Antifragile system for semiconductor processing equipment using multiple specialized sensors and algorithms

Abstract

[0003] Embodiments disclosed herein include processing tools and methods of using the processing tools. In one embodiment, the processing tool includes a chamber and a cartridge for flowing one or more process gases from multiple gas sources into the chamber. In one embodiment, the processing tool further includes a mass flow controller for each of the multiple gas sources and a mass flow meter between the gas source and the cartridge. In one embodiment, the processing tool further includes a first pressure gauge between the mass flow meter and the cartridge, a second pressure gauge fluidly coupled to the chamber, and an exhaust line coupled to the chamber.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Non-provisional Application No. 17 / 019,061, filed September 11, 2020, the entire contents of which are incorporated herein by reference.

[0002] FIELD OF THE DISCLOSURE Embodiments of the present disclosure relate to the field of semiconductor processing, and more particularly to a processing chamber having a control loop sensor and a witness sensor. [Background technology]

[0003] Semiconductor wafer processing becomes more complex as feature sizes in semiconductor devices continue to shrink. A given process can include many different processing parameters (i.e., knobs) that can be individually controlled to provide a desired result on the wafer. For example, the desired result on the wafer can refer to the feature profile, layer thickness, layer chemical composition, etc. As the number of knobs increases, the theoretical process space available for process tuning and optimization becomes much larger.

[0004] Furthermore, once a final process recipe is developed, chamber drift can occur as the process is repeated multiple times on different wafers, causing changes in the results on the wafer. Chamber drift can result from erosion of consumable parts of the chamber, degradation of components (sensors, lamps, etc.), deposition of by-product films on surfaces, etc. Therefore, even after an extensive recipe development process, further tuning is necessary. Summary of the Invention

[0005] The embodiments disclosed herein include a processing tool and a method of using the processing tool. In one embodiment, the processing tool comprises a chamber and a cartridge for flowing one or more processing gases from a plurality of gas sources into the chamber. In one embodiment, the processing tool further comprises a mass flow controller for each of the plurality of gas sources and a mass flow meter between the gas source and the cartridge. In one embodiment, the processing tool further comprises a first pressure gauge between the mass flow meter and the cartridge, a second pressure gauge fluidly coupled to the chamber, and an exhaust line coupled to the chamber.

[0006] In one embodiment, the processing tool comprises a physics tool, a virtual sensor module, and a data model. In one embodiment, the physics tool comprises a control loop sensor and a witness sensor. In one embodiment, the virtual sensor module receives control loop sensor data and witness sensor data as inputs, and the virtual sensor module outputs virtual sensor data. In one embodiment, the data model comprises a statistical model and a physics model. In one embodiment, the virtual sensor data is provided to the data model, and the data model is configured to provide a control effort to the physics tool based at least in part on the virtual sensor data.

[0007] In one embodiment, a method for determining chamber drift is provided. In one embodiment, the method includes providing hardware inputs and process parameter inputs to a physical chamber and a data model, and collecting a witness sensor output from the physical chamber. In one embodiment, the method further includes generating a virtual witness sensor output from the data model, and comparing the witness sensor output to the virtual witness sensor output. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a block diagram of a semiconductor processing tool utilizing a set of control loop sensors according to one embodiment of the present disclosure. [Figure 2A] FIG. 1 is a block diagram of a semiconductor processing tool utilizing a set of control loop sensors and a set of witness sensors according to one embodiment of the present disclosure. [Figure 2B] FIG. 1 is a block diagram of a semiconductor processing tool utilizing a set of control loop sensors and a set of witness sensors including an array of temperature sensors according to one embodiment of the present disclosure. [Diagram 3] FIG. 1 illustrates a perspective view of a physics chamber of a semiconductor processing tool according to one embodiment of the disclosure. [Figure 4A] FIG. 1 is a block diagram of a semiconductor processing tool including a data model and a witness sensor according to an embodiment of the present disclosure. [Figure 4B] 1 is a flowchart of a method of using control loop sensors and witness sensors to provide feedback control of a processing tool, according to one embodiment. [Figure 4C] 1 is a schematic diagram of a processing tool having a witness sensor according to one embodiment. [Figure 4D] FIG. 1 is a schematic diagram of a processing tool with a witness sensor and a mass flow controller controlling a first pressure upstream of a cartridge, according to one embodiment. [Figure 5A] 1 is a flowchart of a method of using a data model to detect chamber drift in a semiconductor processing tool, according to an embodiment of the present disclosure. [Figure 5B] 1 is a flowchart of a method for utilizing virtual sensor data and metrology data to verify the accuracy of a data model in a semiconductor processing tool, according to one embodiment of the present disclosure. [Figure 6] FIG. 1 is a block diagram of an exemplary computer system according to one embodiment of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] A process chamber with a control loop sensor, a witness sensor, and a data model are described. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It will be apparent to one of ordinary skill in the art that the embodiments of the present disclosure may be practiced without these specific details. In other instances, well-known aspects, such as integrated circuit fabrication, are not described in detail in order not to unnecessarily obscure the embodiments of the present disclosure. Furthermore, it should be understood that the various embodiments illustrated in the figures are illustrative representations and are not necessarily drawn to scale.

