Talking about the ion beam technology we usually mean the practical usage of phenomenon which are going on the surface of solids (target) exposed in vacuum under the directed flow of atomic particles - ions or neutrals. These ions are extracted from gas discharge plasma as the spatially restricted beam and accelerated by the electric field to the required energies. The discharge plasma is generated inside the special device - ion source. Usually for the purposes of ion beam milling and ion assistance we use the energy band from about 100 eV to 5000 eV but for some special goals (not optical yet), such as ion implantation, the energies up to a few tens keV may be used. It is important that the ion source is fully separated from a target and as ion source, as ion beam parameters are fully independent of parameters of a target and even of the target presence in a vacuum chamber. It is opposite to some other kinds of ion-plasma technologies. For example, for magnetron sputtering the target to be sputterd is one of the electrodes of system and its parameters define a lot of process and the whole system parameters. Further we'll see that this difference provides a lot of principal technological advantages.
So, what is happened when ions with energy 0,1 - 5 keV bomb the surface of the solid target? (The velocities of ions with such energies in vacuum may be about a few tens thousand meters per second!). Here we don't take in account the charge of ion, so actually we consider the interaction of neutral accelerated atomic particles with the surface of target but we'll use the term "ion" for simplicity. Later you will see that in practical ion beam technology we really use neutral or quasi-neutral atomic particles rather than ions. So ion collides with the surface of target and, because of ion mass and target's atoms masses are comparable (normally the argon, xenon, oxygen, nitrogen or their mixes ions are used), transfers its momentum to target's atom. This atom of target gets the momentum directed inside the volume of target and starts to move inside the target volume. During this moving it transfers in collisions the energy and momentum it has got from the ion to adjacent atoms. They, in turn, transfer the energy and momentum they have got to next atoms and so on. It is obviously that in collisions the first atom shares its momentum and energy with more and more number of atoms, so each following atom gets less energy and smaller momentum than preceding one and this process is fading. It looks like huge micro billiard with a millions balls in pyramid. But by contrast with billiard the forces of interatomic coupling bond these balls (atoms). These forces try to return displaced atom back to its equilibrium position in structure. In case when after all collisions atom has enough energy to overcome the action of these forces, it becomes displaced from its previous place in structure and takes the new place. If atom has not enough energy, it returns back to its previous position after series of fading oscillations.
The atoms on the surface of a target are excited by the same way. And when in the series of secondary collisions they get enough energy to overcome the forces of interatomic coupling, they leave the surface and, having residual momentum directed out of surface, move in vacuum. This process is called "ion sputtering". Note that the coupling forces between surface atoms are usually weaker than between bulk atoms and under the same conditions the first ones have a good possibility to break off their bonds with adjacent atoms .
Fig.1
You can see the process described above on Fig.1. Here I - incident ion, T - target, made from isotropic amorphous or polycrystalline material (it is important!), A -zone where target atoms are displaced from their positions in structure, B - zone where target atoms had been excited but then relaxed to their equilibrium states, and C - surface zone where target atoms may be sputtered from. If we have a lot of ions bombing the target surface, these zones are converted into surface layers and we have two layers: layer A - disturbed layer with the depth dA and layer B with depth dB - the depth of ion influence. Let evaluate this depths. We may say basing on most general physical reasons that the depth of sputtered layer has to be about 1 - 2 interatomic distances and the depth of ion influence for the ions with not very high energy has to be something like 10 interatomic distances. And it will be right! Usually the thickness of sputtering layer is about 0.5 - 1 nm , and the depth of ion influence - 5 - 10 nm. These are a little bit more problems with the disturbed layer. At first, it depth depends on material of target and energy of ions. At second, the state of the real targets surface completely depends on previous treatment. For example, after regular optical polishing of any material we have solid defect layer over the polished surface with the depth at least hundred angstroms. As a matter of fact, it is a very complicated problem to produce defect free surface. At third, adding energy to target atoms we can both create new defects and destroy defects that have been in structure i.e. improve its quality. I would like to stop to talk about this problem because we may spend too much time discussing the different surface and volume defects in the different materials which have been prepared by the different technologies and so on. But you have to pay attention on two important issues: (1) the surface of monocrystal may be amorphized by ion bombardment, and (2) if the thickness of defect layer introduced by previous treatment is more than the depth of defect layer introduced by ion bombardment we can eliminate the first one by ion sputtering (be careful, not always!).
Let's go on. Suppose that ion falls on the surface at the angle a that is read out from an external normal to surface (Fig.2).
Fig.2
In this case the direction of momentum propagation in target will change by the same angle and as a result: (a) the depths of disturbed zone and zone of ion influence will decrease and (b) the surface zone where the atoms are sputtered from will increase (only in one direction - in the plane of picture). Therefore bombarding the surface under the angle we can decrease the thickness of the disturbed layer and increase effectivity of sputtering. It is very easy to show that under the same other conditions the number of sputtered atoms will increase like Seca . But when angle a becomes close to 90o, ions start practically to slide along the surface and the energy and momentum transferred to target's atoms decrease. Correspondingly the number of sputtered atoms decreases too.
The number of atoms sputtered by one incident ion is called as "sputtering yield". So the angular dependence of sputtering yield S should be like shown on the Fig.3.
Fig.3
It is very interesting, that the result which we have got from so plain "billiard" model without any physical formula is exactly the same which has been obtained with the help of complicated theories and in experiments with a lot of amorphous and polycrystalline materials! Now we can determine some other main parameters the sputtering yield depends on. It is easy to understand that they should be (a) the energy E of the incident ion, (b) a target material bonding energy, and (c) atomic masses of incident ion and material of a target. Let's take a look on dependence S (E) for Ta sputtered by Ar ions. The threshold ion energy for sputtering usually is about 20 - 30 eV. With the increasing of ion energy from approximately 40 eV to 80 eV the sputtering yield increases from 0.0001 to 0.1. Further in the energy range 80 eV - 300 eV S increases from about 0.1 to 0.5, and in the range 300 eV - 5000 eV S increases just from 0.5 to 1.5. Many other materials have the same kind of S(E) dependence.
Last two dependencies are not very important for engineering applications and we will not consider them here.
However it would be much easier to work not with the sputtering yield S(atoms/ion) but with more applied parameter - the sputtering rate determined as the mass of target's material or the depth of surface layer sputtered in the unit of time. You can very easy to obtain the expressions for sputtering rate as the depth of surface layer sputtered in the unit of time:
V(nm/s) = 0.104 S(M/d)JCosa ,
and as the mass of target material sputtered from 1 cm2 surface in the unit of time:
V(g/s) = 0.104x10-7 S M JCosa ,
where S - sputtering yield (atoms/ion), M - atomic (molecular) weight (g), d - target density (g/cm3), J - ion current density (mA/cm2) and a - angle of incidence.
Let me say a few words about the real parameters of ion beam sputtering. For argon ions with the current density 3mA/cm2 and energy 200 eV the sputtering rates of optical glasses are in the range of 0.3 - 1.0 nm/s that is good enough for the most engineering applications. Note that only about 5% of ion energy are spending for sputtering, 95% is scattering in other processes, mainly heating the target. However in this case the power density on the surface of target would be 0.6W/cm2, so the target will be heated very weak (usually up to 50o - 90oC). It is one of the main advantages of ion beam treatment - we can work with a lot of temperature sensitive materials! And because of the depth of ion penetration is very small we will heat only thin surface layer (of course, don't forget about heat conductivity). I treated the target made from the optical quartz with diameter 6" and thickness 0.4 inches during 1 hour by unmoving Ar ion beam with ion energy 200 eV and current density about 12 mA/cm2 (the power density was 2.4 W/cm2). The temperature of the target was only about 220oC! It is very good result for the given conditions.
All foregoing is the necessary minimum of information needed to start work with ion beam technologies. But don't think that everything is so easy! These are much more complicated and unsolved both theoretical and experimental problems than simple and successfully solved ones. The part of them is as follows: