KR100265617B1 - Plasma reactor and method using electromagnetic high frequency coupling - Google Patents

Abstract

The plasma reactor chamber is inductively coupled inside the reactor. The antenna generates a high density, low energy plasma inside the chamber to etch the oxygen containing layer on top of the oxygen free layer with high selectivity. The auxiliary RF bias energy applied to the wafer support cathode controls the cathode sheath voltage and the ion energy independent of density. Various magnetic and voltage process enhancement techniques are described according to other etching processes, deposition processes, and etch / deposition bonding processes. The invention described provides a process for the processing of sensitive devices without damage or micro loads, thereby improving yield. Oxygen-containing layer etching on top of the oxygen free layer can be achieved with high selectivity.

Description

Plasma reactor and method using electromagnetic high frequency coupling

1 to 3 are partial cross-sectional views of the plasma processing apparatus of the present invention.

4 is a tuning circuit diagram.

5 is a load circuit diagram.

6 is a load circuit diagram to which RF input power is applied.

7 is a combination diagram of a lookup and load circuit.

8 shows another combination of a tuning circuit and a load circuit.

9 is another combination of the tuning circuit and the load circuit.

10 is a graph of etch rate / deposition rate versus bias voltage for silicon oxide film and silicon.

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

12 shows an alternative embodiment of a DC bias voltage waveform applied during etching.

13 is a cross-sectional plan view of a magnetic array in the dome of the reaction chamber of the present invention.

14 (a)-(d) are magnetic field diagrams of various forms of the device of the present invention.

Figures 15 (a)-(b) illustrate embodiments of Faraday shielding structures useful in the apparatus of the present invention.

Figure 16 is a block diagram of a control system for various components of the apparatus of the present invention.

* Explanation of symbols for main parts of the drawings

5: wafer 10: system

11: vacuum chamber 16A: source chamber

16B: wafer processing unit 17: dome

21: Vacuum pump system G ₁ , G ₂ , G ₃ : Manifold injection source

30: antenna 32I: insulator

39: vacuum pump 47, 76: electromagnet

43: Matching Network 45: Faraday Cloth

52: cathode 86: system regulator

TECHNICAL FIELD The present invention relates to a high frequency (RF) plasma processing reactor, and in particular, a unique plasma reaction using a high frequency (RF) energy source for electromagnetically coupling related high frequency electromagnetic waves to a plasma and a silicon source in contact with the plasma. It relates to a process carried out in the cabin.

The increasing trend toward dense integrated geometries has necessitated very small geometries and devices that are electrically sensitive to energy particle impacts and radiation losses when a small sheath voltage of about 200 to 300 volts is applied to the wafer. Unfortunately, this voltage is less than the voltage applied to the circuit components during the standard integrated circuit fabrication process.

Structures such as MOS capacitors and transistors fabricated for advanced devices have very thin (less than 200 microns) gate oxides. The device can be damaged by charge-up and the gate can collapse. This occurs in plasma success when neutralization of surface charge is not caused by non-uniform plasma potential and / or density or large RF displacement current. Conductors such as interconnect lines can be damaged for the same reasons as above. Etching processes performed in conventional plasma etching chambers are inaccurate because very high aspect ratios, i.e., very deep and very narrow openings and trenches are formed in or filled with various semiconductor materials.

[RF system]

Conventional semiconductor processing systems, such as chemical vapor deposition (CVD) and RIE reactor systems, use RF energy with high frequencies of about 13.56-40.68 MHz at low frequencies of about 10-50 MHz. Below about 1 MHz, ions and electrons are promoted by vibrating the electric field and by any steady-state electric field in the plasma. At these relatively low frequencies, the electrode sheathing voltage on the wafer is typically up to one or more kilovolt peaks, and the loss is much greater than that at nearly 200-300 volts. Above a few MHz, the electrons can still match the change in the electric field. Larger ions cannot respond to changes in the electric field, but are promoted by the steady state electric field. In this frequency range (and actual gas pressure and power levels), the steady state sheath voltage is in the range of hundreds to 1000 volts or more.

[Magnetic field-enhanced]

Preferred methods of reducing the bias voltage in an RF system include applying a magnetic field to the plasma. This B field allows electrons to form in the region near the wafer surface and increases ion current and ion flux density, thus reducing the need for voltage and ion energy. By comparison, a non-magnetic RIE process for etching silicon dioxide was applied with 13.56 MHz RF energy, 10-15 L asymmetric system, 50 millitorr pressure and about (8-10) to 1 anode area to wafer support cathode. A ratio of area, about 800 volts, of the development wafer (cathode) external voltage can be used. Applying a 60 gauss magnetic field can reduce the bias voltage by about 25-30%, or about 500-600 volts, and increase the etch rate by about 50%. However, applying a normal field B parallel to the wafer can develop E x B ion / electron drift and associated plasma density gradients oriented almost across the wafer. Non-uniformity can be achieved by rotating the magnetic field around the wafer, typically by mechanical movement of permanent magnets, or by using electromagnetic coil pairs formed quadrature at 90 ° from the phase, or instantaneously adjusting the current in the coil pairs in steps or otherwise. Otherwise it can be reduced by moving the magnetic field at a controlled rate. However, although the non-uniform tilt can be reduced by rotating this chapter, some non-uniformity usually remains.

In addition, it is difficult to pack two-phase or more coils, especially in one chamber, especially when using a multichamber system of individual self-reinforced reactor chambers surrounding the Helm Holtz coil arrangement and / or the loadlock. It is also difficult to obtain a factor system.

United States Patent No. 4,842,683 to inventor, et al. Discloses a specific reactor system having the ability to temporarily and selectively change the magnetic field strength and direction and designed for multichamber reactor compact systems.

[Microwave / ECR System]

Microwave and microwave ECR (electrocyclone resonance) systems use microwave energy at frequencies greater than 800 MHz, typically 2.45 GHz, to produce plasma. This technique creates a high density plasma, but it is below the minimum response threshold energy for many processes, such as silicon dioxide's RIE. For replenishment the energy-enhanced low frequency power is coupled to the wafer support electrode and coupled to the plasma through the wafer. Thus, the probability of wafer loss is reduced compared to conventional systems.

However, microwave and microwave ECR systems operating at actual power levels for semiconductor wafer processing such as etching or CVD require large waveguides for power transmission, expensive tuners, directional couplers, circuits, and operational pseudo-loads. Additionally, to meet the ECR requirements for microwave ECR systems operating at 2.45 GHz, a magnetic field of 875 gauss, large electromagnets, large power and cooling are required.

Scales of microwave and microwave ECR systems cannot be easily measured. The hardware is suitable for a frequency of 2.45 kHz because this frequency is used in microwave ovens. 915 MHz systems are useful, although expensive. Hardware at other frequencies is not easy or economically useful. Thus, in order to scale 5-6 inch microwave systems that can accommodate larger semiconductor wafers, it is necessary to use a high operating mode. Such scaling at a fixed frequency according to a high operating mode requires very dense process control to avoid mode flipping into high or low order loads resulting in process changes. For example, scaling for 5-6 inch microwave cavities is accomplished by using divergent magnetic fields to spread the plasma flux over a larger area. This method reduces the effective power density and thus the plasma density.

In addition, since the density of plasma generated in the apparatus drops very rapidly to about 2-3 millitorr, the ECR system must operate at very low pressures of 2-3 millitorr. Therefore, a large volume of reaction gas needs to be supplied to the apparatus, and a large vacuum exhaust apparatus is required to remove this large volume of gas.

[HF transmission line system]

United States Patent Application No. 559,947, entitled “VHF / UHV Reactor System”, is incorporated by reference herein. This related application is known for high frequency VHF / UHF reactor systems. In this reactor system, the reactor chamber itself is arranged as a transmission line structure to apply high frequency plasma generated energy to this chamber from mating the network. This particular integrated transmission line structure satisfies the need for very short transmission lines between the matching network and attachment and allows the use of relatively high frequencies of 50 to 800 MHz. This makes it possible to effectively and controllably apply the RF plasma generating energy to the plasma electrode to generate an industrially acceptable etch rate and deposition rate with relatively low ion energy and low external voltage. This relatively low voltage reduces the potential for loss of electrically sensitive small geometry semiconductor devices. In the VHF / UHF system, conventional problems such as the scalability and power limitation described above can be avoided.

Accordingly, it is a primary object of the present invention to provide a plasma reactor and method that uses high frequency AC energy to generate plasma.

The present invention is a vacuum chamber having a source region to overcome the problems of the prior art; A substrate support electrode in the vacuum chamber; A process gas source in the vacuum chamber; Means for electromagnetically coupling RF energy into a source region to generate a plasma from the process gas; Means for capacitively coupling RF energy to a substrate support electrode to control a plasma shielding voltage at the support electrode; Provided is an RF plasma processing apparatus that is constructed and operated including a silicon source in contact with the plasma.

The RF antenna means is preferably near the source region and is coupled with the RF energy source.

The chamber of the present invention preferably comprises an RF cathode in the processing region, an anode defined by the chamber wall thereof, and a third electrode grounded or connected to the RF bias or electrically floating for high plasma processing. It has a three-pole vacuum tube array. The chamber wall defining the third electrode and / or source region is also a silicon source for improving processes such as oxide etching.

By using the apparatus of the present invention, the plasma etching becomes almost infinite in selectivity, and the plasma deposition is free from defects.

The present invention seeks to provide an apparatus and process for plasma processing that improves the selectivity and etch rate variability of the etching process and co-deposition, etching and planarization of the deposited layers.

Within the chamber of the present invention, RF power of LF / VHF (low to ultra high frequency) is preferably used in the range of 100 MHz to 100 MHz. More preferably, LF / HF power in the range of 100 MHz to 10 MHz is used. More preferably, MF (intermediate frequency) is used within the range of 300 Hz to 3 MHz. Preferably, the connecting means is a cylindrical coiled antenna in which the multi-winding, i.e. unwound portion is smaller than [lambda] / 4 (where [lambda] is the wavelength of the high frequency RF excitation energy applied to the coiled antenna during plasma operation).

The invention also encompasses load means connected to the antenna for matching the input impedance of the source to the output impedance of the means for supplying RF energy for the antenna, as well as means connected to the antenna for tuning the antenna to the resonant frequency. The tuning means may be a variable capacitance electrically connected between one end of the antenna and the RF ground. The load means can be a variable capacitance electrically connected between the other end of the antenna coil and the RF ground. RF energy may be applied via a tap at a selected location along the coiled antenna.

In another aspect, the system includes an insulating dome or cylinder that forms a generating zone. Preferably, the coiled antenna surrounds the dome for inductively connecting high frequency electromagnetic energy to the vacuum chamber. The item to be manufactured may be located inside the source or dome, in the lower winding or volume of the antenna or in close proximity, or preferably below the antenna.

The invention also provides a means and reaction for injecting a gas into the chamber which constitutes a gas inlet at the top of the dome, a first ring manifold at the bottom of the dome's source zone and a second ring manifold surrounded by the wafer support electrodes. Means for selectively supplying a solvent diluent, surface stabilizer, and other gases to the chamber.

In another aspect, the alternating current power supply and conditioning system connects alternating current bias power of the same or similar frequency to the wafer support cathode as well as the source coil power, thereby irrelevant to the plasma density regulation affected by the source radio frequency power. It affects the cover voltage regulation and ion energy.

The system provides selected bias frequencies to accomplish various purposes.

First, the upper frequency limit is chosen to prevent "current-induced" losses (very high frequencies can cause charging losses in sensitive devices). The lower frequency limit is chosen to rule out "voltage-liquidity" losses. The lower frequency bias can also produce higher wafer coverage voltage (heated to the lower temperature of the substrate) per unit bias power and make a smaller contribution to the plasma density and thus provide better control of ion density and energy. However, too low a bias frequency causes the ions to follow the RF component of the wafer cladding field, thus controlling the ion energy. The result is a higher peak-to-average energy ratio and a wider (double peak) ion distribution. Very low bias frequencies cause charging of the insulator and inhibit ion-inductive treatment during the bias frequency period. Conveniently, the preferred frequency range to satisfy the above considerations coincides with the source frequency range as described above.

The present invention forms a surface stabilizer coating on the first selected material on the wafer to provide a relatively low etch rate and between the individually selected low and high values for selectively etching the second selected material at a relatively high rate and selectivity. And control means for circulating a DC bias voltage.

In another aspect, the chamber is evacuated by a first vacuum pump connected to the chamber and a second vacuum pump connected to the dome, thereby creating a vertical pressure differential across the dome to generate a flow of neutral particles out of the dome. Wherein the voltage at the wafer support electrode is sufficient to overcome the pressure differential such that charged particles flow towards the chamber.

Another aspect of the invention includes a coiled antenna or other shaped Faraday sheath inserted between the chamber and the coiled antenna or other coupling means to prevent the electrical component of the high frequency electromagnetic energy from coupling with the chamber. A fixed high frequency echo surrounding the coil or other coupling mechanism also concentrates the radiation of high frequency energy into the chamber.

Magnetic enhancement is a permanent or electromagnet array of circumferences that applies a distributed and magnetic mirror shape, selected from a controlled static field, uniformity parallel to the axis of the antenna, to control the positioning and transmission of the plasma downward relative to the wafer. Supplied by Magnets are also mounted around the source and / or chamber to apply a plurality of peaks near the wafer to the chamber to confine the plasma to the wafer zone while substantially removing the magnetic field across the wafer. In addition, the magnetic field separator is mounted surrounding the wafer and the wafer support electrode to disperse the magnetic field from the wafer support cathode.

The system architecture allows for a range of magnitudes by selecting an operating frequency while maintaining low mode operation.

In another aspect, the invention provides a vacuum chamber having a source region; Securing the article or substrate to be treated on an electrode in the chamber; Electromagnetically coupling RF energy into the chamber to control a plasma sheath voltage at the substrate support electrode; Capacitively coupling RF energy into the chamber via the support electrode to control an external voltage at the support electrode; And it is embodied by the process for plasma generation including the process of providing the silicon ion source which is in contact with a plasma.

The method also involves automatically and repeatedly tuning the antenna to the resonant frequency and loading the input impedance to the impedance of the antenna RF energy supply.

In another aspect, the present process for generating a plasma provides a vacuum chamber having a source and a process region, an electrode in the process region and a wall with an electrode in the source region; A process region electrode that is a cathode is a cathode, the wall is an anode, and the electrical connection of the source electrode connects an electrode of the process region, which is a wall of a source electrode and a chamber selected from ground, floating and RF or DC bias; Fixing a processed article on the substrate support electrode; Supply process gas to the chamber; Fluidly coupling RF energy into a source region by the antenna to generate a plasma processing one or more materials on the article by using a cylindrical coil antenna coupled to the means for providing RF energy to the antenna; Providing a silicon source within the chamber. The silicon source is, for example, monocrystalline or polycrystalline polysilicon such as silicon carbide, graphite, or the like. Preferably, the silicon source is a third electrode located in the source region of the chamber. However, the silicon source need not be an electrode and is in contact with the plasma elsewhere in the chamber.

In another aspect, the antenna power and bias power supplied to the electrodes are controlled to selectively effect anisotropic, semi-isotropic and isotropic etching.

The process chamber is particularly useful for etching the second material layer, and more particularly for etching oxygen-containing layers such as oxides on substrates that do not contain oxygen such as silicon. The use of a silicon ion source in the process chamber improves selectivity and etch cross section. The process further drives the reverse voltage to a low value selected to form an etch inhibiting layer, such as a polymer or the like, and to a high value for etching the second material layer at a relatively high rate relative to the substrate. In all cases, the device of the present invention provides high etch rate and high selectivity so that the openings formed down to the substrate in the second layer can have a high aspect ratio.

The process of the present invention also applies sputter deposition of an oxide film, such as a silicon oxide film or a silicate, and first applies RF power to the support electrode to a relatively low level to deposit the oxide film, and RF to the support electrode to deposit the silicon oxide film. It involves a two-step process where the power is applied second to a relatively high level.

Special process aspects include, but are not limited to, etching in the oxide film, etching contact holes in the oxide film formed on a substrate that does not contain oxygen, and etching through holes in the oxide film formed on aluminum or another substrate; Etching oxide films with high ratio isotropic and anisotropic; Etching polysilicon conductors such as gates and the like; Remove photoresist; In particular, unitary silicon is anisotropically etched; Anisotropically etch the photoresist; Low pressure plasma deposition of the nitride film and the oxynitride film; Oxide, oxynitride and nitride films are subjected to isotropic isotropic deposition; Etching metals such as aluminum tungsten and titanium and the like and their compounds and alloys such as tungsten silicide and the like; Deposition of various materials at different deposition rates, sputter deposition of local and rounded surfaces, and planarization.

1 through 3 show a plasma reaction chamber system 10 for processing a semiconductor wafer 5 using an inductive plasma source array, a capacitive connection bias array, a magnetically enhanced plasma source array, and other features of the present invention. It is a partial cross-sectional view. The three figures illustrate a preferred and alternative shape of the present system. Three shapes are shown due to paper space limitations. A typical chamber is the thermal insulation of the chamber shown in the concurrently filed CIP application and includes an integrated transmission line structure. The salient features of the present invention are generally applicable to plasma reaction chambers. In addition, it will be understood by those skilled in the art from the following various features of the present invention that cooperatively enhance the performance of the reaction system, which may be used separately or may be selectively omitted from the system. For example, the processing conditions provided by an inductive plasma source array and a capacitively coupled bias source array often eliminate certain conditions suitable for magnetic enhancement.

The illustrated system 10 includes a vacuum chamber housing 11 formed of anodized aluminum and other suitable material and having side walls 12 and upper and lower walls 13 and 14. Anodized aluminum is suitable because it suppresses arc and sputtering. However, other materials may be used, such as bare aluminum with or without polymer or crystal compensable liner or ceramic.

The upper wall 13 includes a lower chamber wafer processing portion 16B formed between both side walls 12-12 and an upper chamber portion 16A sing central hole 15 formed by the dome 17. The dome 17 preferably takes the form of a cup of one or more walls that are formed of an insulating material, such as quartz, or several other insulating materials, such as alumina and alpha alumina (sapphire). In the preferred arrangement shown in FIG. 1, the dome 17 constitutes a cylindrical wall 17H of insulators, such as quartz, and a cover or top 17T, typically of aluminum or anodized aluminum. For processing such as high selective oxidation etching, silicon or the top wall mechanism contained in silicon, and silicon covering the dome sidewalls are preferred.

As shown in FIG. 1, the internal vacuum of the chamber housing 11 (chamber 16) is connected to the bottom wall 14 and to the vacuum line 19 connected to a vacuum pumping system 21 constituting one or more vacuum pumps. Is controlled by a throttle valve 18 (pressure regulation independent of flow rate).

As shown in system 11, the chamber components, including chamber walls and domes, can be heated and / or cooled to enhance processing performance. For example, the dome 17 may be heated or cooled by liquid or gaseous heat transfer means, or the heating component may directly heat the dome 17.

As described in the part 2 and shown in FIG. 2, the process gas, purge gas, diluent, etc. are three manifolds respectively provided at the bottom of the source (dome), the saucepan 177 of the source and the circumference of the wafer. It can be supplied to the chamber by injection sources G ₁ , G ₂ , G ₃ . Gas is supplied to the chamber 11, for example, via a computer controlled flow regulator (not shown), typically from one or more sources of compressed gas.

In the main gas injection manifold G ₁ , as indicated by arrow 22, an internal vacuum processing chamber (through a correction ring gas manifold 51 mounted integrally on the side or top wall of the top wall 13) ( 16) gas enters. The manifold 23 preferably supplies etch gas and / or deposition gas at slightly upward angles to the chambers / chamber portions 16B and 16A to enhance the etching and / or deposition plasma in the application of RF energy. The upper manifold arrangement G _{2 in} the top plate 17T of the dome 17 is used to inject the reactor and other gases into the chamber 16. In addition, a manifold array G ₃ , which is the periphery of the wafer, is provided for supplying reactors and other gases.

RF energy is supplied to the dome 17 by a source that constitutes at least one coiled or coiled antenna 30 that is output by the RF supply and meshes with the network. Antenna 30 preferably has a multi-winding cylindrical shape. Coil 30 forms a minimum conductor electrical length at a given frequency and a given source (coil) diameter and has an electrical length of less than λ / 4 at an operating frequency. By itself, the antenna 30 is not resonant and is tuned to the resonant frequency for effective inductive coupling with the plasma source by the Faraday law of inductive coupling as described in section 5.

Preferably, gas from the chamber source portion 16A flows down towards the processing substrate 5 and is pumped radially outward from the wafer. For this purpose, a circular vacuum manifold 33 is formed around the cathode transmission line structure 32. That is, between one chamber wall 12 and the other external transmission line conductor 320 and between the chamber base wall 14 on the base and the conductive pumping screen 29 on the top. Manifold screen 29 is sandwiched between vacuum manifold 33 and wafer processing chamber 16B and provides a conductive electrical path between chamber wall 12 and outer conductor 320 of transmission line structure 32. The vacuum manifold 33 forms a circular pumping passage for performing uniform radial pumping of the exhaust gas from the circumference of the substrate 5. The exhaust manifold 33 is connected to the exhaust gas system line 19. The gas flow follows the path 22 from the manifold G ₁ to the dome / source and / or follows the path 24 from the manifold G ₂ to the dome / source and / or from the manifold G ₃ . It follows a path 26 that radially flows inward toward the substrate 5. The overall gas flow follows the path 34 from the upper chamber source portion 16A towards the wafer 5, from the wafer along the path 36 and through the screen 29 to the gas evacuation vacuum manifold 33, And a path from the discharge manifold 33 to the discharge system 21. Conductive manifold screen 29 and cathode transmission line structure are optional. Typically at the lower end of the frequency of interest, the wavelength is very long, so the transmission line structure is unnecessary.

This is in contrast to a conventional RF system arrangement, in which the RF output has two electrodes, typically a wafer support electrode 32C, the upper surface of the support electrode supporting the substrate 5 and the side wall of the reaction chamber. (12), between the top wall 13 and / or the second electrode in the manifold.

Specifically, the antenna 30 is located in the plasma chamber 16A for connecting RF electromagnetic energy with the source chamber 16A and the outside and adjacent locations of the dome 17 to induce an electric field in the process gas. Changing the B (magnetic) component of em energy by inductive coupling of Faraday's law energizes the process gas and thus chamber 16 specified by relatively high density and low energy ions (collectively referred to as chamber 16). Plasma is formed in the (16A and 16B) and the plasma). Plasma is generated in the dome 17 concentrated in a small volume formed in the coiled antenna 30.

Active species such as ions, electrons, free radicals and excited neutrons flow downwardly toward the wafer by diffusion and flow in volume due to the obstructive gas flow described herein. Also, as described in Section 7, a suitable magnetic field can extract ions and electrons toward the wafer, as described below. Alternatively, but preferably, the bias voltage input arrangement 41 shown in FIG. 1 purchases the voltage source 42 and the bias matching network 32 and selectively increases the RF sheath voltage on the wafer. In order to connect with the wafer support electrode 32C, ion energy is selectively increased on the wafer.

Reflector 44, which is a substantially open bottom bath, surrounds the top and sidewall portions of the antenna rather than the bottom portion. The reflector prevents RF energy from radiating out and thereby aggregates and efficiently increases the radiation and dispersion of the plasma output. As described in section (7), the Faraday barrier 45 shown in FIG. 3 is located above and below the antenna 30 to allow storage connected to the plasma, but direct electrical field connection is prohibited, and direct electrical field connection Induces gradients or nonuniformities or accelerates charged particles at high energy.

Optionally, as described in Section 8, optionally, one or more of the electromagnets 47-47 or permanent magnets shown in FIG. 2 are mounted adjacent to the chamber sealer 11 to increase the density of the plasma on the wafer 5. To provide ions, transfer ions to the wafer, or improve plasma homogeneity.

As fully described in Section 4, the present invention provides a high power output that potentially reduces the magnetic component of an inductively coupled electromagnetic force, typically characterized by high density and relatively low energy at frequencies less than microwave or microwave-frequency. It is used to induce a circular electric field inside the vacuum chamber to generate RF energy without being connected through the wafer. In the preferred and illustrated downward plasma source arrangement, RF energy is absorbed away from the wafer in a high plasma density state so that electromagnetic waves do not propagate to the wafer and thus minimize the possibility of loss. Alternatively and optionally, RF bias energy is applied to the wafer support electrode 32C to increase the wafer sheath voltage, so that the ion energy is as required.

Chamber 16 according to the present invention is typically about 0.1 mA to about 50 Torr, and for etching, typically about 0.1 mA to about 200 mA is used to deposit and / or etch a semiconductor wafer using the overall chamber pressure. It is possible. The chamber can be operated at pressures up to 5 milliTorr and, in fact, successfully at 2 millitorr pressures. However, it is desirable in some treatments because of higher pumping rates and faster flow rates. For example, a pressure range of about 5 mA to about 50 mT is suitable for oxidized etching. This relatively high pressure requires a source and wafer sawing enclosed space. The chamber can realize the benefits of a very suitable and enclosed space between the wafer and the lower winding of about 5/2 (m / inch) of the antenna, without the loss of charge. I.e. improved etch rate and selectivity, reduced bias voltage and ion energy amount suitable for a given etch rate and improved etch uniformity of the wafer, e.g., the spacing d between the wafer 5 and the source antenna (cm / inch) is itself in close proximity) to 5/2 (cm / inch), reducing the amount of voltage in half and increasing uniformity from about 2.5% to about 1%.

[2. Gas Process System]

As mentioned, the chambers according to the invention are reactive, i.e. for the purpose of improving the quality of the special materials used in the process and in particular processes that meet the requirements of processes (etching, vapor deposition, etc.) where the gas is seen at different locations, It is provided with a plurality of gas injection sources G ₁ , G ₂ , G _3, shown in FIG. ₂ . First, the chamber comprises a standard radiant gas cracking system G _{1 at} the base / base circumference of the source zone 16B. In the presently preferred shape, the G ₁ injection system comprises a quartz gas distribution ring 51 at the end of the source and a circumferential circular manifold 52 that forms a distribution passage for sending gas to the ring.

The ring has an inwardly directed radial hole 53-53 and, preferably, a stepped sintered ceramic pore gas injection plug 54-54 inserted into the hole to prevent cavity discharge.

The second gas injection array (G ₂ ) is made of the same material as anodized aluminum and has a top of a grounded, floating or biased dome with a central gas injection hole 56 filled with a porous ceramic injection disk 57. The plate 17T is included.

The third gas injection source G ₃ is a ring-shaped gas injection manifold 58 mounted on the circumference of the wafer 5 (or a coupling ring used to support the wafer in place relative to the support (not shown)). And a gas injection unit incorporated in the.

In etching monolithic or polycrystalline silicon, and opening the opening through an oxide-containing layer such as silicon dioxide, on a non-oxygen-containing surface, the silicon dioxide is preferably etched at a faster rate than polycrystalline silicon or other substrates. do. Fluoride containing etching gas is used as an etching liquid. However, since the fluoride etchant usually etches materials such as silicon dioxide, polysilicon, and the like, for example, at the same rate, the etching is almost non-selective to the oxide film rather than silicon. However, during reactive ion etching, the protective polymer layer is formed on the sidewalls and the bottom of the growth openings; Such polymers are formed of carbon and fluorine, and generally contain about 30% carbon and about 60% fluorine.

However, this polymer decomposes in the presence of fluorine atoms. What is desired is to form a carbon rich polymer containing about 50% carbon and about 40% or less fluorine; If a fluorine scavenger is present in the reactor, little free fluorine is present in the plasma and little CF bond is formed in the polymer film. In known reactors, this has been achieved by adding a scavenger for fluorine ions into the plasma itself, such as silane or organosilane, to provide free silicon in the plasma. However, because so many different plasma ions are formed in this way, the process is essentially “dirty” and the nature of the etchant and polymer is varied and uneven, and the process is not entirely renewable. Thus, the present invention provides an apparatus for providing a "clean" silicon source for cleaning free fluorine ions.

According to the invention, such a fluorine scavenger is a reactant for fluorine, which is preferably a silicon source near or within the plasma.

The fluorine stabilizer is pure silicon, such as monocrystalline silicon or polysilicon, or a silicon source therein. The fluorine scavenger is pure silicon such as monocrystalline silicon or polysilicon, or a compound such as silicon carbide, or preferably the third electrode may be made of silicon or silicon containing material. Graphite is also used to clean (scavenges) fluorides; For example, the third electrode is made of graphite.

During plasma etching in the chamber of the present invention, the oxygen present in the silicon oxide film etches the polymer formed on the sidewalls and bottom of the growing trench. However, when the trench depth reaches a polysilicon or other oxygen-containing substrate, no oxygen is present and the polymer remains on the polysilicon surface, preventing further etching.

Preferred etchant here are fluorocarbons such as CF ₄ , C ₂ F ₆ and C ₃ F ₈ , which generate only carbon ions and fluorine ions. Other known fluoride etchant such as CHF ₃ and the like are undesirable because they are generated by adding hydrogen ions to carbon ions and pfuorions.

Thus, according to the present invention, since the carbon-rich polymer film does not decompose on the silicon substrate or the oxygen free substrate or the oxygen free layer, very high selectivity between the oxygen containing layer and the oxygen free substrate or underlayer is achieved to almost infinity. Will be. This polymer is oxygen sensitive and, if no oxygen is present, such as when the etchant reaches the oxygen free substrate, the degradation of the polymer is reduced, and a reduced number of plasma ions protect the underlying layer when bonded to the plasma. A relatively inert polymer surface stabilizer layer is formed.

Preferably the silicon source is located near where the plasma is generated, so that the silicon can clean them when fluorine ions are formed, and the fluorine ions will hardly react with the treated substrate surface. For example, the silicon network is suspended in the plasma region, or silicon is located near the top or wall of the reactor as part of the rf source. The silicon source may be suspended near the substrate surface. However, the polymer will have a higher fluorine content.

