JP2002141341A - Plasma etching method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable processing of a sensitive device without giving any damage and without using microloading with an increased yield and also achievement of etching of an oxygen contained layer provided on a layer not contained by oxygen with a high selectivity. SOLUTION: A processing chamber uses an antenna 30 which is driven by RF energy inductively coupled within the chamber. The antenna 30 generates a plasma of a high density and low energy to etch an oxygen-contained layer provided on a non-oxygen-contained layer within the chamber with a high selectivity. An auxiliary RF bias energy applied to a cathode 32 supporting a substrate controls a cathode sheath voltage, or controls an ON energy regardless of the density. A gaseous silicon or carbon source is passed through an etching gas containing fluorine. In addition to the etching, evaporating and a combination processing thereof, various magnetic and electrical processing techniques can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an RF plasma reactor, and more particularly to a radio frequency (RF) for electromagnetically coupling an associated RF electromagnetic wave to a plasma.
The present invention relates to a plasma reactor using an energy source and a silicon source in contact with the plasma, and a method to be processed in the reactor.

[0002]

2. Description of the Related Art The trend for high density integration is the very small dimensions of components that are electrically sensitive and are susceptible to damage due to energetic particle bombardment when exposed to small wafer sheath voltages on the order of about 200-300 volts. And brought equipment. Unfortunately, such voltages are lower than the voltages that circuit components experience during standard integrated circuit manufacturing processes. MOS capacitors and transistors manufactured for advanced devices have very thin (less than 200 Angstroms thick) gate oxide. These devices can be damaged by charging, which causes gate breakdown. This may occur during plasma processing due to non-uniformity of the plasma potential or plasma density or large RF displacement current when surface charge neutralization does not occur. Conductors such as intermediate connecting lines may also be damaged for the same reason. Very high aspect ratios, i.e. very deep and very narrow openings and trenches, have to be achieved in a conventional plasma etching chamber when they have to be formed or filled with various semiconductor materials. The etching method used is inappropriate. RF system CVD (chemical vapor deposition) reaction system and RIE (Reactive I
on Etching) First, consider a conventional semiconductor processing system such as a reaction system. These systems are about 10-500
High frequency energy from low frequencies of KHz to higher frequencies of about 13.56-40.68 MHz may be used.
Below about 1 MHz, ions and electrons can be accelerated by an oscillating electric field or any steady state electric field generated in the plasma. At such relatively low frequencies, the electrode sheath voltage developed on the wafer is typically above 1 kilovolt peak, which is much higher than the damage threshold of 200-300 volts. Above a few MHz, the electrons can still follow the changing electric field. If the amount of ions is larger than this, it cannot follow the changing electric field and is accelerated by the steady state electric field. In this frequency range (and practical gas pressures and power levels), the steady state sheath voltage ranges from a few hundred volts to over 1,000 volts. A preferred method for lowering the bias voltage of a magnetic field-enhanced RF system is to apply a magnetic field to the plasma. This B-field confines electrons in a region near the surface of the substrate, increasing ion flux density and ion flow, thus reducing voltage and ion energy requirements. By way of comparison, a typical non-magnetic RIE process for etching silicon dioxide includes RF energy applied at 13.56 MHz, an asymmetric system having a volume of 10-15 liters, a pressure of 50 mTorr and a pressure of about (8-10). A substrate (cathode) sheath voltage of about 800 volts is generated using a one to one anode area / substrate supported cathode area ratio. When a magnetic field of 60 Gauss is applied, the bias voltage becomes about 25-30%, 800
From volts down to about 500-600 volts, the etch rate increases by about 50%.

However, when a stationary B field is applied parallel to the substrate, an E × B ion / electron drift and a plasma density gradient associated therewith are generated throughout the substrate. This plasma gradient causes non-uniform etching, deposition and other film properties on the substrate. This inhomogeneity can be reduced by rotating the magnetic field around the substrate, which is usually the mechanical movement of a permanent magnet, or a pair of electromagnetic coils, or pairs of coils, driven 90 degrees out of phase. Can be reduced by momentarily controlling the current to step or otherwise move at a controlled rate in a magnetic field. However, the rotation of the magnetic field reduces the non-uniform gradient, but usually leaves some non-uniformity. Furthermore, it is difficult to set up the coils, especially two or more pairs of coils in the chamber to form a compact system, the Helmholtz coil configuration or individual magnetic enhancements surrounding a common load lock. This is particularly difficult when using a multi-chamber system consisting of a closed reactor chamber. A unique reactor system with the ability to instantaneously and selectively change the strength and direction of the magnetic field and designed for use in small multi-chamber reactor systems was co-assigned in Cheng et al., June 27, 1989. U.S. Patent No. 4,842,
No. 683. Microwave / ECR systems In microwave systems and microwave ECR (Electron Cyclotron Resonance) systems, typically above 800 MHz
The plasma is excited using microwave energy at a frequency of 2.45 GHz. Although this technique produces a high density plasma, the particle energy may be below the minimum reaction threshold energy for many processes, such as reactive ion etching of silicon dioxide. To compensate for this, energy enhancing low frequency power is coupled to the substrate support electrode and to the plasma through the substrate.
In this way, the potential for damage to the substrate is reduced compared to conventional systems.

However, etching or CVD
Microwave systems and microwave ECRs that operate at practical power levels for semiconductor substrate processing include large waveguides for power delivery, expensive tuners, directional couplers, circulators, and dummy dummy for operation. Requires load. In addition, to meet the ECR requirements of a microwave ECR system operating at a commercial frequency of 2.45 GHz, a magnetic field of 875 gauss is required, which requires larger electromagnets, power and cooling specifications. Microwave systems and microwave ECR systems are not easily scaled up or down. Hardware is available for 2.45GHz. This is because this frequency is used in microwave ovens. A 915 MHz system is available, but at a higher cost. Hardware for other frequencies is not readily or economically available. As a result, higher scale modes of operation are needed to expand 5-6 inch microwave systems to process larger semiconductor substrates. Scale-up at constant frequency due to operation in this higher mode requires very strict process control to prevent so-called mode flipping to higher or lower loads and the resulting process changes Becomes Alternatively, for a 5 to 6 inch microwave cavity, this expansion can be achieved by using a diverging magnetic field to spread the plasma flux over a larger area. According to this method, the active power density and therefore the plasma density are reduced.

In addition, ECR systems must be operated at very low pressures, on the order of 2-3 millitorr. This is because the density of the plasma generated in this system increases very quickly to about 2-3 millitorr or more. This requires a large amount of reaction gas to be supplied to the system and requires a large amount of exhaust system to remove these large amounts of gas. RF Transmission Line System As previously mentioned, inventor Collins et al.
U.S. Patent Application No. 559,947 entitled "VHF / UHF Reactor System", which was assigned at the same time as
The MAT file 151-1) is referred to here.
In this application, a high-frequency VHF / UH in which a part of the reaction chamber itself is configured as a transmission line structure for applying high-frequency plasma generation energy from a matching network to the chamber.
An F reactor system is disclosed. This unique integrated transmission line structure satisfies very short transmission line requirements between the matching network and the load, and allows for relatively high frequency specifications from 50 MHz to 800 MHz. This allows for an efficient and controllable application of RF plasma generation energy to the plasma electrode, producing commercially acceptable etch and deposition rates with relatively low ion energy and low sheath voltage. This relatively low voltage reduces the potential for damage to electrically sensitive small size semiconductor devices. This VHF / UHF system prevents various other problems in the prior art, such as the possibility of scaling and power constraints described above.

[0006]

SUMMARY OF THE INVENTION In one aspect, the present invention solves the problems of the prior art. However, the present invention provides a vacuum chamber having a source region, a source of a process gas in the chamber, and a plasma from the process gas. Means for electromagnetically coupling RF energy to the plasma region to generate the plasma, means for capacitively coupling RF energy to the substrate support electrode to control the plasma sheath voltage of the support electrode, and RF including a silicon source in contact with the plasma It is embodied in the configuration and operation of a plasma processing system. Preferably, the RF antenna means is adjacent to the plasma region,
Coupled to an RF energy source. The chamber of the present invention has an RF located in the processing area to enhance plasma processing.
A cathode, an anode limited by a champer wall,
And electrically floating, grounded, or RF
It has a third electrode that can be connected to a bias. The third electrode and / or the champer wall defining the plasma region may be a silicon source for augmenting processes such as, for example, oxide etching. With the apparatus of the present invention, plasma etching can be selected almost infinitely, and plasma deposition can eliminate voids.

[0007]

DETAILED DESCRIPTION OF THE INVENTION The present invention is an apparatus for plasma processing which allows for improved selectivity of the etching process and various etching rates, and allows for simultaneous deposition of the deposited layer, etching and planarization. And a method. In the chamber of the present invention, preferably 100KH
LF / VHF (low to very high frequency) RF power in the range of z to 100 MHz is used. More preferably, MF (medium frequency) RF power in the range of 300 KHz to 3 MHz is used. Preferably, the coupling means is a multi-turn cylindrical coil antenna with an unwinding electrical length of less than λ / 4. Here, λ is the wavelength of the high-frequency RF excitation energy applied to the coil antenna during the plasma operation. The present invention also includes a means connected to the antenna for tuning the antenna to resonance, and likewise an antenna for matching the input impedance of the plasma source to the output impedance of the means for supplying RF energy for the antenna. Including connected load means. The tuning means may be a variable capacitance electrically connected between one end of the antenna and RF ground. RF
Energy can be applied to selected locations on the coil antenna via taps.

[0008] The present invention includes a dielectric dome or cylinder forming a plasma source region. Preferably, a coil antenna surrounds the dome for inductively coupling high frequency electromagnetic energy into the chamber. The article or substrate to be fabricated can be located in the plasma source region or dome, within or near the antenna winding or the bottom winding, or preferably below the antenna. The present invention also provides for selective processing of diluent, passivation, and other gases into the chamber surrounding the gas inlet at the top of the dome, the first ring manifold at the base of the plasma source region of the dome, and the substrate support electrode. Means for supplying gas to a chamber comprising a second ring manifold for supplying gas to the chamber. Further, if the AC power and control system is typically the same as the plasma source coil power,
AC bias power at a frequency close thereto is coupled to the substrate support cathode, thereby providing control of cathode sheath voltage and ion energy independent of plasma density control provided by the plasma source RF power. The system provides a bias frequency selected to serve a number of purposes. First, the upper frequency limit is chosen to prevent "current-induced" damage, which can cause charge-up damage to sensitive devices if the frequency is too high. The lower the bias frequency, the higher the substrate sheath voltage per unit bias power (excluding substrate heating), and its contribution to plasma density is small, thus improving the independent control of ion density and ion energy. However, if the bias frequency is too low, the ions will follow the RF component of the substrate sheath electric field, thereby changing the ion energy. As a result, the peak / average energy ratio increases and the ion energy distribution broadens (2
peak). If the bias frequency is too low, charge up of the insulator will occur, disabling ion induced processing during part of the bias frequency period. The preferred frequency range that satisfies the above considerations corresponds to the preferred plasma source frequency range described above.

