Method and system for automatic frequency tuning of a radio frequency (RF) signal generator for multi-level RF power pulsing

Abstract

SOLUTION: A two-dimension frequency search grid is defined by a first coordinate axis representing an operating frequency set point of the RF signal generator in the first operating state and a second coordinate axis representing an operating frequency set point of the RF signal generator in the second operating state. The RF signal generator has a first output power level in the first operating state and a second output power level in the second operating state. The RF signal generator operates in the multi-level RF power pulsing mode by cycling between the first operational state and the second operational state. An automatic search process is performed within the two-dimension frequency search grid to simultaneously determine an optimal value of the operating frequency set point of the RF signal generator in the first operating state and an optimal value of the operating frequency set point of the RF signal generator in the second operating state.SELECTED DRAWING: Figure 11

Description

[Technical Field]

[0001] [Technical field]

[0002] The present disclosure relates to semiconductor device manufacturing. [Background technology]

[0003] [Related Technology]

[0004] In the fabrication of semiconductor devices (integrated circuits, memory cells, etc.), a series of manufacturing operations are performed to define features on a semiconductor wafer (hereinafter "wafer"). The wafer comprises integrated circuit devices in the form of multi-layer structures defined on a silicon substrate. At the substrate level, transistor elements with diffusion regions are formed. At subsequent levels, interconnect metallization lines are patterned and electrically connected to the transistor elements to define the desired integrated circuit devices. Additionally, patterned conductive layers are insulated from other conductive layers by dielectric materials.

[0005] Many modern semiconductor chip manufacturing processes involve the generation of a plasma, in which ionic and / or radical components derived from the plasma are utilized to directly or indirectly affect changes on the surface of a substrate exposed to the plasma. For example, various plasma-based processes can be utilized to etch material from, deposit material onto, or modify material already present on the substrate surface. Plasmas are often generated by applying radio frequency (RF) power to a process gas in a controlled environment, such that the process gas is excited and transformed into the desired plasma. The characteristics of the plasma are affected by many process parameters, including, but not limited to, the material composition of the process gas, the flow rate of the process gas, the geometric characteristics of the plasma generation region and surrounding structures, the temperature of the process gas and surrounding materials, the frequency of the applied RF power, the magnitude of the applied RF power, and the temporal manner in which the RF power is applied. Therefore, it is important to understand, monitor, and / or control some of the process parameters that can affect the characteristics of the generated plasma, particularly with respect to the delivery of RF power to the plasma generation region. It is in this context that the present disclosure was born. Summary of the Invention

[0006] In one example embodiment, a method for automatic frequency adjustment of an RF signal generator for operation in a multilevel RF power pulsing mode is disclosed. The method includes defining a two-dimensional frequency search grid having a first coordinate axis representing an operating frequency setpoint of the RF signal generator in a first operating state and a second coordinate axis representing an operating frequency setpoint of the RF signal generator in a second operating state. The RF signal generator has a first output power level in the first operating state and a second output power level in the second operating state, the first and second output power levels being different from one another. The RF signal generator is programmed to operate in the multilevel RF power pulsing mode by periodically cycling between the first and second operating states. The method further includes performing an automatic search process within the two-dimensional frequency search grid to simultaneously determine a first optimal value for the operating frequency setpoint of the RF signal generator in the first operating state and a second optimal value for the operating frequency setpoint of the RF signal generator in the second operating state. The method further includes configuring the RF signal generator to operate using a first optimal value for the operating frequency setpoint of the RF signal generator in a first operating state and a second optimal value for the operating frequency setpoint of the RF signal generator in a second operating state.

[0007] In one example embodiment, a system for automatic frequency adjustment of an RF signal generator for operation in a multi-level RF power pulsing mode is disclosed. The system includes a plasma processing chamber having a substrate support structure and an electrode. The system further includes an RF signal generator configured to generate an RF signal and transmit the RF signal through an output of the RF signal generator. The RF signal generator is configured to have a first output power level in a first operating state and a second output power level in a second operating state, the first and second output power levels being different from each other. The RF signal generator is programmed to operate in the multi-level RF power pulsing mode by periodically cycling between the first operating state and the second operating state. The system further includes an impedance matching system having an input connected to the output of the RF signal generator. The impedance matching system has an output connected to the electrode. The impedance matching system is configured to control the impedance at the output of the RF signal generator to enable transmission of the RF signal through the electrode to a plasma generated in the plasma processing chamber. The system further includes a control system programmed to define a two-dimensional frequency search grid having a first coordinate axis representing an operating frequency setpoint of the RF signal generator in a first operating state and a second coordinate axis representing an operating frequency setpoint of the RF signal generator in a second operating state, and the control system is programmed to perform an automatic search process within the two-dimensional frequency search grid to simultaneously determine a first optimum value for the operating frequency setpoint of the RF signal generator in the first operating state and a second optimum value for the operating frequency setpoint of the RF signal generator in the second operating state.

[0008] Other aspects and advantages of the present invention will become apparent from the following detailed description, taken in conjunction with the accompanying drawings, which illustrate the invention. [Brief explanation of the drawings]

[0009] [Figure 1A] 1 is a vertical cross-sectional view of a plasma processing system utilized in the fabrication of semiconductor wafers, in accordance with some embodiments.

[0010] [Figure 1B] 1B is a top view of the plasma processing system of FIG. 1A, according to some embodiments.

[0011] [Figure 1C] FIG. 1 illustrates a control module according to some embodiments.

[0012] [Figure 2] 1 illustrates an RF signal generator system including a primary RF signal generator and a bias RF signal generator in accordance with some embodiments.

[0013] [Figure 3] 1 is an operational diagram illustrating a multi-level RF power pulsing process in which bias RF power is pulsed while maintaining a substantially constant primary RF power in accordance with some embodiments.

[0014] [Figure 4A] FIG. 4 illustrates the reflection coefficient (ΓB1S1) at the output of the bias RF signal generator during the first operating state (S1) in a 2D frequency setpoint grid defined by the first operating frequency setpoint (fB1S1) and the second operating frequency preset point (fB1S2) for the multi-level bias RF power pulsing process example of FIG. 3, in accordance with some embodiments.

[0015] [Figure 4B] FIG. 4 illustrates the reflection coefficient (Γ) at the output of the bias RF signal generator during the second operating state (S2) in a 2D frequency setpoint grid defined by the first operating frequency setpoint (f) and the second operating frequency preset point (f), for the multi-level bias RF power pulsing process example of FIG. 3, in accordance with some embodiments.

[0016] [Figure 5A]FIG. 4 illustrates the voltage (VB1S1) at the output of the bias RF signal generator during the first operating state (S1) in a 2D frequency set point grid defined by the first operating frequency set point (fB1S1) and the second operating frequency preset point (fB1S2) for the multi-level bias RF power pulsing process example of FIG. 3 with a voltage set point of 1000 V for the first operating state (S1), according to some embodiments.

[0017] [Figure 5B] FIG. 4 illustrates the voltage (VB1S2) at the output of the bias RF signal generator during the second operating state (S2) in a 2D frequency set point grid defined by the first operating frequency set point (fB1S1) and the second operating frequency preset point (fB1S2) for the multi-level bias RF power pulsing process example of FIG. 3 with a voltage set point of 200 V for the second operating state (S2), according to some embodiments.

[0018] [Figure 6A] 1 illustrates a 2D frequency search grid with an initial equilateral triangle defined by vertices V1, V2, and V3 located in the lower right quadrant of the 2D frequency search grid, according to some embodiments.

[0019] [Figure 6B] 10A-10C illustrate a 2D frequency search grid depicting a moving vertex V3 moving through the centroid of a current equilateral triangle to form a new vertex V4 of a new equilateral triangle, according to some embodiments.

[0020] [Figure 6C] 10A-10C illustrate a 2D frequency search grid depicting a moving vertex V2 moving through the centroid of a current equilateral triangle to form a new vertex V5 of a new equilateral triangle, according to some embodiments.

[0021] [Figure 6D] FIG. 6C illustrates how the triangulation search algorithm proceeds from FIG. 6C to the point where moving vertex V10 returns to its previous position corresponding to vertex V8, according to some embodiments.

[0022] [Figure 6E] FIG. 10 illustrates the reduction in size of an equilateral triangle after vertex V10 returns to its previous position corresponding to vertex V8, according to some embodiments.

[0023] [Figure 7A] FIG. 10 shows a plot of the cost function (J) evaluated for each new vertex of a defined equilateral triangle over iterations of a triangulation search algorithm to illustrate the minimum cost function (J) convergence criteria for when the cost function (J) can no longer be appreciably minimized and does not oscillate, according to some embodiments.

[0024] [Figure 7B] FIG. 10 shows a plot of the side lengths of equilateral triangles over iterations of a triangulation search algorithm to illustrate the minimum cost function (J) convergence criterion for when the side lengths of equilateral triangles can no longer be appreciably reduced, according to some embodiments.

[0025] [Figure 7C] FIG. 10 shows a plot of the frequency coordinates of a moving vertex in a 2D frequency search grid over iterations of a triangulation search algorithm to illustrate the minimum cost function (J) convergence criterion for when the frequency coordinates of a moving vertex in the 2D frequency search grid can no longer change appreciably and do not oscillate, according to some embodiments.

[0026] [Figure 8] FIG. 6B illustrates isocontours of a cost function (J) evaluated over a 2D frequency search grid and the steps of the triangulation search algorithm of FIGS. 6A-6E required to complete to arrive at an optimal set of setpoint frequency coordinates {FB1S1,FB1S2}, according to some embodiments.

[0027] [Figure 9]1 is an operational diagram illustrating a multi-level RF power pulsing process in which both bias RF power and primary RF power are pulsed synchronously in accordance with some embodiments.

[0028] [Figure 10] 1 is an operational diagram illustrating a multi-level RF power pulsing process in which both the bias RF power and the primary RF power are pulsed asynchronously in accordance with some embodiments.

[0029] [Figure 11] 1 is a flow chart illustrating a method for automatic frequency adjustment of an RF signal generator for operation in a multi-level RF power pulsing mode in accordance with some embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0030] In the following description, numerous specific details are set forth to facilitate understanding of embodiments of the present disclosure. However, it will be apparent to one skilled in the art that embodiments of the present disclosure may be practiced without some or all of these specific details. Additionally, detailed descriptions of well-known processing operations are omitted to avoid unnecessarily obscuring the present disclosure.

[0031] Disclosed herein are embodiments of an automated method and corresponding system that enables semiconductor manufacturing plasma processing recipe steps involving multi-level radio frequency (RF) power pulsing to be frequency tuned for optimal RF stability and performance. The embodiments disclosed herein may be implemented in connection with essentially any semiconductor manufacturing plasma processing tool equipped to provide multi-level pulsing of RF power to an electrode and / or antenna (coil), such as conductor etch semiconductor manufacturing plasma processing tools, among others. Disclosed herein is an automatic frequency tuner that simultaneously optimizes and considers the following quantities, where RF State 1 is a high-power operating state of the RF generator and RF State 2 is a low-power operating state of the RF generator, or vice versa: Reflected RF power for RF state 1, Reflected RF power for RF state 2, Deviation of applied RF voltage / power from setpoint RF voltage / power for RF State 1, and Deviation of applied RF voltage / power from preset set point RF voltage / power for RF State 2.

[0032] The optimization of the above four quantities corresponds to a collective minimization of the above four quantities. The automatic frequency tuner includes an optimizer that uses a search triangulation process to determine predetermined frequency adjustment points on a constrained two-dimensional (2D) frequency search grid that correspond to the collective minimization of the above four quantities through evaluation of a cost function, where a first dimension of the 2D frequency search grid is the setpoint frequency of the RF generator operating in RF State 1 and a second dimension of the 2D frequency search grid is the preset setpoint frequency of the RF generator operating in RF State 2. The result of the triangulation search process is an optimal set of setpoint frequency coordinates within the 2D frequency search grid, the set including: 1) an optimal setpoint frequency for the RF generator operating in RF State 1, and 2) an optimal preset setpoint frequency for the RF generator operating in RF State 2. Using data log parameters available from a user interface (UI) of the plasma processing system, the automatic frequency tuner performs the triangulation search process on the 2D frequency search grid and returns an optimal set of frequency coordinates that satisfy stable RF generator operating conditions.

[0033] In the semiconductor industry, semiconductor substrates may undergo processing operations in various types of plasma chambers, such as capacitively coupled plasma (CCP) and inductively coupled plasma (ICP) processing chambers. In both CCP and ICP processing chambers, radio frequency (RF) power is utilized to excite a process gas within a plasma processing region to convert the process gas into a plasma to which the substrate is exposed. Reactive and / or charged species within the plasma interact with the substrate to alter the substrate's condition, for example, by modifying materials present on the substrate, depositing material onto the substrate, or removing / etching material from the substrate. CCP and ICP processing chambers may include one or more electrodes that receive RF power to generate a plasma within the plasma processing region. CCP and ICP processing chambers may also include one or more electrodes that receive RF power to generate a bias voltage at the substrate (at the wafer level) to attract charged species from the plasma toward the substrate.

[0034] FIG. 1A is a vertical cross-sectional view of a plasma processing system 100 utilized in semiconductor wafer fabrication, according to some embodiments. FIG. 1B is a top view of the plasma processing system of FIG. 1A, according to some embodiments. The vertical cross-sectional view of FIG. 1A corresponds to view AA in FIG. 1B. The plasma processing system 100 of FIG. 1A is an example of an ICP-type plasma processing system. It should be understood that the automatic frequency adjustment method and corresponding system disclosed herein can be performed and implemented in both ICP- and CCP-type plasma processing systems, as well as other types of plasma processing systems in which an RF signal generator is operated to periodically pulse RF power between different power levels and the RF signal generator has different optimal operating frequency set points at different power levels. However, for ease of explanation, the automatic frequency adjustment method and corresponding system are described herein with respect to an example ICP-type plasma processing system, as shown in FIGS. 1A and 1B. An ICP processing chamber may also be referred to as a transformer-coupled plasma (TCP) processing chamber. For ease of discussion herein, the term ICP processing chamber is used to refer to both ICP and TCP processing chambers. It should be understood that plasma processing system 100 represents essentially any type of ICP processing chamber in which an RF signal is transmitted from a coil 101 located outside of processing chamber 103 to a process gas within processing chamber 103 to generate a primary plasma 105 within a plasma processing volume 106 of processing chamber 103, and the primary plasma 105 is used to affect a change in the condition of a substrate 107 held for exposure to components of the primary plasma 105. FIG. 1A shows coil 101 from which an RF signal is transmitted into plasma processing volume 106 to generate primary plasma 105 within plasma processing volume 106 for exposing substrate 107. Coil 101 is also referred to as a primary electrode.

[0035] In some embodiments, substrate 107 is a semiconductor wafer undergoing a manufacturing procedure. However, it should be understood that in various embodiments, substrate 107 can be essentially any type of substrate that undergoes a plasma-based manufacturing process. For example, in some embodiments, the term substrate 107 as used herein can refer to a substrate formed of sapphire, GaN, GaAs, or SiC, or other substrate materials, and can include glass panels / substrates, metal foils, metal sheets, polymeric materials, etc. Also, in various embodiments, substrate 107 as referred to herein can vary in form, shape, and / or size. For example, in some embodiments, substrate 107 as referred to herein can correspond to a 200 mm (millimeter) semiconductor wafer, a 300 mm semiconductor wafer, or a 450 mm semiconductor wafer. Also, in some embodiments, substrate 107 as referred to herein can correspond to a non-circular substrate, such as a rectangular substrate for a flat panel display, among other shapes.

[0036] The plasma processing space 106 of the processing chamber 103 is formed within a surrounding structure 109, below an upper window structure 111, and above a substrate support structure 113. In some embodiments, the surrounding structure 109 is formed of a conductive material (such as a metal) that is mechanically and chemically compatible with the environment and materials present in the plasma processing space 106 during operation of the plasma processing system 100. In these embodiments, the surrounding structure 109 may be electrically connected to a reference ground potential 115. The processing chamber 103 includes a door 151 through which a substrate 107 may be moved into and removed from the plasma processing space 106.

[0037] The substrate support structure 113 is configured to securely support the substrate 107 for exposure to the primary plasma 105 generated in the plasma processing space 106. In some embodiments, the substrate support structure 113 is an electrostatic chuck including one or more clamping electrodes 117 that can be powered by a clamping power supply 119 through an electrical connection 121. The power supplied to the one or more clamping electrodes 117 generates an electrostatic field for clamping the substrate 107 onto the substrate support structure 113. In various embodiments, the clamping power supply 119 can be configured to supply either RF power, direct current (DC) power, or a combination of both RF power and DC power to the one or more clamping electrodes 117. In embodiments in which the clamping power supply 119 is configured to supply RF power, the clamping power supply 119 further includes an impedance matching circuit through which the RF power is transmitted to prevent unacceptable reflection of the RF power from the one or more clamping electrodes 117. In these embodiments, the impedance matching circuit in the clamping power supply 119 includes an array of capacitors and / or inductors.

[0038] The substrate support structure 113 further provides a bias voltage (V bThe plasma processing space 106 may include a bias electrode 123 to which RF bias power can be supplied to generate a bias voltage (V) . The RF power transmitted from the bias electrode 123 to the plasma processing space 106 is referred to as bias RF power. In some embodiments, the bias RF power is generated by a bias RF signal generator 125, transmitted through an electrical connection 127 to an impedance matching system 129, and then transmitted from the impedance matching system 129 to the bias electrode 123 through a transmission rod 131. The transmission rod 131 is electrically isolated from the surrounding structure 109 of the processing chamber 103. The impedance matching system 129 includes an arrangement of bias capacitors and / or inductors configured to ensure that the impedance at the output of the bias RF signal generator 125 is sufficiently close to the load impedance with which the bias RF signal generator 125 is designed to operate (typically about 50 ohms), so that the RF signal generated and transmitted by the bias RF signal generator 125 is transmitted efficiently (i.e., without unacceptable reflections) to the plasma processing space 106.

[0039] The plasma processing system 100 operates by flowing one or more process gases from a process gas source 133 through an array of fluid transport structures 135 into a plasma processing volume 106 and applying RF power from the coil 101 to the one or more process gases to transform the one or more process gases into a primary plasma 105 that is exposed to a substrate 107 to affect a change in the material or surface condition on the substrate 107. Spent process gases and other materials resulting from processing of the substrate 107 are exhausted from the plasma processing volume 106 through one or more exhaust ports 147, as indicated by arrows 149.

[0040] The coil 101 is disposed above the upper window structure 111. In the example of FIGS. 1A and 1B, the coil 101 is formed as a radial coil assembly, with the shaded portion of the coil 101 winding toward the back of the page and the unshaded portion of the coil 101 winding toward the front of the page. It should be understood that the coil 101 of FIGS. 1A and 1B is provided as an example. In various embodiments, the coil 101 may include multiple sections, each spanning a specific corresponding radial range above the upper window structure 111. In these embodiments, the RF power supplied to each section of the coil 101 is independently controlled. It should also be understood that the number of turns (around the center of the upper window structure 111) of the example coil 101 of FIGS. 1A and 1B is provided as an example. In various embodiments, the coil 101 may have any number of turns and any cross-sectional size and shape (circular, elliptical, rectangular, trapezoidal, etc.) as needed to provide the necessary RF signal transmission through the upper window structure 111 to the plasma processing space 106. It should be understood that in various embodiments, the coil 101 may have essentially any configuration suitable for transmitting RF power through the upper window structure 111 to the plasma processing space 106.

[0041] The RF power transmitted from coil 101 to plasma processing space 106 is referred to as plasma primary RF power. The plasma primary RF power is generated by primary RF signal generator 137 and transmitted through electrical connection 139 to impedance matching system 141, and then transmitted to coil 101 through electrical connection 143. In some embodiments, return electrical connection 145 also extends from coil 101 to impedance matching system 141. Impedance matching system 141 includes an arrangement of bias capacitors and / or inductors configured to ensure that the impedance at the output of primary RF signal generator 137 is sufficiently close to the load impedance with which primary RF signal generator 137 is designed to operate (typically about 50 ohms) so that the RF signal provided by primary RF signal generator 137 to coil 101 is transmitted efficiently to plasma processing space 106 without unacceptable reflections.

[0042] The plasma processing system 100 further includes a control system 153 configured and connected to control the operation of the plasma processing system 100. The control system 153 is connected to the process gas source 133 via connection 155 to control the process gas source 133. The control system 153 is connected to the primary RF signal generator 137 via connection 157 to control the primary RF signal generator 137. The control system 153 is connected to the impedance matching system 141 via connection 159 to control the impedance matching system 141. The control system 153 is connected to the bias RF signal generator 125 via connection 161 to control the bias RF signal generator 125. The control system 153 is connected to the impedance matching system 129 via connection 163 to control the impedance matching system 129. The control system 153 is connected to the clamp power supply 119 via connection 165 to control the clamp power supply 119. It should be understood that in various embodiments, any of connections 155, 157, 159, 161, 163, and 165 may be wired, wireless, optical, or a combination thereof. It should be understood that in various embodiments, control system 153 is configured and connected to control essentially any aspect of plasma processing system 100 that is suitable for active control. It should also be understood that in various embodiments, control system 153 is connected to various measuring instruments and sensors and other data acquisition devices located throughout plasma processing system 100 to measure and monitor any parameters related to the operation of plasma processing system 100. It should also be understood that in various embodiments, the data / signal connections between control system 153 and each of the various measuring instruments and sensors and other data acquisition devices may be wired, wireless, optical, or a combination thereof.

[0043] 1C is a diagram illustrating control system 153 according to some embodiments. Control system 153 includes processor 181, storage hardware unit (HU) 183 (e.g., memory), input HU 171, output HU 175, input / output (I / O) interface 173, I / O interface 177, network interface controller (NIC) 179, and data communication bus 185. Processor 181, storage HU 183, input HU 171, output HU 175, I / O interface 173, I / O interface 177, and NIC 179 communicate data with each other via data communication bus 185. Examples of input HU 171 include a mouse, a keyboard, a stylus, a data acquisition system, a data acquisition card, etc. The input HU 171 is configured to receive data communications from multiple external devices, such as the process gas supply 133, the primary RF signal generator 137, the impedance matching system 141, the bias RF signal generator 125, the impedance matching system 129, the clamp power supply 119, and / or any other devices within the plasma processing system 100. Examples of the output HU 175 include a display, a speaker, a device controller, etc. The output HU 175 is configured to transmit data to multiple external devices, such as the process gas supply 133, the primary RF signal generator 137, the impedance matching system 141, the bias RF signal generator 125, the impedance matching system 129, the clamp power supply 119, and / or any other devices within the plasma processing system 100.

[0044] Examples of NIC 179 include a network interface card, a network adapter, etc. In various embodiments, NIC 179 is configured to operate according to one or more communication protocols and associated physical layers, such as Ethernet and / or EtherCAT, among others. Each of I / O interfaces 173 and 177 is defined to provide compatibility between different hardware units connected to the I / O interface. For example, I / O interface 173 may be defined to convert signals received from input HU 171 to a form, amplitude, and / or rate compatible with data communication bus 185. I / O interface 177 may also be defined to convert signals received from data communication bus 185 to a form, amplitude, and / or rate compatible with output HU 175. While various operations described herein are performed by processor 181 of control system 153, it should be understood that in some embodiments, various operations may be performed by multiple processors of control system 153 and / or by multiple processors of multiple computer systems in data communication with control system 153.

[0045] Also, in some embodiments, there is a user interface (UI) associated with the control system 153. The user interface may include a display (e.g., a display screen of equipment and / or processing conditions and / or a graphical software display) and user input devices such as a pointing device, keyboard, touch screen, microphone, etc. In some embodiments, the UI of the control system 153 may provide a UI for one or more of the process gas source 133, the primary RF signal generator 137, the impedance matching system 141, the bias RF signal generator 125, the impedance matching system 129, the clamp power supply 119, and / or any other devices in the plasma processing system 100. More specifically, any of the process gas source 133, the primary RF signal generator 137, the impedance matching system 141, the bias RF signal generator 125, the impedance matching system 129, the clamp power supply 119, and / or any other devices in the plasma processing system 100 may include integrated electronics for controlling their operation before, during, and after processing of the substrate 107. In some embodiments, the integrated electronics in each device in the plasma processing system 100 is connected to the control system 153 via the NIC 179 to allow a process engineer to control any device in the plasma processing system 100 using the control system 153.

[0046] The control system 153 may be configured to execute a computer program, such as a set of instructions, for controlling the operation of the process gas supply 133, the primary RF signal generator 137, the impedance matching system 141, the bias RF signal generator 125, the impedance matching system 129, the clamp power supply 119, and / or any other controllable device in the plasma processing system 100. The control system 153 is programmable to control any of the processes disclosed herein, such as the supply of process gases, temperature settings (e.g., heating and / or cooling), pressure settings, vacuum settings, power settings, RF generator settings, impedance matching system settings, frequency settings, flow rate settings, fluid supply settings, position and operation settings, and wafer movement into and out of the process chamber 103 and other wafer movement tools and / or load locks connected or coupled to the plasma processing system. Additionally, programs stored on a memory device associated with the control system 153 may be used in some embodiments. Software for directing the operation of the control system 153 may be designed or configured in many different ways. Computer programs for directing the operation of control system 153, and therefore plasma processing system 100, may be written in any conventional computer-readable programming language (e.g., assembly language, C, C++, Pascal, Fortran, etc.). Compiled object code or scripts are executed by processor 181 to perform the tasks specified in the programs.

[0047] The control system 153 is defined as electronic equipment having various integrated circuits, logic, memory, and / or software that receive instructions, issue instructions, and control operations. The integrated circuits may include chips in the form of firmware that store program instructions, digital signal processors (DSPs), chips defined as application-specific integrated circuits (ASICs), and / or one or more microprocessors or microcontrollers that execute the program instructions (e.g., software). The program instructions may be communicated to the control system 153 in the form of various individual settings (or program files) that define operational parameters for operating the plasma processing system 100 to perform a predetermined process on the substrate 107. The operational parameters, in some embodiments, may be part of a recipe defined by a process engineer to accomplish one or more process steps during processing of one or more layers, materials, metals, oxides, silicon, silicon dioxide, surfaces, circuits, and / or dies of the substrate 107.

[0048] In some embodiments, the control system 153 is implemented in the “cloud” or as part of a fab host computer system, thereby enabling remote access for controlling the plasma processing system 100. In some embodiments, the control system 153 provides remote access to the plasma processing system 100 to monitor the current progress of a manufacturing operation, examine the history of past manufacturing operations, examine trends or performance indicators from multiple manufacturing operations, adjust parameters associated with a current processing operation, configure processing steps to follow a current processing operation, or initiate a new processing operation. In some examples, access of the control system 153 through a remote computer (e.g., a server) can be utilized to provide process recipes to the plasma processing system 100 over a network (which may include a local network or the Internet). By using a remote computer, the control system 153 can provide a user interface that allows entry or programming of parameters and / or settings, which are then communicated from the remote computer to the plasma processing system 100. In some examples, the control system 153 receives instructions in the form of data, where the instructions specify parameters for each of the processing steps to be performed during one or more operations. It should be understood that the parameters may be specific to the type of process being performed as well as the type of tool the controller is configured to interface with or control. Accordingly, control system 153 may be distributed, such as by having one or more separate controllers that are networked and operate toward a common purpose (such as the process and control described herein). One example of a distributed control system 153 for such purposes is one or more integrated circuits within plasma processing system 100 that communicate with one or more remotely located integrated circuits (e.g., located at the platform level or remotely as part of a remote computer) that cooperate to control the execution of processes within plasma processing system 100.

[0049] 2 shows an RF signal generator system 200 including a primary RF signal generator 137 and a bias RF signal generator 125, according to some embodiments. The primary RF signal generator 137 is connected to provide an RF signal from an output 210 of the primary RF signal generator 137 through connection 139 to an input 217 of an impedance matching system 141, and from an output 223 of the impedance matching system 141 through connection 143 to the coil 101 of the plasma processing system 100. The bias RF signal generator 125 is connected to provide an RF signal from an output 271 of the bias RF signal generator 125 through connection 127 to an input 275 of the impedance matching system 129, and from an output 277 of the impedance matching system 129 through connection 131 to the bias electrode 123 of the plasma processing system 100.

[0050] The impedance matching system 141 comprises a combination of capacitors and inductors connected in an electrical circuit to match the impedance at the output 210 of the primary RF signal generator 137 to a design impedance (typically 50 ohms). The impedance matching system 129 also comprises a combination of capacitors and inductors connected in an electrical circuit to match the impedance at the output 271 of the bias RF signal generator 125 to a design impedance (typically 50 ohms). The impedance matching system 141 further comprises a NIC 239 that enables the impedance matching system 141 to send data to and receive data from systems external to the impedance matching system 141. The NIC 239 of the impedance matching system 141 is in data communication with the NIC 179 of the control system 153, as shown by connection 159. The impedance matching system 129 also comprises a network interface controller 239 that enables the impedance matching system 129 to send data to and receive data from systems external to the impedance matching system 129. NIC 279 of impedance matching system 129 is in data communication with NIC 179 of control system 153, as shown by connection 163. Examples of NICs 239 and 279 each include a network interface card, a network adapter, etc. In various embodiments, NICs 239 and 279 are configured to operate according to one or more network communication protocols and associated physical layers, such as Ethernet and / or EtherCAT, among others.

[0051] The primary RF signal generator 137 includes an oscillator 203 for generating an RF signal. The oscillator 203 is an electronic circuit that generates a periodic oscillating electrical signal (e.g., a sinusoidal electronic signal) having a specific frequency within the RF range. The oscillator 203 can oscillate at a set frequency, which in some embodiments is in the range of up to about 100 megahertz (MHz). As used herein with respect to numerical values ​​disclosed herein, the term "about" means within ±10% of the predetermined numerical value. In some embodiments, the oscillator 203 is configured to oscillate within a frequency range of about 11 MHz to about 15 MHz. In some embodiments, the oscillator 203 is configured to oscillate within a frequency range of about 12.882 MHz to about 14.238 MHz. The output of the oscillator 203 is connected to the input of the power amplifier 205. Power amplifier 205 amplifies the RF signal generated by oscillator 203 and transmits the amplified RF signal through the output of power amplifier 205 to output 210 of primary RF signal generator 137 .

[0052] The primary RF signal generator 137 further includes a control system 209 configured to provide control of all operational aspects of the primary RF signal generator 137. In some embodiments, the control system 209 is configured similarly to the control system 153 of the plasma processing system 100. For example, in some embodiments, the control system 209 includes a processor, a data storage device, an input / output interface, and a data bus over which the processor, the data storage device, and the input / output interface communicate data with each other. The control system 209 is connected to provide control of the oscillator 203, as shown by connection 204. The control system 209 is also connected to provide control of the power amplifier 205, as shown by connection 206. The primary RF signal generator 137 further includes a voltage / current (V / I) sensor 207 connected to an output 210 of the primary RF signal generator 137. The V / I sensor 207 is connected to the control system 209, as shown by connection 208. In this configuration, the V / I sensor 207 provides the control system 209 with a real-time measurement of the voltage and current present at the output 210 of the primary RF signal generator 137 .

[0053] Control system 209 further includes a NIC 211 that enables control system 209 to transmit data to and receive data from systems external to primary RF signal generator 137. NIC 211 of primary RF signal generator 137 is in data communication with NIC 179 of control system 153, as shown by connection 157. Examples of NIC 211 include network interface cards, network adapters, etc. In various embodiments, NIC 211 is configured to operate according to one or more network communication protocols and associated physical layers, such as Ethernet and / or EtherCAT, among others.

[0054] It should be understood that control system 209 is connected and configured to control essentially any aspect of primary RF signal generator 137. Furthermore, it should be understood that control system 209 may be connected and configured to monitor essentially any physical and / or electrical state, condition, and / or parameter at essentially any location within primary RF signal generator 137. Control system 209 is further configured to direct the operation of primary RF signal generator 137 according to a predetermined algorithm. For example, control system 209 is configured to operate primary RF signal generator 137 by executing input and control instructions / programs. The input and control instructions / programs include, among other parameters related to the operation and control of primary RF signal generator 137, a target RF power setpoint and a target frequency setpoint as a function of time.

[0055] The bias RF signal generator 125 includes an oscillator 277 for generating an RF signal. The oscillator 277 is an electronic circuit that generates a periodic oscillating electrical signal (e.g., a sinusoidal electronic signal) having a particular frequency in the RF range. The oscillator 277 can oscillate at a set frequency, which in some embodiments is in a range up to about 100 megahertz (MHz). In some embodiments, the oscillator 277 is configured to oscillate within a frequency range of about 11 MHz to about 15 MHz. In some embodiments, the oscillator 277 is configured to oscillate within a frequency range of about 12.882 MHz to about 14.238 MHz. In some embodiments, the oscillator 277 is configured to oscillate within a frequency range of about 0.9 MHz to about 1.1 MHz. The output of the oscillator 277 is connected to the input of the power amplifier 279. The power amplifier 279 amplifies the RF signal generated by the oscillator 277 and transmits the amplified RF signal through the output of the power amplifier 279 to the output 271 of the bias RF signal generator 125 .

[0056] The bias RF signal generator 125 further includes a control system 281 configured to provide control of all operational aspects of the bias RF signal generator 125. In some embodiments, the control system 281 is configured similarly to the control system 153 of the plasma processing system 100. For example, in some embodiments, the control system 281 includes a processor, a data storage device, an input / output interface, and a data bus over which the processor, the data storage device, and the input / output interface communicate data with each other. The control system 281 is connected to provide control of the oscillator 277, as shown by connection 278. The control system 281 is also connected to provide control of the power amplifier 279, as shown by connection 280. The bias RF signal generator 125 further includes a voltage / current (V / I) sensor 285 connected to the output 271 of the bias RF signal generator 125. The V / I sensor 285 is connected to the control system 281, as shown by connection 282. In this configuration, the V / I sensor 285 provides the control system 281 with a real-time measurement of the voltage and current present at the output 271 of the bias RF signal generator 125 .

[0057] Control system 281 further includes a NIC 283 that enables control system 281 to transmit data to and receive data from systems external to bias RF signal generator 125. NIC 283 of bias RF signal generator 125 is in data communication with NIC 179 of control system 153, as shown by connection 161. Examples of NIC 283 include network interface cards, network adapters, etc. In various embodiments, NIC 283 is configured to operate according to one or more network communication protocols and associated physical layers, such as Ethernet and / or EtherCAT, among others.

[0058] It should be understood that control system 281 is connected and configured to control essentially any aspect of bias RF signal generator 125. Furthermore, it should be understood that control system 281 can be connected and configured to monitor essentially any physical and / or electrical state, condition, and / or parameter at essentially any location within bias RF signal generator 125. Control system 281 is further configured to direct the operation of bias RF signal generator 125 according to a predetermined algorithm. For example, control system 281 is configured to operate bias RF signal generator 125 by executing input and control instructions / programs. The input and control instructions / programs include, among other parameters related to the operation and control of bias RF signal generator 125, a target RF power setpoint and a target frequency setpoint as a function of time.