[0010] To provide context to the embodiments of the present disclosure, a semiconductor processing tool 100 is shown in FIG. 1. The semiconductor processing tool 100 can include a chamber 105. A gas supply system can supply process gases into the chamber 105. For example, gases (e.g., Gas 1 through Gas n) can be flowed into an input line 111 that feeds a gas cartridge 110 that regulates the flow of gas into the chamber 105. Control of the gas flow is defined by a mass flow controller (MFC) 103. The pressure in the chamber 105 is measured by a pressure gauge 117. Conditions in the chamber 105 are controlled by adjusting the MFC 103 to target a desired pressure that is read by the pressure gauge 117. That is, the MFC 103 and the pressure gauge 117 can be considered control loop sensors, since the state of the MFC 103 and the pressure gauge 117 define the process conditions in the chamber 105. However, control loop sensors are not infallible. Control loop sensors often drift (i.e., become miscalibrated). As such, the processing tool 100 is also prone to drift. Drift of the processing tool 100 results in non-uniform wafer results from wafer to wafer. To avoid such non-uniformity, the processing tool 100 requires periodic qualification and tuning performed by a process engineer or other user.

[0011] In addition to being prone to drift, control loop sensors do not provide all of the necessary data that can be used to optimize process results. For example, in radical oxidation processes (sometimes called "radox processes" for short), it is desirable to know parameters such as the rate of gas exiting the cartridge, the composition of the gas exiting the cartridge, and the temperature in the chamber.

[0012] Accordingly, embodiments disclosed herein utilize a semiconductor processing tool that further includes a set of witness sensors. The witness sensors are outside of the control loop. As such, the witness sensors can be used to monitor the control loop sensors. If the control loop sensors drift, a change in the witness sensor output can be identified to alert a process engineer to the drift condition, even though the control loop sensors do not indicate a change in process conditions.

[0013] Additionally, data from the witness sensors can be leveraged to provide virtual sensor data. The virtual sensor data can include calculations of various in-chamber process conditions that cannot be directly measured by physical sensors. The virtual sensor data can be obtained from calculations using the output of the witness sensors and / or control loop sensors. For example, in a Radox process, readings from the witness sensors (e.g., mass flow meters (MFMs) and pressure sensors upstream of the cartridge) and the chamber pressure sensor can be used as inputs to a Bernoulli equation to calculate the flow rate across the inlet orifice of the cartridge.

[0014] Additionally, embodiments disclosed herein may include comparing the witness sensor output to a virtual witness sensor output calculated by a data model of the semiconductor processing tool. The data model may include a statistical model, a physical model of the physical processing tool, or a combination of statistical and physical models. The data model represents a virtual twin of the physical processing tool. That is, for a given set of inputs to the physical processing tool and data model, the output of the physical processing tool should match the output of the data model.

[0015] Thus, if one or more virtual witness sensor outputs differ from one or more witness sensor outputs, it can be determined that the physical chamber has drifted. In such a case, the witness sensor outputs can be fed back into the data model as a training data set to provide an updated data model. The updated data model can then be queried to generate a revised set of process inputs to return the processing chamber to the desired process window.

[0016] In some embodiments where the virtual witness sensor data substantially matches the witness sensor data, further investigations may be used to reconfirm the uniformity between the data model and the physical processing tool. For example, if the virtual witness sensor data matches the witness sensor data, additional calculations of the data model may also be considered correct. For example, additional virtual metrology calculated by the data model, such as, but not limited to, film deposition rate, film composition, etc., may be compared to physical metrology data obtained from the physical wafer. If the physical metrology data matches the virtual metrology data, the data model is confirmed. If the physical metrology data differs from the virtual metrology data, the physical metrology data may be fed back to the data model as a learning data set to provide an updated data model. The updated data model may then be queried to generate a revised set of process inputs to return the processing chamber to the desired process window.

[0017] The use of semiconductor processing tools utilizing witness sensors and / or data models provides several advantages in a manufacturing environment, such as reducing the number of cases where a processing tool becomes miscalibrated and misprocesses a wafer. Additionally, unnecessary downtime due to recalibration and tuning is avoided. Additionally, embodiments may be used to more efficiently design new processes for processing tools.

[0018] 2A, a block diagram of a semiconductor processing tool 200 is shown in accordance with one embodiment. In the illustrated embodiment, the processing tool 200 is described as a radical oxidation tool. However, it should be understood that the embodiments disclosed herein may be suitable for use in other types of processing tools, such as, but not limited to, a plasma etch chamber or a deposition chamber.

[0019] In one embodiment, the processing tool 200 includes a chamber 205. The chamber 205 may be a chamber suitable for providing a sub-atmospheric pressure at which a substrate (e.g., a semiconductor wafer) is processed. In one embodiment, the chamber 205 may be sized to accommodate a single substrate or multiple substrates. Semiconductor substrates suitable for processing in the chamber 205 may include silicon substrates, or any other semiconductor substrate. Other substrates, such as glass substrates, may also be processed in the chamber 205.

[0020] In one embodiment, a gas supply network supplies gas to the cartridge 210 from one or more gas sources (e.g., gas 1, gas 2, gas n, etc.). In certain embodiments, the gas sources can include one or more of oxygen, hydrogen, and nitrogen. Although three gas sources are shown in FIG. 2A, it is understood that an embodiment can include one or more gas sources. The cartridge 210 can include an inlet for receiving gas from line 211 and an outlet for distributing gas into the chamber 210. In the illustrated embodiment, the cartridge 210 is shown as supplying gas into the chamber from the side of the chamber 210. However, it is understood that the cartridge 210 can optionally supply gas into the chamber from above or below the chamber. In some embodiments, the cartridge 210 can also be referred to as a showerhead, especially when the processing tool is a plasma generating tool.

[0021] In one embodiment, the flow of each of the process gases may be controlled by a separate MFC 203. In one embodiment, the MFC 203 may be part of a control loop sensor group. The MFC 203 controls the flow of gas into the input line 211. In one embodiment, a mass flow meter (MFM) 212 is provided upstream of the cartridge 210. The MFM 212 allows the actual flow from the source gas to be measured. Also included upstream of the cartridge 210 is a pressure gauge 213. The pressure gauge 213 allows the pressure in the input line 211 to be measured. The MFM 212 and pressure gauge 213 are outside the control loop and may be considered witness sensors.

[0022] In one embodiment, a chamber pressure gauge 217 may be provided to measure the pressure in the chamber 205. The chamber pressure gauge 217 may be part of a control loop sensor group. In one embodiment, an additional witness sensor is provided along the exhaust line 215 of the processing tool 200. The additional sensors may include a leak detection sensor 216 and additional pressure gauges 218, 219. The leak detection sensor 216 may include a self-contained plasma optical emission spectroscopy (OES) device to measure oxygen leaking into the chamber 205. The pressure gauges 218 and 219 may be provided upstream and downstream of the throttle valve 214, respectively.

[0023] In one embodiment, pressure gauges 213, 217, 218, and 219 can have operating ranges appropriate for typical pressures provided at the locations within the processing tool in which they are located. For example, pressure gauge 213 can operate at a higher pressure range than the pressure ranges of the other pressure gauges 217, 218, 219. Similarly, pressure gauge 218 can operate at a higher pressure range than the pressure range of pressure gauge 219. In certain embodiments, pressure gauge 217 can operate at a range including 1,000T, pressure gauge 217 can operate at a range including 20T, pressure gauge 218 can operate at a range including 100T, and pressure gauge 219 can operate at a range including 10T.

[0024] In one embodiment, the witness sensors (e.g., 212, 213, 216, 218, 219) may be used to provide monitoring of chamber drift. For example, the control loop sensors (e.g., 203 and 217) may become miscalibrated during use of the processing tool 200. As such, the readings of the control loop sensors 203, 217 may remain constant while the results on the wafer (e.g., the deposition rate of the film) change. In such a case, the output of the witness sensor will change, indicating that the chamber has drifted.

[0025] In additional embodiments, witness sensors may be leveraged to implement virtual sensors within the chamber 205. A virtual sensor may refer to a sensor that provides a computationally generated output rather than a direct reading of a physical value (as is the case with physical sensors). Thus, virtual sensors have a powerful ability to determine conditions within the process tool 200 that are difficult or impossible to measure with traditional physical sensors.

[0026] In one embodiment, the virtual sensor can be used to determine the flow rate of the process gas at the outlet of the cartridge 210. Calculating the flow rate at the cartridge 210 is a valuable metric that can be used to control the deposition rate and / or deposition uniformity of the film on the wafer. In a particular embodiment, the flow rate at the cartridge 210 can be calculated using the Bernoulli equation, where the variables are supplied using the outputs of the MFM 212, the pressure gauge 213, the pressure gauge 217, and the known geometry of the cartridge 210. Although an example of the flow rate at the cartridge is provided, it should be understood that other unknowns in the processing tool 200 can be determined using the calculation of the virtual sensor. For example, unknowns such as, but not limited to, the gas composition at various locations in the chamber, the deposition rate across the wafer, the pressure across the wafer, and the film composition across the wafer can be determined using the implementation of the virtual sensor.

[0027] 2B, a diagram of a processing tool 200 according to an additional embodiment is shown. In the illustrated embodiment, the processing tool 200 of FIG. 2B may be substantially similar to the processing tool 200 of FIG. 2A, except that an array of temperature sensors 207 is provided in the chamber 205. For example, the temperature sensors 207 may be thermocouples, etc. In one embodiment, the temperature sensors 207 may be provided on a reflector plate (not shown) of the chamber. The temperature sensors 207 may be considered witness sensors in some embodiments. That is, the temperature sensors 207 may be outside of the control loop.

[0028] The temperature sensor 207 can provide an additional known variable to allow for the implementation of a wider range of virtual sensors. In one embodiment, the temperature sensor 207 can also be used to determine when a steady state has been reached within the chamber 205. This is particularly useful when starting up the processing tool 200 from a cold state, such as ramping up the processing tool 200 after maintenance. For example, the output of the temperature sensor 207 can be monitored in conjunction with one or more pressure gauges 213, 217, 218, 219, or the throttle valve angle, and the chamber can be ready for use when the steady state of the various sensors is reached. In one embodiment, monitoring the chamber reaching a steady state is useful because it eliminates wafer scrap or rework that is typically experienced due to the first wafer effect in processing tools.

[0029] 3, a perspective view of a processing tool 300 is shown in accordance with one embodiment. The processing tool 300 is an illustration of a radical oxidation tool 300. However, it should be understood that other types of processing tools may be used in accordance with embodiments disclosed herein.

[0030] In one embodiment, the processing tool 300 includes a chamber 305 into which a wafer 301 can be placed. The wafer 305 can be inserted into the chamber 305 through a slit valve 309 or the like. As shown, a cartridge 310 or other gas distribution plate is provided adjacent the slit valve 309 to allow gas to flow into the chamber 305 from the side. In one embodiment, an exhaust port 315 is provided at the end of the chamber 305 opposite the cartridge 310. However, it should be understood that more than one exhaust port 315 may be used and the location of the exhaust port may be other locations within the chamber 305.

[0031] In one embodiment, the semiconductor processing tool described herein may include a physical processing tool (such as the processing tools described above) and a digital twin of the physical processing tool. The digital twin may include a data model developed using machine learning. Ideally, when a set of process inputs (e.g., hardware parameters and / or process parameters) are provided to the physical processing tool and the data model, the results on the physical wafer match the results on the virtual wafer. Such a processing tool is beneficial because it can determine chamber drift. Furthermore, if chamber drift is detected, the data model can be updated and queried to determine a new set of process inputs to bring subsequent processing results back to the desired process window.

[0032] 4A, a schematic diagram of a processing tool 400 is shown according to one embodiment. As shown, a data model server 420 may be integrated with the processing tool 400. For example, the data model server 420 may be communicatively coupled to a front-end server 460 by a network connection, as indicated by the arrows. However, in other embodiments, the data model server 420 may be external to the processing tool 400. For example, the data model server 420 may be communicatively coupled to the processing tool 400 via an external network, or the like.

[0033] In one embodiment, the data model server 420 may comprise one or both of the physical model 427 and the statistical model 425. The statistical model 425 and the physical model 427 may be communicatively coupled to a database 430 for storing input data (e.g., sensor data, model data, metrology data, etc.) used to build and / or update the statistical model 425 and the physical model 427. In one embodiment, the statistical model 425 may be generated by performing a physical design of experiments (DoE) and may use interpolation to provide an extended process space model. In one embodiment, the physical model 427 may be generated using real-world physical and chemical relationships. For example, physical and chemical equations for various interactions in a process chamber may be used to build the physical model. In one embodiment, the processing tool 400 may comprise a front-end server 460, a tool control server 450, and tool hardware 440. The front-end server 460 may comprise a user interface 465 of the data model server 420. The user interface 465 provides an interface for process engineers to utilize data modeling to perform various operations such as recipe drift monitoring, as described in more detail below.

[0034] The tool control server 450 can include a smart monitoring and control block 455. The smart monitoring and control block 455 can include modules for providing diagnostics and other monitoring of the processing tool 400. The modules can include, but are not limited to, health checks, sensor drift, fault recovery, and leak detection. The smart monitoring and control block 455 can receive data as input from various sensors implemented in the tool hardware. The sensors can include standard sensors 447 that are typically present in a semiconductor manufacturing tool 400 to enable operation of the tool 400. For example, the sensors 447 can include control loop sensors, as described above. The sensors can also include witness sensors 445 that are added to the tool 400. The witness sensors 445 provide additional information necessary to build highly detailed data models. For example, the witness sensors can include physical sensors and / or virtual sensors. As described above, virtual sensors can utilize data obtained from two or more physical sensors and use calculations to provide additional sensor data not available from physical sensors alone. In a particular example, the virtual sensors may utilize upstream and downstream pressure sensors to calculate the flow rate through a portion of the process tool, such as a gas cartridge. In general, the witness sensors may include any type of sensor, such as, but not limited to, pressure sensors, temperature sensors, and gas concentration sensors. In one embodiment, the smart monitoring and control block 455 may provide data used by the data model server 420. In other embodiments, output data from the various witness sensors 445 may be provided directly to the data model server 420.

[0035] 4B, a schematic diagram is shown illustrating how various sensors are used to provide feedback control to a tool 400, according to one embodiment. As shown, the tool 400 can include one or more control loop sensors 466 and one or more witness sensors 467. Sensor data from the one or more control loop sensors 466 and one or more witness sensors 467 can be provided to a virtual sensor module 468. The virtual sensor module 468 can perform calculations on the sensor data to provide virtual sensor data 469. For example, the virtual sensor module 468 can perform a Bernoulli equation for a Venturi-type sensor to calculate the exact process gas flow at one or more locations in the chamber (e.g., the cartridge exit and / or exhaust into the chamber). Knowing the process gas flow can provide better control over the uniformity on the wafer, as described in more detail below.

