 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method for treating a surface using a
plasma generated by radio frequency wave source, particularly a microwave
or UHF source of energy. In particular, the present invention relates to a
method which utilizes a disk shaped plasma allowing etching, texturing,
vapor deposition and oxidation of the surface. The present invention also
relates to an apparatus for treating the surface of an article, such as an
integrated circuit.
In U.S. application Ser. No. 468,897 filed Feb. 23, 1983 now U.S. Pat. No.
4,507,588, including one of the present inventors, a microwave ion source
apparatus is described which is useful for large surface area treatment.
This apparatus produces a microwave plasma over the surface being
processed. The plasma has a disk shape and hence the name microwave plasma
disk reactor. It can be operated continuously from high pressures (over
one atmosphere) to low (<10.sup.-3 Torr) pressures in any gas, and is
particularly adaptable to many different conditions. This allows the
optimization of a given process. In particular input power, gas flow,
pressure, and the like can be accurately controlled through large
variations, witn a sustained plasma. The present invention is an outgrowth
of this earlier application.
2. Prior Art
The high density requirement of VLSI (Very Large Scale Integration)
technology has provided a driving force for the application of plasma
processing to integrated circuit (IC) fabrication. Current and potential
uses of plasma assisted semiconductor wafer processing include oxidation,
film deposition, and dry etching procedures (Sze, S.M., VLSI Technology,
McGraw Hill, New York (1983)). All of these offer advantages to VLSI
processing because of one or more of several factors, including
anisotropic properties, low temperature which leads to small wafer thermal
stress, or improved material quality. Heretofore most IC plasma processing
has been carried out with dc or rf generated plasmas (less than or equal
to 60 megahertz). However, microwave induced plasmas possess several
advantages and several researchers have suggested the investigation of the
effect of higher frequencies on plasma processing.
Until recently, with the exception of early work on plasma growth of native
oxides in silicon, few experiments have investigated microwave plasma IC
processing. Primarily, this is because microwave discharges usually have
small volumes and, particularly, small surface areas. On the other hand,
microwave discharges have several positive attributes for semiconductor
processing. For example, microwave generated plasmas allow higher plasma
densities at lower pressures when compared to rf or dc plasmas. A
substantially greater degree of anisotropic etching, and therefore high
circuit density, should therefore be possible by processing with microwave
plasmas. Some Japanese experimental work with microwave discharges appears
to confirm this expectation (Suzuki, K. S., S. Okudaira, N. Sadudo and I.
Kanomata, "Microwave Plasma Etching," 16, 1979-1984 (1977); Suzuki, K., S.
Nishimatsu, K. Ninomiya and S. Okudaira, "Microwave Plasma Etching," Proc.
Int'l Ion Engineering Congress ISIAT '83 & IPAT '83, 1645-1656 (1983))
and, in fact, Hitachi has recently introduced a microwave plasma etching
system for commercial use (Electronics, p. 63, Nov. 20, 1980). Also
microwave plasmas, as is the case for rf but not dc, may be electrodeless
which reduces a source of contamination, reduces maintenance and increases
lifetime. Microwave systems are of comparatively low cost, are simple to
operate, and are usually more efficient than dc or rf plasmas. By moving
into the higher frequency range, more excited atomic states and more free
radicals are present in the plasma.
MICROWAVE PLASMAS
This invention concerns itself with the development of microwave plasma
reactors for materials processing. Microwave discharges have been studied
for over thirty years. They first appeared as unwanted gaseous breakdown
inside waveguide and coaxial coupling structures. An early application was
the TR tube which involved igniting a small microwave discharge in a
waveguide (MacDonald, A. D. and S. J. Tetenbaum, "High Frequency and
Microwave Discharges," in Gaseous Electronics, M. N. Hirsch and H. J.
Oskam, eds., Vol. I, Academic Press, New York (1978)). As a result,
initial theoretical and experimental studies focused on understanding the
microwave breakdown process. Most early experimental discharges had small
plasma volumes and thus future applications did not appear promising. Yet,
over the last twenty years many microwave discharge applications have been
and continue to be studied. Examples are frequency converters and harmonic
generators, ion and free radical sources, synthesis of chemicals, spectral
radiation sources, heat sources, and electric propulsion systems for space
applications, i.e. microwave ion engines and microwave electrothermal
engines.
