TWI727005B - Systems and methods for reducing power reflected towards a higher frequency rf generator during a period of a lower rf generator and for using a relationship to reduce reflected power - Google Patents

Abstract

Systems and methods for reducing reflected towards a higher frequency radio frequency (RF) generator during a period of a lower frequency RF generator and for using a relationship to reduce reflected power are described. By tuning the higher frequency RF generator during the period of the lower frequency RF generator, precise control of the higher frequency RF generator is achieved for reducing power reflected towards the higher frequency RF generator. Moreover, by using the relationship to reduce the reflected power, time is saved during processing of a wafer.

Description

System and method for reducing reflected power to higher frequency RF generator during lower frequency radio frequency generator and using a relationship to reduce reflected power

The embodiment of the present invention relates to a system and method for reducing the power reflected to a higher frequency radio frequency (RF) generator during one of the lower frequency radio frequency (RF) generators and using a relationship to reduce the reflected power.

The plasma system is used to control the plasma manufacturing process. A plasma system includes a complex radio frequency (RF) source, an impedance matching, and a plasma reactor. A work piece is placed in the plasma chamber, and then plasma is generated in the plasma chamber to process the work piece. It is important that the work pieces are handled in a similar or uniform manner. In order to process the work piece in a similar or uniform manner, it is important to modulate the complex RF source to match the impedance.

The embodiments described in the text are produced under this context.

Embodiments of the present invention provide devices, methods, and computer programs that reduce the power reflected to a higher frequency radio frequency (RF) generator during one of the lower frequency RF generators and use a relationship to reduce the reflected power. It should be understood that the embodiments of the present invention can be implemented in a variety of ways, such as a process, a device, a system, a piece of hardware, or a method on a computer-readable medium. Several embodiments will be described below.

In some embodiments, an RF frequency of an RF signal generated by a higher frequency RF generator changes during a period of an RF signal of a lower frequency RF generator. For example, a model system is used to determine various frequency values of the RF signal generated by the higher frequency RF generator and apply the various frequency values during a period of the RF signal generated by the lower frequency RF generator.

In several embodiments, a model system is used to modulate an impedance matching network in the presence of a complex load impedance variation generated by the RF signal generated by the lower frequency RF generator. For example, using the model system to calculate an optimal combined variable capacitance value and apply the optimal combined variable capacitance value during the period of the RF signal generated by the lower frequency RF generator.

In various embodiments, the model system is used to calculate modulation trajectories such as modulation polynomials, modulation relationships, and so on. The impedance matching network is pre-characterized instead of using the model system during a wafer processing. The impedance matching network is characterized by the following method: between a real part of the complex load impedance value and the complex load impedance value An optimal combined variable capacitance value is calculated on a two-dimensional grid of an imaginary part, where the complex load impedance value spans a desired operating space. Then, an optimal RF frequency is calculated on a three-dimensional grid of the real part of the complex load impedance value, the imaginary part of the complex load impedance value, and the complex variable capacitance value. Perform a first fitting on various optimal combined variable capacitance values and perform a second fitting on various optimal RF frequencies to obtain a complex polynomial function as a complex solution. An example of the first fitting is an optimal combined variable capacitance value=function (Re(Z Load), Im(Z Load)), where Zload is a load impedance value, and Re is a real value of the load impedance value. The part and Im are an imaginary part of the load impedance value. An example of the second fitting is an optimal RF frequency at a specific optimal combined variable capacitance value=function (Re(Z Load), Im(Z Load), optimal combined variable capacitance value).

Some advantages of the systems and methods described herein include determining the complex RF value during each RF cycle of the lower frequency RF generator to reduce the power reflected to the higher frequency RF generator. A model system is used to determine the complex RF value during each RF cycle of the lower frequency RF generator. The complex RF value is calculated from a complex parameter value, and the complex parameter value is calculated at an output location of the higher frequency RF generator during an RF period of the lower frequency RF generator. During an RF period of the lower frequency RF generator after the RF period of the calculation of the complex parameter value, the determined complex RF value is applied to the higher frequency RF generator. Applying the complex RF value accurately reduces the power reflected to the higher frequency RF generator during each RF cycle of the lower frequency RF generator.

Other advantages of the system and method described herein include not using a model system during wafer processing to determine a complex optimal RF value and/or a complex optimal combined variable capacitance value. The complex optimal RF value and/or the complex optimal combined variable capacitance value are determined in advance before wafer processing. During wafer processing, the processor takes the complex optimal RF value and/or the complex optimal combined variable capacitance value and applies the complex optimal RF value and/or the complex load impedance value determined by the model system. / Or the variable capacitance value of the optimal combination of the plural numbers. The pre-calculation of the complex optimal RF value and/or the complex optimal combined variable capacitance value saves time during wafer processing.

Other aspects will be apparent from the following detailed description with reference to the attached drawings.

The following embodiments illustrate the system and method for reducing the power reflected to the higher frequency radio frequency (RF) generator during one of the lower frequency radio frequency (RF) generators and using a relationship to reduce the reflected power. It should be understood that the embodiments of the present invention can be implemented without some or all of these specific details. In other cases, the known process operations are not described in detail so as not to unnecessarily obscure the embodiments of the present invention.

FIG. 1 is a diagram of an embodiment of a plasma system 100, which illustrates the use of a model system 102 to generate a complex load impedance ZL(P1) for a period P1 of an RF signal generated by an x megahertz (MHz) RF generator. The plasma system 100 includes an x MHz RF generator, a y MHz RF generator 104, an impedance matching network 106, and a plasma chamber 108. The plasma system 100 includes a host computer system 110, a driving component 112, and one or more connection mechanisms 114.

The plasma chamber 108 includes an upper electrode 116, a chuck 118, and a wafer W. The upper electrode 116 faces the chuck 118 and is grounded, such as coupled to a reference voltage, coupled to a zero voltage, coupled to a negative voltage, and so on. Examples of the chuck 118 include an electrostatic chuck (ESC) and a magnetic chuck. The lower electrode of the chuck 118 is made of metal such as anodized aluminum, aluminum alloy, and the like. In various embodiments, the lower electrode of the chuck 118 is a thin metal layer covered with a layer of ceramic. In addition, the upper electrode 116 is made of metal such as aluminum and aluminum alloy. In some embodiments, the upper electrode 116 is made of silicon. The position of the upper electrode 116 is opposite to the lower electrode of the chuck 118 and faces the lower electrode. The wafer W is placed on the upper surface 120 of the chuck 118 for processing, such as depositing material on the wafer W, or cleaning the wafer W, or etching the film deposited on the wafer W, or doping the wafer W , Or implanting ions on the wafer W, or generating a lithography pattern on the wafer W, or etching the wafer W, or sputtering the wafer W, or a combination thereof.

In some embodiments, the plasma chamber 108 is formed by a plurality of additional components, such as the upper electrode extension part surrounding the upper electrode 116, the lower electrode extension part surrounding the lower electrode of the chuck 118, the upper electrode 116 and The dielectric ring between the upper electrode extension, the dielectric ring between the lower electrode and the lower electrode extension is located at the edge of the upper electrode 116 and the chuck 118 to surround an area in the plasma chamber 108 where the plasma is formed. Restriction ring, etc.

The impedance matching network 106 includes one or more circuit elements, such as one or more inductors, or one or more capacitors, or one or more resistors, or a combination of two or more of the foregoing. For example, the impedance matching network 106 includes a series circuit including an inductor coupled in series with a capacitor. The impedance matching network 106 further includes a shunt circuit connected to the series circuit. The shunt circuit includes a capacitor connected in series with an inductor. The impedance matching network 106 includes one or more capacitors, and the corresponding capacitances of the one or more capacitors, such as all variable capacitors, are variable, such as using driving components to vary. The impedance matching network 106 includes one or more capacitors with fixed capacitances, such as capacitors that cannot be changed by the driving component 112. The variable capacitance of one or more variable capacitors of the impedance matching network 106 is the value C1. For example, the complex number of one or more variable capacitors is adjusted to a fixed position relative to the plate to set the variable capacitance value C1. For example, the combined capacitance of two or more capacitors connected in parallel to each other is the sum of the complex capacitances of the plurality of capacitors. For another example, the combined capacitance of two or more capacitors connected in series is the reciprocal of the sum of the reciprocal of the complex capacitances of the complex capacitors. An example of the impedance matching network 106 is provided in the US patent application US 14/245,803.

In some embodiments, the model system 102 includes a computer-generated model of the impedance matching network 106. For example, the model system 102 is generated by the processor 134 of the host computer system 110. The matching network model is derived from a branch of the impedance matching network 106 as representing the branch of the impedance matching network 106. For example, when the y MHz RF generator is connected to the branch circuit of the impedance matching network 106, the matching network model indicates that the circuit of the branch circuit of the impedance matching network 106 is a computer-generated model. For another example, the number of circuit elements in the matching network model is not equal to the number of circuit elements in the impedance matching network 106.

In some embodiments, the number of circuit elements in the matching network model is less than the number of circuit elements in the branch circuit of the impedance matching network 106. For example, the matching network model is a simplified form of the branch circuit of the impedance matching network 106. For another example, the complex variable capacitors of the complex variable capacitors of the branch circuit of the impedance matching network 106 are integrated into a combined variable capacitor and impedance matching network represented by one of the matching network models or multiple variable capacitor elements. The complex fixed capacitances of the plurality of fixed capacitors of the branch circuit of 106 are integrated into a combined fixed capacitance represented by one or more fixed capacitance elements of the matching network model, and/or the complex fixed inductance of the branch circuit of the impedance matching network 106 The complex inductance of the matching network model is integrated into a combined inductance represented by one or more inductance elements, and/or the complex resistance values of the complex fixed resistors of the branch circuit of the impedance matching network 106 are integrated into the matching network A fixed resistance value represented by one or more resistance elements of the model. For example, by taking the reciprocal of each capacitance value to generate a complex reciprocal capacitance value, and the sum of the complex reciprocal capacitance value to generate a reciprocal combined capacitance value, and then taking the reciprocal of the reciprocal combined capacitance value to generate a combined capacitance value to integrate the series The complex capacitance value of the connected complex capacitor. For another example, the complex inductances of the complex inductors connected in series are summed to generate a combined inductance, and the complex resistance values of the complex resistors are integrated to generate a combined resistance. All the fixed capacitances of all the fixed capacitors of the branch circuit of the impedance matching network 106 are integrated into a combined fixed capacitance value of one or more fixed capacitance elements of the matching network model. Other examples of matching network models are provided in the US patent application with application number US 14/245,803. In addition, the method of generating a matching network model from an impedance matching network is contained in a US patent application with application number US 14/245,803.

In some embodiments, the matching network model is generated by the circuit diagram of the impedance matching network 106 with three branches (respectively for x MHz, y MHz, and z MHz RF generators). The three branches are connected to each other at the output 140 of the impedance matching network 106. The circuit diagram starts with various combinations of complex inductors and capacitors. Considering one of the three branches independently, the matching network model represents one of the three branches. The circuit elements are added to the matching network model by an input device, examples of which will be provided below. Examples of additional circuit elements include complex resistors that were not previously included in the schematic diagram and used to explain the power loss in the branches of the impedance matching network 106, and include those that were not previously included in the schematic diagram and used to represent various connected RF bands. The complex inductance of the inductance and the complex capacitors that are not included in the diagram and used to represent the parasitic capacitance value. In addition, some circuit elements are further added by the input device to represent the nature of the branched transmission line of the impedance matching network 106 due to the physical size of the impedance matching network 106. For example, the unwinding length of one or more inductors in the branches of the impedance matching network 106 is not negligible compared to the wavelength of an RF signal passing through the one or more inductors. In order to explain this effect, one inductor in the diagram is split into two or more inductors. Afterwards, some circuit components are removed from the sketch by the input device to generate a matching network model.

In various embodiments, the matching network model and the branch circuit of the impedance matching network 106 have the same topology, such as the multiple connections between the multiple circuit elements, the number of the multiple circuit elements, and so on. For example, if the branch circuit of the impedance matching network 106 includes a capacitor coupled in series with an inductor, the matching network model includes a capacitor coupled in series with an inductor. In this example, the branch circuit of the impedance matching network 106 and the complex inductance of the matching network model have the same value, and the branch circuit of the impedance matching network 106 and the complex capacitor of the matching network model have the same value. For another example, if the branch circuit of the impedance matching network 106 includes a capacitor coupled in parallel with an inductor, the matching network model includes a capacitor coupled in parallel with an inductor. In this example, the branch circuit of the impedance matching network 106 and the complex inductance of the matching network model have the same value, and the branch circuit of the impedance matching network 106 and the complex capacitor of the matching network model have the same value. For another example, the matching network model and the impedance matching network 106 have the same number and the same type of complex circuit elements, and both have the same type of complex connection between the complex circuit elements. Examples of the types of circuit elements include resistors, inductors, and capacitors, and examples of connection types include series, parallel, and the like.

In various embodiments, the model system 102 includes a combination of a matching network model and an RF transmission model. The input of the matching network model is input 142. The RF transmission model is connected in series to the output of the matching network model and has an output 144. The way the RF transmission model is derived from the RF transmission line 132 is similar to the way the matching network model is derived from the impedance matching network 106. For example, the RF transmission model has inductance, capacitance, and/or resistance values derived from the inductance, capacitance, and/or resistance of the RF transmission line 132. For another example, the capacitance value of the RF transmission model matches the capacitance value of the RF transmission line 132, the inductance of the RF transmission model matches the inductance of the RF transmission line 132, and the resistance value of the RF transmission model matches the inductance of the RF transmission line 132.

In some embodiments, the model system 102 includes a combination of an RF cable model, a matching network model, and an RF transmission model. The input of the RF cable model is input 142. The output of the RF cable model is connected to the input of the matching network model, and the output of the matching network model is connected to the input of the RF transmission model. The RF transmission model has an output of 144. The way the RF cable model is derived from the RF cable 130 is similar to the way the matching network model is derived from the impedance matching network 106. For example, the RF cable model has a complex inductance, capacitance, and/or resistance value derived from the complex inductance, capacitance, and/or resistance value of the RF cable 130. For another example, a capacitance value of the RF cable model matches a capacitance value of the RF cable 130, an inductance of the RF cable model matches an inductance of the RF cable 130, and a resistance value of the RF cable model It matches with an inductance of the RF cable 130.

The x MHz RF generator includes an RF power source 121 for generating RF signals. The RF power supply 121 has an output 123, which is also the output of the x MHz RF generator. The output 123 is connected to the input 125 of the impedance matching network 106 via an RF cable 127. The x MHz RF generator is connected to the additional branch of the impedance matching network 106 through the input 125 of the additional branch, which is different from the branch circuit to which the y MHz RF generator is connected at the input 128 of the branch circuit. For example, a combination of one or more resistors, and/or one or more capacitors, and/or one or more inductances included in the additional branch is different from one or more of the branch circuits connected to the input 128. A combination of one resistor, and/or one or more capacitors, and/or one or more inductors. Both the additional branch connected to the input 125 and the branch circuit connected to the input 128 are connected to the output 140.

