Patent Application
20180196349
The present disclosed technology relates to the field of lithography. Various implementations of the disclosed technology may be particularly useful for calibrating lithography models.
Optical Proximity Correction (OPC) enables low-k1 lithography by changing the pattern shapes on the mask to correct for the non-linear process of printing those patterns on a wafer. OPC has continually improved in accuracy over the years by adding more physically based lithography models. Some other design and verification processes such as design for manufacture may also employ lithography models. Lithography models, including rigorous first principle models and fast approximate models, require calibration using measured data. The calibration of lithography models involves function evaluations and data fitting. There are two major concerns for the model calibration. First, the computational cost of function evaluations, especially for complex negative tone developer (NTD) resist models, is very high and dominates the calibration run time. Second, the fitting objective is a multi-modal function of the tuning parameters, and users are generally interested in models other than the model with the best fitting objective. As the model complexity increases, it is challenging to find a convergent solution.
Aspects of the disclosed technology relate to techniques for calibrating lithography models. In one aspect, there is a method comprising: grouping a set of original model candidates into pairs of original model candidates, the original model candidates being derived based on a lithography model for a component of a lithographic process and having different sets of values of model parameters of the lithography model; generating a pair of child model candidates for each of the pairs of original model candidates by performing mutation, crossover, or both on the each of the pairs of original model candidates; deriving a set of new model candidates from the original model candidates and the child model candidates, wherein the deriving comprises: pairing, based on a similarity function, each of the pair of child model candidates with one of the corresponding original model candidates to derive pairs of parent-child model candidates, selecting the child model candidate in each of the pairs of parent-child model candidates as a new model candidate if the child model candidate has a higher fitness based on an objective function, selecting the original model candidate in the each of the pairs of parent-child model candidates as a new model candidate if the original model candidate has a higher fitness based on the objective function or if the original model candidate and the child model candidate are dissimilar based on the similarity function and a parameter associated with the set of original model candidates; and removing one or more new model candidates based on the objective function to keep a number of the new model candidates in each of niches from exceeding a maximum number, the niches being determined based on the similarity function; repeating the grouping, the generating and the deriving by replacing the set of original model candidates with the set of new model candidates until one of one or more predefined conditions is satisfied to obtain a set of final model candidates; and reporting one or more final model candidates in the set of final model candidates based on the objective function and the similarity function, from which one final model candidate is to be selected as a calibrated lithography model for simulating the lithographic process.
The parameter associated with the set of original model candidates may be an average of distances between the original model candidates.
The crossover may employ a uniform crossover scheme. The set of original model candidates may be derived by choosing the values of the model parameters uniformly distributed within predetermined ranges for the model parameters.
The similarity function may be a normalized Euclidian distance function. The objective function may be a root mean square function.
The removing may be performed to keep not only the number of the new model candidates in each of niches from exceeding the maximum number but also a total number of the new model candidates from exceeding a threshold number.
In another aspect, there are one or more non-transitory processor-readable media storing processor-executable instructions for causing one or more processors to perform the above method.
In still another aspect, there is a system, comprising: one or more processors, the one or more processors programmed to perform the above method.
Certain inventive aspects are set out in the accompanying independent and dependent claims. Features from the dependent claims may be combined with features of the independent claims and with features of other dependent claims as appropriate and not merely as explicitly set out in the claims.
Certain objects and advantages of various inventive aspects have been described herein above. Of course, it is to be understood that not necessarily all such objects or advantages may be achieved in accordance with any particular embodiment of the disclosed technology. Thus, for example, those skilled in the art will recognize that the disclosed technology may be embodied or carried out in a manner that achieves or optimizes one advantage or group of advantages as taught herein without necessarily achieving other objects or advantages as may be taught or suggested herein.
Various aspects of the disclosed technology relate to techniques for calibrating lithography models. In the following description, numerous details are set forth for the purpose of explanation. However, one of ordinary skill in the art will realize that the disclosed technology may be practiced without the use of these specific details. In other instances, well-known features have not been described in details to avoid obscuring the present disclosed technology.
Some of the techniques described herein can be implemented in software instructions stored on a computer-readable medium, software instructions executed on a computer, or some combination of both. Some of the disclosed techniques, for example, can be implemented as part of an electronic design automation (EDA) tool. Such methods can be executed on a single computer or on networked computers.
