<%BANNER%>

Quantification of Particle Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics

Permanent Link: http://ufdc.ufl.edu/UFE0042479/00001

Material Information

Title: Quantification of Particle Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics
Physical Description: 1 online resource (121 p.)
Language: english
Creator: Khanna, Aniruddh
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: agglomeration, cmp, control, deagglomeration, defectivity, quality, quantification, slurry, stability, stress
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: As we move towards future technologies of 22 nm nodes and beyond, there are stricter requirements on the amount of defects that are allowed during the fabrication process. Chemical mechanical polishing (CMP) forms a critical part of chip manufacturing process and its performance decides the quality of the device been fabricated. For accurate CMP process, material removal rate and surface finish are of utmost importance. These parameters are greatly determined by the slurry characteristics such as particle size, type, shape, mechanical properties and chemistry. The presence of large (agglomerated) particles in CMP slurries has been shown to increase defects during CMP process. The previous work on studying agglomeration behavior in CMP slurries has been qualitative or at best semi-quantitative in nature. One of the objectives of this work is to study agglomeration behavior of CMP slurries under quantified shear rates. A novel method is developed that uses rheometer as a device to study agglomeration behavior of CMP slurries at different quantified shear rates. The resulting changes in oversize particle distribution of slurries are measured using single particle optical sensing system (SPOS) like Accusizer 780. We measured threshold shear rates, at which each of these slurries starts showing agglomeration behavior under applied shear rate for any duration up to 2500 s. The threshold value increases from acidic (1000 s-1) to basic (2000 s-1) slurries, as a result of increase in inter-particle repulsive forces between silica particles. De-agglomeration i.e. breaking of oversize particles is observed, when these slurries are subjected to less than threshold shear rates. Another objective of this work is to characterize the nature of agglomerates that are formed due to shear-induced agglomeration in various CMP slurries. The nature of agglomerates is determined by analyzing oversize particle distribution curves. The agglomerates formed in neutral and acidic pH slurry are relatively harder than the agglomerates formed in basic pH silica slurry. Finally, agglomeration index (AI) is used to calculate stability ratio of CMP slurries with different chemistries. Stability ratio, unlike agglomeration index, is independent of shear rate and is shown to be a better way to define slurry stability.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Aniruddh Khanna.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Singh, Rajiv K.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042479:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042479/00001

Material Information

Title: Quantification of Particle Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics
Physical Description: 1 online resource (121 p.)
Language: english
Creator: Khanna, Aniruddh
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: agglomeration, cmp, control, deagglomeration, defectivity, quality, quantification, slurry, stability, stress
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: As we move towards future technologies of 22 nm nodes and beyond, there are stricter requirements on the amount of defects that are allowed during the fabrication process. Chemical mechanical polishing (CMP) forms a critical part of chip manufacturing process and its performance decides the quality of the device been fabricated. For accurate CMP process, material removal rate and surface finish are of utmost importance. These parameters are greatly determined by the slurry characteristics such as particle size, type, shape, mechanical properties and chemistry. The presence of large (agglomerated) particles in CMP slurries has been shown to increase defects during CMP process. The previous work on studying agglomeration behavior in CMP slurries has been qualitative or at best semi-quantitative in nature. One of the objectives of this work is to study agglomeration behavior of CMP slurries under quantified shear rates. A novel method is developed that uses rheometer as a device to study agglomeration behavior of CMP slurries at different quantified shear rates. The resulting changes in oversize particle distribution of slurries are measured using single particle optical sensing system (SPOS) like Accusizer 780. We measured threshold shear rates, at which each of these slurries starts showing agglomeration behavior under applied shear rate for any duration up to 2500 s. The threshold value increases from acidic (1000 s-1) to basic (2000 s-1) slurries, as a result of increase in inter-particle repulsive forces between silica particles. De-agglomeration i.e. breaking of oversize particles is observed, when these slurries are subjected to less than threshold shear rates. Another objective of this work is to characterize the nature of agglomerates that are formed due to shear-induced agglomeration in various CMP slurries. The nature of agglomerates is determined by analyzing oversize particle distribution curves. The agglomerates formed in neutral and acidic pH slurry are relatively harder than the agglomerates formed in basic pH silica slurry. Finally, agglomeration index (AI) is used to calculate stability ratio of CMP slurries with different chemistries. Stability ratio, unlike agglomeration index, is independent of shear rate and is shown to be a better way to define slurry stability.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Aniruddh Khanna.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Singh, Rajiv K.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042479:00001


This item has the following downloads:


Full Text

PAGE 1

1 QUANTIFICATION OF PARTICLE AGGLOMERATION DURING CHEMICAL MECHANICAL POLISHING OF METALS AND DIELECTRICS By ANIRUDDH JAGDISH KHANNA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Aniruddh Jagdish Khanna

PAGE 3

3 To my family and friends

PAGE 4

4 ACKNOWLEDGMENTS I would like to express sincere gratitude to my advisor Prof. Rajiv K. Singh for his guidance and support throughout my academic pursuit at the University of Florida. His encouragement for improvement in thought process and constant focus on critical thinking made this work more interesting and helped in developing a great thought process. His words have always encouraged me to work harder and enjoy my work. I would like to acknowledge my supervisory committee: Prof. Stephen Pearton, Prof. D avid Norton, Prof. Timothy Anderson and Prof. Brent Gila. Special thanks to Thomas (Tom) Gassett for his constant moral support through out my stay at University of Florida, without which this work would not have been possible. His constant appreciation of my work encouraged me to work harder and strive for excellence and I am grateful to him for everything he did for me. I am also grateful to my parents for their constant support, blessings and giving me complete freedom for pursuing my academic pursuits and personal interests. I am grateful to my sister Hitu who has always motivated me to work harder, stay focused and give my best in whatever I do. I am also grateful to my eldest sister Ritu, brother in law Pradeep and my cute nephew Krishang, who always l ooks up to me, which in turn drove me to do better. I would like to thank Dr. Purushottam Kumar for his help and numerous intellectually stimulating discussions, academic or otherwise, which helped me extremely in doing this work and am grateful to him for that. A special thanks to Sachin for being a great friend, especially when I needed a friend to discuss all sorts of things here in Gainesville. I have learnt a great deal of things from him and have developed a great thought process from the discussions we had, when it comes to making all kinds

PAGE 5

5 of decisions. I would also like to thank all my friends for all the good times I had in Gainesville and tons of things I learnt from them. I would like to thank all past and present lab mates Dr. Feng Chi Chang, Dr Jaeseok Lee, Dr. Karthik Ramani, Dr. Seung Young, Dr. Sejin Kim, Dr. Taekon Kim and would be doctors Sushant Gupta and Myoung Hwan for their help and numerous intellectually stimulating discussions, academic or otherwise.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES .......................................................................................................... 9 LIST OF ABBREVIATI ONS ........................................................................................... 13 ABSTRACT ................................................................................................................... 14 CHAPTER 1 INTRODUCTION .................................................................................................... 16 Chemical Mechanical Planarization ........................................................................ 16 Slurry Delivery ........................................................................................................ 16 Outline of Research ................................................................................................ 17 2 LITERATURE REVIEW .......................................................................................... 20 Chemical Mec hanical Planarization Process .......................................................... 20 Smoluchowski Theory for Aggregation of Particles ................................................. 21 Slurry Stability ......................................................................................................... 22 Electrostatic St ability ........................................................................................ 23 Stabilization Using Polymers ............................................................................ 24 Agglomeration of Particles in Chemical Mechanical Polishing (CMP) Slurries ....... 27 Defectivity Due to Change in Particle Size .............................................................. 27 Effect of Stress Induced Particle Agglomeration on Defectivity .............................. 29 Effect of Ionic Strength in PumpInduced Agglomeration i n Slurries ....................... 30 Effect of Surfactants in Slurry on Stress Induced Agglomeration ............................ 31 Agglomeration Index for Quantifying Stability of Slurries ........................................ 32 Determination of Agglomeration Index (AI) ............................................................. 33 3 MEASUREMENTS FOR CHEMICAL MECHANICAL POLISHING SLURRIES ...... 54 Introduction ............................................................................................................. 54 Oversize Particle Distribution Measurements ......................................................... 54 Light Ext inction (LE) Method ............................................................................ 55 L igh t Scattering (LS) Method ............................................................................ 56 Rheometer .............................................................................................................. 58 Zeta Potential Measurements ................................................................................. 59

PAGE 7

7 4 STUDY OF AGGLOMERATION BEHAVIOR OF CHEMICAL MECHANICAL POLISHING SLURRY BY RHEOLOGICAL MEASUREMENTS UNDER VARIED SHEAR ENVIRONMENT .......................................................................... 63 Introduction ............................................................................................................. 63 Experimental ........................................................................................................... 64 Results and Discussions ......................................................................................... 65 Summary ................................................................................................................ 67 5 SILICA SLURRY STABILIT Y UNDER DIFFERENT CHEMICAL AND SHEAR ENVIRONMENT ..................................................................................................... 73 Introduction ............................................................................................................. 73 Experimental ........................................................................................................... 74 Results and Discussi on ........................................................................................... 75 Neutral pH ........................................................................................................ 75 Acidic pH .......................................................................................................... 76 Comparison of Normalized Concentration of Silica Slurries with Different pH .. 77 Summary ................................................................................................................ 78 6 CHARACTERIZATION OF THE NATURE OF AGGLOMERATES: HARD VS. SOFT USING RHEOLOGICAL MEASUREMENTS ................................................ 88 Introduction ............................................................................................................. 88 Experimental ........................................................................................................... 89 Results and Discussion ........................................................................................... 90 Basic pH ........................................................................................................... 91 Neutral pH ........................................................................................................ 91 Acidic pH .......................................................................................................... 92 Comparison of Nature of Agglomerates Formed in Slurries with Different pH .. 93 Summary ................................................................................................................ 93 7 MEASUR EMENT OF STABILITY RATIO: A PARAMETER FOR DEFINING SLURRY STABILITY INDEPENDENT OF SHEAR RATES .................................. 103 Introduction ........................................................................................................... 103 Experimental ......................................................................................................... 105 Results and Discussion ......................................................................................... 106 Summary .............................................................................................................. 108 8 CONCLUSIONS ................................................................................................... 112 LIST OF REFERENCES ............................................................................................. 115 BIOGRAPHICAL SKETCH .......................................................................................... 121

PAGE 8

8 LIST OF TABLES Table page 1 1 Various variables in chemical m echanical polishing (CMP) ................................ 19 2 1 Important principles for designing CMP slurries ................................................. 36 2 2 Interconnect surface requirements at near term technology nodes .................... 36 2 3 Agglomeration index of various chemical mechanic al polishing (CMP) slurries 37 3 1 Various particle size analysis measurements ..................................................... 60 4 1 Various shear rates and times for silica slurry .................................................... 69 5 1 Various shear rates and times for silica slurry .................................................... 80 6 1 Various low shear rates and times for silica slurry with pH 10 after being subjected to 3000 s1 and 1000 s ........................................................................ 95 7 1 Different shear rates and times for various silica slurries ................................. 110 7 2 Assigned name for various slurries ................................................................... 110 7 3 Agglomeration index (AI) for different slurries at various shear rates (s1) ........ 110 7 4 Agglomeration indexes (AI) for dif ferent slurries at shear rate of 3000 s1 ........ 110 7 5 Stability ratio for various slurries ....................................................................... 110

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Interaction of electrical double layer ................................................................... 38 2 2 Steric stabilization provided by terminally adsorbed polymer chains .................. 38 2 3 Schematic showing bridging flocculation and re stabilization by adsorbed polymer chains ................................................................................................... 39 2 4 Cumulative concentration vs. particle size for various turnovers for different pumps ................................................................................................................. 40 2 5 Normalized oversize particle distribution for bellows, diaphragm and magnetically levitated centrifugal pumps ............................................................ 41 2 6 Optical images of BD1 and ultra low k wafers polished by maglev and positive displacement pump circulated slurries ................................................... 42 2 7 Cumulative concentration of circulated slurries at pH 2, 7 and 11 with and without 0.1M KCl, and the mean value of normalized oversize particles ............ 43 2 8 Force vs. tip substrate distance between a silica substrate and silica probe in supernatant slurries at pH 2, 7, and 11, with and without 0.1M KCl ................... 44 2 9 Zeta potential of 30 nm silica slurries for different pH, with and without 0.1M KCl ...................................................................................................................... 44 2 10 Correlation between repulsive inter particle forces and mean values of normalized oversize particles for different pH, with and without 0.1M KCl ......... 45 2 11 AFM images and surface roughness of low k wafers polished by circulated silica slurries at different pH with and without 0.1M KCl ..................................... 46 2 12 Comparison of particle agglomeration in unformulated and formulated silica slurries ................................................................................................................ 47 2 13 Plot of zeta potential vs. pH in silica slurry with surfactants ................................ 47 2 14 Plot of zeta potential vs. pH: potential curves of formulated ceria slurries with varying surfactants ............................................................................................. 48 2 15 Comparison of shear induced particle agglomeration in unformulated and formulated ceria slurries ..................................................................................... 48 2 16 Plot of zeta potential vs. pH in ceria slurry with cationic surfactant ..................... 49

PAGE 10

10 2 17 AFM images and surface roughness of silica wafers polished by unformulated and formulated ceria slurry (cationic surfactant) at pH 6 ................... 50 2 18 Schematics for determination of agglomeration index (AI) ................................. 51 2 19 Tail distributions of as received slurry and circulated slurries by positive displacement and magnetically levitated centrifugal pumps ............................... 52 2 20 Effect of external shear stress on agglomer ation index (AI): experimental and modeled tail distributions in positive displacement and maglev pumps .............. 53 3 1 Schematics for optical scheme of a light extinction (LE) type sensor ................. 61 3 2 Detector signal output in LE type sensor, resulting from momentary passage of particle through view volume .......................................................................... 61 3 3 Area of illumination at LE detector, decreased by passage of particle of area A through view volume ..................................................................................... 62 3 4 Detector signal output in light scattering (LS) type sensor, resulting from passage of p article through view volume ............................................................ 62 4 1 Normalized oversi ze particle distribution for 100 s1 and various times ............... 70 4 2 Normalized oversize particle distribution for 1000 s1 and various times ............. 70 4 3 Normalized oversize particle distribution for 1500 s1 and various times ............. 71 4 4 Normalized oversize particle distribution for 2000 s1 and various times ............. 71 4 5 Normalized oversize particle distribution for 3000 s1 and various times ............. 72 5 1 Normalized concentration for 100 s1 and various times at pH~7 ........................ 81 5 2 Normalized concentration for 1000 s1 and various times at pH~7 ...................... 81 5 3 Normalized concentration for 1500 s1 and various times at pH~7 ...................... 82 5 4 Normalized concentration for 2000 s1 and various times at pH~7 ...................... 82 5 5 Normalized concentration for 3000 s1 and various times at pH~7 ...................... 83 5 6 Normalized concentration for 100 s1 and various times at pH~2 ........................ 83 5 7 Normalized concentration for 1000 s1 and various times at pH~2 ...................... 84 5 8 Normalized concentration for 1500 s1 and various times at pH~2 ...................... 84 5 9 Normalized concentration for 2000 s1 and various times at pH~2 ...................... 85

PAGE 11

11 5 10 Normalized concentration for 3000 s1 and various times at pH~2 ...................... 85 5 11 Comparison of deagglomeration of different pH silica slurries for 100 s1 and 100 s ................................................................................................................... 86 5 12 Comparison of deagglomeration of different pH silica slurries for 100 s1 and 1000 s ................................................................................................................. 86 5 13 Comparison of deagglomeration of different pH silica slurries for 100 s1 and 2500 s ................................................................................................................. 87 6 1 Flow chart of experimental procedure ................................................................ 96 6 2 Normalized oversize particle distribution for 100 s for various shear rates at pH~10 ................................................................................................................. 97 6 3 Normalized oversize particle distribution for 1000 s for various shear rates at pH~10 ................................................................................................................. 97 6 4 Comparison of deagglomeration in as received and stressed basic pH silica slurries for 100 s1 and 1000 s ............................................................................ 98 6 5 Normalized oversize particle distribution for 100 s for various shear rates at pH~7 ................................................................................................................... 98 6 6 Normalized oversize particle distribution for 1000 s for various shear rates at pH~7 ................................................................................................................... 99 6 7 Comparison of deagglomeration in as received and stressed neutral pH silica slurries for 100 s1 and 1000 s ................................................................... 99 6 8 Normalized oversize particle distribution for 100 s for various shear rates at pH~2 ................................................................................................................. 100 6 9 Normalized oversize particle distribution for 1000 s for various shear rates at pH~2 ................................................................................................................. 100 6 10 Comparison of deagglomeration in as received and stressed acidic pH silica slurries for 100 s1 and 1000 s .......................................................................... 101 6 11 Comparison of deagglomeration in different pH silica slurries for 100 s1 and 100 s ................................................................................................................. 101 6 12 Com parison of de agglomeration in different pH silica slurries for 100 s1 and 1000 s ............................................................................................................... 102 7 1 Schematic illustration of fitting experi mental and modeled tail distributions to obtain the agglomeration index ......................................................................... 111

PAGE 12

12 7 2 Stability ratio v/s different shear rates (s1) ....................................................... 111

PAGE 13

13 LIST OF ABBREVIATION S AFM Atomic force microscopy AI Agglomeration index CMP Chemical mechanical polishing/planarization IEP Iso electric point LE Light extinction LS Light scattering RMS Root mean square SEM Scanning electron microscopy SPOS Single particle optical sensor

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy QUANTIFICATION OF PARTICLE AGGLOMERATION DURING CHEMICAL MECHANICAL POLISHING OF METALS AND DIELECTRICS By Aniruddh Jagdish Khanna December 2010 Chair: Rajiv K. Singh Major: Materials Science and Engineering As we move towards future technologies of 22 nm nodes and beyond, there are stricter requirements on the amount of defects that are allowed during the fabrication process. Chemical mechanical polishing (CMP) forms a critical part of chip manuf acturing process and its performance decides the quality of the device been fabricated. For accurate CMP process, material removal rate and surface finish are of utmost importance. These parameters are greatly determined by the slurry characteristics such as particle size, type, shape, mechanical properties and chemistry. The presence of large (agglomerated) particles in CMP slurries has been shown to increase defects during CMP process. The previous work on studying agglomeration behavior in CMP slurries has been qualitative or at best semi quantitative in nature. One of the objectives of this work is to study agglomeration behavior of CMP slurries under quantified shear rates. A novel method is developed that uses rheometer as a device to study agglomerat ion behavior of CMP slurries at different quantified shear rates. The resulting changes in oversize particle distribution of slurries are measured using single particle optical sensing system (SPOS) like Accusizer 780. We measured threshold shear rates, at which each of these slurries starts showing

PAGE 15

15 agglomeration behavior under applied shear rate for any duration up to 2500 s. The threshold value increases from acidic (1000 s1) to basic (2000 s1) slurries, as a result of increase in inter particle repulsi ve forces between silica particles. Deagglomeration i.e. breaking of oversize particles is observed, when these slurries are subjected to less than threshold shear rates. Another objective of this work is to characterize the nature of agglomerates that a re formed due to shear induced agglomeration in various CMP slurries. The nature of agglomerates is determined by analyzing oversize particle distribution curves. The agglomerates formed in neutral and acidic pH slurry are relatively harder than the agglom erates formed in basic pH silica slurry. Finally, agglomeration index (AI) is used to calculate stability ratio of CMP slurries with different chemistries. Stability ratio, unlike agglomeration index, is independent of shear rate and is shown to be a better way to define slurry stability.

PAGE 16

16 CHAPTER 1 INTRODUCTION Chemical Mechanical Planarization Chemical m echanical p lanarization (CMP) is the most common planarization technique for both local and global planarization. With the growing miniaturization of electronic devices, there are stricter requirements on the amount of defects that are allowed during the fabrication process. CMP forms a critical part of chip manufacturing process and its performance decides the quality of the device been fabricated. The science of CMP is very different from traditional semiconductor manufacturing technology such as film deposition, ion implantation, lithography, thermal annealing, vapor phase processing [ 1] For accurate CMP process, material removal rate and surface finish are of utmost importance. These parameters are greatly determined by the type of slurry used and slurry characteristics, which are largely determined by externally induced forces and slurry chemistry. There are more than 20 input and output variables that need to be optimized to get a desired CMP performance, some of the variables are listed in Table 11 [2 ] For CMP slurries, the particle characteristics pl ay a big role in determining surface finish and material removal rate. The increase in oversize particle concentration during slurry delivery is one of the major sources of defects during CMP. Slurry Delivery The presence of large oversize particles (>0.5 micron) in CMP slurries could be due to slurry blending or slurry distribution system, which uses pumps for slurry delivery. Feng Chi et al. [4] studied the effect of different pumps like positive displacement, bellows and magneti c levitated centrifugal pump on CMP slurry particle agglomeration

PAGE 17

17 characteristics. They studied the effect of shear induced by these pumps on agglomeration of commonly used CMP slurries. The effect of shear rate on agglomeration becomes severe with changes in slurry chemistry by modifying pH, addition of salt and surfactants. The changes in oversize particle distribution in these slurries were studied using single particle optical sensor (SPOS) system like Accusizer 780. Outline of Research Chapter 3 consi sts of the details regarding the theory of instruments used for characterization of CMP slurries like Accusizer 780, which is a single particle optical sensor (SPOS) system. Rheometer is a device used to study variation in fluid viscosity as a function of shear rate. Zetaplus is a device used to measure zeta potential, the potential that is the closest approximation to the surface potential of particle. Chapter 4 develops a novel technique to study agglomeration behavior of CMP slurries as a function of quantified shear rate. A method is developed to use rheometer as a device to apply different shear rates to CMP slurry and the change in oversize particle distribution is measured using Accusizer 780. Chapter 5 shows application of novel technique to study ag glomeration behavior of CMP slurries with different slurry chemistries under varied shear rates and the change in oversize particle distribution is measured using Accusizer 780. Chapter 6 focuses on characterizing the nature of agglomerates that are forme d when the slurry is subjected to various shear rates. The method developed to determine nature of agglomerates is applied to slurry with different chemistries. The nature of agglomerates is determined by analyzing oversize particle distribution curves. Chapter 7 shows calculation of agglomeration index (AI) for different quantified shear rates using

PAGE 18

18 rheometer and is used to calculate stability ratio of CMP slurry. Stability ratio is measured for slurries with different internal slurry chemistries and calculated at different quantified shear rates. Unlike agglomeration index, stability ratio is independent of shear rate and shown to be a better way to define slurry stability. Chapter 8 gives the conclusions derived from the research work carried out in this dissertation and provides direction for future research.

PAGE 19

19 Table 11. Various variables in chemical mechanical polishing (CMP) [2 ] Input Variables Micro scale Nano scale Parameters Interactions Particle Characteristics Pad Chemo mechanical Size Contact angle Pressure on the pad Dynamics of surface layer formation Size distribution Thickness Shape Particles on the pad Uniformity Mechanical properties Pressure Rate of formation Chemistry Coverage Layer removal mechanism Dispersion Concentration Agglomeration Oversize particles Abrasion frequency Chemical concentration and distribution Slurry Chemistry Chemical and mechanical Oxidizers Contact mode Etching pH stabilizers Direct Mechanical removal Complexing agents Mixed Dispersants Hydroplaning Concentration pH and pH drift Down pressure and linear velocity Pad characteristics Mechanical properties Topography Conditioning Substrate characteristics Feature size Feature density

PAGE 20

20 CHAPTER 2 LITERATURE REVIEW Chemical Mechanical Planarization Process Chemical m ech anical p lanarization (CMP) is a widely used method for planarization of oxide and metal layers in semiconductor industry. CMP has emerged as the fastest growing operation in the semiconductor manufacturing industry and is expected to show growth in the future. The rapid increase of CMP is due to introduction of copper based interconnects for logic and other devices [5, 6] The market size of CMP equipment and consumables i.e. slurries, pads is over $1 billion [7] CMP is both local and global planarization process where the wafer gets planarized as a result of relative motion between a wafer and a polishing pad in the presence of slurry while applying pressure. It has become the most widely used method for polishing a wafer to a required thickness and planarity. One of the unique aspects of CMP is its ability to polish both metal and dielectric layers in multilevel metallization interconnect structures when a well designed slurry and pad is used. The following is the mechanism of CMP that explains process of planarization on a wafer surface: [8] i. Initially chemical reaction by the slurry forms a soft chemically modified surface layer that can be removed easily ii. This soft che mically modified layer is mechanically removed by slurrys abrasive component (generally silica, alumina or ceria particle depending on the substrate to be polished), applied pressure and relative velocity of polishing. Preston equation (31) that gives rate of removal of material is given as: [9 11] R = kPV [2 1] where R = rate of removal

PAGE 21

21 k = constant for given tool and process conditions, including parameters such as oxide hardness, polishing slurry and polishing pad P = applied pressure V = relative velocity between the wafer and polishing pad Smoluchowski Theory for Aggregation of Particles It is clear that agglomeration of abrasive particles in slurry due to shear stresses induced by pumps is a critical issue in CMP process today especially with stricter requirements on defectivity during CMP. A detailed understanding of agglomeration process due to pumps is the need of the day, which would help in predicting slurry behavior and hence its stability. Smoluchowski theory [12 14] gives a detail account of aggregation in colloidal suspension due to shear forces. Particle transport brought by fluid motion gives a tremendous increase in the rate of interparticl e collisions and aggregation brought about in this way is known as orthokinetic aggregation. The fundamental assumption is that the particle collisions are binary and proportional to the particle concentration. The rate of change of k fold aggregates where k = i + j is given by [2 2] where kij represents second order rate constant for aggregation in absence of any electrostatic interactions. The first term on right hand side of equation represents rate of formation of k fold aggregates and second term represents loss of k fold aggregates by collision or aggregation with other aggregates. Assuming that colloidal particles are subjected to uniform laminar shear, kij is given as:

PAGE 22

22 kij = 4/3 G (ai + aj)3 [2 3] where G is shear rate and ai, aj are particle sizes The rate of aggregation will be significantly influenced by the electrostatic interactions between particles in the aqueous system. Electrostatic interactions provide a potential barrier, which hinders the particle agglomeration process. The aggregation rate then becomes [2 4] where W is the stability ratio The term stability ratio is defined as the ratio of rapid aggregation rate in absence of any el ectrostatic interactions to the slow aggregation rate in presence of electrostatic interactions. High stability ratio implies more stable colloidal dispersion, which in our case is CMP slurry. Slurry Stability In CMP, unstable slurry can lead to significan t defects during wafer polishing. The important issues related to slurry performance in CMP are material removal rate, global planarity, surface topography (dishing and erosion), surface defectivity (including roughness, scratches, dents and delamination), and particle contamination. It is critical to understand the mechanism active during CMP for optimally designing slurry formulations. The most important fundamental parameters that need to be optimized for achieving acceptable characteristics in CMP slurr y are listed in Table 21 [8, 15 21] These parameters in turn are obtained after extensive measurements that are detailed elsewhere [8, 15 21] Highly stable CMP slurry is cr itical to reduce process dependent

PAGE 23

23 defectivity like scratches and particle residue. The slurry handling process should not lead to formation of large particle agglomerates. Commonly used methods for achieving slurry stability are: i. Electrostatic interactio n between particles ii. Polymer stabilization Electrostatic Stability When two same charged particles come close to each other in an aqueous system, there is an interaction between diffuse layers that surrounds these particle surfaces. The osmotic pressure causes these particles to move away due to its higher counter ion concentrations ( Figure 21 ) [22, 23] Thickness of these diffuse layers is given by Debye length 1/ [2 5] where r is the relative permittivity, 0 is the permittivity of vacuum, R is the gas constant, T is temperature, F is Faradays constant, Ci is the molar concentration of any ion I in solution phase and Zi is the valence of ion in solution phase. For the case of spherical particles each of size a having Debye length i.e. diffuse layer thickness much greater than particle size i.e. ka << 1, repulsive potential when these two part icles approach close to each other is given by [2 6]

PAGE 24

24 where 0 is the surface potential, H is the nearest distance between particle surfaces and k is Boltzmanns constant. According to Derjaguin LandauVerwey Overbeek (DLVO) theory, the total potential ( V total) in colloidal systems is given as V total = V A + V R [2 7] where V A is Van Der Waals potential due to attraction and V R is repulsive double layer potent ial. The stability of colloidal particles is determined by the balance between V A and V R For similar particles of radius a having centers separated by distance H, Van Der Waals attraction potential V A is given as [2 8] where H is the nearest distance between particle surfaces and A is Hamaker constant. From this equation it is evident that as particle size increases, attraction force between particles increases. For achieving same degree of colloidal stability as smaller size particle s, a thicker stabilizing layer will be required that will give stronger repulsive forces between particles and thereby hindering agglomeration of particles. Stabilization Using Polymers Stabilization of colloidal dispersion in aqueous systems could be achi eved by addition of surfactants i.e. ionic or nonionic polymers. The length of polymer chains that are adsorbed on particle surfaces decides the extent of repulsive forces. Work by Basim et al. [24] showed that when cationic surfactant like trimethyl ammonium bromide (CnTAB) was added to silica slurries, the slurry gets stabilized as a result of additional repulsive forces. When concentration of CnTAB exceeds critical micelle c oncentration (CMC), it results in formation of micelles which thereby gets adsorbed on silica

PAGE 25

25 surfaces. These adsorbed micelles gives rise to repulsive forces when particles interact and thereby preventing coagulation of particles. These kind of repulsion between silica particles were observed for surfactants with carbon chains of CnTAB larger than 8 i.e. n > 8. However, surfactants also provide lubricating effect as observed in case of CnTAB, which results in lower material removal rate. Thus, stabilization of silica slurry by higher repulsion in this case was observed at the cost of low material removal rate. Steric stabilization is another type of polymer stabilization method for colloidal dispersions in aqueous systems [22, 25] The most effective steric stabilizers are polymers that have some affinity for the surface and get adsorb in such a way, that segments of polymer chains extend into dispersion medium. When these particles wit h adsorbed polymers on their surface, approach each other the adsorbed layers come into contact and on further approach there is interpenetration of hydrophilic chains (Figure 22 2 3) [12 ] These chains being hydrated, on overl apping undergo dehydration which increases free energy resulting in repulsion between particles [12] The work by Napper et. al. [2 6] shows that polymer stabilization occurs due to disfavored thermodynamics governed by following equation [2 9] where G F is the change in free energy of two approaching polymer particles, H F and S F are changes in enthalpy and entropy, T is the temperature, subscript F represents flocculation. From thermodynamics, it is known that there should be decrease in free energy G F for flocculation to occur. Varying the right hand side of above equation in

PAGE 26

26 such a way that G F never decreases; we can achieve conditions where it is difficult for flocculation to occur. This could be achieved in three different ways: i. Enthalpic stabilization: Having positive changes of the enthalpy and entropy in such a way that HF is larger than SF and floccu lation is disfavored. However, increasing the temperature can cause flocculation to occur. ii. Entropic stabilization: Having negative changes of the enthalpy and entropy in such a way that HF is smaller than SF and flocculation is disfavored. However, lowering the temperature can cause flocculation to occur. iii. Enthalpic Entropic stabilization: Having positive change of the enthalpy and negative change of entropy so that free energy increases and flocculation is disfavored. Flocculation cannot be achieved in this case even by changing the temperature. For the adsorbed layer it could be fair to assume that these layers have lower Hamaker constant resulting in smaller Van Der Waals attraction. So the work of adsorbed polymeric layers is to limit the attraction between the particles by keeping them finite distance apart, in the range where VA is smaller. So, even if particles come into contact with each other, the aggregates formed are weak and can be easily broken by low amount of shear. Thickness of the adsorbed polymer layer relative to the particle size is the critical factor that decides the degree of steric stabilization. Since attraction potential energy VA is proportional to particle size, larger particles will require thicker stabilizing layers to confer the same degree of stability. Combination of ionic and nonionic surfactants is also employed to achieve steric stabilization. In the work by Palla et al. [27] anionic and nonionic surfactants were used to stabilize alumina particles. The anionic surfactant gets adsorbed on positively charged alumina particles and nonionic surfactant penetrates into anionic surfactant layers owing to hydrocarbon chain interactions. The surfactant adsorption has three important steps:

PAGE 27

27 i. Driving force of the adsorption of ionic surfactants depends on length and concentration of hydrocarbon chains. ii. Hydrocarbon chains then interact with anionic and nonionic surfactants. iii. Extent of polymer solvent interaction should be greater than polymer polymer interaction to help prevent coiled polymers and resulting in steric stabilization when nonionic surfactants are adsorbed onto particle surfaces. Agglomerati on of Particles in Chemical Mechanical Polishing (CMP) Slurries CMP slurry stability plays a critical part while deciding performance of CMP process. During slurry storage or delivery, agglomeration of particles can take place forming larger size particles, which results in defectivity during CMP. CMP slurry consists of abrasive particles, chemicals like oxidizers, buffering agents, dispersants, complexing agents and a synergistic combination of all these is required to form stable CMP slurry. The agglomer ation of CMP slurry could result as a consequence of changes in abrasive particles, slurry chemicals, pH and the external forces like shear induced by pumps employed for slurry delivery. There are two ways in which agglomeration of particles in CMP slurry could take place: i. Changes in slurry chemistry due to changes interparticle interactions due to addition of salts and surfactants ii. External forces acting on CMP slurry during slurry handling Defectivity Due to Change in Particle Size In CMP process, the oxi des and other chemicals soften the top surface and the abrasive particles provide the mechanical force for removal of these soft layers. Cook et al. [11] proposed indentation model for polishing of glass. According to the model, material removal or indentation depth is directly proport ional to particle size, down pressure and inversely to material properties like elastic modulus and hardness. The following equation gives indentation depth Rs as

PAGE 28

28 [2 10] where a is particle size, P down pressure, K surface particle fill factor and E is youngs modulus. So, as particle size increases, it will lead to formation of large Rs and thereby causing increase in defectivity in fragile materials, especially low k substrates. The work by Mahajan and Basim et al. [19] confirm this fact and they observed that the particles greater than 0.5 m resulted in surfac e defectivity. There is a shift in CMP mechanism from contact based to indentation based as particle size increases, which results in defectivity and hence deteriorates the CMP polish performances and material removal rate. Generally, during copper (low k) CMP there are four different types of particleinduced defectivity like killer scratch, triangle scratch, shallow scratch and embedded particle [28] The killer scratch effects copper interconnects by either damaging or cutting them, whereas the triangle scratch results in electrical short circuit by forming metal residual on a subsequent metal layer and scratches could be as deep as 2000 Ao. Shallow trenches and embedded particles influence the dielectric reliability. As we move to 22, 32 and 45 nm node, low k material are commonly used in the interlayer dielectrics. These low k materials show decreasing modulus and hardness as dielectric constant is reduced. Thus, device integration during CMP of copper/low k dielectric could be challenging, as these materials with poor mechanical properties can easily form surface defects. The tolerance for defects and particle densities on wafer is continuously reduced to ensure device reliability. International Technology Roadmap for Semiconductors (ITRS) has forecasted requirement of reduction of these surface

PAGE 29

29 defects for near term technology nodes, given in Tabl e 2 2 [29] Thus, it is critical to have robust CMP slurries to reduced defectivity during CMP of metals and dielectrics. Effect of Stress Induced Particle Agglomeration on Defectivity In the semiconductor industry, CMP process involves use of pumps for slurry blending and slurry delivery. The work by Feng Chi et al. [4] demonstrated that type of pump used, has a significant influence on oversize particle (>0.5 micron) distribution of CMP slurry. Different pumps produce different kinds of stress levels and this was shown to be the reason for difference in oversize distribution of slurries pumped using various pumps. In the absence of shear, the particles in the s lurry are prevented from undergoing agglomeration by electrostatic interactions that generate repulsive forces between these particles. However, the shear flow causes particle to approach each other and when they are close enough the attractive Van Der Waals force exceeds the inter particle repulsive force and the particle undergoes agglomeration. With increase in shear rate, more particles are able to come closer to each other and that results in increased particle agglomeration (Figure 2 2, 2 3) [12] The effect of different pumps on the cumulative concentration of oversize particles, where cumulative concentration is the sum of the number of particles in 0.5 10 m, is shown in Figure 24, 2 5 [30 ] They also showed that the degree of agglomeration and roughness/defect density are positively correlated as shown in Figure 26 [30 ] Further, they observed that for low k wafers, polishing effects were more severe when these circulated slurries were used. As compared to BD1 wafers, ultra low k wafers exhibit poor mechanical properties (el astic modulus and hardness). Among the different pumps that were used, magnetically levitated (maglev) centrifugal pump caused less shear induced defectivity

PAGE 30

30 (e.g. scratches and embedded particles) during CMP of low k dielectrics. Maglev pump was qualitati vely analyzed to be a low shear device. Effect of Ionic Strength in PumpInduced Agglomeration in Slurries Chemical Mechanical Polishing (CMP) slurry is an important consumable in semiconductor industry. The highly stable CMP slurry is critical for CMP per formance and minimizes process dependent defectivity. Stable suspensions in CMP slurry are obtained as a result of repulsive forces between these particles. These repulsive forces result from overlapping of similar surface charges by dissociation of the metal oxide groups (M+OH-). Feng Chi et al. [30] demonstrated that if salts are added to these slurries, the repulsive for ces get diminished between abrasive particles, which results in agglomeration of particles. The introduction of salt increases the polishing rate due to screening the surface charge between abrasive particles and substrate, which leads to increase in frict ion force during polishing. Basim et al. [31] studied effect of salt addition on silica slurry and considered the agglomerates that were formed to be soft but still lead to poor surface quality after CMP. The cumulative concentration and normalized mean value of large size agglomerate concentration was observed to increase by a significant amount. When the slurries with salt were circulated using these pumps, the effect of shear rate on degree of agglomeration was more severe in case of slurry with salt then the one without it. The inter particle and zeta potential measurements confirmed that the repulsive forces in case of salt added slurry was relatively less than the slurry with no salt (Figure 2 7 2 11) [30] This explains the fact that for the same pump, degree of agglomeration will be more in case of slurry with salt then the slurry without salt, as due to screening, repulsive forces are reduced and so same amount of shear rate causes more particle to

PAGE 31

31 approach close enough and then Van Der Waals attraction overcomes repulsive forces and agglomeration takes place. Effect of Sur factants in Slurry on Stress Induced Agglomeration Electrostatic stabilization is one of the methods to achieve stability in colloidal suspensions. Polymer stabilization is another method used to achieve stable colloidal suspension by employing surfactants either ionic or nonionic. Feng Chi et al. [30] studied the effect of surfactant addition to unformulated ceria and sil ica slurry. For silica slurry, both cationic (CTAB) and anionic surfactants (SDS) at pH 9 were used. The cationic surfactant neutralizes the negatively charged silica surface causing screening effect that reduces the magnitude of repulsive forces between s ilica particles. As a result, agglomeration was seen during slurry formulation and handling process. The shear rate had a profound effect on agglomeration and degree of agglomeration was severe in this case as compared to silica slurry without any cationic surfactant. Whereas, when anionic surfactants (SDS) was added to silica slurry, stronger electrostatic interaction between silica particles was observed which reduced the degree of agglomeration due to shear stress during slurry handling. The shear induce d particle agglomeration is determined from the normalized tail distribution of oversize particles and the curves for cationic, anionic and as received silica slurry are as shown in Figure 2 12 [30] The variation in magnitude of surface charges due to surfactants in silica slurry was confirmed by zeta potential measurements as shown in Figure 2 13 [30] Effect of addition of surfactants on stability of ceria slurry was also studied using cationic surfactant like tetramethyl ammonium bromide, dicyclotrimethyl ammonium bromide (C10TAB), anionic surfactants like sodium dodecyl sulfate and polymer

PAGE 32

32 dispersant like polyacrylic acid separately. For ceria slurry, addition of ionic surfactant shifts the isoelectric point (IEP) of ceria to more acidic pH as a result of more negative charges adsorbed on surface. On other hand, addition of anionic surfactant to ceria shifts IEP to more alkaline pH due to the positive charges of hydrophilic heads. These effects were confirmed by zeta potential measurement s given in Figure 214 2 16 [30] The effect of shear during slurry handling on particle agglomeration in ceria slurries was significantly reduced in formulated ceria slurries as adsorption of cationic surfactant provides increased repulsive forces between ceria particles as compared to unformulated one. Thus, addition of cationic surfactant to ceria slurry could increase i ts stability and in contrast could deteriorate the stability of silica slurries resulting in formation of more agglomerates due to shear during slurry handling for the latter and less agglomerates for the former slurry. CMP performance deteriorated when un formulated ceria and silica slurries containing cationic surfactants were used, as evident from increased roughness values ( Figure 2 17 ) [30] whereas for CMP process performed using unformulated silica and formulated ceria slurry, there was significant decrease in roughness values after CMP. Agglomeration Index for Quantifying Stability of Slurries As we move beyond 45 nm to 32 and 22 nm nodes, it is critical to have controllability over agglomeration of abrasive particles to minimize defectivity during chemical mechanical polishing (CMP). A method for quantifying slurry stability is needed to prevent particleinduced defectivity. For slurry stability analysis, inter particle force and/or zeta potential measurements are done. However, these measurements do not fully describe agglomeration process in CMP as a function of different surfactants or other chemicals / addit ives. Besides, various light scattering techniques that measure

PAGE 33

33 particle size distribution are insufficient for determining the degree of agglomeration in CMP slurries as they are unable to detect changes in slurry characteristics. Feng Chi et al. [32] developed a novel method for determining the degree of agglomeration in CMP slurry due to shear induced during slurry handling. They introduced the term agglomeration index (AI), defined as a logarithm of the ratio of external shear rate (G) to the stability ratio (W) i.e. log (G/W). A slurry with lower AI at a given shear rate will be a more stable colloidal suspension, as lower AI implies higher stability ratio W. Now, W is defined as the ratio of rapid coagulation rate to slow coagulation rate where slow coagulation rate is the rate of growth of aggregates in the presence of electrostatic interactions. Higher W value means lower value of slow coagulation rate and hence more stable slurry due to stronger electrostatic interactions. Determination of Agglomeration Index (AI) Feng Chi et al. [3 0 ] determined AI of CMP slu rries in three steps: i. Subject CMP slurries to high shear pumping device like positive displacement pumps ii. Measure oversize particle tail distribution using SPOS Accusizer 780 system iii. Model changes in tail distribution using Smoluchowskis theory; Figure 218 2 20 [3 0 ] shows how AI is determined for CMP slurries. They calculated AI of various types of slurries with different surfactants, particle sizes, different kinds of abrasive particles, salts and pH values using positive displacement pumps and are detailed in Table 23 [30] Acco rding to Smoluchowski slow coagulation theory [12, 14] the total change in the rate of agglomerate concentrations such as singlet (dN1/dt), doublets (dN2/dt), triplets (dN3/dt) and so on is expressed as

PAGE 34

34 [2 10] [2 11] [2 12] Where aggregation constant kij is given as kij = 4/3 G (ai + aj)3 Individual agglomerate concentration can be derived by integrating above equations and is given as [2 13] [2 14] where N0 is total initial particle concentration, t is aggregation time and W is stability ratio. To understand the relation between AI and CMP performance, low k copper wafers were polished using slurries with different AI values. After detail analysis it was found that slurries with AI value greater than 1.8 would form large number of agglomerate particles during slurry handling thereby causing defectivity [32] Thus, AI was used to predict degree of agglomeration in various CMP slurries and lower AI values (< 1.8) were recommended for slurries, in order to achieve surface with minimum defectivity during CMP. It is important to note that these A I values were determined using positive displacement pump and AI will decrease if a low shear rate pump like

PAGE 35

35 magnetic levitated centrifugal pump is used for determining AI, as it is directly proportional to logarithmic value of shear rate G .

