Abstract: A method for etching a stack is described. The method includes etching a first nitrogen-containing layer of the stack by applying a non-metal gas and discontinuing the application of the non-metal gas upon determining that a first oxide layer is reached. The first oxide layer is under the first nitrogen-containing layer. The method further includes etching the first oxide layer by applying a metal-containing gas. The application of the metal-containing gas is discontinued upon determining that a second nitrogen-containing layer will be reached. The second nitrogen-containing layer is situated under the first oxide layer. The method includes etching the second nitrogen-containing layer by applying the non-metal gas.

Abstract: Various embodiments herein relate to methods and apparatus for etching a memory hole in a stack of materials on a substrate. In some cases, the stack includes alternating layers of silicon oxide and silicon nitride. In other cases, the stack includes alternating layers of silicon oxide and polysilicon. In either case, three or more sets of processing conditions are used to etch the substrate. Various processing conditions such as the composition of a reactant mixture, pressure, substrate temperature, and/or plasma generation conditions are varied between the three or more sets of processing conditions to produce high quality etching results with high selectivity, a highly vertical etch profile, and a low degree of bowing.

Abstract: In an approach to AI explaining for natural language processing, responsive to receiving an input text for a machine learning model, an output is generated from the machine learning model. A plurality of alteration techniques are applied to the input text to generate one or more alternate outputs, where each alternate output corresponds to an alteration technique. A variation rate of the alternate output is calculated for each alteration technique. A preferred technique of generating neighboring data of the input text is generated based on a comparison of the variation rate of the alternate output for each alteration technique.

Abstract: A plasma lining structure is used in a process chamber to block direct line-of-sight for plasma generated within to grounded surface. The plasma lining structure includes a plurality of sections to cover at least one or more portions of an inside surface of a plasma confinement structure disposed in the process chamber. The sections of the plasma lining structure are positioned between a plasma region and the sidewall of the plasma confinement structure, when the plasma lining structure and the plasma confinement structure are disposed in the plasma chamber, such that the sections directly face the plasma region.

Abstract: In an approach to AI explaining for natural language processing, responsive to receiving an input text for a machine learning model, an output is generated from the machine learning model. A plurality of alteration techniques are applied to the input text to generate one or more alternate outputs, where each alternate output corresponds to an alteration technique. A variation rate of the alternate output is calculated for each alteration technique. A preferred technique of generating neighboring data of the input text is generated based on a comparison of the variation rate of the alternate output for each alteration technique.

Abstract: A method of etching features in a stack comprising a dielectric material on a substrate is provided. In a step (a) an etch plasma is generated from an etch gas, exposing the stack to the etch plasma, and partially etching features in the stack. In a step (b) after step (a) an atomic layer deposition process is provided to deposit a protective film on sidewalls. The atomic layer deposition process comprises a plurality of cycles, wherein each cycle comprises exposing the stack to a first reactant gas comprising WF6, wherein the first reactant gas is adsorbed onto the stack and exposing the stack to a plasma formed from a second reactant gas, wherein the plasma formed from the second reactant gas reacts with the adsorbed first reactant gas to form the protective film over the stack. In a step (c) steps (a)-(b) are repeated at least one time.

Abstract: Systems and methods for applying three or more states for achieving a high aspect ratio dielectric etch operation are described. In one of the methods, a middle state is introduced between a high state and a low state. The middle state is applied to both a source radio frequency (RF) generator and a bias radio frequency (RF) generator. During the middle state, RF power is maintained to be between a high amount of RF power associated with the high state and a low amount of RF power associated with the low state to achieve the high aspect ratio dielectric etch.

Abstract: Various embodiments herein relate to methods and apparatus for etching a feature in a substrate. Often, the feature is etched in the context of forming a DRAM device. The feature is etched in dielectric material, which often includes silicon oxide. The feature is etched using chemistry that includes WF6. Although WF6 is commonly used as a deposition gas (e.g., to deposit tungsten-containing film), it can also be used during etching. Advantageously, the inclusion of WF6 in the etch chemistry can increase the etch rate of the dielectric material, as well as increase the selectivity of the etch. Unexpectedly, these benefits can be realized without any increase in capping.

Abstract: Methods and apparatus for etching a high aspect ratio feature in a stack on a substrate are provided. The feature may be formed in the process of forming a 3D NAND device. Typically, the stack includes alternating layers of material such as silicon oxide and silicon nitride or silicon oxide and polysilicon. WF6 is provided in the etch chemistry, which substantially reduces or eliminates problematic sidewall notching. Advantageously, this improvement in sidewall notching does not introduce other tradeoffs such as increased bowing, decreased selectivity, increased capping, or decreased etch rate.

Abstract: Methods and apparatus for etching a high aspect ratio feature in a stack on a substrate are provided. The feature may be formed in the process of forming a 3D NAND device. Typically, the stack includes alternating layers of material such as silicon oxide and silicon nitride or silicon oxide and polysilicon. WF6 is provided in the etch chemistry, which substantially reduces or eliminates problematic sidewall notching. Advantageously, this improvement in sidewall notching does not introduce other tradeoffs such as increased bowing, decreased selectivity, increased capping, or decreased etch rate.

