Monte Carlo Simulations of Atomic Layer Deposition on 3D large surface area structures

May 18, 2016, 2:45 PM
8h 15m
Pand

Pand

Speakers

Christophe DETAVERNIER (Ghent University) Filip GEENEN (UGent)Dr Jolien DENDOOVEN (UGent) Véronique CREMERS (Ghent University)

Description

Atomic Layer Deposition (ALD) is a technique for the deposition of uniform thin films with a thickness control on the atomic scale. Due to the self-limited nature of the surface reactions, it is possible to grow uniform thin films with an excellent conformality. Therefore ALD has become a key method for coating and functionalizing 3D large surface area structures such as anodized alumina (AAO), silicon pillars, nanowires and carbon nanotubes. Those structures are of high interest for different applications in electronic devices such as fuel cells, batteries, dye-sensitized solar cells and photocatalytic surfaces. These large surface area substrates often consist of arrays of quasi-1D holes (into which the precursor gas needs to penetrate, e.g. for AAO), or ‘forests’ of pillars (where the precursor gas can reach the surface through the empty 3D space surrounding the pillars). Using a full 3D Monte Carlo model, we compared deposition onto an array of holes versus a forest of pillars. During a simulated TMA cycle, we calculated the TMA exposure dose that is required to achieve ~90% coverage as a function of the height H and the center-to-center distance D for both hole-type (Fig. 1, top) and pillar-type features (Fig. 1, bottom). The required precursor exposure is indicated as a logarithmic color scale. The geometry is expressed in dimensionless numbers H/d and D/d, by taking the ratio of the height H and array spacing D to the diameter d of the hole/pillar. The color plots clearly indicate that the required exposure for an array of holes is determined by the aspect ratio H/d of the individual holes, and is independent of their spacing D in the array (as expected). For the pillars, the required exposure dose increases with decreasing inter-pillar spacing D and converges in the limit of very small D to the exposure of an array of holes (as the remaining space in between nearly touching pillars starts to act as 1D hole-like structures). The black lines in Fig. 1 indicate structures that share the same surface area enhancement factor (SAE). When targeting a specific SAE (e.g. 20x), the pillar geometry is clearly much more ALD friendly, in particular for tall features (large H/d value, e.g. 80) that can then be spaced reasonably far from each other (e.g. D/d ~ 4). Fig. 2 shows a cross section through the color plots in Fig. 1 at D/d = 3, and clearly illustrates that ALD of pillar-type arrays requires 10-30x less exposure dose as compared to hole-type arrays.

Primary author

Véronique CREMERS (Ghent University)

Co-authors

Presentation materials