Influence of O+2 energy, flux, and fluence on the formation and growth of sputtering-induced ripple topography on silicon

Abstract

The formation of ripples on Si(100) by O+2 sputtering at an angle of incidence of 40° and energies from 1 to 9 keV has been studied using secondary ion mass spectrometry and scanning electron microscopy. At 1 keV no ripples are observed. Between 1.5 and 9 keV ripples are observed oriented perpendicular to the ion direction with average wavelengths that increase, from ∼100 to 400 nm, approximately linearly with O+2 energy. Two-dimensional fast Fourier transforms of secondary electron images are used to investigate the frequency distribution of the ripples. For the conditions studied, the distributions of frequencies appear approximately Gaussian. At 1.5 keV, the wavelength and growth rate with sputtered depth are independent of flux for fluxes from 15 to 150 μA/cm2. Accompanying ripple formation are changes in secondary ion yields. The changes occur abruptly at depths that increase, from ∼0.2 to 5.6 μm, with O+2 energy. In contrast, sputtering with Ar+ at 1.5 and 7 keV to depths 5–10 times those that produce ripples with O+2 produce no observable topography. These results are discussed using several existing theories for ripple formation and growth. Ripple growth and the variations in secondary ion yield are modeled by accounting for the change in local angles of incidence as the ripples grow. This model describes well the variation in secondary ion yield assuming an exponential growth rate. Ripple formation is discussed in terms of a balance between roughening (by sputtering-induced surface stress and by the dependence of the sputtering yield on surface curvature) and smoothing (by both diffusion and ion mixing). Variation in ripple wavelength with energy is not simply explained by these theories. Surface smoothing by cascade ion mixing can, however, make the wavelength, as observed, independent of ion flux. Finally, the possibility of formation of ripples by phase separation within the SiOx surface layer is discussed.

Morphology of ion-sputtered surfaces

“…The authors have found that the yield changes exponentially in the first stages of ripple development and saturates for large sputtered depth. Direct evidence on ripple amplitude saturation was obtained by Erlebacher et al [20], who measured the time evolution of the ripple amplitude in experiments bombarding Si (100) [18,21,22], most of these studies were contradicted by subsequent investigations [23] where such chemical component were not present. Furthermore, in Refs.…”

“…The calculation also predicts that the ripple direction is a function of the angle of incidence: for small θ the ripples are parallel to the ion direction, while for large θ they are perpendicular to it. As subsequent experiments have demonstrated [15,18], the BH model predicts well the ripple wavelength and orientation. On the other hand, the BH equation (8) is linear, predicting unbounded exponential growth of the ripple amplitude, thus it cannot account for the stabilization of the ripples and for kinetic roughening, both phenomena being strongly supported by experiments (see Sect.…”

“…Ripple amplitude: Indirect results on the ripple amplitude were presented by Vajo et al [18] in their study of the surface topography induced secondary ion yield changes on SiO 2 surfaces bombarded by O + 2 ions. The authors have found that the yield changes exponentially in the first stages of ripple development and saturates for large sputtered depth.…”

“…In the following we give an order of magnitude estimate for the effective surface diffusion constant D xx and compare it to K, using data from [18,19] for Si(001). Taking Y = 2.6, J = 670 µA/cm 2 , ǫ = 9 keV, a = 100Å, a σ = 2, a µ = 4, and θ = 40…”

Quantum dot and hole formation in sputter erosion

“…While successful in predicting the ripple wavelength and orientation [9], the linear theory fails to explain a number of experimental features, such as the saturation of the ripple amplitude [10][11][12], or the appearance of kinetic roughening [14,15]. To address these questions it has been proposed [16] that these shortcomings can be cured by the inclusion of nonlinear terms and noise, derived from the Sigmund's theory of sputtering.…”