Although the operations of the disclosed methods are described in a particular sequential order for convenient presentation, it should be understood that this manner of description encompasses rearrangements, unless a particular ordering is required by specific language set forth below. For example, operations described sequentially may in some cases be rearranged or performed concurrently. Moreover, for the sake of simplicity, the disclosed flow charts and block diagrams typically do not show the various ways in which particular methods can be used in conjunction with other methods. Additionally, the detailed description sometimes uses terms like “generate”, “group” and “derive” to describe the disclosed methods. Such terms are high-level abstractions of the actual operations that are performed. The actual operations that correspond to these terms will vary depending on the particular implementation and are readily discernible by one of ordinary skill in the art.
Also, as used herein, the term “design” is intended to encompass data describing an entire integrated circuit device. This term also is intended to encompass a smaller group of data describing one or more components of an entire device, however, such as a portion of an integrated circuit device. Still further, the term “design” also is intended to encompass data describing more than one microdevice, such as data to be used to form multiple microdevices on a single wafer.
The execution of various electronic design automation processes according to embodiments of the disclosed technology may be implemented using computer-executable software instructions executed by one or more programmable computing devices. Because these embodiments of the disclosed technology may be implemented using software instructions, the components and operation of a generic programmable computer system on which various embodiments of the disclosed technology may be employed will first be described. Further, because of the complexity of some electronic design automation processes and the large size of many circuit designs, various electronic design automation tools are configured to operate on a computing system capable of concurrently running multiple processing threads. The components and operation of a computer network having a host or master computer and one or more remote or servant computers therefore will be described with reference to
In
The memory 107 may similarly be implemented using any combination of computer readable media that can be accessed by the master computer 103. The computer readable media may include, for example, microcircuit memory devices such as read-write memory (RAM), read-only memory (ROM), electronically erasable and programmable read-only memory (EEPROM) or flash memory microcircuit devices, CD-ROM disks, digital video disks (DVD), or other optical storage devices. The computer readable media may also include magnetic cassettes, magnetic tapes, magnetic disks or other magnetic storage devices, punched media, holographic storage devices, or any other medium that can be used to store desired information.
As will be discussed in detail below, the master computer 103 runs a software application for performing one or more operations according to various examples of the disclosed technology. Accordingly, the memory 107 stores software instructions 109A that, when executed, will implement a software application for performing one or more operations. The memory 107 also stores data 109B to be used with the software application. In the illustrated embodiment, the data 109B contains process data that the software application uses to perform the operations, at least some of which may be parallel.
The master computer 103 also includes a plurality of processor units 111 and an interface device 113. The processor units 111 may be any type of processor device that can be programmed to execute the software instructions 109A, but will conventionally be a microprocessor device. For example, one or more of the processor units 111 may be a commercially generic programmable microprocessor, such as Intel® Pentium® or Xeon™ microprocessors, Advanced Micro Devices Athlon™ microprocessors or Motorola 68K/Coldfire® microprocessors. Alternately or additionally, one or more of the processor units 111 may be a custom-manufactured processor, such as a microprocessor designed to optimally perform specific types of mathematical operations. The interface device 113, the processor units 111, the memory 107 and the input/output devices 105 are connected together by a bus 115.
With some implementations of the disclosed technology, the master computing device 103 may employ one or more processing units 111 having more than one processor core. Accordingly,
Each processor core 201 is connected to an interconnect 207. The particular construction of the interconnect 207 may vary depending upon the architecture of the processor unit 111. With some processor cores 201, such as the Cell microprocessor created by Sony Corporation, Toshiba Corporation and IBM Corporation, the interconnect 207 may be implemented as an interconnect bus. With other processor units 111, however, such as the Opteron™ and Athlon™ dual-core processors available from Advanced Micro Devices of Sunnyvale, Calif., the interconnect 207 may be implemented as a system request interface device. In any case, the processor cores 201 communicate through the interconnect 207 with an input/output interface 209 and a memory controller 210. The input/output interface 209 provides a communication interface between the processor unit 111 and the bus 115. Similarly, the memory controller 210 controls the exchange of information between the processor unit 111 and the system memory 107. With some implementations of the disclosed technology, the processor units 111 may include additional components, such as a high-level cache memory accessible shared by the processor cores 201.
