Experiment-based modelling of a vapor draw ampoule used for low-volatility precursors.

Delivery of low-volatility precursors is a continuing challenge for chemical vapor deposition and atomic layer deposition processes used for microelectronics manufacturing. To aid in addressing this problem, we have recently developed an inline measurement capable of monitoring precursor delivery. Motivated by a desire to better understand the origins of what is now observable, this study uses computational fluid dynamics and a relatively simple model to simulate the delivery of pentakis(dimethylamido)tantalum (PDMAT) from a commercial vapor draw ampoule. Parameters used in the model are obtained by fitting the performance of the ampoule to a limited dataset of PDMAT delivery rates obtained experimentally using a non-dispersive infrared sensor. The model shows good agreement with a much larger experimental dataset over a range of conditions in both pulsed and continuously flowing operation. The combined approach of experiment and simulation provides a means to understand the phenomena occurring during precursor delivery both quantitatively and qualitatively.