Comparison of saturator designs for delivery of low-volatility liquid precursors

• Published in: Journal of crystal growth Vol. 607; p. 127102
• Main Authors: Maslar, James E., Kimes, William A., Khromchenko, Vladimir B., Sperling, Brent A., Kanjolia, Ravindra K.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-04-2023
• Subjects: A1. Precursor transport modeling; A1. Precursor vapor delivery; A1. Precursor vessel design; A3. Chemical vapor deposition processes; B1. Precursors; A1. Precursor vapor delivery; A3. Chemical vapor deposition processes; A1. Precursor transport modeling; A1. Precursor vessel design; B1. Precursors
• ISSN: 0022-0248; 1873-5002
• DOI: 10.1016/j.jcrysgro.2023.127102

•Saturator designs were compared for delivery of low-volatility precursors.•The performance of a bubbler and flow over vessel were characterized.•Mass carryover from each vessel was measured with a non-dispersive gas analyzer.•The impact of various process conditions on mass carryover was examined and modeled.•The bubbler delivered more precursor more efficiently than the flow over vessel.
Numerous low-volatility precursors are utilized in chemical vapor deposition and atomic layer deposition processes. Such precursors are often delivered from one of two common saturator designs: a bubbler or a flow over vessel. Previous reports concerning precursor delivery from such vessels have focused primarily on continuous delivery of moderate to high volatility liquids and solids. Few reports have focused on cyclical delivery of low volatility precursors at reduced pressures. This lack of knowledge concerning such processes can be a hindrance to efficient selection of deposition conditions and vessel design. The objective of this investigation was to compare the performance of these two saturator designs for pulsed injection at reduced pressures using the low volatility liquid precursor μ2-η2-(tBu-acetylene) dicobalthexacarbonyl (CCTBA). The basis of this comparison was the measurement of CCTBA mass carryover per injection as a function of injection number, injection time, carrier gas flow rate, system pressure, and vessel idle time. The mass carryover was determined from absorbance measurements performed using a non-dispersive infrared gas analyzer. The measured mass carryover for both vessels was compared to the theoretical mass carryover determined using a simple analytical model based upon the “bubbler equation”. In the case of the bubbler, this model described the vessel performance well with knowledge of the precursor vapor pressure and vessel head space pressure. In the case of the flow over vessel, this model described the overall vessel performance poorly unless an additional vessel efficiency factor was included, a factor that is difficult to predict a priori. Furthermore, the efficiency factor was not necessarily constant for a series of injections: the efficiency factor tended to decrease from the first injection until a stable value was achieved, a value that depended on the process conditions. This limitation of the model was attributed to the specific flow dynamics associated with the flow over vessel design. Computational fluid dynamics simulations were able to reproduce the mass carryover of the flow over vessel, after estimating the CCTBA-carrier gas binary diffusion coefficient. These simulations also showed that a larger binary diffusion coefficient and a higher vapor pressure both led to an increase in mass carryover but vessel efficiency could not equal that of the bubbler. While these results were obtained with CCTBA, the general relationships between mass carryover and the various process parameters in these saturators are expected to be similar for other low-volatility precursors.