Method and apparatus for ionized physical vapor deposition

[0002] This is a continuation-in-part of U.S. patent application Ser. No. 09/073,141 of John Drewery and Thomas Licata, filed May 6, 1998 now U.S. Pat. No. 6,080,287, hereby expressly incorporated by reference herein.

[0003] This invention relates to the Ionized Physical Vapor Deposition (IPVD) and, more particularly, to methods and apparatus for depositing films, most particularly metal films, onto semiconductor wafer substrates by sputtering the coating material from a target, ionizing the sputtered material, and directing the ionized coating material onto the surface of the substrates.

[0004] Ionized physical vapor deposition is a process which has particular utility in filling and lining high aspect ratio structures on silicon wafers. In ionized physical vapor deposition (IPVD) for deposition of thin coatings on semiconductor wafers, materials to be deposited are sputtered or otherwise vaporized from a source and then a substantial fraction of the vaporized material is converted to positive ions before reaching the wafer to be coated. This ionization is accomplished by a high-density plasma which is generated in a process gas in a vacuum chamber. The plasma may be generated by magnetically coupling RF energy through an RF powered excitation coil into the vacuum of the processing chamber. The plasma so generated is concentrated in a region between the source and the wafer. Then electromagnetic forces are applied to the positive ions of coating material, such as by applying a negative bias on the wafer. Such a negative bias may either arise with the wafer electrically isolated by reason of the immersion of the wafer in a plasma or by the application of an RF voltage to the wafer. The bias causes ions of coating material to be accelerated toward the wafer so that an increased fraction of the coating material deposits onto the wafer at angles approximately normal to the wafer. This allows deposition of metal over wafer topography including in deep and narrow holes and trenches on the wafer surface, providing good coverage of the bottom and sidewalls of such topography.

[0005] Certain systems proposed by the assignee of the present application are disclosed in U.S. patent applications Ser. Nos. 08/844,751, now U.S. Pat. No. 5,878,423, 08/837,551 now U.S. Pat. No. 5,800,688, and 08/844,756 filed Apr. 21, 1997, allowed and hereby expressly incorporated by reference herein. Such systems include a vacuum chamber which is typically cylindrical in shape and provided with part of its curved outer wall formed of a dielectric material or window. A helical electrically conducting coil is disposed outside the dielectric window and around and concentric with the chamber, with the axial extent of the coil being a significant part of the axial extent of the dielectric wall. In operation, the coil is energized from a supply of RF power through a suitable matching system. The dielectric window allows the energy from the coil to be coupled into the chamber while isolating the coil from direct contact with the plasma. The window is protected from metal coating material deposition by an arrangement of shields, typically formed of metal, which are capable of passing RF magnetic fields into the interior region of the chamber, while preventing deposition of metal onto the dielectric window that would tend to form conducting paths for circulating currents generated by these magnetic fields. Such currents are undesirable because they lead to ohmic heating and to reduction of the magnetic coupling of plasma excitation energy from the coils to the plasma. The purpose of this excitation energy is to generate high-density plasma in the interior region of the chamber. A reduction of coupling causes plasma densities to be reduced and process results to deteriorate.

[0006] In such IPVD systems, material is, for example, sputtered from a target, which is charged negatively with respect to the plasma, usually by means of a DC power supply. The target is often of a planar magnetron design incorporating a magnetic circuit or other magnet structure which confines a plasma over the target for sputtering the target. The material arrives at a wafer supported on a wafer support or table to which RF bias is typically applied by means of an RF power supply and matching network.

[0007] A somewhat different geometry employs a plasma generated by a coil placed internal to the vacuum chamber. Such a system does not require dielectric chamber walls nor special shields to protect the dielectric walls. Such a system is described by Barnes et al. in U.S. Pat. No. 5,178,739, expressly incorporated by reference herein. Systems with coils outside of the chamber as well as the system disclosed in the Barnes et al. patent involve the use of inductive coils or other coupling elements, either inside or external to the vacuum, that are physically positioned and occupy space between the planes of the sputtering target and the wafer.

[0008] Whether a coupling element such as a coil is provided inside or outside of a vacuum chamber, dimensions of the system have been constrained by the need for adequate source to substrate separation to allow for the installation of the RF energy coupling elements between the source and the substrate. Adequate diameter must also be available around the wafer for installation of coils or other coupling elements. As a direct result of the increased source to substrate spacing due to the need to allow space for the coupling element, it is difficult to achieve adequate uniformity of deposition with such systems. If the height of the chamber is reduced to improve uniformity there is a loss of plasma density in the central region of the chamber and the percentage of ionization of the coating material is reduced. Also, in practice, the entire system must fit within a constrained radius. As a result, there are frequently problems due to heating arising from the proximity of the RF coils to metal surfaces, which may necessitate extra cooling, which increases engineering and production costs and wastes power.

[0009] An IPVD apparatus with the coil in the chamber has the additional disadvantage that the coils are eroded by the plasma and must therefore consist of target grade material of the same type as that being sputtered from the target.

[0010] Moreover, considerable cooling of coils placed in the vacuum chamber is needed. If liquid is used for this cooling of the coils, there is danger that the coils will be penetrated by uneven erosion or by arcing, causing a resulting leak of liquid into the system, which is highly undesirable and will likely result in a long period of cleaning and requalification of the system. Furthermore, an excitation coil in the chamber also couples capacitively to the plasma, leading to inefficient use of the excitation power and to the broadening of the ion energy spectrum, which may have undesirable effects on the process.

[0011] The miniaturization of semiconductor devices has resulted in a need to form low resistance connections to contacts at the bottoms of high aspect ratio holes of a fraction of a micron in diameter. This has increased the demand for the use of highly electrically conductive metals such as copper over barrier layers of materials such as tantalum and tantalum nitride. The techniques for depositing such materials in the prior art have not been totally satisfactory.

[0012] The deposition of materials by PVD methods has, in the prior art, involved critical designs of sputtering sources to produce plasma concentrations of uniform geometries within sputtering chambers and to directly affect the distribution uniformities of the deposited films. The prior art approaches have resulted in compromises of other performance parameters to those ends.

[0013] As a result of the above considerations and problems, there remains a need for more efficiently coupling energy into the dense coating material ionizing plasma in IPVD processing systems, and to do so without interfering with the optimum dimensions of the chamber and preferably without placing a coil or other coupling element into the vacuum chamber.

[0014] One objective of the present invention is to provide an IPVD method and an IPVD apparatus in which the placement of the coil or other coupling element does not adversely affect the geometry of the chamber of the processing apparatus. Another objective of the present invention is to provide a more efficient and effective method and apparatus for the performance of IPVD.

[0015] According to the principles of the present invention, an IPVD apparatus is provided with a ring-shaped source of coating material for producing a vapor that includes atoms or minute particles of the coating material to a processing space within a vacuum chamber. At the center of the ring-shaped source is provided a coupling element for reactively coupling RF energy into the chamber to produce a high density reactively coupled plasma in the processing space to ionize coating material passing through the processing space. The ions of coating material drift, whether under the influence of electrostatic or electromagnetic fields or otherwise, toward a substrate in the chamber, at the opposite end of the processing space from the source. Those ions that arrive within a certain distance, in the order of a centimeter from the substrate for example, encounter a sheath field and are accelerated toward the substrate so that a high percentage of the coating material arrives on the substrate at angles normal to the substrate, thereby more effectively lining the bottoms and sides of, or filling, small and high aspect ratio features on the surface of the substrate.

[0016] In one embodiment of the invention, a coating material source, preferably a sputtering target, is provided with a central opening in which is placed a dielectric window. Behind the window, outside the vacuum of the chamber, is located a plasma source which includes a coupling element, preferably a coil, which is connected to the output of an RF energy source. The coupling element is configured to couple, preferably inductively, energy supplied from the energy source through the window at the opening at the center of the material source and into the region of the chamber between the coating material source and the substrate, such as a semiconductor wafer, on a substrate support at the opposite end of the chamber from the coating material source.

[0017] The apparatus of the present invention includes an annular sputtering target which surrounds a central ceramic window. This annular target is preferably frusto conical in shape. A magnetron magnet assembly is positioned behind the target to produce a plasma confining magnetic field over the target, preferably in the shape of an annular tunnel on the surface of the annular target surrounding the central opening at its center.

[0018] The coupling element is preferably a coil positioned behind and close to the back outside surface of the dielectric window at the central opening of an annular sputtering target. RF energy of, for example, 13.56 MHZ is applied to the coil to excite a high density inductively coupled plasma in the chamber between the target and the substrate. A main sputtering plasma that is trapped under the field of the magnetron magnets at the surface of the target sputters coating material from the target and into the region of the processing space occupied by the dense secondary plasma, where a substantial portion of the material is stripped of electrons to form positive ions of the coating material. A negative bias voltage is applied to a wafer on the substrate holder, which attracts the positive ions of sputtering material from the region of the secondary plasma and toward and onto the surface of the substrate, with the angles of incidence approaching being perpendicular to the substrate so that they can enter trenches and holes on the wafer substrate to coat the bottoms of these holes and trenches.

[0019] Certain embodiments of the apparatus and method of the invention include an IPVD source that employs a three dimensional coil that energizes a dense inductively coupled three dimensional plasma in three dimensional regions within the chamber. The chamber is operated at a vacuum pressure of between 30 and 130 mTorr to essentially thermalize the plasma, so that ions of coating material can be formed in the plasma and electrically directed perpendicular to and onto the substrate, thereby reducing the effect of target and magnet configuration on coating uniformity. The IPVD source is coupled through a window into the chamber through a high dielectric material such as a TEFLON spacer and then through a dielectric window such as quartz which forms the vacuum barrier closing a circular opening in the chamber wall at the center of an annular target. Inside the chamber is a Faraday shield having chevron shaped slots therein oriented relative to the conductors of the coil to pass inductively coupled RF energy into the chamber while preventing capacitive coupling from the coil to the plasma and avoiding flux compression heating. The shield has integral cooling and is formed of cast copper which is plated with aluminum, so that the shield can be reconditioned by chemically dissolving aluminum coating to remove buildup and then replating the copper shield with aluminum for reuse. The window and shield combination form a replaceable combination. In the combination, the window and shield are spaced so the window is self cleaning adjacent the slits in shield by plasma that forms at this point in the slits.

[0020] The target is preferably frusto conical, with the walls of the truncated cone inclined about 35° to the horizontal or plane of the window. A permanent magnet pack is employed which produces three, and preferably only three, magnetic tunnels over the target surface, with a main central tunnel dominating early in the target's life to erode the mean radius of the annular target and two side tunnels taking over later in the life of the target to erode grooves adjacent the inner and outer rims of the target annulus.

[0021] The apparatus preferably uses a wafer holder mounted for vertical motion on a Z-table motion drive to provide for target to substrate spacing (TSS) of from six to nine inches and to provide for wafer handoff to a transfer arm from a transfer module. The support is provided with an electrostatic chuck, and wafer heating and cooling using Peltier device remote from the support that connects through a GALDEN fluid loop with the support and through another fluid loop with a heat sink. The electrostatic chuck is tripolar with the chuck grid serving as electrodes to provide 2-zone bias to the wafer to attract the ionized sputtered material to the wafer. A shadow ring is provided around the edge of the wafer to provide non-contact edge masking.

