The Radial Arc

(See the Fluxion Presentation for schematics. Skip down to Background for an introduction to arcs.)

The standard in filtered-arc technology is the curvilinear or bent-tube type filter. How do you optimize a bent-tube filter to maximize coating deposition rate? We know that a longer bent-tube results in fewer ions reaching the exit. Ions are imperfectly contained and drift into the walls. So we make the tube shorter or the diameter of the tube larger - but then we must make the bend tighter, in order to preserve elimination of line-of-sight for removing macroparticles (MPs). However, this also reduces ion throughput. So a curvilinear filter is a loose-loose geometry. Efficiency is fundamentally limited.

How do we carry ions through a curved trajectory? This is the fundamental requirement of a filtered-arc source. The magnetic field itself has limited effect on the ions because ions are too heavy. But the much lighter electrons are strongly influenced by the magnetic field. By controlling where the electrons go, we can set up an electric field that in turn influences ion motion - in short, ions follow electrons.

Lucky for us, the arc produces lots of electrons - we can take advantage of this to create a strong electric field for influencing ion motion. Where are there the most electrons? Near to their source - the arc spots on the cathode. Therefore ideally, we create the required bend in the magnetic field as near to the the cathode as possible in order to maximize ion throughput. This is not really possible to do with a bent-tube filter but is a central feature of the Radial Arc geometry.

And the bend in the magnetic field does not have to be large, like in a bent-tube filter. Think about how small electrons and ions are - it can be a very sharp bend indeed. In fact, a sharp bend in the magnetic field helps to reduce ion losses because there is less opportunity for ions to drift into the walls. And, unlike bent-tube filters, the abrupt angle that MPs would have to navigate to escape the Radial Arc further decreases the likelihood that they can reach the substrate. A magnetic field with a sharp bend, which is also located close to the cathode where there are copious electrons, are features of the Radial Arc.

In the Radial Arc, the cathode is closely coupled to the filter duct. The magnetic field curves around the center magnet and the blocking disc protecting it, close to where electrons and ions leave the cathode. We thereby take advantage of higher electron densities near to the cathode to create a stronger electric field for carrying ions around the center blocking disc.

Another advantage of the Radial Arc geometry is radial symmetry. (Hence the name.) In a bent-tube filter, non-symmetrical sideways forces result in a skewed ion beam at the exit of the filter. To provide uniform coating, that small off-center beam must be scanned using additional coils, which only compounds the complexity/cost/impracticality of the bent-tube filter.

In contrast, uniformity is inherent to the Radial Arc geometry. Even though ions feel those same sideways forces, because they travel in a symmetrical ring around the central blocker, any net sideways motion zeroes out and uniform coating is obtained. Furthermore, that uniform coating can be easily controlled (using additional permanent magnets) to create a narrow or wide coating distribution as required by the user.

Sounds easy right? Here’s the catch. The one magnetic field controls six different effects in a filtered-arc source. In addition to ion throughput and coating distribution, these include arc stability, cathode erosion, triggering and the all important MP elimination. These individual effects are criteria that we must have full control over in order to make a useful filtered-arc source. But each of these criteria requires something different from the magnet field; what works for one does not necessarily work for the other criteria. For example, as the parameter space was being explored during development, there were cases where arc stability was achieved, but little coating reached the substrate - and vice versa. Magnetically stabilizing the arc on the cathode is an entirely different phenomenon than maximizing ion throughput - but both are determined by the one magnetic field.

This challenge is made even greater in the Radial Arc because the cathode is closely coupled to the magnet duct (to maximize ion throughput as described above). The arc is exposed to a strong magnet field that has a large effect on spot movement. We had to hope that there was some overlap in the parameter space (all the possible field shapes created by varying four sets of permanent magnets) that allowed for both arc stability and ion throughput to be optimized. Now add in the other four criteria, each one needing something different from that one magnetic field - and you can see the challenge. (This also helps explain why historically, a practical filtered-arc source has been so difficult to create.)

The Radial Arc is a solution that satisfies all six criteria. Maybe you can understand now why it took years of trial and error to get there. It is necessary to have a thorough understanding of what each criteria needs from the magnetic field, along with knowing how to create the desired magnetic field. Control over the magnetic field is achieved in the Radial Arc by changing the relative size, location, orientation of four sets of permanent magnets.

It is understandable, from an engineering point of view, to want to design the magnetic field at the cathode separate from the magnetic field of the duct - precisely because arc stability and ion throughput are two different phenomenon. But in reality, the arc “sees” only one continuous magnet field. If you want to optimize your filter geometry, it seems advisable to look at it from the point of view of the arc. Actually, there is only one continuous magnetic field that extends from the cathode into the magnet filter duct and out the exit.

Here is an example of the consequence of there being only one continuous magnet field in a filtered-arc source. Along with the shape of the magnetic field at the cathode surface, arc stability is determined by the electronic arc current travelling from the arc spots on the cathode, then through the plasma, finally intercepting the anode at some point; the magnetic field lines that the electrons follow must travel from the cathode to the anode. But, these same field lines also determine ion throughput - you don’t want the field lines to intercept internally (a separate anode located close to the cathode or the walls of the duct, for example) - they should travel all the way through the filter in order to lead the ions to the substrate.

You see, there is an inherent conflict between arc stability and ion throughput that is universal to all filters - a tradeoff that must be properly managed. Compounding the challenge is that the arc is less stable the greater the distance it must travel through the plasma to reach the anode (electrons quickly disperse from the arc spots, even with magnetic confinement). Hence, another advantage of the Radial Arc geometry (in addition to or as a result of, the close coupling described previously) is its short length: both arc stability and ion throughput can be better maximized.

So how do we control the arc-stability/ion-throughput balance in the Radial Arc? It is controlled to a fine degree by adding or subtracting permanent magnets behind the cathode that have a polarity that is opposed to the other magnets. These “cathode magnets” cause the field to balloon out around the center housing thereby controlling how much of the field (and the arc current carried through the plasma) intercepts the blocker disc on the center housing. Neat, right?

In addition, the geometry of the Radial Arc provides for a strong magnetic field in a compact device. In contrast to a bent-tube filter that is restricted to one outer magnet, the Radial Arc has inner and outer magnets that compound in strength in the space between them where the ions travel. Field strength is another important factor determining ion throughput.

Another advantage of the Radial Arc geometry is that the line-of-sight elimination for blocking MPs can be easily manipulated. (In contrast, a curvilinear filter geometry requires making a whole new bent-tube.) The exit aperture opening can be easily varied together with the MP blocker disc diameter, all the while maintaining line-of-sight MP elimination. This relationship is an important parameter that controls ion throughput. For example, which is better for maximizing ion throughput, a large aperture opening combined with a large MP blocker disc, or vice versa? The ability to experimentally determine the answer was another feature of the Radial Arc geometry.

In addition, certain cathode materials like graphite (used for depositing ta-C coatings) produce copious hard elastic MPs that bounce around the filter and are therefore more likely to reach the exit. (Metal MPs are typically liquid and don’t bounce.) With a bent-tube filter, this problem can be addressed by combining two filters in series, which results in even greater ion losses. In the Radial Arc, greater MP filtering can be achieved simply by using a smaller exit aperture opening. The aperture plate can be easily changed out.

Why not apply external electric fields using charged surfaces within the magnet duct to reflect ions thereby increasing ion throughput?  My experience is that getting the magnetic field correct obviates any benefit. You’re just going to mess with the arc circuit thereby effecting the arc stability/ion throughput balance - which is ultimately controlled by the magnetic field. So another consequence of the supreme role of the singular magnetic field is that applying external electric fields (using charged surfaces within the magnet duct) isn’t very effective, if you get the magnetic field right.

Every surface in the Radial Arc, including the blocker disc, is at anode potential (and earth grounded). Damage caused by concentrated anodic current is avoided by properly designing the magnet field.

If you’ve read this far, maybe you will appreciate more detail on the use of permanent magnets in the Radial Arc instead of magnet coils. Magnet coils are typically used in filtered-arc sources like the bent-tube filter. There is a profound difference between the two. In fact, using permanent magnets seems like a really dumb idea because field lines are concentrated at the poles of a permanent magnet (instead of looping around as they do for coils):  plasma is concentrated where it can quickly drill through the side of your device! (I’ve got a storage room full of holed development devices to prove it.) The Radial Arc started out using magnet coils - and was heavy and complicated, and didn’t work very well.

I was sure permanent magnets wouldn’t work (for the reasons described above) but also knew that I was wrong often enough (especially about things I was “sure” about) that I tried it anyways. I was encouraged enough at first - it sort of worked. But mostly, there was a path forward made possible by the nature of permanent magnets: they can be moved around, and their relative size, location and orientation more easily manipulated. This allowed a lot of learning to happen and ultimately, the magnetic field to be refined to optimize for the six criteria (described above). For example, through lots of trial and error, I was able to learn how to mitigate the problems caused by plasma concentration (like burning through the walls of the device!) using some field design tricks. This would not have been possible with magnet coils - they are too hard to change in size, shape and location and require problematic electrical and water vacuum feedthroughs. Plus, the Radial Arc retains the beautiful simplicity of unfiltered-arc:  it is compact and does not require additional magnet coil power supplies. One person can mount it, hook up a welder and start coating - like unfiltered-arc.

I hope you enjoyed learning about some of the unique aspects of the Radial Arc geometry. Thank you for your interest.

Background

Arc evaporation is a simple physical vapor deposition technology that produces a large flux of highly ionized vapor, valuable for depositing hard, dense, and well adhered industrial coatings. Arc evaporation also produces macroparticles (MPs) that create defects in the films, relegating this technology to applications that are mostly insensitive to these defects, such as cutting tool coatings.

Many efforts have been pursued over the years to filter out MPs. Although more or less successful at reducing MPs, the usefulness of these filtered-arc sources has been limited by a significant reduction in the coating rate as well as an increase in complexity and cost.

The present convention in filtered-arc, the curvilinear or bent-tube filter, borrows from fusion research to bend the ions around a curve using magnet coils wrapped around the bent-tube.  MPs, unlike the ions, are unaffected by the magnetic field and travel in straight lines, getting caught on the walls of the tube and are thereby prevented from reaching the exit of the filter. The main problem is that the ions are imperfectly confined and mostly don’t make it through the filter. Low deposition rates over small areas, bulky size, and high cost have prevented widespread adoption of filtered-arc technology, despite the advantages of purified ion deposition.

Having worked with a number of different kinds of filtered-arc sources (including a straight source that helped pioneer the ta-C diamond coatings business), I became knowledgeable in the principles of efficient filter construction as well as the limitations of the state of the art. It was clear that a different approach would be required if the benefits of ion deposition were to become more widely available.

What emerged after much rethinking and more than two decades of development, is the Radial Arc source. By redesigning the magnetic field and the cathode/anode geometry, ion transport efficiency and MP elimination is achieved from a compact, symmetrical, uncomplicated, and robust device. The use of permanent magnets instead of magnet coils decreases complexity and size. The ion flux is also inherently distributed, eliminating the need for magnetic rastoring to achieve coating uniformity. The novelty of Radial Arc technology was demonstrated by the awarding of strong patents.

The geometry of the Radial Arc source can be visualized as a the volume created by rotating a ninety degree bent-tube filter around one of its two ends: