Investigation of New Maglev Day-Tank Technology for Effective Slurry Dispersion and Distribution

Transcription

1 Investigation of New Maglev Day-Tank Technology for Effective Slurry Dispersion and Distribution Budge Johl and Hethel Porter th Annual Levitronix CMPUG Symposium Rohm and Haas Electronic Materials CMP Technologies Phoenix, AZ

2 Slurry Delivery and Recirculation Slurry recirculation in fab day-tanks is typically performed by pumps In respect to slurry shear and health all pumps are not created equal Technology has come a long way since the below graph published back in July , 2 Vacuum-Pressure Technology Diaphragm Pump Technology Bellows Pump Technology

3 Additional Published Data Maglev technology showed the best large particle distribution over time with fumed silica slurry 3

4 Historical Large Particle Size Results Diaphragm Pump Large Particle Size Distribution Cumulative Counts > Diameter Time 0 T1Hr-24TO T 2Hr-48TO T3Hr-72TO T 5Hr-120TO Recirculation mode only Particle Size (µ)

5 Historical Large Particle Size Results Large Particle Size Response of Oxide Slurry (MagLev Pump) Cumulative Counts > Diameter 100 T0hrs T336hrs (1700 TO) Particle Diameter (microns) Recirculation mode only 14

6 Historical Response Defectivity and RR Response Oxide Slurry STANDARD (Time 0) LEVITRONIX (Time 336 Hour) = 1,700 Turnovers Normalized Values Recirculation mode only 14 Defect Counts Wafer Results RR

7 Slurry Delivery and Recirculation Since the early pump evaluation back in 1998, new improved diaphragm pumps have been manufactured that exhibit less slurry shear. Improved bellow designs to reduce slurry residence time and shear have been designed and built. Field audits have shown slurry issues from various types of pumps that were continuously recirculating slurry in day-tanks. Past studies with Maglev technology showed no slurry shear effects. Data showed a decrease in overall large particle size distribution coupled with improved wafer results. This leads to the motive behind testing the Maglev day-tank technology.

8 Background Objective of Study: Test efficacy of new Maglev day-tank for effective slurry dispersion and distribution. Recirculated slurry for 24- hours and polished wafers. Slurry was drawn to polisher directly from day-tank. System was configured with a chiller for this study in order to maintain temperature of C. Polishing and analyticals: Wafer defectivity analysis and full slurry analyticals were performed compared to the control. Second part of the study used an alumina slurry to determine efficacy of slurry re-dispersion.

9 MagLev Day-Tank Slurry Recirculation In Tank Mixing

10 Pump Flange Overview Mix Holes* *The quantity, size and configuration of the mix holes is still under review

11 MagLev Day-Tank Overall View of Maglev Day-Tank Maglev Pump Orientation

12 Temperature Results Maglev Day-Tank Temperature Evaluation No Loop Circulation ( C) W/ Loop Circulation ( C) W/ Loop Circulation + Chiller (24 C) Temperature (C) Turned on Chiller Hours

13 Oxide Slurry 6000 rpm Study Day-tank volume 20-gallons Slurry was static for 24-hours before re-dispersion 24-Hour Shutdown followed by Maglev Re-dispersion of Oxide 6000 rpm 28.0 Top Middle Bottom 29.0 % Solids T-0 T-5 T-10 T-15 T-20 T-25 T-30 Time (minutes) Complete re-dispersion occurred between 5 and rpm (15 L/min flow)

14 Oxide Slurry Shutdown Study Day-tank volume 20-gallons Shutdown after 30-minute re-dispersion 28.0 Settling Behavior of Oxide Slurry after 30-minute Re-dispersion in Maglev Tank Top Middle Bottom 29.0 % Solids T-0 T-5 T-10 T-15 T-20 T-25 T-30 Time (minutes) No settling occurred over 30-minutes

15 Alumina Slurry 6000 rpm Study Day-tank volume 20-gallons Slurry was static for 24-hours before re-dispersion Redispersion w/o Circulation of Alumina Slurry MagLev 6000 rpm Top Middle Bottom % Solids T-0 T-20 T-40 T-60 T-80 T-100 T-120 T-140 T-160 T-180 T-300 T-1440 Time (minutes) Total homogeneity was not achieved even after 24-hour of 6000 rpm (15 L/min Flow)

16 Alumina Slurry 8000 rpm Study Day-tank volume 20-gallons Slurry was static for 24-hours before re-dispersion Redispersion w/o Circulation of Alumina Slurry MagLev 8000 rpm 5.0 Top Middle Bottom 9.0 % Solids T-0 T-20 T-40 T-60 Time (minutes) Complete re-dispersion occurred between 50 and rpm 21 L/min internal tank flow

17 Alumina Slurry Re-dispersion with Circulation Study Day-tank volume 20-gallons Slurry was static for 24 hours before re-dispersion Recirculation flow rate: 14 L/min In tank flow rate: 15 L/min Total Percent Solids Hour Shutdown of Alumina Slurry followed by MagLev Redispersion + Recirculation Top Middle Bottom 25.0 T-0 T-5 T-10 T-15 T-20 T-25 T-30 Time (Minutes) Complete re-dispersion occurred at ~ RPM

18 Alumina Slurry Shutdown Study Day-tank volume 20-gallons Total Percent Solids Minutes Shutdown after 30-Minutes of Mixing + Recirculation Alumina Slurry (Maglev Day-Tank) Top Middle Bottom T-0 T-5 T-10 T-15 T-20 T-25 T-30 Time (Minutes) Settling occurred after shutdown

19 Analytical Lab Testing Laboratory Analysis Performed: Large & Mean Particle Size Zeta Potential Conductivity ph Viscosity Density Total Percent Solids

20 Analytical Results Day-tank Volume of Klebosol slurry = 20 gallons ~300 turnovers / 6000 RPM Oxide Slurry Analytical Analysis Control Day-Tank Units Mean Particle Size nm ph Viscosity cps Density g/cm3 Total Solids % Zeta Potential mv Conductivity ms/cm No significant difference between the control and Maglev day-tank observed

21 Large Particle Size Results 1000 Particle Size Response of Oxide Slurry Control Vs. Maglev Tank Control MagLev Day Tank Cumulative Counts > Diameter ` Particle Size (µ) Maglev tank mixing plus recirculation mode (6000 RPM) 15L/min tank + 14 L/min loop

22 Polishing Polishing Tool Used: Applied Materials Mirra Wafer Type: TEOS Blanket Wafers Slurries: Klebosol Pads: MSW1500 (Re-dispersion Study Only) IC1010 Politex Suba Conditioner: DiaGrid Klebosol is a trademark of AZ Electronic Materials IC1010, Suba, and Politex are tradmarks of Rohm and Haas Company or its affiliates, DiaGrid is a registered trademark of Kinik Co. Mirra is a Registered Trademark of Applied Materials, Inc.

23 Polishing Results (Removal Rate) One-way Analysis of Removal Rate By Process Type Normalized RR (A) Control Process Type Day Tank Each Pair Student's t 0.05 Level Number Mean Std Dev Std Err Mean Lower 95% Upper 95% Control Day Tank No significant difference between the control and Maglev day-tank on removal rate

24 Polishing Results (WIWNU) One-way Analysis of WIWNU (%) By Process Type 2 Normalized WIWNU (%) Control Process Type Day Tank Each Pair Student's t 0.05 Level Number Mean Std Dev Std Err Mean Lower 95% Upper 95% Control Day Tank No significant difference between the control and Maglev day-tank on WIWNU

25 Polishing Results (Defectivity) One-way Analysis of Total Defect Count By Process Type Normalized Total Defect Count Control Process Type Day Tank Each Pair Student's t 0.05 Level Number Mean Std Dev Std Err Mean Lower 95% Upper 95% Control Day Tank No significant difference between the control and Maglev day-tank on defectivity

26 Defect Map Control Wafers Maglev Wafers

27 Summary Klebosol 1501 oxide slurry showed no changes in overall normal health parameters after ~300 day-tank turnovers which is ~3X higher than typical fab usage. No settling or thickening was observed at any point in the loop or in the tank. There was no increase in large particle size distribution indicating no particle agglomeration. An actual decrease in the large particle size distribution was observed as shown by the historical data. Additional analytical results showed no significant difference. Statistical analysis showed no significant difference on polishing results (removal rate, WIWNU, and defectivity). Re-dispersion of MSW 1500 alumina slurry was proven to be possible.

28 Future Maglev Day-Tank Improvements Future improvements / modifications include: Increased diameter and / or quantity of mix holes Increased mixing flow rate and mixing action Reduction in turnover time and possible addition of conical face plate for improved draining Alternative designs for smaller and larger pump assemblies (BPS1 & BPS4)

29 Acknowledgements The authors would like to thank the following people for their contributions to this study: Adrian Quantick and Dieter Hudobnik of Levitronix Jimmy Ehler, Tony Brown and Karen Winkelman of Rohm and Haas Electronic Materials, CMP Technologies

30 References [1] 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, January [2] J. Bare, B. Johl, T. Lemke, CMP Slurry Distribution: Vacuum-Pressure vs. Pump Dispense NCCAVS, October, [3] B. Johl, M. Litchy, R. Schoeb, Effect of a Maglev Centrifugal Pump on Slurry Health and Defect Rates PacRim Int. Conference, Korea, [4] B. Johl, R. Singh, Optimum Process Performance Through Better CMP Slurry Management, Solid State Technology, August [5] B. Johl, et al., Dynamic Pot-Life and Handling Evaluation of EPL2362 First Step Copper Slurry, Proc. 8th Int. CMP Conference, Marina Del Ray, CA, February [6] B. Johl, T. Buley, Dynamic Pot-Life and Handling Evaluation of Rodel CUS1351 Copper Barrier Slurry, Proc. 7th Int. CMP Conference, San Jose, [7] B. Johl, et al., Accelerated Aging and Handling Evaluation of Rodel CUS1201 Second Step Slurry in BOC Edwards Bulk Delivery System, Proc. VMIC Conference, [8] R. Singh, B. Johl, Characterization of a Silica Based STI CMP Slurry in a Vacuum-Pressure Dispense Slurry Delivery System and Pump Loop, Proc. VMIC Conference, [9] J.P. Bare, B. Johl, Accelerated Aging and Handling Evaluation of Rodel ILD1300 Oxide CMP Slurry, Proc. AVS N. Calif. CMPUG Annual Symposium, [10] J.P. Bare, B. Johl, Comparison of Humidified vs. Non-humidified Vacuum Pressure CMPS Slurry Distribution System, Proc. VMIC Conference, [11] M. R. Litchy, R. Schoeb, Effect of Particle Size Distribution on Filter Lifetime in Three Slurry Pump Systems, MRS Spring Conference, San Francisco, March 2005 [12] B. Orzechowski, Slurry Distribution, Levitronix CMP Users Conference, Santa Clara, CA, February [13] B. Johl, Slurry Handling, Troubleshooting and Filtration, Levitronix CMP Users Conference, Santa Clara, CA, February [14] B. Johl, H. Porter, Investigation of Valve Effects on Wafer Defectivity using an Oxide Slurry, Levitronix CMP Users Conference, Santa Clara, CA, February 2006.