[0022] The chamber has a removable shield insert in two parts that mechanically float relative to each other to accommodate different expansions due to different heating. The shield assembly is a replaceable subcombination. The apparatus is particularly useful for depositing copper over tantalum and tantalum nitride and for depositing the underlying tantalum and tantalum nitride barrier layer over a patterned wafer, with Ta deposited by ionized PVD and TaN deposited by PVD in the same chamber, followed by deposition of copper by ionized PVD in a similar module attached to a transfer module of the same tool. The copper so deposited is suitable to be followed by any of many methods of Cu fill, particularly by electroplating. The processes are preferably carried out using process parameters, including: pressures, temperatures, gases, bias power and/or voltage levels, sputtering power levels, IC power levels, etc., as described below.

[0023] With apparatus structure according to the invention, the processing chamber can be dimensioned to provide optimum spacing between the coating material source and the substrate to provide both good ionization of sputtered species as well as good uniformity of deposition on the wafers.

[0024] The present invention provides greater freedom of design choice in configuring the processing chamber to optimize the IPVD process, and does so while overcoming the difficulties set forth in the background above.

[0025] These and other objectives and advantages of the present invention will be more readily apparent from the following detailed description of the drawings.

[0026] FIG. 1 is a schematic cross-sectional diagram of one embodiment of IPVD apparatus according to the present invention.

[0027] FIG. 2 is a disassembled perspective view of the housing portion of the IPVD source of the apparatus of FIG. 1.

[0028] FIG. 3 is a disassembled perspective view of the target assembly portion of the IPVD source of the apparatus of FIG. 1.

[0029] FIG. 3A is a fragmentary perspective view of a portion of the cover of the target assembly of FIG. 3.

[0030] FIG. 4 is an enlarged schematic cross-sectional view in circle section area 4 of FIG. 1.

[0031] FIG. 5 is an enlarged cross-sectional view, similar to FIG. 4, illustrating an alternative magnet assembly.

[0032] FIG. 6 is a disassembled perspective view of the Faraday shield and window assembly of the IPVD source of the apparatus of FIG. 1.

[0033] FIG. 6A is a cross sectional view of the circled portion of FIG. 6.

[0034] FIG. 7 is cross sectional view of the RF source assembly of the IPVD source of the apparatus of FIG. 1.

[0035] FIG. 8 is a disassembled perspective view of the RF source assembly of FIG. 7.

[0036] FIG. 9 is an axial cross sectional view of the magnet and cooling manifold mounting assembly of FIG. 13.

[0037] FIG. 10 is an enlarged cross sectional view of a fluid coupling assembly portion of the assembly of FIG. 9.

[0038] FIG. 11 is an enlarged cross sectional view of the mounting connector assembly of the assembly of FIG. 9 for the window and shield assembly of FIG. 6.

[0039] FIG. 12 is an enlarged cross sectional view of a DC contact stub portion of the assembly of FIG. 13.

[0040] FIG. 13 is a top view of the magnet and cooling manifold mounting assembly portion of the IPVD source portion of the apparatus of FIG. 1.

[0041] FIG. 14 is a view of the IPVD source flange and dark space shield of the apparatus of FIG. 1.

[0042] FIG. 15 is a bottom perspective view of the electrostatic chuck wafer support assembly of the apparatus of FIG. 1.

[0043] FIG. 16 is a side dissassembled perspective view of the wafer support assembly portion of the wafer support of FIG. 15.

[0044] FIG. 17 is an axial cross sectional view of the wafer support assembly portion FIG. 16.

[0045] FIG. 18 is a perspective view of the wafer support vertical position adjustment assembly of the apparatus of FIG. 1.

[0046] FIG. 19 is a top perspective view of the vacuum chamber wall assembly of the apparatus of FIG. 1.

[0047] FIG. 20 is a side perspective view of the lower portion of the apparatus of FIG. 1 showing particularly the vacuum chamber wall assembly of FIG. 19 and the gas vacuum system portion of the apparatus.

[0048] FIG. 21 is a gas schematic diagram of the gas vacuum system of FIG. 20.

[0049] FIG. 22 is a disassembled perspective view of the sputter shield portion of the chamber wall assembly of the apparatus of FIG. 1.

[0050] FIG. 23 is a disassembled perspective view of the IPVD source hoist mechanism of the apparatus of FIG. 1.

[0051] An ionized physical vapor deposition apparatus 500 according to one embodiment of the present invention is illustrated in FIG. 1. The IPVD apparatus 500 includes a vacuum chamber 501 bounded by a chamber wall assembly 502. The chamber 501 is provided with an ionized physical vapor deposition (IPVD) source 503 supplying coating material in vapor form into the volume of the sputtering chamber 501 and for ionizing the sputtering material vapor, an electrostatic chuck wafer support system 507 for holding wafers during processing, a wafer handling system 504 for loading and unloading wafers for processing, a vacuum and gas handling system 505 (FIGS. 20-21) for evacuation of the chamber 501 to vacuum, an IPVD source hoist assembly 460 for removal and replacement of the target and for performing other servicing of the source, and a control system 509 which operates the other systems of the apparatus 500 in accordance with the methods and processes described herein and otherwise carried out with the apparatus 500.

[0052] The apparatus 500 is a serviceable module capable of providing features and operating conditions including the following: (1) base vacuum of less than 10⁻⁸Torr, (2) operating inert gas pressure of between 30 and 130 mTorr, (3) provision for reactive gas at partial pressure of 0-50 mTorr, (4) variable substrate to target spacing of 6 to 9 inches, (5) electrostatic chucking with backside gas heating or cooling, and (6) shielding that restricts deposition to removable, cleanable components with surfaces having good adhesion of sputtered material to prevent particle generation.

[0053] The general concepts of the source 503 are described in U.S. patent application Ser. No. 09/073,141, referred to above. The particular implementation of the source 503 includes a target 10 of the conical target type laid out in that application. Essentially the principle objectives of the source 503 include providing the following features and properties: (1) to require minimum operator effort and smallest possible set of tools to perform routine tasks, (2) to provide separation RF and DC power from water to the best extent possible, (3) to provide relative simplicity of design and operation; (4) to allow rapid repair or replacement of the source including quick replacement of the whole internal source assembly, (5) to provide modular internal assemblies, and (6) to maintain RF shielding integrity to prevent leakage of radiation into the operating environment.

