As most of our customers around the world are well aware, the process of sputtering involves charged particles, or ions, (gaseous molecules with an outer electron stripped off containing a positive charge) bombarding negatively charged sputtering target atoms or molecules where the surface charge is derived from an electrical component combined with a magnetic field. This plasma process requires a significant energy or momentum transfer. The kinetic energy of the incoming ions must exceed that of the conventional thermal energy (typically >1ev) for a given species. A non-uniform depletion of the target surface atoms evolves over time as the enhancement from the magnetic component provides a cosine distributed field that must be closed or looped at the ends. This non-homogeneous depletion is referred to as the erosion profile for a given target surface.
The number of actual particles (atoms or molecules) that are ejected per each incident ion collision on the surface is determined by a number of factors, i.e. the incident angle of the impinging ion, ion mass, ion energy, mass of the target atom, binding energy of the atomic bonds of the target material, etc. This number is referred to as the sputter yield for any given elemental material. That is, on average, how many particles of target material are ejected from the target surface per given impinging ion. For a plasma in equilibrium in a fixed environment, the sputter yield is constant for any given material, leaving only the atomic or binding energy as a material variable. For any elemental target composition, this is only dependant on lattice parameters varying within the crystalline lattice. For any given material, these energies can vary significantly based on the specific location of the atomic bond – such as grain boundary, crystal lattice, dislocation, dangling bond, etc. However, on average, over a given geometrical surface area the sputter yield is constant over time.
The question arises then as to what occurs when the sputter yields are NOT constant over time, such as in an alloy where two or more atomic species are present. As expected, each element has its own intrinsic sputter yield so the different atoms are depleting at different rates. This would imply that the resultant films would consist of some composition other than that of the stoichiometry of the initial target. Obviously this would NOT be a desirable effect. Fortunately though this is NOT the case. Herein lies the “magic” of sputtering. Yes, the elemental constituents all do have different specific sputter yields and do deplete the target surface at different rates. But only initially. Initially meaning for a matter of a few seconds or so. During this short period of time, the species with the highest sputter yields, i.e. the lowest binding energy, are sputtered from the target surface at a rate faster than those species of lower sputter yields. So then there are fewer of these atoms present on the target surface. This also implies that there are then more of the lower sputter yield atoms on the target surface, per unit area, present to be sputtered away in the plasma. At this point an equilibrium is reached whereby there are less of the high sputter yield atoms and more of the low sputter yield atoms on the surface available to be struck by the impinging ions. Under these equilibrium conditions the plasma then contains the same percentage of atoms of each species as that of the sputtering target material. Thus the resultant films end up having the same compositional consistency and homogeneity as that of the initial sputtering target.