There is still a bit of “Magic” associated with sputtering, but how the resultant films comprise the same compositional homogeneity as the starting target composition for a given alloy or composite material is not one of them. True, each individual element or molecule in the sputtering target has an associated “Sputter Yield” that defines how many particles are dislodged per each impinging ion, but under equilibrium conditions these factors have a tendency to null out. Let’s take a quick look at how this takes place.
Sputtering is basically the momentum transfer between charged particles (ions) in the plasma and the atoms (or molecules) on the target surface which are atomically bonded to one another. When the impinging ions strike the atomic surface of the target with enough energy, typically 600ev -1000ev, to break the atomic bonds that are holding the lattice structure within the material together, the bonds are broken and the particles are propelled into the plasma – hopefully toward the substrate. However, all atomic bonds are not created equal. Each element, each molecule, each alloy, each composite material has a unique set of bonds holding the individual atomic particles together within the lattice. Each bond energy is unique, differing from one another. These atomic bond energies can be measured and categorized. Conversely the amount of energy required to break these atomic bonds can be measured and categorized as well. However, at any given power density of plasma and for any given ion mass within the plasma there is only a singular striking energy at that target surface. That means that an argon ion within a plasma at 600ev there is a specific momentum associated with the striking energy with which it bombards the target surface. So, given a singular striking energy or impinging momentum within a given plasma, it would seem logical that more of the weaker bonded atoms or molecules would be knocked off of the surface than those with stronger bonds. In fact, this is exactly what happens. So how then can the resultant films, which seemingly should be composed more of weaker bonded atoms than that of the stronger bonded atoms, be the same composition as that of the target material?
Simple. Let’s look at a specific composition and see how this works. Take for example a sputtering target composed of a zirconium silicate (ZrSiO4) composite material for example. If we ignore the oxygen constituent here for a minute (that is a whole different can of worms) and concentrating solely on the ratio of the elemental metallic constituents, i.e. the zirconium and silicon (Si is not really a metal but let’s ignore that as well), we have a one to one (1:1) ratio of Zr to Si in the starting target material. But the atomic bonding energy of Zr is different than that of Si. Zirconium has a weaker atomic bond than that of silicon. So when a single ion of given energy strikes the surface of a target made up of Zr and Si atoms, more of the weaker bonded Zr atoms will dislodge from the target and go into the plasma, eventually ending up on the substrate. This will make the resultant films Zr rich or Si depleted. It will also make the target surface Zr depleted or Si rich. In a very short period of time an equilibrium condition will occur such that there will be less Zr atoms on the target surface and more Si atoms in a given ratio – different than that of the original 1:1 ratio. How different? That depends on something called “Sputter Yield”.
Sputter yield refers to the number of host atoms that are sputtered, or ejected, away from the target surface per a given ion striking the target with a given amount of kinetic energy. In this case, Zr has a statistical sputter yield of 0.7 atoms of Zr departing the target surface per given argon ion arriving at 600 electron volts where as Si has statistically only 0.5 atoms being ejected under the same deposition conditions. In other words, there are 0.7/0.5 or 1.4 times as many Zr atoms leaving the target surface as there are Si atoms. But this is only for an instantaneous amount of time. Because as these “Extra” Zr atoms are leaving, the target surface is becoming more Si rich, because fewer of the Si atoms are leaving per any given unit of time. In a very short period of time, there reaches an equilibrium condition based on the sputter yields. Although the weaker bonded Zr atoms are being emitted from the surface faster there are fewer of them to be ejected. As a result there are more Si atoms on the target surface because they are leaving at a slower rate. At some point this atomic balance reaches an equilibrium condition and the actual rate of departing Zr atoms and Si atoms reach the exact same ratio as the stochiometry of the initial target. There are fewer loosely bonded Zr atoms on the surface and more strongly bonded Si atoms. Therefore they are now departing NOT as the ratio of the sputter yields but as the ratio of the target stochiometry. As a result, the resultant films end up with exactly the same composition as the host target started with.