Automatic impedance matching method for L-type radio frequency matcher with uniform time convergence

Abstract

The invention provides an automatic impedance matching method for an L-type radio frequency matcher with uniform time convergence, which combines feedforward calculation and multivariable Newton extremum search control, introduces an inverse Hessian matrix estimator, eliminates the dependence of algorithm convergence time on initial conditions and a standing-wave ratio-capacitance value function on a Hessian matrix, and improves the impedance matching accuracy of the L-type radio frequency matcher. Therefore, the convergence time is more uniform and can be freely adjusted. The convergence time can be further shortened and the initial matching condition can be optimized by adding feedforward calculation. According to the method provided by the invention, the convergence time is reduced, and the matching convergence time under different initial conditions is relatively uniform and consistent.

Description

Technical Field

[0001] The invention relates to the technical field of radio frequency power supply impedance matchers, in particular to an automatic impedance matching method for an L-type radio frequency matcher with uniform time convergence. Background Art

[0002] The RF power supply impedance matcher is an important component of the RF plasma chamber system. The RF power supply impedance matcher automatically adjusts the matching variable capacitance element through an algorithm to achieve the effect that the overall equivalent impedance of the RF power supply back stage is a pure resistive load. This achieves maximum power transmission and protects the RF power supply from the adverse effects of reflected power.

[0003] In practical applications, the RF power supply impedance matcher: 1) can be configured into different matching network topologies, such as L-type matching circuit, T-type matching circuit, Π-type matching circuit, etc. 2) The load impedance of the connected plasma chamber is unknown. 3) The initial capacitor position conditions may be different for each matching, etc. The above situations will cause the system to have complex and different standing wave ratio-capacitance value function relationships. The convergence time of the traditional gradient-based extreme value search impedance matching algorithm depends on the Hessian matrix information in the standing wave ratio-capacitance value function relationship, which makes it impossible to achieve consistent time convergence under different working conditions through the same set of preset control parameters. This poses a challenge to achieving high-speed, high-precision, stable and reliable impedance matching. Therefore, it is necessary to study an automatic impedance matching algorithm with stronger robustness and better speed accuracy.

[0004] The present application is different from the prior art as follows:

[0005] Technical comparison with patent CN114465593A "Impedance matching method and dual-frequency matching device"

[0006] Patent CN114465593A proposes an impedance matching method and a dual-frequency matcher for a dual-frequency matcher, the method comprising: obtaining a first standing wave ratio of a low-frequency input end and a second standing wave ratio of a high-frequency input end; if the first standing wave ratio does not satisfy the first condition, and / or the second standing wave ratio does not satisfy the second condition, obtaining a first impedance of the low-frequency input end and a second impedance of the high-frequency input end; obtaining the load impedance of the dual-frequency matcher; calculating first data for adjusting the capacitance in the dual-frequency matcher according to the first impedance, the second impedance and the load impedance; adjusting the capacitance in the low-frequency impedance matching network in the low-frequency part according to the first data to achieve impedance matching of the low-frequency part; adjusting the capacitance in the high-frequency impedance matching network in the high-frequency part according to the first data to achieve impedance matching of the high-frequency part. The invention patent focuses on realizing dual-frequency matching control of low frequency and high frequency. The invention patent focuses on the rapidity of the matching process realized by the proposed algorithm and the relative consistency of the convergence time under different initial conditions. There are essential differences between the two in terms of objects and contents.

