Control method and plasma processing device

Abstract

The present invention provides a control method and a plasma processing device. A local plasma using a phased array antenna is controlled. As a control method for a scanning type plasma processing device using a phased array antenna, the method comprises the following steps: observing the light emission of plasma generated inside the processing container from observation windows provided at multiple locations of the processing container; calculating the in-plane distribution of values ​​representing the characteristics of the plasma on a substrate based on the observed data related to the light emission of the plasma; and correcting the scanning pattern of the plasma and / or the plasma density distribution based on the calculated in-plane distribution of values ​​representing the characteristics of the plasma on the substrate.

Description

Technical Field

[0001] The invention relates to a control method and a plasma processing device. Background Art

[0002] A plasma processing device is known that performs plasma processing on a wafer by converting a gas into plasma using the power of electromagnetic waves. For example, Patent Document 1 proposes a method for modifying the reaction rate on a semiconductor substrate in a processing container using a phased array microwave antenna, wherein plasma is excited in the processing container, and a microwave radiation beam is radiated from the phased array microwave antenna, and the beam is directed to the plasma to change the reaction rate on the surface of the semiconductor substrate in the processing container.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2017-103454 Summary of the invention

[0006] Problem that the invention aims to solve

[0007] The present disclosure provides a control method for controlling local plasma using a phased array antenna and a plasma processing apparatus.

[0008] Solutions for solving problems

[0009] According to one embodiment of the present disclosure, a control method is provided for controlling a scanning plasma processing device using a phased array antenna, the control method comprising the following steps: observing the light emission of plasma generated inside the processing container from observation windows provided at multiple locations of the processing container; calculating the in-plane distribution of values ​​representing characteristics of the plasma on a substrate based on observed data related to the light emission of the plasma; and correcting the scanning pattern of the plasma and / or the plasma density distribution based on the calculated in-plane distribution of values ​​representing the characteristics of the plasma on the substrate.

[0010] Effects of the Invention

[0011] According to one aspect, local plasma can be controlled using a phased array antenna. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic cross-sectional view showing an example of a plasma processing apparatus according to an embodiment.

[0013] Figure 2 This is a diagram showing an example of a monopole antenna according to an embodiment.

[0014] Figure 3 This is a diagram showing an example of power phase control performed by a control unit according to one embodiment.

[0015] Figure 4 This is a diagram showing an example of a light emission monitor mechanism according to an embodiment.

[0016] Figure 5 It is a diagram for explaining the emission spectrum measurement involved in one embodiment.

[0017] Figure 6 This is a diagram showing an example of measured values ​​and ideal values ​​of plasma density distribution according to one embodiment.

[0018] Figure 7 It is a diagram schematically showing a series of operations of parameter search.

[0019] Figure 8 It is a diagram for explaining the outline of parameter search.

[0020] Fig. 9 This is a flowchart showing an example of a control method (parameter search) according to one embodiment.

[0021] Fig.10 This is a flowchart showing an example of a control method (substrate processing) according to an embodiment.

[0022] Fig.11 It is a diagram for explaining the effect of the control method involved in one embodiment.

[0023] Fig.12 This is a diagram for explaining a method of learning process prediction data according to one embodiment.

[0024] Description of Reference Numerals

[0025] 1: processing container; 2: phased array antenna; 2a~2g: monopole antenna; 3: mounting table; 5: dielectric window; 6: microwave output unit; 7: power phase control unit; 8: control unit; 9: top plate; 10: plasma processing device; 11: antenna unit; 12: observation window; 13: reflector; 14: motor; 15: spectrometer; 17: beam former; 20: light emission monitor mechanism; 81: storage unit; 82: measurement unit; 83: analysis unit; 84: search unit; 85: synchronization control unit. DETAILED DESCRIPTION

[0026] Hereinafter, the mode for implementing the present disclosure will be described with reference to the drawings. In each of the drawings, the same reference numerals are sometimes given to the same components to omit repeated descriptions.

[0027] [Plasma processing device]

[0028] use Figure 1 A plasma processing apparatus 10 according to one embodiment will be described. Figure 1 1 is a schematic cross-sectional view showing an example of a plasma processing apparatus 10 according to an embodiment. The plasma processing apparatus 10 is described by taking a microwave plasma processing apparatus as an example.

[0029] A plasma processing apparatus 10 according to one embodiment includes a processing container 1 that accommodates a substrate W such as a wafer. The interior of the processing container 1 is a processing chamber 1s for performing a predetermined plasma process such as a film forming process or an etching process on the substrate W by surface wave plasma formed by microwaves.

[0030] The processing container 1 is substantially cylindrical and grounded. The processing container 1 has an upper opening blocked by a top plate 9, thereby maintaining airtightness inside. The processing container 1 and the top plate 9 are formed of a metal material such as aluminum or stainless steel.

[0031] The mounting table 3 on which the substrate W is mounted is supported by a cylindrical support member 4 extending from the center of the bottom of the processing container 1. The mounting table 3 may also be provided with an electrostatic chuck for electrostatically adsorbing the substrate W, a temperature control mechanism, a gas flow path for supplying a heat transfer gas to the back side of the substrate W, and the like. In addition, a high-frequency bias power supply may be electrically connected to the mounting table 3 via a matching device. However, the high-frequency bias power supply may not be provided depending on the characteristics of the plasma processing.

[0032] An exhaust pipe is connected to the bottom of the processing container 1, and an exhaust device including a vacuum pump is connected to the exhaust pipe. When the exhaust device is operated, the processing container 1 is exhausted, thereby reducing the pressure in the processing container 1 to a predetermined vacuum level. A loading and unloading port for loading and unloading substrates W and a gate valve for opening and closing the loading and unloading port are provided on the side wall of the processing container 1.

[0033] A plurality of gas pipes 11 a are formed in the processing container 1 , and the gas supplied from the gas supply unit 21 is supplied to the processing chamber 1 s through the plurality of gas pipes 11 a .

[0034] A phased array antenna 2 is provided on the top plate 9. The phased array antenna 2 is composed of seven monopole antennas 2a to 2g (in Figure 1Only monopole antennas 2a to 2c are shown in the figure. Phased array antenna 2 is formed by a plurality of monopole antennas and is used to irradiate electromagnetic waves into processing container 1. In addition, the number of monopole antennas is not limited to 7, as long as it is 2 or more, preferably 3 or more.

[0035] Monopole antennas 2a to 2g have the same structure. Figure 2 , an example of a monopole antenna 2a is representatively shown, and the illustration and description of the structures of other monopole antennas 2b to 2g are omitted. The monopole antenna 2a is in the shape of a coaxial cable, and has an internal conductor 121, an external conductor 122 outside the internal conductor 121, and a dielectric 123 such as Teflon (registered trademark) provided therebetween. The top end of the monopole antenna 2a constitutes an antenna portion 11, and the antenna portion 11 includes the internal conductor 121 protruding by a length D. The length D of the antenna portion 11 is, for example, several tens of mm to several hundreds of mm.

[0036] The antenna part 11 is exposed from the surface at the same height as the back surface 9a of the top plate 9 of the processing container 1, that is, the end of the dielectric 123, to the internal space of the processing container 1, so that microwaves are radiated from the radiation part 125 into the processing container 1. However, the internal conductor 121 may also be a structure that does not protrude from the dielectric 123.

[0037] like Figure 1 As shown, through this structure, microwaves are output from the microwave output unit 6 , and are radiated into the processing container 1 after the power and / or phase are controlled by the power phase control unit 7 according to the control of the control unit 8 .

[0038] The monopole antennas 2a to 2g are arranged at approximately equal intervals. The distance between the centers of adjacent monopole antennas 2a to 2g is configured to be less than λ / 2 relative to the wavelength λ of the microwave. The dielectric window 5 is configured between the plurality of antenna units 11 and the mounting table 3 in a manner to isolate the two. The dielectric window 5 is a partition that divides the interior of the processing container 1 into a space V above the dielectric window 5 and a space U below the dielectric window 5. The dielectric window 5 is formed of, for example, ceramics such as quartz, alumina (Al2O3), fluorine-based resins such as polytetrafluoroethylene, and polyimide-based resins.

[0039] There is a free space of about tens to hundreds of mm in the space V below the monopole antennas 2a to 2g, and plasma is generated in the space U through the dielectric window 5 provided below the free space. The distance H between the top of the plurality of antenna parts 11 and the upper surface of the dielectric window 5 is greater than λ / 4 with respect to the wavelength λ of the microwave. The plurality of microwaves radiated from the monopole antennas 2a to 2g propagate in the space V. The space V is an atmospheric space, and the space U is a vacuum space.

[0040] The plasma processing device 10 has a control unit 8. The control unit 8 is a computer having a storage unit such as a processor, a memory, an input unit, a display unit, a signal input and output interface unit, etc., and controls each unit of the plasma processing device 10. The storage unit stores a control program and process data. The computer (CPU) executes the control program and controls each unit of the plasma processing device 10 according to the process data. In addition, the computer controls the power phase control unit 7 provided for each antenna unit 11 of the phased array antenna 2, thereby controlling the power and / or phase of the microwave radiated from the antenna unit 11.

[0041] When plasma processing is performed in the plasma processing apparatus 10 having such a structure, first, the substrate W is carried into the processing container 1 through the opening of the gate valve and the carry-in / out port while being held by the transfer arm.

[0042] When the substrate W is transported to the top of the mounting table 3, it is moved from the transport arm to the pusher pin, and is placed on the mounting table 3 by the descent of the pusher pin. After the substrate W is transported in, the gate valve is closed. The pressure of the processing chamber 1s is maintained at a specified vacuum degree by an exhaust device. A specified gas is introduced into the space U below the dielectric window 5. Microwaves with controlled power and / or phase are radiated from the phased array antenna 2. As a result, the electric field is enhanced at a specified position of the dielectric window 5, and the gas in the space U is plasmatized to generate a local plasma. The spatial distribution of the time integral value of the plasma density can be controlled by scanning the generated local plasma, and a uniform or desired plasma treatment can be performed on the substrate W.

[0043] In the plasma processing apparatus 10 according to one embodiment, the power and phase of the microwaves radiated from each of the monopole antennas 2a to 2g are controlled by the power phase control unit 7 under the control of the control unit 8. Thus, the microwaves introduced from each of the monopole antennas 2a to 2g can be interfered to increase the electric field intensity at any position. Thus, by generating plasma intensively, a high degree of control of the plasma density distribution can be performed.

[0044] Since the synthesis of microwaves by phase control described above is not accompanied by mechanical action, it can be controlled at a very high speed. In principle, the focus position C of the synthesis of microwaves can be moved in accordance with time by high-speed control at the same level as the frequency of the microwaves. Thus, the power phase control of the plurality of monopole antennas 2a to 2g can be controlled at a high speed. As a result, the plasma density distribution can be controlled uniformly or freely.

[0045] Figure 3 FIG. 8 is a diagram showing an example of power phase control performed by the control unit 8 according to one embodiment. Figure 3In the example, the control unit 8 uses the power phase control unit 7 to control the phase of the microwaves radiated from the monopole antennas 2a to 2g. These mutually reinforce each other at the focal position C. In order to control the electric field intensity of the microwaves in the combined portion Ar centered at the focal position C, the power of the microwaves radiated from the monopole antennas 2a to 2g is controlled.

[0046] The control unit 8 uses the power phase control unit 7 to control the phase of the microwave at high speed. The surface of the dielectric window 5 is scanned in the radial direction L1 or the circumferential direction L2, etc., so that the focal position C is moved. In addition, the power of the microwave is controlled during the scanning. Thus, the density distribution of the plasma P generated by the power of the microwave transmitted to the bottom of the dielectric window 5 can be freely controlled by controlling the power of the microwave while moving the focal position C and the synthesis portion Ar at a high speed.

[0047] In addition, the control unit 8 can change the phase of the microwave by using the power phase control unit 7. The control unit 8 changes the moving speed of the synthesis part Ar by controlling the speed of the synthesis part Ar. The scanning speed is changed by controlling the synthesis part Ar to move slowly on the outer peripheral side of the dielectric window 5 and to move faster on the inner peripheral side than on the outer peripheral side. In addition, the power of the microwave is controlled during the scanning. Thus, the plasma density distribution can be freely controlled, such as controlling the density of the plasma on the outer peripheral side below the dielectric window 5 to be higher than the density of the plasma on the inner peripheral side.

[0048] [Light-emitting monitor mechanism and control method]

[0049] Next, a method for controlling the power and / or phase of the scanning type plasma processing device 10 using the phased array antenna 2 of the structure is described. It is important to control the power and / or phase output from the phased array antenna 2 in order to control the plasma density distribution to the desired distribution, and to appropriately control the scanning pattern of the local plasma and / or the plasma density distribution. At this time, in the method of repeatedly processing various substrates to search for conditions through repeated trials and then controlling the power and phase according to the search conditions obtained, it is inefficient because a large amount of resources such as substrates, labor, and time are consumed during the condition search. In addition, the same problem also exists in the method of detecting the asymmetry and individual differences (hereinafter also referred to as "mechanical errors") of the plasma processing device 10 and controlling the power and phase according to the detected conditions, and it is inefficient.

[0050] Therefore, the plasma processing apparatus 10 according to one embodiment includes a light emission monitoring mechanism 20, which monitors the light emission of the plasma generated in the processing container 1 from the observation windows 12 provided at multiple locations of the processing container 1. Then, based on the light emission intensity data of the plasma obtained from the light emission monitoring mechanism 20, the power and / or phase output from the phased array antenna 2 are appropriately controlled. Figure 4 and Figure 5 An example of the structure of the light emission monitor mechanism 20 will be described. Figure 4 It is a diagram showing an example of the light emission monitor mechanism 20 according to one embodiment. Figure 5 It is a diagram for explaining the emission spectrum measurement using the emission monitor mechanism 20 according to one embodiment.

[0051] like Figure 4 As shown, the light emission monitor mechanism 20 includes a reflector 13, a motor 14, a spectrometer 15, an optical fiber 16, and a beam former 17. The beam former 17 includes a pinhole pair, a collimating lens, etc., and shapes the light (beam) for observation. The formed beam is collected into the spectrometer 15.

[0052] The reflector 13 is driven by the motor 14 to rotate in the same direction. The spectrometer 15 is connected to the beam former 17 via the optical fiber 16. The light emitted by the plasma observed from the observation window 12 is reflected by the reflector 13, and only the light in the direction defined by the beam former 17 is incident on the spectrometer 15.

[0053] The spectrometer 15 selectively uses the light in the processing container 1 for optical emission spectrometry (OES). Optical emission spectrometry is one of the methods for monitoring the state of plasma. The spectrometer 15 uses optical emission spectrometry to qualitatively characterize the wavelength of the element-specific bright line spectrum (atomic spectrum) obtained by plasma excitation, and quantitatively characterizes it based on the emission intensity.

[0054] like Figure 5 As shown, the light-emitting monitor mechanism 20 is arranged one-to-one with respect to the observation window 12. Figure 5 In the example of FIG. 1 , three observation windows 12 made of quartz are provided on the side wall of the processing container 1. Therefore, three light emission monitoring mechanisms 20 corresponding to the three observation windows 12 are provided. It is sufficient to provide a plurality of observation windows 12 and light emission monitoring mechanisms 20. Provide the same number of light emission monitoring mechanisms 20 as the number of observation windows 12.

[0055] The reflectors 13 of the three emission monitor mechanisms 20 rotate synchronously. By combining these reflectors 13, emission spectrum measurement (angle resolution measurement) can be performed in various directions of the processing container 1. That is, the spectrometer 15 uses the light observed from the three observation windows 12 to measure the emission intensity at each angle. The measured data of the emission intensity (emission intensity data) is sent to the control unit 8. The emission intensity data is an example of data related to the emission of plasma.

[0056] The control unit 8 controls the plasma P (see Figure 3 ) and corresponding to a specific height in the processing container 1, the luminous intensity distribution at each position represented by the two-dimensional coordinates (x, y) is reconstructed. Light is observed from three observation windows 12, an inverse transformation is performed based on the principle of CT scanning, and the result is back-projected, thereby inferring the original luminous intensity distribution. The back-projected image is considered to approximately correspond to the density distribution of various excited particles or the plasma density distribution. Therefore, the back-projected image can be used as information for inferring the distribution of process results. For example, the plasma density distribution on the substrate W can be inferred based on the reconstructed back-projected image. Figure 5 The image R1 shown may be shown as an image of the emission intensity distribution after backprojection, or may be shown as a plasma density distribution estimated based on the emission intensity distribution.

[0057] That is, the control unit 8 calculates the plasma density distribution in the surface of the substrate W based on the luminous intensity data of the plasma observed by the luminous monitor mechanism 20, and corrects the plasma scanning pattern based on the calculated plasma density distribution. In addition, the plasma density distribution is an example of the distribution of the value representing the characteristics of the plasma in the surface of the substrate W. As an example of the value representing the characteristics of the plasma, it can also be any one of the density of the plasma, the value representing the processing result of the substrate W being processed by the plasma, the processing rate of the substrate being processed by the plasma, and the luminous intensity of the plasma. As an example of the processing rate of the substrate, the film forming rate, the etching rate, the ashing rate, the cleaning rate, etc. can be listed. As an example of the value representing the processing result of the substrate W, the thickness of the film formed on the substrate W, the resistance value of the film, the CD value when the substrate W is etched, the etching shape, the etching depth, etc. can be listed.

[0058] [Control of power and / or phase]

[0059] In the generation of local plasma using the phased array antenna 2, Figure 3The situation where the synthetic part Ar is enlarged, for example, the situation where the plasma is deformed when a circular plasma is to be generated. In contrast, in a control method involved in an embodiment described below, the scanning pattern and / or plasma density distribution of the local plasma is controlled (corrected) by controlling the power and / or phase output from each antenna unit 11 of the phased array antenna 2. Thus, a method for controlling the plasma density distribution in the space U to perform a desired plasma treatment on the substrate W is proposed. That is, by controlling the power and / or phase of each antenna unit 11 of the phased array antenna 2, the scanning path, scanning speed, and plasma density distribution of the local plasma are adjusted to correct the spatial distribution of the time-integrated value of the plasma density.

[0060] For example, when the density ρ(x, y, t) of the local plasma is deformed and far from the ideal (for example, it is intended to be a circle but it is not a circle, etc.), even if the scanning pattern of the local plasma is adjusted, the time integral value ∫ρ(x, y, t)dt is likely not to become uniform or to become the desired shape. In this case, the density ρ(x, y, t) of the local plasma is corrected by making the standard deviation σ represented by the equation (1) described later as small as possible. At this time, the Markov chain Monte Carlo method is used to minimize the standard deviation σ.

[0061] In order to perform the control method involved in one embodiment, the density distribution of the plasma generated by the phased array antenna 2 is measured by the method of luminescence spectroscopy, and the measured luminescence intensity data is used to perform parameter search using the Markov chain Monte Carlo method. The parameter search is a technique that uses the power and / or phase of each antenna unit 11 of the phased array antenna 2 as a parameter to obtain a desired plasma density distribution, and obtains the optimal value of the parameter by the Markov chain Monte Carlo method.

[0062] For example, Figure 6 This is a diagram showing an example of estimated values ​​and ideal values ​​of plasma density distribution according to one embodiment. Figure 6 The left side of is an example of an estimated value of plasma density distribution obtained by measuring the emission intensity of plasma and based on the measured emission intensity data. In the following, the estimated value of plasma density distribution is represented by ρ(x, y, t).

[0063] Figure 6 The right side of is an example of an ideal value of plasma density distribution. o (x, y, t) represents the ideal value of plasma density distribution. Plasma density distribution ρ o is the plasma density distribution that we actually want to generate.

[0064] Plasma density distribution ρ, ρ o are the two-dimensional plasma density distribution (x, y) in the space on the substrate W at a certain time t, and are approximately proportional to the plasma density distribution on the substrate W. In the parameter search described below, the power p and phase output from each antenna unit 11 of the phased array antenna 2 are The appropriate values ​​of the parameters are searched so that the plasma density distribution ρ is close to the plasma density distribution ρ o Thus, the processing of the substrate W can be brought into or close to an ideal state.

[0065] The parameter search using the Markov chain Monte Carlo method can also be performed during the development of the plasma processing device 10 or after maintenance. In addition, in order to cope with the difference in plasma density distribution caused by the mechanical error of the plasma processing device 10 and the change in plasma density distribution caused by the aging of the plasma processing device 10, the parameter search can also be performed for each plasma processing device 10 or after a predetermined time has passed. The power p and phase of each antenna unit 11 of the phased array antenna 2 are searched by the Markov chain Monte Carlo method. Thus, the plasma density distribution can be brought close to an ideal value for each plasma processing apparatus 10 in accordance with the adjustment of mechanical errors and changes over time without replacing parts used in the plasma processing apparatus 10 .

[0066] The parameter search using the Markov Chain Monte Carlo method is preferably performed by the control unit 8 during process development or when the process used by the plasma processing apparatus 10 is changed. Figure 4 As shown in FIG. 1 , the control unit 8 includes a storage unit 81, a measuring unit 82, an analyzing unit 83, a searching unit 84, and a synchronization control unit 85. The storage unit 81 stores the measured value of the luminous intensity and / or the estimated value of the plasma density distribution as measured data 181 together with the ideal data 182 of the plasma density distribution. In addition, the storage unit 81 stores data such as process data, a program for executing an algorithm of the Markov chain Monte Carlo method, and appropriate values ​​of parameters.

[0067] The synchronous control unit 85 synchronously controls the motor 14 for rotating the reflector 13, the spectrometer 15 for measuring the luminous intensity, and the power supply system for controlling the power distribution and phase to each antenna unit 11 of the phased array antenna 2. When the reflector 13 is continuously rotated in one direction by the motor 14 after the synchronous control, the power p and phase of each antenna unit 11 of the phased array antenna 2 are firstly Power is supplied by setting an appropriate value (initial value), and gas is supplied into the processing container 1 to generate plasma.

[0068] The spectrometer 15 measures the luminous intensity of the plasma when the reflector 13 is directed toward the observation window 12. The measuring unit 82 acquires the measured luminous intensity data. The analyzing unit 83 reconstructs the plasma density distribution ρ based on the acquired luminous intensity data (refer to Figure 6 The plasma density distribution ρ).

[0069] The search unit 84 executes a Markov chain Monte Carlo method program using the obtained plasma density distribution ρ to determine the parameters to be used next by each antenna unit 11 of the phased array antenna 2. Each antenna unit 11 of the phased array antenna 2 is powered based on the determined parameters.

[0070] Thus, plasma is generated with new parameters. The spectrometer 15 performs spectrum measurement again, and the measuring unit 82 obtains the measured luminous intensity data. The analyzing unit 83 reconstructs the new plasma density distribution ρ based on the obtained luminous intensity data. The searching unit 84 uses the obtained new plasma density distribution ρ to determine the parameters to be used next by each antenna unit 11 of the phased array antenna 2. The control unit 8 repeats a series of actions of parameter search (measurement → analysis → search) until the Markov chain is fully performed.

[0071] Figure 7 1 is a diagram schematically showing a series of actions of parameter search. When the rotation period of the motor 14 deviates from the action timing of each device of the spectrometer 15, the control unit 8, and the microwave output unit 6, the parameter search becomes invalid. Therefore, the state of the motor 14 is always monitored by the synchronization control unit 85 to avoid the timing of a series of actions of parameter search (measurement, analysis, search, and power supply) from being inconsistent. In the following, the series of actions of parameter search (measurement, analysis, search, and power supply) are also referred to as "parameter search actions".

[0072] The synchronous control unit 85 always monitors and controls whether the deviation between the rotation period of the motor 14 and the parameter search operation is not accumulated, so as to stabilize the rotation speed of the motor 14 so that the rotation period of the motor 14 is equal to the step of the parameter search operation. The time required for analysis and search may also vary slightly, so the rotation period of the motor 14 is set in a way that has a margin.

[0073] Until the next power supply starts, the power and phase parameters of each antenna unit 11 of the phased array antenna 2 remain unchanged. That is, until the next power supply starts, the state of the plasma remains basically unchanged. Therefore, the luminous intensity of the plasma can be measured during the period when the state of the plasma remains unchanged, that is, until the next power supply starts. Among them, the luminous intensity of the plasma can be measured only when the reflector 13 is facing the observation window 12.

[0074] Adjust power p and phase So that the estimated value ρ of the plasma density distribution is consistent with the ideal value ρ of the plasma density distribution o Close means: set the power p and phase The matrix A of the vector space is used to minimize the standard deviation σ represented by equation (1), or to make the standard deviation σ smaller than a predetermined threshold.

[0075] The standard deviation σ is expressed by equation (1).

[0076]

Number 1

[0077]

[0078] The matrix A is represented by the following.

[0079]

Number 2

[0080]

[0081] The matrix A is composed of the powers p1 to p2 of the microwaves output from the n antenna units 11. n and Phase The plasma density ρ at a position in the space U on the substrate W that is determined by the two-dimensional coordinates (x, y) is obtained.

[0082] In order to set the power p and phase The search unit 84 uses the Markov chain Monte Carlo method to calculate the power p and phase Parameter search. Figure 8 It is a diagram for explaining the outline of parameter search.

[0083] The search unit 84 calculates the power p and phase of each antenna unit 11 of the phased array antenna 2. Set appropriate initial parameters and start searching. Figure 8 The horizontal axis represents matrix A. Figure 8 The upper curve represents the estimated value ρ of the plasma density distribution obtained as a result of the parameter search and the ideal value ρ of the plasma density distribution. o The standard deviation of the deviations is σ.

[0084] The initial parameters are the power p and phase of each antenna unit 11 of the phased array antenna 2 set in the matrix A. The initial value of . Figure 8 In the example, “a1” represents the power p1 to p2 of the matrix A. n and Phase The standard deviation σ is set as the initial parameter.