When considering plasma processing applications, microwave discharges have
a number of potentially important advantages over lower frequency rf and
dc glow discharges. For example, microwave discharges are very "chemically
active" and can be efficiently maintained over a wide range of operating
pressures. They are electrodeless discharges which are simple to operate
and have low equipment costs.
The lack of metal electrodes in the discharge zone removes a possible
source of chemical contamination and allows the use of chemically active
gases that cannot be easily used in discharges with metal electrodes. It
also has the very practical advantage of being able to quickly cycle the
discharge system to atmosphere air (or other atmospheric gas) between
processing steps or runs. Discharge system life times are lengthened and
allowable discharge intensities can be increased when electrodes are
absent. These attributes, also shared by electrodeless rf generated
plasmas, are usually important advantages in plasma processing
applications. However, at present, a microwave plasma reactor with the
processing versatility that is comparable with dc plasma, barrel reactor
and rf parallel plate processing technologies does not exist. This
invention addresses the lack of such a versatile microwave plasma reactor
by using a microwave plasma disk source.
Work in both plasma assisted etching and plasma assisted oxidation dates
back to the 1960's. Plasma assisted etching has since become a key
processing technology. The plasma assisted growth of native oxides, on the
other hand, has not achieved widespread use to date. This difference is
primarily due to the historical development of requirements for silicon
wafer processing. Specifically, the need to etch Si.sub.3 N.sub.4,
followed by the fine line width requirements of VLSI, provided a strong
impetus for the development of plasma etching techniques. The motivation
for development of plasma oxidation has been less strong, due to the high
quality of presently available thermally grown oxides. Currently, however,
there is renewed interest in plasma assisted oxidation for two reasons.
First it is a low temperature process. This is an extremely advantageous
feature for VLSI because of the corresponding reduction of wafer warpage
and impurity motion. Secondly, there is appreciable interest in III-V
compounds for which thermally grown oxides are not of high quality. In
both cases, plasma grown oxides offer a potential advantage.
Plasma Assisted Oxidation of Silicon
Plasma oxidation of silicon was reported by Ligenza in 1964 (Ligenza, J.
R., "Silicon Oxidation in an Oxygen Plasma Excited by Microwaves," J.
Appl. Phys., 36, 2703-2707 (1965)). An oxide growth rate of 6000 Angstroms
in one hour at temperature below 300.degree. C. was observed in a
microwave generated oxygen plasma. Basic features of Ligenza's results
were confirmed by several groups, although the exact mechanisms for plasma
assisted oxidation are still debated. The importance of a positive bias on
the silicon leads to the conjecture that O.sup.- ions play a critical
role. Later work using ion analysis confirms this hypothesis. Subsequent
investigations showed that similar results could be observed in rf (using
frequencies from 420 KHz to 30 MHz) induced oxygen plasmas and dc arc
oxygen plasmas. A review of oxide formation in plasmas up to 1980 has been
published by Gourrier and Bacal (Gourrier, S., and M. Bacal, "Review of
Oxide Formation in a Plasma," Plasma Chemistry and Plasma Processing, 1,
217-232, (1981)). Regardless of the frequency, the method is often
referred to as "anodic-plasma oxidation" (Katz, L. E., "Oxidation,"
Chapter 4 in VLSI Technology, S. M. Sze, editor, Wiley, New York (1983)).
Recent oxidation experiments indicate that with careful control of
experimental parameters, high quality oxide films can be grown with plasma
systems.
The above review supports the conclusion by Katz, in his review chapter,
that anodic plasma oxidation offers great potential (Katz, L. E.,
"Oxidation," Chapter 4 in VLSI Technology, S. M. Sze, editor, Wiley, New
York, (1983)). He further states that uses should proliferate when
commercial equipment becomes available.
Low temperature plasma grown SiO.sub.2 can be used for less-critical
applications such as passivation, as grown. And with an appropriate low
temperature anneal at least one reported oxide film (Ho and Sugano (Ho, V.
Q. and T. Sugono, "Selective Anodic Oxidation of Silicon in Oxygen
Plasma," IEEE Tran. Elect. Dev., ED-27, 1436-1443, (1980)) appears
suitable for critical applications such as gate oxides.
Plasma Assisted Oxidation of III-V Compounds
Plasma assisted growth of native oxides on high mobility III-V compounds is
another area of commercial concern particularly for InP. The problem in
forming an insulator, suitable for MISFET applications, on the surfaces of
III-V compounds is well known and of long standing. Thermally grown oxides
are generally of poor quality due to the widely differing vapor pressures
of the oxide constituents. Consequently, alternative approaches such as
anodic electrolytic oxidation and anodic plasma oxidation are of interest.
Weinreich reported low temperature oxidation of GaAs in 1966 using a
microwave oxygen plasma (Weinreich, O. A., "Oxide Films Grown on GaAs in
an Oxygen Plasma," J. Appl. Physics, 37, 2924, (1966)) and several other
investigators have since studied plasma grown oxides on GaAs (Chang, R. P.
H. and A. K. Sinha, "Plasma Oxidation of GaAs," Appl. Phys. Lett, 29,
56-58, (1976); Sugono, T., F. Koshiga, K. Yamasolci, and S. Takahashi,
"Application of Anodization in Oxygen Plasma to Fabrication of GaAs
IGFET's," IEEE Tran. Elect Dev., ED-27, 449-455, (1980)). Appreciable
progress has been made and indeed planar GaAs MOSFET integrated circuit
ring oscillators with enhancement/depletion mode gates have been reported,
with anodic plasma oxide gates (Yokoyoma, N., T. Mimura, and M. Fukata,
"Plasma GaAs MOSFET integrated Logic," ED-27, 1124-1128, (1980). However
GaAs is rather handicapped by a high density of surface states which pin
the Fermi energy near the middle of the band gap for both
metal-semiconductor and MIS systems (Wieder, H. H., "Materials Options for
Field Effect Transistors," J. Vac. Sci. Technol., 18, 827-837, (1981).
Therefore enhancement mode MOSFET's are difficult to fabricate. InP,
however, has an appreciably lower surface state concentration, and
enhancement mode MISFET's have been reported by several groups (Lile, D.
L., D. A. Collins, L. G. Meiners, and L. Messich, "N-channel Inversion
Mode InP MISFET," Electron. Lett., 14, 657-659, (1978); Henry, L., D.
LeCrosnier, H. L'Haridon, J. Paugam, G. Pelous, F. Richou, M. Salvi,
"N-Channel MISFET's on Semi-Insulating InP for Logic Applications,"
Electron. Lett., 18, 102-103 (1982)). The first plasma grown native oxide
on InP was reported in 1981 by Kanazawa and Matsunami (Kangawa, K. and H.
Matsunami, "Plasma Grown Oxide on InP," Japanese J. Appl. Phys, 20,
L211-L213, (1981)). Capacitance-voltage measurements on their inductively
coupled rf plasma grown MOS structure show a minimum surface density in
the range 1.sup.-3 .times.10.sup.11 cm.sup.-2 ev.sup.-1. This is
comparable to electrolytically grown oxides on InP. These results are
promising, but further improvements would be in order. The plasma grown
films seem to have a pile-up of P near the interface and a deficiency of P
in the oxide. (However the deficiency is less than in thermally grown InP
oxides). Also a hysterisis was observed in the C-V measurements. This was
reduced on a H.sub.2 anneal, but after the anneal the oxide was leaky.
The possibility of high mobility, enhancement mode MOSFET InP integrated
circuits is attractive. Plasma grown native oxides offer an interesting
fabrication possibility for the gate insulator.
Plasma Assisted Etching
Etching of a semiconductor substrate takes place due to a combination of
physical and chemical processes. The processes are due mainly to ions,
neutral particles and in an indirect way electrons. Plasma etching
reactors are designed to encourage certain reactions while inhibiting
others, and this has led to several different reactor designs. For example
the degree of anisotropic etching is one parameter which is strongly
influenced by reactor design.
Before discussing different reactor designs, it is useful to consider the
etching of silicon in a CF.sub.4 plasma as a case in point. The chemically
active neutral species in the plasma is F, and the simplified steps in the
process are as below (Colburn, J. W. and H. F. Winters, "Plasma-Etching-A
Discussion of Mechanisms," J. Vac. Sci. Technol., 16, 391-403, (1979)).
Chemisorption (F.sub.2)gas.fwdarw.(F.sub.2).sub.ads .fwdarw.2F.sub.ads
Reaction Si+4F.sub.ads .fwdarw.(SiF.sub.4).sub.ads
Desorption (SiF.sub.4).sub.ads .fwdarw.(SiF.sub.4).sub.gas
It is well known that the etch rate increases rapidly when energetic ions
are incident on the wafer. A possible explanation is that ions cause
damage sites on the silicon surface which enhances the dissociative
chemisorption of F.sub.2 (Colburn, J. W. and H. F. Winters,
"Plasma-etching--A Discussion of Mechanisms," J. Vac. Sci. Technol., 16
391-403, (1979)). Therefore, if ions are normally incident on the wafer
the lateral etching of sidewalls proceeds at a much lower rate than is the
case for the horizontal wafer surfaces. Consequently, the etching is
anisotropic, or directional, or vertical, all of which are terms used
rather interchangeably in the literature.
The more normally incident the ions are, the more directional is the
etching. Therefore low plasma pressures are favored for
anisotropic-etching since ions are scattered at high pressures. However,
in conventional rf and dc plasmas, etch rates are greatly reduced at low
pressure due to low plasma densities so there is a trade-off between etch
rate and anisotropic etching. Alternatively, in conventional systems,
anisotropic etching can be increased at higher pressures by increasing the
energy of the incident ions. In this case the cost involves semiconductor
damage and loss of selectivity.
The design of a particular reactor, then, depends on what feature is being
emphasized. For example, consider the following types.
(i) Barrel reactor. Here the wafers are not in direct contact with the rf
discharge, but rather in a perforated metal tunnel which is surrounded by
the discharge. The surface potential of the wafers is not much different
than the surrounding glow, so ion bombardment is not appreciable and
etching is due primarily to long lived radicals. Operating pressures are
typically high, on the order of 200 m Torr, and etching is generally
isotropic. The approach is well suited for multiple wafer processing, but
uniformity problems require overetching (Elliott, D. J., Integrated
Circuit Technology, Chapter 11, McGraw-Hill, (1982)).
(ii) Parallel plate reactor. In this design, also known as the Reinberg
reactor, the wafers are on one plate of a diode plasma system, and in
direct contact with the plasma (Reinberg, A. R., "Radial Flow Reactor,"
U.S. Pat. No. 3,757,733 (1973)). Since the sheath potential is on the
order of one-fourth the applied rf peak to peak voltage, and therefore in
the range of 75 to 250 v., ion bombardment may be significant. Also short
lived radicals take part in the etching in this configuration. Both barrel
and plasmas reactors have received widespread commercial use, with the
latter offering improved etch uniformity and the possibility of
anisotropic etching.
(iii) Reactive Ion Etching (RIE) and Reactive Ion Beam Etching (RIBE).
Reactive ion etching, also known as reactive sputter etching, often
employs apparatus similar to sputtering chambers with the wafer taking the
place of the target. Compared to the parallel plate reactor, RIE systems
are characterized by generally lower pressures (10 m Torr-100 m Torr) and
higher substrate potentials. These combine to produce highly anisotropic
etching. In a report by Lin, et al., a magnetic field was used to achieve
a magnetron RIE system (Lin, I., D. C. Hinson, W. H. Class, and R. L.
Sandstrom, "Low Energy High Flux Reactive Ion Etching by R. F. Magnetron
Plasma," Appl Phys. Lett., 44, 185-187, (1981)). Reactive ion beam
etching, on the other hand, employs apparatus similar to that used for ion
beam milling (Bollinger, L. D., "Ion Beam Etching with Reactive Gases,"
Solid State Technology, 26, 99-108, January 1983)). Whereas ion beam
milling uses inert gases, such as argon, RIBE employs reactive gases. It
offers the possibility of controlling the angle of ion incidence and is
operated at quite low pressure levels, in the range of 1.times.10.sup.-4
Torr. Consequently a very high degree of etching anistropy is achievable.
