Method for controlling the power of an amplifier unit, HF generator for a plasma process, and plasma-generating system

Abstract

A method for controlling the power of an amplifier unit (5) in an HF generator (1) for a plasma process using a controller unit (4) has a plurality of method steps. In a first method step, a multi-dimensional model (15) of the amplifier unit (5) is used (S1), the multi-dimensional model comprising at least one input variable, preferably two input variables, particularly preferably more than two input variables. In a second method step, at least one controller parameter of the controller unit (4) or an error signal (11), which is supplied to an input of the controller unit, or at least one manipulated variable (12), which is output by the controller unit at an output, is continuously modified (S2). The modification process is carried out on the basis of at least one amplification value (14) for the amplifier unit, the amplification value being determined from the multi-dimensional model of the amplifier unit using the current value for the at least one input variable of the multi-dimensional model.

Description

[0001]

[0002] Method for power regulation of an amplifier unit, an RF generator for a plasma process and a plasma generation system

[0003] The invention relates to a method for power control of an amplifier unit in an RF generator for a plasma process, an RF generator for a plasma process, and a plasma generation system with such an RF generator. "RF" here is the abbreviation for "high frequency".

[0004] Surface treatment of workpieces using plasma and gas lasers are industrial processes in which, particularly in a plasma chamber, a plasma is generated with direct current or with a high-frequency alternating signal with a working frequency in the range of a few tens of kHz to the GHz range.

[0005] The plasma chamber is connected to a high-frequency generator (HF generator) via additional electronic components such as coils, capacitors, wires, or transformers. These additional components can be resonant circuits, filters, or impedance matching circuits.

[0006] Plasma processes represent a highly variable load for a high-frequency generator, depending on the conditions in the plasma chamber. In particular, the properties of the workpiece, electrodes, and gas conditions play a role.

[0007] High-frequency generators have a limited operating range with respect to the impedance of the connected electrical load. If the load impedance deviates from a permissible range, the required energy / power cannot be delivered to the consumer. Damage to the RF generator is also possible.

[0008] For this reason, an impedance matching circuit, also called a "matchbox," is frequently used to transform the load impedance to a nominal impedance of the generator output. The technical requirements for the RF generator are also constantly increasing. The purpose of the RF generator is to transfer radio frequency energy into the plasma chamber. This RF energy is used to ionize gases in the chamber, thus generating the plasma. Furthermore, the RF energy helps to stabilize the plasma and control its density and temperature. By adjusting the RF power, different plasma states can be achieved, which are required for various applications. In semiconductor manufacturing and materials processing, plasma chambers are often used for sputtering processes. In this process, ions from the plasma are directed onto a target, also called the "target material," causing atoms to be knocked out of the target and deposited onto a substrate.The RF generator enables control of the energy input into the plasma and thus control of the sputtering process. In chemical vapor deposition (CVD) processes, reactive gases are also introduced into the plasma chamber. This is also known as plasma-enhanced chemical vapor deposition (PECVD). The energy from the RF generator helps to decompose these gases and form reactive particles, which then deposit onto a substrate to form solid layers. The plasma energy can also be used to clean surfaces of contaminants, as the reactive particles in the plasma react with and remove the contaminants on the surface. The RF generator can be a crucial tool for generating and controlling plasma in a plasma chamber, enabling a wide variety of industrial and scientific processes.German patent application DE 10 2022 119 157 A1 discloses a control device for a plasma generation system, configured to control an impedance matching circuit. The impedance matching circuit is connected between an RF generator and a load. The control device is configured to determine a target impedance value for an input terminal of the impedance matching circuit based on: a) a predefinable operating frequency of the RF generator; b) a predefinable target power of the RF generator; and c) model parameters of the RF generator. The target impedance value differs from the nominal impedance in order to increase an operating characteristic of the RF generator, in particular its efficiency, for the predefinable operating frequency and the predefinable target power.

[0009] Modern plasma processes require increasingly dynamic RF power control. The behavior, particularly the gain, of RF amplifiers depends on a multitude of operating conditions. This variability limits the choice of controller parameters to guarantee stability under all conditions, which in turn limits the controller speed. Furthermore, the controller speed itself depends on the operating point and is therefore subject to significant variance. This variance also makes the controller speed highly dependent on the operating point.

[0010] "Controller speed" refers to the ability of a controller to respond to changes in the input signal or system conditions and to make appropriate adjustments to achieve the desired behavior.

[0011] The object of the present invention is therefore to create an improved power control for an amplifier unit of an RF generator in order to enable modern plasma processes to be operated safely and reliably, in particular independently of the operating point, and reproducibly.

[0012] The problem is solved by the method for power control of an amplifier unit in an RF generator for a plasma process according to claim 1. Claims 2 to 18 describe advantageous embodiments of the method. Claim 19 describes an RF generator with such a novel power control. Claim 20 describes a plasma generation system with such an RF generator.

[0013] This method is used to control the power of an amplifier unit in an RF generator for a plasma process. The power control unit includes a regulator. This method can also be referred to as a power control method.

[0014] In a first process step, a model of the amplifier unit, particularly a multidimensional one, is used. The model comprises at least one, preferably two, and especially preferably more than two, of the following input variables:

[0015] - Input power of an RF signal supplied to the amplifier unit;

[0016] - Input amplitude of the RF signal supplied to the amplifier unit;

[0017] - Operating frequency of the RF signal supplied to the amplifier unit;

[0018] - DC supply voltage of the amplifier unit;

[0019] - Load impedance or reflection coefficient of the amplifier unit;

[0020] - Temperature of the amplifier unit;

[0021] - Output power of the amplifier unit;

[0022] - DC input current of the amplifier unit;

[0023] - Ohmic DC input resistance of the amplifier unit;

[0024] - Power / amplitude across the absorber resistor of a power combiner, especially a 3dB hybrid coupler.

