Selective Depositions by ALD

The traditional top-down approach that allows the miniaturization of microelectronic devices leads to an exponential increase in manufacturing costs, due to the multiplication of lithography steps and alignment errors at the atomic scale.

This aggressive miniaturization is based, among other things, on the development of perfectly controlled deposition processes. With this objective, our team in collaboration with the CEA/LETI is currently developing a selective deposition process either on a specific material of a 2D substrate, or on a specific surface (horizontal or vertical) of a 3D substrate.

This original approach is based on the alternation of cycles of deposition by PE-ALD and ALE etching, without lithography and takes advantage of the non-conformity of the deposition step at the atomic scale and/or the anisotropy of the engraving stage.

Selective deposition on a 2D substrate, illustrated in the figure opposite, is based on the deactivation of one material with respect to the other, synonymous with nucleation delay. The incorporation of an etching step in the PE-ALD cycle makes it possible to reset the surface state corresponding to this deactivation, and consequently to promote selective growth on a single material of the 2D substrate. We took advantage of this process to selectively deposit Ta2O5 and TiO2 on metal, to the detriment of Si3N4 or SiO2 surfaces [R. VALLAT et al, JVSTA 2017 and JVSTA 2019].

We have also demonstrated that selective deposition can be carried out on the sides of 3D structures (via or trench) by the combination of a PEALD deposition and an anisotropic etching in a single ALD reactor, according to the following diagram [A. CHAKER et al. Appl. Phys. Lett. 114, 043101 (2019)]. For this, we have equipped our ICP PEALD reactor from Oxford Instruments with an RF polarization kit (13.56 MHz) which makes it possible to extract energetic ions (0-100 W) from the plasma while modulating their energy, and to direct them towards the sample holder. This polarization allows the realization in situ of an anisotropic ALE step in Ar+ plasma. This approach gave rise to the filing of a patent with LETI for the selective growth of spacers at advanced technological nodes (< 7 nm). On the basis of these feasibility demonstrations, our team is working on the optimization of processes for filling structures with a high aspect ratio (gap filling), defining hard masks, manufacturing spacers or even coating nanowires.

More fundamental studies are also underway to understand the nucleation mechanisms allowing precise control of growth mechanisms and kinetics at the atomic scale.