While
Returning now to
Each servant computer 117 may include a memory 119, a processor unit 121, an interface device 123, and, optionally, one more input/output devices 125 connected together by a system bus 127. As with the master computer 103, the optional input/output devices 125 for the servant computers 117 may include any conventional input or output devices, such as keyboards, pointing devices, microphones, display monitors, speakers, and printers. Similarly, the processor units 121 may be any type of conventional or custom-manufactured programmable processor device. For example, one or more of the processor units 121 may be commercially generic programmable microprocessors, such as Intel® Pentium® or Xeon™ microprocessors, Advanced Micro Devices Athlon™ microprocessors or Motorola 68K/Coldfire® microprocessors. Alternately, one or more of the processor units 121 may be custom-manufactured processors, such as microprocessors designed to optimally perform specific types of mathematical operations. Still further, one or more of the processor units 121 may have more than one core, as described with reference to
In the illustrated example, the master computer 103 is a multi-processor unit computer with multiple processor units 111, while each servant computer 117 has a single processor unit 121. It should be noted, however, that alternate implementations of the disclosed technology may employ a master computer having single processor unit 111. Further, one or more of the servant computers 117 may have multiple processor units 121, depending upon their intended use, as previously discussed. Also, while only a single interface device 113 or 123 is illustrated for both the master computer 103 and the servant computers, it should be noted that, with alternate embodiments of the disclosed technology, either the computer 103, one or more of the servant computers 117, or some combination of both may use two or more different interface devices 113 or 123 for communicating over multiple communication interfaces.
With various examples of the disclosed technology, the master computer 103 may be connected to one or more external data storage devices. These external data storage devices may be implemented using any combination of computer readable media that can be accessed by the master computer 103. The computer readable media may include, for example, microcircuit memory devices such as read-write memory (RAM), read-only memory (ROM), electronically erasable and programmable read-only memory (EEPROM) or flash memory microcircuit devices, CD-ROM disks, digital video disks (DVD), or other optical storage devices. The computer readable media may also include magnetic cassettes, magnetic tapes, magnetic disks or other magnetic storage devices, punched media, holographic storage devices, or any other medium that can be used to store desired information. According to some implementations of the disclosed technology, one or more of the servant computers 117 may alternately or additionally be connected to one or more external data storage devices. Typically, these external data storage devices will include data storage devices that also are connected to the master computer 103, but they also may be different from any data storage devices accessible by the master computer 103.
It also should be appreciated that the description of the computer network illustrated in
Electronic circuits, such as integrated circuits (ICs), are used in a variety of products, from automobiles to microwaves to personal computers. Designing and fabricating IC devices typically involves many steps, sometimes referred to as the “design flow.” The particular steps of a design flow often are dependent upon the type of the circuit, its complexity, the design team, and the circuit fabricator or foundry that will manufacture the circuit. Typically, software and hardware “tools” verify the design at various stages of the design flow by running software simulators and/or hardware emulators. These steps aid in the discovery of errors in the design, and allow the designers and engineers to correct or otherwise improve the design.
Several steps are common to most design flows. Initially, the specification for a new circuit is transformed into a logical design, such as a register transfer level (RTL) description of the circuit. With this logical design, the circuit is described in terms of both the exchange of signals between hardware registers and the logical operations that are performed on those signals. The logical design typically employs a Hardware Design Language (HDL), such as the Very high speed integrated circuit Hardware Design Language (VHDL). The logic of the circuit is then analyzed, to confirm that it will accurately perform the functions desired for the circuit. This analysis is sometimes referred to as “functional verification.”
After the accuracy of the logical design is confirmed, it is converted into a device design by synthesis software. The device design, which is typically in the form of a schematic or netlist, describes the specific electronic devices (such as transistors, resistors, and capacitors) that will be used in the circuit, along with their interconnections. The relationships between the electronic devices are then analyzed, to confirm that the circuit described by the device design will correctly perform the desired functions. This analysis is sometimes referred to as “formal verification.” Additionally, preliminary timing estimates for portions of the circuit are often made at this stage, using an assumed characteristic speed for each device, and incorporated into the verification process.
Once the components and their interconnections are established, the design is again transformed, this time into a physical design that describes specific geometric elements. This type of design often is referred to as a “layout” design. The geometric elements, which typically are polygons, define the shapes that will be created in various layers of material to manufacture the circuit. Typically, a designer will select groups of geometric elements representing IC components (e.g., contacts, channels, gates, etc.) and place them in a design area. These groups of geometric elements may be custom designed, selected from a library of previously-created designs, or some combination of both. Lines are then routed between the geometric elements, which will form the wiring used to interconnect the electronic devices. Layout tools (often referred to as “place and route” tools), such as Mentor Graphics' IC Station or Cadence's Virtuoso, are commonly used for both of these tasks.