The silicon source is also located outside the plasma region of the reactor, in which case it is heated to a temperature, for example at least 150 °, to form silicon ions to clean the fluoride ions. In that case, means for regulating the temperature of the silicon source must be provided in the reaction chamber.

Further advantages are achieved if the reactor of the present invention is to be applied as a deposition chamber for depositing various films. For example, after a trench or opening is fabricated in a substrate, the opening or trench is filled with another material to form conductors in the substrate or to form isolation between devices. For example, openings in the substrate are filled with silicon oxide using silane and oxygen. This makes it necessary to carefully control the deposition rate to avoid the formation of voids at the bottom of the trench or at the center of the deposition. This latter phenomenon is known and has occurred in connection with the ECR process. In order to overcome this problem, there is a known method of sputtering the top of the trench with argon prior to the deposition process, ie, shaping the top surface of the trench or opening, whereby the top of the trench is opened so that the void perimeter It will be charged without clogging. However, the ECR process is disadvantageous because very low pressures of about 1 to 2 millitorr are used to maintain high ion density. In ECR, ion density depends a lot on pressure; It peaks at about 1 to 2 millitorr and descends rapidly at 5 to 10 millitorr. Thus, a large amount of gas is supplied to the ECR, so a large vacuum pump is necessary to exhaust the 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 millitorr. Thus a higher pressure, i.e., about 15-30 millitorr, can be maintained in the present inductively coupled plasma, and, incidentally, less gas needs to be supplied to the reactor. Smaller vacuum pumps are needed to exhaust the by-products, and excess gas is required for ECR deposition than 1-2 millitorr operating pressure.

Thus, the device of the present invention can be charged in the simpler and cheaper way with the same result as those obtained in the ECR. Because of the low temperature operation and ability to regulate the ion density of the plasma produced at low pressure, similar processes can be carried out in the apparatus of the present invention without a large amount of gas or a large vacuum pump. Thus, for example, introducing an appropriate amount of silane, oxygen, or argon into the reactor can form a plasma that sputters simultaneously from the top side walls of the trench and fills the trench with a silicon oxide film. The voids in the silicon oxide film deposition are not formed, and the deposition or filling of the top of the trench and the trench is formed in a single process.

As noted above, the various types of gases selected for caustic and deposition, namely surface stabilization, diluent gases, etc., may be passed through one or more sources G ₁ to G ₃ to meet the needs of particular etching and deposition processes and materials. It can be supplied to the chamber.

For example, the present flowable source antenna 30 is very efficient at supplying a very high density plasma and decomposing gas into the source region 16A of the chamber dome. As a result, when the polymer forming agent is supplied to the dome via G ₁ or G ₂ , it does not adhere to the polysilicon surface that is coated and / or protectively coated instead of coating polysilicon inside the highly degradable dome. It can be completely disintegrated. Solution C ₂ F _6, or is introduced into the source region (16A) via the etchant species G ₁ or via the G ₂ or G ₁ and G _3, such as the DF _4, optionally, polymer carbon is pungbun on the substrate It provides silicon to clean the fluoride ions to form.

Because of the high degree of decomposition of the gas in the source region, fluorine-containing gases (even those combined with carbon in fluorine) typically produce free fluorine to etch silicon and thus reduce the etching choice for oxidation. In addition to providing a silicon source to the plasma, when high selectivity is required, a silicon containing additive gas is injected to further clean free fluorine ions and reduce the etching of the oxygen ratio containing substrate. The corrosive gas and the silicon-containing additive gas can be introduced separately via G ₁ or G ₂ or through G ₁ and / or G ₂ in a mixed state. Suitable fluorine consumable silicon containing additive gases include silane (SiH ₄ ), TeO ₃ , dimethylsilane and silicontetrafluor (SiF ₄ ).

Fluorine consumable and polymer forming additive gases may be used together in the same process to improve etch selectivity.

[3. Differential pump process]

2 shows an optional vacuum pumping shape. In addition to the vacuum pumping system 21 connected to or near the bottom of the chamber, the vacuum pump 39 is connected to the source zone 16A inside the dome 17 via a line 38. The flow rates of the pumping systems 39 and 21 depend on the pressure (1) and electrons and ions for the transfer of the vertical pressure difference ΔP _P ie charged particles from the source 16A to the wafer 5 across the source region 16B. It is selected to produce a pressure 2 difference of magnitude smaller than the force F _b exerted by the bias voltage of the same charged particle. As a result of the pressure difference ΔP _P , uncharged particles such as radicals do not reach the wafer 5, but rather flow out of the upper vacuum chamber connection 38. As a result of F _DC · ΔΔP _P ·, charged electrons and ions flow into the treatment zone. This approach is useful and evident for selectively retaining radicals, not ions, out of the wafer processing zone. Such a situation occurs, for example, during etching using polymer forming gas chemistry. However, the polymer is formed in the source region that adheres to the chamber sidewalls and / or does not adhere to the desired wafer surface. And / or fluorine radicals are formed in the source region.

[4. RF power, antenna and bias source]

1) top or antenna source

Referring to FIG. 1, preferably, the operating frequency of the RF power supply 31 suitable for the upper end 30 provides a high density plasma, reduces losses to sensitive devices, and efficiently inductively connects the RF output to the plasma. Is selected to provide. In particular, the upper frequencies of the operating range are limited to minimize circuit-induced losses. The lower limit of the operating frequency is suitably selected for the efficiency of the RF output coupled with the plasma. Preferably, an alternating current output in the range of about 100 Hz to 100 MHz of LF / VHF (low frequency to ultrahigh frequency) is used. More preferably an LF / VHF (low to high frequency) output in the range of 100 Hz to 10 MHz is used. More preferably an MF (intermediate frequency) output of 300 Hz to 3 MHz is used.

2) bottom or bias source

An AC output supply 42 for the wafer support cathode 32C capacitively couples the RF output to the plasma, thereby effectively controlling various factors including cathode coating voltage and ion energy, and cathodic coating voltage and ion energy. Is a controlled independent variable of plasma density control effected by high frequency output. The bias frequency is selected to accomplish a number of purposes. First, the upper frequency limit is chosen to prevent current-induced charging losses in sensitive devices. The lower frequency is chosen to rule out voltage induced losses. Lower frequency bias also produces higher wafer coverage voltages per unit bias output (lower heating) of the substrate and makes a lower contribution to plasma density, thus making a lower contribution to independent regulation of ion density and energy. However, too low a bias frequency allows the ions to follow the RF component of the coated electrons of the wafer, thereby modulating the ion energy. The result will be higher peak to average energy ratios and wider (peak to peak) ion energy distributions. Very low bias frequencies result in insulating charges that inhibit the ion induction process during bias frequency adjustment. Conveniently, it has been found that the above consideration can be satisfied using a bias frequency range that matches the source frequency range as described above.

3) Combined operation of antenna and bias source

A preferred feature of the present invention is to automatically change the base or bias output provided by the output supply 42 to maintain a constant cathode (wafer) covering voltage. At low pressures (<500 mt) in a highly asymmetric system, the direct current bias measured at the cathode 32C is very close to the cathode covering voltage.

The base output can be changed automatically to maintain a constant direct current bias. The base or bias outputs have a fairly small effect on plasma density and ion current density. The top and antenna outputs have a strong effect on plasma density and current density, but have a fairly small effect on the cathode covering voltage.

Since the radio frequency of the source 31 driving the antenna 30 is much smaller than the frequency used for microwave or microwave-ECR applications, the selective small magnets operated by low output currents at low DC currents are small Can be used with heating loads. In addition, as is apparent from the above description, a coaxial cable such as 31C can be used in place of the wave guide.

In addition, the plasma non-uniformity caused by EXB electron drift in other magnetically enhanced or assisted systems here means that the added magnetic field (the magnetic component of the HF field applied through antenna 30 and any magnetic field applied by magnet 81) It is absent because it is substantially parallel to the electric field at the cathode. So there is no EXB drift current in the system.

The self-classification path formed in the permanent magnet can be used to allow B fields in the source (upper chamber 16A) rather than in the substrate.

Optionally, permanent magnets or electromagnets may be placed on the source and / or chamber walls, with a multipolar arrangement around the lower chamber 16B, typically replacing the poles with a north-south-north-south --- north-south arrangement. It is arranged to form a plurality of peak magnetic mirrors. The magnet is, for example, perpendicular to a bar magnet or preferably a horizontal circular magnet. These magnets are used to reduce the number of electrons lost in the wall to improve plasma density and uniformity without requiring wafers in the magnetic field.

4) Combination and Synchronization of RF Sources

As above, both the preferred operating frequency of the top or antenna RF source and the preferred operating frequency of the base or bias RF source fall within the same range. One optional feature is to combine these two RF sources into one single source instead of using two separate sources. More generally, the possibility is to supply all three RF signals (including RF bias to the third or upper electrode) from a single source, or one source suitable for the antenna and base bias and a second source suitable for the third electrode. Use; Or using three separate sources. As separate sources are used, an additional consideration is whether the separate RF signals are the same frequency, and if so, are they combined at any required correlation. Preliminary experiments indicate that the answer to this question depends primarily on the operating frequency chosen. If a single frequency can be selected to suit two or three of the RF sources, and the frequency is not likely to be easily converted for the other process in which the system is used, then a single RF source is a logical choice. However, if other frequencies are needed for the source, or based on the considerations discussed in sections 1 to 3 above, or if the frequencies need to be converted for use in other processes, then separate RF sources will be needed. If there are separate sources and the same frequency is selected, phase looking is not a problem except for relatively low frequencies, such as several hundred kilohertz or less. For example, the source is matched such that the phase angle between the RF voltage injected into the antenna and the RF voltage injected into the base or wafer electrode is maintained at a constant value that appropriately selects the process repetition. At higher frequencies, the operation appears to be independent of phase or frequency locking.

[5. Antenna tuning and load]

1) Synchronization

Typically, the antenna 30 converts the frequency of the generator 31 to synchronize with the antenna; Or to resonate by a separate resonant element connected to the antenna to tune the resonance. For example, this tuning element can be variable inductive ground or variable capacitive ground.

It should be noted that inductive and capacitive tuning reduce the resonance frequency.