[0054] The IPVD source 503 has an annular target 10 and an RF source assembly 450, which energizes an inductively coupled plasma in the chamber 501, in an opening 421 inside of the annulus of the target 10, opposite a 200 mm wafer, or 300 mm wafer, for example, which is to be mounted on an electrostatic chuck 97 of the wafer support system 507. The source 503 includes a source housing assembly 410 that includes a source housing 1, which is preferably an aluminum weldment. The source housing 1 includes structure for mounting the working parts of the source 503 and render the source 503 capable of being mounted to source hoist assembly 460, described below, for installation on, and removal from, the apparatus 500.

[0055] As illustrated in FIG. 2, DC power for the IPVD source 503 is brought into the housing 1 via a connector 2, which mates to a connector 2apermanently mounted on the chamber wall 502 when the source 503 is in the operating position, so DC cannot appear on the target 10 without the source 503 in place on the chamber 501. A negative DC feed 147 passes through watertight strain relieving bushings 149 to a socket 3 mounted in an insulating block 4 inside the housing 1 while a positive feed 148 is connected directly to the housing 1.

[0056] Ports 5 are provided in the top of the housing 1 through which pass conductors 40 (FIG. 7) for RF connections 152 to an RF tuner 96 (FIG. 2) of an inductively coupled plasma (ICP) generator 450 (FIGS. 6-8) described below. The housing 1 is also provided with an interlock switch 6 that detects the presence of the RF tuner unit 96. An interlock switch 7 is provided that is actuated by a pushrod mechanism (not shown), which is actuated by the target 10 pressing on it to ensure that water and power cannot be connected unless the target 10 is in place and locked down. A canted coil spring 8 is provided that makes an electrical connection to a source flange 67 surrounding a circular opening 412 in the top of the wall 502 of the chamber 501. A plurality of, for example three, hand-operated clips 9 are provided at equally spaced intervals around the housing 1, which, when released, allow the source flange 67 to be removed and the target 10 to be changed. Ports 150 are provided for water inlet and outlet, and ports 151 are provided for a connection to a water loop external to the main source housing 1.

[0057] The target 10 is part of a target assembly 420, which is illustrated in FIG. 3. The target 10 is frusto conical, and is designed to minimize manufacturing costs. It has a single upper ‘O’ ring groove 11 defining the upper vacuum coupling and a well-finished surface 427 which makes the lower coupling. The rear 428 of the target 10 is smooth. The target 10 may be monolithic as is usually the case when the target 10 is copper, or it may be formed by bonding a layer of source material to a structural backing plate by one of a number of techniques well known to sputtering engineers. The opposite sides of the target define an included angle of divergence of a cone which is preferably about 110°.

[0058] The 110° included angle of the conical target 10 was selected after extensive computer modeling, using the HPEM code of Kushner et al. This angle leads to optimum uniformity of deposition at the pressures, powers, and target-substrate spacings of, preferably, about six to nine inches. A smaller included angle might also be desirable, but angles lower than 90 degrees are expected to lead to a reduction in deposition rate without much uniformity improvement. Higher included angles are expected to give poorer target utilization and poorer deposition uniformity.

[0059] O ring groove 11 is at the upper extreme of the target 10 at the inner edge of the target annulus around central opening 421 of the target 10. Castellated features 12 are provided to the outside of the O-ring groove 11 to allow the target 10 to assemble to a cooling water system 422 (FIG. 9) without the use of screws. A step 154 is provided in the inside diameter around the opening 421, which, in combination with a similar step in the Faraday shield and window assembly 440 (FIG. 6), described below, prevents deposition of metal on a dielectric window 33 that covers the opening 421 also as described below.

[0060] The target 10 mates with a cooling cover 13 to define the cooling water system 422. This cover 13 has ‘quad ring’ water seals 14 and 15 on its inside on each side of a channel 16. These seals 14, 15 contact the rear of the target 10 when the target 10 and cover 13 are assembled. Bayonet assembly structure 153 is provided to connect the cover 13 to the target 10. To mate the target 10 and cover 13, slots 17 are provided in the cover 13 to allow the cover 13 to be dropped over the castellated features 12 after which the target 10 and cover 13 are rotated with respect to each other to produce a ‘jam jar’ effect that tightens these parts 10, 13 together as they are rotated through about 20°. Water for target cooling enters and exits the cooling cover 13 via ports 155 into manifolds 18 formed in a ring shaped channel 16 in the front surface of the cover 13, as illustrated in FIG. 3A. These ports 155 and manifolds 18 are located 180° from each other around the channel 16 in the cover 13. The manifolds 18 are deeper than the channel 16 and each occupy about a 10° sector of the channel 16. On each side of each manifold 18 is a groove 19athat accepts a “comb” 19. Each comb 19 is a thin metal insert that bears a series of notches 19b. These notches 19bdivide water into separate streams as it enters the main channel 16, avoiding the formation of stagnation regions in the water flow which could lead to a reduction of cooling efficiency or possible local boiling of the water. The design of the combs 19 is determined by computational flow dynamics modeling. The combs 19 may be removed and replaced with others designed for different flow rates versus pressure relationships without removal and replacement of the entire cover 13. At the end of target life, the cover 13 may be removed from the expired target and reused. Water coupling is made through the cover 13 into the manifolds 18 via spring loaded water couplings 69 (FIGS. 9-10) which are described below. DC power is connected via a spring loaded stub 80 (FIG. 12), described below.

[0061] The IPVD source 503 includes a magnetron magnet assembly 430, illustrated in FIG. 4, which includes a magnet pack 20 that connects to the back of the target assembly 420. The magnet pack 20 includes a steel yoke 21 and a series of magnets 22 arranged in three rings including an inner ring 22a, an intermediate ring 22band an outer ring 22c, as shown. The resulting field lines produced by the magnets 22 include a main magnetic tunnel 24a, which affects the target erosion at the beginning of a targets life and runs in a circular path along an intermediate radius of the annular target 10, and inner and outer magnetic tunnels 24band 24c, respectively, which distribute the target erosion toward the inner and outer rims of the target annulus toward the end of the target life. The erosion groove due to this arrangement thereby broadens as the target erodes leading to enhanced target material utilization. The magnets 22 are held in their locations by adhesive bonding. Non-magnetic rings 23 are used to space the magnets correctly.

[0062] An alternative magnetic arrangement is shown in FIG. 5. This simpler design is expected to lead to lower target utilization than that shown in FIG. 4. The magnet assemblies 20aare preferably designed so that at least a portion of a magnetic tunnel traps plasma over the target 10 so that a net erosion of the target 10 takes place at all times over the target life so that no net redeposition takes place on the target 10. One way this can be achieved is by placing the intermediate magnet ring 22bsufficiently far from the target so that its field does not cancel the tunnel formed by field lines of the main tunnel 26 that extend between the opposite poles of the inner and outer magnet rings 22aand 22c.

[0063] The magnet packs contain a number of holes 25 bored parallel to the axis which carry the feedthroughs for water and DC power, leading to the target. The assembly of magnets 22 is coated with a hard polymer coating such as polyurethane or enclosed within non-magnetic metallic or plastic cladding pieces, which are bonded to the magnet pack 20. This coating prevents the magnets 22 and yoke 21 from becoming oxidized in air and also prevents the magnets 22, which may be of sintered construction, from becoming a contaminating particle source.

[0064] The IPVD source 503 also includes a Faraday shield and window assembly 440, which is illustrated in FIG. 6. The shield and window assembly 440 includes a dielectric window 33, formed of a lapped high purity alumina plate 7 mm thick, and a Faraday shield 26, formed of an electrically conductive material such as aluminum or copper. The shield 26 has an integral cooling water channel 27 formed therein and defined between an annular rim portion of the shield 26 and an annular channel cover ring 27abraised or welded to the rim 27aof the shield 26. A number of slots 28 are milled into the shield 26. The slots 28 are chevron shaped in cross section, as illustrated in FIG. 6A. The dimensions of the slots 28 are optimized by computer modeling to balance transparency of the shield 26 to RF energy against minimum transmission of sputtered material from the process region of the chamber 501 to the dielectric window 33.

[0065] Water connection to the channel 27 of the shield 26 is made via stainless steel stubs 29 which thread into the shield 26 and make a water seal via O-rings 30. Each of the stubs 29 has external threads 31 and a conical tapered end 32 on which a smooth finish has been placed. The assembly of the shield 26 with stubs 29 attached is assembled to the window 33 by inserting the stubs 29 through holes 34 in the dielectric window 33, which align with holes 34a, which are 180° apart around the rim of the shield 26 and which communicate with the channel 27. There are two holes 34 through which the water stubs pass, one for water inlet and one for water outlet. Teflon washers 35 and aluminum knurled nuts 36 are screwed onto the stubs 29 and compress O-rings 37 between the Faraday shield and the window 33 forming a vacuum seal around the water stubs 29. O-rings 38 are provided in the nuts 36, which have no sealing function but perform the function of a spring, helping to prevent over-tightening of the nuts 36.

[0066] A dense secondary plasma for ionizing material that has been sputtered from the target 10 is excited by an RF coil assembly 450, illustrated in FIG. 7, which includes a three dimensional coil 39 whose configuration and electrical performance and properties are described in patent application Ser. No. 09/277,526, filed on Mar. 26, 1999, entitled Process Apparatus And Method For Improving Plasma Distribution And Performance in an Inductively Coupled Plasma, invented by Jozef Brcka, an inventor herein, hereby expressly incorporated herein by reference. The RF feeds to the coil are provided by a pair of male connectors 40 that thread into sockets 41 in the coil end pieces 42. Water feed is provided through a pair of threaded pipe couplings 43. These couplings are each surrounded by a flange 44 with an O ring groove 45.

[0067] The coil 39 is mounted within a high dielectric insulating cup 46 made of a material such as TEFLON. The thickness of material of the cup 46 in various areas of the cup surface is calculated to be as thin as possible consistent with the suppression of arcing to the nearest conductive surface. To make this calculation, the effective field and pressure-distance product in air are calculated for the stack consisting of the TEFLON, an air gap, and any other dielectric such as the source window 33. The thickness of TEFLON can be set so that the field is always lower than that required to cause breakdown of air, for reasonable values of the air gap. This calculation allows the cup thickness to be minimized safely. A minimum thickness allows the best coupling to the plasma and lessens the constraints on the dimensions of the coil 39.

[0068] The cup 46 is mounted in an aluminum ring 47. This ring is mounted to the water cooling assembly 422 using shoulder screws 62 and springs 63 which in operation press the ring against the dielectric window 33 which is part of the Faraday shield system 440 described above. This window 33 in turn is pressed against the target upper O ring 48, which is located in the groove 11. This spring loading compensates for tolerance mismatch and mechanical deformation and so holds the vacuum seals together and allows the system to be pumped down.

[0069] The ring 47 also features a canted coil spring 49 and a tube 50 for cooling water. There are two axial through holes 51 for the Faraday shield water stubs 29 to pass through (FIG. 8). There are recesses 52 for TEFLON washers 53.

[0070] An aluminum enclosure 54 is mounted over the coil 39 and held down by a bolt circle so that there is electrical continuity through the canted coil spring 49, to the aluminum ring 47. The purpose of this is to prevent water leaks from reaching the high voltages present on the coil and to act as a primary barrier to RF emission from the source. RF energy can escape only through the dielectric window 33 into the process space of the chamber 501. The enclosure is pierced in four places, two ports 55 carrying the RF feeds and two ports 56 carrying the water feeds. TEFLON parts 57 are passed through these water feed holes and pass over the O rings in grooves 45 making a water seal. This provides secondary containment so that a water leak in the coil water coupling does not lead to water being in contact with the RF elements. TEFLON insulating pieces 58 are used to stand the coil and RF connectors off electrically from the enclosure.