PAGE 36

36 Table 21. Important principles for designing CMP slurries [8, 15 21] Global Planarization: Formation of a thin surface layer which is passivated Reduction of chemical etching Reduction of mechanical polishing Removal (Polishing Rate): Rapid formation of a thin surface layer Control over mechanical/interfacial properties of thin surface layer Stress induction by abrasive to remove the surface layer Indentation based wear Fracture/delamination based removal Slurry Handling: Formulation of stable slurries Control of inter particle and particle surface interactions Steric force based repulsion in ionic systems Surface Defectivity: Rapid formation of a thin surface layer Reduction of mechanical polishing Control of particle size distribution Control of particle size and hardness Selectivity: Top layer chemo mechanical polishing Bottom layer mechanical polishing Minimization of mechanical component in slurry Table 22. Interconnect surface requirements at near term technology nodes [29] Technology node 59 45 36 28 Number of metal levels 12 12 12 13 Average dielectric constant 2.7 3.1 2.4 2.8 2.2 2.6 2.2 2.6 Dishing planarity (nm) 20 16 14 11 Minimum defect particle size (nm) 28.5 22.5 17.5 14

PAGE 37

37 Table 23. Agglomeration index of various chemical mechanical polishing (CMP) slurries Table adapted from reference [30] Type of S lurry Particle Size (nm) Agglomeration Index (AI) Silica: pH 11 30 1.36 Silica: pH 11 + salt 30 1.96 S ilica: pH 7 30 3.92 S ilica: pH 7 + salt 30 4.26 S ilica: pH 2 30 4.36 S ilica: pH 2 + salt 30 4.96 Silica: pH 9 80 1.76 Silica: pH 9 + anionic 80 0.66 S ilica: pH 3 80 2.46 Silica: pH 10 150 4.48 Ceria: pH 5 + cationic 1 70 2.36 Ceria: pH 5 + cationic 2 70 1.96

PAGE 38

38 Figure 21. Interaction of electrical double layer [22, 23] Figure 22. Steric stabilization provided by terminally adsorbed polymer chains [12]

PAGE 39

39 Figure 23. Schematic showing (a) bridging flocculation and (b) restabilization by adsorbed polymer chains [12]

PAGE 40

40 Figure 24. Cumulative concentration vs particle size for various turnovers for (a) bellows, (b) diaphragm and (c) magnetically levitated (maglev) centrifugal pump. (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 41

41 Figure 25 Normalized oversize particle distribution for bellows, diaphragm and magnetically levitated centrif ugal pumps (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 42

42 Figure 26 Optical images of: (a) BD1 and (c) ultra low k wafers polished by maglev pump circulated slurries, and (b) BD1 and (d) ultra low k wafers polished by positive displacement pump circulated slurries (Reprinted with permission from FengChi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 43

43 Fig ure 27. Circulated silica slurries for 1000 turnovers at pH 2, 7 and 11: (a) cumulative concentration without salt, (b) cumulative concentration with 0.1M KCl, and (c) mean value of normalized oversize particles in 0.50 to 1 m (Reprinted with permission from FengChi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 44

44 Figure 28 Force vs. tip substrate distance between a silica substrate and silica probe in supernatant slurries at pH 2, 7, and 11, w ith and without 0.1M KCl (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of F lorida, 2008, pp. 1123 [30] ) Figure 29 Zeta potential of 30 nm silica slurries for different pH, with and without 0.1M KCl (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1 123 [30] )

PAGE 45

45 Figure 210. Correlation between repulsive inter particle forces and mean values of normalized ov ersize particles for different pH, with and without 0.1M KCl (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 46

46 Figure 211. Surface roughness for different pH: (a) RMS and (b) Rmax, and AFM images of low k wafers polished by circulated silica slurries at (c) pH 2, (d) pH 2 with 0.1M KCl, (e) pH 7, (f) pH 7 with 0.1M KCl, (g) pH 11, and (h) pH 11 with 0.1M KCl (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 47

47 Figure 212. Comparison of particle agglomeration in unformulated and formulated silica slurries (Reprinted with permissio n from FengChi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] ) Figure 21 3. Plot of zeta potential vs. pH in silica slurry with surfactants (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 48

48 Figure 214. Plot of zeta potential vs. pH: potential curves of formulated ceria slurries with varying surfactants (Reprinted with permission from Feng Chi Chang, "Ex ternally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] ) Figure 215. Comparison of shear induced particle agglomeration in unformulated and formulated ceria slurries (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 49

49 Figure 216. Plot of zeta potential vs. pH in ceria slurry with cationic surfactant (Reprinted with permission from Feng Chi Chang, "Externally Induced Aggl omeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 50

50 Figure 217. CMP of silica: (a) surface roughness (RMS), (b) Rmax, (c) AFM images of silica wafers polished by unformulated ceria slurry at pH 6, (d) formulated ceria slurry (cationic surfactant) at pH 6 (1000 turnovers), and (d) unformulated ceria slurry at pH 6 (1000 turnovers) (Reprinted with permission from FengChi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," i n Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 51

51 Figu re 2 18. Schematics for determination of agglomeration index (AI): (a) modeling change in tail distribution of oversize particles; and (b) fitting experimental and modeled tail distributions to obtain AI (Reprinted with permission from Feng Chi Chang, "Ex ternally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 52

52 Figure 219. Tail distributions of as received slurry and circulated slurries by positive displacement and magnetically levitated centrifugal pumps (Reprint ed with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 53

53 Figure 220. Effect of external shear stress on agglomeration index (AI): experimental and modeled tail distributions in positive displacement and maglev centrifugal pumps (Reprinted with permission from Feng Chi Chang, "Externally Induced Agglomeration During Chemic al Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123 [30] )

PAGE 54

54 CHAPTER 3 MEASUREMENTS FOR CHE MICAL MECHANICAL POL ISHING SLURRIES Introduction Chemical mechanical polishing (CMP) slurry is one of the most critical consumable in semiconductor industry. CMP slurry is an optim um combination of abrasives like silica, alumina or ceria particles, oxidizers, additives, corrosion inhibitors and other chemicals and their synergistic effects results in optimal removal rates and surface smoothness. The particle size distribution in slurry decides CMP performance. Particles could undergo agglomeration due to shear forces induced during slurry formulation and handling. There are measurement techniques to detect changes in particle size distribution as listed in Table 31 [33] Each of the measurement techniques is based on different principles like light scattering, refraction, microscopy, sedimentation, sieving, and particle counter. These methods have their limitations in terms of determining particles in certain size range from few nanometers to thousand microns. Oversize Particle Distribution Measurements Accusizer 7880 manufactured by PSS Nicomp is a single particle optical sensor system (SPOS) that measures particle size distribution in the range 0.5200 m [34] SPOS detects individual particles in a certain size range as each particle passes through a very thin optical sensing zone. The concentrated suspensions are diluted sufficiently so that particle passes one at a time through de tection zone. Two physical methods are traditionally used to implement SPOS technique namely Light Extinction (LE) and Light Scattering (LS).

PAGE 55

55 Light Extinction (LE) Method [35] When individual particle suspended in a fluid passes through the light beam in a flow channel, there is a decrease in intensity of light that transmits across the flow channel. Measurement of the resulting decrease in intensity of light forms the basis of the LE method. The optical sensing zone or view volume of an LE sensor consists of the walls of the channel and the thickness of the flat ribbon of light that defines the illuminating beam (Figure 31) [35] Light refraction is the physical effect that underlies the LE method. Refractive index of most of the particles is different f rom the fluid they are surrounded by and in most cases it is water. The particles behave like lenses, either causing the light rays that are incident upon them to be refracted or deflected away from the detector on the opposite side of the flow channel. As a result, the particle causes removal of a small fraction of the incident light rays that otherwise would have illuminated the detector. For LE method the useful particle range is 1.3 to 400 m and below 1.3 m, due to prevailing noise, the signal due to particle cannot be reliably detected. Below 1.3 m, scattering is the physical effect that is mainly responsible for the decrease in light intensity reaching the forward detector. In any case, the reduction of the incident light signal at the detector is t he net result and hence the name light extinction. A plot of a typical simplified time dependent signal output VLE of the detector in an LE type sensor is shown in Figure 32 [35] The detected light level is at its maximum level, given by V0 and when VLE is reduced due to particles passing through view volume, it results in a negative going pulse superimposed on the d.c. backgr ound level. The pulse height, VLE being dependent on the size of the particle, increases

PAGE 56

56 monotonically with the particle size. VLE becomes larger as particle size becomes larger (Figure 3 2). The pulse width depends on view volume thickness w, velocit y v of the fluid/particle mixture passing through the sensor (Figure 33) [35] t = w/v = abw/F Now, v depends on fluid flow r ate F, a and b are dimensions of the flow channel. Typically, t = 14 sec, F = 1 ml/sec (60 ml/min) The area of illumination of the LE detector A0 is equal to a times w when there is no particle in view volume as shown in Figure 33. As particle enters view volume, it reduces area A effectively equal to cross sectional area of the particle. For an ideal case, for a spherical diameter d, A equals area of circle of sam e diameter given by d2/4. For particles larger than 3 5 m and smaller than beam thickness w, light refractions dominates LE. LE type sensor has a quadratic response i.e. signal depends on the square of the particle diameter. Below 1.3 m, where light scattering dominates sensor response, the pulse height decreases faster with decreasing diameter (approximately as 4th power of d ). Light Scattering (LS) Method [35] L ight S cattering (LS) method is comp lementary to LE method. When particles less than 1.3 m pass through view volume, they cause scattering of light that results in increase in intensity of light incident on LS detector As in LE method, view volume acts as a source of the scattered light, however, in absence of particle there is little or no light detected from the zone in the absence of a particle. The detector that is utilized for the LS method captures the light that is scattered over a particular, optimized range of solid angles. A plot of a typical time dependent signal output VLS of the detector in an

PAGE 57

57 LS type sensor is shown in Figure 34 [35] Initially with no particle in flow channel output is ideally zero but when a particle passes through the sensing zone, a positive going pulse appears at the output of the de tector. The pulse height ( VLS) depends on size and comp osition of the particle with assumption that solid angle of scattering detections is appropriately chosen, VLS increases monotonically with particle diameter and pulse height becomes larger as particle size becomes larger. Sensitivity is the major advantage of LS method over LE method. Accusizer 780 by PSS Nicomp uses a new patented hybrid design developed by PSS, which combines the advantages of the LE like large particle size range and relative insensitivity to particle comp osition and LS method like high sensitivity and lower diameter limit. Electronically LE and LS signal responses are combined resulting in a single summation signal pulse. Accusizer 780 could be used for particle size analysis of highly concentrated sampl e as well. The dilution system dilutes the sample in two stages till acceptable particle concentration is obtained for SPOS sensor. In the autodilution system, the total extent of initial sample suspension is equal to the product of dilution factors, wher e DF1 and DF2 are dilution factors for first and second stage respectively, as given by [3 1] w here Ci, C1, C2 are initial, first and second stage concentrations respectively. In the present work, 1ml of sample was introduced in the solution chamber and diluted with deionized water. Particles in the slurry are diluted through a twostage dilution system and then flow through sensor zone, a laser diode, which causes a drop or increase in voltage depending upon physical method, extinction or scattering

PAGE 58

58 respectively. Accusizer gives number of particles of different sizes present in the sample and their number are be added up to give cumulative concentration of oversize particles. The cumulative concentration tail has been qualitatively used in the past for determining effect of shear induced during slurry handling on agglomeration behavior of particles. Rheometer Rheometers have been traditionally used to measure the response of a liquid, suspension or slur ry when subjected to applied forces. It is mainly used for fluids whose viscosity values are not constant but rather changes depending on the amount of applied forces e.g. Newtonian and NonNewtonian fluids. Physica UDS 200 can be used as rotational, shear stress, creep and oscillatory rheometer [36] It uses concentric cylinders, cones and plates as measuring systems. The sample is contained in the measuring gap bet ween fixed plate and the rotating cone. The angle between the surface of the cone and the plate is of the order of one degree, thereby making it a shallow cone. The measuring drive is based on a highly dynamical motor driven system without gears and without mechanical force transducer, with torque measurement and optical actuator. The measuring drive is suited to run rotational, strain, and relaxation tests with simultaneous control of the shear rate i.e. preselected number of revolutions and measurement o f the torque (shear stress) acting on the measuring bob. Physics UDS 200 allows user to run controlled stress tests to determine the flow behavior of highly viscous substances, along with measuring the viscosity of fluids. By using Physica UDS 200 rheometer, we are able to apply quantified shear rates to the CMP slurry, which is introduced on the plate. The volume of sample is 3 ml that is subjected to quantified shear rates for by using cone of diameter 75 mm, air gap of 0.05 mm.

PAGE 59

59 Using values of shear rate and viscosity, this rheometer also gives values of shear stress. Zeta Potential Measurements Stabilization of colloidal suspension through electrostatic or polymeric stabilization methods are possible due to the presence of electrical double layer closed to the charged particle surface. However, so far there is no experimental measurement technique that allows measurement of the surface potential. The potential of the shear plane, called as zeta potential, could be measured for an aqueous system by electr o kinetic experiments. Brookhaven Zetaplus is used to measure zeta potential of abrasive particles in CMP slurries. Smoluchowski equation can be applied to measure zeta potential ( ) when thickness of the electric double layer given by Debye length (1/ ) [3 2] where and are viscosity and dielectric constants in the solution and E0 is electric field. VE is the electrophoretic velocity [13, 22] of colloidal particles which could be measured by electrophoresis method, employed by Brookhaven Zetaplus. For the case where double layer thickness 1/ is much greater than particle size, Huckel equation (eqn. 33) could be used to find zeta potential. [3 3]

PAGE 60

60 Table 31. Various par ticle size analysis measurements [33] Method Medium Size Range (micron) Microscopy Optical Liquid/Gas 0.5 400 Electron Vacuum 0.0001 400 Sieving Air 37 8000 Liquid May 00 Sedimentation Gravity Liquid 0.5 100 Centrifuge Liquid 0.01 300 Ultra centrifuge Liquid 0.001 300 Single particle counters Electrical sensing Zone Liquid 0.3 1200 Time of light Gas 0.2 700 Light scattering Diffraction (LALLS) Liquid/Gas (0.1) 0.5 1800 Dynamic (PCS) Liquid 0.002 0.5(1) Gas adsorption Gas/Vacuum 0.005 5

PAGE 61

61 Figure 31 Schematic s for optical scheme of a light extinction (LE) type sensor [35] Figure 32. Detector signal output in LE type sensor, resulting from momentary passage of particle through view volume (in Figure 31) [35]

PAGE 62

62 Figure 33 Area of illumination at LE detector, decreased by passage of particle of area A through view volume [35] Figure 34. Detector signal output in light scattering (LS) type sensor, resulting from passage of particle through view volume [35]

PAGE 63

63 CHAPTER 4 STUDY OF AGGLOMERATI ON BEHAVIOR OF CHEMI CAL MECHANICAL POLISHING SLURRY BY RHEOLOGICAL MEASUREM ENTS UNDER VARIED SHEAR ENVIRONMENT Introduction Chemical mechanical planarization (CMP) is commonly used for both local and global planarization of metals and dielectrics in the semiconductor industry. Due to its unique capability of eliminating topographic variations and simultaneously achieve wafer level global planarization, it has become the most sought after method for planarization of wafers. Highly stable slurry is critical for CMP performance. During polishing process, one of the main reasons for defectivity is the presence of large agglomerates in the slurry. With increased exacting requirements expected for future generation of devices, it is very important to develop highly stable performance slurries for CMP [37] It is critical to reduce and control formation of agglomerates during slurry handling. Feng Chi et al. [4] showed that high shear inducing pumps used for slurry delivery cause a significant increase in concentration of agglomerates. The effect of shear rate is aggravated when salts and surfactants are added to the slurry, which chang es the slurry chemistry, resulting in increased rate of formation of agglomerates in the slurry [30] So far all the stu dies on understanding agglomeration behavior of CMP slurry have been qualitative or at best semi quantitative in nature. A detailed understanding of agglomeration behavior of CMP slurries at quantified shear rates is critical to complete understanding of how shear rates affect agglomeration behavior of CMP slurries. To control agglomeration in slurry it is essential to know the shear rate range in which agglomeration of particles takes place and effect of duration of shear rates on these particles.

PAGE 64

64 Rheometer is a device that is essentially used to study change in behavior of fluids as a result of applied forces like change in viscosity of Newtonian and NonNewtonian fluids as a function of shear rates. Moinpour et al. [38] showed that when CMP slurries were subjected to different shear rates for a prolonged period of times the viscosity changes and drops with time. They believed that these drop in viscosity was due to agglomeration of particles, as larger particles make the fluids less viscous. However, change in normalized particle size distribution, which shows whether it is net agglomeration or deagglomeration taking place as function of shear rates is the key to understand agglomeration behavior of CMP slurries. Physica UDS 200 rheometer uses concentric cylinders, cones and plates as measuring systems [36] The measurement based on highly dynamical motor driven system, allows control of the shear rates acting on the measuring bob, which allows running controlled shear stress tests to determine the flow behavior of viscous substances. This rheometer allows user to apply shear rates up to 5000 s1 and simultaneously maintain desired temperature control. Experimental EM3530K is a silica slurry with 30wt%, 35 nm silica particles loading at pH~10. 10 wt% silica slurry, termed as received slurry, was prepared using EM3530K slurry, keeping the pH~10. The initial oversize particle distribution of this slurry was measured using Accusizer 780, a single particle optical sensor (SPOS) sys tem. Accusizer has two stage dilution system that allows one to use samples with any concentration, as the dilution system is capable of diluting slurry sample to acceptable concentration to allow its photo zone sensor to detect single particles in 0.51 200 m. 3 ml of 10wt% silica slurry was subjected to different shear rates 100 3000 s1 for various times 100 2500

PAGE 65

65 s (Table 41) using rheometer UDS 200. After the slurry is subjected to shear rate, the oversize particle distribution is again measured using Accusizer 780 to observe the resulting change in oversize particle distribution. The primary particle size measur ements by Feng Chi et al. [30] have shown that shear stress does not cause a significant change in primary particle sizes of the abrasives in CMP slurry. Results and Discussions With increase use of chemical mechanical polishing (CMP) in multilevel interconnect fabrication; it is critical to have highly stable slurries for acceptable CMP performance. To understand the effec t of shear on agglomeration behavior of silica slurries, rheometer is used to subject slurries to different shear rates in 100 3000 s1 (Table 41) and the change in oversize particle distribution is measured using Accusizer 780. Normalized concentration is then calculated from these particle size distribution measurements. The normalized concentration is defined as the ratio of cumulative concentration (Cs,G t) of the slurry that is subjected to shear rate G for time t seconds to cumulative concentration (Car) of as received slurry, which is in this case is 10wt% 35 nm silica slurry at pH~10. The slurry that is subjected to shear rate G for any time t seconds is termed as stressed slurry. Normalized concentration = Cumulative concentration of str essed slurry (Cs,G t) Cumulative concentration of as received slurry (Car) [4 1] The normalized concentration shows whether agglomeration or deagglomeration takes place. If the value of normalized concentration is less than 1, it indicates that the oversize particles i.e. agglomerates are breaking due to shear stress, which results in

PAGE 66

66 decrease in cumulative concentration of the stressed slurry. Alternatively, if the value of normalized concentration is greater than 1, it indicates the oversize particles i.e. agglomerates are forming due to shear stress, which results in increase in cumulative concentration of the stressed slurry. The response of slurry changes with shear rates (Figure 41 4 5). For shear rates of 100 1000 s1 for 100 s, 1000 s and 2500 s each, the normalized concentration curves show that net deagglomeration takes place, implying breaking down of oversize particles i.e. agglomerates that are present in the slurry. The fact that these agglomerates breakdown under low shear rat es implies that they are weakly held. When the slurry is subjected to these low shear rates for higher duration, net deagglomeration increases, indicating increased breakage of weakly held agglomerates. When the slurry sample was subjected to 1500 s1 (Fi gure 43), deagglomeration was observed only for the first 100 second. However, for 1000 and 2500 s, normalized concentration became greater than 1 i.e. silica slurry started showing agglomeration. Camp [39] explained that for a fixed solids concentration, same aggregation could be obtained either by a short period of high shear or a long period of low shear. This explains the agglomeration observed at 1500s1 for time 1000 s and 2500 s. The product of shear rate 1500 s1 and time (1000 and 2500 s) is just sufficient enough to cause agglomeration i n the slurry, but small enough to cause agglomeration at lower time 100 seconds. Further, for shear rates of 2000 3000 s1 the normalized concentration was greater than 1 for any duration up to 2500 s that indicates occurrence of shear induced agglomerat ion. The shear rate at which slurry shows agglomeration for any duration is

PAGE 67

67 termed as threshold shear rate. For basic silica slurry, value of threshold shear rate is 2000 s1 as we observed agglomeration for any duration from 100 s up to 2500 s. The value of normalized concentration increases as shear rate is increased from 2000 3000 s1. Also as duration is increased, net agglomeration increases in the slurries, which drives up the values of normalized concentration. This kind of behavior is consistent w ith that predicted by Smoluchowskis theory of agglomeration. The rate of formation of agglomerates is given as: [12] [4 2] and agglomeration growth rate is given as kij = 4/3 G (ai + aj)3 [4 3 ] where G is shear rate. As per equation [43], increase in shear rate will lead to increase in agglomeration rate and thus, for same duration higher shear rates would give larger concentration of oversize particles (Figure 44 4 5). Also the theory considers con tinuous buildup of agglomerates with time, which explains increase in normalized concentration for 2000 s1 3000 s1 from 100 to 2500 s. Summary There is a need to understand changes in behavior of CMP slurries when subjected to various quantified shear rates. In this work, novel method of using rheometer to understand agglomeration behavior of CMP slurries under varied quantified shear rates was demonstrated. The response of CMP slurries to different shear rates has two regimes, one in which it shows deagglomeration and other it shows agglomeration. For shear rates 100 1000 s1, net deagglomeration is observed

PAGE 68

68 due to break down of agglomerates present in the slurry. For shear rate of 1500 s1 de agglomeration was observed for duration of 100 s, howev er, agglomeration was observed for higher duration of 1000 s and 2500 s. As CMP slurry was subjected to higher shear rates of 2000 3000 s1, we observed net agglomeration effect. Increase in shear rates and times lead to increase in concentration of agglomerates. For silica slurry with pH~10, the threshold shear rate i.e. shear rate at which slurry shows agglomeration for any duration, is measured to be 2000 s1. By using this novel method it is possible to quantify shear rates for which CMP slurry will show deagglomeration or agglomeration behavior.

PAGE 69

69 Table 41. Various shear rates and times for silica slurry Shear Rate (s 1 ) Time (s) 100 100, 1000, 2500 1000 100, 1000, 2500 1500 100, 1000, 2500 2000 100, 1000, 2500 3000 100, 1000, 2500

PAGE 70

70 Figure 41. Normalized oversize particle distribution for 100 s1 and various times Figure 42. Normalized oversize particle distribution for 10 0 0 s1 and various times

PAGE 71

71 Figure 43. Normalized oversize particle distribution for 150 0 s1 and various times Figure 44. Normalized oversize particle distribution for 2 0 0 0 s1 and various times

PAGE 72

72 Figure 45. Normalized oversize particle distribution for 3 0 0 0 s1 and various times

PAGE 73

73 CHAPTER 5 SILICA SLURRY STABIL ITY UNDER DIFFERENT CHEMICAL A ND SHEAR ENVIRONMENT Introduction Chemical mechanical planarization (CMP) is most commonly used for eliminating topographic variations and achieving global planarization at wafer level in ultra largescale integration (ULSI) circuit [3, 17, 40, 41] In the l ast decade there has been significant advancements in CMP technology which includes development of more sophisticated processing tools and formulation of novel slurries to further enhance CMP process performance. [42] Slurries with different types of abrasives like silica, alumina, ceria particles and different slurry chemistries like pH, ionic concentration, surfactants are commonly used for controlling material removal rate, defectivity, corrosion and surface fini sh. In case of silica slurry, silica particles is negatively charged under basic pH conditions and with changes in slurry chemistry like reduction of pH, it is seen that there is a reduction in inter particle repulsive forces between silica particles. This leads to increase in agglomeration of abrasive particles that causes a rise in defectivity during CMP process. One of the main causes of the increase in agglomeration is due to the shear stress induced by pumps used for slurry delivery. Chang et al. [4] qualitatively showed that using higher shear rate pump leads to significant increase in agglomeration as compared to low shear rate pump. The agglomeration is aggravated as the slurry pH is reduced or ionic concentration increased for silica slurries [30] The isoelectric point (IEP) of silica is 2 and as the pH is increased there is an increase in repulsive forces as confirmed by zeta potential measurements. They also measured repulsive forces near the silica surface by AFM force measurements. The repulsive forces around silica

PAGE 74

74 surface tend to decrease as the pH is reduced or ionic concentration is increased, as a result of screening of surface charges. A detailed understanding of changes in agglomeration behavior as slurry chemistry is changed at different quantified shear rates, is essential for attaining control over slurry agglomeration due to different amount of shear stress induced by pumps. Rheometer is a device commonly used to study viscosity of fluids as a function of shear rates. In our work we employ Paar Physica UDS 200 Rheometer for applying different quantified shear rates to slurries. The oversize particle distribution of these slurries are measured using Accusizer 780, a single particle optical sensing (SPOS) system. Experimental EM3530K is a silica slurry with 30wt%, 35 nm silica particles loading at pH~10. Two types of 10wt% silic a slurry were prepared using EM3530K slurry, one with pH~7 and other with pH~2. The pH of the silica slurry was altered using 1N nitric acid. Initial oversize particle distribution of these slurries was measured using Accusizer 780, a single particle optic al sensor (SPOS) system. Accusizer has twostage dilution system that allows one to use samples with any concentration, as the dilution system is capable of diluting slurry sample to acceptable concentration to allow its photozone sensor to detect single particles in 0.51 200 m. 3 ml of 10wt% silica slurry was subjected to different shear rates 100 3000 s1 for various times 100 2500 s (Table 51) using rheometer UDS 200. After the slurry is subjected to shear rate, the oversize particle distributio n is again measured using Accusizer 780 to observe the resulting change in oversize particle distribution. The primary particle size measurements by Feng Chi et al.

PAGE 75

75 [30] have shown that shear stress does not cause a significant change in primary particle sizes of the abrasives in CMP slurries. Results and Discussion CMP slurry performance is determined to a large extent by the amount of agglomerates in the slurry. The repulsive inter particle forces in the slurry results in stabilization of slurry and if these forces are reduced as a result of changes in slurry chemistry, it leads to agglomeration. The isoelectric point (IEP ) of silica slurry is around 2 and as the slurry pH is lowered from basic to acidic and approaches IEP, the repulsive inter particle forces decreases [30] Furthermore, the agglomeration problem gets severe once the slurry is subjected to shear stress from pumping devices. In the present work, objective is to study agglomeration and deagglomeration in silica slurries when it is subjected to different shear rates. Neutral pH Neutral pH silica slurry samples are subjected to different shear rates 100 3000 s1 using rheometer UDS 200 for various times 100 2500 s (Table 51) and the change in oversize particle distribution i s measured using Accusizer 780. Normalized concentrations are then calculated from these particle size distribution measurements and that gives the response of slurry for different shear rates (Figure 51 5 5). For shear rates of 100 s1 (Figure 51) for 100 2500 s, the normalized concentration curves show that net deagglomeration takes place, implying breaking down of oversize particles i.e. agglomerates that are present in the slurry. The fact that these agglomerates breakdown under low shear rates i mplies that they are weakly held. When the slurry is subjected to these low shear rates for higher duration, net deagglomeration increases, indicating increased breakage of weakly held agglomerates.

PAGE 76

76 When the slurry sample was subjected to 1000 s1 (Figure 5 2), deagglomeration is observed only for the first 100 second. However, for 1000 s and 2500 s, normalized concentration became greater than 1 i.e. silica slurry started showing agglomeration. Camp [39] explained that for a fixed solids concentration, same aggregation could be obtained either by a short period of high shear or a long period of low shear. This explains the agglomeration observed at 1000 s1 for time 1000 s and 2500 s. The product of shear rate 1000 s1 and time (1000 and 2500 s) is just sufficient enough to cause agglomeration in the slurry, but small enough to cause agglomeration at lesser time 100 seconds. Further, for shear rates of 1500 3000 s1 the normalized concentration was greater than 1 for any duration up to 2500 s, which indicates occurrence of shear induced agglomer ation. The shear rate at which slurry shows agglomeration for any duration is termed as threshold shear rate. For neutral silica slurry, value of threshold shear rate is 1500 s1 as we observed agglomeration for any duration from 100 s up to 2500 s. The value of normalized concentration increases as shear rate is increased from 1500 s13000 s1. Also as duration is increased, net agglomeration increases in the slurries, which drives up the values of normalized concentration. Acidic pH Acidic pH silica slurries are subjected to different shear rates 100 3000 s1 using rheometer UDS 200 for various times 100 2500 s (Table 51) and the change in oversize particle distribution is measured using Accusizer 780. Normalized concentrations are then calculated from these particle size distribution measurements and that gives the response of slurry for different shear rates (Figure 56 5 10). For shear rates of 100 s1 (Figure 51) for 100 2500 s, the normalized concentration curves

PAGE 77

77 show that net deagglomer ation takes place, implying breaking down of oversize particles i.e. agglomerates that are present in the slurry. The fact that these agglomerates breakdown under low shear rates implies that they are weakly held. When the slurry is subjected to these low shear rates for higher duration, net deagglomeration increases, indicating increased breakage of weakly held agglomerates. Further, for shear rates of 1000 s1 3000 s1 the normalized concentration was greater than 1 for any duration up to 2500 s that i ndicates occurrence of shear induced agglomeration. The shear rate at which slurry shows agglomeration for any duration is termed as threshold shear rate. For acidic silica slurry, value of threshold shear rate is 1000 s1 as we observed agglomeration for any duration from 100 s up to 2500 s. The value of normalized concentration increases as shear rate is increased from 1000 s13000 s1. Also as duration is increased, net agglomeration increases in the slurries, which drives up the values of normalized c oncentration. Comparison of Normalized Concentration of Silica Slurries with Different pH The comparison of de agglomeration in three different pH (10, 7 and 2) silica slurries is shown in Figure 511 5 13 for 100 s1 and 100 2500 s. From the three cur ves, it is seen that extent of deagglomeration is the most for basic pH slurry and decreases with from basic to neutral to acidic pH slurries. The extent of deagglomeration indicates that the agglomerates present in basic silica slurry break much easily as compared to neutral and acidic slurry. In other words, the strength of agglomerates increases from basic to acidic pH slurry. The isoelectric point for silica particle is around pH~2, so the surface charge on silica particle decreases as the pH of silic a slurry is lowered from basic to acidic due to screening of surface charges. The reduction in surface charge results in reduced inter particle repulsive forces, which

