Importance of Monitoring Slurry and Ultrapure Chemical Flow in CMP Applications

Transcription

1 Importance of Monitoring Slurry and Ultrapure Chemical Flow in CMP Applications Budge Johl, DOW Electronic Materials 2011 Levitronix Users Group Conference May 11, 2011

2 Introduction As device line widths continue to shrink process requirements are becoming more stringent Flow consistency of CMP slurries Chemical flow in post-cmp cleaning applications Un-optimized liquid flow monitoring can result in higher cost-ofownership (CoO) Monitoring and control of CMP slurry and ultrapure chemical flow is critical for CMP process consistency This led to the motive behind evaluating non-invasive ultrasonic flow monitoring technology for CMP process applications

3 Flow Meter Technologies There are a number of technologies available for monitoring flow including: Paddle wheel Turbine Rotameter Coriolis Vortex Differential pressure Magnetic and ultrasonic

4 Flow Meter Technologies: Limitations Paddle wheel, rotameter and turbine flow meters can lead to particle generation and increased pressure drops Abrasive particles in CMP slurries can effect the performance of paddle wheel, turbine, differential pressure, and vortex technologies over time Coriolis technology and certain ultrasonic devices may be sensitive to presence of bubbles in the process fluid In respect to the above issues the new ultrasonic technology was chosen for evaluation

5 Considerations for Slurry Flow Distribution and Health Metrology Calculating slurry turnovers and potential shear exposure to slurry during its distribution is important for monitoring slurry health Fluid mechanics Flow Instrumental in calculating Reynolds # Equation: Re = ρvd/µ where: ρ (rho) = fluid density v = mean fluid velocity D = pipe diameter µ (mu) = fluid viscosity Flow regimes: Laminar vs. Turbulent; Transitional Note: The Transitional and High Reynolds may produce flow instabilities with chaotic eddies and vortices, resulting in excessive shearing/damage to slurry Slurry Information Necessary for Optimum Slurry Distribution System Designs

6 SDS and Daytank on the Same or Lower Floors than Polisher Acme Polisher Filter Module Instrumentation Backpressure Device Dosing Filter Module Humidifier Supply Drum P4400 AR200 Diagram Courtesy of BOC Edwards

7 SDS and Daytank on the Top Floor Instrumentation Filter Module Humidifier Supply Drum P4400 Daytank Valve Boxes Polishers Acme Ajax Zeus Diagram Courtesy of BOC Edwards

8 ILD Slurry Flow Rate Response: Effects on TEOS RR Many factors come into play on RR Response Including: Down force, platen speed and slurry flow rate

9 % Cu RR Delta from Target Flow Slurry Flow Rate Effects on Cu Removal Rate Improper slurry flows may lead to poor wafer performance Including inconsistent defectivity, removal rate and planarization psi 2.5 psi Slurry Flow Rate (ml / min) Figure 1. Change in Removal Rate with Varying Slurry Flow

10 Flow (ml / min) Delta Pressure (psi) Slurry Flow Monitoring: Filter Lifetime Effects Slurry flow monitoring can be useful in monitoring filter life at various locations in the fab including incoming, global loop and point of use Flow (ml / min) dp (psi) 5 per. Mov. Avg. (dp (psi)) :56:07 9:58:59 10:01:54 10:04:49 10:07:44 10:10:39 10:13:34 Time Figure 2. Slurry Filter Life Monitoring (Correct Filter Change Timing)

11 Ultrasonic Flow Meter Technologies Experimental Data The non-invasive ultrasonic unit in this study was exposed to bubbles and shown to detect and compensate for bubbles in the slurry stream. High and low flow alarms can be set to warn the operator of out of control conditions. The slurry was exposed to > 1,000 turnovers with the unit placed inline in the slurry global loop with no adverse effects on removal rate films or defectivity as shown in Figures 3 & 4. Images Courtesy of Levitronix

12 Removal Rate (Normalized) Ultrasonic Flow Meter Removal Rate Data Control (0 Turnovers) Inline Ultrasonic Flow Meter + (> 1,000 Turnovers) Cu RR TaN RR TE RR Coral RR Films Figure 3. Effects of Inline Flow Meter on RR of Various Films

13 Cu Scratch Defects (Normalized) Ultrasonic Flow Meter Cu Scratch Defectivity Data 1.5 Defects Control (0 Turnovers) Inline Ultrasonic Flow Meter + (> 1,000 Turnovers) Slurry Figure 4. Effects of Inline Flow Meter on Cu Defectivity

14 Cumulative Counts > Diameter Ultrasonic Flow Meter Cumulative Particle Counts Distribution Inline Ultasonic Flow Meter + (> 1,000 Turnovers) Control (0 Turnovers) Note: Decrease in particle size distribution due to MagLev pump as shown in previous studies Size (µ) Figure 5. Effects of Inline Flow Meter on Large Particle Size Distribution

15 Dynamic Flow Control

16 Flow Control BPS-3 MLC or P1 AOD Discharge Dampener 25 Foot Long PFA Tubing Coil Pump NT Flowmeter DI Water Pinch Valve Pr. Regulator BPS-1 MLC Pump Chiller Distribution Loop Filter In Centrifugal Pump Test Only Supply Tank Tool Dispense NT Flowmeter POU Filter Schematic of Slurry Recirculation Loop Test Set-Up Schematic and Figures (6 8) Courtesy: Precision Flow Control paper by Singh et al., CMP-MIC 2007

17 Pr. Drop and Flow Rate Pr. Drop and Flow Rate Slurry Filter Pressure-Drop and Flow Rate Data Number of Slurry Turnovers Number of Slurry Turnovers Pressure Drop, psi Flow Rate, Lpm Pressure Drop, psi Flow Rate, Lpm Figure 6. Filter pressure drop and flow rate data for slurry handling in MLC pump at 7,600 rpm, ~46 turnovers/hour and 5.3 Lpm (backpressure ~31 psi). Slurry recirculated with 1 µ filter. Figure 7. Filter pressure drop and flow rate data for slurry handling in diaphragm pump at ~46 turnovers/hour and ~5.3 Lpm (backpressure ~32 psi). Slurry recirculated with 1 µ filter.

18 Slurry Dispense Flow Data using MLC Pump and Differential Pressure Flowmeter Control System Ratio of Measured Flow Rate to Set-Point Flow Rate Measured Flow Rate, ml/min Qm/Qs y = x R 2 = Set-Point Flow Rate, ml/min Set-Point Flow Rate, ml/min Measured Flow, ml/min Linear (Measured Flow, ml/min) Figure 8. Measured slip stream dispense flow rate and set-point flow rate data for slurry handling in MLC pump at 7,600 rpm, ~46 turnovers/hour and 5.3 Lpm (backpressure ~31 psi) in the global loop. Slurry recirculated with 1 micron nominal rating filter in the global loop.

19 Slurry Flow

20 Slurry Flow

21 Peristaltic Pumps & Slurry Flow

22 Peristaltic Pumps & Slurry Flow

23 Post-CMP Cleaning Post Water Track Cleaning Chemical Filtration DI Filtration CMP Recirculated DI Filtration Post CMP Cleaning Next-generation cleaning applications require more accurate control of ultrapure Post CMP cleaning chemistries Chemical flow monitoring and control is required for process consistency

24 Post CMP Clean

25 Post CMP Clean

26 Post CMP Clean

27 Post CMP Clean

28 Summary No increase in large particle size distribution was observed with the inline ultrasonic flow meter used in this study Other benefits: Easily monitor slurry and chemical flow rates Convenient to monitor filter life Improve wafer process consistency Help prevent issues with alarm capability Post CMP cleaning improvement by consistent chemical flow

29 Summary (continued) The ultrasonic flow measurement technology evaluated in this study showed: Good bubble tolerance with bubble compensation Wide flow range capability Software with real-time charting for easy flow data review Dynamic flow control possible for consistent POU dispense The DOW Electronic Materials copper slurry used in this study was compatible with the fluid dynamic design of this non-invasive ultrasonic technology

30 Future Trends (Conventional vs. Next-Generation) Conventional POU dispense Peristaltic pump system for slurry dispense to the CMP tool Particle contamination from pump tubing Pump frequent maintenance, and reduced uptime Flow variations with tubing wear, feed pressure and time Next-Generation POU dispense Advanced flowmeter technology coupled with: Magnetically Levitated Centrifugal (MLC) pumps Liquid Flow Controllers (LFC) for consistent liquid flow dispense

31 Acknowledgements The authors would like to acknowledge Dr. Arun Reddy and Ken Prygon of DOW Electronic Materials for their contributions to this study Levitronix for providing the LeviFlow ultrasonic flow meters for this study Liquid Flow is Important

32 Avoid a Spike in the Flow I told you there was a Spike in the flow! Cartoon Courtesy of Chemical Processing Weekly

33 References 1. B. Johl, R. Singh, Optimum Process Performance Through Better CMP Slurry Management, Solid State Technology, August B. Johl, Slurry Handling, Troubleshooting and Filtration, Levitronix CMP Users Conference, Santa Clara, CA, February J.P. Bare, B. Johl, Comparison of Humidified vs. Non-humidified Vacuum Pressure CMP Slurry Distribution Systems, Proc. VMIC Conference, J. Bare, B. Johl, T. Lemke, Comparison of Vacuum-Pressure vs. Pump Dispense Engines for CMP Slurry Distribution, Semi Workshop, July 1998 and Semiconductor International, San Francisco, CA, January B. Johl, H. Porter, Investigation of Valve Effects on Wafer Defectivity using an Oxide Slurry, Levitronix CMP Users Conference, Santa Clara, CA, February B. Johl, M. Litchy, R. Schoeb, Effect of a Maglev Centrifugal Pump on Slurry Health and Defect Rates, PacRim Int. Conference, Korea, G.Vasilopoulos, Z. Lin, B. Johl, S. Joshi and B. Chatterjee Copper CMP Defect Reduction Using POU Filtration, Semicon West Technical Program, San Francisco, CA, July J.P. Bare, T. Lemke, Parameters for Monitoring CMP Slurry Stability and Contamination, Micro Magazine, September B. Johl and M. Tseng, Impacts of Point-of-Use Filtration on an ILD CMP Process, 21st International VMIC Conference, Hawaii, October 2004.