[0025] In a second process step, at least one of the following values ​​is continuously modified or adjusted:

[0026] - at least one control parameter of the control unit;

[0027] - an error signal that is fed to an input of the control unit; or

[0028] - at least one manipulated variable that the control unit outputs at an output; depending on at least one gain value for the amplifier unit, wherein the gain value is determined from the multidimensional model of the amplifier unit using a current value for the at least one, preferably two, particularly preferably more than two, input variables of the multidimensional model.

[0029] A "multidimensional model" refers to a model that can provide output variables for more than one input variable, where the

[0030] The input variables each have multiple values, and the output variable determines a single output value from the values ​​of the input variables. The input variables can be arranged in the form of a multidimensional matrix. The number of input variables determines the dimension of the model.

[0031] It is particularly advantageous that a multidimensional model for the amplifier unit, which can also be referred to as a characteristic model, is used for power control of the RF generator's amplifier unit. This multidimensional model allows the operating point, and in particular the current operating point, of the amplifier unit to be determined. At least one of the following input variables is used: input power, input amplitude, operating frequency, DC supply voltage, load impedance, reflection coefficient, amplifier unit output power, amplifier unit DC input current, amplifier unit DC input resistance, power / amplitude across the absorber resistor of a power combiner (especially a 3dB hybrid coupler), or temperature. If at least one of these input variables varies, the amplifier unit's operating point also varies.The operating point of the amplifier unit determines its gain when at least one specified input variable is present. The gain value is either the amplifier unit's actual gain or a value derived from it. This knowledge of the gain value as a function of the current values ​​of the applied input variables can be used to positively influence the controller unit. This can be achieved by modifying a controller parameter of the controller unit, an error signal (for example, a differential signal) fed to the controller unit's input, or at least a manipulated variable output by the controller unit, based on the determined gain value. This allows the maximum or nearly the maximum control reserve to be utilized without affecting the controller unit's performance at other operating points of the amplifier unit, for example, when...If a different gain value is present, the system becomes unstable and / or begins to oscillate. In this case, it is ensured that even in the most critical case, e.g., at the operating point where the amplifier unit has the highest gain, no oscillation occurs, and that the controller unit achieves the fastest possible transient response at other operating points as well. Previously, the stability margin had to be set so that oscillation did not occur at those operating points where the amplifier unit had the highest gain. At other operating points with lower gain, the transient response was correspondingly significantly longer. The described method allows the stability margin to be reduced, and the controller unit operates more dynamically. The term "continuous" means that the modification can be performed in discrete-time steps.A time interval between these steps is preferably constant. A "current value" means that, for input variables in the form of measured values, such as input power, input amplitude, DC supply voltage, load impedance, or temperature, a measured value is used that was determined within a predetermined time interval prior to use. For example, the last measured value or one of the last measured values ​​can be used. For input variables that change rapidly, the time interval is shorter than for input variables that change less rapidly. An input variable that changes rapidly is, for example, the input amplitude of the RF signal or the impedance of the load. An input variable that changes less rapidly is, for example, the DC supply voltage of the amplifier unit or the temperature.For other input variables, such as frequency, the current value, which is stored, for example, in a control unit or a storage unit, can be used.

[0032] One aspect of the continuous modification is that the speed at which the controller unit adjusts to a setpoint is independent of the current operating point, and in particular, independent of the gain value of the amplifier unit at the current inputs. The speed at which the controller unit adjusts to a setpoint is therefore always approximately the same. At the very least, the settling times are much more similar than without this method.

[0033] In one aspect, the amplifier unit is a transistor amplifier. Transistor amplifiers can amplify low-power signals while maintaining good linearity and bandwidth, which is particularly important in RF applications. Transistor amplifiers can also operate very efficiently.

[0034] In one aspect, the amplifier unit comprises exactly one or more transistors.

[0035] In one aspect, the multiple transistors are arranged in parallel and / or in series with each other.

[0036] One or more transistors can be implemented as MOSFETs, particularly LDMOS. MOSFETs offer high switching capability, can be used with currents of 10 A or more, and enable the amplification and / or switching of voltages of 500 V or more. When MOSFETs are used, an antiparallel diode can be provided for each MOSFET, especially in parallel with each MOSFET. The transistors can be silicon carbide (SiC) or gallium nitride (GaN) MOSFETs, which are suitable for fast switching voltages of 500 V or more with high voltage rise and fall times (15 kV / ps or more) as well as for high currents (10 A or more) with high current rise and fall times (10 A / ps or more). An advantage of LDMOS (Laterally Diffused Metal-Oxide Semiconductor) in amplifier units is its high efficiency and robustness at high power levels.LDMOS transistors offer excellent linearization and efficiency, making them ideal for RF applications.

[0037] In one aspect, the control unit regulates the amplifier unit in such a way that the amplifier unit outputs an actual power value that corresponds to a specified target power value or deviates from it by less than a threshold value.

[0038] In one aspect, the RF signal is solely a CW signal. "CW signal" stands for "Continuous Wave Signal." This is a signal that is not pulsed and maintained at a constant value. A CW signal requires a very narrow bandwidth, which makes the system more efficient and causes less interference.

[0039] In one aspect, the RF signal is free of digital modulation. Digital modulation refers to modulation with jumps in the output values. An RF signal free of digital modulation requires a very narrow bandwidth, which makes the system more efficient and causes less interference.

[0040] In one aspect, the multidimensional model outputs the gain value for the amplifier unit, a ratio relative to a reference operating point, or a quantity derived from the gain value after input of at least one, preferably two, and particularly preferably more than two, input variables. In this case, it is particularly advantageous that the multidimensional model determines an updated value for the gain value when a new value for an input variable is supplied.

