Self-Organized NDAs onto a Substrate

First of all, we started from a 200-nm-Ta film as an interlayer between upper Al and a Si substrate to fulfill a typical PAA process in 0.3 M oxalic acid at 40 V. Figure 2.2 shows the top-view scanning electron microscopy (SEM) image of the porous anodic alumina film after a pore-widening treatment. The self-organized nanopores with a uniform size distribution have a pore diameter about 60 nm and an interpore distance about 100 nm. The pores do not show a long-range order but within a pore array domain. Figure 2.3 recorded chronoamperical curves during the anodization process.

At the first stage, the current started high, following a decline as an oxide barrier of PAA formed. The current then reached to a steady level accompanying the growth of pore channel. When the oxide barrier layer at the bottom of the pore approached the Ta–Al interface, anodization of the underlying Ta started. Successively, the currents would decrease gradually and a visible change in color of the sample from metal gloss to charcoal could be observed. Once anodization was completed, the values for current density dropped considerably and remained at a background level.

To investigate the morphological evolution of the resulting structures chronologically, we terminated the anodization process at four different time points, named A, B, C and D in Figure 2.3. After selectively removing upper alumina, the SEM images of these four specimens were displayed in Figure 2.4. It is noted that none of the structure was observed for point A, yet the growth of nanostructure was prohibited. From Figure 2.4b, it is clearly seen that the preliminary tantala nuclei

arose from the scalloped Al concaves as inverted U-shape barrier layer approached the Ta layer. We assume that tantala nuclei commenced growing at the commencement of current decay. Figure 2.4c reveals the fact that the tantala nuclei grew and expanded in horizon and vertical. The residual Al metal around the tantala nanostructures were simultaneously anodized pending it is completely consumed by anodic oxidation.

Finally, the cone-like tantala nanostructures are shown in Figure 2.4d. The average geometric features of tantala nanodot arrays were a period of 100 nm, a base diameter of 80 nm, a height of 50 nm and a density of 8  10⁹ cm^-2. During the growth of nanodot arrays (NDAs), the ionic current was distributed unevenly across the barrier layer, concentrating along nanometer sub-channels, which merged inside the hillocks and resulted in appearance of the root-like structure and distorted hexagonal bases.

The interpore distance of NDA is in common with the relation d=-1.7+2.81Ua,⁶³ which d represents the interpore distance and Ua is a constant for the electrolytes of sulfuric, oxalic, and phosphoric acid solutions. Cross-sectional SEM images in Figure 2.5 reveal that an isolated nanostructure was embedded at the bottom of each alumina pore, i.e., at the interface between the porous alumina and un-oxidation Ta.

The aforementioned information inspired us to propose the mechanism of the formation of NDA and illustrate in Figure 2.6. The anodization behavior of the Ta/Al film on the silicon wafer is different from the case of the Al film directly deposited on semiconductor substrates or foil aluminum. In the first instance, the above aluminum layer oxidized to alumina, accompanied by the outward migration of Al³⁺ and inward diffusion of O^2- driven by the applied electric field, leading to the vertical pore channel growth. Furthermore, the bottom of the alumina film consisted of an array of convex hemispheres during the initial anodization, and the position of nanodot had been decided. The alumina dissolution at the alumina/electrolyte interface is in equilibrium with the alumina growth at the Al/Al2O3 interface. As the oxide barrier

layer at the pore bottom approaches the Ta/Al interface, the O^2- migrating inwards through the alumina barrier layer are continuously injected into the Ta layer and form the tantalum oxide. The O^2- released from the dissociated barrier layer at the Ta2O5/Al2O3 interface are also injected into the Ta2O5 layer, while the released Al³⁺

migrate outwards through the remaining barrier layer and are mostly expelled in the electrolyte. The O^2- injected into the Ta2O5 layer then migrate inwards and the Ta layer is anodized normally to form new oxide at the Ta/Ta₂O₅ interface. In brief, the underlying tantalum oxide by O^2- transported through/from the barrier layer of the initially formed porous alumina without direct contact of Ta with the electrolyte. The tantalum oxide nanodot resulting from oxidation of the Ta layer is accompanied by a volume expansion. Eventually, the aluminum completely transferred into alumina accompanied the end of the all anodic process.

To obtain the insight into the chemical composition of tantala nanodots, XPS depth profiling analysis was performed for the anodized samples after selectively removing the exposed PAA by mixed chromic and phosphoric acid for 10 min. The bombardment of Ar⁺ sputtering was used to examine the inner composition. Each sputter cycle lasted 5 min, which was estimated to remove a 10-nm film of dense anodic tantalum oxide. The measured spectra were calibrated using a Shirley background subtraction. Binding energies were referenced to the carbon-hydrogen peak at 285.0 eV. The Ta⁵⁺ 4f spectrum consists of two doublet peaks (4f^7/2 and 4f^5/2) with binding energies at 26.5 eV and 28.4 eV respectively while Ta⁰ 4f spectrum consists of two doublet peaks (4f^7/2 and 4f^5/2) with binding energies at 21.3 eV and 23.1 eV respectively. Figure 2.7 shows the measured spectrum of the Ta 4f which consists of two doublet peaks (4f7/2 and 4f5/2) have chemical shift relative to the Ta⁵⁺

state at the handbook of XPS. In the beginning of Ar ion sputtering, the shapes of the Ta 4f doublets gave us an evidence for a mere presence of lower valency oxides in the

film composition. It appeared mostly Ta₂O₅ at the surface of nanodot in anodization

[16]. The intensity of the tantalum pentoxide doublet decreases with increasing sputtering time. The shift of the Ta 4f peaks toward to lower valency tantalum oxide with sputtering time increased suggests that there may exist different oxidation states of Ta in the depth of the nanodot in despite of the most stable Ta₂O₅ oxide in anodic tantalum oxidation.⁶⁴ This implies the transition from stoichiometric Ta2O5 at the surface to metallic tantalum with the coexistence of different oxidation states of Ta.

One assumption of various tantalum oxides is that the oxidation occurred too rapidly to form the most stable oxide state. Supposing that unoxidised tantalum does not remain in the hillock composition, it is likely that the presence of metallic tantalum double peak even at the surface of the unsputtered specimen is consistent with metallic tantalum gaps around the hillocks. Yet, the support by Surganov et al. ⁶⁵ indicated that the presence of lower valency tantalum oxides such as TaO, TaO₂ and Ta2O3 in the composition of oxide columns formed in similar work. In other words, the top of the nanodot are Ta₂O₅, and the rest of the nanodot is aluminum-free and composed of tantalum pentoxide mostly and comprises other oxidation states of Ta presumably; the degree of tantalum oxidation decreases from the tops of the nanodot towards the interface.