Search

CN-121978868-A - Training method and using method of optical proximity correction model and electronic equipment

CN121978868ACN 121978868 ACN121978868 ACN 121978868ACN-121978868-A

Abstract

The invention provides a training method, a using method and electronic equipment of an optical proximity effect correction model, and relates to the technical field of photoetching technology, wherein the method comprises the steps of obtaining graphic features of a first graphic, wherein the absolute value of the difference between a Mask Error Enhancement Factor (MEEF) estimated value of a first measurement point of the first graphic and a MEEF theoretical value of the first measurement point is larger than a first preset threshold; the method comprises the steps of selecting a first pattern matched with pattern features from a sample chip, determining a plurality of first measuring points of the first pattern, respectively taking a photoresist linewidth corresponding to each first measuring point as input, taking a mask linewidth corresponding to each first measuring point as output, and training an optical proximity effect correction model to enable a value of a loss function to be smaller than a first preset threshold value, wherein the value of the loss function is positively correlated with an absolute value of a difference between an MEEF estimated value of each first measuring point and an MEEF measured value of the first measuring point.

Inventors

  • ZHANG WENSHENG
  • WANG YI
  • LIU RUIFAN

Assignees

  • 深圳市鹏芯微集成电路制造有限公司

Dates

Publication Date
20260505
Application Date
20241031

Claims (10)

  1. 1. A method of training an optical proximity correction model, comprising: Obtaining a graph characteristic of a first graph, wherein an absolute value of a difference between a Mask Error Enhancement Factor (MEEF) estimated value of a first measurement point of the first graph and a MEEF theoretical value of the first measurement point is larger than a first preset threshold; screening a second graph matched with the graph characteristic from a sample chip; determining a plurality of second measurement points of the second pattern, and And respectively taking the photoresist linewidth corresponding to each second measurement point as input, taking the mask linewidth corresponding to each second measurement point as output, and training the optical proximity effect correction model to ensure that the value of a loss function is smaller than a second preset threshold value, wherein the value of the loss function is positively correlated with the absolute value of the difference between the MEEF estimated value of each second measurement point and the MEEF measured value of the second measurement point.
  2. 2. The method of claim 1, wherein the graphical feature comprises at least one of a line pitch, a line length, and a line width.
  3. 3. The method of claim 1 or 2, wherein the loss function is: Wherein a j represents the MEEF estimated value of the jth second measurement point, B j represents the MEEF measured value of the jth second measurement point, W j represents the weight of the jth second measurement point, and n represents the number of the plurality of second measurement points.
  4. 4. A method according to claim 3, wherein: And under the condition that at least one second measuring point with abnormal MEEF measured value exists in the plurality of second measuring points, the weight of the at least one second measuring point is smaller than that of other second measuring points in the plurality of second measuring points.
  5. 5. A method according to claim 3, further comprising: Dividing the plurality of second measuring points into a plurality of subgroups according to the width of the photoresist, wherein the difference value between the width values of the photoresist line widths of different second measuring points in the same subgroup is smaller than a third preset threshold value; The weights of the second measuring points in the same subgroup are the same, and the weights of the subgroup with the characteristic values being closer to the preset width value determined according to the process requirement are larger, wherein the characteristic values comprise the upper limit, the lower limit or the average value of the width values of the photoresist linewidths of the second measuring points in the subgroup.
  6. 6. The method according to claim 4, wherein: The MEEF measurement value of the at least one second measurement point is greater than a fourth preset threshold value, and/or The variation degree of the MEEF measured value of the at least one second measuring point is larger than a fifth preset threshold value.
  7. 7. The method of claim 1 or 2, wherein the determining the plurality of second measurement points of the second graph comprises: adding a plurality of initial measurement points in the second graph; Screening a plurality of third measurement points from the plurality of initial measurement points, wherein the absolute value of the difference between the MEEF estimated value of each third measurement point and the MEEF theoretical value of the third measurement point is greater than a sixth preset threshold value, and And performing de-duplication operation on the third measurement points with the same environment according to the environments of the plurality of third measurement points, wherein the third measurement points remained after the de-duplication operation are used as the plurality of second measurement points.
  8. 8. A method of using an optical proximity correction model, comprising: Obtaining a desired photoresist linewidth, and And determining the mask plate line width corresponding to the expected photoresist line width according to the expected photoresist line width by using the optical proximity effect correction model obtained by training according to the training method of any one of claims 1-7.
  9. 9. An electronic device, comprising: memory, and A processor coupled to the memory and configured to perform the method of any of claims 1-8 based on instructions stored in the memory.
  10. 10. A computer readable storage medium comprising a computer program, wherein the computer program when executed by a processor implements the steps of the method of any of claims 1-8.

Description

Training method and using method of optical proximity correction model and electronic equipment Technical Field The disclosure relates to the technical field of photolithography processes, in particular to a training method, a using method and electronic equipment of an optical proximity correction model. Background In the field of photolithography, an Optical Proximity Correction (OPC) model may correct a pattern on a reticle so that the pattern projected onto the photoresist meets process requirements. Disclosure of Invention In the related art, even if the OPC model is modified, the pattern projected onto the photoresist still does not meet the process requirements. That is, the OPC model does not accurately correct patterns on the reticle. Analysis has found that in the related art, when the OPC model is trained, data related to Mask Error Enhancement Factor (MEEF) sensitive patterns is removed from the training data. For a certain pattern, if the deviation between the MEEF estimated value and the MEEF theoretical value of at least one measurement point of the pattern is too large, the pattern is considered to be a MEEF sensitive pattern. Therefore, the OPC model in the related art cannot accurately output the DOM corresponding to the MEEF sensitive pattern, that is, cannot accurately correct the pattern on the mask corresponding to the MEEF sensitive pattern. In order to solve the above-described problems, the embodiments of the present disclosure propose the following solutions. According to an aspect of the disclosed embodiments, a training method for an optical proximity correction model is provided, which includes obtaining a pattern feature of a first pattern, wherein an absolute value of a difference between a Mask Error Enhancement Factor (MEEF) estimated value of a first measurement point of the first pattern and a MEEF theoretical value of the first measurement point is greater than a first preset threshold, screening a second pattern matched with the pattern feature in a sample chip, determining a plurality of second measurement points of the second pattern, taking a photoresist line width corresponding to each second measurement point as an input, and taking a mask line width corresponding to each second measurement point as an output, and training the optical proximity correction model so that a value of a loss function is smaller than a second preset threshold, wherein the value of the loss function is positively correlated with the absolute value of the difference between the MEEF estimated value of each second measurement point and the MEEF measured value of the second measurement point. In some embodiments, the graphical features include at least one of line spacing, line length, and line width. In some embodiments, the loss function is: Wherein a j represents the MEEF estimated value of the jth second measurement point, B j represents the MEEF measured value of the jth second measurement point, W j represents the weight of the jth second measurement point, and n represents the number of the plurality of second measurement points. In some embodiments, in the case where at least one second measurement point with abnormal MEEF measurement is present in the plurality of second measurement points, the weight of the at least one second measurement point is less than the weight of other second measurement points in the plurality of second measurement points. In some embodiments, the plurality of second measurement points are divided into a plurality of subgroups according to the size of the photoresist line width, and the difference between the width values of the photoresist line widths of different second measurement points in the same subgroup is smaller than a third preset threshold, wherein the weights of the second measurement points in the same subgroup are the same, and the weights of subgroups with feature values closer to the preset width values determined according to the process requirements are larger, and the feature values comprise an upper limit, a lower limit or an average value of the width values of the photoresist line widths of the second measurement points in the subgroup. In some embodiments, the MEEF measurement at the at least one second measurement point is greater than a fourth predetermined threshold, and/or the variation of the MEEF measurement at the at least one second measurement point is greater than a fifth predetermined threshold. In some embodiments, the determining the plurality of second measurement points of the second graph includes adding a plurality of initial measurement points to the second graph, screening a plurality of third measurement points from the plurality of initial measurement points, wherein an absolute value of a difference between an MEEF estimated value of each third measurement point and an MEEF theoretical value of the third measurement point is greater than a sixth preset threshold, and performing a deduplication operation on third