Introduction

The high-fluence ion bombardment, generally greater than about 10⁷ ions cm², of virtually all solids produces a fascinating and rich variety of surface topographical features. The origin and evolution of such features can usually be traced to local variations of the sputtering process. For example “extrinsic” processes such as the presence of a contaminant can modify sputtering yield so as to produce conelike features. On the other hand, one “intrinsic” process where large concentrations of light ions, such as He ions, can accumulate in solids and form blisters, which burst or exfoliate, can form etch pits which are further contoured by sputtering. In Si wavelike evolution can be suppressed by increasing the background oxygen partial pressure during O⁺ ion bombardment, by using Cs⁺ ion bombardment or by sample rotation during ion bombardment. Other work with a few keV energy Ar⁺ ions is reported not to lead to ripple formation. Similar results have been reported with GaAs, InP , Ge and AlGaAs targets. These indicates the wide range of observations, some conflicting, on repetitive feature development on semiconductors and it is therefore not surprising that there is little understanding of the basic mechanisms which lead to their initiation and evolution.

Experiment

Si(100) wafers have been irradiated at incidence angle 45⁰ with different ion species (Si⁺, Ne⁺, Ar⁺ and Xe⁺), energies (20 and 30keV), high fluences and at target temperatures from 120K to room temperature. The observation is done with SEM and AFM. The ion flux was normally of order 10¹⁵ ions s^-1 cm^-2. The inert gas ions, Ne⁺ Ar⁺ and Xe⁺, were chosen for their expected inability to react chemically with Si as would O₂, N₂ or even Cs, however, because of their mass difference, their penetration ranges, defect production rates, and sputtering yields may be expected to be substantially different. Initial studies revealed that for most species and temperatures easily recognizable topography was not produced up to fluences of about 10¹⁸ ions cm^-2, for low temperatures, and 3 - 10x10¹⁸ ions cm^-2 for room temperatures.

Results

The effect of reducing the target temperature below room temperature is for all ion species to promote the initiation and evolution of topographic features (quasiripples) and to allow attainment of typical feature structures at lower ion fluences. The wavelike topography which evolves, under suitable target temperature conditions, for large ion fluences, is rather independent of the ion species; these wavelike structures have wavelengths, relatively independent of ion species, lying in the region of 0.8 - 1.0μm (i.e 800 - 1000nm) and, except in the case of Xe⁺ ions, not varying with target temperature. For Xe⁺ ions the wavelength varies from 0.44μm, at room temperature, to 1μm at 120K.

Other investigators have argued that wave structures evolve from randomly roughened surfaces due to the competing action of sputtering processes, which depend on surface curvature, and surface diffusion or viscous flow while some other investigators have demonstrated that random atomic removal by sputtering can, in principle, initiate the required roughening. These models predict an increasing wavelength with increasing temperature, in contrast with the present work. According to the BH model, the wavelength should be ion species dependent, at a fixed ion energy, since collision cascade dimensions, which are influential in determining sputtering yield curvature dependence and, hence, selected wavelength, will vary with ion species¹ . Their observation of a definite threshold fluence for ripple or wave formation, as well as indication of no real signs of random roughening² to an amplitude at which wave structures are first observed, is not compatible with a gradually developing dominant wave-vector system.

Neither of the lighter ions, Ne⁺ or Si⁺ incident at 45⁰ and 20-30 keV on to Si at room temperature, generate any major topographic features even at ion fluences greater than 10²⁰ ions cm^-2. Ar⁺-ion bombardment, however produces ripple trains and etch pits and Xe⁺ ions generate extensive low-amplitude transverse wave structure at room temperature. Reduction of target temperature to 200 or to 120 K does allow feature evolution (Ne becomes similar to Ar) and at high enough fluences all species produce faceted transverse wave structures. Differences in defect density, which modify flow processes, and configuration of accumulated implant species may explain these differences in ion species and target temperature behaviors. They proposed a number of speculative reasons for their results.

OEY