<%BANNER%>

Characterization of the Chemical Effects of Ceria Slurries for Dielectric Chemical Mechanical Polishing


PAGE 1

CHARACTERIZATION OF THE CHEMICAL EFFECTS OF CERIA SLURRIES FOR DIELECTRIC CHEMICAL MECHANICAL POLISHING By JEREMIAH TERRELL ABIADE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Jeremiah Terrell Abiade

PAGE 3

This dissertation is dedicated to my wife Kalia and daughter Asha; my parents Candy Grey and Renee McGee; my siblings, Margue rita, Paulette, Jonath an, Candon, Pilar, and Joshua. I also would like to dedicate this work to all of the elders of my family who were denied the educational experiences that I have been so fortunate to enjoy.

PAGE 4

iv ACKNOWLEDGMENTS First and foremost I would like to thank G od for giving me the ability to complete this dissertation and surrounding me with such de lightful people that have made this such an enlightening experience. I would also like to thank my advisor, Dr. Rajiv K. Singh, whose emphasis on sound reasoning, the fundame ntals of materials sc ience, and creative thinking has contributed immeasurably to my graduate experience. My committee, Dr. Brij Moudgil, Dr. Wolfgang Sigmund, Dr. Stephen Pearton, and Dr. Vaneica Young, have been extremely helpful. I would also li ke to thank Dr. Chang-won Park for his many insightful comments and for dutifully main taining the CMP group, and Dr. Valentin Craciun and Dr. Hassan El-Shall for lending their expertise to particular areas of my research. I am also extremely grateful to Dr. Diola Bagayoko of Southern University for convincing me that the Ph.D. was a realistic goal to work toward. I am extremely grateful to the National Science Foundation Alliance for Graduate Education and the Professoriate Program (N SF-AGEP), the UF College of Engineering Minority Fellowship Progra m, NSF-AGEP Dissertation Fe llowship Program, and the Office of Graduate Minority Programs, th e Particle Engineering Research Center (PERC), and the industrial partners of the PERC for financial support throughout my graduate program. Special thanks go to Dr Anne Donnelly, Dr. Dovie Gamble, Dr. Terry Mills, Dr. Jonathan Earle, and Martha Mc Donald for encouragem ent, moral support and assistance with any problems I encountered. I would also like to thank the PERC and MAIC staff members Gary Schieffle, Gil Brubaker, John Henderson, Mike Beasley, Dr.

PAGE 5

v Emilia Hodge, Dr. Kevin Powers, Sophie Leone, Eric Lambers, Kerry Seiben, and Mike Tollon for their assistance throughout. I am ex tremely grateful for the assistance of Mr. James Albury of the ETD Office at the University of Florida. I would also like to extend my gratitude to Professor Brij Moudgil, Professor Paul Holloway, Professor Rolf Hummel, and Professor Robe rt Dehoff who have set great examples of lifetime excellence in both teaching and research. I thank my colleagues, past and present, Uday Mahajan, Zhan Chen, Seung-Mahn Lee, Bahar Basim, Kyo-se Choi, Wonse op Choi, Wonseok Kim, Madhavan Esayanur, Suresh Yeruva, Kyoung-ho Bu, Su-Ho Jung, Sang Hyun Yoon, Han Ho Choi, Karthik Ramani, Anthony Stewart, Robert Cros by, Eboni Westbrooke, Lizandra Williams, Jeremiah Smith, Nakato Kibuyaga, and Samesha Barnes, for their contributions to this research with valuable di scussions and friendship. Of course I would like to thank my family and friends. My wife Kalia and daughter Asha have provided support, comfort, and love throughout this process. My parents, Candy Grey and Renee McGee; my siblings Marguerita, Paulette, Jonathan, Candon, Pilar, and Joshua; and the rest of my family that have s upported me throughout. Finally, thank you to the brothers of the Omega Psi Phi Fraternity, Beta Pi and Omicron Zeta Chapters.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT.....................................................................................................................xi ii CHAPTER 1 INTRODUCTION........................................................................................................1 Background...................................................................................................................1 Multilevel Me tallization........................................................................................2 Chemical Mechanical Polishing............................................................................2 Scope of the Dissertation..............................................................................................5 2 LITERATURE REVIEW.............................................................................................7 Introduction to Multilevel Metallization (MLM).........................................................7 Motivation.............................................................................................................7 Process Flow..........................................................................................................9 Chemical Mechanical Polishing.................................................................................10 CMP Components...............................................................................................12 Polishing pad................................................................................................13 Wafer............................................................................................................15 Slurry............................................................................................................17 Silica-Silica CMP Mechanisms..................................................................................18 Glass Polishing....................................................................................................19 Silica Hydrolysis.................................................................................................20 Silica CMP...........................................................................................................22 Introduction to Shallow Trench Isolation (STI).........................................................30 Ceria-Silica CMP Mechanisms...................................................................................32 Chemical Tooth...................................................................................................33 Surface Defect.....................................................................................................36 Particulate Generation.........................................................................................37 Ceria Surface Chemistry.............................................................................................38

PAGE 7

vii 3 MATERIALS AND METHODS...............................................................................45 Introduction.................................................................................................................45 Samples.......................................................................................................................4 5 Silica Wafers.......................................................................................................45 Ceria Slurries.......................................................................................................45 Ceria Thin Films..................................................................................................46 Functionalized Cantilevers..................................................................................46 Cantilever oxidation.....................................................................................46 Particle attachment.......................................................................................46 CMP Equipment.........................................................................................................47 CMP Tabletop Polishers......................................................................................47 Friction Force Measurement Apparatus..............................................................48 Particle Characterization.............................................................................................49 Zeta Potential Measurement................................................................................49 Particle Size Analysis..........................................................................................51 Slurry Stability.....................................................................................................52 Characterization of Silica Surface Morphology.........................................................53 Atomic Force Microscopy (AFM).......................................................................53 Scanning Electron Microscopy............................................................................55 X-ray Photoelectron Spectroscopy......................................................................56 Spectroscopic Ellipsometry (VASE)...................................................................58 4 THE EFFECT OF PRESSURE, PH AND VELOCITY ON THE SILICA REMOVAL RATE AND SURFACE FINISH: A STATISTICAL DESIGN OF EXPERIMENTS.........................................................................................................61 Introduction.................................................................................................................61 Selecting a Statistical Design..............................................................................63 Role of Pressure, Velocity, and pH in Silica CMP..............................................64 Experimental...............................................................................................................66 Results and Discussion...............................................................................................67 Removal Rate Response......................................................................................67 Surface Finish Response......................................................................................72 Summary.....................................................................................................................74 5 THE EFFECT OF CERIA SLURRY pH ON THE SILICA REMOVAL RATE, FRICTION FORCE, AND SURFACE MORHOLOGY DURING CHEMICAL MECHANICAL POLISHING....................................................................................76 Introduction.................................................................................................................76 Surfactant Based STI CMP.................................................................................78 Fixed Abrasive Pads for STI CMP......................................................................80 Experimental...............................................................................................................83 Results and Discussion...............................................................................................84 Ceria Slurry Characterization..............................................................................84 Silica Removal Rate............................................................................................88

PAGE 8

viii Friction at the Ceria-Silica Interface...................................................................91 Surface Morphology............................................................................................93 Comparison of pH Effects of Ceri a-Silica to Silica-Silica CMP........................94 Summary.....................................................................................................................97 6 NEAR SURFACE ANALYSIS OF POLI SHED SILICA: CHARACTERIZATION OF THE CERIA-SILICA CHEMICAL REACTION................................................99 Introduction.................................................................................................................99 Ceria CMP Models..............................................................................................99 Recent Progress.................................................................................................100 Experimental.............................................................................................................104 Results and Discussion.............................................................................................105 Interaction Force Measurements.......................................................................105 XPS Analysis.....................................................................................................106 FE-SEM Analysis..............................................................................................109 Wear with Silica-Coat ed AFM Cantilevers.......................................................111 Summary...................................................................................................................115 7 SUMMARY, CONCLUSIONS, AND S UGGESTIONS FOR FUTURE WORK..117 Summary...................................................................................................................117 Conclusions...............................................................................................................120 Suggestions for Future Work....................................................................................122 APPENDIX A DEFINITION OF COMMON STATISTICAL TERMS.........................................124 B BOX BEHNKEN DESIGN AND RESPONSES.....................................................125 LIST OF REFERENCES.................................................................................................127 BIOGRAPHICAL SKETCH...........................................................................................134

PAGE 9

ix LIST OF TABLES Table page 6-1 XPS binding energies for silica sa mples before and after polishing......................108 B-1 Box Behnken design for th ree factor experiment...................................................126

PAGE 10

x LIST OF FIGURES Figure page 1-1 Schematic illustration of a typical CMP tool...........................................................11-2 Correlation of system-level variations with microand nanoscale effects and output parameters.....................................................................................................42-1 Cross-section schematic of MLM struct ure with silica dielectric and aluminum metallization.............................................................................................................72-2 Schematic depiction of cumulative increase in topography for non-planar MLM structures.....................................................................................................112-3 Degrees of planarization........................................................................................122-4 The effect of water on the mechanical properties of polyurethane chemistry.......142-5 Current recorded for one stroke duri ng tribological experiments on a tungsten substrate.................................................................................................................172-6 Material removal rate of silica as a function of the concen tration of silica slurries for differe nt particle sizes..........................................................................232-7 Indentation depth as a function of particle size.....................................................242-8 Particle motion as a function of sili ca particle size and concentration..................292-9 Schematic depiction of device isolation:...............................................................312-10 Glass polish rate versus IEP of common abrasives...............................................332-11 Ceria crystal structure............................................................................................392-12 Potential versus distance for a charged surface.....................................................412-13 pH versus acid/base concentration. .......................................................................433-1 SEM analysis of CMP pad.....................................................................................483-2 Schematic of in-situ friction force apparatus.........................................................49

PAGE 11

xi 3-3 Schematic diagram of AFM apparatus..................................................................544-1 Factor space for diffe rent statistical designs..........................................................674-2 Predictions of removal rate as a function of pressure and pH...............................694-3 Interaction profiles for removal rate as a function of pressure and pH.................704-4 Predicted removal rate as a func tion of pressure, pH, and velocity.......................714-5 Predictions of RMS roughness as a function of pressure and pH..........................724-6 Interaction profiles for RMS roughness as a function of pressure and pH............734-7 Predicted removal rate as a func tion of pressure, pH, and velocity.......................745-1 SEM micrograph of fixed abra sive pad surface (200x) [Vel00]............................805-2 Ceria slurry characteris tics as a function of pH ....................................................845-3 Ceria slurry stability...............................................................................................855-4 FE-SEM image of ceria particles...........................................................................865-5 Silica removal rate as a f unction of ceria slurry pH...............................................895-6 Post-CMP slurry characteri stics as a func tion of pH.............................................895-7 Zeta potential versus pH for ceria, silica, and co-dispersions................................905-8 Friction force and removal rate..............................................................................915-9 Qualitative comparison of electrostatic interactions..............................................925-10 AFM topographical analysis of silica wafers.........................................................935-11 RMS roughness of silica wafers as a function of slurry pH..................................945-12 pH dependent ceria-silica a nd silica-silica interactions.........................................956-1 Ratio of Ce M5/M4 white lines versus position....................................................1036-2 Interaction force...................................................................................................1066-3 XPS analysis of silica...........................................................................................1096-4 FE-SEM analysis of silica polished at pH 6.0.....................................................1106-5 FE-SEM analysis of silica polished at pH 2.5.....................................................110

PAGE 12

xii 6-6 Ceria-silica w ear at pH 2.5...................................................................................1126-7 Ceria-silica w ear at pH 4.5...................................................................................1136-8 Ceria-silica w ear at pH 6.0...................................................................................1146-9 Ceria-silica w ear at pH 11.0.................................................................................1146-10 Silica-silica wear versus pH.................................................................................1157-1 Schematic depiction of pH-depende nt ceria-silica removal mechanism.............121

PAGE 13

xiii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CHARACTERIZATION OF THE CHEMICAL EFFECTS OF CERIA SLURRIES FOR DIELECTRIC CHEMICAL MECHANICAL POLISHING By Jeremiah Terrell Abiade December 2004 Chair: Rajiv K. Singh Major Department: Materials Science and Engineering In a little over a decade, chemical me chanical polishing (CMP) has grown from a niche application to an en abling technology for global plan arization of virtually any material. Initially, CMP was limited to plan arization of the silicon substrate during integrated circuit (IC) manufacturing. Howeve r, CMP has been successfully implemented in all phases of IC manufacturing. Since CMP is materials insensitive and the only process capable of producing globally planar surfaces, it is being investigated for new processing techniques like shallow trench isolation (STI), micro electromechanical systems (MEMS) fabrication, and a variety of other applications where the surface properties control device performance. Despite the widespread acceptance of CMP, it remains the least understood step in the microelectronic device fabrication proce ss. The emergence of novel technologies like STI that demand tighter control over the CMP pr ocess have stimulated increased interest in the fundamentals of the CMP process. In th is work, the fundamental properties of ceria abrasives for STI CMP were studied.

PAGE 14

xiv Ceria slurries are promising for STI CMP because of their in trinsic preferential polishing of oxides over nitrides or polysili con. This advantageous behavior has been attributed to a chemical reaction between the ceria abrasive and the oxide substrate. To date, ceria-silica polishing m echanisms have been supported by little to no experimental evidence. As a result of fundamental invest igations, the silica polish rate and surface finish has been found to be a strong function of the ceria slurry pH, with the maximum polish rate occurring near the isoelectric point of ceria. An assortment of complementary surface analytical techniques was used to study the proposed abrasive-surface reaction that is widely attributed to ceria-silica CMP. The ceria-sili ca reaction has been confirmed by x-ray photoelectron spectroscopy (XPS), fi eld emission scanning electron microscopy (FE-SEM), and atomic force microscopy (AFM) Finally, the surface chemistry of ceria has been reviewed and correlated with ceria-silica CMP observations.

PAGE 15

1 CHAPTER 1 INTRODUCTION Background In little more than a decade, chemical m echanical polishing or chemical mechanical planarization (CMP) has grow n from a niche application to an enabling technology for producing globally planar surfaces. Initially, CMP was developed to minimize topographical variations on Si substrates. The technology has sin ce been applied to dielectrics (SiO2, Si3N4, polymers, etc.), metals (W, Al Cu, Pt, Ni), and semiconductors (SiC, GaN, silicon) for integrated circuit (IC) manufacturing and a variety of other emerging technologies such as microelectro mechanical systems (MEMS) and organic light emitting diodes for flat panel displays. Th e use of CMP in such diverse technologies has created interest in fundamental aspects of the CMP process and development of novel designer slurries with tightly controlled polishing characteristics. Dielectric CMP processes essential for manufacturing metal-ox ide-semiconductor field effect transistors (MOS-FETs) for microprocessor or dyna mic random access memory (DRAM) applications are the focus of this research. Figure 1-1 Schematic illustra tion of a typical CMP tool.

PAGE 16

2 Multilevel Metallization Generally, CMP systems consist of three main components: a polymeric polishing pad, the slurry with abrasive particles, and, of course, the surface to be polished (wafer) (figure 1.1). During CMP, the wafer is pressed against the compliant pa d, which serves as a conduit for the abrasive particles that ar e suspended in an aq ueous solution. Surface topography is reduced by simultaneous chemical and mechanical interactions that may be controlled by judicious selection of part icle chemistry and slurry additives. The popularity of CMP in microelectronic device fa brication corresponds di rectly with device shrinkage, increasing wafer diameter, and an exponential increase in the number of IC components. For example, there were approxi mately 2300 transistors on a single chip in Intel’s first generation of micr oprocessors. This number is e xpected to reach one billion by 2007 [ITRS03]. In fact, Intel recently ma nufactured chips with five hundred million transistors [Biz04]. Due to the finite wafer ar ea available to chip manufactures, multilevel metallization (MLM) techniques were adopted. An IC fabricated via MLM techniques is analogous to construction in crowded urban areas like Ne w York City or Tokyo. Once the available real estate has been consumed, th e only option is to build vertical structures of increasing height. In MLM, alternating layers of dielec tric and metal are deposited on the semiconducting substrate. Since topographical variations are additive, each layer must be globally planar before the next is deposited. To date, the CMP process is the only technique capable of produci ng globally planarized serv ices [Ste97, Sin02, Eva02]. Chemical Mechanical Polishing CMP is a deceptively complex process. Ideally, CMP should produce high removal rates (up to 800 nm/min), global planarity, good surface finish, and high selectivity to underlying barrier or isolation materials all w ith minimal defectivity, dishing, or erosion

PAGE 17

3 [Sin02]. An additional compli cating factor is the number of variables and their interactions that influence the CMP proce ss. Despite its importance, the CMP process remains the least understood st ep during IC fabrication [S in02]. In fact, early CMP processes were based on little more than an empirical understanding of CMP phenomena [Ste97, Sin02]. However, narrowing of the CMP process window because of the increasing needs of microelectronic device ma nufacturers has driven the need to reach a more mechanistic understanding of CMP. A majo r hurdle is the lack of understanding of interactions at the pad-pa rticles-wafer interface. Conve ntional CMP processes rely heavily on end-point detection methods to terminate the CMP process before overpolishing ensues [Sin02]. Although the CM P process has matured significantly since its inception, efforts to understand wafer-slurry interactions still deserve considerable attention. This is especially true for emer ging applications such as shallow trench isolation (STI), which requires alternative sl urries with tunable pr operties for enhanced selectivity and plan arity [Sin02, Eva04]. Initially, silica slurries were used to plan arize interlevel dielectric (ILD) layers. However, the advent of new technologies coupled with increasingly stringent requirements for the CMP process has driven development of new abrasive particles for CMP slurries. Ceria slurries have stimu lated widespread interest because of enhancements in material removal rate, su rface finish, and selectivity over slurries formulated with silica abrasives. The replacem ent of local oxidation of silicon with STI has increased the demand for high selectivity slurries capable of removing the overburden of silica with minimal polishing of the underlyi ng silicon nitride, which isolates active areas of the device. This shift to novel slurries is difficult without an understanding of the

PAGE 18

4 slurry-wafer interact ions. To develop CMP processes for STI, a thorough understanding of the variables that control the CMP proce ss must be reached. In the early stages of CMP, empirical cause-and-effect observations of CMP performance as a function of applied pressure, carrier ve locity, and pad type were utilized to explain CMP mechanisms. Years later, fundame ntal investigations were la unched into the effect of process parameters such as slurry pH, abrasive size, and abrasive concentration to further explain CMP phenomena. Two major obstacles to understanding of CMP mechanisms are: the large number of proce ss variables (figure 1.2) and thei r interactions and the lack of sufficiently sensitive t echniques for measuring nano-scale phenomena during the CMP process. The CMP field has matured to th e point of understanding that systematic investigations of microto nano-scale interactions at th e pad-particleswafer interface must be understood to extend th e applicability of the CMP pr ocess to future generations of IC manufacturing. Figure 1-2 Correlation of system-level variations with microand nanoscale effects and output parameters in chemical mechanical polishng [Sin02a].

PAGE 19

5 Scope of the Dissertation This study will focus specifically on dielec tric CMP processes using ceria slurries. Despite the promise of ceria for STI CMP, few studies exist on the fundamentals of slurry-wafer interactions. Ceria–silica CMP mechanisms are assumed to be identical to ceria–glass polishing mechanisms. Furthermore, ceria CMP efforts are toward the use of surfactants to selectively adsorb onto the silicon nitride surface to enhance selectivity with respect to the overlying silica. Materi al removal during CMP takes place because of synergistic chemical and mechanical interactio ns. Ceria slurries are expected to have an enhanced chemical contribution to polishing. Th erefore, investigations in this study are focused on chemical and surface chemical e ffects that control the CMP process. A synopsis of the efforts in this study follows. Chapter 2 reviews the state-of-the-art of the CMP process. Motivations for the implementation of the CMP process are discu ssed. The roles of the pad, slurry, and wafer are covered in greater detail. Further, th e current understanding of silica-silica CMP processes is summarized. A review of STI and STI CMP follows. Ceria behaves differently than silica both in solution and at the pad–wafer interface. Therefore, the surface chemistry of ceria will be reviewed. Chapter 3 details the materials and methods used for this study. This work was approached with the viewpoint that CMP experiments alone are insufficient to explain polishing phenomena. Experiments and meas urements were obtained using standard laboratory CMP equipment and surface sensitive characterization techniques. Chapter 4 features a statistical examina tion of the effect of pressure, pH, and velocity on ceria–silica CMP. Studies that examine these crucial process parameters

PAGE 20

6 neglect interactions between the variables. Using statistical met hods the individual and synergistic effects ma y be ascertained. Chapter 5 examines the effect of pH on the silica removal rate surface finish, and in-situ lateral friction force. Particle characteristics (zeta potential and particle diameter): before and after CMP will be presented. Chapter 6 is a study of the nano-scale phe nomena that affect the CMP process. Atomic force microscopy (AFM), x-ray phot oelectron spectroscopy (XPS), and field emission scanning electron microscopy (FE-SEM) have been used to characterize the extent of particle–wafer intera ctions during ceria-silica CMP. Chapter 7 is a summary and offers conclusi ons of this work. A pH-dependent ceriasilica polishing model is proposed. Chapter 7 ends with some suggestions for future work for further enhancement of the STI CMP process.

