Abstract: Implementations disclosed describe systems and techniques to detect anomalies in a manufacturing operation. The techniques include generating, using a plurality of outlier detection models, a plurality of outlier scores. The outlier scores are representative of a degree of presence, in a plurality of sensor statistics, of an anomaly associated with the manufacturing operation. Individual outlier scores are generated using a respective one of the plurality of outlier detection models. The techniques further include determining, using the outlier scores, a likelihood of the anomaly associated with the manufacturing operation.

Abstract: Implementations disclosed describe a method and a system to perform the method of obtaining a reduced representation of a plurality of sensor statistics representative of data collected by a plurality of sensors associated with a device manufacturing system performing a manufacturing operation. The method further includes generating, using a plurality of outlier detection models, a plurality of outlier scores, each of the plurality of outlier scores generated based on the reduced representation of the plurality of sensor statistics using a respective one of the plurality of outlier detection models. The method further includes processing the plurality of outlier scores using a detector neural network to generate an anomaly score indicative of a likelihood of an anomaly associated with the manufacturing operation.

Abstract: Methods and systems for automated monitoring of sensors associated with a tool used in a manufacturing process are described. Sensor health may be evaluated based on the sensor's actual responses to set point changes. Rather than interrupt operation of the tool to determine whether one or more sensors are behaving in a predictable manner over an applicable range of operating conditions, pairs of set point values and sensor responses may be collected during the manufacturing process and stored in a time-indexed manner. A virtual model may be created for each sensor using selected ones of the indexed pairs and represented as points in an orthogonal coordinate system to identify a predictable operating region corresponding to the operating range. If a sensor response to a set point change, reflected in a stored, time indexed pair, is non-linear or offset relative to the predictable operating region, an alarm may be generated.

Abstract: Implementations disclosed describe a method and a system to perform the method of obtaining a reduced representation of a plurality of sensor statistics representative of data collected by a plurality of sensors associated with a device manufacturing system performing a manufacturing operation. The method further includes generating, using a plurality of outlier detection models, a plurality of outlier scores, each of the plurality of outlier scores generated based on the reduced representation of the plurality of sensor statistics using a respective one of the plurality of outlier detection models. The method further includes processing the plurality of outlier scores using a detector neural network to generate an anomaly score indicative of a likelihood of an anomaly associated with the manufacturing operation.

Abstract: Embodiments provide techniques for compressing sensor data collected within a manufacturing environment. One embodiment monitors a plurality of runs of a recipe for fabricating one or more semiconductor devices within a manufacturing environment to collect runtime data from a plurality of sensors within the manufacturing environment. The collected runtime data is compressed by generating, for each of the plurality of sensors and for each of the plurality of runs, a respective representation of the corresponding runtime data that describes a shape of the corresponding runtime data and a magnitude of the corresponding runtime data. A query specifying one or more runtime data attributes is received and executed against the compressed runtime data to generate query results, by comparing the one or more runtime data attributes to at least one of the generated representations of runtime data.

Abstract: Embodiments of the present disclosure provide a method, system, and computer program product for monitoring a service life of a chamber component. In one example, the method includes receiving one or more power measurements of a semiconductor processing chamber from one or more sensors positioned about the semiconductor processing chamber. The processor compares the one or more power measurements to one or more threshold values corresponding to the service life of the chamber component. The processor determines whether the one or more power measurements exceed the threshold values. If the processor determines that the one or more power measurements exceed the threshold values, the processor takes remedial measures for the service life of the chamber component.

Abstract: Embodiments of the disclosure relate to methods and a system for adjusting the chucking voltage of an electrostatic chuck. In one embodiment, a system for plasma processing a substrate includes a plasma processing chamber, a radio-frequency (RF) matching circuit coupled to the chamber, a sensor and a controller. The chamber includes a chamber body having an inner volume, a bipolar electrostatic chuck disposed in the inner volume and a power supply configured to provide chucking voltage to a pair of electrodes embedded within the electrostatic chuck. When plasma is energized within the chamber by the application of RF power through an RF matching circuit, the sensor is configured to detect a change in an electrical characteristic at the RF matching circuit. The controller is coupled to the power supply and configured to adjust the chucking voltage in response to the change in the electrical characteristic detected by the sensor.

Abstract: Embodiments presented herein provide techniques for predicting the topography of a product produced from a manufacturing process. One embodiment includes generating a plurality of prediction models. Each of the plurality of prediction models corresponds to a respective one of a plurality of positional coordinates of a product produced from a manufacturing process. The method also includes receiving a set of user-specified input parameters to apply to the manufacturing control process. The method further includes generating a graphical representation of a topography map for the product for the user-specified of input parameters based on the plurality of prediction models.

Abstract: Embodiments of the present disclosure provide a method, system, and computer program product for monitoring a service life of a chamber component. In one example, the method includes receiving one or more power measurements of a semiconductor processing chamber from one or more sensors positioned about the semiconductor processing chamber. The processor compares the one or more power measurements to one or more threshold values corresponding to the service life of the chamber component. The processor determines whether the one or more power measurements exceed the threshold values. If the processor determines that the one or more power measurements exceed the threshold values, the processor takes remedial measures for the service life of the chamber component.

Abstract: Embodiments of the present disclosure provide a method, system, and computer program product for monitoring a service life of a chamber component. In one example, the method includes receiving one or more power measurements of a semiconductor processing chamber from one or more sensors positioned about the semiconductor processing chamber. The processor compares the one or more power measurements to one or more threshold values corresponding to the service life of the chamber component. The processor determines whether the one or more power measurements exceed the threshold values. If the processor determines that the one or more power measurements exceed the threshold values, the processor takes remedial measures for the service life of the chamber component.