The method also differs from RIE in that the wafer is not in actual
exposure to the discharge. Therefore there is little interaction with
atoms in short lived, excited states.
(iv) Recent Developments. Geis, et al., (Geis, M. W., G. A. Lincoln. N.
Efreomow, and W. J. Piacentini, "A Novel Anisotropic dry etching
Technique," J. Vac. Sci. Technol. 19, 1390-1393, (1981)) reported a
variation on RIBE in which a Kaufman ion source is used to direct an argon
beam onto a wafer, and also impinging a chemically reactive gas from a jet
near the wafer. This allows a larger flux of reactive gas than in
conventional RIBE. Chinn, et al., refer to the method as chemical assisted
ion beam etching and have reported etching, of GaAs, Ti, Si, and Mo using
similar apparatus (Chinn, J. D., A Fernandez, I. Adesida, and E. D. Wolf,
"Chemically Assisted Ion Beam Etching of GaAs, Ti, and Mo," J. Vac. Sci.
Technol, Al, 701-704, (1983); Chinn, J. D., I. Adesida, and E. D. Wolf,
"Profile Control of Chemically Assisted Ion Beam and Reactive Ion Beam
Etching," Appl. Phys. Lett, 43, 185-187, (1983)). Another approach aimed
at improved control of ion bombardment is triode plasma etching, as
reported by Minkiewicz and Chapman (Minkiewicz, V. J. and B. N. Chapman,
"Triode Plasma Etching," Appl. Phys. Lett. 34, 192 (1979)). Here the
substrate holder is based separately from the two electrodes which sustain
the discharge, but the substrate is still in contact with the plasma.
Improved control of the etch profile is reported. More recently, Mantei
and Wicker reported a triode like system with surface magnetic field
confinement (T. D. Mantei and T. Wicker, "Plasma Etching with Surface
Magnetic Field Confinement," Appl. Phys. Lett. 43, 84 (1983)).
OBJECTS
It is therefore an object to provide an improved method for surface
treatment, for example integrated circuit formation, including etching,
deposition and oxidation. Further still, it is an object to provide a
method which is relatively simple to perform and which is economical,
reliable and produces excellent results. These and other objects will
become increasingly apparent by reference to the following description and
the drawings.
IN THE DRAWINGS
FIG. 1 shows the improved apparatus of the present invention adapted for
treatment of a surface (30) forming part of an integrated circuit, wherein
the surface being treated is exposed to a disk shaped plasma and is
directly biased by a wire (37) with a voltage during the treatment, and is
mounted on an adjustable stand (31, 34) by a sleeve (35).
FIG. 2 is a diagram of the biasing circuit used with the apparatus of the
present invention showing the direct biasing of the integrated circuit
surface (30) being treated.
FIG. 3 is a schematic front cross-sectional view of another embodiment of
plasma treating apparatus, of the present invention wherein the chamber
(16) contains a wide disc shaped plasma (as large as 50 cm) for treatment
of multiple wafers (53) or other articles and particularly showing a
central passage or conduit (55) which allows gas to be removed from a
plasma chamber (16).
FIGS. 4 and 5 are schematic front cross-sectional views showing other
embodiments of the apparatus wherein the ions are either inside the
chamber (16) with the disk plasma or outside the chamber (16).
GENERAL DESCRIPTION
The present invention relates to a method for treating a surface forming
part of an integrated circuit which comprises: providing an ion generating
apparatus including a plasma source employing a radio frequency, including
UHF or microwave, wave coupler which is excited in one or more of its TE
or TM modes of resonance and optionally including a static magnetic field
surrounding the plasma source which aids in coupling at electron cyclotron
resonance and aids in confining the ions in the coupler wherein the plasma
is maintained at a reduced pressure in operation wherein the ion source
apparatus includes an electrically insulated chamber (15) mounted in
closely spaced relationshp to an area (16) of a metallic radio frequency
wave coupler, and gas supply means (18, 19) for providing a gas which is
ionized to form the plasma in the insulated chamber with the surface of
the integrated circuit in position to receive the ions; forming a plasma
disk in the chamber; and contacting a surface of the article with the ions
or free radicals inside the plasma or with ions or neutralized ions
removed from the plasma which treat the surface, wherein the surface has a
suitable voltage potential.
The present invention also relates to a method for treating a surface which
comprises: providing an ion generating apparatus including a plasma source
employing a radio frequency, including UHF and microwave wave coupler
which is excited in one or more of its TE or TM modes of resonance and
optionally including a magnetic field which aids coupling at electron
cyclotron resonance and aids in confining the ions in the coupler wherein
the plasma is maintained at a reduced pressure in operation, wherein the
ion source apparatus includes an electrically insulated chamber (15)
mounted in closely spaced relationship to an area (16) of a metallic radio
frequency wave coupler, and gas supply means (18, 19) for providing a gas
which is ionized to form the plasma in the insulated chamber with the
surface of the integrated circuit in position to receive the ions; forming
a plasma disk in the chamber and attracting the ions to the surface with a
bias means having a suitable voltage potential attached to the surface
which treat the surface.
The present invention also relates to an ion generating apparatus for
treating a surface of an article which comprises:
(a) a plasma source employing a radio frequency, including UHF or
microwave, wave coupler which is excited in one or more of its TE or TM
modes of resonance and optionally including a magnetic field which aids in
coupling at electron cyclotron resonance and aids in confining the ions in
the coupler wherein the plasma is maintained at a reduced pressure in
operation wherein the ion source apparatus includes an electrically
insulated chamber (15) mounted in closely spaced relationship to an area
(16) of a metallic radio frequency wave coupler;
(b) gas supply means (18, 19) for providing a gas which is ionized to form
the plasma in the insulated chamber;
(c) ion attracting means (37) for attachment to the surface for attracting
ions from the plasma to the surface by means of a suitable voltage
potential; and
(d) a platform (31) supporting the surface and electrically insulated from
the ion attracting means.
SPECIFIC DESCRIPTION
FIG. 1 shows the preferred plasma generation apparatus of the present
invention, some of which has elements in common with U.S. Ser. No.
468,897. The principle components of the apparatus are displayed in the
cross-sectional view of FIG. 1. The system, constructed to be operated at
2.4 gigaHertz, consists of a 17.8 cm inside diameter brass cylinder walls
10 forming the microwave cavity 11. A sliding short 12, with brushes 13
contacts cylinder 10 and the adjustable excitation probe 14, provide the
impedance tuning required to minimize reflected power. The sliding short
12 can be moved back and forth along the longitudinal axis of the cavity
11 to adjust its electrical length while the radial penetration of the
excitation probe 14 into the cylinder walls 10 varies the cavity 11 mode
coupling. A quartz dish 15 is shaped like a petri dish and keeps the
working gas in region or chamber 16 while allowing the microwave power to
produce a disk-like plasma adjacent to the field shorting screen 17. The
working gas is introduced into region 16 by means of an annular recess 18
in a base plate 10a, supplied by gas feed tube 19. A conduit 21 is
provided as a microwave inlet port to the cavity 11 for probe 14. The
short 12 is adjusted by means of rods 22 supported by plate 23 outside the
cylinder walls 10 controlled by threaded post 24 and nut 25. This design
source minimizes the plasma volume by creating a thin disk-like plasma
adjacent to the field shorting screen 17. The charged particles, i.e.
electrons and ions, in the disk plasma and the screen 17 form part of a
resonant microwave cavity.
The wafer 30 being treated is mounted on a stainless steel platform 31
which is insulated by a quartz disk 32 having a hole 33 therethrough. The
platform 31 is mounted on a hollow tube 34 which is secured in position by
an sleeve 35 and holding brackets 36 so that the vertical height of the
platform 31 can be adjusted relative to the dish 15. A bias wire 37 is
provided through the tube 34 and hole 33 and is in electrical contact with
the wafer 30 and insulated from the platform 31, tube 34 and sleeve 35.
