U.S. patent application number 13/080244 was filed with the patent office on 2011-07-28 for redundant anode sputtering method.
This patent application is currently assigned to VON ARDENNE ANLAGENTECHNIK GMBH. Invention is credited to Goetz GROSSER, Frank MEISSNER, Falk MILDE, Enno MIRRING, Goetz TESCHNER.
Application Number | 20110180390 13/080244 |
Document ID | / |
Family ID | 37873198 |
Filed Date | 2011-07-28 |
United States Patent
Application |
20110180390 |
Kind Code |
A1 |
TESCHNER; Goetz ; et
al. |
July 28, 2011 |
REDUNDANT ANODE SPUTTERING METHOD
Abstract
A method is provided for coating a substrate with the aid of a
magnetron cathode and two electrodes which are alternately impinged
upon by a positive potential and a negative potential. Also
disclosed is an assembly for coating a substrate, comprising a
vacuum chamber, a magnetron cathode, two electrodes, and a voltage
source. A negative potential is generated at a level that is no
greater than the level of the cathode potential, thus preventing
the electrode that is to be cleaned from being stripped to a
greater extent than the same was coated in the previous half-wave.
The magnetron cathode and the electrodes are connected to the
voltage source via switching elements without being galvanically
such that a negative and a positive voltage generated from the
voltage source can be alternatively applied to the electrodes, the
level of said voltage being no greater than the cathode
voltage.
Inventors: |
TESCHNER; Goetz; (Dresden,
DE) ; MILDE; Falk; (Dresden, DE) ; MIRRING;
Enno; (Amsdorf OT Kleinwolmsdorf, DE) ; MEISSNER;
Frank; (Coswig, DE) ; GROSSER; Goetz;
(Dresden, DE) |
Assignee: |
VON ARDENNE ANLAGENTECHNIK
GMBH
Dresden
DE
|
Family ID: |
37873198 |
Appl. No.: |
13/080244 |
Filed: |
April 5, 2011 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
12092279 |
Jul 31, 2008 |
|
|
|
PCT/DE2006/001942 |
Nov 6, 2006 |
|
|
|
13080244 |
|
|
|
|
Current U.S.
Class: |
204/192.12 |
Current CPC
Class: |
C23C 14/354 20130101;
H01J 37/3438 20130101; H01J 37/3444 20130101; H01J 37/3405
20130101; H01J 37/3408 20130101; H01J 37/32532 20130101; C23C
14/564 20130101 |
Class at
Publication: |
204/192.12 |
International
Class: |
C23C 14/35 20060101
C23C014/35 |
Foreign Application Data
Date |
Code |
Application Number |
Nov 4, 2005 |
DE |
10 2005 053 070.2 |
Feb 15, 2006 |
DE |
10 2006 007 186.7 |
Claims
1. Method for coating a substrate by use of redundant anode
sputtering of a target on a cathode of a magnetron, wherein the
cathode is supplied with a cathode potential; and, besides the
cathode, two electrodes are alternately supplied with a positive
potential or with a negative potential, wherein the negative
potential is generated at a level that is no greater than level of
the cathode potential.
2. Method, as claimed in claim 1, further comprising: generating an
alternating voltage, from which the cathode potential is generated
as a pulsing direct voltage without being electrically isolated;
and wherein each negative half cycle of said alternating voltage is
applied in an alternating manner to one electrode respectively, and
a respective positive half cycle of the alternating voltage is
applied to another respective electrode at a level that is
decreased.
3. Method, as claimed in claim 2, wherein a level of the negative
half cycle at an electrode is decreased.
4. Method, as claimed in claim 3, wherein the level of the
potential at the electrodes is decreased in an adjustable
manner.
5. Method, as claimed in claim 1, further comprising: generating a
direct voltage, which supplies in an electrically non-isolated
manner the cathode with a negative direct voltage as the cathode
potential, and said direct voltage supplies one electrode
respectively with a negative potential and a respective other
electrode with a potential, a level of which is decreased in
comparison to a level of a positive potential of the direct
voltage.
6. Method, as claimed in claim 1, further comprising: generating a
first direct voltage and a second direct voltage, which supplies
alternatingly in an electrically non-isolated manner the cathode
with a negative direct voltage as the cathode potential, and the
first and second direct voltages supply one electrode respectively
with a negative potential and a respective other electrode with a
potential, a level of which is decreased in comparison to a level
of a positive potential of the first and the second direct
voltage.
7. Method, as claimed in claim 6, wherein a level of the negative
potential is decreased in comparison to a level of a negative
potential of the first direct voltage and the second direct
voltage.
8. Method, as claimed in claim 5, wherein a level of the potential
at the electrodes is decreased in an adjustable manner.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application is a divisional of co-pending U.S. Ser. No.
12/092,279, filed Jul. 31, 2008, which is a national stage filing
under section 371 of International Application No.
PCT/DE2006/001942, filed on Nov. 6, 2006, and published in German
on May 10, 2007 as WO 2007/051461 and claims priority of German
application Nos. 10 2005 053 070.2 filed on Nov. 4, 2005, and 10
2006 007 186.7 filed on Feb. 15, 2006, the entire disclosure of
these applications being hereby incorporated herein by
reference.
BACKGROUND ART
[0002] The invention relates to a method for coating a substrate by
the use of redundant anode sputtering of a target on a cathode,
designed as a magnetron. In this case the cathode is supplied with
a negative cathode potential; and, besides the cathode, two
electrodes are alternately supplied with a positive potential
(anode potential) or with a negative potential.
[0003] The invention also relates to a device, which is intended
for coating a substrate by the use of redundant anode sputtering
and which comprises a vacuum chamber, a magnetron cathode, two
electrodes and a voltage source.
[0004] Vacuum coating technology for optical applications requires
transparent oxide layers. According to the current state of the
art, these layers are produced, as a function of the material, with
direct current methods or with alternating current methods.
[0005] To this end, magnetron cathodes--planar or cylindrical--are
provided with material to be deposited and are operated in a pure
argon atmosphere or in a reactive atmosphere in a cathode
sputtering process (sputtering process).
[0006] The major problem with a sputtering process for oxides or
nitrides lies in the fact that unfortunately not only the
substrates but also all other surfaces in the coating zone are
coated with material having poor conductivity. This coating also
takes place on the anodes, which are used in the sputtering
process. This coating of the anodes with a material having poor
conductivity or with a non-conductive material impedes the current
flow and in extreme cases may even totally suppress the flow of
current.
[0007] Outside the vacuum this coating with a material having poor
conductivity may be detected by an increase in the anode voltage.
This additional voltage drop causes a loss of power and leads to
instability in the coating process.
[0008] When higher requirements are imposed on the layer thickness
uniformity of the substrate to be coated, it turns out that the
anode, which is covered with material, results in a non-uniformity
of the layer thickness. The reason for this non-uniformity of the
layer thickness lies in the fact that the anode is not uniformly
coated by the insulating material, so that the current flows
preferably to certain zones of the anodes. This non-uniform current
flow over the length of the anodes is reflected in the plasma
distribution of the sputtering cathode; and then the plasma
concentrates on the area that continues to be the best
conductive.
[0009] In addition, this current distribution is not constant in
terms of time, so that the distribution of the layer thickness
changes over time.
[0010] In order to ensure a stable current distribution on the
anodes, a wide variety of attempts have been made--for example, the
EP 0 632 142, WO 92//09718--however, in the final end without any
outstanding success.
[0011] One solution to these problems is the twin magnetron system,
wherein the discharge is carried out with alternating current
between two identical magnetrons. Both targets are operated
alternatingly as an anode and as a cathode. In the cathode phase
the surface is cleaned of the back coatings from the anode phase,
so that the discharge always meets with an uncoated anode. In this
way the problem of the coated anode is solved for this system.
However, twin magnetrons that are operated in the alternating mode
are associated with not only increased complexity but also with
technological drawbacks.
[0012] If a magnetron is operated as an anode, the anode voltage is
higher than in the case of an anode without a magnet system. The
electrons of the discharge are impeded from penetrating into the
target surface by the magnet system. This impedance by the magnetic
field affects the current distribution in the discharge. The result
is a stationary non-homogeneity in the discharge, which is called
the "cross corner effect" in the scholarly literature.
[0013] It is true that the twin magnetron has significantly
improved the temporal stability, but the local layer thickness
uniformity has become worse in comparison to that in a single
magnetron.
[0014] This drawback can be remedied with a RAS circuit, as
described in U.S. Pat. No. 6,183,605 B1. RAS stands for redundant
anode sputtering--that is, cathode sputtering with an additional
anode.
[0015] To this end, the circuit, depicted in FIG. 1 (state of the
art) is used. In this case the magnetron is connected to the
central tap, and the electrodes are connected to one of the
external connectors respectively of the secondary coil of a
transformer; and its primary coil is fed by the medium frequency
generator Vmf.
[0016] The magnetron always remains negative; and the two
electrodes alternate the polarity.
[0017] Whereas a first electrode acts as the "correct" anode in the
discharge (that is, accepts a positive voltage in relation to the
vacuum tank), the second electrode will have, according to the law
of transformers, twice the voltage of the magnetron and, thus, be
extremely negative. Thus, this second electrode draws positive ions
from the magnetron discharge; and said positive ions lead to the
ion bombardment of the second electrode. The result is that the
electrode is ion-etched.
[0018] In the next half cycle the polarity of the electrodes is
reversed so that at this stage the discharge is provided with a
clean anode.
[0019] The problem here is that owing to the transformer that is
used, the voltage at the negative electrode is permanently
defined--that is, the value of twice the burning voltage of the
magnetron.
[0020] Since the ion density in a magnetron discharge is very high,
the electrode to be cleaned experiences an extensive stripping that
is significantly greater than the coating in the preceding half
cycle.
[0021] This stripping results not only in the abrasion of the
electrodes but also in the contamination of the layers to be
generated with the magnetron sputtering device.
[0022] It has been proposed to make the electrodes of the same
material as the target of the magnetron. However, that leads to
problems in the case of targets that have low conductivity or
targets that are made of brittle materials that cannot be
processed. Owing to these limitations RAS technology, which has
been known for a long time, could not gain acceptance.
[0023] The solution of "hiding" the anode leads to an analogous
situation. The basic idea of this method, which has been practiced
for a long time, is to arrange the anodes behind apertures, so that
the sputtered particles can reach the anode only after multiple
surges. If the opening to the cathode is adequately small, the
dwell time of the anodes can be significantly increased. Of course,
the non-uniformity of the layer thickness has to be accepted,
because for energy-related reasons the electron currents in the
quasi-neutral plasma of the sputter discharge have to be
concentrated into single paths, which in turn result in a degree of
ionization that varies widely and, thus, in locally different
coating rates. In the case of past requirements, which were lower,
this was the way to coat on an industrial scale substrates with
materials having poor conductivity. The major drawback is that the
aforementioned electron paths are locally unstable, so that the
layer thickness distribution on the substrates varies in an
unpredictable way.
[0024] The classic arrangement for anodes is disclosed in U.S. Pat.
No. 4,046,659. In this case the anode carrying rod is located
somewhat further away from the substrate than the target next to
the cathode. This position is good from an electric viewpoint,
because the charge carriers have to travel only a very short
distance, but the anode surface is also directly opposite the
substrate so that all of the particles starting from the anode land
on the substrate. In addition, a sizeable portion of the scattering
vapor travels from the magnetron cathode to this anode.
BRIEF SUMMARY OF INVENTION
[0025] Against this background, the object of the invention is to
increase the quality of the substrate coatings by increasing the
layer thickness uniformity and by decreasing the substrate
contamination caused by the redundant anodes.
[0026] The invention achieves this object with a method, in which
the negative potential is generated at a level that is no greater
than the level of the cathode potential, thus preventing the
electrode to be cleaned from being stripped to a greater extent
than the same was coated in the pervious half cycle.
[0027] A special embodiment of the method provides that alternating
voltage is generated and that the cathode potential is generated
from this alternating voltage as the pulsing direct voltage without
being electrically isolated. Each negative half cycle of this
alternating voltage is applied in an alternating manner to one
electrode respectively, whereas the respective positive half cycle
of the alternating voltage is applied to the other respective
electrode at a level that is decreased. As a result, the voltage at
the electrodes never exceeds the normal anode voltage or the
magnetron burning voltage.
[0028] An advantageous embodiment provides that the level of the
negative half cycle at an electrode is decreased. As a result, it
can be ensured that the material removal from the electrode will be
no higher than that which would result in contamination.
[0029] Therefore, it is especially desirable that the level of the
potential at the electrode is decreased in an adjustable way.
[0030] In addition to generating alternating voltage, it is also
possible to generate direct voltage, which supplies in an
electrically non-isolated manner the cathode with a negative direct
voltage as the cathode potential, whereas this direct voltage
supplies one electrode respectively with a negative potential and
the respective other electrode with a potential, the level of which
is decreased in comparison to the level of the positive potential
of the direct voltage.
[0031] In order to separate the generation of voltage for the
sputtering process from the ion etching of the electrodes, another
embodiment of the invention provides that a first and a second
direct voltage is generated and that this direct voltage supplies
alternatingly in an electrically non-isolated manner the cathode
with a negative direct voltage as the cathode potential, whereas
these direct voltages supply one electrode respectively with a
negative potential and the respective other electrode with a
potential, the level of which is decreased in comparison to the
level of the positive potential of the first and the second direct
voltage.
[0032] In order to achieve the goal of avoiding uncontrolled
etching, it is provided that the level of the negative potential is
decreased in comparison to the level of the negative potential of
the first and the second direct voltage.
[0033] The object of the invention is achieved with a device, in
which the magnetron cathode and the electrodes are connected
without electrical isolation to the voltage source by means of
switching elements in such a manner that a negative and positive
voltage, generated from the voltage source, can be applied
alternatingly to the electrodes; and that the level of said
negative and positive voltage is no greater than the cathode
voltage.
[0034] One embodiment with the generation of an alternating voltage
provides that the voltage source is designed as an alternating
voltage source (V.sub.mf) comprising a first and a second voltage
output. The first voltage output is connected to a cathode of a
first diode (V.sub.1); and the second voltage output is connected
to a cathode of a second diode (V.sub.2). The anodes of the first
(V.sub.1) and the second diode (V.sub.2) are connected together to
the magnetron cathode. The first voltage output is connected
directly to the first electrode; and the second voltage output is
connected directly to the second electrode. Furthermore, the first
electrode is connected to ground by way of a first diode/resistor
series connection (V.sub.zh1, R.sub.zh1); and the second electrode
is connected to ground by way of a second diode/resistor series
connection (V.sub.zh2, R.sub.zh2). In this way the voltages at the
electrodes shuttle between the positive anode voltage and the
negative magnetron burning voltage.
[0035] One embodiment of the inventive device provides that the
voltage source is designed as an alternating voltage source
(V.sub.mf) with a first and a second voltage output. The first
voltage output is connected to a cathode of a first diode
(V.sub.1); and the second voltage output is connected to a cathode
of a second diode (V.sub.2). The anodes of the first (V.sub.1) and
the second diode (V.sub.2) are connected together to the magnetron
cathode. The first voltage output is connected to the first
electrode by means of a third diode (V.sub.3), which, in the event
of a positive voltage at the first voltage output, is poled in the
flow direction, and which is bridged with a first resistor
(R.sub.1). The second voltage output is connected to the second
electrode by means of a fourth diode (V.sub.4), which in the event
of a positive voltage at the first voltage output is poled in the
flow direction and which is bridged with a second resistor
(R.sub.2). The first electrode is connected to ground by way of a
first diode/resistor series connection (V.sub.zh1, R.sub.zh1); and
the second electrode is connected to ground by way of a second
diode/resistor series connection (V.sub.zh2, R.sub.zh2). This
configuration makes it possible to control specifically the etching
procedure on the previously coated electrode.
[0036] The first (R.sub.1) and/or the second resistor (R.sub.2)
is/are designed in a practical way as the adjustable resistor.
[0037] However, it is also possible to design both the first and
the second resistor as variable resistors. To this end, the first
resistor is designed as a variable resistor in the form of a
drain/source zone of a first transistor (V.sub.5), whose gate is
connected to the middle of a series connection, which runs parallel
to the drain/source zone and which comprises a first Zener diode
(V.sub.7) and a third resistor (R.sub.3). The second resistor is
designed as a variable resistor in the form of a drain/source zone
of a first transistor (V.sub.6), whose gate is connected to the
middle of a series connection, which runs parallel to the
drain/source zone and which comprises a second Zener diode
(V.sub.8) and a fourth resistor (R.sub.4).
[0038] An advantageous embodiment provides that the first (V.sub.5)
and the second transistor (V.sub.6) are designed as insulated gate
bipolar transistors (IGBT).
[0039] Furthermore, it is desirable to design the alternating
voltage source (V.sub.mf) as a medium frequency voltage source.
[0040] Another embodiment of the invention provides that the
voltage source is designed as a direct voltage source (V.sub.g1)
with a negative and a positive voltage output. The negative voltage
output is connected to the magnetron cathode by way of a first
switch (S.sub.1); and the positive voltage output is connected to
the magnetron cathode by way of a second switch. The negative and
the positive voltage output is connected to the first and the
second electrode by way of a switch quadripole comprising a third
(S.sub.3), a fourth (S.sub.4), a fifth (S.sub.5), and a sixth
(S.sub.6) in the form of a bridge circuit, where each bridge branch
exhibits one of the switches. The first electrode is connected to
ground by way of a first diode/resistor series connection
(V.sub.zh1, R.sub.zh1); and the second electrode is connected to
ground by way of a second diode/resistor series connection
(V.sub.zh2, R.sub.zh2).
[0041] In order to supply separate voltage for the sputtering and
the ion etching, the voltage source is designed as a first
(V.sub.g11) and as a second direct voltage source (V.sub.g12) with
one negative and one positive voltage output respectively. In this
case both positive voltage outputs are connected together. The
negative voltage output of the second direct voltage source
(V.sub.g12) is connected to the magnetron cathode by way of a first
switch (S.sub.1); and the positive voltage outputs are connected to
the magnetron cathode by way of a second switch. The negative
voltage output of the first direct voltage source (V.sub.g11) and
the positive voltage outputs are connected to the first and the
second electrode by way of a switch quadripole comprising a third
(S.sub.3), a fourth (S.sub.4), a fifth (S.sub.5), and a sixth
(S.sub.6) in the form of a bridge circuit, where each bridge branch
exhibits one of the switches. The first electrode is connected to
ground by way of a first diode/resistor series connection
(V.sub.zh1, R.sub.zh1); and the second electrode is connected to
ground by way of a second diode/resistor series connection
(V.sub.zh2, R.sub.zh2).
[0042] The fulfillment of the problem, proposed by the invention,
can also be effectively facilitated by means of a physical
configuration of the electrodes, which shall be explained below. In
this case it is provided, first of all, that the magnetron cathode
exhibits the shape of an elongated magnetron
[0043] and that the electrodes are shielded by means of a shielding
parallel to the elongation and by a substrate that is situated
opposite the target of the magnetron.
[0044] With such an embodiment it is possible to achieve that the
charge carriers may reach unobstructed the electrodes, which are
switched as the anodes, without causing thereby the formation of
locally different plasma concentrations in front of the electrodes,
acting as the anode.
[0045] One embodiment provides that the electrodes are disposed
with their shielding laterally next to the magnetron.
