Modeling method for drc verification patterns
By identifying and establishing a standard DRC verification model, the problem of time-consuming and costly generation of DRC verification graphics library was solved, enabling fast and accurate generation and updating of graphics library to adapt to process iteration.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI HUAHONG GRACE SEMICON MFG CORP
- Filing Date
- 2022-12-06
- Publication Date
- 2026-05-29
AI Technical Summary
The existing technology requires a lot of time and cost to establish a DRC verification graphic library, and it is difficult to quickly and accurately generate and update DRC verification graphics that change with process iterations.
By reading the text design rules, identifying the test type and layer objects, establishing a standard DRC verification model, and quickly generating a DRC verification graphic library through sample library matching and normalization.
It enables the rapid and accurate generation of DRC verification graphics libraries, reducing testing costs and adapting to rapid changes in process iteration.
Smart Images
Figure CN116227422B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a semiconductor integrated circuit manufacturing method, and more particularly to a modeling method for design rule check (DRC) verification patterns. Background Technology
[0002] DRC is an essential verification step in circuit design, helping layout engineers quickly check whether the circuit layout conforms to the process design rules provided by the foundry.
[0003] When developing DRC verification files, wafer foundries typically build a large number of test patterns, i.e., DRC verification patterns, in reverse according to process design rules to test the accuracy of the DRC verification files. A DRC verification pattern library can be at least several thousand in size, and with rapid process iterations, it will consume a lot of testing costs. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to provide a modeling method for DRC verification graphics, which can establish a corresponding standard DRC verification model based on the read text design rules, thereby enabling the rapid and accurate batch generation of DRC verification graphics and the formation of a DRC verification graphics library.
[0005] To solve the above-mentioned technical problems, the modeling method for DRC verification graphics provided by the present invention includes the following steps:
[0006] Step 1: Read the text design rules.
[0007] Step 2: Identify the text design rules, including the following sub-steps:
[0008] Step 21: Extract the test types contained in the text design rules according to the description format of the test types.
[0009] Step 22: Identify the layer objects contained in the text design rules based on the layer descriptions in the text design rules.
[0010] Step 3: Establish a standard DRC verification model based on the test types and layer objects contained in the text design rules.
[0011] A further improvement is that, in step 21, the test type refers to the relationship between layers in the layout, and each category of the test type has a corresponding fixed description format.
[0012] A further improvement is that the test types include: spacing, enclosure, width, area, and overlap.
[0013] A further improvement is that, in step 22, the layer description is uncertain and will change according to the language habits of the process and design rule setters.
[0014] A further improvement is that, in step 22, when multiple layer objects are identified from the layer description, the layer object with the highest priority is selected as the layer object included in the text design rule.
[0015] A further improvement is that the text design rules include prepositions located between the layers to be detected.
[0016] In step 22, when multiple layer objects are identified from the layer description, the priority parameters include a first word distance between the layer object and the preposition, and a second text matching rate formed by the ratio of the number of characters in the layer object to the number of characters in the entire text design rule.
[0017] The smaller the first word distance, the higher the priority.
[0018] The higher the second text matching rate, the higher the priority.
[0019] A further improvement is that the first word distance is taken as the number of characters between the first letter of the layer object and the preposition.
[0020] A further improvement is that, in step 22, when multiple layer objects are identified from the layer description, the following sub-step is also included:
[0021] Step 221: Extract two-dimensional vectors from each of the identified layer objects and form a layer object matrix. The first dimension vector is a vector composed of the first word distance of each layer object, and the second dimension vector is a vector composed of the second text matching rate of each layer object.
[0022] Step 222: Normalize the layer object matrix to form a layer object normalization matrix.
[0023] Step 223: Multiply the normalization matrix of the layer object by a 2×1 preposition transpose matrix to obtain a one-dimensional layer object vector. Each value in the one-dimensional layer object vector is the final matching value of the corresponding layer object. The layer object with the largest final matching value has the highest priority.
[0024] A further improvement is that, in step 222, the first dimension vector is first transformed, and then the normalization process is performed; the transformation process is to subtract the first word distance corresponding to each layer object in the first dimension vector from the third offset.
[0025] A further improvement is that the normalization process is performed by dividing each element of the column vector by the square root of the sum of the squares of the elements of the corresponding vector.
[0026] A further improvement is that the third offset is the number of characters from the preposition to the end of the character in the text design rule.
[0027] A further improvement is that the first element of the preposition transpose matrix is the third offset, and the second element is 1.
[0028] A further improvement is that step one, defined as the sample library, is included before step one.
