Plasma reactor with inductively coupled source power applicator and a high temperature heated workpiece support
An inductively coupled plasma reactor includes a coil antenna overlying the ceiling and coupled to an RF power source. The reactor further includes gas injection apparatus coupled to a process gas supply that comprises a supply of oxygen gas and a supply of a hydrogen-containing gas. A heated workpiece support in the chamber is configured to operate in a high temperature range including 700 degrees C. or higher.
Latest Applied Materials, Inc. Patents:
- Semiconductor process equipment
- Forming non-line-of-sight source drain extension in an NMOS FINFET using n-doped selective epitaxial growth
- Methods of depositing metal films using metal oxyhalide precursors
- Methods of operating a deposition apparatus, and deposition apparatus
- Data analytics and computational analytics for semiconductor process control
This application claims priority of PCT Patent Application Serial No. PCT/US2006/003250, filed Jan. 30, 2006, entitled SELECTIVE PLASMA RE-OXIDATION PROCESS USING PULSED RF SOURCE POWER, by That Cheng Chua, which claims priority of U.S. patent application Ser. No. 11/050,471, filed Feb. 2, 2005, entitled SELECTIVE PLASMA RE-OXIDATION PROCESS USING PULSED RF SOURCE POWER, by That Cheng Chua.BACKGROUND OF THE INVENTION
The present invention concerns the formation of a thin gate oxide beneath a gate electrode overlying a source-drain channel of a field effect transistor. The gate oxide is extremely thin (on the order of only Angstroms in thickness) and must be at least nearly free of defects in its amorphous silicon dioxide structure, such as dangling silicon or oxygen bonds. Such dangling bonds create quasi-mobile charge in the gate oxide, leading to two problems. One problem is that the electric fields associated with such charge perturb carrier motion in the source-drain channel, preventing a smooth current flow. Another problem is that such charge, to the extent it is mobile, may contribute to leakage between the gate electrode and the source-drain channel. Therefore, processes for forming the gate oxide must be free of any tendency to form defects in the gate oxide. Currently, only thermal oxide growth processes meet such criteria.
Thermal processes for forming the gate oxide layer have worked well in fabrication of semiconductor devices of the larger feature sizes used in the past. Such thermal processes tend to form high quality gate oxide layers that are free of defects such as dangling bonds or contaminant particles. Moreover, the gate oxide thickness tends to be uniform across the gate area. Unfortunately, as feature sizes are becoming much smaller and different gate oxides are employed in the next generation of advanced technologies, the high wafer temperatures required in thermal oxidation processes are problematic in that the sharp junction definitions which are now required become diffused at the higher temperatures (e.g., above 700° C.). Such a distortion of junction definitions and other features can lead to device failure.
One possible solution to this problem is to employ low temperature plasma processing to form the gate oxide layer. In attempting to do this, other problems have been encountered that render plasma processing apparently useless for gate oxide formation. First, at high chamber pressure (e.g., 100 mT), contaminants tend to accumulate in the gate oxide layer during formation, leading to fatal defects in the gate oxide structure such as dangling bonds or mobile charge. In order to reduce such defects, the chamber pressure can be reduce (e.g., down to tens of mT) by increasing the evacuation rate. While this approach can reduce contamination, it suffers from a tendency to increase plasma ion energy so that the gate oxide layer suffers ion bombardment damage, creating the same type of defects that were sought to be avoided, including dangling bonds and mobile charge. Also, the gate oxide thickness is non-uniform. Thus, it would seem that plasma processing is not feasible, and there has appeared to be no way in which a high quality gate oxide could be formed without raising the wafer temperature beyond allowable limits for the latest generation of devices.
Accordingly, what is needed is a low temperature process for forming a very high quality (defect-free) thin gate oxide layer of uniform thickness.SUMMARY OF THE INVENTION
A method of fabricating a gate of a transistor device on a semiconductor substrate includes the steps of forming discrete electrode/insulator layered structures spanning respective source-drain channels regions of the substrate, the layered structures having side walls, and then etching the layered structures to remove oxidation from side walls of the conductive layers of the layered structures. A selective re-oxidation step is performed to restore oxide material removed from side walls of the insulator layers of the layered structures during the etching step. The re-oxidation step consists of:
introducing into a vacuum chamber in which the substrate resides in a process gas that includes oxygen while maintaining a vacuum pressure in the chamber;
forming oxide insulating side layers on the side walls of the insulating layers of the layered structures by generating a plasma in a plasma generation region within the vacuum chamber separated from the substrate by a distance L during successive “on” times, and allowing ion energy of the plasma to decay during successive “off” intervals separating the successive “on” intervals, the “on” and “off” intervals defining a controllable duty cycle;
limiting the duty cycle so as to limit formation of ion bombardment-induced defects in the insulating side layers to a minimum defect density, e.g., less than 1011 cm−2·eV−1 or less than 5×1010 cm−2·eV−1; and
limiting the vacuum pressure so as to limit formation of contamination-induced defects in the insulating side layers.BRIEF DESCRIPTION OF THE DRAWINGS
Such defects can arise in a thermal oxide growth process when the formation of the gate oxide layer 40 is carried out at too low a temperature (e.g., below 800° C.). Such defects can be produced in either a thermal oxide growth process or in a plasma oxide formation process by contaminant particles within the insulating gate oxide layer 40.
The gate electrode 48 may consist entirely of polysilicon. Or, the gate electrode may be a stacked structure as shown in
Plasma processing can be used to form the silicon dioxide gate insulator layer 40 at relatively low temperatures (e.g., below 700° C.). Specifically a plasma formed of oxygen gas can cause silicon oxide to grow on exposed portions of the silicon substrate 42 to form the gate insulator layer 40. However, contaminants can enter the plasma and cause defects in the silicon dioxide gate insulator layer 40 if the chamber pressure is too high. One approach to avoid this problem is to reduce the chamber pressure until the contaminant particle count in the plasma is sufficiently low.
However, a new problem arises upon such a reduction of the chamber pressure, namely an increase in ion bombardment damage in the silicon dioxide gate insulator layer 40 as pressure is decreased. This trend is illustrated qualitatively by the curve labeled “Continuous RF” in the graph of
The increase in ion bombardment-induced defects as chamber pressure is reduced is likely due to the decrease in collision frequency within the plasma, so that energetic ions are less likely to lose energy through inelastic collisions prior to impacting the wafer. At lower chamber pressures, the plasma is less dense so that energetic ions will not tend to lose so much of their energy through numerous collisions with other particles (ions, radicals, atoms, etc.) before impacting the wafer. They therefore impact the substrate with higher energy, causing greater damage.
A further problem with plasma processes for growing the silicon dioxide gate insulator layer 40 is that plasma processing typically produces a non-uniform thickness distribution of the gate insulator layer 40, typically having a variance of about 1.04% across the wafer surface.
In the present invention, contamination-induced defects are eliminated by reducing the chamber pressure to very low levels (on the order of 10 mT). At the same time, ion bombardment-induced defects that would be expected at such a low chamber pressure levels are prevented by using a quasi-remote plasma source and pulsing the RF plasma source power (using a pulsed RF power source). I have discovered that reducing the pulsed RF plasma source duty cycle reduces the density of defects believed to be formed by ion bombardment damage in the silicon dioxide layer. In addition to reducing defect density, pulsing the plasma source power provides a surprisingly uniform distribution of thickness of the gate insulator layer 40, which solves the problem of non-uniform oxide formation in the plasma process.
The qualitative improvement or reduction in the ion bombardment damage defects in the gate oxide layer 40 using pulsed RF plasma source power is illustrated in the curve labeled “pulsed RF” in the graph of
It is believed the improved uniformity of the gate insulator layer 40 realized using a pulsed RF plasma source is attributable to the attenuation of electrical field gradients that tend to build up during the “on” time of the RF source power, the “off” time of the duty cycle being sufficient for these fields to die or relax so that they cannot reach sufficient levels to induce non-uniform growth rates across the wafer.
The effects of pulsing the RF power applied to the coil antenna 16 of
A surprising advantage of the pulsed RF plasma oxidation process is the great improvement in gate insulator thickness uniformity. In the continuous RF source power case, the variance in silicon dioxide thickness across the wafer has been about 1% or greater (when carried out at a high temperature of about 700° C.). Using pulsed RF source power in accordance with the invention, the variance in silicon dioxide thickness is reduced to only 0.16% at the same temperature and to 0.46% a low wafer temperature (30° C.), a dramatic improvement in either case.
Selective Gate Oxidation Process:
Dynamic random access memories (DRAM's) have a gate structure in which the gate insulator layer 40 of
The main problem with the foregoing re-oxidation process is that must be performed at high temperatures (e.g., 800° C.) using rapid thermal processing methods, in order to maintain a high quality silicon dioxide structure. Such a high temperature cannot be used in the latest generation of devices because of their extremely small feature size, in particular the channel length or source/drain spacing.