[0036] In one embodiment, the virtual sensor data 469 may then be provided to the data model server 420. However, in some embodiments, the virtual sensor module 468 may be implemented as part of the data model server 420. The data model server 420 may utilize the process gas flow as a best known flow that can be used to better control, predict, detect drift, find faults, etc. These multiple variables (i.e., sensor data output and virtual data output) may be used as input and training for a data model, which is more effective than using a single point sensor. In one embodiment, the data model server 420 may output a control effort 463 to modify one or more process parameters of the tool 400 to correct for the drift and return the tool 400 to the desired process window.

[0037] In an embodiment, such feedback control actions may be performed on many different types of virtual sensor data. In certain embodiments, as described above, process gas flows may be calculated and used to control the output of the tool. However, the embodiments are not limited to process gas flows. For example, virtual sensors may be utilized to provide gas composition at various locations in the chamber, wafer temperature uniformity, or other process variables that cannot be directly (or easily) measured with physical sensors. In general, the overall effect of using virtual sensors to inform a data model and provide feedback control of the tool is to greatly improve control of the process. Furthermore, providing additional data that physical sensors alone cannot provide allows for optimization of the data learning rate of an intelligent system such as described herein. The multiplicity of inputs, cross-checks from many directions, and expectation of learning in the chamber-specific digital twin leads to a more antifragile system.

[0038] 4C and 4D, a schematic diagram of a processing tool 400 configured to control process gas velocity at the inlet of the chamber 405 (i.e., the outlet of the cartridge 410) is shown, according to one embodiment. In one embodiment, the process gas velocity can be calculated (as a virtual sensor output) using sensor data from the MFM 412, pressure sensor 413A, and pressure sensor 413B as inputs to the Bernoulli equation. In general, the faster the velocity, the further the frame front is from the cartridge 410, and the slower the velocity, the closer the frame front is to the cartridge 410. As one skilled in the art can recognize, the position of the frame front affects the process uniformity. The position of the frame front that provides the optimal process uniformity for a given chamber can be calculated using the data model server 420. That is, the data model server 420 can calculate the optimal pressure delta for a desired process uniformity profile.

[0039] Control of the pressure delta can be implemented by any suitable control of the tool 400. For example, in FIG. 4C, the pressure can be adjusted by varying the total flow rate from the gas panel 408. The total flow rate from the gas panel 408 can be controlled by changes to mass flow controllers (not shown) that adjust the flow of gases out of the gas panel 408. In one embodiment, the mass flow controllers are controlled so that the ratio of process gases (e.g., the ratio of hydrogen gas to oxygen gas) remains constant. In an additional embodiment, a back pressure controller or mass flow controller 404 can be provided at the inlet of the cartridge 410, as shown in FIG. 4D. The mass flow controller 404 can be used to bleed off excess flow and servo the cartridge pressure to obtain the required delta pressure between the cartridge 410 and the chamber 405.

[0040] 5A, a flowchart 570 of a process for utilizing data model 520 in combination with physical chamber 500 to detect chamber drift is provided, according to one embodiment. In one embodiment, data model 520 is calibrated (e.g., through machine learning) such that data model 520 is presumed to function as a digital twin of physical chamber 500. That is, if there is no drift in physical chamber 500, the output of data model 520 is expected to match the output of physical chamber 500.

[0041] In one embodiment, the flowchart 570 can begin with providing inputs 571 to the physical chamber 500 and the data model 520. The inputs 571 can include hardware inputs 572 and / or process parameter inputs 575. Examples of hardware inputs 572 include, but are not limited to, cartridge pore sizes, other cartridge geometries, and chamber geometries. For example, a cartridge having a different geometry can be swapped out and plugged into the chamber 500 to change the physical chamber hardware inputs 572. Examples of process parameter inputs 575 include, but are not limited to, pressure, process gas flow rates, and temperature.

[0042] At block 574, one or more outputs of the physical chamber are compared to one or more outputs of the data model 520. The compared outputs may include sensor data (e.g., control loop sensor data, witness sensor data, and / or virtual sensor data). When the outputs of the data model 520 (e.g., virtual control loop sensor data, virtual witness sensor data, and virtual sensor data) substantially match the outputs of the chamber 500, decision branch 576 is taken and it is determined that the chamber 500 is drift-free.

[0043] If one or more of the outputs of the data model 520 differ from one or more of the outputs of the chamber 500, it can be determined that the chamber 500 has drifted. If chamber drift is detected, branch 577 is selected. In branch 577, one or more chamber outputs from the chamber physics 500 are fed back to the data model 520 as a training data set to provide an updated data model 520. By incorporating the training data set, the updated data model can more accurately model the drifted chamber 500. After the updated data model 520 is generated, the updated data model 520 can be queried for new inputs to bring the drifted chamber physics 500 back to the desired process window. The new inputs are fed back to the input block 571, as shown by branch 578.

[0044] The process illustrated in FIG. 5A may then be repeated any desired number of times. In one embodiment, comparison block 574 may be performed after every wafer is processed. Because the comparison does not require the use of metrology, comparison block 574 may be performed more frequently. In other embodiments, comparison block 574 may be performed at regular intervals or at any desired frequency. For example, comparison block 574 may be performed more frequently during process ramp-up as compared to after the chamber 500 has reached a substantially steady state.

[0045] 5B, a flowchart 580 of a process for performing a secondary check to verify the accuracy of the data model 520 is shown, according to one embodiment. In FIG. 5B, the data model 520 has been verified to be drift-free, for example, using a method as described in FIG. 5A. That is, the virtual witness sensor output substantially matches the physical chamber witness sensor data for a given set of inputs 571. Even if the physical and virtual witness sensor data substantially match, the wafer results may differ. The method of flowchart 580 allows for a comparison of the physical and virtual wafer results.

[0046] If the virtual witness sensor output matches the witness sensor data, then additional calculations of the data model 520 can also be considered correct. The additional calculations can include, but are not limited to, virtual metrology 581, such as film deposition rate, film composition, or any other property typically determined using metrology testing. To validate the data model 520, the virtual metrology 581 calculated by the data model 520 can be compared to physical metrology 582 obtained from a physical wafer processed in the physical chamber 500.

[0047] In block 583, the virtual metrology 581 is compared to the physical metrology 582. If the physical metrology data 582 matches the virtual metrology data 581, branch 585 is taken and the data model 520 is verified in block 584. If the physical metrology data 582 differs from the virtual metrology data 581, branch 586 is taken and the physical metrology data 582 is fed back to the data model 520 as a training data set to provide an updated data model. The updated data model can then be queried to generate a revised set of process inputs to return the processing chamber 500 to the desired process window. The new inputs are fed back to the input block 571, as shown at branch 587.

[0048] In one embodiment, the process of Flowchart 580 may be performed any desired frequency. However, because physical metrology is a time and resource intensive process, an embodiment may perform the process of Flowchart 580 less frequently than the process of Flowchart 570. For example, the process of Flowchart 580 may be performed once per lot, once per planned maintenance, or when a steady state for chamber 500 is deemed to have been reached.

[0049] FIG. 6 illustrates a diagrammatic representation of a machine in the exemplary form of a computer system 600 upon which a set of instructions for causing the machine to perform any one or more of the methodologies described herein may be executed. In alternative embodiments, the machine may be connected (e.g., networked) to other machines in a local area network (LAN), an intranet, an extranet, or the Internet. The machine may operate as a server or a client machine in a client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine may be a personal computer (PC), a tablet PC, a set-top box (STB), a personal digital assistant (PDA), a mobile phone, a web appliance, a server, a network router, a switch, or a bridge, or any machine capable of executing a set of instructions (sequential or otherwise) that specify actions that the machine should take. Furthermore, although only a single machine is illustrated, the term "machine" shall be construed to include any collection of machines (e.g., computers) that individually or jointly execute a set (or sets) of instructions for performing any one or more of the methodologies described herein.

[0050] The exemplary computer system 600 includes a processor 602, a main memory 604 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM (RDRAM), etc.), a static memory 606 (e.g., flash memory, static random access memory (SRAM), MRAM, etc.), and a secondary memory 618 (e.g., a data storage device), which communicate with each other via a bus 630.

