Orientation‐dependent nanostructuring of titanium surfaces by low‐energy ion beam erosion

Regular nanoscopic ripple and dot patterns are fabricated on poly‐crystalline titanium samples by irradiation with 1.5 keV argon ions at normal incidence. The morphology of the nanostructures is investigated by scanning electron microscopy and scanning force microscopy. The ripple structures exhibit a saw‐tooth cross‐section profile. Electron backscatter diffraction experiments are performed to analyze the local grain structure. The study suggests a distinct correlation of the nanostructure morphology to the crystallographic orientation of the titanium surface.


INTRODUCTION
Titanium has emerged a highly relevant material for biological and medical applications because it has attractive physical properties such as low-weight, resistance against corrosion, and a considerable strength combined with the high bio-compatibility of titanium oxide, which can easily be obtained via surface oxidation of a titanium device. 1 A powerful method for customization of the surface properties is nano-patterning allowing to achieve a substantially increased effective surface size, a specific local shape, or a defined surface periodicity. Hence, the combination of a tailored, inorganic surface based on titanium together with functionalized molecules or specific bio-species bears a great potential for innovative, future applications. 2 A flexible processing tool to fabricate different kinds of nanostructured surfaces is ion beam erosion with low-energetic ions (≤ 1.5 keV). 3 The technique allows the defined formation of ordered fields of nanodot or nanoripple structures by direct surface topography modification. The nanostructure formation is kinetically driven considering the interaction of the impinging ions with the surface topography and surface relaxation. [4][5][6][7] The first aspect covers the preferential sputtering in valley regions rather than on hillocks, 8,9 gradient dependent sputtering, 10 and particle-scattering effects. 11 The surface relaxation is usually discussed to result from contributions by viscous flow, ballistic drift, and thermally driven diffusion. 6 Because the contributing factors are of pure physical nature, this process is applicable to a broad variety of materials with an amorphous 12 or crystalline 13 matrix. For amorphous materials as well as crystalline materials, forming an amorphous surface layer under ion irradiation, curvature driven surface relaxation is predominant, 10 which is of minor importance on metal surfaces as result of a much enhanced intrinsic surface mobility. The nanostructure morphology can be controlled by the process parameters (ion energy, fluence, etc) and the process geometry (ion incidence angle). The ion erosion process can be further customized by an in situ chemical surface modification. [14][15][16] For this purpose, a reactive surface species is additionally provided during the erosion process, eg, by co-deposition. This surfactant-assisted ion erosion process can promote the surface structuring. Depending on the surfactant species, local masking effects can alter the topography formation in ion beam erosion also. In advance to the particular process conditions and surfactant effects, the substrate orientation for crystalline materials can influence the nanostructure formation.
For example, in ion beam erosion on misoriented semiconductor materials as GaAs at elevated substrate temperature the formation of nanoripple structures is possible at normal ion incidence. 17 The ripple morphology can be varied by changing the misorientation angle. The driving force for orientation-dependent ripple formation is attributed to the increased adatom diffusion along dense packed crystal planes. 18 Because the diffusion on low-melting metal surfaces is enhanced, structure formation is observed already at room temperature or even below. 19 (A) (B)

EXPERIMENTAL
An ultra high vaccum (UHV) chamber is applied with a base pressure of 2 × 10 −5 Pa. A transformer coupled plasma ion source with argon (5N) gas and a triple grid extraction grid system is used for 1.5 keV ion generation. The distinct, concave grid shape allows the aperture-less generation of a constricted narrow ion beam. 20 At the machining position, the ion beam full width at half maximum is ≈ 5.5 mm with a total beam current of ≈ 1.0 mA. The process pressure is 3 × 10 −3 Pa.
Hence, background gas effects are expected to be negligible.
Polycrystalline pure metal titanium samples were prepared by single-point diamond turning and subsequent polishing, which reveals a flat surface with a low roughness of ≈ 3 nm root mean square (rms).
For ion beam treatment, the sample is mounted onto a water-cooled sample holder. Deterministic processing by scanning the ion beam over a sample area allows homogeneous nanostructuring onto a broad device surface. The experiments are performed under normal ion incidence towards the sample surface. However, a possible misalignment may influence the nanostructure morphology. As already introduced, for many materials, an inclined ion incidence angle is even necessary to allow the formation of regular nanoripple structures. To avoid misinterpretations, experiments with substrate rotation have been carried out also. By that way, an unintended, slight misalignment of the ion incidence angle is eliminated. Another experimental uncertainty can be caused by contaminations during ion beam processing. Contaminations can be introduced due to an impure ion beam itself, eg, due to low-quality process gas or erosion effects inside the ion source.
Because the first factor can be excluded, the latter might contribute because carbon impurities can be formed as the ion beam grids are made of graphite. Further sources of contaminations can be a considerable background gas pressure, 21