[0041] In one aspect, the input variables are the DC supply voltage of the amplifier unit and the operating frequency of the RF signal fed to the amplifier unit. Both input variables can be determined, for example, by a suitable measuring circuit. These two input variables are particularly important because a change in these input variables leads to a, for example, larger shift in the operating point and thus to a change in the gain.

[0042] In one aspect, the input variables are the DC supply voltage of the amplifier unit, the operating frequency of the RF signal fed to the amplifier unit, and the load impedance or reflection coefficient of the amplifier unit. It is particularly advantageous that three input variables are used in this case instead of two, because this allows the operating point and thus the gain value to be determined even more precisely. This enables the control unit to adjust to the setpoint more precisely and quickly.

[0043] Since impedance, especially load impedance, is a complex value, this already involves two input variables.

[0044] In one aspect, the amplification value is a scalar value. This can be generated particularly easily using the multidimensional model and used for continuous modification.

[0045] In one aspect, at least one control parameter of the control unit, or the error signal or the manipulated variable, is compressed or stretched by the gain value.

[0046] In one aspect, an inverse value, also called the "reciprocal," of the gain value is used in the "continuous modification" process step. This is particularly relevant when the gain value is the gain of the amplifier unit, which is typically specified in dB. When the inverse value is used, the controller unit perceives the amplifier unit's gain as constant, even when the input variables of the multidimensional model change. In other words, if the gain, and therefore the gain value, of the amplifier unit is particularly high at a specific operating point, values ​​such as controller parameters, error signals, and manipulated variables—which the controller unit operates with or outputs—can be made particularly small, thus compensating for the excessive gain.If, for example, the amplifier unit has a gain of 65 dB instead of 60 dB at the current operating point, the controller unit would initially output 5 dB too much. In this case, the inverse of 65 dB is used for the ongoing modification. The controller unit therefore knows that the amplifier unit has a gain of 5 dB above the nominal value at this operating point. Consequently, adjusted control signals are output. The controller speed can then be adjusted accordingly. This can be achieved by changing the control signal, the error signal, or the controller parameters.

[0047] In one aspect, at least one control parameter of the control unit, or the error signal fed to an input of the control unit, or at least one manipulated variable output by the control unit, is modified based on the gain value or the inverse of the gain value, preferably multiplied by it. This modification, preferably the multiplication, is preferably performed for each new iteration of the control unit.

[0048] One possible alternative to multiplication could be, for example, polynomial modification or modification using a look-up table.

[0049] In one aspect, the control unit is a PID controller, with at least one of the controller parameters being Kp, Ki or KD.

[0050] In one aspect, the control unit is a PID controller, where at least one of the controller parameters is bO, b1 , ..., bn and aO, a1 , ..., am.

[0051] Controller parameters are specific settings within a controller that determine its behavior and response to changes in the system. For example, the most common controller parameters in a PID controller are:

[0052] - Kp, the proportional value that determines the ratio of the instantaneous error feedback;

[0053] - Ki, the integral value that influences the controller's response to the cumulative error over time;

[0054] -KD, the differential value that responds to the rate of error change.

[0055] In other controller topologies, controller parameters have similar functions to control system behavior. In a discrete-time (DTF) controller, the controller parameters consist of the coefficients of the numerator and denominator polynomials of the transfer function in the discrete-time domain. The numerator coefficients, typically denoted as (b0, b1, ..., bn), determine the influence of current and past input values ​​on the current output. The denominator coefficients, usually denoted as (a0, a1, ..., am), influence how current and past outputs control system behavior. These parameters can be important for defining the controller's dynamic response to the input signal and contribute to the overall system stability and performance.

[0056] At least one, several or all of these controller parameters are then modified based on the gain value or the inverse of the gain value, preferably multiplied by these.

[0057] In one aspect, the error signal arises from the difference between a predefined power setpoint for the RF signal and a measured actual power value of the RF signal. The result of this difference calculation is modified based on the inverse value of the gain, preferably by multiplication, or both the power setpoint and the actual power value are modified based on the inverse value of the gain, preferably by multiplication. In this case, the modification can be performed particularly easily because the control unit comprises precisely one input variable, specifically the difference. The advantage resulting from the "compensation" of the error signal is that jumps in the gain in the multidimensional model, which are inevitably traversed, for example, when the DC supply voltage is adjusted, are smoothed out by the time constant of the control unit.

[0058] In one aspect, the at least one manipulated variable that the control unit outputs is modified based on the inverse value of the gain, preferably by multiplying it. This at least one manipulated variable is, for example, the input power of the RF signal supplied to the amplifier unit, and / or the input amplitude of the RF signal supplied to the amplifier unit, and / or the DC supply voltage of the amplifier unit.

[0059] In one aspect, exactly one control variable, which the control unit outputs at an output, is modified based on the inverse value of the gain value, preferably multiplied by it.

[0060] In one aspect, the most rapidly changeable control variable, which the control unit outputs, is modified based on the inverse value of the gain, preferably by multiplying it. This is particularly relevant to the input amplitude of the RF signal supplied to the amplifier unit. This input amplitude can be changed in the picosecond range. The DC supply voltage can be changed more slowly, particularly in the millisecond range.

[0061] In one aspect, the multidimensional model consists of a lookup table or a mathematical function. This lookup table or mathematical function can be stored in a memory unit. At runtime, this lookup table or mathematical function can be loaded into an FPGA, microcontroller, or DSP. A corresponding gain value can be determined from the lookup table for various input variables. The same applies to the mathematical function, which could, for example, be a polynomial.

[0062] In one aspect, the look-up table or the mathematical function is created by measuring one or more reference amplifier units. The corresponding input variables can be iteratively changed using nested for loops, and the corresponding gain value can be measured for each setting. In another aspect, the look-up table directly stores an inverse value for the gain of the amplifier unit, or a value representing this. This has the advantage that, in this case, no inverse value needs to be calculated at runtime by the controller unit.