The apparatus is mounted in a vacuum bell jar 50 with an opening 51 in a
conduit 52 leading to a vacuum pump (not shown). As shown in FIG. 2, the
bias wire 37 is connected to a voltage power supply 38. The field shorting
screen 17 is electrically connected to the cavity 10. It will be
appreciated that the wafer 30 can be clamped (not shown) on a conductive
surface biased by wire 37. This allows faster treatment of the wafers 30.
All of these variations will be obvious to one skilled in the art. Also
the disk 15 can be provided with silicon semi-conductor electrodes 37a
inside chamber 16 through base plate 10a. The electrodes 37a prevent
contamination of the plasma reactions with metallic electrode materials.
FIG. 3 shows that a large diameter plasma disk can be created in chamber 16
for processing multiple wavers 53. For instance by changing the cavity
resonant frequency to 915 megaHertz and by increasing the chamber 16
diameter, plasma disks of larger diameter (e.g. 50 cm) can be created.
As can be seen from FIG. 3, multiple wafers 53 are biased by wires 54. The
gas is allowed to flow across the wafers 53 and out central conduit 55 to
be removed by vacuum system (not shown) since the flows are small. The
wafers are supported on platforms 56 which form part of the microwave
cavity and are insulated by a quartz disks 57. Alternatively the platforms
56 can be insulating and support the wafers 53 and a field shorting screen
58 forms part of the microwave cavity. As can be seen from FIG. 4, a grid
60 is provided and multiple wafers 61 are biased by wires 62 and circuit
63. Supports 65 hold the grid 60 in place. As in FIGS. 1 and 3 the wafers
61 are insulated by quartz disks. In FIG. 5 wafer 70 is positioned outside
of the chamber 15 on an insulated support 71 and biased by wire 72. Grids
17a and 17b are biased to attract the ions from the chamber, as described
in Ser. No. 468,897. The filament 20 can be used to neutralize electrons
as does the neutralizer 20 described in U.S. Pat. No. 4,507,588. FIGS. 3
to 5 include a gas supply means as in FIG. 1 which is not shown.
The experimental microwave circuit used with the reactor is described in
Ser. No. 468,897. It consists of (1) a 2.45 GHz, cw variable power source,
(2) a circulator and matched dummy load, (3) directional couplers and
power meters that measure incident power, P.sub.i, and reflected power
P.sub.r, (4) the coaxial coupling probe and dc block (when needed) and (5)
the cavity 10 itself.
Several small 17.8 cm diameter reactors were built and tested with the
inert gases, O.sub.2, H.sub.2 and gas mixtures. In all gases a 10 cm
diameter plasma disk could be continually maintained inside the cavity
from pressures as low as 10.sup.-4 Torr to over one atmosphere. This is a
much wider operating pressure range than other microwave, rf and dc plasma
reactors. Depending on the plasma/gas loading, input power varies between
tens of watts to over 1,000 W. In all cases the reactor can be well
matched for efficient operation, i.e. power reflected from the
plasma/cavity was less than 3% of the incident power. The probe and length
tuning can be automated with gearing and small motors. Electron density
measurements made in argon gas with input microwave powers of 80 W to 150
W at pressures of 5.times.10.sup.-4 -10.times.10.sup.-3 Torr with double
Langmuir probes indicate densities of 5.times.10.sup.12 -10.sup.13
/cm.sup.3 in the disk discharge region. These are at least an order of
magnitude higher than those measured in other plasma and microwave
reactors.
The apparatus shown in FIGS. 1 and 2 was operated to carry out an oxidation
of Si wafers. Preliminary oxidation experiments with positive dc current
bias in a O.sub.2 plasma have been carried out with 5 cm and 2.5 cm Si
wafers, demonstrating the feasibility of oxidation with this system. It
can also be used for thin film processing such as for silane decomposition
to deposit silicon or for surface texturing for optical or metal treatment
purposes.
The potential general advantages of a microwave discharge is a chemically
active electrodeless discharge that has low equipment costs and is simple
to operate. Additional advantages appear when one investigates specific
applications of the microwave plasma disk reactor. The strengths of this
reactor are discussed below for its application to plasma oxidation and
etching.