Circuit layout descriptions can be provided in many different formats. The Graphic Data System II (GDSII) format is a popular format for transferring and archiving two-dimensional (2D) graphical circuit layout data. Among other features, it contains a hierarchy of structures, each structure containing layout elements (e.g., polygons, paths or poly-lines, circles and textboxes). Other formats include an open source format named Open Access, Milkyway by Synopsys, Inc., EDDM by Mentor Graphics, Inc., and the more recent Open Artwork System Interchange Standard (OASIS) proposed by Semiconductor Equipment and Materials International (SEMI). These various industry formats are used to define the geometrical information in design layouts that are employed to manufacture integrated circuits. Once the design is finalized, the layout portion of the design can be used by fabrication tools to manufacture the circuit using a photolithographic process.
There are many different fabrication processes for manufacturing a circuit, but most processes include a series of steps that deposit layers of different materials on a substrate, expose specific portions of each layer to radiation, and then etch the exposed (or non-exposed) portions of the layer away. For example, a simple semiconductor device component could be manufactured by the following steps. First, a positive type epitaxial layer is grown on a silicon substrate through chemical vapor deposition. Next, a nitride layer is deposited over the epitaxial layer. Then specific areas of the nitride layer are exposed to radiation, and the exposed areas are etched away, leaving behind exposed areas on the epitaxial layer, (i.e., areas no longer covered by the nitride layer). The exposed areas then are subjected to a diffusion or ion implantation process, causing dopants, for example phosphorus, to enter the exposed epitaxial layer and form charged wells. This process of depositing layers of material on the substrate or subsequent material layers, and then exposing specific patterns to radiation, etching, and dopants or other diffusion materials, is repeated a number of times, allowing the different physical layers of the circuit to be manufactured.
Each time that a layer of material is exposed to radiation, a mask must be created to expose only the desired areas to the radiation, and to protect the other areas from exposure. The mask is created from circuit layout data. That is, the geometric elements described in a design layout define the relative locations or areas of the circuit that will be exposed to radiation through the mask. A mask or reticle writing tool is used to create the mask based upon the design layout, after which the mask can be used in a photolithographic process.
As designers and manufacturers continue to increase the number of circuit components in a given area and/or shrink the size of circuit components, the shapes reproduced on the substrate (and thus the shapes in the mask) become smaller and are placed closer together. This reduction in feature size increases the difficulty of faithfully reproducing the image intended by the design layout onto the substrate. The diffractive effects of light often result in defects where the intended image is not accurately “printed” onto the substrate during the photolithographic process, creating flaws in the manufactured device. One or more resolution enhancement techniques (RETs) are often employed to improve the resolution of the image that the mask forms on the substrate during the photolithographic process. Examples of various resolution enhancement techniques are discussed in “Resolution Enhancement Technology: The Past, the Present, and Extensions for the Future,” Frank M. Schellenberg, Optical Microlithography XVII, edited by Bruce W. Smith, Proceedings of SPIE Vol. 5377, which article is incorporated entirely herein by reference. One of these techniques, “optical proximity correction” or “optical process correction” (OPC), adjusts the amplitude of the light transmitted through a lithographic mask by modifying the design layout data employed to create the mask.
In a conventional OPC process, the edges of the geometric elements in the design are fragmented. The size of the edge fragments in a given layout design depends upon the OPC process parameters, often referred to as the OPC recipe. The “recipe” specifies the size of the edge fragments. While not all edges within a layout design are fragmented in every OPC process, these edges may also be referred to as edge fragments. In a model-based OPC process, the printed image is simulated. That is, the photolithographic process is simulated in order to produce a simulated printed image (referred to as simulated image). This simulated image is compared to the target image. The distances between the target image and the simulated image are often referred to as the edge placement error (EPE). Accordingly, in most conventional model-based OPC processes each edge fragment or unfragmented edge has an associated edge placement error (also referred to as an associated edge placement error value).
Next, the edge fragments are individually moved or adjusted in order to enable the simulated image for the resulting mask to reproduce the target image as much as possible. This process of moving the edge fragments, simulating the image that would be produced using the moved edge fragments, and comparing the simulated image to the target image may be repeated a number of times. Each cycle of moving edge fragments and comparing the new simulated image to target image is referred to as an iteration of the OPC process.
The performance of an OPC process depends in part on the accuracy and predictability of lithography models that mathematically represent the distinct steps in the patterning sequence. These lithography models (sometimes also referred to as OPC models or lithographic process models) include optical models for aerial image formation and resist/etch process models for the photoresist-associated steps such as exposure, post-exposure bake, development, and pattern transfer. The optical models for full-chip simulation are usually quasi-rigorous. For example, the Hopkins optical model may be approximated by a sum of coherent systems approximation method, which is not a dramatic simplification. By contrast, the resist/etch process models tend to be semi-empirical: mathematical formulation is used to provide a transfer function between inputs and measured outputs of interest. Due to the semi-empirical nature, the resist/etch process models need to be calibrated based on measurements on printed wafers. The model parameters are trained or calibrated so that simulation data would fit measurement data. Even some parameters for the optical models such as dose and focus may also need to be calibrated against measurement data.
In a calibration process, the calibration of the optical model parameters may be performed first. With the optical model parameters thus set, a subsequent tuning of the resist model parameters is performed. Alternatively, the tuning of both the optical model parameters and the resist model parameters may be conducted simultaneously in a single lumped calibration process.
The genetic algorithm, as a major family in evolutionary algorithms, is inspired by the natural phenomenon of biological evolution. It relies on the principle of “survival of the fittest.” A genetic algorithm process generally begins with a random population in which potential solutions are encoded into chromosomes. By evaluating each chromosome against the objective functions, its goodness, represented by a fitness value, can be determined. Based on the fitness value, some chromosomes are then selected to form a mating pool that will undergo genetic operations, namely, crossover and mutation, to produce some new solutions.
The chromosomes (called the parent chromosomes) in the mating pool exchange their genes to generate new chromosomes (called the child chromosomes) via crossover, while some genes are changed by a random process called mutation. These two operations keep the balance between the exploration and exploitation of solutions so that a genetic algorithm can generate better solutions based on acquired information and some random processes.
The newly formed chromosomes are then assessed by objective functions, and fitness values are assigned. Usually, fitter child chromosomes will replace some or all of the parent chromosomes such that a new population is obtained. This genetic cycle is repeated until some criteria are met.
As a population-based optimization algorithm, the genetic algorithm can linearly reduce runtime of evaluating the objective functions by executing the evaluations in parallel. The parallelization of function evaluation can be easily implemented as it requires minimal synchronization.
Niching genetic algorithms offers an efficient way to search for multiple optima of a multimodal function by promoting the formation of stable subpopulations in the search space. It is based on the phenomenon of separation and specialization in natural ecosystems. In ecology, a niche is associated with a complex of specific living conditions, a subset of environmental resources, and a species is a set of individuals consuming the resources of a specific niche. Thus, niches are a part of the environment and species are a part of the set of all possible individuals. By analogy, in genetic algorithms the term niche is used for the search space domain, and species for the set of individuals with similar characteristics.
In niching genetic algorithms, individuals of a population are divided into several subpopulations—species, each of which occupies its niche, is related to it during the run of the algorithm, and specializes on the solution of a certain sub-problem of the original problem (searches for the optimum in the niche). Such an approach preserves the variety of individuals in the population and allows the genetic algorithm to analyze several optima simultaneously. To attain this effect, the process of selection of individuals usually takes into account not only the value of the fitness function but also the distribution of individuals in the space of genotypes or phenotypes.
As will be discussed in more detail below, the lithography model calibration tool 300 can receive a set of original model candidates from the input database 305. The original model candidate models are derived based on a lithography model for a component of a lithographic process and have different sets of values of model parameters of the lithography model. The component may be the optical component, the resist component, or the etch component. The grouping unit 310 can group the set of original model candidates into pairs of original model candidates. The generation unit 320 can generate a pair of child model candidates for each of the pairs of original model candidates by performing mutation, crossover, or both on the each of the pairs of original model candidates. The selection unit 330 can derive a set of new model candidates from the original model candidates and the child model candidates. The lithography model calibration tool 300 can repeat the grouping, generating and deriving operations by replacing the set of original model candidates with the set of new model candidates until one of one or more predefined conditions is satisfied to obtain a set of final model candidates. The reporting unit 340 can report one or more final model candidates in the set of final model candidates based on the objective function. Among the one or more final model candidates, one final model candidate may be selected for simulating the lithographic process. The reporting may comprise storing in the output database 355 and/or displaying by a graph unit (not shown in the figure), the one or more final model candidates.
