(A) study on optimization of particle removal in MOCVD chamber

다국어 초록 (Multilingual Abstract)

In the MOCVD process, particles generated from by-products of precursor gas-phase reactions and film peeling caused by plasma treatment are among the major contamination sources that lead to the degradation of semiconductor device performance and a reduction in yield. Accordingly, this study quantitatively analyzed the particle removal mechanism using purge gas in an MOCVD TiN chamber through numerical simulations and experiments to identify the optimal removal performance depending on the purge conditions.

The numerical analysis was performed using ANSYS Fluent 2024, which conducted steady-state compressible flow analysis based on the Navier–Stokes governing equations. The parameters were set to chamber pressure (1, 25, 50 Torr), purge gas flow rate (1000, 2000 sccm), the height between the heater and the showerhead (h₁–h₃ mm), and the showerhead hole geometry (three-stage convergent, convergent–divergent, and jet-type). As a result, it was found that lower pressure increased volumetric flow, strengthened the gas velocity in the chamber, and shortened the residence time, thereby enhancing particle removal performance. In contrast, as the chamber height increased, the expansion of the chamber volume led to the formation of low-velocity and recirculation regions, which in turn increased the probability of particle re-suspension. In addition, a comparison of the showerhead hole geometries showed that the jet-type nozzle maintained the longest jet core and highest outlet velocity, whereas its effect on the overall flow field was limited.

The experiments were conducted using an actual AMAT MOCVD TiN chamber, and variations in the number of particles before and after deposition under each condition were measured using a KLA particle inspection system. As a result, the 1 Torr condition exhibited the lowest average particle count and the smallest variance, indicating the most stable removal performance, while both the mean value and the error range increased with increasing pressure (25 and 50 Torr). In addition, the smallest heater-to-showerhead distance (h₁ mm) showed the lowest average particle count and the narrowest distribution range. These results indicate that a low-pressure environment combined with a smaller chamber volume leads to improved purge-based particle removal efficiency. SEM and EDS analyses confirmed that the detected particles had TiN–C composite compositions, originating from the thermal decomposition by-products of the TDMAT precursor and film fragments. Through these findings, the particle generation and retention mechanisms predicted by the numerical analysis were experimentally validated.

By systematically analyzing the effects of key design parameters—pressure, flow rate, chamber height, and nozzle geometry—on purge-based particle removal efficiency, this study provides practical guidelines that can be directly applied to both the chamber design stage and the optimization of purge conditions.

목차 (Table of Contents)

• Chapter 1. Introduction 1
• 1.1 Research background 1
• 1.2. Review of related research 6
• 1.3. Research objectives and methodology 9
• Chapter 2. Research methodology 11
• 2.1. Theoretical background and numerical methods 11
• 2.2. Modeling, meshing, and simulation conditions 16
• 2.2.1. Chamber geometry and mesh validity 16
• 2.2.2. Boundary conditions 21
• Chapter 3. Results and discussion 24
• 3.1. CFD based flow field analysis 24
• 3.1.1. Showerhead flow 24
• 3.1.2. Pressure variation 27
• 3.1.3. Chamber height variation 30
• 3.1.4. Flow rate variation 38
• 3.1.5. Showerhead nozzle geometry 42
• 3.2. Experimental Analysis of Particle Contamination and Evaluation of Particle Removal Performance in various conditions 47
• 3.2.1. Experimental configuration and particle composition analysis 47
• 3.2.2. Pressure variation 54
• 3.2.3. Chamber height variation 58
• 3.2.4. Flow rate variation 62
• Chapter 4. Conclusion 67
• Reference 71