Measurements and Collision—Radiation Damage Theory of High-Vacuum Sputtering

A brief summary of some of the important experimental and theoretical work related to the subject of metallic sputtering is presented. The need for measurements in high vacuum is indicated and an ion beam which utilized a Philips Ion Gauge discharge ion source to make high-vacuum sputtering ratio measurements is described. Absolute sputtering ratio data for the gas-metal combinations Ag-Kr, Ag-A, Ag-Ne, Ag-He, Cu-Kr, Cu-A, Pb-A, and Pb-He are presented in terms of the number of atoms sputtered per incident ion, $n_a$, versus incident ion energy, $E_0$, for ion energies varying between 400-6100 ev. The data are interpreted by treating the incident ion as a hard sphere which "cools" in a manner similar to a neutron losing energy by collisions in a lattice, each collision producing recoil atoms and atomic displacements near the surface. The number of atoms escaping, or "sputtering," from the metallic surface is reduced from the number displaced by absorption within the metal which is accounted for by a parameter $\alpha$. By use of elementary neutron cooling theory and the Seitz formula for displacements produced by a recoil atom within a solid, the formula for the number of atoms sputtered per incident ion is given by $n_a={\left(\frac{\epsilon E_0}{E_d}\right)}^{\frac{1}{2}}\Sigma \stackrel{n_c}{n=1\mathrm{}}{e}^{-\alpha \surd n}{e}^{-\frac{\mathrm{}(n-1\mathrm{})\mathrm{}\xi }{2}}, n_c=\frac{1}{\xi }\ln \left(\frac{\epsilon E_0}{E_d}\right),$ for the case of ions more massive than the metallic atoms. The effect of ions rebounding from the surface after the first collision is considered to produce effectively two types of incident current particles; (1) incident gas ions and (2) recoil metal atoms. These considerations lead to a modified sputtering ratio formula, which reduces to the above equation when $M_1\geqq M_2$. The displacement energy for the process, $E_d$, is calculated by use of G. K. Wehner's data on sputtering thresholds and the relation $E_t=\frac{E_d}{\epsilon }$. A fair fit to the experimental data is obtained by suitable choice of $\alpha$ in the modified formula for the cases studied.