Device and method for generating photomasks
The present description concerns a method that includes the compression, by a processor, of an image comprising first patterns by transforming the image into a first representation formed of two-point elements. The method also includes the execution, by a neural network, of an inference operation on the first representation to generate a second representation formed of two-point elements. The method further includes the generation of a lithographic mask based on the decompression of the second representation.
1. A method, comprising:
compressing, by a processor, an image comprising first patterns, by transforming the image into a first representation formed of two-point elements, the image covering a surface contained in a plane;
wherein the image is compressed by:
projecting a vertex of a first polygon of the image onto a first point of a first curve running throughout a first surface of the plane; and
projecting the vertex onto a second point of a second curve running throughout a second surface around the first point, the second surface having an area smaller than the area of the first surface;
executing, by a neural network, an inference operation on the first representation to generate a second representation formed of two-point elements; and
generating a lithographic mask based on decompression of the second representation.
2. The method according to claim 1 , further comprising using the lithographic mask to obtain patterns on a photoresist covering a substrate of a wafer by lithography and then performing an etching process to form corresponding patterns in the substrate.
3. The method according to claim 1 , wherein the image is represented in vector format.
4. The method according to claim 1 , wherein the first surface and/or the second surface is a rectangle.
5. The method according to claim 1 , wherein the first surface and/or the second surface is a right-triangle.
6. The method according to claim 1 , wherein the first surface and/or the second surface comprises a first right-triangle and a second right-triangle arranged as a rectangle, and the first representation comprises a bit indicating in which of the first right-triangle and the second right-triangle the first point and/or the second point is located.
7. The method according to claim 1 , wherein the first surface is a rectangle and the second surface comprises at least one right-triangle.
8. The method according to claim 1 , wherein the image is formed of a plurality of polygons, the compression comprising two-point compression of each vertex of each polygon and two-point compression of at least certain perimetric points of each polygon which are not vertices.
9. The method according to claim 8 , comprising comparing a length of each edge of each polygon with a threshold length and, when the length of an edge exceeds the threshold length, compressing at least one perimetric point on this edge.
10. The method according to claim 9 , wherein the threshold length is shorter than a minimum distance separating any two polygons from among the plurality of polygons.
11. The method according to claim 1 , further comprising segmenting the image into a plurality of sub-images, said steps of compressing and generating the mask being applied to each of the sub-images.
12. A method, comprising:
training a neural network to generate lithographic masks by:
generating, at least partially by the neural network, an image representing a lithographic mask;
wherein generating the image representing a lithographic mask comprises projecting a vertex of a first polygon onto a first point of a first curve running throughout a first surface of a plane and projecting the vertex onto a second point of a second curve running throughout a second surface around the first point, the second surface having an area smaller than the area of the first surface;
estimating, by using a lithographic simulation tool, a pattern resulting from the use of the mask;
comparing the resulting pattern and a reference pattern; and
modifying one or a plurality of parameters of the neural network based on the comparison; and
printing the image representing the lithographic mask on a support to produce the lithographic mask.
13. The method according to claim 12 , wherein generating the image comprises:
compressing, by an algorithm implemented by a data processing device, a first image containing first patterns by transforming the image into a first representation formed of two-point elements;
applying the first representation to the neural network to generate a second representation formed of two-point elements; and
generating the image based on a decompression of the second representation.
14. The method according to claim 12 , wherein the neural network is a generative adversarial network comprising a generator network being a first autoencoder and a discriminator network being a second autoencoder.
15. The method according to claim 14 , wherein the first autoencoder and/or the second autoencoder is of U-net type.
16. The method according to claim 12 , wherein the training of the neural network is performed iteratively, and, at each iteration:
a loss is calculated, by using a loss function, based on the comparison between the resulting pattern and the reference pattern; and
modifying one or a plurality of parameters of the network comprises updating of synaptic weights of the neural network based on the calculated loss.
17. The method according to claim 12 , further comprising using the lithographic mask to obtain patterns on a photoresist covering a substrate of a wafer by lithography and then performing an etching process to form corresponding patterns in the substrate.