Abstract: Methods for removing a catalyst particle from a nanotube film used in a photolithographic patterning process are disclosed. The catalyst particle is identified based on its size in the nanotube film. This identification can be done using an inspection device such as a confocal microscope, which permits comparison of images taken in two or more separate focal planes to determine the size of particles. The catalyst particle is then exposed to a first absorption wavelength using a laser, which is selectively absorbed by the catalyst particle and which heats the catalyst particle to remove the catalyst particle from the nanotube film. Optionally, the catalyst particle-free nanotube film can be further exposed to a second absorption wavelength which is selectively absorbed by the film and promotes repair of the film. The resulting nanotube film can be used in a pellicle membrane.

Abstract: A lithography mask including a substrate, a phase shift layer on the substrate and an etch stop layer is provided. The phase shift layer is patterned and the substrate is protected from etching by the etch stop layer. The etch stop layer can be a material that is semi-transmissive to light used in photolithography processes or it can be transmissive to light used in photolithography processes.

Abstract: In various examples, high-precision semantic image editing for machine learning systems and applications are described. For example, a generative adversarial network (GAN) may be used to jointly model images and their semantic segmentations based on a same underlying latent code. Image editing may be achieved by using segmentation mask modifications (e.g., provided by a user, or otherwise) to optimize the latent code to be consistent with the updated segmentation, thus effectively changing the original, e.g., RGB image. To improve efficiency of the system, and to not require optimizations for each edit on each image, editing vectors may be learned in latent space that realize the edits, and that can be directly applied on other images with or without additional optimizations. As a result, a GAN in combination with the optimization approaches described herein may simultaneously allow for high precision editing in real-time with straightforward compositionality of multiple edits.

Abstract: A mask for use in a semiconductor lithography process includes a substrate, a mask pattern disposed on the substrate, and a light absorbing border surrounding the mask pattern. The light absorbing border is inset from at least two edges of the substrate to define a peripheral region outside of the light absorbing border. In some designs, a first peripheral region extends from an outer perimeter of the light absorbing border to a first edge of the substrate, and a second peripheral region that extends from the outer perimeter of the light absorbing border to a second edge of the substrate, where the first edge of the substrate and the second edge of the substrate are on opposite sides of the mask pattern.

Abstract: A mask characterization method comprises measuring an interference signal of a reflection or transmission mask for use in lithography; and determining a quality metric for the reflection or transmission mask based on the interference signal. A mask characterization apparatus comprises a light source arranged to illuminate a reflective or transmissive mask with light whereby mask-reflected or mask-transmitted light is generated; an optical grating arranged to convert the mask-reflected or mask-transmitted light into an interference pattern; and an optical detector array arranged to generate an interference signal by measuring the interference pattern.

Abstract: Approaches are presented for training an inverse graphics network. An image synthesis network can generate training data for an inverse graphics network. In turn, the inverse graphics network can teach the synthesis network about the physical three-dimensional (3D) controls. Such an approach can provide for accurate 3D reconstruction of objects from 2D images using the trained inverse graphics network, while requiring little annotation of the provided training data. Such an approach can extract and disentangle 3D knowledge learned by generative models by utilizing differentiable renderers, enabling a disentangled generative model to function as a controllable 3D “neural renderer,” complementing traditional graphics renderers.

Abstract: Apparatuses, systems, and techniques to determine pixel-level labels of a synthetic image. In at least one embodiment, the synthetic image is generated by one or more generative networks and the pixel-level labels are generated using a combination of data output by a plurality of layers of the generative networks.

Abstract: Apparatuses, systems, and techniques to generate complete depictions of objects based on a partial depiction of the object. In at least one embodiment, an image of a complete object is generated by one or more neural networks, based on an image of a portion of the object, using an encoder of the one or more neural networks trained using training data generated from output of a decoder of the one or more neural networks.

Abstract: A method of fabricating a mask is provided. The method includes providing a hard mask layer disposed on top of absorber, a capping layer, and a multilayer that are disposed on a substrate. The method includes forming a middle layer over the hard mask layer, forming a photo resist layer over the middle layer, patterning the photo resist layer, etching the middle layer through the patterned photo resist layer, etching the hard mask layer through the patterned middle layer, and etching the absorber through the patterned hard mask layer. In some embodiments, etching the hard mask layer through the patterned middle layer includes a dry-etching process that has a first removal rate of the hard mask layer and a second removal rate of the middle layer, and a ratio of the first removal rate of the hard mask layer to the second removal rate of the middle layer is greater than 5.

Abstract: Approaches are presented for training an inverse graphics network. An image synthesis network can generate training data for an inverse graphics network. In turn, the inverse graphics network can teach the synthesis network about the physical three-dimensional (3D) controls. Such an approach can provide for accurate 3D reconstruction of objects from 2D images using the trained inverse graphics network, while requiring little annotation of the provided training data. Such an approach can extract and disentangle 3D knowledge learned by generative models by utilizing differentiable renderers, enabling a disentangled generative model to function as a controllable 3D “neural renderer,” complementing traditional graphics renderers.

Abstract: A method for manufacturing a reticle is provided. The method includes forming a first reflective multilayer over a mask substrate. The method also includes forming a capping layer over the first reflective ML. The method further includes depositing a first absorption layer over the capping layer. In addition, the method includes depositing an etch stop layer over the first absorption layer. The method also includes forming a second reflective multilayer (ML) over the etch stop layer. The method further includes forming a second absorption layer over the second reflective ML. In addition, the method includes forming an opening through the second absorption layer and the second reflective ML until the etch stop layer is exposed. The method also includes etching the etch stop layer and the first absorption layer through the opening until the capping layer is exposed.

Abstract: A reticle and a method for manufacturing the same are provided. The reticle includes a mask substrate, a reflective multilayer (ML), a capping layer and an absorption composite structure. The reflective ML is positioned over a front-side surface of the mask substrate. The capping layer is positioned over the reflective ML. The absorption composite structure is positioned over the capping layer. The absorption composite structure includes a first absorption layer, a second absorption layer, a third absorption layer and an etch stop layer. The first absorption layer is positioned over the capping layer. The second absorption layer is positioned over the first absorption layer. The third absorption layer is positioned over the second absorption layer. The etch stop layer is positioned between the first absorption layer and the second absorption layer. The first absorption layer and the second absorption layer are made of the same material.

Abstract: Systems and methods are provided for tracking sharing of an electronic content. An exemplary method may include receiving a request to access content associated with a web address by a user. Based on a unique identity assigned to the user and the web address, a unique tracking web address may be generated. This tracking web address may be shared with additional users. As other users request content associated with the tracking web address, information regarding the sharing of the electronic content may be determined and stored, allowing for tracking of sharing behavior of users.

Abstract: Systems and methods are provided for tracking sharing of an electronic content. An exemplary method may include receiving a request to access content associated with a web address by a user. Based on a unique identity assigned to the user and the web address, a unique tracking web address may be generated. This tracking web address may be shared with additional users. As other users request content associated with the tracking web address, information regarding the sharing of the electronic content may be determined and stored, allowing for tracking of sharing behavior of users.

Abstract: The present invention relates generally to object annotation, specifically to polygonal annotations of objects. Described are methods of annotating an object including steps of receiving an image depicting an object, generating a set of image features using a CNN encoder implemented on one or more computers, and producing a polygon object annotation via a recurrent decoder or a Graph Neural Network. The recurrent decoder may include a recurrent neural network, a graph neural network or a gated graph neural network. A system for annotating an object and a method of training an object annotation system are also described.

Abstract: An apparatus that has a shape and uses a method to securely interweave with chain-link fences and the like without additional fasteners or mechanisms of similar nature. The shape is comprised of an orifice and a gap merging into each other. The interweaving method is a set of orientation-dependent operations established with the relative locations of three elements: apparatus, entanglement and 2 bent wires forming the entanglement.

Abstract: The present invention relates generally to object annotation, specifically to polygonal annotations of objects. Described are methods of annotating an object including steps of receiving an image depicting an object, generating a set of image features using a CNN encoder implemented on one or more computers, and producing a polygon object annotation via a recurrent decoder or a Graph Neural Network. The recurrent decoder may include a recurrent neural network, a graph neural network or a gated graph neural network. A system for annotating an object and a method of training an object annotation system are also described.

Abstract: The present invention relates generally to object annotation, specifically to polygonal annotations of objects. Described are methods of annotating an object including steps of receiving an image depicting an object, generating a set of image features using a CNN encoder implemented on one or more computers, and producing a polygon object annotation via a recurrent decoder or a Graph Neural Network. The recurrent decoder may include a recurrent neural network, a graph neural network or a gated graph neural network. A system for annotating an object and a method of training an object annotation system are also described.

Abstract: A reticle and a method for manufacturing the same are provided. The reticle includes a mask substrate, a reflective multilayer (ML), a capping layer and an absorption composite structure. The reflective ML is positioned over a front-side surface of the mask substrate. The capping layer is positioned over the reflective ML. The absorption composite structure is positioned over the capping layer. The absorption composite structure includes a first absorption layer, a second absorption layer, a third absorption layer and an etch stop layer. The first absorption layer is positioned over the capping layer. The second absorption layer is positioned over the first absorption layer. The third absorption layer is positioned over the second absorption layer. The etch stop layer is positioned between the first absorption layer and the second absorption layer. The first absorption layer and the second absorption layer are made of the same material.

Abstract: An apparatus that has a shape and uses a method to securely interweave with chain-link fences and the like without additional fasteners or mechanisms of similar nature. The said shape is comprised of an orifice and a gap merging into each other. The said interweaving method is a set of orientation-dependent operations established with the relative locations of three elements: apparatus, entanglement and 2 bent wires forming the said entanglement.