PAGE 21

7 CHAPTER 2 LITERATURE REVIEW Introduction to Multilevel Metallization (MLM) Motivation It is impossible to provide a comprehens ive review of CMP for IC fabrication without discussing multilevel metallizati on (MLM), figure 2-1. Modern ICs contain Figure 2-1 Cross-section schematic of MLM st ructure with silica di electric and aluminum metallization [Sti97]. several million components (transistors, resist ors, and capacitors) on a single chip. To form a circuit, these elements must be connected with some form of wiring. MLM provides a more efficient means of routing me tal lines. Historicall y, each new generation of IC has involved a shrink of the device size [Wil93]. Device sh rinkage results in improved speed, increased packing density, and greater functionality. MLM also provides

PAGE 22

8 a means of separating long, winding metal lines resulting in increased yield and device reliability. CMP is an enabling technol ogy for the production of MLM structures. Combined, these factors are responsible for th e decline in prices a nd product dimensions for seemingly omnipresent items like the de sktop or laptop comput er, mobile phone, and calculator. As the device size decreases, a nd the number of components in the circuit increases, the interconnect becomes a limiting factor in chip performance. The most widely used measure of IC perfor mance is the product of the resistance in ohms (Equation 2.1) and capacitance in fara ds (Equation 2.2), commonly called the RC time constant (Equation 2.3), measured in seconds, which is the time required for a voltage at one end of a metal line to reach ~ 63% of its final value at the opposite end [Ste97]. R = l/wd = l/A (2.1) C = wl/t (2.2) RC = l2/td (2.3) Where is the resistivity of the metal line, l, w, d, and A are the leng th, width, thickness, and cross sectional area respectively and is the permittivity of the dielectric, which equals the insulator dielectric constant ( r) times the permittivity of free space ( o), and t is the thickness of the insulator. According to (Equation 2.3), the time constant may be lowered by using metal lines of lower resistivit y (i.e. replacing Al with Cu), insulators with lower dielectric constants (i.e., porous silica, or polymeric materials instead of SiO2), or by decreasing the length of the meta l lines. Equation 2.3 does not show a direct dependence on line width, but as devices are continually miniaturi zed, the metal line thickness (d) and interlevel (ILD) thickness (t ) are also scaled to smaller dimensions,

PAGE 23

9 causing an increase in the time constant. Scali ng the device dimensions is one alternative for improving device speed, since device delay decreases with decreasing feature size [Ste97]. However, for 0.5 m features, interc onnect delay accounts for ~ 25% of the total delay, while for 0.25 m dimensions, ~ 50% of the delay is due to the interconnect [Ste97]. With continued scaling, interc onnect performance dominates device speed resulting in diminishing return s as the dimensions are aggr essively reduced. Any change that decreases l, or increases t and d will enhance device speed by decreasing the interconnect delay. Since RC is proportional to l2, the most dramatic decreases in interconnect delay can be achieved by avoidi ng long metal lines. This is possible using MLM [Wil93, Ste97]. Process Flow The process flow for MLM is as follows [Wil93]: silicides (metal-Si compounds) are formed on the source, gate, and drai n of MOS transistors as well as on doped polysilicon films that are used as resistors or contacts in bipolar ci rcuits. Next, the premetal dielectric (ILD0) is deposited. The ILD0 and subsequent levels of dielectric (ILD1, 2, 3, ...) are different materials. Boron phosphate silicate glass (BPSG) [Ols93] is commonly chosen for the ILD0 layer whereas silica (SiO2), silicon nitride (Si3N4), or silicon oxynitride (SiOxNy) are used for the ILD. Depending on the planarity requirements and thermal budget, planariza tion ensues via thermal flowing [Wil93], etchback [Che97], or CMP [Ste97]. The ILD0 la yer is then patterned and etched to form contact holes through the oxide to the underlying structures or silicide. The holes or vias (V1, 2, …) are then filled by barrier film if none exists on silicide, then a conducting film (Al or W) is deposited. Currently, W is the metal of choice because of its ability to conformally fill high aspect ratio vias [Ire97] The W is etched back to form a plug. The

PAGE 24

10 interconnect metal (M1, 2, …) is then deposit ed. Cu has replaced Al as the metal of choice for the metal interconnect becaus e of its low resistivity and improved electromigration resistance resulting in fast er, more reliable devices [Lee03]. Vertical structures are built by replica ting the procedure of dielectri c deposition, etching to form vertical vias, patterning and interconnect deposition. The MLM pro cess is completed by depositing a final dielectric la yer that hermetically protec ts the underlying layers from corrosive ions, moisture, and solder chemi cals. For feature sizes of ~100 m, 12 metal layers are required when us ing Al interconnects and SiO2 as the ILD. In contrast, only 6 levels are required using Cu interconnects and a low dielectric constant material [Sin98]. MLM is not a simple process. A variety of t ools, materials, and t echniques are involved. Stringent control must be maintained over each sequence in the process. Chemical Mechanical Polishing Initially, it may seem counter -intuitive to remove any of the dielectric or metal layers that have been so carefully de posited, especially by exposing the pristine environment of microelectronic fabs to su ch a dirty process as CMP. Topographical variations or thickness non-uniformities are deleterious to device performance [Ols93]. Also, in multilevel structures necessary fo r ICs, topographical variations are additive (stack-up effect) [Vie01]. There are seve ral sources of topography generation during fabrication. Non-uniform deposition may occu r if curved wafers are used. Wafer curvature may be because of intrinsic stress in the substrate [Ste97]. The fabrication process itself may contribute to non-planarit y. Repetitive cycles of deposition, patterning, and etching are known to generate steps at the surface [Ste97]. Auto-focus errors, residual lens aberrations, and resist thic kness variations may also produce non-planar surfaces during lithography [Gry93, Ste97]. Planar ization has its benefits at each level of

PAGE 25

11 the MLM structure [Ols93]. Planarization of the pre-metal and interlevel dielectric improves the patternability of metal inte rconnects, decreases line resistance, and increases the reliability of the device. Also, as topography is minimized before ILD and metal deposition, the stack-up effect is re duced, which reduces the complexity of subsequent planarization steps. Along with the benefits of dielectric planarization, metal planarization results in an increase in step coverage, device speed and device yield, and a reduction in manufacturing cost. Figure 2-2 Schematic depiction of cumula tive increase in topography for non-planar MLM structures [Vie01]. Several techniques capable of achieving some level of planarization exist. For example, thermal flow of BPSG films is used to planarize the pre-metal dielectric layer. After chemical vapor deposition (CVD) of the BPSG film, an anneal step is conducted to soften and flow the film over voids or thickness non-uniformities [Ols93]. This technique is not universally applicable because of the presence of metals with low melting

PAGE 26

12 temperatures like Al. Resist etchback (R EB) or reactive ion etching (RIE) uses a sacrificial spun-on material as a planariz ing agent [Ols93, Ste97] After dielectric deposition onto the interconnect, photoresist is spun onto the entire wafer. The smooth surface of the resist is transf erred to the underlying dielec tric layer by a non-selective blanket etchback process. The photoresist etch ra te is selected to match that of the oxide etch rate so that the surface is removed uni formly. A second layer of dielectric is then deposited to achieve the desired dielectric thickness. This tech nique is undesirable because several attempts are necessary to ac hieve planarization and the plasma used for etching is composed of hazardous gases [Mur 93]. Thermal flowing, reactive etching, and all other planarization techniques have one common flaw: none can produce universal globally planarized surfaces. CMP Components CMP is the only technique capable of producing global planarization. Figure 5.3 shows schematically the difference in pl anarization length for local and global planarization. Generally, CMP systems consis t of only 3 main components, pad, slurry, and wafers. Currently, intense research is being conducted to understand the fundamentals of the process. Figure 2-3 Schematic depiction of the degrees of planarization.

PAGE 27

13 Polishing pad Until recently, CMP polishing pads have received little attention. Although the polishing pad structure determines the polish rate and planarity of a CMP process, pad effects and properties are poorly underst ood. Generally, manufacturing processes for CMP pads are empirically derived. Because of the ability to control pad characteristics, polyurethane impregnated felts or cast polyu rethane foams with filler materials are the most desirable materials for commercial CMP pads [Ste97, Jai94]. A hydroxylterminated resin known as polyol is reacted with di-isocyanate, in the presence of a catalyst and foaming agent to form polyuret hane, a condensation polymer [Jai94]. Filler material may be added to achieve the desired mechanical properties. During the CMP process, the pad has 3 functions [Ste97]. Pores in the bulk transport slurry to the pad–wafer interface, the pad asperities provide points of contact to form the abrasive–wafer interface, and the pa d also transports abraded material away from the pad–particles–wafer in terface. Appreciable remova l rates are expected only when the abrasive particles are embedded in the CMP pad and slide against the wafer surface [Zha99]. The difficulty in engineering the performance of CMP pads is related to ambiguity on the effect of pad properties on the CMP performance. Specific gravity, compressibility, hardness, shear modulus, dyna mic shear modulus, pad asperity height, width, and distribution, and pad roughness cont rol planarization efficiency. To achieve planar surfaces, the pressure at high features should exceed the pressure at low features. Empirical results indicate hard pads with low compressibility are more effective in achieving planar surfaces, because they do not conform to wafer topography. However, considerable difficulty exists in determini ng what pad properties should be measured to determine pad hardness. The dynamic nature of the CMP process further complicates the

PAGE 28

14 measurement. Through diffusion, water mo lecules break hydrogen bonds in the polyurethane structure, which l eads to pad softening (figure 5.4) [Hog73]. The pad performance, in terms of plan arity and removal rate, is de pendent on the its modulus of elasticity (G) [Ste97]. To maintain the pressu re differential at high and low features, high G values are desirable for CMP pads. N H C O O H N C O OHydrogen Bond N C O O H O H O C N O H O H HBond breakage by water moleculesCase A: low water content Case B: high water content N H C O O H N C O OHydrogen Bond N C O O H O H O C N O H O H HBond breakage by water moleculesCase A: low water content Case B: high water content Figure 2-4 Schematic depicition of the effect of water on the mechanical properties of polyurethane chemistry adapted from [Hog73]. Polyurethane responds elastically and vi scoelastically to shear stresses during CMP. CMP phenomena occur in short timeframes–approximately micro to milliseconds. Further insight into planariz ation behavior is possible by measuring the dynamic shear modulus (G ) under periodic stress. G increases with measurement frequency. Planarity has also been shown to increase with pad ro tation speed. At higher frequencies, the pad

PAGE 29

15 has less time to deform, decreasing the pa d sensitivity to wafer topography. Early measurements of pad properties were conducted in dry conditions. More recent measurements have shown a dependence of pad properties on water content, temperature, polishing cycles, and slurry chemistry [S te97, Mur94, Jai94]. Water soaking and pad heating are known to lower the mechanical pr operties of the pad. Alteration of the pad properties leads to inconsistent CMP perfor mance. To maintain reproducible conditions, CMP pads are conditioned between runs or dur ing the process. In our studies, a diamond pad conditioner was used between CMP runs The conditioner reproduces the original rough surface and also removes slurry particles from the previous CMP run. Conditioning extends pad lifetime and prevents glaz ing–plastic wear of the pad surface. Fixed abrasive pads–wherein the pad is co ated with slurry particles–are receiving increased attention. With fixed abrasive pads, slurry handling is no longer an issue. Fixed abrasive pads are also less sensitive to wafer topography, which may increase the planarization capability of the CMP process. Si nce slurry stability is not a concern, more diverse slurry chemistries may be investig ated. Fixed abrasive pads are potentially applicable to STI [Gor03], Cu CMP [Ngu01], and MEMS [Kul04] fabrication processes where over-polishing, erosion, and pattern de nsity effects complicate the conventional CMP process. Wafer Virtually all materials are candidates for planarization via CMP. Knowledge of the chemical, structural, and mechanical propertie s of the surfaces to be polished is key for designing materials-specific CMP processes. The mechanisms for dielectric and metal CMP differ significantly. Metal CMP may be considered a re petitive corrosion/erosion process [Ste97], wherein the slurry chemistry is tailored to form a modified surface layer

PAGE 30

16 or oxide scale. This scale is then rapidly removed by repetitive abrasive contact. To control removal rate, planarity, defectivity, and selectivity considerable latitude exists in metal CMP processes. Several chemical addi tives are used in metal CMP slurries to control the kinetics of layer formation and the mechanical propertie s of the surface layer [Ste97, Sin02, Lee03]. Metal CMP slurries are more chemically active than dielectric slurries. Generally, dielectric CMP processes are based on diss olution of the substrate or wafer via slurry additives or abrasive pa rticle indentation. Chemical processes in dielectric CMP are largely due to the presen ce of water. For example, little polishing is observed in oil or alcohol based slurries [Coo90] even with diamond abrasives [Ste97, Coo90]. Few techniques exist that are capable of measuring the modified surface layer properties in-situ. This problem is particularly glaring in dielectric CMP. However, for metal CMP, electrochemistry and surface sens itive thin-film techniques have been used to determine the properties of the metal su rface layer. Bielmann demonstrated the active– passive behavior in W [Bie98] by measuring the current vs. time while rubbing a sphere against the W surface and while motionless. Also, x-ray photoelectron spectroscopy (XPS), chrono-amperometry, potentio-dynami c measurements, and x-ray reflectivity techniques were applied to Cu CMP sy stem. To achieve planarization without contributing to surface topography, the CMP proce ss must interact with only the topmost atomic layers. For dielectric CMP, and sili ca in particular, attempts to characterize surface layer or ‘gel’ layer formation ar e conducted ex-situ. Ellipsometry [Fran99], infrared reflectivity [Wal96], and atomic fo rce microscopy (AFM) [Adl00] indicate the presence of a softened surface layer w ith thickness in the nanometer range.

PAGE 31

17 Figure 2-5. Current recorded for one stroke during tribological e xperiments on a tungsten substrate [Bie98]. Slurry CMP slurries consist of abrasive particle s 50-500nm in diameter suspended in deionized (DI) water. Abrasive particles in the slurry are responsible for mechanical abrasion, the M in CMP. Highly re active particles like ceria (CeO2) and zirconia (ZrO2) may also contribute to the chemical co mponent via particle–substrate bonding. The abrasive particles are also responsible fo r pad glazing during the actual CMP process. Residual particles must be removed between each CMP run. The abrasive particles may be characterized by measuring a variety of pr operties. The particle type, hardness, size, shape, isoelectric pH–the pH at which the surface is neutral–ch emical reactivity, and solubility in the slurry environment all aff ect the planarization ra te, surface finish, and defectivity of the process. The trend in slu rry formulation is toward more ‘gentle’ CMP processes [Sin02]. Hard abrasives like alumina (Al2O3), which were originally used in W and Cu CMP are being replaced by ceria (limited cases), silica, or abrasive-free slurries that contain little to no particles. For di electric CMP ceria is gaining widespread attention. Selectivity, enhanced removal rate and improved surface finish are the factors

PAGE 32

18 that have driven interest in ceria-based slurries. Oxidizers, corrosion inhibitors, complexing agents, stabilizers, dissoluti on agents, and surfactants are among the chemical additives found in CMP slurries [Si n02]. Dielectric CMP slurries tend to be less complex. Dispersants–to inhibit particle agglomeration dissolution agents–to enhance surface solubility and surfactants–to enhan ce polish selectivity tend to be the only additives in CMP slurries. Traditionally, sl urries for silica CMP consisted of silica particles. The most intense research efforts in CMP ar e in the areas of slurry formulation and particle properties. A tremendous potential for commercialization ex ists in CMP slurry manufacturing. In less than te n years the CMP slurry market has grown from non-existent to a billion dollar enterprise [Sin03]. Currently, the slurry market accounts for ~ 40% of the worldwide nano-particle market share. These facts are wonderful for industrial companies or the researcher who develops marketable CMP slurry. However, for the academic researcher, the tremendous profitabi lity has led to re luctance to release proprietary information, which complicates efforts to understand CMP fundamentals. What is available in the open and patent literature does not sufficiently explain CMP mechanisms. Fundamental investigations into slurry properties are necessary, especially in a setting where a sustained effort can be ma intained. Considerable room exists in the academic arena to contribute to the knowledge of CMP slurry development, particularly in areas like STI and MEMS that will be introduced in the near future. Silica-Silica CMP Mechanisms Aside from planarization of silicon wafers the most fundamental studies in CMP exist in oxide CMP for back-end planarization after pre-metal and interlevel dielectric deposition [Ste97]. CMP of silica with silica abrasives is a mature process and may be

PAGE 33

19 considered a model for other dielectric CM P processes. This section will cover the current understanding of sili ca–silica CMP fundamentals. Much of the knowledge in oxide CMP is based on over a century of gla ss finishing for optical lens applications [Cook90, Ste97, Kel98]. A significant amount of polishing data exists for many abrasive types diamond and oxides of iron, cerium, chro mium, silicon, zirconium, titanium, zinc, lanthanum, nickel, and aluminum [Cor63, Iz u79, Izu84, Coo90, Ste97]. Therefore, it is worthwhile to begin this section of th e review with glass polishing mechanisms. Although these glasses are not used in IC ma nufacturing, the variance in their physical properties is useful for unders tanding polishing mechanisms in glass, including that of silica for ILDs in IC manufacturing [Ste97]. Oxide CMP may be considered a specialized form of glass polishing. Glass Polishing The mechanisms for glass polishing may be grouped into three categories: wear, flow, and chemical [Hol64, Isu79]. The wear mechanism suggests surface smoothing is wholly mechanical because of increased contac t at high points resulting in minimal wear at low points on the surface. The flow mechan ism suggests the glass surface flows during polishing from peak to valley regions, until a smooth surface is achieved. Finally, glass polishing via a chemical mechan ism is attributed to increas ed dissolution at high surface features because of the high pressures exer ted by the abrasive particles. Izumitani determined the glass polishing mechanism by examining the polish rate as a function of glass hardness (bulk), softeni ng point–a measure of a gla ss’ resistance to flow, and chemical durability [Izu79]. Among the three, only chemical durability influenced the polish rate. The chemical durabil ity of a glass is measured by the weight loss in water or dilute acids. Weight loss occurs because heavy modifier ions (Pb2+ and La+) are leached

PAGE 34

20 from the glass and replaced by lighter hydroge n ions [Ste97]. The influence of chemical durability on polish rate indicates that chem ical reactions influence the glass polishing mechanism. Chemical modification of the gl ass surface alters the surface properties, including hardness. Izumitani measured the effect of leaching time on Vickers microhardness and polish rate versus surface ha rdness of glasses. Here the hardness was shown to decrease with leaching time and the po lish rate correlated well with the surface hardness of the glass [Izu79]. Glass polishing is crucially dependent on the presence of water [Coo90]. Polishing rates are negligible in hydrocar bon liquids like kerosene [Izu84], paraffin [Cor63], or oil [Izu79] and non-hydroxyl containing liquids like formamide [Cor63]. Furthermore, the polish rate in dry conditions or oil solutions correlated with the bulk hardness of the glass [Izu79]. Surface modification is not expected in dry or oily conditions. In th ese conditions, wear dominates the glass polishing mechanism. The surface quality of gl ass polished in solutions other than water or in dry conditions was poor and polish rate s were significantly lower. Therefore, optimum glass polishing occurs due to synerg istic interactions between chemical and mechanical interactions. Silica Hydrolysis From the preceding discussion two points s hould be emphasized. First, water-based slurries are essential, both for effective pl anarization and desired surface finish. Second, during CMP the abrasive particles only intera ct with the topmost, chemically modified layers of the surface to be polished. In the case of silica, this modified layer results because of hydrolysis, which proceeds as follows [Ile79]. SiO2 + 2H2O (SiO2)x-1 + Si(OH)4 (2.4)

PAGE 35

21 In the hydrolysis reaction, Equation 2.4, inte ractions between siloxane bonds (Si— O—Si) and water determine the behavior of silica during CMP. Budd described this reaction, in three stages, as the a ttack of the solvated hydrogen ion (H3O+) on the negative sites of silica [Bas02a Bud61]. First, a water molecule from the environment attaches to the Si—O—Si) bond by formi ng a hydrogen bond between its hydrogen and the oxygen atom of the silica surface; lone pair orbitals of the oxygen from water interact with the Si atoms. Second, hydrogen bound to water transfers a pr oton to the oxygen of silica, while an electron is simultaneously tr ansferred from the oxygen of water to the Si atom. Finally, two new bonds are formed (Si—O—H), by destroying the original siloxane bond. The dissolution of silica o ccurs if all four bridging oxygen bonds are hydrated, forming the solute species Si(OH)4 (silanol) [Ile79, Coo90, Tro94]. The rate of this reaction is controlled by the diffusion of water into the silica su rface, the pH of the polishing slurry, and the indent depth of the ab rasive particle into the silica surface [Ile79, Coo90]. The dissolution rate of silica incr eases rapidly above pH 9.0. Therefore, formulating slurries in the ba sic pH regime enhances the removal rate and the surface finish [Cook90]. According to Budd, the OHion acts a catalyst for water attack on the silica network [Bud61]. OHions create excess electrons resulting in hi gher negative surface potentials, which increases the affinity for attack by H3O+. The diffusion of water is known to aff ect the surface hardness [Izu79, Hir87], fatigue [Han90, Han91], and structure of s ilica [Tro94, Dav96, Wal96, Adl01]. However, attempts to measure the properties, specifi cally the thickness and hardness, of the chemically modified layer on silica formed during CMP have been inconclusive. Trogolo and Rajan have reported, using transmi ssion electron microscopy (TEM) and Fourier

PAGE 36

22 transform infrared reflectivity (FTIR), the presence of two near-s urface regions, 3 and 1520 nm below the surface after CMP [Tro94] The 3nm surface layer possessed density lower than the bulk value, with the 15-20 nm layer displaying an increased density with a gradual return to bulk values ~ 20nm below the surface. However, Wallace et al., using x-ray reflectivity, reported the opposite [Wal96] According to their measurements, the near-surface region after CMP is compacte d, resulting in an increased density. They further dispute the validity of a 20 nm region of increased density, because the accompanying increase in electron density due to water diffusion was not observed. Adler observed a short-range repulsion at th e silica surface by direct measurement of surface forces with AFM [Adl01]. This behavior was attributed to the formation of a ‘gel layer’ because of swelling of the silica surf ace as a result of water diffusion. Since in-situ techniques capable of measuring these phenom ena are not readily available, research efforts in oxide CMP have centered largely on particle effects. Fundamental observations have also been reported by m easuring variations in the in -situ friction force during CMP conditions [Mah99a, Mah99b, Mah00, Cho03, Cho04a, Cho04b, Cho04c]. Silica CMP In the last five to six years, considerab le progress has been made in CMP research. Our group in the Particle Engineering Research Center (PERC) has b een at the forefront, particularly in oxide CMP. The properties of th e abrasive particles in CMP slurries have a significant impact on the polishing behavior. Alteration of the particle size, size distribution, and/or solids lo ading changes the number of abrasives available for polishing at the pad-wafer interface. This a ffects the polish rate and surface finish by modulating the load per abrasive particle. Therefore, the poli shing behavior as a function

PAGE 37

23 of particle size, size distribution, solids loading, and the contact characteristics at the pad–wafer interface should be understood. The particle size effect has been stud ied by a number of researchers with conflicting results. Brown and Jairath reported an increase in polish rate with an increase in particle size [Bro81, Jai94] Izumitani reported the opposite, a decrease in polish rate with an increase in particle size [Izu79] In total disagreement, Cook and Sivaram concluded that the polish rate was not in fluenced by the particle size [Coo90, Siv92]. Despite the numerous studies on the effect of particle size in oxide CMP, considerable confusion remained. By studying the effect of particle size (0.2 m-1.5 m) and solids loading (0.2-15 wt%) simultaneously, Mahaja n, a former member of the CMP group in the PERC, was able to clarify the particle size effects [Mah00]. According to his results, the polish rate was highest for larger particles at the lowest solids loadings. Except for 0.2 m particles, the polish rate decreased w ith increasing solids loading (figure 5.5). Furthermore, silica surfaces polished with la rger particles showed significant surface deformation, whereas, 0.2 m abrasive particles produced smoother surfaces. Figure 2-6 Material removal ra te of silica as a function of the concentration of silica slurries for different particle sizes [Mah00].

PAGE 38

24 To explain the CMP behavior as a function of particle size, two mechanisms were proposed [Bie98]. The first is a contact or surface area based mech anism, in which the polish rate depends on the total contact ar ea between the abrasive particles and the surface undergoing CMP. The following expressi on gives the dependence of contact area on particle diameter: A C0 1/3 -1/3 (2.5) where A is the contact area, C0 is the abrasive concentration, and is the particle diameter. According to Equation 2.5, surface ar ea controls the removal rate for smaller size particles. Also, more part icles are available for polishing due to tighter packing at the pad–wafer interface. For the second, an inde ntation–based mechanism, polishing occurs due to penetration of the surface by the abrasi ve particles. The indent volume may be expressed as follows: V C0 -1/3 4/3 (2.6) where V is the indent volume. Other results have also indicated th at the indent depth increases with increasing abrasive particle size [Coo90, Qin04], figure 2-6. Figure 2-7 Schematic depiction of the indenta tion depth as a functi on of particle size. CMP processes must produce planar surf aces with minimal surface deformation. Surface deformation or defectivity may be cla ssified as scratches, pits, film delamination,

PAGE 39

25 and/or chemical or particul ate residue [Ste97]. The presence of microscratches has a deleterious effect on micr oprocessor performance [Sch95]. Earlier glass polishing experiments indicated that the surface deformation was less when slurries were formulated with monosized particles [Coo90]. A small percentage of oversized particles may be responsible for pitting or scratches af ter CMP. To determine the role of oversized particles during polishing, Ba sim formulated slurries with both hard and soft agglomerates for silica CMP [Bas02a]. The e ffect of hard agglomerates was studied by adding small amounts of oversized particles to the otherwise stable monosized, and welldispersed slurry [Bas00, Bas02a]. Soft agglom erates were grouped in to 3 categories: 1) dry aggregates or partially di spersed slurries; 2) polymer fl occulated slurries, which were created by adding 0.5 mg/g of polyethylene oxide (PEO); 3) coagulat ed slurries, which were produced by adding, 0.2M, 0.4M, and 0.6M NaCl [Bas02a, Bas02b]. In the case of hard agglomerates [Bas00, Bas02a], oversized particles were added to the slurry until the particle size measuremen t tool was able to detect their presence (~1.0 wt%). The baseline slurry consisted of 0.14 m sized silica par ticles at 12 wt%. 0.5 m, 1.0 m, and 1.5 m Geltech particles were used to repr esent hard agglomerates. For all slurries except those spiked with 1.5 m pa rticles, the removal rate was less than that obtained using the baseline slurry. In all cases, the surface finish worsened when oversized particles were used. Scratche s and pits were observed on the surface. Furthermore, surface deformation worsened as the concentration of coarse particles was increased. Basim explained these observations by considering the load distribution per particle. According to her results, coarse particles acts as pillars at the pad-wafer interface, decreasing the opportunity for smaller, primary slur ry particles to participate in

PAGE 40

26 the polishing process [Bas02a]. This asser tion has two consequen ces: first, larger particles experience an increased load in comparison to smaller particles, and second, the total contact area at the pad–wafer interface de creases, which is because of a decrease in the effective number of abrasives at the pa d–wafer interface. The p itting depth after CMP increased as the coarse particle size increas ed, because of the increased pressure and greater penetration depth of the larger particles [Bas02]. Hard agglomerates may be filtered as the slurry is pumped to the pad-wafer interface. However, even after filtering, de fectivity may still be high [Ewa99]. Scratches or pits in this case may be attributed to tr ansient or soft agglomerates formed during the CMP process [Bas02]. Van der Waals or w eak physical bonds are responsible for aggregation in the case of dry agglomerat es. Post–CMP particle size measurements indicate agglomerated particles are broken apart during the polishing process. During polishing, these interc onnected particles experienced th e majority of the load. Higher removal rates were obtained at the expense of surface quality. The effect of PEO addition was also studied for dispersed slurries, in which the polymer molecules are believed to completely surround each abrasive particle a nd flocculated slurries in which the mean particle size shifted to 5.82 m from 0.20 m. For flocculate d slurries the polish rate slightly decreased with a significant increase in defectivity. However, for well–dispersed polymer containing slurries, su rface deformation was comparable to the baseline values, with diminished removal rate. AFM force m easurements indicated that the polymer acts as a lubricant at the pad–wafer interface, which is responsible for the removal rate decrease [Bas02a, Bas02b].