[0059] In some embodiments, the control system 209 of the primary RF signal generator 137 is programmed to determine the real-time reflection coefficient (i.e., gamma (Γ)) at the output 210 of the primary RF signal generator 137, where Γ=V r / V f , V r is the complex amplitude of the reflected RF signal, and V f is the complex amplitude of the forward RF signal. In some embodiments, the control system 209 of the primary RF signal generator 137 is also programmed to determine a voltage standing wave ratio (VSWR) at the output 210 of the primary RF signal generator 137, where VSWR=|V max | / |V min |=((1+|Γ|)) / ((1-|Γ|)), |V max |=|V f |+|V r |, |V min |=|V f |-|V r|. Minimization of the reflected RF power associated with the RF signal generated by the primary RF signal generator 137 occurs when the reflection coefficient at output 210 of the primary RF signal generator 137 is as close to zero as possible. Also, minimization of the reflected RF power associated with the RF signal generated by the primary RF signal generator 137 occurs when the VSWR at output 210 of the primary RF signal generator 137 is as close to one as possible, where one is the smallest possible VSWR. In some embodiments, the control system 209 is programmed to calculate the real-time reflection coefficient and / or VSWR at output 210 of the primary RF signal generator 137 using a real-time measured voltage at output 210 of the primary RF signal generator 137. The real-time reflection coefficient and / or VSWR at output 210 of the primary RF signal generator 137, determined using voltage measurements taken at output 210 of the primary RF signal generator 137, may be utilized in a cost function to optimize the frequency setpoint of the primary RF signal generator 137.

[0060] Similarly, in some embodiments, control system 281 of bias RF signal generator 125 is programmed to determine the reflection coefficient (i.e., gamma (Γ)) and VSWR at output 271 of bias RF signal generator 125. Minimization of reflected RF power associated with the RF signal generated by bias RF signal generator 125 occurs when the reflection coefficient at output 271 of bias RF signal generator 125 is as close to zero as possible. Also, minimization of reflected RF power associated with the RF signal generated by bias RF signal generator 125 occurs when the VSWR at output 271 of bias RF signal generator 125 is as close to one as possible, where one is the smallest possible value for VSWR. In some embodiments, control system 281 is programmed to calculate the real-time reflection coefficient and / or VSWR at output 271 of bias RF signal generator 125 using a real-time measured voltage at output 271 of bias RF signal generator 125. The real-time reflection coefficient and / or VSWR at output 271 of bias RF signal generator 125, determined using voltage measurements taken at output 271 of bias RF signal generator 125, may be used as a feedback signal to minimize the reflection coefficient to as close to zero as possible and / or minimize the VSWR to as close to one as possible at output 271 of high frequency RF signal generator 202. The real-time reflection coefficient and / or VSWR at output 271 of high frequency RF signal generator 202, determined using voltage measurements taken within high frequency RF signal generator 202, may also be used in a cost function to optimize the frequency setpoint of bias RF signal generator 125.

[0061] The plasma processing system 100 offers several advantages for plasma process control in various plasma-based semiconductor processing applications (e.g., plasma etching, etc.). The plasma processing system 100 provides independent control of plasma density (ion flux / radical flux) and ion energy. Specifically, plasma density can be controlled to some extent by plasma primary RF power transmitted from the primary RF signal generator 137 to the coil 101 and through the upper window structure 111 into the plasma processing volume 106. And ion energy can be controlled from the bias RF signal generator 125 to the bias electrode 123 and into the plasma processing volume 106. Independent control of plasma density (which is directly related to ion flux and radical flux) and ion energy is particularly useful in some semiconductor processing applications. For example, this is useful in patterning applications where high plasma density is required to achieve a desired etch rate and low ion energy is required to reduce damage to one or more materials (e.g., photoresist material) present on the substrate. It should be understood that in addition to patterning applications, many other plasma-based semiconductor processing applications can also benefit from independent control of plasma density and ion energy. For example, independent control of plasma density and ion energy can be useful in high aspect ratio (HAR) etching applications, where high ion energy and directionality are required to maintain the etch front at the bottom of deep trenches and / or holes. In some embodiments, the plasma primary RF power / frequency and bias RF power / frequency may need to be controlled simultaneously and differently to achieve the desired results. For example, in some embodiments, to obtain high plasma density with low ion energy, the plasma primary RF power needs to be high, and at the same time, the bias RF power needs to be low.

[0062] In some situations, the plasma primary RF power transmitted from the coil 101 through the upper dielectric window 111 to the plasma processing space 106 does not provide sufficient density at the substrate 107 level to obtain the required etch rate and / or etch selectivity. One reason for this is that the density of the primary plasma 105 generated by the plasma primary RF power transmitted from the coil 101 decreases with increasing distance from the coil 101. Therefore, as the distance between the coil 101 and the substrate support structure 113 increases, it becomes more difficult to achieve the required plasma density at the substrate 107 level. Also, the lower frequency bias RF power applied to the bias electrode 123 may increase the DC bias voltage (V) on the substrate 107 without significantly contributing to the plasma density near the substrate 107. b ). Furthermore, it may not be possible to simply increase the plasma primary RF power supplied to the coil 101 beyond a specified maximum amount (e.g., about 3 kW (kilowatts)) due to potential damage caused by overheating of the upper window structure 111. Also, shortening the distance between the coil 101 and the substrate support structure 113 may require costly redesign of the process chamber 103, potentially causing problems with plasma uniformity at the substrate 107 level, and present other challenges.

[0063] It is possible to provide an increase in plasma density at the substrate 107 level without causing an increase in ion energy at the substrate 107 level. The bias electrode 123 can be utilized to deliver a specially controlled RF signal to the plasma processing space 106 to generate a supplemental plasma density 154 (see FIG. 1A) locally at the substrate 107 level. At the substrate 107 level, the effective plasma density is the sum of the plasma density generated by the plasma primary RF power supplied to the coil 101 and the plasma density generated by the bias RF power supplied to the bias electrode 123. In various embodiments, it is possible to generate a supplemental plasma density 154 locally at the substrate 107 with or without increasing the ion energy at the substrate 107 level. The bias RF power applied by the bias RF signal generator 125 at the substrate 107 level is controlled to generate a supplemental plasma density 154 at the substrate 107 level (i.e., directly above the substrate). Generally, the bias voltage (V b ) is inversely proportional to the frequency (f) of these RF signals (V b ∝1 / f). Bias RF power (P b ) is the bias voltage (V b ) and bias current (I b ) product, i.e. (P b =V b *I b ), so the bias voltage (V b ) is lower, the same bias RF power (P b ), the bias current (I b ) must be correspondingly high. Therefore, the predetermined bias RF power (P b ) to achieve higher plasma density, a lower bias voltage (V b ) and a correspondingly higher bias current (I b ) and the bias voltage (V b ) is inversely proportional to the frequency (f) of the bias RF signal, so the predetermined bias RF power (P b) for lower bias voltage (V b ), the frequency (f) of the bias RF signal can be increased. Thus, to obtain an increase in the replenishment plasma density 154 generated at the substrate 107 level, the bias voltage (V b ) low, a higher frequency (f) RF signal can be supplied to the bias electrode 123. Then, to obtain an increase in the replenishment plasma density 154 generated at the substrate 107 level, the bias voltage (V b ) high, a lower frequency (f) RF signal can be supplied to the bias electrode 123.

[0064] In some embodiments, the control system 153 is programmed to direct operation of the primary RF signal generator 137 and / or the bias RF signal generator 125 according to a multi-level RF power pulsing process. Figure 3 is an operational diagram illustrating a multi-level RF power pulsing process in which bias RF power is pulsed while maintaining a substantially constant primary RF power, according to some embodiments. The primary RF power supplied from the primary RF signal generator 137 to the coil 101 is held substantially constant at a primary power level (PP1) for the duration of the process, as shown by line 301. The bias RF power supplied from the bias RF signal generator 125 to the bias electrode 123 is pulsed between a low bias power level (BP1) and a high bias power level (BP2) for the duration of the process, as shown by line 303. The high bias power level (BP2) occurs during a first operating state (S1). The low bias power level (BP1) occurs during a second operating state (S2). The first operating state (S1) and the second operating state (S2) occur periodically (alternately and sequentially) over the duration of the process. In various embodiments, the duration of the predetermined first operating state (S1) is specified by an operator, and the duration of the predetermined second operating state (S2) is specified by an operator. In various embodiments, the duration of the predetermined first operating state (S1) can be the same as or different from the duration of the predetermined second operating state (S2). For example, in some embodiments, the duration of the predetermined first operating state (S1) is substantially equal to the duration of the predetermined second operating state (S2). Alternatively, in various embodiments, the duration of the predetermined first operating state (S1) is shorter than the duration of the predetermined second operating state (S2). Alternatively, in various embodiments, the duration of the predetermined first operating state (S1) is longer than the duration of the predetermined second operating state (S2). Additionally, the high bias power level (BP2) and the low bias power level (BP1) are specified by the operator.Transitions between the high bias power level (BP2) and the low bias power level (BP1), and vice versa, cause changes in the impedance of the plasma. Thus, for each of the first operating state (S1) and the second operating state (S2), which correspond to the high bias power level (BP2) and the low bias power level (BP1), respectively, the bias RF signal generator 125 has a different optimal operating frequency set point. For the first operating state (S1), the bias RF signal generator 125 has a first optimal operating frequency set point (F. B1S1 ) and for the second operating state (S2), the bias RF signal generator 125 has a second optimum operating frequency preset point (F B1S2 )

[0065] The sum of the duration of one first operating state (S1) and one second operating state (S2) gives the duration of a bias RF power pulse cycle. In some embodiments, the bias RF power is rapidly changed between the high bias power level (BP2) and the low bias power level (BP1). For example, in some embodiments, the duration of the bias RF power pulse cycle is on the order of 100 microseconds. However, in some embodiments, the duration of the bias RF power pulse cycle may be shorter or longer than 100 microseconds. The key point is that the duration of the bias RF power pulse cycle is very short so that very rapid switching occurs between the high bias power level (BP2) and the low bias power level (BP1). When the bias RF power level is rapidly changing, only one of the first operating state (S1) and the second operating state (S2) can be adjusted using the impedance matching system 129 to maximize the supplied bias RF power. The other of the first operating state (S1) and the second operating state (S2) needs to be adjusted using the frequency of the bias RF signal generator 125 to maximize the bias RF power supplied. For example, in some embodiments, the first operating state (S1) (high bias power) is adjusted using one or more capacitor settings in the impedance matching system 129 to maximize the bias RF power supplied, and the second operating state (S2) (low bias power) is adjusted using the frequency setting of the bias RF signal generator 125 to maximize the bias RF power supplied or the sustained voltage at the wafer plane.

[0066] To maximize RF power delivery to the plasma in both the first operating state (S1) and the second operating state (S2), a first operating frequency setpoint (f B1S1 ) and the second operating frequency preset point (f B1S2) must be set to simultaneously minimize the reflected RF power at output 271 of bias RF signal generator 125 in both the first operating state (S1) and the second operating state (S2), and simultaneously minimize the difference between the applied voltage / power and the setpoint voltage / power measured at output 271 of bias RF signal generator 125 in both the first operating state (S1) and the second operating state (S2). Thus, in the multi-level bias RF power pulsing process example of FIG. 3, there is a two-dimensional frequency setpoint grid for bias RF signal generator 125, with the horizontal axis of the two-dimensional frequency setpoint grid representing the first operating frequency setpoint (f B1S1 ), and the vertical axis of the two-dimensional frequency setpoint grid is the second operating frequency preset point (f B1S2 ) is the optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} are present in a two-dimensional frequency setpoint grid that includes: 1) an optimal value of a first operating frequency setpoint (F B1S1 ), and 2) the optimum value of the second operating frequency preset point (F B1S2 ). If a person is tasked with adjusting the bias RF signal generator 125 for the multi-level RF power pulsing process of FIG. 3, it is necessary to find an optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} must be selected: the minimum reflection coefficient (Γ) at the output 271 of the bias RF signal generator 125 during the first operating state (S1); the minimum reflection coefficient (Γ) at the output 271 of the bias RF signal generator 125 during the second operating state (S2); a minimum deviation of the voltage / power at the output 271 of the bias RF signal generator 125 from the setpoint voltage / power for the first operating state (S1); and The minimum deviation of the voltage / power at the output 271 of the bias RF signal generator 125 from the setpoint voltage / power for the second operating state (S2).

[0067] FIG. 4A illustrates a first operating frequency setpoint (f) for the example multi-level bias RF power pulsing process of FIG. 3, according to some embodiments. B1S1 ) and the second operating frequency preset point (f B1S2 ), the reflection coefficient (Γ) at the output 271 of the bias RF signal generator 125 during the first operating state (S1) is B1S1 4B illustrates the first operating frequency setpoint (f) for the example multi-level bias RF power pulsing process of FIG. 3, according to some embodiments. B1S1 ) and the second operating frequency preset point (f B1S2 ), the reflection coefficient (Γ) at the output 271 of the bias RF signal generator 125 during the second operating state (S2) is B1S2 4A shows the reflection coefficient (Γ) at the output 271 of the bias RF signal generator 125 during the first operating state (S1). B1S1 ) to find the optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2 However, FIG. 4B shows that the reflection coefficient (Γ) at the output 271 of the bias RF signal generator 125 during the second operating state (S2) is B1S2 ) to find the optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} indicates that there are a limited number of values ​​allowed.

[0068] FIG. 5A illustrates the first operating frequency setpoint (f) for the multi-level bias RF power pulsing process example of FIG. 3 at a voltage setpoint of 1000 V for the first operating state (S1), according to some embodiments. B1S1 ) and the second operating frequency preset point (f B1S2 ) at the output 271 of the bias RF signal generator 125 during the first operating state (S1). B1S15B illustrates the first operating frequency setpoint (f) for the multi-level bias RF power pulsing process example of FIG. 3 at a voltage setpoint of 200 V for the second operating state (S2), according to some embodiments. B1S1 ) and the second operating frequency preset point (f B1S2 ) the voltage (V) at the output 271 of the bias RF signal generator 125 during the second operating state (S2). B1S2 5A shows the optimum set of setpoint frequency coordinates {F B1S1 ,F B1S2 5B illustrates that multiple values ​​of setpoint frequency coordinate {F B1S1 ,F B1S2} indicates that there are a limited number of values ​​allowed.

[0069] Based on the data shown in Figures 4A, 4B, 5A, and 5B, an optimal set of setpoint frequency coordinates {F B1S1 , F B1S2 It should be appreciated that manually determining the setpoint frequency coordinates {F B1S1 , F B1S2 It should be understood that there is a limited number of preset operating frequencies {F B1S1 ,F B1S2 This is due to the interaction between the first operating state (S1) and the second operating state (S2) of the bias RF signal generator 125, which limits the number of allowable combinations of .

[0070] The prior art has been trying to find an optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} for the first operating state (S1) and the second operating state (S2) of the multi-level RF power pulsing process of FIG. B1S1 , F B1S2 Identifying or selecting the} was previously done by an RF specialist / engineer manually performing a series of frequency scans and then selecting the appropriate preset frequency after each scan. This traditional manual frequency scan procedure can take up to 20 minutes per process run, considering the time required to create each frequency scan recipe.

[0071] The automatic frequency adjustment method and corresponding system disclosed herein are directed to finding an optimal set of setpoint frequency coordinates {F B1S1 , F B1S2}. The automatic frequency adjustment method eliminates the need for a manual frequency scanning procedure to identify the optimum set of setpoint frequency coordinates {F B1S1 ,F B1S2} corresponds to the minimum achievable value of the cost function (J), as shown in Equation 1.

[0072]

number

[0073] The cost function (J) in Equation 1 is calculated by calculating the average reflection coefficient (

number

number

number

number

number

number

number

number

number

number

[0074] The cost function (J) in Equation 1 is calculated by calculating the first operating frequency setpoint (f B1S1 ) and the second operating frequency preset point (f B1S2 ) can be evaluated at any point on the 2D frequency setpoint grid defined by: operating the primary RF signal generator 137 to generate the plasma 105; Set the high power / voltage state (BP2) of the bias RF signal generator 125 in Match Cap adjustment mode (e.g., the first operating state (S1) in the example of FIG. 3); Set the low power / voltage state (BP1) of the bias RF signal generator 125 (e.g., the second operating state (S2) in the example of FIG. 3) in the Manual adjustment mode; The preset frequency (F') of the bias RF signal generator 125 for the first operating state (S1) B1S1 ) to the desired value, The preset frequency (F') of the bias RF signal generator 125 for the second operating state (S2) B1S2 ) to the desired value, For approximately 5 seconds or so, measure the average reflectance (

number

number

number

number

[0075] The average values ​​of the reflection coefficient and voltage / power for the cost function (J) are obtained from the RF generator's data logger by filtering out the transient portion of the process to obtain steady-state values. The cost function (J) is minimized using systematic cost function (J) evaluations carefully placed on a 2D frequency setpoint grid according to a triangulation search algorithm. For this reason, the 2D frequency setpoint grid is also referred to as a 2D frequency search grid. The triangulation search algorithm begins by placing an equilateral triangle of side length (L) in the lower right quadrant of the 2D frequency search grid. FIG. 6A shows a 2D frequency search grid with an initial equilateral triangle placed in the lower right quadrant of the 2D frequency search grid and defined by vertices V1, V2, and V3, in accordance with some embodiments. The location of the initial equilateral triangle in the lower right quadrant is determined by determining whether the lower right quadrant of the 2D frequency search grid is located within the optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2}. In this example embodiment, the initial side length (L) of the equilateral triangle is set to 200 kHz in order for the triangulation search algorithm to progress through the 2D frequency search grid quickly enough. However, it should be understood that in other embodiments, the initial side length (L) of the equilateral triangle may be set to a value other than 200 kHz.

[0076] Once the initial equilateral triangle is located, the cost function (J) is evaluated at all three vertices V, V, and V of the equilateral triangle. B1S1 ,f B1S2The cost function (J) in {} is evaluated by performing the above-described plasma processing recipe steps on a test wafer having a blanket film of target material exposed to plasma (e.g., a blanket film of target material to be etched). After the cost function (J) is evaluated at all three vertices V1, V2, and V3, the values ​​of the cost function (J) for the three vertices V1, V2, and V3 are compared to identify the vertex having the maximum value of the cost function (J). The vertex identified as having the maximum value of the cost function (J) is designated as the moving vertex.

[0077] Once a moving vertex is determined for the current equilateral triangle, the moving vertex is geometrically moved linearly within the 2D frequency search grid from its current position through the centroid of the current equilateral triangle until it is moved to a position where it forms a new equilateral triangle with the other two (non-moved) vertices of the current equilateral triangle, i.e., until it is separated from the other two (non-moved) vertices of the current equilateral triangle by the side length (L). In the example of FIG. 6A, vertex V3 is determined to have the maximum value of the cost function (J). Therefore, vertex V3 is designated as the moving vertex. FIG. 6B shows a 2D frequency search grid depicting the moving vertex V3 moving through the centroid of the current equilateral triangle to form a new vertex V4 of a new equilateral triangle. At this time, the new equilateral triangle formed by vertices V1, V2, and V4 is considered to be the current equilateral triangle. The triangulation search algorithm prevents vertices from moving outside the 2D frequency search grid.

[0078] The cost function (J) is then evaluated at the new vertex V4. Next, the values ​​of the cost function (J) for the three vertices V1, V2, and V4 of the current equilateral triangle are compared to identify the vertex with the maximum value of the cost function (J). The vertex identified as having the maximum value of the cost function (J) is designated as the next moving vertex. In the example of FIG. 6B, vertex V2 is determined to have the maximum value of the cost function (J). Therefore, vertex V2 is designated as the next moving vertex. FIG. 6C shows a 2D frequency search grid depicting the moving vertex V2 moving through the center of gravity of the current equilateral triangle to form a new vertex V5 of the new equilateral triangle. At this time, the new equilateral triangle formed by vertices V1, V4, and V5 is considered to be the current equilateral triangle. The cost function (J) for the new vertex V5 is evaluated. Next, the values ​​of the cost function (J) for the vertices V1, V4, and V5 of the current equilateral triangle are compared to identify the vertex with the maximum value of the cost function (J) as the next moving vertex.

[0079] The process of evaluating the cost function (J) at the three vertices of the current equilateral triangle, comparing the values ​​of the cost function (J) for the three vertices of the current equilateral triangle to determine the next moving vertex, and moving the next moving vertex to form a new equilateral triangle is repeated until the moving vertex returns to a previously occupied vertex position in the 2D frequency search grid. Figure 6D shows how the triangulation search algorithm progresses from Figure 6C up to the point where moving vertex V10 returns to the previously occupied vertex position corresponding to vertex V8. At this point, the minimum value of the cost function (J) is located within the region bounded by the set of iteration frequency coordinates corresponding to vertices V8 and V10.

[0080] If the moving vertex is found to repeat its previous position in the 2D frequency search grid, the size of the equilateral triangle is reduced by halving the size of its current side length (L). Thus, after the latest moving vertex is moved to occupy its previous position in the 2D frequency search grid, that previous position in the 2D frequency search grid becomes an anchor position for reducing the size of the equilateral triangle. With the anchor positions set, the other two vertices of the most recently formed equilateral triangle are each linearly moved toward the anchor positions to form a new equilateral triangle with a side length (L) that is half the size of the side length (L) of the previous equilateral triangle. Figure 6E shows the reduction in the size of the equilateral triangle after vertex V10 returns to its previous position corresponding to vertex V8. Vertex V8 is set as the anchor position for reducing the size of the equilateral triangle. To reduce the size of the equilateral triangle, vertex V6 is linearly moved toward vertex V8 to form a new vertex V11 until the linear distance between vertices V6 and V8 is half the side length (L) of the previous equilateral triangle. Also, vertex V7 is moved linearly toward vertex V8 to form new vertex V12 until the linear distance between vertices V7 and V8 is one-half the side length (L) of the previous equilateral triangle. At this point, vertices V8, V11, and V12 form a new equilateral triangle for the continuation of the triangulation search algorithm.

[0081] Once the size of the equilateral triangle is reduced, the triangulation search algorithm continues the iterative process of evaluating the cost function (J) at the three vertices of the current equilateral triangle, comparing the values ​​of the cost function (J) for the three vertices of the current equilateral triangle to determine the next moving vertex, and moving the next moving vertex to form a new equilateral triangle, in the same manner as described with respect to Figures 6A-6D, until the moving vertex returns to its previous position in the 2D frequency search grid. At this point, the size of the equilateral triangle is then reduced again, in the same manner as described with respect to Figure 6E. As shown in Figures 6A-6E, the process of moving vertices to create new equilateral triangles in the 2D frequency search grid and reducing the size of the equilateral triangle continues until a set of minimum cost function (J) convergence criteria is met.

[0082] When certain criteria for minimum cost function (J) convergence are met, the final optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} is the frequency that defines the centroid of the final equilateral triangle in the triangulation algorithm. In some embodiments, the convergence criteria for the minimum cost function (J) include: The cost function (J) can no longer be appreciably minimized and does not oscillate. The side length (L) of an equilateral triangle can no longer be appreciably reduced. · The frequency coordinates of moving vertices in a 2D frequency search grid no longer change appreciably and do not oscillate.

[0083] 7A shows a plot of the cost function (J) evaluated for each new vertex of a defined equilateral triangle over iterations of a triangulation search algorithm to illustrate the minimum cost function (J) convergence criteria for when the cost function (J) can no longer be appreciably minimized and does not oscillate, according to some embodiments. Oscillation of the cost function (J) is when the value of the cost function (J) moves in a repeating pattern as iterations of the triangulation search algorithm continue to be performed. In some embodiments, the cost function (J) can no longer be appreciably minimized when the change in the cost function value evaluated for sequential positions of the moving vertex within the two-dimensional frequency search grid is about 5% or less.

[0084] 7B shows a plot of the side lengths of equilateral triangles over iterations of the triangulation search algorithm to illustrate the minimum cost function (J) convergence criterion for when the side lengths of equilateral triangles can no longer be appreciably reduced, according to some embodiments. In some embodiments, the side lengths of equilateral triangles can no longer be appreciably reduced when new equilateral triangles formed during the triangulation search algorithm have side lengths of about 5 kilohertz or less within the two-dimensional frequency search grid.

[0085] FIG. 7C shows a plot of the frequency coordinates of a moving vertex in a 2D frequency search grid over iterations of the triangulation search algorithm to illustrate the minimum cost function (J) convergence criterion for when the frequency coordinates of the moving vertices in the 2D frequency search grid can no longer change appreciably and no longer oscillate, according to some embodiments. Oscillation of the frequency coordinates of the moving vertices is when the positions of the moving vertices move in a repeating pattern as iterations of the triangulation search algorithm continue to run. If oscillations in the values ​​of the moved vertex coordinates are detected, the triangulation search algorithm is stopped and restarted using a different set of vertices for the initial equilateral triangle. In some embodiments, the frequency coordinates of a moving vertex in the 2D frequency search grid can no longer change appreciably when the change in the position of the moving vertex is approximately 1 kilohertz or less in each frequency coordinate dimension of the two-dimensional frequency search grid.

[0086] FIG. 8 illustrates isocontours of the cost function (J) evaluated over a 2D frequency search grid and the final optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2 6A-6E, and the steps of the triangulation search algorithm required to complete to arrive at the final optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2 It should be understood that {F} represents the output of the automatic frequency adjustment method. The triangulation search algorithm typically requires approximately 10-20 iterations to meet the minimum cost function (J) convergence criterion. Each iteration (i.e., each evaluation of the cost function (J) for a new vertex) takes approximately 5 seconds to run. Thus, the triangulation search algorithm typically requires approximately 3-4 minutes to run, which is much shorter than the 20-30 minutes required for a manual frequency scanning procedure. Also, the optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} is much more reliable than the setpoint frequency coordinates determined by a manual frequency scanning procedure. It should be appreciated that because the triangulation search algorithm is systematically optimization-oriented and searches for the minimum cost function (J) within a 2D frequency search grid, the use of the triangulation search algorithm avoids the time and expense of having to generate the complete data sets shown in FIGS. 4A, 4B, 5A, and 5B.

[0087] While the above discussion of the automatic frequency adjustment method is described in the context of FIG. 3 in which the primary RF power is held substantially constant while the bias RF power level pulses between a low bias power level (BP1) and a high bias power level (BP2), it should be understood that the automatic frequency adjustment method disclosed herein is not so limited. For example, the automatic frequency adjustment algorithm disclosed herein may be utilized in processes in which both the primary RF power and the bias RF power are pulsed between different power levels, where the automatic frequency adjustment method generates an optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2}, and an automatic frequency adjustment method is applied to determine an optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2} is applied to determine

[0088] 9 is an operational diagram illustrating a multilevel RF power pulsing process in which both the bias RF power and the primary RF power are synchronously pulsed, according to some embodiments. The bias RF power is pulsed as described with respect to FIG. 3. The primary RF power supplied to the coil 101 from the primary RF signal generator 137 is pulsed between a low primary power level (PP1) and a high primary power level (PP2) for the duration of the process, as shown by line 401. The high primary power level (PP2) occurs during a first operating state (S1). The low primary power level (PP1) occurs during a second operating state (S2). The first operating state (S1) and the second operating state (S2) are the same as those described with respect to FIG. 3. The high primary power level (PP2) and the low primary power level (PP1) are specified by an operator. The transition between the high primary power level (PP2) and the low primary power level (PP1), and vice versa, causes a change in the impedance of the plasma. Thus, for each of the first and second operating states (S1) and (S2), which correspond to the high primary power level (PP2) and the low primary power level (PP1), respectively, the primary RF signal generator 137 has a different optimum operating frequency set point. For the first operating state (S1), the primary RF signal generator 137 has a first optimum operating frequency set point (F P1S1 ) and for the second operating state (S2), the primary RF signal generator 137 has a second optimum operating frequency preset point (F P1S2 )

[0089] The sum of the duration of one first operating state (S1) and the duration of one second operating state (S2) gives the duration of a primary RF power pulse cycle. In some embodiments, the primary RF power is rapidly changed between a high primary power level (PP2) and a low primary power level (PP1). For example, in some embodiments, the duration of a primary RF power pulse cycle is on the order of 100 microseconds. However, in some embodiments, the duration of a primary RF power pulse cycle may be shorter or longer than 100 microseconds. The key is that the duration of a primary RF power pulse cycle is very short so that very rapid switching occurs between the high primary power level (PP2) and the low primary power level (PP1). When the primary RF power level is rapidly changing, only one of the first operating state (S1) and the second operating state (S2) can be adjusted using the impedance matching system 141 to maximize the delivered primary RF power. The other of the first operating state (S1) and the second operating state (S2) needs to be adjusted using the frequency setpoint of the primary RF signal generator 137 to maximize the delivered primary RF power. For example, in some embodiments, the first operating state (S1) (high primary power) is adjusted using one or more capacitor settings in the impedance matching system 141 to maximize the delivered primary RF power, and the second operating state (S2) (low primary power) is adjusted using the frequency setpoint of the primary RF signal generator 137 to maximize the delivered primary RF power.

[0090] To maximize RF power delivery to the plasma in both the first operating state (S1) and the second operating state (S2), a first operating frequency setpoint (f P1S1 ) and the second operating frequency preset point (f P1S2) must be set to simultaneously minimize the reflected RF power at output 210 of primary RF signal generator 137 in both the first operating state (S1) and the second operating state (S2), and simultaneously minimize the difference between the applied voltage / power and the setpoint voltage / power measured at output 210 of primary RF signal generator 137 in both the first operating state (S1) and the second operating state (S2). Thus, in the multi-level RF power pulsing process example of FIG. 4, there is a two-dimensional frequency setpoint grid for primary RF signal generator 137, with the horizontal axis of the two-dimensional frequency setpoint grid representing the first operating frequency setpoint (f P1S1 ), and the vertical axis of the two-dimensional frequency setpoint grid is the second operating frequency preset point (f P1S2 ) is the optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} are present in a two-dimensional frequency setpoint grid that includes: 1) an optimal value of a first operating frequency setpoint (F P1S1 ), and 2) the optimum value of the second operating frequency preset point (F P1S2 ). In the automatic frequency adjustment method, the optimum set of setpoint frequency coordinates {F P1S1 ,F P1S2} corresponds to the minimum achievable value of the cost function (J), as shown in Equation 2.

[0091]

number

[0092] The cost function (J) in Equation 2 is calculated by calculating the average reflection coefficient (

number

number

number

number

number

number

number

number

number

number

[0093] The cost function (J) in Equation 2 is calculated by calculating the first operating frequency setpoint (f P1S1 ) and the second operating frequency preset point (f P1S2 ) can be evaluated at any point on the 2D frequency setpoint grid defined by: Turn off the bias RF signal generator 125, Set the high power / voltage state of the primary RF signal generator 137 in Match Cap adjustment mode (for example, the first operating state (S1) in the example of FIG. 4); Set the primary RF signal generator 137 to a low power / voltage state (e.g., the second operating state (S2) in the example of FIG. 4) in Manual adjustment mode; The preset frequency (F') of the first operating state (S1) P1S1 ) to the desired value, Preset frequency (F') of the second operating state (S2) P1S2 ) to the desired value, For approximately 5 seconds or so, measure the average reflectance (

number

number

number

number

[0094] The average values ​​of the reflection coefficient and voltage / power for the cost function (J) are obtained from the RF generator's data logger by filtering out the transient portion of the process to obtain steady-state values. The triangulation search algorithm described above with respect to Figures 6A-6E finds an optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} is performed to determine the optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} is the output of the frequency adjustment process of the primary RF signal generator 137. The optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} is determined, an optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2}, the same automatic frequency adjustment method described above is used to determine the optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} while operating the primary RF signal generator 137.

[0095] 10 is an operational diagram illustrating a multi-level RF power pulsing process in which both the bias RF power and the primary RF power are pulsed asynchronously, according to some embodiments. The bias RF power is pulsed in the opposite manner to that described with respect to FIG. 3. Specifically, as shown by line 503, a low bias power (BP1) occurs during a first operating state (S1) and a high bias power (BP2) occurs during a second operating state (S2). Also, similar to the embodiment of FIG. 9, a high primary power (PP2) occurs during the first operating state (S1) and a low primary power (PP1) occurs during the second operating state (S2). The automatic frequency adjustment method described above with respect to FIG. 9 finds an optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} for the primary RF signal generator 137. P1S1 ,F P1S2} is determined, an optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2}, the same automatic frequency adjustment method described above with respect to FIG. 3 is used to determine the optimal set of setpoint frequency coordinates {F P1S1 ,F P1S2} while operating the primary RF signal generator 137.

[0096] 11 is a flow chart illustrating a method for automatic frequency adjustment of an RF signal generator for operation in a multi-level RF power pulsing mode, according to some embodiments. The RF signal generator of FIG. 11 may be any of the bias RF signal generators 125 of the primary RF signal generator 137. The method includes plotting a first coordinate axis (e.g., f B1S1 or f P1S1 ), and a second coordinate axis (e.g., f B1S2 or f P1S2 The method includes step 1101 for defining a two-dimensional frequency search grid having a first output power level in a first operating state and a second output power level in a second operating state. The first and second output power levels are different from one another. The RF signal generator is programmed to operate in a multi-level RF power pulsing mode by periodically cycling between the first operating state and the second operating state, as shown in FIGS. 3, 9, and 10. The method further includes step 1103 for performing an automatic search process within the two-dimensional frequency search grid to simultaneously determine a first optimum value for an operating frequency setpoint of the RF signal generator in the first operating state and a second optimum value for an operating frequency setpoint of the RF signal generator in the second operating state. The method further includes step 1105 for configuring the RF signal generator to operate using the first optimum value for the operating frequency setpoint of the RF signal generator in the first operating state and the second optimum value for the operating frequency setpoint of the RF signal generator in the second operating state.

[0097] In some embodiments, the automatic search process is directed by the control system 153 of the plasma processing system 100 including the RF signal generator. In some embodiments, the automatic search process includes multiple iterations, each iteration including operating the RF signal generator in a multi-level RF power pulsing mode to generate a plasma in the plasma processing system 100 using a different set of operating frequency setpoint coordinates within a two-dimensional frequency search grid. In some embodiments, each iteration of the automatic search process includes an empirical evaluation of the RF power delivery from the RF signal generator to the plasma. In some embodiments, a plasma is generated to be exposed to the substrate 107 in each iteration of the automatic search process, the substrate 107 having a film of target material present on an upper surface of the substrate 107 exposed to the plasma. In some embodiments, the RF signal generator in this method is the bias RF signal generator 125, and the plasma processing system 100 further includes a primary RF signal generator 137 separate from the bias RF signal generator 125. In these embodiments, each iteration of the automatic search process includes operating the primary RF signal generator 137 to generate an additional plasma in the plasma processing system 100. In some embodiments, the RF signal generator of this method is primary RF signal generator 137. In some embodiments, plasma processing system 100 further comprises a bias RF signal generator 125 that is separate from primary RF signal generator 137. In these embodiments, each iteration of the automatic search process for automatically frequency tuning primary RF signal generator 137 for operation in the multi-level RF power pulsing mode is performed with bias RF signal generator 125 turned off.

[0098] In some embodiments, the automated search process finds an optimal set of frequency coordinates (e.g., {F B1S1 ,F B1S2} or {F P1S1 ,F P1S2}) within the two-dimensional frequency search grid. The optimal set of frequency coordinates is a first optimal value (e.g., F B1S1 or F P1S1 ) and a second optimum value of the operating frequency preset point of the RF signal generator in the second operating state (e.g., F B1S2 or F P1S2 ) In some embodiments, the triangulation search process includes evaluating a cost function (J) at vertices of equilateral triangles formed sequentially within a two-dimensional frequency search grid. In some embodiments, the evaluation of the cost function (J) for predetermined vertices of predetermined equilateral triangles within the two-dimensional frequency search grid is performed using frequency coordinates (e.g., {F' B1S1 ,F' B1S2} or {F' P1S1 ,F' P1S2}) for a period of time. In some embodiments, the period of operation of the RF signal generator in the multi-level RF power pulsing mode for evaluating the cost function (J) is about 5 seconds or less. In some embodiments, the evaluation of the cost function for predetermined vertices of predetermined equilateral triangles in the two-dimensional frequency search grid is performed by calculating an average (e.g., (

number

number

number

number

number

number

number

number

[0099] In some embodiments, the triangulation search process further includes sequentially forming equilateral triangles within the two-dimensional frequency search grid by: 1) identifying a specific vertex of the current equilateral triangle that has the largest cost function (J) value among the three vertices of the current equilateral triangle; and 2) linearly moving the specific vertex of the current equilateral triangle through the centroid of the current equilateral triangle while maintaining the positions of the two other vertices of the current equilateral triangle until the specific vertex reaches a new vertex position within the two-dimensional frequency search grid that corresponds to the formation of a new equilateral triangle defined by the position of the new vertex and the positions of the two other vertices whose positions were maintained. This process is described by way of example with reference to FIGS. 6A-6E . In some embodiments, the triangulation search process continues sequentially forming equilateral triangles within the two-dimensional frequency search grid until the new equilateral triangle is a repetition of the previous equilateral triangle, at which point the triangulation search process reduces the size of the new equilateral triangle and resumes sequentially forming equilateral triangles within the two-dimensional frequency search grid.

[0100] In some embodiments, the triangulation search process continues to sequentially form equilateral triangles within the two-dimensional frequency search grid and reduce the size of each new equilateral triangle as it repeats a previous equilateral triangle until convergence criteria are met. In some embodiments, the convergence criteria for the triangulation search process include a change in cost function value evaluated for sequential positions of the moved vertex within the two-dimensional frequency search grid of no more than about 5%, a side length of the new equilateral triangle within the two-dimensional frequency search grid of no more than about 5 kilohertz, and a change in the position of the moved vertex of no more than about 1 kilohertz in each frequency coordinate dimension of the two-dimensional frequency search grid.

[0101] Disclosed herein is a system for automatic frequency tuning of an RF signal generator for operation in a multilevel RF power pulsing mode. The system includes a plasma processing chamber 103, which includes a substrate support structure 113 and electrodes 123 and / or 101. The system further includes an RF signal generator configured to generate an RF signal and transmit the RF signal through an output of the RF signal generator. The RF signal generator is either a primary RF signal generator 137 or a bias RF signal generator 125. The RF signal generator is configured to have a first output power level in a first operating state and a second output power level in a second operating state, the first and second output power levels being different from each other. The RF signal generator is programmed to operate in the multilevel RF power pulsing mode by periodically alternating between the first and second operating states. The system further includes an impedance matching system having an input connected to the output of the RF signal generator. The impedance matching system is impedance matching system 129 when the RF signal generator is bias RF signal generator 125. The impedance matching system is impedance matching system 141 when the RF signal generator is primary RF signal generator 137. The impedance matching system has an output connected to electrode 123 or 101. The impedance matching system is configured to control the impedance at the output of the RF signal generator to enable transmission of an RF signal through electrode 123 or 101 to a plasma generated in plasma processing chamber 103.

[0102] The system further includes a first coordinate axis (e.g., f B1S1 or f P1S1 ), and a second coordinate axis (e.g., f B1S2 or f P1S2The RF signal generator includes a control system 153 programmed to define a two-dimensional frequency search grid having a first optimum value for the operating frequency setpoint of the RF signal generator in a first operating state and a second optimum value for the operating frequency setpoint of the RF signal generator in a second operating state. In some embodiments, the control system 153 is programmed to configure the RF signal generator to operate using the first optimum value for the operating frequency setpoint of the RF signal generator in the first operating state and the second optimum value for the operating frequency setpoint of the RF signal generator in the second operating state.

[0103] In some embodiments, the automated search process includes multiple iterations, and for each iteration, the control system 153 is programmed to direct operation of the RF signal generator in a multi-level RF power pulsing mode to generate a plasma in the plasma processing chamber 103 using a different set of operating frequency setpoint coordinates within the two-dimensional frequency search grid. In some embodiments, the control system 153 is programmed to empirically evaluate the RF power delivery from the RF signal generator to the plasma in each iteration of the automated search process. In some embodiments, a substrate 107 is disposed on the substrate support structure 113, the substrate 107 having a film of target material present on an upper surface of the substrate 107 exposed to the plasma generating region 106 in the plasma processing chamber 103. In some embodiments, a plasma is generated to be exposed to the substrate 107 in each iteration of the automated search process.

[0104] In some embodiments, the RF signal generator of the system is the bias RF signal generator 125. In some embodiments, the system further includes a primary RF signal generator 137 that is separate from the bias RF signal generator 125. In these embodiments, the control system 153 is programmed to direct operation of the primary RF signal generator 137 to generate additional plasma in the plasma processing chamber 103 during each iteration of the automatic search process. In some embodiments, the RF signal generator of the system is the primary RF signal generator 137. In some of these embodiments, the system further includes a bias RF signal generator 125 that is separate from the primary RF signal generator 137. In these embodiments, the control system 153 is programmed to turn off the bias RF signal generator 125 during each iteration of the automatic search process to automatically frequency tune the primary RF signal generator 137 for operation in the multi-level RF power pulsing mode.

[0105] In some embodiments, the control system 153 determines an optimal set of frequency coordinates (e.g., {F B1S1 ,F B1S2} or {F P1S1 ,F P1S2}) within the two-dimensional frequency search grid. The optimal set of frequency coordinates is a first optimal value (e.g., F B1S1 or F P1S1 ) and a second optimum value of the operating frequency preset point of the RF signal generator in the second operating state (e.g., F B1S2 or F P1S2 In some embodiments, the control system 153 is programmed to perform the triangulation search process by evaluating a cost function (J) at the vertices of equilateral triangles formed sequentially within a two-dimensional frequency search grid.

[0106] In some embodiments, the control system 153 selects frequency coordinates (e.g., {F' B1S1 ,F' B1S2} or {F' P1S1 ,F' P1S2}) to direct operation of the RF signal generator in a multi-level RF power pulsing mode for a period of time, thereby evaluating a cost function (J) for predetermined vertices of a predetermined equilateral triangle in a two-dimensional frequency search grid. In some embodiments, the control system 153 is programmed to set the period of operation of the RF signal generator in the multi-level RF power pulsing mode for evaluating the cost function (J) to about 5 seconds or less. In some embodiments, the control system 153 calculates the average (e.g., (

number

number

number

number

number

number

number

number

[0107] In some embodiments, as part of the triangulation search process, the control system 153 is programmed to perform a process of sequentially forming equilateral triangles within a two-dimensional frequency search grid by: 1) identifying a specific vertex of the current equilateral triangle that has the largest cost function value among the three vertices of the current equilateral triangle; and 2) linearly moving the specific vertex of the current equilateral triangle through the centroid of the current equilateral triangle while maintaining the positions of the two other vertices of the current equilateral triangle until the specific vertex reaches a new vertex position within the two-dimensional frequency search grid that corresponds to the formation of a new equilateral triangle defined by the position of the new vertex and the positions of the two other vertices whose positions have been maintained. This process is described by way of example with reference to FIGS. 6A-6E . In some embodiments, as part of the triangulation search process, the control system 153 is programmed to continue the process of sequentially forming equilateral triangles within the two-dimensional frequency search grid until the new equilateral triangle is a repeat of the previous equilateral triangle, at which point the control system 153 is programmed to reduce the size of the new equilateral triangle and resume the process of sequentially forming equilateral triangles within the two-dimensional frequency search grid.

[0108] In some embodiments, as part of the triangulation search process, control system 153 is programmed to continue the process of sequentially forming equilateral triangles within the two-dimensional frequency search grid and reducing the size of each new equilateral triangle as it repeats a previous equilateral triangle until convergence criteria are met. In some embodiments, control system 153 is programmed to apply convergence criteria including: a change in cost function value evaluated for sequential positions of the moving vertex within the two-dimensional frequency search grid of no more than about 5%, the new equilateral triangle having a side length within the two-dimensional frequency search grid of no more than about 5 kilohertz, and a change in the position of the moving vertex of no more than about 1 kilohertz in each frequency coordinate dimension of the two-dimensional frequency search grid.

[0109] In some embodiments, the automatic frequency adjustment method disclosed herein for tuning multi-level RF power pulsing recipe steps for optimal RF stability and performance is implemented as a software module integrated into the recipe editor of control module 153. In these embodiments, the automatic frequency adjustment method is implemented as computer program instructions stored in computer memory within control module 153. The software module prompts the user to select a tuner (if necessary) and automatically tunes each multi-level RF power pulsing step of the manufacturing process. This procedure requires the user to place an optimal wafer in plasma processing apparatus 100. Once plasma processing apparatus 100 has completed the automatic frequency adjustment method, the operator can view the optimal set of setpoint frequency coordinates {F B1S1 ,F B1S2 You will be prompted and asked to accept the

[0110] It should be appreciated that the automatic frequency adjustment method disclosed herein provides a frequency adjustment procedure for recipe creation and development, eliminating the need for an RF expert / engineer to manually recommend stable operating frequency conditions. In other words, if a process engineer is developing / creating a recipe with seemingly unstable conditions, the automatic frequency adjuster disclosed herein can provide a stable operating point for advancing process development. Furthermore, utilizing the automatic frequency adjustment method disclosed herein may lead to the exploitation of new, advantageous processing conditions that may result in better on-wafer metrics (e.g., CD, bow). Because implementation of the automatic frequency adjustment method does not require hardware modifications to the plasma processing apparatus 100, feedback implementation of the automatic frequency adjustment method can be achieved at low cost. The automatic frequency adjustment method disclosed herein can be implemented in software without requiring manufacturing or engineering modifications to existing processing tools.

[0111] As described herein, manual frequency adjustment procedures performed by an RF specialist / engineer can take up to 20 minutes per process step (including frequency scan recipe preparation). The automatic frequency adjustment method disclosed herein has been demonstrated to complete in 2-2.5 minutes, eliminating the need for an RF specialist / engineer to be present at the tool. Given the wide range of applicability of the automatic frequency adjustment method disclosed herein, it is envisioned that the automatic frequency adjustment method may provide a basis for application of automatic frequency adjustment to other semiconductor manufacturing processing tools, particularly tools that utilize inter-level pulsing.

[0112] It should be understood that the embodiments described herein may employ various computer-implemented operations, including data stored in a computer system. These operations require physical manipulation of physical quantities. Any operations described herein that form part of the embodiments are useful machine operations. The embodiments further relate to hardware units or apparatus for performing these operations. An apparatus may be specially configured for a special-purpose computer. When defined as a special-purpose computer, the computer may operate for the special purpose while also performing other processes, program execution, or routines not included in the special purpose. In some embodiments, operations may be processed on a general-purpose computer selectively activated or configured by one or more computer programs stored in computer memory, cache, or retrieved over a network. When data is retrieved over a network, the data may be processed by other computers on the network (e.g., a cloud of computing resources).

[0113] Various embodiments described herein may be fabricated as computer-readable code on a non-transitory computer-readable medium. The non-transitory computer-readable medium is any data storage hardware unit that can store data so that it can be read by a computer system. Examples of non-transitory computer-readable media include hard drives, network-attached storage (NAS), ROM, RAM, compact disc-ROMs (CD-ROMs), CD-recordables (CD-Rs), CD-rewritables (CD-RWs), magnetic tapes, and other optical and non-optical data storage hardware units. The non-transitory computer-readable medium may include a tangible computer-readable medium distributed over network-coupled computer systems so that the computer-readable code is stored and executed in a distributed manner.

[0114] While the present disclosure includes certain details for ease of understanding, it will be apparent that certain changes and modifications may be made within the scope of the appended claims. For example, it should be understood that one or more features from any embodiment disclosed herein may be combined with one or more features of any other embodiment disclosed herein. Accordingly, these embodiments are to be considered illustrative and not limiting, and the claims are not to be limited to the details set forth herein, but may be modified within the scope and equivalents of the described embodiments.

Claims

1. 1. A method for automatic frequency adjustment of a radio frequency (RF) signal generator for operation in a multi-level RF power pulsing mode, comprising: defining a two-dimensional frequency search grid having a first coordinate axis representing an operating frequency setpoint of the RF signal generator in a first operating state and a second coordinate axis representing an operating frequency setpoint of the RF signal generator in a second operating state, the RF signal generator having a first output power level in the first operating state and a second output power level in the second operating state, the first and second output power levels being different from one another, and the RF signal generator being programmed to operate in the multi-level RF power pulsing mode by periodically alternating between the first operating state and the second operating state; performing an automatic search process within the two-dimensional frequency search grid to simultaneously determine a first optimum value for the operating frequency setpoint of the RF signal generator in the first operating state and a second optimum value for the operating frequency setpoint of the RF signal generator in the second operating state; configuring the RF signal generator to operate using the first optimum value and the second optimum value; A method comprising:

2. 2. The method of claim 1, wherein the automatic search process is directed by a control system of a plasma processing system that includes the RF signal generator.

3. 3. The method of claim 2, wherein the automatic search process includes multiple iterations, each iteration including operating the RF signal generator in the multi-level RF power pulsing mode to generate a plasma in the plasma processing system using a different set of operating frequency setpoint coordinates within the two-dimensional frequency search grid.

4. 4. The method of claim 3, wherein each iteration includes an empirical evaluation of RF power delivery from the RF signal generator to the plasma.

5. 4. The method of claim 3, wherein the plasma is generated to be exposed to a substrate in each iteration, the substrate having a film of target material present on an upper surface of the substrate exposed to the plasma.

6. 4. The method of claim 3, wherein the RF signal generator is a bias RF signal generator, the plasma processing system includes a primary RF signal generator separate from the bias RF signal generator, and each iteration includes operating the primary RF signal generator to generate additional plasma in the plasma processing system.

7. The method of claim 3 , wherein the RF signal generator is a primary RF signal generator.

8. 8. The method of claim 7, wherein the plasma processing system includes a bias RF signal generator separate from the primary RF signal generator, and each iteration is performed with the bias RF signal generator turned off.

9. 2. The method of claim 1, wherein the automated search process is a triangulation search process performed within a two-dimensional frequency search grid to determine an optimal set of frequency coordinates within the two-dimensional frequency search grid that corresponds to minimizing a cost function defined to characterize RF power delivery from the RF signal generator to a plasma load, the optimal set of frequency coordinates including the first optimal value and the second optimal value.

10. 10. The method of claim 9, wherein the triangulation search process includes evaluating the cost function at the vertices of equilateral triangles formed sequentially within the two-dimensional frequency search grid.

11. 11. The method of claim 10, wherein evaluating the cost function for predetermined vertices of a predetermined equilateral triangle in the two-dimensional frequency search grid is performed by operating the RF signal generator in the multi-level RF power pulsing mode for a period of time using frequency coordinates for the predetermined vertices in the two-dimensional frequency search grid as operating frequency setpoints for the RF signal generator in the first and second operating states.

12. 12. The method of claim 11, wherein the period of time is about 5 seconds or less.

13. 12. The method of claim 11 , wherein evaluating the cost function for the predetermined vertices of the predetermined equilateral triangle in the two-dimensional frequency search grid comprises calculating the sum of an average of a reflection coefficient of an RF signal generated by the RF signal generator in the first operating state over the time period, an average of a reflection coefficient of an RF signal generated by the RF signal generator in the second operating state over the time period, an average of a relative error of a voltage generated by the RF signal generator in the first operating state over the time period, and an average of a relative error of a voltage generated by the RF signal generator in the second operating state over the time period.

14. 14. The method of claim 13, wherein the triangulation search process further includes identifying a particular vertex of the current equilateral triangle that has the largest cost function value among the three vertices of the current equilateral triangle, and sequentially forming equilateral triangles within the two-dimensional frequency search grid by moving the particular vertex of the current equilateral triangle linearly through the center of gravity of the current equilateral triangle while maintaining the positions of the two other vertices of the current equilateral triangle, until the particular vertex reaches a position of the new vertex in the two-dimensional frequency search grid that corresponds to the formation of a new equilateral triangle defined by the position of the new vertex and the positions of the two other vertices whose positions are maintained.

15. 15. The method of claim 14, wherein the triangulation search process continues the process of forming successive equilateral triangles in the two-dimensional frequency search grid until the new equilateral triangle is a repeat of the previous equilateral triangle, at which point the triangulation search process reduces the size of the new equilateral triangle and resumes the process of forming successive equilateral triangles in the two-dimensional frequency search grid.

16. 16. The method of claim 15, wherein the triangulation search process continues the process of sequentially forming equilateral triangles in the two-dimensional frequency search grid and reducing the size of the new equilateral triangles when the new equilateral triangles repeat the previous equilateral triangles until a convergence criterion is met.

17. 17. The method of claim 16, wherein the convergence criteria include a change in the cost function value evaluated for sequential positions of a moving vertex within the two-dimensional frequency search grid of about 5% or less, the convergence criteria include the new equilateral triangle having a side length within the two-dimensional frequency search grid of about 5 kilohertz or less, and the convergence criteria include a change in position of the moving vertex of about 1 kilohertz or less in each frequency coordinate dimension of the two-dimensional frequency search grid.

18. 1. A system for automatic frequency adjustment of a radio frequency (RF) signal generator for operation in a multi-level RF power pulsing mode, comprising: a plasma processing chamber having a substrate support structure and an electrode; an RF signal generator configured to generate an RF signal and transmit the RF signal through an output of the RF signal generator, the RF signal generator configured to have a first output power level in a first operating state and a second output power level in a second operating state, the first and second output power levels being different from one another, and the RF signal generator programmed to operate in the multi-level RF power pulsing mode by periodically alternating between the first operating state and the second operating state; an impedance matching system having an input connected to the output of the RF signal generator, the impedance matching system having an output connected to the electrode, the impedance matching system configured to control the impedance at the output of the RF signal generator to enable transmission of the RF signal through the electrode to a plasma generated in the plasma processing chamber; a control system programmed to define a two-dimensional frequency search grid having a first coordinate axis representing an operating frequency setpoint of the RF signal generator in the first operating state and a second coordinate axis representing an operating frequency setpoint of the RF signal generator in the second operating state, the control system being programmed to perform an automatic search process within the two-dimensional frequency search grid to simultaneously determine a first optimum value for the operating frequency setpoint of the RF signal generator in the first operating state and a second optimum value for the operating frequency setpoint of the RF signal generator in the second operating state; A system comprising:

19. 20. The system of claim 18, wherein the control system is programmed to configure the RF signal generator to operate using the first optimum value and the second optimum value.

20. 20. The system of claim 18, wherein the automated search process includes multiple iterations, and for each iteration, the control system is programmed to direct operation of the RF signal generator in the multi-level RF power pulsing mode to generate a plasma in the plasma processing chamber using a different set of operating frequency setpoint coordinates within the two-dimensional frequency search grid.

21. 21. The system of claim 20, wherein the control system is programmed to empirically estimate RF power delivery from the RF signal generator to the plasma at each iteration.

22. 21. The system of claim 20, wherein a substrate is disposed on the substrate support structure, the substrate having a film of target material present on an upper surface of the substrate exposed to a plasma generating region in the plasma processing chamber, and the plasma is generated to expose the substrate in each iteration of the automatic search process.

23. 21. The system of claim 20, wherein the RF signal generator is a bias RF signal generator, the system further comprising a primary RF signal generator separate from the bias RF signal generator, and the control system is programmed to direct operation of the primary RF signal generator to generate additional plasma in the plasma processing chamber during each iteration of the automatic search process.

24. 21. The system of claim 20, wherein the RF signal generator is a primary RF signal generator.

25. 25. The system of claim 24, further comprising: a bias RF signal generator separate from the primary RF signal generator; The control system is programmed to turn off the bias RF signal generator during each iteration of the automatic search process.

26. 20. The system of claim 18, wherein the control system is programmed to execute the automated search process as a triangulation search process within the two-dimensional frequency search grid to determine an optimal set of frequency coordinates within the two-dimensional frequency search grid that corresponds to minimizing a cost function defined to characterize RF power delivery from the RF signal generator to the plasma, the optimal set of frequency coordinates including the first optimal value and the second optimal value.

27. 27. The system of claim 26, wherein the control system is programmed to perform the triangulation search process by evaluating the cost function at vertices of equilateral triangles formed sequentially within the two-dimensional frequency search grid.

28. 28. The system of claim 27, wherein the control system is programmed to evaluate the cost function for predetermined vertices of a predetermined equilateral triangle within the two-dimensional frequency search grid by directing the RF signal generator to operate in the multi-level RF power pulsing mode for a period of time using frequency coordinates for the predetermined vertices within the two-dimensional frequency search grid as operating frequency setpoints for the RF signal generator in the first and second operating states.

29. 29. The system of claim 28, wherein the control system is programmed to set the period of time to about 5 seconds or less.

30. 29. The system of claim 28, wherein the control system is programmed to evaluate the cost function for the predetermined vertices of the predetermined equilateral triangle in the two-dimensional frequency search grid by calculating the sum of an average of a reflection coefficient of an RF signal generated by the RF signal generator in the first operating state over the time period, an average of a reflection coefficient of an RF signal generated by the RF signal generator in the second operating state over the time period, an average of a relative error of a voltage generated by the RF signal generator in the first operating state over the time period, and an average of a relative error of a voltage generated by the RF signal generator in the second operating state over the time period.

31. 31. The system of claim 30, wherein the control system is programmed to: as part of the triangulation search process, identify a particular vertex of the current equilateral triangle that has the largest cost function value among the three vertices of the current equilateral triangle; and perform a process to sequentially form equilateral triangles within the two-dimensional frequency search grid by moving the particular vertex of the current equilateral triangle linearly through the center of gravity of the current equilateral triangle while maintaining the positions of the two other vertices of the current equilateral triangle, until the particular vertex reaches a position within the two-dimensional frequency search grid that corresponds to the formation of a new equilateral triangle defined by the position of the new vertex and the positions of the two other vertices whose positions have been maintained.

32. 32. The system of claim 31 , wherein as part of the triangulation search process, the control system is programmed to continue the process of forming successive equilateral triangles in the two-dimensional frequency search grid until the new equilateral triangle is a repeat of a previous equilateral triangle, at which point the control system is programmed to reduce the size of the new equilateral triangle and resume the process of forming successive equilateral triangles in the two-dimensional frequency search grid.

33. 33. The system of claim 32, wherein as part of the triangulation search process, the control system is programmed to continue the process of sequentially forming equilateral triangles within the two-dimensional frequency search grid and reducing the size of the new equilateral triangles when the new equilateral triangles repeat the previous equilateral triangles until a convergence criterion is met.

34. 34. The system of claim 33, wherein the control system is programmed to apply convergence criteria including a change in the cost function value evaluated for sequential positions of a moving vertex within the two-dimensional frequency search grid of about 5% or less, the convergence criteria including the new equilateral triangle having a side length within the two-dimensional frequency search grid of about 5 kilohertz or less, and the convergence criteria including a change in position of the moving vertex of about 1 kilohertz or less in each frequency coordinate dimension of the two-dimensional frequency search grid.

Images