TWI665711B - Plasma processing device - Google Patents

Abstract

The present invention provides a pulse wave modulation method that changes the power of radio frequency used in plasma processing to switch between high and low levels in response to modulation of the pulse wave working ratio. In this plasma processing device, for example, when high / low pulse wave modulation is applied to a radio frequency used for plasma generation, a weighted average weight variable K is set to 0.5 <K <1 in a matcher. On the RF power supply line of the plasma generation system, during the on period of the pulse wave, the internal reflection wave T _on is still generated with a certain power PR _H. On the other hand, the reflected wave power PR _L during the pulse wave off period T _off is reduced. By adjusting the value of K, the balance of the reflected wave power in the pulse wave on period T _on and the reflected wave power in the pulse wave off period T _off can be arbitrarily controlled.

Description

Plasma processing device

The present invention relates to a technique for performing plasma treatment on an object to be treated, and more particularly to a plasma treatment device for a pulse wave modulation method that modulates the power of radio frequency used in plasma treatment with a pulse wave of a certain frequency.

Generally speaking, a plasma processing device manufactures a plasma of a processing gas in a processing container capable of being evacuated in a vacuum, and arranges the plasma in the processing container by the gas phase reaction or surface reaction of the radicals and ions contained in the plasma Deposit thin films on the object to be processed, or perform microfabrication such as cutting materials or films on the surface of the object to be processed.

In a capacitive coupling type plasma processing apparatus, an upper electrode and a lower electrode are arranged in parallel in a processing container, and an object to be processed (semiconductor wafer, glass substrate, etc.) is placed above the lower electrode. RF produced by plasma (generally above 13.56MHz). As a result of the application of a radio frequency, the electrons are accelerated by the radio frequency electric field between the upper and lower electrodes, and a plasma is generated by the collision ionization of the electrons colliding with the processing gas. In addition, the following RF bias method is also commonly used: a low-frequency (generally 13.56 MHz or lower) radio frequency is applied to the lower electrode of the object to be processed, and The negative bias voltage or sheath voltage generated on the electrodes accelerates the ions in the plasma and introduces them into the substrate. By the RF bias method, the self-plasma accelerates the ions and causes them to collide with the surface of the object to be treated, which can promote surface reactions, anisotropic etching, or film modification.

In recent years, in order to improve the yield and processing accuracy of dry etching, to prevent, for example, charging damage (damage of the gate oxide film caused by charge accumulation), or to suppress micro-load effects (pattern-based geometric structure or pattern density) Local difference in etching speed), and the technology of modulating the RF for plasma generation and / or the RF for bias with a certain frequency pulse wave is popular.

Generally speaking, this kind of pulse wave modulation responds to the working ratio of the modulated pulse wave. During the period when the pulse wave is turned on, the power of the radio frequency receiving the modulation is turned on in a certain position. During the period when the pulse wave is turned off, The power of the radio frequency is turned off at the zero level. Therefore, for example, when the power pulse wave of the radio frequency for plasma generation is modulated, the plasma is generated during the pulse-on period and the etching is performed, and the plasma disappears and the etching is temporarily stopped during the pulse-wave period. In this case, the matcher provided on the transmission line for the plasma generation radio frequency measures the load impedance during each period of the pulse wave on period, and variably controls the reactance of the variable reactance element provided in the matching circuit so that The load impedance measurement value is consistent or similar to the matching point (generally 50Ω).

[Xizhi technical literature]

[Patent Literature]

Patent Document 1: Japanese Patent Application Laid-Open No. 2012-9544

Patent Document 2: Japanese Patent Application Publication No. 2013-33856

As one form of the pulse wave modulation of the capacitive coupling plasma processing device as described above, there is a method as follows: according to the modulation of the pulse wave working ratio, the power of the radio frequency is controlled to a certain level during the pulse wave on period The standard is the high level, and the power of the power of the radio frequency is controlled to a certain low level lower to a higher level during the period when the pulse wave is off. Here, the low level is selected to be a value higher than the lowest level necessary to maintain the plasma generation state.

In these high / low pulse wave modulation methods, the pulse wave is still in the processing container during the closed period, so that a certain amount of plasma electrons and ions and even free radicals exist respectively. Taking advantage of this phenomenon, the low level of the power of the radio frequency and other process parameters are set to appropriate values to control the chemical or physical effect of electrons, ions and / or free radicals on the surface of the object to be treated. An effect of improving a predetermined etching characteristic in this etching process.

However, in the high / low pulse wave modulation method, if the frequency of the modulated pulse wave is set to a high value (generally 1 kHz or more), the variable control of the variable reactance element in the matcher becomes impossible. Track modulation pulses. Therefore, matching is performed only during the high pulse period which mainly affects the plasma processing, and the secondary low pulse period must be excluded from the matching object. As a result, a large reflected wave is generated on the RF power supply line during a period of low pulses that are not matched at all. Therefore, it is difficult to control the power of the radio frequency stably and accurately at a preset low level, and the expected effect of the process of the high / low pulse wave modulation mode is weakened, and the burden of the radio frequency power supply and the like is also increased.

In order to solve the problems of the conventional technology as described above, the present invention provides a plasma processing device, which can interact between the high level and the low level of the power of the radio frequency used in the plasma processing in response to the modulation of the pulse wave working ratio ( In particular, the pulse modulation method switched at high speed is efficient and utilized as expected.

The plasma processing apparatus of the present invention manufactures a plasma generated by a radio frequency discharge of a processing gas in a vacuum-exhaustable processing container that stores the processed object in and out so that the processed object can be moved in and out. The object to be processed in the processing container performs desired processing. The plasma processing device includes: a first radio frequency power source, which outputs a first radio frequency; a first radio frequency power modulation unit, which repeatedly repeats a first operation at a certain working ratio. And in the second period, modulating the output of the first radio frequency power supply with a modulation pulse of a certain frequency so that the power of the first radio frequency in the first period becomes a high level, and the second radio frequency in the second period The power becomes a lower level than the high level; a first radio frequency power supply line for transmitting the first radio frequency output from the first radio frequency power source to a first electrode arranged in or around the processing container; And a first matcher, measuring the impedance of the load observable from the first radio frequency power supply on the first radio frequency power supply line, so that the measured value of the load impedance in the first period and the Weighted average of the weighted average value of values measured in the load impedance during obtained, the output impedance of the first RF power.

In the above device configuration, the balance of the reflected wave power in the high pulse wave period and the reflected wave power in the low pulse wave period can be arbitrarily controlled by adjusting the value of the weight variable of the weighted average. With this, the power of the reflected wave during the low pulse period is arbitrarily reduced, and the load power can be set to an arbitrary value that increases this part And respond to process requirements. In addition, the self-reflected wave reduces the burden of a circulator and the like for protecting the RF power supply, and the resistance of the reflected power of the RF power supply itself can realize the compactness and simplification of the hardware and the efficiency of the power consumption around the RF power supply.

If the plasma processing device according to the present invention has the structure and function as described above, the power of the radio frequency used in the plasma processing can be changed between the high level and the low level according to the working ratio of the modulated pulse wave ( In particular, the pulse modulation method of high-speed) switching is realized efficiently and as expected.

10‧‧‧ chamber

12‧‧‧ Insulation board

14‧‧‧ base support desk

16‧‧‧ base (lower electrode)

18‧‧‧ electrostatic chuck

20‧‧‧ electrode

22‧‧‧Switch

24‧‧‧DC Power

26‧‧‧focus ring

28‧‧‧Inner wall members

30‧‧‧Refrigerant Room

32a, 32b‧‧‧Piping

34‧‧‧Gas supply line

36‧‧‧ (plasma generation system) RF power supply

38‧‧‧ (ion introduction system) RF power

40, 42‧‧‧ Matcher

43, 45‧‧‧ RF power line

44‧‧‧ Power Stick

46‧‧‧upper electrode (shower head)

48‧‧‧ electrode plate

48a‧‧‧gas ejection hole

50‧‧‧ electrode support

50a‧‧‧Air vent

52‧‧‧Gas buffer room

54‧‧‧Gas supply pipe

56‧‧‧Processing gas supply source

58‧‧‧mass flow controller (MFC)

60‧‧‧Open and close valve

62‧‧‧DC Power Supply Department

64‧‧‧ insulator

66, 68‧‧‧ DC Power

70‧‧‧ switch

72‧‧‧Main Control Department

74‧‧‧DC power supply line

76‧‧‧Filter circuit

78‧‧‧ exhaust port

80‧‧‧ exhaust pipe

82‧‧‧Exhaust

84‧‧‧ moved out of the entrance

86‧‧‧Gate Valve

90A, 90B‧‧‧RF Oscillator

92, 92A, 92B‧‧‧ Power Amplifier

94A, 94B‧‧‧ Power Control Department

96A, 96B‧‧‧RF Power Monitor

98A, 98B‧‧‧ matching circuit

100A, 100B, 102A, 102B‧‧‧ Motor

104A, 104B‧‧‧Matching Controller

106A, 106B‧‧‧Impedance sensors

107A, 107B‧‧‧V _pp detector

110A, 110B‧‧‧RF voltage detector

112A, 112B‧‧‧RF current detector

114A, 114B‧‧‧ Load impedance instantaneous value calculation circuit

116A, 116B‧‧‧ Arithmetic average operation circuit

118A, 118B‧‧‧weighted average operation circuit

120A, 120B‧‧‧moving average calculation circuit

122, 122A, 122B‧‧‧Load Power Measurement Department

124A, 124B‧‧‧ RF output control department

126A, 126B‧‧‧ (for pulse wave on period) control command value generation unit

128A, 128B‧‧‧ (for pulse wave off period) control command value generation unit

130A, 130B‧‧‧ Comparator

132A, 132B‧‧‧amplifier control unit

134A, 134B‧‧‧ Controller

136A, 136B‧‧‧switching circuit

140‧‧‧hole

142‧‧‧etch mask

144‧‧‧First SiO ₂ layer

146‧‧‧First SiN layer

148‧‧‧Second SiO ₂ layer

150‧‧‧Second SiN layer

152‧‧‧third SiO ₂ layer

154‧‧‧The third SiN layer

156‧‧‧Semiconductor substrate

cw‧‧‧ cooling water

PA‧‧‧Processing space (plasma generation space)

SH _DC ‧‧‧High electric field area (DC sheath)

SH _RF ‧‧‧ Plasma Cover (Ion Cover)

SW‧‧‧Switch control signal

W‧‧‧Semiconductor wafer

X _H1 , X _H2 ‧‧‧Reactance element

FIG. 1 is a cross-sectional view showing a configuration of a capacitive coupling type plasma processing apparatus of a dual-frequency superimposition method according to an embodiment of the present invention.

FIG. 2 is a waveform diagram showing a typical combination of waveforms in the case where high / low pulse wave modulation is applied to the RF for plasma generation.

Fig. 3 is a block diagram showing the configuration of a radio frequency power supply and a matcher for plasma generation.

FIG. 4A is a block diagram showing a configuration example of an impedance sensor provided in the matching device of FIG. 3.

FIG. 4B is a block diagram showing another configuration example of the impedance sensor.

FIG. 5A is a Smith chart showing the matching effect when the weighting variable K of the weighted average operation in the embodiment is selected such that K = 1.

FIG. 5B is a Smith chart showing the matching effect when the weighted variable K of the weighted average operation is selected such that 0.5 <K <1.

FIG. 6A is a waveform diagram showing the waveforms of each part when K = 1 is selected.

FIG. 6B is a waveform diagram showing the waveforms of each part when 0.5 <K <1 is selected.

FIG. 7 is a block diagram showing a configuration in the RF output control section of FIG. 3. FIG.

FIG. 8 is a block diagram showing the configuration of the RF power monitor and the power control section of FIG. 7.

FIG. 9 is a cross-sectional view for explaining the HARC process in the embodiment.

FIG. 10A is a graph showing the dependence of the pulse wave off period of one process characteristic (etching amount) obtained in the first experiment of the example.

FIG. 10B is a graph showing the dependence of the pulse wave closing period on one of the process characteristics (necking CD) obtained in the first experiment.

FIG. 10C is a graph showing the dependence of the pulse wave closing period of one of the process characteristics (middle Ox bowing CD) obtained in the first experiment.

FIG. 10D is a graph showing the dependence of the pulse-off period of one of the process characteristics (selection ratio) obtained in the first experiment.

FIG. 10E is a graph showing the dependence of the pulse wave closing period on one of the process characteristics (the aspect ratio change rate) obtained in the first experiment.

FIG. 11A is a graph showing the upper DC voltage dependency of one process characteristic (etching amount) obtained by the second experiment of Example 2. FIG.

FIG. 11B is a graph showing the upper DC voltage dependency of a process characteristic (reduced diameter CD) obtained by the second experiment.

FIG. 11C is a graph showing the upper DC voltage dependency of one of the process characteristics (middle Ox bending CD) obtained in the second experiment.

FIG. 11D is a graph showing the upper DC voltage dependency of one process characteristic (selection ratio) obtained in the second experiment.

FIG. 11E is a graph showing the upper DC voltage dependency of one of the process characteristics (aspect ratio change) obtained in the second experiment.

Fig. 12 is a graph showing the relationship between the load power that can be set in the RF power supply and the reflected wave power.

FIG. 13 is a diagram for explaining a mechanism of generating an abnormal discharge inside the upper electrode when pulse wave modulation of on / off is applied to both the RF for plasma generation and the RF for ion introduction.

FIG. 14 is a diagram for explaining a case where a high / low pulse wave modulation is applied to a radio frequency for plasma generation, and an on / off pulse wave modulation is applied to a radio frequency for ion introduction, and no abnormal discharge inside the upper electrode is generated. Diagram of mechanism.

FIG. 15 is a diagram showing an example of monitoring information obtained when an abnormal discharge inside the upper electrode occurs in the plasma processing apparatus of FIG. 1. FIG.

16 is a diagram showing an example of monitoring information obtained when the abnormal discharge inside the upper electrode is not generated in the plasma processing apparatus of FIG. 1.

FIG. 17 is a graph showing an experimental result of the presence or absence of occurrence of abnormal discharge inside the upper electrode in the plasma processing apparatus of FIG. 1.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

[Configuration of Plasma Processing Device]

FIG. 1 shows a configuration of a plasma processing apparatus according to an embodiment of the present invention. This plasma processing apparatus is constituted as a capacitive coupling (parallel flat plate) plasma etching apparatus of a lower dual radio frequency overlapping application method, and has, for example, a cylinder made of aluminum whose surface is treated with an aluminum oxide film (anodized). Shaped vacuum chamber (processing container) 10. The chamber 10 is grounded.

A cylindrical base support table 14 is arranged at the bottom of the chamber 10 through an insulating plate 12 made of ceramics or the like, and a base 16 made of, for example, aluminum is provided above the base support table 14. The susceptor 16 (first electrode) constitutes a lower electrode, and a semiconductor wafer W is mounted on the susceptor 16 as a processing object, for example.

An electrostatic chuck 18 for holding a semiconductor wafer W is provided on the top surface of the base 16. The electrostatic chuck 18 sandwiches an electrode 20 made of a conductive film between a pair of insulating layers or sheets. The electrode 20 is electrically connected to the DC power source 24 through a switch 22. The semiconductor wafer W can be held on the electrostatic chuck 18 with an electrostatic attraction force by a DC voltage from the DC power source 24. A focusing ring 26 made of, for example, silicon is arranged around the electrostatic chuck 18 on the top surface of the base 16 to improve the uniformity of etching. A cylindrical inner wall member 28 made of, for example, quartz is attached to the sides of the base 16 and the base support 14.

A refrigerant chamber 30 extending in the circumferential direction is provided inside the base support table 14, for example. In this refrigerant chamber 30, a refrigerant at a predetermined temperature, such as cooling water (cw), is circulated and supplied from an external quench unit (not shown) through pipes 32a and 32b. The processing temperature of the semiconductor wafer W on the susceptor 16 can be controlled by the temperature of the refrigerant. Further, a heat transfer gas such as He gas from a heat transfer gas supply mechanism (not shown) is supplied between the top surface of the electrostatic chuck 18 and the back surface of the semiconductor wafer W through a gas supply line 34.

The base 16 is electrically connected to the RF power sources 36 and 38 through the matching devices 40 and 42 and a common power supply conductor (such as a power supply rod) 44 respectively. One radio frequency power source 36 (first radio frequency power source) outputs radio frequency _HF of a certain frequency f _HF (for example, 40 MHz) suitable for generating plasma. The other radio frequency power source 38 (second radio frequency power source) outputs a radio frequency LF of a certain frequency f _LF (for example, 12.88 MHz) suitable for introducing ions from the plasma to the semiconductor wafer W on the base 16.

In this way, the matcher 40 and the power supply rod 44 constitute a part of the RF power supply line (RF transmission line) 43 that transmits the RF HF for plasma generation from the RF power source 36 to the base 16. On the other hand, the matcher 42 and the power supply rod 44 constitute a part of a radio frequency power supply line (radio frequency transmission path) 45 that transmits radio frequency LF for ion introduction from the radio frequency power source 38 to the base 16.

On the ceiling of the chamber 10, an upper electrode 46 is provided parallel to the base 16 and facing each other. This upper electrode 46 is composed of an electrode plate 48 having a plurality of gas ejection holes 48a, for example, made of a silicon-containing material such as Si, SiC, and the like, and an electrode support 50 supporting the electrode plate in a detachable manner 48, composed of a conductive material such as aluminum whose surface is treated with an aluminum oxide film. A processing space or a plasma generation space PA is formed between the upper electrode 46 and the base 16.

The electrode support 50 has a gas buffer chamber 52 inside, and a plurality of vent holes 50 a communicating with the gas ejection holes 48 a of the electrode plate 48 from the gas buffer chamber 52 on the bottom surface thereof. The gas buffer chamber 52 is connected to a processing gas supply source 56 via a gas supply pipe 54. A process gas supply source 56 is provided with a mass flow controller (MFC) 58 and an on-off valve 60. When a predetermined processing gas (etching gas) is introduced from the processing gas supply source 56 into the gas buffer chamber 52, the processing gas is directed from the gas ejection hole 48a of the electrode plate 48. The semiconductor wafer W on the susceptor 16 is sprayed out in a shower PA toward the plasma generation space PA. In this way, the upper electrode 46 (second electrode) also functions as a shower head that supplies the processing gas to the plasma generation space PA.

In addition, a passage (not shown) for circulating a refrigerant such as cooling water is also provided inside the electrode support 50, and the entire upper electrode 46, especially the electrode plate 48, is adjusted to a predetermined temperature by an external quencher unit. . Furthermore, in order to further stabilize the temperature control of the upper electrode 46, a configuration may be adopted in which a heater (not shown) made of, for example, a resistance heating element is mounted inside or on the top surface of the electrode support 50.

In this embodiment, a DC power supply unit 62 is provided for applying a DC voltage V _dc having a negative polarity to the upper electrode 46. Therefore, the upper electrode 46 is mounted on the upper portion of the chamber 10 in an electrically floating state by a ring-shaped insulator 64. The ring-shaped insulator 64 is made of, for example, alumina (Al ₂ O ₃ ), hermetically closes the gap between the outer peripheral surface of the upper electrode 46 and the side wall of the chamber 10, and physically supports the upper electrode 46 without being grounded. .

The DC power supply unit 62 includes two DC power supplies 66 and 68 having different output voltages (absolute values), and a switch 70 that selectively connects the DC power supplies 66 and 68 to the upper electrode 46. The DC power source 66 outputs a negative polarity DC voltage V _dc1 (eg, -2000 ~ -1000V) with a relatively large absolute value, and the DC power source 68 outputs a negative polarity DC voltage V _dc2 (eg, -300 ~ 0V) with a relatively small absolute value. The switch 70 is operated by receiving the switching control signal SW from the main control unit 72, and is between a first switching position where the DC power source 66 is connected to the upper electrode 46 and a second switching position where the DC power source 68 is connected to the upper electrode 46. Switch. Further, the switch 70 may have a third switch position that blocks the upper electrode 46 from the DC power sources 66 and 68.

A filter circuit 76 provided in the middle of the DC power supply line 74 between the switch 70 and the upper electrode 46 is configured to pass the DC voltage V _dc1 (V _dc2 ) from the DC power supply unit 62 while passing through it, and is applied to the upper electrode 46. On the other hand, the radio frequency that has entered the DC power supply line 74 through the processing space PA and the upper electrode 46 from the base 16 flows to the ground line and does not flow to the DC power supply portion 62 side.

In addition, in the chamber 10, a DC grounding member (not shown) made of a conductive material such as Si, SiC and the like is mounted at an appropriate place facing the plasma generation space PA. This DC grounding part is kept grounded by a ground wire (not shown).

An annular space formed between the base 16 and the base support 14 and the side wall of the chamber 10 becomes an exhaust space, and an exhaust port 78 of the chamber 10 is provided at the bottom of the exhaust space. Here, the exhaust port 78 is connected to the exhaust device 82 through an exhaust pipe 80. The exhaust device 82 has a vacuum pump such as a turbo molecular pump, and can decompress the interior of the chamber 10, especially the plasma generation space PA, to a desired vacuum degree. In addition, a gate valve 86 that opens and closes the loading / unloading port 84 of the semiconductor wafer W is mounted on the side wall of the chamber 10.

The main control section 72 includes one or more microcomputers, and controls various sections in the device, especially the RF power sources 36 and 38, and the matchers 40 and 42 according to the software (program) and recipe information stored in the external memory or the internal memory. , MFC58, on-off valve 60, DC power supply unit 62, exhaust device 82, etc., and the entire operation (program) of the device.

In addition, the main control unit 72 is also connected to an operation panel (not shown) for a human-machine interface including an input device such as a keyboard and a display device such as a liquid crystal display, and an external device for storing or storing various programs, recipes, and setting data A memory device (not shown) and the like. In this embodiment, although the main control unit 72 is shown as one control unit, a form in which a plurality of control units share the functions of the main control unit 72 in parallel or hierarchically may be adopted.

The basic operation of the monolithic dry etching in this capacitively coupled plasma etching apparatus is performed as follows. First, the gate valve 86 is opened, and the semiconductor wafer W to be processed is carried into the chamber 10 and placed on the electrostatic chuck 18. Next, an etching gas (generally a mixed gas), which is a processing gas, is introduced into the chamber 10 from the processing gas supply source 56 at a predetermined flow rate and flow rate. The pressure is the set value. Further, from the RF power sources 36 and 38, a radio frequency HF (40 MHz) for plasma generation and a radio frequency LF (12.88 MHz) for ion introduction are superimposed on the base 16 at a predetermined power, respectively. In addition, a DC voltage is applied from the DC power source 24 to the electrode 20 of the electrostatic chuck 18, and the semiconductor wafer W is fixed to the electrostatic chuck 18. The etching gas sprayed from the shower head of the upper electrode 46 is discharged under the radio frequency electric field between the two electrodes 46 and 16 to generate a plasma in the processing space PA. The processed film of the main surface of the semiconductor wafer W is ion-etched by the radicals contained in the plasma.

In this plasma etching apparatus, the following first (plasma generation system) power modulation method can be used in a predetermined etching process: the power of radio frequency HF used for plasma generation from the radio frequency power source 36 is provided in, for example, The modulation pulse MS with a certain frequency f _S and a variable operating ratio D _S selected in the range of 1 kHz to 50 kHz is modulated.

In this first power modulation mode, there are two modes of on / off pulse wave modulation and high / low pulse wave modulation. Here, the on / off pulse wave modulation is changed according to the modulation of the pulse wave MS working ratio, so that the power of the radio frequency HF used for plasma generation is set to the on-target state in the pulse pulse on period. During the period when the wave is turned off, the power of the radio frequency HF is turned off at the zero level. On the other hand, the high / low pulse wave modulation becomes, in response to modulating the working ratio of the pulse wave MS, the power of the RF HF is controlled to a high level during the pulse wave on period, and the RF HF is adjusted during the pulse wave off period. The power is controlled at a lower level which is higher and lower. Low level, choose a higher value than the lowest level necessary to maintain the state of plasma generation. In addition, for lower levels, higher levels are generally chosen to have significantly lower values (below 1/2).

In addition, in this plasma etching apparatus, in a predetermined etching process, the power of the radio frequency LF for ion introduction output from the radio frequency power source 38 can also be used to modulate the pulse wave MS second (ion introduction system) Power modulation mode. Similar to the first power modulation method, the second power modulation method also has two modes of on / off pulse wave modulation and high / low pulse wave modulation.

FIG. 2 shows an example of waveforms of various parts in a case where the plasma generation system and the ion introduction system are synchronized and pulse wave modulation is performed simultaneously. As shown in the figure, between the period T _{C of the} modulation pulse wave MS, the pulse wave on period (the first period) T _on and the pulse wave off period (the second period) T _off , there is T _C = T _on + T _off relationship. If the frequency of the modulation pulse MS is f _S , then T _C = 1 / f _S and the operating ratio D _S is D _S = T _on / (T _on + T _off ).

The example shown in the figure shows a case where the RF HF for plasma generation is subjected to high / low pulse wave modulation, and the RF LF for ion introduction is subjected to on / off pulse wave modulation. Furthermore, the DC voltage V _dc applied from the DC power supply unit 62 to the upper electrode 46 may be synchronized with the modulation pulse MS. In the example shown in the figure, a DC voltage V _{dc2 having a} small absolute value is applied to the upper electrode 46 during the pulse-on period T _on , and a DC voltage V _{dc1 having a} large absolute value is applied to the pulse-off period T _off .

[Composition of RF power supply and matching device]

FIG. 3 shows the configuration of the RF power source 36 and the matching device 40 of the plasma generating system according to this embodiment.

The RF power source 36 includes: an RF oscillator 90A, which is generally used to produce a basic radio frequency of a certain frequency (for example, 40 MHz) suitable for generating a plasma with a sine wave waveform; a power amplifier 92A, which is the basic RF output from this RF oscillator 90A The power is amplified with a controllable gain or magnification; and the power supply control section 94A directly controls the RF oscillator 90A and the power amplifier 92A according to a control signal from the main control section 72. Since the autonomous control unit 72, not only the control signal indicating the output mode of RF and the modulation pulse MS are given to the power control unit 94A, but also the control signal and power setting value of the general power on and off, power chain relationship, etc. Power control unit 94A. When pulse wave modulation (especially high / low pulse wave modulation) is performed on the RF HF for plasma generation, the power supply control section 94A (the first RF power modulation section) forms a pulse under the control of the main control section 72. Wave modulation department.

An RF power monitor 96A is also provided in the unit of the RF power source 36. Although not shown in the figure, this RF power monitor 96A includes a directional coupler, a traveling wave power monitor section, and a reflected wave power monitor section. Here, the directional coupler takes out signals corresponding to the power of the traveling wave propagating in the forward direction and the power of the reflected wave propagating in the reverse direction on the RF power supply line 43 respectively. The traveling wave power monitor unit generates a traveling wave power measurement value signal representing the power of the traveling wave included in the traveling wave on the radio frequency power supply line 43 based on the traveling wave power detection signal taken out by the directional coupler. Traveling wave power The measured value signal is given to the power control unit 94A in the radio frequency power supply 36 for power feedback control purposes, and is given to the main control unit 72 for display display purposes. The reflected wave power monitor unit measures the power of the reflected wave from the plasma in the chamber 10 and returns to the radio frequency power source 36. The reflected wave power measurement value obtained by the reflected wave power monitor unit is given to the main control unit 72 for display use, and is given to the power source control unit 94A in the radio frequency power supply 36 as a monitor value for power amplifier protection.

The matcher 40 includes a matching circuit 98A connected to the radio frequency power supply line 43 and has a plurality of, for example, two controllable reactance elements (such as a variable capacitor or a variable inductor) X _H1 , X _H2 ; a matching controller 104A The reactance of the reactance elements X _H1 and X _H2 is controlled by an actuator such as a motor (M) 100A and 102A; the impedance sensor 106A measures the impedance of the load including the impedance of the matching circuit 98A on the RF power line 43 ; V _pp detector 107A measures the peak-to-peak value V _pp of the radio frequency HF on the radio frequency power supply line 43 on the output terminal side of the matching circuit 98A. The internal configuration and function of the impedance sensor 106A and the function of the V _pp detector 107A will be described in detail later.

The RF power source 38 (Figure 1) of the ion introduction system differs only in the frequency of the RF LF and the RF HF, and is the same as the RF power source 36 of the plasma generation system described above, and includes: an RF oscillator 90B, a power amplifier 92B, The power source control unit 94B and the power monitor 96B (none of which is shown in the figure). In addition, the matching device 42 has a matching circuit 98B, motors (M) 100B and 102B, a matching controller 104B, an impedance sensor 106B, and a V _pp detector 107B (the same as the matching device 40 of the plasma generating system). None of the above).

[Composition of the impedance sensor]

FIG. 4A shows a configuration example of the impedance sensor 106A included in the matching device 40 of the plasma generation system. This impedance sensor 106A includes an RF voltage detector 110A, an RF current detector 112A, a load impedance instantaneous value calculation circuit 114A, an arithmetic average calculation circuit 116A, a weighted average calculation circuit 118A, and a moving average calculation circuit 120A. .

The RF voltage detector 110A and the RF current detector 112A respectively detect the voltage and current of the radio frequency HF on the radio frequency power supply line 43. The load impedance instantaneous value calculation circuit 114A calculates the instantaneous value of the load impedance Z on the RF power line 43 based on the voltage detection signal JV and the current detection signal JI obtained by the RF voltage detector 110A and the RF current detector 112A, respectively. JZ. The load impedance instantaneous value calculation circuit 114A may be an analog circuit, but it should be constituted by a digital circuit.

When the arithmetic mean calculation circuit 116A applies high / low pulse wave modulation to the radio frequency HF for plasma generation, in each cycle of the modulation pulse wave MS, the pulse wave ON period T _on is determined by The instantaneous value of the load impedance Z obtained by the load impedance instantaneous value calculation circuit 114A is sampled at a predetermined sampling frequency f _C , and the arithmetic mean value aZ _on of the load impedance Z within the pulse on period T _on is calculated, and will be during the pulse off period. T _off by the instantaneous value of the load impedance calculating circuit 114A to obtain the instantaneous value of the load impedance Z of the above-described JZ sampling sampling frequency f _C, the pulse wave calculating the arithmetic mean of the load impedance in the off period T _oFF of the Z aZ _off.

In the case where pulse wave modulation is applied to the RF HF for plasma generation, the arithmetic mean calculation circuit 116A, in each period of the modulation pulse wave MS, will borrow only within the pulse on period T _on the momentary load impedance value calculation circuits 114A to obtain the instantaneous value of the load impedance Z of JZ to the predetermined sampling sampling frequency f _C, the pulse wave calculating the arithmetic mean of the load impedance turned _on in the period T is Z aZ _on.

The main control unit 72 (FIG. 1), in synchronization with the modulation pulse MS, supplies the monitoring signal JS specifying the sampling time or monitoring time and the clock CK _{1 for} sampling to the arithmetic mean calculation circuit 116A. Here, in the case where the monitoring signal JS applies high / low pulse wave modulation to the RF HF for plasma generation, both the pulse wave on period T _on and the pulse wave off period T _off specify the monitoring time described later, respectively. T ₁ , T ₂ , in the case of applying pulse wave modulation on / off to the radio frequency HF, only the monitoring period T ₁ for the pulse wave on period T _on is specified. The arithmetic mean calculation circuit 116A is synchronized with the sampling clock CK _{1 of} several tens of MHz and requires high-speed and large-scale signal processing. Therefore, FPGA (Field Programmable Gate Array) can be suitably used.

The weighted average calculation circuit 118A is preferably composed of a CPU. When high / low pulse wave modulation is applied to the radio frequency HF for plasma generation, the pulse wave on period T _on obtained by the arithmetic average calculation circuit 116A is applied. in the arithmetic mean of the load impedance Z aZ _on, T _off in arithmetic average of the load impedance Z aZ _off during the pulse off, the desired weight to weight (weighting variables K) a weighted average, one obtains the load impedance The weighted average of the period bZ. The main control unit 72 gives the weighted variable K and the clock CK ₂ used for the weighted average operation to the weighted average operation circuit 118A.

In the case where pulse wave modulation is applied to the RF HF, the weighted average calculation circuit 118A is not operated, and the arithmetic average of the load impedance Z in the pulse on period T _on output from the arithmetic average calculation circuit 116A is not operated. The value aZ _on is sent to the moving average calculation circuit 120A in the subsequent stage without passing through the weighted average calculation circuit 118A.

The moving average calculation circuit 120A is preferably constituted by a CPU. When high / low pulse wave modulation is applied to the RF HF for plasma generation, based on the continuous majority load impedance Z obtained by the weighted average calculation circuit 118A Weighted average value bZ of one period, and calculate the moving weighted average value cZ of the load impedance Z, and output this moving weighted average value cZ as the measured value MZ of the load impedance Z.

In addition, the moving average calculation circuit 120A, when pulse wave modulation is turned on / off to the radio frequency HF, is based _on the load impedance Z in the continuous majority pulse on period T _on output from the arithmetic average calculation circuit 116A. arithmetic mean aZ _on, dZ calculated moving average, a moving average of this measured value dZ MZ output impedance Z as a load. The main control unit 72 supplies the set values of the moving interval L and the moving pitch P and the clock CK ₃ to the moving average calculation circuit 120A.

MZ from the mobile measurement value of the load impedance of the output average value calculating circuit 120A, CK ₃ updated in synchronization with the clock. Generally, the load-side impedance measurement value MZ includes the absolute value and phase measurement value of the load impedance Z.

FIG. 4B shows another configuration example of the impedance sensor 106A. As shown in the figure, the weighted average calculation circuit 118A may be provided after the moving average calculation circuit 120A. In this configuration example, when high / low pulse wave modulation is applied to the RF HF for plasma generation, the moving average calculation circuit 120A is based on a continuous majority (n during a) pulse wave T _on the opening of the arithmetic mean of the load impedance Z aZ _on, off, and arithmetic average pulse period T _off of the load impedance Z of the aZ _off, and turned on during the calculation of pulse in the T _on T _off the movement of the load impedance Z load impedance Z during the moving average eZ _on, and the average value of the pulse wave off eZ _off.

The weighted average calculation circuit 118A shifts the moving average value eZ _on of the load impedance Z in the pulse wave on period T _on obtained by the moving average calculation circuit 120A and the movement of the load impedance Z in the pulse wave off period T _off The average value eZ _{off is a} weighted average with the desired weight (weight variable K), to obtain a weighted moving average fZ of the load impedance Z, and output the weighted moving average fZ as the load impedance measurement value MZ.