[0051] Processor 602 represents one or more general-purpose processing devices, such as a microprocessor, a central processing unit, etc. More specifically, processor 602 may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor implementing other instruction sets, or a processor implementing a combination of instruction sets. Processor 602 may also be one or more special-purpose processing devices, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. Processor 602 is configured to execute processing logic 626 to perform the operations described herein.

[0052] Computer system 600 may further include a network interface device 608. Computer system 600 may also include a video display device 610 (e.g., a liquid crystal display (LCD), a light emitting diode display (LED), or a cathode ray tube (CRT)), an alphanumeric input device 612 (e.g., a keyboard), a cursor control device 614 (e.g., a mouse), and a signal generation device 616 (e.g., a speaker).

[0053] The secondary memory 618 may include a machine-accessible storage medium (or more specifically, a computer-readable storage medium) 632 having stored thereon one or more sets of instructions (e.g., software 622) embodying any one or more of the methods or functions described herein. The software 622 may also reside, completely or at least partially, within the main memory 604 and / or within the processor 602 during its execution by the computer system 600, with the main memory 604 and the processor 602 also constituting machine-readable storage media. The software 622 may further be transmitted or received over a network 620 via the network interface device 608.

[0054] While machine-accessible storage medium 632 is shown in the exemplary embodiment to be a single medium, the term "machine-readable storage medium" shall be interpreted to include a single medium or multiple media (e.g., a centralized or distributed database and / or associated caches and servers) that store one or more sets of instructions. The term "machine-readable storage medium" shall also be interpreted to include any medium that can store or encode a set of instructions that can be executed by a machine to cause the machine to perform any one or more of the methods of the present disclosure. The term "machine-readable storage medium" shall therefore be interpreted to include, but is not limited to, solid-state memory, and optical and magnetic media.

[0055] According to one embodiment of the present disclosure, a machine-accessible storage medium has stored thereon instructions that cause a data processing system to perform a method for processing wafers using insights from a data model and / or a method for updating or building the data model.

[0056] Thus, a method for using a data model to process a wafer in a processing tool has been disclosed.

Claims

1. A processing tool, comprising: Chamber, a cartridge for flowing one or more process gases from a plurality of gas sources into the chamber, the cartridge receiving one or more process gases from the plurality of gas sources via input lines; a mass flow controller for each of said plurality of gas sources; a mass flow meter between the gas source and the cartridge; a first pressure gauge between the mass flow meter and the cartridge, the first pressure gauge being configured to measure a pressure in the input line upstream of the cartridge; a second pressure gauge fluidly coupled to the chamber; an exhaust line coupled to the chamber; and a virtual sensor module that receives as inputs control loop sensor data from the control loop sensors and witness sensor data from the witness sensors, and outputs virtual sensor data; Equipped with the control loop sensor comprises the mass flow controller and the second pressure gauge; the witness sensor comprises the mass flow meter and the first pressure gauge; Processing tools.

2. The processing tool of claim 1 , further comprising an array of temperature sensors on a reflector in the chamber.

3. a throttle valve in the exhaust line; a third pressure gauge in the exhaust line between the throttle valve and the chamber; and a fourth pressure gauge in the exhaust line on an opposite side of the throttle valve from the third pressure gauge; The processing tool of claim 1 , further comprising:

4. 4. The processing tool of claim 3, wherein the first pressure gauge is optimized to detect pressures in a first pressure range, the third pressure gauge is optimized to detect pressures in a second pressure range, and the fourth pressure gauge is optimized to detect pressures in a third pressure range, the first pressure range having a maximum pressure that is greater than a maximum pressure of the second pressure range, and the maximum pressure of the second pressure range is greater than a maximum pressure of the third pressure range.

5. The processing tool of claim 1 , wherein the processing tool performs a radical oxidation process.

6. The processing tool of claim 5 , wherein the multiple gas sources comprise an oxygen source gas and a hydrogen source gas.

7. The processing tool of claim 1 , wherein the cartridge injects gas into the chamber from a side of the chamber.

8. The processing tool of claim 1 , wherein the cartridge and the exhaust line are at opposite ends of the chamber.

Images