Simulating near field image in optical lithography

1. A method for determining a near field image for optical lithography, comprising:

receiving a thin mask image indicative of a photomask feature, wherein the thin mask image is determined without considering a mask topography effect associated with the photomask feature;

determining, from the thin mask image by a processor, a near field image using an artificial neural network (ANN), wherein

the ANN uses the thin mask image as input,

the ANN comprises at least one of multilayer perceptron (MLP) model and a convolutional neural network (CNN) model,

input data for the ANN comprises image data of a sampled point of multiple sampled points in the thin mask image, and

the image data comprises at least one of: image intensity of the thin mask image at the sampled point, and a value of a vector image determined from the thin mask image; and

performing a photolithography simulation based on the near field image to determine an aerial image.

2. The method of claim 1 , wherein the photomask feature comprises at least one of: a mask pattern, an edge of the mask pattern, a corner of the mask pattern, and an area of the mask pattern.

3. The method of claim 1 , wherein determining the near field image using the ANN comprises:

determining the near field image by adding the thin mask image to outputs of the ANN, wherein the outputs are indicative of difference between the thin mask image and the near field image and the ANN uses a vector image determined from the thin mask image as input.

4. The method of claim 1 , wherein the multiple sampled points in the thin mask image are sampled in accordance with a sampling scheme comprising one of a concentric circle area sampling (CCAS), a concentric square sampling (CSS), and a uniform sampling.

5. The method of claim 4 , wherein parameters associated with the ANN comprises a weight associated with the sampled point and the weight is determined based on a distance between the sampled point and another sampled point of the multiple sampled points.

6. The method of claim 5 , further comprising:

determining whether a match exists between the near field image determined using the ANN and a defined image indicative of the photomask feature, wherein the defined image is determined using a rigorous electromagnetic simulation technique; and

based on a determination that no match exists between the near field image and the defined image, updating the parameters associated with the ANN.

7. The method of claim 5 , further comprising:

determining whether a cost value associated with the near field image is minimized; and

based on a determination that the cost value associated with the near field image is not minimized, updating the parameters associated with the ANN.

8. The method of claim 1 , wherein the near field image comprises a complex value.

9. The method of claim 1 , wherein determining the near field image using the ANN comprises:

determining gradient data associated with the near field image, wherein the gradient data comprises a gradient of the near field image with respect to the thin mask image.

10. An apparatus for determining a near field image for optical lithography, comprising:

a processor; and

a memory coupled to the processor, the memory configured to store instructions which when executed by the processor become operational with the processor to:

receive a thin mask image indicative of a photomask feature, wherein the thin mask image is determined without considering a mask topography effect associated with the photomask feature and the photomask feature comprises at least one of a mask pattern, an edge of the mask pattern, a corner of the mask pattern, and an area of the mask pattern;

determine, from the thin mask image, a near field image and gradient data associated with the near field image using an artificial neural network (ANN), wherein the gradient data comprises a gradient of the near field image with respect to the thin mask image, and wherein the ANN comprises at least one of multilayer perceptron (MLP) model and a convolutional neural network (CNN) model and the ANN uses the thin mask image as input; and

perform a photolithography simulation based on the near field image.

11. The apparatus of claim 10 , wherein the instructions operational with the processor to determine the near field image using the ANN further comprise instructions to:

determine the near field image by adding the thin mask image to outputs of the ANN, wherein the outputs are indicative of difference between the thin mask image and the near field image, the ANN uses a vector image determined from the thin mask image as input, and the near field image comprises a complex value.

12. The apparatus of claim 10 , wherein

input data for the ANN comprises image data of a sampled point of multiple sampled points sampled in the thin mask image in accordance with a sampling scheme comprising one of a concentric circle area sampling (CCAS), a concentric square sampling (CSS), and a uniform sampling,

the image data comprises at least one of: image intensity of the thin mask image at the sampled point, and a value of a vector image determined from the thin mask image, and

parameters associated with the ANN comprises a weight associated with the sampled point and the weight is determined based on a distance between the sampled point and another sampled point of the multiple sampled points.

13. The apparatus of claim 12 , wherein the memory further comprises instructions when executed by the processor become operational with the processor to:

determine whether a cost value associated with the near field image is minimized; and

based on a determination that the cost value associated with the near field image is not minimized, update the parameters associated with the ANN.

14. A non-transitory computer-readable medium storing a set of instructions which when executed by an apparatus using a processor become operational with the processor for determining a near field image for optical lithography, the non-transitory computer-readable medium comprising instructions to:

receive a thin mask image indicative of a photomask feature, wherein the thin mask image is determined without considering a mask topography effect associated with the photomask feature and the photomask feature comprises at least one of a mask pattern, an edge of the mask pattern, a corner of the mask pattern, and an area of the mask pattern; and

determine, from the thin mask image, a near field image using an artificial neural network (ANN), wherein

the ANN comprises at least one of multilayer perceptron (MLP) model and a convolutional neural network (CNN) model and the ANN uses the thin mask image as input,

input data for the ANN comprises image data of a sampled point of multiple sampled points sampled in the thin mask image in accordance with a sampling scheme comprising one of a concentric circle area sampling (CCAS), a concentric square sampling (CSS), and a uniform sampling,

the image data comprises at least one of: image intensity of the thin mask image at the sampled point, and a value of a vector image determined from the thin mask image, and

parameters associated with the ANN comprises a weight associated with the sampled point and the weight is determined based on a distance between the sampled point and another sampled point of the multiple sampled points; and

perform a photolithography simulation based on the near field image.

15. The non-transitory computer-readable medium of claim 14 , wherein the instructions to determine the near field image using the ANN further comprise instructions to:

determine the near field image by adding the thin mask image to outputs of the ANN, wherein the outputs are indicative of difference between the thin mask image and the near field image, the ANN uses a vector image determined from the thin mask image as input, and the near field image comprises a complex value.

16. The non-transitory computer-readable medium of claim 14 , further comprising instructions to:

determine whether a cost value associated with the near field image is minimized; and

based on a determination that the cost value associated with the near field image is not minimized, update the parameters associated with the ANN.

17. The non-transitory computer-readable medium of claim 14 , wherein the instructions to determine the near field image using the ANN further comprise instructions to:

determine gradient data associated with the near field image, wherein the gradient data comprises a gradient of the near field image with respect to the thin mask image.