Abstract: Provided herein are low resistance metallization stack structures for logic and memory applications and related methods of fabrication. In some implementations, the methods involve providing a tungsten (W)-containing layer on a substrate; and depositing a molybdenum (Mo)-containing layer on the W-containing layer. In some implementations, the methods involve depositing a Mo-containing layer directly on a dielectric or titanium nitride (TiN) substrate without an intervening W-containing layer.

Abstract: Provided herein are low resistance metallization stack structures for logic and memory applications and related methods of fabrication. The methods involve forming bulk conductive films on thin low resistivity transition metal layers that have large grain size. The bulk conductive films follow the grains of the low resistivity transition metal films, resulting in large grain size. Also provided are devices including template layers and bulk films.

Abstract: Multi-pixel sensors such as camera sensors may be configured to capture two-dimensional and/or three-dimensional images of the interior of a process chamber or other fabrication tool. The sensors may be configured to capture pixelated electromagnetic radiation intensity information from within the interior of such process chamber before, during, and/or after processing of a substrate in the chamber. Such sensors may also be utilized for control, predictive, and/or diagnostic applications.

Abstract: Methods, systems, and computer programs are presented for determining the recipe for manufacturing a semiconductor with the use of machine learning (ML) to accelerate the definition of recipes. One general aspect includes a method that includes an operation for performing experiments for processing a component, each experiment controlled by a recipe, from a set of recipes, that identifies parameters for manufacturing equipment. The method further includes an operation for performing virtual simulations for processing the component, each simulation controlled by one recipe from the set of recipes. An ML model is obtained by training an ML algorithm using experiment results and virtual results from the virtual simulations. The method further includes operations for receiving specifications for a desired processing of the component, and creating, by the ML model, a new recipe for processing the component based on the specifications.

Abstract: A showerhead includes a plurality of plenums and a plurality of through holes positioned in the plurality of plenums. The plenums are stacked in a sequential order in an axial direction perpendicular to a semiconductor substrate. The plenums extend radially fully across the semiconductor substrate. The plenums are disjoint from each other and are configured to respectively supply a first metal precursor, a second metal precursor, and a reactant via the respective plenums without intermixing the first metal precursor, the second metal precursor, and the reactant in the plenums. The through holes of the respective plenums are arranged in a radial direction, which is perpendicular to the axial direction, in the same sequential order as the sequential order of the plenums. The through holes of the plenums open along a flat surface at a bottom of the showerhead. The flat surface extends radially fully across the bottom of the showerhead.

Abstract: Described herein are methods of filling features with tungsten, and related systems and apparatus, involving inhibition of tungsten nucleation. In some embodiments, the methods involve selective inhibition along a feature profile. Methods of selectively inhibiting tungsten nucleation can include exposing the feature to a direct or remote plasma. Pre-inhibition and post-inhibition treatments are used to modulate the inhibition effect, facilitating feature fill using inhibition across a wide process window. The methods described herein can be used to fill vertical features, such as in tungsten vias, and horizontal features, such as vertical NAND (VNAND) wordlines. The methods may be used for both conformal fill and bottom-up/inside-out fill. Examples of applications include logic and memory contact fill, DRAM buried wordline fill, vertically integrated memory gate and wordline fill, and 3-D integration using through-silicon vias.

Abstract: Various embodiments herein relate to systems and methods for predictive maintenance for semiconductor manufacturing equipment.

Abstract: Methods, systems, and computer programs are presented for determining the recipe for manufacturing a semiconductor with the use of machine learning (ML) to accelerate the definition of recipes. One general aspect includes a method that includes an operation for performing experiments for processing a component, each experiment controlled by a recipe, from a set of recipes, that identifies parameters for manufacturing equipment. The method further includes an operation for performing virtual simulations for processing the component, each simulation controlled by one recipe from the set of recipes. An ML model is obtained by training an ML algorithm using experiment results and virtual results from the virtual simulations. The method further includes operations for receiving specifications for a desired processing of the component, and creating, by the ML model, a new recipe for processing the component based on the specifications.

Abstract: A method includes arranging a substrate in a processing chamber, and exposing the substrate to a gas mixture including a first metal precursor gas and a second metal precursor gas to deposit a first metal precursor and a second metal precursor onto the substrate at the same time. The method further includes purging the processing chamber, supplying a reactant common to both the first metal precursor and the second metal precursor to form a layer of an alloy on the substrate, and purging the processing chamber.

Abstract: Provided herein are methods and apparatuses for forming metal films such as tungsten (W) and molybdenum (Mo) films on semiconductor substrates. The methods involve forming a reducing agent layer, then exposing the reducing agent layer to a metal precursor to convert the reducing agent layer to a layer of the metal. In some embodiments, the reducing agent layer is a silicon- (Si-) and boron- (B-) containing layer. The methods may involve forming the reducing agent layer at a first substrate temperature, raising the substrate temperature to a second substrate temperature, and then exposing the reducing agent layer to 10 the metal precursor at the second substrate temperature. The methods may be used to form fluorine-free tungsten or molybdenum films in certain embodiments. Apparatuses to perform the methods are also provided.

Abstract: Methods, systems, and computer programs are presented for predicting the performance of semiconductor manufacturing equipment operations. One method includes an operation for obtaining machine-learning (ML) models, each model related to predicting a performance metric for an operation of a semiconductor manufacturing tool. Further, each ML model utilizes features defining inputs for the ML model. The method further includes an operation for receiving a process definition for manufacturing a product with the semiconductor manufacturing tool. One or more ML models are utilized to estimate a performance of the process definition used in the semiconductor manufacturing tool. Additionally, the method includes presenting, on a display, results showing the estimate of the performance of the manufacturing of the product.

Abstract: Described herein are methods of filling features with tungsten, and related systems and apparatus, involving inhibition of tungsten nucleation. In some embodiments, the methods involve selective inhibition along a feature profile. Methods of selectively inhibiting tungsten nucleation can include exposing the feature to a direct or remote plasma. In certain embodiments, the substrate can be biased during selective inhibition. Process parameters including bias power, exposure time, plasma power, process pressure and plasma chemistry can be used to tune the inhibition profile. The methods described herein can be used to fill vertical features, such as in tungsten vias, and horizontal features, such as vertical NAND (VNAND) wordlines. The methods may be used for both conformal fill and bottom-up/inside-out fill. Examples of applications include logic and memory contact fill, DRAM buried wordline fill, vertically integrated memory gate/wordline fill, and 3-D integration using through-silicon vias.

Abstract: Methods, systems, and computer programs are presented for determining the recipe for manufacturing a semiconductor with the use of machine learning (ML) to accelerate the definition of recipes. One general aspect includes a method that includes an operation for performing experiments for processing a component, each experiment controlled by a recipe, from a set of recipes, that identifies parameters for manufacturing equipment. The method further includes an operation for performing virtual simulations for processing the component, each simulation controlled by one recipe from the set of recipes. An ML model is obtained by training an ML algorithm using experiment results and virtual results from the virtual simulations. The method further includes operations for receiving specifications for a desired processing of the component, and creating, by the ML model, a new recipe for processing the component based on the specifications.

Abstract: Provided herein are low resistance metallization stack structures for logic and memory applications and related methods of fabrication. The methods involve forming bulk conductive films on thin low resistivity transition metal layers that have large grain size. The bulk conductive films follow the grains of the low resistivity transition metal films, resulting in large grain size. Also provided are devices including template layers and bulk films.

Abstract: Various showerheads and methods are provided. A showerhead may include a faceplate partially defined by a front surface and a back surface, a back plate having a gas inlet, a first conical frustum surface, and a second conical frustum surface, a plenum volume fluidically connected to the gas inlet and at least partially defined by the gas inlet, the back surface of the faceplate, the first conical frustum surface, and the second conical frustum surface, and a baffle plate positioned within the plenum volume, and having a plurality of baffle plate through-holes extending through the baffle plate. The second conical frustum surface may be positioned radially outwards from the first conical frustum surface with respect to a center axis of the showerhead, and the second conical frustum surface may be positioned along the center axis farther from the gas inlet than the first conical frustum surface.

Abstract: Defects on a substrate comprising electronic components can be classified with a computational defect analysis system that may be implemented in multiple stages. For example, a first stage classification engine may process metrology data to produce an initial classification of defects. A second stage classification engine may use the initial classification, along with manufacturing information and/or prior defect knowledge to output probabilities that the defects are caused by one or more potential sources.

Abstract: Provided herein are methods and apparatuses for reducing line bending when depositing a metal such as tungsten, molybdenum, ruthenium, or cobalt into features on substrates by periodically exposing the feature to nitrogen, oxygen, or ammonia during atomic layer deposition, chemical vapor deposition, or sequential chemical vapor deposition to reduce interactions between metal deposited onto sidewalls of a feature. Methods are suitable for deposition into V-shaped features.

Abstract: Provided herein are methods of depositing tungsten (W) films without depositing a nucleation layer. In certain embodiments, the methods involve depositing a conformal reducing agent layer of boron (B) and/or silicon (Si) on a substrate. The substrate generally includes a feature to be filled with tungsten with the reducing agent layer conformal to the topography of the substrate including the feature. The reducing agent layer is then exposed to a fluorine-containing tungsten precursor, which is reduced by the reducing agent layer to form a layer of elemental tungsten. The conformal reducing agent layer is converted to a conformal tungsten layer.

Abstract: Described herein are methods of filling features with tungsten, and related systems and apparatus, involving inhibition of tungsten nucleation. In some embodiments, the methods involve selective inhibition along a feature profile. Methods of selectively inhibiting tungsten nucleation can include exposing the feature to a direct or remote plasma. In certain embodiments, the substrate can be biased during selective inhibition. Process parameters including bias power, exposure time, plasma power, process pressure and plasma chemistry can be used to tune the inhibition profile. The methods described herein can be used to fill vertical features, such as in tungsten vias, and horizontal features, such as vertical NAND (VNAND) wordlines. The methods may be used for both conformal fill and bottom-up/inside-out fill. Examples of applications include logic and memory contact fill, DRAM buried wordline fill, vertically integrated memory gate/wordline fill, and 3-D integration using through-silicon vias.

Abstract: Provided herein are low resistance metallization stack structures for logic and memory applications and related methods of fabrication. In some implementations, the methods involve providing a tungsten (W)-containing layer on a substrate; and depositing a molybdenum (Mo)-containing layer on the W-containing layer. In some implementations, the methods involve depositing a Mo-containing layer directly on a dielectric or titanium nitride (TiN) substrate without an intervening W-containing layer.