Vi. Shulga et P. Sigmund, ANALYSIS OF THE PRIMARY PROCESS IN ISOTOPE SPUTTERING, Nuclear instruments & methods in physics research. Section B, Beam interactions with materials and atoms, 119(3), 1996, pp. 359-374
Knockon sputtering from Mo isotopic mixtures has been studied by binar
y-collision and (in part) molecular-dynamics simulation. Differential
and integrated partial sputter yields have been determined for homogen
eous, mostly polycrystalline targets. Simulations have been carried ou
t on Mo-103-Mo-92 and Mo-100-Mo-50 mixtures and accurate scaling relat
ions with mass ratio have been established. The reliability of the sim
ulation code was tested by computation of absolute elemental sputter y
ields, and the influence of key input parameters on these yields was d
etermined. Relative sputter yields of isotopic mixtures do not depend
sensitively on most of those parameters, with the notable exception of
the target-target interatomic potential, the surface barrier, and the
treatment of nonbinary collisions. Most of our simulations refer to A
r bombardment, and ion energies covered range from less than 100 eV to
100 keV. At high ion energies sputtering is found to be preferential
in the light species in agreement with predictions from transport theo
ry, and the magnitude of this effect appears consistent with the few a
vailable experimental results. A weak dependence of the yield ratio on
emission angle is found which is identified as a surface scattering e
ffect. At low ion energies very pronounced preferential sputtering is
found which occasionally even may go in the opposite direction, i.e.,
preferential emission of the heavier species. This effect is governed
by threshold processes and characterized by a predominant contribution
from primary recoil atoms to the sputtered-particle flux. It is sensi
tive to the angle of incidence. Although the relative importance of th
reshold processes decreases rapidly with increasing energy their signa
tures remain visible at fairly high energies because isotope effects i
n higher generations of recoil atoms are weak. In the 1-5 keV range wh
ich is important for numerous applications and where most experimental
and previous simulational work was carried out, the two types of proc
esses compete. This complicates the analysis and explains why it has b
een difficult to reconcile discrepancies in the past.