[0063] In one aspect, values ​​that lie between two gain values ​​in the look-up table are interpolated. This eliminates the need for an extremely fine-grained look-up table, which would require additional memory.

[0064] In one aspect, for each pass of the control unit to generate at least one manipulated variable, the latest values ​​for at least one, several or all input variables are used to determine the at least one gain value.

[0065] In one aspect, the amplifier unit is a class A, B, AB, C, D, E, F or F amplifier. -1 or a combination of these amplifier classes, e.g. operation in AB in one amplifier range and C or D in another amplifier range.

[0066] Class A amplifiers are characterized by excellent linearity and signal quality.

[0067] Class B amplifiers offer higher efficiency by conducting current only during half of the signal cycle, which reduces heat generation.

[0068] Class AB amplifiers are a compromise between classes A and B, offering better efficiency than class A and lower distortion than class B.

[0069] Class C amplifiers are very efficient because they do not conduct for more than half of the input signal.

[0070] Class D amplifiers often achieve even higher efficiency through the use of switching techniques.

[0071] Class E amplifiers also offer high efficiency in RF applications through resonance and switching time optimization. Class F amplifiers increase efficiency and output power by utilizing harmonic oscillations and are suitable for extremely high-frequency applications.

[0072] Class F' 1, also called “Class F inverse”, represents an extended version of Class F, which is even more efficient through optimized control of harmonics and waveguide impedances, improving both efficiency and gain.

[0073] In one aspect, jumps in the gain value from the multidimensional model that exceed a threshold are not addressed by modification, because the power control could otherwise be negatively affected by the dead time of the controller unit. Specifically, jumps occurring within a time window smaller than a predetermined threshold are disregarded after at least one controller parameter of the controller unit, the error signal, or at least one manipulated variable has been adjusted in the modification process step. For example, several values ​​from the LUT can be filtered by a low-pass filter after being read. This would reduce jumps accordingly. In other words, the values ​​are read from the LUT, and the gain is only allowed to change more slowly than a defined time.

[0074] Also disclosed is a method for power control of an amplifier unit in an RF generator for a plasma process with a control unit, comprising the following process steps:

[0075] - Use of data from a particularly multidimensional model of an amplifier characteristic curve or its inverse, wherein the multidimensional model comprises at least one, preferably two, particularly preferably more than two, subsequent input variables:

[0076] • Input power of an RF signal supplied to the amplifier unit;

[0077] • Input amplitude of the RF signal supplied to the amplifier unit;

[0078] • Operating frequency of the RF signal supplied to the amplifier unit; • DC supply voltage of the amplifier unit;

[0079] • Load impedance or reflection coefficient of the amplifier unit;

[0080] • Output power of the amplifier unit;

[0081] • DC input current of the amplifier unit;

[0082] • Ohmic DC input resistance of the amplifier unit;

[0083] • Power / amplitude across the absorber resistor of a power combiner, especially a 3dB hybrid coupler;

[0084] • Temperature of the amplifier unit.

[0085] - Continuous adjustment of the control parameters of the controller unit or an error signal supplied to the controller unit, or a manipulated variable output by the controller unit, depending on the current operating points and the data of the multidimensional model.

[0086] In one aspect, the RF generator can comprise a signal generator, an amplifier unit, a regulator unit, a control device, and a storage unit. The RF generator is designed to generate a power-controlled RF signal for a plasma process and therefore includes the following features:

[0087] - The control unit is designed to receive a power setpoint for the RF signal and an actual power value for the RF signal and to control the signal generator and amplifier unit in such a way that the actual power value of the generated RF signal corresponds to the power setpoint of the RF signal or deviates from it by less than a threshold value.

[0088] - the storage unit comprises a multidimensional model of the amplifier unit, wherein the model comprises at least one, preferably two, particularly preferably more than two, of the following input variables:

[0089] • Input power of an RF signal supplied to the amplifier unit;

[0090] • Input amplitude of the RF signal supplied to the amplifier unit;

[0091] • Operating frequency of the RF signal supplied to the amplifier unit;

[0092] • DC supply voltage of the amplifier unit; • Load impedance or reflection coefficient of the amplifier unit;

[0093] • Output power of the amplifier unit;

[0094] • DC input current of the amplifier unit;

[0095] • Ohmic DC input resistance of the amplifier unit;

[0096] • Power / amplitude across the absorber resistor of a power combiner, especially a 3dB hybrid coupler;

[0097] • Temperature of the amplifier unit.

[0098] - the control device is designed to continuously determine an amplification value for the amplifier unit from the multidimensional model of the amplifier unit using a current value for at least one, preferably two, particularly preferably more than two, of the input variables of the multidimensional model.

[0099] - The control unit is designed to continuously adjust at least one of the following values ​​depending on the continuously determined gain value:

[0100] • at least one controller parameter of the controller unit or

[0101] • an error signal that can be fed to an input of the control unit or

[0102] • at least one control variable that the control unit outputs at an output.

[0103] This allows not only for particularly rapid adjustment to a desired power setpoint, but also for maintaining or quickly regaining the desired power setpoint even if the operating point of the amplifier unit changes, for example, due to a change in load impedance or reflection coefficient. This underscores the suitability of the RF generator for modern plasma processes.

[0104] The plasma generation system comprises the RF generator described above. Furthermore, the plasma generation system includes an impedance matching device with an input and an output connection. The input connection of the impedance matching device is connected to an output connection of the RF generator via a first cable connection. The output connection of the impedance matching device can be connected to a load, in particular an electrode in a plasma chamber, via a second cable connection.

[0105] In one aspect, the impedance matching device is located in a separate housing and at a distance from the RF generator. Alternatively, the impedance matching device can also be integrated into the housing of the RF generator.

[0106] The following description of the development is purely exemplary and refers to the drawings. They show:

[0107] Figure 1: an embodiment of the plasma generation system with a

[0108] RF generator;

[0109] Figures 2, 3, 4: various embodiments describing the use of a multidimensional model for an amplifier unit of the RF generator to achieve optimized power control with a controller unit;

[0110] Figures 5A, 5B:

[0111] Differences between a previous benefit scheme and the benefit scheme described here; and

[0112] Figure 6: a flowchart describing the procedure for power control of an amplifier unit in the RF generator.

[0113] Figure 1 shows a plasma generation system 100 comprising an RF generator 1, a central control device 50, an impedance matching circuit 60, and a load, in particular a load 70, especially in the form of a plasma chamber. The RF generator 1 is configured to provide an RF signal, in particular in the form of a high-frequency signal, with a nominal power PNent and a frequency fo, and to output it at an output terminal 2. The impedance matching circuit 60 comprises an input terminal 60a, wherein the RF generator 1 is connected to the input terminal 60a via a first cable connection 61 at its output terminal 2. The impedance matching circuit 60 further comprises an output terminal 60b. The output terminal 60b is connected to the at least one load 70 via a second cable connection 62.The first and / or second cable connection 61, 62 can comprise one or more cables, for example, connected in series and / or in parallel. Coaxial cables are preferably used.

[0114] The consumer 70, in particular the plasma chamber, comprises at least one electrode 71 for generating a plasma 72. The electrode 71 is connected to the output terminal 60b of the impedance matching circuit 60. In this embodiment, a camera system 73 is also arranged in the plasma chamber, which is configured to observe the plasma 72.

[0115] The central control device 50 is preferably a processor and / or a programmable logic device, e.g., an FPGA and / or microcontroller and / or ASIC, which is programmed according to its suitability or design. The central control device 50 may also include a storage device, among other things.

[0116] The central control device 50 is configured to control the RF generator 1, in particular to activate or deactivate it. Additionally or alternatively, the central control device 50 is also configured to change the power and / or amplitude and / or frequency of the RF signal by appropriately controlling the RF generator 1.

[0117] The central control device 50 is preferably also configured to control the impedance matching circuit 60. In particular, the central control device 50 is configured to change the transformation ratio within the impedance matching circuit 60 and / or to specify an impedance at the output terminal 60b. Additionally or alternatively, the central control device 50 is configured to specify the impedance at the input terminal 60a, which acts on the RF generator 1.

[0118] The plasma generation system 100 also includes a first measuring unit 80. The first measuring unit 80 is preferably arranged between the RF generator 1 and the impedance matching circuit 60. The first measuring unit 80 is configured, for example, to measure power transmitted from the RF generator 1 towards the impedance matching circuit 60 and power reflected back towards the RF generator 1. Alternatively, the first measuring unit 80 can also be configured to measure the impedance at the input terminal 60a of the impedance matching circuit 60.

[0119] For this purpose, the first measuring unit 80 includes, for example, a directional coupler unit. Using the directional coupler unit, the first measuring unit 80 can measure the power of a forward and reversed high-frequency signal on the first cable connection 61 in order to calculate the respective power or impedance at the input terminal 60a. Alternatively, the first measuring unit 80 can also include a current sensor and a voltage sensor. The central control device 50 is configured to calculate the respective power or impedance at the input terminal 60a, as seen by the RF generator 1, based on the measurement result from the directional coupler unit or the current and voltage sensors.

[0120] The plasma generation system 100 preferably also includes a second measuring unit 81. The second measuring unit 81 is preferably arranged between the impedance matching circuit 60 and the load 70. The second measuring unit 81 is configured, for example, to measure the power transmitted from the impedance matching circuit 60 to the load 70 and to measure the power reflected back towards the impedance matching circuit 60. Alternatively, the second measuring unit 81 can also be configured to measure the impedance at the output terminal 60b of the impedance matching circuit 60.

[0121] For this purpose, the second measuring unit 81 includes, for example, a directional coupler unit. Using the directional coupler unit, the second measuring unit 81 can measure the power of a forward and reverse high-frequency signal on the second cable connection 62 in order to calculate the respective power or impedance at the output terminal 60b. Alternatively, the second measuring unit 81 can also include a current sensor and a voltage sensor. The central control device 50 is configured to calculate, based on the measurement result of the directional coupler unit or the current and voltage sensors, the respective power or impedance at the output terminal 60b that the consumer 70 sees.

[0122] The first and / or second measuring unit 80, 81 can also be arranged within the impedance matching circuit 60. The first measuring unit 80 is located at the input terminal 60a and the second measuring unit 81 at the output terminal 60b.

[0123] The plasma generation system 100 preferably includes an operating unit 90, which has a visualization device. The operating unit 90 is preferably a screen, in particular a touchscreen. In addition to a screen, the operating unit 90 may also include input devices such as a keyboard and / or mouse. The operating unit 90 may also be a web server that provides data and receives user input. The central control device 50 is configured to receive input from the operating unit 90.

[0124] The central control device 50 is preferably configured to receive setpoint specifications, for example for the power of the RF signal, from the operating unit 93. Additionally or alternatively, the frequency of the RF signal can be received from the operating unit 90. A desired impedance at the output terminal 60b of the impedance matching circuit 60 can also be received via the operating unit 90. From this, corresponding control variables for the RF generator 1 and control data for the impedance matching circuit 60 can be generated and transmitted to it.

[0125] Preferably, the control unit 90 is directly connected to the RF generator 1. The RF generator 1 comprises a control unit 3 and a regulator unit 4. Furthermore, the RF generator 1 comprises an amplifier unit 5, a signal generator 6, and a power supply unit 7. The amplifier unit 5 can have one or more power amplifiers, in particular transistor amplifiers.