As previously noted, various examples of the disclosed technology may be implemented by one or more computing systems, such as the computing system illustrated in
It also should be appreciated that, while the grouping unit 310, the generation unit 320, the selection unit 330 and the reporting unit 340 are shown as separate units in
With various examples of the disclosed technology, the input database 305 and the output database 355 may be implemented using any suitable computer readable storage device. That is, either of the input database 305 and the output database 355 may be implemented using any combination of computer readable storage devices including, for example, microcircuit memory devices such as read-write memory (RAM), read-only memory (ROM), electronically erasable and programmable read-only memory (EEPROM) or flash memory microcircuit devices, CD-ROM disks, digital video disks (DVD), or other optical storage devices. The computer readable storage devices may also include magnetic cassettes, magnetic tapes, magnetic disks or other magnetic storage devices, holographic storage devices, or any other non-transitory storage medium that can be used to store desired information. While the input database 305 and the output database 355 are shown as separate units in
Initially, in operation 410 of the flow chart 400, the grouping unit 310 groups a set of original model candidates into pairs of original model candidates. The original model candidates are derived based on a lithography model for a component of a lithographic process. The lithography model may be an optical model, a mask model, a resist model or a topographical model. The set of original model candidates are derived based on the lithography model and have different sets of values of model parameters of the lithography model.
The horizontal bias term has been discussed in U.S. patent application Ser. No. 14/493,073, entitled “Horizontal development bias in negative tone development of photoresist,” naming Deng, Yunfei et al. as inventors, which is incorporated herein by reference.
As noted previously, the set of original model candidates have different sets of values of the model parameters. For example, one original model candidate may have a set of the model parameter values including a=0.01, b=0.3, s=100 and d=50, while another original model candidate may have a set of the model parameter values including a=0.8, b=0.3, s=100 and d=75. Still another one may share the same values for the model parameters of the horizontal bias term with the former but has different values for at least one other model parameter (e.g., S5 for the fifth modeling term in MF22 shown in
The lithography model calibration tool 300 may receive the set of original model candidates from the input database 305. The set of original model candidates may be derived by choosing the values of the model parameters uniformly distributed within predetermined ranges for the model parameters. Alternatively, model candidates derived using different sampling processes or based on users' experience may be included in the set of original model candidates.
Using the same set of original model candidates, the disclosed model calibration process can lead to the same calibration result, which is preferred in some circumstances.
The grouping unit 310 may divide the original model candidates into groups randomly. An alternative way for grouping is based on genome similarity. For example, a normalized Euclidian distance function may be used to measure the similarity of genomes (model candidates). In some embodiments of the disclosed technology, the grouping unit 310 may divide the original model candidates into pairs. That is, each of the groups of the original model candidates has two members.
In operation 420, the generation unit 320 generates a pair of child model candidates for each of the pairs of original model candidates by performing mutation, crossover, or both on the each of the pairs of original model candidates. A variety of crossover techniques can be employed by the generation unit 320. The single-point crossover technique selects one crossover point on two parents. All model parameter values beyond that point in either parent are swapped between the two parents as illustrated in
The uniform crossover technique uses a fixed mixing ratio between two parents. Unlike single- and two-point crossover, the uniform crossover technique can allow the parents to contribute at the model parameter level rather than at the segment level, as shown in
Mutation is a genetic operator that may be used to maintain genetic diversity from one generation of a population to the next. It is analogous to biological mutation. The generation unit 320 may use mutation to alter one or more model parameters in an original model candidate from its initial state according to a user-definable mutation probability. Like crossover, mutation can have different types.
In operation 430, the selection unit 330 derives a set of new model candidates from the original model candidates and the child model candidates.
The next two operations of the flowchart 1000 perform crowding. In operation 1020, the child model candidate in each of the pairs of parent-child model candidates is selected as a new model candidate if the child model candidate has a higher fitness based on an objective function. The objective function measures goodness or fitness of the original/child model candidates by comparing values calculated using the model candidates with measured values. The objective function may, for example, calculate a sum of squared errors or may be a root mean square function. In this case, the smaller value of the objective function a model candidate, the higher fitness.
