Order-of-magnitude differences in retention of low-energy Ar implanted in Si and SiO2

Abstract

The retention of 1 and 5 keV Ar implanted at 45° in Si and 4.3 nm SiO2 on Si was studied at fluences between 3 × 1014 and 1.5 × 1016 cm−2. X-ray photoelectron spectroscopy (XPS) served to monitor the accumulation of Ar as well as the removal of SiO2. Bombardment induced changes in oxygen chemistry caused the O 1s peak position to move toward lower binding energies by as much as 2.2 eV. Plotted versus depth of erosion, the fluence dependent changes in oxygen content, and peak position were similar at 1 and 5 keV. The Ar content of Si increased with increasing exposure, saturating at fluences of ∼2 × 1015 cm−2 (1 keV) and ∼6 × 1015 cm−2 (5 keV). Much less Ar was retained in the SiO2/Si sample, notably at 1 keV, in which case the low-fluence Ar signal amounted to only 8% of the Si reference. The results imply that essentially no Ar was trapped in undamaged SiO2, i.e., the Ar atoms initially observed by XPS were located underneath the oxide. At the lowest fluence of 5 keV Ar, the retention ratio was much higher (43%) because the oxide was already highly damaged, with an associated loss of oxygen. The interpretation was assisted by TRIM(SRIM) calculations of damage production. Partial maloperation of the ion beam raster unit, identified only at a late stage of this work, enforced a study on the uniformity of bombardment. The desired information could be obtained by determining x,y line scan profiles of O 1s across partially eroded SiO2/Si samples. Fluence dependent Ar retention in Si was described using an extended version of the rapid relocation model which takes into account that insoluble implanted rare-gas atoms tend to migrate to the surface readily under ongoing bombardment. The range parameters required for the modeling were determined using TRIM(SRIM); sputtering yields were derived from the literature. The other three parameters determining the Ar signal, i.e., (1) the thickness w of the near-surface Si region devoid of Ar, (2) the relocation efficiency Ψrlc, and (3) the effective attenuation length L in XPS analysis were varied within reasonable limits until the calculated retention curves for 1 and 5 keV Ar in Si agreed with experimental data to better than 8%, using the same XPS sensitivity factor throughout. Results: w = 1.4 ± 0.1 nm, Ψrlc = 6.6 ± 0.5, and L = 2.7 ± 0.2 nm. Combining experimental and calculated data, it was found that the Ar trapping efficiency of the damaged oxide is intimately correlated with the loss of oxygen. The calculated stationary areal densities of all retained Ar are compared with results obtained by high-resolution medium-energy ion scattering spectrometry. Attractive areas of future research in rare gas retention and nanobubble formation are sketched briefly.