PAGE 78

78 disturbs the silica slurry stability and in the process making it more susceptible to ag glomeration. The reduction of repulsive forces increases the inter particle attractive forces between silica particles and thus, increasing the strength of agglomerates. The relative increase in strength of agglomerates in acidic and neutral pH slurry expl ains the reduced deagglomeration in these slurries as compared to the basic slurry. The threshold shear rates i.e. shear rate for which agglomeration occurs for any duration increases from acidic (1000 s1) to neutral (1500 s1) to basic (2000 s1) pH sl urries, as the inter particle repulsive forces increases with increase in pH, thus requiring more shear stress to induce agglomeration. From the normalized concentration curves of the three slurries it is evident that concentration of larger particles in 15 micron range is much higher in neutral and acidic slurry as compared to basic slurry. The increase in concentration of larger particles is again due to increased attraction between silica particles, which causes severe agglomeration when subjected to hi gh shear rates. The higher concentration of oversize particles in turn increases the roughness and defect density of the surface being polished. Summary With increase in applications of CMP for planarization of various surfaces, there are stricter requirements to have controllability over the defects created during the CMP process. The formation of large agglomerates due to shear induced by pumps during sl urry delivery is a known problem and an understanding of the nature of agglomerates formed due to shear will be instrumental in finding a way to reduce the formation of these agglomerates. From the extent of deagglomeration it is evident that the agglomer ates in neutral and acidic pH silica slurries are relatively harder than the agglomerates present in the basic pH slurry. The threshold shear rate for silica slurries

PAGE 79

7 9 increases from acidic (1000 s1) to neutral (1500 s1) to basic (2000 s1) pH slurries, a s the inter particle repulsive forces increases with increase in pH. The concentration of larger particles in 1 5 micron range is much higher in neutral and acidic slurry as compared to basic slurry, which is due to increased attraction between silica particles. Also, it was seen for basic and neutral pH slurries, that application of lesser shear rate (1500 s1 for basic and 1000 s1 for neutral pH slurry) for prolonged period (1000 to 2500 s) could also result in agglomeration.

PAGE 80

80 Table 51. Various shear rates and times for silica slurry Shear Rate ( s 1 ) Time (s) 100 100, 1000, 2500 1000 100, 1000, 2500 1500 100, 1000, 2500 2000 100, 1000, 2500 3000 100, 1000, 2500

PAGE 81

81 Figure 51. Normalized concentration for 100 s1 and various times at pH~7 Figure 52. Normalized concentration for 1000 s1 and various times at pH~7

PAGE 82

82 Figure 53. Normalized concentration for 1500 s1 and various times at pH~7 Figure 54. Normalized concentration for 2000 s1 and various times at pH~7

PAGE 83

83 Figure 55. Normalized concentration for 3000 s1 and various times at pH~7 Figure 56. Normalized concentration for 100 s1 and various times at pH~2

PAGE 84

84 Figure 57. Normalized concentration for 1000 s1 and various times at pH~2 Figure 58. Normalized concentration for 1500 s1 and various times at pH~2

PAGE 85

85 Figure 59. Normalized concentration for 2000 s1 and various times at pH~2 Figure 510. Normalized concentration for 3000 s1 and various times at pH~2

PAGE 86

86 Figure 511. Comparison of deagglomeration of different pH silica slurries for 100 s1 and 100 s Figure 512. Comparison of deagglomeration of different pH silica slurries for 100 s1 and 1000 s

PAGE 87

87 Figure 5 13. Comparison of deagglomeration of different pH silica slurries for 100 s1 and 2500 s

PAGE 88

88 CHAPTER 6 CHARACTERIZATION OF THE NATURE OF AGGLOM ERATES: HARD VS. SOF T USING RHEOLOGICAL ME ASUREMENTS Introduction Since its inception, over the past two decades chemical mechanical planarization (CMP) has become a dominant technique for achieving both local and global planarization during manufacturing of integrated circuit (IC). CMP slurries are complex mixtures of oxidizers, abrasives, corrosion inhibitors, buffers and surfactants. The stability of slurry is critical for CMP performance. Changes in slurry chemistry or application of external forces can significantly affect the stability of these slurries. It has been shown that shear induced agglomeration due to pumps leads to formation of large particles (agglomerates) that results in significant defects like microscratches and other defects on CMP polished wafer surfaces [4, 4345] It has emerged t hat pumps used for slurry delivery results in increase in concentration of oversize (>0.5 m) particles with few studies focusing on the effects of various distribution systems such as different pumps and vacuum pressure dispense technologies [46 52] Silica based CMP slurries are colloidal suspensions that are s tabilized by repellent electrostatic forces existing between silica particles. However, application of strong external forces can help these particles overcome opposing electrostatic forces and once these particles are close enough, attractive Van Der Waal s forces become large enough to overcome inter particle repulsive forces, thereby leading to agglomeration of silica particles [53, 54] Feng Chi et al. showed that effect of shear stress on agglomeration in slurry gets severe with changes in internal slurry chemistry like pH, salt addition and surfactants. Isoelectric point (IEP) of silica slurry is 2.0, so an increase in pH gives increase in stabilization of silica slurry by increasing repulsive forces

PAGE 89

89 between particles. However, the nature of agglomerat es formed due to shear induced by pumps needs to be characterized in order to understand type of agglomerates formed i.e. hard or soft and the resulting defects. Even a small fraction of hard agglomerates result in deformation on the wafer surfaces, resulting in defective microprocessors [55] Basim et al. [56] showed that presence of hard agglomerates tend to create critical defects on the oxide film and also changes the polishing mechanism. However, some of the agglomerates can be removed by filtration of CMP slurries; nevertheless, there are still defects that are created due to the presence of transient i.e. soft agglomerates that are formed either during the polishing process [57] or during slurry handling. It is critical to characterize the nature of agglomerates formed due to shear induced agglomeration during slurry handling, in order to determine further course of action to reduce defectivity. Experimental EM3530K is a silica slurry with 30wt%, 35 nm silica particles loading at pH~10. Three types of 10wt% silica slurries were prepared using EM3530K slurry with pH~10, 7, 2 an d these slurries are termed as received slurries. The pH of the silica slurry was altered using 1N nitric acid. Initial oversize particle distribution of these as received slurries were measured using Accusizer 780, a single particle optical sensor (SPOS) system. 3 ml of 10wt% silica slurries were subjected to different shear rates 3000 s1 for 1000 s using rheometer UDS 200 and these slurries are termed stressed slurries. After the slurries are subjected to 3000 s1 for 1000 s, the oversize particle distri bution of stressed slurries are measured using Accusizer 780 to observe the resulting change in oversize particle distribution. The stressed slurries are then subjected to low shear rates of 100 1000 s1 for 100 s and 1000 s (Table 61) and oversize part icle distribution is

PAGE 90

90 then measured and the normalized concentration is calculated. The entire experimental process is summarized in flow chart given in Figure 61. Results and Discussion The formation of agglomerates due to shear induced by pumps is a comm on problem in CMP industry. The coarser/hard agglomerates present in the slurry during slurry preparation can be filtered out; however, transient agglomerates formed during slurry delivery resulting from shear induced by pumps could result in creation of defects during CMP. The nature of agglomerates formed due to shear induced agglomeration needs to be characterized to understand the type of defects resulting from the use of the stressed slurry. For forming agglomerates, slurry was subjected to shear rates of 3000 s1 for 1000 s, as agglomeration were observed for the slurries with pH ranging from basic to acidic. The stressed slurries are then subjected to low shear rates for 100 1000 s1 for 100 s and 1000 s to see if we observed deagglomeration behavi or as observed for as received slurries in previous chapters. The comparison of de agglomeration for same shear rate and time in as received slurry and stressed slurry gives an insight on the nature of agglomerates in stressed slurry. If the extent of dea gglomeration of stressed slurry is more than that of as received slurry, it indicates that agglomerates in stressed slurry are weaker than those in as received slurries and hence for same low shear rate and time gives increased break down of agglomerates and these agglomerates can be termed as soft agglomerates. Alternatively, if the extent of de agglomeration in stressed slurries is less than that in as received slurries, it indicates agglomerates are stronger in stressed slurry, so the extent of break dow n is reduced and the agglomerates are termed as hard agglomerates.

PAGE 91

91 Basic pH The normalized concentration of stressed slurries that are subjected to 100, 500 and 1000 s1 for 100 s and 1000 s are show in figure 62 and 63. In previous chapters it was observed that when as received basic pH slurry is subjected to 100 s1 and 1000 s1 for 100 s and 1000 s they undergo deagglomeration. Hence, we subjected stressed basic pH slurries to these low shear rates of 100 s1, 500 s1 and 1000 s1 for 100 s and 1000 s and calculated normalized concentration from their oversize distribution. The deagglomeration was observed for these stressed slurries as well. However, to characteri ze the nature of agglomerates in stressed slurries, it is critical to compare the extent of deagglomeration in stressed slurries with that in as received slurries. Figure 6 4 shows the comparison of deagglomeration in stressed slurry with that in as rece ived slurry under 100 s1 for 100 s and 1000 s. It is observed that extent of deagglomeration is more in stressed than as received slurry. This indicates that the agglomerates formed due to shear induced agglomeration in basic pH slurries are much weaker than that in as received slurries and hence can be termed as soft agglomerates. Neutral pH The normalized concentration of stressed slurries that are subjected to 100, 300 and 500 s1 for 100 s and 1000 s are shown in figure 65 and 66. In previous chapters it was observed that when as received neutral pH slurry is subjected to low shear rate for 100 s and 1000 s they undergo deagglomeration. Hence, we subjected stressed neutral pH slurries to these low shear rates of 100 s1, 300 s1 and 500 s1 for 100 s and 1000 s and calculated normalized concentration from their oversize distribution. The deagglomeration was observed for these stressed slurries as well. However, to characterize the nature of agglomerates in stressed slurries, it is critical to compare the

PAGE 92

92 extent of deagglomeration in stressed slurries with that in as received slurries. Figure 6 7 shows the comparison of deagglomeration in stressed slurry with that in as received slurry under 100 s1 for 100 s and 1000 s. It is observed that extent of de agglomeration is less in stressed than as received slurry. This indicates that the agglomerates formed due to shear induced agglomeration in neutral pH slurries are much stronger than that in as received slurries making it difficult to break them at low shear rates. Hence, these agglomerates can be termed as hard agglomerates. Acidic pH The normalized concentration of stressed slurries that are subjected to 100, 300 and 500 s1 for 100 s and 1000 s are show in figure 68 and 69. In previous chapters it was observed that when as received acidic pH slurry is subjected to low shear rate for 100 s and 1000 s they undergo deagglomeration. Hence, we subjected stressed acidic pH slurries to these low shear rates of 100 s1, 300 s1 and 500 s1 for 100 s and 1000 s and calculated normalized concentration from their oversize distribution. The deagglomeration was observed for these stressed slurries as well. However, to characterize the nature of agglomerates in stressed slurries, it is critical to compare the extent of deagglomeration in stressed slurries with that in as received slurries. Figure 6 10 shows the comparison of deagglomeration in stressed slurry with that in as received slurry under 100 s1 for 100 s and 1000 s. It is observed that extent of deagglomeration is less in stressed than as received slurry. This indicates that the agglomerates formed due to shear induced agglomeration in acidic pH slurries are much stronger than that in as received slurries making it difficult to break them at low sh ear rates. Hence, these agglomerates can be termed as hard agglomerates.

PAGE 93

93 Comparison of Nature of Agglomerates Formed in Slurries with Different pH From above discussion it is clear that shear induced agglomerates in basic pH slurries are soft and can be br oken easily by subjecting to low shear. However, shear induced agglomerates in neutral and acidic pH silica slurries are relatively harder. To characterize the strength of shear induced agglomerates we compare the deagglomeration of the stressed slurries for three different pH for 100 s1 and time 100 s, 1000 s. The normalized concentration at these shear rates and times are given in Figure 611 and 612 It is observed that extent of deagglomeration for basic pH stressed slurry is more than that for neutr al and acidic pH slurry. Also extent of deagglomeration for neutral pH slurry is more than that for acidic pH slurry. These indicates that strength of agglomerates increases from basic to acidic pH slurry and it can be explained based on increased attract ion or decrease in repulsive forces between silica particles at acidic pH. Thus, agglomerates in acidic pH stressed slurries are harder, followed by agglomerates in neutral pH stressed slurries that are hard too but relatively less hard than those present in acidic pH stressed slurries. The agglomerates in basic pH stressed slurries are the weakest of all and are termed as soft agglomerates. So in acidic and neutral pH slurries not only is the concentration of oversize particles more than that in basic pH s lurry but also the oversize particles are harder in acidic and neutral pH slurries. This explains the observed relative increase in the number of defects when acidic and neutral pH slurries are used instead of basic pH slurry. Summary With increase in appl ication of CMP for planarization of various surfaces, there are stricter requirements to have controllability over the defects created during CMP. The

PAGE 94

94 formation of large agglomerates due to shear induced by pumps during slurry delivery is a known problem and an understanding of the nature of agglomerates formed due to shear will be instrumental in finding a way to reduce the formation of these agglomerates. A method was developed to characterize the type of agglomerates formed due to shear induced agglomer ation. The method involves subjecting as received slurries to high shear rate and time i.e. a condition in which shear induced agglomeration occurs and then subjected this stressed slurry to low shear rates and time to cause deagglomeration. The comparison of percentage deagglomeration in stressed slurries with that in as received slurries gives an insight into the nature of agglomerates. It was observed that for basic pH silica slurry agglomerates formed due to shear induced agglomeration are softer. The agglomerates formed in acidic and neutral pH slurries due to shear induced agglomeration are hard agglomerates, with the ones in acidic pH being the hardest. Thus, in addition to increase in concentration of oversize particles in acidic and neutral pH slurries, presence of hard agglomerates further aggravates formation of defects during CMP when these slurries are used.

PAGE 95

95 Table 61. Various low shear rates and times for silica slurry with pH 10 after being subjected to 3000 s1 and 1000 s Shear Rate ( s 1 ) Time (s) 100 100, 1000 3 00 100, 1000 500 (for pH 2 and 7) / 1000 (for pH 10) 100, 1000

PAGE 96

96 Figure 61. Flow chart of experimental procedure Measure oversize particle distribution using Accusizer 780 and cumulative concentration is Cs,G t Calculate normalized concentration Subject stressed slurry to low shear rates 100 1000 s 1 for 100 s and 1000 s Measure oversize particle distribution using Accusizer 780 and cumulative concentration of stressed slurry is Cs As Received (As R) Slurry 10wt%, 35 nm silica slurry at different pH Stressed slurry Subject As R slurry to 3000 s1 for 1000 s

PAGE 97

97 Figure 62. Normalized oversize particle distribution for 100 s for various shear rates at pH~10 Figure 63. Normalized oversize particle distribution for 1000 s for various shear rates at pH~10

PAGE 98

98 Figure 64. Comparison of deagglomeration in as received and stressed basic pH silica slurries for 100 s1 and 1000 s Figure 65. Normalized oversize particle distribution for 100 s for various shear rates at pH~7

PAGE 99

99 Figure 66. Normalized oversize particle distribution for 1000 s for various shear rates at pH~7 Figure 67. Comparison of deagglomeration in as received and stressed neutral pH silica slurries for 100 s1 and 1000 s

PAGE 100

100 Figure 68. Normalized oversize particle distribution for 100 s for various shear rates at pH~2 Figure 69. Normalized oversize particle distribution for 1000 s for various shear rates at pH~2

PAGE 101

101 Figure 610. Comparison of deagglomeration in as received and stressed acidic pH silica slurries for 100 s1 and 1000 s Figure 611. Comparison of deagglomeration in different pH silica slurries for 100 s1 and 100 s

PAGE 102

102 Figure 6 12. Comparison of deagglomeration in different pH silica slurries for 100 s1 and 1000 s

PAGE 103

103 CHAPTER 7 MEASUREMENT OF STABI LITY RATIO: A PARAME TER FOR DEFINING SLURRY STABILITY INDEPENDEN T OF SHEAR RATES Introduction Chemical mechanical planarizat ion (CMP) process has lead to a growth of multi million dollar industry of CMP consumable especially slurry and nano particles used for CMP slurry formulation. CMP consumables have attracted lot of attraction of researchers in the semiconductor industry especially slurry formulation. Large amount of work in being carried out to formulate various kinds of specialties slurries for different substrates. Initial application of CMP was found in the fabrication of circuit interconnect wiring, so called back end processes [8, 58] Also, the synergistic application of dry etchi ng and CMP was used for fabrication of advanced tungsten plug in addition to copper damascene interconnect structures [59, 60] With expansion of copper, gallium nitride and other low k advance applications, it is seen that each of these technologies requires multiple planarization steps and there is a need for creati ng a well controlled CMP process. Low k dielectric films have much lower mechanical properties, which poses new challenges in developing integration schemes like achieving sufficient polishing while maintaining very low down forces during CMP process [17, 6169] Especially lot of efforts are directed towards stabilization of slurry and formulating highly stable slurries that involves minimum large size (> 0.5 m) agglomerates. These slurries have slow agglomerate growth rate in response to shear rates that these slurries are subjected to, during slurry delivery as a result of circulation using pumps. [30, 56] Feng Chi et al. [32] proposed a term agglomeration index (AI), which is defined as logarithm of the ratio of shear rate to stability ratio, where stability ratio is defined as ratio of number of collisions between particles to the nu mber of collisions leading to

PAGE 104

104 agglomeration. So the inverse of stability ratio defines collision efficiency, so higher the value of stability ratio, lower is the collision efficiency and hence lower are the number of collisions that lead to agglomeration t hereby forming large particles in the slurry. Slurry with a lower AI value will be more stable as a result of higher value of stability ratio. Higher value of stability ratio implies lower value of slow coagulation rate, which means that electrostatic interactions are strong enough to keep particles apart due to repulsive forces between them and thereby reduce the rate of formation of agglomerates. They were able to relate AI values to CMP performance in an excellent way by polishing wafers using various sl urries and then characterized those wafers by atomic force microscopy (AFM). They concluded that slurries having AI < 1.8 were stable ones and did not cause defectivity due to presence of large size agglomerates formed due to shear induced by pumps during slurry circulation. For silica slurry, 35 nm slurry at pH ~10 with AI of 1.36 was the most stable one and caused the least defectivity whereas 35 nm slurry at pH~2 with AI of 4.96 and 135 nm slurry at pH ~10 with Ai of 4.48 were the most unstable slurries and caused large number of defects. AI values were determined using a high shearing positive displacement pump. If we make use of any other pump with causes different amount of shear stress, we would obtain different AI values, as AI is a function of shear rate applied to the slurry. Now, if an unstable slurry having AI value > 1.8 (AI value obtained using same positive displacement pump that was used by Feng Chi et al.) is subjected to a low shear stress causing pump, we could obtain an AI value less than 1.8 indicating that it is a stable slurry [32] So unstable slurry could have AI value less than 1.8 if a low shear pump is used and AI value greater than 1.8 if a positive displacement pump is used. Hence,

PAGE 105

105 there is an urgent need to find a parameter that is independent of shear rate in order to define slurry stability. Generally, zeta potential, particle size measurements are used to monitor stability of slurries qualitatively. However, these measurements are unable to quantify the degree of agglomeration and/or slurry stability as and when the slurries are formulated. In a modification to the novel method developed by Feng Chi et al. [32] we subject CMP slurries to quantified shear rates using a rheometer, in contrast to a positive displacement pump used in their work. After slurries are subjected to shear by rheometer, we measure their oversize particle distribution using Accusizer 780, which is a single particle optical sensing (SPOS) system. In order to determine AI, Smoluchowski theory of agglomeration was used to model changes in oversize distribution and from AI we can easily calculate stability ratio W. Experimental Agglomeration Index (AI) Measurement: AI of CMP slurries is determined using a rheometer Physica Paar UDS 200 AI of slurry was calculated in the following three steps: i. Subject CMP slurries to various shear rates 1500 3000 s1 (Table 7 1) using rheometer UDS 200 ii. Measure oversize particle distribution using SPOS system Accusizer 780 iii. Model changes in oversize particle distribution using Smoluchowski theory of agglomeration We selected shear rate of 1500 3000 s1 and 1000 s condition as we were able to obtain agglomeration for all kinds of slurries for these shear rates and time, which is required for modeling the changes in oversize particle distribution. We used various types of slurry for our experim ental work with different pH (2, 7 and 10), ionic concentration, particle sizes all with 10weight percent particle loading. The oversi ze

PAGE 106

106 particle distribution was then measured using Accusizer 780 (SPOS) This instrument has a twostage auto dilution syst em which allows one to use sample of any concentration, as it can dilute the sample to the acceptable concentration for single particle size analysis and can give particle size distribution in 0.51 200 m. Results and Discussion Smoluchowskis theory of agglomeration was used to model growth of agglomerates in order to obtain particle size distribution [12, 70] The theory assumes all particle collisions are binary in nature, proportional to the particle concentration and there is no breaking of agglomerates taking place.Rate of growth of agglomerates singlet, doublets and triplets can be expressed as [7 1] [7 2] [7 3] where aggregation constant, kij = 4/3 G (ai+aj)3 is a function of particle size (a) aggregates, i and j folds and the shear rate ( G) From the above equations individual particle concentrations of aggregates can be obtained as follows: [7 4] [7 5]

PAGE 107

107 where N0 is the initial total particle concentration, t is the aggregation time and W is the stability ratio, which is defined as the ratio of rapid coagulation rate to slow coagulation rate in presence of electrostatic interactions between abrasive particles [13] Figure 71 shows fitting of experimental and simulated curves. Simulation of growth of aggregates is done using initial oversize particle distr ibution and then by curve fitting experimental and modeling curves we determine AI value. CMP slurry stability is determined to a large extent by the amount of agglomerates in the slurry. Agglomeration index for different slurries at various shear rates were calculated by following the steps described above and is given in Table 72, 7 3 and 7 4. Now, AI is defined as logarithm of the ratio of shear rate to stability ratio of slurry and since we subject the slurry to quantified shear rate using rheometer, w e can easily calculate value of stability ratio. AI = log (G/W) [7 6] Simplifying further and solving for W, we get W = G / 10(AI) [7 7] where G is shear rate. The values of stability ratio are calculated for the slurries at diff erent shear rates. From Figure 72 it is seen that stability ratio of a slurry is constant for a given slurry condition and independent of shear rate that is applied to the slurry. The values of stability ratio for various slurries are detailed in Table 7 5. It is seen that silica slurry of 35 nm primary particle size at pH ~10 and no salt has the highest stability ratio which makes it the most stable of all the slurries tested and silica slurry of 35 nm particle size at pH ~2 with 0.1M KCl is the most unst able form of slurry and susceptible to severe shear induced agglomeration. Also changing the primary particle size

PAGE 108

108 significantly influences slurry stability and the values of stability ratio shows that 135 nm silica slurry at pH~10 is significantly more un stable than 35 nm silica slurry at pH ~10. These conclusions regarding stability of various silica slurries is in agreement with the conclusions derived in previous work by FengChi et al., however, the stability ratio which is used to define slurry stabil ity here is independent of shear rate and hence a better way to define slurry stability. From equation [76] it is evident that AI value is a function of shear rate and one will obtain different AI values when different shearing device is used. Hence, it undermines the effort of finding AI values for measuring slurry stability if a different shearing device is used. However, stability ratio being independent of shear rate is a better way to define slurry stability. The stability ratio is also defined as ratio of the number of collisions between the particles to the number of collisions leading to agglomeration. The inverse of stability ratio gives collision efficiency, so lower is the value of stability ratio, higher is the value of collision efficiency, which implies more collisions between particles leading to agglomeration. Summary Chemical mechanical planarization (CMP) is the widely used method for surface planarization. CMP slurry is a critical component of CMP process and presence of large particl es significantly affects the amount of defects, surface finish and material removal rate. There is an urgent need for a parameter that defines slurry stability independent of shear rate. In the present work, agglomeration index (AI) was computed for various slurries at different shear rates. AI was used to calculate stability ratio for various slurries at different shear rates. However, stability ratio was found to be constant for a given slurry condition, independent of shear rate applied to the slurry. Comparison of

PAGE 109

109 the values of stability ratio shows that silica slurry of 35 nm primary particle size at pH ~10 and no salt, is the most stable and the silica slurry of 35 nm particle size at pH ~2 with salt is the most unstable form of slurry and susceptible to severe shear induced agglomeration. Also, 35 nm silica slurry at pH ~10 is significantly more stable than 135 nm silica slurry at pH ~10.

PAGE 110

110 Table 71. Different shear rates and times for various silica slurries Shear Rates ( s 1 ) Time (s) 1 5 00 1000 20 00 1000 3000 1000 Table 72. Assigned name for various slurries Slurry Type Assigned Name Silica: pH 10, 35 nm A Silica: pH 7, 35 nm B Silica: pH 2, 35 nm C Silica: pH 2 + 0.1M KCl, 35 nm D Table 73. Agglomeration index (AI) for different slurries at various shear rates (s1) Shear Rate ( s 1 ) A B C D 1500 1.97 2.30 2.61 3.07 2000 2.09 2.42 2.74 3.17 3000 2.27 2.60 2.91 3.39 Table 74. Agglomeration indexes (AI) for different slurries at shear rate of 3000 s1 Slurry Type AI Silica: pH 10 + 0.1M KCl, 35 nm 2.44 Silica: pH 7 + 0.1M KCl, 35 nm 2.81 Silica: pH 10, 135 nm 3.28 Table 75. Stability ratio for various slurries Slurry Type Stability Ratio Silica: pH 10, 35nm (A) 16.53 Silica: pH 10 + 0.1M KCl, 35nm 10.8 Silica: pH 7, 35nm (B) 7.61 Silica: pH 7 + 0.1M KCl, 35nm 4.65 Silica: pH 2, 35nm (C) 3.64 Silica: pH 2 + 0.1M KCl, 35nm (D) 1.28 Silica: pH 10, 135nm 1.5 6

PAGE 111

111 Figure 71. Schematic illustration of fitting experimental and modeled tail distributions to obtain the agglomeration index Figure 72. Stability ratio v/s different shear rates (s1)

PAGE 112

112 CHAPTER 8 CONCLUSIONS Chemical mechanical planarization (CMP) is the most commonly used method used to achieve local as well as global planarization of metals and dielectrics. With increasing miniaturization of devices there are more exacting requirements on the performance of CMP process where material removal rate and surface finish decides CMP performance to a large extent. Presence of large oversize particles (> 0.5 micron) in the slurry is one of the main reasons for resulting defects during CMP process. One of the main reasons for presence of large particles in the slurry is due to usage of pumps during slurry delivery. The shear rate of pumps is known to cause agglomeration problem in CMP slurries, with increasing shear rate leading to increase in agglomeration in the slurries. However, so far all the work on shear induced agglomeration has been qualitative or at best semi quantitative in nature. In the present work, a novel technique to study agglomeration behavior as a function of quantified shear rates is developed. Rheometer was used as a means to subject slurry to quantified shear rates for different times. The resulting change in oversize particle distribut ion was measured using Accusizer 780, a single particle optical sensing (SPOS) system. The behavior of silica slurry when subjected to different shear rates falls into two regimes, one deagglomeration and other agglomeration. There is a threshold shear rate for each slurry condition, above which agglomeration is observed and below which deagglomeration is observed. The deagglomeration increased as the slurry pH was increased and away from its isoelectric point at pH~2. Based on percentage deagglomeratio n the strength of the agglomerates present in the as received basic pH slurry is seen to be less than that for neutral and acidic pH

PAGE 113

113 slurries. The increase in attractive forces between silica particles as the pH is lowered towards isoelectric point leads t o formation of stronger agglomerates in acidic and neutral pH slurries, with agglomerates in acidic pH slurry being stronger than that in neutral pH slurries. The threshold shear rate to initiate agglomeration in basic pH silica slurry is 2000 s1, for neutral pH is 1500 s1 and for acidic pH is 1000 s1 The decrease in threshold shear rate was seen to be due to decrease in repulsive forces between the interacting silica particles as the slurry pH is reduced. The nature of agglomerates hard or soft could determine the kind of defects created during the CMP process. The large particles that are present in the slurry could be filtered out, however, the transient agglomerates are generated due to shear induced by pump, which create defects on CMP polished wafer surfaces. To understand the nature of agglomerates, the slurry is subjected to high shear rates and times such as 3000 s1 and 1000 s, corresponding to the conditions for which agglomeration was observed for all slurries. Subsequently, the stressed slurry is subjected to lower shear rates and times such as 100 1000 s1, 100 s and 1000 s, corresponding to the conditions for which de agglomeration is observed for as received slurries. Based on comparison of percentage deagglomeration in stressed and as re ceived slurries, it is qualitatively showed that all the agglomerates formed due to shear induced agglomeration in basic pH slurries are soft agglomerates, whereas as those formed in neutral and acidic pH stressed slurries are hard agglomerates, with agglomerates in acidic pH slurry being stronger than that in neutral pH slurries. In this way a unique method is developed to characterize the nature of shear induced agglomerates.

PAGE 114

114 Finally, the agglomeration index (AI) was explored further to calculate stabilit y ratio, a parameter for defining slurry stability independent of shear rate. Stability ratio showed that it is constant for a given slurry condition and does not change with shear rates. The values of stability ratios are 1.26 16.53 depending upon slurr y conditions and higher the value of stability ratio implies higher is the slurry stability. Based on comparison of stability ratios with previous work on agglomeration index and CMP polished wafers using silica slurries; it is seen that for 10wt% particl e loading, 35 nm basic pH silica slurry is the most stable and 35 nm acidic pH slurry and 135 nm basic pH slurry are the most unstable among the set of slurries that were studied. Thus, AI was used to calculate stability ratio that defines CMP slurry stabi lity under varied shear environment.

PAGE 115

115 LIST OF REFERENCES [1] S. M. Sze, VLSI Technology 2nd ed. New York: McGraw Hill, 1988. [2] R. K. Singh and R. Bajaj, "Advances in chemical mechanical planarization," MRS Bulletin vol. 27, pp. 743 751, 2002. [3] I. M. Steigerwald, Murarka, S.P., Gutmann, R.J., Chemical Mechanical Planarization of Microelectronic Materials New York: John Wiley & Sons, 1997. [4] F. C. Chang, S. Tanawade, and R. K. Singh, "Effects of Stress Induced Particle Agglo meration on Defectivity during CMP of Low k Dielectrics," Journal of the Electrochemical Society vol. 156, pp. H39H42, 2009. [5] A. E. Braun, "CMP becomes gentler, more efficient," in Semiconductor International vol. 24, Nov. 2001, pp. 5461. [6] V. Sha nnon, Smith, D.C., "Copper Interconnects for HighVolume Manufacturing," in Semiconductor International vol. 24, May 2001, pp. 9397. [7] S. Wolf, Tauber, R.N., Silicon Processing for the VLSI Era: Process Technology vol. 1, 2nd ed. Sunset Beach, CA: Lat tice Press, 1999. [8] F. B. Kaufman, Thompson, D.B., Broadie, R.E., Jaso, M.A., Guthrie, W.L., Pearson, D.J., Small, M.B., "Chemical Mechanical Polishing for Fabricating Patterned W Metal Features as Chip Interconnects," Journal of Electrochemical Society vol. 138, pp. 34603465, Nov. 1991. [9] F. Preston, "The Traction of Glass Polishing," Journal of the Society of Glass Technology vol. 12, pp. 3 6, 1928. [10] N. J. Brown, Baker, P.C., Maney, R.T., "Optical Polishing of Metals," presented at Contemporar y Methods of Optical Fabrication, SPIE The International Society for Optical Engineering Bellingham, Bellingham, WA, 1981. [11] L. M. Cook, "Chemical processes in glass polishing," Journal of NonCrystalline Solids vol. 120, pp. 152171, 1990. [12] M. E limelech, J. Gregory, X. Jia, and W. R.A, "Particle Deposition and Aggregation Measurement, Modelling and Simulation," Elsevier, 1998. [13] H. Stechemesser, Dobias, B., Coagulation and Flocculation, Second ed. Boca Raton: CRC Press, Taylor and Francis Gr oup, 2005. [14] K. O. Higashitani, Ryuji; Hosokawa, Gijiro; Matsuno, Yoshizo "Kinetic theory of shear coagulation for particles in a viscous fluid Journal of Chemical Engineering of Japan vol. 15, pp. 299304, 1982.

PAGE 116

116 [15] U. Mahajan, Bielmann, M., Singh, R.K., "Dynamic Lateral Force Measurements during Chemical Mechanical Polishing of Silica," Electrochemical and Solid State Letters vol. 2, pp. 80 82, Feb. 1999. [16] M. Bielmann, Mahajan, Y., Singh, R.K., "Effect of Particle Size during Tungsten Chemical Mechanical Polishing," Electrochemical and Solid State Letters vol. 2, pp. 401403, Aug. 1999. [17] R. K. Singh, Bajaj, R., Moinpour, M., Meuris, M., Chemical Mechanical Polishing 2000 Fundamentals and Materials Issues vol. 613, 2001. [18] M. BIelmann, Mahajan, Y., Singh, R.K., Agarwal, P., Mischler, S., Rosset, E., Landolt, D., Chemical Mechanical Polishing Fundamentals and Challenges vol. 566. Warrendale, PA., 2000. [19] G. B. Basim, J. J. Adler, U. Mahajan, R. K. Singh, and B. M. Moud gil, "Effect of Particle Size of Chemical Mechanical Polishing Slurries for Enhanced Polishing with Minimal Defects," Journal of The Electrochemical Society vol. 147, pp. 35233528, 2000. [20] S. M. Lee, Choi, W., Craciun, V., Jung, S. H., Singh, R.K., "Electrochemical Measurements to Understand the Dynamics of the Chemically Modified Surface Layer Formation During Copper CMP," presented at Materials Research Society, San Francisco, CA, April 2002. [21] S. Mischler, Debaud, S., Landolt, D., "Wear Accele rated Corrosion of Passive Metals in Tribocorrosion Systems," Journal of the Electrochemical Society vol. 145, pp. 750758, Mar. 1998. [22] R. J. Hunter, Foundations of colloid science / Robert J. Hunter Oxford ; New York : Oxford University Press, 2001. [23] M. J. Rosen, Surfactants and Interfacial Phenomena: John Wiley & Sons, Inc., 2004. [24] G. B. Basim, I. U. Vakarelski, and B. M. Moudgil, "Role of interaction forces in controlling the stability and polishing performance of CMP slurries," Journal of Colloid and Interface Science, vol. 263, pp. 506515, 2003. [25] I. D. Morrison, Ross, S., Colloidal suspensions: Suspensions, Emulsions, and Foams : John Wiley and Sons Inc., 2002. [26] D. H. Napper, Polymeric stabilization of colloidal dispersions Academi c Press, London, New York, 1983. [27] B. J. Palla and D. O. Shah, "Stabilization of High Ionic Strength Slurries Using the Synergistic Effects of a Mixed Surfactant System," Journal of Colloid and Interface Science vol. 223, pp. 102111, 2000.

PAGE 117

117 [28] T. Y. Teo, W. L. Goh, V. S. K. Lim, L. S. Leong, T. Y. Tse, and L. Chan, "Characterization of scratches generated by a multiplaten copper chemical -mechanical polishing process," Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures vol. 22, pp. 65 69, 2004. [29] "Interconnect," International Technology Roadmap for Semiconductors pp. 1 67, 2007. [30] F. C. Chang, "Externally Induced Agglomeration During Chemical Mechanical Polishing of Metals and Dielectrics," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2008, pp. 1123. [31] G. B. Basim and B. M. Moudgil, "Effect of soft agglomerates on CMP slurry performance," Journal of Colloid and Interface Science, vol. 256, pp. 137142, 2002. [32] F. C. Chang and R. K. Singh, "Method for Quantifying the Degree of Agglomeration in Highly Stable Chemical Mechanical Polishing Slurries," Electrochemical and Solid State Letters vol. 12, pp. H127H130, 2009. [33] P. Bowen, "Particle Size Distribution Measurement from Millimeters to Nanometers and from Rods to Platelets," Journal of Dispersion Science and Technology vol. 23, pp. 631 662, 2002. [34] D. F. Nicoli, K. Hasapidis, P. O'Hagan, D. C. McKenzie, J. S. Wu, Y. J. Chang, and B. E. H. Schade, "HighResolution Particle Size Analysis of Mostly Submicrometer Dispersions and Emulsions by Simultaneous Combination of Dynamic Light Scattering and SingleParticle Optical Sensing," in Particle Size Distribution III vol. 693, ACS Symposium Series : American C hemical Society, 1998, pp. 5276. [35] P. Nicomp, "Accusizer 780 APS Manual," PSS Nicomp. [36] P. Paar, "Rheometer UDS 200 Manual." [37] "Interconnect," International Technology Roadmap for Semiconductors pp. 1 178, 2002. [38] R. A. A. Sorooshian, H. K. C hoi, M. Moinpour, A. Oehler, A. Tregub "Effect of Particle Interaction on Agglomeration of Silica Based CMP Slurries," presented at Materials Research Society (MRS), San Francisco, 2004. [39] T. R. Camp, "Flocculation and flocculation basins," presented at ASCE, 1953. [40] L. ShonRoy, "CMP: Market Trends and Technology," in Solid State Technology vol. 43, 2000, pp. 67. [41] K. D. Beyer, IBM Micronews vol. 5, 1999.

PAGE 118

118 [42] T. Tucker, Plenary Talk, "Key Issues related to CMP Market Segments," in Chemical Mec hanical Polishing 2001American Vacuum Society San Jose, October 2001. [43] G. Vasilopoulos, et al., "STI CMP Defect Reduction with Slurry Filtration," presented at VLSI Multilevel Interconnect Conference (VMIC), June 2000. [44] T. Nishiguchi, "CMP Consu mable Technologies in PERL Hitachi," presented at ERC Retreat & IAB Meeting, 2001. [45] K. Nicholes, et al., "Analysis of Wafer Defects Caused by Large Particles in CMP Slurry using Light Scattering and SEM Measurement Techniques," presented at Eight Inter national Chemical Mechanical Planarization for ULSI Multilevel Interconnection Conference (CMP MIC), Marina Del Ray, CA, Feb. 2003. [46] B. Johl, Singh, R., "Optimum Process Performance Through Better CMP Slurry Management," in Solid State Technology vol 46, Aug. 2003. [47] B. Johl, et al., "Dynamic Pot Life and Handling Evaluation of EPL2362 First Step Copper Slurry," presented at 8th International CMP Conference, Marina Del Ray, CA, Feb. 2003. [48] B. Johl, Buley, T., "Dynamic Pot Life and Handling Evaluation of Rodel CUS1351 Copper Barrier Slurry," presented at 7th International CMP Conference, San Jose, 2002. [49] B. Johl, et al., "Accelerated Aging and Handling Evaluation of Rodel CUS1201 Second Step Slurry in BOC Edwards Bulk Delivery System," prese nted at VMIC Conference, 2001. [50] B. Johl, Singh, R., "Characterization of a Silica Based STI CMP Slurry in a Vacuum Pressure Dispense Slurry Delivery System and Pump Loop," presented at VMIC Conference, 2001. [51] B. Johl, Bare, J.P., "Accelerated Aging and Handling Evaluation of Rodel ILD1300 Oxide CMP Slurry," presented at AVS N. California CMPUG Annual Symposium, CA, 1997. [52] B. Johl, Bare, J.P., "Comparison of Humidified vs. Nonhumidified Vacuum Pressure CMPS Slurry Distribution System," presented at VMIC Conference, 1998. [53] B. M. Moudgil, "Stability of Colloidal Suspensions for CMP Applications," presented at Levitronix CMP Users Conference, Santa Clara, CA, Feb. 2005. [54] J. Bare, Johl, B., Lemke, T., "Comparison of Vacuum Pressure vs. Pump D ispense Engines for CMP Slurry Distribution," in Semiconductor International Jan. 1999.

PAGE 119

119 [55] J. Schlueter, presented at SemiCon Southwest, 1995. [56] G. B. Basim, "Formulation of Engineered Particulate Systems for Chemical Mechanical Polishing Applications," in Materials Science & Engineering, vol. Ph.D. Gainesville, Florida: University of Florida, 2002, pp. 1141. [57] "Material Safety Data Sheet for Semi Sperse 12 and 25 Aqueous Dispersions," Cabot Corporation Microelectronics Division, Aurora, IL 2000. [58] W. J. Patrick, W. L. Guthrie, C. L. Standley, and P. M. Schiable, "Application of Chemical Mechanical Polishing to the Fabrication of VLSI Circuit Interconnections," Journal of The Electrochemical Society vol. 138, pp. 17781784, June 1991. [59] F. W hite, W. Hill, S. Eslinger, E. Payne, W. Cote, B. Chen, and K. Johnson, "Damascene stud local interconnect in CMOS technology," presented at Electron Devices Meeting, 1992. Technical Digest., International, 1992. [60] J. Givens, Geissler, S., Cain, O., Cla rk, W., Koburger, C., Lee J., "A Low Temperature Local Interconnect Process in a 0.25 m Channel CMOS Logic Technology with Shallow Trench Isolation," presented at Proceedings of the 11th International VLSI Multilevel Interconnection Conference, 1994. [61] S. Hayashi, Poust, B., Heying, B., Goorsky, M.S., "Silicon Carbide and Related Materials 2003," Trans Tech Publications pp. 1605, 2004. [62] C. Lian, Xing, H., Wang, S., McCarthy, L., Brown, D., "DC Characteristics of AlGaAs/GaAs/GaN HBTs Formed by Direc t Wafer Fusion," IEEE Electron Device Letters vol. 28, pp. 810, 2007. [63] A. Tauzin, Akatsu, T., Rabarot, M., Dechamp, J., Zussy, M., Moriceau, H., Michaud, J.F., "Transfers of 2inch GaN films onto sapphire substrates using Smart CutTM technology," Ele ctronics Letters vol. 41, pp. 668670, 2005. [64] L. Jay, Ma, Y., Wang, Y., Bajaj, R., Redekar, F., "Process and Material Challenges for 0.1 m Copper Technology," presented at Presented at 5th International Symposium on Chemical Mechanical Polishing, Lak e Placid, NY, Aug. 2000. [65] R. Bajaj, Redekar, F., Wijekoon, K., "Unlocking the Damascene Puzzle," presented at VMIC Conference, Santa Clara, CA, Jun. 2000. [66] S. V. Babu, Cadien, K.C., Yano, H., ChemicalMechanical Polishing 2001Advances and Future Challenges vol. 671. Warrendale, PA, 2001. [67] S. Tsai, [CD ROM], presented at Proceedings IITC Conference, IEEE Electron Devices Society, Piscataway, NJ, 2002.

PAGE 120

120 [68] C. C. Chang, [CDROM] presented at Proceedings IITC Conference, IEEE Electron Devices S ociety, 2002. [69] K. C. Yu, [CD ROM], Piscataway, NJ., 2002. [70] W. B. Russel, Saville, D. A., Schowalter, W. R., Colloidal Dispersions United Kingdom: Cambridge University Press, 1989.

PAGE 121

121 BIOGRAPHICAL SKETCH Aniruddh Jagdish Khanna was born in 1983, in Mumbai. He received Bachelor of Technology (B.Tech) in Polymer Engineering and Technology from Institute of Chemical Technology (ICT) formerly called University Department of Chemical Technology (UDCT), Mumbai, Indi a, in 2005. Aniruddh pursued further studies at University of Florida and obtained Master of Science in materials science and engineering in May 2007. A niruddh continued for doctoral research in materials science and engineering under the guidance of P rof. Rajiv Singh at University of Florida. His dissertation focused on quantifying particle agglomeration during chemical mechanical polishing (CMP) of metals and dielectrics. In July 2008, Aniruddh began his studies in concurrent degree program at Warring ton College of Business Administration, University of Florida and obtained Master of Science in management in May 2010. Aniruddh was the President of Electrochemical Society at University of Florida for the year 2009. Aniruddh was selected as the honorary member of the Epsilon Lambda Chi Engineering Leadership Circle in 2010, a society for the leaders of the college of engineering at University of Florida, whose membership is restricted to only top 1% of the college of engineering students. His presentati on at 218th Electrochemical Society conference was awarded second prize among student presenters He received Doctor of P hilosophy in materials science and engineering from University of Florida in fall of 2010.