As a result, when capacitive or inductive tuning uses a toilet, it is required to tailor the system to the higher required resonant frequencies to compensate for the reduction in resonant frequencies.

Autotuning is preferred and can be implemented by using an impedance phase / magnitude detector to drive the tuning / load variable. Referring to Figures 16 and 9, an optionally reflected output bridge or VSWR bridge is used to drive the tuning and load variable, but repetition is also required.

2) load

Conductive, capacitive or inductive load means L can be used to match the generating antenna 30 with the impedance of the RF generator 31 and the coaxial cable 31C to which it is connected. For example, a tap or wiper may be ohmically connected to the output impedance location of another generator at or along 50 ohms or 300 ohms. Optionally, the variable derivative or variable capacitor can be connected to the generator output impedance point 50 on the antenna.

3) Tuning and load circuit

4 and 9, it is provided that the tuning means T is integrated in the generating antenna 30 for tuning in order to resonate the source. An integrated load means L is also provided for coupling the input impedance of the generating antenna 30 to the combined output generator 30 (or transmission line 31C). Referring to FIG. 4, on the one hand, tuning The means T is a variable capacitor electrically connected between one end of the antenna 30 and RF ground.

As shown in FIG. 5, in other respects, the load means L is a variable capacitor electrically connected between one end of the antenna 30 and RF ground. The load means may also be a variable position tap 60 for applying the RF input to the antenna. See also FIG.

In the preferred combination shown in FIG. 7, the tuning means T is a variable capacitor electrically connected between one end of the antenna 30 and RF ground and the load means L is between the other end of the antenna and RF ground. Another variable capacitor is electrically connected. In this arrangement, the RF input can be applied to the antenna via a tap, i.e., a tap applied along or at either end of the antenna. Referring to FIG. 8, optionally, the RF input connection 66 may be positioned substantially at the connection, as shown in FIG. 9, between the variable capacitive L of the load and the end of the antenna 30. Can be.

[6. Source / Bias Process Control]

The present invention also uses a sufficiently high bias voltage to periodically increase the etching rate of the material, such as silicon dioxide, and the selectivity of the etching of the silicon dioxide with respect to materials such as silicon, and to provide a high silicon dioxide etching rate. Increased by pulsing to a lower value.

1) Pulsed / Modulated High Bias Etch Rate and Selectivity

Referring to FIG. 10, the etch rate of a material, typically silicon dioxide (SiO ₂ ), is increased with the bias voltage. Thus, increasing the bias voltage increases the etching rate of the oxide. Unfortunately, however, the etch rate of materials such as silicon / polysilicon bonded to the integrated circuit structure increases with the bias voltage. Thus, using a bias voltage of sufficient magnitude to provide a very high silicon dioxide etch rate is also effective at unnecessarily high silicon etch rate (although lower than oxide etch rate) and reduces selectivity.

Quite clearly, when etching silicon dioxide, it is required to have a high oxide etch rate characteristic of a high direct current bias voltage (V _h ) combined with a relatively low silicon etch rate characteristic of a low direct current bias voltage (V ₁ ). And thus high oxide selectivity is required.

Referring to the DC bias voltage waveform 70 shown in FIG. 11, the hypothesized opposite goal represented in the previous paragraph combining the V _h and V ₁ characteristics results in a substantially high baseline DC bias voltage V _h and voltage. By pulsing or modulating periodically to a low value (V ₁ ), a polymer forming etching process (a process of forming an etching dominant polymer in a material such as silicon) is performed. V ₁ is at or below the cross point / voltage 68, or between the silicon etch and silicon deposition shown in FIG. 10, while the oxide is at or above the cross point / voltage 69. As a result, the protective polymer is deposited on silicon to squeeze the etch while recovering at high etch voltage V _h , but there is no or insufficient deposition on the oxide to sufficiently squeeze the etch of the oxide at V _h . Preferably, V ₁ is characterized by deposition on the polymer, but at least by a slight etching of the oxide. In a preferred embodiment of the invention, the parameters ie V _h (high DC bias voltage), V ₁ (low DC bias voltage), P _W (pulse width of low voltage V ₁ ), P _rp (low voltage) And the pulse repetition rate or combined width of the high voltage pulse) are -400 V, -225 V, about 0.1 second and about 1 second, respectively.

2) dual frequency bias

An alternative approach is the DC bias voltage wave form 7 shown in FIG.

Relatively low frequency voltage variables are superimposed on the fundamental bias voltage frequency. For example, the low frequency (T ₂ ≤ ₂₅ kHz) (preferably 5-10 kHz) is superimposed or mixed with the base radio frequency (T ₁ ≤ ₂ kHz). Silicon oxide is an insulator; Typically silicon / polysilicon has only a very thin negative oxide layer. Thus, the low frequency (T ₂ ) direct current bias voltage variable does not appear on the oxide surface because it is charged. However, essentially non-insulated polysilicon responds to the low frequency (T ₂ ) in a manner similar to that previously described by forming a protective layer during the low voltage circuit 72 (V ₁ ) of the low frequency (T ₂ ) cycle. do. This low frequency shaping layer is etched during the variable high voltage skew 73 of the high frequency T ₁ cycle. As mentioned, the non-insulating nature of silicon dioxide prevents the etch from interfering with deposition during the low voltage application of T ₂ and the oxide etch proceeds without decay during the high voltage of the T ₁ cycle.

In brief, the protective layer is silicon during the low voltage phase 72 of the low frequency cycle T ₂ , which suppresses silicon etching during the high voltage phase 73 of the high frequency cycle T ₁ , which quickly etches oxides without deposition inhibition. Is formed on the phase. The result is a high silicon oxide etch rate, a relatively low overall silicon etch rate and a high etch selection for oxide, similar to the results of the pulse / modulating approach. Note that the pulse / modulating approach is presently preferred for the dual frequency bias approach because of the precisely controlled nature of the previous approach.

[7. Faraday Cloth]

Consider the characteristics of a typical coiled antenna 30 having a load capacitor L at the injection end and a tuning capacitor T at the far end and a much higher voltage at the lower end and farther end at the input end. At the bottom base coil winding is connected with a low voltage RF input. Typically the plasma is applied by an electrostatic field coupled with a relatively high voltage winding near the forging of the body and the voltage winding starts the plasma by electrostatically initiating the collapse of the gas. After the initial collapse, the coupling to the plasma is mainly electromagnetic, or inductive. This operation is well known. Under steady-state conditions, electrostatic and electromagnetically inductive connections are typically present. Although electromagnetic connections dominate, some processes are sensitive to electrostatic fields. For example, etching of polysilicon requires low energy particles and low energy impact to avoid etching other oxides.

1 and 15, the chamber optionally incorporates a Faraday sheath 45 to reduce the steady state electrostatic field. In one embodiment, shown in FIG. 15 (a), the structure includes a cylindrical arrangement of grounded spaces, “single” surrounding the dome wall 17W and the antenna 30, extending axially the posts or bars. Faraday coating 45S. Single sheaths vary from the shape of large spaces to shapes with very small gaps between the sheaths.

Figure 15 (b) shows a so-called "complete" Faraday coating 45F with a pair of rigid covering spaces so that one bar overlaps the other gap as well. This excludes the line of observation passages suitable for the electric field line through the sheath, thereby eliminating the electrostatic field.

Although various shapes of Faraday sheaths 45S and 45F are possible, the presently preferred shape is a cylindrical shape, which is a vertical cross-sectional view shown in FIG. 1, which is flanged outward, electrically conductive, and an open end. Single or double-walled perforated surfaces 46, 47, and 48 may be provided around the top, inside (source) and base of the antenna, while the ground plane 49 (which may be solid) is located on the outside of the antenna. Expand. This shape allows the axial magnetic component of the em wave at the antenna 30 to induce a plasma that is within the plane of the antenna and consequently a parallel closed loop electric field. However, sheath 45 prevents the direct field component of high-frequency electromagnetic energy from connecting to the plasma, capacitively aside the field component in order to ground. Without the sheath 45, the voltage transformed along the antenna is connected to the plasma in accordance with the Maxwell equation suitable for capacitive displacement current coupling. This can lead to inhomogeneities and gradients in plasma density and energy across the wafer 5, resulting in process inhomogeneities and high energy charged particles. Faraday's law, expressed in an integrated form, requires that changes in the magnetic field throughout the surface create an electric field enclosed on that surface. Maxwell's equations describing phenomena in particular shapes say that the induced vortex of the electric field is proportional to the rate of negative change of the magnetic field. In the case of sine wave excitation, the vortex of induced E is proportional to the peak amplitude as well as the radiation frequency of the changing B field.

In simple terms, discontinuous, crevices or separate Faraday coatings minimize the effect of shortening the coating on the em field across the coil, reducing eddy current losses, and allowing the connection of radio frequencies. That is, the rim is axially oriented to the plasma to induce a closed loop electric field that generates the plasma.

However, direct connection of the electric field (which changes along the antenna) to the plasma is excluded, thereby precluding any combined loss of plasma non-uniformity and non-uniform treatment suitable for high energy charged particles.

[8. Magnetic field limitation and strengthening]

1) Limited

In order to eliminate losses (reduced plasma density) in the wall portion 17W of the cylindrical / domed source, the magnet arrangement provides for creating a circumferential annular (thin) field. In a preferred arrangement, ie in the horizontal configuration shown in FIG. 13, this magnetic field is caused by a closely spaced “bucket” or cylindrical multipole arrangement of axial permanent magnets or electromagnets 76-76. Provided, the magnets are magnetized across a small range to form each hermetically sealed and optional pole, circumferential-NSNS- magnetic field (B).

The multipole arrangement produces a plurality of cusp magnetic mirrors 77 in the dome wall portion. Optionally, the arrangement may be a horizontal ring magnet. This magnet induces electron reduction of the wall 17W, thus improving plasma density and uniformity without exposing the wafer to the magnetic field.

Optionally and similarly, permanent magnets or electromagnets are installed in a multipolar arrangement around the lower chamber 16A, typically alternately arranged in north-south-north-south --- north-south, and multiple chamber walls. Create a magnetic mirror with tips. The magnet may for example be a vertical bar magnet or preferably a horizontal ring magnet. Such magnets can be used to reduce electron loss in the wall, thus improving plasma density and homogeneity without exposing the wafer to magnetic fields. In addition, the radial arrangement of the magnets is mounted on the top plate 17T of the top or cylindrical source of the dome to reduce losses at the top.

Referring to FIG. 3, on the other hand, the plasma in the substrate treatment zone 16B is separated from the plasma in the generation or source zone by installing a general parallel grid of magnets at the base / top of the treatment zone. The magnet grid is closely magnetized to NS across a small scale to provide a -NS-NS-NS magnetic field that is a planar array with magnetic lines oriented on one side of the rod and ending on the other, such as the Bercat arrangement discussed above. It constitutes a generally parallel bar magnet (78-78) with a spacing. The generally planar magnetic filter 79 produced across the opening 15 of the source defines a magnetic field in the plane / area of the plate and does not penetrate into the source or wafer processing zone.

By the formula F = qV × B, the high energy / high velocity electrons in the source are bent or exhaled into a larger area by this magnetic field 79 than in the case of ions and cannot penetrate into the substrate processing zone. This reduces the high energy electron density in the treatment zone 16B and the plasma density in that zone. Treatment zones and source zones are separated.

This filter self-limiting approach is particularly useful for separating plasma zones in dense systems. That is, for example, it is used to provide radial high density without high density of ions in the substrate while maintaining compactness. In contrast, conventional approaches require increasing the distance between the substrate and the source at the expense of compactness.

In one preferred arrangement, the filter magnet confinement is mounted to a machined aluminum plate with a bar hole magnet for cooling and a long thin magnet.

Bucket self-limiting arrangements and filtration self-limiting arrangements can be used together.

2) strengthen

As noted above, one or more (preferably at least two) permanent magnets or electromagnets 81-81 shown in FIG. 3 are induced by the RF plane frequency radiated from the antenna and the vertical plane of the coiled antenna. It can be used to form a static and generally magnetic field axis perpendicular to the full length. Preferably, as in the following, one of three magnetic field forms is used: homogeneous, divergent or magnetic mirror.

Referring to FIG. 14 (a), a homogeneous and uniform magnetic field axis 82 acting perpendicular to the wafer 5 by magnets 81-81 restricts the driving of electrons to the wall. Because of the ions that are not subject to high frequency field conversion, the ions follow electron deprivation and are concentrated in the plasma on the wafer. For maximum effect, this and other static magnetic fields can be tuned to resonate with high frequency electromagnetic fields, Ω = 2πF = Be / m, where B is the magnetic flux density and e and m are the charge and the weight, respectively.

Emitted storage 83 is shown in Figure 14 (b). By preserving the magnetic kinetic energy, the slope of the magnetic axis tends to convert the circular transmission energy into the transmission energy of the axis and tends to drive electrons and ions from the stronger magnetic field to the weaker magnetic field. The dispersed magnetic field can be used to push electrons and ions out of the plasma generation zone and to concentrate the plasma on the wafer.

Referring to FIGS. 14 (c) and 14 (d), the bulging or aiding magnetic field 84 (see also fifteen (c)) and the tip or acting magnetic field 85 (fifteen (d) Respectively). Each effect of these so-called “magnetic mirrors” is similar to the effect of the axis's divergent magnetic field, ie, charged particles are driven towards a central region, which is a relatively weak magnetic field in a relatively strong magnetic field (t end here).

Selectively positioning magnets and selecting and converting the strength of the magnetic field formed by a single magnet or coupled magnets creates a uniform, divergence, or magnetic mirror field coupled in a limited way to increase the density of the dolasma on the wafer. . In the case of a magnetic mirror field, a preferred wafer location suitable for maximum plasma density improvement is in close proximity to the bulge or tip, providing maximum plasma density improvement.

The magnetic field axis is used in the volume of the antenna to improve plasma generation, but is also preferably used to remove the magnetic field from the wafer. An annular disk of high magnetic permeability material (such as nickel or steel for soft iron) can be fitted over the wafer 5 and below the plane of the magnet and antenna.

3) extraction

Appropriate magnetic fields can be used to extract ions and electrons towards the wafer.

[9. Control system]

The following description is used herein with reference to the control system shown in FIG.

FIG. 16 is a block diagram illustrating an example system for controlling various components, including an output supply. Here, the system regulator 86 includes an antenna output supply 31, an impedance bridge 87, an antenna 30, a bias output supply 42, an impedance bridge 88, a matching network 43, and It is connected to the cathode (32). Process antenna output parameters and direct current bias, i.e., those selected for ion flux density and ion energy, are supplied as injected into the regulator 86. The regulator 86 also controls parameters such as gas flow, chamber pressure, electrode or wafer temperature, chamber temperature, and others. The regulator can be scheduled by signaling the initial air conditioning (1) and load (1) conditions on the Tsp1 and Lsp1 lines connected to the antenna 30. The regulator can also be scheduled by signaling the initial air conditioning 2 and load 2 conditions on the Tsp ₂ and Lsp ₂ lines connected to the matching network 43. Typically, these conditions are chosen to optimize the plasma initiation (gas separation) conditions. The output may first be applied to the antenna 30 or the cathode 32 or may be applied to both places at the same time.

The regulator 86 outputs the output to determine the point on Psp ₁ to the antenna output supply 31 and the Psp ₂ to the bias power supply 42 simultaneously or continuously (in any order).

Avalanche breakdown occurs rapidly in the gaseous state, producing a plasma. The regulator 86 adjusts the front output P _f1 echo output P _r1 towards / from the antenna and adjusts the front output P _f2 and echo output P _r2 towards / from the cathode 52. The direct current bias current (negative-to-anode direct current voltage) is also controlled as shown by the regulator 86. The regulator sets the coiled (1) tuner and load (1) parameters to either (a) the forward output (P _f1 ) and the echo output (P _r2 ) or (b) the magnitude of the impedance of Z ₁ | It is controlled by diverging certain points on the Tsp ₁ and Lsp ₁ lines based on this. The bridge 87 supplies the impedance magnitude and phase angle information to the regulator. Antenna 30 has echo output P _f1 substantially zero and impedance (magnitude and phase | Z ₁ | <phi) is coiled power supply output impedance (zero echo power condition and complex impedance power are minimized or As a result, it can be connected: Optionally, it is a complex number of VSWR (voltage sustain wave ratio) or echo constant can be minimized). The regulator 86 adjusts the cathode 32 and adjusts the matching network 43 tuner 2 and load (2) parameters to (a) the forward power (P _f2 ) and the echo power (P _r2 ) or (b) impedance. Adjust by diverging fixed points on Tsp ₂ and Lsp ₂ lines based on magnitude (│Z ₂ │) and impedance phase below phi ₂ . Bridge 88 provides impedance magnitude (│Z ₂ │) and phase (<phi ₂ ) information to regulator 86. The matching occurs, similar to antenna matching, when the echo power P _r2 is substantially zero and the impedance (magnitude Z ₂ ) and the phase is less than phi ₂ is the output impedance of the bias power supply 504. The direct current bias is regulated by a regulator, and the regulator 86 converts the output power of the bias power supply to obtain the desired degree of direct current bias. The regulator 86 subtracts the measured value of the direct current bias from the desired amount of direct current bias. If the difference is negative, the output of the bias power supply is increased. If the difference is positive, the output of the bias power supply is reduced. (The output of the higher bias power supply produces a more negative DC bias.) Proportionally-cumulative, or proportionally-cumulatively-inductive controller or other controller is used in accordance with the present method.

Optionally, instead of the preferred embodiment of adjusting the output of the bias power supply 42 to maintain a constant direct current bias, a constant bias power supply output may be used.

In addition to the DC bias servo-matching technique described above, the operation tuning method can also be performed by servoing to a peak-to-peak RF voltage. The latter method is advantageous for certain etching treatment methods that require sufficient conductive surface area for the cathode and anode, for example, to supply current for driving the device. The use of polymer coating technology prevents current from passiveizing this inductive area and satisfying the device and obtaining correct readings. In contrast, the peak-to-peak RF voltage method is ineffective and ineffective at low frequencies in the particularly preferred frequency range. The measurement can be made where the matching network 43 is closer to the chamber than to the cathode.

The controller 86 may be a central controller or a distributed system of controllers.

Turn on / turn off continuity is important for sensitive wafer device structures. It is generally desirable to turn the source on first and off last.

This is because the cover voltage change is minimized in this way. For some applications it is desirable to turn the bias on first.

[10. Transmission line structure]

U.S. Patent No. 599,947, Appropriate Coaxial / Transmission Line Design, which requires injection into a wafer in a low characteristic impedance matching network through a return passage along a short transmission line and a transmission line, is described in detail as a corresponding application. This design requirement is satisfied by the integrated transmission line structure shown in FIG. 1, which constitutes the cathode 32C, the concentrated annular conductor 320, and the non-porous low loss insulator 32I, wherein the insulator 32I is a cathode. Surround the 32C and insulate the cathode from the concentrated annular conductor 320 and install a process gas that may yield. For example, Teflon manufactured or modified materials are preferred because they have high polarization strength, low polarization constant and low loss. The input side of this structure is connected to the matching network in the following manner. The insulated cathode 32c and the outer conductor 320 provide a separate current path between the matching network 43 and the plasma 16. One reversible current path is from the matching network along the outside of the cathode 32C to the plasma coating at the chamber (electrode) surface. The second reversible passageway is from the plasma to the matching network along the upper inner surface of the chamber wall 12 and along the conductive discharge manifold screen 29 and via the inside of the outer conductor 320. Note that the exhaust manifold screen 29 is part of a uniform radial gas pumping system and a return passage for RF current.

During the application of alternating energy, the RF current path reciprocates between the right direction and the reverse direction. Because of the coaxial cable type of the construction of the transmission line structure 32, in particular, due to the higher internal impedance of the cathode 32C (to the outside) and the higher impedance towards the outside of the conductor 320 (to the inside), The RF current is applied to the outer surface of the cathode 32C and to the inner surface of the outer conductor 320 in a coaxial transmission line manner. Surface effects concentrate RF current near the surface of the transmission line, reducing the effective cross section of the current path. The use of large wafers, for example wafers with a diameter of 4 to 8 inches, and complementarily the use of large diameter cathodes 32C and large diameter outer conductors 320 is a large cross section, low impedance, efficient along the transmission line structure. Provide a current path.

Also, if the coaxial transmission line structure ending at pure resistance is equal to the characteristic impedance (Z _O ), the matching network shows a constant impedance (Z _O ). That is, it is independent of the length of the transmission line. However, this is not the case here. This is because the plasma operates over a range of pressures and powers, and constitutes another gas, which collectively converts the load impedance Z ₁ that the plasma provides at the end of the transmission line 32. Since the load Z ₁ deviates from the non-ideal (i.e., no loss) transmission line 32, standing waves increase the resistive and polarized losses between the transmission line and the matching network 43. . Although the matching network 43 can be used to remove any standing waves and loss occurs continuously from the injection end of the matching network to the amplifier or power supply 42, the matching network, transmission line input 32 and inside the chamber Polazma constitutes a resonance system that increases the resistive and polarized losses between the transmission line 32 and the matching network 43. In short, the load impedance Z ₁ deviates from the loss, but the loss becomes minimum when Z ₁ to Z ₀ .

In order to reduce losses due to mismatches in load, the coaxial transmission line structure 32 is designed to have a characteristic impedance Z _{0 that} is most appropriate for the range of load impedances combined with plasma operation. Typically, for the above operating parameters (eg bias frequency in the range of about 0.3 to 3 MHz) and the material of interest, the series of equivalent RC load impedances Z ₁ imparted to the transmission line by the plasma is about 10 ohms. Resistance values in the range of from 100 ohms and capacities in the range of about 50 picoprads to about 400 picoprads. In conclusion, as appropriate, the transmission line characteristic impedance Z ₀ is selected to be centered within the load impedance range of about 30 to 50 ohms.

The transmission line 32 needs to be very short in order to avoid conversion of the plasma impedance encountered with the matching network. Preferably, the transmission line is smaller than lambda / 4, more preferably about (0.05 to 0.1) x lambda value.

Also, for efficient connection of power, the inner diameter (cross section) of the return conductor 320 should not be significantly larger than the outer diameter (cross section) of the center conductor 32C.

In short, the chamber incorporates a transmission line structure that connects power from the matching network 31 to the plasma. The transmission line structure is preferably (1) likewise much shorter than [lambda] / 4 at the frequency of interest, or, optionally, almost equal to [lambda] / 2 integer value, in order to prevent unnecessary transmission of plasma impedance, between the plasma and the matching network. With a characteristic frequency (Z ₀ ) selected to surpass the loss due to the presence of standing waves on the line at; (3) Use the cross section of the outer conductor passageway that is not substantially larger than the cross section of the center conductor.

[11. Chamber temperature control]

Temperature controllers that can be incorporated within the process chamber system 10 include the use of a fluid thermal conductor to maintain the internal and / or external temperature of the gas injection manifold above or below a constant value or within a range; Resistance heat of the cathode 32C; Fluid heat transfer heating or cooling of the cathode 32C; The use of gas thermal conductors between the wafer 15 and the cathode 32C; The use of fluid thermal conductors to heat or cool the chamber wall portions 12-14 and / or the dome 17; And, of course, but not limited to, mechanical or electrostatic means for connecting the wafer 15 to the cathode 32C. Such devices are described in US Pat. Nos. 4,872,947 and 4,842,683, which are incorporated by reference.

For example, a closed loop heat exchanger 90 capable of recirculation may be a block of wafer support / cathode 32C, as shown by flow passage 91, for cooling (and / or heating) the wafer support. It can be used to flow a fluid, preferably a dielectric fluid, through the pedestal. For silicon oxide etching, a temperature of dielectric fluid of -40 ° C is used. As with the long term, heat transfer between the wafer 5 and the wafer support 32 is enhanced by an inert gas such as helium (He) on the wafer and the support surface.

Chamber walls and domes may be heated and / or cooled by air convection (air injection) and / or heat exchangers of dielectric fluid. For example, a closed circuit heat exchanger 92 recirculates the dielectric fluid at a temperature adjusted from heating to cooling, ie, from + 120 ° C. to −150 ° C. along the passage 93 through the chamber side wall. Similarly, dome side wall 17W and top 17T may be heated and / or cooled by two heat exchangers 94 and 96 that recycle fluid along two passages 95 and 97, respectively.

In the dielectric thermal control system, the coiled antenna 30 is mounted between two walls 17W of the dome submerged in the recirculating dielectric fluid.

In another embodiment of the dome's dielectric fluid thermal controller, the coil of antenna 30 is comprised of plastic or Teflon manufactured at high temperature, thermally conductive thermoresist is applied between the merged antenna and the dome, and the hollow coil is a coil The dielectric fluid flows through and is heated and / or cooled. Because RF energy is applied to the coil and approaches the source plasma, the dielectric oil must have good dielectric and insulating properties and high boiling point in addition to having a high specific heat and density suitable for a sufficient heat exchanger at an acceptable flow rate. One suitable dielectric fluid is silther, commercialized by DuPont.

[12. Three electrode configuration]

Referring to FIG. 1, in the presently preferred embodiment, the chamber incorporates only three electrode arrays that contribute to novel process control and enhancement. The arrangement includes an upper electrode comprising a cathode (preferably wafer support electrode 32), an anode (preferably chamber sidewalls and bottom) and a dome top plate 17T. The upper electrode includes various shapes and includes various materials; Ie conductive material, preferably aluminum; Polarizable coating materials such as anodized aluminum; It may be in the form of a silicon or silicon-containing material, such as an aluminum-silicon alloy, or a silicon wafer, of course, but not limited to. Contains sacrificial silicon elements such as

1) Grounded Third Electrode

The grounded top plate 17T improves the grounding auxiliary plate suitable for the bias voltage (compared to the conventional auxiliary provided by the wall 12) and as a result improves the excitation of ions from the source 164 to the treatment zone 16B. Thus increasing processing speed (such as etching speed). In addition, the grounded top plate improves the bonding of the plasma generated at the wafer and the source.

2) biased third electrode

Using a third RF-biased electrode (using a silicon-containing element or coated electrode) in combination to supply the glass silicon to the source plasma improves various processing characteristics including etch rate and selectivity. Preferred by the strong binding properties of the source plasma, silicon enters the gaseous state and binds with / exhausts with free fluorine. (The binding properties of the source plasma occur at high concentrations when the fluorine containing gas phase is used for etch oxidation. This not only increases the etch rate of the bonded wafer material, such as polysilicon, but also increases the etch rate of the oxide, so that the oxide Decreases to multiple selectivity). Fluorine ions removed by the glass silicon form a carbon rich polymer. As a result, the oxide etch rate is increased, the oxide selectivity with respect to the polymer is increased, the anisotropy and perpendicularity of the oxide etch are improved, and the micro load is reduced.

In addition, the free silicon affects the polymerization reaction to form a carbon-rich polymer, which is easily etched in the presence of oxygen, such as when etching an oxygen-containing layer, and the opening is made of polysilicon or the like. Etching is further inhibited when oxygen is not present, such as when reaching an oxygen free layer or substrate.

As a result, the oxide film selectivity is increased relative to the oxygen-free substrate.

[13. Process Example]

Examples of the processes effectively performed in the reactor of the present invention are given below.

Example 1

Polysilicon Etching on Silicon Oxide

The polysilicon on the silicon oxide of the silicon wafer has a pressure in the range of about 2 mt to 20 mt; 50 cc chlorine (Cl ₂ ) caustic gas flow rate (only in the manifold (G ₁ )); Source power of 1500 watts; A bias voltage of -20 volts; And etched in a chamber with three electrodes using a grounded top electrode (no silicon) and having a 3500 to 400 microsecond polysilicon etch rate, vertical etching profile, and selectivity less than 100: 1 for the oxide of polysilicon. to provide.

Example 2

Silicon Oxide Deposition

Two-step bias stuffer deposition of silicon dioxide into the high aspect ratio openings formed in the silicon oxide film may be at a pressure ranging from about 2 mt to 10 mt (both steps); A gas flow rate of about 200 cc of argon, about 90 cc of oxygen, and about 45 cc of silane (in both stages and only in the manifold (G ₁ )); Grounded upper electrode (both steps); And in a chamber with three electrodes using a voltage of about -20 volts (at the first stage) and a voltage of about 100 to 200 volts (at the second stage), whereby deposition of less than 7500 kW per minute during the first stage And total oxide deposition (deposition with profile controlled sputtering) per minute during the second step. Argon sputtering abraded the upper edge of the openings, ensured that the openings were completely filled with silicon dioxide without voids, and finally flattened the silicon oxide film. Thus, filling and planarization can be carried out continuously in the same deposition chamber of the present invention.

An important problem in semiconductor fabrication is that silicon dioxide must be etched to a selected thickness when the underlying layer is polysilicon or another oxygen free material. High selectivity is required, so that the silicon oxide film will be etched at a relatively high rate before any exposed silicon layer is etched away.

However, since both materials are etched at about the same rate when exposed to an etching plasma, it is known to add a fluoride source to an etching gas such as CHF ₃ to form a polymer coating layer on the substrate. However, forming a polymer layer on polysilicon makes etching more difficult in small device geometries. An important concept in this respect is the "micro load", which is defined as 1- (etch rate ratio), where the etch rate ratio is the ratio of the etch rate in the small structure of the wafer to the etch rate in the large structure. Thus, if the etching process has desirable etching characteristics of large and small structures at the same rate, the micro load will be 1-1 / 1 = 0. In a process where a small structure is etched at a slower rate than a large structure, the micro load is close to 1.0.

The difficulty with the described etch application is that in order to obtain high etch selectivity, a relatively large amount of amphoteric polymer forming gas must be used in the plasma, but the polymer layer will eventually have a micro load greater than zero. Typically, it is expected to achieve only a 10: 1 selectivity with a micro load of about 0.1. However, many applications require selectivity as high as 30: 1 or 40: 1 with near-zero micro loads.

[2. Silicon in source area]

For high density plasma sources, one of the decomposition products that etch polysilicon is fluorine. Silicon is used to remove free fluorine radicals from the source region. The silicon may be a coating on the third electrode 17T or on the inner wall 17W of the chamber. If the sacrificial silicon is on the wall, the thickness of the silicon is determined in consideration of the frequency at which RF energy is supplied to the plasma from the antenna 30. This parameter should be chosen to ensure that enough energy is electromagnetically coupled through the chamber wall. If silicon is incorporated into the third electrode 17T, the silicon thickness is not a threshold. In any case, if the silicon is made to be useful for removing free fluorine from the source region, it is possible to form volatile compounds that are easily removed from the chamber.

The silicon scavenger material covers itself with the polymer during the etching process; The polymer may be removed by heating to a high temperature, or by electrically biasing the silicon, the impact that occurs on the silicon surface is increased and the polymer is sputtered from the surface, thus exposing the free silicon again.

Example 3

Silicon Oxide Etching on Polysilicon

Silicone compounds on polysilicon may have a pressure of 2 to 30 mt; Gas chemical flow rates of 30 to 60 sccm for CHF ₃ , 6 to 18 sccm for CO or CO ₂ , and 100 to 200 sccm for argon; (Only in the inlet manifold G ₁ ); 200 watt source power; A 200 volt bias voltage; A silicon disk 17S is mounted and etched in a chamber with three electrodes using the top electrode 17T biased by 2 MHz and 1000 Watts of RF energy. Silicon oxide is etched at a flow rate of 8000 kPa per minute with a 50: 1 selectivity of oxide to poly. Optionally, the silicon content is supplemented by silica coated onto the crystal dome wall 17M.

Example 4-3

Etching the Oxygen-Containing Layer on an Oxygen-Free Substrate

A series of etching processes have been performed in the reactor of the present invention which etches oxygen containing layers such as silicon oxide and the like on various oxygen free substrates. After formation of the oxide film pattern using a conventional photolithography technique, an etching process using C ₂ F ₆ was carried out in the reactor of the present invention using a grounded third silicon electrode.

The absence is summarized in Table 1 below.

TABLE 1

Another series of etch using CF ₄ was performed except that borophosphosilicate glass (BPSG) was replaced as the oxygen containing layer. The results are summarized in Table 2 below.

TABLE 2

Selectivity can be controlled by varying gas flow, source power and silicon substrate conditions to optimize selectivity for a particular silicon oxide film or glass on a particular substrate. Selectivity up to 100: 1 can be achieved in the reactor of the present invention by using a silicon scavenger for fluorine ions.

[14. Other structure]

1) plasma control

A preferred form of the present invention is to automatically change the "bottom" current to maintain a constant cathode (wafer) covering voltage. At low pressures of less than 500 mt in highly asymmetrical systems, the direct current bias measured at the cathode is almost similar to the cathode coating voltage. Floor power can be automatically converted to maintain a constant direct current bias. Floor power has a very small effect on plasma density and ion current density. Top or antenna power has a very significant effect on plasma density and current density, but very little on cathode coating voltage. Therefore, it is required to use the top power to form the plasma and ion current density, and to use the bottom power to form the cathode covering voltage.

2) differential bias

As an option for biasing the wafer 5 with respect to ground, the bias matching network 43 and top plate 17T may not be grounded and as indicated by the dashed line 50 in FIGS. Reference is made to different things. Referring to FIG. 2, the top plate has a voltage V _T -SS between the top plate and the wafer, which is almost twice the magnitude of the voltage V _TW between the top plate and the wall portion 12, and between the wafer and the wall portion. Differently driven and tailored to nearly double the voltage (V _{SS -W} ) magnitude. This balanced unusual drive reduces the interaction of the wall with the plasma and increases the interaction, ie, ion excitation, between the source region 16A and the process region 16B.

3) shift structure

The plasma reaction system of the present invention is shown in FIG. 1 in a conventional direction with a substrate 5 over an electrode (cathode) and an antenna 30 surrounding the dome 17 over the electrode. Conveniently, the power supplied to the antenna 30 is referred to as "antenna" or "source" or "top" power, and the power supplied to the electrode / cathode 32 is referred to as "bias" or "floor" power. It was. Such representation and designation is only convenient and the system may be altered. That is, it can be understood that the upper electrode 32 and the antenna located below it can be shaped or oriented in other ways, such as horizontally without complement. In the modified shape, the plasma can be generated at the antenna 3a and transmitted to the substrate mounted above the antenna in the same manner as mentioned in the specification. That is, the transmission of the active species may be caused by diffusion and bulk flow, or may be selectively assisted by a magnetic field with an axial gradient. This method does not depend on gravity and so does not have a significant effect on direction. The changed direction can be used to minimize the functionality of the particles produced in the plasma generation zone on the gas phase or on the surface and fall into the substrate. Gravity thus minimizes the possibility that the smallest of these particles are driven upwards with respect to the potential gradient of gravity to the substrate surface.

4) High and low pressure operation and variable spacing

Other chamber designs are useful for high and low pressure operation. The spacing d between the wafer support cathode 32C and the coiled bottom or winding of the antenna is inconvenient for high and low pressure operation. For example, an operating pressure of 500 millitorr to 50 torr preferably uses an interval d of about 5 cm or less, while an operating dynamic pressure of 0.1 millitorr to 500 millitorr or less results in an interval d of 5 cm or more. It is preferable. The chamber may utilize a fixed spacing d, as shown, or devise a variable spacing, such as coral exchangeable or stretchable upper chamber portions.

Reaction system 10 includes high and low pressure deposition of materials such as silicon oxide and silicon nitride; Etching of low pressure anisotropic reactive ions of materials such as silicon dioxide, silicon nitride, silicon, polysilicon, glass and aluminum; High pressure plasma etching of such materials; And beneficial for processes such as CVD wear that simultaneously deposit such materials and use etched walls, and includes planarization of substrate topography. These other processes for which the reaction system 10 can be used are described in US patent application Ser. No. 07/560530 and referred to by reference.

[15. Device Embodiment

The working embodiment of the system is merged with the source shape and the antenna shape as shown in FIG. The fine 5 inch quartz crystal source chamber 17 is 12 inches in diameter. The 2 MHz, 13 inch diameter, 4 inch high, 13 turns coiled antenna ends at both ends (with grounded variable capacitors L and T) spaced about 2.5 inches from the ground plane and surrounds the source. Active load matching is supplied to the variable storage battery L (10 to 3000 picoparr variable storage batteries, 5 KV acceleration). The capacitive winding of the antenna for resonance is also provided by a wound storage battery T (5-100 picoparr, 15 KV acceleration). The operation uses a 2 kilowatt RF energy source, so 2 MHz extends to the wafer and supplies 2 inches of plasma down (under the source). This provides the wafer with a plasma density of (1-2) × 10 12 / cm 3 and an ion saturation current density of 10-15 mmA / cm 2. A 2 MHz bottom or bias, 600 watts applied to a 5 inch wafer mounted on a support electrode that is nearly 1 inch below the antenna, provides a 200 volt cathode covering voltage.

As indicated above, the reaction system embodying the present invention includes low pressure chemical vapor deposition (CVD) including reactive ionizer (RIE), high pressure plasma etching, sputter facet deposition and planarization, and high pressure isotropic CVD. It is useful for several plasma treatments. Other applications also include, but are not limited to, sputter etching, ion beam etching, or electron ion or active neutral plasma sources.

Claims (23)

A plasma etching method comprising the steps of: a) providing a vacuum chamber for forming and maintaining a plasma, b) providing an article to be processed by the plasma on a support in the chamber, c) fluorine-containing etching in the chamber Supplying a gas, d) coupling RF energy to the chamber to form and maintain a plasma of the etching gas in the chamber, and e) supplying a source of silicon or carbon to the etching gas. Plasma etching method characterized in that. The method of claim 1, wherein the silicon source is a silicon containing gas. The method of claim 1, wherein the silicon source is a solid silicon source that activates to provide silicon ions to the plasma. 4. The plasma etching process of claim 3, wherein the solid silicon source is activated by adjusting the temperature of the source. 4. The method of claim 3, wherein the solid silicon source is activated by contacting the plasma. 4. The method of claim 3, wherein the silicon source is activated by electrically biasing the source. The method of claim 1, wherein the substrate comprises an oxygen containing layer on an oxygen free layer. 8. The plasma etching method of claim 7, wherein the etching method selectively etches the oxygen-containing layer. 8. The plasma etching method of claim 7, wherein the etching method provides a selectivity enhancing polymer layer on the substrate. 10. The method of claim 9, wherein the polymer comprises at least 50 wt% carbon and at most 40 wt% fluorine. The method of claim 1, wherein the RF energy is inductively coupled to the chamber by a coil antenna. The method of claim 1, wherein the RF energy is capacitively coupled to the chamber through the substrate support, which is an electrode. 12. The method of claim 11, wherein the temperature of the chamber walls is controlled independently. The method of claim 4, wherein the silicon or carbon source is heated to a temperature of at least 150 ° C. 6. 14. The plasma etching method of claim 13, wherein said temperature is controlled by heating to at least 150 &lt; 0 &gt; C. A plasma etching reactor, comprising: a vacuum chamber; A source of process gas for the vacuum chamber; A support for a substrate to be processed in the vacuum chamber; At least one RF power supply; A coil antenna for inductively coupling an RF signal from the RF power supply to the vacuum chamber to generate a plasma from the processing gas; And a source of carbon or silicon capable of supplying carbon or silicon ions to the plasma. 17. The plasma etching reactor of Claim 16, wherein said processing gas is a fluorine containing etching gas. 17. The plasma etch reactor of Claim 16, wherein the carbon or silicon source is a solid source in contact with the plasma. 17. The plasma etching reactor of claim 16, wherein the carbon or silicon source is a solid source having an independently controlled temperature to maintain an active surface. 17. The plasma etch reactor of Claim 16, wherein the carbon or silicon source is an electrically biased solid source. 17. The plasma etch reactor of Claim 16, wherein the carbon or silicon source is a gaseous compound of silicon. 17. The plasma etch reactor of claim 16, further comprising temperature control means for the dielectric walls of the chamber. 17. The plasma etching reactor of claim 16, wherein a second RF power supply is connected to a support for the substrate for biasing the substrate.