[0126] The signal generator 6 is configured to generate an RF signal to be amplified and to supply it to the amplifier unit 5. The amplifier unit 5 is configured to amplify the RF signal to be amplified and to output it at the output terminal 2 of the RF generator 1. The power supply unit 7 is configured to supply the amplifier unit 5 with electrical energy. For this purpose, the amplifier unit 5 provides, in particular, a DC supply voltage 8.

[0127] The control unit 4 is configured to receive a setpoint, for example, for the power of the high-frequency signal, and to generate and output corresponding manipulated variables. Preferably, the control unit 4 receives a power setpoint 9 for the RF signal and an actual power value 10 for the RF signal, or directly receives the corresponding difference, here the error signal 11, between these two power values ​​9 and 10. In one aspect, the control unit can receive an actual value and calculate an error signal, or it can work directly with the actual value. The control unit 4 is further configured to generate manipulated variables 12, in particular in the form of a frequency for the RF signal, an amplitude for the RF signal supplied to the amplifier unit 5, and / or an input power 13 for the RF signal supplied to the amplifier unit 5, and to transmit them to the signal generator 6.Additionally or alternatively, the control unit 4 is configured to generate a manipulated variable, in particular in the form of a DC supply voltage 8 for the amplifier unit 5, and to transmit it to the power supply unit 7. This ensures that the amplifier unit 5 operates at a specific operating point, in which the amplifier unit 5 has a predetermined gain value 14. This allows the control unit 4 to regulate the RF signal at the output of the amplifier unit 5 particularly quickly to a desired power setpoint 9.

[0128] The controller unit 4 can be integrated into the control unit 3. Alternatively, the controller unit 4 can be implemented separately from the control unit 3, for example, in a separate processor, microcontroller, or FPGA. The central control device 50 can also be arranged separately from the control unit 3 or integrated into the control unit 3.

[0129] The power setpoint 9 for the RF signal can be specified by the central control unit 50 or by the operating unit 90. The power setpoint 9 for the RF signal can also be represented as a waveform used for a specific plasma process. A waveform comprises a time-dependent profile of the RF signal. This waveform, which in particular includes a power setpoint 9, is replicated as accurately as possible by the control unit 4 so that the RF signal at the output of the amplifier unit 5 corresponds approximately to the RF signal of the waveform or deviates from it by less than a specified threshold. This ensures that reproducible results are achieved in the plasma chamber 70. The waveform can also be loaded from a storage unit (not shown) by the control unit 3.

[0130] Figures 2, 3, and 4 illustrate the power control method in more detail. These figures show the use of a multidimensional model 15 for the amplifier unit 5 of the RF generator 1 to achieve optimized power control with the controller unit 4. In this embodiment, the multidimensional model 15 of the amplifier unit 5 comprises two input variables. A first input variable is the magnitude of the DC supply voltage 8 of the amplifier unit 5. A second input variable is the magnitude of the input power 13 of the RF signal supplied to the amplifier unit 5, which is generated by the signal generator 6. Other input variables are conceivable, such as an input amplitude of the RF signal supplied to the amplifier unit 5, an operating frequency of the RF signal supplied to the amplifier unit, a load impedance and / or a reflection coefficient of the amplifier unit 5, and / or a temperature of the amplifier unit 5.In this embodiment, the two input variables are plotted on the X-axis and the Y-axis. The Z-axis shows the gain value 14, which the amplifier unit 5 generates when the two corresponding input variables are present.

[0131] Figure 2 shows a control computer comprising the controller unit 4 and the multidimensional model 15. The control computer receives the power setpoint 9 and the actual power value 10 of the RF signal. The difference between the power setpoint 9 and the actual power value 10 is then calculated, and the result is output as an error signal 11. This error signal 11 is fed to the controller unit 4. Depending on the magnitude of the error signal 11, the controller unit 4 calculates the value for at least one manipulated variable 12 and outputs this manipulated variable 12. The manipulated variable 12 represents the adjustable values ​​of the RF generator 1.This includes, for example, the input power 13 of the RF signal supplied to the amplifier unit 5, and / or the input amplitude of the RF signal supplied to the amplifier unit 5, and / or the operating frequency of the RF signal supplied to the amplifier unit 5, and / or the DC supply voltage 8 of the amplifier unit 5. In the embodiment of the control device shown in Figure 2, this manipulated variable is continuously modified as a function of at least one gain value 14 for the amplifier unit 5. This gain value 14 is determined from the multidimensional model 15 for the amplifier unit 5 using a current value for the input power 13 of the RF signal supplied to the amplifier unit 5 and the DC supply voltage 8 of the amplifier unit 5. In this case, the multidimensional model 15 is a lookup table.The gain value 14 is a scalar value and can be multiplied by the manipulated variable 12. Specifically, the reciprocal of the gain value 14 is multiplied by the manipulated variable 12. This multiplication can occur with all manipulated variables 12, or with one manipulated variable, or with selected manipulated variables 12. Preferably, the manipulated variable 12 that can change most rapidly is multiplied by the inverse of the gain value 14. More generally, instead of multiplication, a modification based on the reciprocal of the gain value 14 can also be performed.

[0132] Figure 3 shows another embodiment of the control computer. In contrast to Figure 2, in the embodiment shown in Figure 3, the error signal 11, which is fed to the input of the control unit 4, is continuously modified. This error signal 11 is also modified depending on at least one gain value 14, which is taken from the multidimensional model 15 for the amplifier unit 5 for the corresponding input variable. The input variables here are again the input power 13 of the RF signal fed to the amplifier unit 5 and the DC supply voltage 8 of the amplifier unit 5. Other input variables could also be used, such as the input amplitude of the RF signal fed to the amplifier unit 5, the operating frequency of the RF signal fed to the amplifier unit 5, the load impedance or reflection coefficient of the amplifier unit 5, or the temperature of the amplifier unit 5.A particular advantage here is that only one quantity needs to be continuously modified. In principle, it would also be conceivable to modify the error signal 11 by directly multiplying the target power value 9 and the actual power value 10 by the inverse value of the gain value 14. More generally, a modification based on the inverse of the gain value 14 is also possible. Multiplication is a special case of this.