In the case where pulse wave modulation is applied to the RF HF for plasma generation, the weighted average calculation circuit 118A is not operated, and the load in the pulse on period T _on output from the moving average calculation circuit 120A is not operated. The moving average value eZ _{on of the} impedance Z is output as the load impedance measurement value MZ without change.

The matching device 42 (FIG. 1) of the ion introduction system is the same as the impedance sensor 106A in the matching device 40 of the plasma generation system, and includes an impedance sensor 106B having the following elements: an RF voltage detector 110B, The RF current detector 112B, the load impedance instantaneous value calculation circuit 114B, the arithmetic average calculation circuit 116B, the weighted average calculation circuit 118B, and the moving average calculation circuit 120B (all are not shown above). In this impedance sensor 106B, the mode (high / low or on / off) of the pulse wave modulation applied to the radio frequency LF for ion introduction is switched, and the weighted average calculation circuit 118B and the movement are switched in the same manner as described above. Signal processing in the average calculation circuit 120B.

[The role of the matcher]

Here, the function of the matching device 40 of the plasma generation system when high / low pulse wave modulation is applied to the power of the RF HF for plasma generation will be described. In addition, on / off pulse wave modulation is applied to the power of the radio frequency LF for ion introduction under the same modulation pulse wave MS.

In this case, on the RF power supply line 43 of the plasma generation system, starting from the RF power source 36, not only during the pulse-on period T _on , but also during the pulse-off period T _off , the plasma load in the chamber 10 continues. Ground RF HF. However, in the iontophoresis system, the power of the radio frequency LF is turned on and off in synchronization with the modulation of the operating ratio of the pulse wave MS, so the plasma load that can be observed by the matcher 40 of the plasma generation system during the pulse wave turn-on period T _on and the pulse T _off changes significantly during wave off. Therefore, if the frequency of the modulated pulse wave MS is set to a high value (generally 1 kHz or more), the matching unit 40 of the plasma generating system is controlled by the matching controller 104A through the motors 100A and 102A to make the The reactance of the reactance elements X _H1 and X _H2 is variable and the automatic matching operation becomes impossible to track the modulation pulse MS.

In this embodiment, even if the frequency of the modulated pulse wave MS is increased to such an extent that the automatic matching action of the matcher 40 cannot be tracked, the special signal processing in the impedance sensor 106A described later is still in the pulse wave on period The balance of the degree of matching or mismatch between T _on and T _off during the pulse wave off period can effectively and stably use high / low pulse wave modulation.

In this case, the main control unit 72, for the RF power source 36 of the plasma generation system, alternately repeats the preset high-level power and the preset low-level power radio frequency in response to the modulation of the operating ratio of the pulse wave MS. The HF output method is given a predetermined control signal, set value, and timing signal to the power supply control section 94A. Then, the main control unit 72 gives the monitoring signal JS, the weight variable K, and the setting values L and P for the moving average calculation to the impedance sensor 106A in the matcher 40 for the high / low pulse wave modulation. , and clock _{CK 1, CK 2, CK 3} .

On the other hand, the main control unit 72 repeats the preset on-level (on-state) and zero-level alternately to the RF power source 38 of the ion introduction system so that the power of the radio-frequency LF responds to the modulation of the operating ratio of the pulse wave MS. (Closed state), a predetermined control signal, a set value, and a timing signal are given to the power control unit 94B. Then, the main control unit 72 gives the monitoring signal JS necessary for the on / off pulse wave modulation to the impedance sensor 106B in the matcher 42, the set values L and P for the moving average calculation, and the time pulse CK. ₁ , CK ₂ , CK ₃ . However, no weight variable K is given.

In the matching device 40 of the plasma generating system, as shown in FIG. 6A or FIG. 6B, in each period of the modulation pulse wave MS, the monitoring time is set in the pulse on period T _on and the pulse off period T _off respectively. T ₁ and T ₂ . Preferred aspect, for the pulse wave is turned on during T _on, in addition to rapid changes immediately after the start of power and the reflected wave other than the transition time interval immediately before the end of the RF power supply line 43, the set monitoring time T _1. Similarly, in the pulse wave off period T _off , the monitoring time T _{2 is} set in a period other than the transition time immediately after the start and immediately before the end.

The arithmetic mean calculation circuit 116A in the impedance sensor 106A adjusts the instantaneous value of the load impedance Z obtained by the load impedance instantaneous value calculation circuit 114A during the pulse wave on period T _on during each period of the modulated pulse wave MS. JZ sampling to sampling clock CK _1, calculates the pulse wave period T turned _on in the load impedance Z of the arithmetic mean _on aZ, in the pulse wave by closing the momentary load impedance value calculation circuit 114A is obtained within the period T _off load The instantaneous value JZ of the impedance Z is sampled at the sampling clock CK ₁ , and the arithmetic mean value aZ _off of the load impedance Z in the pulse wave off period T _off is calculated.

The weighted average calculation circuit 118A calculates the arithmetic mean value aZ _on of the load impedance Z in the pulse-on period T _on obtained by the arithmetic average calculation circuit 116A and the calculation of the load impedance Z in the pulse-off period T _off The average value aZ _{off is a} weighted average with a desired weight (weight variable K), and a weighted average value bZ of one period of the load impedance is obtained. Here, the weight variable K is an arbitrary value selected in the range of 0 ≦ K ≦ 1, and the weighted average value bZ is expressed by the following formula (1).

bZ = K ＊ aZ _on + (1-K) ＊ aZ _off ‧‧‧‧ (1)

The moving average calculation circuit 120A applies high / low pulse wave modulation to the RF HF for plasma generation, based on the continuous majority (n) load impedance Z obtained by the weighted average calculation circuit 118A. A periodic weighted average bZ, and a predetermined moving interval L and a predetermined moving interval P are used to calculate a moving weighted average cZ of the weighted average bZ. For example, when the frequency f _S of the modulation pulse MS is 1000 Hz, when the moving interval L is set to 10 msec and the moving pitch P is set to 2 msec, 10 consecutive one-cycle weighted average values bZ are calculated every 2 msec. 1 Moving average cZ.

The moving average calculation circuit 120A outputs the moving weighted average cZ as the load impedance measurement value MZ. This impedance measurement values of the MZ load, relative to the value given from the main control unit 72 of the weight calculating circuit 118A weighted average weight of K variables, regardless of the work becomes the MS pulse modulation ratio D _S.

The matching controller 104A of the matcher 40 can track and respond to the load impedance measurement value MZ output by the moving average calculation circuit 120A from the impedance sensor 106A at a period of the clock CK _3. The matching controller 104A drives and controls the motor 100A, 102A and variably control the reactances of the reactance elements X _H1 and X _H2 in the matching circuit 98A so that the phase of the load impedance measurement value MZ becomes zero (0) and the absolute value becomes 50Ω, which is the same as the matching point Z _S or approximate.

In this way, the matching device 40 performs a matching operation so that the load impedance measurement value MZ output from the impedance sensor 106A is consistent with or similar to the matching point Z _S. That is, the load impedance measurement value MZ becomes a matching target point. Thus, the pulse wave open arithmetic average load impedance period T _on of the Z, aZ arithmetic mean period _on, and the pulse wave is turned on T _off of the load impedance Z aZ _OFF, in response to the value of the weighted average of the weighting variables K, self- The matching points Z _S are offset by a ratio of (1-K): K.

Here, if the weight variable K given to the impedance sensor 106A of the matching unit 40 by the autonomous control unit 72 is set to K = 1, the right side of the above-mentioned weighted average calculation formula (1) is equal to The weight K of aZ _on becomes the maximum value “1”, and the weight (1-K) of aZ _off with respect to the second term becomes the minimum value, that is, zero “0”. In this case, as shown in the Smith chart of FIG. 5A, the arithmetic average value aZ _on of the load impedance Z in the pulse-on period T _on is identical or approximate to the matching point Z _S. On the other hand, the arithmetic mean pulse off period T _off of the load impedance Z of the aZ _off, since the matching point furthest from _S Z offset.

In the case of setting K = 1 in this way, as shown in the waveform diagram of FIG. 6A on the RF power supply line 43 of the plasma generating system, the pulse wave is turned on during the period T _on , and the power is slightly completely matched, so the power of the reflected wave PR _{H is} almost present, and the power PF _{H of the} traveling wave remains unchanged to become the load power PL _H. On the other hand, during the pulse off period T _off , the matching is shifted the most, so the power PR _{L of the} reflected wave becomes Very high, the power PF _{L of the} traveling wave is significantly higher than the load power PL _{L by} this part.

In addition, as for the radio frequency power source 36 of this embodiment, regarding the control of the power of the radio frequency HF, it is possible to selectively implement a PF control that keeps the power PF of the traveling wave to a certain value, and subtract the power of the self-propagating wave PF from the reflected wave The net input power (load power) of the power PR is maintained at any one of a certain PL control. In the case where high / low pulse wave modulation is applied to the power of the radio frequency HF, it is suitable to use a PL control that can set a low level of power that is set to a low value at least during the pulse wave off period T _off to the PL control. . However, if the PL control is used under the condition of K = 1, as in the conventional technique, since the pulse wave is not completely matched during the _off period T _off , the power PR _{L of the} reflected wave changes significantly as shown in FIG. 6A Big.

In this embodiment, the above-mentioned problem can be handled by setting the weight variable K to 0.5 <K <1. That is, in the case of 0.5 <K <1, in the right side of the above-mentioned weighted average expression (1), the weight K with respect to aZ _on of the first term becomes smaller than the maximum value “1”, and relative to The weight (1-K) of aZ _off in the second term becomes larger than the minimum "0" by this part. Thus, the Smith chart shown in FIG. 5B, the arithmetic mean pulse open period T _on the load impedance Z of the matching point Z from aZ _{S on} the bias, and pulse off period T _off of the load impedance Z of the arithmetic The average value aZ _off approaches the offset point towards the matching point Z _S.

Here, the matching point Z _S is located _on a straight line (middle point) connecting the load impedance measurement values (arithmetic mean values) aZ _on and aZ _off in the two periods T _on and T _{off on the} Smith chart. Further, the value of K farther away from one (or closer to 0.5), the pulse load impedance measurement value T _on the period of waves open aZ _on the farther from the matching point Z _S, the pulse wave off the load impedance measured value T _off DURING aZ _off the Close to the matching point Z _S.

In this way, when the weight variable K is set to 0.5 <K <1, as shown in the waveform diagram of FIG. 6B, the RF power supply line 43 of the plasma generation system also uses a certain power PR during the pulse on period T _on . _H generates a reflected wave. On the other hand, the power PR _L of the reflected wave in the off period T _off of the pulse wave is more reduced than in the case of K = 1. By adjusting the value of K, the balance of the reflected wave power PR _H in the pulse-on period T _on and the reflected wave power PRL in the pulse-off period T _off can be arbitrarily controlled.

Thereby, the power PR _L of the reflected wave in the off period T _off of the pulse wave can be arbitrarily reduced, and the load power PL _{L can be} set to increase any part of this value in response to the process requirements. In addition, it can reduce the burden of the circulator and the like used to protect the RF power source 36 from self-reflected waves, or the tolerance of the reflected power of the RF power source 36 itself. . Furthermore, by reducing the power PR _{L of the} reflected wave, as described later, the PL control for maintaining the net radio frequency power (load power) PL of the input plasma load to a set value can be performed more accurately and efficiently.

In addition, the weight variable K is not limited to a range of 0.5 <K ≦ 1, and may be set within a range of 0 ≦ K ≦ 0.5. In the case of K = 0.5, either of the weight K of aZ _on relative to the first term and the weight (1-K) of aZ _off relative to the second term in the right side of the above-mentioned weighted average expression (1) become equal to 0.5, although not shown, the matching point on the Smith chart of the pulse wave located Z _S open load impedance measurement value T _on the measured value of the load impedance during aZ T _off and _on during the pulse off _{off on} aZ midpoint.

In addition, when 0 ≦ K <0.5, the weight K of aZ _on with respect to the first term in the right side of the above-mentioned weighted average expression (1) is greater than the weight (1-K) with respect to aZ _off of the second term. Is small, so the measured load impedance aZ _on during the pulse-on period T _on is relatively far away from the matching point Z _S , and the measured load impedance aZ _off during the pulse-off period T _off is relatively close to the matching point Z _S . In this case, the power PR _L of the reflected wave during the pulse-off period T _off is relatively small, and the power PR _H of the reflected wave during the pulse-on period T _on is relatively large.

In this way, in this embodiment, the duty ratio D _{S of the} pulse wave MS can be adjusted independently, and the reflected wave power PR _H in the pulse on period T _on and the reflected wave power in the pulse off period T _off can be arbitrarily controlled. The balance of PR _L (or the degree of matching or non-matching). The main control unit 72 arbitrarily sets the weight variable K in the range of 0 ≦ K ≦ 1 in the process recipe, and can switch the weight variable K in each process, or switch the weight variable stepwise or continuously in one process. K.

In addition, in the matcher 42 of the ion introduction system, the on / off pulse wave modulation is applied to the radio frequency LF, so the main control unit 72 does not give the weighted variable K to the impedance sensor 106B as described above, and the weighted average operation is performed. Circuit 118B is not functioning. The moving average calculation circuit 120B calculates the moving average based _on the arithmetic mean aZ _on of the impedance Z in the continuous majority of pulse-on periods T _on output from the arithmetic average calculation circuit 116B in each cycle of the clock CK ₁ . dZ, and outputs the moving average dZ as the measured value MZ of the load impedance Z.

The matching controller 104B of the matching device 42 can track and respond to the load impedance measurement value MZ outputted by the clock CK ₃ with the moving average calculation circuit 120B from the impedance sensor 106B. The matching controller 104B drives and controls the motor 100B, 102B and variably control the reactances of the reactance elements XL1 and XL2 in the matching circuit 98B so that the phase of the load impedance measurement value MZ becomes zero (0) and the absolute value becomes 50Ω, which is the same as or close to the matching point Z _S. In this case, the arithmetic average value aZ _on to the moving average value cZon of the load impedance Z in the pulse on period T _on always becomes the matching target point.

[Structure of the main part in the power supply control part]

FIG. 7 and FIG. 8 show the structure of the main parts in the power source control section 94A of the RF power source 36 of the plasma generating system.

As shown in FIG. 7, the power supply control unit 94A includes a load power measurement unit 122A and a radio frequency output control unit 124A. The load power measurement section 122A calculates the load power of the load (mainly plasma) from the traveling wave power detection signal S _PF and the reflected wave power detection signal S _PR obtained by the RF power monitor 96A. Measured value of _PL M _PL (M _PL = S _PF -S _PR ).

The load power measurement unit 122A may have any of an analog operation circuit and a digital operation circuit. That is, the load power measurement value M _{PL of the} analog signal can be generated by taking the difference between the analog traveling wave power detection signal S _PF and the analog reflected wave power detection signal S _PR , or the traveling power detection signal S _PF and reflected wave power detection signal S _PR are respectively converted into digital signals and the difference between the two is taken to generate a digital signal load power measurement value M _PL .

As shown in FIG. 8, the RF output control unit 124A includes a first control command value generation unit 126A for the pulse wave on period (first period), and a second control command value for the pulse wave off period (second period). The generating unit 128A; the comparator 130A compares the traveling wave power detection signal S _PF from the RF power monitor 96A with the first control command value C _on or the second control command value from the first control command value generating unit 126A. The second control command value C _{off of the} generating unit 128A is compared to generate a comparison error ER _on or ER _off . The amplifier control unit 132A variably controls the power amplifier 92 in response to the comparison error ER _on or ER _off from this comparator 130A. Gain or amplification; and a local controller 134A that controls each of the sections within the RF output control section 124A.

Here, the first control command value generation unit 126A inputs the load power measurement value M _PL given from the load power measurement unit 122A and the load power set value PL _H (or PL _on) given from the main control unit 72 by the controller 134A. ), The first control command value C _on used for the feedback control of the power PF of the traveling wave is generated in the pulse wave on period T _on in each period of the modulated pulse wave MS.

On the other hand, the second control command value generation unit 128A inputs the load power measurement value M _PL from the load power measurement unit 122A and the load power set value PL _L from the controller 134A to modulate each cycle of the pulse wave MS. T _off during the closing of the pulse generator is applied to the traveling wave power PF feedback control a second control command value used C _off.

The first control command value generation unit 126A and the second control command value generation unit 128A are preferably configured by digital circuits. In this case, by setting a digital-to-analog (D / A) converter in each output section, the first control command value C _on and the second control command value C _{off can} be output as analog signals.

The first control instruction value C _on output from the first control instruction value generation unit 126A and the second control instruction value C _off output from the second control instruction value generation unit 128A are given to the comparator interactively by the switching circuit 136A. 130A. The switching circuit 136A operates under the control of the controller 134A, and selects the first control command value C _on from the first control command value generation unit 126A in the pulse wave on period T _on in each period of the modulation pulse wave MS. is transmitted to the comparator 130A, a second selection control instruction from the second control command value generating portion 128A is closed during the pulse C _off value T _off is transmitted to the comparator 130A.

Thus, the comparator 130A, to each of the pulse wave period becomes the MS tune, the opening C _on the comparison period T is generated _on the traveling wave power detection signal S _PF and the first control command value which compares the error in the pulse wave ie 1Comparison error ER _on (ER _on = C _on -S _PF ), during the pulse wave off period T _off , the traveling wave power detection signal S _{PF is} compared with the second control command value C _off to generate its comparison error, which is the second Comparison error ER _off (ER _off = C _off -S _PF ).

The amplifier control unit 132A operates under the control of the controller 134A, and variably controls the gain or amplification factor of the power amplifier 92A during the pulse on period T _on during each period of the pulse wave MS to make the first comparison. The error ER _{on is} close to zero to control the output of the RF power source 36. During the pulse wave off period T _off , the gain or amplification of the power amplifier 92A is variably controlled so that the second comparison error ER _{off is} close to zero to control the RF power source 36 Output.

For the power amplifier 92A, a linear amplifier (linear amplifier) is appropriately used. The comparator 130A uses, for example, a differential amplifier. In the comparator 130A, a certain proportional relationship may be established between the difference (C _on -S _PF ) or (C _off -S _PF ) of the input signal and the comparison error ER _on or ER _{off of the} output signal.

The RF power source 38 of the iontophoresis system is the same as the RF power source control unit 94A in the RF power source 36 of the plasma generating system except that the frequency of the RF LF and the RF HF of the plasma generating system are different. The structure and function of the load power measurement unit 122B and the RF output control unit 124B (neither of which is shown).

[Effect of PL control in the embodiment]

In the plasma processing apparatus of this embodiment, when any one of the RF power sources 36 and 38 is supplied with RF HF for plasma generation or RF LF for ion introduction into the chamber 10, it is used to charge The net RF power of the load (mainly plasma), that is, the load power PL, is controlled by the PL control that the pulse-on period T _on and the pulse-off period T _{off are} individually maintained at the set values.

In the following, the effect of the PL control in this embodiment will be described with respect to the case where high / low pulse wave modulation is applied to the power of the RF HF for plasma generation. In addition, on / off pulse wave modulation is applied to the power of the radio frequency LF for ion introduction under the same modulation pulse wave MS.

In this case, the main control section 72 gives the power control section 94A of the radio frequency power supply 36 of the plasma generation system to the control signals and load power setting values PL _H and PL _L necessary for high / low pulse wave modulation. And give the modulation pulse wave MS as the timing signal for pulse wave modulation. In addition, PL _{H is} a first load power setting value that specifies the power level (high level) of the radio frequency HF during the pulse-on period T _on . On the other hand, PL _{L is} a second load power setting value that specifies the power level (low level) of the radio frequency HF during the pulse-off period T _off . The radio frequency power supply 36 performs the following PL control on the radio frequency HF output from the power supply 36 with high / low pulse wave modulation.

First, in the future, the load power setting values PL _H and PL _{L of the} autonomous control unit 72 are set in the controller 134A in the radio frequency output control unit 124A. The controller 134A gives load power set values PL _H and PL _L and required control signals and clock signals to the first control command value generation unit 126A and the second control command value generation unit 128A.

A first control command value generation section 126A, modulation in each cycle of the pulse wave MS, the load power from the load power measuring portion measurement value M _PL 122A only capturing between open period T _on is used in the pulse wave in the return Grant signal. Here, although the instantaneous value or representative value of the load power measurement value M _PL may be used as the feedback signal, the average value (preferably a moving average value) of the load power measurement value M _PL is generally used as the feedback signal.

Specifically, for the load power measurement value M _PL given from the load power measurement unit 122A between the pulse wave on period T _on , a moving average value AM _PL of the complex cycle minutes of the modulation pulse wave MS is obtained, and this moving average value is obtained. AM _{PL is} compared with the load power set value PL _H to obtain the comparison error or deviation. In the next or subsequent cycle, it is determined that the deviation approaches zero at a suitable speed, and it is decided to apply a traveling wave during the pulse on period T _on The target value of the feedback control of the power PF is the first control command value C _on . In order to determine the first control command value C _on , a conventional algorithm commonly used in the technology of feedback control or feedforward control can be used.

On the other hand, the second control command value generation unit 128A captures the load power measurement value M _PL given by the load power measurement unit 122A only during the pulse wave off period T _off in each cycle of the modulation pulse wave MS. Use it for feedback signals. However, although an instantaneous value or a representative value of the measured load power value M _PL may be used as the feedback signal, an average value (preferably a moving average value) of the measured load power value M _PL is generally used as the feedback signal.

Specifically, a moving average value BM _PL of one cycle point or a plurality of cycle points is obtained for the load power measurement value M _PL given from the load power measurement section 122A between the pulse wave off period T _off , and this moving average value BM is obtained. _{PL is} compared with the load power set value PL _L to obtain a comparison error or deviation. In the next or subsequent cycle, the deviation of the travelling wave during the _off period T _off is determined so that the deviation approaches zero at an appropriate speed. The target value of the feedback control of the power PF is the second control command value C _off . In order to determine the second control command value C _off , a conventional algorithm commonly used in feedback control or feedforward control can be used.

As described above, the comparator 130A, in each period of the modulated pulse wave MS, transmits the traveling wave power detection signal S _PF and the first control command from the first control command value generating section 126A within the pulse on period T _on . The value C _{on is} compared to generate its comparison error (the first comparison error) ER _on , and the travelling wave power detection signal S _PF and the second control command from the second control command value generation unit 128A are generated during the pulse wave off period T _off . The value C _{off is} compared to generate its comparison error (second comparison error) ER _off . Then, the amplifier control unit 132A variably controls the gain or amplification factor of the power amplifier 92A during the pulse-on period T _on during each period of the modulation pulse wave MS so that the first comparison error ER _on approaches zero. The gain or amplification of the power amplifier 92A is variably controlled during the pulse wave off period T _off so that the second comparison error ER _off approaches zero.

In this way, in the radio frequency power source 36 that outputs radio frequency HF with high / low pulse wave modulation, feedback control is applied to the power PF of the traveling wave propagating in the forward direction on the radio frequency power supply line 43 so that The measured value M _PL of the load power PL obtained by the power monitor 96 and the load power measurement unit 122 is the same as or similar to the first load power set value PL _H within the pulse on period T _{on and} within the pulse off period T _off It is the same as or similar to the second load power set value PL _L. That is, the output of the RF power source 36 is subjected to independent feedback control during the pulse-on period T _on and the pulse-off period T _off .

According to the feedback control of the independent two systems based _{on the} pulse on period T _on and the pulse off period T _off , the reflected wave power PR synchronized with the modulated pulse MS can be simply and accurately tracked to the traveling wave. The periodic variation of the power PF makes it easy to keep up with the rapid load variation that occurs when the modulation pulse MS reverses. Thereby, even if the frequency of the modulated pulse wave MS is increased, the load power PL can still be stably maintained as an individual set value PL _H during any of the pulse on period T _on and the pulse off period T _off . , PL _L.

On the other hand, in the radio frequency power supply 38 of the ion introduction system that applies pulse wave modulation on / off to the radio frequency LF, the power supply control section 94B (second radio frequency power modulation section) is used to modulate the pulse wave MS. In each cycle, feedback control for PL control is applied to the power PF of the traveling wave only during the pulse-on period T _on . The controller 134B in the power supply control section 94B keeps the second control command value generating section 128B used during the pulse wave off period in a completely inactive or non-active state, and generates only the first control command value used during the pulse wave on period. The unit 126B operates. This case, the first control command value generating section 126B, the pulse wave is turned on to give an indication of the RF power level of HF in the period T _on (ON level) of the set value of the load power PL _on.

The comparator 130B, during each period of the modulation pulse wave MS, transmits the traveling wave power detection signal S _PF from the RF power monitor 96B and the first control command value generation unit 126B during the pulse wave on period T _on . The first control command value C _{on is} compared to generate its comparison error (first comparison error) ER _on , and it is substantially stopped during the pulse wave off period T _off . In addition, the amplifier control unit 132B variably controls the gain or amplification factor of the power amplifier 92B during the pulse-on period T _on during each period of the modulation pulse wave MS so that the first comparison error ER _oo approaches zero. During the pulse wave off period, T _off is substantially stopped.

In the RF power source 38 that performs on / off pulse wave modulation, PF control can also be performed. In this case, the controller 134B may give the comparator 130B a traveling wave power set value (PF _S ) as a comparison reference value.

[Example of Etching Process]

The inventor of this case performed an experiment using a high- / low-pulse-modulated HARC (High Aspect Ratio Contact) process using the plasma etching apparatus of FIG. 1 to verify that T _on as the value of the parameter T _on the upper portion of the DC voltage of the RF power during the period length T _on the pulse wave in the closed (load power) PL _L, or pulse off, given the role of the various characteristics of the process.

In this experiment, as shown in FIG. 9 (a), it is prepared to form fine holes 140 in the surface layer portion of the multilayer film structure by the first etching step until halfway (until the depth d ₁ of the third SiO ₂ layer 152 is reached). ) Semiconductor wafer W as a sample. Then, on the semiconductor wafer W of this sample, as shown in FIG. 9 (b), the depth of the fine holes 140 is extended to the second part up to the lower part of the third SiO ₂ layer 152 (up to the depth d ₂ ). In the etching step, the following experiments were performed: applying high / low pulse wave modulation to the radio frequency HF for plasma generation, and applying on / off pulse wave modulation to the radio frequency LF for ion introduction, so that the The magnitude (absolute value) of the DC voltage (upper DC voltage) V _dc is variable in synchronization with the modulation pulse MS. In FIG. 9, 142 is an etching mask (photoresist), 144 is a first SiO ₂ layer, 146 is a first SiN layer, 148 is a second SiO ₂ layer, 150 is a second SiN layer, and 152 is a third SiO ₂ Layers, 154 is a third SiN layer, and 156 is a semiconductor substrate.

The process characteristics selected as the evaluation object in this experiment are: [1] the increment of the depth of the hole 140 in the second etching step (d ₂ -d ₁ ), that is, the etching amount; [2] the shrinkage near the entrance of the hole 140 Incremental diameter (reduced diameter CD); [3] incremental increase in bending in the second SiO ₂ layer 148 (middle Ox curved CD); [4] selection ratio (increment of depth of hole 140 d ₂ -d ₁ / The reduction of the thickness of the mask by d _m ); and [5] the amount of change in the aspect ratio (the increment of the depth of the hole 140 d ₂ -d ₁ / the middle Ox bend CD).

The experiment of the second etching step, in more detail, includes the case where the RF HF power PL _L in the pulse off period T _off is set to 0 W and the case where the power is set to 200 W. The pulse off of various process characteristics is compared. The first experiment of period dependency and the second experiment of upper DC voltage dependency comparing various process characteristics. In addition, the case where the power PL _L of the radio frequency HF in the pulse wave off period T _off is 0 W with high / low pulse wave modulation is the same as the pulse wave modulation applied with on / off.

As the etching conditions of the main fixed value common to the first experiment and the second experiment, the etching gas was C ₄ F ₆ / NF ₃ / Ar / O ₂ = 76/10/75 / 73sccm, the chamber pressure was 15mTorr, and the lower electrode temperature was 60 ℃, pulse duration T _on opening of 100μs, introduction of the LF RF pulse duration T _on the opening of the plasma power of 10000W, pulse duration T _on the opening of the plasma generating power was 1000W RF HF , The absolute value | V _dc | of the upper DC voltage V _dc in the pulse-on period T _on is 500V.

<Parameters and experimental results of the first experiment>

During the pulse off the first experimental dependence of comparing various process characteristics of the pulse wave closes the upper period T _off of the DC voltage V _dc of the absolute value | V _dc | during fixed 900V, the pulse off T _off (tune The frequency f _S and operating ratio D _{S of the} variable pulse wave MS are selected as parameters, and T _off = 25 μs (f _S = 8 kHz, D _S = 80%), T _off = 100 μs (f _S = 5 kHz, D _S = 50% ), T _off = 150μs (f _S = 4kHz, D _S = 40%), T _off = 233μs (f _S = 3kHz, D _S = 30%), T _off = 400μs (f _S = 2kHz, D _S = 20 %).

The results obtained in the first experiment are shown in graphs in FIGS. 10A to 10E. As shown in FIG. 10A, [1] The increase in the depth of the hole 140 (etching amount: d _2- d ₁ ), when the power PL _L of the radio frequency HF is either 0W or 200W, the pulse wave off period T _off In the range of 25 μs to 400 μs, all beams are in the range of about 700 to 750 nm, which is not so different. In this way, if high / low pulse wave modulation is used at PL _L = 200W, the same amount of etching or etching rate can be obtained as in the case of using on / off pulse wave modulation.

As shown in FIG. 10B, [2] Reducing the diameter CD, if the pulse-off period T _{off is} gradually increased from 25 μs to 400 μs, the power PL _L of the RF HF is 0W and stops at about 22.0 ~ In the range of 23.0, when the power PL _L of the frequency HF is 200 W, it is gradually reduced to approximately 18.0 nm stepwise from about 22.0 nm. In this way, if high / low pulse wave modulation is used at PL _L = 200W (especially if f _S is 3 kH or less, and T _off is 233 μs or more), compared with on / off pulse wave modulation In this case, the diameter reduction CD is greatly increased.

As shown in FIG. 10C, [3] The middle Ox bending CD, if the pulse-off period T _{off is} gradually increased from 25 μs to 400 μs, the power PL _L of the RF HF is 0 W and stops at about 36.0. In the range of ~ 37.0, when the power PL _L of the frequency HF is 200 W, it is significantly reduced from about 37.0 nm to about 34.0 nm (however, if T _off becomes 233 μs or more, it hardly decreases). In this way, if high / low pulse wave modulation is used at PL _L = 200W (especially if f _S is 3 kHz or less and T _off is 233 μs or more), compared with the on / off pulse wave modulation In this case, the middle Ox-bending CD also increased significantly.

As shown in FIG. 10D, if [4] the selection ratio, if the pulse-off period T _{off is} gradually increased from 25 μs to 233 μs, the power PL _L of the radio frequency HF is 0 W and either the case of 200 W It increases from about 2.5 to about 4.2 at a slightly same rate of change, and becomes saturated once T _off exceeds 233 μs. In this way, if the high / low pulse wave modulation is used at PL _L = 200W, the selection ratio is increased to the same extent as when the on / off pulse wave modulation is used.

As shown in FIG. 10E, if [5] the aspect ratio change amount, if the pulse-off period T _{off is} gradually increased from 25 μs to 400 μs, the power PL _L with respect to the RF HF stops at about 0 W. In the range of 80 to 85, when the power PL _L of the frequency HF is 200 W, it is greatly increased from about 80 to about 130 (however, once T _off exceeds 233 μs, it is saturated). In this way, if high / low pulse wave modulation is used at PL _L = 200W (especially if f _S is 3 kH or less, and T _off is 233 μs or more), compared with on / off pulse wave modulation In this case, the aspect ratio change rate has increased significantly.

<Parameters and experimental results of the second experiment>

In the second experiment comparing the upper DC voltage dependency of various process characteristics, the pulse-off period T _off (the frequency f _{S of the} pulse wave MS and the operating ratio D _S ) were fixed at T _off = 233 μs (f _S = 3kHz, D _S = 30%), the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | is used as a parameter, and | V _dc | = 500V, 900V, and 1200V are selected in three stages.

The results obtained in the second experiment are shown in graphs in FIGS. 11A to 11E. As shown in FIG. 11A, the [1] increment of the depth of the hole 140 (etching amount: d _2- d ₁ ), if the absolute value of the upper DC voltage V _dc during the pulse-off period T _off | V _dc | stage If the power is increased to 500V, 900V, 1200V, the power PL _L of the RF HF will decrease linearly from about 760nm to about 680nm, and the power PL _L of the RF HF will decrease gradually from about 700nm when the power is 200W. To about 680nm. In this way, in the case of using high / low pulse wave modulation with PL _L = 200W, even if the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | increases, the depth of the hole 140 is The increase (etching amount) does not necessarily increase, but instead tends to decrease. However, it does not necessarily deteriorate compared to the case where the on / off pulse wave modulation is used.

As shown in FIG. 11B, [2] Reducing the diameter CD, if the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | is gradually increased to 500 V, 900 V, and 1200 V, When the power PL _L of the radio frequency HF is 0W, it is gradually reduced to about 20.0nm from about 23.0nm, and when the power PL _L of the frequency HF is 200W, it is gradually reduced to about 10.0nm from the low level. 17.8nm. In this way, when the pulse wave modulation with high / low pulses is used with PL _L = 200W, the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | Diameter CD, and the diameter reduction CD is increased compared to the case of using on / off pulse wave modulation.

As shown in FIG. 11C, [3] The middle Ox bending CD, if the absolute value of the upper DC voltage V _dc in the pulse off period T _off | V _dc | is gradually increased to 500V, 900V, 1200V, the relative When the power PL _L of the radio frequency HF is 0W, it is gradually reduced from about 37.5nm to about 35.5nm. When the power PL _L of the frequency HF is 200W, it is gradually reduced at a lower level from about 35.2nm. To approximately 33.5 nm (however, if | V _dc | becomes 900V or more, it will hardly decrease). As described above, in the case where high / low pulse wave modulation is used with PL _L = 200W, the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | generally increases as the increase increases. Ox-bent CDs are more improved than those with on / off pulse wave modulation in the middle of Ox-bent CDs.

As shown in FIG. 11D, the [4] selection ratio changes the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | in the range of 500V to 1200V, and the power PL _L at RF HF is The 0W case and the 200W case are still within the range of about 4.1 ~ 4.5. In this way, if the high / low pulse wave modulation is used with PL _L = 200W, the selection ratio is improved to the same extent as when the on / off pulse wave modulation is used.

As shown in FIG. 11E, if [5] the aspect ratio change amount, if the absolute value of the upper DC voltage V _dc in the pulse-off period T _off | V _dc | is gradually increased to 500V, 900V, 1200V, then Relative to the case where the power PL _L of the radio frequency HF is 0W, if | V _dc | becomes 900V or more, it rises from about 80nm to about 92nm. The case where the power PL _L of the radio frequency HF is 200W is at a higher level from about 99nm It rises to about 132 nm (however, if | V _dc | becomes 900V or more, it will be saturated). In this way, if high / low pulse wave modulation is used at PL _L = 200W, the aspect ratio change rate is greatly increased compared to the case where the pulse wave modulation is turned on / off.

<Evaluation of Experiment>

As described above, it is judged that in the HARC (High Aspect Ratio Contact) process shown in FIG. 9, the RF HF for plasma generation is applied with high / low pulse wave modulation, compared with the application. The on / off pulse wave modulation has advantages in various process characteristics, in particular, it can ensure a high selection ratio and effectively suppress bending. Consider this.

In pulse wave modulation, if the pulse wave is switched from the pulse wave on period to the pulse wave off period in each cycle of the modulation pulse wave, the effect of ion introduction is weak, and the plasma reaction product is deposited on the mask. Therefore, the low-speed pulse wave / low operating ratio (the pulse wave off period is long) can be said to be a range suitable for improving the selection ratio of the mask to the material to be etched or the target film. However, the pulse wave off period has little effect on the etching. Therefore, if the pulse wave off period is increased to more than necessary, the time required for plasma processing becomes longer, which may cause a reduction in productivity.

In addition, if the depth-to-width ratio of a hole in a HARC ground is increased, the etching time will be longer. Therefore, when the on / off pulse wave modulation is used, even if the selection ratio with the mask is ensured, due to the long time to the side wall of the hole With the incidence of ions, bending becomes more likely to occur, so it is ultimately difficult to obtain a good processed shape.

Immediately after switching from the pulse wave on period to the pulse wave off period, in the processing space of the chamber, electrons, ions, and free radicals are reduced in different proportions. Compared with the case where the electron disappears in 10 μs and the ion disappears in a short time of about 100 μs, free radicals still exist after a time of about 1 ms. I think that the mask surface protective film is formed by the free radicals reacting with the surface layer of the mask during this closing period.

In the high / low pulse wave modulation, the RF HF used in the plasma generation during the pulse wave off period still excites the processing gas, generating ions and free radicals. In this case, compared with the radio frequency LF used for ion introduction, the acceleration energy given to the ions is small, so the proportion of influence on the etching is small. On the other hand, a lot of free radicals are generated, and in the case of the lower dual-frequency overlapping application method, because LF is turned off and HF's power is weak, ions can be introduced into the bottom of the hole by drawing in the radicals with a moderate RF bias. As a result, the deposition of the reaction product on the side wall of the hole is promoted, and a side wall protective film effective for suppressing bending can be formed.

In addition, as described above, when using high / low pulse wave modulation, it is known that the technique of increasing the absolute value of the upper DC voltage during the pulse wave off period is higher than the pulse wave on period in synchronization with the modulation pulse wave. The improvement of various process characteristics, especially the improvement of the diameter reduction, the improvement of the intermediate bending CD, and the improvement of the vertical shape are effective.

That is, I think that by increasing the absolute value of the upper DC voltage during the pulse-off period, any function can be operated (for example, by increasing the energy of the electrons implanted in the material to be etched and the mask). The effect of extending the side wall protective film to the bottom side in the hole, or the effect of suppressing the lowering of the mask shoulder (by this, reducing the incidence of ions from the tilt component that induces bending).

In any case, in the case of HARC process, when high / low pulse wave modulation is applied to the radio frequency used for plasma generation, the frequency of the modulated pulse wave should be above 1kHz (preferably 2kHz ~ 8kHz, more preferably 2kHz ~ 3kHz) ), The power PL _{L of the} RF generation HF for the plasma generation during the pulse wave off period should be set to a certain level of RF (for example, 100W or more, preferably 200W or more).

At this point, in the plasma etching apparatus of this embodiment, the matching device 40 (first matching device) of the plasma generation system operates as follows: With the impedance sensor 106A having the structure and function described above, the RF The impedance of the plasma load observable from the RF power source 36 is measured on the power supply line 43. The measured value of the load impedance during the pulse on period T _on and the measured value of the load impedance during the pulse off period T _on are obtained. The weighted average measurement value obtained by weighting the desired weighted average is matched with the output impedance of the RF power source 36. In this case, the value by adjusting the weight weighted average of the weight variable (K), and can control the pulse wave is turned on T _off in the period of the reflected wave power period T _on of the PR _H Close pulse wave reflected power PR _H of Balance, so the power PR _L of the reflected wave in the off period T _off of the pulse wave can be arbitrarily reduced, and the load power PL _{L can be} set to an arbitrary value which increases this part.

As an example, as shown in FIG. 12, the radio frequency power supply (the allowable limit value of the reflected wave power is 1200 W) of the radio frequency power supply 36 of the plasma generation system is used, as compared with that during the pulse wave on period T _on . In the case of a conventional matching method (a method of obtaining a slight perfect match in the pulse wave on period T _on ) with a reflection coefficient Γ of Γ = 0.0, by using a matching method of an implementation form of Γ = 0.2 and Γ = 0.3, And the settable range of the load power PL _L in the pulse off period T _off can be greatly expanded from about 230W (Γ = 0.0) to about 300W (Γ = 0.2) and further to about 350W (Γ = 0.3). This phenomenon means that the size of the radio frequency power supply 36 becomes possible from another viewpoint. The reflection coefficient Γ is given by Γ = (PR _H / PF _H ) ^1/2 .

[Example of countermeasures against discharge of upper electrode]

Generally speaking, in a hole etching process such as the HARC process, if the aspect ratio is increased, positive ions will easily accumulate at the bottom of the hole, and the straightness of ions in the hole will decrease, making it difficult to obtain a good etched shape. In this regard, the plasma etching apparatus shown in FIG. 1 includes a DC power supply unit 62. By applying a negative DC voltage to the upper electrode 46, electrons emitted from the upper electrode 46 to the plasma generation space PA are directed toward the base (lower portion). The semiconductor wafer (to-be-processed object) W on the electrode 16 is accelerated, and the electrons accelerated at a high speed are supplied deep inside the hole, and the positive ions accumulated at the bottom of the hole are electrically neutralized, so it can be avoided as described above. The problem that the straightness of ions in the pores is reduced.

However, by applying a DC voltage of a negative polarity to the upper electrode 46, a discharge (abnormal discharge) of a gas is generated in the upper electrode 46, particularly in the gas ejection hole 48a to the vent hole 50a, and the upper electrode 46 may be damaged. These abnormal discharges inside the upper electrode are likely to cause pulse wave modulation that is on / off for both the RF HF for plasma generation and the LF for ion introduction.

In this case, as shown in FIG. 13, during the _off period T _off , both the RF power source 38 for ion introduction and the RF power source 36 for plasma generation are turned off. On the other hand, the upper electrode 46 is removed from the DC power source section 62. A negative DC voltage V _{dc1 having a} large absolute value is applied. In this way, near the surface of the upper electrode 46, a high electric field region (hereinafter referred to as a "DC sheath") SH _DC that accelerates electrons (e) in the throwing direction and accelerates ions (+) in the attracting direction is generated. This DC protective layer SH _DC accelerated electrons (e) enter the semiconductor wafer W on the susceptor 16 and neutralize the positive charges accumulated at the bottom of the hole. At this time, the plasma disappears in the plasma generation space PA, so that a plasma protective layer (ion protective layer) SH _RF is hardly formed on the surface of the semiconductor wafer W. This state continues through the pulse wave off period T _off .

Then, if the pulse wave off period T _{off is} changed to the pulse wave on period T _on , the two RF power sources 36 and 38 are turned on at the same time, and the two radio frequency HF and LF are applied to the base 16. Thereby, a plasma for processing gas is manufactured in the plasma generation space PA, and a plasma protective layer SH _RF is formed in the chamber 10 to cover the surface of the semiconductor wafer W. In this case, the plasma protective layer SH _RF suddenly appears from a state that did not exist substantially before this, and rapidly grows toward the upper electrode 46 (the thickness of the protective layer increases). The growth rate of the plasma protective layer SH _RF mainly depends on the increase rate of the voltage (peak-to-peak value) V _pp of the radio frequency LF for ion introduction with a relatively low frequency to a saturation value.

On the other hand, in the upper electrode 46, the absolute value of the DC voltage applied from the DC power supply section 62 has changed from a larger value before | V _dc1 | to a smaller value | V _dc2 |, but still emits electrons (e ) And accelerate toward the semiconductor wafer W. However, it is different from the time T _off when the pulse wave is turned _off . In this case, the plasma protective layer SH _RF on the semiconductor wafer W grows rapidly in a direction that increases the thickness, that is, the electric field strength. The electron (e) is strongly rebounded by the growing plasma sheath SH _RF . Then, the electrons (e) rebounded by the plasma shroud SH _RF bounce toward the upper electrode 46 this time, and resist the electric field of the DC shroud SH _DC and enter the gas ejection hole 48a of the electrode plate 48 of the upper electrode 46. A situation in which a discharge is caused deep inside.

When an abnormal discharge occurs inside the upper electrode in this way, when the electrons (e) emitted from the upper electrode 46 are accelerated toward the semiconductor wafer W side, and the plasma protective layer SH _{RF on} the semiconductor wafer W side is accelerated When the rebounded electron (e) decelerates, the electric field of the DC protective layer SH _DC on the upper electrode 46 side acts on the electron (e) with the same force. Therefore, the frequency and velocity of the electrons entering the gas ejection holes 48a of the upper electrode 46 are almost independent of the size of the DC protective layer SH _DC , but rely on the plasma protective layer SH _RF to return the electrons (e) to the upper electrode 46 side. The strength of the bullet is the growth rate of the plasma sheath SH _RF .

In addition, even if the positive ions (+) generated in the upper part of the plasma generation space PA are introduced into the DC protective layer SH _DC and splashed against the surface of the upper electrode 46 (electrode plate 48), the upper electrode 46 is not caused. Internal abnormal discharge.

In the plasma etching apparatus of FIG. 1, as described above, the abnormal discharge inside the upper electrode 46 can be changed from on / off pulse modulation to high / Low pulse wave modulation, which is effectively avoided.

In this case, as shown in FIG. 14, during the _off period of the pulse wave T _off , the RF power source 36 remains on, and the RF HF for plasma generation is applied to the base 16 at a low level of power, so space is generated in the plasma. The PA plasma does not disappear but remains at a low density, and the surface of the semiconductor wafer W is covered with a thin plasma protective layer SH _RF . At this time, electrons (e) accelerated to a high speed by the large electric field of the DC protective layer SH _DC from the upper electrode 46 side receive a reverse electric field or force on the plasma protective layer SH _RF . However, since the plasma protective layer SH _{RF is} thin and the reverse electric field is weak, electrons (e) pass through the plasma protective layer SH _RF and are incident on the semiconductor wafer W. This state continues through the pulse wave off period T _off .

In addition, if the pulse wave off period T _{off is} changed to the pulse wave on period T _on , the radio frequency power source 38 is turned on and the radio frequency LF for ion introduction is applied to the base 16, and the radio frequency power source 36 changes the power of the radio frequency HF from Prior to this, the low level changed to a high level. As a result, the density of the plasma generated in the plasma generation space PA is rapidly increased, and the thickness of the plasma protective layer SH _RF covering the surface of the semiconductor wafer W is further increased. However, in this case, the plasma protective layer SH _{RF does} not appear suddenly and grows rapidly, but only increases the thickness from the existing state, so its growth rate is quite stable, and it will accelerate from the upper electrode 46 side. The force of the high-speed electron (e) rebound is not so great. Therefore, the electron (e) rebounded by the plasma protective layer SH _RF has a low initial velocity of splashing, so it cannot penetrate through the DC protective layer SH _DC and enter the gas ejection hole 48 a of the electrode plate 48 of the upper electrode 46. Therefore, no abnormal discharge occurs inside the upper electrode 46.

In addition, it was confirmed that, when an abnormal discharge occurs inside the upper electrode 46 during the pulse on period T _on , the peak-to-peak value V _{pp of} the radio frequency LF for ion introduction related to the growth rate and thickness of the plasma protective layer SH _RF is supplied at the radio frequency. The electric wire 45 is greatly changed. In the plasma etching apparatus of this embodiment, V _pp detectors 107A and 107B are provided in the matching devices 40 and 42 (FIG. 3). V _pp detector within the matcher 42 107B, as determined by ion on the RF feed line 45 for introducing LF RF peak-to-peak value V _pp, by the CPU within the main control unit 72 or the matching controller 104B, V _pp parse The measured information can be used to obtain monitoring information indicating whether an abnormal discharge has occurred inside the upper electrode 46 (FIG. 15, FIG. 16).

Here, the monitoring information in FIG. 15 is information obtained when an abnormal discharge occurs inside the upper electrode 46 (an example). As shown in the figure, it is known that the variation rate of V _pp frequently and significantly (a few% or more) jumps upward in the determination interval set in the monitoring period. Generally, as the frequency of occurrence of abnormal discharge increases, the rate of change in V _pp tends to increase. The rate of change of V _pp on the vertical axis of the graph shown is given by, for example, the following formula (2).

V _pp change rate = 100 × (V _pp-max -V _pp-ave ) / V _pp-ave ... (2)

V _pp-max is set to the average of the maximum value determined during the constant sampling interval T _S to the V _pp, V _pp-ave for the period T _S V _pp in the sample.

The monitoring information in FIG. 16 is information obtained when an abnormal discharge does not occur inside the upper electrode 46 (an example). By the determination interval V _pp , the fluctuation rate is stable at several% or less (the example in the figure is 1% or less). In addition, immediately after the start of the monitoring period and immediately before the end, it is the time point of plasma ignition and disappearance, and the V _pp change rate increases regardless of the presence or absence of abnormal discharge, so the self-judgment interval is excluded.

The inventors of this case performed experiments in which the gas pressure, the frequency f _{S of the} pulse wave modulation, and the operating ratio D _S were changed as parameters in the HARC process as described above, and the upper and middle portions of each pulse wave modulation were investigated. The presence or absence of abnormal discharge inside the electrode. In this experiment, the fluorocarbon-based gas was used for the etching gas in the same manner as in the above-mentioned embodiment, and the power of the RF HF for plasma generation during the pulse-on period T _on was 2000 kW, and the power of the RF LF for ion introduction was 14000 kW. , So that the power of the radio frequency HF in the pulse off period T _off is 100W. In addition, as the parameters, five kinds of gas pressures are selected: 10mTorr, 15mTorr, 20mTorr, 25mTorr, and 30mTorr. The frequency of pulse wave modulation f _{S is} selected from 3 kinds of 4kHz, 5kHz, and 10kHz. The operating ratio D _{S is} selected from 20%, 30%, 40%, 50%, 60% of 5 kinds.

17A and 17B, the results of this experiment are shown in a table format. In the table, ○ is the case where the V _pp change rate in the above monitoring information is less than 2% (allowable value), and it indicates the determination result of “no abnormal discharge”. × is the case where the variation rate of V _{pp in the} above monitoring information exceeds 2% (allowable value), which indicates the determination result of “abnormal discharge”.

FIG. 17A shows a case where RF HF for plasma generation and pulse wave modulation of on / off are applied to both sides. In this case, all the parameters (gas pressure, pulse wave modulation frequency f _S , operating ratio D _S ) are fully variable, and the result of “with abnormal discharge” (×) is widely distributed.

FIG. 17B shows a case where high / low pulse wave modulation is applied to the radio frequency HF for plasma generation, and on / off pulse wave modulation is applied to the radio frequency LF for ion introduction. In this case, the fully variable region covering (gas pressure, pulse wave modulation frequency f _S , operating ratio D _S ) is always "no abnormal discharge" (○).

In this way, by selecting a modulation mode of applying high / low pulse wave modulation to the radio frequency HF for plasma generation and applying on / off pulse wave modulation to the radio frequency LF for ion introduction, it is possible to effectively Avoid abnormal discharges inside the upper electrode 46. However, for this method, a preferable aspect must have a technology that can accurately and stably maintain the power (load power) of the RF HF for plasma generation to the optimal set value during the pulse-off period T _off . In this regard, as described above, by adjusting the value of the weighting coefficient K in the impedance sensor 106A of the matcher 40, it is preferable to use arbitrary control of the reflected wave power PR _H and the pulse wave in the pulse on period T _on during the off period T _off of the reflected power PR _L balance of the art, and the RF power source 36 to a pulse wave in the closed T _off is applied to the load power PL _L separate control of feedback techniques.

[Other embodiments or modifications]

As mentioned above, although the suitable embodiment of this invention was described, this invention is not limited to the said embodiment, Various deformation | transformation can be performed within the range of the technical thought.

In the present invention, when the first (plasma generation system) power modulation method, the second (ion introduction system) power modulation method, and the upper DC application method are combined, each mode can be arbitrarily selected. In addition, it can also be a form that does not apply pulse wave modulation at all to the power of radio frequency LF for ion introduction, and applies high / low pulse wave modulation to radio frequency HF for plasma generation, or vice versa. The radio frequency HF does not apply pulse wave modulation at all and applies high / low pulse wave modulation to the power of the radio frequency LF for ion introduction. Further, it may be a form using only one of the first power modulation method or the second power modulation method, or a form not using the upper DC application method.

In the above embodiment (FIG. 1), radio frequency HF for plasma generation is applied to the base (lower electrode) 16. However, a configuration in which radio frequency HF for plasma generation is applied to the upper electrode 46 may be used.

The present invention is not limited to a capacitive coupling type plasma etching device, and can also be applied to a capacitive coupling type plasma that performs any plasma treatment such as plasma CVD, plasma ALD, plasma oxidation, plasma nitridation, and sputtering. The processing device is further applicable to an inductively coupled plasma processing device in which a radio frequency electrode (antenna) is provided around the chamber. The object to be processed in the present invention is not limited to semiconductor wafers, and may be various substrates for flat panel displays, organic ELs, solar cells, photomasks, CD substrates, printed substrates, and the like.

Claims (11)

A plasma processing device for manufacturing a plasma generated by radio frequency discharge of a processing gas in a vacuum-evacuable processing container that stores the processed object in and out of the processed object, and processes the plasma under the plasma. The object to be processed in the container performs desired processing. The plasma processing device includes: a first radio frequency power source that outputs a first radio frequency; a first radio frequency power modulating unit that repeatedly repeats the first and the first at a certain working ratio. In the second period, the output of the first radio frequency power source is modulated with a modulation pulse of a certain frequency so that the power of the first radio frequency becomes high in the first period, and the power of the first radio frequency in the second period A lower level lower than the high level; a first radio frequency power supply line for transmitting the first radio frequency output from the first radio frequency power source to a first electrode disposed in or around the processing container; and A matching device for measuring the impedance of the load observable from the first radio frequency power supply on the first radio frequency power supply line, so that the measured value of the load impedance in the first period and the second period are by a desired weight; A weighted average value of measured values weighted average measurement of the load impedance obtained, the output impedance of the first RF power. The plasma processing device according to any one of the scope of application for a patent, which includes: a second radio frequency power source for outputting a second radio frequency; and a second radio frequency power supply line for outputting the first radio frequency from the second radio frequency power source. Two radio frequencies are transmitted to the first electrode; and a second radio frequency power modulation section modulates the output of the second radio frequency power source with the modulation pulse wave so that the power of the second radio frequency is turned on in the first period State or high level, the power of the second radio frequency in the second period becomes the off state or a lower level lower than the high level. For example, the plasma processing apparatus according to any one of the second patent application scope, wherein the second radio frequency has a frequency suitable for introducing ions from the plasma into the object to be processed. For example, the plasma processing device according to any one of claims 1 to 3, wherein the first radio frequency power source includes: a first RF power monitor that detects the first radio frequency on the first radio frequency power supply line. The power of a traveling wave propagating in a forward direction from the power source toward the first electrode, and the power of a reflected wave propagating in the reverse direction from the first electrode toward the first radio frequency power source, respectively, generate a travel representing the power of the traveling wave. Wave power detection signal, and reflected wave power detection signal indicating the power of the reflected wave; a first load power measurement unit, the traveling wave power detection signal and the reflected wave power obtained from the RF power monitor Detecting a signal to obtain a measured value of a load power supplied to a load including the plasma; and a first radio frequency output control section, for the second period of each period of the modulation pulse wave, for the traveling wave The feedback power is controlled so that the measured value of the load power obtained by the load power measurement unit is consistent with or similar to a predetermined load power setting value. For example, the plasma processing device according to any one of claims 1 to 3, wherein the first radio frequency power source includes: a first RF power monitor that detects the first radio frequency on the first radio frequency power supply line. The power of a traveling wave propagating in a forward direction from the power source toward the first electrode, and the power of a reflected wave propagating in the reverse direction from the first electrode toward the first radio frequency power source, respectively, generate a travel representing the power of the traveling wave. Wave power detection signal, and reflected wave power detection signal indicating the power of the reflected wave; a first load power measurement unit, the traveling wave power detection signal and the reflected wave power obtained from the RF power monitor Detecting a signal to obtain a measured value of a load power supplied to a load including the plasma; and a first RF output control section, during the first and second periods in each period of the modulation pulse, The power of the traveling wave is individually subjected to feedback control in the first period and the second period so that the measured value of the load power to be obtained by the load power measurement section and the first and second periods are Individually The given first and second load power settings are identical or similar, respectively. For example, the plasma processing apparatus according to any one of claims 1 to 3, wherein the first radio frequency includes a frequency suitable for the generation of the plasma. For example, the plasma processing apparatus according to any one of claims 1 to 3, wherein the object to be processed is placed on the first electrode. For example, the plasma processing device according to any one of claims 1 to 3, wherein the power of the second radio frequency in the second period is higher than the minimum power necessary to maintain the plasma generation state. . For example, the plasma processing device according to any one of claims 1 to 3, which includes a DC power supply section, which is synchronized with the modulation pulse and applies a negative polarity to the second electrode only during the second period. DC voltage. For example, the plasma processing device according to any of claims 1 to 3 includes a DC power supply unit, and applies a negative electrode to an electrode opposed to the object to be processed through a plasma generating space in the processing container. The synchronous DC voltage is synchronized with the modulation pulse, and the absolute value of the DC voltage is increased during the second period compared to the first period. For example, the plasma processing device according to any one of claims 1 to 3, wherein the frequency of the modulated pulse wave is 2 to 8 kHz, and the operating ratio is 20 to 80%.