A pulsed RF plasma can be employed to carry out the gate insulator selective re-oxidation process to deposit a high quality oxide layer having a very low defect density. As in the embodiments described above the defect density is reduced by reducing the duty cycle of the pulsed RF plasma. The process can be performed at a reduced temperature, a significant advantage. The selective re-oxidation process is carried out by providing a hydrogen gas source 62 in the reactor of
Maintaining a separation between the ion generation region and the substrate of a distance L (block 114-1 of
Generating the plasma by applying source power only during successive “on” times, allowing ion energy of said plasma to decay during successive “off” intervals separating the successive “on” intervals, the “on” and “off” intervals defining a controllable duty cycle (block 114-2);
Limiting the duty cycle so as to limit formation of ion bombardment-induced defects in the insulating layer (block 114-3); and
Limiting the vacuum pressure in the chamber so as to limit formation of contamination-induced defects in the insulating layer (block 114-4 of
After the oxide insulating layer is formed, a conductive gate electrode is deposited over the insulating layer (block 122 of
introducing into a vacuum chamber in which the substrate resides a second process gas that includes oxygen and a reducing agent such as hydrogen while maintaining a vacuum pressure in the chamber (block 134-1 of
forming oxide on the insulating layer by generating a plasma in a plasma generation region within the vacuum chamber (block 134-2);
maintaining a separation between the ion generation region and the substrate of a distance LD (block 134-3);
applying plasma source power only during successive “on” times, allowing ion energy of the plasma to decay during successive “off” intervals separating the successive “on” intervals, the “on” and “off” intervals defining a second controllable duty cycle (block 134-4);
limiting the duty cycle so as to limit formation of ion bombardment-induced defects in the insulating layer (block 134-5) to a defect density less than about 5×1010 cm−2·eV−1; and
- limiting the vacuum pressure so as to limit formation of contamination-induced defects in the insulating layer (block 134-5).
The foregoing gate re-oxidation process may be applied to a gate electrode not having tungsten material, such as a gate that consists only of polysilicon. In such a case, the use of hydrogen gas may not be needed.
In some cases, the gate oxide 40 formed by the pulsed RF plasma process described above is implanted with a species such as nitrogen to enhance its performance, or to reduce current leakage or change the dielectric constant, for example. In such a case, the defect density is increased slightly. For example, if the gate oxide is deposited with a defect density of about 5×1010 cm−2·eV−1, then following the introduction (or implantation) of nitrogen, the defect density may become 1011 cm−2·eV−1 but this is still a great improvement considering the very low temperature at which the oxide was deposited.
While the invention has been described in detail by specific reference to preferred embodiments, it is understood that variations and modifications thereof may be made without departing from the true spirit and scope of the invention.
19: A plasma reactor, comprising:
- a vacuum enclosure defining a chamber and comprising a cylindrical side wall and a ceiling;
- a coil antenna overlying said ceiling;
- an RF power source;
- an RF impedance match coupled between said RF power source and said coil antenna;
- gas injection apparatus coupled to the interior of said chamber and a process gas supply coupled to said gas injection apparatus and comprising a supply of oxygen gas and a supply of a hydrogen-containing gas; and
- a heated workpiece support in said chamber configured to operate in a high temperature range including 700 degrees C. or higher.
20: The reactor of claim 19 wherein said high temperature range includes 800 degrees or higher.
21: The reactor of claim 19 wherein said hydrogen containing gas comprises molecular hydrogen gas.
22: The reactor of claim 19 wherein said ceiling has a dome shape and said coil antenna conforms with said dome shape.
23: The reactor of claim 19 wherein said ceiling is flat.
24: The reactor of claim 19 wherein said heated workpiece support comprises a heater with multiple independent heating zones for heating.
25: The reactor of claim 19 wherein said heated workpiece support comprises a dual zone heater comprising:
- a central heating element having a circular shape underlying a central portion of said heated workpiece support;
- an annular outer heating element around and concentric with said central heating element.
26: The reactor of claim 19 wherein said RF power source comprises an RF generator capable of producing pulsed RF power at a selected pulse duty cycle.
27: The reactor of claim 19 wherein said RF power source comprises an RF generator with a gate and a pulse generator coupled to said gate.
International Classification: C23F 1/00 (20060101);