[0046] It is especially desirable to use the dark space shield,
which already exists in any event, as the shielding of the
electrodes by connecting the shielding to a dark space shield of
the magnetron.
[0047] Since the shielding is heated as a consequence of the
particle bombardment of the plasma, it is expedient to provide the
shielding with water cooling.
[0048] In order to achieve even higher efficiency, it is provided
that the electrodes are also arranged on the narrow sides of the
elongated magnetron, preferably as the ring electrodes.
[0049] In an optimal design the shielding envelops the electrodes
while simultaneously leaving unobstructed a slit. In this case it
is desirable for production purposes that the shielding consists of
a rectangular tube having a slit.
[0050] Especially practical is the arrangement of the shielding
laterally next to the magnetron, where the slit is configured on
the side facing away from the magnetron.
[0051] In the case of an elongated magnetron it may be desirable to
arrange the electrodes with their shields on the side of the tube
magnetron that faces away from the substrate, and that the slits of
the shields, which are located opposite each other, point at each
other.
[0052] In order to achieve even higher efficiency, it is provided
that the spacing between the electrodes is adjusted in relation to
each other, and that the distance between the electrodes and the
shield is adjusted so as to prevent the formation of plasma.
[0053] Therefore, it is advantageous that the spacing between the
electrodes ranges from 4 to 10 mm; and that the distance between
the electrodes and the shield ranges from 4 to 10 mm.
[0054] Finally it is also possible to cool the electrodes directly
by designing the electrodes in the shape of a tube and so as to
convey coolant.
BRIEF DESCRIPTION OF THE DRAWING FIGURES
[0055] The invention is explained in detail below with reference to
one embodiment. In the associated drawings
[0056] FIG. 1 depicts a circuit configuration for redundant anode
sputtering in accordance with the state of the art.
[0057] FIG. 2 depicts an inventive circuit configuration with a
simple diode coupling to the alternating voltage source.
[0058] FIG. 3 depicts an inventive circuit configuration with an
adjustable potential at the electrodes.
[0059] FIG. 4 depicts an inventive circuit configuration with
variable resistors.
[0060] FIG. 5 depicts an inventive circuit configuration with a
direct voltage source.
[0061] FIG. 6 depicts an inventive circuit configuration with two
direct voltage sources.
[0062] FIG. 7 depicts an inventive circuit configuration, as
depicted in FIG. 3, with a bridging R/C series connection.
[0063] FIG. 8 depicts an inventive circuit configuration with a
direct voltage source and variable resistors.
[0064] FIG. 9 is a front view of a magnetron cathode with
electrodes, which are laterally arranged.
[0065] FIG. 10 depicts the insertion of the electrodes into the
vacuum chamber.
[0066] FIG. 11 is a bottom view of the magnetron cathode.
[0067] FIG. 12 is a front view of a magnetron cathode with a
ring-shaped arrangement of the electrodes.
[0068] FIG. 13 is a bottom view of the configuration, according to
FIG. 12.
[0069] FIG. 14 depicts an inventive configuration with a tube
magnetron.
[0070] FIG. 15 is a bottom view of the configuration, according to
FIG. 14.
[0071] FIG. 16 is a front view of a second inventive embodiment
with the use of a tube magnetron with a slit that is located
externally.
[0072] FIG. 17 is a bottom view of the embodiment, according to
FIG. 16.
[0073] FIG. 18 is a front view of a third inventive embodiment with
the use of a tube magnetron with a slit that is located
internally.
[0074] FIG. 19 is a bottom view of the embodiment, according to
FIG. 18.
[0075] FIG. 20 is a cross sectional view of the electrodes with
shielding; and
[0076] FIG. 21 is a cross sectional view of the electrodes with
shielding and cooling.
DETAILED DESCRIPTION
[0077] The known state of the art uses, as depicted in FIG. 1, a
transformer, which has the drawbacks described in the introductory
part.
[0078] As depicted in FIG. 2, a magnetron cathode can also be
operated by means of two diodes--V1 and V2--with two additional
electrodes. In this case the voltages at the electrodes shuttle
only between the positive anode voltage and the negative magnetron
burning voltage.
[0079] If the medium frequency generator Vmf supplies the pole 1
with a voltage, which is negative compared to pole 2 of the medium
frequency generator Vmf, then the diode V1 conducts, the diode V2
blocks and the magnetron discharge ignites in the event of an
adequate voltage level between the cathode and the electrode 2. At
this point in time the electrode 2 acts as the anode of the
magnetron discharge. Then the voltage at this electrode adjusts, in
conformity with the conditions of the magnetron discharge, to +20
to +150V.
[0080] The electrode 1 is connected directly to the pole 1 of the
Vmf. Therefore, apart from the conduction losses in the diode V1,
said electrode has the same negative voltage in relation to the
electrode 2 as the cathode.
[0081] Therefore, the effect of the ion etching on the electrode 1
takes place, but the voltage is only half the value of the
configuration with the transformer in accordance with the state of
the art in FIG. 1.
[0082] In the next half cycle the relationship reverses, so that at
this point the pole 2 of the medium frequency generator Vmf is
negative in relation to the pole 1 of the medium frequency
generator Vmf. Therefore, at this stage the diode V2 conducts; the
diode V1 blocks; and the electrode 1 acts as the anode of the
magnetron discharge and at the electrode 2, which is now at the
same potential as the cathode.
[0083] The technical implementation demands, besides the two
diodes, also the wiring elements, which limit the overvoltages,
and, in order to ignite the discharge reliably, a diode/resistor
combination, which is referred to as Vzh1 and Rzh1 and Vzh2 and
Rzh2 in FIG. 2.
[0084] Despite the voltage that is reduced to half of that in the
original design, the etching effect on the electrodes is still very
strong. Therefore, it is desirable to be able to control the
etching procedure.
[0085] The circuit in FIG. 3 shows the solution:
[0086] The diodes V3 and/or V4 are inserted into the lines to the
electrodes 1 and/or 2. These diodes are connected parallel to the
variable resistors R1 and/or R2.
[0087] If the medium frequency generator Vmf supplies the pole 1
with a voltage, which is more negative than that supplied the pole
2, then the diode V1 conducts, the diode V2 blocks and the
magnetron discharge ignites between the cathode and the electrode
2.
[0088] The diode V4 becomes conductive, whereas the diode V3
blocks. In this way a current flow between the electrode 1 and the
pole 1 from the medium frequency generator Vmf can still take place
only by way of the resistor R1. Depending on the size of the
resistor, the electrode 1 is impinged on with fewer ions and, thus,
is not etched as much.
[0089] In the next half cycle of the medium frequency voltage the
relationship reverses; and the electrode 1 becomes the anode,
whereas the electrode 2 is etched.
[0090] Owing to the size of the variable resistors R1 and/or R2,
the flow through the electrodes 1 and/or 2 and, thus, the etching
ion bombardment may be adjusted individually in such a manner as is
necessary for the process.
[0091] In order to have the freedom of electronic control, it is
desirable to use modern semiconductor components as the variable
resistance.
[0092] The circuit in FIG. 4 shows the variant, in which an IGBT
(IGBT: insulated gate bipolar transistor) is used as the variable
resistor. The Zener diodes V7 and/or V8 become conductive, starting
from a defined voltage, so that a current flows over the resistor
R3 and/or R4. The resulting voltage drop drives the IGBT V5 and/or
V6, so that exactly enough voltage remains over the Zener diodes.
As a result, the voltage at the electrode becomes less by a
constant amount than that of the Zener voltage of V7 and/or V8 at
the magnetron cathode.
[0093] The IGBT containing protective diodes may be used for this
circuit so that V3 and V5 and/or V4 and V6 are combined into one
power module.
[0094] Similarly a conventional constant current circuit, which
sets a defined ion current, can be used.
[0095] Whereas all of the aforementioned circuits must absolutely
have a medium frequency generator, the RAS principle (RAS:
redundant anode sputtering) is also possible with pulsing
devices.
[0096] The circuit in FIG. 5 shows the configuration. The classical
H bridge comprising the switches S3 to S6 cyclically reverses the
polarity of the electrodes so that the cleaning effect is generated
again.
[0097] The switches S1 and S2 offer another option that is not
available with a medium frequency feed.
[0098] In the variants, according to FIG. 2 to FIG. 4, the cathode
potential fluctuates between zero and the negative burning voltage
of the cathode during the entire operating period. The side
diagrams depicting the characteristics of the voltages over time
illustrate this behavior.
[0099] In the case of highly insulating materials like SiO.sub.2,
this permanent orientation of the potential is deleterious. It
leads to charges on the target, which in turn end in over-arcing
with electric arcs. A more reliable remedy against these charges is
to reverse the polarity of the target surface adequately fast so
that the charge carriers in the plasma are neutralized (see
Szczyrbowsko and Teschner "Reactive Sputtering of SiO2 Layers . . .
", SVC 1995).
[0100] Time 0: all switches are open.
[0101] Time 1: S1, S4, S5 closed. The discharge ignites and burns
between the cathode and the electrode 2. The electrode 1 is etched.
The electrode 1 is supplied with the same voltage as the
cathode.
[0102] Time 2: S1 and S4 are opened. The discharge is
interrupted.
[0103] Time 3: S2 is closed. The charged cathode is now more
positive than the electrode 1 and, thus, extracts the electrons
from the remaining plasma and, in so doing, discharges itself
[0104] Time 4: All of the switches are open.
[0105] Time 5: S1, S3, S6 closed. The discharge ignites and burns
between the cathode and the electrode 1. The electrode 2 is etched.
The electrode 2 is supplied with the same voltage as the
cathode.
[0106] Time 6: S1 and S3 are opened. The discharge is
interrupted.
[0107] Time 7: S2 is closed. The charged cathode is now more
positive than the electrode 2 and, thus, extracts the electrons
from the remaining plasma and, in so doing, is discharged.
[0108] Time 8: All of the switches are open.
[0109] Thereafter the cycle repeats.
[0110] IGBTs, which are present in the pulse circuits, are used for
the switches.
[0111] If in this case the object is also to achieve an ion etching
process that can be controlled, the circuit, shown in FIG. 6, can
be modified. The current supply is divided into 2 current
supplies--the first for sputtering and the second for ion etching
of the electrodes. Since the two current supplies can be adjusted
independently of each other, the voltage level at the electrodes
can also be adapted to the technological requirements during the
etching phase.
[0112] The anodes 2 and 3 in FIG. 7 are connected to an RC element
from C1 and R3. The capacitor C1 serves to return the positive
voltage to the negative branch, which is blocked by the diodes.
Whereas without a capacitor the diodes emit their stored charge
into the etching discharge and, thus, always result in a minimum
etching procedure, the charge of the diodes is conveyed past the
plasma by way of the RC element so that it is possible to totally
suppress the etching.
[0113] The resistor R3 in the RC element serves to limit the pulse
currents to the allowable value for the diodes used in the diode
module 8.
[0114] Another modification of the RAS principle is the switched
double anode system, which is shown in FIG. 8. In contrast to the
original RAS principle, in this case, instead of the medium
frequency supply 9, a direct current supply is used. In this case,
the minus pole of the direct current supply is permanently
connected to the magnetron cathode 1. The anodes 2 and 3 are
connected to the cathode by way of the resistors R3 and R4. In
addition, each anode is connected to the plus pole of the direct
current supply by way of a switch, shown as the IGBTs V5 and V6 in
FIG. 8.
[0115] If the switches V5 and V6 are closed, this configuration
functions as a classical direct current sputtering system.
[0116] If one of the switches is opened, then the potential of the
electrode, which is connected to the open switch, becomes highly
negative owing to the resistor, which connects this electrode to
the cathode, so that this electrode can no longer act as the anode
of the gas discharge, but rather acts as the additional cathode.
The other electrode, which is connected to the closed switch, takes
over the entire anode current of the discharge in the period of
time in which a switch is opened.
[0117] Since this additional cathode is no longer promoting the
magnetic field, very little current is contributed. However, the
plasma cloud that is generated by the magnetron cathode delivers a
sufficiently high amount of positive ions that an ion extraction
takes place owing to the negative potential. This ion extraction
leads to etching and, thus, to the cleaning of the anode surface.
One of the two switches must always be closed so that the
sputtering discharge finds its anode.
[0118] FIG. 8 shows an example of the pulse train, during which the
opening times of the switches t.sub.off5 and t.sub.off6 vary,
because with different opening times the electrode coating that may
or may not vary can be counteracted. The cleaning effect can be
metered by adjusting the opening times individually.
[0119] This individual adjustment is especially important because
the coating of the electrodes changes as a function of the supplied
cathode power and the prevailing working pressure.
[0120] The logical cycle duration for the repeated opening of the
switches depends on the material properties of the resulting layer.
They are in a range of a few Hertz up to 100 kHz. In the case of
highly insulating layers it must be prevented due to the fast
cleaning, that the anode can be completely coated in one
period.
[0121] The combinations of diode and resistor V1/R1 and/or V2/R2
serve to promote the initial ignition of the magnetron
discharge.
[0122] The diodes V3 and/or V4 belong to the respective IGBT and
serve to protect it against polarity reversal.
[0123] In order to achieve a uniform coating with a magnetron
system, the construction of the anode must be designed, inter alia,
in such a manner that the charge carriers may reach the anode
without any obstruction and without thereby causing locally
different plasma concentrations to develop in front of the
anodes.
[0124] This system was modified (see also FIG. 9 to FIG. 11) to the
extent that the two anode carrying rods 2 and 3, which are
necessary for the RAS system, were mounted on the two longer sides
of the cathode 1 and were separated from the substrate by means of
the areas, connected to the dark space shield 4 and 5. Another area
was inserted on the side of the anode-carrying rods that face away
from the target, so that the cross section of the part 5 exhibits a
U-shape.
[0125] The surfaces of the parts 4 and 5 are exposed to the
particle bombardment of the plasma, so that in the case of high
performance cathodes they are designed as water cooled sheet metal
plates.
[0126] FIG. 10 is a side view from the left hand side of FIG. 9;
FIG. 11 is a bottom view of FIG. 9.
[0127] The electric wiring is rendered only symbolically. The
respective upper anode 3 is connected to a pole of the medium
frequency supply 9. The respective bottom anode 2 is connected to
the other pole of the medium frequency supply 9. The diode module 8
is connected to the two poles of the medium frequency supply. The
anodes of the diode module 8 are connected to the magnetron cathode
1.
[0128] The effect of the anode and the amount of the injected
etching power can be controlled by the opening in the area 4. For
the function of the gas discharge it is necessary that the anode
drop in front of the anode carrying rods can be implemented in its
entirety. This means for the design instructions that there may be
no additional parts that restrict the plasma in a space interval of
40 to 80 mm in front of the anodes 2 and 3. In FIG. 21 this measure
is labeled a.
