USF Libraries
USF Digital Collections

Consumable process development for chemical mechanical planarization of bit patterned media for magnetic storage fabrication

MISSING IMAGE

Material Information

Title:
Consumable process development for chemical mechanical planarization of bit patterned media for magnetic storage fabrication
Physical Description:
Book
Language:
English
Creator:
Bonivel Jr., Joseph T.
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Tribology
Coefficient of friction
Wear
Superparamagentic limit
Magnetic hard drive fabrication
Dissertations, Academic -- Mechanical Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: As the superparamagnetic limit is reached, the magnetic storage industry looks to circumvent the barrier by implementing patterned media (PM) as a viable means to store and access data. Chemical mechanical polishing (CMP) is a semiconductor fabrication technique used to planarize surfaces and is investigated as a method to ensure that the PM is polished to surface roughness parameters that allow the magnetic read/write head to move seamlessly across the PM. Results from this research have implications in feasibility studies of utilizing CMP as the main planarization technique for PM fabrication. Benchmark data on the output parameters of the CMP process, for bit patterned media (BPM), based on the machine process parameters, pad properties, and slurry characteristics are optimized. The research was conducted in a systematic manner in which the optimized parameters for each phase are utilized in future phases. The optimum results from each of the phases provide an overall optimum characterization for BPM CMP. Results on the CMP machine input parameters indicate that for optimal surface roughness and material removal, low polish pressures and high velocities should be used on the BPM. Pad characteristics were monitored by non destructive technique and results indicate much faster deterioration of all pad characteristics versus polish time of BPM when compared to IC CMP. The optimum pad for PM polishing was the IC 1400 dual layer Suba V pad with a shore hardness of 57, and a k-groove pattern. The final phase of polishing evaluated the slurry polishing properties and novel nanodiamond (ND) slurry was created and benchmarked on BPM. The resulting CMP output parameters were monitored and neither the ND slurry nor the thermally responsive polymer slurry performed better than the commercially available Cabot iCue slurry for MRR or surface roughness. Research results indicate CMP is a feasible planarization technique for PM fabrication, but successful implementation of CMP for planarizing PM must address the high initial start up cost, increase in the number of replacement pads, and increase in polishing time to reach the required surface roughness for magnetic storage devices.
Thesis:
Dissertation (PHD)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Joseph T. Bonivel Jr..
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004770
usfldc handle - e14.4770
System ID:
SFS0028060:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Consumable Process Development for Chemical Mechanical Planarization of Bit Patterned Media for Magnetic Storage Fabrication b y Joseph T. Bonivel Jr. A dissertation submitted in partial fulfillment of the requir ements for the degree of Doctor of Philosophy Department of Mechanical Engineering College of Engineering University of South Florida Major Professor: Ashok Kumar, Ph.D Delcie Durham, Ph.D C. Fred Higgs III, Ph.D Garrett Matthews, Ph.D Geoffrey Okogbaa, Ph.D Frank Pyrtle Ph.D Date of Approval: October 25, 2010 Keywords: tribology, coefficient of friction, wear, superparamagentic limit, magnetic hard drive fabrication Copyrig ht 2010, Joseph T. Bonivel Jr

PAGE 2

DEDICATION This thesis is dedicated to my family who has always lived by the mantra I love you all my heart. & Thank you for everything you have done for me

PAGE 3

ACKNOWLEDGEMENTS First and foremost I would like to thank and acknowledge Jesus Christ for without him none of this would be possible. I would like to thank my research advisors Dr. Ashok Kumar, and Dr. C. Fred Higgs III for their guidance friendship, and counseling throughout this dissertation process. Without your guidance I would have not been able to complete this process I would also like to sincerely thank Mr. Bernard Batson who has been a cornerstone to my experience s at USF. I would like to thank the NSF GK 12, NSF Bridge to Doctorate, Sloan Foundation, and MRSEC CMU, for funding my research. I would like to thank Dr. Garrett Matthews, Dr. Delcie Durham, Dr. Frank Pyrtle Dr Geoffrey Okogbaa for serving as my committee members for my dissertation and being patient with me through this process, your expertise is invaluable. I would also like to thank Dr. Philip Voglewede, Dr. Philip LeDuc for their encouragement over the years. Thanks go to my CMU and NMRL USF labmates. I also t hank my REU students Justin and Yusuf for their tireless work. To my colleagues, Alisha, Brandon, Boone, Frank, Eric, and Tarah thank you for your encouragement support, and willingness to help me during this process. Lastly and most importantly I want to acknowledge and thank my friends and family who supported me during all of my schooling: my mother Carolyn, father Joe Sr, sister Shon, brother Marlon, Dennis, Bernie P, Moiya, and the USC and USF members of Phi Beta Sigma fraternity Inc I love you all.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES ..... ................................ ................................ ................................ ........ vi LIST OF FIGURES ... ................................ ................................ ................................ ...... viii ABSTRACT ... ............ ................................ ................................ ................................ ...... xiv CHAPTER 1: MAGNETIC STORAGE DEVICES: PATTERNED MEDIA .................. 1 1.1 Foreword ................................ ................................ ................................ .......... 1 1.2 Introduction ................................ ................................ ................................ ........ 3 1.2.1 Fundamentals of Read/Write Magnetic Hard Drives .......................... 3 1.2.2 Magnetic Hard Drive Fundamentals ................................ ................... 4 1.3 Technologies to Avoid Superparamagnetic Limit ................................ ............. 6 1.3.1 Perpendicular Recording ................................ ................................ ..... 7 1.3.2 Heat Assisted Magnetic Recording ................................ ..................... 8 1.3.3 Patterned Media Data Storage ................................ ............................ 8 1.3.3.1 Background on Patterned Media ................................ .......... 9 1.3.3.2 Patterned Media and Areal Density ................................ ... 10 1.3.3.3 Fabrication of Patterned Media ................................ .......... 11 1.4 Outline of this Dissertation ................................ ................................ .............. 12 CHAPTER 2: CHEMICAL MECHANICAL PLANARIZATION ................................ 15 2.1 Foreword ................................ ................................ ................................ ........ 15 2.2 Introduction: Development of CMP ................................ ............................... 17 2.2.1 Multilevel Metallization ................................ ................................ .... 19 2.2.2 Interconnect Fabrication ................................ ................................ ... 21 2.2.3 Multilevel Metallization Challenges ................................ ................. 22 2.2.4 Copper Emergence ................................ ................................ ............ 24 2.2.4.1 Subtractive Etch ................................ ................................ 26 2.2.4.2 Damascene Process ................................ ............................ 26 2.2.5 Options for Planarization ................................ ................................ .. 27 2.3 Chemical Mechanical Planarization (CMP) ................................ .................... 28 2.3.1 Applications of CMP ................................ ................................ ........ 30 2.3.2 History of CMP ................................ ................................ ................. 31 2.3.3 CMP Process ................................ ................................ ..................... 33 2.3.4 Material Removal Mechanism ................................ .......................... 35 2.3.4.1 Mechanical Aspects of CMP ................................ ............. 36 2.3.4.2 Chemical Aspects of CMP Material Removal ................... 37 2.3.4.3 Governing Factors of CMP Process ................................ ... 38 2.3.5 Process Parameters ................................ ................................ ............ 41

PAGE 5

ii 2.3.6 Consumable Characteristics ................................ .............................. 42 2.4 Tribology of CMP Process ................................ ................................ ............... 44 2.4.1 CMP Process ................................ ................................ ..................... 44 2.4.2 Tribo Metrology of CMP ................................ ................................ .. 45 2.4.2.1 Coefficient of Friction ................................ ........................ 45 2.4 .2. 2 Boundary Lubrication ................................ ........................ 47 2.4.2. 3 Mixed Lubrication ................................ ............................. 47 2.4.2 .4 Hydrodynamic Lubrication ................................ ................ 47 2.4.3 Acoustic Emission ................................ ................................ ............ 48 2.4.4 End Point Detection ................................ ................................ .......... 48 2.4.4.1 EPD, COF, and AE Signal ................................ ................. 50 2.5 Modeling in CMP ................................ ................................ ............................ 52 2.5.1 Preston Model ................................ ................................ ................... 52 2.5. 2 Mechanical Models ................................ ................................ ........... 53 2.5.2.1 Empirical Wear Modeling Studies ................................ ..... 53 2.5.2.2 Contact Mechanics Modeling ................................ ............ 54 2.5.3 Fluid Mechanics Models ................................ ................................ ... 55 2.5.4 Hybrid Models (PAML) ................................ ................................ ... 56 2.6 Challenges During the CMP Process ................................ ............................... 57 2.6.1 Non Planarity ................................ ................................ .................... 58 2.6.1.1 Dishing and Erosion ................................ ........................... 59 2.6.1.2 Oxide and Metal Loss ................................ ........................ 59 2.6.2 Surface Scratches ................................ ................................ .............. 60 2.6.3 Delamination ................................ ................................ ..................... 61 2.7 Conclusion Research Objectives ................................ ................................ ...... 62 CHAPTER 3: CMP PROCESS MACHINE PARAMETER OPTIMIZATION .............. 65 3.1 Foreword ................................ ................................ ................................ ........ 65 3.2 Patterned Media Data ................................ ................................ ....................... 67 3.3 Experimental ................................ ................................ ................................ .... 70 3.3.1 Candidate Samples ................................ ................................ ............ 70 3.3.2 Mechanical Properties ................................ ................................ ....... 74 3.3.3 WYKO Surface Profiler ................................ ................................ .... 78 3.3.4 CETR Benchtop Chemical Mechanical Polishing Tester ................. 84 3.3.4.1 Experimental Procedure: Process Parameters ................... 86 3.3.4.2 Experimental Procedure: Material Removal Rate ........... 87 3.3.4.3 Transmission Electron Microscopy (TEM) ....................... 89 3.3.4.4 Consumables ................................ ................................ ...... 90 3.3.4.5 Optimization of CMP Experimentation ............................. 91 3.4 Results and Discussion ................................ ................................ .................... 92 3.4.1 CMP of the Patterned Media Surface ................................ ............... 92 3.4.1.1 MRR and Pressure ................................ ............................. 92 3.4.1.2 MRR and Velocity ................................ ............................. 93 3.4.1.3 Stribeck Curve ................................ ................................ ... 95 3.4.2 BPM Pre/Post CMP Mechanical Properties ................................ ..... 98 3.4.2.1 Slurry Chemistry Characterization ................................ .. 100

PAGE 6

iii 3.4.2.2 TEM Analysis ................................ ................................ .. 101 3.4.3 Surface Roughness Characterization ................................ .............. 104 3.4.4 Statistical Analysis of Variance (ANOVA) ................................ .... 107 3.5 Conclusion and Remarks ................................ ................................ ............... 109 CHAPTER 4: BIT PATTERNED MATRIX CMP PAD CHARACTERIZATION ...... 112 4.1 Foreword ................................ ................................ ................................ ...... 112 4.2 Introduction: CMP Pads ................................ ................................ ................ 114 4.2.1 Pad Materials ................................ ................................ .................. 114 4.2.2 Effect of Pad Geometry ................................ ................................ .. 115 4.2.2.1 Effect of Pores ................................ ................................ .. 115 4.2.2.2 Effect of Grooves ................................ ............................. 116 4.3 Pad Characterization ................................ ................................ ...................... 117 4.3.1 Ultrasound Transmission ................................ ................................ 118 4.3.2 Surface Characterization ................................ ................................ 119 4.3.3 Wafer and Pad Surface Roughness ................................ ................. 120 4.3.4 Ex Situ and In Situ CMP Pad Characteristics ............................... 121 4.3.5 Material Removal ................................ ................................ ............ 121 4.4 Experimental Set Up ................................ ................................ ...................... 122 4.4.1 CMP Pads ................................ ................................ ........................ 122 4.4.2 Ultrasound Transmission Testing System (UTS) ........................... 124 4.4.3 CETR CMP Polishing ................................ ................................ ..... 126 4.5 Results and Discussion ................................ ................................ .................. 127 4.5.1 COF and Pad Polishing ................................ ................................ ... 127 4.5.1.1 Stribeck Curve ................................ ................................ 128 4.5.1.2 COF and the MRR ................................ ........................... 129 4.5.1.3 COF and Polish Time ................................ ....................... 131 4.5.1.4 BPM Surface Roughness and COF ................................ .. 132 4.5.2 Polish Time Metrics ................................ ................................ ........ 133 4.5.2.1 Material Removal ................................ ............................. 133 4.5.2.2 BPM Surface Roughness ................................ ................. 135 4.5.2.3 Pad Roughness ................................ ................................ 136 4.5.2.4 Pad Thickness ................................ ................................ .. 137 4.5.3 Metrics Discussion ................................ ................................ .......... 138 4.5.4 Surface Morphology Characterization ................................ ............ 139 4.5.4 Pad Morphology Discussion ................................ ........................... 144 4.5.5 WTWNU ................................ ................................ ......................... 144 4.6 Conclusion and Remarks ................................ ................................ ............... 145 CHAPTER 5: SYNTHESIS OF NOVEL CMP SLURRY ................................ ............. 148 5.1 Foreword ................................ ................................ ................................ ...... 148 5.2 Introduction ................................ ................................ ................................ .... 150 5.2.1 Effect of Slurry on Planarization (Surface Quality) ....................... 150 5.2.2 Chemical Effect of Slurry of Material Removal Rate .................... 152 5.2.3 Mechanical and Material Pr operties ................................ ............... 153 5.2.4 Particle Size and Hardness ................................ .............................. 156

PAGE 7

iv 5.2.5 Abrasive Particle ................................ ................................ ............. 157 5.2.5.1 Commercial Slurry Abrasive Synthesis ........................... 157 5.2.5.2 Surface Quality Based on Abrasives ................................ 159 5.3 Novel Nanodiamond (ND) Slurry Synthesis ................................ ................. 160 5.3.1 Hybrid Particle Synthesis ................................ ................................ 161 5.3.2 Particle Characterization ................................ ................................ 162 5.3.2.1 NIPAM Dynamic Light Scattering [188] ........................ 162 5.3.2.2 Transmission Electron Microscopy (TEM) ..................... 163 5.3.2. 3 Post CMP Surface Characterization ................................ 163 5.4 Experimental Conditions for ND Slurry and Particle Slurry Testing ............ 163 5.5 Results and Discussion ................................ ................................ .................. 164 5.5.1 TEM Imaging ................................ ................................ .................. 165 5.5.2 COF of the Slurry Abrasives ................................ ........................... 166 5.5.3 MRR Versus the Abrasive Particle ................................ ................. 168 5.5.4 Surface Quality and Roughness ................................ ...................... 169 5.6 Analysis of NIPAMND Abrasive Concentration ................................ .......... 171 5.6.1 COF and the Abrasive Particle Concentration ................................ 172 5.6.2 MRR and Abrasive Particle Concentration ................................ ..... 173 5.6.3 Surface R oughness and A brasive P article C oncentration ............... 174 5.7 Conclus ions and Remarks ................................ ................................ .............. 174 CHAPTER 6: CMP MODELING OF MICROSTRUCTURAL VARIATION ............. 176 6.1 Foreword ................................ ................................ ................................ ...... 176 6.2 Introduction ................................ ................................ ................................ .... 177 6.3 Crystallography ................................ ................................ .............................. 178 6.4 E xperimental Design ................................ ................................ ...................... 179 6.4.1 CMP Simulation ................................ ................................ .............. 181 6.5 Simulation Results ................................ ................................ ......................... 182 6.6 Conclusions and Remarks ................................ ................................ .............. 183 CHAPTER 7: MULTIPHYSICS DISCUSSION OF BPM CMP ................................ .. 184 7.1 Foreword ................................ ................................ ................................ ...... 184 7.2 Pad Based Wear ................................ ................................ ............................. 185 7.3 Slu rry Based Wear ................................ ................................ ......................... 187 7.4 Mixed Polishing ................................ ................................ ............................. 190 CHAPTER 8: CONCLUSION AND FUTURE WORK ................................ ................ 194 8.1 Conclusions ................................ ................................ ................................ .... 194 8.2 Future Work ................................ ................................ ................................ ... 199 REFERENCES .......... ................................ ................................ ................................ ...... 204 APPENDIX A: PHASE I MACHINE PARAMETERS ................................ ................. 218

PAGE 8

v APPENDIX B: PHASE II: PAD CHARACTERIZATION DATA .............................. 227 ABOUT THE AUTHOR ................................ ................................ ....................... End Page

PAGE 9

vi LIST OF TABLES Table 2. 1 Interconnection delay ................................ ................................ ....................... 23 Table 2. 2 Disadvantages of CMP ................................ ................................ .................... 28 Table 2. 3 Advantages of CMP ................................ ................................ ......................... 29 Table 2. 4 Applications of CMP technology ................................ ................................ .... 31 Table 2. 5 Factors governing output of CMP process ................................ ...................... 41 Table 2. 6 End point d etection methods ................................ ................................ ............ 49 T able 3. 1 Metrics for SEMATECH s amples ................................ ................................ ... 73 Table 3. 2 Nanoindentation results of BPM c opper ................................ ......................... 78 Table 3. 3 Comparison of VSI and PSI m odes in Wyko surface p rofiler ......................... 82 Table 3. 4 Process parameters for CETR t ests ................................ ................................ .. 87 Table 3. 5 CMP pad material properties ................................ ................................ ........... 90 Table 3. 6 Statistical ANOVA table ................................ ................................ ................. 91 Table 3. 7 Nanoindentation results of BPM c opper ................................ .......................... 99 Table 3. 8 Mechanical properties from the slurry ................................ ........................... 100 Table 3. 9 RMS surface roughness for CMP process parameters ................................ ... 104 Table 3. 10 ANOVA table for pressure and velocity of BPM CMP .............................. 108 T able 4. 1 CMP pad characteristics ... 123 Table 4. 2 Process p arameters for pad characterization polishing ................................ .. 126 Table 4. 3 W T WNU percentage based on BPM wafer CMP ................................ ......... 145 Table 5 1 CMP process output and slurry mechanisms [171] .......... 151

PAGE 10

vii Table 5. 2 Slurry abrasive characteristics ................................ ................................ ....... 158 Table 5. 3 Slurry details ................................ ................................ ................................ .. 164 Table 5. 4 Process conditions for slurry testing ................................ .............................. 164 Table 5. 5 Surface roughness for slurries ................................ ................................ ........ 169 Table 5. 6 WYKO surface profile data for number of particles ................................ ..... 174

PAGE 11

viii LIST OF FIGURES Figure 1. 1 Traditi onal longitudinal read/w rite hard drive ................................ ................. 4 Figure 1. 2 Scaling factors used to increase areal density [10] ................................ ........... 6 Figure 1. 3 Perpendicular recording ................................ ................................ .................... 8 Figure 1. 4 Patterned media schematic[21] ................................ ................................ ......... 9 Figure 1. 5 Need for BPM planarization ................................ ................................ ........... 11 Figure 1. 6 SEM of nanocolumns for patterned media [26] ................................ ............. 12 Figure 2. 1 Historical comparison of the trend of microprocessors [3] ............................ 17 Figure 2. 2 MOS capacitor configuration ................................ ................................ ......... 18 Figure 2. 3 MLM s cheme [12] ................................ ................................ .......................... 20 Figure 2. 4 SEM image MLM roughness [12,17 ] ................................ ............................ 24 Figure 2. 5 BEOL IC fabrication ................................ ................................ ...................... 26 Figure 2. 6 Global and surface planarity for planarization processes [23] ....................... 27 Figure 2. 7 Schematic of CMP p rocess ................................ ................................ ............. 34 Figure 2. 8 Three body abrasion on CMP process ................................ ............................ 35 Figure 2. 9 Three body abrasion during CMP ................................ ................................ .. 37 Figure 2. 10 Degrees of planarity in CMP process ................................ ........................... 40 Figure 2. 11 Stribeck curves generated from COF data [33] ................................ ............ 46 Figure 2. 12 AE signal to CMP polish layer change ................................ ......................... 51 Figure 2. 13 Cu CMP surface characteristics ................................ ................................ .... 58 Figure 2. 14 CMP challenges ................................ ................................ ............................ 59

PAGE 12

ix Figure 2. 15 Optical image of surfaces scratches from CMP ................................ ........... 60 F igure 2. 16 Delamination from copper CMP ................................ ................................ .. 62 Figure 2. 17 CMP process factors ................................ ................................ ..................... 63 Figure 3. 1 Machine process parameters optimized for PM polishing ............................. 65 Figure 3. 2 Conventional longitudinal magnetic storage ................................ .................. 67 Figure 3. 3 Read/write slider head on PM ................................ ................................ ........ 68 Figure 3. 4 CMP p rocess ................................ ................................ ................................ ... 69 Figure 3. 5 MIT 854 p attern s ublevel ................................ ................................ ............... 71 Figure 3. 6 Patterned media configurations ................................ ................................ ...... 72 Figure 3. 7 Optical microscope images of BPM at 30x and 500x ................................ .... 73 Figure 3. 8 FIB image of patterned media configuration ................................ .................. 74 Figure 3. 9 MTS Nano Indenter XP ................................ ................................ ............... 75 Figure 3. 10 Typical nanoindentation curve [119] ................................ ............................ 75 Figure 3. 11 Nanoindentation into BPM ................................ ................................ ........... 77 Figure 3. 12 Wyko NT9100 surface profiler ................................ ................................ .... 79 Figure 3. 13 Interference microscope [124] ................................ ................................ ...... 80 Figure 3. 14 CETR benchtop tester ................................ ................................ ................... 84 Figure 3. 15 Four point probe station ................................ ................................ ................ 88 Figure 3. 16 Pressure vs MRR for BPM CMP ................................ ................................ .. 93 Figure 3. 17 MRR vs. relative velocity ................................ ................................ ............. 94 Figure 3. 18 Stribeck curve ................................ ................................ ............................... 96 Figure 3. 19 MRR vs. COF ................................ ................................ ............................... 98 Figure 3. 20 Initial multigranular cross section of copper BPM ................................ ..... 101

PAGE 13

x Figure 3. 21 Post polish TEM of Cu BPM ................................ ................................ ...... 102 Figure 3. 22 Dislocation motion in aluminum [145] ................................ ...................... 103 Figure 3. 23 Microcrack formations after polishing ................................ ....................... 103 Figure 3. 24 Initial surface roughness for BPM prior to polishing ................................ 105 F igure 3. 25 1 Psi post CMP surface roughness ................................ ............................. 105 Figure 3. 26 3 Psi post CMP surface roughness ................................ ............................. 106 Figure 3. 27 6 Psi post CMP surface roughness ................................ ............................. 106 Figure 3. 28 Delaminated edge SEM image ................................ ................................ ... 107 Figure 3.29 Residuals plot ................................ ................................ .............................. 109 Figure 4. 1 Pad characterization based on optimized machine input parameters ........... 113 Figure 4. 2 Schematic of CMP polyurethane pad pores [159] ................................ ........ 116 Figure 4. 3 UTS schematic for CMP pads [138] ................................ ............................ 119 Figure 4. 4 Jeol JSM6490 SEM surface morph o logy tool ................................ .............. 120 Figure 4. 5 FIB image of initial thickness of BPM ................................ ......................... 122 Figure 4. 6 Stribeck curve for pad polishes. ................................ ................................ ... 128 Figure 4. 7 MRR vs COF for CMP of BPM p ads ................................ ........................... 130 Figure 4. 8 Polish t ime versus COF for BPM CMP pads ................................ .............. 131 Figure 4. 9 BPM wafer roughness vs. COF ................................ ................................ .... 132 Figure 4. 10 MRR vs. polish time ................................ ................................ ................... 134 Figure 4. 11 BPM surface roughness vs. polish time ................................ ..................... 135 Figure 4. 12 Pad roughness vs. polish time ................................ ................................ .... 136 Figure 4. 13 Pad thickness vs. polish time ................................ ................................ ...... 138 Figure 4. 14 As received and conditioned UTS and SEM images of pad (2) ................. 140

PAGE 14

xi Figure 4. 15 UTS, SEM, and BPM surface roughness of pad (2) 10 polishes ............... 141 Figure 4. 16 UTS, SEM, and BPM surface roughness for pad (2) 20 polishes .............. 142 Figure 4. 17 UTS, SEM, and BPM surface roughness of pad (2) 30 polishes ............... 143 Figure 4. 18 UTS, SEM, and BPM surface roughness of p ad (2) 40 polishes ............... 143 Figure 5. 1 Slurry optimization schematic ................................ ................................ ...... 149 Figure 5. 2 Pad/wafer interface reactions with the slurry ................................ ............... 153 Figure 5. 3 Contact modes with abrasive weight concentrations ................................ .... 155 Figure 5. 4 Example of saturati on for MRR ................................ ................................ ... 156 Figure 5. 5 Slurry p articles ................................ ................................ .............................. 166 Figure 5. 6 COF vs. BPM wafers ................................ ................................ .................... 167 Figure 5. 7 COF vs. b lanket c opper w afers ................................ ................................ ... 167 Figure 5. 8 MRR vs. w afer type for analysis of slurry abrasives ................................ .... 168 Figure 5. 9 WYKO surface profiler images fo r BPM polishing ................................ ..... 170 Figure 5. 10 WYKO images of slurry polishing of blanket copper wafers .................... 171 Figure 5. 11 COF vs. number of particles for NIPAMND slurry ................................ ... 172 Figure 5. 12 MRR vs. n umber of abrasive particles for NIPAMND slurry .................... 173 Figure 6. 1 Hysitron n anoindenter ................................ ................................ .................. 180 Figure 6. 2 Deterministic surface topography ................................ ................................ 180 Figure 6. 3 Contour map of hardness ................................ ................................ .............. 181 Figure 6. 4 Voxelized surface topography ................................ ................................ ...... 182 Figure 6. 5 Cumulative wear rate simulation of CMP ................................ .................... 182 Figure 7. 1 CMP process parameters ................................ ................................ .............. 184 Figure 7. 2 Pad based wear ................................ ................................ ............................. 186

PAGE 15

xii Figure 7. 3 Slurry based wear ................................ ................................ ......................... 188 Figure 7. 4 Three body wear ................................ ................................ ........................... 191 Figure 8. 1 Process schematic for BPM CMP ................................ ................................ 195 Figure 8. 2 Future work ................................ ................................ ................................ .. 200 Figure A. 1 1 Psi 0.2 relative velocity ................................ ................................ ............ 219 Figure A. 2 1 Psi 0.8 relative velocity ................................ ................................ ............ 220 Figure A. 3 1 Psi 1.2 relative velocity ................................ ................................ ............ 221 Figure A. 4 3 Psi 0.2 relative velocity ................................ ................................ ............ 221 Figure A. 5 3 Psi 0.8 relative velocity ................................ ................................ ............ 222 Figure A. 6 3 Psi 1.2 relative velocity ................................ ................................ ............ 222 Figure A. 7 6 Psi 0.2 relative velocity ................................ ................................ ............ 223 Figure A. 8 6 Psi 0.8 relative velocity ................................ ................................ ........... 223 Figure A. 9 6 Psi 1.2 relative velocity ................................ ................................ ............ 224 Figure A. 10 COF vs. MRR for BPM ................................ ................................ ............. 225 Figure A. 11 SEM of BPM delamination at 6 Psi ................................ .......................... 226 Figure B. 1 Pad 1 (10 p olishes) ................................ ................................ ....................... 228 Figure B. 2 Pad 1 (20 p olishes) ................................ ................................ ....................... 229 Figure B. 3 Pad 1 (30 p olishes) ................................ ................................ ....................... 229 Figure B. 4 Pad 1 (40 p olishes) ................................ ................................ ....................... 230 Figure B. 5 Pad 1 (50 p olishes) ................................ ................................ ....................... 230 Figure B. 6 Pad 2 (10 p olishes) ................................ ................................ ....................... 231 Figure B. 7 Pad 2 (20 p olishes) ................................ ................................ ....................... 231 Figure B. 8 Pad 2 (30 p olishes) ................................ ................................ ....................... 232

PAGE 16

xiii Figure B. 9 Pad 2 (40 p olishes) ................................ ................................ ....................... 232 Figure B. 10 Pad 2 (50 p olishes) ................................ ................................ ..................... 233 Figure B. 11 Pad 3 (10 p olishes) ................................ ................................ ..................... 233 Figure B. 12 Pad 3 (20 p olishes) ................................ ................................ ..................... 234 Figure B. 13 Pad 3 (30 p olishes) ................................ ................................ ..................... 234 Figure B. 14 Pad 3 (40 p olishes) ................................ ................................ ..................... 235 Figure B. 15 Pad 3 (50 p olishes) ................................ ................................ ..................... 235 Figure B. 16 SEM morphology evolution ................................ ................................ ...... 236 Figure B. 17 Pad (2) SEM morphology evolution ................................ .......................... 237 Figure B. 18 Pad (3) SEM morphology evolution ................................ .......................... 238 Figure B. 19 Pad (1) UTS characterization ................................ ................................ ..... 239 Figure B. 20 Pad (2) UTS characterization ................................ ................................ ..... 240 Figure B. 21 Pad (3) UST characterization ................................ ................................ ..... 241 Figure B. 22 CMP p ad r oughness vs.COF ................................ ................................ ...... 242 Figure B. 23 BPM w afer r oughness vs. COF ................................ ................................ 242 Figure B. 24 Wafer roughness vs. pad roughness ................................ ........................... 243 Figure B. 25 MRR vs. p ad roughness ................................ ................................ ............. 243

PAGE 17

xiv A BSTRACT As the superparamagnetic limit is reached, the magnetic storage in dustry looks to circumvent the barrier by implementing patterned media (PM) as a viable means to store and access data. Chemical mechanical polishing (CMP) is a semiconductor fabrication technique used to planarize surfaces and is investigated as a method to ensure that the PM is polished to surface roughness parameters that allow the magnetic read/write head to move seamlessly across the PM. Results from this research have implications in feasibility studies of utilizing CMP as the main planarization technique for PM fabrication. Benchmark data on the output parameters of the CMP process, for bit patterned media (BPM), based on the machine process parameters, pad properties, and slurry characteristics are optimized. The research was conducted in a systematic manner in which the optimized parameters fo r each phase are utilized in future phases. The optimum results from each of the phase s provide an overall optimum characterization for BPM CMP. Results on the CMP machine input parameters indicate that for optimal surface roughness and material removal, low polish pressures and hi gh velocities should be used on the BPM. Pad characteristics were monitored by non destructive technique and results indicate much faster deterioration of all pad characteristics versus polish time of BPM when compared to IC CMP The optimum pad for PM p olishing was the IC 1400 dual layer Suba V pad with a shore hardness of 57, and a k groove pattern. The final phase of

PAGE 18

xv polishing evaluated the slurry polishing properties and novel nanodiamond (ND) slurry was created and benchmarked on BPM. The resulting CMP output parameters were monitored and neither the ND slurry nor the thermally responsive polymer slurry performed better than the commercially available Cabot iCue slurry for MRR or surface roughness. Research results indicate CMP is a feasible planar ization technique for PM fabrication, but successful implementation of CMP for planarizing PM must address the high initial start up cost, increase in the number of replacement pads, and increase in polishing time to reach the required surface roughness fo r magnetic storage devices.

PAGE 19

1 CHAPTER 1 : MAGNETIC STORAGE DEVICES: PATTERNED MEDIA 1.1 Foreword The hard disk drive is by far the most important member of storage hierarchy in modern computers [1] The magne tic hard disk drive (HDD) currently plays the most influential role in the storage industry and this role is continually growing due to capacity, performance, and price. Areal density, also sometimes called bit density, refers to the amount of data that c an be stored in a given amount of hard disk platter. Since disk platters surfaces are two dimensional, areal density is a measure of the number of bits that can be stored in a unit of area [2] Since the inception of the original RAMAC by IBM in 1956, a variation of scaling laws have been us ed to increase the area l density of HDD [1] In current longitudinal magnetic recording media, high areal density and low noise are achieved by averaging several hundred weakly coupled fer romagnetic grains per bit cell [3] The scaling laws that enable smaller bit and grain sizes will eventually prompt a sp ontaneous magnetization reversal process, which destroys the data, when the stored energy per particles competes directly with the thermal energy, at which point the maximum reliable areal density is reached, this point is the superparamagnetic limit. The growth rate of magnetic storage density has increased to compound growth rate of 100% per year. A t this rate of areal density, the physical limit of areal density known as the superparamagnetic limit will soon be reached [4] To elude the superparamagnetic limit new technologies must be developed in order to continue the increase in storage capa city or the risk of losing valuable data The obje ct ives for this chapter include:

PAGE 20

2 1) Detail ing the need for increasing the areal s torage density for magnetic storage device s 2) Provide fundamental understanding on the feasibility of patterned media storage devices 3) Detail fabrication challenges in making patterned media storage devices and the need for chemical mechanical planarization a s a fabrication step 4) Provide a systematic layout for this dissertat ion

PAGE 21

3 1.2 Introduction 1.2.1 Fundamentals of Read/Write Magnetic Hard D rives Data is re ad and written on magnetic disks due to the electromagnetic physics phenomena. In 1820 physicist Hans Christian Oersted observed that an electrical current flowing in a wire moved the needle of a compass located near this wire. When the electrical current was shut down, the compass needle went back showing the location of orth pole. Oersted concluded that all conductors (wires) create a magnetic field around them when an electrical current is flowing. When the direction (polarity) of this electrical current is reversed, so is the polarity of the magnetic field [5] In 1831 another physicis t called Michael Faraday found out that t he inverse was also true, if a strong enough magnetic field was created near a wire, electrical current would be produced (inducted) in the wire. If the direction of this magnetic field was reversed, the direction of the ele ctrical current was reversed as well [6] To understand how data is read and written on hard disk drives (HDD) and other magnetic devices, it is important to note two electromagnetic properties: 1) All conductors create magnetic fi elds around them when there is an electrical current flowi ng 2) A strong magnetic field can generate (induct) electrical current on a wire. The HDD read/write head is made up of a U shaped conductive material with a coil wrapped around it On the process of writing data to the hard disk drive, an electrical current is applied to the coil, creating a magnetic field around the read/write read. This field magnetizes the platter surface right below the head, aligning the magnetic particles to the left or to the right, depending on the polarity of the electrical current that was

PAGE 22

4 applied. R eversing the electrical current polarity will also reverse the polarity of the magnetic field. A stored bit is a sequence of magnetized particles. I n the p rocess of reading data from the hard disk drive, when the head passes on a magnetized area either a positive or a negative current will be inducted, allowing the drive control circuit to read the stored bits. Figure 1.1 contains a basic schematic of a rea d/write magnetic hard drive [7] Figure 1. 1 Traditional longitudinal read/w rite hard drive 1.2.2 Magnetic Hard Drive F undamentals The hist ory of hard disks is intertwined with the history of computing within the integrated circuit industry. The concept of storing large amounts of data on magnetic media was already in practice in the early 1950s with magnetic drum memories H owever, the vol umetric density was limited by the relatively low surface to volume ratio of such devices meaning these drum memories cou ld not hold much data [8, 9] In a magnetic disk, data is stored on a recording medium ( comm only referred to as media), which is responsive to the presence of strong magnetic fields, but stable in their absence. The storage density that a given medium can sustain is det ermined by a variety of factors:

PAGE 23

5 1) S ize and uniformity of the magnetic dipoles in the material 2) O rientation of the domains, 3) C oercivity 4) T emperature stability of the media. 5) Distance between the magnetic read/write head and the media. Since the magnetic field drops off as the cube of the distance between the head and the media, writing and reading sma ller spots depends on lowering the distance between the head and the magnetic media. Traditionally, the main component of this has been flying height [9] This requires the read/write head to fly at nanometers above the surface in order for the HDD to be efficient. The a real density, also somet imes called bit densit y, refers to the amount of data that can be stored in a given amount of hard disk platter. Since disk platters surfaces are two dimensional, areal density is a measure of the number of bits that can be stored in a unit of area [2] Since the inception of the original RA MAC by IBM in 1956, a variation of scaling laws have been used to increase the area l density of HDD, as shown in figure 1.2 and with a growth of 100% per year the limit to the scaling laws will soon be reached [1]

PAGE 24

6 Figure 1. 2 Scaling factors used to increase areal density [10] T he latest push from the consumer market is to achieve 100 Gb/in 2 of areal storage and this storage density presents a fundamental processing issue. The problem with scaling down the feature size is fundamental ly physics problem : decreasing bit size while achievi ng satisfactory signal to noise r equires decreasing grain size. Grain size, however, cannot be shrunk signific antly below its present state of the art value, ~100 Angstroms (A o ) without the magnetization of the gra ins becoming thermally unstable o r superparamagnetic [11] The scaling laws that enable smaller bit and grain sizes will eventually prompt a spontaneous magnetization reversal process which destroys the data. When stored energy per particles competes directly with thermal energy, the maximum reliable areal density is reac hed. T his point is the superparamagnetic limit. Details in the equations and boundary conditions of the superparamagnetic limit are beyond the s cope of this dissertation but can be found in literature [4, 10 13] 1.3 Technologies to A void S uperparamagne tic L imit To circumvent this barrier, three major technologies are being proposed by magnetic storage experts. They include :

PAGE 25

7 1) Perpendicular recording 2) H eat a ssisted magnetic recording 3) Patterned media It is known that the thermal energy required to reverse the magnetization of a magnetic region is proportional to the size of the magnetic region and the magnet ic coercivity of the material. This means the larger the magnetic region the higher the coercivity of the material and the less likely the material will spontaneously de magnetize by local thermal fluctuations (avoidance of the superparamagnetic limit) [14] 1.3.1 Perpendicular R ecording Perpendicular recording uses a higher coercivity mate rial through which the magnetic are enhanced [14 16] Perpendicular recording represents the shortest technological leap as evidenced by the recent roll out the world's first perpendicular hard disk drive by Toshiba [17] In this method the magnetic domain is vertically aligned, shown in figure 1.3 so that more data can be stored but as storage growth increases this technology will ultimately succumb to the superparama gnetic limit

PAGE 26

8 Figure 1. 3 Perpendicular recording 1.3.2 Heat A ssisted M agnetic R ecording In heat assisted magnetic recording (HAMR), a laser is used to heat up the media to reduc e its coercivity thus sufficiently allow ing switching of the field by the media head [18] A t almost the same instant as the laser is heating the media the information is being written onto the media HAMR takes advantage of high stability magnetic compounds such as iron platinum alloy and t hese materials can store single bits in a much smaller area without being limited by the same superparamagnetic effect [19] HAMR can theoretically create an areal density of 1 Tb/in 2 but requires a recorded mark size of approximately 25 nm. For this density the grain size in the recording medium must be less than approximately 5 nm to obtain a sufficient signal to noise ratio [18 20] 1.3.3 Patterned M edia D ata S torage The concept of lithographically patterning a hard disk was originally introduced to improve head tracking and signal to noise ratio, but it is now clear that patterning offers the possibility of much higher areal densities than conventional hard disk media

PAGE 27

9 A patterned medium consists of an array of Each element is a s ingle magnetic domain, with uni axial magnetic anisotropy so that the magnetization points in one of only two directions at remanence, representing 1 bi t of data as shown in figure 1.4 Figure 1. 4 Patterned media schematic [21] The direction of magnetization in each dot (upward or downward) corresponds to the digital sign al of "0" or "1". no transition noise in the read/write process. Additionally, the dot size, which determines the areal dot density, can be decreased ultimately to the critical grain size of thermal stability. An ultra high recording density beyond 1 0 Tbit/inch 2 i s therefore expected to be achieved by applying the patterned media to the recording system. 1.3.3.1 Background on P atterned M edia The advantages of patterning recording media were recognize d as early as 1963 by Shew et al. They showed that discrete patterned tracks on a hard disk platter could reduce the c ross talk and noise problems associate d with head positioning errors and allow increased tracking tolerances [11] More recently, Lambert et al. have used patterned magnetic films to e xplore narrow track recording [22] It was shown that

PAGE 28

10 patterned media can be used to provide feedback information to a head servomotor and that M3 mag n etized patterned media can be used as a read only storage system [23] The first studies of regular arrays of sub micron patterned magn etic islands were presented in a series of papers by Smyth et al. The group investigated the c ollective switching properties of lithographically define d permalloy (NiFe) islands and compared their results with micro magnetic calculations [24] Gibson et al. investigated the individual switching characteristic s of similar permalloy particles using magn etic force microscopy (MFM) [25] These permalloy pa rticles behaved identically to single domain particles, and would reverse their magnetizati on under the influence of an MFM tip, Th e use of patterned magnetic islands as a sin gle bit per island discrete recording medium was not the focus of the aforementioned papers but the feasibility for such configurations is evident [11] 1.3.3.2 Patterned M edia and A real D ensity Patterned media data storage technology aims to increase areal storage density by using advanced semiconductor processing techniques This techniques are used to fabricate nano magnetic structures for the purpose of isolating indiv idual grains for magnetic domains into regular patterns [26, 27] This techn ology would allow for storage of one bit per cell or grain, which is different from conventional drives where each bit is stored across a few hundred magnetic grains. Without the proper understanding of the prominent tribological issues that exist in the fabrication and successful operation of patterned media, this technology will remain in research laboratories. Conceivably, a patterned media disk drive will consist of a magnetic slider head that reads and writes information onto a spinning disk during f light. The disk will

PAGE 29

11 be comprised of micro to nanoscale magnetic structures that must be planarized to prevent lateral collision between the slider and coarse topography on the disk surface. Figure 1.5 shows that the individual magnetic domains are initially rough after fabrication until a planarization step is employed. Figure 1. 5 Need for BPM planarization 1.3.3.3 Fabrication of P atterned M edia Some of the fabricatio n requirements for realizing the patterned m edia recording system include: highly ordered dot arrays with high aspect ratio, formation of the dot arrays in the desired position, mass productivity and low cost, and planarization techniques after fabricatio n for suf ficient read/write clearance. The integrated circuit industry has tackled the s e fabrication limitations and the magnetic storage hard drive industry is utilizing those techniques to fabricate patterned media. Fabrication of patterned media data storag e devices utilizes semi conduc ting manufacturing techniques. These techniques include but are not limited to electrodeposition, evaporation and liftoff, etching processes and chemical mechanical polishing (CMP) for planarization [26] Figure 1.6 depicts a n example of fabricated

PAGE 30

12 columns (which consequently have an uneven surface due to the height differential of the columns and thereby need planarization) is shown. Figure 1. 6 SEM of nanocolumns for patterned media [26] There exists a need to understand the fundamental polishing mechanisms for planarization of PM. Optimization technique for planarization o f the PM is needed to reduce waste and sustainability of consumables. The nanodots or squares fabricated for patterned media have the capacity circumvent the superparamagnetic limit but t he re exists little data that optimizes the parameters for polishing of these structure s. CMP has been used since the 1920s for planarization of multiple materials and serves as a viable candidate to planarize these surfaces to the nanometer roughness they require for the read/write head fly height. 1.4 Outline of this D is sertation This dissertation is divided into three main sections; the first phase of this dissertation is based on benchmarking the CMP machine process parameter s on bit patterned matrix configurations. The second phase focuses on the consumables of the CMP process and their use for bit pattern matrix configurations. The last phase is a

PAGE 31

13 completely separate study which is focused on modeling of the abrasive portion of the CMP process and detailing the change in mechanical properties as it affects the CMP process. Chapter 2 explains the evolution of the CMP process from the integrated circuit (IC) industrial perspective. The chapter details the basic mechanisms of the CMP process from both a mechanical and chemical aspect, and present in detail the tribolo gical mechanisms active during CMP. The chapter details the various models develop ed to predict the multi physics phenomena of the CMP process. This chap t er also serves to introduce the rest of the results for the phases studied during this dissertation after a full understand ing of the process parameters and output from the CMP process is fully undertaken. Chapter 3 focuses on benchmarking the bit patterned matrix data utilizing a multitude of metrology tools and analytical techniques. The chapter detai ls the effects of the coefficient of friction, lubrication regimes, material removal rate, and resulting surface topographies from polishing the bit patterned media configurations. Finally the chapter offers a statistical analysis to determine the critica l parameter in material removal from a statistical standpoint while evaluating the CMP process for bit patterned configurations versus stand copper thin film polishing. Chapter 4 is the first of a two chapter investigation in the consumables during the C MP process. This chapter deals with the CMP pad and compares three commercially popular pad architectures while detailing their material removal rates, surface qualities, and coefficient of friction during bit patterned media polishing. The pad life is c haracterized for bit patterned configurations and a novel non destructive technique

PAGE 32

14 utilizing ultrasound detection is de tailed and used to monitor pad material properties while scanning electron microscopy is used to monitor the surface characteristics of the pad during polishing. The resulting surface qualities of the wafers are obtained through a surface profiler and a correlation between the pad characteristics and CMP output parameters is detailed. Chapter 5 is the second half of the consumable proces s investigation for bo th bit patterned configurations. This chapter focuses on understanding the polishing phenomena of the slurry used during CMP and development of novel polymer ND slurry to be used for either blanket or bit patterned copper CMP. The c oefficients of friction, material removal r ate and surface quality are compared for three different slurries and an investigation into the abrasive polymer ND slurry concentration is elucidated. Chapter 6 reflects a separate modeling study which focuse s the evolution mechanical properties during the CMP p rocess. The deterministic microstructural variation during polishing is incorporated into a particle augmented lubrication model developed at Carnegie Mellon University and the results of the model are presented in this chapter. Chapter 7 discusses the interaction between the consumables and the machine process parameters. The influence of the optimized bit patterned media CMP is discussed in detail. The chapter provides a detailed account how each p arameter affects the output parameters of the CMP process. Chapter 8 summarizes the work done in this dissertation along with suggestions for future work with bit patterned CMP process parameters and optimization techniques.

PAGE 33

15 CHAPTER 2: CHEMICAL MECHANICAL PLANARIZATION 2.1 Foreword CMP has been described as an enabling technology because the high degree of planarization generated with the CMP eases the burden of advanced lithography and etching techniques. The drawback to the CMP process is that it is not stable and well controlled and several issues ultimately affect the chip performance. The semiconductor device industry has been focused on implemented an increase in the number of transistors on a chip thereby increasing the device densit y. This approach is emulated by the magnetic storage drive community for the aforementioned reasons of current areal density reaching the superparamagnetic limit. Both industries are striving to put greater numbers of features on a smaller area. For a complete and through historical development of the CMP process, the rest of the chapter explanations on CMP is based on the CMP app roach from the semiconductor manufacturing industrial standpoint. The research objectives of this chapter are: 1) P rovide a fundamental background on the development of integrated circuit industry 2) Detail the fabrication steps of integrated chips 3) Detail the emergence of CMP as a planariazation step for many metals in particular for copper in integrated chips

PAGE 34

16 4) Detail the fundame ntal removal mechanisms and the multi physics phenomena of the CMP process 5) Outline and detail various models developed to predict the CMP process 6) Provide detail into the forthcoming chapters on CMP and its effect on bit patterned matrix configurations

PAGE 35

17 2.2 Introduction : D evelopment of CMP During the semiconductor fabrication and microelectronics revolution, the industry h as been focused on rapidly increasing the nu mber of devices per chip and shrinking the critical dimensions of these electronic components. Law y 1.5 to 2 years i n the semiconductor industry [28] In conjunction with the de mand for transistors from Moore, the microprocessor performance in terms of millions of instructions per second (MIPS) will also double in the same time frame [29] Figure 2.1 gives a historical comparison of the trend of microprocessors according to the semiconductor industry performance [30] Figure 2. 1 Historical comparison of the trend of microprocessors [3] The International Technology Roadmap for Semiconductors (ITRS) predicts that by 2011 over one billion transistors will be integrate d on a single monolithic die [31] of the art int egrated circuits (ICs) contain tens of millions of transistors, which are used to amplify and switch electronic signals, capacitors, that are used to block

PAGE 36

18 direct currents and allow alternate curr ents to travel through the chip, an d resistors, that are used to produce voltage, on a single chip. A tra nsistor is typically a MOSFET (M etal O xide S emiconductor Field Effect Transistor), which consist s o f a source, gate, and a drain. These devices can b e made to operate faster by red ucing the size of devices and hav ing the devices packed densely into a given chip reducing the distance the carriers ha ve to travel to interact [32, 33] The minimum feature size decreases the size of the device itself and this translates into reduction in intermediate pitch or spacing between features. Figure 2.2 is a basic schematic of the basic MOS capacitor. Figure 2. 2 MOS capacitor configuration Shrinking of device dimensions has become a crucial caveat for both the semiconductor industries and the magneti c storage hard drive industry. However, t he explosion in the number of transistors f abricated on a single IC has placed extreme demands on electrically interconnecting these devices in the manner necessary to perform the logical operatio ns of a modern microprocessor. The transit time, (T r ) of electrons in a device with velocity (V) is directly proportional to the length of the gate (L g ) T ransit

PAGE 37

19 time is the ratio of gate length to velocity of the electr on, and the transit times dictate the frequency of operation [33, 34] Understandably the efforts of the semiconductor industry are concentrat ed on reducing the gate length of the d evices themselves with current devices having gate length s on the order of nanometers [35] 2. 2. 1 Multilevel M etallization on tran sistor scaling has provided ever increasing transi stor performance and density. Scientists and engineers but each time the technology reached the predicted barriers, scaling did not stop I nstead im aginative new solutions were developed [36] The fabrication of these small devices faces several design, manufacturing, a nd process control challenges. Once the devices are fabricated on the silicon substrate in the preferred orientation they need to be connected to a device network between each other and connected to the outside wor ld via interconnect materials. Metallization is the fabrication step in which proper interconnection of circuit elements is made; it is the general name for the technique of coating metal on the sur face of non metallic objects. The metal layers deposited, typically copper, are vacuum d ep osited by one of four methods: filament evaporation, flash evaporation, electron bea m evaporation, or sputtering [37] Details of the deposition process are outside the scope of the research for this dissertation; however, further details can be found within the references. The metals connecting the devices at the silicon substrate level are deemed contact or first level metalli zation, and the metals that connect the devices to the outer world are the s econd level metallization [33, 34] The materials required for metallization need to be selected

PAGE 38

20 based on their mechanical, electrical, and chemical properties since these properties dictate the frequency of flow of the charge. The ITRS dictates that the minimum and maximum number of layers in a multi layer metallization (MLM) scheme needs to be 13 and 17 res pectively. Figure 2.3 shows the cross section image of a seven level multi l ayer metallization scheme with the silicon dioxide interleve l dielectrics (ILD) labeled [38, 39] Figure 2. 3 MLM s cheme [12] As previously mentioned chip manufacturers wanted to make chips with higher speed while simultaneously reduc ing the device dimensions. A sing le MOSFET is shown in figure 2.2 with gate (G), source (S), and drain (D) connections labeled. I nterconnection of this MOSFET with other devices on the chip is not shown and is

PAGE 39

21 accomplished through the polysilicon/metal alloy (silicide) gate level a nd several metal SiO 2 ILD levels joine d together by vertical vias [40] Fabrication of interconnects was vital to the emergence of CMP and is briefly detailed in the following section. 2.2.2 Interconnect F abrication The relentless competitor and customer driven demand for increased circuit densit y, functionality and versatility has led to evolutionary and revolutionary advances the integrated circuit (IC) [39] connecting circuit elements and distributing pow er [29] In order t o incorporate and acco mmodate the improvements in decreased feature size, increased device speed, and mor (BEOL) processes This made BEOL processes equally as import ant as the front end of line (FEOL) processes to reduce gate oxide thickness and channe l length in the MLM layers [41] In order to achieve high device frequency and low feature sizes of the devices the interconnect delay had to be reduced so that the signals could pass faster through the metal layers, thus making device s function at greater speed s The measure of the interc onnect delay is the RC delay or the time delay (T RC ) in seconds, or the frequency of charge flow associated with the interconnect materials is computed as product of resistance (R) of the metal lines and the capacitance (C) of the insulating interconnects Substituting for resistance in terms of wiring dimensions and material properties, RC time delay can be written as equation (2.1) [42] :

PAGE 40

22 (2 1) resistivit y of the metallic interconnect, k is the dielectric constant of the insulator, o is the permittivity of vacuum, L is the length of the interconnect line, P is the pitch between interconnect lines, and T is the thickne ss of the line. Any changes in the variables will increase or decrease the interconnect delay. R eduction of the RC delay leads to an increa se in the performance of ICs. Physically a reduction in the RC delay translates to a reduction in the length of the interconnect wiring This is why the IC fabrication industry has been increasing the number of metallization layers. 2.2.3 Multilevel M etallization C hallenges I t can be seen fr om equation ( 2. 1) that there is an increase in the RC delay with the decrease in inter connect wiring pitch. I n order to decrease the RC delay several options were explored: 1) Cu has replaced Al as interconnect wiring materia ls due to its lower resistivi ty 2) S everal novel low k m aterials are being explored 3) M ultilevel metallization scheme of wiring is being implemented. Table 1 calculates the RC time constants calculated for a few metals of given R s (sheet resistance) and 1 mm length on 1 m thick Si O 2 [43]

PAGE 41

23 Table 2. 1 Interconnection delay It is evident from table 2.1 that semiconductor manufactures would use Al or Cu as the interconnect material due to their low RC delay and advantageous material properties. In order to produce the multilevel metallization schemes in figure 2.3 for IC devices the top most layer of the previous metallization layers must be optically flat and in more recent devic es, atomically smooth [44] This is because if there exists any resid ual roughness at the previous layer, it will get compounded as the layers increase and after a couple of compounded layers, the roughness will be so high that lithography (patterning) will experience issues with the depth of foc us and any further processi ng will not be possible. To compound the problem of lithographic patterning, t he irregular surface anomalies cause the variation of the thickness in fine line widths (sub 0.5 m) depending upon photo resist thickness. An effectively planar ized surface off ers enormous benefits such as:

PAGE 42

24 1) Higher photolithography and dry etch yields, 2) Elimination of step coverage concerns 3) Minimization of prior level defects, 4) Elimination of contact interruption, undesired contac ts, and electro migration effects 5) Reduction of high contact resistance and inhomogeneous met allization layer thickness 6) L imitation in the stacking hei ght of metallization layers [45] Figure 2.4 shows SEM images of a) unpolished and b) polished MLM schemes. Figure 2. 4 SEM image MLM roughness [12, 17] 2.2.4 Copper E mergence It can be seen from figure 2.4 that it is not possible to proceed with further processing steps as depth of focu s issues come up during photolithography. S everal other subsequent processing challenges, such as voids with in interconnect layers due to compounded roughness, also occur. Thus, the CMP process becomes a crucial

PAGE 43

25 processing step in device fabrication in order to achieve successful fabrication of MLM structure. Besides increasing the number of metallization layers, reducing the resistance of the metal interconnects and the capacitance of the dielectr ic layer reduces the RC delay. Several electrical and m echanical properties are deemed optimal for selection of the correct materials for the metallic interconnects. F or brevity the optimal properties for IC fabrication can be found in [46] Based on the necess ities of the IC industry, copper was chosen to replace aluminum as the material for metallic inte rconnects to further reduce RC delay for future application trends [47, 48] Choosing electrically superior copper over aluminum comes with an inherent drawback Due to the difficulty in dry etching fine line copper, the d amascene ap proach was developed to make the metallization lay ers [49] These layers are fabricated throu gh two different BEOL processes: the subtrac tive process and the dama scene process show in figure 2.5

PAGE 44

26 Figure 2. 5 BEOL IC fabrication 2.2.4.1 Subt ractive E tch In the subtractive process the metal leads are patterned by subtractive etching followed by the deposition of an interlevel dielectric (ILD) to insulate and passivate the metal lines. This process was used for aluminum or tun gsten interconnects. 2.2.4.2 Damascene P rocess As Cu can not be effectively etche d due its ability to form non toxic volatile by products and its property of diffusion in neighboring materials, present day MLM structures are fabricated using the damascene process In the damascene process a thermal oxide layer i s grown on the bare silicon in the preferred orientation. The oxide layer is then etched us ing photolithography techniques. A layer of metal is then deposited onto the etch ed dielectric and is polished to ensure planarity as the MLM scheme is complied (f igure 2.5)

PAGE 45

27 2.2.5 Options for P lanarization Several technologies exist that achieve local and global planarity and are utilized by the semiconductor and magnetic storage industry. Techniques such as spin on deposition (SOD), reflow of boron phosphorous silicate glass (BPSG), spin etch planarization (SEP), reactive ion etching and etch back (RIE EB), SOD + EB have been di scussed in detail by Zantye [45] The different degrees of global a nd loc al surface planarity from each fabrication process can be seen from figure 2.7 [50] These are the promin ent output parameters of several competing technologies presently being used to achieve local and global planarization. Figure 2. 6 Global and surface planar ity for planarization processes [23]

PAGE 46

28 2.3 Chemical M echanical P lanarization (CMP) Chemical mechanical planarization (CMP) is the process of smoothing and polishing a surface by the aid of chemical and mechanical forces Presently, CMP is the only technique that offers both local and global planarity on the surface of the wafer. The plasma enhanced chemical vapor deposited oxides have limited capability of gap filling and are restricted in their gap filling ability below patterns having 0.3 m feature size. High density plasma deposited oxides have acceptable gap fi lling capabilities; however, they produce variation in surface topography on the local as well as global level. Even though spin on deposited (SOD) doped and undoped oxides and polymeric materials have acceptable ability for gap filling, the disadvantages and advantages of the CMP technique are listed in Table 2.2 and 2.3 respectively. Table 2. 2 Disadvantages of CMP Disadvantages Comments Multi physics phenomenon Poor control of process variables and variability in consumables leads to fine tuning for proper polishing parameters New Defects New defects from CMP can affect die yield (*crucial for sub 0.25 features) Process developments Endpoint detection difficult to control, therefore need for additional process control and metrology High cost of ownership Costly equipment and consumables with high turnover rate for consumables

PAGE 47

29 Table 2. 3 Advantages of CMP Benefits Comments Planarization Global and local planarization Material selection Metals and non metals Planarization of multi material services (MLM) Achieves pol ishing of multiple materials on same polishing step (* polish rate varies) High surface r emoval rate Can removal extremely rough surfaces globally and locally for tight design restrictions and MLM Metal patterning CMP is an a lternative means of patterning metal eliminating need to plasma etch difficult to etch metals and alloys Metal step coverage Reduction in surface topography reduces metal step coverage Increased IC reliability Contributes to increasing IC reliability, speed, yield (lower defect density) of sub 0.5m and circuits. Reduce defects CMP is a subtractive process and can remove surface defects. No hazardous gases Does a not use hazardous gas common in dry etch process.

PAGE 48

30 2.3.1 Applications of CMP Historically CMP has been used to polish a variety of metals and was generally taken from nature as a polishing method used to produce beautifully finished stones from years of exposure to mild chemicals and mechanical forces of nature. Modern CMP originated fro m methods on polishin g glass for optical devices [51] Manufacture of telescopes, microscopes, eyeglasses and various lenses was well understood and established prior to invention of transistors. IBM developed CMP for the semiconductor community during the 1980s. IBM initially applied the CMP process to the silicon dioxide inter level dielec tric planarization for the integrated circuit industry. CMP is now utilized in planarizing the interlayer dielectric (ILD) and met als used to form interconnections between devices [14]. With the successful implementation of CMP for local and global planarization of silicon dioxide (SiO 2 ), removal of excessive tungsten (interconnect) from the horizontal surfaces on the wafer pattern proved to be an asset for subsequent Al metallization [42, 52, 53] CMP was develop ed with a two fold approach of planarizing oxide and removing the via fill metal from the horizontal surfaces. The major ap plications of CMP are given in t able 2.4 [45]

PAGE 49

31 Table 2. 4 Applications of CMP technology The application of CMP in electronic device fabrication is significant in both memory and microprocessor device fabrication [46] From table 2.4 it can be seen that CMP is emerging in all fields of study including microelectromechanical systems (MEMS) and various other electronic device fabrication. As devices continuously shri nk with the technology advancement in device manufacturing, the output specifications of the CMP process have become more stringent. 2.3.2 History of CMP Zantye et al covered the basic history of CMP emerging as a technology from the an optical lens po lishing technology to the modern day semiconductor multiphysics problem [41] The first semiconductor CMP machine was an innovation of the optical

PAGE 50

32 lens polishing machine. The proper polishing abrasives in presence of the slurry chemicals were used to achieve a superior degree of precision and flatness to meet the d emands of the semiconductor industry. By supplementing mechanical polishing with high hardness abrasives, such as silica in an alkaline medium, there are significant gains in material removal and reduction in the process time. A further improvement to the CMP process was made at IBM in the late 19 70s and early 1980s. The new process was faster than the previous silica based polishing method and resulted in ultra flat, ultra smooth surfaces to meet the stringent requirements of the IC industry [41] The slurry was later tailored to reduce defects and surface non pla narity introduced by the etching and deposition processes. The IBM process was then applied for trench isolation in the late 1980s in Japan for various logic and DRAM devices. There was widespread industrial implementation of different variations of the CMP process by companies like NEC, Nati on Semiconductor, Hitachi, etc. This led to the introduction of the first commercial polisher designed specifically for CMP by Cybeg in Japan in 1988. Later, International SEMATECH identified CMP as a technology cri tical for the future of IC manufacturing and launched a project to develop competitive, advanced CMP tools in the US [18]. Throughout the history of the CMP process there have been advances in the types and capabilities of each polisher. There have been a total of three different generations of needs of the IC industry. The first generation CMP tools based on rotational platen had low throughput values of about 10 18 w afers/hour [54] The second generation of CMP tools emphasized evolutionary improvements to the original designs and the second

PAGE 51

33 generation of polishers included multi wafer platen polishers and sequential rotational systems. The third generation equipment designs were modified to stay in production for long periods of time by giving them adaptability to future technology modifications. The third generation of polishers included sequential linear polishers, orbital polishers, rotary inverted polishers, and pad feed polishers [41, 45] While the details of these polishers are not detailed in this dissertation, the evolution of these polishers has been critical in the variabi lity of CMP to be utilized in many industries 2 .3.3 CMP P rocess Wear is the phenomenon of material removal from a surface due to interaction with a mating surface, either through micro fracture, chemical dissolution, or melting the contacting surface [55] CMP abrasive wear is usually divided into two types: two body and three body abrasion. The situation when exactly two bodies are involved in the interaction is known a s two body abrasion. Two body abrasive wear is caused by the displacement of material from a solid surface due to hard particles sliding along the surface or when rigidly held grits pass over the surface like a cutting tool. Two body abrasive wear is a c omplex process often involving high strain and plastic deformation and fracture of micro volumes of the material, which might be described as the removal of discrete surface by a harder substance which tends to gauge, score, or scratch. In the case of pla stic contact between hard and sharp material and a relatively softer material, the hard material penetrates the softer one causing fracture (e.g., two body abrasion). This fracture can lead to micro cutting and ultimately material removal. In three body a brasive wear the particles from two body collisions or introduced wear abrasives are

PAGE 52

34 free to roll as well as slide over the surface. It is through the wear process that raw materials can be turned into the electronic instruments used every day. The CMP pr ocess shown in figure 2.7 involves mounting a wafer with a thin film of metal or oxide deposited on it onto a spindle. A downward force is exerted on the wafer pressing it onto a rotating polymer pad while a liquid containing colloidal abrasive particles and dissolved chemicals, labeled slurry, is introduced in the space between the pad and the wafer. Figure 2. 7 Schematic of CMP p rocess This process employs both solid on solid (e.g., pad on wafer surface two body abrasion) and solid on liquid (pad, wafer, with abrasive nanoparticles) three body abrasion wear to polish the surface of thin films to atomic smoothness as shown in figure 2.8.

PAGE 53

35 Figure 2. 8 Three body abrasion on CMP process The slurry, which includes abrasive nanoparticle s, polishes (or wears) the film surfaces by the combined action of chemical corrosion and mechanical removal Therefore the CMP process can be described as a process which uses the combination of mechanical energy from the pad and abrasives and chemical energy from the slurry chemicals to polish and remove material from the wafer surface. The major consumables for bit patterned matrix configurations investigated in this dissertation are the polymer polishing pads and an initial investigation into the slurry abrasives with a novel slurry developed. Details for the pads and slurry will be discussed in subsequent c hapters 2.3.4 Material R emoval M echanism As mentioned above, the CMP process involves both chemical and mechanical components acting in synergy to bring about removal of excess material as well as planarization of the surface topography. It is importa nt to understand the mechanism of material removal during CMP. Studies by Ahmadi et al, have characterized removal during CMP in to four different categories [56, 57] :

PAGE 54

36 1) A brasive wear 2) A dhesive wear 3) C orrosive wear 4) E ro sive wear This section provides insight into the chemical and mechanical aspects of CMP, which are responsible for material removal. 2.3.4.1 Mechanical A spects of CMP The mechanical aspects of the CMP process deal solely in abrasive wear of the mating surfaces, whether it is wafer to pad abrasion or slurry abrasive to pad abrasion. From figure 2.8, two body wear abrasion occurs when the abrasive particles from the slu rry interact with the wafer surface, and also when the pad surface asperities (surface protrusions) slide against the wafer surface. As the roughness of the pad is on the order of microns and the size of the abrasive particles is on the order of nanometer s, a significant amount of the two body abrasion takes place between the pad and the wafer asperities. The interaction of the pad and wafer asperities, and the slurry abrasives leads to three body abrasion Three body abrasions are much more complicat ed than the standard two body abrasion but results in lower removal rates and can lead to a reduction in surface defects In three body abrasion the abrasive particles that come in contact with the wafer asperities are held in place under a given pressur e by the pad asperities. As the abrasive particles are dragged across the wafer surface under pressure applied, ploughing, and cutting processes occur simultaneously resulting in the material removal from the

PAGE 55

37 wafer surface. Figure 2.9 is a simplified gra phical representation of the interface during CMP. Figure 2. 9 Three body abrasion during CMP The relative velocity of the wafer/pad interface aids in the removal process through the momentum transfer of abraded particles from the wafer and the abraded slurry particles as the centrifugal forces force these particles from the interfaces. This erosi ve wear is a function of the fluid motion which is covered in the fluid interfaces modeling section of this dissertation. 2.3.4.2 Chemical A spects of CMP M aterial R emoval chemistry and the ability of the chemically enhanced slurry to modify the wafer surface through corrosive wear prior to abrasion by the particles. The slurry must be able to dissol ve the abraded material, thereby avoiding re deposition of the removed material onto the wafer surface. As CMP was initially developed to polish dielectrics (oxide layers in MOSFETS), the oxide layers were first hydrolyzed by the chemicals in the slurry an d the abrasive particles abraded the surface. The same is true for metal polishing; the difference lies in the chemical make up for the slurries. The abrasives in the slurry

PAGE 56

38 provide another mechanical abrasion to the surface, but also the abrasive partic les bond themselves chemically to the wafers surfaces and remove material as they separate from the surface, through adhesive wear [41, 49] Therefore the abrasives have a chemical role in removal to accompany their mechanical abrasion. In the case of copper CMP, the slurries can be acidic and alkaline in nature [41, 58 61] The metallic copper surface is modified by the active ingredients and pH of the slurry. The ingredients and the pH cause a reaction that dissolves the copper oxides and hydroxides and the rate of oxidation of the copper depends on the particular formulation o f the slurry and concentration of oxidizers and complexing agents of the slurry. The formed surface copper compounds will then be abraded off the surface by the abrasive nanoparticles and the pad asperities. For copper CMP, the abrasive particles only pr ovide mechanical action and the chemical nature of the particles plays no part in the material removal. The abraded copper compounds (from the dissolved oxides and hydroxides and the abraded particles) are carried away along with the dispensed slurry. Th e dissolution is very crucial to the removal process, as it avoids re deposition of the material onto the wafer surface. The material removal rate (MRR) depends equally on the mechanical as well as chemical aspects during CMP. 2.3.4.3 Governing F actors of CMP P rocess The CMP process is a multi physics process with many factors governing the final material removal rate and surface quality. It is a process that is judged by its ability to have local and global planarity. Wafer planarization may be classifie d into three categories of planarity. These are summarized below and shown in figure 2.10:

PAGE 57

39 1) Surface smoothing: feature corners are smoothed and high aspect ratio holes are filled 2) Local planarity: surfaces are flat locally, but the surface height may va ry across the die 3) Global planarity: the surface is flat across the entire stepper field. The requirement for surface smoothing and local planarity comes from metal step coverage, which is defined by the ratio of thinnest point in metal film to th e thickest point in metal film [62] The requirement for global planarity increases when the circuit dimensions reach sub 0.5. Very few planarization schemes obtain the global pla narity offered by CMP due to the stringent restrictions required to meet the depth of field requirements of lithography tools in the sub 0.5 micron regime. Planarization over many micrometers is needed to eliminate metal etch residuals, and planarization o ver several millimeters is required to alleviate photolithographic depth of focus limitations [53] The varying degrees of planarity are shown in figu re 2. 10.

PAGE 58

40 Figure 2. 10 Degrees of planarity in CMP process The governing factors behind the planarity resu lts and the multi physics phenomenon can be categorized into process parameters and consumable characteristics. Table 2.5 gives an overall list of aspects that govern the output of the C MP process.

PAGE 59

41 Table 2. 5 Factors governing output of CMP process Process Parameters Consumable Characteristics 1) Load applied (Psi) 2) Angular velocity Polishing Pad Wafer carrier 3 ) Slurry flow rate 1) Wafer Contour and size Bit pattern density Pattern dimensions Chemical compatibility of underlying layers to slurry components 2) Pad Bulk characteristics Surface characteristics Groove design Groove dimensions 3) Slurry pH / Zeta potential Particle size and distribution Additives Oxidizer and concentration 2.3.5 Process P arameters The a pplied pressure and velocities of the wafer carrier and polishing pad are the most crucial machine process input variables which impact CMP performance. From aw adopted from glass polishing, pressure and velocity during polishing dictate the re moval rate during the process [51] The pressure and velocity also dictate

PAGE 60

42 the friction characteristics at the interface and determine the regime of lubrication as de tailed by the Stribeck curve [63] High pressure and velocities result in high shear forces applied on the wafer surface, which can induce delamination, or peeling of the deposition layer from the silicon substrate, at the weakly adhered surfaces involving ILDs [64] The other process parameters that have an effect on the removal rate include slurry flow r ate and pad surface temperature. T hese factors have been previously investigat ed by Mudhivarthi and Zantye [41, 46] 2.3.6 Con sumable C haracteristics The characteristics of the wafer being polished have a signific ant effect on the CMP process. The size and shape of the wafer determines the contact area and the polishing uniformity across the w afer surface during polishing. The wafer shape also changes the interaction between pad and the wafer at different pressure settings resulting in a change in contact dynamics at the surface [65] Looking at the die level, the pattern density and dimensions along with the layout of the pattern in a die affects th e uniformity within the die resulting in change in uniformity of polish and generation of post CMP characteristics [66] Th is process of pattern density and uniformity becomes important in the magnetic storage hard drive community as the flight height of magnetic head and the ability to access memory is dependent on the pattern density and planarity of the surface. The polis hing pad is consumable of CMP, which provides a major part of the mechanical component to polishing with a 550 million dollar economic impact. The asperities, or roughness of the pad surface, directly determine the contact area for material removal (this is also dependent on the pressure applied). A secondary function of the pad asperities is to prevent the abrasive particles of the slurry from sliding off the

PAGE 61

43 pad due to centrifugal forces of rotation and to have an efficient pad wafer contact. The abras ive particles, which are held at the contact by the pad asperities, are the only particles available to provide active mechanical component during CMP. Typically a polishing pad is constituted of two sections: the top section of the polishing pad, which c onsists of grooves and surface asperities, and the bottom bulk portion of the pad, which supports the upper portion and helps in achieving polishing uniformity [67] The two sections of the pad are either fabricated together or applied together for specific CMP applications. The polishing pad is fundamental in material removal through two body abrasion and must also transport the slurry effectively to the polishing surface [68] The dimensions of the pad groove, such as the width and depth of the groove, along with the groove pattern are also important to have a uniform slurry distribution on the pad surface. Out of all the consumables the slurry and its chemical constituents have the most influential economic impact with profits and sells above two billion dollars a year [69] CMP slurry is comprised of oxidizers, complexing agents, abrasive particles, and The slurry plays a critical role in modifying the surface being planarized, abrading the surface, and also dissolving the abraded debris. The concentrati ons of its various constituents significantly influence the output of CMP. The particle size distribution, zeta potential, uniform dispersion, and other characteristics need to be maintained and monitored continuously to avoid formation of agglomerated pa rticles or chipped particles. If the slurry characteristics are not closely monitored the wafer surface will end up being severely scratched, hampering the device yield and impacting the overall performance of the device. The slurry also acts as a barrier chemical selectivity layer for the different

PAGE 62

44 underlying layers (ILD and barrier layers) that are not supposed to be polished. The barrier layer is created by utilizing suitable additives and pH conditions. Consequently this is where much of the research in polishing has been focused. Successful implementation of the CMP process significantly depends on optimizing the process parameters and selections of the consumables to ensure product performance and reliability. 2.4 Tribology of CMP P rocess In ord er to truly understand the material removal mechanisms and the output during the CMP process, understanding of the tribological aspects of polishing is fundamental. Tribology is the science and engineering of i nteracting surfaces in relative motion. It i ncludes the study and application of the principles of friction, lubrication and wear. It is easy to ascertain that CMP is a tribology process from the applications of two body and three body abrasion, to the friction regimes during polishing, and the res ulting material removal or wear during polishing. 2.4.1 CMP P rocess CMP is an abrasion process which involves rubbing of wafer and pad surfaces in the presence of chemical slurry and abrasive particles. During the CMP process, low friction and efficien t lubrication are desirable, but the optimization of the process focuses on highly controllable material removal as well as a great surface quality. Since the pressure and velocity are the major contributors to the removal process during CMP it is evident that the frictional forces would play a pivotal role in understanding and improving the CMP process. There are various analytical and theoretical models to predict frictional and removal characteristics during lubricating sliding contacts which is

PAGE 63

45 closel y related to the CMP process [70] The interface during CMP process is much more complicated as compared to the studies on the frictional characteristics considered in the aforementioned models. The models fail to incorporate the abrasive particles f or three body wear and the chemical component of the slurry during polishing; therefore the applications of these models find minimal application in CMP predictions The best way to understand the interfacial characteristics (dynamics, material removal, a nd surface tribology) is to study the frictional characteristics in situ or during the process. One of the most influential parameters for the CMP process is the coefficient of friction during polishing. 2.4.2 Tribo M etrology of CMP The coefficient of friction (COF) and the contact acoustic emission signal (AE) are crucial in characterizing the friction characteristics of a system consisting of sliding surfaces. The study of these parameters along with the wear rate of the surfaces and pad wear are ter [71] 2.4.2.1 Coefficient of F riction The coefficient of friction is the r atio of the tangential force that is resisting motion to the normal load [55] The numerical co efficient representing the fric tion at the polishing interfaces reflects the nature of abrasive wear (two body, three body wear). The COF is influenced by several parameters including the materia l properties of the properties, the kinematic parameters of the polishing process, slurry viscosity, and chemical properties of the slurry and its ability to alter the surface.

PAGE 64

46 As mentioned previously the COF is used to generate Stribeck curves which offer an efficient means to monitor tribological processes [63] The Stribeck curves sho wn in figure 2.11 are generated using COF data and the Sommerfeld number. Figure 2. 11 Stribeck curves generated from COF data [33] The COF and Sommerfeld number are used to determine the lubrication regime at the polishing interface. The Sommerfeld nu mber is defined in equation (2 .2 ), (2.2) w here is the viscosity of the lubricant, U is the relative velocity, p is the applied eff is the effective lubricant film thickness. The Sommerfeld number can be calculated since the process parameters dictate the pressure and velocity. The viscosity of the slurry is calculated based on the sp ecifications while the effective lubricant film thickness is the variable that does not remain constant and is estimated to be pad surface roughness based on literature [72] To account for deviations of the slurry film thickness on different grooved pads, a dimensionless factor has also been suggested.

PAGE 65

47 Based on figure 2.11 there are three different major regions of lubrication that can contribu te or negate material removal, boundary lubrication, mixed or partial lubrication, and hydrodynamic lubrication at a lubricated frictional interface. Studies have shown that there are minor regimes such as hyd rostatic and elastohydrodynamic lubrication regimes that are not investigated in this thesis due to their small contribution to the theory behind the removal rate [55, 73] 2.4.2. 2 Boundary L ubrication Boundary lubrication consists of two body abrasion on solid solid contact between the wafer and pad during boundary lubrication, where the removal process is dominated by surface abrasion. In this regime, polishing results in severe surface damage due to the aggressive abrasion by slurry particles and the polishing pad. 2.4. 2. 3 Mixed L ubrication The mixed lubrication regime consists of a thin film of slurry which partially supports th e applied pressure, and thus prevents the aggressive abrasion seen from boundary lubrication which has no lubrication. For optimization proposes the CMP process should be conducted in this regime to ensure reduction in surface damage. 2.4.2 .4 Hydrodyn amic L ubrication The hydrodynamic lubrication regime is a mode of polishing res ulting from the entire applied pressure being supported by the interfacial slurry fluid layer. This will result in a very low COF, and therefore a very low removal rate as there is little abrasion. Therefore knowledge of the polishing regime is high ly beneficial to understanding the polishing process.

PAGE 66

48 The importance of the process parameters becomes apparent based on the lubrication regimes for polishing. Too little slurry and the surfaces will be abraded with high COF and thermal stresses resulting in a highly polished material and high surface damage. If there is too much slurry they hydrodynamic lubrication regime will dominant and the result surface wi ll have poor material removal. Optimization of the process parameters in pat terned media configurations is detailed in Chapter 3. 2.4.3 Acoustic E mission The AE signal is a caveat of the COF and can be monitored during the CMP process. The AE signal i s an estimate of the acoustic energy dissipated at the interface due to the mechanical interactions of sliding surfaces and abrasive particles at the interface. The shear generated by the down pressure and platen rotation brings about a strain in the thin film that is being polished and thus is also responsible for material removal. If the shear force is sufficient enough to overcome the interfacial adhesion of the thin film and the buried layer, the interfacial adhesion energy is dissipated in the form o f acoustic vibrations. Higher AE signals indicate intense mechanical interactions or aggressive abrasion at the interface and lower signal indicates a smooth, mild polishing resulting in lower shear forces and less damaged wafer surface. A noisy signal c ould indicate the presence of slurry agglomerates or delamination at the interface which is beneficial in knowing when to stop polishing 2.4.4 End P oint D etection The end point detection parameters of the CMP process ensure that the goal of the process f or wafer uniformity, removal rate, and removal stability are within process specifications. In line and/or in situ metrology that can assess th e polish quality of the

PAGE 67

49 CMP polishers and product wafers immediately can reduce the wafer test time (production time that ensures process specifications are met by polishing first few wafers empirically). There are numerous ways to determine the end point detection (EPD) of the CMP process. A summary of the end point detection methods is listed in the table 2.6 Table 2. 6 End point d etection methods Methods To determine EPD Techniques Optical Interferometry, reflectance, spectral reflectivity Electrical Friction sensing Impedance and conductance (non friction sensing) Acoustic sensing Generation of acoustic signal from abrasive grinding process From table 2.6, most of the methods involve monitoring a signal which contains a signature indicative of an appropriate stopping point. T he optical method s utilize a variation of techn iques involving light to determine if process parameters are met. While the optical techniqu e was the first utilized, it had the limitation of having to add an extra tool to the CMP process altering the overall effect of the CMP process to take measuremen ts. E lectrical EPD systems fall into two subcat egories, systems that monitor COF and those that do not. The systems that do not sense friction have typically been proven unsuccessful outside research premises [74] The components f or the systems that do not monitor COF are intrusive to the process and require electrical connections to the wafer during CMP or modifications to the platen/carrier which a ffect the performance of the

PAGE 68

50 tool itself. The methods that measure the COF are passive and have provided solid reliable and viable data included in this dissertation. 2.4.4.1 EPD COF and AE S ignal During the process of CMP, work is done on the wafer by the pad and wafer carriers. This work is done by friction. A s the two surfaces must pass across each other and friction is resisting that motion and causes material removal. In metal CMP, the materi al removal will eventually lead to the exposure of the underlying ILD which has a different COF than the metal. The EPD for these systems is based on monitoring the changes in the motor current to infer the state of the friction between the wafer and the p ad [74] As mentioned before the concept behind the AE detection is that the grinding action that takes place during polishing generates an acoustic signal. The challenge in AE EPD is that the signal must be demodulated to yield information about the polishing process. The demodulating methods involve detecting and analyzing the amplitude and frequency of the spectral peaks or the acoustic wa ve velocity. The COF and the AE can be constantly monitored and recorded to determine EPD of the polishing process [45] Changes in the AE and COF are indications that a new polishing surface interface has been encountered due to the change in properties compared to the film that has been removed. Figure 2.12 is a graphical depiction of raw data acquired for copper CMP, the transition to the dielectric layer is shown as the signal increases.

PAGE 69

51 Figure 2. 12 AE signal to CMP polish layer change From figure 2.12 the copper layer is deposited onto the barrier tantalum layer and then polished as the AE signal transitions the onse t of polishin g of the ILD SiO 2 layer is evident. Either in situ or post CMP analysis of the COF data allows for the calculation of the time to remove a particular layer such as copper, and allows for calculation of MRR [75] Careful monitoring of these parameters can ensure that the samples are not over polished and will avoid surface de fects. By running a variance sequential probability ratio test (SPRT) on the COF signal data, the EPD can be detected more effectively [75] This will allow for not only the removal rate and EPD to be determined but also the uniformity of the polish can be estimated from the COF data. This can be d one by monitoring the time for the COF transition from one end point to the next end point level. The longer the transition time, the higher the non uniformity [76] The same SPRT analysis can be done on the AE signal to determine process induced defects such as delamination and generation of m icro scratches

PAGE 70

52 2.5 Modeling in CMP It is evident from previous sections that CMP is a multi physics problem that incorporates multiple disciplines such as; chemistry, fluid mechanics, particle dynamics, solid mechanics, and physics which all combine to contribute to the removal at the wafer pad interface The lack of detailed knowledge on these effects on the 2 billion dollar semiconductor and magnetic hard drive industry has the driven industry to empirically tune the process This has lead to sev eral process mod els developed to help to optimize the process and predict the MRR and surface quality after polishing. 2.5.1 Preston M odel The most basic and referenced model to describe the CMP process was first proposed by Preston for glass polishing [51] From section 2.3.1, the polishing of silicon dioxide, which is a form of glass is approximated by P und in equation ( 2. 3): (2.3) where K p is the Preston coefficient which incorporates several unknown variables such as chemistry effects in the CMP process, P is the pressure applied, V is the relative velocity of the wafer/pad interface, and H is the hardness of the wafer surface. equation states that the material removal rate (MRR) is directly proportional to the pressure and the relative velocity and inversely proportional to the hardness of the wafer surface. global MRR [77] From equation (2.3) the pressure applied is shown in equation (2.4) :

PAGE 71

53 (2.4) where L is the load applied, and A is t he contact area on the pad. The contact area is not necessarily the geometric area or the actual area surface, because wafer surfaces which are mostly patterned have severe topographies. In these cases the assumed contact area will not be the geometric a rea of the wafer being polished. This caveat has lead in the surface topography based solely on the mechanical interactions of the wafer pad interface. 2.5.2 Mechanical M odels A numbe r of sophisticated wafer surface wear models have been developed to account for various physical phenomena that take place during CMP. The mechanical models that have be en de veloped can be broken into two major categories empirical wear studies and conta ct mechanics models 2.5.2.1 Empirical W ear Modeling S tudies The models that have employed an empirical wear modeling approach, typically account for the pressure and velocity components in the MRR and reference directly back to the Preston equation [77 82] The empirical models do not of ten take into account all of the multiphysics phenomena in the process, and historically th ese models negate the slurry lubrication regime, the abrasive particle dynamics in the system and the che mical effects of the slurry. D ue to these limitations mode ls developed for one tribosystem may not be applicable to a different tribosystem [77]

PAGE 72

54 2.5.2.2 Contact M echanics M odeling The studies that have taken contact mechanics approach toward the CMP analysis assum e that the wafer and pad surfaces are in direct sliding contact during the CMP process. These models predict that all MRR is due to the pressure and velocity process parameters. determine the wear parameters, and the equation has been proven to be non linear although the major parameters in the removal process remains unchanged [83 86] Understanding of the contact models used in many approaches to CMP involves understanding of basic elastic contact Greenwood and Williamson provided a theory of elastic contact between two mating surfaces that a number of CMP mechanical models incorporat e. The Greenwood and Williamson (GW) model and other statistical based models represent the contacting surfaces (pad or wafer) as a probability distributions function (PDF) of surface heights, and by using the PDF calculate the number of asperities contac ting the surface. The reaction force (pressure) between the two forces is calculated and utilized to determine MRR. The weakness of the statistical models is in the model s inability to accurately determine the contact area distribution as a PDF is used to estimate the surface topography [87 90] To alleviate this approach researchers have used a deterministic approach in which the actual pad and surface topographies are found using metrology tools. Since the deterministic models are based upon the geometry of the contacting surfaces, it allows for both forces response and contact area distribut ion to be predicted [91 95]

PAGE 73

55 Bot h methods of modeling exploiting only the mechanical aspects fail to incorporate the polishing regime, slurry abrasive particles and their interaction with the wafer surface, or the fluid mechanics that play a role in MRR. 2.5.3 Fl uid M echanics M odels A set of models has analyzed the CMP process based solely on fluid me chanics modeling of the wear. These approaches assume that the wafer and pad surfaces are completely separated by the slurry and fall into the hydrodynamic lubrication regime. Nanz and Camilletti and Steigerwald et al noted the importance of the slurry flow field in the overall CMP process as well as the need for in depth understanding of the slurry flow at the wafer/pad interface [59, 92, 96] Many of the fluid based models incorporate the GW model to solve for a contact stress distribution across the wafer and solve for the film thickness to input into the Reynolds equation, a simplified form of the N avier Stokes equation 2.5 The Reynolds equation is used to relate the slurry pressure field, film thickness distribution, and shear rate to the CMP process Reynold s 1 D equation shown in equation 2.6 is used to analyze film thickness and pressure dis tribution of viscous fluids through small gaps (e.g., wafer/pad interface) The full Navier Stokes equation is shown in equation 2.5, (2.5) w here v is the flow velocity, is the flu id density, p is the pressure, is the deviatoric stress tensor, and f represents body forces (per unit volumes). Equation (2.5) is

PAGE 74

56 simplified for Newtonian fluids, incompressible, constant viscosity, stea dy state flow, in equation shown in equation 2.6, ( 2.6 ) w here p is the hydrodynamic pressure, h is the local film thickness, is the dynamic viscosity of t he slurry, U is the relative velocity of the bottom surface (pad), and x is the downstream distance. The slurry pressure distribution is key parameter in the fluid mechanics modeling of CMP because it dictates whether the wafer and pad surfaces are in con tact (negative pressure) or are completely separated by the fluid (positive pressure). Results from the solutions of the Reynolds equation are then combined with mass transport theory in order to predict the material removal rate distribution over the sur face of the wafer. Most of the aforementioned fluids modeling studies neglected the effect of the abrasive particles on the rheology of the slurry. S tudies assumed that the wafer was fixed and the pad was the only rotating surface, which is not the case in the industrial process in which the pad and wafer rotate about different axes. These studies do not account for the wafer bending, pad deflection, slurry particle entrapment on the pad, or the 2.5.4 Hybrid M odels (PAML) In order to fully incorporate all of the physical aspects of the CMP process several hybrid models have been developed to try to successfully predict process outcomes. These models have combined the contact mechanics and abrasive particle into one model, while other studies have integrated the fluid mechanics and contact mechanics into their

PAGE 75

57 predictions [78, 81, 82, 97 99] While these studies included one or two of the main physical aspects the model by Terrell and Higgs captures all combinations of the physics of CMP. This model contains the fluid continuum modeling for the slurry flow and pressure distribution the deterministic contact mechanics modeling allowing for true forces response and contact area distribution to be predicted, particle dynamics modeling of the abrasive par ticles in the slurry, and first principle wear modeling. This model was unique in the fact that it uses first principle approaches to model the C MP process, including the actual measured surface of the sample to be polished, the model has the variability to change the micro scale input parameters, and it is one the first models that allow for greater prediction of local wear phenomena. The model d id not incorporate surface properties and characteristics and the analysis of the hardness variation within the PAML was investigated and discussed in subsequent chapters. 2.6 Challenges D uring the CMP P rocess Although CMP has become the choice for loc al and global planarization, it also comes with its own inherent challenges as with any planarization process. The challenges for the process start with the number of materials utilized for copper/low k integration. Throughout this research project the ch allenges faced are: 1) L ocal and global non planarity 2) E tching and contamination 3) Microscratching 4) Delamination

PAGE 76

58 The following sections will detail the challenges faced when polishing bit patterned matrix configurations. 2.6.1 Non P lanarity The post polis h planarity is a result of several factors and cannot generally be limited to one factor. The etch and dissolution rate of the slurry, pad asperities contact area, and non uniform pressure distribution on the patterns with line width and density issue are a few of the issues [100, 101] Figure 2.13 shows the various post CMP surface characteristics that result due to the factors at different pa ttern densities and line widths. Figure 2. 13 Cu CMP surface characteristics

PAGE 77

59 From the figure 2.13 the second and third step prove to be the most critical in compared to the global planarization step. There are four major surface defects that result in a surface that deviates from a planar surface: 1) Dishing 2) Erosion 3) Oxide loss 4) Metal loss 2.6.1.1 Dishing and E rosion A schematic illustration of dishing and erosio n defects is shown in figure 2.14 The dishing effect is characterized by high polishing rates in localized regions where the pattern is significantly different from its surround ing. The formation of trough shaped dish has been attributed to excessive over polishing in these areas and an efficient EPD helps to prevent this defect. Figure 2. 14 CMP challenges 2.6.1.2 Oxide and M etal L oss During copper CMP, a minuscule amount of over polish ing is required to remove all metallic residues on the dielectric surface to ensure electrical or magnetic isolation between adjacent components. Oxide loss is the loss of the field oxide next to an array of

PAGE 78

60 th in metal lines separated by a wide oxide pattern [102, 103] Met al loss is the total loss of thickness of metal lines separated by a thin oxide pattern. All of these post CMP characteristics affect the electrical properties of the interconnect structure, chip reliability, and induce non planarity over the wafer surfa ce, which in turn causes photolithography issues negating the initial use of CMP. The process parameters and consumables that cause these characteristics need to be understood in detail to avoid reduction in device yield. 2.6.2 Surface S cratches The mechanical interaction of the abrasive particles with in the surface of the wafer (two body abrasion) leads to plastic deformation and surface damage to the wafer surface. Although the plastic deformation is necessary to complete the fundamental job of CMP process scratches, both macro and micro, can form due to the deep indentation and dragging of the abrasive particles as seen in figure 2.15 Figure 2. 15 Optical image of surfaces scratches from CMP Many factors such as p article size distribution, and formations of agglomerates due to slurry pH are possible factors contributing to surface scratches. A number of the surface scratches can be removed during the final buffer polishing, but the deeper

PAGE 79

61 scratches are permanent and cannot be removed from the surface. As with dishing and erosion the surface scratches reduce the usable yield of the wafers and impact electrical and magnetic schematic designs. Development of novel slurries to reduce scratches while maintaining per formance continues to be a driving factor in research and is presented in chapter 5 of this dissertation. 2.6.3 Delamination In conjunction with replacing aluminum with copper, another way to reduce RC delay is introduction of materials that have lower dielectric constant (low k) than SiO 2 The drawback of introducing low k materials is that th ey are mechanically weak materials [46, 104 109] The low k materials cannot withstand the shear forces applied by the shearing motions of the platen and wafer carrier during CMP. Their interfacial adhesion energies are low and even m oderate frictional forces can induce failure of these interfaces [110] Th e d elamination shown in figure 2.16 has not been attributed to one factor but studies have proven that in order to reduce the delamination low pressures and velocities are ideal to during the CMP process.

PAGE 80

62 Figure 2. 16 Delamination from copper CMP Reducing the press ure and velocity leads to an increase in the number of polishes needed to complete a cycle for an individual wafer and an increase in processing time ultimately costing industry money. Therefore fundamental studies are under current investigation to help understand the delamination phenomena. The aforementioned defects destroy the devices and increase the cost of production. These defects along with post CMP characteristics such as dishing and erosion need to minimized if a successful implementation of the process is to be achieved [46] Yield and throughput of the polishing process is highly dependent on the process parameters and consumable characteristics which is the focus of this dissertation. 2.7 Conclusion R esearch O bjectives The CMP process is a multidisciplinary problem in which the output metric are not direct ly tied with any one input metric. As shown in figu re 2.17 the process must be broken into several sub factions and investigated individually in order to fully understand

PAGE 81

63 Figure 2. 17 CMP process factors F ig ure 2.17 shows the CMP process broken into its individual subsections. The bolded subsection s are detailed and outlined in this dissertation. All of the output factors are characterized based on the input parameters. The objectives of this research are to broken into phases and are as follows : The first phase of this research will provide: 1) Provide benchmark data CMP on BPM configurations 2) To optimize the MRR and surface quality based on input parameters 3) Evaluate mechanical properties evolution during CMP of BPM

PAGE 82

64 4) Complete a statistical analysis on the input paramete rs to determine which parameter in paramount in BPM CMP. Following optimization of the machine parameters Phase II will: 1) Investigate pad wear on BPM CMP 2) Perform a p arametric study of pad wear, pad roughness, COF, MRR, surface morphology 3) Give a qualitative analysis on the pad life, surface characteristics of the wafer, pad, and polishing regime 4) Determine optimal polishing pad for CMP of BPM from three commercially a vailable pads. Finally the optimized machine parameters and optimal pad ar e used in phase III which will: 1) Develop and investigate new nanodiamond (ND) slurry for BPM CMP 2) Determine the MRR and surface quality based on the new ND slurry. 3) Compare and contrast the novel ND slurry versus industrial slurry CMP This disser tation also contains a separate investigation into m icrostructu r al variation in me chanical properties during polishing and evolution of the MRR and surface roughness due to the variation. A model is developed to predict the evolution for standard abrasive grinding process and the resulting surface qualities and MRR are reported. The model u ses deterministic pad and wafer surface s to check validate the PAML model developed at CMU, based on incorporation of microstructural variation.

PAGE 83

65 C HAPTER 3 : CMP PROCESS MACHINE PARAMETER OPTIMIZATION 3.1 Foreword Patterned media data storage technology aims to increase are al storage density by utilizing chemical mechanical planarization (CMP ) as a planarization technique to create atomically smooth surfaces f or the read/write head to fly across. Due to the novelty of the patterned media process there is limited data on the planarizing process of the patterne d media (PM) structures. This phase of research focuses on improving the output parameters of material removal and local and global planarity, based on the machine input parameters of pressur e and velocity from figure 2.17, and detailed in figure 3.1. Figure 3. 1 Machine process parameters optimized for PM polishing These machine parameters have been proven to be the main contributing factors in material removal and surface quality for integrated circuit CMP, and therefore shall be investigated in this chapter A secondary analysis is performed based on the evolution of th e mechanical properties during polishing, while a tertiary statistical analysis of variance (ANOVA) on the fundamental machine input parameters is analyzed and compared against Preston equation for MRR on patterned SEMATECH copper wafers.

PAGE 84

66 In this resear ch, bit patterned matrix (BPM) SEMATECH media samples of Cu are polished at various pressure and speeds and the material removal rate and surface quality are analyzed as an initial case study on the feasibility of CMP for patterned media fabrication. The pre polishing and post polishing hardness and elastic modulus were obtained through nanoi n dentation and d iscussed in further detail The purpose is to optimize and benchmark data on CMP of patterned media configurations. In particular the goals of this r esearch are as follows: 1) Provide benchmark data CMP on BPM configurations 2 ) To optimize the MRR a nd surface quality based on input parameters 3 ) Evaluate mechanical properties evolution during CMP of BPM 4) Complete a statistical analysis on the input parameters to determine which parameter in paramount in BPM CMP

PAGE 85

6 7 3.2 Patterned M edia D ata As described in chapter 1, the data storage industry aims to increase areal storage density by using p atterned media (PM) as a new storage technology. PM uses advanced semiconduct or processing techniques to fabricate nano magnetic structures for the purpose of isolating individual grains for magnetic domains into regular patterns [26, 27] This technology would allow for storage of one bit per cell or grain as shown in figure 1.4 in chapter 1 PM is different from conventional longitudinal drives shown in figure 3.1 where each bit is stored acros s a few hundred magnetic grains. Figure 3. 2 Conventional longitudinal magnetic storage As recently as A ugust 2010, Toshiba successfully created a 2.5 Tbit hard disk using PM. The fabrication of the disk is the first successful PM hard disk created at the time, but it lacked the functionality of writing or reading data on the disk due to re ad/write head fly height issues [17] Without proper understanding of the prominent tribological issues that exist in the fabrication and successful operati on of the read/write head in patterned media, this technology will contain r em ain in research laboratories. Conceivably, a patterned media disk drive will consist of a magnetic slider head that

PAGE 86

68 reads and writes information onto a spinning disk during flig ht. The Toshiba disk was comprised of micro to nanoscale magnetic structures that must be planarized to prevent lateral collision between the slider and coarse topography on the disk surface as shown in figure 3.3 Figure 3. 3 Read/write slider head on PM Fabrication of patterned media dat a storage devices utilizes semiconducting manufacturing techniques. These techniques include and are not limited to electrodeposition, evaporation and liftoff, etching processes and chemical mechanical polishing (CMP) for planarization [26] As discussed in chapter 2 the CMP process, shown in f igure 3. 4 is a vital interim fa brication step for integrated circuits (IC) and data storage devices where it is used to planarize thin film sur faces down to atomic smoothness.

PAGE 87

69 Figure 3. 4 CMP p rocess The material removal rate (MRR) affects the surface top ography and thereby performance and reliability. The MRR corresponding to CMP is gi ven by the rudimentary Preston e quation, which contains the load applied, the relative velocity of the pad to the wafer carrier the Preston coefficient that includes chemical dependencies, an d the hardness of the material. Tribological MRR models mainly account for the mechanical removal [ 99, 111] and weakly account for the chemically induced removal [66, 112] by using a sophisticated form of the well known Preston e quation (3.1) ( 3. 1) where k is the Preston coefficient which accounts for the chemical and mechanical removal based on polishing experiments, P app is the applied pressure on the w afer, U is the relative velocity between the wafer carrier and polishing pad, H is the hardness of the material being removed, and MRR is the material removal rate (typically in nm/min). Since CMP is predominantly used for semiconductor IC applications, t here is limited data available related to the polishing of thin films in PM configurations for

PAGE 88

70 advanced data storage applications. Therefore, this effort will provide key benchmark data from CMP patterned media experiments that will answer questions about the viability of producing a tomically smooth PM by CMP 3.3 Experimental 3.3 .1 Candi date S amples Unless otherwise noted, a ll processing and characterization steps were co mpleted in the Nanotechnology Research and Education Center (NR E C) at the University of South Florida. The wafer s unless otherwise noted were purchased from SEMATECH Inc. type copper patterned SEMATCH wafers, the wafers first had to be diced into their individual repeating patterns for polishing and mapping purposes Prior to dicing a thin layer of SU 8 negative photoresist was dep osited using the laurel spinner on the SEMATCH wafers in order to ensure no oxidation or mineral deposition on the wa fers during the dicing process. The wafers were diced along the individual dicing axe s and resu wafer coupons. The photoresist was removed using acetone for characterization and experimentation. The wafers utilized a MIT 853 pattern deve loped and fabricated by Park et al shown in f igure 3. 5 [113 115]

PAGE 89

71 Figure 3. 5 MIT 854 pattern s ublevel The Cu/low K CMP test mask set consists of three layers and thus three masks M1 mask, via mask, and M2 mask enabling the study of multilayer issues in Cu CMP. However, the M1 mask itself is purposely designed to be efficiently used for characterizing single layer polishing behaviors. This M1 single mask contains al l of the relevant structures for probing ele ctrical or magnetic b ond structures all within the same M1 layer [116] Details of the pattern specifications and fabrication are not incorporated within t his research but can be found in the references [116] This test mask design is concerned with the following aspects of copper chemical mechanical polishing (CMP) with either conventional oxide (e.g., SiO2) or low K dielectrics as the recession layer : 1) Intralevel Metal 1 (M1) polishing pattern effects resulting from various pattern factors created by combinations of different line wi dths and line spaces (e.g., density, pitch) and combinations of structures

PAGE 90

72 2) Interlayer multilevel effects of polishing pattern effects with non uniform topography on a layer below; effects on Metal 2 (M2) polishing of surface topog raphy generated by M1 polishing 3) Intralayer (lateral) and interlayer capacitance and resistance variations from polishing non uniformity (e.g., dishing and erosion) [116] The MIT 854 samples are fabricated in a reversed patterned media configuration which is different from that of the patterned media configurations mentioned in chapter 1 as shown by the gold me tal representation in figure 3.6 Figure 3. 6 Patterned media configurations The planarization of repeating patterns on a Cartesian grid is fundamentally similar to conventional PM and the interest of this dissertation lies in the fundamental science with CMP of these types of heterogeneous matrices T herefore through the remainder of the experimentation the reversed configuration will be notated as a bit patterne d matrix (BPM) configuration. Table 3.1 gives the parameter s for the BPM characterized samples.

PAGE 91

73 Table 3. 1 Metrics for SEMATECH s amples Characteristic BPM SEMATECH Sample Parameter Value: English (metric) Diameter 6 in .0009 (152.4 mm) Length 1 in .0039 (25.4 mm) Width 1 in .0039 (25.4 mm) Initial film thickness 3.937 E 05 (1.02 m) Figure 3.7 is an optical image of the BPM of the MIT 854 pattern at two differen t magnifications, and figure 3.8 is a focused ion beam (FIB) image of the patterned media configuration of the wafers. Figure 3. 7 Optical microscope images of BPM at 30x and 500x

PAGE 92

74 Figure 3. 8 FIB image of patterned media configuration Figure 3.8 was taken at a magnification of 2000x with an accelerating voltage of 30 kV. From the figure s above the individual bit pattern matrix (BPM) is shown in detail. This matrix configuration is analogous to the magnetic patterned media storage configurations utilized by the magnetic storage indu stry. The copper BPMs utilized throughout this dissertatio n were fabricated using some of the same semiconductor fabrication techniques as magnetic storage industry. 3.3 .2 Mechanical P roperties The mechanical properties of the thin film being polished pl ay an important role during CMP. The surface scratches, being one of the most critical aspects that determine polishing performance depend on the mechanical properties of the sliding surface. The harder the surface being polished the harder the abrasive needed to polish the surface. This relationship can lead to scratches, cracking, dishing, and erosion of the underlying material during polishing. Following the removal of the photoresist, the mechanical properties o f the BPMs were measured by nanoi n dentation using the Nano Indenter XP (MTS System Corporation Oak Ridge, TN) shown in figure 3.9. The elastic modulus and hardness were taken using a MTS Nano Indenter with Testworks 4 software

PAGE 93

75 Figure 3. 9 MTS Nano I n dente r XP Nanoindentation is similar to conventional hardness tests, but is performed on a much smaller scale using very sensitive load and displacement sensing equipment. The force required to press a three sided Berkovich shaped diamond indenter into th e candidate samples is recorded as a function of indentation depth. The load and unload displacement data obtained in the nanoindentation tests were analyzed according to the method of Oliver and Pharr [117, 118] An example of a standard nanoi n dentation curve i s sho wn in figure 3.10. Figure 3. 10 Typical nanoi n dentation curve [119]

PAGE 94

76 The stiffness and thereby modulus of elasticity is calculated from the unloading portion of the load displa cement curve shown in figure 3.10 The continuous stiffness measurement (CSM) technique was used for measuring absolute and depth dependent hardness and modulus values. Equation 3.2 represents the equation utilize d by most standard nanoi n dentors; (3.2) where S is the contact stiffness, and A is the contact area In order to use equation 3.2 two keys assumptions are made: 1) Deformation upon unloading is purely elastic thus implying that the entire load is recovered and there is no plastic deformation into the wafer surface 2) Contact between the rigid indenter and the sample is modeled using equation The deformation of the sample and of the indenter tip can be combined and given as a reduce d elastic modulus as shown in equation 3.3 [117, 119] ; (3.3) w here E r i and s represents indenter and sample respectively. The hardness of the thin film being indented can be determined as the ratio of the maximum load, P and the area of contact, A equation 3.4 shown below.

PAGE 95

77 (3.4) The depth of penetration fo r the indenter was fixed at ~50 % of the sample film thickness. The calculations of mechanical properties were performed at 50 % of the indentation depth ( e.g. ~25 % of the film thickness) [120 122] At this depth the substrate effects as well as the effect of the surface oxide is avoided when calculating the mechanical properties [123] Figure 3.11 shows an AFM image of the indented samples. Figure 3. 11 Nanoi n dentation into BPM Using the optical camera in the nanoi n dentor recognizable regions were chosen in the MTS nanoindentor for pre and post CMP measurements. A 40 micron by 40 micron indention square was created for statistical averaging of the nine inden ts done per sample. Mechanical p roperty v alues were calculated by averaging a number of separate indentations at various depth specifications. Initially the instrument was calibrated with the standard sample (fus ed silica) provided by MTS along with other single crystal metal samples.

PAGE 96

78 V alues o f the ter CMP have been tabulated in t able 3.2. Details of the mechan ical properties impact on CMP will be discussed in later sections of this chapter Table 3. 2 Nanoindentation results of BPM c opper Patterned Copper Sample Elastic Modulus (GP a) Hardness (GP a) Unpolished 121.657 3.152 1.249 0.121 1 Psi 129.828 1.090 1.440 0.058 3 Psi 135.419 0.752 1.609 0.061 6 Psi 138.594 1.292 1.962 0.208 3.3 .3 WYKO S urface P rofiler The surface topography and planarity of the wafers was measured using Wyko NT9100 surface profiler by Veeco Ins truments Inc shown in figure 3.12

PAGE 97

79 Figure 3. 12 Wyko NT9100 surface profiler Wyko surface profiler systems are non contact optical profilers that use two technologies to measure a wide range of surface heights phase shifting interferometry and vertical scanning interferometry Phase shifting interferomet ry (PSI) mode allows you to measure smooth surfaces and small steps, while vertical scanning interferometry (VSI) mode allows you to measure rough surfaces and steps up to several millimeters high [124] Phase shifting interferometry (PSI) has typically been used to accurately measure previously smo oth surfaces. In phase shifting interferometry, a white light beam is filtered and passed through an interferometer objective and onto the test surface. The interferometer acts as a beam splitter reflecting half of the incident beam on to the reference sur face within the interferometer and the other half to the test surface The beams reflected from the test surface and the reference surface recombine to form interference fringes. These fringes are the alternating light and dark bands that can be seen when the sur face is in focus. Figure 3.13 shows a diagram of an interference microscope.

PAGE 98

80 Figure 3. 13 Interference microscope [124] During the measurement, a piezoelectric transducer (PZT) moves the reference surface a small, known amount to cause a phase shift between the test and reference beams. The system records the intensity of the resulting interference pattern at many differen t relative phase shifts, and then converts the inten sity to phase data by integrating the intensity data The phase data is processed to remove phase ambiguities between adjacent pixels, and the relative surface height can be calculated from the phase dat a shown in equation 3.5 ; (3.5) y ) is the phase data. This technique for re solving surface heights is reliable when the fringe pattern is sufficiently sampled. When the surface height difference between adjacent measurement points is be introduced and the phase data

PAGE 99

81 cannot be reliably reconstructed [124] Thus, conventional phase shifting interferometry is limited to fai rly smooth, continuous surfaces and due to the rougher BPM surfaces encountered during polishing the vertical scanning interferometry (VSI) technique is employed The basic interferometric prin ciples are similar in both techniques: light reflected from a reference mirror combines with light reflected from a sample to produce interference fringes, where the best contrast fringe occurs is typically at the best focus. However, in VSI mode, the whi te light source is filtered with a neutral density filter, which preserves the short coherence length of the white light, and the system measures the degree of fringe modulation, or coherence, instead of the phase of the interference fringes [125] The system scans through focus (starting above focus) as the camera captures frames of interference data at evenly spaced intervals. As the system scans downward, an interference signal for each point on the surface is recorded. The system uses a series of advanced computer algorithms to demodulate the envelope of the fringe signal. Finally the vertical position corresponding to the peak of the interference signal is extracted for each point on the surface. Table 3.3 has a comparison of both modes and the resolutions ranges for scans.

PAGE 100

82 Table 3. 3 Compariso n of VSI and PSI modes in Wyko surface p rofiler Process Parameter VSI PSI Light Neutral Density filter for white light Narrow bandwidth filtered light Scanning and Focus Vertically scans the objective actually moves through focus phase shift at a single focus point the objective does not move Data processing Processes fringe modulation data from the intensity signal to calculate surface heights Processes phase data from t he intensity signal to calculate surface heights Vertical Resolution 3nm single scan <1nm scans averaged 3 A single scan 1 A scans averaged Range 2 mm 160 nm From table 3.3 the range refers to the greatest vertical distance the profiler can accurately measure. Given the high initial roughness of the BPM configurations from the MIT 854 pattern the VSI mode was chosen as the mode used in the WYKO NT9100. The Wyko NT9100 has several o utputs for the roughness of the surface being examined, the average roughness, R a and the root mean squared (RMS) roughness, R q are both displayed in the results throughout this dissertation. R a represents the two dimensional roughness average, the arit hmetic mean of the absolute values of the surface

PAGE 101

83 departures from the mean plane. R a is normally used to describe the roughness of machined surfaces. The effect of a single spurious, non typical peak or valley will be averaged out and have only a small i nfluence on the value. This statistic cannot detect differences in spacing or the presence or absence of infrequently occurring high peaks a nd deep valleys; therefore, it give s no information as to the shape of the irregularities or surface [124] The R a value is useful in determining the global pl anarity of the surface as it is an average of the hills and valleys of the surface. R q represents the root mean square (RMS) roughness, obtained by squaring each height value in the dataset, then taking the square root of the mean. RMS roughness is generally used to describe the finish of optical surfaces. It has statistical significance because it represents the standard deviation of the surface heights. RMS roughness cannot detect differences in spacing or the presence or absence of infrequently occurring high peaks and deep valleys; therefore, these statistics give no information as to the shape of the irregularities or surface. A surface with a high spatial frequency may have the same Rq as a surface with a low spatial frequency, but may behave radically differently. Because height values are squared in the calculation, the RMS rough ness statistics are more sensitive to peaks and valleys than average roughness statistics [124] The RMS roughness will be utilized for local planarity and is also insightful for the global planarity of the surface. Prior to taking any measurements the VSI mode is calibrated against a 10m step hei ght standard that is supplied by VEECO This calibration was done daily and the system would only allow further measurements if the calibration was within 0.5% of the calibration sample.

PAGE 102

84 3 3 4 CETR B enchtop C hemical M echanical P olishing T ester All of the CMP tests ran on BPM in this dissertation were done with a benchtop CMP tester (C ETR Inc, Ca) show n in figure 3.14 The machine process parameters discussed in the chapter are inputs for the CMP process on the CMP tester. Figure 3. 14 CETR b enchtop t ester The CMP tester has several sensors (force sensor, acoustic emission (AE) sensor, and electrical sensor), which are used for in situ monitoring and optimizing of the CMP process. A strain gauge force sen sor (0 2 00 N) can record both vertical and frictional force and the coefficient of friction ( COF ) is monitored during the process. The system is also equipped with a high frequency acoustic emission ( AE ) sensor, which can detect the delamination, endpoint, and debris during polishing. As mentioned in chapter 2 the AE analysis is a powerful technology that can be deployed within a wide

PAGE 103

85 range of usable applications of non destructive testing The AE sensor works on the basis of elasticity of the materials. All solid materials have certain elasticity and t hey become strained or compressed under external forces and spring back to original form (given no plastic deformation) when released. Higher input force s and, thus, the ela stic deformation, results in hig her elastic energ ies If the elastic limit is exceeded a f racture occurs immediately given it is a brittle material, otherwise fracture will occur after a given amount of plastic deformation in ductile materials like copper If the elastically strained material contains a defect, e.g. a welded joint defect, a non metallic inclusion, incompletely welded gas bubble or similar, cracks may occur at heavily stressed spots, ra pidly relaxing the material by fast dislocation motion This rapid release of elast ic energy is what we call an AE event. It produces an elastic wave that propagates and can be detected by appropriate sensors and analyzed. The impact at its origin is a wideband movement (up to some MHz). The frequency of AE testing of metallic objects is usually between 100 and 300 kHz typical values for ultrasound The acoustic emission sensor employed in this tester has a frequency range between 0.5 to 5 KHz. The AE sensor, in conjunction with COF, has been used to detect the delamination, endpoin t, and debris during polishing. The CMP tester can hold a pad up to 15.24 cm diameters. The upper carriage can hold sample wafers up to 40mm X 40mm. The upper carriage is connected to a vertical linear motion system that has a travel length of 150mm and can oscillate on the pad during polishing. The CMP tester is a testing tool and thus the following assumption s are made when different pads, slurri es and materials were evaluated;

PAGE 104

86 1) Due to the lack of uniformity on the surface of the coupon, an average of the material removal rate measured at different points from the center to the edge of each BPM coupon and was assumed to be the MRR 2) During the in situ de tection of MRR using COF the non uniformity of the sample surface brought about gradual tapering of the signals even for blanket samples. The end point was assumed to have occurred when more than 70 % of the material was removed from the surface and fur ther material removal brought significant change in the COF signals Further details of the CMP be nchtop tester are found in previous literature on the tester [126, 127] 3.3 .4 .1 Experimental P rocedure: P rocess P arameters Influence of machine parameters such as down force, relative velocity, coefficient of friction (COF) and material removal rate (MRR) were observed. Based on previous studies and models of the CMP process, relative velocity and pressure were chosen as the influential parameters to vary for the BPM copper CMP [78, 79, 84, 85, 96, 104, 128, 129] The initial downward force applied to wafer from the CETR tester is cal culated using this equation (3.6 ): F= (P*A)/2.2046 (3.6 ) where P is the pressure in P si and A is t he estimated contact the area ( in ) The forces applied to the wafers ran ged from 1 6 pounds per square inch (Psi) After polishi ng the wafers are cleaned with acetone and blown dry with compressed nitrogen to avoid any surface deposition from the slurry or oxidation of the newly polished copper surfaces

PAGE 105

87 (although the oxidation is in evita ble to avoid, reducing the oxidation is the closest solution for wear, surface quality, and mechanical property measurements ) Table 3.4 shows the variation in process parameters for the polishing experiments. Table 3. 4 Process parameters for CETR t ests # Process Parameter Metric 1. Pressure 1, 3, 6 Psi 2. Platen speed 50 300 RPM increments of 50 RPMs 3. Wafer carrier speed 50 300 RPM increments of 50 RPMs 4. Slider movement 3 mm/s 5. Slurry flow rate 75 ml/min 6. Time 120 sec 7. Pad Rodel IC 1000 Suba IV k groove 8. Specimen coupons The polishing experiments were run for 120 seconds and all experiments were repeated five times with a statistical average shown in the results section. 3.3.4 .2 Experimental P rocedure: Material R emo val R ate The initial thickness of the Cu films were obtained using three different methods to ensure that the beginning thickness was accurate as shown in table 3.1. Initially the copper mask was etched using hydrofluoric acid to the silicon substrate, and then the step edge was measured using a Dektak 150 stylus profil mentor by Veeco. A second

PAGE 106

88 measurement was taken by a Cascade 4 Point Probe Station, with a Keithley current source, and a HP digital multimeter shown in figure 3.15 Figure 3. 15 Four point probe station The second measurement captured the sheet resistance at six fixed points on the diced wafer in the specified patterned region. E q uation 3.7 contain s the sheet resistance equation used for calculation of the sample. (3.7) where R s t is this thickness, V m is the voltage, and I s is the current. The v alue of 4.53 is a constant that is equal to the pi divided by the logarithm of 2. This value is derived from equation 3.8 which is the resistivity of thin film layers [130] (3 .8)

PAGE 107

89 The bulk resistivity of copper is 1.68e 6 ohm cm and knowing this resistivity the thickness can be solved for given that all other parameters are inputs. During experimentation the average of nine tests is taken for the final thickness measurement with the error shown accordingly in the graphs The final thickness measurement s of the samples were further verified by a Qua nta 200 3D dual beam focused ion beam (FIB) by FEI. The procedures for these measurements are detailed below in the transmission electron microscopy (TEM) section The FIB me asurements were taken after individual section s to be analyzed by the TEM were It should also be noted that in later chapters the use of the COF for EPD is used to determine the final MRR and by knowing the initial thickness a removal rate can be calculated. 3.3 .4 .3 Transmission Electron Microscopy (TEM ) With the decrease in feature size below the sub micron range, t ransmission electron microscopy (TEM) has become the most important tool for detailed physical failure analysis and material analysis The focus ion beam (FIB) has become a necessary tool uti lized for TEM sample preparation. The samples were prepared using the FIB [131] The technique involves gener ally no sample preparation as long as sample sizes are able to fit inside the FIB specimen chamb er. A metal line of platinum is deposited over the area of interest and a large stair step FIB trench is cut on one side of the area of interest and a rectangular FIB trench is cut on the other side of the area of interes t. Prior to final thinning the sa mple is tilted to >45 and then the bottom, left side, and a portion of the right si de of the are cut free [132] Then the sample is tilted back to its starting position and the specimen is thinned to electron transparency. Since the specimen is to be used for high resolution electron

PAGE 108

90 microscopy (HREM), a final FIB cut is performed 1 2 with respect to the plane o f the specimen surfa ce, therefore the thinnest portion of the specimen lies in the area of int erest [132] The finalized sample is imaged using a Technai F20 TEM by FEI Inc, with an accelerating voltage of 200kV. Results from the TEM imaging are discussed in subsequent sections of this chapter. 3.3 .4 .4 Consumables A si x inch circular portion of a Rodel, Inc IC 10 00 Suba IV A 4 perforated pad was used for polishing. Table 3.5 contains the prope rties of the pad tested for the experiment s Table 3. 5 CMP pad material properties Rodel IC 1000 Suba IV A 4 Parameter Value: E nglish (metric) Diameter Specific gravity 630 800 (kg/cm 3 ) Thickness 50 mils (1270m) Hardness 57 (Shore D) Compressibility 2 25% The pad is conditioned for 20 minutes followed by 1 minute of polishing on a dummy sample, then another 20 minutes of conditioning, followed by another 1 minute of polishing on a dummy sample, and then the final 10 minutes of conditioning of the pad followed by the actual experiments with copper slurry The conditioning of the pad is used to increase the roughness of the pad to help in the material removal process and reduce glazing of the pad The fundamentals of the conditioning process are further

PAGE 109

91 explained in chapter 4. During conditioning and dummy polishing deionized (DI) water is used as the lubricant instead of the slurry mixture used in the actual experiments. The slurry used in these experiments was Cabot iCue 5001which contains 500 nm precipitated alumina oxide abrasive nanoparticles. Nine hundred milliliters of Cabot slurry was combined with 100ml of hydro gen peroxide during polishing of the BPM. A magnetic stirrer was utilized to disperse the mixture and particles in the slurry The abrasive particles are amorphous in nature and the hardness of the alumina oxide particles can lead to scratches on the sur face as discussed in chapter 2. The Cabot iCue slurry is developed for copper CMP and has a low selectivity and etch rate on the silicon dioxide dielectric layer. D ue to priority reasons the distribution, shape, and mechanical properties of the abrasive silica particles are not disclosed but the pH of the slurry is maintained at 7.64. Further details on the slurry as a consumable and their interactions during CMP are detailed in chapter 5. 3. 3 4 .5 Optimization of CMP E xperimentation A statistical analysis of variance (AN OVA) was done with two factors and three level s as the experimental design. The factors analyzed are the pressure and velocity to test the effect these factors had on the MRR during CMP Table 3. 6 described below contains the spec ificat ions for polishing of the wafer. Table 3. 6 Statistical ANOVA table # Factor Level 1. Pressure 1, 3, 6 Psi 2. Relative Velocity 0.2, 0.8, 1.2 m/s 3. Time 120 seconds

PAGE 110

92 Each test was completed three times giving a total of 27 total experiments and values were averaged when completing the ANOVA. The levels chosen were based on IC CMP for the CMP benchtop tester in order to compare the values to previ ously attained values blanket materials. These factors were chosen to optimize the CMP process for BPM fabrication. 3 .4 Results and D iscussion The results from benchmark experimentation on CMP of BPM are detailed below. Analysis on the MRR, lubrication regime, mechanical properties, and a statistical analysis is presented below. 3.4 .1 CMP of the P atterned M edia S urface Results of the CMP of the BPM surface are detailed below. The interactions of the output parameters are discussed in the conclusion. Results have an indication on the feasibility of CMP as a fabrication technique. 3.4.1.1 MRR and P ressure As the pressure is increased and th e velocities of pad and carrier are held constant the MRR is als o increased, shown in figure 3.16 and this directly correlates to the material removal from Preston equation.

PAGE 111

93 Figure 3. 16 Pressure vs MRR for BPM CMP As the pressure increases the number of asperiti es from the pad and abrasive particles that come into cont act with the wafer surface asperities increases causing an increase in the MRR. Consequently the increase in pad to wafer surface contact pushes the po lishing regime closer to the boundary lubrication regime which is characterized by high MRR and high surface defects due to lack of a lubricating fluid during polishing. It should be noted that similar tests done on blanket copper wafers at 3 Psi and 1. 1 m/s relative velocity yields a MRR from 135 nm/min to 200 nm/min which is on the higher end of the BPM CMP MRR 3.4.1.2 MRR and V elocity Similar to section 3.4.1.1 the pressure values are held constant and the velocities are increased to determine the effect of velocity change on BPM CMP. Figure 3.17 shows that as the velocity is increased and pressure is held constant the MRR are also increased.

PAGE 112

94 Figure 3. 17 MRR vs. relative velocity The increase in velocity introduces more abrasive particles from the slurry into contact with the wafer surfaces and pad asperities leading to a higher removal rate [84] The increase in the v elocity does not have the large increases in MRR as with the pressure increase and this can be attributed to the fact that as the velocity increase the number of particles interacting with surface will plateau depending on the concentration of particles in the slurry and further increase in the speed yields little t o no effect on the MRR. It should be noted that the both the 1 Psi and 6 Psi curves are linear in fashion with respect to the MRR, while there is a decrease in the MRR at high speeds of the 3 Psi polish. It is also interesting to note that for the relative velocity plots the linear trend estimations are not exact. The relationship is not a linear relationship and this fact has been determi ned by several other authors [56, 83, 133 135]

PAGE 113

95 Quantitative results indicate that the pressure increase has a more pi votal impact on the MRR than the velocity based on the increases in the values for the MRR from figures 3.16 and 3.17. 3.4.1 .3 Stribeck C urve The C OF at the interface is a function of various factors such as surface to pography of the pad and wafer and the machine process parame ters. The Stribeck curve can be calculated by first calculating the Somme r feld number shown in equation 2.2 and reproduced below; (2.2) wafer average the pad wafer region. Determination of U and are fairly straightforward as the latter can be measured experimentally for a given slurry, while the former depends on tool geometry and angular velocities of the wafer and the platen [136] Based on knowledge of the Cabot iC ue 5001 slurry the viscosity ranges from 1 3 cP, and the average value of 2 cP is used for calculation of Sommerfeld number. This number will shift the Sommerfeld number to the left or right based on the other parameters but will not affect the overall im portance of the regime. Previous dual emission laser induced fluorescence (DELIF) experimental results by Coppeta and L u have shown that the slurry film thickness in the pad wafer region ranged from 20 to 40 m [137, 138] Li et al. proposed using the surface roughnes s of

PAGE 114

96 in the Sommerf eld equation as their results indicated a pad [136] In this study through WYKO surface profiler the pad roughness was found to be closer to ~ thickness. This approximation resulted in the calculated Sommerfeld number to shift to the right or to the left, in the data but had no effect on the trends of the Stribeck curves [136] The film is considered to distribute the pressure and eliminate the effect caused by different grooves. Therefore the wafer pressure is defined as the applied down force divided by the wafer area [136] Figure 3.18 is the Stribeck curve for the polishing parameters for experimentation in table 3.4. Figure 3. 18 S tribeck curve From figure 3.18 it is evident that all three polishing pressure s undergo transitions into all three lubrication regimes. All of the process start with the boundary lubrication regime in which the fluid carries little or none of the pressure applied In this regime the

PAGE 115

97 pad asperities and the wafer asperities are in direct contact and all of material removal is due to mechanical interactions of the pressure and velocity. Consequently due to the lack of slurry lubrication the COF is hi ghest in this regime and the thermal energy dissipated in the case must be very high resulting in non uniform and inconsistent material removal. All curves then transition to the hydrodynamic or partially lubricated regime, in which the pad to wafer conta ct and the slurry interaction play a role in MRR. The slurry along with the abrasive nanoparticles within slurry interacts with the wafer surface causing chemical corrosion and mechanical removal. This regime has high MRR and low surface defects; the lon ger a process is in this regime the better the results for the output parameters. The final transition occurs at the end of polishing to the fully lubricated regime in which the slurry contributes to all of the MRR, because the pad and wafer are not in co ntact. Polishing in this regime results in smooth surfaces but to increase the MRR, the velocities of the pad and wafer must be increased up to a threshold value. These regimes from the Stribeck curve are consistent with the calculated Stribeck curves f rom literature but for the BPM wafers the COF values are all larger than standard copper CMP [45, 46] The blanket copper wafers have a greater decline in the COF over the course of polishing and this fact could be due the planarization of the thin film in blanket copper CMP, whereas the B PM configuration polishing is not fully planarizing the surface but rat her a smoothing su rface as seen in figure 2.13 This leads to a fairly co nstant value of the COF during polishing The COF also gives a mea sure of polishing intensity at the interface which would result in heat dissipation generation of thermal energy from mechanical interaction [126, 139] Thus, measure of the COF at the

PAGE 116

98 interface during CMP gives vital information about the polishing and removal mechanism. Figure 3.19 gives the MRR versus the COF for the th ree recorded pressures. Figure 3. 19 MRR vs. COF The trend from figure 3.19 indicates that the lower the COF the lower the MRR during polishing. The MRR and COF values indicate that as the pressure is increased the MRR a nd COF are also increased. Utilizing the fact that the COF can be monitored in situ a qualitative model can be developed for the BPM CMP based on these results. T he results do not indicate the overall surface quality but the trend followed for material r emoval. CMP of BPM follows closely to experimentation by Zantye for COF and MRR trends done on silicon dioxide and other ILDs [45] 3.4 .2 BPM Pr e/ P ost CMP M echanical P roperties Table 3.8 below gives the pre and post CMP mechanical properties evolution during the CMP process for the increases in pressure. 0 50 100 150 200 250 300 350 0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 MRR (nm/min ) COF MRR vs COF 1 Psi 3 Psi 6 Psi

PAGE 117

99 Table 3. 7 Nanoindentation results of BPM c opper Patterned Copper Sample Elastic Modulus (GP a) Hardness (G P a) Unpolished 121.657 3.152 1.249 0.121 1 Psi 129.828 1.090 1.440 0.058 3 Psi 135.419 0.752 1.609 0.061 6 Psi 138.594 1.292 1.962 0.208 Table 3.7 shows the results from the nanocharacterization of the BPM CMP experiments. From the table it is evident that as pressure increases during polishing the elastic modulus and the hardness increase as well. Increase in the mechanic al properties arises from two possible reasons. The number of dislocations is increased as the abrasives particles permanen tly deform particles and then are removed due to the velocity and removal of the slurry. As the dislocation density is increased, the surfaces become harder to indent and permanently deform Another possible reason is due to the multigranular structure. Focused ion beam (FIB) and TEM images clearly depict several grains within the patterned structure. Orientation of these grains plays a role in the hardness due to several crystallographic effects. The critically resolved shear stress (CRSS) required t o cause slip in a crystal (grain) depends heavily on the orientation of that grain. Within the multigranular structure seen during polishing several orientations exists and as the BPM is polished through shearing of the surface the resulting surface will have an orientation different from the previous grain. The new grain orientation along with the dislocation motion to

PAGE 118

100 the grain boundaries increases the mechanical properties of the material through work hardening. These enhanced mechanical properties ar e advantageous for the BPM ability to withstand down pressure and shear during CMP. 3.4.2.1 Slurry Chemistry C haracterization To ensure that the slurry chemistry did not have strengthening effect on the copper surfaces, each BPM sample was allowed to sit in the Cabot iCue slurry and nanoi n dentation was done on each sample to determine the effects of the slurry on the mechanical prope rties of the samples. Table 3.9 shows the results from the experimentation. Table 3. 8 Mechanical properties from the slurry BPM Sample Elastic Modulus Hardness ( GPa ) Unpolished 121.657 3.152 1.249 0.121 5 minutes slurry 120.9212.758 1.228 0.221 30 minutes in slurry 119.241 3.56 1.256 0.48 1 hour in slurry 118.441 4.21 1.205 0.19 10 hours in slurry 117.021 1.51 1.18 1.13 24 hours in slurry 113.231 1.93 1.17 0.59 Based on table 3.8 the effect of the slurry alone has as detrimental effect on the mechanical properties of the BPM wafer coupons. The data in table 3.8 is based on nine indents averaged from the MTS indenter. It is evident that the Cabot iCue slurry weakens the copper on the silicon substrate and this is beneficial from a comm ercial aspect to

PAGE 119

101 increase the MRR during CMP, but this does not account for the increase in the properties seen in table 3.7 3.4 .2.2 TEM A nalysis Prior to polishing t he a TEM image was taken using the Technai F20 TEM shown in figure 3.20 shows that the patterned wafers are multi granular. The two major factors that affect the mechanical properties of a metal are the s ize of the grains coupled with the grain boundaries and grain orientation [140 142] Figure 3. 20 Initial multigranular cross section of copper BPM Given that the act of polishing is mechanical work on the metals, the increase in mechan ical properties is produced by work hardening and plastic deformation of the metal. For copper with large grains (micron size), the plastic yielding occurs by generation of dislocations from internal sources. The increase in stress results from th e pile up of dislocations causing the activation of sources in the adjacent grains and the

PAGE 120

102 resulting s train hardening arises from the accumulation of dislocations as seen in figu re 3.21 [143, 144] Figure 3. 21 Post po lish TEM of Cu BPM The increase in the theoretical shear strength of metals from shearing or indentation has been reported in literature and the dislocation m otion has been documented accordingly [145] Minor et al, provided evidence of the dislocation motion and increase in shear strength as shown in Figure 3.22.

PAGE 121

103 Figure 3. 22 Dislocation motion in aluminum [145] The grain boundaries disrupt the movement of dislocations in a crystal and the disruption leads to larger applied forces need to cause the crystal to deform and lead to mic ro cracks as seen in figure 3.23 Figure 3. 23 Microcrack formation s after polishing The micro cracks seen bel ow the surface in figure 3.19 was witnessed in all cases of polishing in the samples, even at the optimized polishing parameters. The formation of these cracks will cause the PM to lose any data written on it and future research will

PAGE 122

104 have to investigate either a means to reduce and remove the cracks or stronger materials to circumvent cracking. 3.4 .3 Surface R oughness C haracterizat ion Surface roughness and the overall surface quality play vital roles in the performance of the magnetic devices. It is imperative that the devices have atomically smooth surfaces so that the slider and magnetic head does not crash into the data on the devices. Table 3.9 contains the final average surface roughness values after CMP, the initial surface roughness, RMS, for the BPM configurations was 352.73 1.24 nm. Table 3. 9 RMS surface roughness for CMP process paramete rs Relative Velocity (m/s) 1 Psi 3 Psi 6 Psi 0.2 336 .88 1.95 nm 32 2.34 3 .24 nm 339 .00 6.96 nm 0.8 334 .0 4 0.86 nm 201.00 0.38 nm 222.4 2 4.95 nm 1.2 304 .17 10. 98 nm 195.88 0.32 nm 195 .57 2.57 nm From table 3.9 the polishing parameters for 3 Psi had the least amount of standard deviation Figures 3.24 contain the initial 3D surface topography taken by the Wyko NT 9 100 by Veeco, Inc. Figures 3.25 3.27 contain the final characterization images taken by white light interferometry using the VEECO NT 9100 surface profiler for the optimum polishing conditions using the pol ishing parameters from table 3.4 The images for each polish are contained in the appendix of this diss ertation but the optimum images for each pressure are shown below.

PAGE 123

105 Figure 3. 24 Initial surface roughness for BPM prior to polishing Figure 3. 25 1 Psi post CMP surface roughness

PAGE 124

106 Figure 3. 26 3 Psi post CMP surface roughness Figure 3. 27 6 Psi post CMP surface roughness T he final RMS surface roughness, Rq, from the 3 Psi, 1.2 m/s polishing is of 196.26 nm after polishing 120 seconds This median pressure and highest velocity

PAGE 125

107 yielded the optimal surface roughness for polishing at 120 seconds using the parameters for standard polishing in the IC industry. It should be noted that the 6 Psi polishing parameters had the highest M RR but the surface roughness was contaminated with debris from surface scratches and delamination at the edges due to over polishing. Figure 3.28 shows the delaminated SEM images after polishing at 6 Psi and 1.2 m/s. Figure 3. 28 Delaminated edge SEM image After determining optimal machine input parameters for the surface roughness and MRR for BPM CMP, a statistical analysis was conducted to determine the importance of each parameter. 3.4 .4 Statist ical A n alysis of V ariance (ANOVA) Table 3.10 contains the r esults from the ANOVA done on the 3 2 factorial test design on patterned media CMP using an alpha of 0.5.

PAGE 126

108 Table 3. 10 ANOVA table for pressure and velocity of BPM CMP ANOVA SOURCE SS Df MS F F Critical A 323210.5 2 161605.2 1141.19 3.354141 B 8609.056 2 4304.528 30.39682 3.35414 AB 3984.611 4 996.1528 7.034425 2.727765 Error 3823.5 27 141.6111 Total 339627.6 35 F actor A is the pressure and factor B is the velocity, the statistical analysis also accounts for the interaction between the two variables (namely the multiplication of AB) as a further factor for analysis. The results indicate that the pressure is the dominant factor in the MRR as the experime ntal results have proven. The velocity plays a less significant role in the MRR and these results were also reported by Tseng et al [84] The results indicate that the interaction of the pressure and velocity plays a s ignificant role but is simplified to a linear relationship as reported by [83, 134, 135] A residual versus fitted val ues graph is show in figure 3.29 to ensure there were no compounded errors in the calculations.

