|
Description  |
|
|
FIELD OF THE INVENTION
This invention pertains to an apparatus and method for forming integrated
circuit substructure which facilitates depositing a planar aluminum layer.
BACKGROUND OF THE INVENTION
High-performance, large-area integrated circuits (IC's) often require
several levels of interconnect. Planarization processes which smooth and
flatten the surface of an IC at various stages of fabrication are becoming
essential, both for high-resolution photolithography and for adequate
coverage of steps by thin films. Numerous techniques exist for planarizing
the insulating layers in multilevel IC's, e.g., spin-on insulators,
spin-on sacrificial layers followed by nonselective etching, and RF bias
sputter deposition. CO.sub.2 lasers have been used in both CW and pulsed
operation to rapidly flow phosphosilicate glass over aluminum
interconnects.
Planarization of the insulating layers does not by itself yield a fully
planar multilevel interconnect process. Severe step-coverage problems
still occur where metals are deposited over deep vertical vias in an
insulator. The problem is magnified if vias are vertically stacked. These
difficulties are typically mitigated by tapering the via walls (which
consumes valuable area), limiting their depth/width ratio, and forbidding
stacked vias. Alternatively, selective deposition techniques (tungsten
chemical vapor deposition or lift-off or metal pillar fabrication) show
promise for filling deep vias. Yet another solution is to planarize each
metal film prior to patterning. Metal planarization could be combined with
insulator planarization to yield a fully planar IC process; it might also
be used by itself in specialized multilevel interconnect structures where
power and ground planes are interspersed between signal levels (e.g.,
"silicon PC boards" for advanced IC packaging). One metal planarization
technique is RF bias sputtering.
One impediment to planarization of aluminum layers by RF bias sputtering is
the presence of oxygen in the system which combines with the aluminum to
form aluminum oxide. The melting point of aluminum oxide is much higher
than that of aluminum and an aluminum film contaminated with aluminum
oxide resists planarization.
Magnetron sputter devices are characterized by crossed electric and
magnetic fields in an evacuated chamber into which an inert, ionizable
gas, such as argon, is introduced. The gas is ionized by electrons
accelerated by the electric field. The magnetic field confines the ionized
gas, which forms a plasma in proximity to a target structure. The gas ions
strike the target structure, causing emission of atoms that are incident
on a workpiece, typically a substrate in a coating process. Generally, the
magnetic field is established by a permanent magnet structure, although
electromagnetic devices are increasingly being employed for this purpose.
In coating applications, the magnetron sputtering devices are frequently
employed to deposit metals in the manufacture of electronic integrated
circuit type devices. It is also known to deposit magnetic materials in
the manufacture of high density magnetic discs of a type used for magnetic
disc memories.
In prior art magnetron sputtering devices, uniform coating thickness across
a substrate was obtained by moving the substrates during coating. Moving
the substrates also assisted in obtaining step coverage, i.e., conformal
coating over step-type transitions. Of course, there are many problems in
moving a substrate during operation of a sputtering device. It is also
desirable in certain instances to co-deposit different materials,
particularly materials which are difficult or impossible to alloy; that
is, materials which are not adapted to being on a single target. In all
instances, it is desirable to operate the sputtering device at as high a
rate as possible.
Sputter sources incorporating only permanent magnets, the typical prior art
arrangement, do not enable the plasma confining magnetic field to change
over the life of the target. In consequence, the impedance of the sputter
device, i.e., the ratio of the discharge voltage which establishes the
electric field to the discharge current flowing in the plasma, decreases
steadily as the target erodes during use. The power supplies necessary to
provide the electric field are therefore relatively complicated and
expensive in an attempt to match the varying sputter device impedance over
the target life.
As the target surface erodes during use, the target has a tendency to
create a shadow for material emitted from the source. Thereby, the gross
efficiency of the sputter device decreases as the target erodes during
use. Because of the shadowing effect, the rate at which material is
deposited on a substrate decreases usually in a non-linear manner, as the
target erodes.
One attempt to minimize the reduced deposition rate caused by the shadow
effect involves revolving an assembly including the permanent magnet about
an axis of the sputtering device. Revolving the magnet assembly results in
a substantial improvement in the efficiency of the sputtering process near
the end of target life, but a decrease in the impedance of the device
still has been observed as the target erodes. In addition, the rate at
which material is sputtered from the target also decreases as the target
erodes with this approach. Of course, rotating the permanent magnet
structure is mechanically complex.
While many of the problems associated with the permanent magnet arrangement
have been obviated by using electromagnets, the electromagnet devices have
generally had the disadvantage of using single targets, having relatively
narrow widths of approximately one inch. There has been recently developed
systems wherein the targets have been configured as assemblies having
plural target elements, generally concentric with each other. In one
configuration, the targets are both planar elements; in a second
configuration, an inner target is planar and an outer target is concave,
having an emitting surface defined by a side wall of a frustum of a cone.
These prior art devices are effective to enable material to be deposited
uniformly over a large area workpiece, such as a substrate being coated.
It has been observed that the relative contributions of the two targets on
the workpiece change differentially as the targets erode during use. In
other words, the amount of material reaching the workpiece from the first
target changes relative to the amount of material reaching the workpiece
from the second target as the targets are being consumed or eroded. Thus,
designing a controller for multiple element target assemblies to achieve
uniform impact of material on the workpiece during the useful life of the
target assemblies is complex, and not straightforward. This is
particularly the case for uniform deposition across relatively large area
workpieces, such as a six-inch integrated circuit wafer or a hard computer
storage magnetic disc. The system is also complex because of the need and
desire to control the impedances of the plasma discharges during the
changing conditions that occur as the targets erode.
OBJECT OF THE INVENTION
It is an object of the invention to provide an apparatus and method of
forming a structure which forms a barrier to oxygen between an aluminum
layer and oxygen-bearing layers to facilitate planarizing the aluminum
layer.
SUMMARY OF THE INVENTION
Oxygen-bearing layers of an integrated circuit, such as layers of silicon
dioxide, have the potential to contaminate a layer of aluminum formed on
the oxygen-bearing layer. A layer of refractory metal silicide formed over
the oxygen-bearing layer passivates the oxygen-bearing layer so that a
further layer of aluminum can be planarized. In particular a layer of
tantalum silicide of thickness at least 200 angstroms forms a durable
layer which survives the high temperature of planarizing the aluminum
layer.
In accordance with the present invention, a cathode sputter magnetron
device is controlled so that material is uniformly supplied to a workpiece
having a relatively large area over the lives of plural geometrically
spaced targets from which material is sputtered, wherein each target is
subjected to a separate plasma discharge that is confined to the
associated target by a separate magnetic field. In accordance with one
aspect of the invention, the uniformity is attained by controlling the
relative powers of the separate plasma discharges so that the relative
powers change as a function of target erosion condition.
I have found that changing the relative powers of the separate plasma
discharges enables uniformity to be maintained over the lives of the
targets. It is postulated that the change in the relative powers of the
plasma discharges provides the desired uniformity because the degree of
self-shadowing in the target elements changes differentially during
consumption of the targets. The erosion profiles of the targets are such
that the outer target, which erodes faster than the inner target, develops
self-shadowing at a much higher rate than the inner target. Because the
outer target erodes faster than the inner target, the outer target
requires more power to compensate for the resulting loss of deposition
efficiency as target erosion progresses.
In accordance with another important aspect of the invention, the
impedances of the separate discharges are controlled as the targets erode.
The impedances are controlled by varying each separate confining magnetic
field. Each magnetic field is derived by an electromagnet which is
supplied with a variable current that controls the impedance of each
discharge. The impedance of a first of the discharges is compared with a
set value therefor. The current applied to the electromagnet for the first
discharge is controlled in response to the comparison. The current applied
to the electrcmagnet for a second of the discharges is preferably
controlled so it is a constant factor of the current applied to the
electromagnet for the first discharge.
Preferably, the relative powers and the impedances of the discharges are
controlled simultaneously to achieve the maximum desired uniform result.
The powers of the discharges for the first and second targets are adjusted
so that as target erosion occurs the amount of power supplied to the
second target relative to the amount of power supplied to the first target
increases, to overcome the tendency of the targets to cause material to be
differentially incident on the workpiece as the targets erode.
According to a further feature of the invention, a cathode sputter target
is held in situ and easily removed from a support structure therefor by
providing bayonet slots in the target support structure, in combination
with pins in the targets which engage the slots.
In the original work that was done in connection with the sputter coater of
the invention, magnetic fields from a pair of magnetic circuits, one for
each target discharge, were combined in a single intermediate pole piece
member between a pair of the targets. I also found that the flux fields
must be additively combined in the intermediate pole piece member to
provide proper operation. I found that the intermediate pole piece is
preferably tapered to provide optimum performance.
While one aspect of the invention is directed to a sputtering device, it is
also applicable to targets per se, and in particular to a target assembly
having, in the preferred embodiment, a concave surface defined by a side
wall of a frustum of a cone. In the preferred embodiment, the concave
surface is included at approximately 45.degree. relative to a base of the
cone, an angle which has been found to provide excellent step coverage for
large area targets. Such a second target is used with a first target
element initially having a planar emitting surface with a circular
perimeter having a radius R.sub.2. The concave surface has inner and outer
radii of R.sub.3 and R.sub.4, respectively, where R.sub.2 <R.sub.3
<R.sub.4. Preferably, the first target is formed as a ring having an inner
radius of R.sub.1, where R.sub.1 <R.sub.2 .
In another aspect of the invention, I have found that by applying an R.F.
bias to the substrate in addition to heating the substrate, the quality of
the coating is improved. In general, while low power R.F. bias improves
the quality of the coating, high power R.F. bias can cause damage to the
substrate due to contacting the plasma to the substrate. A magnetic mirror
in the vicinity of the substrate, which can take the form of a coil around
the substrate, can be used to move the plasma away from the substrate
thereby increasing the R.F. bias power levels acceptable without damaging
the substrate.
These and further constructional and operational characteristics of the
invention will be more evident from the detailed description given
hereinafter with reference to the figures of the accompanying drawings
which illustrate one preferred embodiment and alternatives by way of
non-limiting examples.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a schematic diagram of a sputtering device including a pair of
target elements in combination with a controller in accordance with a
preferred embodiment of the present invention;
FIG. 2 is a cross-sectional view of a target assembly schematically
illustrated in FIG. 1, taken along the lines 2--2, FIG. 3;
FIG. 2A is the left half of FIG. 2.
FIG. 2B is the right half of FIG. 2.
FIGS. 3 and 4 are respectively top and bottom views of the assembly
illustrated in FIG. 2;
FIG. 5 is a schematic diagram of details of the controller illustrated in
FIG. 1; and
FIG. 6 is a schematic diagram of details of the controller illustrated in
FIG. 5.
FIG. 7 is a cross-sectional view of a cooling ring assembly for the inner
cathode showing the bayonet cutout.
FIG. 8 is a cross-sectional view of a cooling ring assembly for the outer
cathode showing the bayonet cutout.
FIG. 9 is a partial cross-sectional view of the inner cathode.
FIG. 10 is a partial cross-sectional view of the outer cathode.
FIG. 11 shows a schematic cross-section of an integrated circuit structure
according to the invention.
DESCRIPTION OF THE PREFERRED EMBODIMENT
Reference is now made to the schematic diagram, FIG. 1, wherein magnetron
sputtering apparatus 11 is illustrated as including vacuum chamber 12,
containing enclosed sputter coating processing or depositing volume 13 in
which workpiece 14 is fixedly mounted to a heated chuck 15. Magnetic
mirror coil 17 is mounted behind the substrate so that magnetic field
lines are perpendicular to the substrate. Typically, substrate 14 is part
of an integrated circuit wafer having a relatively large diameter, such as
four to six inches, on which material is deposited for electrical
interconnection by subsequent removal of selected areas of the deposited
material. In such a situation non-magnetic material is usually deposited
on the substrate.
It is to be understood, however, that the invention is applicable to
depositing magnetic materials on substrate 14 to form devices such as
magnetic disc memories. Certain modifications of the specific structure
described in connection with FIGS. 2-4 are generally necessary to provide
optimum results for deposition of magnetic materials. Each target for
sputtering magnetic materials includes a relatively thin magnetic strip
mounted on a non-magnetic, metallic holder. The magnetic strips are
relatively thin, between one-quarter and one-half inch, so magnetic field
lines are not affected materially by them. The magnetic material is
saturated to minimize the effect thereof on the magnetic flux flowing
through it. Layers of different materials can be deposited on substrate 14
by the apparatus illustrated in FIG. 1, by suitable choice of target
materials for cathode assembly 15.
Chamber 12 includes metallic electrically conducting and grounded exterior
housing 16, made of a material having high electric conductivity. Housing
16 is a part of an anode assembly and is generally formed as a cylinder
having an axis concentric with substrate 14, in turn coaxial with target
cathode assembly 15. Targets in cathode assembly 15 are maintained at
negative high voltage potentials relative to ground by DC power supply 18.
To establish a plasma in processing volume 13 in the vicinity of cathode
assembly 15, an inert gas, typically argon, is supplied to the processing
volume from pressurized inert gas source 19. The processing volume is
evacuated by vacuum pump 20. The combination of gas source 19 and vacuum
pump 20 maintains processing volume 13 at a relatively low pressure, such
as 7 millitorr.
In the illustrated embodiment, cathode assembly 15 includes two target
elements 22 and 23, respectively, having planar, annular atom emitting
surface 24 and concave atom emitting surface 25, shaped as a side wall of
frustum of a cone having a base 47 at right angles to the longitudinal
axis of disc shaped target element 22. Surface 25 is inclined throughout
its length at an angle of 45.degree. relative to base 47. Target elements
22 and 23 are concentric to each other, having coincident axes along axis
27 of substrate 14. The particular configurations of target elements 22
and 23 are described in detail infra, in connection with FIGS. 2-4.
Separate plasma discharges are established and confined over target
elements 22 and 23. The separate discharges are confined by separate,
variable magnetic fields coupled to target elements 22 and 23 by magnetic
(preferably iron) pole piece assembly 28 in response to magnetic fields
derived from solenoidal electromagnets 29 and 30. Pole piece assembly 28
and coils 29 and 30 are axially symmetric with and concentric to axis 27,
with coil 30 being located outside of coil 29.
Pole piece assembly 28 includes disc shaped base 32, disposed at right
angles with respect to axis 27, in combination with central stud 33, and
rings 34 and 35. Stud 33 extends along axis 27, while rings 34 and 35 are
concentric with axis 27, with the stud and each ring extending
longitudinally from base 32 towards substrate 14. Stud 33 is centrally
located, in a cylindrical space within coil 29, while ring 34 extends
between coils 29 and 30. Ring 35 is outside of coil 30 and target element
23. Ring 35 includes inwardly directed flange 36, at right angles to axis
27. Ring 34 is proximate the outer diameter of annular target element 22
and a lower surface of target element 23, while central stud 33 is
proximate the inner diameter of target element 22.
Separate, independently controlled currents are supplied to electromagnet
coils 29 and 30 by DC power supplies 37 and 38, respectively. Power
supplies 37 and 38 are separately controlled in response to signals
derived from controller 39 so that as target elements 22 and 23 erode
during use, the currents supplied to coils 29 and 30 change to maintain
the discharge impedances relatively constant.
To establish the separate discharges, DC power supply 18 maintains target
elements 22 and 23 at different negative DC high voltage levels -E.sub.a
and -E.sub.b, respectively. The detailed structures of pole piece assembly
28 and for supplying DC power to target elements 22 and 23 are described
infra in connection with FIGS. 2-4.
Controller 39 responds to indications of the erosion of the target assembly
including target elements 22 and 23 and the impedance of the plasma
discharge associated with one of the target elements to control the power
and impedance of the discharges as the target elements erode. The target
erosion can be determined by the total energy supplied to target elements
22 and 23, or by deriving an electric signal proportional to the current
supplied to coils 29 and 30, or by an on-line measurement of deposition
uniformity using commercially available eddy current loss measuring
devices. The discharge impedance is measured in response to the voltage
and current in the discharge. In the described embodiment, the total
energy supplied to target element 23 is computed to derive the target
erosion indication.
To these ends, DC power supply 18 includes conventional devices for
monitoring the voltage levels -E.sub.a and -E.sub.b and the currents Ia
and Ib fed by supply 18 to the leads carrying voltages -E.sub.a and
-E.sub.b. Controller 39 responds to the measurement signals from supply
18, i.e., signals E.sub.am, E.sub.bm, I.sub.am and I.sub.bm, and a signal
indicative of the total time that a target assembly has been used to
compute energy supplied to and dissipated by the target assembly and the
impedance of the discharge for target cathode 23. In response to the
computed signals, controller 39 supplies set point signals I.sub.fls and
I.sub.f2s to coil power supplies 37 and 38. In addition, controller 39
derives signals for power set point values P.sub.as and P.sub.bs of power
supply 18. Power supply 18 is constructed so that it is a constant power
device whereby the power supplied by it to target elements 22 and 23 is
constant as a function of the discharge voltage and current for the
elements. Thereby, the currents and voltages coupled by supplies 18 to
target elements 22 and 23 vary as functions of the values of P.sub.as and
P.sub.bs. As the target assembly including elements 22 and 23 erodes, the
ratio of the power in the discharges associated with the elements changes.
Initially, the ratio of the power in the discharge for elements 22 and 23
is relatively low; the power ratio of the discharges for elements 22 and
23 increases as the target elements erode. For example, in one actual
configuration, the initial ratio of the power supplied to the discharges
for target elements 22 and 23 is 1:5, while the final ratio is 1:12; the
power P.sub.b supplied to target element 23 exceeds the power P.sub.a
supplied to target element 22.
In general, the DC currents supplied to coils 29 and 30 and the
construction of pole piece assembly 28 establish magnetic flux lines in
target elements 22 and 23 which intersect emitting surface 24 and pass in
a first generally vertical direction, e.g. upwardly, through the boundary
of annular emitting surface 24 in proximity to the outer diameter of the
emitting surface. The same flux lines pass in a second generally vertical
direction, e.g. downwardly, through emitting surface 24 in proximity to
the inner radius of the emitting surface. Similarly, the flux lines which
pass through emitting surface 25 toward axis 27 in proximity to the outer
radius of the emitting surface also pass back into target element 23 at
the inner radius of the target element. Thereby, separate plasma
discharges are contained above the emitting surfaces 24 and 25 and the
erosion profiles of target elements 22 and 23 are centered on the emitting
surfaces of the targets. The angle between the magnetic field lines
traversing the boundaries defined by surfaces 24 and 25 is maintained very
low by the magnetic pole piece assembly 28 so that the magnetic field is
very uniform over emitting surfaces 24 and 25. It is important to maintain
the plasma density as uniform as possible immediately over emitting
surfaces 24 and 25 to provide uniform erosion from the emitting surfaces,
and thereby minimize the tendency for a "V" erosion profile which induces
target self-shadowing by emitted material. Self-shadowing is a phenomenon
whereby material emitted or sputtered from the target collects on the
target and has a tendency to prevent escape of further material from the
target toward the substrate.
The magnetic field coupled by coil 29 to pole piece assembly 28 causes
magnetic flux to flow through a first magnetic circuit. The flux in the
first magnetic circuit flows axially along ring 34, thence radially inward
through target element 22 and slightly above emitting surface 24 thereof.
From target element 22 and the space immediately above emitting surface
24, the magnetic flux flows radially inward to stud 33, thence axially
along stud 33 to base 32. In base 32, the first magnetic circuit is
completed by the flux flowing radially back to ring 34.
The magnetic flux established by electromagnet 30 flows through a second
magnetic circuit. The flux in the second magnetic circuit flows axially
through ring 34, into target element 23. The magnetic flux flows in target
element 23 and slightly above emitting surface 25 thereof, thence into
ring 35 through flange 36. In ring 35, the magnetic flux flows axially
back to base 32, where it flows radially inwardly to ring 34 to complete
the second magnetic circuit. The directions of windings of electromagnets
29 and 30 and the polarities of the currents applied to the electromagnets
by power supplies 37 and 38 are such that the first and second magnetic
circuit fluxes in ring 34 flow in the same direction. The flux level in
ring 34 is kept below saturation; ring 34 is appreciably thicker than ring
35 for that reason.
If target elements 22 and 23 are magnetic, sufficient current is supplied
by power supplies 37 and 38 to electromagnets 29 and 30 to saturate the
magnetic targets so that fringing fields subsist above the targets to
confine the plasma immediately above emitting surfaces 24 and 25.
Targets 22 and 23 are located relative to each other and spaced from
substrate 14 to enable material to be coated uniformly across the surface
of the substrate. The relative sputter rates from surfaces 24 and 25 are
adjusted during the life of device 11 through the adjustment of power set
points P.sub.as and P.sub.bs which respectively cause supply 18 to supply
powers P.sub.a and P.sub.b to targets 22 and 23. The values of P.sub.as
and P.sub.bs maintain uniform deposition on different ones of substrates
14 as emitting surfaces 24 and 25 of targets 22 and 23 erode.
Target elements 22 and 23, as well as pole piece assembly 28, are cooled,
in a manner described infra in detail, in connection with FIGS. 2-4. The
same structure which cools target elements 22 and 23 supplies DC operating
voltages from supply 18 to them. The structures which supply cooling fluid
to pole piece assembly 28 also assist in supporting the pole piece
assembly.
Reference is now made to FIGS. 2-4 of the drawing, wherein there are
illustrated detailed views of cathode assembly 15. It is noted from a
comparison of FIGS. 2 and 3 that the cross-sectional view of FIG. 2 is
along a rather circuitous path indicated by the dotted line 2--2, FIG. 3;
such a cross-sectional view enables the most important features of cathode
assembly 15 to be most clearly illustrated.
Disc like target element 22, in addition to including planar, annular
emitting surface 24, includes tapered interior surface 41 that flares
outwardly from axis 27 as it extends in a generally longitudinal direction
of target 22 toward planar surface 42, opposite from and parallel to
emitting surface 24. The outer perimeter of target 22 includes an axially
extending segment 43 which intersects surface 42, as well as radially
extending rim 44, which is disposed parallel to surfaces 24 and 42.
Extending generally axially between surface 24 and rim 44 is an outer
perimeter surface including beveled surface 45. On axially extending wall
segment 43 are two diametrically opposed holes 46, each of which receives
a (preferably) non-magnetic pin to assist in holding target element 22 in
situ; preferably the pins in cut-out segments 46 are formed of a
beryllium-copper alloy.
Target element 23 is formed as a ring having concave emitting surface 25,
in combination with base 47 and cylindrical side wall 48. Base 47 and side
wall 48 are respectively at right angles to and parallel with axis 27.
Concave emitting surface 25 is formed as a wall of a frustum of a cone
that is inclined 45.degree. with respect to base 47 and wall 48 throughout
the length of the surface. Diametrically opposed holes 49 in side wall 48
receive non-magnetic beryllium-copper alloy pins to hold target element 23
in situ.
Target elements 22 and 23 are arranged so that the outer radius of planar
annular emitting surface 24, having a radius R.sub.2, is less than the
inner radius R.sub.3, of inclined emitting surface 25. Of course, the
outer radius R.sub.4 of emitting surface 25 is greater than radius
R.sup.3, and the inner radius, R.sub.1, of surface 24 is less than radius
R.sub.2.
On target element 22, there is an 0.03 inch flat 419 parallel to axis 27
formed at the intersection of the emitting surface 24 and the tapered
interior surface 41. The inner radius R.sub.1 at the emitting surface 24
thus formed is 0.49 inches. The inner radius R.sub.5 at the annular rear
surface 42 is 0.72 inches. There is another 0.03 inch flat 421 parallel to
the axis 27 formed at the intersection of the emitting surface 24 and the
beveled surface 45.
The outer radius R.sub.2 thus formed at the emitting surface 24 is 3.125
inches. The beveled surface 45 is beveled at an angle of 34 degrees to the
axis 27 or an angle A of 56 degrees to the emitting surface 24. The radius
R.sub.6 to the axially extending wall segment 43 is 2.72 inches and the
thickness T3 of the axially extending wall segment 43 is 0.375 inches. The
overall thickness T.sub.4 of the target element T.sub.4 is 0.600 inches.
The pin hole 46 is a distance H of 0.162 inches above the annular rear
surface 42.