The specific advantages of the reactor for plasma oxidation can be grouped
into two categories; those concerned with (1) efficiency, and (2)
experimental (and processing) versatility. The reactor requires no
additional tuning stubs or other equipment to couple (i.e. match)
microwave energy into the reactor cavity over a large pressure range
(preferably 10.sup.-4 Torr to one-half atmosphere). Once microwave energy
is in the cavity, the efficiency of coupling into the plasma is high,
especially for high density (.gtoreq.10.sup.11 /cm.sup.3) operation. The
plasma volume may be adjusted by varying the quartz disk size and platform
positions. The plasma is only created in a thin layer over the surface
being processed and extra plasma volume requiring extra power is not
produced. Thus, the available input power requirements and associated
energy and equipment costs are reduced. This is, of course, important for
a process such as plasma oxidation where energy requirements appear to be
large even for small wafers.
Experimental versatility is evidenced by the ability to scale to large
plasma disks allowing large wafer and multiple wafer treatment. This
versatility is also evidenced by the ability to operate a high density
oxygen plasma over a wide range of pressures (10.sup.-4 Torr to 10 Torr)
and flow rates by simply adjusting the input power. The substrate is
easily biased relative to the plasma. Since high density
(.gtoreq.10.sup.12 /cm.sup.3) discharges can be produced without a
magnetic field, it appears that fast electrons (which are produced by
electron cyclotron resonance discharges (Loncar, G., J. Musil and L.
Bardos, "Present States of Thin Oxide Films Creation in a Microwave
Plasma," Czech. J. Phys. B30, 688-707, (1980)) will not be generated.
Thus, oxide films that are undamaged by fast electrons should result.
The above mentioned advantages indicating good efficiency and versatility
apply also to etching applications. However, several other important
features should be noted. They are (1) the potential of low pressure
operation with high electron, ion, and other species densities, (2)
separate control over the energy and flux of the ions and other species on
the wafer, and (3) the potential of reducing surface and near surface
damage. Higher low pressure ion, electron, and free radical densities are
desirable. Measurements on the plasma disk reactor taken at low pressure
(5.times.10.sup.-4 -10.sup.-3 Torr) show plasma densities
(>5.times.10.sup.12 cm.sup.-3) are an order of magnitude higher than for
systems reported by others, particularly the Suzuki et al references
discussed in the prior art. Thus, it is believed that the microwave plasma
disk reactor has corresponding higher anisotropic etch rates.
In addition to the possibility of low pressure, high density operation, the
microwave plasma reactor is attractive for etching because the wafer
platform is independently biased. Since the platform is not an integral
part of the plasma generation process (as opposed to conventional rf
parallel plate reactors) the discharge electron, ion and reactive ion
species densities can be adjusted by varying the input power and cavity
tuning, while separately applied dc and/or rf voltages can be applied to
the wafer 30 as is shown in FIG. 1. Thus this reactor has similarities to
triode plasma and multipole plasma etching systems where energy and flux
of the ions and reactive species can be controlled independently (or
quasi-independently) without the presence of high voltages in the
processing chamber. These systems have greater control over the type,
number, and energy of the species incident on the substrate. The microwave
disk reactor essentially offers a single electrode (the wafer) version of
the triode system. However, it does not have the electrode (cathode)
lifetime and contamination limitations and also should be able to operate
at higher species concentrations.
As is shown in FIG. 5, the microwave plasma disk reactor can be configured
for a reactive ion beam etching (RIBE). In this case, a colliminated beam
of ions is drawn from the discharge and directed at the wafer. This
microwave system would differ from other RIBE systems by the way the ions
are produced. Most RIBE systems employ a Kaufman ion source, which
produces a plasma with a dc discharge. The electrodeless microwave plasma
ion source would not have the problem of short filament lifetime and
electrode contamination problems associated with the Kaufman ion source.
In addition, it can also be easily operated over a wider higher pressure
(10.sup.-4 -10.sup.-2 Torr) range and at higher beam densities. Both of
these features are expected to be advantages for etching applications.
Substrate surface damage is often a problem with conventional parallel
plate rf reactors. This is because the applied rf voltage produces a large
dc potential relative to the plasma. The microwave plasma disk reactor
should have little or no microwave field adjacent to the substrate.
Microwave exc | | |