[0133] Figure 4 shows another embodiment of the control computer. In contrast to Figure 2, in the embodiment shown in Figure 3, at least one control parameter of the controller unit 4 is modified. This modification again takes place depending on at least one gain value 14 for the amplifier unit 5, wherein the gain value 14 is derived from the multidimensional model 15 of the amplifier unit 5 using a current value for at least one, preferably two, particularly preferably more than two, of the input variables of the multidimensional model 15. The input variables here are again the input power 13 of the RF signal supplied to the amplifier unit 5 and the DC supply voltage 8 of the amplifier unit 5.Other input variables can also be used, such as the input amplitude of the RF signal supplied to the amplifier unit 5, the operating frequency of the RF signal supplied to the amplifier unit 5, the load impedance or reflection coefficient of the amplifier unit 5, or the temperature of the amplifier unit 5. Preferably, the controller unit 4 is a PID controller. In this case, the at least one controller parameter to be modified is K. P , K, or K D Here too, at least one control parameter is preferably multiplied by the reciprocal of the gain value 14, which results from the multidimensional model 15 for the respective current input variables. More generally, a modification based on the reciprocal of the gain value 14 can also be performed. Multiplication is a special case of this.

[0134] Figure 5A shows the transient response of a conventional control computer, and Figure 5b shows the transient response of a control computer using the power control method described here. In both figures, the power is controlled to a setpoint 9, which corresponds to 10% of the maximum power that the amplifier unit 5 can deliver. Figure 5A shows that the power setpoint 9 is reached after approximately 85 ps. Figure 5B shows that the power setpoint 9 is reached after approximately 20 ps. Furthermore, the power control method described here ensures that no oscillation or overshoot occurs. The power control method described here is therefore significantly more efficient than that of a conventional control computer.

[0135] Figure 6 shows a flowchart describing the procedure for power control of an amplifier unit 5 in the RF generator 1.

[0136] The method serves to control the power of the amplifier unit 5 in the RF generator 1 for a plasma process and utilizes a controller unit 4. The description of the method begins at "Start" and ends at "End". In a first method step Si, a multidimensional model 15 of the amplifier unit 5 is used. The multidimensional model 15 comprises at least one, preferably two, and particularly preferably more than two, of the following input variables:

[0137] - Input power 13 of one of the RF signals supplied to the amplifier unit 5;

[0138] - Input amplitude of the RF signal supplied to amplifier unit 5;

[0139] - Operating frequency of the RF signal supplied to amplifier unit 5;

[0140] - DC supply voltage 8 of the amplifier unit 5;

[0141] - Load impedance or reflection coefficient of amplifier unit 5; and / or

[0142] - Output power of amplifier unit 5;

[0143] - DC input current of amplifier unit 5;

[0144] - Ohmic DC input resistance of amplifier unit 5;

[0145] - Power / amplitude across the absorber resistor of a power combiner, especially a 3dB hybrid coupler;

[0146] - Temperature of amplifier unit 5.

[0147] In a second process step S2, at least one of the following values ​​is continuously modified or adjusted:

[0148] - at least one controller parameter of controller unit 4 or

[0149] - an error signal 11 which is supplied to an input of the control unit 4, or - at least a manipulated variable 12 which the control unit 4 outputs; depending on at least one gain value 14 for the amplifier unit 5, which is determined from the multidimensional model 15 of the amplifier unit 5 using a current value for at least one, preferably two, particularly preferably more than two, of the input variables of the multidimensional model 15.

[0150] The development is not limited to the described embodiments. Within the scope of the development, all described and / or drawn features can be combined with one another as desired, unless otherwise specified.

Claims

28 Claims 1. Method for power control of an amplifier unit (5) in an RF generator (1) for a plasma process with a control unit (4), comprising the following method steps: - Using (Si) a multidimensional model (15) of the amplifier unit (5), wherein the multidimensional model (15) comprises at least one, preferably two, particularly preferably more than two, of the following input variables: • Input power (13) of an RF signal supplied to the amplifier unit (5); • Input amplitude of the RF signal supplied to the amplifier unit (5); • Operating frequency of the RF signal supplied to the amplifier unit (5); • DC supply voltage (8) of the amplifier unit (5); • Load impedance or reflection coefficient of the amplifier unit (5); • Output power of the amplifier unit (5); • DC input current of the amplifier unit (5); • Ohmic DC input resistance of the amplifier unit (5); • Power / amplitude across the absorber resistor of a power combiner, especially a 3dB hybrid coupler; • Temperature of the amplifier unit (5); - Continuous modification (S2) of at least one of the following values: • at least one controller parameter of the controller unit (4); or • an error signal (11) which is fed to an input of the control unit (4); or • at least one manipulated variable (12) that the control unit (4) outputs at an output; depending on at least one gain value (14) for the amplifier unit (5), wherein the gain value (14) is derived from the multidimensional model (15) of the amplifier unit (5) is determined using a current value for at least one, preferably two, particularly preferably more than two, of the input variables of the multidimensional model (15).

2. Method for power control according to claim 1, characterized in that the multidimensional model (15) outputs the gain value (14) for the amplifier unit (5) and / or a ratio relative to a reference operating point or a quantity derived from the gain value after input for at least one, preferably two, particularly preferably more than two, of the input variables.

