Apparatus for Evaporation Arc Stabilization During the Initial Clean-Up of an Arc Target

This application is related to application Ser. No. 492,831 now U.S. Pat. No. 4,430,184, filed May 9, 1983, by William M. Mularie entitled "Evaporation Arc Stabilization". It is also related to an application entitled "Improved Apparatus and Method for Evaporation Arc Stabilization for Non-Permeable Targets Utilizing Permeable Stop Ring" filed by Anthony R.T. Lefkow on even date herewith. Furthermore, it is also related to an application entitled "Improved Method and Apparatus for Evaporation Arc Stabilization Including Initial Target Cleaning" filed by Charles F. Morrison, Jr. on Sept. 12, 1983. All of the above applications are assigned to the assignee of the present application.

This invention relates to arc stabilization processes and devices which may be employed, for example, in arc coating systems. Such coating systems are disclosed in U.S. Pat. Nos. 3,625,848 and 3,836,451 to Alvin A. Snaper and U.S. Pat. Nos. 3,783,231 and 3,793,197 to L. Sablev, et al. These systems are characterized by high deposition rates and other advantageous features. However, these advantages can be somewhat offset due to instability of the arc. That is, the arc involves currents of about 60 amperes, or more, concentrated into a cathode spot so small that current densities are 10.sup.3 to 10.sup.6 amperes per square inch. The voltages are 15 to 45 volts. Thus, power densities at the tiny cathode spot are in the order of megawatts/inch.sup.2. Accordingly, local violence is an understatement. The target surface under the cathode spot flash evaporates from the intense heat. It is the evaporated target material which deposits as the coating on a substrate. The cathode spot migrates about the target surface in a random, jerky motion with reported velocities of many meters per second. Because of this random movement, damage to the device and contamination of the coating can occur if the spot moves off the target surface.

Different solutions to the arc instability problem have been proposed. Thus, in Sablev, et al., U.S. Pat. No. 3,793,179, a shield is placed close to the edge of the target. In particular, it is placed at a distance from the target which represents less than a mean free path of the gas present. In an arc discharge, gas and plasma are generated at the cathode spot with sufficient violence that local mean-free paths may occasionally be reduced to a few thousandths of an inch. When such a blast of local high pressure is blown under the shield, which is spaced at several millimeters (.about.80 thousandths of an inch), there is a finite possibility the arc can migrate under the shield. When this happens, there will be arc damage to the cathode, contamination of the evaporant, or the arc will extinguish.

Sablev, et al., U.S. Pat. No. 3,783,231, (copy submitted herewith) apparently addresses the foregoing problem by providing a feedback mechanism of some complexity that emphasizes the frustrations caused by the problem. The feedback involves the utilization of a magnetic field to retain the cathode spot on the target surface. U.S. Pat. No. 2,972,695 (copy submitted herewith) also discloses the use of a magnetic field for cathode spot retention.

It is an object of the present invention to provide in an arc evaporation process, stabilization of the arc for permeable targets in such a manner as to avoid the inadequacies and complexities of the prior art approaches.

It is a further object of this invention to provide improved arc stabilization during the initial clean-up of an arc target.

It is a further object of this invention to provide arc stabilization of the above type for use in arc coating processes.

Other objects and advantages of this invention will be apparent from a reading of the following specification and claims taken with drawing.

FIG. 1 is a schematic diagram in cross-section which illustrates the uniform erosion pattern which results when a non-permeable target is confined by a N-ring in accordance with above-mentioned, co-pending application Ser. No. 492,831, now U.S. Pat. No. 4,430,184.

FIG. 2 is a schematic diagram in cross-section illustrating the erosion pattern which results when a permeable target is confined by an N-ring.

FIG. 3 is a schematic diagram in cross-section of an illustrative embodiment of an arc stabilization apparatus for a permeable target in accordance with the present invention.

FIGS. 4(a), 4(b), 5(a), and 5(b) are schematic "iron filings" diagrams in cross-section which illustrate various conditions of permeable and non-permeable targets.

FIG. 6 is a schematic diagram in cross-section of an illustrative embodiment of an arc stabiliation apparatus for use during initial clean-up of a target.

Reference should be made to the drawing where like reference numerals refer to like parts.

Prior to discussing in detail an illustrative embodiment of the present invention, the invention of before-mentioned, co-pending application Ser. No. 492,831 now U.S. Pat. No. 4,430,184 will be discussed in detail with respect to FIG. 1 since the latter invention is employed in the present invention.

In FIG. 1, an arc coating system for a substrate 12 comprises an anode 14 where, if desired, the anode and substrate may be the same member, a non-permeable target 16 of conductive or insulative material, a cathode 18 where, if desired, the target and cathode may be the same member, an arc confinement and retaining ring 20, and a power source 22, which is DC if the target is conductive and RF if it is insulative. If desired, separate rings may be used for the arc containment function and the target retention function. In accordance with the invention, the confinement ring is composed of boron nitride or a similar material as discussed in more detail hereinafter.