[0071] TEFLON combs 59 are used with insulating screws to hold the coil down to the cup 46. This leads to consistent behavior between different sources.

[0072] As illustrated in FIG. 9, cooling water is distributed via components mounted on a large plastic plate 60 which also supports the magnet assembly 430, which is mounted thereto on spacers 61, and the RF source assembly 450, which is mounted thereto using the shoulder screws 62. A main aluminum manifold block 64 is mounted to the plate 60 and brings in water from the exterior of the source via threaded connections. A secondary manifold 65 allows the flow of water through the coil to be checked by a flow sensor mounted outside the source and also allows connection of the water flow to a cooling channel in the source flange 67 (FIG. 14).

[0073] Cooling water to the target 10 passes through stub assemblies 68, as illustrated in FIG. 10. The assemblies 68 are designed so that water couplings 69 can be assembled through holes no larger than the outer diameter of the coupling tubes themselves; which allows the holes 25 in the magnet pack to be of minimum diameter, which minimizes the disturbance of the magnetic field that these holes might cause. The assemblies 68 are spring loaded, the springs 70 acting between metal cups 71 that are pressed into the plastic mounting block 60 and washers72 held in place on the water tubes 69 by retaining rings 73. In this way in operation the lower end of the cooling tube bears on an O-ring 74 that is forced into a specially designed detail in the target cooling cover 13. The detail is a doubly tapered bore similar to that used in ISO and SAE standard fluid connecting glands.

[0074] The Faraday shield 26 is cooled via special couplings 75 that also make a DC connection 423 to the shield, as illustrated in FIG. 11. The coupling features a tapered recess with an inward facing O-ring 76. The entire fitting is spring loaded via springs 77 and when assembled, the conical surface of the Faraday shield, water stub 29 is forced hard against the O-ring 76 making a water seal. The coupling also has a canted coil spring 77, which performs a double function. Firstly it makes electrical connection to the shield 26. Secondly it performs a latching function, engaging a shallow recess in the stub 29 which holds the shield 26 approximately in place after it has been installed to the water housing 422. Finally this coupling bears a tapped hole 78 for making electrical connections. The coupling inserts into the RF assembly through the TEFLON washers 53. Support blocks 79 are mounted on the block 60 and support the coupling, and they also apply pressure to the springs 77. These blocks are fitted at their upper end with contact fingers 156 which ground the entire assembly, and through it the Faraday shield, when the source is assembled, by making contact to the source housing 1.

[0075] The block 60 also contains supports for two non-water-related items. One is the DC contact stub 80, which is spring loaded similarly to the target water couplings, as illustrated in FIG. 12. In this case the stub 80 is forced against the target cooling cover, contact being ensured by another canted coil spring 81. The upper end of the stub is fitted with a male connector 158. This DC connector stub assembly is surrounded by a plastic tube 159 that protects it from spray in the event of a water leak in the source. The other item is a springloaded pin 166 (FIG. 9) that passes through the magnet pack and bears against the target cooling cover 13 (FIG. 3). When the source 503 is assembled, the upper end of this pin 166 bears on another plunger assembly described above, which in turn activates a microswitch 7. The activation of this switch 7 indicates that the target assembly 420 is correctly made up and that it is therefore safe to turn on the cooling water.

[0076] The water distribution assembly 422 contains a number of couplings and a considerable length of TEFLON hose. In the event of a leak developing, it is desirable that water be drained to the exterior of the source and taken somewhere it can be detected. Therefore, the block 60 contains a pattern 82 of drainage grooves that terminate at small holes 83 (FIG. 2) in the source housing 1.

[0077] Vacuum leak checking of the source should be carried out. The inner target O ring as well as the window seal O-rings are rather relatively inaccessible. For this reason, item 60 is also provided with stainless steel capillary tubes 84, which are embedded into some of the grooves 82 as shown. During leak checking, these tubes 84 can be used to feed helium to the vicinity of the inaccessible O-rings.

[0078] Referring to FIG. 14, a source flange and dark space ring assembly 470 is illustrated. This assembly 470 includes the flange 67 on which is mounted the fixed parts 86 corresponding to the clips 9. A swaged-in cooling water tube 471 is provided. The upper side bears a recess 87 with an O-ring groove 88 into which is inserted a TEFLON insulator 89, which is itself provided with an O-ring groove 90. Weep holes 472 are provided leading from the recess 87 to the outside diameter of the flange, which provide drainage and visual warning in the event of a cooling water leak. The O-ring groove 90 bears an O-ring 90athat interfaces to the target 10, forming the outer vacuum seal. The underside of the flange 67 has another recess into which is mounted the dark space shield 91. This shield 91 is held in place by shoulder screws 92 that tighten into slot features 93. To remove the shield 91, the screws 92 are slightly loosened and the shield 91 then slightly rotated and lifted off. The shield 91 is designed so as to not contact the flange 67 in regions near the O-ring groove 88, to avoid overheating of that O-ring. The source flange 67 interfaces electrically to the process chamber 501 using a canted coil spring 94. An O-ring 95 makes the vacuum seal.

[0079] The autotuner 96 is a commercial unit that mounts to the top of the source housing 1 (FIG. 2). Large female connectors 40asupply power to the RF assembly via the connectors 40 (FIGS. 7-8).

[0080] There are several features or considerations relating to the IPVD source 503. Assembly and dismantling for repairs is one. In assembly, the fully assembled source interior (less Faraday shield assembly 440 and target assembly 420) is dropped into the upturned source housing assembly so that the DC connector 158 (FIG. 12) inserts into the socket 4 (FIG. 2). Six screws are inserted and water connected to four ports 151, 152. Electrical feeds are plugged in. The source 503 is then ready for use. Dismantling is the reverse of assembly. This simple assembly technique is an advantage of the modular construction of the source 503. Routine target replacement is another such feature or consideration. After cooling water has been blown out, the source is inverted and the clips 9 released. The flange assembly is then lifted off. The target then can be lifted out and then the Faraday shield assembly can be removed. No tools are needed except to release the dark space shield 91.

[0081] The electrostatic chuck 507 and the wafer transfer system 504 cooperate in the transfer of wafers from one to the other. Chuck assembly 507 includes a service support assembly 480 as illustrated in FIG. 15, which includes the wafer support, holder or chuck 97. A suitable chuck 97 may be obtained from INVAX Inc or other sources. A fluid passage is provided for the passage of cooling fluid, for example a GALDEN brand perfluorinated fluid. The chuck 97 is of the tripolar type, having two embedded, electrically isolated, electrodes for the application of a chucking voltage, while RF bias can be applied to the chuck body by way of the electrostatic chuck electrodes. The RF is thereby coupled through to the embedded electrodes and thus to the wafer. All metal parts of the chuck are aluminum coated with a proprietary dielectric. Back side gas can be provided through a central hole. A thermocouple is mounted to the rear of the chuck.

[0082] The chuck 97 has a number of counterbored holes and is mounted to the stainless steel base 98 using screws; there are polyimide ‘vespel’ insulators that protect the chuck from damage by the screws and provide electrical isolation. An insulating block 99 isolates the chuck from the base.

[0083] FIG. 16 shows the mounting of the chuck to its support structures. The stainless steel base 98 is fitted with rest pegs 101, which support a ring 102. The ring has features which work with those on the rest pegs to give an accurate alignment of the ring with the chuck. This ring has sockets 103 that are fitted with ceramic lift pins 104, which pass through holes in the chuck. The ring 102 rests on the pegs 101 during process. As the table is lowered to the wafer transfer position, the ring intercepts a separate assembly as described below, and is lifted off the pegs, causing the pins to rise through the chuck and to lift the wafer off the chuck ready for transfer to a handler. The base has a short downward extending tube at the end of which is a flange 119 that can be clamped to the z drive assembly 490 described below.

[0084] Two configurations of shields for this table are shown in FIG. 25. In the simplest case, a stainless shield 105 rests on a step on the base 98 and shields the chuck from metal deposition. A system that may be better is also shown. Here a grounded shield 106 is supplemented by a ring 107, which rests directly on the chuck. This ring may be made of aluminum or stainless steel and may or may not be coated with a dielectric material, possibly of high dielectric constant similar to that used in the chuck dielectric. This ring couples to the RF power that is applied to the chuck through the chuck dielectric. Advantages of this are that the shield can be in very close proximity to the chuck, thereby more effectively blocking metal deposition, and that RF power is applied to the ring causing it to attain the same bias as the wafer, which lessens the distortion of electric fields near the wafer edge. The ring overlaps but is separated from the grounded shield. This provides a convoluted path for metal deposition and keeps material from being deposited on the chuck.

[0085] The wafer transfer mechanism 504, illustrated in FIG. 18, is attached to the base of the chamber wall assembly 502 (FIG. 19) using the threaded pegs 108. A ring 109 is held in a raised position by springs 110. It is held in an accurate position by the features 111 and 112. Slotted pegs 113 are mounted to this ring. When the table or chuck is lowered, the pegs 101 descend into the slots 114 in the slotted pegs. The ring 102 is lifted and the pins 104 raised. When the pins have lifted 11 mm, the pegs 101 reach the bottom of the slots 114. Usually the transport system would now insert a pick to collect the wafer. Further lowering of the table compresses the springs 110 causing the entire system including the wafer on the pins to be lowered onto the pick. The pick is now removed, bearing the wafer. Bellows 115 form a vacuum barrier while allowing the table to be raised and lowered. Service support assembly 480 includes a rigid assembly 116, formed by three silver plated brass tubes 117 with nylon spacers 118 at regular intervals and a silver plated plate at each end, is fixed to the chuck with screws. It has a triple function that includes (1) mechanical support of the chuck voltage wires, thermocouple wires, temperature control fluid tubes, and backside gas tube, (2) RF power conduction to the chuck body, and (3) conduction of nitrogen purge gas to the space behind the chuck. At low operating temperatures, water condensation hinders chuck operation unless a gas purge is provided to sweep out moisture. In this case there are small cross holes in the support tubes near the interface to the chuck. Purge gas is connected at the lower end of the tube.

[0086] The vertical elevator or Z drive system 490 for vertically moving and adjusting the height of the chuck 97 is illustrated in FIG. 18. This consists of a tube 120 with an upper flange that can be coupled to the table flange 119 using clamps 121. A rail 122 is mounted to the tube which rides in sliders 123 which are supported on a bracket 124. A ball nut 125 mounted to a flange 126 at the lower end of tube 120 is driven by a lead screw 127. This screw is turned by a motor and gearbox combination 128. The motor is fitted with a brake (not shown) that prevents backdriving of the motor when it is turned off. The motor control system is conventional. Position of the table is determined using an encoder mounted at the upper end of the lead screw. The bracket 124 is mounted to the process chamber.

[0087] A backside gas delivery system and components enclosure 129 is mounted below flange 126. This enclosure contains electronics for conditioning of signals running to and from the chuck. The table RF autotuner 130, a commercial unit, is mounted to the exterior.

[0088] In one embodiment, valves and a universal pressure controller (UPC) are mounted to the exterior of enclosure 129. These components control the backside gas delivery. In the future, these components will be moved to a position elsewhere. A plasma arrestor is used to form an insulating break in the backside gas system and avoid plasma formation in the gas line.

[0089] The vacuum and gas handling system 505 is illustrated in the diagram of FIG. 21. This includes a chamber and pumping system. The chamber 501 is contained within a chamber wall assembly 502 which includes particularly a vacuum tight chamber enclosure 131 as illustrated in FIGS. 19 and 20, which is of a conventional stainless steel design. The enclosure 131 has an internal isolation valve 132 for coupling to a handler of the transfer system 504. There are upper and lower flanges for the source and table assemblies and appropriate flanges for vacuum metering, process gas input, etc. An eight inch conflat flange in the base couples to an isolation valve 142, and this is connected to a cryogenically cooled panel 143 operating at about 110K and a turbomolecular pump 144. The speed of the turbo pump can be controlled to allow for processing at relatively high gas pressures in low speed (about 24000 RPM) operation while rapid pump-down from atmospheric pressure is provided at high pump speeds (350 Is⁻¹nominal at about 56000 RPM). The turbo pump is backed by a non oil sealed type rotary pump 145 which is also used to pump the chamber down to 100 mTorr or so after venting at which pressure the isolation valve may be opened. The vacuum arrangement is quite conventional except for the use of the water pump/variable speed turbo pump combination.

[0090] During recovery after the venting of the chamber 501, internal halogen lamps provide bake-out capability. Process plasma is also used to raise the temperature of the module interior, assisting the bake-out. In this way pressures of 10⁻⁸torr or lower are achieved within several hours after a typical target change procedure.

[0091] The chamber has external water cooling channels for cooling during normal operation. A flange is provided for mounting of the source hoist. The chamber is locally stiffened around this flange by the addition of external welded ribs, to support the loads exerted when the source is raised by the hoist. A housing for the DC connector 162 for the source is also mounted on the chamber.

[0092] The gas systems for modules with and without reactive gas are shown in the schematic diagram of FIG. 21. Delivery of argon gas to the chamber is via a simple flange. Reactive gas, if used, passes into a branched ‘spider’ tube 133, which carries it into the process space. The tubes end in two positions on opposite sides of the table assembly. Small covers 134 rest on the ends of the gas tubes to prevent deposition of sputtered material on the gas tubes.

[0093] A sputter shield assembly 495 is provided as illustrated in FIG. 22. These are among five shields that are subject to removal and cleaning. These are the Faraday and dark space shields described above, the table shields described above, and two chamber shields 136 and 137. These chamber shields are supported on an armature 138. Upper shield 136, which typically rises to a higher temperature than lower shield 137, is supported at three points on the pegs 139 rising from the armature. Features 140 on the shield 138 rest on these pegs, centering the shield in the chamber but allowing radial motion. This avoids the buildup of stresses in the shield due to thermal expansion and deposition of hot material with subsequent differential thermal expansion. Such stresses can lead to particles being released into the system. The shield requires no tools to remove it. The lower shield 137 is inserted into the armature 138 and is supported by it. Self centering behavior arises from the chamfer 141 in the armature. No tools are required to remove the shield.

[0094] Pumping of the space within the shields should be under control throughout the life of a sputter target regardless of the history of depositions in the chamber, that is, pumping should be at the same rate regardless of the number of wafers coated. In most sputtering systems this is not well controlled. Pumping in these systems may occur through gaps whose dimensions change with chamber temperature or through holes in the shields provided for other purposes. This may lead to process issues particularly in the case of reactive sputtering. This design avoids these problems because the gap is set by the length of the pegs 139. These are short and outside the region of highest heat load, and so are not significantly altered in length by thermal expansion. As a result, the gap between the shields 136 and 137 is well controlled and pumping occurs all around the circumference of the process region in a controlled way. The height of shield 136 controls the gap between the source dark space shield and the shield 136. This gap is designed to be small, preferably of the order of one millimeter to reduce the passage of metal through the gap. The shield 136 fits in a recess in the dark space shield forming a convoluted path that intercepts metal that enters the gap between the shields before the unprotected regions of the chamber are reached. Good dimensional specification is necessary to achieve this as is attention to the thermal expansion of the parts.

[0095] The wafer is loaded through slot 146 in the lower shield. During processing, this slot is closed off by the table shield 105 or 106. In future versions of the hardware, the slot will have a surround 147 that should intercept sputtered metal more efficiently than at present.

[0096] The source hoist 460 is illustrated in FIG. 23. The source hoist 460 is a pneumatic lift capable of raising a weight of about 200 pounds. It is required to lift the source off the chamber; it must then rotate about a vertical axis. There must also be a motion allowing the source to be turned over for target removal. The source must be lowered for easier access to the target. Considerable rigidity of the assembly is required in order that the source be maintained level when lifted off the chamber. In the arrangement used there is a fixed vertical shaft 170 that carries an upper and lower assembly 171 and 172, respectively. Both of these items contain linear bearings which can slide and rotate on shaft 170. A pneumatic cylinder 173 allows the raising and lowering of the source. Item 171 is intended to move vertically and also to rotate while item 172 rotates only, being kept aligned with upper assembly 171 by the cylinder. Rollers 174 act as bearings on which item 172 rotates. There are recessed features 174 into which these rollers can drop. This provides some resistance to the rotation of the assembly at certain points, allowing the operator to detect the correct positions for operation.

[0097] Assembly 176 contains a pair of opposed angular bearings. Shaft 177 passes through these and is retained by a cap 178 and a spring washer 179 that preloads the bearings. This arrangement leads to a rotational joint with very high radial rigidity. The assembly 176 also contains a cam wheel feature that, in conjunction with a plunger 180, ensures that the source can only be rotated in one direction and can be locked in its upright and inverted orientations. The hoist mounts to the source using the flange 181.

[0098] Preferred process parameters for the deposition of tantalum and tantalum nitride in one module 500 and then deposition of copper in another module 500 of the same cluster tool are attached as appendix A hereto.

[0099] Those skilled in the art will appreciate that the implementation of the present invention herein can be varied, and that the invention is described in preferred embodiments. Accordingly, additions and modifications can be made without departing from the principles and intentions of the invention.