3. Method for power control according to claim 1 or 2, characterized in that the DC supply voltage (8) of the amplifier unit (5) and the operating frequency of the RF signal supplied to the amplifier unit (5) are used as input variables.

4. Method for power control according to claim 1 or 2, characterized in that the DC supply voltage (8) of the amplifier unit (5) and the operating frequency of the RF signal supplied to the amplifier unit (5) and the load impedance or the reflection factor of the amplifier unit (5) are used as input variables.

5. Method for power control according to one of the preceding claims, characterized in that the gain value (14), the ratio relative to a reference operating point or the quantity derived from the gain value is each a scalar value.

6. Method for power control according to one of the preceding claims, characterized in that an inverse value of the gain value (14) is used in the process step “continuous modification” (S2).

7. Method for performance regulation according to one of the preceding claims, characterized in that: a) the at least one control parameter of the control unit (4); or b) the error signal (11) which is supplied to an input of the control unit (4); or c) the at least one manipulated variable (12) which the control unit (4) outputs at an output; is modified on the basis of the gain value (14) or the inverse value of the gain value (14), preferably multiplied by it.

8. A method for power control according to one of the preceding claims, characterized in that the controller unit (4) is a PID controller and that the at least one controller parameter is Kp, Ki or KD; or that the controller unit (4) is a DTF controller and that the at least one controller parameter is bO, b1 , ..., bn and aO, a1 , ..., am.

9. A method for power control according to one of the preceding claims, characterized in that the error signal (11) results from the difference between a power setpoint (9) for the RF signal and an actual power value (10) of the RF signal, wherein: a) the result is modified on the basis of the inverse value of the gain value (14), preferably multiplied by it; or b) both the power setpoint (9) and the actual power value (10) are modified on the basis of the inverse value of the gain value (14), preferably multiplied by it.

10. Method for power control according to one of the preceding claims, characterized in that the at least one manipulated variable (12) which the control unit (4) outputs at an output is: a) input power (13) of the RF signal supplied to the amplifier unit (5); b) Input amplitude of the RF signal supplied to the amplifier unit (5); and / or c) DC supply voltage (8) of the amplifier unit (5); specified and accordingly based on the inverse value of the gain value (14) modified, preferably multiplied with it.

11. Method for power control according to one of the preceding claims, characterized in that exactly one manipulated variable (12) which the control unit (4) outputs at an output is modified on the basis of the inverse value of the gain value (14), preferably multiplied by it.

12. Method for power control according to one of the preceding claims, characterized in that the most rapidly changeable manipulated variable (12) which the control unit (4) outputs at an output is modified on the basis of the inverse value of the gain value (14), preferably multiplied by it.

13. Method for performance control according to one of the preceding claims, characterized in that the multidimensional model is (15) is a look-up table or a mathematical function.

14. Method for power control according to claim 13, characterized in that an inverse value for the gain value (14) for the amplifier unit (5) is stored in the lookup table.

15. Method for power control according to claim 13 or 14, characterized in that values ​​which lie between two gain values ​​(14) in the lookup table are interpolated.

16. Method for power control according to one of the preceding claims, characterized in that for each pass of the control unit (4) to generate the at least one manipulated variable (12), the latest values ​​for at least one, several or all input variables are used to determine the at least one gain value (14). 32 17. Method for power control according to one of the preceding claims, characterized in that at least one input variable is an averaged input variable, wherein a plurality of measured values ​​for this input variable are averaged.

18. Method for power control according to one of the preceding claims, characterized in that the amplifier unit (5) is an amplifier of class A, B, AB, C, D, E or F, F' 1 it.

19. RF generator (1) comprising a signal generator (6), an amplifier unit (5), a control unit (4), a control device (3) and a storage unit for generating a power-controlled RF signal for a plasma process with the following features: - the control unit (4) is designed to receive a power setpoint (9) for the RF signal and an actual power value (10) of the RF signal and to control the signal generator (6) and the amplifier unit (5) such that the actual power value (10) of the generated RF signal corresponds to the power setpoint (9) of the RF signal or deviates from it by less than a threshold value; - the storage unit comprises a multidimensional model (15) of the amplifier unit (5), wherein the multidimensional model (15) comprises at least one, preferably two, particularly preferably more than two, of the following input quantities: • Input power (13) of an RF signal supplied to the amplifier unit (5); • Input amplitude of the RF signal supplied to the amplifier unit (5); • Operating frequency of the RF signal supplied to the amplifier unit (5); • DC supply voltage (8) of the amplifier unit (5); • Load impedance or reflection coefficient of the amplifier unit (5); • Output power of the amplifier unit (5); 33 • DC input current of the amplifier unit (5); • Ohmic DC input resistance of the amplifier unit (5); • Power / amplitude across the absorber resistor of a power combiner, especially a 3dB hybrid coupler; • Temperature of the amplifier unit (5); - the control device (3) is designed to continuously determine an amplification value (14) for the amplifier unit (5) from the multidimensional model (15) of the amplifier unit (5) using a current value for at least one, preferably two, particularly preferably more than two, of the input variables of the multidimensional model (15); - the control device (3) is designed to continuously adjust at least one of the following values ​​depending on the continuously determined gain value (14): • at least one controller parameter of the controller unit (4); • an error signal (11) that can be fed to an input of the control unit (4); • at least one manipulated variable (12) that the control unit (4) outputs at an output.

20. Plasma generation system (100) with an RF generator (1) constructed according to claim 19, characterized by the following features: - an impedance matching device (60) with an input terminal (60a) and an output terminal (60b) is provided; - the input terminal (60a) of the impedance matching device (60) is connected to an output terminal (2) of the RF generator (1) via a first cable connection (61); 34 - the output terminal (60b) of the impedance matching device (60) can be connected to a consumer (70), in particular an electrode (71) in a plasma chamber, via a second cable connection (62).

Images