In operation 1030, the original model candidate in the each of the pairs of parent-child model candidates is selected as a new model candidate if the original model candidate has a higher fitness based on the objective function or if the original model candidate and the child model candidate are dissimilar based on the similarity function and a parameter associated with the set of original model candidates. The parameter associated with the set of original model candidates defines the range of similarity. Because the parameter is associated with the current generation, the similarity standard may change with the evolution. One example of the parameter may be an average of the original model candidate distances, which is discussed in detail below.
As a result of the operations 1020 and 1030, a child replaces a parent only if it is similar to the parent; otherwise, the child will be added to the new generation if it has a higher fitness. This allows the population to grow in size. This may prevent good model candidates from being eliminated, increase diversity of the population, and reduce chances of pre-mature convergence.
In operation 1040 of the flowchart 1000, one or more new model candidates are removed based on the objective function to keep a number of the new model candidates in each of niches determined based on the similarity function from exceeding a maximum number. This is a niche clearing technique. In general, clearing is based on the concept of limited resources in the environment and the necessity of sharing them among all individuals of the subpopulation but provides them only to the best representatives. In the present case of lithography model calibration, the niche clearing performed by the operation 1040 can help to prevent a few model candidates from dominating the entire population along with the operations 1020-1030.
Additionally, one or more new model candidates may be removed based on the objective function to keep the total number of the new model candidates from exceeding a threshold number.
Niches may be defined using niche radius. The new model candidates that have a distance from a certain point equal to or smaller than the niche radius may be defined as a niche. That certain point may be set as the new model candidate that has the smallest value of the objective function. The niche radius may be predefined by users or set to be a default number.
where dij represents the distance (or similarity) between the genome Gi and the genome Gj determined according to, e.g., Eq. (1). Here, dij is the same as Dist(i, j).
The first inequality defines whether the parent-child pairs should be (c1,p1) and (c2,p2), or (c1,p2) and (c2,p1). After the pairing, whether a child genome replaces the corresponding parent genome in the new generation or is added to the new generation is dependent upon the distance between the two genomes and their relative fitness. Div(P) varies from generation to generation. Accordingly, the above deterministic crowding technique is referred to as adaptive deterministic crowding.
In operation 440 of the flow chart 400, the lithography model calibration tool 300 decides whether one or more predefined conditions are satisfied. If none of the one or more predefined conditions is satisfied, the operations 410-430 are repeated by replacing the set of original model candidates with the set of new model candidates. The one or more predefined conditions may be total number of generations, total run time, convergence of diversity (the diversity of the entire population becomes lower than a certain threshold value), convergence of objective function values (the objective function values of the genomes in all of the niches stop improving for a certain number of generations), or any combination thereof.
If one of the one or more predefined conditions is satisfied, the process moves to operation 450. In the operation, the reporting unit 340 reports one or more final model candidates in the set of final model candidates based on the objective function and the similarity function. For example, the reporting unit 340 may select the fittest final model candidate in each of the top five niches for the reporting. The reporting may comprises storing the one or more final model candidates in the output database 355 and/or displaying them on a display unit. Users may analyze the one or more final model candidates and select one as a calibrated lithography model for their lithographic processes. The calibrated lithography model can be used by OPC and/or other design and layout verification techniques.
The disclosed technology improves lithography model calibration, a computer-related technology by offering high scalability for parallelization of function evaluation) and capabilities of specifying multiple starting points, increasing population diversity and avoiding pre-mature convergence. Experimental results have shown that a lithography model calibrated using an embodiment of the disclosed technology can readily identity multiple different global optima while a traditional approach always converges to 1 particular global optima and that each of the multiple global optima has a higher fitness than the particular global optima. Moreover, the disclosed technology can achieve significant runtime reduction without sacrificing the quality of the result.
While the disclosed technology has been described with respect to specific examples including presently preferred modes of carrying out the disclosed technology, those skilled in the art will appreciate that there are numerous variations and permutations of the above described systems and techniques that fall within the spirit and scope of the disclosed technology as set forth in the appended claims. For example, while a particular resist model has been employed above to illustrate how the disclosed technology may be implemented, it should be appreciated that various examples of the disclosed technology may be implemented to calibrate other lithography models such as a topographic model for modeling physical effects related to non-uniform wafer topography and an optical model.
