final result. For example, if the simulation result 114 consists of the intensity value, simulator 104 uses the intensity value to determine EPE. If the simulation result 114 is final (e.g., EPE), simulator 104 does not need to perform any computations for this reticle layout portion during the current simulation, i.e., the current simulation can be skipped for this reticle layout pattern.
Layout correction analyzer 116 is responsible for making the determined corrections for each reticle layout portion and comparing the corrected layout with the target layout. If the corrected mask layout is not sufficiently pre-compensated (i.e., it is unlikely to yield the desired image within a predetermined tolerance level), the process is repeated for the corrected reticle layout portion. The process may be performed multiple times, iteratively, until the predetermined tolerance level is reached.
Accordingly, OPC module 100 reduces costs associated with the simulation process by ensuring that the expensive computations are performed only once for matching geometrical layout patterns of the reticle layout. In addition, the reuse technique employed by OPC module 100 has no negative effect on OPC accuracy because this reuse technique does not eliminate iteratively performed simulations, thereby taking into account the matching geometry and various other factors that can contribute to computations performed during each iteration.
Referring to
In one embodiment, a single control point is identified at processing block 202. In another embodiment, a group of control points is identified at processing block 202. The group of control points may include a predetermined number of control points (e.g., 12 or 24). The control points may be positioned in a line or in any other way (e.g., they may form a cross, a grid, etc.).
At processing block 204, processing logic determines geometry (also referred to herein as a geometrical layout pattern) in the vicinity of the control points. In one embodiment, the geometry is determined by defining a support region around the control points using an optical radius and analyzing reticle layout segments within the support region. The optical radius may be specified by the user or defined programmatically. The support region may be of any shape (e.g., rectangular, square, circle, oval, etc.).
Referring to
It should be noted that the geometry in the vicinity of control points can be determined in any other manner known in the art without loss of generality.
Returning to
Next, processing logic determines whether the cache contains matching geometry data (decision box 208). If the determination is positive, processing logic retrieves a simulation result associated with the matching geometry from the cache (processing block 214). In one embodiment, in which a single control point is used, the simulation result includes an intensity value computed based on segments included in the geometry. In another embodiment, in which a group of control points is used, the simulation result may include a set of intensity values computed at corresponding control points from the group, a cumulative intensity value or other lithography and etch result value computed for the entire group of control points, or any combination of the above.
If the determination made at decision box 208 is negative, processing logic computes a simulation result (e.g., intensity, EPE, etc.) for the current geometry (processing block 210) and stores the simulation result, together with data identifying the geometry, in the cache for future reuse (processing block 212).
As described above, the cache of simulation results may be a lookup table, a tree data structure, a sparse matrix data structure, a hash table, or any other data structure for mapping one piece of data to another piece of data.
Referring to
The selection of parameters to be included in the lookup table 400 depends on the cost associated with the computation of this particular parameter, on how frequently this parameter is used during the OPC process, and some other factors.
In one embodiment, the size of the lookup table 400 is limited. In this embodiment, when it is necessary to add data to the lookup table 400 after it has already reached its size limit, any method known in the art can be used to select data in the lookup table 400 for removal. For example, the data may be selected for removal randomly (a Monte-Carlo method), based upon how recently each value has been used, based on the count indicating how many matches each element has had, etc.
Some target layouts may consist of patterns that are only partially similar. For such layouts, it may be useful to reuse partial simulation results. Partial simulation results correspond to partial geometries. For example, a partial simulation result may be a contribution of a specific partial geometry to intensity computed at a corresponding control point.
Referring to
At processing block 508, the cache is searched for a match of the partial geometry using data identifying the partial geometry. In one embodiment, the partial geometry data includes coordinates of each segment within the partial geometry. For example, if the partial geometry includes a single edge, the partial geometry data will consist of the coordinates of this edge.
At decision box 510, a determination is made as to whether a match is found. If the determination is positive, processing logic retrieves a partial simulation result (e.g., convolution) associated with the matching partial geometry data (processing block 512) and stores it in a temporary storage (processing block 514) until the results are determined for all portions of the geometry (decision box 516). Then, a cumulative simulation result (e.g., intensity) for the entire geometry is computed (processing block 518). In one embodiment, processing logic ensures, prior to reusing the partial simulation result, that the partial simulation result retrieved at processing block 512 can be used with the current partial geometry by verifying that the direction of the group of control points associated with the matching partial geometry is the same as the direction of the control points associated the current partial geometry and the segments of the matching partial geometry have the same relative location with respect to the control points as the segments of the current partial geometry.
Alternatively, if the determination made at decision box 510 is negative, i.e., no matching partial geometry is found in the cache, then processing logic computes the simulation result for the current partial geometry (processing block 520) and stores it in the cache (processing block 522).
In the foregoing specification, the invention has been described with reference to specific embodiments thereof. It will, however, be evident that various modifications and changes can be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.
Number | Name | Date | Kind |
---|---|---|---|
4762396 | Dumant et al. | Aug 1988 | A |
5502654 | Sawahata | Mar 1996 | A |
5655110 | Krivokapic et al. | Aug 1997 | A |
5723233 | Garza et al. | Mar 1998 | A |
5825647 | Tsudaka | Oct 1998 | A |
5879844 | Yamamoto et al. | Mar 1999 | A |
6016357 | Neary et al. | Jan 2000 | A |
6049660 | Ahn et al. | Apr 2000 | A |
6077310 | Yamamoto et al. | Jun 2000 | A |
6120952 | Pierrat et al. | Sep 2000 | A |
6128067 | Hashimoto | Oct 2000 | A |
6187483 | Capodieci et al. | Feb 2001 | B1 |
6243855 | Kobayashi et al. | Jun 2001 | B1 |
6249904 | Cobb | Jun 2001 | B1 |
6263299 | Aleshin et al. | Jul 2001 | B1 |
6269472 | Garza et al. | Jul 2001 | B1 |
6301697 | Cobb | Oct 2001 | B1 |
6370679 | Chang et al. | Apr 2002 | B1 |
6425117 | Pasch et al. | Jul 2002 | B1 |
6453452 | Chang et al. | Sep 2002 | B1 |
6453457 | Pierrat et al. | Sep 2002 | B1 |
6467076 | Cobb | Oct 2002 | B1 |
6499003 | Jones et al. | Dec 2002 | B2 |
6591207 | Naya et al. | Jul 2003 | B2 |
20020113234 | Okuda et al. | Aug 2002 | A1 |
Number | Date | Country |
---|---|---|
09319067 | Dec 1997 | JP |
WO 0165315 | Sep 2001 | WO |
Number | Date | Country | |
---|---|---|---|
20040019872 A1 | Jan 2004 | US |