Mechanism for LPC Pooling and Release in CMP Slurry Lines

Transcription

1 Mechanism for LPC Pooling and Release in CMP Slurry Lines Kelly A. Barry Product Manager Michael A. Fury, Ph.D. Director of Market Development Vantage Technology Corporation, 1731 Dell Avenue, Campbell, CA USA NCCAVS CMPUG March 19,

2 Outline Experimental setup Observations Conclusions & Recommendations 2

3 Premise Vantage performed a simple experiment using fluorescent latex spheres to help visualize particle behavior in slurry lines The data yielded some unexpected results Our hypothesis has implications for the design of slurry loops, SDS systems, and CMP polishers 3

4 SlurryScope Field Observations 24 Particles/mL >1um Minutes 1 Day 12:00:00 12:28:48 12:57:36 Time 13:26:24 13:55:12 14:24:00 11,000 Flow Rate (ml/min) 9,000 7,000 5,000 2,500 2,000 1,500 1, Minutes 4 Hours 12:14:51 12:16:11 12:17:31 12:18:52 12:20:12 12:21:32 12:22:52 12:24:12 12:25:32 12:26:53 12:28:13 12:29:33 12:30:53 12:32:13 12:33:33 12:34:54 12:36:14 12:37:34 12:38:54 12:40:14 12:41:34 12:42:55 12:44:15 12:45:35 12:46:55 12:48:15 4

5 Vantage Lab Observation Total Particle Count >0.8µm Flow Rate ml/min Periodic LPC spikes not caused by change in flow rate 5

6 Fluorescent Particle Experiment 3 inches Low point 4 feet Test loop: 20 of ¼ tubing all circulating at 15 ml/min P SlurryScope IN OUT Low point 6

7 UV Fluorescent Latex Particles UV fluorescent latex microspheres, 1-5µm Glow bright green under UV light Sonicated for 6 min Recirculated in pump loop for SlurryScope measurement 7

8 Particle Count and Size: 0h-24h Total Particle Count TPC > 1.2um gradually decreases, indicating settling of larger particles TPC um gradually increases, possibly indicating latex particles being broken up in the pump 8

9 Particle Count and Size: 0h-56h Flow Rate Total Particle Count Two events of periodic large LPC spikes not (all) due to flow rate Flow disruptions are thought to be air bubbles moving through the LFC 9

10 Latex Settling in Tubes Latex particles stuck at top of pump output tube Some latex particles settled at low point in this tube Many latex particles settled at low point of tube at input to pump Flow direction Slurry particles are much more dense than latex 10

11 Pump Input 3/8 Tubing Settling occurs at input point (horizontal) 11

12 Dead Leg Settling: 1/8 and 1/4 Stagnant 1/8 leg (latex) shows settling as line becomes horizontal Active uphill ¼ legs (latex) do not show settling Stagnant ¼ leg (latex) shows settling Stagnant ¼ leg (DIW only) 12

13 Connector Settling Settling in open 1/8 valve Settling in active ¼ connector Settling in stagnant leg of active ¼ connector 13

14 Low Point Settling Settling at low point of active ¼ line 14

15 Horizontal Line Settling Settling in active ¼ line occurs as the orientation changes from vertical to horizontal Settling in active ¼ line is more spread out at the low point of the line 15

16 Vertical Line Accumulation Latex clustering was also observed along vertical orientations may indicate occlusions or surface abnormalities inside the line 16

17 Vantage Lab Observation Total Particle Count >0.8µm Flow Rate ml/min Periodic LPC spikes with no change in flow rate 17

18 Proposed Mechanism Particles accumulate in a zone with impaired flow until that zone reaches full capacity A portion of the accumulated particles are swept back into the slurry flow Accumulation cycle begins again Event that triggers the start of a cycle may be physical, such as an air bubble or a line bump End of a cycle may be a return to more uniform flow, or a sustainable metastable condition 18

19 Bottle With Fluorescent Latex After 24 hours loop recirculation at 15 ml/min, there is no latex left in the bottle except for a small amount trapped in the filter. This bottle was bright green when experiment started. Fluid level ~200 ml in 500 ml bottle 50µm filter paper in housing for large debris 19

20 Particle Count and Size: 0h-56h Flow Rate Total Particle Count No more particles in source bottle after 24 hours 20

21 Implications for Plumbing Design Regions of imperfect slurry flow collect particles and become a source for random or periodic LPC release Low linear velocity allows particles to accumulate even in horizontal runs Surface imperfections in tubing promote particle accumulation even in vertical runs Flow diameter changes create particle accumulation zones 21

22 Implications for SlurryScope Tubing diameter, length & orientation are significant LPC issues at 15 ml/min Large particles don t go uphill readily at 15 ml/min Large particles settle readily in low spots at 15 ml/min Lost LPC = missed slurry particle events & inaccurate data Tap-off line to SlurryScope must be kept short (<1m), horizontal or downhill and 1/8 to best capture LPC events accurately 22

23 Conclusions & Recommendations Periodic LPC spikes may be the result of regions of imperfect slurry flow somewhere in the system Continuous slurry monitoring facilitates detection and elimination of the root causes Linear slurry flow velocity below the recommended minimum has LPC consequences in the SDS, in the polisher, and in the metrology Locating the SlurryScope as close as possible to the main slurry line tap-off point is critical for accurate LPC monitoring 23