[0007] Technical comparison with patent CN114914146A "RF matching device and its control method, semiconductor process equipment"

[0008] Patent CN114914146A discloses a radio frequency matcher and a control method thereof, wherein the radio frequency matcher comprises: a power detection unit, an impedance adjustment unit, an analog control unit, an analog interface, a micro control unit, a touch screen display unit and a transmission unit; the power detection unit is connected to the analog control unit; the analog control unit is connected to the impedance adjustment unit; the impedance adjustment unit comprises an adjustable capacitor; the analog interface is connected to the analog control unit; the transmission unit is respectively connected to the micro control unit and the analog interface; the micro control unit is connected to the touch screen display unit, so that the status information of the radio frequency matcher can be intuitively displayed on site of the semiconductor process equipment and a touch screen operation means can be provided, thereby improving the flexibility of the control operation for the radio frequency matcher. The invention patent focuses on explaining a structural design scheme, and the adjustment of the capacitor is based on an analog control method. The present invention focuses on explaining a digital, fast algorithm that can have a relatively consistent convergence time under different initial conditions. There is an essential difference between the two in terms of content. Summary of the invention

[0009] In order to solve the above technical problems, the present invention proposes an L-type RF matching device automatic impedance matching method with uniform time convergence, which can achieve more consistent matching convergence time under different working conditions, and the matching speed and matching accuracy can be arbitrarily configured.

[0010] To achieve the above object, the technical solution adopted by the present invention is:

[0011] An automatic impedance matching method for an L-type radio frequency matching device with uniform time convergence is specifically as follows:

[0012] Step 1, initialize the program variables;

[0013] Step 2: Determine whether the standing wave ratio is greater than the feedforward calculation threshold standing wave ratio σ op , if yes, go to step 3, otherwise, go to step 4;

[0014] Step 3: Perform open-loop feedforward calculation to obtain the desired feedforward capacitance value for matching. Use the desired feedforward capacitance value to drive the variable capacitor and return to step 2;

[0015] Step 4, perform multivariable Newton extreme value search control calculation, use the capacitance value obtained by the multivariable Newton extreme value search to drive the variable capacitor, and adjust the capacitance value to approach the matching point. Execute step 5;

[0016] Step 5: Determine whether the standing wave ratio is less than the end threshold standing wave ratio σstop If yes, stop the calculation, otherwise, go to step 4.

[0017] As a further improvement of the present invention, step 1 includes: setting nominal parameters of the impedance matching network, setting the amplitude of the sinusoidal signal, controller gain parameters, and filter parameters.

[0018] As a further improvement of the present invention, the open-loop feedforward calculation method described in step 3 has the following process:

[0019] 1) Determine the system angular frequency ω and the nominal value of the matching network inductance L1;

[0020] 2) Calculate the plasma chamber load impedance under the current control cycle and Calculate by solving the following equation;

[0021]

[0022] Among them, Re[] represents the real part of the complex number, Im[] represents the imaginary part of the complex number, and represents the capacitance values ​​of variable capacitor 1 and variable capacitor 2 under a certain control cycle, and They represent the overall resistance and reactance of the matcher and plasma chamber load as seen from the input side of the matcher, and need to be measured by sensors. and represent the calculated resistance and reactance of the plasma chamber load respectively.

[0023] 3) Calculate the expected matching capacitance value for the next control cycle And drive the variable capacitor, calculated by solving the following equation:

[0024]

[0025] in, and They represent the feedforward capacitance values ​​that need to be set for capacitor No. 1 and capacitor No. 2 in the next control cycle respectively;

[0026] 4) Determine whether the standing wave ratio is greater than the feedforward calculation standing wave ratio threshold σ op ; If so, return to 2), otherwise, end this round of feedforward calculation.

[0027] As a further improvement of the present invention, the multivariable Newton extreme value search control calculation is performed in step 4, specifically in the following manner:

[0028] 1) The system standing wave ratio signal passes through a high-pass filter to obtain two signals A.1 and B.1. The high-pass filter is in the form of a first-order, such as:

[0029]

[0030] Among them, ω h Represents the cutoff frequency of the high-pass filter;

[0031] 2) The A.1 signal is multiplied by the M(t) signal and then passed through a low-pass filter to obtain the A.2 signal. The low-pass filter is in the form of a first-order filter, such as:

[0032]

[0033] Among them, ω l Represents the cutoff frequency of the low-pass filter;

[0034] 3) The B.1 signal is multiplied by the N(t) signal and passed through a low-pass filter to obtain the B.2 signal; the B.2 signal passes through the Hessian inverse matrix estimator to obtain the B.3 signal;

[0035] N(t) is an n-dimensional matrix, and the matrix elements are written as:

[0036]

[0037] Among them, i,j∈{1,2,...,n}.a i ,a j ∈R,ω i / ω j is a rational number, and ω i ≠ω j .