Abstract: Training data including a first and second group of questions each associated with an answer is read into memory by a computer. A class of answers including the answer to a question for each question is determined, where each class of answers has a class label that is associated with each of the questions, and each of the questions are classified into a respective class of answers, accordingly. First and second training data is generated including the first and second groups of questions and corresponding classes of answers for use in first and second classifiers, respectively. Each question of the first and second group of questions is classified by the second and first classifiers, respectively, where the classifying generates corresponding classification results. The first or second training data is updated based on the classification results to generate corresponding updated first or second training data, respectively.

Abstract: Training data including a first and second group of questions each associated with an answer is read into memory by a computer. A class of answers including the answer to a question for each question is determined, where each class of answers has a class label that is associated with each of the questions, and each of the questions are classified into a respective class of answers, accordingly. First and second training data is generated including the first and second groups of questions and corresponding classes of answers for use in first and second classifiers, respectively. Each question of the first and second group of questions is classified by the second and first classifiers, respectively, where the classifying generates corresponding classification results. The first or second training data is updated based on the classification results to generate corresponding updated first or second training data, respectively.

Abstract: Various embodiments herein relate to methods and apparatus for etching a feature in a substrate. Often, the feature is etched in the context of forming a DRAM device. The feature is etched in dielectric material, which often includes silicon oxide. The feature is etched using chemistry that includes WF6. Although WF6 is commonly used as a deposition gas (e.g., to deposit tungsten-containing film), it can also be used during etching. Advantageously, the inclusion of WF6 in the etch chemistry can increase the etch rate of the dielectric material, as well as increase the selectivity of the etch. Unexpectedly, these benefits can be realized without any increase in capping.

Abstract: Systems and methods for applying three or more states for achieving a high aspect ratio dielectric etch operation are described. In one of the methods, a middle state is introduced between a high state and a low state. The middle state is applied to both a source radio frequency (RF) generator and a bias radio frequency (RF) generator. During the middle state, RF power is maintained to be between a high amount of RF power associated with the high state and a low amount of RF power associated with the low state to achieve the high aspect ratio dielectric etch.

Abstract: Various embodiments herein relate to methods and apparatus for etching a feature in a substrate. Often, the feature is etched in the context of forming a DRAM device. The feature is etched in dielectric material, which often includes silicon oxide. The feature is etched using chemistry that includes WF6. Although WF6 is commonly used as a deposition gas (e.g., to deposit tungsten-containing film), it can also be used during etching. Advantageously, the inclusion of WF6 in the etch chemistry can increase the etch rate of the dielectric material, as well as increase the selectivity of the etch. Unexpectedly, these benefits can be realized without any increase in capping.

Abstract: Various embodiments herein relate to methods and apparatus for etching a feature in a substrate. Often, the feature is etched in the context of forming a DRAM device. The feature is etched in dielectric material, which often includes silicon oxide. The feature is etched using chemistry that includes WF6. Although WF6 is commonly used as a deposition gas (e.g., to deposit tungsten-containing film), it can also be used during etching. Advantageously, the inclusion of WF6 in the etch chemistry can increase the etch rate of the dielectric material, as well as increase the selectivity of the etch. Unexpectedly, these benefits can be realized without any increase in capping.

Abstract: Systems and methods for applying three or more states for achieving a high aspect ratio dielectric etch operation are described. In one of the methods, a middle state is introduced between a high state and a low state. The middle state is applied to both a source radio frequency (RF) generator and a bias radio frequency (RF) generator. During the middle state, RF power is maintained to be between a high amount of RF power associated with the high state and a low amount of RF power associated with the low state to achieve the high aspect ratio dielectric etch.

Abstract: Training data including a first and second group of questions each associated with an answer is read into memory by a computer. A class of answers including the answer to a question for each question is determined, where each class of answers has a class label that is associated with each of the questions, and each of the questions are classified into a respective class of answers, accordingly. First and second training data is generated including the first and second groups of questions and corresponding classes of answers for use in first and second classifiers, respectively. Each question of the first and second group of questions is classified by the second and first classifiers, respectively, where the classifying generates corresponding classification results. The first or second training data is updated based on the classification results to generate corresponding updated first or second training data, respectively.

Abstract: Training data including a first and second group of questions each associated with an answer is read into memory by a computer. A class of answers including the answer to a question for each question is determined, where each class of answers has a class label that is associated with each of the questions, and each of the questions are classified into a respective class of answers, accordingly. First and second training data is generated including the first and second groups of questions and corresponding classes of answers for use in first and second classifiers, respectively. Each question of the first and second group of questions is classified by the second and first classifiers, respectively, where the classifying generates corresponding classification results. The first or second training data is updated based on the classification results to generate corresponding updated first or second training data, respectively.

Abstract: Training data including a first and second group of questions each associated with an answer is read into memory by a computer. A class of answers including the answer to a question for each question is determined, where each class of answers has a class label that is associated with each of the questions, and each of the questions are classified into a respective class of answers, accordingly. First and second training data is generated including the first and second groups of questions and corresponding classes of answers for use in first and second classifiers, respectively. Each question of the first and second group of questions is classified by the second and first classifiers, respectively, where the classifying generates corresponding classification results. The first or second training data is updated based on the classification results to generate corresponding updated first or second training data, respectively.