Abstract: Embodiments of the disclosure relate to methods and a system for adjusting the chucking voltage of an electrostatic chuck. In one embodiment, a system for plasma processing a substrate includes a plasma processing chamber, a radio-frequency (RF) matching circuit coupled to the chamber, a sensor and a controller. The chamber includes a chamber body having an inner volume, a bipolar electrostatic chuck disposed in the inner volume and a power supply configured to provide chucking voltage to a pair of electrodes embedded within the electrostatic chuck. When plasma is energized within the chamber by the application of RF power through an RF matching circuit, the sensor is configured to detect a change in an electrical characteristic at the RF matching circuit. The controller is coupled to the power supply and configured to adjust the chucking voltage in response to the change in the electrical characteristic detected by the sensor.

Abstract: Embodiments of the present disclosure provide a method, system, and computer program product for monitoring a service life of a chamber component. In one example, the method includes receiving one or more power measurements of a semiconductor processing chamber from one or more sensors positioned about the semiconductor processing chamber. The processor compares the one or more power measurements to one or more threshold values corresponding to the service life of the chamber component. The processor determines whether the one or more power measurements exceed the threshold values. If the processor determines that the one or more power measurements exceed the threshold values, the processor takes remedial measures for the service life of the chamber component.

Abstract: Embodiments provide techniques for compressing sensor data collected within a manufacturing environment. One embodiment monitors a plurality of runs of a recipe for fabricating one or more semiconductor devices within a manufacturing environment to collect runtime data from a plurality of sensors within the manufacturing environment. The collected runtime data is compressed by generating, for each of the plurality of sensors and for each of the plurality of runs, a respective representation of the corresponding runtime data that describes a shape of the corresponding runtime data and a magnitude of the corresponding runtime data. A query specifying one or more runtime data attributes is received and executed against the compressed runtime data to generate query results, by comparing the one or more runtime data attributes to at least one of the generated representations of runtime data.

Abstract: Embodiments presented herein provide techniques for predicting the topography of a product produced from a manufacturing process. One embodiment includes generating a plurality of prediction models. Each of the plurality of prediction models corresponds to a respective one of a plurality of positional coordinates of a product produced from a manufacturing process. The method also includes receiving a set of user-specified input parameters to apply to the manufacturing control process. The method further includes generating a graphical representation of a topography map for the product for the user-specified of input parameters based on the plurality of prediction models.

Abstract: Methods and systems for automated monitoring of sensors associated with a tool used in a manufacturing process are described. Sensor health may be evaluated based on the sensor's actual responses to set point changes. Rather than interrupt operation of the tool to determine whether one or more sensors are behaving in a predictable manner over an applicable range of operating conditions, pairs of set point values and sensor responses may be collected during the manufacturing process and stored in a time-indexed manner. A virtual model may be created for each sensor using selected ones of the indexed pairs and represented as points in an orthogonal coordinate system to identify a predictable operating region corresponding to the operating range. If a sensor response to a set point change, reflected in a stored, time indexed pair, is non-linear or offset relative to the predictable operating region, an alarm may be generated.

Abstract: Methods and apparatus for semiconductor manufacturing process monitoring and control are provided herein. In some embodiments, apparatus for substrate processing may include a process chamber for processing a substrate in an inner volume of the process chamber; a radiation source disposed outside of the process chamber to provide radiation at a frequency of about 200 GHz to about 2 THz into the inner volume via a dielectric window in a wall of the vacuum process chamber; a detector to detect the signal after having passed through the inner volume; and a controller coupled to the detector and configured to determine the composition of species within the inner volume based upon the detected signal.

Abstract: Described herein are methods for fabricating dual-damascene interconnect structures. In one embodiment, the interconnect structures are fabricated with a dual-damascene method having trenches then vias formed. The method includes novel liner depositions after the trench and via etches. The method includes etching trenches in a dielectric layer. Next, the method includes depositing a first liner layer on the dielectric layer. Next, the method includes etching vias in the dielectric layer and an etch stop layer. Next, the method includes depositing a second liner layer on the first liner layer. The second liner layer is deposited on the exposed surfaces of the first liner layer, dielectric layer, etch stop layer, and the first metal layer. Then, a second metal layer is deposited on the second liner layer.

Abstract: Methods for etching silicon-based antireflective layers are provided herein. In some embodiments, a method of etching a silicon-based antireflective layer may include providing to a process chamber a substrate having a multiple-layer resist thereon, the multiple-layer resist comprising a patterned photoresist layer defining features to be etched into the substrate disposed above a silicon-based antireflective coating; and etching the silicon-based antireflective layer through the patterned photoresist layer using a plasma formed from a process gas having a primary reactive agent comprising a chlorine-containing gas. In some embodiments, the chlorine-containing gas is chlorine (Cl2).

Abstract: A method, a system and a computer readable medium for integrated in-vacuo repair of low-k dielectric thin films damaged by etch and/or strip processing. A repair chamber is integrated onto a same platform as a plasma etch and/or strip chamber to repair a low-k dielectric thin film without breaking vacuum between the damage event and the repair event. UV radiation may be provided on the integrated etch/repair platform in any combination of before, after, or during the low-k repair treatment to increase efficacy of the repair treatment and/or stability of repair.

Abstract: A substrate comprising a resist layer overlying a dielectric feature, is processed in a substrate processing chamber comprising an antenna, and first and second process electrodes. A process gas comprising CO2 is introduced into the chamber. The process gas is energized to form a plasma by applying a source voltage to the antenna, and by applying to the electrodes, a first bias voltage having a first frequency of at least about 10 MHz and a second bias voltage having a second frequency of less than about 4 MHz. The ratio of the power level of the first bias voltage to the second bias voltage is sufficient to obtain an edge facet height of the underlying dielectric feature that is at least about 10% of the height of the dielectric feature.