[0038] The form of the Hessian inverse matrix estimator is as follows:

[0039]

[0040] Where ρ is a positive gain, Γ is the inverse matrix of the Hessian matrix, is an estimate of the Hessian matrix.

[0041] 4) After multiplying the A.2 signal and the B.3 signal, take the negation and pass it through the integrator to get the C.1 signal; add the C.1 signal and the S(t) signal to get the C.2 signal.

[0042] The integrator expression is:

[0043]

[0044] Where diag[] represents a diagonal matrix, K gnis the integral gain.

[0045] The S(t) signal is written as:

[0046] S(t)=[a1 sin(ω1t a n sω n t)] T

[0047] 5) Signal C.2 is used as the position reference signal of the variable capacitor of the matching network to drive the variable capacitor to operate, generate the calculated standing wave ratio of the next cycle, and repeat step 1).

[0048] The benefits of this application are:

[0049] The present invention realizes the automatic impedance matching process of the L-type RF matcher through the above algorithm flow, and has good matching speed and matching accuracy. Compared with the prior art, this solution combines open-loop feedforward calculation with multivariable Newton extreme value search control calculation, taking into account the rapidity, accuracy, uniformity and adjustability of the matching process, and using a set of preset control parameters, it can achieve rapid convergence of the matching process under uniform time. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the implementation process of an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of a program framework of an embodiment of the present invention;

[0052] Figure 3 A circuit topology of a matching system according to an embodiment of the present invention;

[0053] Figure 4 The multivariable Newton extreme value search control algorithm described in the embodiment of the present invention;

[0054] Figure 5 The conventional extreme value search control matching effect of the embodiment of the present invention;

[0055] Figure 6 The uniform time convergence effect of the inventive algorithm described in the embodiment of the present invention;

[0056] Figure 7 This is the overall effect of the inventive algorithm described in the embodiment of the present invention. DETAILED DESCRIPTION

[0057] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments:

[0058] like Figure 1As shown, the embodiment of the present invention applies a uniform time convergence L-type radio frequency matching device automatic impedance matching method, including the following steps. The L-type impedance matching network is as follows Figure 2 shown.

[0059] Step 1: Initialize program variables.

[0060] Step 2: Determine whether the standing wave ratio is greater than the feedforward calculation threshold standing wave ratio σ op , if yes, go to step 3. Otherwise, go to step 4.

[0061] Step 3: Perform open-loop feedforward calculation to obtain the desired feedforward capacitance value for matching. Drive the variable capacitor with the desired feedforward capacitance value and return to step 2.

[0062] Step 4: perform multivariable Newton extreme value search control calculation, use the capacitance value obtained by the multivariable Newton extreme value search to drive the variable capacitor, and adjust the capacitance value to be close to the matching point. Execute step 5.

[0063] Step 5: Determine whether the standing wave ratio is less than the end threshold standing wave ratio σ stop If yes, stop the round of calculation. Otherwise, go to step 4.

[0064] The programmed execution flow chart is as follows Figure 3 shown.

[0065] Specifically, in this embodiment, the open-loop feedforward calculation method described in step 3 has the following process:

[0066] 1) The system angular frequency is set to ω = 2πf = 2π × 13.56 × 10 6 rad / s, the nominal value of the matching network inductor is L1=0.7μH.

[0067] 2) Set the control cycle to 10ms and the sampling measurement cycle to 0.5ms. Calculate the plasma chamber load impedance under the current control cycle and Calculate this by solving the following equation:

[0068]

[0069] Among them, Re[] represents the real part of the complex number, and Im[] represents the imaginary part of the complex number. and It represents the capacitance values ​​of variable capacitor 1 and variable capacitor 2 in the current control cycle, and the variable capacitors can be calibrated in advance.