On target element 23, there is a first 0.03 inch flat 427 parallel to axis
27 formed at the intersection of the emitting surface 25 and the base 47
and a second 0.03 inch flat 429 parallel to the base 47 formed at the
intersection of the side wall 48 and the emitting surface 25. The inner
radius R.sub.3 of the ring thus formed is 3.38 inches, the outer radius
R.sub.4 is 4.84 inches, and the thickness T.sub.2 from the second flat 429
to the base 47 is 1.470 inches. The center of the holes 49 are a distance
D of 0.352 inches above the base 47. There is an angle B of 45 degrees
between the emitting surface 25 and the base 47.
As illustrated in FIGS. 2, 2A and 2B, pole piece assembly 28 includes
several individual structures whereby central pole piece stud 33,
intermediate pole piece ring 34 and outer pole piece ring 35 are mounted
on and secured to base 32 by screws 51. Coils 29 and 30 are mounted on
base 32, with current being supplied to the coils from supplies 37 and 38
by identical feedthrough assemblies 52.
As illustrated in FIGS. 2, 2A and 2B, one of assemblies 52 includes
electric insulating sleeve 53 having a relatively thick metallic coating
54 on the interior wall thereof into which is threaded metal screw 55 that
bears against metal flat washer 56. A terminal lug (not shown) is
connected to a lead between the head of screw 55 and washer 56 to a
terminal of power supply 37. To electrically insulate the lug from the
remainder of the sputtering device, dielectric washer 57 is interposed
between washer 56 and the top surface of sleeve 53.
To assist in providing the desired magnetic field shape, central pole piece
stud 33 is cylindrically shaped, having an upward, inwardly inclined
segment that is capped by magnetic metallic (preferably ferromagnetic
stainless steel) pole piece insert 69. Upper portion 58 of studs 33 and
insert 69 are both inclined with respect to axis 27 by the same angle as
the inclination angle of inner surface 41 of target 22. As a result, there
is a constant spacing between portion 58 and insert 69 to assist in
preventing penetration of plasma and sputtered metal into the underside of
the source. Cap 58 is held in situ on stud 33 by non-magnetic, preferably
austenitic stainless steel screw 59.
Ring 34 includes upper and lower segments, having walls parallel to axis
27, and a central segment with an interior wall that is inclined outwardly
with respect to axis 27. Magnetic field saturation in the lower portion of
ring 34 is avoided because of the relatively large cross sectional area
presented by it to the magnetic flux flowing through it.
Ring 35 has walls of constant thickness throughout substantially the entire
length thereof. At the upper end of ring 35 is inwardly extending flange
36, formed of two separate abutting metallic elements, namely exterior
magnetic pole piece insert 61 and shield 62 are spaced from outer wall 48
of target 23, whereby a gap having a constant separation between the
target and pole pieces is established.
To couple magnetic flux from intermediate ring 34 to both of targets 22 and
23, middle pole piece insert 64 is mounted by metal non-magnetic,
preferably austenitic stainless steel, screws 65 on the top surface of the
intermediate ring. Pole piece 64 is configured to provide a constant gap
between it and the opposing surfaces 45 and 47 of targets 22 and 23. To
this end, pole piece insert 64 includes an outwardly tapered inner
cylindrical like wall 365 which extends from a plane below the plane of
target surface 24 to the top of the pole piece insert. The top of pole
piece 64 is defined by planar annulus 66, disposed parallel to bottom
surface 47 of target 23. Surface 66 extends radially outward from axis 27
from a point just outside of the intersection of emitting surface 25 and
planar surface 47 of target 23 to a point approximately one-quarter of the
length of surface 47 in the radial direction. The geometry provides a
constant gap between pole piece insert 64 and each of targets 22 and 23.
Target cathodes 22 and 23 are maintained at different high voltage negative
potentials relative to grounded pole piece assembly 28, with target 22
being maintained at a voltage of -E.sub.a and target 23 being maintained
at a potential of -E.sub.b. In the presence of plasma, electric lines of
force exist along surfaces 24 and 25 of targets 22 and 23, respectively,
as well as in the previously mentioned gaps between targets 22 and 23 and
the adjacent pole piece elements, namely central pole piece insert 69 on
central pole piece 33, middle pole piece insert 64, and outer pole piece
insert 61 and shield 62.
Target 22 is supplied with a voltage of -E.sub.a by axially extending
metal, non-magnetic | | |