In addition, the y MHz RF generator includes an RF power supply 122 for generating RF signals. The y MHz RF generator includes sensors 124 connected to the output 126 of the y MHz RF generator, such as complex impedance sensors, complex current and voltage sensors, complex reflection coefficient sensors, complex voltage sensors, and Current sensor, etc. The output 126 is connected to the input 128 of the branch circuit of the impedance matching network 106 via the RF cable 130. The impedance matching network 106 is connected to the plasma chamber 108 through an output 140 and an RF transmission line 132. The RF transmission line 132 includes an RF rod and an RF outer conductor surrounding the RF rod.

The driving component 112 includes a driving device such as one or more transistors and a motor. The motor is connected to the variable capacitor of the impedance matching network 106 through the connecting mechanism 114. Examples of the connecting mechanism 114 include one or more rods, a plurality of rods connected to each other by gears, and the like. The connecting mechanism 114 is connected to a variable capacitor of the impedance matching network 106. For example, the connecting mechanism 114 is connected to a variable capacitor as part of the branch circuit, which is connected to the y MHz RF generator through the input 128.

It should be noted that in the case where the impedance matching network 106 includes more than one variable capacitor connected to the branch circuit of the y MHz RF generator, the driving component 112 includes a plurality of separate motors for controlling more than one variable capacitor. Each motor is connected to a corresponding variable capacitor by a corresponding connecting mechanism. In this example, the plural connection mechanisms may be referred to as the connection mechanism 114.

In some embodiments, an example of an x MHz RF generator includes a 2 MHz RF generator, an example of a y MHz RF generator includes a 27 MHz RF generator, and an example of a z MHz RF generator includes a 60 MHz RF generator. Device. In various embodiments, an example of an x MHz RF generator includes a 400 kHz RF generator, an example of a y MHz RF generator includes a 27 MHz RF generator, and an example of a z MHz RF generator includes a 60 MHz RF generator. .

It should be noted that in the case where three RF generators such as x, y, and z MHz RF generators are used in the plasma chamber 100, the x MHz RF generator is connected to the input 125, y MHz of the impedance matching network 106 The RF generator is connected to the input 128 of the impedance matching network 106, and the third of the complex RF generators is connected to the third input of the impedance matching network 106. The output 140 is connected to the input 125 by an additional branch of the impedance matching network 106 and the output 140 is connected to the input 128 by a branch circuit of the impedance matching network 106. The output 140 is connected to the third input through the third circuit branch of the impedance matching network 106.

The host computer system 110 includes a processor 134 and a memory device 137. The memory device 137 stores the model system 102. The processor 134 can receive the model system 102 from the memory device 137 and execute it. Examples of the host computer system 110 include a notebook computer, or a desktop computer, or a tablet, or a smart phone, etc. As used in the text, terms such as central processing unit (CPU), controller, application-specific integrated circuit (ASIC), or programmable logic device (PLD) can replace the term processor and be used interchangeably with the term processor . Examples of memory devices include read-only memory (ROM), random access memory (RAM), hard disks, volatile memory, non-volatile memory, redundant array of storage disks, flash memory, etc. The sensor 124 is connected to the host computer system 110 through a network cable 136. Examples of the network cables used in the text are cables that use serial, parallel, or universal serial bus (USB) protocols to transmit data.

During the period P1 of the RF signal generated by the x MHz RF generator, the y MHz RF generator, which has a higher frequency than the x MHz RF generator, operates under the complex radio frequency value RF1(P1)o, where o is an integer greater than zero . Examples of the complex radio frequency value RF1(P1)o include RF1(P1)1, RF1(P1)2, RF1(P1)3, and so on. For example, the processor 134 provides a formula including the complex radio frequency value RF1(P1)o and the complex power level used in the period P1 to the y MHz RF generator.

In various embodiments, each of the x and y MHz RF generators receives a clock signal from the processor 134 in the host computer system 110 or a clock source such as an oscillator. During the period P1 of the x MHz RF generator, the y MHz RF generator generates an RF signal with a complex period. For example, after receiving the clock signal, the x MHz RF generator generates an RF signal with a period P1 during a clock period of the clock signal, wherein the period P1 repeats during the clock period. For example, the RF signal generated by the x MHz RF generator is repeated in the period P1. Furthermore, in this example, after receiving the clock signal, the y MHz RF generator generates the RF signal with a complex period in the period P1 during the clock period of the clock signal. For example, the RF signal generated by the y MHz RF generator repeatedly oscillates a plurality of times during the period P1, and the period P1 is an oscillation of the RF signal generated by the x MHz RF generator.

During the period P1 of the RF signal generated by the x MHz RF generator, the y MHz RF generator receives the recipe through the network cable 138 connected to the y MHz RF generator and the host computer system 110, and then the y MHz RF generator The digital signal processor (DSP) provides the formula to the RF power supply 122. The RF power supply 122 generates an RF signal having a complex radio frequency value RF1(P1)o and a complex power level specified in the formula.

The impedance matching network 106 is initialized to have a combined variable capacitance value C1. For example, the processor 134 sends a signal to the driving device of the driving component 112 to generate one or more current signals. One or more current signals are generated by the driving device and sent to one or more stators of the driving assembly 112 corresponding to one or more motors. One or more rotors of the driving component 112 that are in contact with the corresponding one or more stator electric fields rotate to move the connecting mechanism 114 to change the combined variable capacitance value of the branch circuit of the impedance matching network 106 to C1. The branch circuit with the combined variable capacitance value C1 of the impedance matching network 106 receives the RF signal with the complex radio frequency value RF1(P1)o from the output 126 through the input 128 and the RF cable 130. In addition, the additional branch of the impedance matching network 106 receives the RF signal from the output 123 of the x MHz RF generator through the RF cable 127 and the input 125. After receiving the RF signal from the x and y MHz RF generators, the impedance matching network 106 matches the impedance of the load connected to the impedance matching network 106 with the impedance of the source connected to the impedance matching network 106 to generate a modified RF signal . Examples of loads include the plasma chamber 108 and the RF transmission line 132. Examples of sources include RF cable 127, RF cable 130, x MHz RF generator, and y MHz RF generator. The modified signal is provided to the chuck 118 via the output 140 of the branch circuit of the impedance matching network 106 through the RF transmission line 132. When the modified signal and one or more process gases such as oxygen-containing gas, fluorine-containing gas, etc. are provided to the chuck 118, plasma is generated or maintained in the gap between the chuck 118 and the upper electrode 116.

During the time that the RF signal with radio frequency RF1(P1)o is generated, the impedance matching network 106 has a combined variable capacitance value C1, and the x MHz RF generator generates the RF signal during the period P1, the sensor 124 is at the output 126 The complex voltage reflection coefficient Γmi(P1)n is sensed and the complex voltage reflection coefficient Γmi(P1)n is provided to the processor 134 through the network cable 136, where n is an integer greater than zero. For example, during the period P1, the sensor 124 is at a plurality of predetermined periodic time intervals, such as every 0.3 microsecond, every 0.5 microsecond, every 0.1 microsecond, a fixed fraction of one microsecond, every 0.v microsecond, etc. Measure the voltage reflection coefficient Γmi(P1)n, where n is the number of time intervals and is also equal to the number of complex voltage reflection coefficients Γmi(P1)n, and v is a real number greater than 0 and less than 10. For another example, during the period P1, the sensor 124 measures the voltage reflection coefficient Γmi(P1)1 at 0.3 microseconds from the beginning of the period P1 and measures the voltage reflection at 0.6 microseconds from the beginning of the period P1. The coefficient Γmi(P1)2. An example of the voltage reflection coefficient includes a ratio of the voltage reflected from the plasma chamber 108 to the y MHz RF generator and the voltage supplied in the RF signal generated by the y MHz RF generator.

For another example, the period P1 of the 400 kHz RF signal is divided into 8 sub-periods such as ΔT1, ΔT2, ΔT3, ΔT4, ΔT5, ΔT6, ΔT7, ΔT8. Each of these sub-periods is a short time interval equal to P1/8, or about 0.v microseconds, etc. In some embodiments, since the 400 kHz frequency varies between 350 and 450 kHz, the processor 134 will make the duration of each of these sub-periods longer or shorter and the processor 134 will increase or decrease this The number of these sub-periods. The beginning of the period P1 of the 400 kHz RF signal is detected by the processor 134 and the beginning of the period marks the beginning of the sub-period ΔT1, and each additional ΔT2 to ΔT8 sequentially follows the sub-period ΔT1. Measure the eight measured values of the voltage reflection coefficient Γmi(P1)n related to the 60 MHz RF generator, such as Γmi(P1)1, Γmi(P1)2, Γmi(P1)3, Γmi(P1)4, Γmi (P1)5, Γmi(P1)6, Γmi(P1)7, Γmi(P1)8. In some embodiments, the eight measured values are measured in the period P1. In various embodiments, the eight measured values Γmi(P1)1, Γmi(P1)2, Γmi(P1)3, Γmi(P1)4, Γmi(P1)5, Γmi(P1)6, Γmi(P1) )7, and Γmi(P1)8 are measured in multiple periods of the 400 kHz RF signal, such as period P1, period P(1+1), period P(1+2), etc. It should be noted that the eight measurement values are an example. In some embodiments, any number of measurement values of the voltage reflection coefficient can be measured in the period P1 or in multiple periods.

The processor 134 calculates the complex impedance Zmi(P1)n from the complex voltage reflection coefficient Γmi(P1)n. For example, the processor 134 calculates impedance Zmi(P1)1 by applying equation (1) and solving Zmi(P1)1, where equation (1) is Γmi(P1)1 = (Zmi(P1)1 – Zo) /(Zmi(P1)1 + Zo) and Zo is the characteristic impedance of the RF transmission line 132. For another example, the processor 134 calculates impedance Zmi(P1)2 by applying equation (2) and solving Zmi(P1)2, where equation (2) is Γmi(P1)2 = (Zmi(P1)2 – Zo )/(Zmi(P1)2 + Zo). Impedance Zo is provided to the processor 134 through input devices. Input devices such as mouse, keyboard, stylus, keypad, buttons, touch screen, etc. are through input/output interfaces such as serial interface, parallel interface, and USB interface. Connected to the processor 134. In some embodiments, the sensor 124 measures the complex impedance Zmi(P1)n and provides the complex impedance Zmi(P1)n to the processor 134 through the network cable 136.

The complex impedance Zmi(P1)n is provided by the processor 134 to the input 142 of the model system 102 and propagated forward by the model system 102 to calculate the complex load impedance ZL(P1)n at the output 144 of the model system 102. The processor 134 initializes the model system 102 so that the model system 102 has a combined variable capacitance value C1 and a complex radio frequency value RF1(P1)o. For example, the processor 134 uses one or more circuit elements of the model system 102 to propagate the impedance Zmi(P1)1 forward to generate the load impedance ZL(P1)1. For example, the model system 102 is initialized to have a radio frequency RF1(P1)1 and a combined variable capacitance value C1. When the model system 102 includes a series connection combination of a resistance element, an inductance element, a fixed capacitance element, and a variable capacitance element, the processor 134 calculates the impedance Zmi (P1) received at the input 142 of the model system 102 1. A complex impedance across the resistance element, a complex impedance across the inductive element, a complex impedance across the variable capacitance element with a variable capacitance value C1, and a complex impedance across the fixed capacitance element The sum of the complex impedances in one direction produces the load impedance ZL(P1)1. For another example, the processor 144 uses one or more circuit elements of the model system 102 to propagate the impedance Zmi(P1)2 forward to generate the load impedance ZL(P1)2. For example, the model system 102 is initialized to have a radio frequency RF1(P1)2 and a combined variable capacitance value C1. When the model system 102 includes a series combination of a resistance element, an inductance element, a fixed capacitance element, and a variable capacitance element, the processor 134 calculates the impedance Zmi (P1) received at the input 142 of the model system 102 2. A complex impedance across the resistance element, a complex impedance across the inductive element, a complex impedance across the variable capacitance element with a variable capacitance value C1, and a complex impedance across the fixed capacitance element The sum of the complex impedances in one direction produces the load impedance ZL(P1)2.

In various embodiments, instead of measuring a voltage reflection coefficient at the output 126, a voltage reflection coefficient is measured at any point on the RF cable 130 from the output 126 to the input 128 (including the output 126). For example, the sensor 124 is connected to a point between the RF power source 122 and the impedance matching network 106 to measure a voltage reflection coefficient.

In some embodiments, the processor 134 weights each of the measured complex voltage reflection coefficients Γmi(P1)n according to a pre-assigned weight. As will be explained below, the complex weight applied by the processor 134 to the complex voltage reflection coefficient Γmi(P1)n is received by the processor 134 through the input of the input device and is determined based on engineering knowledge and/or process conditions. The weighted complex voltage reflection coefficient wΓmi(P1)n can be applied to the model system 102 instead of applying the complex voltage reflection coefficient Γmi(P1)n to determine the complex load impedance, where each w is a pre-assigned weight.

2 is a diagram of an embodiment of the model system 102. The model system 102 is initialized to have a complex radio frequency value RF1(P1)o and a variable capacitance value C1 to determine the complex radio frequency value RF(P1)n. For each of the complex radio frequency values RF(P1)n, a voltage reflection coefficient Γ(P1)n at the input 142 during the period P1 is the minimum value. The processor 134 calculates the complex radio frequency value RF(P1)n from the complex load impedance ZL(P1)n and the model system 102. For each of the complex radio frequency values RF(P1)n, the voltage reflection coefficient Γ(P1) is the minimum value among the multiple values of the voltage reflection coefficient Γ(P1). For example, the processor 134 uses the model system 102 to propagate the load impedance ZL(P1)1 backward to determine the radio frequency value RF(P1)1, wherein the model system 102 is initialized to have the radio frequency RF1(P1)1 and the variable capacitance value C1 , Where the radio frequency value RF(P1)1 is generated at the input 142 as an input impedance Z1 during the period P1. The way that the processor 134 calculates a voltage reflection coefficient Γ(P1)1 from the input impedance Z1 is similar to the way of using equation (1) above. In addition, the processor 134 uses the model system 102 to propagate the load impedance ZL(P1)1 backward to determine the radio frequency value RF(P1)1_1. The model system 102 is initialized to have the radio frequency value RF1(P1)1 and the variable capacitance value. C1, where the radio frequency value RF(P1)1_1 is generated at the input 142 as an input impedance Z2 of the period P1. The way that the processor 134 calculates a voltage reflection coefficient Γ(P1)2 from the input impedance Z2 is similar to the way of using equation (1) above. The processor 134 determines that the voltage reflection coefficient Γ(P1)1 is smaller than the voltage reflection coefficient Γ(P1)2 and determines the radio frequency value RF1(P1)1 to make the voltage reflection coefficient Γ(P1)1 the smallest value.

For another example, the processor 134 uses the model system 102 to propagate the load impedance ZL(P1)2 backward to determine the radio frequency value RF(P1)2, wherein the model system 102 is initialized to have the radio frequency RF1(P1)2 and the variable capacitance value C1, where the radio frequency value RF(P1)2 is generated at the input 142 and an input impedance Z3 of the period P1 is generated. The way that the processor 134 calculates a voltage reflection coefficient Γ(P1)3 from the input impedance Z3 is similar to the way of using equation (2) above. Furthermore, the processor 134 uses the model system 102 to propagate the load impedance ZL(P1)2 backward to determine the radio frequency value RF(P1)2_2, wherein the model system 102 is initialized to have the radio frequency RF1(P1)2 and the variable capacitance value C1 , Wherein the radio frequency value RF(P1)2_2 at the input 142 generates an input impedance Z4 of the period P1. The way that the processor 134 calculates a voltage reflection coefficient Γ(P1)4 from the input impedance Z4 is similar to the way of using equation (2) above. The processor 134 determines that the voltage reflection coefficient Γ(P1)3 is smaller than the voltage reflection coefficient Γ(P1)4 and determines the radio frequency value RF1(P1)2 to make the voltage reflection coefficient Γ(P1)3 the smallest value.

It should be noted that the value ZL(P1)1 is determined from the value Zmi(P1)1, and the value Zmi(P1)1 is measured at the end of the first time period such as t1, etc., from the start measurement of the period P1, And measured during the period P1. The value ZL(P1)2 is determined from the value Zmi(P1)2, and the value Zmi(P1)2 is measured at the end of the second time period such as t2, measured from the first time period, and during the period P1 Period measurement. The second time period t2 is connected after the first time period t1 and has a length equal to the first time period t1. In various embodiments, the voltage reflection coefficient Γ(P1)1 is the minimum value among all the complex voltage reflection coefficients during the first time period t1, and the voltage reflection coefficient Γ(P1)2 is all the complex voltage reflection coefficients during the second time period t2. The minimum value of the coefficient.

In some embodiments, the processor 134 executes a nonlinear least squares optimization procedure to solve and calculate the complex radio frequency value RF(P1)n from the load impedance ZL(P1)n and the model system 102. For each of the complex radio frequency values RF(P1)n, the voltage reflection coefficient Γ(P1)n during the period P1 is the minimum value. In various embodiments, the processor 134 supplies predetermined equations to solve and calculate the complex radio frequency value RF(P1)n from the load impedance ZL(P1)n and the model system 102.

In various embodiments, a value of the radio frequency of the model system 102 that minimizes the voltage reflection coefficient Γ at the input 142 is referred to herein as a favorable RF value.

In some embodiments, an RF value is sometimes referred to as a "parameter value" in the text. In addition, a capacitance value is sometimes referred to as a "measurable factor" in the text.

In various embodiments, in addition to calculating the complex radio frequency value RF(P1)n, the processor 134 can also calculate a value of the combined variable capacitance value Coptimum(P1) during the period P1, or the processor 134 can calculate the combination of the period P1 during the calculation period. The variable capacitance value Coptimum(P1) is a value but does not calculate the complex radio frequency value RF(P1)n. For example, the processor 134 calculates the combined variable capacitance value Coptimum(P1) that makes the weighted average of one of the complex voltage reflection coefficients Γ(P1)n at the input 142 the minimum value. For example, the processor 134 calculates a weighted average of one of the complex voltage reflection coefficients Γ(P1)n. The processor 134 uses the model system 102 to propagate the load impedance ZL(P1)n backward to determine the combined variable capacitance value Coptimum(P1) that minimizes the weighted average of the complex voltage reflection coefficient Γ(P1)n. For example, the processor 134 uses the model system 102 to propagate back any of the complex load impedances ZL(P1)n, such as ZL(P1)1 or ZL(P1)2, to determine the value of the complex voltage reflection coefficient Γ(P1)n. The weighted average has a combined variable capacitance value Coptimum(P1) of a first value. When any of the complex load impedances ZL(P1)n propagates backward, the model system 102 is initialized to correspond to any of the complex radio frequency values RF1(P1)n and the variable capacitance value C1. For example, when the load impedance ZL(P1)1 propagates backward, the model system 102 is initialized to the corresponding radio frequency value RF1(P1)1, and when the load impedance ZL(P1)2 propagates backward, the model system 102 is initialized to the corresponding radio frequency value RF1(P1)1. Radio frequency value RF1(P1)2. Continuing to further illustrate, the processor 134 uses the model system 102 to propagate back any of the complex load impedances ZL(P1)n to determine the other that the weighted average of the complex voltage reflection coefficient Γ(P1)n has a second value. A combined variable capacitance value Coptimum(P1)2. The processor 134 determines that the first value is lower than the second value and determines that the combined variable capacitance value Coptimum(P1)1 is the optimal combined variable capacitance that minimizes the weighted average of the complex voltage reflection coefficient Γ(P1)n The value Coptimum(P1). It should be noted that the weight of each of the complex voltage reflection coefficients Γ(P1)n used to generate the weighted average is received by the processor 134 from the input device.

In various embodiments, the sensor 124 generates q measured values of the voltage reflection coefficient Γmi(P1)q instead of obtaining n quantities of the voltage reflection coefficient Γmi(P1)n from the sensor 124 (FIG. 1). Measured value, where q is an integer greater than n and greater than zero. The processor 134 propagates the complex voltage reflection coefficient Γmi(P1)q forward through the model system 102 to generate q values of the load impedance ZL(P1)q at the output 144 of the model system 102. The model system 102 is initialized to have a variable capacitance value C1 and a complex value RF1(P1)o. The processor 134 divides the complex load impedance ZL(P1)q into n identical sections and calculates an average value of the complex load impedance in each of the n sections. For example, the processor 134 calculates a first average value of 10 measurement values ZL(P1)1 to ZL(P1)10 and calculates a second value of 10 measurement values ZL(P1)11 to ZL(P1)20 The average value, where 1, 10, 11, and 20 are all examples of q. The first average value is an example of one of the complex load impedance ZL(P1)n and the second average value is an example of the other one of the complex load impedance ZL(P1)n.

In some embodiments, instead of minimizing the voltage reflection coefficient Γ(P1)n, another parameter, such as the power reflection coefficient, is minimized at the input 142.

3 is a diagram of an embodiment of the plasma system 100, which illustrates the use of the model system 102 to generate a complex load impedance ZL(P(1+m) during a period P(1+m) of the RF signal generated by the x MHz RF generator m)) n, where m is an integer greater than zero. The period P(1+m) is after the period P1. For example, a first oscillation of an RF signal generated by an x MHz RF generator is followed by a second oscillation of the RF signal. The second oscillation and the first oscillation are continuous and there is no other oscillation between the first and the second oscillation. The second oscillation has a time period of P2 and the first oscillation has a time period of P1. In some embodiments, the time length of the period P2 is equal to the time length of the period P1. For another example, the first oscillation of the RF signal generated by the first oscillation of x MHz RF generator is not immediately followed by the second oscillation of the RF signal but immediately followed by one or more oscillations, the one or more The oscillation is followed by the (1+m)th oscillation of the period P(1+m). The (1+m)th oscillation and the first oscillation are not continuous and there are one or more intermediate oscillations between the first oscillation and the (1+m)th oscillation. In some embodiments, the amount of time of the clock cycle covered by the period P(1+m) is equal to the amount of time of the clock cycle covered by the period P1.

During the period P(1+m) of the RF signal generated by the x MHz RF generator, the processor 134 modifies the formula to include the complex radio frequency value RF(P1)n and provides the complex radio frequency value RF(P1)n to y MHz RF generator. In addition, the processor 134 determines the one-step variable capacitance value Cstep1 for the period P(1+m). For example, the processor 134 detects the beginning of the period P(1+m) of the 400 kHz RF generator and targets a first part of the period P(1+m) of the RF signal, such as the first part of the period P(1+m). The RF value RF(P(1)1) is supplied during the 1/8 period. Continuously, for a second part of the period P(1+m) of the RF signal, such as the second 1/8 of the period P(1+m) The radio frequency value RF(P(1)2, etc.) is supplied during part of the period. The second part of the period P(1+m) is continuous with the first part of the period P(1+m). The variable capacitance value Cstep1 of this step is at There is a difference from the value C1 to the value Coptimum(P1).

It should be noted that when one or more capacitance values corresponding to one or more variable capacitors of the impedance matching network 106 are changed from C1 to Coptimum(P1), the one or more variable capacitors are relative to y MHz RF The change of the RF frequency of the RF signal generated by the generator moves in a sufficiently slow manner. The processor 134 controls the driving component 112 so that the combined variable capacitance value of the impedance matching network 106 is set at the value Cstep1 instead of setting the combined variable capacitance value of the impedance matching network 102 at the value Coptimum(P1). The time required for the impedance matching network 106 to reach the variable capacitance value Coptimum (P1) (such as the level of seconds) is greater than the time required for the y MHz RF generator to generate an RF signal with a complex radio frequency value RF(P1)n . For example, a y MHz RF generator needs microsecond-level time from the complex radio frequency RF1(P1)o to the complex radio frequency value RF(P1)n. Therefore, it is difficult to directly reach the variable capacitance value Coptimum(P1) from the value C1 and at the same time from the complex radio frequency value RF1(P1)o to the complex radio frequency value RF(P1)n so that the voltage at the output 126 of the y MHz RF generator can be reflected The coefficient Γ(P1)n is the minimum value. Therefore, during the period P(1+m), the variable capacitance value of the impedance matching network 106 is adjusted along the direction toward the variable capacitance value Coptimum(P1) with a complex number of steps such as Cstep1.

The processor 134 further controls the y MHz RF generator so that the y MHz RF generator operates at the complex radio frequency value RF(P1)n during the period P(1+m). For the complex radio frequency RF(P1)n and the variable capacitance value Cstep1, the y MHz RF generator generates an RF signal with the complex radio frequency value RF(P1)n, and this RF signal passes through the branch circuit of the impedance matching network 106. In addition, the additional branch of the impedance matching network 106 receives the RF signal from the output 123 of the x MHz RF generator through the RF cable 127 and the input 125. After receiving the complex RF signals from the x and y MHz RF generators, the impedance matching network 106 generates a modified signal, and the modified signal is provided to the bottom electrode 118. When the complex value RF(P1)n is used to replace the complex value RF(P1)o, compared to the period P1, a smaller amount of power is reflected to the y MHz RF generator during the period P(1+m).

During the period P(1+m), when the y MHz RF generator generates a complex radio frequency value RF(P1)n RF signal and the combined variable capacitance value is Cstep1, the sensor 124 measures the complex number at the output 126 Voltage reflection coefficient Γmi(P(1+m)n. For example, the period P(1+m) of the 400 kHz RF signal is divided into 8 sub-periods such as ΔT1, ΔT2, ΔT3, ΔT4, ΔT5, ΔT6, ΔT7, ΔT8. Each of these sub-periods is a short time interval equal to P(1+m)/8, or about 0.v microseconds, etc. In some embodiments, since the frequency of 400 kHz is between 350 and Between 450 kHz, the processor 134 will make each of these sub-periods longer or shorter and the processor 134 will increase or decrease the number of these sub-periods. The period P(1+ The start of m) is detected by the processor 134 and the start of the period marks the beginning of the sub-period ΔT1, and each additional ΔT2 to ΔT8 sequentially follows the sub-period ΔT1. Measure the voltage associated with the 60 MHz RF generator The eight measured values of reflection coefficient Γ(P(1+m)n such as Γmi(P(1+m))1, Γmi(P(1+m))2, Γmi(P(1+m))3 , Γmi(P(1+m))4, Γmi(P(1+m))5, Γmi(P(1+m))6, Γmi(P(1+m))7, Γmi(P(1 +m)) 8. In some embodiments, the eight measured values are measured in the period P(1+m). In various embodiments, Γmi(P(1+m))1, Γmi (P(1+m))2, Γmi(P(1+m))3, Γmi(P(1+m))4, Γmi(P(1+m))5, Γmi(P(1+m) ))6, Γmi(P(1+m))7, Γmi(P(1+m))8 are in multiple periods of 400 kHz RF signal such as period P(1+m) and period P(1+m +1) and the period P(1+m+2), etc. It should be noted that the eight measurement values are an example. In some embodiments, the period P(1+m) may be more or more Measure any number of measurement values of the voltage reflection coefficient during each period.

During the period P(1+m), the sensor 124 provides the complex voltage reflection coefficient Γmi(P(1+m))n to the processor 134 via the network cable 136. The processor 134 generates the complex impedance Zmi(P(1+m))n from the complex voltage reflection coefficient Γmi(P(1+m))n and the above-mentioned self-complex voltage reflection coefficient Γmi(P1)n generates the complex impedance Zmi (P1)n is the same way. For example, the processor 134 generates the impedance value Zmi(P(1+m))1 from the voltage reflection coefficient Γmi(P(1+m))1, and the voltage reflection coefficient Γmi(P(1+m))1 is in the period P (1+m) is measured during the first time period t1 starting from the period P(1+m). In addition, the processor 134 generates an impedance value Zmi(P(1+m))2 from the voltage reflection coefficient Γmi(P(1+m))2, and the voltage reflection coefficient Γmi(P(1+m))2 is in the period P (1+m) is measured at the end of the second time period t2. The second time period t2 starts at the end of the first time period t1 and the first time period t1 starts from the period P(1+m).

Furthermore, when the model system 102 is set to have a complex radio frequency value RF(P1)n for the period P(1+m) and a combined variable capacitance value Cstep1 for the period P(1+m), the complex impedance Zmi(P( 1+m))n through the forward propagation of the model system 102 to generate the complex load impedance ZL(P(1+m))n at the output 144 of the model system 102 is the same as the input 142 from the model system 102 The complex impedance Zmi(P1)n generates the complex load impedance ZL(P1)n at the output 144 in the same way.

In various embodiments, the combined variable capacitance value Cstep1 is closer to the combined variable capacitance value Coptimum(P1) than the combined variable capacitance value C1. For example, the combined variable capacitance value Cstep1 is greater than the combined variable capacitance value C1 and the combined variable capacitance value Coptimum (P1) is greater than the combined variable capacitance value Cstep1. For another example, the combined variable capacitance value Cstep1 is smaller than the combined variable capacitance value C1 and the combined variable capacitance value Coptimum(P1) is smaller than the combined variable capacitance value Cstep1.

In some embodiments, the processor 134 receives the voltage reflection coefficient and generates a complex number corresponding to the load voltage reflection coefficient at the output 144 of the model system 102, such as ΓL(P1)n, ΓL(P(1+m))n, etc. instead The voltage reflection coefficients received from the sensor 124, such as Γmi(P1)n, Γmi(P(1+m))n, etc., generate impedances such as impedance Zmi(P1)n, Zmi(P(1+m))n, etc. . The way in which the complex number corresponding to the load voltage reflection coefficient is applied to the output 144 of the model system 102 is related to the way the load impedance such as ZL(P1)n, ZL(P(1+m))n, etc. is applied to the output of the model system 102 The same way. There is no need to convert from voltage reflection coefficient to impedance, and vice versa.

In some embodiments, the processor 134 weights each of the measured voltage reflection coefficients Γmi(P(1+m))n according to a pre-assigned weight. The complex weight applied by the processor 134 to the complex voltage reflection coefficient Γmi(P(1+m))n is received by the processor 134 through the input of the input device and determined based on engineering knowledge and/or process conditions. The weighted complex voltage reflection coefficient wΓmi(P(1+m))n can be applied to the model system 102 instead of applying the complex voltage reflection coefficient Γmi(P(1+m))n to determine the complex load impedance ZL(P(1+m))n m)) n, where each w is a pre-assigned weight.

In various embodiments, the value Coptimum(P1) and the value Cstep1 are applied to the plasma system 100 but the complex radio frequency value RF(P1)n is not determined and the complex radio frequency value RF(P1)n is not applied to the plasma system 100.

4 is a diagram of an embodiment of the model system 102. The model system 102 is initialized to have a complex radio frequency value RF(P1)n and a variable capacitance value Cstep1 to determine the complex radio frequency value RF(P(1+m))n . For each of the complex radio frequency values RF(P(1+m))n, a voltage reflection coefficient Γ(P(1+m))n at the input 142 during the period P(1+m) is the minimum . The processor 134 calculates the complex radio frequency value RF(P(1+m))n from the complex load impedance ZL(P(1+m))n and the model system 102. For each of the complex radio frequency values RF(P(1+m))n, the voltage reflection coefficient Γ(P(1+m))n at the input 142 is derived from the voltage reflection coefficient Γ(P(1+m )) The smallest value among the multiple values of n. For example, the processor 134 uses the model system 102 to propagate the load impedance ZL((P(1+m))1 back to determine the radio frequency value RF(P(1+m))1, wherein the model system 102 is set to have a radio frequency RF1 (P1)1 and the variable capacitance value Cstep1, where the radio frequency value RF(P(1+m))1 generates an input impedance Z5 during the period P(1+m) at the input 142. The processor 134 calculates from the input impedance Z5 The way of a voltage reflection coefficient Γ(P(1+m))5 is similar to the way of using equation (1) above. Furthermore, the processor 134 propagates the load impedance ZL((P(1+m) backwards through the model system 102). ))1 to determine another radio frequency value RF(P(1+m))1_1, where the model system 102 is set to have a radio frequency value RF(P1)1 and a variable capacitance value Cstep1, and the other radio frequency value RF(P (1+m)) 1_1 generates an input impedance Z6 during the period P(1+m) at the input 142. The processor 134 calculates a voltage reflection coefficient Γ(P(1+m))6 from the input impedance Z6. Similar to the above method using equation (1). The processor 134 determines that the voltage reflection coefficient Γ(P(1+m))5 is smaller than the voltage reflection coefficient Γ(P(1+m))6 and determines the radio frequency value RF (P(1+m))1 is the minimum value of the voltage reflection coefficient Γ(P(1+m))5.

For another example, the processor 134 uses the model system 102 to propagate the load impedance ZL((P(1+m)2) backwards to determine the radio frequency value RF(P(1+m)2), wherein the model system 102 is set to have a radio frequency The value RF(P1)2 and the variable capacitance value Cstep1, wherein the radio frequency value RF(P(1+m)2) is generated at the input 142 at an input impedance Z7 of the period P(1+m). The processor 134 self-input impedance The way Z7 calculates a voltage reflection coefficient Γ(P(1+m))7 is similar to the way of using equation (1) above. In addition, the processor 134 propagates the load impedance ZL((P(1) backwards through the model system 102). +m)2) to determine the radio frequency value RF(P(1+m)2_1), where the model system 102 is set to have radio frequency RF(P1)2 and variable capacitance value Cstep1, where the radio frequency value RF(P(1+ m) 2_1) An input impedance Z8 of the period P(1+m) is generated at the input 142. The way the processor 134 calculates a voltage reflection coefficient Γ(P(1+m))8 from the input impedance Z8 is similar to the above Using equation (1), the processor 134 determines that the voltage reflection coefficient Γ(P(1+m))7 is less than the voltage reflection coefficient Γ(P(1+m))8 and determines the radio frequency value RF(P(P(1+m)) 1+m)2) is the one that minimizes the voltage reflection coefficient Γ(P(1+m))7.

It should be noted that the value ZL(P(1+m)1) is determined by the load value Zmi(P(1+m))2, and the load value Zmi(P(1+m))2 is determined during the second time period as Measure at the end of t2, etc., where t2 starts at the end of the time period t1 and the time period t1 starts from the period P(1+m). The second time period of the period P(1+m) is connected after the first time period of the period P(1+m). The voltage reflection coefficient Γ(P(1+m))5 is the minimum value of all the complex voltage reflection coefficients in the first time period of the period P(1+m), and the voltage reflection coefficient Γ(P(1+m))7 It is the minimum value among all the complex voltage reflection coefficients during the second time period of the period P(1+m).

In some embodiments, the processor 134 performs a nonlinear least squares optimization procedure to solve and calculate the complex radio frequency value RF(P(1+m)n from the complex load impedance ZL(P(1+m))n and the model system 102 m))n. For each of the complex radio frequency values RF(P(1+m))n, the voltage reflection coefficient Γ(P(1+m))n during the period P(1+m) is the minimum value. In various embodiments, the processor 134 applies predetermined equations to solve and calculates the complex radio frequency value RF(P(1+m))n from the complex load impedance ZL(P(1+m))n and the model system 102.

In some embodiments, in addition to finding the complex radio frequency value RF(P(1+m))n, the processor 134 can also find the combined variable capacitance value Coptimum(P(1+m) during the period P(1+m) )), or the processor 134 finds a value of the combined variable capacitance value Coptimum(P(1+m)) during the period P(1+m) but does not find the complex radio frequency value RF(P(1+m) )n. For example, the processor 134 calculates the combined variable capacitance value Coptimum(P(1+m)) that makes the weighted average of one of the complex voltage reflection coefficients Γ(P(1+m))n at the input 142 the minimum value. For example, the processor 134 calculates a weighted average of one of the complex voltage reflection coefficients Γ(P(1+m))n. The processor 134 uses the model system 102 to propagate the load impedance ZL(P(1+m))n backwards to determine the combined variable capacitance that minimizes the weighted average of the complex voltage reflection coefficient Γ(P(1+m))n The value Coptimum(P(1+m)). For example, the processor 134 uses the model system 102 to propagate back any of the complex load impedances ZL(P(1+m))n such as ZL(P(1+m))1 or ZL(P(1+m)) 2 and so on to determine the combined variable capacitance value Coptimum(P(1+m))1 such that the weighted average of the complex voltage reflection coefficient Γ(P(1+m))n has a first value. When any of the complex load impedances ZL(P(1+m))n propagates backward, the model system 102 is initialized to correspond to any of the complex radio frequency values RF(P(1)n and the variable capacitance value Cstep1. For example, when the load impedance ZL(P(1+m))1 propagates backward, the model system 102 is initialized to the corresponding radio frequency value RF(P1)1, when the load impedance ZL(P(1+m))2 propagates backward When time, the model system 102 is initialized to the corresponding radio frequency value RF(P1)2. To further illustrate, the processor 134 uses the model system 102 to propagate back any of the complex load impedance ZL(P(1+m))n It is used to determine another combined variable capacitance value Coptimum(P(1+m))2 such that the weighted average of the complex voltage reflection coefficient Γ(P(1+m))n has a second value. The processor 134 determines The first value is lower than the second value and determines the combined variable capacitance value Coptimum(P(1+m))1 to make the weighted average of the complex voltage reflection coefficient Γ(P(1+m))n the smallest value. The best combination variable capacitance value Coptimum(P(1+m)). It should be noted that the weight of each of the complex voltage reflection coefficients Γ(P(1+m))n used to generate the weighted average is determined by the processor 134 Received from the input device.

In various embodiments, the sensor 124 generates q measured values of the voltage reflection coefficient Γmi(P(1+m))q instead of obtaining the voltage reflection coefficient Γmi(P( 1+m)) n measured values of n. The processor 134 propagates forward the complex voltage reflection coefficient Γmi(P(1+m))q through the model system 102 to generate the load impedance ZL(P(1+m))q at the output 144 of the model system 102 Numerical value. The model system 102 is initialized to have a variable capacitance value Coptimum(P1) and a complex value RF1(P1)n RF1(P1)n. The processor 134 divides the complex load impedance ZL(P(1+m))q into n identical sections and calculates an average value of the complex load impedance in each of the n sections. For example, the processor 134 calculates a first average of 10 measurement values ZL(P(1+m))1 to ZL(P(1+m))10 and calculates 10 measurement values ZL(P(1 +m)) a second average value from 11 to ZL(P(1+m))20, where 1, 10, 11, and 20 are all examples of q. The first average value is an example of one of the complex load impedance ZL(P(1+m))n and the second average value is an example of the other one of the complex load impedance ZL(P(1+m))n.

In some embodiments, instead of minimizing the voltage reflection coefficient Γ(P1)n, another parameter, such as the power reflection coefficient, is minimized at the input 142.

Figure 5 is a diagram of an embodiment of the plasma system 100, which illustrates the use of a capacitance value Coptimum (P(1+m)) and the use of a complex radio frequency value RF(P(1+m))n in the x MHz RF generator The wafer W is processed during a period P(1+m+q) of the generated RF signal, where q is an integer greater than zero. The period P(1+m+q) is after the period P(1+m) of the RF signal generated by the x MHz RF generator. For example, a second oscillation of the RF signal generated by an x MHz RF generator is followed by a third oscillation of the RF signal. The third oscillation and the second oscillation are continuous and there is no other oscillation between the second and the third oscillation. The third oscillation has a period P3 and the second oscillation has a period P2. In some embodiments, the time length of the period P3 is equal to the time length of the period P2. For another example, the second oscillation of the RF signal generated by the x MHz RF generator is not immediately followed by the third oscillation of the RF signal but is followed by one or more oscillations, and the one or more oscillations are followed by the period The (1+m+q) oscillation of P(1+m+q). The (1+m+q)th oscillation and the second oscillation are not continuous, and there are one or more intermediate oscillations between the second oscillation and the (1+m+q)th oscillation. In some embodiments, the amount of time of the clock cycle covered by the period P(1+m+q) is equal to the amount of time of the clock cycle covered by the period P(1+m).

During the period P(1+m+q) of the RF signal generated by the x MHz RF generator, the processor 134 modifies the formula during the period P(1+m+q) to include the complex radio frequency value RF(P(1+ m))n and provide the complex radio frequency value RF(P(1+m))n to the y MHz RF generator. For example, the processor 134 detects that the period P(1+m+q) of the 400 kHz RF generator starts and targets a first part of the period P(1+m+q) of the RF signal such as the period P(1+m+ The radio frequency value RF(P(1+m))1 is applied during the first 1/8 part of q). Continuously, for a second part of the period P(1+m+q) of the RF signal, such as the second 1/8 part of the period P(1+m+q), the radio frequency value RF(P(1+m )) 2 and so on. The second part of the period P(1+m+q) is continuous with the first part of the period P(1+m+q). When the complex value RF(P(1+m))n is used to replace the complex value RF(P1)n, there is less during the period P(1+m+q) compared to the period P(1+m) The amount of power is reflected to the y MHz RF generator.

In addition, the processor 134 controls the driving component 112 so that the combined variable capacitance value of the branch circuit of the impedance matching network 102 is set to the value Cstep2, instead of setting the combined variable capacitance value of the impedance matching network 102 to the value Coptimum At (P1), the variable capacitance value is set at the value Coptimum(P1) as a step toward the value Coptimum(P(1+m)). It should be noted that in some embodiments, the combined variable capacitance value Cstep2 is equal to the combined variable capacitance value Coptimum(P(1+m)).

During the period P(1+m+q) of the RF signal generated by the xMHz RF generator, when the combined variable capacitance value of the impedance matching network 106 is Cstep2, the y MHz RF generator generates a complex radio frequency value RF( An RF signal of P(1+m))n. The RF signal with a complex radio frequency value RF(P(1+m))n passes through the branch circuit of the impedance matching network 106. In addition, the additional branch of the impedance matching network 106 receives the RF signal from the output 123 of the x MHz RF generator through the RF cable 127 and the input 125. After receiving the complex RF signals from the x and y MHz RF generators, the impedance matching network 106 generates a modified signal, and the modified signal is provided to the bottom electrode 118 for the period P(1+m+q) During the processing of wafer W.

In various embodiments, the combined variable capacitance value Cstep2 is closer to the combined variable capacitance value Coptimum(P(1+m)) than the combined variable capacitance value Cstep1. For example, the combined variable capacitance value Cstep2 is greater than the combined variable capacitance value Cstep1 and the combined variable capacitance value Coptimum (P(1+m)) is greater than the combined variable capacitance value Cstep2. For another example, the combined variable capacitance value Cstep2 is smaller than the combined variable capacitance value Cstep1 and the combined variable capacitance value Coptimum (P(1+m)) is smaller than the combined variable capacitance value Cstep2.

In various embodiments, the value Coptimum(P(1+m)) and the value Cstep2 are applied to the plasma system 100 but the complex radio frequency value RF(P(1+m))n is not determined and the complex radio frequency value RF( P(1+m))n is applied to the plasma system 100.

Fig. 6 shows the embodiment of Figs. 602 and 604, which are used to illustrate the complex period of an RF signal 606 generated by the y MHz RF generator and an RF signal 608 generated by the x MHz RF generator. Within a period of time. Graph 602 shows the complex power value of the RF signal 606 on the y-axis versus the time t on the x-axis. Graph 604 shows the complex power value of the RF signal 608 on the y-axis versus the time t on the x-axis. The time axis x of the two RF signals 606 and 608 are the same. For example, 10 periods of the RF signal 608 and a period P1 of the RF signal 606 are generated in a time interval t2. In addition, 10 periods of the RF signal 608 and a period P2 of the RF signal 606 occur in a period of time between t2 and t4. In addition, 10 periods of the RF signal 608 and a period P3 of the RF signal 606 are generated in a period between time t4 and t6. Each period of an RF signal generated by an RF generator is sometimes referred to herein as an RF cycle. In each period of the RF signal 606, 10 periods of the RF signal 608 occur. Furthermore, the period P1 is immediately followed by the period P2 of the RF signal 606. The period P2 is immediately followed by the period P3 of the RF signal 606.

In some embodiments, more than one period of the RF signal 608 may occur within a period of the RF signal 606, such as 100 periods, 200 periods, any number of periods between 100 and 200, and so on. Such a ratio between the periods of the RF signal 608 and the RF signal 606 is the ratio between the frequencies of the RF signal 608 and 606.

The period P2 and the period P1 are continuous, and the period P3 and the period P2 are continuous. In addition, the period P3 and the period P1 are not continuous. Between periods P1 and P3, there is an oscillation of period P2.

FIG. 7A is an embodiment of the diagram 700, which is used to illustrate the generation of complex optimal combined variable capacitance values Coptimum, such as Coptimum1, Coptimum2, Coptimum3, etc., from the complex value of the load impedance Zload for various process conditions of the plasma chamber 108. The graph 700 shows an imaginary part of the load impedance Zload, such as reactance, etc. (Im(Zload) on the y-axis), and a real part of the load impedance Zload, such as resistance, etc. (Re(Zload)). Examples of multiple process conditions include various frequency values of the operation of the x MHz RF generator, or various frequency values of the operation of the y MHz RF generator, or the gap between the upper electrode 116 and the chuck 118, or the plasma chamber 108 The temperature or the pressure in the plasma chamber 108, or the power value of the RF signal generated by the x MHz RF generator, or the power value of the RF signal generated by the y MHz RF generator, the gas in the plasma chamber 108 Chemicals, or a combination of two or more of the above. For example, a process condition 1 includes a frequency value frq1 of an RF signal generated by an x MHz RF generator, a power value pwr1 of an RF signal generated by an x MHz RF generator, and a power value of an RF signal generated by a y MHz RF generator. A frequency value frq1, a power value pwr2 of the RF signal generated by the y MHz RF generator, a temperature tmp1 in the plasma chamber 108, a pressure pr1 in the plasma chamber 108, a gap gp1 (mm, mm) , And a chemical of two process gases. A process condition 2 includes a frequency value frq2 of the RF signal generated by the x MHz RF generator, a power value pwr2 of the RF signal generated by the x MHz RF generator, and a frequency of the RF signal generated by the y MHz RF generator Frequency value frq3, a power value pwr3 of the RF signal generated by the y MHz RF generator, a temperature tmp1 in the plasma chamber 108, a pressure pr1 in the plasma chamber 108, a gap gp1 (mm), and two The chemical in the process gas. The value Zload1 corresponds to process condition 1 and the value Zload2 corresponds to process condition 2. Similarly, a value of ZloadQ corresponds to the process condition Q, where Q is an integer greater than zero. For example, ZloadQ is an impedance measured between the output 140 of the impedance matching network 106 and the chuck 118 when the plasma chamber is operated under the process condition Q. In various embodiments, the plasma chamber 108 is operated with a limited number of complex process conditions Q and does not operate outside this limited number of conditions.

7B is a diagram of an embodiment of the model system 102, which is used to illustrate the generation of a complex optimal value Coptimum that makes a voltage reflection coefficient Γ at the input 142 of the model system 102 zero. The processor 134 uses the model system 102 to propagate various values of Zload backward from the output 144 of the model system 102 to determine the complex optimal value Coptimum that makes the voltage reflection coefficient Γ at the input 142 zero. The complex value of Zload is input by the processor 134 from an input device or pre-programmed to be generated by the processor 134, and the complex value of Zload is restricted based on the complex number process conditions. For example, the Zload measured at a point between the output 140 of the impedance matching network 106 and the chuck 118 is Zload1 when there is process condition 1 in the plasma chamber 108. For another example, the Zload measured at the point between the output 140 of the impedance matching network 106 and the chuck 118 is Zload2 when there is process condition 2 in the plasma chamber 108. In this example, the complex value of Zload is limited to Zload1 and Zload2 when the process conditions are limited to process conditions 1 and 2. The plasma chamber 108 is not operated using process conditions other than the above process conditions. In some embodiments, the plasma chamber 108 cannot be operated with process conditions other than the predetermined process conditions.

For each value of Zload, the processor 134 uses the model system 102 to determine a value of the optimal combined variable capacitance value Coptimum. For example, for the value Zload1, the capacitance value Coptimum1 that makes the voltage reflection coefficient Γ at the input 142 of the model system 102 zero is determined. Furthermore, for the value Zload2, the capacitance value Coptimum2 that makes the voltage reflection coefficient Γ at the input 142 of the model system 102 zero is determined.

In some embodiments, the zero value of another parameter at the input 142, such as the power reflection coefficient, is reached instead of reaching the zero value of the voltage reflection coefficient Γ.

FIG. 7C shows an embodiment of table 720 and polynomial (1), both of which are generated by the processor 134. The table 720 contains the correspondence between the complex load impedance value Zload and the complex optimal combined variable capacitance value Coptimum. For example, the processor 134 applies the model system 102 as explained above with reference to FIG. 7B to determine, for a value ZloadQ, a capacitance value CoptimumQ that enables the voltage reflection coefficient Γ at the input 142 of the model system 102 to be zero, where Q is greater than zero. Integer. The value ZloadQ is one of the complex values Zload and the value CoptimumQ is one of the complex values Coptimum. The processor 134 stores the table 720 in the memory device 137. Table 720 is an example of the relationship between the complex load impedance value Zload and the complex capacitance value Coptimum.

In some embodiments, the processor 134 generates a polynomial (1) representing the relationship between the complex optimal combined variable capacitance value Coptimum and the complex load impedance value Zload to replace the table 620 or generate a representative table 620 The polynomial (1) of the relationship between the complex optimal combination variable capacitance value Coptimum and the complex load impedance value Zload. The complex variable capacitance value Coptimum is a function of the real part of Zload and the imaginary part of Zload, and the function is determined by fitting the function to the complex value Coptimum on the graph 600 (FIG. 6A). The function represented by the polynomial (1) is fitted by the processor 134.

FIG. 8A is an embodiment of the graph 800, which is used to illustrate the generation of the complex optimal RF values RFoptimum1, RFoptimum2, RFoptimum3, etc. from the complex optimal capacitance value Coptimum and the complex load impedance value Zload. The graph 800 shows the real part of the complex load impedance value Zload on the x-axis, the imaginary part of the complex load impedance value Zload on the y-axis, and the complex optimal capacitance value Coptimum on the z-axis. The optimal capacitance value Coptimum1 and the load impedance value Zload1 correspond to the optimal RF value RFoptimum1. In addition, the optimal capacitance value Coptimum2 and the load impedance value Zload2 correspond to the optimal RF value RFoptimum2, and the optimal capacitance value Coptimum3 and the load impedance value Zload3 correspond to the optimal RF value RFoptimum3.

FIG. 8B is an embodiment of the model system 102, which is used to illustrate the generation of the complex optimal RF value RFoptimum from the complex optimal capacitance value Coptimum and the complex load impedance value Zload. The processor 134 applies the load impedance value ZloadQ at the output 144 of the model system 102 and initializes the model system 102 to have the value CoptimumQ, and then the model system 102 propagates the value ZloadQ back to determine the voltage at the input 142 of the model system 102 The reflection coefficient Γ is an optimal RF value RFoptimumQ with a minimum value such as a non-zero value, where Q is an integer greater than zero. For example, the processor 134 is initialized to propagate the load impedance value Zload1 back to the model system 102 with the value Coptimum1 to determine a first RF maximum that enables the voltage reflection coefficient Γ at the input 142 of the model system 102 to be a first value. Good value RFA. In addition, the processor 134 uses the model system 102 to propagate the load impedance value Zload1 backward to determine a second RF optimal value RFB that enables the voltage reflection coefficient Γ at the input 142 of the model system 102 to be a second value. The processor 134 compares the first value with the second value to determine that the first value is the minimum of the two, and further determines the value RFA as the one that enables the voltage reflection coefficient Γ at the input 142 to be the minimum. The value RFA is an example of the value RFoptimum1. For another example, the processor 134 is initialized to propagate the load impedance value Zload2 back to the model system 102 with the value Coptimum2 to determine a first RF that enables the voltage reflection coefficient Γ at the input 142 of the model system 102 to be a first value. The best value RFC. In addition, the processor 134 uses the model system 102 to propagate the load impedance value Zload2 backward to determine a second RF optimal value RFD that enables the voltage reflection coefficient Γ at the input 142 of the model system 102 to be a second value. The processor 134 compares the first value with the second value to determine the first value to be the minimum of the two, and further determines the value RFC to be the one that enables the voltage reflection coefficient Γ at the input 142 to be the minimum. The value RFC is an example of the value RFoptimum2. The value RFoptimumQ is one of the complex values RFoptimum.

For another example, the processor 134 applies the load impedance value ZloadQ at the output 144 of the model system 102 and initializes the model system 102 to have the value CoptimumQ, and then propagates the value ZloadQ backwards through the model system 102 to determine whether it can be generated by an RF generator The combination of a voltage reflection coefficient polynomial Γ1 of a state S1 of an RF signal and a voltage reflection coefficient polynomial Γ2 of a state S2 of an RF signal generated by an RF generator has a minimum value such as a non-zero value, An optimal RF value RFoptimumQ of zero value etc. An example of this combination of complex voltage reflection coefficients is A*Γ1 + B*Γ2, where A is a coefficient between 0 and 1 and B is another coefficient between 0 and 1. The coefficients A and B are provided to the processor 132 by the user through the input device. An example of B is (1-A). For example, the processor 134 is initialized to propagate the load impedance value Zload1 back to the model system 102 with the value Coptimum1 to determine a first value for the combination of the voltage reflection coefficients Γ1 and Γ2 at the input 142 of the model system 102 to have a first value. An RF best value RFA. In addition, the processor 134 uses the model system 102 to propagate the load impedance value Zload1 backward to determine a second RF optimal value RFB that enables the combination of the voltage reflection coefficients Γ1 and Γ2 at the input 142 of the model system 102 to have a second value. The processor 134 compares the first value with the second value to determine that the first value is the minimum of the two values, and further determines that the value RFA is the polynomial A*Γ1 + (1 -A)*Γ2 is the minimum value. The value RFA is an example of the value RFoptimum1.

In some embodiments, another parameter at the input 142, such as the power reflection coefficient, or the combination of the parameters of the states S1 and S2, can be minimized instead of minimizing the voltage reflection coefficient Γ or the combination of the voltage reflection coefficients Γ1 and Γ2, or In addition to minimizing the voltage reflection coefficient Γ or the combination of the voltage reflection coefficients Γ1 and Γ2, another parameter at the input 142 such as the power reflection coefficient or the combination of the parameters of the states S1 and S2 is also minimized.

In various embodiments, an RF signal generated by an RF generator during the state S1 has a power level greater than that of the RF signal during the state S2. The power level is, for example, one or more power quantities, The root mean square power amount of one or more power amounts, a power level of a pulse line of an RF signal, etc. Similarly, during the state S1, an RF signal has a frequency level greater than that of the RF signal during the state S2. The frequency level is, for example, one or more frequency quantities, or the root mean square of one or more frequency quantities. The amount of frequency and so on. In these embodiments, the state S1 is referred to as a high state in the text and the state S2 is referred to as a low state in the text.

In some embodiments, an RF signal generated by an RF generator during the state S2 has a power level greater than that of the RF signal during the state S1. Similarly, in these embodiments, the RF signal during the state S2 has a frequency level greater than that of the RF signal during the state S1. The frequency level is, for example, one or more frequency quantities, one or more The root mean square frequency of the frequency and so on. In these embodiments, the state S1 is referred to as a low state in the text and the state S2 is referred to as a high state in the text.

In various embodiments, an RF signal generated by an RF generator during the state S2 has a power level equal to that of the RF signal during the state S1.

In various embodiments, regardless of whether a power level of an RF signal generated by an RF generator during the state S2 is greater than or less than a power level of the RF signal during the state S1, a frequency of the RF signal during the state S2 The level is greater than or less than a frequency level of the RF signal during the state S1.

In some embodiments, the level used in the text, such as a frequency level, a power level, etc., includes one or more values, and the first state, such as state S1, state S2, etc., has a complex value Exclude the complex value of one level of the second state that is different from the first state, such as state S1, state S2, and so on. For example, none of the complex power values of an RF signal during the state S1 is equal to the complex power values of the RF signal during the state S2. For another example, none of the complex frequency values of an RF signal during the state S1 is equal to the complex power value of the RF signal during the state S2.

FIG. 8C is an embodiment of the table 820. The table 820 includes the corresponding relationship between the complex load impedance value Zload, the complex optimal capacitance value Coptimum, and the complex optimal radio frequency value RFoptimum, all of which are used by the processor 134 using the model system 102 produced. For example, the processor 134 applies the model system 102 to the value ZloadQ and the capacitance value CoptimumQ as explained above with reference to FIG. 8B to determine the value RFoptimumQ that enables the voltage reflection coefficient Γ at the input 142 of the model system 102 to be the minimum value, where Q is greater than An integer of zero. The processor 134 stores the table 820 in the memory device 137.

[表II]

[表III]

應注意， R1至R5為電阻值而X1至X5為電抗值。更應注意，表I中的複數值RFoptimum係於模型系統102受到初始化以具有最佳電容值Coptimum1時所產生。又，表II中的複數值RFoptimum係於模型系統102受到初始化以具有最佳電容值Coptimum2時所產生。又，表III中的複數值RFoptimum係於模型系統102受到初始化以具有最佳電容值Coptimum3時所產生。Other examples of the table generated by the processor 134 using the model system 102 as explained above with reference to FIG. 8B are provided below: [Table I]

[Table II]

[Table III]

It should be noted that R1 to R5 are resistance values and X1 to X5 are reactance values. It should be noted that the complex value RFoptimum in Table I is generated when the model system 102 is initialized to have the optimal capacitance value Coptimum1. In addition, the complex value RFoptimum in Table II is generated when the model system 102 is initialized to have the optimal capacitance value Coptimum2. In addition, the complex value RFoptimum in Table III is generated when the model system 102 is initialized to have the optimal capacitance value Coptimum3.

For each value of the load impedance ZloadQ and the optimal capacitance value Coptimum1, the processor 134 finds a column in Table I to find a value Re(Zload) and a row in Table I to find a value Im(Zload), and then based on The values Re(Zload) and Im(Zload) find the optimal value RFoptimumQ. Similarly, for each value of the load impedance ZloadQ and the optimal capacitance value Coptimum2, the processor 134 finds a column in Table II to find a value Re(Zload) and finds a row in Table II to find a value Im(Zload) , And then find the optimal value RFoptimumQ based on the values Re(Zload) and Im(Zload). Furthermore, for each value of the load impedance ZloadQ and the optimal capacitance value Coptimum3, the processor 134 finds a column in Table III to find a value Re(Zload) and finds a row in Table III to find a value Im(Zload), Then find the optimal value RFoptimumQ based on the values Re(Zload) and Im(Zload).

In various embodiments, the two representations of RFoptimumQ and RFoptimum can be used interchangeably herein. In addition, in these embodiments, ZloadQ and Zload can be used interchangeably in this text. Moreover, in these embodiments, Coptimum and CoptimumQ can be used interchangeably in this document.

In some embodiments, the processor 134 looks up tables I, II, and III approximately to generate a polynomial RFoptimumQ = function 3 (Re(Zload), Im(Zload), CoptimumQ), where function 3 is a function. For example, the complex RFoptimumQ value, the complex value of Re(Zload), and the complex value of Im(Zload) in the best fit tables I to III, the processor 134 generates the complex CoptimumQ value in Tables I to III to generate the polynomial RFoptimumQ = Function 3(Re(Zload), Im(Zload), CoptimumQ). The lookup tables I to II and the polynomial RFoptimumQ = function 3 (Re(Zload), Im(Zload), CoptimumQ) are stored in the memory device 137.

FIG. 8C also generates an embodiment of polynomial (2). Each of Table 820 and polynomial (2) is an example of the relationship between the complex load impedance value Zload, the complex optimal capacitance value Coptimum, and the complex optimal radio frequency value RFoptimum. In some embodiments, the processor 134 generates polynomial (2) instead of generating table 820 or generates polynomial (2) in addition to generating table 820. The complex RF value RFoptimum is a function of the complex combined variable capacitance value Coptimum, the real part of the complex Z load value, and the imaginary part of the complex Z load value. This function is achieved by fitting the function to the graph 800 (Figure 8A) The complex value of RFoptimum is determined. The function represented by the polynomial (2) is fitted by the processor 134.

[表V]

9 is a block diagram of an embodiment of the model system 102, which is used to illustrate the generation of complex optimal values Coptimum and RFoptimum that make the voltage reflection coefficient at the input 142 of the model system 102 zero. The voltage reflection coefficient Γ at the input 142 of the model system 102 depends on the complex load impedance value Zload, the complex optimal capacitance value Coptimum such as the variable capacitor position, etc., and the complex RF frequency optimal value RFoptimum. For each load impedance ZloadQ, there is a single combination of the optimal capacitance value CoptimumQ and the RF frequency optimal value RFoptimumQ determined by the processor 134 to enable Γ=0 at the input 142 of the model system 102. For example, the processor 134 applies the load impedance value ZloadQ to the output 144 of the model system 102, and then uses the model system 102 to propagate the value ZloadQ backward to determine the voltage reflection coefficient Γ at the input 142 of the model system 102 to be Zero optimal RF value RFoptimumQ and optimal capacitance value CoptimumQ. The optimal capacitance value CoptimumQ and the RF frequency optimal value RFoptimumQ are sometimes referred to as the modulation value in the text. Using complex modulation values, the impedance matching network 106 modulates the load impedance at the output 140 of the impedance matching network 106 so that the voltage reflection coefficient Γ at the input 128 of the impedance matching network 106 is zero, which is equal to the input 128 An impedance at 50 + 0j Ω, where j is a complex number. Using the model system 102, the processor 134 pre-calculates or generates a look-up table or polynomial function to find the complex modulation value. Examples of lookup tables are: [Table IV]

[Table V]

During the plasma process, for each of the complex load impedances Zload, the processor 134 finds a column in Table IV to find a value of Re(Zload) and finds a row in Table IV to find a value of Im(Zload) Then find the best capacitance value CoptimumQ based on the values Re(Zload) and Im(Zload), such as Coptimum11, or Coptimum12, or Coptimum13, or Coptimum14, or Coptimum15, or Coptimum21, or Coptimum22, or Coptimum23, or Coptimum24, or Coptimum25, Or Coptimum31, or Coptimum32, or Coptimum33, or Coptimum34, or Coptimum35, or Coptimum41, or Coptimum42, or Coptimum43, or Coptimum44, or Coptimum45, or Coptimum51, or Coptimum52, or Coptimum53, or Coptimum54, or Coptimum55, etc. Similarly, for each of the complex load impedances Zload, the processor 134 finds a column in table V to find a value of Re(Zload) and finds a row in table V to find a value of Im(Zload), and then Find the RF frequency optimal value RFoptimumQ based on the values Re(Zload) and Im(Zload), such as RFoptimum11, or RFoptimum12, or RFoptimum13, or RFoptimum14, or RFoptimum15, or RFoptimum21, or RFoptimum22, or RFoptimum23, or RFoptimum24, or RFoptimum25, or RFoptimum31 , Or RFoptimum32, or RFoptimum33, or RFoptimum34, or RFoptimum35, or RFoptimum41, or RFoptimum42, or RFoptimum43, or RFoptimum44, or RFoptimum45, or RFoptimum51, or RFoptimum52, or RFoptimum53, or RFoptimum54, or RFoptimum55. It should be noted that for each CoptimumQ value in Table IV and for each RF optimal value RFoptimumQ in Table V, the voltage reflection coefficient at the input 142 of the matching network model 102 is zero.

In some embodiments, the processor 134 approximates the lookup tables IV and V by generating the following polynomial functions: Coptimum = function 1(Re(Zload), Im(Zload))... Equation (3) RFoptimum = function 2( Re(Zload), Im(Zload))... Equation (4) where function 1 is a function of Re(Zload) and Im(Zload), and function 2 is a function of Re(Zload) and Im(Zload). For example, to best fit the complex values of Re(Zload) and Im(Zload), the processor 134 generates the complex number Coptimum in Table IV to generate the polynomial (3). For another example, the processor 134 best fits the complex numbers RFoptimum, Re(Zload), and Im(Zload) in Table V to generate equation (4). The look-up tables IV and V and equations (3) and (4) are stored in the memory device 137.

FIG. 10 is a block diagram of an embodiment of a plasma system 1000, which is used to illustrate the application of complex optimal values RFoptimum and Coptimum based on complex load impedance values. The plasma system 1000 includes a y MHz RF generator. In some embodiments, the y MHz RF generator is a 400 kHz RF generator, or a 2 MHz RF generator, or a 27 MHz RF generator, or a 60 MHz RF generator. During the processing of the wafer W in the plasma chamber 108, the sensor 124 measures the voltage reflection coefficient Γmi at the output 126 of the y MHz RF generator. The processor 134 receives the voltage reflection coefficient Γmi through the network cable 136 and applies equation (1) to convert the voltage reflection coefficient Γmi into an impedance value Zmi.

The processor 134 applies the impedance value Zmi at the input 142 and propagates the impedance value Zmi forward by the model system 102 to generate the load impedance value ZloadQ at the output 144 in a manner similar to the self-complex value Zmi(P1)n (Figure 1). The method of complex load impedance value ZL(P1)n. The processor 134 receives a table A such as table I, or table II, or table III, or tables IV and V, or table 820 from the memory 137 and determines the value CoptimumQ and the value RFoptimumQ corresponding to the value ZloadQ from the table A. For example, when determining that the load impedance at the output 144 of the model system 102 is Zload1, the processor 134 accesses the table A from the memory 137 and determines the value Coptimum1 and the value RFoptimum1 corresponding to the value Zload1 from the table A. For another example, when determining that the load impedance at the output 144 of the model system 102 is Zload2, the processor 134 accesses the table A from the memory 137 and determines the value Coptimum2 and the value RFoptimum2 corresponding to the value Zload2 from the table A. For another example, when the capacitance value of the impedance matching network 106 and the model system 102 is set to Coptimum1 and the load impedance at the output 144 of the model system 102 is determined to be R1 and the reactance of the load impedance is determined to be X1, the processor The value RFoptimum111 determined by 134 from Table I corresponds to the values R1 and X1. For another example, when a resistor that determines the load impedance at the output 144 of the model system 102 is R1 and the reactance of the load impedance is X1, the processor 134 determines from Table IV that the value Coptimum11 corresponds to the values R1 and X1. Also, in this example, the processor 134 determines from the table V that the value RFoptimum11 corresponds to the values R1 and X1.

For another example, the processor 134 applies polynomial (1) to the value ZloadQ to calculate the value CoptimumQ and applies polynomial (2) to the complex values ZloadQ and CoptimumQ to determine the value RFoptimumQ. For example, the processor 134 applies polynomial (1) to the value Zload1 to calculate the value Coptimum1 and applies polynomial (2) to the complex values Zload1 and Coptimum1 to determine the value RFoptimum1. For another example, the processor 134 applies polynomial (1) to the value Zload2 to calculate the value Coptimum2 and applies polynomial (2) to the complex values Zload2 and Coptimum2 to determine the value RFoptimum2. For another example, the processor 134 receives the polynomial RFoptimumQ = function 3 (Re(Zload), Im(Zload), CoptimumQ) from the memory device 137 and applies the polynomial to the values R1 and X1 and Coptimum1 to generate the value RFoptimum111. In this example, the capacitance value of the impedance matching network 106 and the model system 102 is set to Coptimum1. For another example, the processor 134 receives the equation (3) from the memory device 137 and applies the equation (3) to the values R1 and X1 to determine the value Coptimum1. Also, in this example, the processor 134 receives the equation (4) from the memory device 137 and applies the equation (4) to the values R1 and X1 to determine the value RFoptimum1. For another example, the processor 134 determines whether the RF signal generated by the y MHz RF generator is a multi-state signal. For example, in a recipe provided to the processor 134, it is specified that the RF signal needs to have two states S1 and S2. In this example, the capacitance value of the impedance matching network 140 and the model system 102 is set to Coptimum1. The processor 134 pre-determines the value Coptimum1. In order to minimize the combination of the voltage reflection coefficient polynomial Γ1 of the state S1 of the RF signal generated by the y MHz RF generator and the voltage reflection coefficient polynomial Γ2 of the state S2 of the RF signal, the optimal The RF value RFoptimumQ is provided to the y MHz RF generator.

The processor 134 modifies the formula to include the value RFoptimumQ in the formula and sends the formula to the y MHz RF generator via the network cable 138. After receiving the value RFoptimumQ, the DSP of the y MHz RF generator controls the RF power supply 122 to generate an RF signal with a frequency value RFoptimumQ or within a predetermined range of the frequency value RFoptimumQ. After the RF power supply 122 receives the signal indicating that the RF signal is to be generated, it generates the RF signal and sends the RF signal to the input 130 of the impedance matching network 106 through the RF cable, wherein the RF signal has a frequency value of RFoptimumQ or It falls within the predetermined range of the frequency value RFoptimumQ.

Furthermore, in some embodiments where the combined variable capacitance value of the impedance matching network 106 is to be changed, the processor 134 sends a signal representing the value CoptimumQ to the driving device of the driving component 112 to generate one or more current signals. For example, when the application table I, II, or III, or the polynomial RFoptimumQ = function 3 (Re(Zload), Im(Zload), CoptimumQ), the impedance matching network 106 and the model system 102 are set to the values CoptimumQ, RFoptimumQ It is determined by the value CoptimumQ, so there is no need to reach the value CoptimumQ. In this example, the value ZloadQ is determined when the model system 102 is initialized 102 to have the optimal value CoptimumQ. For another example, when applying Tables IV and V, or equations (3) and (4), the impedance matching network 106 and the model system 102 are not set to the value CoptimumQ and are set to another combined variable capacitance value. The combined variable capacitor value can be adjusted to reach the value CoptimumQ.

The driving device generates one or more current signals based on the capacitance value CoptimumQ and sends them to the corresponding one or more stators of the driving component 112 corresponding to one or more motors. One or more rotors of the driving component 112 that are in contact with the corresponding one or more stator electric fields rotate to move the connecting mechanism 114 to change the combined variable capacitance value of the branch circuit of the impedance matching network 106 to CoptimumQ. The branch circuit of the impedance matching network 106 with the combined variable capacitance value CoptimumQ receives the RF signal with the radio frequency value RFoptimumQ from the output 126 through the input 128 and the RF cable 130, and adjusts the impedance of the load connected to the impedance matching network 106 It is matched with the impedance of the source connected to the impedance matching network 106 to generate a modified signal. Examples of sources include a y MHz RF generator and RF cable 130. The modified signal is provided from the output 140 of the branch circuit of the impedance matching network 106 through the RF transmission line 132 to the chuck 118. When the modified signal and one or more process gases are provided to the chuck 118, plasma is generated or maintained in the gap between the chuck 118 and the upper electrode 116 to process the wafer W.

By using table A such as table I, or table II, or table III, or table IV and V, or table 820, etc., or polynomial A such as polynomial (2), or polynomial RFoptimumQ = function 3(Re(Zload), Im (Zload), CoptimumQ), or equations (3) and (4) to generate complex values RFoptimumQ and CoptimumQ can increase the operating speed of the plasma system 1000 for processing wafer W. For example, after the sensor 124 measures the voltage reflection coefficient Γmi, there is no need to use the model system 102 to determine the complex values RFoptimumQ and CoptimumQ. In contrast, before the sensor 124 measures the voltage reflection coefficient Γmi, the complex values RFoptimumQ and CoptimumQ are pre-stored in Table A and/or polynomial A has been generated. Once the sensor 124 measures the voltage reflection coefficient Γmi, the processor 134 can take the complex values RFoptimumQ and CoptimumQ from Table A and/or the processor 134 can apply the polynomial A to calculate the complex values RFoptimumQ and CoptimumQ. When measuring the voltage reflection coefficient Γmi, it is not necessary to use the model system 102 to calculate the complex values RFoptimumQ and CoptimumQ, which can save time during the processing of the wafer W. In addition, applying the complex values RFoptimumQ and CoptimumQ to the plasma system 1000 can reduce the power reflected to the y MHz RF generator to improve the efficiency of processing the wafer W.

In some embodiments, any of the complex-valued RFoptimumQ or the complex-valued CoptimumQ falls outside the physically accessible space. For example, the frequency modulation range of a 60 MHz RF generator is from 57.00 MHz to 63.00 MHz, and the value RFoptimum1 determined by the model system 102 is lower than 57 MHz or higher than 63 MHz. In such cases, in terms of scaled distance, the optimal operating conditions are located on the boundary of the restricted space closest to the out-of-bounds solutions such as RFoptimumQ, CoptimumQ, etc. Example of scaling distance = [(capacitor position) – (CoptimumQ)]^2 + k^2 * [(RF frequency) – (RFoptimumQ)]^2, where k is provided by the input device to the processor 134 as input A predetermined value.

In various embodiments, the processor 134 weights each of the complex number of measured voltage reflection coefficients Γmi according to a pre-assigned weight. The complex weight applied by the processor 134 to the complex voltage reflection coefficient Γmi is determined by the processor 134 through the input device receiving input and is determined based on engineering knowledge and/or process conditions. The weighted complex voltage reflection coefficient wΓmin can be applied to the model system 102 instead of applying the complex voltage reflection coefficient Γmi to determine the complex load impedance ZloadQ, where each w is a pre-assigned weight.

FIG. 11 is an embodiment of FIG. 1100, which is used to illustrate the variation of the input impedance of the impedance matching network 106 at the input 128 when the y MHz RF generator is a 60 MHz RF generator. The real and imaginary parts of the galaxy are calculated from the input impedance and show changes over time due to the effect of the RF signal generated by the x MHz RF generator. The graph 1100 shows the real part of Gara on the x-axis and the imaginary part of Gara on the y-axis. The graph 1100 shows that the real part and the imaginary part of Gara form a pattern. As shown in Figure 1100, a complete cycle of the pattern requires a period of x MHz RF generator or about 2.5 microseconds. In some embodiments, the complete period needs to be greater than or less than 2.5 microseconds, such as 2 microseconds, 3 microseconds, a range between 2.5 to 4 microseconds, and a range between 1 to 2.5 microseconds. Scope etc.