[0070] and They respectively represent the total resistance and total reactance of the matcher and plasma chamber load viewed from the input side of the matcher, which need to be measured by sensors. and represent the calculated resistance and reactance of the plasma chamber load respectively.

[0071] 3) Calculate the expected matching capacitance value for the next control cycle And drive the variable capacitor. Calculate by solving the following equation:

[0072]

[0073] in, and They respectively represent the feedforward capacitance values ​​that need to be set for capacitor No. 1 and capacitor No. 2 in the next control cycle.

[0074] 4) Set the feedforward calculation standing wave ratio threshold σ op =1.60. Determine whether the standing wave ratio is greater than the feedforward calculation standing wave ratio threshold σ op ; If yes, return to step 2). Otherwise, end this round of open-loop feedforward calculation.

[0075] In this embodiment, the multivariable Newton extreme value search control algorithm is executed in step 4, such as Figure 4 As shown, the specific implementation is:

[0076] 1) The system standing wave ratio signal φ passes through a high-pass filter to obtain two signals A.1 and B.1. The high-pass filter is designed as:

[0077]

[0078] 2) The A.1 signal is multiplied by the M(t) signal and passed through a low-pass filter to obtain the A.2 signal. The low-pass filter is designed as:

[0079]

[0080] 3) The B.1 signal is multiplied by the N(t) signal and passed through a low-pass filter (which is the same as the low-pass filter described in the previous step) to obtain the B.2 signal; the B.2 signal passes through the Hessian inverse matrix estimator to obtain the B.3 signal.

[0081] The N(t) signal is designed as:

[0082]

[0083] The form of the Hessian inverse matrix estimator is as follows:

[0084]

[0085] Among them, the design ρ=23.

[0086] 4) After multiplying the A.2 signal and the B.3 signal, take the negation and pass it through the integrator to get the C.1 signal; add the C.1 signal and the S(t) signal to get the C.2 signal.

[0087] The integrator expression is:

[0088]

[0089] The S(t) signal is designed as:

[0090] S(t)=[4sin(16πt n 2πt)] T

[0091] 5) The C.2 signal is used as the position command signal of the variable capacitor of the matching network to drive the variable capacitor to operate and generate the calculated standing wave ratio of the next cycle. Repeat step 1).

[0092] In order to better illustrate the practical effect of the present invention, a group of impedance matching systems using traditional sinusoidal extreme value search control is set as a control group. The parameter configuration of the traditional sinusoidal extreme value search scheme is as follows:

[0093] The cutoff frequency of the first-order high-pass filter is set to 6.28 rad / s, the cutoff frequency of the two first-order low-pass filters is set to 10 rad / s, and the extremum search gains are set to K1 = -625 and K2 = -2130. The amplitudes of the sinusoidal excitation signals are set to a1 = 4 and a2 = 10, and the angular frequencies of the sinusoidal signals are set to ω1 = 10πrad / s and ω2 = 12πrad / s.

[0094] In this embodiment, the capacitance starting points of the two test experiments are set to be [160pF, 240pF] and [240pF, 240pF] respectively. The capacitance value and the standing wave ratio graph of the traditional algorithm at two different starting points are shown as follows: Figure 5 The uniform time convergence effect of the algorithm of the present invention under two different starting points is shown in Figure 6 By comparison Figure 5 and Figure 6 The convergence time of the method of the present invention can be observed under the test conditions of two different starting points. Taking another capacitance starting point [350pF, 350pF], the overall effect of the method of the present invention is as follows: Figure 7 shown.

[0095] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. An automatic impedance matching method for an L-type radio frequency matching device with uniform time convergence, characterized in that: Specifically: Step 1, initialize the program variables; Step 2: Determine whether the standing wave ratio is greater than the feedforward calculation threshold standing wave ratio σ op , if yes, go to step 3, otherwise, go to step 4; Step 3: Perform open-loop feedforward calculation to obtain the desired feedforward capacitance value for matching. Use the desired feedforward capacitance value to drive the variable capacitor and return to step 2; Step 4, perform multivariable Newton extreme value search control calculation, use the capacitance value obtained by the multivariable Newton extreme value search to drive the variable capacitor, and adjust the capacitance value to approach the matching point. Execute step 5; Step 5: Determine whether the standing wave ratio is less than the end threshold standing wave ratio σ stop If yes, stop the calculation, otherwise, go to step 4.

2. The method for automatic impedance matching of an L-type radio frequency matching device with uniform time convergence according to claim 1, characterized in that: The step 1 includes: setting nominal parameters of the impedance matching network, setting the amplitude of the sinusoidal signal, controller gain parameters, and filter parameters.

3. The method for automatic impedance matching of an L-type radio frequency matching device with uniform time convergence according to claim 1, characterized in that: The open-loop feedforward calculation method described in step 3 has the following process: 1) Determine the system angular frequency ω and the nominal value of the matching network inductance L1; 2) Calculate the plasma chamber load impedance under the current control cycle and Calculate by solving the following equation; Among them, Re[] represents the real part of the complex number, Im[] represents the imaginary part of the complex number, and Indicates the capacitance values ​​of variable capacitor 1 and variable capacitor 2 in the current control cycle, and They represent the total resistance and total reactance of the matcher and plasma chamber load as seen from the input side of the matcher, which need to be measured by sensors. and represent the calculated resistance and reactance of the plasma chamber load respectively. 3) Calculate the expected matching capacitance value for the next control cycle And drive the variable capacitor, calculated by solving the following equation: in, and They represent the feedforward capacitance values ​​that need to be set for capacitor No. 1 and capacitor No. 2 in the next control cycle respectively; 4) Determine whether the standing wave ratio is greater than the feedforward calculation standing wave ratio threshold σ op ; If so, return to 2), otherwise, end this round of feedforward calculation.

4. The method for automatic impedance matching of an L-type radio frequency matching device with uniform time convergence according to claim 1, characterized in that: The multivariable Newton extreme value search control calculation is performed in step 4, specifically in the following manner: 1) The system standing wave ratio signal passes through a high-pass filter to obtain two signals A.1 and B.

1. The high-pass filter is in the form of a first-order, such as: Among them, ω h Represents the cutoff frequency of the high-pass filter; 2) The A.1 signal is multiplied by the M(t) signal and passed through a low-pass filter to obtain the A.2 signal. The low-pass filter is in the form of a first-order filter, such as: Among them, ω l Represents the cutoff frequency of the low-pass filter; 3) The B.1 signal is multiplied by the N(t) signal and passed through a low-pass filter to obtain the B.2 signal; the B.2 signal is passed through the Hessian inverse matrix estimator to obtain the B.3 signal; N(t) is an n-dimensional matrix, and the matrix elements are written as: Among them, i,j∈{1,2,...,n}.a i ,a j ∈R,ω i / ω j is a rational number, and ω i ≠ω j . The form of the Hessian inverse matrix estimator is as follows: Where ρ is a positive gain, Γ is the inverse matrix of the Hessian matrix, is an estimate of the Hessian matrix. 4) After multiplying the A.2 signal and the B.3 signal, take the negation and pass it through the integrator to get the C.1 signal; add the C.1 signal and the S(t) signal to get the C.2 signal. The integrator expression is: Where diag[] represents a diagonal matrix, K gn is the integral gain. The S(t) signal is written as: S(t)=[a1 sin(ω1t a n s ω n t)] T 5) Signal C.2 is used as the position reference signal of the variable capacitor of the matching network to drive the variable capacitor to operate, generate the calculated standing wave ratio of the next cycle, and repeat step 1).

Images