[0129] Therefore, the two legs of the part 5 are designed to this
length. The ends of these legs support the end plates 4, which
permit the cross section, in which the anodes 2 and 3 are located,
to become a rectangular tube, which is opened with a slit parallel
to the longitudinal expansion of the target. The width of the slit
was marked with the reference letter e in FIG. 13. The charge
carriers penetrate from the plasma space into this slit. Since the
ions and the electrons exhibit different ranges of motion, it is
possible that in the case of a small slit width e preferably the
electrons reach the anodes, but the ions are impeded.
Correspondingly the etching effect at the anodes decreases. On the
other hand, a small slit width e causes the substrate to be well
shielded against impurities from the anode.
[0130] In order to make sure that the discharge in front of the
anodes 2 and 3 propagates in the direction of the slit opening e,
the distance b between the anodes 2 and 3 and the dark space shield
5 is to be adjusted to 4 to 10 mm, so that the plasma is quenched
therein. In order to prevent a direct plasma formation between the
anodes 2 and 3, the distance between the anodes 2 and 3, which is
marked with the reference letter c in FIG. 21, is also to be
adjusted to 4 to 10 mm.
[0131] In order to be able to define the electrically effective
anode, the parts of the anodes 2 and 3 that are not to make contact
with the plasma, are covered with the insulators 6. Owing to the
proximity to the plasma space, insulating materials--for example,
quartz glass or ceramic--are to used that can withstand the
stresses, which are unavoidable in the plasma, like the ion
bombardment and ultraviolet radiation.
[0132] An even higher uniformity of the layer thickness can be
achieved if the magnetron cathode is surrounded by anodes on all
sides. FIG. 5 is a bottom view of FIG. 12, where the lateral anodes
2 and 3 were configured as rings 2 and 3. In the middle of the
rings there is located the magnetron cathode 1. The rings are
housed in the above described tube cross section 4 and 5. In
contrast to the rod design, in the ring system the cover also
extends over the face sides of the cathode so that the dark space
shields 4 and 5 also form rings.
[0133] In the case of high performance cathodes the anodes must be
water cooled, because approximately 10% of the entire discharge
power is transferred to the anodes. In the case of rod anodes the
cooling water is conveyed back and forth in the rod owing to the
double tube design. In the case of the ring anodes the water is fed
into the ring in a T-shaped segment and, after passage through the
entire ring, is carried away again at the same T-shaped
segment.
[0134] The ring design has an additional advantage over the rods.
The ring design simplifies the insulated attachment of the anodes.
FIG. 18 shows that the above described spacing c between the ring
anodes 2 and 3 in relation to each other and the distance b of the
ring anodes 2 and 3 from the dark space shield 5 are adjusted by
ceramic cylinders 10. In this case they are short ceramic rods,
which are distributed over the length of the anode rings. Owing to
the adjusted spacing, which causes a dark space quenching of the
plasma on the inside of the anode rings 2 and 3, these ceramic rods
have no influence on the homogeneity of the plasma on the front
side of the anodes.
[0135] FIG. 9 to FIG. 13 show the systems, in which the magnetron
cathode 1 is a planar magnetron. However, the same technique can be
applied to a cylindrical magnetron.
[0136] FIG. 14 to FIG. 19 show that the ring anodes 2 and 3 are
situated in a plane behind the cylindrical cathode 1, facing away
from the substrate. The electrical wiring and the geometrical
configuration of the anode tubes 2 and 3 in the dark space shield 4
and 5 are identical to those described above.
[0137] FIG. 14 and FIG. 15 show that the dark space shield 4 and 5
and the anode rings 2 and 3 envelop the entire device of the
cylindrical magnetron--that is, including the holders for the
target tube. In this case the slit opening e points outwardly.
[0138] FIG. 16 and FIG. 17 show that the dark space shield 4 and 5
and the anode rings 2 and 3 envelop only the space below the target
tube. In this case the slit opening e points outwardly.
[0139] FIG. 18 and FIG. 19 show that the dark space shield 4 and 5
and the anode rings 2 and 3 envelop only the space below the target
tube. In this case the slit opening e points inwards.
[0140] The control of the etching of the anodes by adjusting the
slit width e in the above described way can also be used to change
locally the plasma intensity by means of the locally different slit
width and, thereby, to influence the layer thickness distribution
on the substrate. Hence, the non-homogeneity, which may have been
caused by other influencing factors, may be compensated.
[0141] It follows from the description that the etching of the
anodes may be decreased, but not totally suppressed with the slit
width e.
[0142] The ideal situation would be that the undesired coating of
the anodes and the cleaning by means of etching were exactly
balanced. However, this state cannot be implemented by mechanical
means. For this reason the circuits in FIG. 3 and FIG. 4 were
expanded to include another branch, as depicted in FIG. 7.
* * * * *