Full text loading...
AFM images and histograms. (a) Contact mode topography AFM image of a ta-C sample grown on silicon and irradiated with 1 GeV lead at 1011 ions/cm2. The number of hillocks in the depicted area of 200 × 200 nm2 is within 10% in good agreement with the applied ion fluence. This suggests that each projectile impact results in the creation of a hillock. (b) Line scan along the red line marked in the AFM image. Hillock height is about 3 nm and width at half maximum is about 8 nm. (c) Height distribution of hillocks on samples irradiated with several ion species and energies. The data were obtained by automatically detecting hundreds of hillocks in multiple AFM scans.
MD simulations of hillock formation. The top frame shows an overview of the simulation cell shortly (6 ps) after insertion of the thermal spike. Atoms are colored according to potential energy in the left half of the frame (−7.5 eV red to −4 eV blue, see color scale at bottom of figure). Atoms are colored according to a locally averaged temperature (raverage = 0.5 nm) with blue corresponding to 0 K and red atoms to 3500 K and above. The horizontal dashed line indicates the upper boundary of the frozen bottom later of atoms. The vertical dashed lines indicate the extent of the thermal spike at the time of insertion (r = 3 nm). The bottom four frames show the hillock region for selected simulation runs at 1.7 keV/nm, 3.35 keV/nm, 6.7 keV/nm, and 10.1 keV/nm after a relaxation time of about 30 ps. The potential energy distribution of the atoms in the spike is narrower than in the bulk due to a prevalence of sp2 hybridized atoms. The spike and hillock regions have cooled down considerably and no further change in hillock height was observed in the simulations. The hillock height clearly gets larger with increasing energy loss. Very little or no sputtering can be observed if dE/dx ≤ 10.1 keV/nm.
Comparison of experimental and simulated hillock height. The experimental hillock heights are plotted as a function of the electronic energy loss dE/dx, while the simulated data points use (dE/dx)eff = 0.235 × dE/dx where (dE/dx)eff denotes the effective energy loss providing local lattice heating. Both simulation and experiment agree reasonably well with a linear dependence of hillock height on energy loss. Furthermore, both data sets exhibit a similar y-axis intersections corresponding to an energy loss threshold for hillock formation of 10 ± 2 keV/nm.
Article metrics loading...