At the Materials Research Society (MRS) Spring Conference in San Francisco the other day an attendee stopped by our Trade Show Booth in the Exhibition Hall at the Moscone Center where we had on display a multi piece linear target bonded to a copper backing plate and asked why stray material didn’t sputter out from the “gap” between the individual tiles and contaminate the resultant films? It might make sense to try to explain the answer to this query here. At least briefly.
Why doesn’t material sputter from the “Gap” in multi piece construction sputtering target assemblies and thus contaminate the resultant films? In the case of magnetically enhanced cathode assemblies, generically referred to as magnetron cathodes, it doesn’t. Magnetron cathode designs are incorporated in 99% of the existing planar cathode assembles and virtually all of the rotatable designs. For planar diode sputtering, this could be an issue, but we won’t address that here.
When certain materials need to be supplied in what is referred to as a “Multi Piece Construction” configuration for a verity of reasons such as overall size, expansion differential between the target material and backing plate, brittleness of the target, magnet configuration, etc. it is common to provide a spacing, or “gap” between the individual target segments. This gap is to compensate for the expansion of the target material as it heats up during ion bombardment and associated atomic bond breaking (an exothermic reaction) during the plasma deposition process. A similar analogy might be adding expansion strips in lengthy concrete sidewalks at various intervals to keep the cement from buckling or cracking as it expands (summer time) and contracts (winter time).
These segmented gaps vary a bit in length, depending on the coefficients of expansion for both the sputtering target and the backing plate and the target thickness, but are typically around 0.015″ or so.For an impinging ion to traverse through this gap, strike a solder or backing plate atom and propel it back toward the substrate would require a trajectory nearly perpendicular (normal) to the target surface. However, this is statistically impossible based on the magnetic field present at the target surface. This magnetic enhancement (M), which also multiplies the deposition rate many fold from just the electronic component (B) in the MxB relationship, provides the field that is directed from the permanent magnet contained within the cathode assembly beneath the target in a cosine distribution. This means that the magnetic field, and thus the impinging ions, are directed at a 45 degree angle toward the target surface. This makes it virtually impossible for an ion, arriving at a 45 degree angle into the gap, to traverse the distance from the top of the target to the bottom of the target, strike a solder atom, break the atomic bond and have it traverse backwards up through the gap toward the substrate. This path is mathematically eliminated based on the physics of the design. Think of bank shots on a pool table.
Some customers request that the targets have a 45 degree bevel at the adjoining gap surfaces, but this is a bit superfluous. It doesn’t cause a problem, but it also doesn’t offer any benefit. It does add to the cost of producing the target however since it adds to the overall length of the target tiles (by 1.414 times the thickness of the target times two times the number of gaps required) plus the additional machining costs required to produce the unnecessary angles.