In operaton, source 22 is energized across the anode and cathode to strike an arc between the anode and target in a well known manner. Arc initiating means (not shown) may be employed to initiate the arc. Target material is then flash evaporated from the cathode spot formed at the root of the arc on the target surface and deposited on the substrate as a coating. Confinement ring 20 contacts target 16 and surrounds an exposed area constituting an evaporation surface of the target to thereby confine the cathode spot to the evaporation surface in such a manner that continuous, stable operation is effected for the entire lifetime of the target even though the ring may be overlayed with evaporated target material many mils thick.

Briefly, the confinement ring is made of a material such that the ring is characterized by (a) a low absolute value of secondary emission ratio where preferably the ratio of the target is greater than that of the ring and (b) a low surface energy of the ring relative to that of the evaporant.

The secondary emission ratio .delta. is defined as the number of secondary electrons produced by a primary electron, or other charged particle, incident upon a target, Thus, ##EQU1## The number of electrons emitted by the target is dependent not only on the particular target material but also on the energy of the primary charged particle. In accordance with one aspect of the invention of the co-pending application, the secondary emission ratio of the confinement ring, .delta. (confinement ring), should be less than one at the mean energies of the charged primary particles typically found in arc coating processes such as disclosed in the aforementioned Snaper and Sablev patents where these mean particle energies may extend from about 20-100 eV and typically are about 40-60 eV. Moreover, the secondary emission ratio of the target, .delta. (target) is preferably greater than .delta. (confinement ring).

One class of materials which, generally speaking, has a .delta.<1 is the nitrides and boron nitride, in particular. Boron nitride has been observed to be effective with metal targets. It also can be used with insulative targets, many of which have a high secondary electron emission ratio due to a substantial oxide content. Titanium nitride, even though quite conductive electrically, also restricts arc paths.

In general, a further aspect of the invention of the co-pending application for effecting arc confinement is the low surface energy (.gamma.) of the confinement ring relative to that of the evaporant--that is, .gamma.(ring)<.gamma. (evaporant). In this regard, BN (surface energy of 600-700 ergs/cm..sup.2) is not wet by most metals at the temperatures encountered in the arc coating processes described in the aforementioned Snaper and Sablev patents. Only the oxides of lead, bismuth, copper and antimony will wet and attack BN, the surface energies of these oxides ranging from 100-300 ergs/cm..sup.2. However, BN and materials similar thereto are suitable for the practical applications encountered in arc coating processes.

In summary, confinement ring 20 should be formed from or coated with materials having the foregoing characteristics. The nitride compounds are particularly suitable and, in particular, the nitrides of boron and titanium. In this regard, it should be noted if the nitride compound constitutes less than 100% of the confinement ring composition, the ring will nevertheless function in the desired manner as long as the entire composition has the above-discussed characteristics. Hereinafter, a ring of the type described in the co-pending application will be termed an "N-ring".

As can be seen in FIG. 1, as long as the target 16 is non-permeable, the erosion pattern 24 obtained with the confinement ring of the co-pending application is quite uniform. However, if a permeable target 16 is employed, the erosion pattern 24 is not uniform, as can be seen in FIG. 2. Examination of the target of FIG. 2 leads to the conclusion that the arc is influenced to move toward the edge of the permeable target, for there is no reason to expect it to move specifically toward the N-ring 20 otherwise.

The assignee of the present application has conducted experiments, one of which involved placing an N-ring on a large sheet of permeable material such that the plane of the ring was parallel to that of the sheet, the ring being in the approximate center of the sheet. An arc was struck within the ring and the resulting erosion pattern was normal--that is, is corresponded to that of FIG. 1.

The above principles are incorporated in the FIG. 3 embodiment of the present invention where a permeable target 16 is surrounded by a flat ring 26 made of a permeable material such as soft iron or Permalloy or the target material itself. In fact, any material considered to be permeable may be used, such materials including but not limited to iron; nickel; cobalt; and alloys thereof with small amounts of optional additives; ferrites; steel; etc. Further, the ring 26 may comprise an integral extension of the target itself. Hereinafter ring 26, whether it be a separate member, as shown in FIG. 3 or an integral extension of the target will be termed a "P-ring".

N-ring 20 is disposed around the periphery of target 16 and contributes to the confinement of the arc on the target surface. The N-ring also retains ring 26 and target 16 in place via a bolt 28 which is threaded into cathode body 18.

Although there is no intent to be limited to a particular theory of operation, the following considerations apparently are applicable to P-rings of the present invention. It has been observed a vacuum arc struck on a non-permeable target wanders randomly about, most often leaving the target for other areas of the cathode within a second or so. A permeable target loses the arc to other cathode areas in milliseconds. Obviously, for uncontaminated coatings, the arc must remain only on the target.

The early literature of magnetic fields applied to the vacuum arc indicates the arc moves most readily in the direction of greatest magnetic field density. Assuming this is the mechanism for forcing the arc to the edge of the target in FIG. 2, it appears the arc moves away from the permeable material that reduces field density.

Another insight relating to the containment mechanism is obtained from the work of Naoe and Yamanaka ("Vacuum-Arc Evaporations of Ferrites and Compositions of their Deposits" Japanese Journal of Applied Physics, Vol. 10, No. 6, June 1971, copy submitted herewith), who arc-evaporated ferrite composites from a cup-shaped ferrite target. They were attaining a melted portion of the target, and the oxide materials behaved very differently from the metals. They reported a very stable arc that moved in a very slow circular motion at the approximate center of the cup. They made no reference to this as unusual behavior, but gave considerable detail of the arc motion as observed visually. Implications relative to general arc containment were lacking.

To better understand this phenomenon the assignee of the present invention has used a DC current through a wire to generate a cylindrical magnetic field of the type that the arc appears to produce close to the target. This wire has been brought into the proximity of various geometries of permeable targets and "iron filings diagrams" produced to give insights to the magnetic influence of these permeable materials on the magnetic field.

It should be noted this is somewhat different from the application of a magnetic field to interact with the arc field as applied in aforementioned U.S. Pat. Nos. 2,972,695 and 3,783,231, for no external field is applied. When current is passed through the wire, magnetic flux is generated symmetrically around the wire. When two wires are placed in parallel with current in the same direction through them, the wires are pulled toward each other. The field generated between the wires is cancelled, for the flux direction is different on the right versus the left of the wire.

In actuality, an arc is quite unique, and modeling it as a wire with current flowing is not a true indicator of what an arc will do. If sufficient current is permitted to flow in the arc, it will divide itself into two simultaneous arc spots moving independently about the cathode surface. This is quite the reverse of the wires which move together as current flows. Simple explanation of this difference between the arc and the wire is sometimes given in terms of the electrons being free to move sideways in space in the arc, but being confined within the wire. They try to move sideways in the wire, creating equal and opposite forces on the wires. Thus the wire moves opposite to the direction the arc moves where the magnitude of the force moving an arc toward the edge of a permeable target is significant relative to the random motion forces that typically move the arc about a non-permeable target.

FIGS. 4(a), 4(b), 5(a), and 5(b) show the nature of the filing diagrams from the various conditions of permeable and non-permeable targets. In FIG. 4 less and less flux density 30 is seen as the iron plate 32 is approached for these lines (of which only the cross sections are seen) are drawn into the iron, for they travel more easily there. In the case of the aluminum target 34, the flux remains very constant as the plate is approached, as can be seen in FIG. 4(b). Moving to the target edge, as in FIG. 5(b), makes no change in the case of the aluminum target. However, as shown in FIG. 5(a) movement to the edge of the iron target produces a reasonably strong flux outboard of the target, and virtually none inboard. With the force on the arc thus outward, it is quite understandable that the arc races for the outside edge. The forces will be nil when the arc is perfectly in the target center, but normal random migration from the arc action will quickly push it off center. As it gets closer to an edge, the force toward the edge multiplies. It is thus quite logical that erosion would occur as shown in FIG. 2, for the arc is trapped between the induced electromagnetic force outward, and the "N"-ring which prevents the arc from moving further outward. There is relative freedom of motion only into the plane of FIG. 2. Thus the arc moves around the target perpendicularly to the trap. The attempt here is not to fully define the arc motion, but only to show certain aspects of it can be influenced significantly for purposes of control.

Furthermore, it follows that the presence of P-ring 26 of FIG. 3, the lines of force travel as easily through ring 26 as they do through target 16 disposed within the inner periphery of the ring. Hence, the flux density within the ring is substantially equal to that outside the ring's inner perphery. Thus, there is no outward force on the arc as is the case in FIG. 2 device. Accordingly, the arc randomly migrates over the entirety of the permeable target surface in the same manner it migrates over the non-permeable target of FIG. 1 to thereby effect uniform erosion of the permeable target.

Reference should now be made to FIG. 6, which illustrates a further application of the invention. Due to work performed by the assignee of the subject application, it has been established that a P-ring surrounding a non-permeable target is very effective in containing an arc on such a target once the target is ultra-clean. As will now be described, the FIG. 6 configuration enables the target to be ultra-cleaned by an arc so that the P-ring can then be employed to effect its arc retention function where, in FIG. 6, N-ring 20 may be (a) a separable cover on or (b) a coat painted, sputtered, evaporated, or otherwise applied to P-ring 26.

The literature repeatedly stresses the need for ultra-clean conditions for the target assembly, but gives little description of the behavior of an unclean target contaminated with oxides and the like. It is indicated the oxides are evaporated (or exploded) from the surface by an arc before meaningful metal evaporation can start. Most of the removal of these contaminants occurs during the initial cleaning phase of the target with the substrate in some instances, removed from the system. During this time, so-called type-one cathode spots are formed which do not effectively evaporate the target. Only after the contaminants have been effectively removed, are the type-one spots replaced by so-called type-two spots which are effective to evaporate the target. At this time, the substrate may be placed in the system preparatory to coating thereof if not already in place.

The N-ring appears to restrain the arc to the target during the initial cleaning phase even when the target is assembled in a dirty state. The P-ring is not that absolute a repellant, so that the arc may run upon the target and the P-ring for the first few starts of the arc during the initial cleaning phase. Once the target is sufficiently clean to evaporate target material, it coats the portion of the ring immediately adjacent the target, and the control of the arc becomes quite absolute. However, during the initial cleaning phase, the arc can move onto the back and sides of the cathode body 18 where it may damage the structure, or evaporate contaminating metals into the chamber.

In certain applications, the N-ring tends to (a) be somewhat fragile, (b) have a rather limited life time, (c) be expensive and (d) lose target material deposited thereon due to its low wettability. Hence, it is desirable to take advantage of the N-ring's arc retention capability during the initial clean-up (that is, the N-ring appears to hold type-one arcs for many target materials, perhaps all, at least all that have been tested by the assignee to date) and then utilize a P-ring once the target has been arc-cleaned.

In general, two items are of major concern. First, it is necessary to protect non-target areas of the cathode from eroding, especially, near the insulators. Thus, the arc cleaning activity must be kept on the target and clamp assembly. Secondly, the arc extinction frequency must be reduced, such that a few strikes will lead to a clean surface. With present methods, over a hundred strikes may be needed on a large aluminum cathode with light P-ring containment.

The severity of the arc-cleaning problem is a function of the target material. For example, aluminum is very bad, zirconium is very good. The observation of the assignee is that type-one arcs have a hierarchy of functions. First, they search out and attack oxided areas. This is reasonable, for such compounds as the metal oxides tend to be ready electron emitters. It appears only this oxide search mode can easily jump the P-ring. When the violent flashes from the obvious oxide areas are gone, progress is made through many more strikings of the arc, usually with longer and longer periods of operation before extinction. During most of this period there is almost no evaporated metal coming from the discharge--even though it may last for many seconds. It appears to still be operating as a type-one arc which has only a tiny evaporation output. This is probably the explosion of sharp points and ridges which can give vast numbers of electrons at quite low arc voltages.

In the oxide phase of clean-up, the arc easily trangresses the P-ring onto the non-target parts of the cathode. In the point and ridge phase, the P-ring holds the discharge, but the discharge extinguishes very often. Only when the type-one arc has eliminated all ready sources of electrons does it change into a type-two arc and proceed with flash evaporation of the target surface.

Because, as stated above, the N-ring will hold the oxide phase discharge, the embodiment of FIG. 6 may be employed to optimize the respective advantages of N-ring 20 and P-ring 26. When the N-ring is employed as a separable cover, it would be in place during the initial arc, clean-up phase until the type-two arc is established on the target. The cover would then be removed from the target to uncover it. There would then be a second arc clean-up phase conducted which removes the oxide areas, etc., covered by the N-ring 20 during the first cleanup phase. There are less than before, however, and the total clean-up time is less than that which would be required in the absence of the N-ring cover. Moreover, since this cover would be removed after the clean-up (to thereafter allow the P-ring to effect the arc retention function, as discussed above) the life time of the cover would be substantially increased. Furthermore, the loss of target material deposited thereon would not be a problem.

When the N-ring cover 20 is applied as a thin film coating, this should be done for each new target. Although this may be inconvenient in certain situations, there are certain advantages especially when the coating is made from TiN on a similar material. The BN paints do not adhere very well. Radio frequency sputtering of BN onto the ring 19 is better. The most acceptable approach is the use of titanium nitride--the very material that many arc systems are designed to produce. An extra iron ring 26 can be coated along with a batch of tools, for example. This ring can then serve with the next target, which may be titanium. The TiN bonds very tightly to the iron ring. Even though this coating is electrically conductive, it does reject the arc for the reasons discussed hereinbefore. It is only necessary for this nitride coating to be exposed during the oxide phase of the target clean-up, so this method can be very practical. Furthermore, the TiN coating does not result in flaking.

It is to be understood that the above detailed description of the various embodiments of the invention is provided by way of example only. Various details of design and construction may be modified without departing from the true spirit and scope of the invention as set forth in the appended claims.