Fig. 12 is an embodiment of Fig. 1200, which is used to illustrate the Fourier transformation of a voltage reflected to the y MHz RF generator when the y MHz RF generator is a 60 MHz RF generator, which is expressed as y MHz RF The fraction of the forward feed power supplied by the generator. Figure 1200 shows the square of the voltage versus the frequency of the RF signal generated by the y MHz RF generator. The square of the voltage is a representative quantity of the power reflected to the y MHz RF generator. In some embodiments, the reflected power at the fundamental frequency of the RF signal generated by the y MHz RF signal is filtered by the system and method described in the article. The small reflected power peak at the fundamental frequency in the Fourier spectrum of Fig. 1100 is shown in Fig. 1200. In addition, there are large reflected power peaks at intermodulation frequencies such as 60 MHz ± 400 kHz. The system and method described in the article apply the model system 102 to reduce the power reflected to the y MHz RF generator at various frequencies, such as y MHz ± x MHz intermodulation frequency, 60 MHz ± 400 kHz intermodulation frequency Wait. The system and method described in the article find the best combination of variable capacitance value and complex radio frequency value to minimize all reflected power not only at the fundamental frequency but also at other frequencies such as the intermodulation frequency of y MHz ± x MHz.

In order to reduce the reflected power, in some embodiments, the front-feed and reflected waveform data of the y MHz RF generator are collected at a rate to capture the changes during one of the x MHz RF generators. For example, such data collection is performed in at least 2.5 microseconds at a rate of at least one billion samples per second. Then analyze the collected data in the range of a complex number such as 0.1 microseconds, and decompose the 2.5 microseconds during x MHz into 25 separate impedance measurement values. Figure 11 shows the result of analyzing the multiple 0.1 microsecond segments of the waveform, where a time difference between the multiple segments is 0.03 microseconds, so there is a slight overlap between different points. Next, calculate an average of the power reflection coefficient such as |Γ|^2 to obtain an average power reflected to the y MHz RF generator during a period of x MHz. In the model system 102, the processor 134 changes the combined variable capacitance value and the RF frequency, and the processor 134 records how the power reflection coefficient changes for every 25 impedance measurement values to the memory device 137. Next, the processor 134 determines the value of the capacitor position and/or the RF frequency of the y MHz RF generator that can minimize the overall variable capacitance value of the impedance matching network 106 such as the average power reflection coefficient. In various embodiments, the total calculation time will be greater than 2.5 microseconds but can achieve a power delivery improvement on a time scale of about a few milliseconds. By using the model system 102, the y MHz RF generator can be modulated to an RF frequency so as to reach the minimum average value of the power reflection coefficient |Γ|^2 on average during a period of x MHz RF frequency. For a period of the RF signal generated by the x MHz RF generator, the same combined variable capacitance capacitor value and RF frequency are used.

In some embodiments, the frequency of the RF signal generated by the y MHz RF generator is modulated within a single RF cycle of the RF signal generated by the x MHz RF generator. For example, one RF period of the RF signal generated by the x MHz RF generator, such as a 2.5 microsecond period, is segmented into five segments each of 0.5 microseconds. A different y MHz RF frequency is applied in each of the complex sections, and each of the complex frequencies is an optimal value determined by the model system 102 with respect to the optimal value of the combined variable capacitance value of the model system 102 frequency. For another example, the 2.5 microsecond period of one cycle of the RF signal generated by the x MHz RF generator is segmented into four segments each of 0.625 microseconds, and y is determined during each of the four segments. A different frequency of the RF signal generated by the MHz RF generator. The complex frequency determined from the model system 102 can minimize a reflected power coefficient at the output 126 or input 128 (FIG. 1) of the y MHz RF generator during each segment. For another example, the RF frequency of the RF signal generated by the y MHz RF generator at x MHz is modulated by some simple functions such as sine wave, cosine wave, etc. The processor 134 obtains the 25 initial measurement values obtained at the output 126 of the y MHz RF generator to calculate the amplitude and phase of the frequency modulation, and reduce the power reflection coefficient of the period average. In several embodiments, the frequency of the y MHz RF generator is adjusted in microseconds, sub-microseconds, or milliseconds.

In some embodiments, the complex RF signals generated by the x MHz RF generator and the y MHz RF generator have complex states. For example, the x MHz RF generator has operating states S1 and S2 and the y MHz RF generator is the same. The power level of an RF signal generated by an RF generator during the state S1 is greater than the power level of an RF signal generated by the RF generator during the state S2. For example, an RF generator in the state S1 generates a complex power envelope of an RF signal that has a higher power level than a complex power envelope of the RF signal during the state S2.

In various embodiments, the complex RF signals generated by the x MHz RF generator and the y MHz RF generator are continuous. For example, each of the x and y MHz RF generators has a single state.

It should be understood that in some of the foregoing embodiments, an RF signal is supplied to the lower electrode of the chuck 118 and the upper electrode 116 is grounded. In various embodiments, an RF signal can be supplied to the upper electrode 116 and the lower electrode of the chuck 118 is grounded.

The embodiments described in the text can be implemented using various computer system configurations. The computer system includes a handheld hardware unit, a microprocessor system, a microprocessor system or programmable consumer electronic devices, a microcomputer, a host computer, etc. The embodiments described in the article can also be implemented in a distributed computing environment. In the distributed computing environment, tasks are performed by a plurality of remote processing hardware units linked via a computer network.

In some embodiments, the controller is part of the system, and the system may be part of the above examples. Such systems may include semiconductor process equipment, which includes a process tool or a plurality of process tools, a process chamber or a plurality of process chambers, a process platform or a plurality of process platforms, and/or specific process components (wafer platform, gas flow system) Wait). This system is integrated with some electronic devices that are used to control the operation of the system before, during and after semiconductor wafer or substrate processing. These electronic devices are called "controllers", which can control various components or sub-components of the system. Depending on the process requirements and/or system type, the controller can be programmed to control any process disclosed in the article, including process gas delivery, temperature setting (such as heating and/or cooling), pressure setting, vacuum setting, power setting, RF Generator setting, RF matching circuit setting, frequency setting, flow rate setting, fluid delivery setting, position and operation setting, wafer transfer entering or leaving the tool and other transfer equipment and/or load interlocking connected to or interfacing with the system mechanism.

In summary, in various embodiments, the controller can be defined as an electronic device with various integrated circuits, logic, memory and/or software, which can receive instructions, issue instructions, control operations, enable cleaning operations, Enable endpoint measurement, etc. An integrated circuit may include a chip in the form of firmware that stores program instructions, a digital signal processor (DSP), a chip defined as an ASIC, PLD, one or more microprocessors, or the ability to execute program instructions (such as software ) Microcontroller. The program commands can be commands in the form of various independent settings (or program files) that communicate with the controller, and are defined as operating parameters used for processing on or against semiconductor wafers. In some embodiments, the operating parameters are one or more during the manufacturing process of one or more films, materials, metals, oxides, silicon, silicon dioxide, surfaces, circuits, and/or wafers by the process engineer. Part of a recipe defined by multiple process steps.

In some embodiments, the controller is a part of a computer integrated into the system, coupled to the system, connected to the system via a network, or a combination thereof, or the controller is coupled to the computer. For example, the controller can be located in the "cloud" or in all or part of the factory host computer system, which allows users to remotely access the wafer process. The controller can enable remote access to the system to monitor the current progress of manufacturing operations, view the history of past manufacturing operations, view driving force or performance metrics from multiple manufacturing operations, change existing process parameters, set process steps to conform to existing processes, Or start a new process.

In some instances, the remote computer (or server) can provide process recipes to the system via a computer network, and the network includes a local area network or the Internet. The remote computer may include a user interface, which allows the user to enter or program parameters and/or settings, and then communicate with the system from the remote computer. In some instances, the controller receives commands in the form of settings for processing wafers. It should be understood that the setting is specifically for the type of process to be performed on the wafer and the type of tool that the controller uses to interface or control. Therefore, as described above, a decentralized controller may include one or more discrete controllers that are interconnected by a network and work toward a common purpose as described in the process. An example of a distributed controller for such purposes is one or more integrated circuits on the process room, which are connected with one or more integrated circuits located remotely (for example, at the platform level or part of a remote computer) Communication and joint control of the process in the process room.

Without limitation, in various embodiments, the system may include a plasma etching chamber, a deposition chamber, a rotating rinse chamber, a metal plating chamber, a clean room, an edge etching chamber, a physical vapor deposition (PVD) chamber, and a chemical vapor deposition chamber. (CVD) chamber, atomic layer deposition (ALD) chamber, atomic layer etching (ALE) chamber, ion implantation chamber, orbital chamber, and any other semiconductor processing chamber related to or used for the manufacture of semiconductor wafers.

It should be noted that although the above operation refers to a parallel plate plasma chamber such as a capacitively coupled plasma chamber, etc., in some embodiments, the above operation can be applied to other types of plasma chambers such as inductively coupled plasma (ICP ) Reactor, transformer coupled plasma (TCP) reactor, conductor tool, plasma chamber of dielectric tool, plasma chamber containing electron cyclotron resonance (ECR) reactor, etc. For example, the x MHz radio frequency generator, the y MHz radio frequency generator, and the z MHz radio frequency generator are coupled to the inductance in the ICP plasma chamber. Examples of inductor shapes include solenoids, dome-shaped coils, plate-shaped coils, and the like.

As mentioned above, depending on the process operation that the tool intends to perform, the controller can communicate with one or more of the following: circuits or modules of other tools, components of other tools, cluster tools, interfaces of other tools, adjacent Tool, proximity tool, tool located in the factory, host computer, another controller, or material transportation tool used to load and unload the wafer container into and out of the tool position and/or loading interface in the semiconductor manufacturing factory.

Considering the above-mentioned embodiments, it should be understood that certain embodiments can perform various computer-implemented operations involving data stored in a computer system. These computers need to manipulate physical quantities to perform operations.

Some embodiments also relate to hardware units or devices used to perform such operations. The equipment can be specially constructed for special purpose computers. When a computer is defined as a computer for special purposes, in addition to being able to operate for special purposes, this computer can also perform other processes, program executions, or other subprograms that are not for special purposes.

In some embodiments, operations can be performed by a selectively activated computer or can be configured by one or more computer programs stored in computer memory or obtained from a computer network. When the data is obtained from a computer network, the data can be processed by other computers on the computer network, such as electrical terminal computing resources.

One or more of the embodiments described in the text can also be made into a computer-readable code on a non-transitory computer-readable medium. The non-transitory computer-readable medium can be any data storage hardware unit, such as a memory device, that can store data and can be read by a computer system later. Examples of non-transitory computer-readable media include hard disks, network attached storage (NAS), ROM, RAM, compact disc-ROM (CD-ROM), recordable CD (CD-R), and rewritable CD (CD-RW), magnetic tape and other optical and non-optical storage hardware units. In some embodiments, the non-transitory computer-readable medium may include a computer-readable physical medium dispersed in a network-coupled computer system, so the computer-readable code is stored and executed in a distributed manner.

Although some of the above method operations are described in a specific order, it should be understood that, in various embodiments, other steps may be performed between method operations or the method operations may be adjusted to slightly differ in time of occurrence, or may be changed Method operations are distributed to systems that allow method operations to be performed at various intervals, or method operations may be performed in a different order than shown in the text.

It should also be noted that without departing from the scope of the various embodiments described herein, in one embodiment, one or more features from any of the above embodiments may be combined with one or more features of any other embodiment. Combine.

In order to allow those skilled in the art to understand the present invention clearly, the foregoing embodiments have been described in detail, and it should be understood that certain changes and modifications can be made within the scope of the attached patent application. Therefore, these embodiments should be regarded as illustrative rather than restrictive, and the embodiments are not limited to the details described in the text, and these embodiments can be modified within the scope and equivalents of the attached application.

100‧‧‧Plasma system 102‧‧‧Model system 104‧‧‧RF generator 106‧‧‧Impedance matching network 108‧‧‧Plasma chamber 110‧‧‧Host computer system 112‧‧‧Drive components 114‧ ‧‧Connecting mechanism 116‧‧‧Upper electrode 118‧‧‧Clamp 120‧‧‧Upper surface 121‧‧‧RF power supply 122‧‧‧RF power supply 123‧‧‧output 124‧‧‧sensor 125‧‧‧ Input 126‧‧‧Output 127‧‧‧RF cable 128‧‧‧Input 130‧‧‧RF cable 132‧‧‧RF transmission cable 134‧‧‧Processor 136‧‧‧Network cable 137‧‧‧Memory Body device 138‧‧‧Network cable 140‧‧‧Output 142‧‧‧Input 144‧‧‧Output 602‧‧‧Figure 604‧‧‧Figure 606‧‧‧RF signal 608‧‧‧RF signal 700‧‧ ‧Picture 720‧‧‧Table 800‧‧‧Picture 820‧‧‧Table 1000‧‧‧Plasma system 1100‧‧‧Picture 1200‧‧‧Picture

The embodiments of the present invention can be understood with reference to the following description of the drawings.

FIG. 1 is a diagram of an embodiment of a plasma system, which illustrates the generation of complex load impedance ZL(P1)n during a period P1 of a radio frequency (RF) signal generated by an x megahertz (MHz) RF generator.

2 is a diagram of an embodiment of a model system which is initialized to have a complex radio frequency value RF1(P1)o and a variable capacitance value C1 to determine the complex radio frequency value RF(P1)n.

Figure 3 is a diagram of an embodiment of a plasma system, which is used to illustrate the use of a model system to generate a complex load impedance ZL(P) during a period P(1+m) of an RF signal generated by an x MHz RF generator (1+m))n.

Figure 4 is a diagram of an embodiment of a model system which is initialized to have a complex radio frequency value RF(P1)n and a variable capacitance value Cstep1 to determine the complex radio frequency value RF(P(1+m)) n.

Fig. 5 is a diagram of an embodiment of a plasma system, which is used to illustrate the use of a capacitance value Coptimum(P(1+m)) and the use of a complex radio frequency value RF(P(1+m))n at x MHz During a period P(1+m+q) of an RF signal generated by the RF generator, a wafer is processed.

Figure 6 shows a complex embodiment of the complex diagram, which is used to illustrate the complex period of an RF signal generated by the y MHz RF generator and the complex period occurs during a period of an RF signal generated by the x MHz RF generator Inside.

FIG. 7A is an embodiment of a diagram, which is used to exemplify the complex value of the load impedance Zload used for various process conditions of a plasma chamber to generate the complex optimal combined variable capacitance value Coptimum.

FIG. 7B is a diagram of an embodiment of a model system, which is used to illustrate the generation of a complex optimal combined variable capacitance value Coptimum that makes a voltage reflection coefficient Γ at the input of the model system zero.

FIG. 7C is an embodiment of a table and a polynomial, both of which are generated by the processor applying the model system before processing the wafer.

FIG. 8A is an embodiment of a diagram, which is used to illustrate the generation of the complex optimal RF value from the complex optimal combined variable capacitance value Coptimum and the complex load impedance value Zload.

FIG. 8B is an embodiment of a model system, which is used to illustrate the generation of an optimal RF value RFoptimumQ from an optimal combination of variable capacitance value CoptimumQ and a load impedance value ZloadQ.

FIG. 8C is an embodiment of a table. The table includes the correspondence between the complex load impedance value Zload, the complex optimal capacitance value Coptimum, and the complex optimal radio frequency value RFoptimum.

FIG. 9 is a diagram of an embodiment of a model system, which is used to illustrate the generation of the optimal RF value RFoptimumQ and the optimal combined variable capacitance value CoptimumQ that make a reflection coefficient of the input of the model system zero.

FIG. 10 is a block diagram of an embodiment of a plasma system, which is used to illustrate the application of the optimal values RFoptimumQ and CoptimumQ based on the load impedance value ZloadQ.

FIG. 11 is an embodiment of a diagram, which is used to illustrate the variation of an input impedance of an impedance matching network when the y MHz RF generator is a 60 MHz RF generator.

FIG. 12 is an embodiment of a diagram, which is used to illustrate the Fourier transformation of a voltage reflected to the y MHz RF generator when the y MHz RF generator is a 60 MHz RF generator.

100‧‧‧Plasma system

102‧‧‧Model System

104‧‧‧RF generator

106‧‧‧Impedance matching network

108‧‧‧Plasma Room

110‧‧‧Host computer system

112‧‧‧Drive components

114‧‧‧Connecting mechanism

116‧‧‧Upper electrode

118‧‧‧Chuck

120‧‧‧Upper surface

121‧‧‧RF power supply

122‧‧‧RF power supply

123‧‧‧output

124‧‧‧Sensor

125‧‧‧input

126‧‧‧output

127‧‧‧RF cable

128‧‧‧input

130‧‧‧RF cable

132‧‧‧RF transmission line

134‧‧‧Processor

136‧‧‧Network cable

137‧‧‧Memory device

138‧‧‧Network cable

140‧‧‧Output

142‧‧‧input

144‧‧‧Output

Claims (26)

A method for reducing the power reflected toward a radio frequency (RF) generator, comprising: receiving a first plurality of measured input parameter values, the first plurality of measured input parameter values being related to the operation of the radio frequency (RF) generator Sensed between an output of the RF generator and an input of the impedance matching network when the first complex radio frequency value and an impedance matching network has a first variable measurable factor; initialize one or more Computer models such that the one or more computer models have the first variable measurable factor and the first complex radio frequency value; the one or more computer models have the first variable measurable factor And the first complex radio frequency value, use the one or more computer models to determine the first complex output parameter value from the input parameter value measured by the first complex number; use the first complex output parameter value and the one or more A computer model determines a first complex favorable radio frequency value, wherein each of the first complex favorable radio frequency values makes a reflection coefficient at an input of the one or more computer models a minimum value, wherein the first complex favorable radio frequency value is determined to be a minimum value. The step of a plurality of favorable radio frequency values includes: one of the first plurality of output parameter values from an output of the one or more computer models passes through the one or more computer models and propagates back to the one or more computer models The input to determine that the value of the reflection coefficient is the lowest among the complex reflection coefficient values at the input of the one or more computer models; and from the backward propagation step, one of the first complex favorable radio frequency values is determined , Which makes the reflection coefficient at the input of the one or more computer models have the lowest value; controls the RF generator so that the RF generator operates at the first complex favorable radio frequency value to reduce the direction toward the RF The power reflected by the generator. For example, the method of reducing the power reflected to the radio frequency (RF) generator in the first item of the scope of patent application, The step of receiving the first plurality of measured input parameter values is performed during a first RF cycle of a second RF generator, and the initialization step is performed during the first RF cycle, wherein The one or more computer models include a model of the impedance matching network, wherein the reflection coefficient is a minimum value for the first RF period, and the step of controlling the RF generator is performed in the second RF generator Is executed during a second RF period, and the method further includes: receiving a second plurality of measured input parameter values during the second RF period of the second RF generator, the second plurality of measured input parameters The value is sensed between the output of the RF generator and the input of the impedance matching network when the RF generator is operating at the first plurality of favorable radio frequency values; the one or more are initialized for the second RF period Computer models so that the one or more computer models have the first plural favorable radio frequency value; when the one or more computer models have the first plural favorable radio frequency value, use the one or more computer models from the first Two complex numbers of measured input parameter values determine a second complex output parameter value for the second RF cycle; use the second complex output parameter value and the one or more computer models to determine a second complex favorable radio frequency value, wherein the first Each of the two complex favorable radio frequency values makes the reflection coefficient of the second RF cycle at the input of the one or more computer models a minimum value; and a third RF in the second RF generator During the period, the RF generator is controlled so that the RF generator operates at the second plurality of favorable radio frequency values. For example, in the method for reducing the power reflected toward a radio frequency (RF) generator in the second patent application, the third RF period is continuous with the second RF period. For example, the method for reducing the reflected power toward a radio frequency (RF) generator in the scope of the patent application, wherein the third RF period is after the second RF period and between the second RF period and the third RF period There are one or more intermediate RF cycles. For example, the method for reducing the power reflected toward a radio frequency (RF) generator in the second patent application, wherein the second RF period is continuous with the first RF period. For example, the method for reducing the reflected power toward a radio frequency (RF) generator in the second scope of the patent application, wherein the second RF period is after the first RF period and between the first RF period and the second RF period There are one or more intermediate RF cycles. For example, the method for reducing the reflected power toward a radio frequency (RF) generator in the first item of the patent application further includes calculating an optimal variable capacitance value by using the first complex output parameter value and the one or more computer models, where The optimal variable capacitance value is such that a weighted reflection coefficient at the input of the one or more computer models is the minimum value. For example, the method for reducing the power reflected toward a radio frequency (RF) generator in the first patent application, wherein the first complex number of measured input parameter values is received from a sensor connected to the output of the RF generator Wherein the output of the RF generator is connected to the input of the impedance matching network via an RF cable. For example, the method for reducing the power reflected toward a radio frequency (RF) generator in the first patent application, wherein the one or more computer models include a model of an RF cable, a model of the impedance matching network, and an RF transmission line The model of, wherein the RF cable is connected between the output of the RF generator and the input of the impedance matching network, and the RF transmission line is connected to an output of the impedance matching network. For example, the method for reducing the power reflected to a radio frequency (RF) generator in the scope of the patent application, wherein the first complex output parameter value is determined by the following steps: the input parameter value measured by the first complex number Propagate forward through the model of the RF cable so that the model of the RF cable Generates a first complex intermediate value at an output of the impedance matching network; propagates the first complex intermediate value forward through the model of the impedance matching network to generate a second complex intermediate value at an output of the impedance matching network of the model ; And the second complex intermediate value is propagated forward through the model of the RF transmission line to generate the first complex output parameter value. For example, the method for reducing the power reflected toward a radio frequency (RF) generator in the ninth patent application, wherein the reflection coefficient is the minimum value at an input of the model of the RF cable. For example, the method for reducing the power reflected toward a radio frequency (RF) generator in the scope of the patent application, wherein the first complex number of measured input parameter values includes a complex impedance value, and the first variable measurable factor Contains the capacitance of the impedance matching network. For example, the method for reducing the reflected power toward a radio frequency (RF) generator in the first patent application, wherein the one or more computer models are initialized so that when the one or more computer models have the impedance matching network A combined capacitance has the first variable measurable factor. A controller includes: a processor configured to: receive a first plurality of measured input parameter values, the first plurality of measured input parameter values are operated by a radio frequency (RF) generator operating on the first complex number RF value and an impedance matching network with a first variable measurable factor are sensed between an output of the RF generator and an input of the impedance matching network; initialize one or more computer models to Make the one or more computer models have the first variable measurable factor and the first complex radio frequency value; make the one or more computer models have the first variable measurable factor and the first In the case of a complex radio frequency value, use the one or more computer models to determine the first complex output parameter value from the input parameter value measured by the first complex number; Utilize the first complex output parameter value and the one or more computer models to determine a first complex favorable radio frequency value, wherein each of the first complex favorable radio frequency values is the input of one of the one or more computer models A reflection coefficient is a minimum value, wherein in order to determine the first complex number of favorable radio frequency values, one of the first complex number output parameter values is obtained from an output of the one or more computer models through the one or more computer models Backward propagation to the input of the one or more computer models, wherein one of the first complex output parameter values is propagated backward to determine the value of the reflection coefficient at the input of the one or more computer models The complex reflection coefficient value is the lowest, wherein one of the first complex output parameter values is propagated backwards to determine one of the first complex favorable radio frequency values, which makes the input of the one or more computer models The reflection coefficient at the position has the lowest value; and controlling the RF generator so that the RF generator operates at the first plurality of favorable radio frequency values to reduce the power reflected toward the RF generator; and a memory device connected to The processor. For example, the controller of the 14th item of the scope of patent application, wherein the first complex number of measured input parameter values are received during a first RF period of a second RF generator, wherein the one or more computer models are The structure is initialized during the first RF period, wherein the one or more computer models include a model of the impedance matching network, wherein the reflection coefficient is a minimum value for the first RF period, and the RF generator It is controlled during a second RF cycle of the second RF generator, where the processor is constructed to: During the second RF period of the second RF generator, a second plurality of measured input parameter values is received, and the second plurality of measured input parameter values is due to the fact that the RF generator is operating on the first complex number. The radio frequency value is sensed between the output of the RF generator and the input of the impedance matching network; the one or more computer models are initialized for the second RF cycle so that the one or more computer models have the A first complex number of favorable radio frequency values; when the one or more computer models have the first complex number of advantageous radio frequency values, the input parameter values measured from the second complex number using the one or more computer models are directed to the second RF The period determines the second complex output parameter value; the second complex output parameter value and the one or more computer models are used to determine the second complex advantageous radio frequency value, wherein each of the second complex advantageous radio frequency values makes the one or The reflection coefficient of the second RF cycle at the input of a plurality of computer models is a minimum; and during a third RF cycle of the second RF generator, the RF generator is controlled so that the RF generator is at The second complex number operates at a favorable radio frequency value. For example, the controller of item 14 of the scope of patent application, wherein the first plurality of measured input parameter values are received from a sensor connected to the output of the RF generator, wherein the output of the RF generator is It is connected to the input of the impedance matching network via an RF cable. For example, the controller of claim 14, wherein the one or more computer models include models generated by the processor, and the one or more computer models include a model of an RF cable, and a model of the impedance matching network Model, and a model of an RF transmission line, wherein the RF cable is connected between the output of the RF generator and the input of the impedance matching network, and the RF transmission line is connected to one of the impedance matching network Output. For example, in the controller of item 17 of the scope of patent application, in order to determine the value of the first complex number output parameter, the processor is constructed to: propagate the measured input parameter value of the first complex number forward through Pass the model of the RF cable to generate a first complex intermediate value at an output of the model of the RF cable; propagate the first complex intermediate value forward through the model of the impedance matching network to A second complex intermediate value is generated at an output of the model of the impedance matching network; and the second complex intermediate value is propagated forward through the model of the RF transmission line to generate the first complex output parameter value. Such as the controller of the 17th patent application, wherein the reflection coefficient is the minimum value at an input of the model of the RF cable. For example, the controller of item 14 of the scope of patent application, wherein the first complex number of measured input parameter values includes a complex impedance value, and the first variable measurable factor includes the capacitance of the impedance matching network. For example, the controller of claim 14, wherein the one or more computer models are initialized to have the first variable capacity when the one or more computer models have a combined capacitance of the impedance matching network The measured factor. A plasma system includes: a radio frequency (RF) generator configured to operate on a first complex radio frequency value to generate an RF signal; an impedance matching network having an RF generator connected to the RF generator via an RF cable An input of output, wherein the impedance matching network is constructed with a first variable measurable factor; a host computer is connected to the RF generator, wherein the host computer is constructed to: receive the first complex quantity Measured input parameter values, the first complex number of measured input parameter values are sensed between the output of the RF generator and the input of the impedance matching network; initialize one or more computer models to make The one or more computer models have the first variable measurable factor and the first complex radio frequency value; When the one or more computer models have the first variable measurable factor and the first complex radio frequency value, the one or more computer models are used to determine the first complex number of input parameter values measured A complex output parameter value; using the first complex output parameter value and the one or more computer models to determine a first complex favorable radio frequency value, wherein each of the first complex favorable radio frequency values makes the one or more computers A reflection coefficient at one input of the model is a minimum value. In order to determine the first complex number favorable radio frequency value, one of the first complex output parameter values is passed from an output of the one or more computer models through the one or A plurality of computer models are propagated backward to the input of the one or more computer models, wherein one of the first complex output parameter values is propagated backward to determine the value of the reflection coefficient in the one or more computer models The input of is the lowest in the complex reflection coefficient value, where one of the first complex output parameter values is propagated backwards to determine one of the first complex favorable radio frequency values, which makes it in the one or more The reflection coefficient at the input of the computer model has the lowest value; and the RF generator is controlled so that the RF generator operates under the first plural favorable radio frequency value to reduce the power reflected toward the RF generator. For example, the plasma system of the 22nd patent application, wherein the input parameter value measured by the first complex quantity is received during a first RF period of a second RF generator, wherein the one or more computer models The system configuration is initialized during the first RF cycle, wherein the one or more computer models include a model of the impedance matching network, wherein the reflection coefficient is a minimum value for the first RF cycle, and the RF generation The device is controlled during a second RF period of the second RF generator, The host computer is configured to receive a second plurality of measured input parameter values during the second RF period of the second RF generator, and the second plurality of measured input parameter values are related to the RF generator The device is sensed between the output of the RF generator and the input of the impedance matching network when operating at the first complex favorable radio frequency value; the one or more computer models are initialized for the second RF period to make the One or more computer models have the first complex number of favorable radio frequency values; when the one or more computer models have the first complex number of favorable radio frequency values, use the one or more computer models to measure from the second complex number The input parameter value determines the second complex output parameter value for the second RF period; the second complex output parameter value and the one or more computer models are used to determine the second complex favorable radio frequency value, wherein the second complex favorable radio frequency value is Each makes the reflection coefficient of the second RF cycle at the input of the one or more computer models a minimum value; and controls the RF generation during a third RF cycle of the second RF generator So that the RF generator operates at the second complex favorable radio frequency value. For example, the plasma system of item 23 of the scope of patent application, wherein the first complex number of measured input parameter values is received from a sensor connected to the output of the RF generator. For example, the plasma system of item 23 of the scope of patent application, wherein the one or more computer models are generated by the host computer, and the one or more computer models include a model of an RF cable and a model of the impedance matching network , And a model of an RF transmission line, wherein the RF cable is connected between the output of the RF generator and the input of the impedance matching network, and the RF transmission line is connected to an output of the impedance matching network . For example, in the plasma system of the 25th patent application, in order to determine the first complex output parameter value, the host computer is constructed to: propagate forward the input parameter value measured by the first complex number Pass the model of the RF cable to generate a first complex intermediate value at an output of the model of the RF cable; propagate the first complex intermediate value forward through the model of the impedance matching network to A second complex intermediate value is generated at an output of the model of the impedance matching network; and the second complex intermediate value is propagated forward through the model of the RF transmission line to generate the first complex output parameter value.

Images