Over the past decades, nickel mono-silicide NiSi has become one of the preferred materials to enhance contacting in silicon-based semiconductor devices. However, the formation of a useable NiSi contacting layer is challenged by the continued miniaturization in the micro-electronics industry. As the silicide thickness is reduced, it is more likely to agglomerate [D. Deduytssche et al., J. Appl. Phys. 98, 033526 (2005)]. Therefore it is crucial to stabilize silicide films, either by altering the microstructure of the NiSi phase or by exploring the use of alternative phases that are more resistant to agglomeration. However, De Keyser et al. [Appl. Phys. Letters. 96, 173503 (2010)] reported that below a critical thickness of 5 nm, annealing of ultrathin Ni films results in surprisingly stable silicide layers.

In this work, we systematically investigated (1) the influence of Ni thickness on the formation of nickel silicides in the range between 0 and 15 nm, and (2) the influence of alloying elements (e.g. Al, Co, Pd, Ge and Pt) on the phase formation sequence and on the critical thickness phenomenon. Research on such ultra-thin films is very challenging using conventional lab-based XRD techniques. Therefore, the silicidation process was investigated in detail through XRD-based pole figure measurements at the DiffAbs beamline of the SOLEIL synchrotron. These measurements allow unambiguous phase identification, even if the formed silicide would exhibit a preferential orientation to the monocrystalline substrate. Furthermore, these synchrotron based pole figure measurements provided detailed texture information that helps to understand the observed shift in critical thickness and the morphological stability of the ultra-thin silicide layers.

For an as deposited layer thickness above the critical thickness, we observed the sequential formation of Ni -> θ-Ni2Si (epitaxial) -> δ-Ni2Si -> NiSi, which agglomerates. Samples in the ultrathin regime (i.e. below the critical thickness, e.g. for 3nm pure Ni) instead exhibit a formation sequence identified as Ni -> θ-Ni2Si (epitaxial) -> NiSi2 (epitaxial). This finally formed NiSi2 is remarkably stable and does not suffer from agglomeration.

The phase sequence changes at a critical thickness for pure Ni films around 5nm as-deposited thickness. Through alloying of the initial Ni layer, we could influence this critical thickness to higher (Al) and lower (Ge, Pd, Pt) values. Our pole figure measurements also showed that the interface alignment of these silicide grains with their surroundings, and the energetic cost of creating such an interface, plays a crucial role during the phase formation.

In summary, we report that the phase formation sequence during the Ni salicidation process severely changes at a critical thickness of 5nm. Pole figure measurements show that the thicker films form intermediate Ni-rich phases which subsequently form agglomerating NiSi. Thinner films, however, show a simplified formation sequence which results in the formation of epitaxial, non-agglomerating NiSi2. Secondly, we demonstrate that this critical thickness can be influenced to higher or lower values through the addition of alloying elements, an effect that can be understood through the effect of mixing entropy. Furthermore, crystallographic alignment matching at the interface is a crucial parameter controlling the phase formation kinetics in these sub-10nm thick films.