[0085] exist Figure 7 In the first cycle (t1 to t2) shown in the figure, the power p and phase The first power supply is performed from each antenna unit 11 of the phased array antenna 2. In the next step of the parameter search, i.e. Figure 7 In the next cycle (t2-t3) shown in FIG. 1 , it is determined probabilistically whether to move to another position or stay at the current position. The closer the standard deviation σ is to 0, the higher the probability of staying at the current position. In addition, the stay mentioned here can also be set as not staying in time but only adding a value to the histogram representing the stay time on the computer. In this way, the processing time can be shortened. In addition, moving to another position means changing the power and / or phase of the parameter. Figure 8 In the example, a2 and a2' represent the standard deviation σ when the parameters of matrix A are changed. The direction when moving to another position is randomly determined. In other words, it is randomly determined whether the standard deviation σ changes from a1 to a2 or from a1 to a2'.

[0086] Specifically, the probability of moving to another position is determined by the change in the standard deviation σ, which is the objective function. The change in the objective function is determined by the power p and / or phase p supplied from each antenna unit 11 of the phased array antenna 2 in each cycle. is determined by the amount of change.

[0087] Figure 8 The histogram at the bottom of represents the sum of the dwell time. The histogram represents the power p and / or phase of each antenna unit 11 of the phased array antenna 2 supplied with power in each cycle. The position where the standard deviation σ stays when it changes is used to reflect the objective function.

[0088] The search unit 84 continues to repeat the parameter search operation for a predetermined number of times, such as 1000 times. The predetermined number of times is the number of times a histogram representing the sum of the time the standard deviation σ stays is obtained, and it can also be preset. However, the present invention is not limited to this, and the search unit 84 can also end the parameter search operation when the stay time becomes greater than the threshold. The standard deviation σ corresponding to the longest stay time is likely to be the smallest standard deviation σ that can be achieved, and the power p and phase represented by the matrix A corresponding to the standard deviation σ at this time are It is determined that the plasma density distribution ρ is closest to the plasma density distribution ρ o The power p and phase

[0089] [Control method of parameter search]

[0090] Next, refer to Fig. 9 An example of the control method (parameter search) SW1 executed by the plasma processing apparatus 10 will be described. Fig. 9 1 is a flowchart showing an example of the control method SW1 according to one embodiment.

[0091] When this process starts, the synchronization control unit 85 synchronously controls the motor 14 that rotates the reflector 13, the spectrometer 15 that measures the light emission intensity, and the power supply system that controls the power distribution and phase to each antenna unit 11 of the phased array antenna 2 (step S1).

[0092] Next, the measuring unit 82 uses the spectrometer 15 to measure the emission intensity of the plasma from the light incident from the multiple observation windows 12 at multiple angles (step S3). Next, the analyzing unit 83 analyzes the measured emission intensity data and reconstructs the distribution of the emission intensity to estimate the plasma density distribution (step S5).

[0093] Next, the search unit 84 uses the Markov chain Monte Carlo method to perform parameter search based on the analysis results (step S7). Next, the search unit 84 determines whether the parameter search action shown in steps S3 to S7 has been repeated a preset number of times (step S9). When the search unit 84 determines that the parameter search action shown in steps S3 to S7 has not been repeated a preset number of times, it returns to step S1 and performs the parameter search action of steps S3 to S7 again. When the search unit 84 determines in step S9 that S3 to S7 have been repeated a preset number of times, based on the histogram of residence time produced by repeating the parameter search action, the power p1 to p2 represented by the matrix A with the longest residence time is selected. n and Phase The next parameter is determined, and the density distribution of the local plasma is corrected thereby (step S11).

[0094] Thus, by determining the power p1~p n and Phase Each antenna unit 11 of the phased array antenna 2 is controlled during the next substrate processing.

[0095] As described above, a control method involved in one embodiment controls a scanning plasma processing device 10 using a phased array antenna 2, and the control method has the following steps: observing the light emission of plasma generated inside the processing container 1 from observation windows 12 arranged at multiple locations of the processing container 1; calculating the in-plane distribution of values ​​representing characteristics of the plasma on a substrate (for example, plasma density distribution) based on the observed data related to the light emission of the plasma; and correcting the scanning pattern and / or plasma density distribution of the plasma based on the calculated in-plane distribution of values ​​representing characteristics of the plasma on the substrate.

[0096] According to the plasma processing apparatus 10 of the present embodiment, the scanning pattern of the local plasma and / or the plasma density distribution are corrected by changing the power and phase of the microwaves supplied from each antenna unit 11 of the phased array antenna 2. Thus, microwave synthesis control can be freely performed based on microwave power phase control, and the time integral value of the plasma density can be corrected to an ideal spatial distribution.

[0097] For example, Figure 6 As shown in the example on the left side of , due to the influence of the non-uniformity of gas flow and the structure (installation error, processing tolerance, etc.) that cannot be expressed in the electromagnetic field simulation or plasma simulation, a plasma density distribution ρ different from the assumed ideal value on the right side is sometimes generated. In this case, it is necessary to readjust the power distribution and phase of each antenna unit 11 of the phased array antenna 2. In this case, based on the spectral measurement, the density distribution (luminous intensity distribution, etc.) of the chemical species that has a strong correlation with the process results is obtained. Therefore, even if the substrate W is not actually processed, the plasma density experienced by a certain point on the substrate W at a certain time can be inferred, and the two-dimensional plasma density distribution on the substrate W can be inferred. The film formation rate and etching rate experienced by a certain point on the substrate W at a certain time can be inferred through the inferred plasma density distribution.

[0098] In addition, by referring to the estimated distribution of process speeds such as film formation rate and etching rate, it is possible to lock in appropriate power and phase parameters with less substrate consumption, so that it becomes or approaches Figure 6 The ideal value of the plasma density distribution is shown in the example on the right.

[0099] In addition, for example, Fig.11 The estimated value of the plasma density distribution shown on the left side is not as Fig.11 If the final film formation amount or etching amount is not uniform as the ideal value shown on the right side of the substrate W, the final film formation amount or etching amount becomes non-uniform in the surface of the substrate W. In this case, in order to make the final film formation amount or etching amount uniform in the surface of the substrate W, it is necessary to readjust the local plasma irradiation time and the like. In this case, the scanning condition, that is, the plasma scanning pattern, is corrected so that the time integral value of the plasma density calculated based on the plasma light emission distribution measured by the spectrum measurement becomes uniform at all positions in the surface of the substrate. The power p and phase are adjusted by the power phase control unit 7. The plasma scanning pattern can be corrected by adjusting the power p and phase. As a result, it is expected that the uniformity of the film formation amount or etching amount within the surface of the substrate W can be improved. According to the control method involved in this embodiment, the time required to adjust the power p and phase can be reduced. The number of times the process is actually performed based on the parameters can reduce the process evaluation of repeated trials, reduce the consumed substrates, and save labor, time and cost.

[0100] In addition, you can also Fig. 9 In the “process of correcting the plasma scanning pattern” shown in step S11 , the power and / or phase output from at least one antenna unit 11 in the phased array antenna 2 is controlled.

[0101] In addition, the “step of correcting the plasma scanning pattern” may be performed by controlling the supply timing of the power output from at least any one of the antenna units 11 in the phased array antenna 2 .

[0102] In addition, the “process of correcting the plasma scanning pattern” may be performed at least one of before processing the substrate W, after maintenance of the plasma processing apparatus 10 , and after the process data used in the plasma processing apparatus 10 is changed.

[0103] In addition, based on the data related to the light emission of plasma observed when processing one substrate W, the "process of correcting the plasma scanning pattern" may be performed in the initial process of the substrate W to be processed after the processing of the substrate W. The control method involved in this embodiment may also be performed at the time of process development, after maintenance of the plasma processing device 10, and the like. In addition, the control method involved in this embodiment may also be used for real-time control of controlling the power and / or phase of each antenna unit 11 of the phased array antenna 2 when processing a next substrate based on the light emission intensity data when processing a previous substrate.

[0104] [Control method during substrate processing]

[0105] Next, refer to Fig.10 A control method (substrate processing) SW2 executed in the plasma processing apparatus 10 will be described. Fig.10 1 is a flowchart showing an example of the control method SW2 according to one embodiment.

[0106] First, the control unit 8 carries the substrate W into the processing container 1 and places the substrate W on the mounting table 3 to prepare the substrate W (step S21 ). Next, the control unit 8 supplies a predetermined gas from the gas supply unit 21 (step S23 ).

[0107] Next, the control unit 8 uses the power phase control unit 7 to control the power p and phase of the microwave radiated from each antenna unit 11 of the monopole antennas 2a to 2g. Adjust to the determined power p1~p n and Phase The microwaves with adjusted power and phase are supplied (step S25). The microwaves with controlled power and phase generate interference in the space V, and the microwaves can be synthesized at a predetermined position of the dielectric window 5. Thus, the local plasma generated between the dielectric window 5 and the mounting table 3 is scanned to process the substrate W (step S27). Next, the control unit 8 carries out the processed substrate W (step S29).

[0108] Next, the control unit 8 determines whether to end the processing of the substrate W (step S31). If there is a next substrate W, the control unit 8 determines not to end the processing of the substrate W, and prepares the next substrate W (step S33), returns to step S23, executes steps S23 to S27, and processes the next substrate W by local plasma.

[0109] In step S31, when there is no next substrate W, the control unit 8 determines that the processing of the substrate W is to be completed, and the present processing is completed. In addition, when the present processing is completed, the output of the microwave is stopped, and the supply of the gas is stopped.

[0110] [study]

[0111] Reference Fig.12 To illustrate the method of learning and inferring process prediction data. Fig.12 It is a diagram for explaining learning and estimation of process prediction data in substrate processing according to one embodiment.

[0112] exist Fig.12 In the learning phase shown in the upper part of , parameters (power, phase), target function, RF application time, and serial number of the plasma processing device 10 are used as input data. The parameters are power and phase controlled by the power phase control unit 7. The target function can also be luminous intensity data acquired at a series of action timings or a plasma density distribution ρ inferred based on the luminous intensity data. The RF application time is an example of elapsed time. The elapsed time can also be the time that has passed since the plasma processing device 10 was a new product or in an initial state after maintenance. The serial number of the plasma processing device 10 is an example of an identification number unique to the plasma processing device 10.

[0113] The learning device 100 is a device for performing machine learning such as deep learning. By inputting input data to the learning device 100, the change of the objective function corresponding to the elapsed time of each plasma processing device 10 is learned. After learning, the learning completion model obtained by learning is applied to the inference device 110. The learning device 100 and the inference device 110 can be the same device or different devices. For example, the learning completion model is composed of a set of learning completion programs and learning completion values, i.e., parameters.

[0114] exist Fig.12 In the inference stage shown in the lower part of FIG, parameters (power, phase), target function, RF application time, and serial number of the plasma processing apparatus 10 are input as input data to the inference device 110 equipped with the learned model. The inference device 110 predicts the change of the target function corresponding to the elapsed time of the plasma processing apparatus 10 identified by the input serial number, and outputs the prediction data of the process.

[0115] As an example of process prediction data, we can cite Figure 6 and Fig.11 The estimation device 110 can predict the change of the target function with time for each plasma processing device 10, and thus can output an appropriate plasma density distribution image in which the plasma density distribution is changed according to the mechanical error and the change over time of the plasma processing device 10. Thus, based on the output plasma density distribution image, it is possible to predict the area where the plasma density distribution is easy to change and the area where it is difficult to change without actually performing the process.

[0116] In addition, plasma density distribution is an example of process prediction data, but is not limited thereto. Other examples of process prediction data include, in the case of film formation, distribution of film thickness, film formation rate, and prediction data of the process indicating other film formation characteristics. Other examples of process prediction data include, in the case of etching, etching shape, etching depth, etching rate distribution, and prediction data of the process indicating other etching characteristics.

[0117] The control method and plasma processing apparatus involved in one embodiment disclosed herein should be considered to be illustrative in all aspects and not restrictive. The above-mentioned one embodiment can be modified and improved in various ways without departing from the attached claims and their gist. The matters described in the above-mentioned multiple embodiments can adopt other structures within the scope of non-contradiction, and can be combined within the scope of non-contradiction.

[0118] In addition, the plasma processing device disclosed in the present invention is illustrated by an example of a plasma processing device that radiates microwaves, but is not limited to this. The phased array antenna of the plasma processing device disclosed in the present invention is not limited to radiating microwaves, but can also radiate electromagnetic waves with a frequency of 100 MHz or more, such as UHF (Ultra High Frequency). The phased array antenna is more preferably able to radiate electromagnetic waves with a frequency in the range of 1 GHz to 3 GHz. In addition, the antenna constituting the phased array antenna is not limited to a monopole antenna, and other antennas such as a helical antenna and a patch antenna can be used.

Claims

1. A control method for controlling a scanning plasma processing device using a phased array antenna, the control method comprising the following steps: observing light emission of plasma generated inside the processing container from observation windows provided at a plurality of locations of the processing container; calculating, based on the observed data related to the light emission of the plasma, an in-plane distribution of a value representing a characteristic of the plasma on a substrate; as well as Based on the calculated in-plane distribution of the value representing the characteristic of the plasma on the substrate, a scanning pattern of the plasma and / or a plasma density distribution are corrected, Wherein, the control method further comprises the following steps: The power and phase output from the antenna portion of the phased array antenna, the plasma density distribution of the plasma processing device corresponding to the elapsed time, and the elapsed time are input as input data to an inference device equipped with a learning completion model obtained by learning the change in the plasma density distribution of the plasma processing device corresponding to the elapsed time, so as to predict the change in the plasma density distribution of the plasma processing device corresponding to the elapsed time.

2. The control method according to claim 1, wherein: In the step of correcting the scanning pattern and / or plasma density distribution of the plasma, power and / or phase output from at least one antenna unit of the phased array antenna is controlled.

3. The control method according to claim 1 or 2, wherein: In the step of correcting the scanning pattern and / or plasma density distribution of the plasma, supply timing of power output from at least one antenna unit of the phased array antenna is controlled.

4. The control method according to claim 1 or 2, wherein: The step of correcting the scanning pattern and / or plasma density distribution of the plasma is performed at least at any one of before processing a substrate, after maintenance of the plasma processing apparatus, and after process data used by the plasma processing apparatus is changed.

5. The control method according to claim 1 or 2, wherein: Based on data related to the light emission of the plasma observed when processing one substrate, a step of correcting the scanning pattern of the plasma and / or plasma density distribution is performed in an initial step of a substrate to be processed after the processing of the substrate.

6. The control method according to claim 1 or 2, wherein: The value indicating the characteristic of the plasma is any one of a density of the plasma, a value indicating a result of processing a substrate by the plasma, a processing rate of the substrate by the plasma, and light emission intensity of the plasma.

7. A plasma processing device, comprising: Handling containers; a gas supply unit for supplying gas into the processing container; a phased array antenna that irradiates electromagnetic waves toward the interior of the processing container; a light emission monitoring mechanism for observing light emission of plasma generated in the processing container from observation windows provided at a plurality of locations of the processing container; and a control unit that calculates an in-plane distribution of a value representing a characteristic of the plasma on a substrate based on data related to the light emission of the plasma observed by the light emission monitoring mechanism, and corrects a scanning pattern of the plasma and / or a plasma density distribution based on the calculated in-plane distribution of the value representing the characteristic of the plasma on the substrate, in, The control unit also performs the following processing: The power and phase output from the antenna portion of the phased array antenna, the plasma density distribution of the plasma processing device corresponding to the elapsed time, and the elapsed time are input as input data to an inference device equipped with a learning completion model obtained by learning the change in the plasma density distribution of the plasma processing device corresponding to the elapsed time, so as to predict the change in the plasma density distribution of the plasma processing device corresponding to the elapsed time.

Images