PAGE 127

109 Figure 3 29 Residuals plot And the statistical analysis is verified for the process paramet ers. Based on the values for the inputs of pressure and velocity and the resulting value of the MRR a linear regression model was developed as shown in equation 3.9; (3.9) where Y represents the MR R, and X1 and X2 represent the pressure and the velocity respectively. This equation does not take into the account the slurry chemistry or the Preston coefficient and is detailed here as an accompaniment to the statistical analysis. 3.5 Conclusion and R emarks CMP tests were run on copper bit patterned media in magnetic data storage device configurations, in order to understand the viability of CMP as the planarization technique on the new data storage configuration. The data on the CMP of BPM copper wafers was presented an d detailed aspects of the MRR w ere determined. The pressure and velocity

PAGE 128

110 were optimized to determine the parameters to induce the best surface quality and for material removal based o n pressure and velocity was verified through statistical analysis. Results indicate the best surface roughness of 196.26 nm occurs at polishing at 3 Psi, 1.2 m/s, and for the duration of 120 seconds. The highest MRR of 320 nm/min occurs at the highest pressure of the experiment of 6 Psi, 1.2 m/s, and 120 second polishing time. The values of the MRR for BPM are considerably lower for the same process parameters of blanket copper wafers with a percent difference for the low end of polishing of 3.11% 40% difference in the overall MRR. At all three velocities for the 6 Psi polishing test resulted in delamination of the patt erned media from the substrate, indicating that this polishing pressure should be avoided. Polishing at 6 Psi will result in failure o f the magnetic hard drive to the magnetic read/write head crashing while trying to access a grain or from crashing into the delaminated edges of the PM surface. The overall low surface roughness and repeatability of the low polishing pressures and high velocities indicate that the required atomic surface roughness can be achieved on the PM configurations The mechanical properties were characterized before and after polishing and results indicate an increase in mechanical propertie s with no depreciable change in grain size. The cause for the increase in mechanica l properties is linked to work hardening through the plastic deformation from the mechanical work done during polishing as evidenced through metallurgical studies [145, 146] T EM analysis shows dislocation motion and pile up at grain boundaries further verifying that the metallic copper is reaching the theoretical shear strength through the dislocation pile up [145]

PAGE 129

111 Optimizati on based on a statistical analysis mechanical properties, and metrology studies provided results that yield promising initial ramifications on the feasibility of using CMP as a planarization technique The machine parameters have been optimized based for BPM CMP based on IC CMP. This chapter serves to be the foundation for polishing BPM as literature has proven that the machine input parameters provide the greatest influence on the output parameters f or polishing [42, 59, 84, 85] The consumables utilized in the CMP process must next be optimized based on the machine input optimization in this chapter. The next phase characterizes the pads used for the BPM CMP process as detailed in figure 2.17.

PAGE 130

112 CHAPTER 4 : BIT PATTERNED MATRIX CMP PAD CHARACTERIZATION 4.1 Foreword A stable and predictable CMP process requ ires full control of the consumable parameters shown in figure 2.17 Luo argues that based on his particle scale model, the micro scale polishing pad topography, and nano scale abrasive size distribution in the slurry are the two most important parameters for the CMP process (consequently the slurry size distribution is covered in chapter 5 of this dissertation) [147] The polishing pad is arguably the most important component of the CMP system and has an economic impact of 550 million dollar s annually [148] The pad plays a crucial role in both the mechanical and chemical aspects of the polish. The mechanical properties of the pad will determine the polish rates and planarization ability of the process. The surface of the polishing pad, with its pores, grooves, and compressibility play an important role in t he mechanical removal of the reaction p roducts from the wafer surface [149] T he pad also carries the slurry on top of it, executes the polishing action, and transmits the normal and shear forces during polishing. At the pad/ wafer interface, the slurry acts on the wafer and forms a compound with the material that is being polished. This c ompound is removed when the abrasive particles collide with pad and wafer asperities. The material removed, is then washed away due to the constant slurry flow on the pad. This chapter serves to focus on char acterization of the pad and pad life in comparison to the material removal and resulting surface quality of the BPM wafer

PAGE 131

113 surface, based on the optimized parameters from chapter 3, the machine input parameters are used for pad characterization as shown in figure 4.1 Figure 4. 1 Pad characterization based on optimized machine input parameters T he objectives for this chapter are: 1) Investigate pad wear on BPM CMP 2) Perform a p arametric study of pad wear, pad roughness, COF, MRR, surface morphology 3) Give a qualitative analysis on the pad life, surfa ce characteristics of the wafer, pad and polishing regime 4) Determine optimal polishing pad for CMP of BPM from three commercially available pads.

PAGE 132

114 4 .2 Introduction : CMP P ads Polishing pads are major consuma bles affecting the within wafer and wafer to wafer non uniformity (WIWNU and WTWNU respectively) in planarization technology. WIWNU is where there is non homogeneity of polishing at different areas of the same wafer, and WTWNU occurs when there is non hom ogeneity of polishing when one wafer is compared to another [150] Polishing pads are composed of either a matrix of cast polyurethane foam with filler material to control hardness or polyurethane impregnated felts [42] The role of the pad is to transport media of slurry to the polishing reaction point and to support polishing pressure derived from down force to the wafer [151 154] The pad also transfers the shear force of the slurry to the wafer surface and eliminates polishing residue from the polishing point to allow new polishing reactions [153] The combination of the many duties of the pad results in the properties an d behavior of the pads directly affect ing the CMP output parameters. Unfortunately pad fabrication technology has not kept pace with t he continual progress of device fabrication processes. Specific problems include short pad life, inconsistent process results and extreme variability within each pad and from pad to pad; this requires costly adjustments of the CMP system and process parame ters [154] 4.2.1 Pad M aterials Based on the microstructure, pads can be divided into four categories [155] Type (1) pads typically are polymer impregnat ed felts. The microstructure of a pad of Type (2 ) is characterized by non woven polyester fibers infused with polyureth ane. Porome t rics form the Type (3 ) pads and pads of this type display a poro us layer on a substrate. T ype (4 ) pads are filled polymer s heet s and have a closed foam structure with macro pores

PAGE 133

115 For any polishing technique t he choice of pad depends on the nature of the material to be polished and polishing output requirements [156] Most commercial polishing pads for metal or oxi d es are of viscoelast ic nature and, are mostly of type (1) [93, 157] Polyurethanes are condensation polymers prepared by reaction of isocyanate and a polyol in the presence of a catalyst and a foaming agent [158] The intrinsic polymer properties, like glass transition temperature (Tg) and mechanical properties like elastic modulus, compressive strength, shear modul us etc., are strongly dependent on the molecular structure of the isocyanate and polyols. A large selection of commercially available isocyanates and polyol combinations, along with very versatile chemistry of urethanes, makes it possible to synthesize po lymers with specifically tailored properties [159] Synthesis of the pads is not specifically covered in this dissertation but can be found in literature review [159 163] 4.2.2 Effect of P ad G eometry The polishing process involves, intimate contact between the asperities on the wafer surface and the pad material, in the presence of slurry. The mechanical p roperties of both surfaces play a significant role in the final planarity and polishing rate. The wafer su rface is hard and brit tle, while pads tend to be made of relatively softer materials for optimal polishing. 4.2.2.1 Effect of P ores The polyureth ane pad consists of pores with in the pad. T hese pores can be either closed or open pore systems. In either case t he pores on the pad surface aid in slurry transport, to all parts of the wafer surface thereby ensuring chemical erosion. The cell walls of the foam mechanically remove the reaction products from the wafer surface and

PAGE 134

116 the pores of the pad enable transport of the reaction products from the interior of the wafer surface to outside as seen in figure 4.2 [159] Figure 4. 2 Schematic of CMP polyurethane pad pores [159] The pores on the pad surface e nhance both the chemical and mechanical aspects of the process [159] For a pad with an open pore structure, increasing the number of pores increases the cell wall scraping and henceforth the mechanical abrasion. Alternatively the closed pad structure is not interconnected to the w afer surface and therefore does not aid in the mechanical abrasion. The closed pore structures (pores t hat have a dead end) cannot aid slurry transpo rt and/or product removal, but assists in supporting the pressure applied to the pad for stability. It is important to know the total pore volume as well as the fraction of open cells The pad design should involve optimization of both to aid in support of the pad structure, the slurry movement through the pad, and mechanical abrasion [159] 4.2.2.2 Effect of G rooves The grooves or perforations on the polishing pads have a significant impact on the polishing mechanism and outcome parameters [164, 165] Grooves on the pad allow for effective slurry flow under the wafer surface and thus are very crucial for an effective

PAGE 135

117 CMP pro cess. Phillipossian et al. carried out fundame ntal tribological studies on CMP of pads with different groove types at various slur ry abrasive concentrations [166] The COF data w as fitted as a function of Sommerfeld number and a tribological mechanism indicator was developed to index and des cribe the change in COF. Stribeck curves were generated using the friction data for a variety of groove and pad types. From the shapes of individual curves, the authors hypothesized that some of the pads polished in partial lubrication regime and some in boundary lubrication at lower Sommerfeld numbers and transitioned to partial lubrication regimes. Consistent removal rates and uniformity were observed as long as the polishing regime is in boundary lu brication regime, although polishing in the boundary lubrication regime is aggressive and may induce delamination of the surface. 4.3 Pad C haracterization There is ongoing research to investigate the dependence of various pad material p roperties on the CMP process. Various resear chers have focused on the macro e ffect of the wafer shape, mechanical properties, and polishing pad profile on the MRR, and findin gs include that there is a drop in the MRR as a function of time due to varying the mechanical response under conditions of critical shear [167] During the CMP process the surface of the CMP pad gets loaded with debris from the polishing operation, which leads to glazing on the surface. This means that there are no asperiti es to hold the abrasive grits, which leads to inefficient polishing and possible micro scratching on the surface wafer. The phenomenon response under conditions of shear. Other researchers have focuse d on the wafer planarity and determined that th e wafer planarity is a function of pad stiffness, which is

PAGE 136

118 determined by the elastic properties of the pad material [167, 168] As previously stated it has been shown that t he pad m ay be directly responsible for several process defects li ke WIWNU and WTWNU and t he techniques used for characteri zation of the CMP polishing pads such as, dynamic mechanical analysis (DMA) are destructi ve and therefore do not yield ex situ information on the pads which requires a new and novel way to analyze the pad characteristics, properties, and life. This chapter serves on qualitative character ization of pad properties through a novel nondestructive technique and their effect on the output parameters. Secondly this research aims to predict pad life for BPM configurations to reduce waste and increase the sustainability of the pads. 4.3 .1 Ult rasound T ransmission A novel non destructive ultrasound t ransmi ssion system ( UTS ) developed at USF has be en effectively used for evaluation of the CMP pads [169] This technique works on the principle of ultrasound permeability through absorbing viscoelastic medium. The tr ansmitted ultraso und signal is used to determine how the material properties of the pad vary over its geometry [170] By sending an ultrasonic wave through the pad and measuring the change in transmitted signal at different spots, one can create a UTS map of the underlying pad structure. The UTS amplitude can be monitored as a function of time, height above pad and depth below its surf ace application of compression at different transmitted frequencies [170] The regions of polishing pad having variations in specific gravit y transmit different amplitudes of UTS at a same frequency. The amplitude of the ultrasound permeability with in a pad obtained as a result of UTS is normalized to against the average UTS amplitude to estimate the comparative variations in specific gravit y in the different regions of the same polishing pad. The output of the

PAGE 137

119 measurement is a Doppler diagram in which different colors correspond to different amplitudes of UTS within the pad. This picture may be correlated with the pad life and performance in order to predict these variables. This could increase yield by reducing the number of rejected wafers as the pad ages prematurely or lengthen production time by indicating which pads have more desirable properties The UTS system developed at the Center for Microelectronics Research (U. South Florida) is comprised of two key elements; a resonance circular piezoelectric transducer as an emitter of acoustic vibrations of se lected amplitude and frequency, and an acoustic probe as a receiver of ultrasonic vibrations. A schematic diagram for the UTS system is shown in Figure 4.3 with the transducer and acoustic probe labeled respectively Figure 4. 3 UTS schematic for CMP pads [138] 4 .3 .2 Surface C haracterization The surface morphology of the pads was characterized using a JEOL JSM6490 scanning electron microscope (SEM) shown in figure 4 .4

PAGE 138

120 Figure 4. 4 Jeol JSM6490 SEM surface morph o logy tool The SE M supports ultrastru ctural analyses of surfaces, 3D organization, and has a 3.0 nm resolution. Prior to SEM, the polyurethane pads were coated with a thin layer, ~10 nm, of gold using the HUMMER X sputtering system at the Nanomaterials and Research Engine ering Center (NREC) at the University of South Florida (USF). The SEM is used to investigate the surface morphology, cross section of the pa ds and the effect of polishing on the pores and grooves of the pads 4.3 .3 Wafer and P ad S urface R oughness As de tailed in chapter 3 section 3 .3, t he WYKO NT9100 surface profiler is used to determine the surface profile roughness in three dimension s for the pad surfaces and the BPM wafer surfaces d etailed in chapter 3 section 3.3 .1. Knowledge of the resulting roughness parameters wafer will lead to conclusions about how the wear life of the pads affects the overall surface roughness of each wafer. Optimization of this parameter can lead to predetermination of unusable pads which ca n reduce WTWNU thereby decrea sing waste.

PAGE 139

121 During polishing pad r oughness is expected to decrease along with the MRR during pol ishing. The removal of the pores and grooves significantly reduces the a bility of the pad to remove material and a prediction of the pad life is therefore v ital for sustainability. The decreas e in the roughness of the pad is detailed by the WYKO NT9100 A statistical WT WNU measurement is calculated by dividing the standard deviation of the wafer surface roughness by the average surface roughness of each pol ish. This value will indicate how repeatable the polish parameters are for each pad and polish set. 4.3 .4 Ex S itu and I n S itu CMP P ad C haracteristics As described in chapter 3 section 3 .3. 4 the CETR benchtop tester allows for in situ monitoring of th e forces applied for polishing. The F x F z forces are utilized to determine the COF during polishing of all pads and wafers. Knowledge of the COF and the polishing parameters along with the pad roughness allows for calculation of the Sommerfeld number du ring pad wear. This calculation allows for elucidation on the lubrication regime and thereby the polishing mechanism during the CMP process. 4.3 .5 Material R emoval The material removal on the BPM copper wafers was determined using the technique de scribed in chapter 3 section 3.3 .4.2. The four point probe was used to determine the thickness of the wafers before and after polishing and this change over time was deemed the material removal rate. A second method was also used when the BPM wafers were polished to the silicon substrate. Figure 4.5 shows the initial thickness measurement done by the FIB machine described in chapter 3.

PAGE 140

122 Figure 4. 5 FIB image of initial thickness of BPM By knowing the initial thickness of 1. level of the BPM the amount of time used to polish to the substrate can also be used to determine the MRR. 4.4 Experimental S et U p Details on the experimental set up for characterization of the CMP pads are des cribed below. All experimentation was done on a statistical analysis with testing done a minimum of five times for each pad and experiment. 4.4.1 CMP P ads Three different CMP pads were tested for experimentation. These commercial pads are a ll utilize d in either copper or other metal polishing. The MRR and pad wear is characterized by the techniques mentioned in section 4.2 of this dissertation. Details of each individual pad are shown in table 4.1 (note k grooves are concentric pores)

PAGE 141

123 Table 4. 1 CMP pad characteristics CMP Pad Type Diameter Specific Gravity Thickness Hardness Compress ibility Groove type (1) Rodel dual layer IC 10 00 81 cm 630 800 (kg/cm 3 ) 55 59 shore D 2.25% k groove (2) Rodel dual layer IC 14 00 Suba V 81 cm 0.75g/cm 3 57 shore D 0.7 6.6% k g roove (3) RD 2003 matrix foam 57 1 5 cm n/a ~10 shore D n/a xy groove It should also be noted that discussions of the pads will be noted by the numerical value assigned in table 4.1 (e.g. Rodel IC 10 00 will be referred to as pad (1), etc). The shore D value of pads (1) and pad (2) are much greater than pad (3) for the same scale. The shore A value of hardness was converted to shore D value based on literature [171] Prior to polishing experiments all commercial pads had to be conditioned for polishing. Pad conditioning is an important process to r estore the pad properties that deteri orate over time from stagnant pads The surface interactions involved in the process of polishing are influenced by the pad texture resulting from conditioning. The proc ess of con ditioning is used to: 1) M aintain the roughness of the pad and promote effective slurry distribution 2) R emove unwanted products after polishing (glazing)

PAGE 142

124 Various pad conditioning methods have been used to improve the pad properties and stabilize the removal rates. The most effective method found was using diamond as the abrasive material [149] The p roperties of the conditioner such as diamond density and diamond mounting play a major role. The other input variables for the conditioning process include parameters such as the conditioning down force and relative speed of rotation (rpm) of the pad plat en and the conditioner. The process of conditioning can be quantified in terms of MRR, pad roughness, and wear of pad. It has been found that a pad conditioned before the first polished wafer doubled the removal rate compared to the unconditioned pads. C onditioning maintains the removal rate by maintaining the asperity height an d density on the pad surface. U neven pad wear results in uneven distribution of the pressure affecting the planarization uniformity and removal rate. During the process of condi tioning, the conditioner disk rotates about its axis and simultaneously moves linearly towards and away from the center of the pad for a uniform conditioning of the pad surface All pads were conditioned with TBW grid abrade 2 diamond pad conditioners, f or 20 minutes followed by 1 minute of polishing on a dummy sample, then another 20 minutes of conditioning, followed by another 1 minute of polishing on a dummy sample, then the final 10 minutes of conditioning of the pad followed by the actual experiments with Cabot iCue 5001 copper slurry (details of this slurry are in chapter 3 section 3.3.4.4). 4.4.2 Ultrasound T ransmission T esting S ystem (UTS ) The UTS system consists of a flat square table that can accommodate polishing The center of the table has a circular hole that allows the two screws that holds the pad to pass through. One side of the table has a slot which

PAGE 143

125 enabl the radial direction of the polishing pad. The transducer has a hole in the center and has trenches in the sides which help in generation of the vacuum which is used to hold th e pad on the surface during UTS measurement. The piezoceramic transducer emits resonance ultrasound vibrations at 26 KHz (first resonant frequency of the piezoceramic transducer), while a 7 mm diameter quartz rod or a pinducer housed in aluminum casing al igned directly above the transducer acts as the receiver. The received ultrasound frequency is then converted in to electrical energy and the raw output is seen on the oscilloscope. The signal from the probe and reference input are both sent to a lock in amplifier which records the amplitude of the received signal at the same frequency as the emitted signal. Different sections of the polishing pad are scanned be rotating the pad using the mounting screws by a step motor with 2 degrees angular steps, and moving the emitting piezoceramic transducer and the aligned receiver with another step motor at 7mm radial steps with help of a threaded spindle and screw. There is a provision for vertical movement of the receiving pinducer with the help of a vertically positioned spindle and screw. The measurements are taken at a distance of 100 m below the pad surface to cross linked polymer material of the polishing pad, all mea consideration. The details of the UTS set up, measurement techniques, characterization procedures and operation have already been published in literature [169, 170] The UTS

PAGE 144

126 liner below it. It is assumed that the bottom PSA and the plastic liner are uniform and will have a similar effect on the ultrasound transmission. These effects could then be filtered out when the entire data is normalized against the mean UTS The area s of interest ( e.g. showing the highest or lowest ultrasound transmission over the entire pad ultrasound transmissions were ). T hese areas were imaged under high m agnification in the SEM to look at the surface characteristics of the pads. 4.4.3 CETR CMP P olishing In order to fully understand and characterize the pad, the process mac hine input parameters are kept constant during polishes of the pad. The input parameters are based on the optimized data from chapter 3. Table 4. 2 contains the process input parameters for pad characterization. T able 4. 2 Pro cess p arameters for pad characterization polishing # Parameter Conditions 1. Pressure applied 2 Psi 2. Platen rotation 200 RPM 3. Carrier rotation 200 RPM 4. Slider movement 3 mm/s 5. Slurry flow rate 75 mL/min 6. Time 120 Seconds 7. Pad Varying 8. Specimen

PAGE 145

127 The pads undergo a design of experiments that consists of: 1) Ultrasound testing of the pad at each polish 2) SEM of the pads HT and LT sections based on the UTS output for surface morphology 3) If needed the pads are conditioned and steps (1) and (2) are repeated 4) The pad then polish the specimen using the process parameter in table 4.2 5) Steps 1 4 are repeated for each pad up ~ 100 polishes or 200 minutes This process described above begins with the as received pad s and completed after all polishes on the BPM are completed. For benchmarking purposes the tests are run until a) the BPM is fully removed from the silicon substrate or b) the pad life has been exceeded without conditioning ( e.g. the pad is visibility wo re or the thickness is half of the initial thickness of the pads based on micrometer readings ) [42, 168, 172] 4.5 Results and D iscussion For each polish and pad including the conditioning of the pads, the average COF, MRR, wafer surface roughne ss, and pad surface roughness are recorded at 20 minute intervals Wyko surface profile images of the wafer and pad surface roughness, 2 D profile, surface data including any scratches or delamination, a histogram of the height distribution, and a 3 D image were recorded for all polishes, this data is present ed in this dissertation for pad (2), the images for the other pads is referenced in the appendix of this dissertation. A table of the graphical values is shown in the conclusions section. 4.5.1 COF and P ad P olishing The importance of the COF during all polishing process was detailed in chapter 3 and verified by literature [73] During characterization of the pad, the COF helps to

PAGE 146

128 determine the polishing regime, as well as gives a n indication of the MRR, and pad life for each of the pads tested. 4.5.1.1 Stribeck C urve The Stribeck curves were calculated for the three pads for values of the COF from 10 polishes to 10 0 polishes (20 minutes to 200 minutes) The COF was monitored in situ and the average value was taken for calculation. The Sommerfeld number was calculated using equation 2.2 and the input parameters from table 4.2 along with using the Cabot Microelectronics value for the viscosity of the slurry and a slurry film t hickness approximation based on the pad roughness. T he Stribec k curve is plotted in figure 4.6 for changes pad roughness during polishing to locate the polishing regime Figure 4. 6 Str ibeck curve for pad polishes The lubrication regimes and characteristics for each regime during polishing were covered in chapter 2. Based on figure 4.6 it can be seen that p ad (1) operated predominately in the fully lubricate d regime; and this regime has the lowest MRR along 0 0.2 0.4 0.6 0.8 1 1.2 0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 0.0007 COF Sommerfeld Number Stribeck Curve Pad (1) Pad (2) Pad (3)

PAGE 147

129 with t he least amount of surface defects. Pad (2) polished in the partial lubrication regime, most preeminent for CMP because it allows for high MRR along with l ow surface defects. Pad (3) borders the boundary l ubrication regime as well as the partial lubrication regime. The boundary lubrication regime is good for the material removal rate but because of the two body abrasion has high surface defects It should be noted that for BPM the least amount of surface defects does not equate to the lowest RMS surface roughness. This is due to the face that the patterned surface is only planarized for each high asperity leaving the differential between the remaining asperity height s and valley s high and therefore leavin g the root mean square surface roughness high 4.5.1.2 COF and the MRR A model between the COF and the MRR for composite non heterogeneous materials has not been developed due to the changes in pads, material properties, process parameters, and slurr ies (which includes chemistries and abrasive particles). During polishing a correlation was investigated to find a relationship that can be used as a parameter to optim ize the MRR and COF. Figure 4.7 has the MRR versus the COF data.

PAGE 148

130 Figure 4. 7 MRR vs CO F for CMP of BPM p ads Although a predicting model cannot be drawn from figure 4.6, there does exist a relationship between the COF and the MRR. The higher the COF the greater the MRR for all three pad s regardless of layering, groove type, or polishing parameters. This relationship could be described by the action of mechanical polishing during CMP. As the high asperities come into co ntact with the pad there is two body abrasion that happens and during this interacti on between the pad/wafer surfaces the COF is highest. Alternatively, the three body abrasion system inherently cannot have a value as high for COF as the two body abrasion due to the lubricating fl uid as a medium. The regime with the high est COF would be boundary lubrication and this regime has the highest MRR while also producing the greatest surface defects. This fact is further proven by the high material removal during polishing without slurry and the higher COF value, the lack of lubrication shows that the boundary lubrication regime dominates with

PAGE 149

131 high material removal and high COF, but his regime has the highest amount of surface defects. 4.5.1.3 COF and P olish T ime The COF is a direct corollary to the MRR during polishing and in situ measureme nts of the COF during polishing lends to predictable determination of the MRR during polishing. Figure 4.8 details the COF versus polish time for the pads. Figure 4. 8 Polish t ime versus COF for BPM CMP pads The initial value for the COF for all three pads is roughly 0.4 based on figure 4.8. All pads are still in the break in portion of polishing and no difference in COF is appreciable (although the MRR is different for all pads). After 20 minutes of polishing all pads exhibit an increase in the COF and this corresponds to increase in the MRR during polishing for all pads seen in section 4.5.2.1. After break in for pad (1) the decrease in the COF remains study during polishing for 200 min. This correlates to a 0 0.2 0.4 0.6 0.8 1 1.2 0 50 100 150 200 250 COF Time (min) COF vs Polish Time Pad (1) Pad (2)

PAGE 150

132 steady a nd repeatable polishing regime for pad (1). Pad (2) has another COF increase in polishing between 40 and 50 polishes, due to introduction of a new wafer coupon for polishing. Pad (2) then has the same d ecrease in COF as seen in pad (1) only shifted for higher COF values. The COF values for pad (3) follow the same pattern as the other two pads for the but as the pad deteriorates there is no relationship to be determined for COF and polish time for pad (3), and the introduction of pad particle s into the wafer/pad interface causes anomalies in the COF values. 4.5.1.4 BPM S urface R oughness and COF The ability to determine the output parameters by monitoring in situ parameters is paramount for BPM CMP. It has been shown the COF is an indicator o f several process parameters such as the MRR and lubrication regime. Figure 4.9 contains the plot of the BPM surface roughness versus the COF. Figure 4. 9 BPM wafer roughness vs. COF 0 50 100 150 200 250 300 350 0 0.2 0.4 0.6 0.8 1 1.2 Wafer Roughness (nm) COF BPM Wafer Roughness vs COF Pad (1) Pad (3) Pad (2) wafer (1) Pad (2) wafer (2)

PAGE 151

133 Based on figure 4.9, as the COF is decreased the wafer surface roughness is also decreased for pad (1) and pad (2). As the pads polish the surface the number of contacting asperities becomes worn and the contact area is decreased. As the contact area is decrease d the COF decreases and the worn surfaces become smoother. The trend does not hold for pad (3) and again this is linke d to the deterioration of the pad. Pad (1) has the lowest COF values and the gradual increase in polish time results in a lower COF, and thereby wafers roughness. Pad (2) follows this same trend but broken into two separate wafer coupons. The consistently higher values of COF seen from pad (3) arise from the abraded pad particles becoming trapped in the wafer/surface interface instead of being removed by the angular velocity. These particles lead to a stagnant value COF during polish time. 4.5.2 Polish T ime M etrics The amount of time required to polish the BPM to atomic roughness based on IC CMP elucidates the overall efficiency of the process. Pad characterization will serve to dictate the output parameters during polishing. The data reported in this chapter for the pads is based on only initial conditioning of the pad. Standard CMP practices reconditions the pads throughout the pro cess cycle, but for benchmarking purposes the pad is conditioned only that the onset of utilization. 4.5.2.1 Material R emoval The CMP process is employed to remove unwanted material from surfaces to ensure local and global planarity during polishing. Th e MRR is plotted vers us the polish time in figure 4. 10

PAGE 152

134 Figure 4. 10 MRR vs polish time From figure 4. 10 the polish rate decreases with time for polishing of BPM Based on the Stribeck curve, p ad (1) operated in the fully lubricated regime which was distinguished by low material removal but extremely smooth resultant surfaces. Although lower MRR than pad (2), pad (1) showed a linear decrease in MRR over the 200 minute polish time Pad (2) wh ich operated in the partial lubrication regime had the highest MRR of the three pads. It should be noted that pad (2) polishes two different wafer coupons ; the first coupon reached the atomic surface roughness within 75 minutes of polishing and was subse quently replaced with a new coupon for pad (2) to polish Pad (3) operated on the border of the boundary lubrication regime (wafer to pad contact) rapidly deteriorates due to the low Shore hardness value of the pad and possibly the groove pattern and afte r 100 minutes of polishing the pad no longer removes appreciable material. 0 100 200 300 400 500 600 0 50 100 150 200 MRR (nm/min) Polish time (min) MRR vs Polish Time Pad (1) Pad (2) Pad (3)

PAGE 153

135 4.5.2.2 BPM S urface R oughness The second overall goal for CMP is to polish surface to an atomic smoothness Figure 4 .11 contains the resulting BPM surface roughness versus polish time for the pads tested. Figure 4. 11 BPM surface roughness vs. polish time From figure 4. 11 as polish time increases the BPM surface roughness decreases. This is expected and f ollows the trend from figure 4.9 in which the MRR rate is initially very high and as the BPM becomes planarized the MRR decreases as there is less material to remove Pad (1) takes the longest time to p olish the wafer to atomically smooth surfaces requiring at nearly 200 minutes. Pad (2) completed polishing to the required 10 nm surface roughness in roughly 75 minutes, after which a second coupon was polished to try and determine if a second BPM could b e polished to the required roughness value. This resulted in the second coupon reaching a surface roughness of 0 50 100 150 200 250 300 350 0 50 100 150 200 Wafer Surface Roughness (nm ) Polish Time (min) Wafer Surface Roughness vs Polish Time Pad (1) Pad (2) Pad (3)

PAGE 154

136 roughly 125 nm, far below the initialized value of 352 nm. Pad (3) was unable to polish the surfaces below 200 nm roughness and as mentioned pr eviously deteriorated rapidly during polishing. 4.5.2.3 Pad R oughness The relation of the pad roughness to the resulting surface roughness provides valuable data into the regime and how the pad is actually polishing the material. The pad roughness will ultimately determine how smooth the BPM surface will be due to the pad/wafer interface asperities dominating the MRR as shown in chapter 3. As mentioned previously the pad roughness has been directly linked to the material removal rate and efficiency of the pad. The pad roughness is plotted versus the polish time in figure 4.12 Figure 4. 12 Pad roughness vs. polish time 0 2 4 6 8 10 12 0 50 100 150 200 250 Pad roughness (micron) Polish time (min) Pad Roughness vs Polish Time Pad (1) Pad (2) Pad (3)

PAGE 155

137 For pads (1) and (2) there is a direct relationship between the pad roughness and polishing time. As the pads continue to polish the BPM the glazing effect reduces the pad roughness and thereby the MRR and resulting BPM surfac e roughness The increase in pad roughness seen in figure 4.12 for pad (3) results from deterioration of pad (3) during polishing. The abraded pad particles were embedded on the surface of the pad during polishing and the resulting imaging seen in the appendix reflects this embedment. A secondary indicator for pad (3) r oughness phenomena arises from the fact that t he MRR did not increase for pad (3) as would be seen if the imaging were a correct representation of the pad surface without deterioration. 4.5.2.4 Pad T hickness Industrial standards indicate that once the pad has reached half the original thickness the pad must replaced in order to reduce cost of losing useable wafer coupons. Replacing the pads is a costly procedure and benchmark evidence on the life of the pads for BPM is critic al to reduce waste. Figure 4.13 shows the reduction in pad thickness over time.

PAGE 156

138 Figure 4. 13 Pad thickness vs. polish time The initial thickness is seen as the first data point on figure 4.13 for each pad. The pad thickness decreases with polish time as would be expected. The rate at which the pads lose thickness is much faster than would be expected from IC CMP. The pads tested have a pad life from 400 700 minutes and results indicate replacement of the p ads at 100 minutes, 200 minutes, and 50 minutes for pads (1 3) respectively. This dramatic change in pad life must be taken as a cost for polishing BPM and is due to the high initial roughness of the BPM surfaces when polishing as opposed to the relativel y smooth surfaces for IC CMP. 4.5.3 Metrics D iscussion It has been shown in the previous sections that the rougher the pad, the higher the COF, and the higher the COF, the higher the MRR, and pad roughness to the MRR. This knowledge can prove benefici al from an economic standpoint because optimal pad

PAGE 157

139 roughness for BPM is essential for polishing regimes and vendors can specifically supply pads with the required roughness values to improve sustainability and reduce polishing times. For the pads testing and utilized to polish BPM configurations the highest removal rate was obtained by the k groove pad along combined with the highest surface roughness. All three trends are linear in fashion and the values for polishes rates are presented in the appendix. Pad (2) obtained the highest MRR with a value of 480.27 16.7 nm/ min 4.5.4 Surface M orphology C haracterization A qualitative understandi ng of the surface morphology evolution during polishing is needed to predict pad life and wear characteristics; thi s is paramount for sustainability and process optimization during the CMP process. The UTS and SEM machines were utilized to characterize each polish for the BPM from table 4.2. Figures 4.10 4.15 show the representative surface evolution of pad (2) durin g polishing for 80 minutes of polishing (time required to polish wafer coupon to atomic roughness) Pads (1) and pad (3) images are depicted in the appendix of this dissertation.

PAGE 158

140 Figure 4. 14 As received and conditioned UTS and SEM images of pad (2) The UTS amplitude is monitored as a function of height above pad and depth below its surface application of compression at different transmitted frequencies The normalized U TS figures are pictorial represen tation of the chang es in specific gravity (which can be related to the density) during polishing with the darker red areas deemed HT and the lighter colors deemed LT. From figure 4.14 after conditioning the pad the as received pad t he specific gravity increases along with the viscoelastic properties. The surface characteristics from the SEM follow the UTS images.

PAGE 159

141 Figure 4. 15 UTS SEM, and BPM surface roughness of pad (2) 10 polishes In comparison to figure 4.14, figure 4.15 show s a representation of the UTS value increase in the number of areas of HT. This increase in HT corresponds to the increase in pad roughness as seen in section 4.5.2. The SEM images show a rougher morphology of the polishing surface while the WYKO images i ndicate the surface roughness from the BPM wafer being polished.

PAGE 160

142 Figure 4. 16 UTS SEM and BPM surface roughness for pad (2) 20 polishes After 20 polishes as seen in figure 4.16 the UTS graph indicates that the pad continues to become denser during polishing and the compression properties increase while the SEM morphology reflects the change in the surface characteristics. It should be noted that the pad has a reduction in roughness, an d the surface can be seen to contain slurry remnants and partial glazing of the surface The wafer surface continues to become further planarized and the results match well with the data from section 4.5.2. This trend i s continued below in figure 4.17 an d a summary of the result s is described after figure 4.18

PAGE 161

143 Figure 4. 17 UTS SEM, and BPM surface roughness of pad (2) 30 polishes Figure 4. 18 UTS SEM, and BPM surface roughness of pad (2) 40 polishes The material properties of the pad begin to deteriorate after 40 polishes for pad (2) based on figure 4.15. This decrease is also characterized b y the SEM picture and the overall surface quality for 40 polishes of pad (2) has reduced the BPM wa fer surface down to approximately 10 nm. The SEM images of the pad clearly depict slurry

PAGE 162

144 remnants pad glazing, and changes in the pore sizes and grooves due to the polishing of the rough BPM surface. There remain a number of good HT regions in the pad a fter 40 polishes therefore the pad is still suitable for polishing and therefore a new coupon was polished utilizing this pad. 4.5.4 Pad M orphology D iscussion The optimal polish time to planar ize the BPM wafers for pad (2) was discovered to be 75 minutes for the process parameters in table 4.2. A parametric study was conducted to determine how each output parameter would affect each other to determine other methods for EPD and polishing requirements. The next step in determining the pad wear and polishi ng (beyond the UTS and SEM as these characterization techniques are costly and require the machine to be shut down in order to process this information). 4.5.5 WT WNU The ability to detect the amount of variation that a particular pad will yield on the w afer surface is paramount and the ability to have accurate and repeatable results for given process conditions reduces cost and waste to the fabricat ion industry. A statistical WT WNU measurement is calculated using by dividing the standard deviation of th e wafer surface roughness by the average surface roughness of each polish. This value will indicate how repeatable the polish parameters are for each pad and polish set. Table 4.3 contains the W T WNU values for each polish set (in increments of 10), up to the deterioration point of the pad or the planarity of the wafer

PAGE 163

145 Table 4. 3 W T WNU percentage based on BPM wafer CMP WIWNU 10 20 30 40 50 60 70 Pad (1) 1 9.79% 2 7.48% 8.7% 6.87% 4.47% 11 .32% 10.99 % Pad (2) 14.32% 16.39% 8.54% 3.34% 11.47% 16.42% 13.47% Pad (3) 10.49% 9.61% 27.21% 28.01% From table 4.3, pads (1) and (2) follo w a trend in which increase the polish time results in the surfaces having a lower WTWNU measurement. Pad (3) had the highest W T WNU and this could be due to the rapid decline of the pad, as indicated by the steady W T WNU up to 30 polishes. This table indicates that pad (2) produces the greatest number of repeatable results for polishing the copper BPM wa fers. 4.6 Conclusion and R emarks A benchmark parametric study on CMP pads for BPM copper wafe rs was completed to determine the effect on output parameters during the CMP process based on optimization of the machine parameters from chapter 3. Resu lts indicate that as the pad is polished over time the MRR, COF BPM surface roughness, pad thickness, and pad roughness are de creased The process input parameters were held constant for polishing and a Stribeck curve was created for the pads based on th e change in slurry thickness during polishing. An analysis of the lubrication regime which helps dictate polishing output parameters indicate that pad (1) operates mostly in the full lubrication regime, pad (2) operates in the partial lubrication regime, and pad (3) operates in the boundary regime.

PAGE 164

146 Results also show that a force sensor is easily implemented into most standard polishing systems and can be used to determine the COF and analyze the pad life, MRR, and surface quality based on the COF during polishing. The pad life, pad material removal, and pad roughness are characterized for 100 200 minutes of polishing at the process parameters in table 4.2) for each of the three commercially available pads and results indicate that for BMP copper CMP, the Rodel dual layer IC 1000 Suba V k groove pad provides superior MRR, wafer roughness, and pad life for polishing with Cabot iCue 5001 and Cabot iCue 5003 slurries. The surface morphology evolution of the pad was characterized using a novel non destructi ve ultrasound technique and scanning electron microscopy. The UTS readings are crucial indicators of the pad life and provide critical insight into the evolution of this morphology through a cost effective means. The UTS characteriza tion and SEM characte rization were able to detect that the pad material properties are inversely proportional to the porosity of the pad. This means that as the pads are poli shed over time the pores in the pad are worn away, and the lower material characteristics of the pad l ead to a lower porosity and henceforth a lower MRR and COF. Several other researchers have found this aspect true for blanket copper and dielectric polishing, but the novelty in this research is in proving that the BPM wafers pad life is much lower than p ad life for blanket wafers, with an average percent difference of 63.63% [167, 168] The experimental pad morphology life time without reconditioning was characterized for all three pad s with images from the SEM, UTS and WYKO in the appendix of this dissertation.

PAGE 165

147 A statistical WT WNU measurement is calculated by dividing the standard deviation of the wafer surface roughness by the average surface roughness of each polish. This value will indicate how repeatable the polish parameters are for each pad and polish set. Pad (1) provid ed the lowest WTWNU measurement, indicating that the repeatability of the pro cess is optimal utilizing pad (1), although not optimal for MRR and surface roughness characteristics Results for the optimal machine parameters and pad are next utilized in the slurry characterization seen in the forth coming chapter.

PAGE 166

148 CHAPTER 5: SYNTHESIS OF NOVEL CMP SLURRY 5.1 Foreword The economic impact of the slurry during CMP for the semiconductor industry rises over 2 billion dollars a year. There is signi ficant research into manufacturing new slurries that have the capability of high removal rate, excellent global planarization, corrosion prevention (in case of metal, especially copper), good surface finish, low probability of defects and high selectivity. The chemical and mechanical interactions of the slurry is the least understood mechanism during the CMP process and ongoing research has yet to adequately explain these mechanisms. CMP is a process that is influenced by numerous slurry parameters such a s pH, solution chemistry, charge type, concentration and size of abrasives, complexing agents, oxidizers, buffering agents, surfactants, corrosion inhibitors, etc [42, 173, 174] The specific and proprietary nature of slurry manufacturing makes it difficult to elucidate the exact effects of slurry on the particula r thin films that are polished by it. The slurry interactions at the pad wafer interface are probably therefore, the least understood mechanisms in entire semiconductor fabrication process technology [154] Due to lack of understanding of the mechanisms for polishing and the economic impact the slurry has, CMP slurry has continued to be a catalyst for research and development. The ability to chemically et ch a specific material and polish that material while essentially leaving the underlying material alone are interests of both the industrial and academic relevance Based on the

PAGE 167

149 optimized machine parameters and pads the output parameters based on the slur ry interaction are investigated as shown in figure 5.1. Figure 5. 1 Slurry optimization schematic The research objective s of this chapter are: 1) Develop and investigate new nanodiamond (ND) slurry for BPM CMP 2) Determine the MRR and surface quality based on the new ND slurry. 3) Compare and contrast the novel ND s lurry versus industrial slurry CMP

PAGE 168

150 5.2 Introduction In order to create slurry that is novel in use and effective for the CMP process, an inves tigation into the most important aspects of the slurry is given in the following sections. While not all of the chemical aspects are investigated for this slurry, this research is a feasibility study in utilizing nanodiamond (ND) particles in polymer matr ix as a polishing solution or a final buffing solution for the CMP process. 5.2 .1 Effect of S lurry on P lanarization ( S urface Q uality) In order to achieve the strict requirements of the magnetic storage and semiconductor industries on the removal rate a nd surface roughness during CMP, the effects of the slurry must be investigated. Table 5.1 lists the output parameters for global planarization and the mechanisms by which these parameters are achieved by the slurries [175] T he slurry parameters must be optimized s o that the mechanical removal of the material is minimized because excessive mechanical removal produces high frictional forces and can thus damage the surface topography. An initial study into the surface characterization of new and novel slurry was unde rtaken based parameters in the table below.

PAGE 169

151 Table 5. 1 CMP process output and slurry mechanisms [175] Global planari ty Removal Rate Surface defecti vity Selectivity Slurry handling Formation of thin passivation layer Rapid formation of thin surface layer Rapid formation of thin surface layer Top layer chemomechanica l polishing Formation of stable slurries M inimize chemical etching Control of mech anical/interfacial properties of the surface layer M inimize mechanical polishing Bottom layer mechanical polishing Control of interparticle and particle surface interactions Minimize mechanical polishing Stress induction by abrasion to remove surface layer Control of particle size and hardness Reduction of mechanical component in slurry Steric force based repulsion in ionic systems Indentation based wear Control of particle size distribution Fracture/delamination based removal In order to make effective slurry for CMP there are several issues that must be considered before specific slurry design the slurry must : 1) M inimize the frictional forces 2) Maintain c onstant local polishing pressure 3) Reduce excess chemic al etching The frictional forces must be minimized in order to ensure th at the amount of surface defects is reduced. High frictional forces indicate that polishing is going in boundary lubrication regime which is a two body abrasion mode in which the pad and wafer are in direct contact; consequently this would indicate that t he slurry is not contributing to material removal. Lowering the pressure and increasing the polishing

PAGE 170

152 times helps to lower the COF and increases global planarity [175] Ensuring that the local pressure is constant is paramount because variable pressure leads to variable polishing rates and non uniformity in the wafer surface. Excessive chemical etching adversel y affects surface planarity and induces defects on the surfa ce such as corrosion [175, 176] The novel slurry developed in this chapter seeks to lower the COF as compared to industrial slurry, while reducing the chemical etching on BPM matrices. 5.2.2 Chemical E ffect of S lurry of M aterial R emoval R ate The re action of the slurry chemica ls on the metal to be polished, the mechanical abrasion of the particles, the interplay of the different complexing agents, oxidizers, and corrosion inhibitors are all intertwined into one process during polishing. The re have been numerous studies on the effect of the chemicals in the slurry on the wafer surfa ce I t has been concluded that the reaction rate and the creation of a passivation layer on the surface can be increased up to a limit by adding oxidizers and corrosive inhibitors [78, 153] Creation of a passivation layer weakens the metal surface allowing for the abraded particles to strike the surface and cause material removal. Figure 5. 2 is a diagram of these interactions.

PAGE 171

153 Figure 5. 2 Pad/wafer interface reactions with the slurry 5.2.3 Mechanical and M aterial P roperties T he particle concentration s all play a role in the overall material removed duri ng polishing. The generalized abrasive material removal rate (MRR) for CMP has been mode led in the literature and is shown in equation 5.1 [153] (5.1) The variable n is number of active abrasives taking part in the process and Vol removed is the volume of material removed by each abrasive. To estimate the total volume of material removed, it is necessary to estimate the total area of the pad / wafer a nd wafer / abrasive co ntact. The area of active abrasive co ntact is given by equation 5.2, (5.2)

PAGE 172

154 where A is the area of contact, x is diameter of abrasive, and on the passivating film made by the abrasive particle [129] Based on the above equations many models have been created that suggest the influence of the abrasive particle size on the material removal rate and surface finish. Jairath et al. and Xie et al observed that the polishing rate increases with both particle size and concentration [177, 178] Contrary to these findings, Bielmann et al. have found that decreased particle size led to higher polishing rates or no effect on the MRR [179, 180] Mahajan, Lee and Sign have pro posed that the CMP process is based on two removal mechanisms: an indentation based wear which dominates for large abrasive particles and a contact based mechanism which dominates for small abrasives [181] In either ca se there is a saturation point for abrasives in which an increase the amount of particles and/or the particles size results in there n o longer being an increase in the amount of material removed. A qualitative explanation for this was given by Luo, the total contact area between the wafer and pad surface asperities is occupied by the active abrasives when saturation occurs, and a furthe r increase in concentration cannot increase the number of abrasives in the cont act area, as shown in figure 5.3 [147]

PAGE 173

155 Figure 5. 3 Contact modes with abrasive weight concentrations From figure 5.3, an increase in particle size as well as concentration will lead to the same effects from the saturation of the contact area. Therefore there are two transitions of material removal regions with the increase in abrasive weight concentration or size. First the transition from a rapid increase at a small abrasive concentration to a slower linear increase region. The second transition is from the linear increase region to the saturation region at larger abrasive concentrations T his example is shown in figure 5.4

PAGE 174

156 Figure 5. 4 Example of saturation for MRR The figure gives an example of the effect seen in literature for increase the particle concentration is this effect shown for polishing on SiO2 by Singh et al [175] As shown by Singh, Bajaj and Mahajan et al., the removal rate of the silica increases with increase in particle size and concentration at low particle concentrati on s, however after a particular threshold for every given particle size the mechanism of removal changes and there is considerable decrease in removal rate with increase in particle concentration [181] This effect will be investigated for the novel slurry created in this chapter. 5.2.4 Particle S ize and H ardness As mention ed in section 5.2.3, the pa rticle size has the same effect as the concentration on the removal rate during CMP, an d a similar figure to figure 5.2 could be drawn for particle size. The particle size also has effects on the overall surface quality after polishing. An i ncrease in pa rticle size or hardness also gives rise to surface defects

PAGE 175

157 such as micro scratches that cause fatal long term device failure. Bigger and harder particles would cause deeper micro scratches, which will be very difficult to eliminate even by the final buffi ng CMP step. Therefore in the creation of the ND slurry, the particle size is kept small. Although diamond is the hardest natural material, the inclusion of the softer polymer matrices helps to reduce the effects after polishing of the harder abrasive pa rticles. 5.2.5 A brasive P article The abrasive particle in the CMP slurry serves as the mechanical mechanism for abrasion during polishing. Without the abrasive particles, the slurry would only aid in chemical corrosion of the surfaces. Selection of t he correct abrasives requires knowledge of the mechanical properties, hardness, and fabrication of the nanoabrasives. 5.2.5.1 Commercial S lurry A brasive S ynthesis The quality of the post CMP wafer surface is significantly dependent on the characteristic s of the abrasive particles present in the slurry. There are several abrasive particle type options and the table below outlines the abrasive s and the synthesis technique.

PAGE 176

158 T able 5. 2 Slurry abrasive characteristics Slurry Abrasive Synthesis Type Fumed (F), colloidal (C), Sol (S) Abrasive Use Characteristics Remarks Silica F, C,S Oxides: F,C Cu: C W: F, C Sizes vary, of medium hardness Low selectivity Polishes vast number of materials low selectivity Alumina F, C Dielectrics Copper Strong Lewis acidic surface Very Hard Amorphous Low MRR on dielectrics, Hardness can lead to surface scratches Ceria SiO 2 Low scratches and high MRR Lewis acidic surface Expensive High selectivity for SiO 2 Titania, Zirconia C, S FRAM, d ielectrics oxides Additives have high hardness and selectivity abrasives High selectivity Low MRR Used w ith other abrasives The synthesis method determines the size of the abrasive particles. Fumed abrasives tend to be chained particles that are larger in size than colloidal abrasives, which consist of discrete particles in dispersion that precipitate from a solution For the

PAGE 177

159 same solids concentration, the removal rate using a fumed abrasive is higher than that using a colloidal abrasi ve due to larger particle size. For this reason t he defect density using a fumed abrasive is also higher and the c olloidal abrasive having a uniform particle size is preferred. However, to achieve the same remova l rate as using a fumed abrasive, the solids concentration of colloidal slurry must be almost three times higher, thereby increasing the cost of the slurry. 5.2.5.2 Surface Q uality B ased on A brasives The generation of surface scratches depends on a wide variety of factors such as the process conditions as mentioned in previous sections of this chapter. Of particular interest in the present research are the characteris tics of the abrasive particles and their effect on the surface quality. A comparison of ND polymer slurry synthesized to the abrasive s used for commercial copper CMP helps to evaluate the surface quality and feasibility of the ND polymer slurry during polishing. Several other commercial abrasives have been detailed in literature and are not characterized in this dissertation [136, 156, 176, 182, 183] A lumina particles h ave been used for copper CMP due to their low selectivity and material removal of the dielectric layer below copper, but the high hardness value has lead to severe surface scratches. These particles typically have a Mohns hardness value of 9, which is onl y below the hardness of diamond which values at 10. The alumina a brasive particles can agglomerate in the slurry T he effective size of the particles can be much higher than the specification of the slurry and the agglomerated particles can make deep scratches in the surface. The deep scratches result in defects that cannot be removed by any other post processing techniques. Commonly

PAGE 178

160 used ceramic abrasive particles are much harder than the low dielectric constant materials and copper. These particles can easily scratch the surface and, if agglomerated can result in permanent scratch defects. Thus, the inherent nature of the particle plays a significant role. T he abrasive particles that result in low friction at the interface are bene ficial to the process due to the fact that lower friction helps reduce surface damage during CMP [184] The particle residue encountered after polishing along with post CMP cleaning are parameters that must also be investigated for any slurry fabricated [135, 185] 5.3 Novel N anodiamond (ND) S lurry S ynthesis Knowledge of the slurry chemistry, particle size, particle concentration, and hardness are properties that must be investigated in order to create any new or novel slurr y. The slurry developed in this dissertation contains composite particles that are inherently soft due to the presence of polymer. This alternative approach involves using responsive polymer microgels to entrap the ND particles and utilize the new slurry for CMP. In this approach, the N D particles will on the surface of the polymeric microgel (hybrid composites) and will conceivably prevent aggressive abrasions of the ND on the wafer surface resulting in smoother surfaces and reduced surface damage. The ND slurry composite is hypothesized to provide a cushioning effect to the wafer due to the soft nature of the polymer and yet achieve appreciable CMP material removal due to the abrasion of the hard ND particles. The polymer particles exhibit controllable surface hardness and chemical nature and hence are hypothesized to prevent aggressive scr atching, particle residue, and apply high mechanical stress during polishing. The incorporation of functional groups onto polymer latex surfaces to form new hybrid

PAGE 179

161 materials represents an emerging discipline for the synthesis of novel materials with divers e architectures. Polymer p article synthesis and characterization was initially conducted and verified by Dr Cecil Coutinho under the guidance of Dr. Vinay Gupta in the Department of Chemical Engineering at the University of South Florida 5.3.1 Hybrid P article S ynthesis Unless otherwise noted, all chemicals were purchased from Sigma Aldrich (WI) and used without further purification. The monomer nisopropylacrylamide (NIPAM TCI ) was recrystallized from hexane before use. With the goal of developing nove l slurry for CMP applications, polymer siloxane (hybrid) microgels were formed by the surfactant methylenebisacrylamide (0.2g) as the cross linker. Following purging with N 2 for 1h, the reaction mixture was heated in an oil bath to 75C and the ionic initiator potassium persulfate (0.1g) was added to instigate polymerization. After an initial polymerization of 2 hours, 3 (trimethoxysilyl ) propyl methacrylate (1g) was adde d to the reaction mixture and the polymerization continued for a further 90 minutes. The m icrogels formed were collected and purified by repeated centrifugation (7800g 30minutes) and re dispersed with deionized water [186 188] The nanodiamond (ND) particles were acquired from International Technology Center (ITC Raleigh NC). The particles were suspended in a water solution with 1 wt% concentration. The ND particles are 98% pure and are all in cubic phase and were measured to be 5 3 nm. The ND slurry was c omposed of the softer NIPAM microgel matrix and the harder fused ND particles. For characterization of

PAGE 180

162 were immersed in 4 ml of deionized water. The solution was put into an ice bath and was allowed to chill for twenty minutes. was immersed in 2 ml of deionized waters and the solution was put on a hot magnetic stirring plate and heated and stirred at 40C for twenty mi nutes. Heating NIPAM allows the thermally responsive polymer to denature the polymeric chains that create the spheres seen from TEM imaging. Once the NIPAM is receptive to the ND particles the two solutions are mixed together on the stirring plate for 30 minutes. The solution is then centrifuged at 5000 RPM for an additional 30 min ute s. The new slurry labeled NIPAMND is settled and dispersed in 3ml of deionized water. 5.3.2 Particle C haracterization The NIPAM particles along with the ND particles w ere characterized to ensure synthesis and fabrication of the particles was successful. Dr. Coutinho completed the dynamic light scattering characterization of the NIPAM particles while the rest of the characterization was completed during the research of this dissertation. 5.3.2.1 NIPAM D ynamic L ight S cattering [188] Microgel sizes and polydispersities were determined via dynamic light scatterin g (DLS) using a Zetasizer Nano S (Malvern PA ). Samples were sonicated prior to a nalysis. A 1ml of the microgel solution was placed into a cuvette and allowed to thermally equilibrate to a certain temperature for 10 min utes before each measurement. Data fitting was done using a multi modal algorithm supplied by Malvern. The collected correlelogr ams were fitted to diffusion co efficients and converted to a hydrodynamic diameter using the Einstein Stokes equation [186 188]

PAGE 181

16 3 5.3.2.2 Transmission E lectron M icroscopy (TEM) The NIPAMND composites were examined using TEM to visually determine the extent of ND loading and dispersion within the polymer matrix. A drop of the sample solution was diluted with ethano l and then was placed on a Formvar coated Cu TEM grid that was examined using a FEI Morgagni 268D. 5.3.2. 3 Post CMP S urface C haracterization Qualitatively, the surface quality of the BPM and blanket copper surfaces post CMP was examined by WYKO NT 9100 and a Leitz Ergolux optical microscope. The resulting surface roughness was measured from the WYKO NT 9100. The removal rates were calculated from four point probe measurements on the sample and details of those calculations are in chapter 3, section 3.2 .4.2. The initial and final thickness measurements were calculated as an average after nine different readings were taken. The bench t op CMP tested provided real time measurements of the friction coefficient during polishing and the average value after th e process had reached state has been reported. 5.4 Experimental C onditions for ND S lurry and P article S lurry T esting Three different slurries were used as a comparison for testing the validity of the N IPAMN D slurry. The slurries were tested on the MIT 854 BPM copper wafer and a blanked wafer of copper. The blanket wafer was fabricated at University of South Florida NREC and is a silicon substrate with, a 5nm layer of barium on top of the silicon substrate, as well as a 10 nm layer o f tantalum, and finally electrodeposited with 10m thick copper. The Cabot iCue 5001 colloidal alumina commercial slurry, the NIPAM polymer matrix, and the developed NIPAMND slurry have their properties displayed in

PAGE 182

164 table 5.3. All the slurries formulated were to have an equal amount of weight percentage of abrasive particles. Table 5. 3 Slurry details Slurry Name Particle type Particle Size (nm) Hardness (Mohrs) Wt% Cabot iCue 5001 Alumina 20 6 nm 9 1.5 NIPAM Hybrid polymer 500 20 nm 0 1.5 NIPAMND Hybrid ND 500 40 nm ND 10 1.5 The slurries were then employ ed for perfo rming CMP on the BPM copper wafer wafers. The testing of the slurry samples was carried out at the process conditions summarized in table 5.4. Table 5. 4 Process conditio ns for slurry testing # Parameter Value 1. Pressure 4 Psi 2. P laten speed 200 RPM 3. Carrier speed 200 RPM 4. Slider movement 3 mm/s 5. Slurry flow rate 75 mL/min 6. Time 6 0 seconds 7. Pad Rodel IC 14 00 Suba K groove pad 8. Specimen 5.5 Results and D iscussion As with chapter 4, the amount of data acquired for analysis through the WYKO NT9100 greatly exceeds the amount of reasonable space in this dissertation therefore images and characterization for the surface profile presented in the section are the images

PAGE 183

165 th at reflect the most important information of all the data. Additional data is presented in the appendix of this dissertation. The blanket copper polish experimentation was done to compare to the IC industry to determine if the ND slurry was a feasible me ans for either IC fabrication or BPM fabrication. The blanket results due not elucidate any information on BPM CMP. 5.5.1 TEM I maging The hybrid particles were synthesized using precipitation polymerizations, and all three slurries were characterized us ing TEM, and the average size from the TEM calculations is shown in table 5.3. Figure 5.5 contains the image for the three abrasive particles from the TEM imaging.

PAGE 184

166 Figure 5. 5 Slurry p articles The agglomeration for the particles in all three figures seen in above is cause for concern for the result surface quality. The massive agglomeration of the Cabot slurry particles helps further validate the necessity of new and novel slurry. 5.5.2 COF of the S lurry A brasives T he values from the CMP benchtop tester for COF during polishing for the abrasives against the BPM and blanket copper wafers are shown in figure s 5. 6 and figure 5. 7 respectively

PAGE 185

167 Figure 5. 6 COF vs BPM wafers Figure 5. 7 COF vs. blanket copper w afers From the figures above it is evident that the softer NIPAM particles have the lowest COF, whereas the Cabot and the NIPAMND particles have coefficients that are similar. This is to be expected since the hardness values ( e.g. the amount each particle will indent into the surf ace and possibly abrade the material) are larger than that for the

PAGE 186

168 NIPAM alone. Hardness is t he ability for a surface to resist plastic deformation, and this resistance can be linked to the COF. The h ardness of the NIPAM lends to the indentation based we ar r egime developed by Mahajan et a l but the size leads to the contact based wear, and this is optimal for the polishing regimes of the NIPAMND. The fact that the alumina particles have a higher COF than the NIPAMND even though the NIPAMND has harder ab rasives than the alumina particles is promising and elucidates further research into the NIPAMND slurry as a technique for CMP. 5.5.3 MRR Versus the A brasive P article The MRR for the slurries must be known in order to determine th e feasibility of t he p rocess. Figur e 5. 8 contains the MRR versus the various slurries for the polishing parameters in table 5.4 Figure 5. 8 MRR vs w afer type for analysis of slurry abrasives

PAGE 187

169 F igure 5.8 details the interactions of the slurry with the wafer surface s The NIPAM slurry achieves the lowest MRR, and this is expected due to the inability of the particles to impinge upon the harder copper surface and remove material. The removal mechanism during this polishing is sole ly due to the asperities of the pad and the wafer asperities. The NIPAMND slurry has a significantly higher MRR than the NIPAM alone, but this value still falls short of the Cabot slurry MRR. The difference in the MRR for the Cabot slurry and the NIPAMND cannot be explained solely by the hardness of the abrasive as it can with the COF data. The Cabot slurry has a chemical aspect that both the NIPAM and NIPAMND slurry do not contain, as this commercial slurry has oxidizers and passivating agents in order t o further break up the copper surface to help the abrading particles. In order to develop fully commercialized NIPAMND slurry the chemical aspects of the slurry and the interacting surface must be fully understood and investigated. 5.5.4 Surface Q uality and R oughness The resulting surface roughness (RMS) values from polishing of the blanket copper wafers and the BPM wafers are shown in table 5.5. The initial roughness value for the BPM is 352.73 1.24 nm and for the blanket copper wafers the initial value is 50.92 2.46 nm Table 5. 5 Surface roughness for slurries Slurry BPM Roughness (RMS) Blanket Cu (RMS) Cabot iCue 5001 308.23 3.45 nm 27.37 3.16 nm NIPAM 343.07 2.21 nm 41. 75 2.08 nm NIPAMND 344.25 2.72 nm 38.18 2.31 nm

PAGE 188

170 From the table above, the Cabot and the NIPAM ND slurries result in very similar surfaces although the MRR is muc h greater with the Cabot slurry The surface roughness of the blanket copper wafer with respect to the NIPAM slurry is defect free but the MRR and the resulting RMS values indicate that the NIPAM slurry is ineffective without a harder abrasive agent for copper CMP. The WYKO images of three slurries for the BPM matrix and blanket copper wafers are shown in figures 5. 9 and 5.10 respectively. Figure 5. 9 WYKO surface profiler images for BPM polishing From figure 5.9 t he NIPAM and the NIPAMND had very similar surface roughness values for the BPM wafer polish ing although t h e MRR for the slurries differed. The theory behind the matching surface roughness values arises from the ND particles that were no t removed during post CMP clean up. These particles embedded themselves in the BPM configurations and continued to cause scratches on the surface or were embedded in the matrix configurations themselves.

PAGE 189

171 Figure 5. 10 WYKO images of slurry polishing of blanket copper wafers From figure 5. 10 the results are indicative of the MRR results. The NIPAMND slurry achieves greater local and global planarity then the NIPAM particles themselves while still lacking in over all polish quality of the commercially available Cabot slurry. The NIPAMND surface was further hampered by ND embedment in the surface and this is an issue to address in the post CMP cleanup process. 5.6 Analysis of NIPAMND A brasive C oncentration With a successful synthesis and polishing using the NIPAMND slurry, an initial investigation into two different abrasive concentrations and the saturation point is studied. Due to the success of the blanket copper wafer CMP, only these wafers are included in t he study of the saturation point and the process outputs. The nomenclature used for the number of ab rasive particles is as follows:

PAGE 190

172 1) Particle (1): 6.07x 10 19 ND particles in the solution 2) Particle (2): 1.21x10 20 ND particles in the solution 3) Particle (3) : 1.81x10 20 ND particles in the solution 4) Particle (4): 3.642x 10 20 ND particles in the solution T he values for the number of ND particles in the solution were arbitrarily chosen T he novelty of the slurry requires inherent baseline testing for the number of abrasives. The same synthesis technique of the NIPAM slurry as mentioned in section 5.3 is d one for this research parameter. T he amount on ND concentration is varied. The same process parameters and characterization techniques as table 5.4 are investigated. 5.6.1 COF and the A brasive P article C oncentration Figure 5.11 shows the variation in the COF with the different abrasive concentrations used in parametric study. Figure 5. 11 COF vs number of particles for NIPAMND slurry

PAGE 191

173 Figure 5.11 details that the COF value does not significantly increase for the parameters from table 5.4, above particle (3) number. This could be due to saturation of the particles on the contact area of the wafer, with full coverage increasing the number of particles impinging the surface will not increase active particles in the polishing region and thereby the COF. 5.6.2 MRR and A brasive P article C oncentration Figure 5.12 shows the MRR versus the abrasive particle concentration for both the blanket copper wafers and the BPM wafers. Figure 5. 12 MRR vs n umber of abrasive particles for NIPAMND slurry Figure 5.12 indicates that the saturation point for the number of particles that are active in the contact area during polishing is at particle number (3). Above this number of particles the MRR no longer increases linearly and the resulting slurry particle concentration is not adv antageous for polishing. 0 10 20 30 40 50 60 70 80 90 100 0 1 2 3 4 5 MRR (nm/min) Number of Particles MRR vs Number of Particles

PAGE 192

174 5.6.3 Surface Roughness and Abrasive Particle C oncentration Table 5.6 contains the surface roughness values for the abrasive concentrations used during experimentation. T able 5. 6 WYKO surface profile data for number of particles Number of Abrasive Particles Surface Roughness (RMS) Particle (1) 38.18 nm Particle (2) 34.17 nm Particle (3) 31. 21 nm Particle (4) 30.54 nm From table 5.6 and the data fro m figures 5.11 and 5.12 increasing the number of particles and thereby the cost of the slurry does not significantly increase the output of the deliverables of the slurry ( e.g. MRR and surface roughness). For the NIPAMND slurry the optimal number of particles for copper CMP rem ains between particle number (2) and (3). 5.7 Conclusions and R emarks There has been tremendous amount of research and development of novel slurries funneling from the 1 billion dollar economic impact that the slurries have on CMP. Although there is a multitude of slurries for various applications no slurry exists that has been able to polish several different materials locally and globally without a detrimental effect to the surface finish. In this chapter new novel hybrid pol y mer based ND slurr y wer e developed for polishing on both BPM copper wafers and blanket copper wafers. The process parameters remained constant during polishing and the resulting COF, MRR,

PAGE 193

175 and surface roughness were monitored. The particles were characterized by DLS and TEM. T he ND s were suspended within the thermally responsive NIPAM polymer matrix and provided better results for polishing the then previously published NIPAM polishing results because the hard ND particles were a better abrasive for we ar than the softer NIPAM s lurry [186] Consequently the NIPAMND slu rry contained a higher COF than the NIPAM slurry and was comparable to the commercial slurries in the COF. This higher COF may be attributed to the hardness value of the ND particles. Upon successful synthesis and testing of the NIPAMND slurry, two diff erent concentrations were created to test the saturation limit for the ND particles and the removal rate. It was determined that beyond particle number (3) the MRR and surface roughness outputs remain nearly constant for polishing of the blanket wafers. Discussion of future work will be in chapter 7.

PAGE 194

176 CHAPTER 6: CMP MODELING OF MICROSTRUCTURAL VARIATION 6.1 Foreword With the rapid change of materials systems and decreased feature size, thin film microstructure and mechanical properties have become critical parameters for microelectronics reliability. This requires inherent knowledge of the mechanical properties of ma terials and an in depth understanding of the tribological phenomena involved in the manufacturing process. CMP is a semi conductor manufacturing process used to remove or planarize ultra thin metallic, dielectric, or barrier films (copper) on silicon wafe rs. The material removal rate (MRR), which ultimately effects the surface topography, corresponding to CMP is given by the standard Preston e quation, that contains the load applied, the velocity of the pad, the Preston coefficient which includes chemical dependencies, and the hardness of the material. Typically the hardness, a bulk material constant, is taken as a constant throughout the wafer and thereby included in the Preston coefficien t. Through metallurgy studies (on the micro and nano scale) it has been proven that the hardness is dependent upon grain size and orientation. This research served to first relate the crystallographic orientation of a crystal to a hardness value The second objective of this chapter is to use the hardness variation in the previously developed particle augmented mixed lubrication (PAML) model to verify the surface topography and MRR during CMP.

PAGE 195

177 6.2 Introduction Wear is the phenomenon of material removal from a surface due to in teraction with a mating surface either through micro fracture, chemical dissolution, or melting the contacting surface. In the case of plastic contact between hard and sharp material and a rel atively softer material the hard material penetrates the softer one causing fracture; this fracture can lead to micro cutting and ultimately material removal. As mentioned above CMP is utilized in the semiconductor industry for planarization of thin film metal layers on a silicon substrate. For most reliability and performance tests, knowledge of the thin film constitutive mechanical behavior is required. Mechanical properties of thin films often differ f rom those of the bulk materials due to the small grain sizes attributed to the deposition methods. Small sized grains typically contain high grain boundary volume fractions that can lead to an increase or decrease in resulting hardness depe ndent on the volume fraction [189] This can also be partially explained by the nanocrystalline structure of thin films and the fact that these films are attached to a substrate. Most research on mechanical properties has concentrated on measurements of hardnes s as function of gr ain size; however this relationship has not be en extensively investigated in relation to CMP and the resulting MRR. Thin film mechanical properties can be measured by tensile testing of freestanding films and by the micro beam cantilever deflection techni que but the easiest way is by means of nanoindentation (chapter 3, section 3.2.2) since no special sample preparation is required and tests can be performed quickly and inexpensively. For most reliability and performance tests, knowledge of the thin fil m constitutive mechanical behavior is required. Mechanical properties of thin films often differ from those of the

PAGE 196

178 bulk materials, due to the small grain sizes attr ibuted to the deposition and various annealing methodologies [105, 110] 6.3 Crystallography The re are two major factors that affect the hardness of a material; the size of grains coupled with types of grain boundaries and the individual grain orientati on, ( e.g. the crystallography). The grain boundaries disrupt the movement of dislocations in a crystal and the disruption leads to la rger applied forces required to cause the crystal to deform. This leads to a larger yield stress for plastic deformation and the smaller the grains the harder the material, this relationship is known as the Hall Petch relationship which relates grain size to yield strength, however the re is a limit to dependence on size of the grain on a micro nano scale as the relationship begins to break down for grains smaller than 1 micron and the hardness to grain size relationship on the scale has yet to be completel y investigated [190, 191] The orientation of the grain will determine how a dislocation will move. The presence of dislocations strongly influences many of the properties of real materials. The critically resolved shear stress (CRSS) is a characteristic property of a material T he slip system that it is activated under CRSS and can be measured by orienting a single crystal sample with respect to the applied stress and calculating the yield stress. Copp er is a face centered cubic (FCC ) structure which contains 12 different slip systems The CRSS effect s the yield stress of the material thereby affecting the hardness, the predominant slip plane for copper is [111]. Values of CRSS for FCC metals range from 0.34 M Pa to 0.69 MPa. The material slip system for copper is {111} <110 > and the CRSS for copper is 0.64.

PAGE 197

179 The orientation variation is dependent on the deposition method and more specifically the time and temperature at which t he target is deposited on the subs t rate. Sputtering copper directly on to silicon wafer leads to less variation while electroplating copper yields a greater variation in orientations Annealing the wafers after the deposition process causes the variation in grain size to decrease and the predominant orientation <111> results [8] 6.4 Experimental D esign A blanket set of polished orientation (100) 1 10 ohm cm SSP 4850um Prime silicon wafers were deposited with .75 of copper using the Materials Research Science Engineering Center (MRSEC) at Carnegie Mellon University. One set was electroplated and the other set utilized sputtering as the deposition method. Both sets were then anne aled at 450 degrees centigrade for 13 hours in order to allow the grains to grow on the order of several microns. Following annealing the sample was then polished using the Strabraugh chemical mechanical polisher in order to remove any oxide layers that m ay have formed from the annealing process. A Hysitron t riboi n dentor show n in figure 6.1 was then used in order to obtain the deterministic surface topography of the sputtered and electroplated surface. X Ray diffraction and orientation imaging microscopy (OIM) were utilized after annealing to determine the orientations of the grains in the sample, these were found to be the predominant <111> orientation. A raster can was implemented in order to scan a 30 m by 30 m area and provide the profilome try of th at surface. Figure 6.2 shows the surface topography of the wafer.

PAGE 198

180 Figure 6. 1 Hysitron n anoi n denter Figure 6. 2 Deterministic surface topography A nanoi n dentation method was utilized to determine the hardness variation throughout the 30 sample and a contour plot of the hardness versus (x, y) position is shown in figure 6.3 along with a hardness grid corresponding to the gray scale on the plot An indent ation depth less than 10% of the film thickness was done to avoid indentation size effects (ISE).

PAGE 199

181 Figure 6. 3 Contour map of hardness The gray scale shown here has hardness values of 1.2 GPa, the white correspond to values i n the range of 2.0 GPa, and the black with values ranging near 1.75 GPa. Elastic modulus variation was shown to range from 115 GPa to 132 GPa. 6.4.1 CMP S imulation The coefficients of load, P app equal to 100 microN ewtons, U pad and U wafer equal to 10 RPMs and Preston coefficient, k equal to 1, are all initialized in the PAML model. The Preston coefficient is set to 1 therefore the effects of the slurry are negated for this simulation. The authors determined a thorough understanding of the mechanical abr asion and the resulting effect from variable hardness is a first case scenario. Future simulations will incorporate slurry chemistry and colloidal particle effects. The experimental surface topography was imported and a tech nique developed by Dickerell e t al is utilized to convert the experimental dat a into volume pixels (voxels) Each voxel, shown in figure 6.4 contains the x y and z position in three dimensional space of each grain along with the corresponding value of hardness from experiments. A r andom pad surface topography is generated and contact is initiated with the wafer surface. The stress on individual voxels is calculated and these stresses are used to calculate individual and

PAGE 200

182 cumulative wear rates on the wafer surface. The wear distance from each voxel is calculated and then subtracted from the wafer surface. Figure 6. 4 Voxelized surface topography 6.5 Simulation R esults The results from the numerical si mulation are shown in fig ure 6 .5 Figure 6. 5 Cumulative wear rate simulation of CMP

PAGE 201

183 The MRR difference between the variable hardness and bulk modulus value (constant hardness) is 2.17 m 3 /s. Although this value is not large the chemical effects and particle dynamics of the slurry have not been incorporated in this initial investigation. This wear is calculated by the aforementioned numerical simulations. Of greater importance is the resulting surface topography as this relates directly to the viability of the integrat ed circuit or media storage device fabrication. Incorporation of the variation of hardness resulted in a surface topography with a difference in roughness from the bulk constant hardness value of 6 nm 6.6 Conclusions and R emarks A two part investigation was conducted in order to determine if a previously developed chemical mechanica l polishing (CMP) model, PAML, could be enhance d through further experimental validation. The first part involved relating the critically resolved shear stress ( CRSS) of a single crystal to an individual hardness value. An investigation relating the CRSS to the hardness value was conducted based on the orientations and hardness values from experimentally found properties. C urrently there is not an empirical mode l or equation to relate the CRSS to the hardness value. The second part of this investigation utilized the variation in hardness values from the initial study and incorporated these results into a particle augmented mixed lubrication (PAML) numerical mode l that incorporates all the mechanical physics of chemical mechanical polishing (CMP). Incorporation of the variation of hardness resulted in a surface topography with a difference in roughness from the bulk const ant hardness value of 6nm. The MRR of the process differs by 2.17 m 3 /s.

PAGE 202

184 CHAPTER 7: MULTIPHYSICS DISC US S ION OF BPM CMP 7.1 Foreword The feasibility of utilizing the CMP process for planarizing patterned media has been investigated and benchmark data has been reported. The CMP process is a multi physics process in which the machine inputs, pad and slurry characteristics affect the output parameters as shown in figure 2.16 and repeated below. Figure 7. 1 CMP process parameters

PAGE 203

185 Based on figure 7.1, the machine input parameters, pad characteristics, and slurry characteristics were evaluated to help determine the optimum polishing pressure, pad and wafer velocities, and CMP pad. A separate evaluation on the role of the CMP slurry based on the characteristics of the aforementioned parameters was also conducted During the CMP process the machine input parameters dictate the removal mechanism, resulting MRR, surface r oughness, and surface defects. The inputs for pressure and velocity are the key parameters which determine the outputs. Due to the complexity of the process, the relationship between each parameter and resulting output is not straightforward and this cha pter discusses the interactions between all of the parameters and introduces a conclusion on the feasibility of using CMP for the PM planarization process. 7 .2 Pad B ased W ear From chapter 3, it has been reported that as the pressure is increased and the velocity is held constant, the MRR increases at a faster rate than if the pressure is held constant and velocity is increased. The resulting MRR from the two cases offers insight into the fundamental science of CMP polishing, namely, pad based wear versu s slurry based wear as the main polishing theory. For the case of increasing pressure and constant velocity the main polishing mechanism is the pad, and is deemed pad based wear. The increase in pressure causes an increase in two body abrasion as the wafe r and pad are forced into contact due to the high pressures. Figure 7.2 shows the schematic of two body abrasion during polishing.

PAGE 204

186 Figure 7. 2 Pad based wear Two body abrasion results in a high COF, high MRR, high surface defects, and the boundary lubrication regime (from the Stribeck curve). It is useful to note that in the calculation of the qualitative Sommerfeld number during polishing, increasing the pressure decreases the overall Sommerfeld number shifting the curve further into boundary lubrication which was detailed in chapter 2.4.2.1. Pad based wear also results in microcrack formation during polishing as seen and explained in chapter 3. For the case of BPM, the high pressure s and fatigue of polishing the metal causes shearing of the crystals, causing work hardening on the BPM and thereby causing an increase in the mechanical properties during polish. This increase is only momentary as the crack propagation occurs through the fatigue of the metal during polishin g. The cracks formed during polishing weaken the material and also reduce the reliability of the BPM in magnetic storage hard drive use. Pad based wear also results in the highest pad wear during polishing as evidenced in chapter 4 through pad thickness and pad roughness studies. The rougher pads initially start in the boundary lubrication regime and result in higher MRR. The pads then slowly

PAGE 205

187 transition into the partial lubrication and finally full lubrication regimes during polishing ( based on Sommerf eld number and COF ). The high pad wear reduces the pad roughness and pad life but also plays a vital role in reducing the wafer surface roughness due to the high MRR. During the CMP process the two body wear is most beneficial at the start of the process due to the high wear rates but is also detrimental to the pad life. There is little slurry int eraction with pad based wear, due to increase in pressure the entrenched fluid (slurry) e polishing pad. T he refore interaction of the wafer surface and slurry only results from the chemical corrosion of the wafer surface and not from the abrasive particles. 7.3 Slurry B ased W ear The opposite is true as the velocity is increased and pressure is held constant during CMP. The inc rease in the relative velocity which is correlated to the angular velocities of the pad and wafer carrier increases the entrenchment of the slurry to the wafer surface. As velocity is increased the Sommerfeld number is increased shifting the polishing regime further to the hydrodynamic lubrication or full lubrication regime. This regime has the slurry fluid supporting the entire load (pressure) with little or no wafer to pad contact. This regime is also two body abrasion, but the abrasion oc curs between the abrasive nanoparticles and the wafer surface as shown in figure 7.3.

PAGE 206

188 Figure 7. 3 Slurry based wear The full lubrication regime results in lower MRR but less surface defects and lower surface roughness values. The wear from the abrasive particles is based on indentation or ploughing. During ploughing the slurry abrasives are harder than the wafer surfa ce being polished T he harder slurry abrasives such as alumina, silica, Titania zirconia, or nanodiamond plastically plough through the surface during polishin g by striking the surface causing the surface to plastically deform The material removal is co mpleted through the angular velocities of the pad and wafer carrier during which the ploughed wafer surface particles are removed from the surface through centripetal acceleration. Increasing the velocity during CMP of BPM increases the number of collisio ns between the slurry abrasives and the wafer surface thereby increasing ploughing. This serves to also remove the abraded material more quickly than at slower speeds. The evidence of the ploughing surfaces can be seen in chapter 3 for the high polish p abrasive particle has ploughed through the surface causing surface defects The slurry

PAGE 207

189 based wear has lower MRR than the pad based wear but the resulting surface defects and rou ghness values are better. The slurry interaction during slurry based wear is pivotal as the wafer surface is only in contact with the slurry during polishing. The chemistry, particle size, particle distribution, and particle material properties all con tribute to the MRR and resulting wafer surface. An important caveat of the slurry chemistry and particles is the abrasive agglomeration during polishing. In both cases of polishing for NDs and alumina abrasive the particles agglomerated to form larger pa rticles which are detrimental to the CMP process. As the larger particles interact with the surface the ploughing of these particles has two effects on the surface: 1) T he bigger particle removes a greater amount of material then the non agglomerated par ticles resulting in a reduction in the surface quality of the surface 2) A s the particles plough the surface the bigger agglomerated particles begins to deposit particles onto the BPM during polishing consequently both effects are undesirable for BPM CMP During polishing of the BPM the MRR was increased with an increase in the hardness of the abrasive particle and an increase in the number of particles in a solution. Due to the regime of polishing, low MRR, and high initial surface roughness of the BP M, the amount of slurry required to polish the surface to an atomically smooth surface is increase thereby increasing cost. As the slurry is the most expensive consumable conducting polishing in this regime of low pressure and high velocity is not cost ef fective for all materials.

PAGE 208

190 pad surface even though the pad is not in direct contact with the wafer surface and not being worn away (allowing for softer and less expensive pad s). Glazing is a CMP phenomenon in which the slurry abrasive become entrapped in the pad surface along with chemical etchants and wafer surface particles. The glazing effect reduces the surface roughness of the pad through the two b ody abrasion of the pa d surface by abrasive particle s (akin to the wafer surface). Results from the nanoparticle deposits on the pad surfaces were characterized in the SEM images in the appendix of this dissertation. The glazing effect decreases the pad thickness over time an d reduces the number of usable pads during polishing. 7.4 Mixed P olishing The final set of parameters that produce an overall effect on the output parameters of figure 7.1 was a variation in the pressure and the velocity during polishing. The elastohydro dyna mic regime or partial lubrication regime combines the optimum settings of pressure and velocity to ensure a balance between pad based wear and slurry based wear or the boundary lubrication regime and the full lubrication regime respectively. The mixed lubrication regime combines the contact of the pad based wear with the entrenchment of the slurry during polishing. The resulting wear is based on three body wear shown in figure 7.4.

PAGE 209

191 Figure 7. 4 Three body wear Based on this research the optimal input polishing parameters for the BPM matrices result from using relatively low polish pressure and high velocities. The Stribeck curve from these parameters on BPM CMP results in the partially lubricated polishing regime (as the Sommerfeld number is optimized for the input parameters and the in situ COF is monitored). This regime results in a median MRR from the pad based contact and atomic surface roughness of the BPM from the slurry chemistry and nanoparticle abrasive interaction. As with the slurry based wear the slurry nanoparticles must be harder than the surface they are polishing in order to ensure ploughing the wafer surfa ce. The surface chemistry must etch the surface to weaken the surface during polishing to aid in the MRR. Characteristics of the slurry for the partial lubrication regime remain unchanged from the fully lubricated regime in part because the overall task to be completed by this type of wear remains the same, namely reducing the surface roughness during polishing.

PAGE 210

192 During polishing of the partially lubricated regime the pad and wafer surface are in direct contact in combination with entrenchment of the slu rry particles ploughing the surfaces of the wafer and the pad during CMP. This is the optimal configuration for BPM polishing T he pad based wear and slurry based wear interactions were coupled during polishing of the BPM and reduction of the pad lifetim e was evident as the pad lifetime for BPM CMP falls below the IC CMP industry average of 400 700 minutes of polishing for each pad. Typical lifetimes for the commercial pads tested ranged from 80 200 minutes, far below th e industry average. SEM and UTS r esults indicate the glazing effect during polishing which is indicative of slurry based wear, while the reduction in pad thickness and pad roughness coupled with the decrease in MRR are indicators of pad based wear. Due to the atomic surface roughness re quired, high initial surface roughness of BPM and the entrenchment of the slurry into the medium between the pad and the wafer during polishing, the glazing onset was much earlier than for typical IC CMP pads dictating the need for new fabrication technol ogy for polishing pads of BPM configurations. In order to incorporate CMP as the main planarization technique for the PM configurations several process advantageous and limitations must first be well understood. CMP is able to polish the PM configuration, on both a local and global planarization scheme, to the required surface roughness values for the read /write head to work efficiently, but the drawback of the process comes at the high cost of this planarization. The lower pressures and high initial surf ace roughness of the PM result in a higher number of polishing steps to reach required outputs. The increase in process steps arises from the need to begin polishing based on pad based wear, where the contact

PAGE 211

193 mechanics of the pad and wafer dominate polish ing to wear down the high surface roughness. The pad needs to transition to the partially lubricated regime in which the process still maintains an appreciable MRR but the PM surface begins to become smoother through the entrenchment of the slurry as an a dditional removal mechanism utilizing both chemical and mechanical effects. The final polish should be maintained in the full lubrication regime in which is dictated by a low MRR and a lower surface roughness. The final polish results in slurry based wea r. It should be noted that during the course of the transitions to polish to atomically smooth surfaces the pads will need to be replaced depending on the groove type and hardness value ( k groove and shore hardness about 55 shore D for the optimized pad utilized in this experiment) for optimal output parameters. The slurry will also be a provide high cost for PM CMP as the abrasive for polishing must be harder than the wafer surface (Mohrs hardness of 9 10) and the slurry chemistry for the material must be customized for the PM. CMP is a viable and feasible process for planarization of the surface if there is no other methodologies are available but the magnetic storage industry must be willing to incur the initial high start up cost to utilize this tech nology.

PAGE 212

194 CHAPTER 8 : CONCLUSION AND FUTURE WORK 8.1 Conclusions As the superparamagnetic limit is reached, the magnetic storage industry looks to circumvent that barrier by implementing polymer and patterned media (PM) as a viable means to store a nd access data [26, 27] The issue with PM is the ability of the magnetic Chemical mechanical planarization (CMP) is a semiconductor fabrication technique used to planarize surfaces in the multi level meta llization schemes for the integrated circuit (IC) industries. The CMP technique is thereby employed to ensure that the PM is polished to surface roughness requirements of the magnetic storage industry that will allow the magnetic read/write head to move s eamlessly across the PM. Due to the novelty of PM fabrication, data on the output parameters of the CMP process based on the machine process parameters, pad properties, and slurry characteristics is extremely limited, and therefore benchmark data on a specific reversed patterned matrix was conducted and compared to standard IC fabrication CMP. A reverse bit patterned media (BPM) matrix is tested in this research for be nchmarking purposes. The BPM is fabricated with silicon as the substrate, followed by barium and tantalum adhesion layers, and silicon dioxide with copper in the PM configuration. The planarization of repeating patterns on a Cartesian grid in the reverse BPM configuration is fundamen tally similar to PM configurations, and the interest of this dissertation lies in the fundamental science with CMP of these types of heterogeneous matrices (assuming

PAGE 213

195 another material is the recession ). The research was conducted in a systematic manner sho wn in figure 8.1 [192 194] Figure 8. 1 Process schematic for BPM CMP Based on figure 8.1, the process parameters were first determined for BPM CMP. These parameters include the machine parameters, pad characteristics and slurry characteristics. The parameters were then characterized individually and optimized based on the previous integration of parameters. T he machine process parameters were first investigated, followed by the pad characteristics during polishing, and finally investigation into the slurry characteristics on BPM was conducted. Following the characterizat ion of the process, the quantitative and qualitative data from the experiments was then interpreted to determine the best applications for CMP on

PAGE 214

196 BPM based on the systematic approach utilized in solving the problem. Results from this research has implicati ons in feasibility studies of utilizing CMP as the main planarization technique for PM magnetic hard drive fabrication, sustainability in the consumables of the CMP process for fabrication, and practicability of different slurry designs to polish BPM. A separate investigation was conducted which modeled the evol ution of the mechanical properties during CMP process to determine the importance of microstructure during polishing. This investigation looked at the process parameters and evaluated the mechanic al properties to determine the best set of applications utilizing a simulation I t should still be noted that the input parameters for both sets of experiments was based on IC CMP. The input parameters for pressure and velocity dictate the resulting mate rial removal and surface roughness of the CMP process [41, 42, 59, 83, 85, 178] Results from a two factor three level statistical analysis of variance, quantitative data are reported, and Stribeck curves of polishing of the BPM indicate that pressure is major driving factor in the material removal during polishing. The optimal machine parameters for the surface roughness and material removal ar e at low polish pressure s and high relative velocities These parameters ensure there is three body abrasion between the pad, wafer surface, and abrasive particles in the slurry and polishing remains in the partial lubrication regime [55] The mechanical properties during polishing were monitored and an increase in the shear strength during polishing is attribut ed to dislocation motion toward grain boundaries during polishi ng. This phenomenon has been reported in previous literature for metal nanoi n dentation and polishing [14 5, 146, 195] The

PAGE 215

197 dislocation motion also leads to micro cracks in the BPM and future research will address this problem. Using the mach ine parameters from previously mentioned optimization, the consumables were characterized to determine their effect on polishing of BPM. The MRR, COF, wafer surface roughness, pad roughness, and pad thickness were monitored and results indicate a deterior ation of all the parameters versus polish time. This deterioration is directly linked to the sustainability life of each pad. A new non destructive ultrasound technique for evolution of the pad properties during polishing followed closely to quantitative data and this technique can be utilized to test future pads for enhanced pad life. The optimum pad for B PM polishing was the IC 1400 dual layer Suba V pad with a shore hardness of 57, and a k groove pattern. The softer, polyurethane matrix foam pad with x y groove pattern was inadequate to polish PM configuration and should not be utilized in future PM fabrications. The resulting pad life for PM polishing indicates a dramatic decrease is pad life for polishing of PM when compared to the IC CMP pads. Th e feasibility of the magnetic storage industry utilizing CMP as the planarization process must incorporate the increase in costs for pad replacement. The final consumable analyzed was the slurry for CMP of BPM. Again, using the optimized machine parame ters and optimized pad for polishing, the slurry polishing properties were evaluated and a novel nano diamond (ND) slurry was created to benchmark the data on BPM and blanket copper polishing. The process parameters remained constant during polishing and t he resulting COF, MRR, and surface roughness were monitored. The particles where characterized by DLS and TEM. The NDs were suspended within the thermally responsive polymer matrix and provided better results for

PAGE 216

198 polishing than the previously published p olymer polishing results because the hard ND particles were a better abrasive for wear than the softer polymer slurry. Upon successful synthesis and testing of the ND slurry, four different concentrations were created to test the saturation limit for the ND particles and the removal rate. It was determined that beyond inclusion of 1.81x10 20 ND particles in the solution, there is not an appreciable increase in the output metrics for polishing. Neither the ND slurry nor the polymer slurry performed better than the commercially available Cabot iCue slurry for MRR or surface roughness, partially due to the chemical effects of the Cabot slurry on the surface of the BPM. Although the commercially available slurry outperformed the ND slurry, the new ND slurry o ffered an improvement in MRR and surface roughness for blanket copper wafers when compared to polishing with the patented thermally responsive polymer [186 188] Based on a cost benefit analysis to incorporate the ND slurry to current CMP process metrics, the ND slurry is not cost effective for BPM CMP, but could be potentially used as a buffer step in the final polish of the PM. A separate study on the importance of microstructure was conducted to determine how microstructural variation affects the polishing output metrics. A two part investigation was cond ucted in order to determine if a previously developed particle augmented m ixed lubrication (PAML) model could be enhance d further through experimental validation. The first part involved relating the critically resolved shear stress (CRSS) of a single crystal to an individual hardness value, and results indicate that currently there is not an empirical model or equation to relate the CRSS to the hardness value. The second part of this investigation utilized the variation in hardness values from the initial study and incorporated these results into the PAML simulation. Incorpor ation

PAGE 217

199 of the variation of mechanical properties resulted in a surface topography with a difference in roughness from the bulk constant hardness value of 6nm, while t he MRR of the process differs by 2.17 m 3 /s. 8.2 Future W ork The benchmarking of the data for BPM CMP involves a systematic approach to solving for the optimized input parameters enhance the output parameters. The approach utilized in this dissertation took a methodical approach in which variations in each process parameter were tested individ ually, an optimization was found, and that optimized factor was imparted into the next phase as seen by the arrows in figure 8.1. This approach did not take into account a reversal in the input parameters and how the metrics would affect each other. This means that the selection of the Rodel IC1400 pad as the optimized pad is a consequences of the optimized machine parameters of low polish pressure and high velocity. A design of experiments which incorporates the effects of the individual pads or slurry on the different machine input parameters would increase the knowledge on the interactions CMP process properties. Figure 8.2 depicts a schematic of the current and future work for the process.

PAGE 218

200 Figure 8. 2 Future work The green arrows represent the work and path currently taken in this research. The yellow arrows represent the interaction of the process parameters on each other and the resulting optimization that needs to be undertaken to gain further knowledge on the proc ess. Research conducted was based on CMP of copper PM. While this is an option the magnetic storage industry has investigated most of the PM fabrication technology utilizes Co 70 Cr 18 Pt 12 nickel ferrite, or other polymer or highly magnetic metal films [11, 23, 26, 196] It should be noted that copper PM offers a worst case scenario for PM CMP, as the copper is the hardest material utilized for BPM out of the choices currently used. T hi s worst case scenario will ultimately result in a decrease in the pressures applied and number of polishing steps required for other materials in PM configuration, as copper has the highest mechanical properties and therefore greatest resis tance to plastic deformation. Consequently since the polishing steps would be de creased, the pad life would be increased based on replacing the copper with less hard and less stiff polymers and metals. The slurry interaction cannot be fully understood for other material s as the slurry

PAGE 219

201 chemistry utilized in experiments was specifically designed for copper CMP, although the abrasives in the slurry would remain effective. Research in this dissertation is a benchmark study on the CMP for PM as a intern fabrication step, du e to the novelty of the project, the amount of future work described herein is extensive and the interactions between the phases and parameters warrants future work that may not be fully understood in this dissertation. Future work for the process optimiz ation of BPM CMP would include indexing the geometrical shape of the PM being polished. The shapes of the PM, columns or bit pattern matrix form, have an effect on the tribological interaction during CMP. The PM shape determines the area of contact betwe en the wafer and pad along with the abrasives. Thus, the amount of surface asperity interaction and the particle wafer interaction depends also on the PM shapes. The fluid film that is in contact with the wafer surface also is dependent on the PM shape. Scarfo et al., conducted polishing tests at different process conditions on different wafer samples with concave, convex and intermediate surface contours [65] The study determined that the change in coefficient of friction is directly linked to changes within the shape. The variability in the shape leads to a change in the Sommerfeld number due to changes in the process conditions because of the resulting contact area and pressure changes [65] This study did not take into account shapes other than BPM configuration, based on the assumption that BP M wafers are single level wafers and the variability could be negligible. Future studies should incorporate any PM shape variability. Future work for pad characterization would incorporate the mechanical property evolution of the pads during BPM polish ing. Characterization and evaluation of the elastic modulus, shear modulus, and pad hardness during each polish step along with

PAGE 220

202 results from ultrasound system would lead to a better model for prediction pad life for BPM CMP. The commercial slurry perfo rmed better than the ND slurry created in this research, due to the effects of the slurry chemistry on the surface. Future work for the slurry would incorporate an etchant and oxiders in the slurry chemistry to help weaken the surface prior to polishing. The interaction of the thermally responsive polymer and ND particles with the slurry chemistry must be fully understood to optimize material removal. This would require inherent knowledge of chemistry and surface mechanics making it a multidisciplinary p roblem to be solved. The simulation work done in chapter 6, took into account only the mechanical interactions of the polishing process. It is a mechanical polishing simulation with the evolution of microstructure displayed during abrasive polishing. The abrasive module predi cted that the MRR during polishing changes as each copper grain is sheared away and a new grain is reveal ed which may have a different orientation resulting in a different hardness value. Further studies into the grain boundary interaction during polishin g warrants research for future work. Most grains will get harder due to the dislocations piling up at the grain boundaries while others will become weaker as they are removed from the polishing surface. This evolution of the mechanical properties due to grain boundary orientation parameters needs to be properly understood and implemented in the PAML simulation. Future work for the simulation would also incorporate the slurry chemistry interaction with the surface to accurately predict the MRR and surface roughness.

PAGE 221

203 Based on the benchmark data on the three phases of this research, CMP is a feasible process to planarize the BPM to atomically smooth surfaces. In order to incorporate the CMP process the magnetic storage industry must understand the tribolo gical iss ues that must first be overcome: 1) In order to polish the PM to atomic smoothness, low polish pressures and high velocities need to be incorporated. This will result in multiple polish steps (increasing cost) to insure local and global planarity. 2) T he CMP pads utilized to polish the PM will have a much shorter pad life than the IC CMP and this will result in an increase in the number of replacement pads as the optimal pad for polish ing will last roughly half of the pad life for IC CMP. 3) The commercial slurry for copper CMP is an adequate substitute for PM CMP, but a change in the material for polishing results in further research in the slurry chemistry for multiple materials. Successful implementation of CMP for the planarization step in PM fabricat ion must address the high initial start up cost, increase in the number of replacement pads, and increase in polishing time to reach the required surface roughness for magnetic storage devices.

PAGE 222

204 REFERENCES 1. Thompson, D. and J. Best, The future of magnetic data storage technology. IBM Journal of Research and Development, 2000. 44(3): p. 311 322. 2. Grochowski, E. and R. Halem, Technological impact of magnetic hard disk drives on storage systems. IBM Systems J ournal, 2003. 42(2): p. 338 346. 3. Weller, D. and A. Moser, Thermal effect limits in ultrahigh density magnetic recording. IEEE Transactions on Magnetics, 1999. 35(6): p. 4423 4439. 4. Skumryev, V., et al., Beating the superparamagnetic limit with excha nge bias. Nature, 2003. 423(6942): p. 850 853. 5. Dibner, B., Oersted and the Discovery of Electromagnetism. 1962. 6. Faraday, M., Experimental researches in chemistry and physics 1859: Taylor & Francis. 7. How Perpendicular Recording works 2006 [ cited 2010 September 15]; Available from : http://www.hardwaresecrets.com/article/347 8. Daniel, E., C. Mee, and M. Clark, Magnetic recording: the first 100 years 1999: Wiley IEEE Press. 9. Abram ovitch, D. and G. Franklin, A brief history of disk drive control. Control Systems Magazine, IEEE, 2002. 22(3): p. 28 42. 10. Thompson, D. and J. Best, The future of magnetic data storage tech n ology. IBM Journal of Research and Development. 44(3): p. 311 322. 11. White, R., R. Newt, and R. Pease, Patterned media: a viable route to 50 Gbit/in< sup> 2 and up for magnetic recording? Magnetics, IEEE Transactions on, 2002. 33(1): p. 990 995. 12. Majetich, S. and Y. Jin, Magnetization directions of indiv idual nanoparticles. Science, 1999. 284(5413): p. 470. 13. Weller, D. and A. Moser, Thermal effect limits in ultrahigh density magnetic recording. Magnetics, IEEE Transactions on, 2002. 35(6): p. 4423 4439.

PAGE 223

205 14. Piramanayagam, S., Perpendicular recording m edia for hard disk drives. Journal of Applied Physics, 2007. 102: p. 011301. 15. Khizroev, S. and D. Litvinov, Perpendicular magnetic recording 2004: Kluwer Academic Publishers. 16. Hoagland, A., History of magnetic disk storage based on perpendicular magnetic recording. IEEE Transactions on Magnetics, 2003. 39(4): p. 1871 1875. 17. Merritt, R. Toshiba Ships "Perpendicular Disk Drive 2005; Available from: http://infor mationweek.com/story/showArticle.jhtml?articleID=169300286 18. Challener, W., et al., Light delivery techniques for heat assisted magnetic recording. Jpn. J. Appl. Phys., Part, 2003. 1(42): p. 981 988. 19. Rottmayer, R., et al., Heat assisted magneti c recording. IEEE Transactions on Magnetics, 2006. 42(10): p. 2417 2421. 20. Gavrila, H., Heat assisted magnetic recording. Journal of Optoelectronics and Advanced Materials, 2008. 10(7): p. 1796. 21. Fabrication of magnetic nanodot arrays by using elec trochemical deposition process [cited 2010 August 12]; Available from: http://www.ec.appchem.waseda.ac.jp/research/rese05.htm 22. S. E. Lambert, I.L.S., A. M. Patlach, M. T. and M. Krounbi, IEEE Trans. Mag, 1987. MAG 23,(3690). 23. Lambert, S., et al., Beyond discrete tracks: Other aspects of patterned media. Journal of Applied Physics, 2009. 69(8): p. 4724 4726. 24. J. F. Smyth, S.S., D. R. Fredkin, D. P. Kern, and H.S. S. A. Rishton, M. Cali, T. R. Koehler, J. Appl. Phys, 1991. 69(5262). 25. G. A. Gibson, J.F.S., S. Schultz, D. P. Kern, IEEE Trans. Mag. 1991. MAG 27(5187). 26. Ross, C., et al., Fabrication of patterned media for high density magnetic storage. Journal o f Vacuum Science & Technology B: Microelectronics and Nanometer Structures, 1999. 17: p. 3168. 27. Terris, B., T. Thomson, and G. Hu, Patterned media for future magnetic data storage. Microsystem technologies, 2007. 13(2): p. 189 196. 28. Moore, G., Cram ming more components onto integrated circuits. Proceedings of the IEEE, 1998. 86(1): p. 82 85.

PAGE 224

206 29. Yu, A., Future of microprocessors. IEEE Micro, 1996. 16(6): p. 46 53. 30. Gordon, M., Solid State Circuits Conference, Digest of Technical Papers. ISSCC., IEEE International, 2003. 1: p. 20 23. 31. Davis, J. and J. Meindl, Interconnect technology and design for gigascale integration 2003: Kluwer Academic Pub. 32. Sze, S. and K. Ng, Physics of semiconductor devices 2007: Wiley Blackwell. 33. Mayer, J. a nd S. Lau, Electronic Materials Science : Macmillan Press. 34. Thompson, S., et al., 130nm logic technology featuring 60nm transistors, low k dielectrics, and cu interconnects. Intel Technology Journal, 2002. 6(2): p. 5 13. 35. Chau, R., et al. 30nm Physi cal Gate Length CMOS Transistors with 1.0 ps n MOS and 1.7 ps p MOS Gate Delays 2000: IEEE; 1998. 36. Kuhn, K. CMOS scaling beyond 32nm: challenges and opportunities 2009: ACM. 37. Ghate, P., J. Blair, and C. Fuller, Metallization in microelectronics. Thin Solid Films, 1977. 45(1): p. 69 84. 38. Quirk, M. and J. Serda, Semiconductor manufacturing technology 2000: Prentice Hall Upper Saddle River, NJ. 39. Ryan, J., et al., The evolution of interconnection technology at IBM. IBM Journal of Research and Development, 1995. 39(4): p. 371 382. 40. Hopwood, J., Ionized physical vapor deposition of integrated circuit interconnects. Physics of Plasmas, 1998. 5: p. 1624. 41. Zantye, P., A. Kumar, and A. Sikder, Chemical mechanical planarization for microelect ronics applications. Materials Science and Engineering: R: Reports, 2004. 45(3 6): p. 89 220. 42. Steigerwald, J., S. Murarka, and R. Gutmann, Chemical mechanical planarization of microelectronic materials 1997: Wiley Interscience. 43. Murarka, S., et al., Advanced multilayer metallization schemes with copper as interconnection metal. Thin Solid Films, 1993. 236(1 2): p. 257 266. 44. Xuan, Y., et al., Atomic layer deposited nanostructures for graphene based nanoelectronics. Applied Physics Letters, 2008. 92: p. 013101.

PAGE 225

207 45. Zantye, P., Processing, reliability and integration issues in chemical mechanical planarization 2005, University of South Florida. 46. Mudhivarthi, S., Process optimization and consumable development for Chemical Mechanical Planarization (CMP) processes. 2007. 47. Bakli, M., et al., Materials and processing for 0.25 [mu] m multilevel interconnect. Microelectronic Engineering, 1997. 33(1 4): p. 175 188. 48. Stavreva, Z., et al., Chemical mechanical polishing of copper for in terconnect formation. Microelectronic Engineering, 1997. 33(1 4): p. 249 257. 49. Zantye, P., et al., Metrology and characterization of application specific chemical mechanical polishing pads. Journal of Vacuum Science & Technology A: Vacuum, Surfaces, an d Films, 2005. 23: p. 1392. 50. Shon Roy, L., A. Wiesnoski, and R. Zorich, Advanced Semiconductor Fabrication Handbook. Integrated Circuit Engineering Corporation, Scottsdale, AZ, 1998: p. 5 58. 51. Preston, F., The theory and design of plate glass polis hing machines. J. Soc. Glass Technol, 1927. 11: p. 214256. 52. Stein, D., J. Cecchi, and D. Hetherington, Atomic force microscopy, lateral force microscopy, and transmission electron microscopy investigations and adhesion force measurements for elucidati on of tungsten removal mechanisms. J. Mater. Res, 1999. 14(9): p. 3696. 53. Landis, H., et al., Integration of chemical mechanical polishing into CMOS integrated circuit manufacturing. Thin Solid Films, 1992. 220(1 2): p. 1 7. 54. Li, S. and R. Miller, C hemical mechanical polishing in silicon processing 2000: Academic Pr. 55. Bhushan, B. and P. Ko, Introduction to tribology. Applied Mechanics Reviews, 2003. 56: p. B6. 56. Ahmadi, G. and X. Xia, A model for mechanical wear and abrasive particle adhesion during the chemical mechanical polishing process. Journal of the Electrochemical Society, 2001. 148: p. G99. 57. Mazaheri, A. and G. Ahmadi, Modeling the effect of bumpy abrasive particles on chemical mechanical polishing. Journal of the Electrochemical Society, 2002. 149: p. G370.

PAGE 226

208 58. Luo, Q., D. Campbell, and S. Babu, Chemical mechanical polishing of copper in alkaline media. Thin Solid Films, 1997. 311(1 2): p. 177 182. 59. Steigerwald, J., et al., Chemical processes in the chemical mechanical polishing of copper. Materials Chemistry and Physics, 1995. 41(3): p. 217 228. 60. Li, Y., M. Hariharaputhiran, and S. Babu, Chemical mechanical polishing of copper and tantalum with silica abrasives. Journal of Materials Research, 2001. 16(4): p. 1066 10 73. 61. Ein Eli, Y., M. Auinat, and D. Starosvetsky, Electrochemical and surface studies of zinc in alkaline solutions containing organic corrosion inhibitors. Journal of power sources, 2003. 114(2): p. 330 337. 62. Wolf, S. and R. Tauber, Silicon Proces sing for the VLSI Era, Volume 1, Process Technology 1986: Lattice Press, Sunset Beach, CA., USA. 63. Stribeck, R., Characteristics of Plain and Roller Bearings (in German), Zeit. VDI, 1902. 76: p. 1341 1432. 64. Hosali, S. and E. Busch, Analyzing damag e from ultralow k CMP. Solid State Technology, 2005. 48(11): p. 33. 65. Scarfo, A., et al., In situ measurement of pressure and friction during CMP of contoured wafers. Journal of the Electrochemical Society, 2005. 152: p. G477. 66. Luo, J. and D. Dornfe ld, Optimization of CMP from the Viewpoint of Consumable Effects. Journal of the Electrochemical Society, 2003. 150: p. G807. 67. Li, W., et al., The effect of the polishing pad treatments on the chemical mechanical polishing of SiO2 films. Thin Solid Fil ms, 1995. 270(1 2): p. 601 606. 68. Nishioka, T., K. Sekine, and Y. Tateyama. Modeling on hydrodynamic effects of pad surface roughness in CMP process 1999. 69. Micheal Fury, R.R., Steven Holland, CMP consumables market reaches middle age. 2008. 70. Ludema, K., Friction, wear, lubrication: a textbook in tribology 1996: CRC. 71. Li, Y., Microelectronic applications of chemical mechanical planarization 2007: Wiley Interscience. 72. Philipossian, A. and S. Olsen, Effect of slurry flow rate on pad life during interlayer dielectric CMP. Journal of the Electrochemical Society, 2004. 151: p. G436.

PAGE 227

209 73. Ludema, K. and W. Friction, Lubrication: A Textbook in Tribology 1996, CRC Press, Boca Raton. 74. Bibby, T. and K. Holland, Endpoint detection for CMP. Journal of Electronic Materials, 1998. 27(10): p. 1073 1081. 75. Das, T., et al., Online end point detection in CMP using SPRT of wavelet decomposed sensor data. IEEE Transactions on Semiconductor Manufacturing, 2005. 18(3). 76. Gitis, N., Tribology iss ues in CMP. 2007. 77. Terrell, E., Application of a particle augmented mixed lubrication model to chemical mechanical polishing: Modeling and experimentation 2007, University Microfilms International, P. O. Box 1764, Ann Arbor, MI, 48106, USA. 78. Luo, J. and D. Dornfeld, Material removal mechanism in chemical mechanical polishing: theory and modeling. IEEE Transactions on Semiconductor Manufacturing, 2001. 14(2). 79. Chekina, O., L. Keer, and H. Liang, Wear Contact Problems and Modeling of Chemical Mec hanical Polishing. Journal of the Electrochemical Society, 1998. 145: p. 2100. 80. Seok, J., et al., Multiscale material removal modeling of chemical mechanical polishing. Wear, 2003. 254(3 4): p. 307 320. 81. Castillo Mejia, D. and S. Beaudoin, A locall y relevant wafer scale model for CMP of silicon dioxide. Journal of the Electrochemical Society, 2003. 150: p. G581. 82. Fu, G., et al., A plasticity based model of material removal in chemical mechanicalpolishing (CMP). IEEE Transactions on Semiconductor Manufacturing, 2001. 14(4): p. 406 417. 83. Tseng, W. and Y. Wang, Re examination of Pressure and Speed Dependences of Removal Rate during Chemical Mechanical Polishing Processes. Journal of the Electrochemical Society, 1997. 144: p. L15. 84. Tseng, W., J. Chin, and L. Kang, A comparative study on the roles of velocity in the material removal rate during chemical mechanical polishing. Journal of the Electrochemical Society, 1999. 146: p. 1952. 85. Zhao, B. and F. Shi, Chemical mechanical polishing: Thre shold pressure and mechanism. Electrochemical and Solid State Letters, 1999. 2: p. 145.

PAGE 228

210 86. Wei, C. and R. Harrington, Computation of the parameters of multiconductor transmission lines in two dielectric layers above a ground plane. Department of Electric al and Computer Engineering, Syracuse Univ., Tech. Rep. TR 82 12, 1982. 87. Jiang, B. and G. Muldowney. Computational Solid Mechanics Modeling of Asperity Deformation and Pad Wafer Contact in CMP 2007: Warrendale, Pa.; Materials Research Society; 1999. 88. Pdra, P. and S. Andersson, Wear simulation with the Winkler surface model. Wear, 1997. 207(1 2): p. 79 85. 89. Sawyer, W., Surface shape and contact pressure evolution in two component surfaces: application to copper chemical mechanical polishing. Tr ibology Letters, 2004. 17(2): p. 139 145. 90. Feng, C., et al., A Contact Mechanics Based Model for General Rough Pads in Chemical Mechanical Polishing Processes. Journal of the Electrochemical Society, 2009. 156: p. H601. 91. Terrell, E., M. Kuo, and C. Higgs. An approach to modeling particle based and contact based wear in CMP 2007: Warrendale, Pa.; Materials Research Society; 1999. 92. Terrell, E. and C. Higgs III, A particle augmented mixed lubrication modeling approach to predicting chemical mechanical polishing. Journal of Tribology, 2009. 131: p. 012201. 93. Johnson, K., Contact mechanics. 1985. Cambridge Uni. 94. Johnson, K., Contact mechanics 1987: Cambridge Univ Pr. 95. Johnson, K., Contact mechanics and the wear of metals. Wear, 1995 190(2): p. 162 170. 96. Nanz, G. and L. Camilletti, Modeling of chemical mechanical polishing: a review. IEEE Transactions on Semiconductor Manufacturing, 1995. 8(4): p. 382 389. 97. Tichy, J., A.A.L.D.D. G. Dalmaz, and M. Priest, Contact mechanics and lubrication hydrodynamics of chemical mechanical planarization in Tribology and Interface Engineering Series 2001, Elsevier. p. 63 68. 98. Zhao, Y. and L. Chang, A micro contact and wear model for chemical mechanical polishing of silicon wafers. Wear, 2002. 252(3 4): p. 220 226. 99. Higgs III, C., et al., A mixed lubrication approach to predicting CMP fluid pressure modeling and experiments. Journal of the Electrochemical Society, 2005. 152: p. G193.

PAGE 229

211 100. Chang, S., et al., Pattern effects on planariza tion efficiency of Cu electropolishing. Jpn. J. Appl. Phys., Part, 2002. 1(41): p. 7332. 101. Nguyen, V., H. VanKranenburg, and P. Woerlee, Dependency of dishing on polish time and slurry chemistry in Cu CMP. Microelectronic Engineering, 2000. 50(1 4): p. 403 410. 102. Pan, J., et al. Copper CMP integration and time dependent pattern effect 1999. 103. Li, S. and R. Miller, Chemical Mechanical Polishing in Silicon Processing, vol. 63 2000, Academic Press, New York. 104. Borst, C., et al., Chemical mechanical polishing mechanisms of low dielectric constant polymers in copper slurries. Journal of the Electrochemical Society, 1999. 146: p. 4309. 105. Volinsky, A., J. Vella, and W. Gerberich, Fracture toughness, adhesion and mechanical properties of lo w K dielectric thin films measured by nanoindentation. Thin Solid Films, 2003. 429(1 2): p. 201 210. 106. Yang, L., et al., Improving electrical performance of Cu= porous ultra low k dielectrics single damascene lines. Electronics Letters, 2004. 40: p. 12 107. Balakumar, S., et al., Peeling and delamination in Cu/SiLK (TM) process during Cu CMP. Thin Solid Films, 2004. 462: p. 161 167. 108. Shen, L., et al., Determination of the hardness and elastic modulus of low k thin films and their barrier layer fo r microelectronic applications. Microelectronic Engineering, 2003. 70(1): p. 115 124. 109. Wang, Y., et al., The mechanical properties of ultra low dielectric constant films. Thin Solid Films, 2004. 462: p. 227 230. 110. Volinsky, A., N. Moody, and W. Ge rberich, Interfacial toughness measurements for thin films on substrates. Acta Materialia, 2002. 50(3): p. 441 466. 111. Shan, L., S. Danyluk, and J. Levert. Interfacial Pressure Measurements at Chemical Mechanical Polishing Interfaces 1999: Warrendale, Pa.; Materials Research Society; 1999. 112. Sundararajan, S., et al., Two dimensional wafer scale chemical mechanical planarization models based on lubrication theory and mass transport. JOURNAL ELECTROCHEMICAL SOCIETY, 1999. 146(2): p. 761 766. 113. Par k, T., T. Tugbawa, and D. Boning. Pattern dependent modeling of electroplated copper profiles 2002: IEEE.

PAGE 230

212 114. Park, T., T. Tugbawa, and D. Boning. Overview of methods for characterization of pattern dependencies in copper CMP 2000: Citeseer. 115. Park, T., et al. Pattern and process dependencies in copper damascene chemical mechanical polishing processes 1998: Citeseer. 116. Tae Park, T.T., Duane Boning (MIT); Steve Hymes, Konstantin Smekalin and (SEMATECH) (2000) MIT/SEMATECH 854 AZ Copper on Lo w K Chemical Mechanical Polishing (CMP) Characterization Test Chip International SEMATECH Technology Transfer. 117. Oliver, W. and G. Pharr, Improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. Journal of Materials Research, 1992. 7(6): p. 1564 1583. 118. Taylor, C., et al. Investigation of ultralow load nanoindentation for the patterning of nanostructures 2002. 119. Pharr, G., Measurement of mechanical properties by ultra low loa d indentation. Materials Science and Engineering A, 1998. 253(1 2): p. 151 159. 120. Nix, W., Mechanical properties of thin films. Metallurgical and Materials Transactions A, 1989. 20(11): p. 2217 2245. 121. Cammarata, R., et al., Nanoindentation study o f the mechanical properties of copper nickel multilayered thin films. Applied Physics Letters, 2009. 56(19): p. 1862 1864. 122. Fang, T. and W. Chang, Nanomechanical properties of copper thin films on different substrates using the nanoindentation techniq ue. Microelectronic Engineering, 2003. 65(1 2): p. 231 238. 123. Suresh, S., T. Nieh, and B. Choi, Nano indentation of copper thin films on silicon substrates. Scripta Materialia, 1999. 41(9): p. 951 958. 124. Zecchino, C.L.a.M., WYKO Surface Profilers T echnical Reference Manual 1999, Veeco Metrology Group: Tuscon, Arizona. p. 1 3. 125. Amstutz, H., Surface Texture: The Parameters Warner & Swasey, Sheffield Measurement Division). 126. Sikder, A. and A. Kumar, Mechanical and tribological properties of interlayer films for the damascene Cu chemical mechanical planarization process. Journal of Electronic Materials, 2002. 31(10): p. 1016 1021.

PAGE 231

213 127. Ronay, M., Second order normal force brought about by polishing with a polyurethane pad. Journal of the Elec trochemical Society, 2004. 151: p. G847. 128. Vijayakumar, A., et al., Polishing mechanism of tantalum films by SiO2 particles. Microelectronic Engineering, 2003. 70(1): p. 93 101. 129. Jeng, Y. and P. Huang, Impact of abrasive particles on the material removal rate in CMP. Electrochemical and solid state letters, 2004. 7: p. G40. 130. Fuchs, K. The conductivity of thin metallic films according to the electron theory of metals 1938: Cambridge Univ Press. 131. Overwijk, M., F. Van den Heuvel, and C. Bul le Lieuwma, Novel scheme for the preparation of transmission electron microscopy specimens with a focused ion beam. Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures, 1993. 11: p. 2021. 132. Giannuzzi, L. and F. Stevie, A review of focused ion beam milling techniques for TEM specimen preparation. Micron, 1999. 30(3): p. 197 204. 133. TSENG, W. and Y. WANG. The Intercorrelation Between Microstructure and Chemical Mechanical Polish of Metal Thin Films 1999: Materials Resea rch Society. 134. Shi, F. and B. Zhao, Modeling of chemical mechanical polishing with soft pads. Applied Physics A: Materials Science & Processing, 1998. 67(2): p. 249 252. 135. Zhang, F. and A. Busnaina, Submicron particle removal in post oxide chemical mechanical planarization (CMP) cleaning. Applied Physics A: Materials Science & Processing, 1999. 69(4): p. 437 440. 136. Li, Z., et al., Determining the effects of slurry surfactant, abrasive size, and abrasive content on the tribology and kinetics of copper CMP. Journal of the Electrochemical Society, 2005. 152: p. G299. 137. Coppeta, J. and C. Rogers, Dual emission laser induced fluorescence for direct planar scalar behavior measurements. Experiments in Fluids, 1998. 25(1): p. 1 15. 138. Lu, J., et al., Measurements of slurry film thickness and wafer drag during CMP. Journal of the Electrochemical Society, 2004. 151: p. G241. 139. Sikder, A., et al. Delamination studies in Cu ultra low k stack 2003. 140. Bonivel, J., et al. The Effect of Microstru cture on Chemical Mechanical Polishing Process of Thin Film Metals in Proceedings of the STLE/ASME International Joint Tribology Conference 2007: ASME.

PAGE 232

214 141. Fracassi, F., R. d'Agostino, P. Favio J. Electrochem Soc, 1992. 139(9): p. 2636 2643. 142. Boni vel, J., et al. THE Effect of Microstructure on Chemical Mechanical Polishing Process of Thin Film Metals in Materials Research Society 2009. San Francisco. 143. Cottrell, A., Theory of brittle fracture in steel and similar metals. Trans. Met. Soc. AIME, 1958. 212(192): p. 56. 144. Ebrahimi, F., et al., Mechanical properties of nanocrystalline nickel produced by electrodeposition. Nanostructured Materials, 1999. 11(3): p. 343 350. 145. Minor, A., et al., A new view of the onset of plasticity during the nanoindentation of aluminium. Nature Materials, 2006. 5(9): p. 697 702. 146. Shan, Z., et al., Mechanical annealing and source limited deformation in submicrometre diameter Ni crystals. Nature Materials, 2007. 7(2): p. 115 119. 147. Luo, J., Integrated modeling of chemical mechanical planarization/polishing (CMP) for integrated circuit fabrication: from particle scale to die and wafer scales 2003, UNIVERSITY OF CALIFORNIA. 148. Kim, I., The semiconductor industry. Chemical Engineering Progress, 2002. 9 8(7): p. 12 14. 149. Hooper, B., G. Byrne, and S. Galligan, Pad conditioning in chemical mechanical polishing. Journal of Materials Processing Technology, 2002. 123(1): p. 107 113. 150. Zantye, P., et al. Investigation of mechanical integrity and its eff ect on polishing for novel polyurethane polishing pad 2004. 151. Wang, D., et al., Von Mises Stress in Chemical Mechanical Polishing Processes. Journal of the Electrochemical Society, 1997. 144: p. 1121. 152. Srinivasa Murthy, C., et al., Stress distrib ution in chemical mechanical polishing. Thin Solid Films, 1997. 308: p. 533 537. 153. Runnels, S. and L. Eyman, Tribology Analysis of Chemical Mechanical Polishing. Journal of the Electrochemical Society, 1994. 141: p. 1698. 154. Sikder, A., et al. Evalu ation of mechanical and tribological behavior and surface characteristics of CMP pads 2001: Warrendale, Pa.; Materials Research Society; 1999. 155. Oliver, M., Chemical mechanical planarization of semiconductor materials 2004: Springer Verlag.

PAGE 233

215 156. Mano cha, C., Chemical mechanical planarization: Study of conditioner abrasives and synthesis of nano zirconia for potential slurry applications. University of South Florida, 2008. 157. Hepburn, C., Polyurethane elastomers. C. Hepburn, Applied Science Publishe rs, Barking(Essex) 1982, 402, 1982. 158. Saunders, J. and K. Frisch, Polyurethanes: chemistry and technology. Interscience, New York, 1962. 1. 159. Bajaj, R., et al. Effect of polishing pad material properties on chemical mechanical polishing (CMP) processes 1994: MATERIALS RESEARCH SOCIETY. 160. Szycher, M., Szycher's handbook of polyurethanes 1999: CRC. 161. Wolf, R., et al., Polyurethane pad covering for gel filled articles 1997, Google Patents. 162. Cruz, J., et al., Composite polish pad for CMP 1999, Google Patents. 163. Sperling, L., Introduction to physical polymer science 2001: Wiley Online Library. 164. Doy, T., et al., Impact of novel pad groove designs on removal rate and uniformity of dielectric and copper CMP. Journal of the E lectrochemical Society, 2004. 151: p. G196. 165. Muldowney, G. and D. James. Characterization of CMP pad surface texture and pad wafer contact 2004. 166. A. Philiposian S.O., Chemical Mechanical Planarization. Japan Journal of Applied Physics, 2003. 4 2: p. 6371. 167. Lu, H., Y. Obeng, and K. Richardson, Applicability of dynamic mechanical analysis for CMP polyurethane pad studies. Materials Characterization, 2002. 49(2): p. 177 186. 168. Lu, H., et al., Quantitative analysis of physical and chemical changes in CMP polyurethane pad surfaces. Materials Characterization, 2002. 49(1): p. 35 44. 169. Totzke, D.G., Ultrasound of chemical mechanical planarization pads in Electrical Engineering 2000, University of South Florida: Tampa. 170. Totzke, D., et al. Non destructive characterization of CMP pads using scanning ultrasonic transmission 2001.

PAGE 234

216 171. Performance Plastics Shore Conversions : Brokenhurst. p. 2. 172. Stein, D., et al., Optical interferometry for surface measurements of CMP pads. Journal o f Electronic Materials, 1996. 25(10): p. 1623 1627. 173. Kondo, S., et al., Abrasive free polishing for copper damascene interconnection. Journal of the Electrochemical Society, 2000. 147: p. 3907. 174. Tsai, T. and S. Yen, Localized corrosion effects an d modifications of acidic and alkaline slurries on copper chemical mechanical polishing. Applied Surface Science, 2003. 210(3 4): p. 190 205. 175. Singh, R., et al., Fundamentals of dielectric and metal chemical mechanical planarization (CMP). MRS Bull, 2 002. 27: p. 752. 176. Tsai, T., Y. Wu, and S. Yen, A study of copper chemical mechanical polishing in urea hydrogen peroxide slurry by electrochemical impedance spectroscopy. Applied Surface Science, 2003. 214(1 4): p. 120 135. 177. Jairath, R., et al. Consumables for the chemical mechanical polishing (CMP) of dielectrics and conductors 1994: Materials Research Society. 178. Xie, Y. and B. Bhushan, Effects of particle size, polishing pad and contact pressure in free abrasive polishing. Wear, 1996. 200( 1 2): p. 281 295. 179. Bielmann, M., U. Mahajan, and R. Singh, Effect of particle size during tungsten chemical mechanical polishing. Electrochemical and solid state letters, 1999. 2: p. 401. 180. Cook, L., Chemical processes in glass polishing. Journal of Non Crystalline Solids, 1990. 120(1 3): p. 152 171. 181. Mahajan, U., S. Lee, and R. Singh. Effect of slurry abrasive size on polish rate and surface quality of silicon dioxide films 2000: The Electrochemical Society. 182. Basim, G. and B. Moudgil, Effect of soft agglomerates on CMP slurry performance. Journal of Colloid and Interface Science, 2002. 256(1): p. 137 142. 183. Basim, G. and B. Moudgil, Slurry design for chemical mechanical polishing, KONA Power Technol. Jpn, 2003. 21: p. 178. 184. Cho i, W., et al., Effect of slurry ionic salts at dielectric silica CMP. Journal of the Electrochemical Society, 2004. 151: p. G185. 185. Zhang, F., A. Busnaina, and G. Ahmadi, Particle adhesion and removal in chemical mechanical polishing (CMP) and post CMP cleaning. Journal of Electrochemical Society, 1999. 146(7).

PAGE 235

217 186. Coutinho, C., et al., Novel ceria polymer microcomposites for chemical mechanical polishing. Applied Surface Science, 2008. 255(5): p. 3090 3096. 187. Coutinho, C. and V. Gupta, Formation a nd properties of composites based on microgels of a responsive polymer and TiO2 nanoparticles. Journal of Colloid and Interface Science, 2007. 315(1): p. 116 122. 188. Coutinho, C., Multi Functional Composite Materials for Catalysis and Chemical Mechanica l Planarization in Chemical and Biomedical Engineering 2009, University of South Florida: Tampa. 189. Volinsky, A., et al. Microstructure and mechanical properties of electroplated Cu thin films 2001: Warrendale, Pa.; Materials Research Society; 1999. 190. Hall, E., The deformation and ageing of mild steel: III discussion of results. Proceedings of the Physical Society. Section B, 1951. 64: p. 747. 191. Petch, N., The cleavage strength of polycrystals. J. Iron Steel Inst, 1953. 174(1): p. 25 28. 192. Carney, C. and D. Durham, Optimization of hardness by the control of microwave power in TiN thin film deposited by electron cyclotron resonance assisted sputtering in a nitrogen plasma. Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 1999. 17: p. 2535. 193. Carney, C. and D. Durham, Establishing the relationship between process, structure, and properties on titanium nitride films deposited by electron cyclotron resonance assisted reactive sputtering. II. A process model. Journal of V acuum Science & Technology A: Vacuum, Surfaces, and Films, 1999. 17: p. 2859. 194. Kim, S. and D. Durham, Microscopic studies of flank wear on alumina tools. Journal of Tribology, 1991. 113: p. 204. 195. Gane, N., The direct measurement of the strength o f metals on a sub micrometre scale. Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 1970. 317(1530): p. 367 391. 196. Albrecht, M., et al., Magnetic coercivity patterns for magnetic recording on patterned media. A pplied Physics Letters, 2003. 83(21): p. 4363 4365.

PAGE 236

218 A PPENDIX A : PHASE I MACHINE PARAMETERS

PAGE 237

219 APPENDIX A (CONT) A .1 Machine Parameters WYKO I mages Figure A. 1 1 Psi 0.2 relative velocity

PAGE 238

220 APPENDIX A (CONT) Figure A. 2 1 Psi 0.8 relative v elocity

PAGE 239

221 APPENDIX A (CONT) Figure A. 3 1 Psi 1.2 relative velocity Figure A. 4 3 Psi 0.2 relative velocity

PAGE 240

222 A PPENDIX A (CONT) Figure A. 5 3 P si 0.8 relative velocity Figure A. 6 3 P si 1.2 relative velocity

PAGE 241

223 APPENDIX A (CONT) Figure A. 7 6 Psi 0.2 relative velocity Figure A. 8 6 Psi 0.8 relative velocity

PAGE 242

224 APPENDIX A (CONT) Figure A. 9 6 Psi 1.2 relative velocity

PAGE 243

225 APPENDIX A (CONT) A .2 COF V ersus MRR Figure A. 10 COF vs. MRR for BPM

PAGE 244

226 APPENDIX A (CONT) A .3 SEM C haracterization of BPM S urfaces Figure A. 11 SEM of BPM delamination at 6 Psi

PAGE 245

227 APPENDIX B: PHASE II: PAD CHARACTERIZATION DATA

PAGE 246

228 APPENDIX B (CONT) B .1 Pad R oughness E volution Figure B. 1 Pad 1 (10 p olishes)

PAGE 247

229 APPENDIX B (CONT) Figure B. 2 Pad 1 (20 p olishes) Figure B. 3 Pad 1 (30 p olishes)

PAGE 248

230 APPENDIX B (CONT) Figure B. 4 Pad 1 (40 p olishes) Figure B. 5 Pad 1 (50 p olishes)

PAGE 249

231 APPENDIX B (CONT) Figure B. 6 Pad 2 (10 p olishes) Figure B. 7 Pad 2 (20 p olishes)

PAGE 250

232 APPENDIX B (CONT) Figure B. 8 Pad 2 (30 p olishes) Figure B. 9 Pad 2 (40 p olishes)

PAGE 251

233 APPENDIX B (CONT) Figure B. 10 Pad 2 (50 p olishes) Figure B. 11 Pad 3 (10 p olishes)

PAGE 252

234 APPENDIX B (CONT) Figure B. 12 Pad 3 (20 p olishes) Figure B. 13 Pad 3 (30 p olishes)

PAGE 253

235 APPENDIX B (CONT) Figure B. 14 Pad 3 (40 p olishes) Figure B. 15 Pad 3 (50 p olishes)

PAGE 254

236 APPENDIX B (CONT) B.2 SEM P ad C haracterization Figure B. 16 SEM morphology evolution

PAGE 255

237 APPENDIX B (CONT) Figure B. 17 Pad (2) SEM morphology evolution

PAGE 256

238 APPENDIX B (CONT) Figure B. 18 Pad (3) SEM morphology evolution

PAGE 257

239 APPENDIX B (CONT) B.3 UTS Characterization Figure B. 19 Pad (1) UTS characterization

PAGE 258

240 APPENDIX B (CONT) Figure B. 20 Pad (2) UTS characterization

PAGE 259

241 APPENDIX B (CONT) Figure B. 21 Pad (3) UST characterization

PAGE 260

242 APPENDIX B (CONT) B.4 Pad C haracterization R esults Figure B. 22 CMP p ad r oughness vs COF Figure B. 23 BPM w afer r oughness vs. COF

PAGE 261

243 APPENDIX B (CONT) Figure B. 24 Wafer roughness vs. pad roughness Figure B. 25 MRR vs. p ad roughness

PAGE 262

A BOUT THE AUTHOR Joe completed his bachelors at University of South Carolina (USC) in mechanical engineering and he completed his M asters at USC in biomechanical engineering Following USC, he switched his research interest and went to Carnegie Mellon University (CMU) to study CMP from a materials science aspect. At CMU he was a fellow of the material research and science engineering center (MRSEC) He obtained his 2 nd Masters from Carnegie Mellon University (CMU) in 2008 He then joined Dr. magnetic storage devices. He has received numerous awards and grants to tr avel to China and Brazil to present his research on CMP while publishing 2 journal papers with another 3 papers submitted at this time. He will accept a position in the department of defense upon graduation and looks to return to academia after gaining ind ustry experience.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004770
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Bonivel Jr., Joseph T..
0 245
Consumable process development for chemical mechanical planarization of bit patterned media for magnetic storage fabrication
h [electronic resource] /
by Joseph T. Bonivel Jr..
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Dissertation (PHD)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: As the superparamagnetic limit is reached, the magnetic storage industry looks to circumvent the barrier by implementing patterned media (PM) as a viable means to store and access data. Chemical mechanical polishing (CMP) is a semiconductor fabrication technique used to planarize surfaces and is investigated as a method to ensure that the PM is polished to surface roughness parameters that allow the magnetic read/write head to move seamlessly across the PM. Results from this research have implications in feasibility studies of utilizing CMP as the main planarization technique for PM fabrication. Benchmark data on the output parameters of the CMP process, for bit patterned media (BPM), based on the machine process parameters, pad properties, and slurry characteristics are optimized. The research was conducted in a systematic manner in which the optimized parameters for each phase are utilized in future phases. The optimum results from each of the phases provide an overall optimum characterization for BPM CMP. Results on the CMP machine input parameters indicate that for optimal surface roughness and material removal, low polish pressures and high velocities should be used on the BPM. Pad characteristics were monitored by non destructive technique and results indicate much faster deterioration of all pad characteristics versus polish time of BPM when compared to IC CMP. The optimum pad for PM polishing was the IC 1400 dual layer Suba V pad with a shore hardness of 57, and a k-groove pattern. The final phase of polishing evaluated the slurry polishing properties and novel nanodiamond (ND) slurry was created and benchmarked on BPM. The resulting CMP output parameters were monitored and neither the ND slurry nor the thermally responsive polymer slurry performed better than the commercially available Cabot iCue slurry for MRR or surface roughness. Research results indicate CMP is a feasible planarization technique for PM fabrication, but successful implementation of CMP for planarizing PM must address the high initial start up cost, increase in the number of replacement pads, and increase in polishing time to reach the required surface roughness for magnetic storage devices.
590
Advisor: Ashok Kumar, Ph.D.
653
Tribology
Coefficient of friction
Wear
Superparamagentic limit
Magnetic hard drive fabrication
690
Dissertations, Academic
z USF
x Mechanical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4770