RESULTS AND DISCUSSION
Four samples are fabricated by argon ion beam processing (Table 1).
Samples #1 and #2 are placed each on one half of the same piece of titanium and samples #3 and #4 in the same way on another piece of titanium. The effect of titanium revetment and sample rotation are investigated: Sample #1 is fabricated without revetment and without rotation, while sample #2 is made with revetment and rotation.
The removal depth is determined at the defined mask edge by CM ( Figure 1B). For sample #2, the argon ion fluence is roughly doubled and the removal depth scales in good agreement with the fluence.
Hence, there is no apparent difference in the ion etching rate. This result indicates, that the possible occurrence of contaminations due to peripheral sputtering from surrounding areas, in particular from the silicon mask covering sample #1, has no significant influence on the titanium etch behaviour. Indeed, the formation of metal silicides in surfactant-assisted ion beam processing can strongly influence the etching characteristics. Furthermore, the formation of the nanostructures can even be reliant on a specific silicide surface coverage, which forms during surfactant-assisted processing in a dynamic equilibrium. 15 In the fabrication of samples #3 and #4, the effect of sample rotation only is investigated, and no effect could be observed. Hence, the sample alignment during processing is supposed to be sufficiently accurate.
The average removal depth of sample #1 is In Figure 3 the cross-section geometry of ordered nanoripple structures is shown. Note, the nanostructures on samples #2-#4, which are fabricated with titanium revetment, do not differ from those of sample #1 qualitatively. The ripples exhibit a saw-tooth cross-section.
The larger the structure period, the better the geometrical specificity.
Within the saw-tooth geometry, the ripple height h and the surface roughness R q are correlated via To give an impression about the structure sizes, at sample #1, the  Figure 3. From , the saw-tooth slope angles and can be separated via As a result, the ripple topography can be fully described by the geometrical parameters: period, height, and both slope angles.
EBSD experiments are performed to analyze the local crystal structure ( Figure 4A). Up to 1155 K titanium is thermodynamically stable in the hexagonal -phase. The MTEX toolbox for MATLAB is applied for EBSD data evaluation and grain recognition. The rhombohedral unit cell is described by the lattice vectors a 1 and a 2 , which lie in-plane in the dense packed (00.1) plane, and lattice vector c, which is directed parallel to [00.1] (see the structure inset in Figure 4b).  Figure 2A). For grain #03, the Ψ-angle is almost 90 • , and no nanostructures are observed but a smooth surface. The arrows in Figure 4C illustrate the direction of the c-vector projected to the surface plane. The nanostructure propagation direction is oriented along the projected c-vectors. Hence, not only the nanostructure morphology but also their in-plane alignment are found to be correlated to the crystallographic orientation of the titanium grain structure.
Qian and Zhou 24 reported on the nanoripple formation at titanium surface after irradiation with midenergy gallium ions (30 keV). However, despite ripples were also observed at normal ion incidence, an incidence angle of 15 • was necessary, in which a crystal orientation dependent formation was obtained. Argon ion irradiation experiments at nickel films reveal even higher incidence angles of 70 • or 80 • to be necessary for ripple formation. 23 However, the nickel grain sizes are very tiny, and the ripples formed at high incidence angles comprise a large amount of grains, which indicates that those formation conditions are not correlated to any crystal alignment effect but the distinct experimental process geometry at grazing ion incidence. In contrast, in the present study already at normal ion incidence, the local crystal structure determines the nanostructure formation. If surface adatom diffusion is the major driving force, crystal facets with higher diffusion coefficients will determine the forming etch morphology corresponding to the equilibrium crystal etch shape. Actually, the adatom diffusion on the hexagonal dense packed titanium (00.1) plane is predicted to be exceedingly fast. 25 In this view, the top facets of the ripples would be (00.1). Indeed, the experiments show increasing ripple terraces, if the Ψ-angle becomes lower, ie, the c-axis nears the surface normal. Also the in-plane ripple alignment dependency on the surface-projected c-axis matches this view. However, more systematic investigations are necessary to prove this hypothesis for normal ion incidence and to study the limitations due to specific process conditions.

CONCLUSIONS
In summary, the formation of regular nanostructures is observed on titanium surfaces after low-energy argon ion irradiation at normal incidence. With respect to the periodicity, continuous ripple patterns or regular arrays of nanodots are obtained. The nanostructure period and the surface roughness, which is directly related to the nanostructure height, are correlated and follow a power-law scaling behaviour. The investigations suggest, that the nanostructure topography is determined by the local crystallographic alignment of the grains at the surface rather than by the process conditions of