Nikolaos Cheimarios

Enabling a commercial computational fluid dynamics code to perform certain nonlinear analysis tasks

Abstract In this work we enable the commercial computational fluid dynamics code Fluent, to successfully trace a complete solution branch, even past turning points. Here the so-called Recursive Projection Method (RPM) is implemented as a computational shell “wrapped” around Fluent, in conjunction with a pseudo-arc-length method for convergence on the unstable branch. The case study is a mixed convection flow in a stagnation point chemical vapor deposition (CVD) reactor. Multiple steady states coexist over a range of inlet Reynolds numbers, due to the competition of the two dominant physical mechanisms: forced and free convection. Continuation on the solution branch reveals a curve consisting of a stable branch, dominated by free convection, followed, past the first turning point, by an unstable branch. Past a second turning point, follows another stable branch dominated by forced convection. Taking the problem a step further, it is augmented with a chemical model describing the deposition of silicon (Si) from silane (SiH 4 ), silylene (SiH 2 ) and hydrogen (H 2 ). The solution branch does not alter since the gas mixture is dilute and the carrier gas, in this case nitrogen (N 2 ), and the precursor, in this case SiH 4 , are of similar molar masses; the concentration differences cannot lead to solutal convection. Results for the mass fraction distribution inside the reactor and the film growth rates are reported in all parts of the solution branch.

A multi-parallel multiscale computational framework for chemical vapor deposition processes

Abstract A multiscale computational framework for coupling the multiple length scales in chemical vapor deposition (CVD) processes is implemented for studying the effect of the prevailing conditions inside a CVD reactor (macro-scale) on the film growth on a wafer with predefined topography (micro-scale). A multi-parallel method is proposed for accelerating the computations. It combines domain decomposition methods for the macro-scale (reactor scale) model, which is based on partial differential equations (PDEs), and a synchronous master-worker scheme for the parallel computation of the boundary conditions (BCs) for the PDEs; BCs are coming from the micro-scale model describing film growth on the predefined topography.

Multiscale modeling of chemical vapor deposition of silicon

Abstract A multiscale computational study of chemical vapor deposition (CVD) of silicon (Si) from silane (SiH4) is presented. The macro-scale of the bulk of a CVD reactor is coupled with the micro-scale of a predefined topography consisting of trenches on the wafer surface through a multiscale modeling framework. The coupling is implemented by the correction of the boundary condition for the consumption of species on the wafer which takes into account the existence of the micro-topography. The effect of the wafer temperature on the Si film uniformity inside the trenches as well as the effect of the micro-topography on the species consumption on the wafer are studied.