Polycrystalline silicon (poly-Si) have been gaining the importance for thin-film transistor (TFT) applications since it has a higher carrier mobility and current carrying capacity than hydrogenated amorphous silicon (a-Si:H). Among the many different methods for producing poly-Si films, silicide mediated crystallization (SMC) of amorphous silicon offers the advantage of producing large-grained poly-Si films at temperatures low enough for glass substrates while maintaining a high throughput without the need for expensive machinery. For further applications of poly-Si, the location and orientation of grains need to be controlled as well as the grain size. These issues become more critical as grains become larger, since there are fewer grains within a channel. Control of the grain orientation in poly-Si is difficult because nucleation is a stochastic process. However, if an external template for crystallization is provided, both the orientation and location of crystal grains can be controlled. In this study, we report on the results of using cold-rolled and annealed Ni tapes as a template for crystallization of a-Si:H thin films. We find that crystallization is mediated by $NiSi_2$ formation, and that all Si grains have their ＜110＞ axis oriented the same direction within just few degrees. Furthermore, nearly all of them had the same rotation about the ＜110＞ except for the presence of twins and/or type A-B formations. Despite the use of the nickel substrate, the Ni concentration within the Si film was below the detection limit of energy-dispersive x-ray spectroscopy ($10^{19} cm^{-3}$). This low-Ni contamination level is attributed to the presence of an oxide layer between the Ni substrate and the Si film.
Amorphous silicon (a-Si) films deposited on oxidized silicon wafers were crystallized to a highly textured form using contact printing of rolled and annealed nickel tapes. Crystallization was achieved by first annealing the a-Si film in contact with Ni tape at 600 ℃ for 20 min in a flowing forming gas( 90% $N_2$, 10% $H_2$) environment, then removing the Ni tape and further annealing the a-Si film in vacuum for 2 hr at 600 ℃. At the crystallization front, A needle-like morphology which was typical of NiSi2 mediated crystallization was observed. All needles, regardless of their points of origin, belonged to the same ＜111＞ family of directions except for the presence of twins and/or type A-B formations, leading to formation of nearly single crystalline regions up to ~20 ㎛ in radius. Furthermore, the orientational relation between Si grains and Ni tape was observed to be Si＜110＞ ∥ Ni＜110＞. We have answered 3 major questions in orientation-controlled crystallization of a-Si:(1)why the entire films were not transformed to silicide but crystallized ,(2)how the crystallinity of Ni tape was transmitted from Ni tape to a-Si, and (3)what the crystallization mode and exact crystal orientations of crystallized Si were. The sequential growth of various-species silicide was observed as follows: $Ni_2Si$ ⇒ $Ni_3Si_2$ ⇒NiSi ⇒ $NiSi_2$. After this sequential growth, Si grains were nucleated on $NiSi_2$, and $NiSi_2$ particles migrated through a-Si matrix toward Si ＜111＞ directions. As the result of $NiSi_2$ migration, the interface between NiSi and crystalline Si has formed. This interface have been thermodynamically stable at 600 ℃. Because of this interface, the entire films cannot be transformed to silicide but be crystallized by $NiSi_2$. In the sequential growth of silicides, we have observed the orientation relationship among the intermediate silicide such as $Ni_2Si$ ＜001＞ ∥ $Ni_3Si_2$ $＜1\bar{1}\bar{2}＞$ ∥ NiSi ＜010＞ ∥ Si ＜111＞. The d-spacings of zone axis of each silicides were nearly identical to each other within 4 $%$. It means that the process of transmitting the Ni crystallinity to a-Si may be the sequential growth from $Ni_2Si$ to $NiSi_2$, satisfying with the lowest misfit energy between silicide. Through this sequential growth mode, the 3 crystallographic orientations of Si grains would be controlled by Ni tape. The crystal orientations of crystallized Si were observed that one of ＜110＞s was off ~ 21 ℃ from film normal and one of ＜111＞s was in the film plane. We have observed that, after facing with the interface or free surface, the crystallization needles were kept migrating with sustaining their crystal orientation. This observation was explained by that, when the crystallization needle reached the interface or free surface, it changed the direction to other ＜111＞ directions with sustaining its crystal orientation. By the successive changes of migration direction, the crystallizing needles can grow up to $~ 40 ㎛. We identify that the orientation-controlled crystallization of a-Si is ascribed to the characteristic(nearly single crystalline phase)of the cold-rolled and annealed Ni tape.