Further, the present invention provides a DC bias voltage with a periodic pulse between selected low and high values to provide a passivation coating on the first selected material on the substrate to be etched. Passivating coatings), for example, including control means for forming a polymer coating, providing a relatively low etch rate of the substrate material and a relatively high etch rate of the second material, eg, a layer of the second material on the substrate. And selective etching with selectivity. Also, the chamber is evacuated by a first vacuum pump means connected to the chamber body and a second vacuum pump means connected to the dome, thereby establishing a flow of neutral particles out of the dome. And the voltage at the substrate support electrode is sufficient to overcome this pressure difference so that the charged particles flow toward the chamber body. The present invention also includes a differently configured conductive Faraday shield interposed between the coil antenna or other coupling means and the reaction chamber to prevent coupling of the electric field component of the high frequency electromagnetic energy to the chamber. Also, high frequency energy radiation is concentrated in the chamber by a high frequency reflector arranged to surround the coil or other coupling means.

The magnetic enhancement is provided by a surrounding permanent magnet or electromagnet arrangement, which applies a controlled static magnetic field parallel to the axis of the antenna selected from a uniform diverging magnetic mirror arrangement and a plasma below the substrate. Control the position and movement of the. Also, a magnet may be mounted around the plasma source and / or the chamber to apply a multi-pole cusp magnetic field to the chamber near the substrate, thereby confining the plasma to the substrate area and substantially eliminating the substrate magnetic field. . Further, a magnetic shunt can be placed around the substrate and the substrate support electrode to redirect the magnetic field from the substrate support electrode. This system configuration allows the size to be scaled up and down while maintaining low mode operation by selecting the operating frequency. As a processing aspect, the present invention provides a vacuum chamber having a plasma source region and a processing region, supporting an article or substrate to be processed on an electrode in the chamber, and applying a plasma sheath voltage at a substrate support electrode. Provide a silicon ion source that electromagnetically couples RF energy to the chamber for control and capacitively couples RF energy to the chamber through the support electrode to control the sheath voltage at the support electrode and contacts the plasma The present invention is embodied in a plasma generation process.

The present invention also includes the automatic repetitive tuning of the antenna to resonance and the loading of its input impedance to the impedance of the RF energy source for the antenna. The plasma generation process provides a vacuum chamber having a plasma source region and a processing region and a wall, an electrode in the processing region and an electrode in the plasma source region, wherein the processing region electrode is a cathode, the wall is an anode, and the plasma source electrode is The electrical connection is selected from ground, floating and RF or bias to electrically connect the electrodes in the processing area, chamber walls and plasma electrodes, support the article to be processed on the substrate support electrode, An antenna that generates a plasma for processing one or more materials on an article using a cylindrical coil antenna connected to a means for supplying a processing gas to the antenna and applying RF energy to the antenna. To inductively couple RF energy into the plasma source region and to control sheath voltage at the support electrode. Capacitively coupled RF energy into the chamber via, and a processing method and a silicon source in the chamber. The silicon source is preferably a third electrode located at the plasma source in the chamber, which need not be the electrode,
It may be anywhere in the chamber in contact with the plasma.

The antenna power and the bias power supplied to the electrodes are controlled so as to selectively perform anisotropic, semi-anisotropic and isotropic etching. The processing chamber of the present invention is particularly useful for etching a layer of a second material, particularly an oxygen-containing layer such as an oxide on an oxygen-free substrate such as silicon. The use of a silicon ion source in the processing chamber results in improved selectivity and increased etch profile. In addition, the process of the present invention periodically cycles the bias voltage to a selected low value for forming an etch-suppressing layer on silicon, such as a polymer, and a high value for etching a second material at a high rate with respect to the substrate. And a driving method. In all cases, the apparatus of the present invention provides high etch rates and high selectivity, which requires creating high aspect ratio openings in the second layer toward the substrate. The process in the reactor of the present invention comprises sputter deposition of an oxide such as silicon oxide or silicate, and first applying relatively low RF power to the support electrode to deposit the oxide, and secondly depositing the silicon oxide. It involves two steps: vapor deposition, sputter faceting the silicon oxide, and applying a relatively high level of RF power to the support electrode to planarize the layer.

[0013] A specific method comprises etching an contact hole in an oxide formed on an oxygen-free substrate and etching through a hole in an oxide formed on aluminum or another substrate. Etching of polysilicon; etching of polysilicon conductors such as gates; removal of photoresist; anisotropic etching of single crystal silicon; anisotropic photoresist etching; nitrogen and oxidation Low pressure plasma deposition of nitrogen; high pressure isotropic conformal deposition of oxides, oxynitrides and nitrides; etching of compounds such as aluminum, tungsten and titanium and compounds and alloys such as tungsten silicide; and various deposition rates; Localized with flattening
Includes, but is not limited to, deposition of various materials, including blanket sputter facet deposition. 1. Overview FIGS. 1-3 illustrate an inductive plasma source device, a magnetically enhanced plasma source device, a capacitively coupled bias device, and a plasma reactor 10 using other aspects of the present invention for processing a semiconductor substrate 5. It is an outline sectional view. The three figures illustrate preferred features and other features of the system.
Three drawings are used due to space limitations. This illustrated chamber is a variation of that shown in the co-pending, part-continuation application with an integrated transmission line structure. Important features of the present invention can be widely applied to plasma reactors. In addition, those of ordinary skill in the art will appreciate that the various features of the present invention that enhance the performance of the reactor can be used individually and that the system can be selectively omitted from the following description. You can also. For example, processing conditions provided by a bias source capacitively coupled to an inductive plasma source device often eliminate the need for magnetic enhancement.

The illustrated system 10 includes a side wall 12, a top wall 1
3. A vacuum chamber housing 11 made of anodized aluminum or other suitable material having a bottom wall 14.
including. Anodized aluminum is preferred because it suppresses arcing and sputtering. However, other materials, such as bare aluminum with or without a polymer, quartz, or ceramic liner, suitable for this process may be used. Top wall 13 is wall 12
-12, the lower chamber substrate processing unit 16 formed during
It has a central opening 15 between B and the upper chamber plasma source 16A formed by the dome. The dome is preferably configured as an inverted single-wall or double-wall cup formed by a dielectric such as quartz or some dielectric material such as aluminum or alpha alumina (sapphire). it can. In the embodiment shown in FIG. 1, the dome 17 is a cylindrical wall 17 made of a dielectric material such as quartz.
W and normal aluminum or anodized aluminum cover or top wall 17T. For purposes such as highly selective oxide etching, silicon or silicon containing top wall means and silicon covered dome sidewalls are preferred.

As shown in FIG. 1, the reduced pressure exhaust inside the chamber housing 11 (chamber 16)
Is controlled by a throttle valve 18 (regulates pressure independently of flow rate) in a vacuum line 19 connected to a vacuum pump system 21 consisting of one or more vacuum pumps connected to the vacuum pump. As described in Section 11, chamber components, including chamber walls and domes, can be heated and / or cooled for processing performance. For example, the dome can be heated or cooled by a liquid or gas heat transfer medium, or the dome can be heated directly using a heating element. As shown in Section 2 and illustrated in FIG. 2, the process gas, purge gas, diluent, etc., are provided at the base of the plasma source (dome), the top plate 17T of the plasma source, and three manifolds located around the periphery of the substrate, respectively. It can be supplied to the chamber by injection sources G1, G2 and G3. These gases are supplied, for example, from one or more sources of pressurized gas to the chamber 11 via a computer controlled flow controller (not shown). In the main intake manifold G1, gas enters the internal vacuum processing chamber 16 as shown at 22 via a quartz ring gas manifold 51 mounted inside or integral with the top wall 13. The manifold 23 preferably supplies an etching gas or a deposition gas at a slightly upward angle to the chamber portions 16B and 16A to generate an etching or deposition plasma after the application of RF energy. The top manifold device G2 in the top plate 17T of the dome 17 supplies a reactive gas or other gas to the chamber 1
6 can be used to take in. Further, a manifold device G3 for supplying a reactive gas and other gases around the substrate may be provided.

RF energy is supplied to the dome by a plate source consisting of at least one turn antenna 30 or coils fed by an RF supply and matching network 31. The antenna 30 preferably has a multi-turn cylindrical configuration. The coil 30 defines a minimum conductor electrical length for a given frequency and plasma source (coil) diameter, and preferably at the operating frequency, a quarter wavelength (<
λ / 4) or less. The antenna 30 itself is not a resonator, but is tuned to resonance as described in Section 5 for effective inductive coupling with the plasma source according to Faraday's law of inductive coupling. Preferably, the gas flow from the chamber plasma source 16A flows downward toward the substrate 5 and is then drawn radially outward from the substrate. For this purpose, around the cathode transmission line structure 32, between the chamber wall 12 on one side and the outer transmission line conductor 320 on the other side and between the bottom chamber bottom wall 14 and the top conductive pumping screen 29. Annular vacuum manifold 33 can be formed. Manifold screen 29 is interposed between vacuum manifold 33 and substrate process chamber 16B and provides a conductive path between chamber wall 12 and outer conductor 320 of transmission line structure 32.
The vacuum manifold 33 forms an annular pumping channel for uniformly drawing out the exhaust gas from the periphery of the substrate 5 in the radial direction. Vacuum manifold 33 is in communication with exhaust system line 19. The gas flow is directed radially inward toward the substrate 5 along the passage 22 from the manifold G1 to the dome / plasma source and / or along the passage 24 from the manifold G2 and / or the passage 26 from the manifold G3. Flows to The overall gas flow is
Along the upper chamber plasma source 16A and the substrate 5
Next, the exhaust manifold 33 is passed from the substrate along the passage 3 through the screen 29 and from the exhaust manifold 33 along the passage 37 to the exhaust system 21. It should be noted that the conductive manifold screen 29 and the cathode transmission line structure are optional. Usually, the wavelength is very long on the low frequency side of interest, so that no transmission line structure is required.

This is in contrast to the configuration of a conventional RF system, where the RF power is supplied by two electrodes, typically a wafer support electrode 32C whose upper surface supports the substrate 5 and the side walls 12, top wall 13 and And / or between a second electrode which is a manifold 23. In particular, the antenna 30 is located outside and adjacent to the dome 17 and the plasma chamber 16A to couple RF electromagnetic (em) energy to the plasma source chamber 16A to induce an electric field in the process gas. . From the Faraday's law of inductive coupling, the em energy changes B
The (magnetic) component energizes the process gas to form a chamber 16.
A plasma having a characteristic of relatively high density and low energy ions is formed therein (reference numeral 16 collectively refers to the chambers 16A, 16B and the plasma). This plasma is generated in the dome 17 concentrated in a small volume formed in the coil antenna 30.
Active species, including ions, electrons, free radicals, and excited neutrals, migrate downstream toward the substrate by diffusion and bulk flow by the gas flows described herein. Further, as described in Section 7, an appropriate magnetic field can be used to extract ions and electrons toward the substrate as described below. This is optional, but the bias energy input device 41 of FIG. 1 consisting of a plasma source 42 and a bias matching network 43
Preferably, coupled to 2C, the plasma sheath voltage of the substrate is selectively increased, thereby selectively increasing the ion energy of the substrate.

The reflector 44, which is basically an open-ended box, surrounds the antenna at its top and side, but does not surround the bottom of the antenna 30. This reflector 44 prevents the emission of RF energy into free space, thereby concentrating the emission and dissipation of power in the plasma to increase efficiency. As described in detail in Section 7, the Faraday shield 45 of FIG. 3 can be located inside, above, and below the antenna 30 to allow a magnetic field to couple to the plasma but to disable direct electric field coupling. ing. Direct electric field coupling may induce tilt or non-uniformity in the plasma, or may accelerate charged particles to high energies.
As described in Section 8, one or more electromagnets 47-47 of FIG. 2 may optionally be used to increase the plasma density on the substrate 5, transport ions to the substrate, or improve plasma uniformity. Alternatively, a permanent magnet can be mounted close to the chamber enclosure 11. As described in detail in Section 4, the present invention uses the magnetic force component of inductively coupled electromagnetic energy, typically at frequencies much lower than microwave or microwave ECR frequencies, and may be potentially damaging. High power RF
A circular electric field is induced in the vacuum chamber to generate a plasma having the characteristics of high density and relatively low energy without coupling energy to the substrate 5. In the preferred downstream plasma source configuration shown, the RF energy is completely absorbed away from the substrate at high plasma densities, such that the waves do not propagate to the substrate and thus minimize the potential for damage. Optionally and optionally, RF bias energy is applied to substrate support electrode 32C to increase the substrate sheath voltage, and thus the ion energy, as needed.

Chamber 16 can process substrates, including semiconductor wafers, by vapor deposition and / or etching using a total chamber pressure of about 0.1 mTorr to about 50 mTorr, and typically 0.1 mTorr to 200 mTorr for etching. . This chamber could operate at pressures below 5 millitorr and in fact operated normally at 2 millitorr. However, high pressures are preferred for certain types of processing in that pumping speed and flow rate are increased. For example, a pressure range of about 5 mTorr to about 50 mTorr is suitable for oxide etching. At such relatively high pressures, the distance between the plasma source and the substrate must be reduced. The chamber of the present invention comprises a substrate 5 and an antenna 30.
Between the bottom turns and about 5 cm / 2 in. And a very suitable small spacing d performed well without charge-up damage to sensitive devices. Therefore, the benefits of such a very small spacing,
That is, an improvement in etching rate and selectivity, a reduction in bias voltage and ion energy conditions for a certain etching rate, and an improvement in energy uniformity on the substrate are achieved. For example, the distance d between the substrate 5 and the source antenna 30 is 10 cm / in. Reducing from 5 cm / 2 in (which is itself a small spacing) halved the required voltage and increased the uniformity from about 2.5% to about 1%. 2. Gas processing system As mentioned above, this chamber is injected with reactive gas, purge gas, etc. into different places to improve the processing according to the conditions of each processing (etching, vapor deposition etc.) and the material used for the processing A plurality of gas injection sources G1, G2, and G3 (FIG. 2) for performing the operation are incorporated. First, the chamber has a standard radial gas distribution system G1 around the base / bottom of the plasma source region 16A. In a preferred configuration, the G1 injection system comprises a quartz gas distribution ring 51 at the bottom of the plasma source and a peripheral annular manifold 52 forming a distribution channel supplying gas to the ring. This ring has an inward facing radial hole 53-53
And preferably has a stepped sintered ceramic porous gas diffusion plug 54-54 inserted into said hole to prevent hollow cathode discharge.

A second gas injector G2 is grounded, made of a material such as anodized aluminum having a central intake hole 56 filled with a porous ceramic diffusion disk 57.
Or consist of a floating or biased dome top plate 17T. The third gas injection source G3 is supplied from a ring-shaped intake manifold 58 attached to the periphery of the substrate 5 (or a gas inlet incorporated in a clamp ring (not shown) used to hold the substrate on the support base). Become. When etching openings through an oxygen-containing layer, such as silicon dioxide, over the surface of a non-oxygen-containing layer, including single crystal or polysilicon, silicon dioxide is preferably etched at a much higher rate than polysilicon or other substrates. An etching gas containing fluoride is used as an etchant. However, since fluoride etchants usually etch materials such as silicon dioxide and polysilicon, for example, at the same rate, the etch has little selectivity for oxides rather than silicon. However, during reactive ion etching, a protective layer of polymer forms on the sidewalls and bottom of the growing opening. Such polymers are formed from carbon and fluorine, and generally contain about 30% carbon and about 60% fluorine.

The polymer is separated in the presence of a fluorine atom. All you need is about 50% more carbon and about 40%
The following is to form a polymer having a high carbon content including fluorine. If a scavenger for fluoride ions is placed in the reactor, there will be very few free fluoride ions in the plasma and no CF bonds will form in the polymer membrane. Was found. Success in conventional reactors has been obtained by adding a scavenger for fluoride ions to the plasma itself, such as silane or organosilane, to provide free silicon in the plasma. However, because many different plasma ions are formed, the process is inherently "dirty", the properties of the etchant and polymer are different and non-uniform, and the process is quite reproducible. Absent. Thus, the present apparatus, which provides a "clean" silicon source to remove free fluoride ions, is superior. According to the present invention, such a fluoride ion scavenger is a reactant for fluoride. And it is preferably a silicon source in or near the plasma. The fluoride scavenger is pure silicon such as single crystal silicon, polysilicon or silicon carbide, etc.
Alternatively, it is preferable that the third electrode is made of silicon or a material containing silicon. Graphite can also be used to remove fluoride ions. For example, the third electrode is made of graphite.

During plasma etching in the chamber of the present invention, oxygen in the silicon oxide film readily etches the polymer formed on the sidewalls and bottom of the growing trench. However, when the depth of the trench reaches polysilicon and other oxygen-free substrates, no oxygen is present,
Also, the polymer remains on the polysilicon surface and protects the substrate from further etching. Preferred etchants here a fluorocarbon such as CF _4, C ₂ F ₆ and C ₃ F _8, they generate only the fluoride ions and carbon ions. Other known fluoride, for example, CHF ₃ is not preferred. Because they generate hydrogen ions in addition to carbon and fluorine ions. According to the invention, a very high selectivity between the oxygen-containing layer and the non-oxygen-containing substrate or lower layer is obtained almost infinitely. This is because carbon-rich polymers do not decompose on silicon surfaces or oxide-free substrates or oxygen-free layers. These polymers are sensitive to oxygen, and in the absence of oxygen, for example, when the etchant reaches an oxygen-free substrate, polymer separation decreases, and when coupled to the plasma with reduced fluorine ions, the layers passivate. A relatively inert polymer is formed, protecting the underlying layer.

The silicon source is preferably located near where the plasma is generated, so that the silicon can remove fluoride ions. Few fluoride ions react with the surface of the substrate to be treated. For example, a silicon mesh is suspended in the plasma area, or silicon is placed near the reactor wall or top as part of the RF power supply. The silicon source can also be suspended near the surface of the substrate, but polymers with high fluoride content may result. The silicon source can also be located outside the plasma region of the reactor, and in each case heated to a temperature that forms silicon ions to remove fluoride ions, for example, at least about 150 ° C. or higher. .
In that case, means for adjusting the temperature of the silicon source must be provided in the reaction chamber. There are other advantages when the reactor of the present invention is used as a deposition chamber for depositing various films. For example, after a trench or opening is created in a substrate, the trench or opening is filled with another material to form continuity or isolation between devices in the substrate. For example, the openings in the substrate are filled with silicon oxide using silane and oxygen. This requires careful control of the deposition rate to avoid creating voids at the bottom corner of the trench, or at the center of the deposit. This latter phenomenon is well known and addressed in ECR processing. To avoid this problem, it is known to sputter the top of the trench with argon before vapor deposition, ie to facet the top of the trench or opening. Thereby, the top of the trench can be opened without closing around the gap. However, ECR processing is disadvantageous because very low pressures, on the order of 2-3 mTorr, are used to maintain high ion density. The ion density in ECR is very dependent on pressure. The ion density is maximized at about 1-2 mTorr and drops sharply at 5-10 mTorr pressure. Therefore, a large amount of gas
And a large vacuum pump is needed to evacuate excess gas.

The apparatus of the present invention for forming an inductive plasma maintains a high ion density up to a pressure of about 30 mTorr. Thus, very high pressures, about 15-30 milliTorr, are maintained in the inductively coupled plasma of the present invention, and concomitantly, little gas needs to be supplied to the reactor. And 1-2 necessary for vapor deposition in ECR
A small vacuum pump is required to remove by-products and excess gas other than millitorr pressure operation. The device of the present invention provides trenches that meet the results in an equivalent, but simple and inexpensive manner to those obtained in ECR. Due to the low temperature operation and ability to regulate the ion density of the plasma generated at low pressure, a similar method can be achieved without the need for large volumes of gas or large vacuum pumps in the apparatus of the present invention. Thus, the introduction of appropriate amounts of silane, oxygen, and argon, for example, into the reactor, sputters the top, sidewalls of the trench and forms a plasma that fills the trench with silicon oxide. The creation of voids in the silicon oxide deposition is avoided, and the formation of the top of the trench and the deposition or filling in the trench can be done in a single processing step.

As mentioned above, various types of gases selected from etchants and deposition species, passivation species, diluent species, etc., may be used to meet the specific etching and deposition processes and material requirements of G1. -Supplied to the chamber via one or more sources of G3. For example, the inductive plasma source antenna 30 provides a very dense plasma and is very effective at separating gases in the dome plasma source region 16A of the chamber. As a result, when polymer-forming species are provided to the dome via G1 or G2, highly isolated species may coat the interior of the dome at the expense of polysilicon coverage and / or Since it is separated, it does not adhere to the surface of the polysilicon to be coated for protection. One solution to the etching species such as C ₂ F ₆ or CF ₄ G1 or G2, or G1
And G3 through the plasma source region to provide silicon to remove fluoride ions to form preferentially a carbon-rich polymer on the substrate. Due to the high separation of gases in the plasma source region, fluorine containing gases (fluorine may be present in the fluorine containing gas with carbon) typically produce free fluoride ions etching silicon and silicon oxide. . Thus, the etch selectivity for the oxide is reduced. In addition to providing a source of silicon in the plasma,
When high selectivity is required, additional gases, including silicon, are inserted to further remove free fluoride ions and to reduce etching of oxygen-free substrates. The etching gas and the additional gas containing silicon are G1 and G
2 can be introduced separately or G1
And / or mixed and introduced via G2. Suitable fluorine-consuming silicon-containing additive gases are silane (SiH ₄ ), TEOS, diethyl lan and silicon tetrafluoride (S
iF ₄ ).

The fluorine consuming silicon source gas and the polymer additive gas can be used together in the same process to enhance etch selectivity. 3. Differential Pumping FIG. 2 shows an alternative vacuum pumping configuration. In addition to the vacuum pumping system 21 connected at or near the bottom of the chamber, a vacuum pump 39 is connected via line 38 to the plasma source region 16A in the dome 17. The flow rates of the pumping systems 39 and 21 are selected such that they generate a pressure difference ΔP _{p in a} direction perpendicular to the plasma source region 16A. This pressure difference ΔP _p (1) prevents the movement of uncharged particles from the plasma source 16A to the substrate 5, and (2) is smaller than the force F _b applied to charged particles such as electrons and ions by the bias voltage. Due to ΔP _p , uncharged particles such as radicals do not arrive at the substrate 5, but rather primarily flow out of the top vacuum connection 38.
Since F _DC > ΔP _p , charged electrons and charged ions mainly flow to the processing region. Obviously, this method is effective when it is desired to selectively place radicals, not ions, outside the substrate process region. This situation may be, for example, (1) using the chemistry of the polymer forming gas, but where the polymer is formed in the plasma source region and adheres to the sidewalls of the chamber and / or does not adhere well to the desired substrate surface, or (2) This occurs when fluorine groups are formed in the plasma source region. 4. RF Power, Top and Bias Source 1) Top or Antenna Source In FIG. 1, preferably the operating frequency of the RF power source 31 of the top plasma source 16A generates a dense plasma to minimize damage to sensitive equipment. And to provide efficient inductive coupling of RF power to the plasma. That is, frequencies above this operating range are limited to minimize "current-induced" damage. The lower limit of the operating frequency is chosen to increase the efficiency of the RF power coupling to the plasma. Suggested limitations are described above. 2) The AC power supply 42 of the lower or bias source substrate-supporting cathode 32C inductively couples RF power into the plasma to thereby control cathode sheath voltage and ion energy, etc., which are controlled independently of plasma density control performed by high frequency power. Controls various elements, including: The bias frequency is selected to achieve many goals. First, the upper frequency limit is selected to prevent current induced charge-up damage to sensitive devices. Lower frequencies are selected to eliminate, in part, voltage-induced damage. In addition, when the frequency bias is low, the substrate sheath voltage (excluding heating) per unit bias voltage of the substrate is high, and the contribution to the plasma density is small, so that the independent control of ion density and ion energy is improved. However, if the bias frequency is too low, the ions will follow the RF component of the substrate sheath electric field, thereby modulating the ion energy. As a result, the peak / average energy ratio increases and the ion energy distribution (between peaks) widens. If the bias frequency is too low, insulation charge-up will occur, disabling ion-induced processing as part of the bias frequency control.

Applicants have discovered that the above considerations are satisfied by using a bias frequency range corresponding to the plasma source frequency range described above. 3) Connection operation of upper antenna source and bias source A preferred feature of the present invention is to automatically change the lower or bias power supplied by power supply 42 to maintain a constant cathode sheath voltage. At low pressures (<500 mTorr) in highly asymmetric systems, the DC bias measured at cathode 32C is an approximation of the cathode sheath voltage. The lower power can be changed automatically to maintain a constant DC bias. The effect of lower or bias power on plasma density and ion flow density is very small. Top or antenna power has a very large effect on plasma density and current density, but very little on cathode sheath voltage. Therefore, it is desirable to use the upper power to define the plasma density and the ion flow density, and to use the lower power to define the cathode sheath voltage. Nevertheless, since the high frequency of the power supply 31 driving the antenna 30 is much lower than that used for microwave or microwave ECR applications, an optional smaller magnet operated at lower DC current by a cheaper power supply is needed. It can also be used. In this case, the associated heat load is also reduced. Further, as is apparent from the above description, a coaxial cable such as 31C can be used instead of the waveguide. Furthermore, plasma non-uniformities caused by ExB electron drift in other magnetically enhanced or magnetically assisted systems are not present here. this is,
Applied magnetic field (RF ignited via antenna 30)
Both the magnetic component of the field and any static magnetic field applied by the magnet 81) are substantially parallel to the electric field of the cathode. Therefore, there is no E × B drift in this system.

A B field is generated in the plasma source (upper chamber 16A) using a magnetic shunt formed of a material having a high magnetic permeability, and is not generated on the substrate. Optionally, a permanent magnet or electromagnet may be provided in the lower chamber 16.
A multi-cusp magnetic mirror can be created in the plasma source and / or chamber wall in a multi-pole arrangement of alternating pole configurations, typically NSNS... NS, around B.
The magnet can be a vertical bar magnet or preferably, for example, a horizontal ring magnet. Such magnets can be used to reduce electron loss to the wall, thereby improving plasma density and plasma uniformity without exposing the substrate to a magnetic field. 4) Coupling and Synchronization of RF Power Sources As mentioned above, the preferred frequency of operation of the upper or antenna RF power source and the preferred frequency of operation of the lower or bias RF power source are conveniently in the same range. A configuration that can be selected here is to combine these two RF power supplies into one power supply instead of using them separately. More generally speaking, three RF
Signal (including RF bias to third or upper electrode)
Can be supplied from a single power supply, or one power supply can be used for the antenna and the lower bias, the second power supply can be used for the third electrode, or three separate power supplies can be used. If separate power supplies are used, it must be taken into account that the frequencies of the separate RF signals must be equal or, if they must be, locked to some desired phase relationship. That is. Preliminary studies have shown that the answers to these questions depend primarily on the operating frequency chosen. A single RF power source is a logical choice if one frequency can be selected for two or three RF power sources, and if that frequency cannot be changed for another process in which the system is used. Will be.

However, the above sub-paragraphs 1-3
Based on the considerations discussed above, separate RF power sources will be required if different frequencies are required for these plasma sources, or if the frequencies must be changed for use in different processes. If there is a separate power source and the same frequency is selected, phase synchronization is probably not an issue, except at relatively low frequencies, such as below a few hundred KHz. For example, the power supply may maintain the phase angle between the RF voltage input to the antenna and the RF voltage input to the lower or substrate electrode at a constant value selected to optimize the repeatability of the process. At higher frequencies, operation appears to be independent of phase or frequency synchronization. 5. Antenna tuning and load 1) Tuning Normally, the antenna 30 is either (1) changed by changing the frequency of the oscillator 31 to resonate with the antenna, or (2) another antenna connected to the antenna to tune to resonance. Are tuned to resonance by the resonant element of For example, the tuning element can be a variable inductance-to-ground or a variable capacitance-to-ground.

It should be noted that inductive tuning and capacitive tuning lower the resonant frequency. Therefore, it is desirable to configure the system to have the desired highest resonance frequency to account for the reduction in resonance frequency when using capacitance or inductance tuning variables. Automatic tuning is preferred and can be performed by driving a tuning / load variable with an impedance phase / amplitude detector. See FIGS. 6 and 9. Also, a reflected power bridge or a VSWR bridge can be used to drive both tuning and load variables, but requires repetition. 2) The impedance of the plasma source antenna 30 can be matched to the impedance of the RF generator 31 and the connecting coaxial cable 31C using the loading conductive, capacitive or inductive load means L. For example, a tap or wiper can be in ohmic contact near 50 or 300 ohms or other generator output impedance location on the antenna. Also, a variable inductance or variable capacitance can be connected to the generator output impedance point 50 on the antenna. 3) Tuning circuit and load circuit In FIGS. 4 and 9, preferably, the tuning means T which is integrated with the plasma source antenna 30 to synchronize the plasma source with resonance
Is provided. Further, the integrated load means L changes the input impedance of the plasma source antenna 30 to the associated oscillator 31.
(Or the transmission line 31C). In FIG. 4, in one aspect, tuning means T is a variable capacitance electrically connected between one end of antenna 30 and RF ground.

As shown in FIG. 5, in another aspect, the load means L can be a variable capacitance electrically connected between one end of the antenna and RF ground. Also, this load means can be a variable position tap 60 that applies RF input power to the antenna. Please refer to FIG. In the preferred combination shown in FIG. 7, the tuning means T is a variable capacitance electrically connected between one end of the antenna 30 and RF ground, and the load means L is electrically connected between the other end of the antenna and RF ground. Is another variable capacitance connected to. In this configuration, the RF input power can be applied to the antenna via a tap, ie, along the antenna or via a tap provided at either end thereof. Please refer to FIG.
Further, the RF power input connection section 66 can be arranged at the connection section between the load variable capacitance L and the end of the antenna 30 as shown in FIG. 6. Source / Bias Process Control Also, the present invention provides a high silicon dioxide etch rate using a sufficiently high bias voltage, and by periodically pulsing the bias voltage to a low value, the etch rate of materials such as silicon dioxide is improved. And the discovery that the etch selectivity of silicon dioxide is increased compared to materials such as silicon. 1) Pulse / Modulation Bias-etch rate and selectivity
In FIG. 10, the etching rate of a material such as silicon dioxide SiO ₂ usually increases as the bias voltage increases. Therefore, increasing the bias voltage increases the oxide etching rate. Unfortunately, however, the etch rate of related materials in integrated circuit structures, such as silicon / polysilicon, also increases with bias voltage. Thus, with a bias voltage large enough to provide a very high silicon dioxide etch rate, the silicon etch rate will be too high (albeit somewhat lower than the oxide etch rate) and the selectivity will be reduced. . When etching silicon dioxide, a combination of a high oxide etch rate characteristic of a high DC bias voltage V _h and a relatively low silicon etch rate characteristic of a low DC bias voltage V ₁ , and thus a high oxide selectivity It is very clear that gaining degree is very desirable.

Here, the DC bias voltage waveform 70 shown in FIG.
Referring to FIG._h And V₁ Combining the characteristics of
The seemingly contradictory objective shown in the previous paragraph is actually high
Baseline DC bias voltage V_h Using this voltage
Low value V₁ To periodically pulse or modulate
Therefore, polymer formation etching process (materials such as silicon
Process to form an etch-inhibiting polymer on the surface)
Achieved. V₁ Silicon etching and silicon deposition
Between the intersection / voltage 68 (FIG. 10) and the oxide
The intersection / voltage is 69 or more. As a result, the protective polymer
Is deposited on silicon and the high-speed etching voltage V_h Return to
While suppressing the etching,_h Oxide edge
Deposition that adds significant suppression to the etching
Does not occur, or occurs insufficiently. Preferably, V
₁ Is characterized by deposition on polysilicon, but at least
Slight etching of the oxide. One embodiment of the present invention
In the example, the parameter V_h (High DC bias voltage
Pressure), V₁ (Low DC bias voltage), P_w (Low voltage V
₁ Pulse width), and P_rp(Low voltage pulse and high voltage pulse
Pulse repetition rate or combined width)
-400V, -225V, about 0.1 second, and about 1 respectively
Seconds. 2)2 frequency bias An alternative method is illustrated by the DC bias voltage waveform 71 in FIG.
Shown. Relatively low frequency voltage fluctuation is the basic bias voltage
Superimposed on frequency. For example, low frequency T _Two <25
KHz (preferably 5-10 KHz) based on high frequency T₁ 
<2 MHz. Silicon oxide
It is an insulator. Silicon / polysilicon is usually very thin
It has only a unique oxide layer. Therefore, low frequency
Number T_Two DC bias voltage fluctuation is seen on the oxide surface
Absent. This is because it is charged. Only
And basically uninsulated polysilicon isLowfrequency
Number T_Two Cycle low voltage excursion 72 (V
₁ A) forming a protective layer in between
In a similar manner the low frequency T_Two Reacts to. At this low frequency
The formed layer ishighFrequency T₁ High voltage with variable cycle
During excursion 73, etching is disabled. Previous
As described above, the insulating property of silicon dioxide causes T_Two of
Inhibition of etching suppression deposition during low voltage excursion
And oxide etching is T₁ Of the high voltage part of the cycle
Proceed without being restrained for a period.

[0033] That is, the protective layer on the silicon is formed in the low voltage excursions 72 of the low frequency cycle T _2, a high voltage excursion 7 high frequency cycle T ₁ to rapidly etch the oxide without inhibiting deposition
3 suppresses silicon etching. The result is a high etch rate for silicon oxide, a relatively low etch rate for silicon, and a high etch selectivity for oxides, as in the pulse / modulation method described above. It should be noted that the pulse / modulation method is currently preferred over the two-frequency bias method. This is because the former can perform precise control. 7. Load capacitor L at the input end of Faraday shield , tuning capacitor T at the other end,
Also consider the coil configuration of a typical antenna 30 having a relatively low voltage at the input and a much higher voltage at the other. The bottom winding of the coil near ground is connected to the low voltage RF input. Typically, the plasma is exposed to a relatively high voltage winding electrostatic field near the tuning end that initiates the plasma by initiating the decomposition of the gas electrostatically. Subsequent to the onset of decomposition, the coupling to the plasma is mainly electromagnetic or inductive. Such an operation is well known. Under steady state conditions, there is usually both electrostatic and inductive coupling. Electromagnetic coupling is dominant, but some types of processing are more sensitive to electrostatic fields. For example, etching of polysilicon requires low energy particles and low energy bombardment to prevent oxide etching.

Referring to FIGS. 1 and 15, the chamber of the present invention may optionally include a Faraday shield 45 to reduce the steady state electrostatic field. The structure in the embodiment shown in FIG. 15A is a "single" Faraday consisting of a grounded spaced axially extending post or bar or other cylindrical arrangement surrounding the dome wall 17W and antenna 30. This is called a shield 45S. The single shield can take a variety of forms from large spaced configurations to very small spacing between portions of the shield. FIG. 15B shows a so-called "all" Faraday shield 45F consisting of a pair of concentric shields spaced so that one bar overlaps the other. This eliminates the line of sight of the electric field lines passing through the shield,
This shunts the electrostatic field. Faraday shield 4
Various configurations are possible for 5S and 45F,
The presently preferred configuration is an open-ended, cylindrical, open-ended end configuration with an outwardly directed flange shown in a vertical cross-sectional view in FIG. Field surfaces 46, 47, 48 with single-wall or double-wall openings are the top, inner surface (source) of the antenna
And extends around the bottom surface, with the ground side 49 (which may not be open) located outside the antenna. According to this configuration, the magnetic component in the axial direction of the electromagnetic wave from the antenna 30 generates the plasma 16.
To induce a closed-loop electric field parallel to the plane of However, the shield 45 capacitively shunts the direct electric field component to ground and prevents the direct electric field component of the high frequency electromagnetic energy from coupling to the plasma. Shield 45
, The voltage that fluctuates along the antenna 30 couples to the plasma according to Maxwell's equation for capacitive displacement current coupling. Thereby, the plasma density and the substrate 5
Inhomogeneities and gradients are induced in the energy of GaN, which may cause non-uniform processing and high-energy charged particles.
According to Faraday's law in integral form, a changing magnetic field through a surface must produce a closed electric field at that surface. Maxwell's equation, which describes this phenomenon in differential form, shows that the rotation of the induced electric field is proportional to the negative time rate of change of the magnetic field. For sinusoidal excitation, the induced E is proportional to the changing B-field radiation frequency and its peak amplitude.

That is, a discontinuous or slitted or split Faraday shield minimizes the short-circuit effect of the shield on the changing electromagnetic field from the coil, reduces eddy current losses, and reduces the high frequency axial magnetic field. Enables coupling to the plasma and induces a closed-loop electric field that creates the plasma, but disallows direct coupling of this electric field (which changes on the antenna) to the plasma, thereby causing plasma non-uniformity and high energy Eliminate losses such as processing non-uniformity on charged particles. 8. Restriction and Enhancement of Magnetic Field 1) Restriction In order to reduce the loss (reduction in plasma density) at the wall 17W of the cylindrical / dome plasma source, a magnetic device for generating a peripheral annular (shallow) magnetic field is provided. FIG.
In a preferred configuration, shown in a horizontal cross-section, this field is provided by an axial permanent magnet or a "bucket" or cylindrical multipole array of closely spaced electromagnets 76-76, each magnet having its short length. A magnetic field B is formed around the alternating magnetic pole which is magnetized in the direction and is closed.
This multipole arrangement creates a multi-cusp magnetic mirror 77 on the dome wall. This arrangement can also be a horizontal ring magnet. Such a magnet reduces the electron loss of the wall 17W,
Improve plasma density and uniformity without exposing the substrate to a magnetic field.

Similarly, a permanent magnet or an electromagnet is placed around the lower chamber 16A, usually by NSNS ...
A multi-cusp magnetic mirror can also be generated on the chamber wall by arranging in a multi-pole arrangement of alternating NS configuration.
These magnets may be vertical bar magnets and preferably may be, for example, horizontal ring magnets. Such magnets can be used to reduce wall electron loss, thereby increasing the density and uniformity of the plasma without exposing the substrate to a magnetic field. In addition, a radial array of magnets can be mounted on the top of the dome or top plate 17T of the cylindrical plasma source to reduce losses at the top. Referring to FIG. 3, the substrate processing area 16
The plasma in B can be generated or decoupled from the plasma in the plasma source region 16A by placing a substantially planar magnet grid at the bottom of the plasma source region and at the top of the processing region. The magnetic grating is formed by a close, nearly parallel magnetic bar 78 similar to the bucket configuration described above.
-78, which is NS-magnetized in the short direction and the magnetic field lines exit from one bar and end at the next bar.
NS-NS-NS-Provide a magnetic field. The resulting substantially planar magnetic filter 79 over the opening 15 of the plasma source limits the magnetic field to this plane and plate area and does not penetrate the plasma source or the substrate area.

From the relationship F = qV × B, the high energy / high speed electrons of the plasma source are bent or repelled to a higher degree than the ions by the magnetic field 79 and cannot penetrate the substrate process region. As a result, the density of high-energy electrons in the process region 16B decreases, and the plasma density in the region decreases. The process region and the plasma source region are decoupled. This filter-based magnetic limitation is particularly effective in decoupling the plasma region of small systems. That is, for example, it provides a high substrate density without increasing the ion density on the substrate, while maintaining compactness. In one preferred arrangement, the filter magnetic restriction is performed using a machined aluminum plate with hollow bars for air cooling and elongated magnets. The bucket magnetic limiting configuration and the filter magnetic limiting configuration can also be used together. 2) Enhancement As described above, one or more (preferably at least two) permanent magnets or electromagnets 8 shown in FIG.
The horizontal plane and high frequency RF of the antenna coil using 1-81
A substantially axial electrostatic field can be formed perpendicular to and through both of the electric fields induced by the radiating antenna. Preferably, one of three types of magnetic fields, a uniform magnetic field, a divergent magnetic field, or a magnetic mirror, is used, as described below.

Referring to FIG. 14A, the magnet 8
A homogeneous axial uniform magnetic field 82 applied perpendicular to the substrate 5 by 1-81 restricts the movement of electrons to the wall. Since the ions cannot follow the high-frequency magnetic field fluctuation, the ions gather in the plasma on the substrate according to the shortage of the electrons. For maximum efficiency, this or other static magnetic field can be tuned to resonance with a high frequency electromagnetic field. Ω = 2πF
= Be / m, where B is the magnetic flux density, and e and m are the charge and mass of the electron, respectively. FIG. 14B schematically shows the divergent magnetic field 83 in the axial direction. By preserving the magnetic moment, the axial gradient of the magnetic field converts circular translational energy to axial translational energy, attempting to drive electrons and ions from a strong magnetic region to a weaker magnetic region. The diverging magnetic field can be used to push electrons and ions out of the plasma generation region and concentrate the plasma on the substrate. FIG. 14 (C) and FIG.
To explain (D), a bulging magnetic field or a supporting magnetic field 84 (FIG. 15C) and a cusp-shaped or counter magnetic field 85 (FIG. 15D) are shown. The effect of each of these so-called "magnetic mirror" magnetic fields is similar to that of the axially divergent magnetic field. The charged particles are driven from a relatively strong magnetic field region to a relatively weak central region.

By selectively positioning the magnets and selecting and varying the strength of the magnetic field provided by the magnet or magnets, the associated uniform divergent or magnetic mirror field is controlled. And increase the plasma density of the substrate. In the case of a magnetic mirror magnetic field, the preferred substrate location for maximum plasma density enhancement is overhang or on or close to the cusp, thereby providing maximum plasma density enhancement. There is a case where it is desired to improve the generation of plasma by using an axial magnetic field for the volume of the antenna, but to eliminate the magnetic field on the substrate. An annular disk made of a highly permeable material (such as nickel or steel for soft iron) can be interposed on the lower substrate 5 in the plane of the magnet and antenna. 3) Extraction Using a suitable magnetic field, ions and electrons can be extracted in the direction of the substrate. 9. Control System The following definitions are used for the control system shown in FIG. Psp Power set point Pf Forward power (measured by directional coupler placed on power supply) Pr Reflected power (measured by directional coupler placed on power supply) | Z | Impedance magnitude <phi impedance phase Tsp tuning setting Point Lsp Load Set Point Tfb Tuning Feedback (Measured Value) Lfb Load Feedback (Measured Value) FIG. 16 is a block diagram of a representative system for controlling various components including a power supply. Here, the system controller 86 is interfaced with the antenna power supply 31, the impedance bridge 87, the antenna 30, the bias power supply 31, the impedance bridge 88, the matching network 43, and the cathode 32. Process parameters, antenna power and DC bias selected for ion flux density and ion energy are provided as inputs to controller 86. The controller 86 also controls gas flow, chamber pressure, electrode or substrate temperature, chamber temperature, and other parameters. The controller 86 has a Tsp connected to the antenna 30
By issuing a signal on _one line and the Lsp _one line, the initial tuning 1 and load 1 conditions can be set. The controller can also set initial tuning ₂ and load ₂ conditions by issuing signals on the Tsp ₂ and Lsp ₂ lines connected to matching network 43. Typically, these conditions are selected to optimize the onset of the plasma (gas breakdown).
Power can first be applied to either the antenna 30 or the cathode 32 or both simultaneously. The controller 86 sequentially simultaneously or on Psp ₂ line to Psp ₁ line and a bias power source 42 to the antenna power supply 31 (sequential Which may be the first) emits power setpoint.

The avalanche breakdown occurs rapidly in the gas,
A plasma is generated. Controller 86 is antenna 3
0 (Pf₁) And reflected power (Pr₁)
And the forward power (Pf_Two) And
And reflected power (Pr_TwoMonitor). DC bias (cathode
-Anode DC voltage) is also shown to controller 86
Watched. The controller 86 has (a) forward power
Pf₁ And reflected power Pr ₁ Or (b) impedance
Size | Z₁ | And impedance phase <phi₁Nozomi
Line based on TSP₁ And Lsp₁ Set point above
Tuning coil by emitting₁ And load₁ The parameters of
Adjust the parameters. Bridge 87 is connected to controller 86
Gives information on the magnitude and phase angle of the impedance. antenna
30 is the reflected power Pr₁ Is almost zero,
-Dance (size and phase | Z₁| <Phi) is coil power supply
Matches when the output impedance is complex conjugate.
ing. (Zero reflected power condition and conjugate impedance condition
Occur simultaneously. Therefore, the reflected power is minimized
Or the impedance is matched,
The result will be the same. Alternatively, VSWR (voltage standing
Wave ratio) or reflection coefficient is minimized. )controller
86 indicates (a) forward power Pf_Two And reflected power Pr_Two There
Is (b) the magnitude of impedance | Z_Two | And Impeder
Phase <phi_TwoLine Tsp based on one of_Two You
And Lsp_Two This emits a set point on the cathode 3
2 and matching network tuning_Two And load_Two Parameters
To adjust. Bridge 88 is connected to controller 86
-The size of the dance | Z_Two | And phase <phi_TwoGive the information
You. As with antenna matching, the reflected power Pr_Two Gaho
When zero, the impedance (magnitude and phase
｜ Z_Two | <Phi_Two) Is a bias power supply 504 output impedance
Matching occurs when the complex conjugate of the DC bus
The eas is monitored by the controller 86. Conte
The roller 86 is changed by changing the output power of the bias power supply.
To obtain the measured DC bias. The controller 86 is a DC bus
Subtract DC bias measurement from desired value of ias
You. If the difference is negative, the output of the bias power supply 42
Can be raised. If the difference is positive, the bias power
(The higher the output of the bias power supply, the lower the DC
Ias grows in the negative direction. According to this method,
Proportional control, proportional integral control, or proportional integral derivative control
Alternatively, other controls can be used.

In place of this embodiment in which the output of the bias power supply 42 is adjusted to maintain a constant DC bias,
A constant bias power supply output can also be used. In addition to the DC bias servo matching technique described above, automatic tuning can be performed by servoing to the peak-to-peak RF voltage. This latter method may be useful, for example, in certain etching processes that require sufficient conductive area on the cathode and anode to provide current for driving the instrument. Using polymer coating technology, these conductive regions are passivated, preventing current from saturating the meter and providing an effective reading. In contrast, the peak-to-peak RF voltage method is unaffected at low frequencies, which are particularly relevant for the preferred frequency range. The measurements can be obtained with a matching network 43 close to the chamber instead of the cathode. Controller 86
May be a central controller or a distributed system of controllers. The turn-on / turn-off sequence is important for obtaining a sensitive substrate device structure. In general, it is preferable to turn on the plasma source first and turn it off last. This is because this method can minimize the change in sheath voltage. In some applications, it may be better to turn on the bias first. 10. As described in detail in the patent application US Pat. No. 559,947 with reference to the transmission line structure , proper coaxial / transmission line design includes feeding through low characteristic impedance, short transmission from the matching network to the wafer. A return path is required along the line, and along the transmission line. This design condition is to insulate the cathode 32C, the concentric annular conductor 320, and the cathode surrounding the cathode 32C from the concentric annular conductor 320, and replace the non-porous low loss insulator 3 that replaces the process gas that may yield.
2I is satisfied by the integral transmission line structure 32 shown in FIG. For example, materials such as Teflon ^™ and quartz are preferred because of their high dielectric strength, low relative dielectric constant, and low loss. The input side of this structure is connected to a matching network in the manner described below. The insulated cathode 32C and outer conductor 320 are matched to the matching network 43 and the plasma 16
To provide separate current paths. One reversible current path is from the matching network along the perimeter of cathode 32C to the plasma sheath on the surface of the chamber (electrode). Second
The reversible path from the plasma 16 is along the inner portion of the top of the chamber habit 12 and then along the conductive exhaust manifold screen 29 through the interior of the outer conductor 320 to the matching network. It should be noted that the exhaust manifold screen 29 forms part of the uniform radial gas pumping system and return path for the RF current.

During the application of AC energy, the direction of the RF current path alternates between the illustrated direction and the opposite direction. Since the transmission line structure 32 is a coaxial cable type structure, and more specifically, the internal impedance of the cathode 32C is high (compared to its outside), the RF current is transmitted to the outside and outside of the cathode 32C in the form of a coaxial transmission line. It flows on the inner surface of the conductor 320. The skin effect concentrates the RF current near the surface of the transmission line, reducing the effective area of the current path. The use of a large substrate, for example, 4-8 inches in diameter, and a corresponding large diameter cathode 32C and large diameter outer conductor 320 increases the effective cross section and allows low impedance current to flow through the transmission line structure. Also, if the coaxial transmission line structure 32 is terminated with a pure resistance equal to its characteristic impedance Z ₀ , the matching network will have a constant impedance Z ₀ regardless of the length of the transmission line. However, this is not the case in practice. This is because the plasma operates over a range of pressures and powers and consists of various gases, which collectively change the load impedance Z ₁ that the plasma provides at the end of the transmission line 32. Load Z ₁ is not ideal (i.e. non-lossless)
Because it is not matched to the transmission line 32, standing waves on the transmission line increase resistive, dielectric, and other losses between the transmission line and the matching network. The matching network 43 can be used to eliminate the standing wave and losses from the input of the matching network to the amplifier or power supply 42, but the matching network, transmission line 32, and the plasma in the chamber are coupled to the transmission line 32 and matching network. A resonance system which increases resistance loss, dielectric loss, and other loss between the components 43 is formed. That is, the load impedance Z ₁ does not match the loss, but the loss is minimized when Z ₁ ≒ Z ₀ .

In order to eliminate the loss due to the load mismatch, the coaxial transmission line structure 32 has a characteristic impedance Z most suitable for the range of the load impedance associated with the plasma operation.
Designed to have ₀ . Usually, the operating parameters described above: for the material (eg bias frequency range is from about 0.3-3MHz) are discussed and equivalent series RC load impedance Z ₁ given to the transmission line from the plasma of about 10 ohms Consisting of a resistance in the range of 100 ohms and a capacitance in the range of about 50 picofarads to about 400 picofarads. Therefore, the optimum value of the transmission line characteristic impedance Z ₀ is selected in the middle of the load impedance range, that is, about 30 ohms to 50 ohms. The transmission line 32 must be very short to avoid deformation of the plasma impedance seen by the matching network. Preferably, the transmission line is much shorter than a quarter wavelength (λ / 4). More preferably, it is about (0.05-0.1) λ.
In order to perform power coupling more efficiently, the inner diameter (cross-sectional dimension) of the return conductor 320 must not be significantly larger than the outer diameter (cross-sectional dimension) of the central conductor 32C.