[0029] The basic sample library contains: test type format samples and layer format samples.
[0030] A further improvement is that, in step 21, the test types contained in the text design rules are extracted by matching the description format of the test type with the test type format samples in the sample library.
[0031] A further improvement is that, in step 22, layer objects are identified by matching the layer descriptions in the text design rules with the layer format samples.
[0032] A further improvement is that step three also includes: generating DRC verification graphics in batches based on the standard DRC verification model and forming a DRC verification graphics library.
[0033] After reading the text design rules, this invention identifies the text design rules according to the description format of the test type and the layer description, thereby accurately identifying the test type and layer object of the text design rules and forming a standard DRC verification model. Therefore, this invention can unify text design rules with the same layer object but different layer descriptions to the same standard and form a standard DRC verification model, thereby enabling fast and accurate batch generation of DRC verification graphics and the formation of a DRC verification graphics library.
[0034] With the accelerating pace of process iteration, the number of DRC verification graphics in the DRC verification graphics library is large and needs to change constantly with the process. This invention can greatly reduce testing costs by rapidly and accurately generating DRC verification graphics libraries in batches. Attached Figure Description
[0035] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:
[0036] Figure 1 This is a flowchart of the modeling method for DRC verification graphics according to an embodiment of the present invention;
[0037] Figure 2 This is a detailed flowchart of the modeling method for DRC verification graphics according to a preferred embodiment of the present invention;
[0038] Figure 3A This is a schematic diagram of the text design rules in the modeling method of DRC verification graphics according to a preferred embodiment of the present invention;
[0039] Figure 3B This is a schematic diagram of multiple layer objects identified in the layer description of the text design rules in the modeling method of DRC verification graphics in a preferred embodiment of the present invention;
[0040] Figure 4A This is a schematic diagram of the layer object matrix in the modeling method of DRC verification graphics according to a preferred embodiment of the present invention;
[0041] Figure 4B This is a schematic diagram of the normalization process of the layer object matrix to form a layer object normalization matrix in the modeling method of DRC verification graphics in a preferred embodiment of the present invention;
[0042] Figure 4C This is a schematic diagram illustrating the process of multiplying the normalized matrix of a layer object by the preposition transpose matrix to obtain a one-dimensional layer object vector in the modeling method of DRC verification graphics according to a preferred embodiment of the present invention. Detailed Implementation
[0043] like Figure 1 The diagram shown is a flowchart of the modeling method for DRC verification graphics according to an embodiment of the present invention; as shown Figure 2 The diagram shown is a detailed flowchart of the modeling method for DRC verification graphics according to a preferred embodiment of the present invention; the modeling method for DRC verification graphics according to the present invention includes the following steps:
[0044] Step 1: Read the text design rules.
[0045] In some preferred embodiments, the step before step one includes:
[0046] Define the sample library.
[0047] The basic sample library contains: test type format samples and layer format samples.
[0048] like Figure 2 As shown, this step corresponds to step S001: defining the basic sample library for subsequent matching, including:
[0049] (1) Test type format sample
[0050] (2) Basic layer format sample
[0051] Step 1 corresponds to Figure 2 Step S002: Read the text design rules.
[0052] Step 2: Identify the text design rules, including the following sub-steps:
[0053] Step 21: Extract the test types contained in the text design rules according to the description format of the test types.
[0054] In this embodiment of the invention, step 21, the classification of test types refers to the relationships between layers, which can generally be divided into spacing, enclosure, width, area, overlap, etc. These relationships between layers have fixed descriptive formats in text design rules and exhibit characteristics of machine language.
[0055] Figure 2 In step 21, which corresponds to step S003, the test type is matched according to the fixed test type description format based on the sample library defined in step S001.
[0056] Step 22: Identify the layer objects contained in the text design rules based on the layer descriptions in the text design rules.
[0057] In this embodiment of the invention, the layer description refers to the test object layer of the aforementioned test type. Depending on different processes and the language habits of engineers who formulate design rules, there will be various forms of description. This uncertainty has the characteristics of natural language.
[0058] Since text design rules have both standardized machine language and natural language descriptive features, this embodiment of the invention extracts and identifies these two aspects through steps 21 and 22 respectively.
[0059] In some preferred embodiments, in step 22, when multiple layer objects are identified from the layer description, the layer object with the highest priority is selected as the layer object included in the text design rule.
[0060] The text design rule contains prepositions located between layers to be detected. In step 22, when multiple layer objects are identified from the layer description, the priority parameters include a first word distance between the layer object and the preposition, and a second text matching rate formed by the ratio of the number of characters in the layer object to the total number of characters in the entire text design rule.
[0061] The first word distance is the number of characters between the first letter of the layer object and the preposition.
[0062] The smaller the first word distance, the higher the priority.
[0063] The higher the second text matching rate, the higher the priority.
[0064] Figure 2 In this step, step 22 corresponds to step S004: based on the sample library defined in step S001, and combined with the specific context of the text design rules, the layer description is vectorized, and after normalization, the final layer object is identified according to the highest matching value.
[0065] like Figure 3A The diagram shown is a schematic of the text design rules in the modeling method of DRC verification graphics according to a preferred embodiment of the present invention. Figure 3A Let's take a text design rule as an example to illustrate. Figure 3A The text design rule in the dashed box 101 is “Minimum spacing of P+ to N+ACT inside 5V PW (PW in TGO2) is 0.1um”. The dashed box also includes descriptions of the test type and layer objects, which are extracted through steps 21 and 22, respectively.
[0066] In step 21, the extracted test type description format is fixed as: CHKTYPE lay1 PV lay2 CHKVAL;
[0067] Wherein, CHKTYPE: represents a fixed combination of check type characters, such as {minimum, spacing of}, {minimum, enclosure of}, etc. The CHKTYPE in the dashed box 101 is "Minimum spacingof" as shown in the box 102, which corresponds to the Chinese term "minimum spacing".
[0068] lay1 / lay2: These represent the descriptions of the layers to be inspected, i.e., the specific descriptions of layer 1 and layer 2. They have the characteristics of natural language. It is not necessary to identify lay1 and lay2 in step 21, but it is necessary to identify them in step 22. More detailed descriptions will follow.
[0069] PV: indicates prepositions between layers to be inspected, such as to, of, on, beyond, etc. Figure 3A In the text, PV corresponds to "to" shown in box 103.
[0070] CHKVAL: This represents the constraint value of the design rule, and it is usually located at the end of the text rule description. Figure 3A In the diagram, CHKVAL corresponds to “0.1um” shown in box 104, which means the minimum spacing is 0.1 micrometers.
[0071] Since boxes 102, 103, and 104 all have fixed description formats, they can be directly extracted through step 21. In some preferred embodiments, since the sample library has already been set in step S001, in step 21, the test types contained in the text design rules are extracted by matching the description format of the test type with the test type format samples in the sample library.
[0072] like Figure 3B The diagram shown illustrates the identification of multiple layer objects in the text design rules of the modeling method for DRC verification graphics according to a preferred embodiment of the present invention. It can be seen that layer 1 (lay1) before the preposition "to" can only describe one layer object, and therefore can be directly identified. However, layer 2 (lay2) after the preposition "to" can describe multiple layer objects, requiring identification, which is explained in detail below:
[0073] like Figure 3B As shown, there are 5 layer objects described by lay2, which are enclosed in boxes 1, 2, 3, 4 and 5 respectively. Layer object "1" corresponds to "N+", layer object "2" corresponds to "N+ACT", layer object "3" corresponds to "N+ACT inside 5V PW", layer object "4" corresponds to "PW", and layer object "5" corresponds to "TGO2". Obviously, these descriptions are all natural language descriptions and may change according to habits. Therefore, it is necessary to identify a correct layer object through step 22 of this embodiment of the invention.
[0074] In some preferred embodiments, since the sample library has been set in step S001, in step 22, the layer objects are identified by matching the layer descriptions in the text design rules with the layer format samples. That is, layer objects "1" to "5" can be identified by matching them with the layer format samples.
[0075] In this embodiment of the invention, step 22, when multiple layer objects are identified from the layer description, further includes the following sub-steps:
[0076] Step 221: Extract two-dimensional vectors from each of the identified layer objects and form a layer object matrix. The first dimension vector is a vector composed of the first word distance of each layer object, and the second dimension vector is a vector composed of the second text matching rate of each layer object.
[0077] As described earlier, due to the inherent uncertainty of natural language in describing layer objects, conventional keyword matching methods often result in matching multiple layer objects. However, based on the descriptive characteristics of text design rules, we can analyze that layer objects closer to the PV preposition have higher priority; and layer objects with a higher character count relative to the total character count of the text design rule also have higher priority.
[0078] Based on the above analysis, the description of a single layer object can be transformed into a two-dimensional vector (pwd, lmr):
[0079] pwd represents the first-dimensional vector, which is the vector composed of the first word distances of each layer object;
[0080] lmr represents the second-dimensional vector, which is the vector composed of the second text matching rate of each layer object.
[0081] (pwd, lmr) represents the layer object matrix. Figure 4A In this context, the layer object matrix is represented by matrix 201.
[0082] The literal meaning of "pwd" indicates the distance of the layer object from the preposition "PV". In this embodiment of the invention, the distance corresponding to "pwd" is represented by word distance, that is: the first word distance is taken as the number of characters between the first letter of the layer object and the preposition. Figure 4A As shown, the first-dimensional vector has Figure 3B The first word distances corresponding to the five layer objects in the image are represented by pwd1, pwd2, pwd3, pwd4 and pwd5 respectively.
[0083] The literal meaning of LMR is the text match rate of a layer object relative to the entire design rule, also known as the second text match rate. It is calculated as the percentage of consecutive characters in the layer object relative to the total number of characters in the entire text design rule. The calculation formula is:
[0084] lmr=lay_wnum / rule_wnum (1);
[0085] Here, `lay_wnum` represents the number of consecutive characters in the layer object, and `rule_wnum` represents the number of characters in the entire text design rule.
[0086] Depend on Figure 4A As shown, the second-dimensional vector has Figure 3B The second text matching rate corresponding to the 5 layer objects in the image is represented by lmr1, lmr2, lmr3, lmr4 and lmr5 respectively.
[0087] Figure 4A In China, according to Figure 3B Calculate the pwd and lmr values for each layer object and substitute the corresponding values into matrix 201 to obtain matrix 201a on the right.
[0088] From the above, it can be seen that... Figure 3B When extracting the lay2 layer, 5 layer objects can be matched. By performing two-dimensional vector extraction on each of these 5 layer objects, a 5*2 matrix can be formed, where each row represents a layer object and the number of rows represents the number of matched layer objects.
[0089] Step 222: Normalize the layer object matrix to form a layer object normalization matrix.
[0090] In some preferred embodiments, the first dimension vector is first transformed, and then the normalization process is performed; the transformation process involves subtracting the first word distance corresponding to each layer object in the first dimension vector from a third offset. In some embodiments, the third offset is the number of characters from the preposition to the end of the character in the text design rule.
[0091] The normalization process is performed by dividing each element of the column vector by the square root of the sum of the squares of the elements of the corresponding vector.
[0092] That is: combination Figure 4B The normalized matrix of the layer object corresponding to marker 202 is shown. To facilitate subsequent sorting, the generated layer object matrix is normalized column by column: first, the elements of the first column are transformed by matrix transformation, and the third offset D is uniformly subtracted. D represents the word distance from the PV preposition to the end of the character of the entire text rule. Then, the matrix is divided by the square root of the sum of the squares of the elements in each column.
[0093] After the transformation, all elements in the layer object normalization matrix 202 will be between 0 and 1, where:
[0094] <1> The closer the element in the first column is to 1, the closer the corresponding layer object is to the PV preposition.
[0095] <2> The closer an element in the second column is to 1, the more characters the corresponding layer object occupies in the entire text design rule.
[0096] Step 223: Multiply the normalization matrix of the layer object by a 2×1 preposition transpose matrix to obtain a one-dimensional layer object vector. Each value in the one-dimensional layer object vector is the final matching value of the corresponding layer object. The layer object with the largest final matching value has the highest priority.
[0097] In some preferred embodiments, the first element of the preposition transpose matrix is the third offset, and the second element is 1.
[0098] That is: combination Figure 4C As shown, after the layer object matrix is normalized, it is multiplied by the transpose matrix of the preposition PV, i.e., the preposition transpose matrix corresponding to mark 203, to compress the two-dimensional layer vector into a one-dimensional layer vector as shown in mark 204, thus obtaining the final matching value of each layer; then, as shown in box 205, the final matching values of the layers are arranged in descending order, and the layer object with the highest matching value is the one to be extracted.
[0099] Step 3: Establish a standard DRC verification model based on the test types and layer objects contained in the text design rules.
[0100] exist Figure 2 In the corresponding preferred embodiment, step three corresponds to step S005:
[0101] Test types and identified layer objects extracted based on steps S003 and S004;
[0102] Establish a standard DRC test model:
[0103] Test the constraint values of the layer object type, for example:
[0104] min_space pp ndif5v 0.1.
[0105] at last, Figure 3A The text design rule corresponding to the dashed box 101, “Minimum spacing of P+to N+ACTinside 5V PW(PW in TGO2)is 0.1um”, is represented by the standard “min_spacepp ndif5v 0.1” in the standard DRC verification model.
[0106] In this embodiment of the invention, step three further includes: generating DRC verification graphics in batches based on the standard DRC verification model and forming a DRC verification graphics library.
[0107] After reading the text design rules, this embodiment of the invention identifies the text design rules according to the description format of the test type and the layer description, thereby accurately identifying the test type and layer object of the text design rules and forming a standard DRC verification model. Therefore, this embodiment of the invention can unify text design rules with the same layer object but different layer descriptions to the same standard and form a standard DRC verification model, thereby enabling fast and accurate batch generation of DRC verification graphics and the formation of a DRC verification graphics library.
[0108] As the pace of process iteration accelerates, the number of DRC verification graphics in the DRC verification graphics library is large and needs to change constantly with the process. The embodiments of the present invention can greatly reduce testing costs by rapidly and accurately batch-forming the DRC verification graphics library.
[0109] The present invention has been described in detail above through specific embodiments, but these are not intended to limit the invention. Many modifications and improvements can be made by those skilled in the art without departing from the principles of the invention, and these should also be considered within the scope of protection of the present invention.
Claims
1. A modeling method for DRC verification graphics, characterized in that, Includes the following steps: Step 1: Read the text design rules; Step 2: Identify the text design rules, including the following sub-steps: Step 21: Extract the test types contained in the text design rules according to the description format of the test types; Step 22: Identify the layer objects contained in the text design rules based on the layer descriptions in the text design rules; The layer descriptions are uncertain and may change depending on the language habits of the process and design rule setters; When multiple layer objects are identified from the layer description, the layer object with the highest priority is selected as the layer object included in the text design rule; The text design rules contain prepositions located between the layers to be detected; When multiple layer objects are identified from the layer description, the priority parameters include a first word distance between the layer object and the preposition, and a second text matching rate formed by the ratio of the number of characters in the layer object to the number of characters in the entire text design rule; The smaller the first word distance, the higher the priority. The higher the second text matching rate, the higher the priority. Step 3: Establish a standard DRC verification model based on the test types and layer objects contained in the text design rules.
2. The modeling method for DRC verification graphics as described in claim 1, characterized in that: In step 21, the test type refers to the relationship between layers in the layout, and each category of the test type has a corresponding fixed description format.
3. The modeling method for DRC verification graphics as described in claim 2, characterized in that: The test types include: spacing, enclosure, width, area, and overlap.
4. The modeling method for DRC verification graphics as described in claim 2, characterized in that: The first word distance is the number of characters between the first letter of the layer object and the preposition.
5. The modeling method for DRC verification graphics as described in claim 2, characterized in that: In step 22, when multiple layer objects are identified from the layer description, the following sub-steps are also included: Step 221: Extract two-dimensional vectors from each of the identified layer objects and form a layer object matrix. The first dimension vector is a vector composed of the first word distance of each layer object, and the second dimension vector is a vector composed of the second text matching rate of each layer object. Step 222: Normalize the layer object matrix and form a layer object normalization matrix; Step 223: Multiply the normalization matrix of the layer object by a 2×1 preposition transpose matrix to obtain a one-dimensional layer object vector. Each value in the one-dimensional layer object vector is the final matching value of the corresponding layer object. The layer object with the largest final matching value has the highest priority.
6. The modeling method for DRC verification graphics as described in claim 5, characterized in that: In step 222, the first dimension vector is first transformed, and then the normalization process is performed; the transformation process is to subtract the first word distance corresponding to each layer object in the first dimension vector from the third offset.
7. The modeling method for DRC verification graphics as described in claim 6, characterized in that: The normalization process is performed by dividing each element of the column vector by the square root of the sum of the squares of the elements of the corresponding vector.
8. The modeling method for DRC verification graphics as described in claim 6, characterized in that: The third offset is the number of characters from the preposition to the end of the character in the text design rule.
9. The modeling method for DRC verification graphics as described in claim 8, characterized in that: The first element of the preposition transpose matrix is the third offset, and the second element is 1.
10. The modeling method for DRC verification graphics as described in claim 1, characterized in that: Before step one, there is also the step of defining the sample library; The sample library contains: test type format samples and layer format samples.
11. The modeling method for DRC verification graphics as described in claim 10, characterized in that: In step 21, the test types contained in the text design rules are extracted by matching the description format of the test type with the test type format samples in the sample library.
12. The modeling method for DRC verification graphics as described in claim 10, characterized in that: In step 22, the layer object is identified by matching the layer description in the text design rules with the layer format sample.
13. The modeling method for DRC verification graphics as described in claim 1, characterized in that: Step three also includes: generating DRC verification graphics in batches based on the standard DRC verification model and forming a DRC verification graphics library.