DuPont de Nemours, Inc. · NYSE:DD
DuPont de Nemours — DuPont — Technical Presentation (February 2004)
31 slides from DuPont de Nemours's technical presentation, February 2004
Slide 1: Advances in Reducing PFC Emissions and Gas Costs: Qualification of C-C4F8 Chamber Clean Gas
ADVANCES IN REDUCING PFC EMISSIONS AND GAS COSTS: QUALIFICATION OF C-C4F8 CHAMBER CLEAN GAS IN A NOVELLUS CONCEPT-2 CHEMICAL VAPOR DEPOSITION TOOL Allen Evans, Lance Nevala, Michael Sledz – Advanced Micro Devices Charles C. Allgood - DuPont
Slide 2: Agenda
AGENDA AMD PFC Reduction Goals and plans PECVD Plasma Clean PFC reduction options DuPont Zyron® 8020 DOE Plasma Clean Optimization Process related data observations Conclusions PFC Emissions Reductions/Next Steps
Slide 3: AMD PFC Reduction Strategy
AMD PFC REDUCTION STRATEGY AMD has a goal to reduce PFC emission 50% from the 1995 baseline by 2010 The main focus for this reduction is to develop processes that produce less emissions Fab 25, located in Austin Texas, is a large portion of the total yearly emissions for AMD Fab 25 PECVD cha
Slide 4: Fab 25 PECVD PFC Strategies
FAB 25 PECVD PFC STRATEGIES A major portion of the Dielectric depositions are performed using the Novellus Concept-2 Sequels The current chamber cleaning gas is C3F8 Options for PFC reductions Attempt to further Optimize the C3F8 process Convert to NF3 as the cleaning gas Find another al
Slide 5: Fab 25 PECVD PFC Strategies: NF3 and C3F8 Optimization
FAB 25 PECVD PFC STRATEGIES NF3 Would provide the lowest possible emissions Would require $1.0M – $1.25M in equipment upgrades Would increase the cost of ownership ( gas cost and tool maintenance) C3F8 Optimization The current process was a balance of process requirements and PFC emissions
Slide 6: Fab 25 PECVD PFC Strategies: C4F8
FAB 25 PECVD PFC STRATEGIES C4F8 DuPont made a presentation to AMD in the summer of 2001 about Zyron® 8020 – C4F8 for chamber cleans The product sparked interest because of the potential for both a cost savings and a substantial PFC emissions reduction Initial testing was performed on a Concep
Slide 7: Fab 25 C4F8 Testing
FAB 25 C4F8 TESTING AMD obtained some concept and feasibility data for C4F8 from Novellus – this became the center point for the experiment The team decided to vary three variables Total Flow % O2 RF Power A Box-Behnken Design of Experiment (DOE) was chosen to obtain curvature data withi
Slide 8: DOE Parameters
C4F8 DOE PARAMETERS Exp # % O2 Total Flow(sccm) Power 1 62 2176 2500 2 77 2720 2500 3 62 2720 2000 4 62 3264 2500 5 77 3264 3000 6 77 2720 2500 7 62 2720 3000 8 92 2176 2500 9 77 2176 2000 10 77 3264 2000 11 77 2176 3000 12 77 2720 2500 13 92 2720 2000 14 92 2720 3000 15 92 3264 2500
Slide 9: Experiment Setup
EXPERIMENT SETUP The tests were performed on a single-chamber Novellus Concept 2 Sequel tool running Silane oxides DuPont attached a MIDAC FTIR into the exhaust port at the outlet of the chamber pump for data collection For each test, 25 wafers were run using the same 11,500A silane oxide proc
Slide 10: Experimental Measurements
EXPERMENTAL MEASUREMENTS The data for both PFC emissions and total clean time were obtained from the FTIR spectra The total high pressure clean time was defined as the point at which the SiF4 emissions had decreased 90% from the peak value The emissions were then calculated by integrating the
Slide 11: DOE Results
C4F8 DOE RESULTS DOE # O2 flow C4F8 flow Power Pressure Clean time kg CE 1 1349 827 2500 3.0 15.2 136.4 2 2094 626 2500 3.0 11.4 73.4 3 1686 1034 2000 3.0 15.3 214.4 4 2024 1240 2500 3.0 15.2 241.7 5 2513 751 3000 3.0 10.6 80.24 6 2094 626 2500 3.0 11.0 70.14 7 1686 1034 3000 3.0 15.2 192.6 8 2002 1
Slide 12: DOE Model
C4F8 DOE MODEL 15.83 Clean Time 11.12 9.52 255.9 Emissions 89.32143 23.727 2176 3264 2000 3000 62 92 77 2720 2500 % O2(-1,1) Total Flow(-1,1) Power(-1,1)
Slide 13: DOE SiF4 Spectra
DOE SIF4 SPECTRA SiF4 Emissions Profiles 18000 16000 14000 C4F8 Center Point 12000 C3F8 POR DOE 2 SiF4, PPMV 10000 DOE 5 DOE 11 8000 Less SiO2 to etch in low C3F8 III 6000 pressure clean for C4F8 4000 2000 0 1 6 101 106 111 116 121 126 131 136 141 146 151 156 161 166 171 176 181 186 191 11 16 21 26
Slide 14: DOE Analysis and Learning
DOE ANALYSIS AND LEARNING The DOE data pointed to a process with lower C4F8 flows, higher RF power, and O2 in the 81 – 82% range Single variable tests were then run looking for the minimum endpoint time using Optical emissions spectrometry (OES) As an alternative, data from the DOE was used to
Slide 15: Clean Time Optimization
C4F8 CLEAN TIME OPTIMIZATION C4F8 %O2 OPTIMIZATION WITH C4F8=500sccm C4F8 FLOW OPTIMIZATION AT 82% O2 13:55 13:40 13:40 13:33 HP Clean EP Time 13:26 13:26 HP Clean EP Time 13:12 13:19 13:12 12:57 13:04 12:43 12:57 12:28 12:50 12:43 12:14 12:36 12:00 12:28 11:45 12:21 78 81 84 550 500 450 400 % O2 C4
Slide 16: Clean Time Optimization
C4F8 CLEAN TIME OPTIMIZATION Several interesting observations were made while analyzing the C4F8 low flow cleans The SiF4 graphs showed higher SiF4 amounts removed during the high pressure clean The low pressure clean showed lower SiF4 peaks, indicating less oxide remained for this clean The
Slide 17: OES Endpoint for C3F8 Clean
AMD¢1 OES ENDPOINT FOR C3F8 CLEAN C2DUAL.RUN Ver. Dual 3.3 4 aRRay[@i_i2_i68.01 12718/01 02:23p | B ARRAY[Gi_12_18.62 12718701 G2: © ARRAYS ] D aRRay[a_ ] Chan.A HV. 5 5.98) 7430 BERT. (#3! au s 1a. OBB SE 7a a53-o : i 3 - 80) g 80.@!- 4 Wavelength a f. a 1-s.e8 1 69.6 . s s A Run Time t_10.0 £ 40.0
Slide 18: OES Endpoint for C4F8 Clean
C2DUAL.RUN Yer. Dual 3.3 A ARRAY(G2_61_04.01 @1704/02 @1:3ip | B ARRAY[G2_01_04.02 61704702 61:31p | ¢ arRay[e D aARRAY[d @ Wavelength 5 2.00/7 NTs 50.0 F20F- 3-.400 49.9 B Wavelength 3 464-6] i_s.se 30.4] > Chan. HY f-5.26 20.9) ~2O96 r-5. : [__-sae@ e Chan.B HY ¢-7.60 19.0 nh -168. 8) - Notes Mode
Slide 19: C3F8 High Pressure Clean
C3F8 HIGH PRESSURE CLEAN
Slide 20: High Pressure Clean
C4F8 HIGH PRESSURE CLEAN
Slide 21: Process Comparison
C4F8 PROCESS COMPARISON 50 Wafers of 11,500A Silane Oxide C4F8 C3F8 HP EP Clean Time 13:32 13:37 LP Clean Time 3:02 7:16 Total Time 16:34 20:53 Cost per Wafer $0.073 $0.276 Emissions 0.13 1
Slide 22: Oxide Process Characterization
OXIDE PROCESS CHARACTERIZATION Once the final process was defined, film characterization work was undertaken to ensure no deleterious affects with the C4F8 clean Collected thickness and uniformity data after a clean Collected chamber temperature data during and after the clean Sent wafers ou
Slide 23: Silane Oxide Process Data: Thickness Profile
C4F8 SILANE OXIDE PROCESS DATA S ILA N E O XID E T H IC KN E S S P R O F ILE A F T E R C LE A N 11650 11600 11550 C4F8 Ra nge =137A 11500 C4F8 C3F8 11450 C3F8 Ra nge =162A 11400 11350 11300 Wa f e r #
Slide 24: Silane Oxide Process Data: Uniformity
C4F8 SILANE OXIDE PROCESS DATA S ILA N E O XID E WIT H IN - WA F E R UN IF O R M IT Y A F T E R C LE A N 0.85 0.8 0.75 C4F8 C3F8 0.7 0.65 0.6 Wa f e r #
Slide 25: Silane Oxide Process Data: Chamber Temperatures
C4F8 SILANE OXIDE PROCESS DATA C H A M B E R T E M P E R A T UR E S F O R T H E C 4 F 8 P LA S M A C LE A N 420.00 400.00 380.00 S howe r he a d Te mp 360.00 He a t e rbloc k Te mp 340.00 P la sma Cle a n 320.00 300.00 T i me
Slide 26: In-Film Particle Data
C4F8 IN-FILM PARTICLE DATA N O VELLUS S EQ UE L S ILA N E O XID E IN - F ILM P A R TIC LE C OUN T 40 30 C4 F8 C4 F8 20 10 0 -1 0 -20 -30 Da t e
Slide 27: TOF-SIMS Results for Silane Oxide Wafer
TOF-SIMS RESULTS FOR SILANE OXIDE WAFER Surface Concentrations File Name C4F8 Center C4F8 Edge C3F8 Center C3F8 Edge B 1.8E+13 1.2E+13 1.7E+13 1.6E+13 C 2.6E+15 2.6E+15 2.4E+15 1.8E+15 O 8.9E+15 6.3E+15 8.1E+15 9.7E+15 F 3.5E+12 3.1E+12 4.3E+12 3.9E+12 Na 8.9E+09 3.1E+09 3.5E+10 3.9E+09 K 2.9E+09 3.
Slide 28: Conclusions
CONCLUSIONS The new C4F8 plasma clean for PECVD chambers has several advantages Estimated 87% reduction of PFC emissions as compared to the current C3F8 process 70% weight reduction of gas consumed, which results in a significant cost savings Reduction of the total clean time, improving thro
Slide 29: Estimated Effects of C4F8 Clean
ESTIMATED AFFECTS OF C4F8 CLEAN AMD PFC Emissions Fab 25 C4F8 Dielectric Deposition Cleans Fab 25 C3F8 Dielectric Deposition Cleans 1.4 All other sources 1.2 MMTCE / MMTCE(1995) 1.0 0.8 0.6 *Projection 0.4 0.2 0.0 1995 1996 1997 1998 1999 2000 2001 2002* 2003*
Slide 30: Next Steps
NEXT STEPS There is currently one dual-chamber Novellus Sequel tool running production with the C4F8 clean Once good data is obtained from this tool across a PM cycle, the process can be migrated to other tools The conversion requires new MFCs (lower flow) and a change to the endpoint script
Slide 31: Questions?
| QUESTIONS?