That is, the chamber contains a transmission line structure for coupling the power from the matching network 31 to the plasma 33. This transmission line structure is preferably (1) very short or almost equal to a half wavelength compared to a quarter wavelength at the frequency of interest to prevent deformation of the plasma impedance. An outer conductor path having a characteristic impedance Z ₀ selected to suppress losses due to the presence of standing waves on the line between the matching networks, and (3) the cross-sectional dimension is not much greater than the cross-sectional dimension of the central conductor Is used. 11. Chamber Temperature Control Temperature control functions that can be incorporated into the reactor chamber system 10 include fluid transfer to maintain the internal or external temperature of the intake manifold above or below a certain value, or within a certain range. Use of heat medium, resistance heating of cathode 32C, fluid heat transfer heating or cooling of cathode 32C, use of gas heat transfer medium between substrate 15 and cathode 32C, chamber wall 12-14 or dome 17.
Including but not limited to the use of a fluid heat transfer medium to heat or cool the substrate 15 and mechanical or electrostatic means to restrain the substrate 15 to the cathode 32C. Such a function is referred to herein as the co-assigned 19
U.S. Pat. No. 4,872,947, issued Oct. 10, 1989, and U.S. Pat.
No. 842,683.

For example, using a recirculating closed loop heat exchanger 90, a fluid, preferably a dielectric fluid, is passed through the block and pedestal of the substrate support / cathode 32C as shown schematically in flow path 91 to provide substrate support. The body can be cooled (or heated). For silicon oxide etching, a dielectric fluid temperature of, for example, -40C is used. As described above, heat transfer between the substrate 5 and the substrate support 32 is enhanced by an inert gas heat transfer medium such as helium at the interface between the substrate and the support. The chamber walls and the dome can be heated or cooled by air convection (blown air) and / or a dielectric fluid heat exchanger. For example, the closed circuit heat exchanger 92 recirculates the dielectric fluid from the heating to the cooling along the passage 93 to the sidewall of the chamber at a controlled temperature, for example, in the range of + 120C to -150C. Similarly, the dome side wall 17W and the top wall 17T are connected to the passages 95 and 97.
Can be heated or cooled by heat exchangers 94, 96 which recirculate the fluid along. In an alternative dielectric thermal control system, the antenna coil 30 is positioned between the dome double wall 17W and immersed in a recirculating dielectric fluid. In another alternative dome dielectric fluid thermal control method, the coil of the antenna 30 is encapsulated in high temperature plastic or Teflon ^™ , a thermally conductive grease is applied between the encapsulated antenna and the dome, and the hollow coil is insulated. The fluid is heated or cooled by flowing the fluid through the coil. Because RF energy is also applied to the coil and close to the plasma, dielectric oils have good dielectric and insulating properties in addition to high specific heat and density for efficient heat transfer at acceptable flow rates. Must have a higher boiling point. A suitable dielectric fluid is Siltherm sold by DuPont. 12. Three Electrode Configuration Referring to FIG. 1, in the presently preferred embodiment, the chamber incorporates a unique three electrode configuration that allows for new process control and improvements. This configuration consists of a cathode (preferably, a substrate support electrode 32), an anode (preferably, the chamber side and bottom walls) and a top electrode, wherein the top electrode is a top electrode that is (or includes) a dome top plate 17T. Become. As shown in FIG. 1, the top electrode is floating, grounded, or connected to an RF power source 40. The top electrode includes various configurations and can be made of various materials. That is, a sacrificial silicon member 17S made of a conductive material (preferably aluminum), a dielectric coated material such as anodized aluminum, a silicon or silicon-containing conductive material such as an aluminum-silicon alloy, or a silicon wafer. Including, but not limited to the silicon substrate. 1) Grounded Third Electrode Grounded top plate 17T enhances the ground reference plane of the bias voltage (relative to the conventional reference provided by wall 12) so that the plasma source 16A to the process area 16B Enhance ion extraction and thus increase process rates (such as etch rates). In addition, the grounded top plate improves the coupling between the plasma (generated by the plasma source) and the substrate. 2) Biased third electrode RF using a biased third electrode in combination with free supply of silicon to the source plasma (using or containing a silicon-containing member). Various process characteristics, including etch rate and selectivity, are improved. With the help of the strong dissociation properties of the source plasma, silicon enters the gas phase and combines / removes with free fluorine. (Due to the dissociation properties of the source plasma, the use of fluorine-containing gas chemistry for oxide etching, for example, results in high concentrations. This increases the oxide etch rate but also increases the etch rate of related substrate materials such as polysilicon. Thus, the selectivity of the oxide to polysilicon is reduced.) The removal of fluoride ions by free silicon forms a carbon-rich polymer. As a result, the oxide etch rate is increased, the oxide selectivity over polysilicon is increased, and the oxide etch anisotropy and vertical profile are enhanced. In addition, free silicon affects the polymerization reaction, forming a carbon-rich polymer, which is easily etched in the presence of oxygen to etch the oxygen-containing layer, but in the absence of oxygen For example, when the openings reach a substrate such as an oxygen-free layer or polysilicon, the suppression is increased and the selectivity of the oxide with respect to the oxygen-free substrate is increased. 13. Processing Examples Examples of processes that are advantageously achieved in the reactor of the present invention are as follows.

Example 1 Polysilicon on silicon oxide
The polysilicon over the silicon oxide on the etched silicon wafer is subjected to a 50 cc chlorine (Cl ₂ ) etchant gas flow (manifold G1 only) in a three electrode chamber of the present invention at a pressure in the range of about 2 mTorr to about 20 mTorr. This was done with a 1500 watt power supply, a 20 volt bias voltage, and a grounded top electrode (no silicon). The result was a polysilicon etch rate of 3500-4000 angstroms / min, a vertical etch profile, and a polysilicon selectivity over oxide of 100: 1 or greater. Example 2. Silicon Oxide Deposition Two step bias sputter deposition of silicon dioxide into high aspect ratio openings formed in a silicon oxide layer on a silicon wafer is performed in a three electrode chamber at a pressure in the range of about 2 mTorr to about 10 mTorr (both steps). ),
A gas flow rate of about 200 cc of argon / about 90 cc of oxygen / about 45 cc of silane (only manifold G1 in both steps), 20
This was performed using a power supply of 00 watts (both steps), a top electrode grounded (both steps), a bias voltage of about -20 volts (first step) and about 100-200 volts (second step). . As a result, in the first step (without sputtering), a deposition of 7500 Å / min or more was obtained, and in the second step, a pure oxide deposition (profile-controlled sputtering deposition) of about 4000 to 5000 Å / min was obtained. The upper corner of the opening was processed by argon sputtering, followed by filling the opening with silicon dioxide without voids, thereby flattening the silicon oxide layer. This filling and planarization can be performed continuously in the same deposition chamber of the present invention.

A significant challenge in semiconductor fabrication is that when the underlying layer is polysilicon or other oxygen-free material,
Etching a selected thickness of silicon dioxide. Although the exposed silicon is hardly etched, a high selectivity is required because the silicon oxide is etched at a relatively high rate. However, when exposed to the plasma, since the two materials are etched at approximately the same rate, to make the coating of the polymer on the substrate, applying an etching gas such as CHF ₃ fluoride source is known is there. This creates a passivated coating on the polysilicon, although the silicon oxide to be etched is continuous. However, the formation of a polymer layer on polysilicon makes it difficult to etch the geometry of small devices. An important concept in this regard is "microloading", which is defined as 1- (etch rate ratio). Here, the etching rate ratio is a ratio between the etching rate in a small shape of the wafer and the etching rate in a large shape of the wafer. Therefore, if the etching process has the desired property of etching large and small features at the same rate, this microloading will be 1-1 / 1 = 0.
become. In processes where small features are etched at a much slower rate than large features, microloading approaches 1.0.

A difficulty with the above-described etching application is that relatively high amounts of polymer forming gas must be used in the plasma to obtain high etch selectivity, but the polymer layer is much more than zero. Large microloading. Usually about 0.
In the case of microloading of 1, a selectivity ratio of 10: 1 or more cannot be expected. However,
30: 1 or 4 with microloading close to 0
Many applications require a selectivity ratio of about 0: 1. Use of Silicon in the Plasma Region For high density plasma sources, one of the separation products that etches polysilicon is fluorine. Free fluorine groups can be removed from the plasma source region using silicon. Silicon can be in the form of a coating on the third electrode 17T or on the inner wall 17W of the chamber. If the sacrificial silicon is on a wall, the thickness of the silicon becomes an issue with the frequency at which RF energy is supplied to the plasma from antenna 30. These parameters are
It must be chosen to ensure that sufficient energy is electromagnetically coupled through the chamber walls. When silicon is included in the third electrode 17T, the thickness of the silicon is not so important. In any event, if silicon is used to remove free fluorine from the plasma source region, volatile compounds are formed that are easily removed from the chamber.

The silicon scavenger material itself is coated with a polymer during the etching process. The polymer can be sputtered from the surface because the polymer can be removed by heating at elevated temperatures or by increasing the impact on the silicon surface by electrically biasing the silicon, thereby freeing the silicon. Exposure again. Silicon Oxide on Polysilicon Etching Silicon oxide on polysilicon is applied in a three electrode chamber of the present invention at a pressure of about 2 mTorr to about 30 mTorr, CHF ₃ , 30-60 sccm / CO or CO ₂ -18 sccm.
/ Ar, 100-200 sccm (manifold G1 only) gas chemistry flow rate, 2000 watts power supply, 200 volt bias voltage, top electrode 17T with 2 MHz, 1000 watts of RF energy attached to them. This was performed using a silicon disk 17S. The silicon oxide was etched at a rate of 8000 Å / min and the oxide selectivity to polymer was 50: 1. Also, the silicon-containing body can be reinforced by a silica coating on the quartz dome wall 17W. Oxygen-Containing Layer Etching on Non-Oxygen-Containing Substrates A series of etching processes were performed by the reactor of the present invention for etching oxygen-containing layers such as silicon oxide on various non-oxygen-containing substrates. After patterning the oxide layer using conventional photolithographic techniques, etching using C ₂ F ₆ was performed in the reactor of the present invention using a grounded third silicon electrode. The results are summarized in Table 1 below.

Table 1Example Silicon oxide thickness substrate Selectivity 4 5000-10000} Polysilicon 25: 1 5 5000-10000} WSi 25:16 PECVD Si_ThreeN_Four 15: 1 75000-10000 LPCVD Si_ThreeN_Four 15:18 5000-20,000 {TiN / Al * 15:19 5000-20000} Al 30: 1 10 5000} Single-crystal silicon 30: 1 * 2-3000 CF_Four Another series of etchings using BPSG (Boro
-Phosph-Silicate-Glass)
Other than the above, the same conditions were used. The results are shown in Table 2 below.
Is summarized in Table 2Example Silicon oxide thickness substrate Selectivity 11 5000-10000 WSi 30: 1 12 5000-25000 {p + doped single crystal 30: 1 silicon 13 5000-25000} p + doped single crystal 30: 1 silicon Selectivity depends on gas flow, power supply and silicon plate. Change the conditions
Can be adjusted by special oxidation on special substrate
The selectivity of silicon or glass can be made appropriate.
Using a silicon scavenger of fluorine ion of the present invention
In the reactor, selectivities of up to 100: 1 can be achieved.
You.14． Other features 1)Plasma control A feature of this invention is that it automatically changes the "lower" power.
Maintaining a constant cathode (substrate) sheath voltage.
You. Low pressure (<50 in highly asymmetric systems)
0 millitorr), the DC bias measured at the cathode is
This is an approximate value of the cathode sheath voltage. Lower power is automatic
To maintain a constant DC bias.
You. Bottom power is almost impossible for plasma density and ion flow density.
No effect. Top or antenna power is plasma tight
Although it has a very strong effect on the temperature and ion flow density,
The effect on the sheath voltage is very small. Accordingly
The upper power to determine the plasma density and ion flow density.
And use the lower power to determine the cathode sheath voltage.
Is desirable. 2)Differential bias Instead of biasing the substrate 5 against ground, FIG.
And the bias matching network as shown by the dotted connection 50 in FIG.
Network 43 and top plate 17T out of ground
However, each other can be a reference. Figure 2
To illustrate, the top plate is the one between the top plate and the substrate.
Pressure V_T-SSIs the voltage V between the top plate and the wall 12_TW Large
The voltage V between the substrate and the wall at about twice the magnitude_SS-WAbout the size of
It is differentially driven and balanced to be doubled. This
Balanced differential drive enhances plasma-wall interaction
Between the plasma source region 16A and the substrate region 16B
Interaction, increase the ion extraction. 3)Alternative configuration Fig. 1 shows a conventional plasma reactor system according to the present invention.
(Vertical). Substrate 5 on electrode 32 (cathode)
Yes, antenna 30 surrounds dome 17 above electrode
In. For convenience, the power supplied to the antenna 30 is “A”.
Antenna "or" plasma source "or" top "power
And the power supplied to the electrode / cathode 32 is
We have called them “ass” or “lower” power.
The names and names are for convenience only and may not be
The system is inverted, i.e. the antenna is
It can be configured below the poles or it can be modified
Place it in another way without adding it (for example, horizontally
Can be placed). In other words, this reactor system
Works regardless of orientation. Plasma in inverted configuration
Is generated by the antenna 30 and transported upward to
To the substrate 5 located above the antenna in the manner described in
U. In other words, the transport of active species is based on diffusion and bulk flow.
It occurs. Or possibly with an axial gradient
Generated with the help of a magnetic field. This process depends on gravity
And is therefore relatively insensitive to direction. Invert
The direction is, for example, a plasma generation region in a gaseous state, or
Minimizes the possibility of particles formed on the surface falling on the substrate.
This is useful in that it can be minimized. Then by gravity
Only the smallest of these particles is gravitational potential
It rises toward the substrate surface against the gradient. 4)High and low pressure operation and variable spacing The design of the chamber of the present invention is both high pressure operation and low pressure operation
It is effective for The substrate support cathode 32C and the antenna
The distance d between the lower coil or winding surface is high pressure operation
And low pressure operation. for example
Suitable for high pressure operation of 500 mtorr-50 mtorr
The distance d is less than about 5 cm and is 0.1 mTorr
5 cm for low pressure operation in the range less than -500 mTorr
Larger intervals d are preferred. Shown in chamber
A fixed distance d can be used as
Interchangeable or nestable upper chamber
Variable spacing designs can also be used. Reactor system 1
0 is the high pressure of materials such as silicon oxide and silicon nitride and
Low pressure deposition, silicon dioxide, silicon nitride, silicon,
Low materials such as polysilicon, glass and aluminum
Pressure anisotropic reactive ion etching, high pressure plastic
Including zuma etching and planarization of substrate topography
CVD including simultaneous deposition and etchback of such materials
It is effective for processing such as faceting. Reactor
These processes that can use the stem 10 and their processing
For other processing, please refer to the VHF / UHF PLASM
A PROCESS FOR USE IN FORMING INTEGRATED CIRCUIT ST
RUCTURES ON SEMICONDUCTOR WAFARS (on semiconductor wafer
VHF / UHF plasma pump used to form integrated circuits
Collins Others, July 31, 1990, entitled "Roses"
U.S. patent application Ser. No. 07 / 560,530 (AMAT file).
No. 151-2). 15.Example of device The embodiment of the system of the present invention includes a plasma source structure shown in FIG.
Configuration and antenna configuration. 5-inch high quartz
The diameter of the plasma source chamber 17 is 12 inches.
You. 2MHz, 13 inches in diameter, 4 inches in height, 13 volumes
At both ends (with a grounded variable capacitor L
About 0.25 inch from (below) ground plane
Terminates at intervals and surrounds the plasma source. Anti
Variable load matching with variable capacitor L (10-3000 picof
Powered by a variable capacitor, rated 5kV)
I have. Capacitive tuning to antenna resonance is also
Sensor T (5-100 picofarads, 15kV rating)
Provided. 2 kW 2 MHz source RF energy
When the operation using energy is performed, 2 inches downstream (plus
An extending plasma is provided to the wafer (below the source).
This is 1-2X10 ¹²/cm^Three Plasma density and substrate
10-15 mA / cm downstream^Two Provides ion saturation current density of
I do. Place it on the support electrode about 2 inches (downstream) below the antenna.
2 MHz, 600 W applied to a 5-inch substrate placed
Bottom of unit or bias is 200 volt cathode
Provides sheath voltage.

It will be apparent to one skilled in the art that the present invention is not limited to the use of a dome. Rather, the present invention is applicable to almost any configuration having a plasma source region and a process region. This includes, for example, a "stepped" dome-shaped chamber configuration as shown, or a non-stepped configuration in which the plasma source region and the process region or chamber portion have approximately the same cross section. As mentioned above, the above-described reactor embodying the present invention includes reactive ion etching (RIE), high pressure plasma etching, low pressure chemical vapor deposition (CVD) including sputter facet deposition and planarization and high pressure conformal isotropic C
Provides unique effects for various plasma processing such as VD. Applications include, but are not limited to, sputter etching, ion beam etching, or an ion or active neutral plasma source for electrons. Having described preferred embodiments of the apparatus and process of the present invention, those skilled in the art will readily be able to adapt, modify and extend the apparatus and methods within the scope of the appended claims.

[Brief description of the drawings]

FIG. 1 is a sectional view of a plasma reactor of the present invention.

FIG. 2 is a cross-sectional view of the plasma reactor of the present invention.

FIG. 3 is a sectional view of the plasma reactor of the present invention.

FIG. 4 shows a tuning circuit.

FIG. 5 is a diagram showing a load circuit;

FIG. 6 is a diagram showing a load circuit to which an RF input is applied;

FIG. 7 is a diagram showing coupling between a tuning circuit and a load circuit;

FIG. 8 is another embodiment showing the coupling of the tuning circuit and the load circuit.

FIG. 9 is another embodiment diagram showing the coupling between the tuning circuit and the load circuit.

FIG. 10: Etching / deposition rates for silicon oxide and silicon bias voltage

FIG. 11 is a waveform diagram of a DC bias voltage applied during etching.

FIG. 12 is another diagram of a DC bias voltage waveform applied during etching.

FIG. 13 is a horizontal plan view of the magnet array in the dome of the reaction chamber according to the present invention.

14A to 14D are diagrams showing magnetic field lines of various shapes of the system of the present invention.

FIGS. 15A and 15B are diagrams showing an embodiment of a Faraday shield useful for the system of the present invention.

FIG. 16 is a block diagram of a control system of various components in the system of the present invention.

Continuing on the front page (72) Inventor Kennis S. Collins United States 95112 San Jose North 9th Street, California 871 (72) Inventor Jang Long Yang United States 95032, California United States 95032 Los Gatos Leroy Avenue 16788 (72) Inventor Jerry Yuen Kui Won United States 94539 California Fremont Cougar Sarkle 44994 (72) Inventor Jeffrey Marks United States of America 95129 San Jose Cielo Vista Way 4730 (72) Inventor Peter Earl Keswick United States of America 94560 Newark Joaquin Marie Ta Avenue 6371 A (72) Inventor David W. Grocher 94087 Sunnyvale Robin Way 927 California USA Term (Reference) 5F004 AA05 BA20 BB07 BB11 BD04 BD05 CA02 CA03 CA04 DA01 DA16 DA26 DB01 DB02 DB06 DB07 DB09 DB17

Claims (6)

[Claims] 1. A plasma etching method, comprising: a) providing a vacuum chamber for maintaining a plasma in a vacuum chamber; b) on a support device connected to an RF energy source in the vacuum chamber. Providing a silicon substrate to be treated by said plasma; c) forming a plasma by a coil antenna;
Inductively coupling RF energy to the vacuum chamber; d) providing a fluorine-containing etching gas to the vacuum chamber; e) etching during processing of the substrate to remove fluorine ions in the plasma. Passing a gaseous silicon or carbon source in addition to the gas. 2. The method according to claim 1, wherein said fluorine-containing etching gas is fluorocarbon. 3. The method according to claim 1, wherein said etching gas is fluorine, hydrogen and carbon. 4. The method of claim 1, wherein the substrate has an oxygen-containing material on the substrate to be etched. 5. The method of claim 1, wherein a solid source of silicon or carbon is in the plasma region of the vacuum chamber and is independently treated to maintain its surface to react. 6. The method of claim 1, wherein the solid source of silicon or carbon is independently heated to maintain the surface to be reacted.