PAGE 41

27 The effect of coagulation on slurry perf ormance can be simulated by adding ionic salts to the slurry. In the studi es of Basim, coagulated slur ries were formulated by adding different concentrations of NaCl. Salts enhance dissolution of the silica surface [Hou92, Dov92] and increase electrosta tic interactions at the padwafer interface by screening charge on the opposing surfaces [Mah99b, Bas 02a]. For 0.2 M, the particle size was unchanged. However, the removal rate incr eased along with the maximum pitting depth. For 0.4 and 0.6 M solutions, irreversible coa gulation ensued. The particle diameter increased in both cases along with an increa se in root mean square roughness, and maximum pitting depth. Furthermore, the pa rticle size before and after CMP was identical. Meaning, the particle s remain coagulated during po lishing, or re-coagulate after polishing in the post-CMP slurry. To this point, the effect of particulate properties has been covered. The research reviewed, although beneficial, has not comple tely uncovered the oxide CMP mechanism. A precise description of how th e abrasive particle interacts with the surface layer from the micro to the nano-scale is needed. In me tal CMP, surface layer formation is easily quantifiable. The layer properties may even be tuned for the desired CMP application. In dielectric CMP, and oxide CMP in particular it is extremely difficult to quantify the extent of surface modification that occurs in the CMP environment. Mechanistic investigations have instead focused on inte rfacial interactions. In-situ friction force measurements are the most widely used diagnostic tool for oxide CMP interfacial phenomena [Mah99a, Mah99b, Sik01, Hom 03, Cho03, Cho04a, Cho04b, Cho04c]. From a micro and nano-scale viewpoint, the surface removal rate (RR) can be estimated by the following equation [Sin02]:

PAGE 42

28 RR = Kc( xScPc)V (2.7) where Kc is a constant, x is the depth of indentation by a particle, Sc is the fraction of the pad surface covered by abrasive particles, Pc is the contact area of the pad on the surface (< 1% [Bas02a, Bas03]), and V is the linear velocity difference between the particle and the wafer. Sc is dependent on particle size, concen tration, and the state of dispersion. Pc is a function of the applied pressure, relative ve locity, and mechanical properties of the pad, including pad degradation. In-situ friction for ce measurements may be used to estimate Sc and Pc [Sin02]. Using a sapphire (Al2O3) substrate and silica abrasive s Choi et al. attempted to measure the contact area as a function of particle size, down pressure, and particle concentration [Cho04b]. Sapphire a hard, inert material, was used to determine the specifics of mechanical contact in the absence of surface roughness effects. The normalized fractional surface coverage (Sc) is a strong function of particle size. For smaller sized particles, the pad was almost completely covered with particles [Bas02, Cho03, Cho04b], which corres ponds to a value for Sc near 1. Sc showed a weak dependence on concentration and no pre ssure or down load dependence. These measurements confirm Basim’s assertion that ove rsized particles act as pillars at the padwafer interface [Bas00, Bas03]. By varying the particle size and particle concentration, Choi et al. confirmed ea rly polishing results. Friction force measurements showed analogous transitions from high to low friction (figure 2-7) [Cho03, Cho04c] as a function of par ticle size as the removal rate (figure 2-6). For 0.2 m particles, the fric tion force steadily incr eased with particle concentration. However, larger sized part icles 0.5 m, 1.0 m, and 1.5 m showed a

PAGE 43

29 transition from high to low friction as part icle concentration increased. For larger particles, the transition occurred at lower part icle concentrations. Increasing the particle concentration results in an increase in the number of abrasives available for polishing at the pad-wafer interface. For 0.2 m particles, more efficient packing is achieved as evidenced by an increase in the fraction of the polishing pad covered by abrasive particles [Cho03, Cho04c]. As a result, pr essure at the pad-wafer interface is more uniformly distributed. This assertion is confirmed by the reduction in the root mean square (RMS) surface roughness as a function of particle c oncentration after polishing. According to these results, 0.2 m particles are comple tely embedded into the polishing pad and remove material by sliding on the wafer surface. Time (sec) 010203040506070 Friction Force (N) 1 2 3 4 5 6 7 8 9 30 15 12 10 8.0 5.0 2.0 1.0 0.5,0.2 (wt.%) (a) (c) (d) Time (sec) 010203040506070 Friction Force (N) 1 2 3 4 5 6 7 8 9 30 15 12 10 8.0 5.0 2.0 1.0 0.5,0.2 (wt.%) (a) (c) (d) Figure 2-8 Dynamic particle motion as a functi on of silica particle size and concentration. As particle size increase, the sliding to rolling transition occurs at lower solids concentration. a) 0.2 m sliding mo tion, b) 0.5 m sliding to rolling transition occurs at 8 wt%, c) 1.0 m s liding to rolling transition occurs at 5 wt%, d) 1.5 m sliding to rolling tran sition occurs at 4 wt% [Cho03, Cho04b]. In the particle size range (0.5-1.5 m), th e transition from high to low friction may be considered the boundary between pre dominantly sliding particle motion to

PAGE 44

30 predominantly rolling motion [Cho04c]. Again the surface or contact area of the particles may be invoked to explain these phenomena. Th e decrease in surface area with increasing particle diameter results in less interacti on area between particles and wafer. According to a CMP model proposed by Zhao and Shi [Zha99], the material removal mechanism is dependent on a threshold pressure (Pth). Pth marks the onset of sliding to rolling or vice versa. For P < Pth, material removal rates are expect ed to be negligible and polished surfaces will have a ‘shot blast a ppearance’. Conversely, for P > Pth, the particles are embedded in the compliant CMP pad and surfac es with scratches are produced. Using insitu friction force measurements, theoretical predictions of the exis tence of a threshold pressure were experimentally validated. Introduction to Shallow Trench Isolation (STI) The ability to produce surfaces with global planarity has resulted in the application of CMP in emerging technologies. Applica tions where the surface dominates device performance (i.e. optoelectronic devices, hi gh k dielectrics, packaging, and flat panel displays) are candidates for CMP [Ste97]. Shallow trench isolation (STI) is critically dependent on the planarization cap ability of the CMP process. The continued march down the Inte rnational Technology Roadmap for Semiconductors (ITRS) has profound implications on materials selection, processing, and device architecture. Specifically, the demands for smaller geometry and higher circuit density require more effectiv e isolation of active regions of complementary metal oxide semiconductor circuits [Lin99]. Traditionall y, local oxidation of silicon (LOCOS) has been the method of choice for device isola tion. In the LOCOS pro cess, silicon nitride (Si3N4), a barrier to oxidation, is selectively deposited onto the sacrificial or pad oxide. The silicon substrate is then ther mally oxidized in regions without Si3N4. After removal

PAGE 45

31 of the Si3N4 by chemical etching, the thermal oxi de provides electrical isolation by separating active regions of the device. LOCOS contributes to nonplanarity and reduces device packing density, which is unacceptable for device dimensions less than 0.5 m [Ste97, Lin99, Sch99]. The STI process allows dense, more functi onal, high-speed devices to be fabricated at lower costs [Zha01]. Furthermore, the in troduction of nonplanarity because of ‘bird’s beaks’ associated with LOCOS is alleviated with STI (figure 2-9) In the STI process, trenches are etched into the Si3N4 mask layer and the Si substrate. Chemical vapor deposition (CVD) or spin on glass (SOG) techniques are used to fill the trenches with SiO2. CMP or a combination of RIE and CMP then remove the overlying oxide. The planarized trench-fill oxi de isolates the device after removal of the Si3N4 mask. To reduce process complexity, single-step STI CMP processes are being developed [Zha01]. Shallow Trench Isolation(STI) Si3N4 SiO2 Si Si3N4 SiO2 SiLocal Oxidation of Silicon (LOCOS) Si3N4 SiO2 SiO2 ------isolated region SiO2Si3N4 Si substrate Si3N4 SiO2 SiO2 ------isolated region SiO2Si3N4 Si3N4 SiO2 SiO2 SiO2 ------isolated region SiO2Si3N4 Si substrate-----Si3N4 SiO2isolated region SiO2Si3N4 -----Si3N4 SiO2isolated region SiO2Si3N4-----Si3N4 SiO2isolated region SiO2Si3N4 Si3N4 SiO2isolated region SiO2Si3N4 SiO2-----isolated region Si substrate SiO2-----isolated region SiO2-----isolated region-----isolated region Si substrateIdeal result of STI CMP step STI structure after complete nitride and oxide removal (a) (c)(d) (b)Si substrate Figure 2-9 Schematic depiction of device isola tion: a) LOCOS b) STI before CMP c) STI after CMP d) STI structure after nitride and oxide removal. STI is not possible without CMP. An ideal STI CMP process stops at the nitride layer and requires uniform removal of the tr ench-fill oxide (figure 2-9c). Naturally, STI

PAGE 46

32 CMP slurries must have high selectivity to ox ide with respect to th e nitride, and should leave the surface planar and def ect-free. Over polishing of th e nitride or erosion of the oxide causes damage to the active region a nd degrades isolation performance [Lin99]. Since STI CMP is a device level process, it must be more tightly controlled [Sch99]. Defectivity in the form of scratches, forei gn contaminants, and particle residue should be suppressed. Traditional ILD consumables ar e not sufficient for the STI CMP process [Hos97, Sch99]. New slurry and pad chemistrie s with tunable propert ies are required to meet the stringent planarity, removal rate, and surface finish requirements of the STI process. For this application, ceria-based sl urries outperform silicabased slurries. Ceria slurries preferentially remove the oxide la yer with minimal polishing of the underlying nitride. To date, the ceria-silica polishing m echanism is unclear. Currently, the ceria CMP literature is applications oriented. In th is study, ceria-silica polishing is studied to understand the fundamental interactions that lead to material removal. With this knowledge, the STI CMP process can be more effectively controlled. Ceria-Silica CMP Mechanisms The ceria based CMP literature is analogous to that of the silica based literature approximately 5 years ago. Rigorous fundamental investigations in the manner of silicasilica CMP experiments have not been conducte d. Despite the relative immaturity of the ceria-silica CMP process, three competing m echanisms exist. The first two, ‘chemical tooth’ and a surface defect model, are base d on results from glass polishing. The third, specifically for ceria-silica CMP, is more m echanical in nature. None of the models conclusively explain the particle-surface inte ractions during CMP that lead to surface removal.

PAGE 47

33 Chemical Tooth Why do ceria abrasives polish silica so much faster than silica-based slurries? This is the fundamental question in ceria-silica CMP. For STI a pplications, understanding the origins of preferential polishing of oxide to nitride surfaces is also important. The chemical tooth model is the most widely accepted explanation of the ceria-silica polishing mechanism. Much has been made of the enhanced chemical reactivity of ceria abrasives, hence the name chemical tooth. Howe ver, little experimental validation of this concept has been presented. According to the chemical tooth model, the material removal rate is controlled by the re lative rates of [Coo90]: 1. water penetration into the surface layer 2. surface dissolution under particle load 3. adsorption of solute onto the abrasive particle surface 4. re-deposition of solute products onto surface 5. surface dissolution between particle impacts SiO2CeO2TiO2ZnO Al2O3IsoelectricpHPolish rate 2 6 10 SiO2CeO2TiO2ZnO Al2O3IsoelectricpHPolish rate 2 6 10 Figure 2-10 Glass polish rate versus IEP of common abrasive compounds [Coo90].

PAGE 48

34 Ceria slurries, by far, are the most effi cient in removing the glass or pure silica surface (figure 2-10). In fact, Cook estimated that one SiO2 molecule is removed for every twenty-four collisions be tween the ceria abrasive and s ilica substrate, compared to five hundred million collisions for silica abrasives [Coo90]. Among the properties of abrasive particles (Gibbs free energy of formation, bond dissociation energy, metal– oxygen bond strength, aqueous solubility product, Z/Rionic, Z/Reff, pK1, and isoelectric pH), only those related to surface charge, i.e. dissociation constant (pK), effective ionic field strength (Z–cationic charge/Reff–separation distance between the cation and the adsorbed proton), and charge density, whic h is related to the coordination number correlated with polishing ra tes [Coo90]. Although a direct co rrelation was not found, the product of the log of the single oxide bond st rength and the difference between IEP and slurry pH offers some predictive capability on polishing activity fo r prospective polishing compounds [Coo90]. The chemical tooth model depicts CMP as a molecular scale process in which material removal is achieved by displacing silica tetrahedra from the surface into the abrasive slurry. Net removal occurs if th e solute products are bound by the abrasive or swept away by the turbulent motion of the fl owing slurry. Otherwise, re-deposition may occur. Due to particle impact at the sili ca surface, water penetrat es the silica network, equation 2.4 proceeds in the forward di rection–creating solu te species Si(OH)4 (silicic acid), which is released into the solution, and hydroxylated surface species are created with one, two, or three remaining linkages to the silica network. This process is repeated as each abrasive particle trav erses the pad-wafer interface. In the case of glass polishing and STI CMP, polishing is done near the isoele ctric pH (point of zero charge) of ceria,

PAGE 49

35 which is ~ 6-8. According to Cook, the extent of particle-surface bonding is controlled by a hydrolysis reaction of the form: Si – O+ Ce – OH Ce – O – Si + OH(2.8). Silica, with an IEP ~ 2-3, remains negatively charged across the pH range. Because of its median IEP, ceria is amphoteric–able to adsorb positive and negative charges. As the particle moves away from the abrasive si te, the Ce–O–Si bond, equation 2.8, is strained and finally ruptured. If the st rength of the Ce–O > Si–O, s ilica accretion at the particle surface occurs and vice versa. This bond sc ission and release of surface species is expected to happen only when one Si–O bond remains of th e tetrahedrally coordinated silica [Coo90]. The silicic acid molecule ma y re-bond to the glass surface, form a complex with active sites on the abrasive part icle via hydroxyl exchange, or form colloids in the solution [Coo90, Oss02]. If particle-surface bonding ind eed occurs, residual Ce–O bonding should be present on the silica surface, an important point reserved for further discussion. The chemical tooth model is based on interactions betw een surface functional groups at the particle and silica surface. Ther efore, the nature of the abrasive-silica interactions should be strongly dependent on pH. Osseo-Asare has examined the relationship between the silica removal rate and the IEP of the abrasive material [Oss02]. In th is work, silica removal is achieved by surface complexation of the abrasive by silicic acid. O sseo-Asare stipulated that the slurry pH should be greater than the IEP of the sili ca surface and near the IEP of the abrasive particle. Furthermore, the IEP that demonstrates maximum silicate adsorption corresponds with the IEP for maximum polishi ng (ceria). The central feature of OsseoAsare’s theoretical formulation is th e exchange of hydrogen and hydroxyl ions

PAGE 50

36 (protonation and de-protonation) to form neut ral hydroxyl groups at the abrasive surface. In the case of an abrasive particle with IEP < slurry pH (M-OH M-O+ H+), there is a lower concentration of neutral sites. Conversely, when the abra sive particle IEP > slurry pH (M-OH + H+ M-OH2 +), there is also a lower concen tration of neutral sites and the main reaction is the transfer of protons to the silica surface [Oss02]. Therefore, the most favorable conditions for maximum silicate removal by ceria particles occur at neutral pH, which is near the IEP of th e ceria abrasives [Oss02]. Surface Defect Nearly 50 years ago, a general model fo r glass polishing was proposed based on lattice defects (ionic vacancies) at the ab rasive surface [Kal56a, Kal56b, Kal98]. Despite its longstanding presence in the literature, the defect-based glass polishing mechanism has been virtually disregarded in the literature [Kal98]. The centr al features of this theory are: 1. The abrasive should be softer than the surface to be polished 2. Minute bits of the abrasive are rubbed off during polishing 3. The abrasive hardness should exceed some minimum value to ensure that the particle is not pulverized during polishing. According to Kaller, the abrasive partic les are bound to the glass surface via a tribochemical reaction and movement of the polishing disk results in breakage of the solid bonds [Kal91]. Glass removal is expected to be a molecular-scale process. And similar to the chemical tooth theory, the removed glass bonds to the abrasive particles at high-energy defect sites [Kal91, Kal98]. At the time this th eory was proposed, there was little opportunity for experimental validation because of the l ack of sufficiently sensitive analytical techniques. Forty years later, Ka ller presented milling and BET surface area

PAGE 51

37 data as a function of time to prove the fractioni zation or creation of pa rticle debris during the polishing process [Kal98]. In the surf ace-defect theory, intrinsic defects are considered to have a minimal effect on polishi ng. Defects or distortio ns may be created in a variety of ways: 1) reduction/oxidation reactions, i.e., Ce4+ to Ce3+, incorporation of foreign metal ions or molecules (i .e., alloying of iron oxide with Ni2+), 3) rapid quenching from high temperatur e during synthesis, and 4) freezing in of thermal vibrations after long dwell times at high temp erature [Kal98]. These defects are expected to be responsible for the gripping or bonding of the abrasive part icle to the surface. Furthermore, the defect density and hence the particle reactivity are expected to increase as the particle is abraded due to the exposure of freshl y created, more reactive sites [Kal91, Kal98]. Particulate Generation The first two ceria-silica mechanisms are general models for all metal oxide abrasive particles for glass polishing. The accel erated removal rates for ceria with respect to silica have been attributed to its me dian IEP [Cook90, Oss02] or multiple valence states [Kal91, Kal98]. The final model is base d on more mechanical interactions and is proposed only for ceria-silica pol ishing as an alternative to the chemical tooth theory [Hos01]. In this model, silica particulates are cleaved from the surface during wear and bound by the abrasive particle. In support of this assertion, the particle size and composition of the post-CMP slurry was anal yzed. Hoshino observed an increase in particle size from the original 80 nm to 600 nm lumps of a white paste-like substance. Inductively coupled plasma atomic emi ssion spectroscopy, scanning electron microscopy, infrared spectroscopy, and x-ray microana lysis were conducted. According to their analysis, the large, irregularly shaped, white paste was SiO2 cleaved from the surface. In

PAGE 52

38 their opinion, if the chemical tooth mechanism was operative then silicic acid should be isolated in the post-CMP slurry. Since lumps of SiO2 were present in the slurry they assumed they must have been directly cleaved from the surface. The interpretation of Hoshino et al. is complicated by the presence of an organic surfactant, the slurry pH pH 8, and the dynami cs of colloidal silica stability in this pH regime. The presence of the surfactant likely complicates the analysis of the ceria-silica polishing mechanism, since surfactants have been demonstrated to change the characteristics of contact at the pad-partic les-wafer interface [Bas02a, Sin02]. Also, at ~ pH 5 colloidal silica rapidly a ggregates, with particle growth ensuing at ~ pH 7 [Ile79]. Recalling the formulation of Cook, in the post-CMP dispersion; the solute species, Si(OH)4, is free to form a complex with the ab rasive particle, which could change the shape of the original ceria pa rticle or coalesce and form colloids, which may explain the growth of the silica particle. To this point, three models with diverging mechanisms have been proposed. However, each has a central feature ceria-si lica bonding that is believed to occur upon particle-wafer impact. Therefor e, the surface chemistry of ceria should be considered in all proposed mechanisms for ceria-silica CMP. Ceria Surface Chemistry Cerium, atomic number 58, is one of the rare earth elements in the lanthanide series, which are characterized by the filling of 4f-orbitals [Sam73]. The chemistry of all lanthanides is similar. With the exception of cerium, the lant hanides have a valence of 3. However, cerium may also have a valence of 4, forming two oxides–Ce2O3 (cerrous oxide) and CeO2 (ceric oxide). Ceria is an extremel y efficient polisher of glass. A fact that has been ascribed to enhanced chemical reactivity between ceria and silica even in

PAGE 53

39 the absence of chemical additives to the liqui d portion of the slurr y. Ceria has a fluorite crystal structure, with the oxygen atoms coordi nated to four cerium atoms and the cerium atoms coordinated to eight oxygen atoms. The cerium atoms assume face-centered cubic (FCC) crystal symmetry with oxygen inserted in to the eight tetrahedral spaces within the FCC unit cell (figure 2-11) [Eva04]. Figure 2-11 Ceria crystal struct ure normal view (left), di splaced by one-half a lattice parameter (right) [Eva04]. The size of abrasive partic les has continually decrease d as more specialized CMP processes are introduced. Because of the high surface area of smaller particles, surface chemical effects are important for all CMP app lications. This is especially true for such reactive abrasives as ceria. The transition to nanometer scale pa rticles from the submicron regime has profound implications on th e CMP process from slurry formulation, handling, and transport, to actual polishi ng, and finally post-CMP cleans. A thorough understanding of the abrasive surface chem istry allows tailoring of the surfaces for enhanced dispersion before use, reactivity a nd desired polishing during use, and efficient removal after use. Stable micron-sized dispersions may be obtained by simple manipulation of the pH. However, in the sub-micron to nanometer scales, dispersion agents like surfactants, polymers, or excess ions (electrolytes) are required [Nab93]. Nanosized particles, in particular, cannot be kept apart from each other through

PAGE 54

40 electrostatic repulsions, because the inte raction potentials are less than 15kBT (Boltzmann’s constant temp erature) [Nab93, Ver48], ~8KBT for ceria [Ver48]. Therefore, in water nanoparticles will experience multiple collisions, which should result in aggregation due to particle-particle bondi ng [Nab93]. This is especially true for ceria, which demonstrates three states of stability based on the chemical treatment sequence [Nab93]. The loss of stability, i.e. agglomeration, has negative impacts on the particle’s polishing capacity be cause of the reduction in surf ace area and loss of active sites. Nabavi et al. have investigated the su rface chemistry of ceria [Nab93]. Applying Pauling’s definition of formal valence charge, the multi-site complexation or ‘MUSIC’ model was used to determine the acidbase behavior of surface hydroxyls [Nab93, Hie89]. In this treatment, atoms in an ionic cr ystal give each of its neighbors a number of electrons, v = Z/CN where z is the atomic num ber of the atom and CN equals the number of nearest neighbors. For n metals, a charge of Z = nv results. Its nearest neighbors in a bulk crystal compensate this charge. A hypothetical free surface may be created by splitting the crystal along a plane of oxygen atoms, creating dangling bonds. Surface passivation is achieved by adso rption of aqueous species (H+, OH-)–general protonation or de-protonation equilibria. Through such reac tions an electric char ge is acquired at the particle-liquid interface. Additional complications arise when additi onal ions are present. This process is driven by the surface energy of the oxide in water. Ion adsorption may be hindered by steric repulsion or steric poi soning, which lowers the surf ace energy by binding of ions from the solution. In addition to chemical reactions at the surface, charge may be

PAGE 55

41 acquired by isomorphous substiti on within the lattice (i.e., Si4+ replaced by Al3+) and adsorption of a hydrophobic species or surfactant ions [Stu92]. Thus the electrical state of a surface depends on the spatial distribution of electronic or ionic charges in the nearsurface region. Using the Gouy-Chapman model this concept is idealized as the electrical double layer. In this model, one layer is de picted as a region of fixed charge rigidly bound to the surface while the other layer is dist ributed more diffusively in the liquid in contact [Stu92]. The potential drop across the mobile portion of th e double layer (figure 2-12) zeta potential ( ), is responsible for the electrok inetic phenomena, particle motion in the presence of an electric field, whic h may be related to dispersion stability. Distance (x)Potential ( ) Shear plane Distance (x)Potential ( ) Shear plane Figure 2-12 Potential versus dist ance for a charged surface [Stu73]. The surface chemistry of ceria and its effect on dispersion stability is particularly intriguing. In the work of Naba vi et al., nanosized ceria part icles were obtained through precipitation of Ce4+(NO3 -)4 salt of the form CeO2(HNO3)0.5(H2O)4 at very low pH

PAGE 56

42 [Nab93]. Simply raising the pH caused irrevers ible aggregation and transformation of the particle surfaces well below the predicted po int of zero charge (pzc). Three dispersion states were obtained based on the affinity for proton (H+) adsorption and also binding and release of nitrates present from the nitric acid based precursor solution. The acid-base equilibria at the ceria partic le surfaces were investigated through titration of the surface groups with a base (NaOH) or acid (HClO4). The pH was measured and he concentration of free nitrate ions (NO3 -) at each step of the tit ration was also measured. In Nabavi’s work, the behavior of the di spersion was related to the difference in hydroxyl and proton consumption in dispersions of acidic and basic pH and the presence or lack of nitrates as counter-ions or covalently bound surface species [Nab93]. According to their results, tit rating the dispersion above pH 5.0 results in a loss of active sites that is reflected in an excess amount of hydroxyl consumption with respect to proton consumption in the reverse cycle. The transiti on between different su rface states occurred in narrow pH ranges, generating a hysteresis-t ype curve for pH versus volume of acid or base used (figure 2-13). The horizontal dist ance between the curves corresponds to an excess consumption of hydroxyl during the first part of cycle [Nab03] According to their results, these phenomena may be explained as follows. The or iginal dispersion contains a high concentration of nitrate ions as either counter-ions or strongly bound surface species. The counter-ions are released becau se charge is cancelled by a progressive release of protons and adsorp tion of hydroxyls. Covalently bound nitrates were released up to pH 6.0 in an exchange reaction of the form: Ce-O-NO2 + OHCe-OH + NO3 (2.9).

PAGE 57

43 Particle aggregation ensues, which results in a loss of surface. The remaining surface sites determine the electrical ch arge characteristics of the di spersion. This is reflected by a shift in the pzc from the theoretically predicted value of 7.9 to 10.0. Particle aggregation results in flo cculation of the dispersion. The dispersion stability of nanoparticles ceria is thus related to the pr esence of strongly bound ionic surface species, which influences the ability to adsorb aque ous species from the liquid environment. V[Base/Acid] (ml)pH 0.5 1.52.5 3.5 3 5 7 9 V[Base/Acid] (ml)pH 0.5 1.52.5 3.5 3 5 7 9 V[Base/Acid] (ml)pH 0.5 1.52.5 3.5 3 5 7 9 Figure 2-13 pH versus acid/base concentrati on. Horizontal distance between curves is related to excess consumption of OHduring acid to base cycle [Nab93]. These observations have wide-ranging imp lications for the va rious technological applications of ceria particles. For CMP in pa rticular, irreversible changes in the particle dispersability, reactivity, and su rface area should be reflected in the polishing behavior as a function of pH. Three major questions form the basis of this research. 4. Why does ceria polish so much faster th an other abrasives especially silica? 5. Why does ceria preferentially polis h oxides over other materials?

PAGE 58

44 6. How can the polishing characteristics of cer ia slurries be controlled for optimum STI polishing? There are major gaps in the literature c oncerning ceria abrasives for silica CMP. As previously mentioned, too few studies have focused on the fundamental interactions between the ceria particle surf ace and the surface of the sili ca substrate. Most studies complicate mechanistic studies by including su rfactants into the ceria-based slurries. However, recent publications indicate the fo cus is shifting towards more fundamental investigations. The surface structure of indi vidual ceria particles has been studied by electron energy loss spectroscopy (EELS) [Gil 04]. These measurements have confirmed the presence of Ce3+ at the surface of bulk CeO2 particles. In this dissertation, ceria-silica interactions will be investigated, specifi cally surface interactions. To confirm the existence or absence of chemical reactions between the ceria particles and silica substrates extensive surface analysis has al so been conducted. The major application of this research is in STI CMP. Therefore, general strategies for STI CMP are also proposed.

PAGE 59

45 CHAPTER 3 MATERIALS AND METHODS Introduction This research involves polishing experiments using ceria-based slurries and a silica film. In-depth analysis of the effect of pH on ceria surface chemistry, silica removal rate, ceria-silica contact characteristics, ceria particle morphology, and silica surface morphology is discussed. The in-situ friction force is reported us ing a novel apparatus that was fitted to our tabletop polisher. Slurry stability, part icle size, and zeta potential measurements are reported using a variety of light scattering instruments. Surface analysis using x-ray photoelectron spectro scopy, scanning electron microscopy, atomic force microscopy, and variable angle spectroscopic ellipsomet ry results are presented. Samples Silica Wafers Silica (SiO2) samples for all experiments had a nominal thickness of 2 m. Plasma enhanced chemical vapor deposition (PECVD) was used to the deposit the oxide layers on 1 mm thick, p-type silicon substrates with (001) orientation. The silica wafers were obtained from Silicon Quest International. Ceria Slurries Ceria particles were obtained from Sams ung/Corning. The slurries were dispersed in DI water at pH 4.5. KOH or HCL were used to adjust the pH to the desired value. The concentration of the particles in suspension was between 2 and 3wt%.

PAGE 60

46 Ceria Thin Films Ceria thin films were grown on Si substr ates using pulsed laser deposition (PLD). The PLD ceria thin films were used to m easure the force measurements with a 7.3 m silica particle. The depos ition was conducted at 7000 C in 100 mtorr of O2. A KrF laser of wavelength 248 nm was used. The laser energy was 500 mJ and the pulse rate was 5 Hz. During PLD, a pulsed laser stri kes a solid target. Target materials are removed in the form of a plume. Substrates of choice may be coated by exposure to the plume, which is created by laser bombardment. The major advant ages of PLD include; film stoichiometry close to that of the target, low contamin ation levels in deposited films, and high deposition rates. Functionalized Cantilevers Cantilever oxidation Silicon tapping mode atomic force mi croscopy (AFM) cantilevers were functionalized by thermal oxidation at 10000 C for 2 hours in air. The cantilevers were furnace cooled. After oxidation, the cantilevers were used to abrade the ceria thin films at different pH values. The wear profile and cantilever composition was measured before and after the experiment with a scanning electron microscope. Particle attachment AFM cantilevers can be further functionaliz ed by attaching a particle for force measurements. A micromanipulator fitted onto an optical microscope was used to attach a 7.3 m silica to a silicon nitride contact mode AFM tip. The spring constant of the cantilever was 0.13.

PAGE 61

47 CMP Equipment CMP Tabletop Polishers Polishing experiments were performed on a Struers Rotopol 31 and TegraPol 35 tabletop polishers. The TegraPol 35 polisher allows control of platen velocity and pressure down to 2 psi. The ceria po lishing slurry is delivered by a Struers Multidoser unit that is connected to the polisher. The head and platen speed were 150 rpm and the center of the sample to be polished was placed at a distance of 8.9 cm from the center of the table. This setup leads to a linear velocity of 137.44cm/s. Perforated IC 1000/Suba IV (figure 3-1) stacked pads by Rodel Inc. and Rohm and Haas were us ed for all polishing tests. During polishing, the slurry is con tinuously stirred by a magnetic bar, and pumped to the pad-wafer interface at a flow rate of 100 mL/min by the Multidoser unit. The pad was conditioned before each polishing r un with a grid-abrade diamond pad conditioner manufactured by TBW. The diamond conditioner minimi zes pad glazing (figure 3-1d) and ensures reproducible pad conditions for each polish run. The sample holder for the polishing experiments is a stainless steel cylinder 2.25 inches in diameter and 1.125 inches in height. The flat surface of the cy linder has a recessed, machined area of 1 inch2 and 0.03 inches deep. The recessed area ensures the sample holder holds the wafer during polishing experiments. A porous carrier film s upplied by Rodel slightly raises the wafer

PAGE 62

48 Ra: 7.60 m Rq: 11.97 m Rmax: 57.34 m (b) (a) (c) Ra: 7.60 m Rq: 11.97 m Rmax: 57.34 m Ra: 7.60 m Rq: 11.97 m Rmax: 57.34 m (b) (a) (c) (a) (b) (c) (a) Ra: 7.60 m Rq: 11.97 m Rmax: 57.34 m (b) (a) (c) Ra: 7.60 m Rq: 11.97 m Rmax: 57.34 m Ra: 7.60 m Rq: 11.97 m Rmax: 57.34 m (b) (a) (c) (a) (b) (c) (a) Figure 3-1 SEM analysis of CMP pad: a) cross-section of IC 1000/Suba IV stacked polishing pad (40 ) [Obe98] b) Plane view imag e with profilometer scan showing pad roughness [Cho03] c) image of fresh pad [Ste97] d) image of used pad [Ste97]. over the recessed area. Additi onally, when wet, the film ensures the wafer remains stationary during the experiment. Friction Force Measurement Apparatus The friction force studies were conducted using the tabletop Struers Rotopol 31 polisher. An apparatus, developed by us, was f itted to the polisher (figure 3-2). Glue was used to mount the silica samples (0.8 in2) on a rectangular alum inum block. Circular washers were then loaded onto the block to achieve the desired dow n force. The entire assembly is then placed in a restraining fixt ure that ensures movement occurs only in one direction. During the experiments, the lateral force applied by the aluminum block on the restraining fixture is measured by a force tr ansducer (Sensotec model 31 load cell). The load cell is connected to a da ta acquisition system. A data poi nt was taken every 250 ms.

PAGE 63

49 PLAN VIEW PROFILE Restraining Fixture Down Force Polishing Pad Force Transducer Force Transducer Wafer Wafer Sample Holder PLAN VIEW PROFILE Restraining Fixture Down Force Polishing Pad Force Transducer Force Transducer Wafer Wafer Sample Holder PLAN VIEW PLAN VIEW PROFILE Restraining Fixture Down Force Polishing Pad Force Transducer Force Transducer Wafer Wafer Sample Holder PLAN VIEW PROFILE Restraining Fixture Down Force Polishing Pad Force Transducer Force Transducer Wafer Wafer Sample Holder PLAN VIEW Figure 3-2 Schematic of in-situ friction force apparatus. Particle Characterization The particle size and surface charge ar e two properties that have a profound influence on the slurry stability and result ant polishing characteristics. Multiple light scattering techniques were used to measure th e particle size and zeta potential, and to determine the stability of the ceria slurries. Zeta Potential Measurement Almost all solids acquire a surface charge in polar liquids like water. The various ways of charge accumulation have been covered in Chapter 2. The surface charge or surface charge density plays a crucial role in sl urry stability. This is especially true for nano-sized particles. Because of dynamic motion and continuous particle-particle collisions, attractive forces are always present. These forces originate from the momentary, induced dipoles that arise when particles approach each other [Zeta]. London, van der Waals, and/or dispersion forces are common terms for these short-range attractive forces. To maintain slurry stability, electrostatic and/or st eric repulsive forces

PAGE 64

50 can be applied. Because of their surface charge particles move with an average velocity (V) in an electric field (E), whic h is referred to as electrophoresis. A Zeta potential analyzer manufactured by Brookhaven Instruments was used for zeta potential measurements as a function of ceria slurry pH. There is not an instrument that is capable of directly measuring the zeta potential. Instead, the velocity of the moving particles is measured. The electrophoretic mobility (e) is then calculated as a function of velocity (V) a nd electric field (E). e = V/E (3.1) The particle, adsorbed ions, counter-ions, so lvent molecules bound to ions on particle and/or counterions, and surfactan ts or polymers if any exist, may be considered a single unit. The velocity is measured experimentally as the charged particle is drawn to an oppositely charged electrode. The zeta potential is then calculated from the mobility. The concept of surface charge density leads to a simplified model referred to as the electric double layer. In this approach, th e particle surface is populated by counter-ions with opposite charge of the particle and co-ions that possess the same charge as that on the particle surface. Thermally driven coll isions between particles and the solvent molecules cause continuous agita tion of the adsorbed ions. This results in an electrostatic potential ( ) that decays exponentially for planes separated from the particle surface (figure 3-2). = oex (3.2) The electric double layer thickness ( -1), which is on the order of nanometers, determines the distance from the partic le surface over which various electrical potentials are significant [Stu73]. This thickness is a function of temperature, fluid dielectric constant,

PAGE 65

51 and the ionic strength or free ion concentration. For high ionic strength conditions, the electrical double layer is compressed and th e particle-particle interactions are more attractive. Thus the zeta potential is the potentia l between the bulk of the liquid and the surface of shear or shear plane that extends out from the surface of the particle. This potential is a direct measure of the re pulsive force between two charged bodies. However, the zeta potential and mobility depends on the model chosen. For a particle of radius a, the Huckel limit is applied when a << 1. It is practically impossible to satisfy this limit for most particle systems. Theref ore, the zeta potential analyzer utilizes the Smoluchowski limit for a >> 1, where zeta potential ( ) is proportional to the product of the fluid viscosity ( ) and the mobility divided by the dielectric constant ( ) of the fluid [Hun01]. = e/ (3.3) Particles move in a direction perpendicular to the focused laser of wavelength 670 nm (red). The Doppler shift due to the scattering of the inci dent beam is determined and mobility and zeta potential are calculated as described above. Particle Size Analysis A Microtrac UPA 150 particle size analyzer manufactured by Honeywell was used to determine the average particle diameter as a function of pH. Particles were sonicated in solutions of pH 2.5-10.5. The background solution used in the particle size analyzer was of the same pH as the ceria slurry. This in strument also utilizes dynamic light scattering for measurement purposes. Again, the veloc ity and particle motion direction are randomized by collisions with solvent molecu les. The velocity di stribution of a large number of mono-sized particles will a pproach a well known functional form. After

PAGE 66

52 compensating for thermal and viscosity eff ects, the velocity di stribution becomes a unique function of the particle diameter. The scattered light by moving particles is proportional to the particle velocity. A dispersion of particles will have a unique distribution of frequency shifts. Spectra from small particles have more high frequency comp onents than spectra from large particles. Care must be taken to minimize thermal and fluid viscosity contribu tions. Both increased temperatures and decreased fluid viscositie s will increase the high frequency components of the spectrum causing an underestimation of th e particle size. The particle distribution is computed directly from the measured frequency spectrum obtained from the Doppler shifted scattered light. Slurry Stability Slurry stability is an anomalous property. CMP slurries must not only be stable at the point of use, but also must maintain st ability during shipping and while on the shelf. In this work, a Hach model 2100AN laboratory turbidimeter is used to determine the stability of the slurries at di fferent pH values. This instrument measures turbidity from 010,000 Nephelometric Turbidity Units (NTU). A tungsten filament lamp is used to provide the incident beam. Lenses and apertu res focus the light. Four detectors are used: 90 degree, forward-scatter, transmitted, and b ack-scattered light dete ctors. In the Ratio On mode, mathematical calculations are used to average signals from each detector for improved linearity, calibration stability, and wider measur ing range, and to perform measurements of colored solutions. A nephelometer is an instrument that measures the amount of cloudiness in a dispersion. The intensity of sca ttered light changes as the particles settle out of solution.

PAGE 67

53 High NTU values correspond to di spersed or cloudy so lutions. Measurements with this instrument are presented as normalized NTU. Characterization of Silica Surface Morphology Atomic Force Microscopy (AFM) Roughness is a surface property that aff ects device performance. Even well polished surfaces have topographical variations that can be considered surface roughness. Atomic force microscopy (AFM), a hybrid of scanning force microscopy allows imaging of surface topographical variations with subnanometer resolution [Ste92]. In this study, the topography after CMP was compared to as received samples using AFM. AFM, although time consuming, is a robus t technique that allows ch aracterization of insulating and conducting surfaces in liquids or in air. AFM uses a sharp tip mounted on a flexib le cantilever. Repulsive van der Waals forces between atoms on the tip and substrat e cause deflection of the cantilever. The force between the tip and sample are a func tion of the separation distance. Therefore, surface topography is obtained by monitoring the cantilever deflection. Raster scanning during AFM is controlled by a piezoelectric tube, which maintains a constant separation between the tip and sample. Cantilever defl ection due to surface topography is measured using light from a laser diode that reflects fr om the back of the can tilever to a positionsensitive photodiode (figure 3-3). The AF M image is generated by monitoring the position of the scanner.

PAGE 68

54 Figure 3-3 Schematic diagram of AFM apparatus. In this study, a Digital Instruments Na noscope III AFM apparatus was used. The measurements were conducted in contact mode. Contact mode tips have four Si3N4 cantilevers with spring cons tants that vary from 0.06 to 0.58 N/m. For topographical imaging, 10 m scan sizes were used to qua ntify the extent of scratching and surface damage after CMP. The software bundled w ith the AFM calculates Root Mean Square (RMS), average (Ra), and peak-to-valley (Rpv) roughness values. The RMS roughness is used in this study. RMS = ( (Yi 2/N))1/2 (3.4) Ra = Yi/N (3.5) Rpv = Rmax – Rmin (3.6) In all three equations, Y is the height va lue and N corresponds to the total number of measurements. For all scans, a plane fit and flatten data filter were applied to the AFM

PAGE 69

55 images using the AFM software. With these filters, scanning arti facts can be removed and the image quality is improved for display purposes. Force measurements are also possible usi ng the AFM. They may be conducted with the standard AFM tip or with cantilevers that have been functionalized by attaching a micron-sized particle. In this study, a 7.3 m di ameter silica particle was used to measure the interaction force as a function of separa tion distance at various pH values using a ceria thin film. Further functionalization is possible using specialty coatings. To further study the ceria-silica mechanisms in the absence of particles, chemical vapor deposition was used to coat silicon tapping mode cantile vers with a 100 nm th ick silica layer. Tip wear against a ceria surface was compared at different pH values. Scanning Electron Microscopy Scanning electron microscopy (SEM) provi des a highly magnified image of the surface of a material that is similar to what one would expect if one could actually see the surface. SEM provides both topographical a nd elemental composition. Resolution down to a few nanometers is possible with magnification of ~300,000X. During SEM analysis, electrons are fl ooded onto the specimen surface with a variety of output signals. El ectrons with energies of ~ 50 eV are called secondary electrons [Bin92]. Secondary electrons result from inelastic scattering between the primary electron beam and atomic electr ons of the specimen [Bin92]. The surface sensitivity of SEM is due to the shallow es cape depth of secondary electrons. Inelastic scattering in the specimen bulk results in el ectrons with insufficient energy to escape from depths greater than a few nanomet ers. Backscattered electrons, including characteristic Auger electrons, have energies in excess of 50 eV [Bin92]. Backscattered electrons result from elastic scattering of th e primary electron beam. A further utility of

PAGE 70

56 the backscattered electron signal is co mposition mapping. The backscattered electron intensity increases with atomic number, yielding brighter images [Bin92]. Characteristic x-rays are also emitted as an excited atom decays to its ground state. The x-ray emission single can be sorted by energy or wavelengt h in an energy disper sive or wavelength dispersive x-ray detector. The distribution of en ergies or wavelengths is characteristic of the elements that produce them. There are three types of emission capable of producing beams for SEM analysis: thermionic, LaB6, and field emission. Field emission provides the brightest emitted beam with the sharpest electron energies, which limits chromatic aberrations of magnetic defocusing lens [Bin92]. Field emission SEMs of fer enhanced resolution at lower applied voltages. In this study, a SEM equipped with a field emission gun (JEOL JSM 6335F) will be utilized along with the energy disper sive x-ray detector (EDX) for compositional analysis of the polished surfaces. SEM analysis can be limited by the samp le properties. To minimize charging, insulating samples must be imaged at vo ltages below 2 keV or a thin (< 10 nm) conductive coating must be applied. Also, the specimen should be stab le at high vacuum and be of appropriate size for the sample holder. X-ray Photoelectron Spectroscopy X-ray photoelectron spectroscopy (XPS)is based on the photoelectric effect explained by Einstein in 1905 [Bru92]. Highenergy x-rays ionize an atom producing free photoelectrons. KE = h – BE (3.7) Since the energy of the incoming photons is known, by measuring the kinetic energy (KE) of the ejected electron, the binding energy (B E) of the electron to the

PAGE 71

57 particular atom can be determined. The bindi ng energy of the electron is sensitive to the chemical state of the atom. Therefore, X PS is a powerful surface analysis tool that provides semi-quantitative elemental composition information. Since XPS peaks are due to the ionization of core level electrons, they can be used to detect all elements except hydrogen and helium. Liquids, gases, and solid s can be analyzed. Generally, XPS spectra give information from the topmost atomic layers (2-20) [Bru92]. The number of peaks in an XPS spectrum corresponds to the number of occupied energy levels in the ionized atoms whose binding energy is lower than the energy (h ) of the incident radiation. The peak position de termines the respectiv e binding energies of the electrons in subatomic or bitals and provides a means of identifying atoms present in the specimen. Additional features in the spectrum are due to the involvement of passive electrons (non-photoele ctrons) [Bru92]. Unlike valence electrons, core level el ectrons do not partic ipate in chemical bonding. Therefore, the elemental binding ener gies have characteristic values that provide unique signatu res of elements. Quantitative analysis (at%) of the relative amounts of elements present is possible by comp aring the relative peak intensities. This requires knowledge of the photoionization cro ss section for each element and control of experimental artifacts that affect peak in tensities [Bru92]. XPS provides only semiquantitative elemental composition information b ecause of: variations in peak widths and peak line shapes, inco rrect background subtraction, and in strumental artifacts. Coupled, these factors introduce an uncertainty in the measurement that may be as high as 30%. However, for well characterized XPS peaks with well known standa rds the uncertainty may be reduced to only a few at. %.

PAGE 72

58 In addition to quantitative analysis, th e chemical environment of the ejected photoelectrons may be determined as well. Slight shifts in the binding energy of XPS peaks may be related to the respective chemical environment. In general, differences in the binding energy are seldom more than a fe w electron volts. Other features of the XPS spectrum may also be utilized to identify elem ents in the specimen. Shake-up satellites are created because valence electrons may be excited to empty orbital levels during relaxation. This results in photoe lectrons of diminished ener gy, hence, a shift to higher binding energy. Since valence electrons pa rticipate in chemical bonding, additional chemical state information is obtained. Auge r electrons are also generated when an electron from a higher energy level fills the core hole left by the ejected photoelectron. By changing the radiation source, overlapp ing XPS and Auger peaks may be resolved. The primary use of XPS is for surface an alysis. Electrons travel short distances before losing energy in collisions with atoms of the specimen. Therefore, only near surface photoelectrons contribute to XPS p eaks. The background in a XPS spectrum may be attributed to electrons from the bulk that have lost energy due to inelastic collisions. Additional surface sensitivity is possible by utilizing a grazing incidence angle, which further minimizes the escape depth of elasti cally scattered electrons. The XPS system used in these experiments is a Perkin -Elmer PHI 5100 manufactured by Physical Electronics. Spectroscopic Ellipsometry (VASE) Spectroscopic ellipsometry is a surface se nsitive thin-film measurement technique that measures changes in the polarization st ate of light after refl ection or transmission from a surface of interest. Large amounts of data may be obtained using probe beams of various wavelengths and angles of inciden ce. A broad range of sample materials and

PAGE 73

59 structures are suitable for this technique. As with all optical tec hniques the property of interest (in this case film thickne ss) is not measured directly. Ellipsometers measure the change in the polarization state ( and ) of the light reflected from the specimen surface. The pa rameters may be related to the Fresnel reflection coefficients (Rp, Rs). = Rp/Rs = tan ( ) ei (3.8) Ellipsometry is a very precise and reproducible technique since it measures the ratio of two values. The sensitivity of the measurement is enhanced by the complex component of equation 3.6 that gives phase information. Initially, the ellipsometer measures the reflected or transmitted intensity or the polarization state of the re flected light. A model is de veloped based on the probe wavelength, polarization state, a ngle of incidence, thickness, and optical constants. Next, the model is fitted to the experimental data by varying physical parameters and calculating data that closely matches the meas ured optical data. The ellipsometer in our lab, manufactured by J. A. Woollam Co. Inc., uses the Levenberg-Marquardt multivariate regression algorithm. Finally, the best-fit model should be evaluated to ensure the fit is unique and is physically reasonable. This is achieved by monitoring the MSE value. For a unique, reasonable fit the MSE value tends towards some unique minimum value that approaches zero. Although ellipsometry is a robust techni que that requires very little sample preparation, measurements should be made with caution. The roughness of the measured spot of the specimen should be less than 10% of the beam’s wavelength. Also, the thickness of the film should vary no more than 10% over the width of the beam spot. The

PAGE 74

60 presence of adsorbed contamination layers may slightly affect the measurement. Angstrom level measurements should also be made with care since the optical constants of thin films rarely are identical to bulk values.

PAGE 75

61 CHAPTER 4 THE EFFECT OF PRESSURE, PH, AND VE LOCITY ON THE SILICA REMOVAL RATE AND SURFACE FINISH: A STATIS TICAL DESIGN OF EXPERIMENTS Introduction Chemical Mechanical Polishing (CMP) be gan as a proprietary process at IBM [Bey00]. Initially used for planarizati on of silicon, the CMP process has been successfully applied to all steps in integrated circuit manufacturing. In fact, CMP is an enabling technology for multilevel, interconnect ed structures that have allowed the continued shrinkage of microe lectronic devices [Ste97]. Becau se of its ability to produce globally planar surfaces, CMP is being implemented in a number of emerging technologies. Although CMP is simple to perf orm, designing an optimum process for a new set of materials is not a trivial tas k. Often, conventional CMP consumables (pads, slurries, equipment) are not applicable to the new application. Furt hermore, the use of CMP in new applications is complicated by the lack of fundamental understanding of general CMP processes that have been in us e for well over a decade. Hence, conventional CMP processes can not be strictly relied upon when developing strategies for novel CMP applications. Lacking a defined starting point for new CMP applications, fundamental investigations of the nature of those reviewed in Chapter 2 for silica-silica CMP are often necessary to develop an efficient, reliable CM P process. Singh et al have identified over forty variables microand nano-scale effects, and output parameters that control the CMP process [Sin02]. Common CMP experiments investigate the effects of only a few

PAGE 76

62 variables at a time by varying one parameter while all others remain fixed. After an optimum response is received at one level, a nother parameter is varied and so on until the desired polish rate, surface finish, planarity, and/or sele ctivity are achieved. A thorough investigation of the influen ce of each variable and its e ffect on the respective output parameters is impossible using the one vari able at a time experimental approach. Statistically designed experiments are an attractive alternative to testing one variable at a time. Statistical methods are crucial tools for industrial CMP practiti oners [Par04]. The competitiveness of the microelectronics fabr ication industry requires that resources be used efficiently, that processes are reproducible with minimal lot-to-lot variation, and that manufacturing delays be minimized to ensu re maximum profit. Statistically designed experiments based on sound scientific and engi neering principles allow companies like chip manufacturers to conti nually increase production at lowe r costs, resulting in higher profits. In laboratory testing, st atistical methods are usually ap plied only in data analysis (i.e. measurement average and standard de viation from measurement average). CMP research, with its many variables and inter-r elated interactions, is a prime area for implementation of statistically designed experiments. The basic principle of statistically designe d experiments is that all measurements are subject to variation [Mas89]. Variati ons in measurements may be because of experimenter bias, changes in the experimental ambient, or measurement error. Failure to minimize error can often result in the wrong in terpretation of data. For example, if an output parameter has a large vari ation about the measurement m ean it is not clear that the observations are significant, which results in the obscuring of the true effects of the

PAGE 77

63 variables in question. Statistical experimental designs are e ffective in eliminating bias, ensuring the measured output is gained from the experiment, and conservation of resources. Each observation generally provides information on all the factors of interest in the experiment. Conventional laboratory pr ocedures limit the numbe r of variables that may be studied. Also, since only one variable is changed at a time, the evaluation of the effect of interactions among experimental va riables is not possible. Statistical designs allow for the estimation of the magnitude of uncontrollable variation, and modeling of relationships between measurements of intere st and variables believed to influence these measurements [Mas89]. Selecting a Statistical Design Mason, Gunst, and Hess identified five sta tistical design criteria : 1) consideration of objectives, 2) factor effects [Appendix A], 3) precision, 4) efficiency and 5) randomization [Mas89]. While formulating the obj ectives of the experiment the nature of the anticipated conclusions, the manner in wh ich concepts are defined, and the manner in which observations are measured and reported should be considered. For factor effects, systematic errors should be eliminated, c ovariates should be measured [Appendix A], variable relationships should be identified, and th e entire experimental range should be explored to ensure a meaningful experiment al design is produced. Precise experiments include an estimation of inherent uncertainty associated with the e xperiment, replication of experimental runs, adjustment for c ovariates, and blocking [Appendix A]. Precise experimental designs also ensure that the deta il of the response and its relationship to the factors is clear. Efficient experiments are characterized by the examination of multiple factors with screening de signs. Finally, the experiments should be randomized.

PAGE 78

64 Randomization protects against unknown sources of bias like instrume ntal drift and helps validate assumptions required to apply certain statistical techniques [Mas89]. Statistically based experimenta tion is widely practiced in various fields. Basim et al investigated the effect of pa rticle size, applied load, and particle concen tration on the silica-silica CMP removal rate and surface fini sh using a central composite design with three factors at five leve ls [Bas03b]. Park used three designs: a nested design, a completely randomized block design, and a f actorial design to analyze gate polysilicon critical dimensions for integrated circuit de vices [Par03]. Martin et al used a fractional factorial design to optimize cryo-preservati on conditions for rat liv er slices and to evaluate the day-to-day variation in the le vels of cryo-preservation [Mar03]. Tye has reviewed the applicatio ns of statistical design of expe riments in drug discovery [Tye04]. Finally, Zaman et al. used the Box-Behnken met hod to investigate the solubility of metal ions dissolved from kaolin in the presen ce of different anionic dispersing agents [Zam03]. Role of Pressure, Velocity, and pH in Silica CMP Contact mechanics, wherein the polishing pr ocess is believed to occur because of wear due to direct wafer-part icle contact, are most commonly utilized to explain the CMP process. In fact, most mathematical CMP m odels are based on an empirical relationship (equation 4.1) initially proposed for glass polishing [Pre27]: RR = KpPV (4.1) where RR is the material removal rate, P is the applied pressure, V is the relative velocity between the wafer and the pad, and Kp is a process dependent c onstant [Coo90, Bas02a, Qin04]. Hence, the pressure and velocity are the most studied variables in CMP experiments. As the applied pressure is increased, the contact area at the pad-wafer

PAGE 79

65 interface increases. The increase in pad-wafer co ntact results in an increase in the number of abrasives available for polishing. Particle s not embedded in the polishing pad have a negligible effect on the material removal rate [Zha99]. According to Basim, a decrease in particle size or an increase in solids load ing results in a significant increase in the removal rate while an increase in the applie d pressure does not significantly impact the removal rate [Bas02a]. Due to the viscoelas tic response of the polis hing pad to applied loads, the actual contact pressure at th e particle-wafer interface is not changed significantly. Improvements in surface quality we re predicted for smaller particles, lower solids loadings, and lowe r applied pressures. The removal rate and surface finish are al so affected by the relative velocity between the pad and wafer. The pad-wafer c ontact area is influenced by the rotational velocity. Using in-situ friction force measur ements, Choi investig ated the effect of velocity on the pad-wafer contact area [Cho03] According to his results, at low solids loading (< 5 wt%) the pad-particle contact area is greater than the pa rticle-wafer contact area, which results in a decrease in friction force with increasing velocity. However, the increase of friction force with platen velo city when the solids loading exceeded 5 wt% was attributed to increased shear as the num ber of particles at the pad-wafer interface increased. Choi’s analysis did not consider the mechanical properties of the pad in his analysis. Generally, CMP pads are made of polyurethane or polyur ethane impregnated felts [Ste97]. Therefore, the response to an applie d load is time-dependent. For maximum padwafer contact ~1% of the pad asperities are ac tually in contact with the wafer [Bas03]. As the velocity is increased, the pad response time may affect the pad-wafer contact area.

PAGE 80

66 This may be noticeable in particle-free a nd low solids loading conditions. There may not be sufficient pad asperities in contact with the wafer to ensure sufficient abrasives are available for polishing. However, as the numbe r of particles at the pad-wafer interface increases, the viscoelasticity of the pad ma y have less of an effect on the pad-wafer contact area since sufficient particles may be present. The pH also has an effect on the silica removal rate and surface finish. However, in the pH regime of this study (2 -6), the pH effects on the silic a surface properties should be negligible. As reviewed in Ch apter 2, the pH has a significant effect on the ceria surface chemistry, which controls the stability and reac tivity of the abrasive particle. Above pH 6, rapid agglomeration ensues, which results in a reduction in the total ceria surface area. This results in a decrease in the density of reactive sites on the ceria particle surface, which lowers the chemical reactivity of the ab rasive particles. Gene rally, shallow trench isolation (STI) CMP is conducted below pH 7. Th erefore, this study is restricted to the pH regime in which the ceria slurry should be stable. Experimental My Design Expert software version 6.0.5 wa s used for statistical modeling of the effect of pressure, pH, and ve locity on the removal rate a nd surface finish during CMP of silica with ceria abrasives. To formulate an effective statisti cal design, the general outcome should be known. Therefore, an initia l study of the effect of pH was conducted at a fixed pressure and velocity. The Box Be hnken method was used for this three factor design (pressure, velocity, and pH). B ox Behnken designs are alternatives to 3k factorials (figure 4-1c, which are a combination of the central composite (figure 4-1a)and Box Behnken designs (figure 4-1b) [Mas89]. Box Be hnken designs for 3 factors are efficient, rotatable [Appendix A], and require fewer r uns than central composite or fractional

PAGE 81

67 factorial designs. For three factors, only fifteen experiments are required by the Box Behnken method compared to 17 for central composite designs and 27 for a 33 factorial design [Mas89]. In the Box Behnken design, all po ints may be envisioned as occupying a sphere. Extreme edges of the cube created by two level factorial designs are not considered [Box78]. The experimental design for this study, which uses the Box Behnken method is shown in Appendix B. (a) (b) (c) Figure 4-1 Factor space for different statisti cal designs: a) central composite b) BoxBehnken c) fractional factorial [Mas89]. Results and Discussion Removal Rate Response Scientific problems are usually expressed as relations of one or more variables (i.e. material removal rate = f(pressure, pH, velo city)). Using statistic al design methodology,

PAGE 82

68 this is accomplished by an empirically derived response function (Appendix A). `Generally, knowledge of the form of the response func tion (f) is obtained by fitting models to data obtained from the statistical design. Response functi ons are important for multiple reasons [Mas89]: 1. The function is characterized in the re gion of interest to the experimenter. 2. They are used to model the relations hip between factors and observations. 3. Gives concise summary of experiment. 4. Statistical inferences can be made on the sensitivit y of responses to factors of interest. 5. Factor levels can be determined for which response variable is optimum. 6. Factor levels can be determined that simultaneously optimize several responses. Contour plots are a popular graphical repres entation of a response surface that can be fitted to a statistical design. For simple re sponse functions, contour plots are generated by plotting one factor versus the other (two level experiment) and fixing the response. When the response function is more comple x, matrices of the re sponse variables are calculated and the numerical va lues of the response are plotted as a function of the factors. The optimum response can then be located by fixing the e xperimental variables to the desired values. Figures 4-2 (a-c) shows the contour plots for pH versus pressure at 75, 112.5, and 150 RPM, respectively. At 75 RPM figure 4-2a the region that predicts the highest removal rate is at ~ 6 psi and pH values less than 3. The silica removal rate should be in the range of 200-800 nm/min. This criterion is only met for low pH values and applied pressures in

PAGE 83

69 (a) (b) (c) (a) (b) (c) Figure 4-2 Predictions of removal rate as a f unction of pressure and pH at: a) 75 RPM b) 112.5 RPM c) 150 RPM. low to mid-pressure regimes < 6 psi at 75 RPM. At 112.5 RPM, the contour plot for pH versus pressure changes, figure 4-2b. The rem oval rate is predicted to increase uniformly as the pH and pressure incr eases. In contrast to the 75 RPM case, at 112.5 RPM the minimum removal rate criteria is achievable at all values of pressure and pH. Figure 4-2c (150 RPM) shows a similar trend to that pred icted in Figure 4-2b. Howe ver, pH values in excess of 3 are required to achieve the mini mum removal rate criterion. If we compare figures 4-2a-c, the optimum region for removal rate as function of pressure and pH is clear only in figure 4-2a. However, in figures 4-2b and 4-2c, the removal rate increases in the same direction (increasing pH and pressure). The lack of a contour with a maximum response indicates the optimum removal rate is outside of the experimental region of figures 4-2b and 4-2c for the resp ective experimental designs. Figures 4-3 (a-c) are the inte raction profiles for the vari ables under investigation. Pressure increases at pH 6.0 have a more pr onounced effect on the removal rate than at

PAGE 84

70 (a) (b) (c) (a) (b) (c) Figure 4-3 Interaction profiles fo r removal rate as a function of pressure and pH at: a) 75 RPM b) 112.5 RPM c) 150 RPM. pH 2.0 for the velocity range in this study. From 2-10 psi, the removal rate increases fivefold at 75 RPM (figure 4-3a) and two-fold at 112.5 (figure 4-3b) and 150 RPM (figure 43c) at pH 6.0. An increase in applied pressure only results in a sli ght enhancement of the removal rate at pH 2.0 for all velocities. To understand this behavior the actual contact pressure at the pad-wafer interface should be considered. For slurry pH values << than the IEP of ceria (~ 6.1-6.3 for these slur ries), the particle size remains unchanged. However, as the slurry pH approaches the IEP of the ceria abrasives, the effective particle diameter increases because of agglomeration, which occurs as particle-particle repulsive forces are reduced. Now, as the applied pressu re increases, the number of particles at the pad-wafer interface and the pad-wafer contact area increases. For smaller particles (low pH case), the fraction of the pad that is covered with abrasives or fractional surface coverage increases [Cho04b]. In this case, the actual contact pressure at the pad-particles-

PAGE 85

71 wafer interface only slightly increases. However, in the case of pH 6.0 large agglomerates are formed, the fractional surface coverage is less [Cho04b]. Hen ce the actual contact pressure or load per particle is higher in the high pH case, which is evidenced by the removal rate response as a functi on of applied pressure. Figure 4-4 shows the predicted removal rate response for all three variables (pressure (A), pH (B), and velocity (C)). A negative sign on the factor or variable corresponds to the minimum variable in the ra nge, with positive signs being used for the maximum value. The numerical values are th e removal rate responses for the respective combinations of pressure, pH, and veloci ty. The maximum removal rate should be reached for high pressure, pH, and velocity values (10 psi, pH 6.0, and 150 RPM). Figure 4-4 Predicted removal rate as a function of pressure, pH, and velocity.

PAGE 86

72 Surface Finish Response A successful CMP process requires more th an just a high rem oval rate. Among the many parameters that may be considered, su rface finish is perhap s second in importance only to the removal rate. For integrated ci rcuit manufacturing, the surface roughness or topographical variations must be minimized to ensure conformal deposition of successive materials and error-free lithography [Wil93] In STI CMP for DRAM structures, the surface roughness must be minimized to ensure the read-write heads of recording media do not crash into the storage material. (a)(b) (c) (a)(b) (c) Figure 4-5 Predictions of RMS roughness as a fu nction of pressure and pH at: a) 75 RPM b) 112.5 RPM c) 150 RPM Figures 4-5 (a-c), show the predicte d root mean square (RMS) roughness in nanometers as a function of pressure a nd pH for 75, 112.5, and 150 RPM respectively. For the 75 RPM case (figure 4-5a), the lowe st roughness values ar e achieved for pH 3.0 and when the applied pressure ~ 4.0 psi. An increase in the velocity to 112.5 RPM (figure 4-5b) results in a further decr ease in the lowest predicted roughness. Again this optimum

PAGE 87

73 region is located at low pH (< 4.0) and low pressure (< 5 psi). Further improvements in surface finish are predicted as the velocity is increased to 150 RPM, figure 4-5c. This optimum region is also located at low pH (< 4.0) and low pressure (< 4 psi). As the rotational velocity is increased, the regions of factor space that predict the lowest RMS roughness values are found at increasingly lower pH and pressure values. The indentation depth of a single particle is proportional to the applied pressu re and the particle diameter [Coo90]. An increase in pressu re, pH, or both in tandem, re sults in deeper penetration into the silica surface. The interaction profiles for the surface roughness predictions, figures 4-6 (a-c), show that there is a strong interaction between the experimental variables at low velocities (figure 4-6a). The mutual effects weaken as the rotational velocity is increased (figure 4-6c and 4-6c). According to this design planarity should increase with velocity. (a)(b) (c) (a)(b) (c) Figure 4-6 Interaction profiles for RMS roughness as a function of pressure and pH at: a) 75 RPM b) 112.5 RPM c) 150 RPM

PAGE 88

74 Figure 4-7 shows the predicted surface finish as a function of pressure (A), pH (B), and velocity (C). The optimum regi ons that produce minimal surface deformation occur at low pressure and pH values and high rotational velocities. The statistical results have been explained using actual CMP phe nomena. Hence, this statistical model accurately predicts the effect of pressure, pH, and veloci ty on the removal rate and surface finish for ceria-silica CMP. Figure 4-7 Predicted removal rate as a function of pressure, pH, and velocity. Summary A statistical design based on the Box Be hnken method was used to predict the removal rate and RMS roughness after CMP as a function of pressure, pH, and velocity. Both the removal rate and su rface finish are predicted to be strongly pH and velocity dependent. The applied pressure is expected to have only a slight effect on the material

PAGE 89

75 removal rate for most of th e experimental conditions examined. However, the surface roughness does depend strongly on th e applied pressure, since th e indentation or particle penetration depth is proportional to the a pplied pressure. Using statistical methodology, the isolated and joint effects of three variables were simultaneously investigated. Statistical designs are a powerful means of process optimization. However, the goal of this study is to unders tand the mechanisms of the ceria -silica interactions during CMP that lead to material removal. Therefor e, the remainder of this study will focus on the fundamentals at the ceria particle silica wa fer interface. In the next chapter, chemical and surface chemical effects are investigated.

PAGE 90

76 CHAPTER 5 THE EFFECT OF CERIA SLURRY PH ON THE SILICA REMOVAL RATE, FRICTION FORCE, AND SURFACE MORHOLOGY DURING CHEMICAL MECHANICAL POLISHING Introduction A statistical design based on the Box Be hnken method was used in Chapter 4 to predict the effect of pressu re, pH, and velocity on the silica removal rate and surface finish after CMP with ceria slurries. Statisti cal designs are important methods for process optimization or materials selection. However, statistical methods are limited in usage by the depth of knowledge of the system under investigation. Furthermore, mathematical rather than statistical models are required for articulating the materialistic responses based on the properties of the components of the system in question (i.e. surface layer hardness, particle hardness, pad hardness, etc.). Hence, fundamental experiments are required and the one-factor-at-a-time laborat ory testing technique is unavoidable. The rapid growth of the semiconductor industry is made possible by a combination of innovative device integration schemes, technological advan ces, and judicious materials selection. The continued scaling of microel ectronic devices has re sulted in unprecedented improvements in microprocessor performance a nd an increased density of memory chips. The advances in device integration and the in troduction of new mate rials to integrated circuit manufacturing have also driven the development of novel pr ocessing techniques. Vertically stacked or hierarchal type stru ctures that are critica lly dependent on global planarity are utilized in a variet y of manufacturing techniques.

PAGE 91

77 Hence, CMP is increasingly relevant and has stimulated significant research interest in the academic arena. Shallow trench isolation (S TI) is an example of a novel manufacturing technique that is critically reliant u pon CMP. Traditionally, silica-base d slurries were used for the interlevel dielectric CMP step during IC fabr ication. The popularity of STI has driven the interest in new CMP consumables (pads, slur ries, and slurry additives), to meet the stringent planarity, removal rate, selectivity, and surface finish requirements. Significant improvements in device characteristics are re alized by replacing c onventional isolation techniques with STI. However, STI CMP is much more complex. For a successful STI CMP process, the oxide to nitride polish sel ectivity, oxide erosion or dishing, and nitride loss must be controlled. Ideally, the STI CMP pr ocess will come to a halt immediately as the oxide is cleared and the ni tride is exposed. Ceria-based sl urries are promising for high selectivity polishing as they intrinsically polish oxides faster than nitr ides or bare silicon [Eva04]. The CMP process should also result in uniform removal of the trench oxide. Deviations in oxide thickness in excess of 3% trench oxide erosion greater than 50 nm, and nitride overpolishing may cause damage to the underlying silicon substrate [Sch99, Zha01]; all are deleterious to device performance. The “holy grail” of sorts in STI CMP is an appropriately high selectivity, i.e. high oxide polish rate with little to no nitride remova l. If the selectivity is low, nitride loss may occur. However, if the selectivity is too hi gh, the oxide removal rate is uncontrollable and oxide erosion results. To achie ve the desired oxide to nitr ide selectivity (> 100:1), two methods have been adopted. In the first me thod, surfactants (anionic, nonionic, or a mixture of the two) are used to passivate or lubricate the nitrid e surface. In the case

PAGE 92

78 where the slurry pH is less than the IEP of the silicon nitride surface, an anionic surfactant in the ceria-based slur ry is expected to selectivel y adsorb on the nitride surface. The mechanisms of the adsorption are not unde rstood, but electrostatic attraction because of opposite charge at the nitride and surfactant surfaces is expected to play a role. Tseng, S-D Kim, Katoh, and Park have studied the eff ect of surfactants on the silica removal rate for high selectivity polishing [Tse01, Kim02, Kat03, Par04b]. Fixed abrasive pads are used in the second appr oach [Sim01, Gag02, Kul04]. Surfactant Based STI CMP Tseng et al. studied the effect of two nonionic surfactants with similar molecular weights: SHE, which has a siloxane h ead group, and CHE, which has a methyl methacrylate head group [Tse01] Silica-based slurries were used for the studies. The superior polish rate of CHE-based slurries wa s attributed to an accel eration of chemical processes during polishing. The results of Ts eng and co-workers are insufficient for the STI CMP process. The maximum removal rate of the surfactant based slurries was less than that for silica-based slurries with no a dditives. Since the oxide removal rate is not accelerated, the oxide to nitr ide selectivity should remain low. Using atomic force microscopy, Basim determined that the additi on of surfactants lowe rs the removal rate because of increased lubrication at the pad-wafer interface [Bas02a]. S-D Kim et al compared chip scale and wafer-to-wafer thickness variations for a ceria-based high selectivity slurry (selectivity values were not given) and a silica-based low selectivity slurry [Kim02] According to their results, CMP with the conventional silica slurry resulted in great er nitride overpolishing and oxi de erosion. An in-situ filter was required for the high selectivity slu rry to minimize scratches on the polished surfaces. Katoh et al. studied the effect of su rfactant concentration a nd ceria particle size

PAGE 93

79 on the oxide and nitride removal rate [Kat03, Par04b]. For larger particles (120 nm), the polish rate without surfactant was twice that of the smaller particles (50 nm) for both oxide and nitride polishing. However, as th e acrylic acid (wate r-soluble surfactant) concentration was increased to 0.10 wt%, the ni tride polish rate rapi dly saturated at ~ 5 nm/min, for the 120 nm and 50 nm (primary part icle size) slurries. The oxide polish rate was not affected until the su rfactant concentrat ion reached 0.40 wt% for the 120 nm particles. The smaller particles showed sim ilar polishing behavior for oxide and nitride surfaces. These observations were attributed to the difference in partic le velocity in the viscous surfactant as a function of particle size. In their interpretation abrasive particles freely move through the fluid film, impinge on the surface, and remove material. According to their conclusions, the velocity of larger particles remains significant enough to reach the surface until a critical fluid thickness is reached at higher surfactant concentrations. This assertion is problematic for several reasons. First, free-flowing particles that are not embedded in the pad are not believed to contribute to the material removal rate [Zha99]. Second, the suppression of the nitr ide removal rate for 120 nm at 0.10 wt% acrylic acid does not agr ee with the proposed explan ation. Finally, because of agglomeration, the average particle size was nearly identical up to 0.4 wt% acrylic acid, negating particle size effects. It is more like ly that the acrylic acid passivates the nitride surface resulting in a saturation of the nitr ide removal rate at very low surfactant concentrations. For the oxide CMP case, the en hanced removal rate for larger particles may occur because of the greater penetration depth of abrasive particles embedded into the polishing pad. An interesting aspect of this study by Katoh and co-workers is the

PAGE 94

80 possibility of nitride passivat ion using acrylic acid, which re sults in enhanced oxide to nitride selectivity. Fixed Abrasive Pads for STI CMP STI CMP with high selectivity slurries is problematic because of increased oxide erosion. Robust endpoint detection schemes are n eeded to stop polishing as the nitride is exposed [Sim01]. Fixed abrasive (FA) pads are a revolutionary approach to STI CMP [Sim01]. Using FA pads, problematic issues a ssociated with particle-based slurries such as handling, lifetime, and stab ility may be alleviated. Genera lly, FA CMP slurries consist of pH-adjusted water and a separate chem ical for enhanced selectivity. A further complication of STI CMP with particle-based slur ries is the requirement of an initial etch step to minimize difficult-to-polish topography [Ste97]. CMP with FA pads allows direct polishing and the additional advantage of reduced oxide erosion [Sim01]. The FA pad is a composite of ceria abrasives and a resin binder supported by polym er backing (figure 51). The composites have a well-defined shape and are regularly spaced to ensure transport of the liquid slurry and polish debris. The pl anarization efficiency is a function of the morphology of the composite. Figure 5-1 SEM micrograph of fixed ab rasive pad surface (200x) [Vel00].

PAGE 95

81 Simpson et al. studied three different fixe d abrasive pads supplied by 3M that were classified as slow, medium, or fast accord ing to their respectiv e oxide removal rates [Sim01]. The slow pad consisted of cyli ndrically shaped ceria/polymer composite structures. The oxide removal rate for the slow pad was less than 10 nm/min. Hence, this pad was deemed suitable for CMP of thin oxi de overfills of less than 20 nm [Sim01]. Simpson and co-workers characterized this pad as tough, meaning micro-scale topography was required to initia te polishing. The slow pad showed high selectivity to topography, which means the removal rate at low areas was negligible. Oxide dishing was eliminated with an overall improvement in oxide and nitride uniformity after CMP. The polishing capability of the slow pad was critically dependent on the quality of the oxide deposition [Sim01]. The pad classified as medium was structurally similar to the slow pad except that it releases more ceriu m during polishing [Sim01]. The medium pad also produced very low oxide removal rate s (~20 nm/min). Unlike the slow pad, the medium pad did not provide a suitable ove rpolish process window (excessive oxide dishing), but may be applicable to borophos phosilicate glass polishing [Sim01]. The fast polishing pad was used for isolation trenches with thick overlying oxi de layers. The oxide polish rate was significantly higher (> 200 nm/min). The fast pad produced poor oxide to nitride selectivity and excessi ve oxide dishing, and, was ther efore deemed unsuitable for STI CMP. Gagliardi et al have also studied FA CMP for STI. LProline, an amino acid, was used to tailor the oxide-to-nit ride polish selectivity, elimin ate nitride erosion, and oxide dishing [Gag02]. In their study the polishing t ool was specifically configured for the FA study. A roll of fixed abrasive was loaded onto the CMP machine. After polishing a

PAGE 96

82 wafer, the FA pad is moved ~ 3 mm. After a certain number of wafers are polished, each subsequent wafer is exposed to an abrasive surface that has been used in an identical manner [Gag02]. The abrasives were shaped like pyramids and were 63 m tall and 125 m wide. The combination of FA pad and L-Proline yielded a robust CMP process with well-defined polish characteristics. The oxide and nitride removal rates were controlled by the addition of L-Proline. An oxide to ni tride selectivity in excess of 100 was obtained for a variety of pH conditions and amino acid concentrations. For precise process control of particle-bas ed and fixed abrasive STI CMP solutions, a fundamental understanding of the ceria-silica interactions is necessary. However, very few investigators have focused on the mechan isms of the ceria-silica CMP process. In chapter 2, the current unders tanding of the ceria-silica pol ishing process was reviewed. Despite the lack of fundament al ceria-silica CMP investigat ions, three models can be found in the glass polishing and CMP literature. The chemical tooth model is most widely relied upon for explanations of the ceriasilica CMP mechanism. However, nothing in th e manner of experimental proof has been extended to validate the proposed theory. Tw o other less-referenced models are also available. One, which I have termed the pa rticulate generation model, depicts the ceriasilica polishing mechanism as more mechanical in nature. The other, which I have termed surface defects model, stipulates that effi cient polishing compounds must have surface defects for desired polishing. In itial ceria-silica reactions are a central theme of all three proposed models. In Chapter 2, the effect of pH on the ceria surface chemistry was reviewed. Since the ceria polish rate is believ ed to be enhanced by its intrinsic chemical reactivity and the solution pH affects the ce ria surface chemistry and chemical reactivity;

PAGE 97

83 the CMP performance should be strongly pH depende nt. In this chapter, the effect of the ceria slurry pH on the silica re moval rate, friction force at the pad-ceria-silica interface, and silica surface morphology is presented. Experimental All polishing slurries contained ceria a nd DI water only. To adjust the pH to desired values (2-6.5), HCl and KOH were us ed. The slurry was delivered at a constant rate of 100 ml/min. In-situ lateral friction force measurements and polishing experiments were conducted on a Struers Rotopol 31 tablet op polisher. A more complete description of the CMP polisher and friction force appara tus may be found in Chapter 3. The sample size for polishing and friction tests was 1.0 x 1.0 in., and the samples were ~2 m thick oxides deposited on p-type silicon (100) by chemical vapor deposition. The down pressures for polishing experiments and fricti on force measurements were 8.8 and 3.5 psi, respectively. All samples were polished for 30s. Slurry samples were collected at pH 2.0 and 6.0 during experimental runs. Rodel IC 1000/ Suba IV stacked pads were used for all experiments. To ensure reproducible condi tions, a Grid-Abrade diamond pad conditioner was used to roughen the po lishing pad before each run. Dynamic light scatteri ng techniques were used to me asure the ceria particle size, zeta potential, and stability. A scanning elec tron microscope (SEM) equipped with a field emission gun was also used for particle size measurements. A Microtrac UPA 150 particle size analyzer was used to measure the particle size befo re and after CMP. A turbidimeter Model 2100AN manufactured by Hach Company was used to assess the slurry stability at pH 2.0, 6.0, 6.5, and 10.5. Zeta potential measurements on the slurries before and after CMP were performed on an apparatus manufactured by Brookhaven Instruments. Material removal rates were calculated by measuring the oxide thickness

PAGE 98

84 before and after CMP with a variable angle spectroscopic ellipsomet er. AFM was used to characterize the morphology of the polished surfaces after CMP. Results and Discussion Ceria Slurry Characterization In the pH range studied, the silica substrate possesses a predominantly negative charge because of deprotonation of surface hydroxyls. Si-OH Si-O-+H+ (5.1) Conversely, the ceria abrasives go from predom inantly positive to mo stly neutral as the pH of the slurry approach es its isoelectric point. Ce-OH2 + Ce-OH+H+ (5.2) Figure 5-2 Ceria slurry characteristics as a f unction of pH a) zeta po tential b) particle diameter. The IEP for silica is between pH 2-3 and pH 6-7 for ceria [Coo90]. The effect of pH on the ceria slurry properties is shown in Figures 5-2a and 5-2b In the absence of surfactants or other dispersing agents, electrostatic mean s are solely responsible for the dispersion state of the slurry. Figure 5-2a indicates th e IEP of the ceria abrasives used is ~ pH 6.0. As the pH is increased from 2-6.5, the electros tatic interactions become less repulsive. As

PAGE 99

85 expected, slurry agglomerati on ensues, resulting in an increase in abrasive size and a decrease in slurry stability (Figures 5-2b and 5-3) Even at pH 10.5, the ceria slurry remains unstable with the partic les rapidly settling out of solu tion in less than 10 minutes. Since direct imaging is possible, SEM m easurements were conducted, (figure 5-4). The primary particle size as determined by SE M is ~ 60 nm. The particle size values determined by light scattering should be c onsidered the secondary or agglomerated particle size. The particle si ze measurements via light scatte ring were consistent even after sonication. Figure 5-3 Stability of the ceria slurry as a function of time for different pH values. Vertical axis is displayed as normalized nephelometric turbidity units (NTU). The lack of ceria stability at pH 10.5 is une xpected, especially in light of the zeta potential measurements. At pH 10.5, the zeta potential is ~ -40 mV, approximately equal in magnitude to the low pH case. Therefore, it is not unreasonable to expect that ceria dispersions at pH 10.5 should be as stable as dispersions at pH 2.5. The actual state of the dispersions used for turbidity and zeta potenti al measurements and surface chemistry of

PAGE 100

86 ceria as studied by Nabavi et al. [Nab93] may shed some light on these contradictory observations. Figure 5-4 FE-SEM image of ceria particles. The dispersions for zeta potential measurem ents are much more dilute than for turbidity measurements. For turbidity measur ements, the ceria slurry is added to the turbidity cell as is, since the concentration is less than 3 wt%. Howe ver, for zeta potential measurements the specimen holder was initiall y filled with pH 10.5 DI water and then a drop of the ceria slurry was added to the speci men. The inter-particle spacing is greater for the dilute suspension and the likelihood of particle-particle collis ion diminishes as the concentration of the dispersi on decreases. The samples for turbidity measurements are more representative of the actual slurries used for the CMP experiments. The zeta potential is directly related to the density of charged sites at the abra sive particle surface. Nabavi’s investigation of the ceria surface chem istry as a function of pH is relevant to this discussion.

PAGE 101

87 Nabavi demonstrated that stable ceria di spersions could only be obtained in certain pH ranges [Nab93]. Originally, Nabavi and co -workers set out to measure the surface charge of the nanoscale ceria particles through titra tion of the surface groups with either a base or acid. Surprisingly, it was found that titrations in the pH range where charge reversal was expected could not be performed as cycles, because the dispersion did not return to its original state. This behavior is demonstrated by an excess consumption of hydroxyls (OH-) in acid to base titration as compar ed to the consumption of protons (H+) in the reverse cycle. According to their resu lts, three ranges of pH were identified that correspond to three unique dispersion characteristics (figure 2-10). Range A extends up to ~ pH 5. The titration cycles in this range were completely reversible and the dispersion remained stab le, which indicates the surface charge is constant. Range B extends from pH 7 to 10. Disp ersions titrated from pH 3.4 to any pH in this range back to pH 3.4 resulted in an excess hydroxyl consumption of 40 [Nab93]. Flocculation ensued as the boundary between range A and B was crossed. The dispersions remained unstable even afte r titrating back to pH 3.4. Interestingly, successive titration cycles from range A to B and back did not show the hysteresis (excess hydroxyl consumption), which indicates the number of sites able to consume a hydroxyl ion is: 1) different from the original dispersion and 2) stabil ized after the initial titration cycle. Range C extends beyond pH 10. The initial titration cycle in this range resulted in an excess consumption of 50 hydr oxyl ions [Nab93]. The step at pH 10 was sharp for slow titrations. Nabavi et al s uggested that the excess use of hydroxyls above pH 10 is associated with th e kinetics of a slow reaction, which suggests a more stable dispersion may be obtainable at pH 10 if the liquid environment was changed rapidly

PAGE 102

88 from acidic (pH < 5) to basi c pH (> 10). The dispersions in ranges B and C remained unstable even after returning to the original acidic pH for 40 days. At pH 2, re-dispersion took 2 days and at pH 0.5, re-dispersion occu rred immediately. This agglomeration was attributed to particle-particle condensation reactions that resu lt in a loss of surface area [Nab93]. In light of this discussion, it is more reasonable to assume that the surface chemistry instead of the solids concentration is responsible for the lack of stability at pH 10.5 as measured by the turbidimeter. The cer ia slurry was received in an aqueous dispersion at pH 4. To adjust the pH, KOH or HCl was added drop by drop until the desired pH value was reached. Since, the ceria slurries did not require significant dilution to perform the turbidity tests there was su fficient time for the c ondensation reaction to occur as the pH was adjusted from 4 to 10.5. However, for the zeta potential measurements a small amount of the slu rry was immediately exposed to a pH 10.5 environment, which prevented particle-p article bonding and is evidenced by the appreciable zeta potential value. It should be noted that it ma y be possible to temporarily disperse the flocculated slurri es using an ultrasonic bath. Silica Removal Rate Figure 5-5 shows the removal rate as a function of slurry pH. During CMP, a magnetic stirrer is used to counteract slurry settling, which makes it possible to polish wafers with unstable slurries. From pH 2.0 to 6.0 there is a five-fold increase in the removal rate. The silica polish rate continua lly decreases from pH 2.5 to the maximum value at pH 6. At pH 6.5, the removal rate begi ns to decrease. According to the chemical tooth theory as interpreted by Osseo-Asare [Oss02] the silica removal rate should be

PAGE 103

89 Figure 5-5 Silica removal rate as a function of ceria slurry pH. Figure 5-6 Post-CMP slurry characteristics as a function of pH a) zeta potential and b) particle size. highest at the IEP of the abra sive particles. The chemical tooth theory depicts silica polishing as an adsorptive removal process, which is controlled by the interactions between surface functional groups [Coo90, Oss02]. If the ceria particle s adsorb silicate species during polishing; the particle size and zeta potenti al of the post CMP slurry

PAGE 104

90 should be significantly changed. The ceria slurry was collected during CMP at pH 2.0 and 6.0, and the zeta potential (c ircular data points) and pa rticle size (A2.0 and A6.0) were re-measured (Figures 5-6a and 5-6b) After CMP at pH 6.0 the particle size increased significantly. Also, the zeta potential is much more negative than if ceria alone was present in the slurry. At pH 6.0 the zeta potential on the ceria and silica surfaces is expected to be approximately 10 mV and -60mV respectively. The measurements of particle size and zeta potential after CMP i ndicate the abrasive particle adsorbs silica during polishing or that the si lica coats the ceria particle while in solution. The zeta potential of co-dispersions of ceria and silica has been reported [Eva04], figure 5-7 Mixtures of ceria and silica in an aqueous e nvironment seem to beha ve like pure silica, which indicates that the ceria particle forms the core and the silica is the shell of the ceria-silica composite. Figure 5-7 Zeta potential versus pH for ceria, silica, and co-d ispersions [Eva04].

PAGE 105

91 Figure 5-8 Friction force (right axis) and removal rate (left) at pad-ceria particles-silica wafer interface. Friction at the Ceria-Silica Interface In-situ friction force measurements have b een used extensively to understand silicasilica CMP mechanisms [Mah99a, Mah99b, Cho01, Cho03, Cho04a, Cho04b, Cho04c]. The friction force is known to be strongl y dependent on interfacial electrostatic interactions, dynamic surface conditions, prop erties of the opposing surfaces, and the abrasive particle size, which all influence the contact area between the opposing surfaces. Figure 5-8 shows the in-situ friction force at the pad-particles-wafer interface along with the material removal rate as a function of pH. The maximum in the friction force is highest near pH 5.0. Changing the slurry pH not only influences the slurry stability; the extent of the electrostatic interactions betw een the abrasives and the silica substrate are also affected. The electrosta tic interaction between two su rfaces is proportional to the product of the charge on each surface (F ~ q1*q2), where F is the electrostatic force and q

PAGE 106

92 is the charge on the opposing surfaces. Therefor e, the product of the ze ta potential on the ceria abrasive and silica substrate as a functi on of pH may be used as an indicator of the magnitude of the electrostatic interactions (Figure 5-9). Chen has demonstrated the validity of this assumption using the lin ear superposition approximation (LSA) for calculation of the electrica l interaction energy ( GE) between a silicon plate and a spherical particle [Chen00, Chen03]. The LSA states that the pot ential distribution ( ) in overlapping double layers is just the summation of potential distributions in two isolated double layers [Chen03]. = 1exp(x) + ( 2exp[(d-x)] (5.3) Where is the inverse of the electrical double thickness, x is the position of surface 1, and d is the separation distance between su rfaces 1 and 2. Chen adopted the method of Derjaguin [Der87] to derive the expressi on for electrical interaction energy: GE = 4 a 12exp(x) (5.4) Figure 5-9 Qualitative comparison of the extent of the electrostatic interactions using the product of the zeta potential on the ceria and silica surfaces.

PAGE 107

93 where is the dielectric constant of the liquid media and a is the particle diameter. The extent of the electrostatic interactions can be crudely estimated by examining the product of the zeta potential on the silica wafer and th e ceria particles. In the pH range studied, ceria-silica interactions are predominantly attr active with a maximum at the apex of the friction force vs. pH curve. The particle si ze increase and the reduction in electrostatic attraction between the silica and ceria surfaces which reduces the contact area, may be responsible for the decrease in the friction force above pH 5, wher eas the removal rate increases up to pH 6. An additional contri buting factor may be operational at pH 6.0 other than mechanically controlled wear. Surface Morphology The result of AFM analysis of the polishe d silica surface is depicted in Figure 4Root mean square (RMS) roughness values were belo w 1 nm for the as-received silica samples and polished surfaces. The RMS roughness was near ly identical after polishing at pH 2.0 and 4.5. Because of slurry agglomeration, the RMS roughness of silica polished at pH 6.0 was nearly identical to that of the as-received samples. Ther efore, if ceria slurries are required at pH 6.0, a smaller particle size is needed to improve the silica surface quality. (a)(b) (c)(d) (a)(b) (c)(d) Figure 5-10 AFM topographical analys is of silica wafers a) no polishing b) polished at pH 2.0 c) polished at pH 4.5, and d) polished at pH 6.0.

PAGE 108

94 Figure 5-11 RMS roughness of silica wafe rs as a function of slurry pH. Comparison of pH Effects of Ce ria-Silica to Silica-Silica CMP Silica-silica CMP may be considered a model system for other dielectric CMP processes. The chemical effects during silicasilica CMP are also pH-dependent. In silicasilica CMP, the primary chemical contribution to the polishing process is hydrolysis and dissolution of the uppermost silica layers (e quation 2.4). Since silica-silica CMP provides the foundation for dielectric CMP processes, it is worthwhile to discus s the effect of pH on silica-silica interactions in light of the results of this chapter. Mahajan measured the in-situ friction force and silica removal rate as a function of pH for abrasive-free slurries [Mah99b, Mah00] The slurry pH was adjusted to 3.0, 7.0, and 11.0. In these findings, the frictional for ces at the pad-wafer interface were highest

PAGE 109

95 for the acidic pH condition. Mahajan attributed these results to electr ostatic interactions between the pad and wafer. Since the polishing pad has an IEP near that of silica, increasing the slurry pH results in an incr ease in repulsive for ces at the pad-wafer interface. Since no particles were used in thes e experiments, the silic a removal rate as a function of pH was very low and relatively c onstant except for very basic pH conditions. Contrary to friction force measurements, the silica removal rate in creased at pH 11. As reviewed in chapter 2, the sol ubility of silica increases rath er rapidly above pH 9.0 [Ile79, Coo90]. According to Mahajan, the increased diss olution of the silica network at high pH was responsible for the enhanced polishing. p H 024681012 Silica removal rate (nm/min) 0 200 400 600 800 1000 1200 2% ceria 10% silica p H 024681012 Friction force (N) 0 2 4 6 8 10 12 14 16 2% ceria 10% silica Figure 5-12 pH dependent ceria-s ilica and silica-silica [Cho03, C ho04d] interactions a) removal rate versus pH for 2 wt% cer ia slurry at 8.8 psi and 10 wt% silica slurry at 10.5 psi and b) friction force versus pH both measurements conducted at 3.5 psi. Choi studied the effect of pH on sili ca-silica polishing and measured the friction force for particle-based slurries [Cho03, Cho04d] These results are presented in figure 512 along with the silica removal rate and fric tional forces for ceria-based slurries. In figure 5-12a, the silica removal rate as a function of pH is plotted for silica and ceria slurries. For a true comparison, the results of figure 5-5 have been extended to pH 10.5.

PAGE 110

96 For the pH range displayed, the silica particle size is cons tant. The removal rate for the silica slurries is much lower than that obtai nable with ceria slurries. Also, the transition from high to low removal is fairly shallow, with the difference being approximately ~150 nm/min from pH 2.0 to pH 10.5. In contrast, the silica removal rate for ceria polishing increases from a minimum of < 200 nm/min to a maximum of over 1 m/min at pH 6.0. The removal rate further decreases as the sl urry pH is increased. At pH 10.5, a slight increase is seen. For ceria-silica polishing, three distinct regions are present in the removal rate versus pH curves. The first, from pH 2.5-6.0 shows a steady increase in the polish rate with increasing pH. The second, pH 6.5-8.5 show s a decrease in the polish rate with increasing pH. Finally, the polish rate increas es at pH 10.5. For sili ca-silica polishing, only two distinct regions are present. Initially, the polish rate steadily decreases with increasing pH. As in the ceria-silica case, an increase is seen at pH 10.5, which may be attributed to the enhanced silica dissolution kinetics in very basic media. Both Mahajan and Choi suggested the domi nant interactions during si lica-silica polishing are electrostatic in acidic to neut ral pH conditions and enhanced indentation of the softened silica surface because of increased dissoluti on in basic pH conditions. Figure 5-12a clearly shows that ceria abrasives even at lower solids loading and applied pressure; possess some intrinsic properties that contribut e to enhanced ceria-silica interactions that result in increased silica removal. Figure 5-12b shows the measured friction force at the pad-wafer interface for both ceria and silica slurries. Th e silica data was obtained fr om Choi et al [Cho03, Cho04d]. Since, the silica particle size remains constant over the enti re pH range, th e silica-silica

PAGE 111

97 in-situ friction force measurements are controll ed by the electrostatic interactions and the dissolution of the silica surface. However, the ceria particles aggregate as the slurry pH approaches the IEP of ceria. Therefore, along wi th electrostatic intera ctions, silica surface dissolution, and the proposed ceria -silica chemical reaction; th e change in particle size must be considered. The effect of particle agglomeration on the ceria-silica frictional forces was discussed earlier in this chapter. Perhaps the most interes ting feature of figure 5-12b is the difference in the ma gnitude of the ceria and silica frictional forces at pH 5-6. The friction force value for ceria slurries is a bout 3 times that for silica slurries for pH 56. Frictional forces are a function of the c ontact area at the inte rface between the two opposing surfaces. If ceria-silica bonding does occur, the frictional forces should be enhanced. In fact, Cook suggested that fric tional forces should be proportional to the fraction of the contact area in which bonding occurs [Coo90]. Summary Ceria-silica interactions during CMP are strongly pH dependent. The rate of chemical reaction between the abrasives and th e substrate, electrostatic interactions, the abrasive size because of agglomeration, and silica adsorption via ceria surface complexation may contribute separately or join tly to the observations in this work. The abrasive particle size along with the extent of electrostatic inte ractions between the abrasives and the silica surface affects th e contact area. In-situ friction force measurements have demonstrated the effect of pH on both. Additionally particle size and zeta potential measurements indicate dissolved silica is indeed adsorbed onto the abrasives during polishing. Furthermore, the additional chemical contribution for ceria slurries was demonstrated by comparing the si lica removal rates and in-situ friction force responses as a function of slurry pH.

PAGE 112

98 The results in this chapter may be interpre ted as experimental proof of the validity of a portion of the chemical tooth theory that was first proposed for glass polishing [Coo90] and adapted to ceria-silica CMP [O ss02]. In accordance with prediction, the material removal rate is highest near the IEP of the abrasive particles. To this point, the central facet of all cer ia polishing theories, the proposed particle-surface reaction, has not been verified. To determine the extent of the chemical reaction between ceria and silica and provide further interpretation of the results in this chapter, the next chapter’s focus is silica surface analysis after CMP.

PAGE 113

99 CHAPTER 6 NEAR SURFACE ANALYSIS OF POLISH ED SILICA: CHARACTERIZATION OF THE CERIA-SILICA CHEMICAL REACTION Introduction Ceria CMP Models A major feature of all ceria-silica chem ical mechanical polishing (CMP) models: surface defect, chemical tooth, and particulate ge neration, is the belief in the existence of a chemical reaction between the ceria partic le and the silica wafer [Coo90, Kal91, Kal98, Hos01]. The difference in the mechanisms is in the processes after ceria-silica contact. For the polishing model based on surface defect s, Kaller stipulated that [Kal91, Kal98]: 1. The abrasive particles should be softer than the substrate 2. The abrasive is slightly worn away, which exposes high-energy defect sites abrasive-substrate bonding occurs 3. Material is removed on a molecular scal e because of movement of the workpiece and subsequent rupture of the particle-substrate bonds 4. Portions of the substrate removed duri ng CMP remain on the abrasive surface. In his formulation, Kaller suggested intrin sic defects like oxygen vacancies did not contribute to the chemical reactivity of the ab rasive particles; hence, their effect on the polishing process is negligible. According to the chemical tooth mo del proposed by Cook [Coo90] and further developed by Osseo-Asare [Oss02], material removal during polishing is controlled by the interactions between surface functional groups at the abrasive and substrate surfaces. The relevant features of the chemical tooth model are:

PAGE 114

100 1. Water initially penetrat es the silica network. 2. Surface dissolution occurs because of simultaneous synergistic actions of hydrolysis and particle indentation. 3. The abrasive particle adsorbs the silica solute species. 4. The entire process is controlled by the respective surface charges. Osseo-Asare further stipulated that the most favorable c onditions for silica removal by ceria are met when the slurry pH is near the ceria IEP. In the particulate generation model proposed by Hoshino et al, the ceria particles remove silica as discrete lumps or part iculates. Infrared spectroscopy, electron microscopy, and x-ray microanalysis were used to characterize the size, shape, and chemical composition of the post-CMP slurr y. According to the size analysis, 80 nm ceria particles and a 600 nm lump of silica was present after CMP. Additionally, the pH shifted from a pre-CMP valu e of 8 to 7.4 after CMP. In the previous chapter the silica removal ra te, friction force, and surface finish was shown to be strongly dependent on the ceria sl urry pH. Also, the properties of the ceria dispersion are strongly pH dependent. The surf ace chemistry of ceria was correlated with the characteristics of the slurry (i.e. particle size, slurry st ability, and zeta potential). To further develop our understanding of the ce ria-silica polishing process, the surface chemical effects as a function of pH will be discussed in light of spectroscopic and microscopic analysis of polished substrates. Recent Progress There are few fundamental investigations into the propertie s of ceria for CMP applications. However, the increasing popularity of shallow trench isolation (STI) has stimulated interest in the fundamental pol ishing mechanisms. Also, re-examination of

PAGE 115

101 earlier investigations from glass polishing have contribute d to the understanding of ceriasilica interactions and are helpful in understanding the CMP mechanisms. Sabia and Stevens determined separately the respective chemical, mechanical, and the chemical mechanical contributions to glass polishing using cer ia abrasives [Sab00]. Sabia and Stevens assumed the material rem ove rate is just the summation of the respective contributions from the mechanic al, chemical, and chemical mechanical portions. The chemical effect was studied by performing corrosion tests for 24 hours at 950C. The dissolved species from ceria slurries was used as a corrosive agent. No glass weight loss was detectable; therefore, Sa bia and Stevens assumed there is no purely chemical contribution during ceria-silica pol ishing [Sab00]. The mechanical component of polishing was studied usi ng ethanol-based ceria slurri es. Ethanol has a similar viscosity (1.078 cP at room temperature) and dispersion performance as water. Substituting ethanol for water, Sabia was able to isolate the mechanical nature of the abrasive particles. The processing history (t hermal treatments, milling, etc.) of the ceria abrasives significantly affected the mechani cal interactions during polishing. According to their results, the particle size did not aff ect the mechanical interactions. The chemical mechanical component was extracted by subtra cting the removal rate in the non-aqueous solution from the removal rate in water. Particle-substrate attraction dominated the chemical mechanical component of the overall polishing process [Sab00]. Sabia and Stevens concluded that contri butions to the chemical mech anical component of polishing not related to surface charge or solubility are associated with mechanically induced particle-substrate in teractions and interfacial chemical reactions [Sab00].

PAGE 116

102 Kelsall [Kel98] and Sabia [Sab00] pred icted that redox (re duction/oxidation) processes at the ceria particle surface contribut e to the chemical reactivity of the ceria particles. CeO2 has been shown to be thermodynamically unstable in aqueous environments [Pour66]. In light of Pourbaix’s description, Kelsall predicted that ceria particles can be considered mixtures of the dominant Ce4+ and the reduced Ce3+ [Kel98]. Using electron energy-loss spectroscopy (EELS) in a field emission gun, scanning transmission electron microscope (FE-STEM), Gilliss et al. determined the valence state for single nanosized ceria particles [Gil04] by measuring several points from the center to the surface of the particle. The oxidation state of the cerium ion was determined from the ratio of the cerium white-line intensity (Ce M5/M4). The white-line intensity ratio of the Ce M5/M4 peaks was determined by taking the ratio of the sum of the intensities of the M5 and M4 white lines. The measurements (figure 6-1) show a gradual increase of the Ce M5/M4 ratio, from 0.85 to 1.1, with increasing dist ance from the particle center. The Ce M4,5 edge occurs at the ionizat ion threshold, which is appr oximately the binding energy of the 3d shell [Gil04]. These core -loss peaks are a result of the 3d104fn 3d94fn+1 transitions. CeO2 (Ce4+) is characterized by unfilled 4f states. The transition to Ce2O3 (Ce3+) occurs with the production of oxygen vacancies and the lo calization of an electron in the 4f state [Sko02]. Therefore, during CMP, ceria particles shoul d not be considered pure, stoichiometric CeO2. Rather, the ceria particles cons ist of a mixed valence state of predominantly Ce4+ in the interior and Ce3+ at the surface.

PAGE 117

103 M5/M4 Distance from particle center (nm) M5/M4 Distance from particle center (nm) Figure 6-1 Ratio of Ce M5/M4 white lines versus position [Gil04]. Recently, Chandrasekaran reported the results of x-ray photoelectron spectroscopy (XPS) analysis of ceria particles after STI CMP of silica (SiO2) and silicon nitride (Si3N4) wafers [Chan04]. The elemental composition before CMP and after CMP of oxide and nitride wafers was obtained from XPS data. As expected, no silicon was detected on the ceria particles before CMP. However, after oxide CMP, considerable Si was present (~ 14 at%). After 5 nanometers of sputtering, th e silicon content decreased to ~ 12.20 at%, approximately equal to the cerium content ( 12.80 at%). Perhaps most interesting, was the XPS analysis of the ceria particles after nitride CMP. No silicon was detected after nitride CMP, even though the silicon to nitride ratio is 3:4. Until recently, very little was known about the ceria-silica in teractions that lead to material removal during CMP. In the past, the proposed chemical reaction between ceria and silica was supported by circumstantial ev idence, at best. To understand the ceriasilica polishing mechanism, cons iderable attention has been focused on the ceria particles

PAGE 118

104 themselves. For a complete understanding of th e ceria-silica CMP proces s, it is necessary to understand the morphological change s of the substrate after polishing. Experimental The development of surface analytical tools with nanometer to atomic scale resolution has revolutionized materials characterization. Now, with these powerful techniques, theories originally proposed decad es ago can be tested from a materialistic aspect. In this chapter, a combination of surface analytical methods was applied, in conjunction, to interpret the observations in Chapter 5. The characteristics of each measurement technique were de scribed in detail in Chapter 3. If ceria-si lica bonding is a reality, as many have asserted for years a nd recent results indicate; the silica surface should contain some residual cerium. Therefore, after CMP, XPS was used to analyze the silica substrates that were polished at pH 2.5, 3.5, and 6.0. The minimum and maximum removal rates and the silica and ceria IEP is near pH 2.5 and 6.0, respectively. An aluminum anode (1483.6 eV), was used for x -ray generation. To ensure the XPS results were not distorted because of residual ceria particles, field emission scanning electron microscopy (FE-SEM) was conducted. The atomic force microscope (AFM) may be used for topographical, force, and wear measurements. The AFM can be furt her functionalized by depositing desired coatings on the atomically sharp cantilevers Tapping mode cantilevers fabricated from silicon and supplied by Digital Instruments were thermally oxidized. The oxidation was conducted at 10000 C for 2 hours. The cantilevers were then furnace cooled. The silicacoated cantilevers abraded the ceria film for 30 minutes at pH 2.5, 4.5, 6.0, and 11.0. To minimize the sliding speed and to avoi d re-tracing of the wear track, a 50 m scan size was used and the scan rate was 0.1 Hz. The interaction force was measured using a 7.3

PAGE 119

105 m silica particle and a ceria thin film that was deposited using pulsed laser deposition (PLD). The silica particle was attached to the long, thick silicon nitride cantilever (spring constant k = 0.13). The silicon nitride cantile vers are commonly used for contact mode topographical imaging. Further confirmation of the friction force and zeta potential measurements was obtained via AFM Results and Discussion Interaction Force Measurements The forces relevant to AFM are ultimately of electromagnetic origin [Gar02]. In the case of CMP, different inter-particle, surface, and environmental effects give rise to specific interactions that are strongly distance depende nt. The sum of the forces (Ftotal) acting on an AFM tip or partic le in this case is the su mmation of the long-range and short-range interactions [Gar02]. Ftotal = Felectrostatic + Fvan der Waals + Fchemical (6.1) Where Felectrostatic refers to the long range interacti ons related to charge on the opposing surfaces, Fvan der Waals refers to bonding that occurs becau se of interactions of fluctuating dipoles, and Fchemical refers to the short-range bindi ng forces typical of chemical bonds. Using a liquid cell filled w ith a solution of desired chemistry, non-contact AFM force measurements are possible. DI water solutions were formulated at the desired pH values using KOH and HCl. Interaction forces are measured by monitoring the displacement of the AFM cantilever as it approaches th e surface. For strongly adhering surfaces, hysteresis is seen in the approaching and attracting force versus distance curves. Figure 6-1 shows the force versus distance curves for the respective pH conditions (2.5, 4.5, 6.0, and 10.5). The force was normalized by the particle radius. At pH 2.5 there is only a slight, short range attraction that oc curs at separation di stances less than 5 nm.

PAGE 120

106 As expected, the dominant interaction between the silica particle and the ceria thin film is electrostatic and occurs at pH 4.5. This meas urement agrees well with the zeta potential, friction force, and qualitative estimation of the extent of the electrostatic interactions via the product of the ceria and silica zeta poten tials discussed in Chapter 5. A moderate attraction also exists at pH 6.0. In basic pH conditions the ceri a-silica electrostatic interactions are entirely repulsive. se p aration distance ( nm ) 020406080 -0.10 -0.05 0.00 0.05 0.10 pH 2.5 pH 4.5 pH 6.0 pH 11 Figure 6-2 Interaction force measurements, th e retracting portion of the curve: force versus separation distance for a ceria thin film and silica particle at different pH values. XPS Analysis Because of its sensitivity to local atomic bonding, XPS was used to probe the surface region of the polished substrat es. The surface sensitivity of XPS is due to the shallow

PAGE 121

107 escape depth of the photoelectrons. Informati on is obtained from only the first few atomic layers. The concentration of chemical sp ecies on the surface may be determined by comparing the relative peak intensities of th e constituent elements at their respective atomic binding energies. The as-received silic a sample shows the presence of Si, C, and O atoms (figure 6-3a). The adventitious C 1s peak can be used for calibration of the atomic binding energies, i.e. for correction of peak shifting because of charging. Figure 6-3b shows a typical XPS spectrum of the s ilica substrate after polishing at pH 6.0 along with a plot of Ce content (at%) versus polis h pH (6-3c). The Ce 3d peak after polishing at pH 6.0 is also presented (figure 6-3d). In comparison to the Si, C, and O photoelectron peaks, the Ce peak is rela tively broad and of low inte nsity. The most intense Ce photoelectron peak is exp ected at 882 eV for Ce 3d5. The atomic bonding energies of Si 2p3, Si 2s, C 1s, O 1s, Ce 3d5, and Ce 3d3 are: 107.5, 158.5, 289, 537, 889, and 907.5. Ce (< 1 at %) is detected on all silica samples analyzed with XPS (pH 2.5, 3.5, and 6.0). Table 6-1 shows the binding energies for th e XPS peaks as a f unction of polishing condition. As shown in figure 6-3c, silica polis hed at pH 6.0 containe d the most residual cerium. In the as received silica sample, the C 1s peak is shifted more than 3 eV. With a correction for charging, the XPS peaks for the as received silica samples are in agreement with well-known standards [M ou95]. The XPS peaks for the samples polished with ceria slurries at different pH valu es are shown in table 6-1. The binding energies of the peaks of interest occur at virtually identical pos itions for the samples polished at pH 2.5, 3.5, and 6.0. However, a difference of 1-1.5 eV ex ists in the binding energy of the Si 2p3 peak for the as received and polished silica samples. This discrepancy may be attributed to increased charging of the polished silica sa mples, which is confirmed by the higher

PAGE 122

108 binding energy of the C 1s peak for samples after CMP. The 3d peak (figure 6-3d), the most intense in the Ce series, was the only X PS peak attributed to cerium surface species. If we assume uniform charging, the actual position of the Ce 3d5 peak is 884-885 eV, which is still considerably higher than the expected position for CeO2. The closest matching ceria peak occurs at 883.6 for Ce2O3, which suggests the cerium compound is a reduced form of ceria. So, is this surface specie s particle contaminati on or a layer of ceria that has formed during polishing? Table 6-1 XPS binding energies for silica samp les before and after polishing with ceria slurries at different pH. XPS peaks As received pH 2.5 pH 3.5 pH 6.0 Standard Si 2p3 106.55 107.5 108 107.5 103.3, 103.6 SiO2 C 1s 288 288.5 289.5 289 284.6 O 1s 536 536.5 537 536 529.2 CeO2 532.5 SiO2 Ce 3d5 ________ 889 888.5 889 881.8, 882.4 CeO2

PAGE 123

109 Figure 6-3 XPS analysis: a) si lica before CMP b) silica afte r polishing at pH 6.0 c) Ce content versus polish pH d) Ce 3d peak after polishing at pH 6.0. FE-SEM Analysis To answer this question a FE-SEM equipped with an energy dispersive spectrometer (EDS) for chemical analysis was used to image the surfaces polished at pH 2.5 and 6.0. In figure 6-4a, what have been termed cerium rich ‘island’ type structures are clearly evident. EDS confirms that the “islands” ar e actually cerium-rich, figure 6-4b. Particle contamination was not detected at pH 6.0. However, at pH 2.5, fi gure 6-4a shows an interconnected network of particles on the silica surface. The isla nd-like cerium-rich regions were also found on the silica substrate polished at pH 2.5. The EDS spectrum at pH 2.5 confirms the particles are ceria, figure 6-5b. The IE P of silica is near pH 2.5. Therefore, the silica surface is predominantly neutral. This may indicate adsorption as

PAGE 124

110 predicted by Osseo-Asare. However, in this case the silica surface serves as the adsorbent and the abrasive particle is the adsorbat e. Interestingly, a nonuniform coating, which may be dissolved silica, is present on the abrasive particles ( Figure 6-5b). EDS results indicate only a small amount of cerium can be found on the surfa ces of the polished substrates for both pH condi tions, which is in agreement with the XPS elemental analysis. (b) (a) (b) (a) Figure 6-4 FE-SEM analysis of silica polishe d at pH 6.0: a) SEM image showing Ce-rich ‘islands’ b) EDS spectrum. (b) (a) (b) (a) Figure 6-5 FE-SEM analysis of silica polis hed at pH 2.5: a) SEM image showing CeO2 agglomerates b) EDS spectrum.

PAGE 125

111 Wear with Silica-Coated AFM Cantilevers During CMP, there are many complicated phenomena that determine the overall CMP performance. This is especially tr ue for CMP using ceria-based slurries. Mechanistic studies to elucidate CMP polishi ng fundamentals are simplified when issues like slurry dispersion and polishing pad effect s can be isolated. To further understand the silica-ceria CMP mechanism, w ear measurements were conduc ted at several pH values using silica coated cantilevers and a ceria thin film. SEM image and compositional analysis was conducted before and after wear of each tip at every pH condition. With this approach, the ceria-silica interactions can be studied in the absence of complicating factors that are inherent to th e CMP process. Figure 6-6 (a-d) shows the result of wear experiments conducted at pH 2.5. Figure 6-6a shows the silicon canti lever after thermal oxidation and 6.6b is an SEM image after wear for 30 minutes. Figures 6-6c and 6-6d are the EDS spectra before and after wear at pH 2.5. The tip shape is distinctly changed by the cantilever-film abra sion. The tip shape has changed from extremely sharp to blunted, which indicates the silica coating was worn away during abrasion. This is confirmed by EDS, which shows the O to Si ratio is decrease d from stoichiometric si lica to a nearly 1:1 ratio.

PAGE 126

112 70 at% O 30 at% Si 53 at% O 47 at% Si(b) (d) (c) (a) 70 at% O 30 at% Si 53 at% O 47 at% Si(b) (d) (c) (a) Figure 6-6 Cantilever-film wear at pH 2.5 a) tip after thermal oxidation b) tip after wear against ceria thin film c) EDS spectra of oxidized tip before wear d) EDS spectrum of oxidized tip after wear. Figure 6-7 (a-b) shows the silicon cantilev er before and after wear at pH 4.5 along with the respective EDS spectra (figure 6-7c and 6-7d). Simila r to the pH 2.5 case, the tip is blunted after wear at pH 4.5. However, the EDS spectra are si gnificantly different. After wear at pH 2.5, the O content decrea sed and the Si content increased, which indicates the oxide layer is being removed exposing the Si cantilever. However, after wear at pH 4.5 the Si content decreased and the O content in creased (figure 6-7d). Visual inspection of figures 6-6b and 6-7b indicate that wear at pH 4.5 is gr eater than at pH 2.5, despite the EDS results. This result is in qua ntitative agreement with the polishing results discussed in chapter 5. The rem oval rate increases significantly

PAGE 127

113 74 at% O 26 at% Si 93 at% O 7 at% Si(b) (d) (a) (c) 74 at% O 26 at% Si 93 at% O 7 at% Si(b) (d) (a) (c) Figure 6-7 Cantilever-film wear at pH 4.5 a) tip after thermal oxidation b) tip after wear against ceria thin film c) EDS spectra of oxidized tip before wear d) EDS spectrum of oxidized tip after wear. Figure 6-8 (a-d) shows the result of image and compositional analys is after wear at pH 6.0. Again, tip blunting is observed. The EDS spectrum in this case shows a considerable amount of cerium af ter wear (figure 6-8d). Also, de bris or what appear to be particulates are present after abrasion (figure 6-8b). Hen ce, the ceria-silica reaction is unambiguously confirmed. The ceria thin film is the only source of Ce. Figure 6-9 shows the wear result for pH 11. Similar to the pH 4.5 result, the O content increased while the silicon content decreased after wear. Tip wear at pH 11 is mo st extensive. This may be because of the enhanced silica dissolu tion in highly basic pH conditions.

PAGE 128

114 Thermal oxide before wear55 at% O 40 at% Si 5 at% Ce 69 at% O 31 at% Si (b) (d) (a) (c) Thermal oxide before wear55 at% O 40 at% Si 5 at% Ce 69 at% O 31 at% Si (b) (d) (a) (c) Figure 6-8 Cantilever-film wear at pH 6.0 a) tip after thermal oxidation b) tip after wear against ceria thin film c) EDS spectra of oxidized tip before wear d) EDS spectrum of oxidized tip after wear. Thermal oxide before wear 73 at% O 27 at% Si 86 at% O 14 at% Si(b) (d) (a) (c)Thermal oxide before wear 73 at% O 27 at% Si 86 at% O 14 at% Si(b) (d) (a) (c) Figure 6-9 Cantilever-film wear at pH 11.0 a) tip after thermal oxidati on b) tip after wear against ceria thin film c) EDS spectru m of oxidized tip before wear d) EDS spectrum of oxidized tip after wear.

PAGE 129

115 Figure 6-10 (a-c) shows the tip wear ag ainst silica substrat es under the same experimental conditions as the ceria-silica wear This result demonstrat es the effect of pH on silica dissolution. The debris present in figure 6-10a is likely dust or some other contaminant, compositional analysis indicate d that no silicon or cerium species were present. The silica-silica wear changes from mechanical to chemical plus mechanical as the pH is increased. The wear at pH 11 is significantly gr eater than in the acidic pH regimes. As in the ceria-silica case the highest wear occurs at pH 11. pH 2.5 pH 11 pH 6.0(a)(b) (c) Figure 6-10 Cantilever wear against a silica substrate a) pH 2.5, b) pH 6.0, c) pH 11. Summary The proposed ceria-silica chemical reacti on was verified using XPS, SEM, and AFM measurements. Force measurements we re conducted with a ceria thin film deposited by PLD and a silica particle using an AFM. The most extensive interaction was electrostatic in nature an d occurred at pH 4.5. This is in agreement with the measurements of zeta potential and friction force and the semi-quantitative predictions

PAGE 130

116 based on the product of the ceria and silica zeta potentials presented in Chapter 5 of this work. XPS analysis indicates that the most exte nsive reaction occurs at pH 6.0, which is the pH that gave the highest silica removal rate. To ensure that the cerium species were not from residual particles, SEM analysis was conducted on the polished samples. Surprisingly, cerium-rich, island-type stru ctures were found s cattered on the silica surface. At pH 2.5, a small contribution to the cerium signal was due to the presence of residual cerium particles. The ceria thin films were also used to compare the wear of silica coated AFM cantilevers at pH 2.5, 4.5, 6.0, and pH 11.0.The rmally oxidized AFM cantilevers were rastered against ceria thin f ilms and silica substrates in various pH environments. SEM image and compositional analysis was conduc ted after cantilever-f ilm abrasion. For all conditions, the tip assumed a blunted appearan ce from the original sharp shape. Ce was only detected at pH 6.0. Also, the most extens ive tip wear occurred at pH 11.0. In the case of silica-silica, the tip shape indicates a transition from primarily mechanical to chemically-assisted wear at high pH.

PAGE 131

117 CHAPTER 7 SUMMARY, CONCLUSIONS, AND SUGG ESTIONS FOR FUTURE WORK Summary The continued miniaturization of microelect ronic devices requires tighter control of the chemical mechanical polishing (CMP) pro cess. Also, the implementation of CMP in the manufacturing process of other areas like shallow trench isolation (STI), micro electromechanical systems (MEMS), and flat panel displays requires a new set of CMP consumables (pads, polishers, slurries, etc.). To meet the stringent planarity requirements of these novel processes, the f undamental interactions at th e pad-slurry-wafer interface must be understood. This is especi ally critical in STI CMP. The STI CMP process requires the global planarity of conventional interlevel dielectric (ILD) CMP, with the additional complication of preferential polishing of oxide layers to nitride or poly silicon. The factors that cont rol oxide erosion, nitride overpolishing, particulate resi due, and other sources of de fectivity (surface scratches from particles or pad) must be known and eliminated. Traditionally, this has been accomplished by altering the chemistry of the polishing slurry. Ceria-based slurries and fixed abrasive pads, which consist of ceria abrasives surrounded by a polymeric material, are two proposed solutions for STI CMP. Theref ore, the most critical interactions during STI CMP occur at the ceriasilica interface. A thorough unders tanding of the mechanisms at the particle-substrate interface will allow tighter control of the STI CMP process. The ceria-silica interactions that determine th e output of the CMP process are reported.

PAGE 132

118 Chapter 1 started the dissertation with the motivation for implementation of CMP in integrated circuit (IC) manu facturing. The growth of CMP to other areas like STI, were also discussed. In Chapter 2, the curr ent understanding of th e CMP process was summarized. This chapter also included a basi c description of the multilevel metallization process that is critically de pendent on CMP. Silica CMP may be considered a specialized form of glass polishing. Hence, glass polishi ng mechanisms were reviewed. For silicasilica CMP, the effect of part icle size, concentration, and pa rticle size distribution on the silica removal rate and surface finish was cove red. The basics of shallow trench isolation along with three proposed models for ceria-s ilica CMP were detail ed. Chapter 2 ended with a discussion on the ceria surface chemis try. Chapter 3 detailed the materials and methods used in this work. In Chapter 4 the effect of pressure, pH, a nd velocity were studied using statistically based experimental methods based on the B ox Behnken design. There are more than forty variables, effects, and output parameters involved in CMP process design. It is impossible to investigate the effects of each simultaneously, using conventional experimental techniques. Sta tistically based experimentati on allows prediction of the optimum combination of the respective variables for the desired performance. Furthermore, mutual or joint effects can be examined with statistically based techniques that are not readily apparent in laboratory experiments. The optimum removal rate and surface finished was found to be strongly de pendent on the slurry pH and rotational velocity of the polisher. The pressure effect on the removal rate was slight. However, the applied pressure has a pronounced effect on the surface finish.

PAGE 133

119 A major feature in all ceria-silica polis hing models is the proposed chemical reaction between the abrasive particles and the silica substrate. The chemical tooth theory is the most widely accepted model to explai n the removal of sili ca by ceria. In this formulation, the polishing performance is co ntrolled by the interactions between surface functional groups on the abrasive and substrate surface. To date, no experimental evidence has been provided in support of this proposed mechanism. There is also a lack of fundamental studies on ceria-silica polishi ng. In Chapter 5, the effect of pH on the silica removal rate, friction force, and su rface morphology was studied. The removal rate, friction force, and surface roughness are depe ndent on the slurry pH, with the maximum removal rate occurring just below the isoe lectric (IEP) point of ceria. A qualitative comparison of the extent of the electrostati c interactions was obt ained by analyzing the product of the ceria and silica zeta potentials at various pH values. The maximum in friction force occurred in the pH range wh erein the electrostatic attraction was most extensive. Zeta potential and particle size analysis of the post-CMP slurry indicated the ceria abrasives adsorb silicate species duri ng polishing. The effect of pH on the ceria surface chemistry was correlated with the CMP observations. In Chapter 6, the silica substrates polished with ceria slurries at different pH values were examined by x-ray photoelectron sp ectroscopy (XPS), field emission scanning electron microscopy (FE-SEM), and atomic force microscopy (AFM). If the ceria particle indeed reacts with the silica substrate during polishing, residual cerium should be present on the silica surfaces af ter CMP. XPS analysis determined ceria was present on the surface after polishing. To ensure the residua l cerium was not because of particulate impurities, FE-SEM analysis was conducted. At both pH 2.5 and pH 6.0, isolated patches

PAGE 134

120 termed cerium-rich ‘islands’ were found on th e silica substrates. El emental analysis by XPS and energy dispersive spectroscopy (EDS) indicated the most cerium, and therefore, the most extensive ceria-s ilica reaction was at pH 6.0. To understand the ceria-silica polishing mechanisms, a ceria thin film deposited by pulsed laser deposition and thermally oxidized silicon cantilevers were used for wear measurements as a function of pH. The cantile vers were rastered against the ceria thin film at pH 2.5, 4.5, 6.0, and pH 11.0. These measurements provide further confirmation of the ceria-silica polishing mechanism. From the change in cantilever shape and composition as measured by EDS, the abrasion mode is similar at all pH values. However, at pH 6.0, residual cerium was dete cted on the cantilever tip. The presence of cerium in the absence of particles confirms the ceria-silica reac tion indeed occurs. Conclusions Three models have been proposed for ceria-silica polishing : chemical tooth [Coo90, Oss02], surface defects [Kal91, Kal98], and particulate gene ration [Hos01]. In light of these findings, it seems Cook’s mode l provides the most accurate description of the ceria-silica polishing mechanism. The exis tence of a chemical reaction between the ceria particle and the silica surface has been confirmed, which was predicted in all three models. Zeta potential measurements of the pos t-CMP slurry indicate that a ceria-silicate surface complex indeed forms, which was pr edicted by the chemical tooth theory. And electron microscopy has confirmed the existence of Ce3+ ions at the ceria particle surface [Gill04], which are formed by reduction of the Ce4+ ions. The conversion of Ce4+ to Ce3+ ions is facilitated by the formation of oxygen vacancies [Sko02]. The reduction of ceria may be the driving force for the solid-s tate ceria-silica r eaction. Particle size

PAGE 135

121 measurements indicate that silic a is accumulated on the ceria particles as predicted by the particulate generation mechanism. Several ceria-silica interfacial phenomena ar e considered to contribute to the ceriasilica CMP mechanism: the extent of chemi cal reaction between the abrasives and the substrate, the magnitude of electrostatic inte ractions, the change in particle size with slurry pH, and surface chemical phenomena due to ceria surface properties, which lead to the formation of a ceria-silica complex. The en hanced polishing of silica by ceria may be explained in three ways. Figure 7-1 shows the silica polish rate along with schematic diagrams for the three mechanisms (che mical or dissolution, surface chemical or physicochemical, and/or mechanical) that may control the silica removal rate. ceria p H 024681012 Silica removal rate (nm/min) 0 200 400 600 800 1000 1200 2% ceria III III ceria p H 024681012 Silica removal rate (nm/min) 0 200 400 600 800 1000 1200 2% ceria ceria p H 024681012 Silica removal rate (nm/min) 0 200 400 600 800 1000 1200 2% ceria III III Dissolution Surface Chemical MechanicalHighest in regions I and III. Higher in region II. Highest in region II. Lowest in region III. Dissolution Dissolution Surface Chemical Surface Chemical Mechanical MechanicalHighest in regions I and III. Higher in region II. Highest in region II. Lowest in region III. Figure 7-1 Schematic depiction of pH-dep endent ceria-silica removal mechanism. In the chemical mechanism (top schema tic in figure 7-1), ceria-silica bonding results in dissolution of the silica substrate si milar to the dissolution of silica by water in very acidic media. The AFM wear measuremen ts presented in chap ter 6 indicate ceria does not chemically modify silica. Therefor e, any dissolution of the substrate may be

PAGE 136

122 attributed to the slurry pH during polishing. Si lica dissolution is expected to be highest in very basic or acidic media (regions I and III in figure 7-1). In the surface chemical mechanism, ceria-silica bonding results in in creased shear at the abrasive-substrate interface, which leads to higher removal rates. The reactivity of the ce ria particles results in an enhancement of the mechanical contri bution to polishing. According to the results of this work, the surface chemical mechanis m predominates in region II, the area of maximum polishing. Finally, because of partic le instability and enhanced abrasivesubstrate interaction, a purely mechanical m echanism is possible in which the increased number of abrasives available for polishing are responsible for the enhanced polishing rate. Mechanical contributions are expected to be highest in region II because of slurry instability and lowest in region III because of repulsive electrostatic interactions. Even for reactive abrasive particles like ceria, di electric CMP is primarily mechanical in nature. Suggestions for Future Work STI CMP will continue to be a fertile research area for the foreseeable future. The efforts in this study can be further extended by studying the effect of ce ria particle size on the silica removal rate, friction force a nd surface morphology. The ceria-silicon nitride interactions are also important for ST I CMP and should be studied. To enhance selectivity, the nitride removal rate should be low with respect to the oxide. This can be accomplished by inhibiting the abrasive-nitride interaction. Surfactant-based slurries are currently in development for STI CMP. However, the fundamentals of surfactant adsorption are not well understood. By studyi ng surfactant-nitride bonding, mechanisms of surfactant adsorption can be understood. Fourier transform infrared spectroscopy (FTIR) is a surface sensitive technique th at identifies chemical bonding. FTIR is a

PAGE 137

123 promising, well understood technique for in-situ characterization of surfactant adsorption on nitride surfaces. The possibili ty of reduced defectivity is an additional benefit of surfactant-based slurries. To tailor surfactant adsorption for desire d STI CMP performance, an understanding of the silicon nitride surface layer properties is required. The slurry chemistry can then be tailored to inhibit the surface modification pr ocess that favors nitride removal over nitride passivation. Recent studies have suggested that conversion of the nitride to an oxide layer is required for chemically assisted remova l of the nitrides su rface. Along with FTIR, spectroscopic ellipsometry and nano-indenta tion are ideal techniques for determining surface layer chemical and mech anical properties. This information is valuable for STI CMP with ceria-based slurries and fixed abrasive pads. Although ceria-based slurries have ma ny promising attributes, STI polishing processes that utilize fixed abrasive pads seem most promising. Further research into the morphology of the abrasive-polymer composite is needed. Also since slurry stability is not a concern with fixed abrasive pads, mo re extensive investigations into pH and chemical additive effects can be conducte d. The proposed studies in the two preceding paragraphs are also directly applicable to STI CMP with fixed abrasive pads.

PAGE 138

124 APPENDIX A DEFINITION OF COMMON STATISTICAL TERMS [MAS89] Bloc k – group of homogeneous experimental units. Confounding – one or more effects that can not be unambiguously attributed to a single factor Contour plot – series of lines or curvew that id entify values of the factorfor which the response is constant Covariate – an uncontrollable variable that infl uences the response but is unaffected by any experimental factors Design – complete specification of experi mental test runs including blocking, randomization, repeat tests, replicati on, and assignment of factor-level combinations to experimental units Effect – change in the average response between two factor-level combination or between two experimental conditions. Factor – a controllable experiment al variable that is though t to influence the response Homogeneous experimental units – are as uniform as possibl e on all characteristics that could affect the response Interaction – existence of joint effects in which the effect of each factor depends on the levels of the other factors. Level – specific value of a factor Repeat tests – two or more observations that ha ve the same levels for all factors Replication – repetition of an entire experiment or a portion of an experiment under two or more sets of conditions Response – outcome or result of an experiment Response surface – geometric representation obtained when a response variable is plotted as a function of one or mo re quantitative factors. Test run – single combination of factor levels that yields an observation on the response Unit – entity on which a measurement or an observation is made

PAGE 139

125 APPENDIX B BOX BEHNKEN DESIGN AND RESPONSES Experimental variables: A: Pressure (2-10 psi) B: pH (2-6) C: Rotational velocity (75-150 RPM) Responses: Silica removal rate (RR) nm/min RMS surface roughness (nm) RR = 436.0 + 141.0 A + 139.13 B + 151.13 C + 47.50 A2 29.75 B2 94.25 C2 + 71.25 A*B + 83.25 A C + 213.0 B* C RMS roughness = 0.31 + 0.21 A + 0.16 B – 0.019 C + 0.25 A2 + 0.21 B2 + 0.047 C2 – 0.16 A *B + 0.10 A C.

PAGE 140

126 Table B-1 Box Behnken design fo r three factor experiment. Std Run Factor A Pressure (psi) Factor B pH Factor C Velocity (RPM) Response RR (nm/min) Response RMS (nm) 5 1 2 4 75 180 0.5 14 2 6 4 112.5 440 0.3 11 3 6 2 150 90 0.35 12 4 6 6 150 810 0.8 7 5 2 4 150 342 0.2 10 6 6 6 75 108 0.7 15 7 6 4 112.5 438 0.32 2 8 10 2 112.5 405 0.95 9 9 6 2 75 240 0.4 4 10 10 6 112.5 810 0.9 13 11 6 4 112.5 430 0.3 3 12 2 6 112.5 360 0.9 6 13 10 4 75 270 0.8 1 14 2 2 112.5 240 0.3 8 15 10 4 150 765 0.9

PAGE 141

127 LIST OF REFERENCES Adl01 J. J. Adler, Interactions of Non-Ideal Sur faces in Particulate Systems Ph.D. Dissertation, University of Fl orida, Gainesville, Fl., (2001). Bas00 G. B. Basim, et al., J. of Electrochem. Society, 147, (2000), p. 3523. Bas02a G. B. Basim, Formulation of Engineered Particulate Systems for Chemical Mechanical Polishing Applications Ph.D. Dissertation, Univ ersity of Florida, Gainesville, Fl., (2002). Bas02b G. B. Basim and B. M. Moudgil, J. of Colloid and Interface Science v256 (1), (2002), p. 137. Bas03 G. B. Basim, I. Vakarelski, and B. M. Moudgil, J. of Colloid and Interface Science, v263 (2003), p. 506. Bas03b G. B. Basim, S. C. Brown, I. U. Vakarleski, and B. M. Moudgil, J. of Dispersion Science and Technology v24 (3&4), (2003), p. 499. Bey00 K. Beyer, Future Fab Intl. v9 (2000). Bie98 M. Bielmann, Chemical Mechanical Polishing of Tungsten, Master’s Thesis, University of Florida, Gainesville, Fl., (1998). Bin92 J. B. Bindell, in Encyclopedia of Materials Characterization edited by C. R. Brundle, C. A. Evans Jr., and S. Wilson, Butterworth-Heinemann, Boston, (1992), p. 699. Biz04 San Jose/Silicon Valley Business Journal, http://sanjose.bizjournals.com/sa njose/stories/20 04/08/30/daily2.html Last visited 10/16/04. Box78 G. Box, W. Hunter, and J. Hunter, Statistics for Experimenters Wiley and Sons, New York, (1978). Bro81 N. J. Brown, P. C. Baker, and R. T. Maney, Proc. SPIE, v306, (1981), p. 42. Bro84 N. Brown and L. Cook, Tech. Digest, Topical Meeting on the Science of Polishing, Optical Society of America, (1984), paper TuB-A4.

PAGE 142

128 Bru92 C. R. Brundell, in Encyclopedia of Materi als Characterization edited by C. R. Brundle, C. A. Evans Jr., and S. Wilson, Butterworth-Heinemann, Boston, (1992), p. 282. Bud61 S. M. Budd, Phys. Chem. Glasses v2 (1961), p. 111. Chan04 N. Chandrasekaran, Mat. Res. Soc. Symp. v816 (2004), p. K9.2.1. Che97 Y. L. Cheng et al., J. of Electrochem. Soc. v144 (1), (1997), p. 315. Chen00 Z. Chen, Silicon Wafer–Contaminant Interactio ns in Dilute Hydroflouric Acid Solutions and Related Fundamentals in Colloid and Interface Science, Ph.D. Dissertation, University of Flor ida, Gainesville, FL, (2001) p. 32. Chen03 Z. Chen, and R. K. Singh, J. Electrochem. Soc. v150 (11), (2003), p. G667. Cho01 W. Choi, S-M, Lee, and R. K. Singh, Mater. Res. Soc. Symp. Proc., v671 (2001), p. M5. Cho03 W. Choi, Study of Interfacial Interacti ons During Chemical Mechanical Polishing (CMP) of Dielectric Silicon Dioxide Ph.D. Dissertation, University of Florida, Gainesville, Fl., (2003). Cho04a W. Choi, et al., J. of Electrochem. Soc., v151, (3), (2004), p. G185. Cho04b W. Choi, et al., J. of Electrochem. Soc., v151, (5), (2004), p. G368. Cho04c W. Choi, et al., J. of Electrochem. Soc., v151, (8), (2004), p. G512. Cho04d W. Choi, et al., Electrochem. and Solid-State Letters., v7, (7), (2004), p. G141. Coo90 L. M. Cook, J.l of Non-Cryst. Solids, Cor63 Cornish, D., and Watt, L., Br. Sci. Instr. Res. Assoc. Rep. (1963), R295. Dav96 K. M. Davis, et al., J. of Non-Cryst. Solids, v203 (1996), p. 27. Der87 B. V. Derjaguin, N. V. Churaev, and V. M. Muller, Surface Forces, Consultants Bureau, New York, (1987). Dov92 P. M. Dove, and S. F. Elston, Geochim. Cosmochim. Acta v56, (1992), p. 4147. Eva02 D. Evans, MRS Bulletin, v27 (10), (2002), p. 779. Eva04 D. R. Evans, MRS Symp. Proc. v816 edited by, D. Boning, M. Voss, and A. Philloposian, Materials Research Soci ety Proceedings, San Francisco, Ca, (2004) p. K9.1.1.

PAGE 143

129 Ewa99 R. Ewasiuk, S. Hong, and V. Desai, in Chemical Mechanical Polishing in IC Device Manufacturing III edited by Y. A. Arimoto, R. L. Opila, J. R. Simpson, K. B. Sundaram, I. Ali, Y. Homma, Electrochemical Society Proceedings Series, Pennington, NJ, (1999), p. 408. Fran99 J. Franz, P. Erichson, and J. Kross, Optik v110 (9), (1999), p. 426. Gag02 J. Gagliardi et al., CMP-MIC (2002). Gar02 R. Garcia and R. Perez, Surface Science Reports v47 (2002), p. 197. Gil04 S. R. Gilliss, J. Bentley, and C. B. Carter, MRS Symp. Proc. v818 Materials Research Society Proceedings, San Francisco, (2004), p. M1.8.1. Gor03 V. R., Gorantla, et al., J. of the Electrochem.l Soc., v150 (12), (2003), p. G821. Gry93 G. W. Grynkewich, and J. N. Helbert, in Handbook of Multilevel Metallization for Integrated Circuits,” Noyes Publications, New Jersey, 1993, Ch. 7. Han90 W-T. Han, and M. Tomozawa, J. of Non-Cryst. Solids, v122 (1990), p. 90. Han91 W-T. Han, and M. Tomozawa, J. of Non-Cryst. Solids, v127 (1991), p. 97. Hie89 T. Hiemstra, W. H. Va n Riemsdijk, and G. H. Bolt, J. of Colloid and Interface Sci. v133 (1989), p. 133. Hir87 K. Hirao and M. Tomozawa, J. Am. Ceram. Soc., v70, (7), (1987), p. 497. Hog73 J. M. Hogan, et al., J. Cellular Plastics September/October (1973), 219. Hol64 L. Holland, The Properties of Glass Surfaces Chapman & Hall, London, (1964). Hom03 Y. Homma, et al., J. of Electrochem. Soc. v150 (12), (2003), p. G751. Hos97 S. Hosali, D. Evans, et al., CMP–MIC Conference ISMIC–200P, Feb. (1997), p. 52. Hos01 T. Hoshino et al., J. of Non-Cryst. Solids, v283 (2001), p. 129. Hou92 W. A. House, and D. R. Orr, J. Chem. Soc. Faraday Trans., v88 (1992), p. 233. Hun01 R. J. Hunter, Colloids and Surfaces A: Physicochemical and Engineering Aspects v195 (1-3), (2001), p. 205. Ile79 R. Iler, The Chemistry of Silica Wiley, New York, (1979).

PAGE 144

130 Ire97 P. J. Ireland, Thin Solid Films v304, (1-2), (1997), p. 1. ITRS03 International Technology Ro admap for Semiconductors (2003). Izu79 T. Izumitani, in Treatise on Materials Science and Technology, eds. M. Tomozawa and R. Doremus, Academic Press, New York, (1979), p. 115. Izu84 T. Izumitani, Tech. Digest, Topical Meeti ng on the Science of Polishing, Optical Society of America, April 17 (1984), TuB-A1. Jai94 R. Jairath, et al., MRS Symp. Proc., v337, (1994) p.121. Kal56a A. Kaller, Naturwissenschaften v43 (7), (1956), p. 156. Kal56b A. Kaller, Silikattechnik v7 (10), (1956), p. 380. Kal91 A. Kaller, Glastech. Ber. v64 (9), (1991), p. 241. Kal98 A. Kaller, Glastech. Ber. Glass Sci. Tech. v71 (6), (1998), p. 175. Kat03 T. Katoh, H-G. Kang, U. Paik, and J-G. Park, Jpn. J. of Appl. Phys. v42 (2003), p. 1150. Kel98 A. Kelsall, Glass Technology, v39 (1), February (1998), p. 6. Kim02 S-K. Kim et al, J. Vac. Sci. Tech. B v20 (3), (2002), p. 918. Kim03 S-K. Kim et al, Jpn. J. of Appl. Phys. v42 (2003), p. 1227. Kul04 M. Kulawski, MRS Symp. Proc. v816 (2004), p. K7.2.1. Lee03 S-M. Lee, Characterization of Chemical Interactions During Chemical Mechanical Polishing of Copper, Ph.D. Dissertation, University of Florida, Gainesville, Fl., (2002) Lin99 C. F. Lin et al, Thin Solid Films v347 (1999), p. 248. Mah99a U. Mahajan, M. Bielmann, and R. K. Singh, Electrochem. and SolidState Letters 2, (1), (1999), p.46. Mah99b U. Mahajan, M. Bielmann, and R. K. Singh, Electrochem. and SolidState Letters 2, (2), (1999), p.80. Mah00 U. Mahajan, Fundamental Studies on Silicon Dioxide Chemical Mechanical Polishing, Ph.D. Dissertation, University of Florida, Gainesville, Fl., (2000). Mar03 H. Martin, B. Bournique, B. Blanchi, and C. Lerche-Langrand, Analytical Biochemistry v319 (2003), p.56.

PAGE 145

131 Mar03b A. Martin et al, Mat. Res. Soc. Symp. Proc v767 (2003), p. F5.10.1. Mas89 R. L. Mason, R. F. Gunst, and J. L. Hess, Statistical Design and Analysis of Experiments with Applications to Engineering and Science, J. Wiley and Sons, New York, (1989). Mou95 J. F. Moulder et al, Handbook of X-ray Photoelectron Spectroscopy: A Reference Book of Standard Spectra fo r Identification and In terpretation of Xray Photoelectron Spectroscopy Data, eds. J. Chastain and R. C. King II., Physical Electronics, Ed en Prairie, MN (1995). Mur93 S. P. Murarka, Metallization Theory and Pr actice for VLSI and ULSI, Butterworth-Heinemann, Boston (1993). p. 100. Mur94 S. P. Murarka, and R. Gutmann, 1994 Annual Report of the New York State SCOE, Semiconductor Research Corporation, Research Triangle Park, NC (1994). Nab93 M. Nabavi, O. Spalla, and B. Cabane, J. Colloid and Interface Sci. v160 (1993), p. 459. Ngu01 V. H. Nguyen, et al., Microelectronic Engineering v55 (1-4), Mar (2001), p. 305. Obe98 Y. K. Obeng et al., Electrochemical Society Proceedings Series, 98-7 151, (1998). Ols93 J. Olsen, and F. Moghadam, in Handbook of Multilevel Metallization for Integrated Circuits,” Noyes Publications, New Jersey, 1993, Ch. 6. Oss02 K. Osseo-Asare, J. of Electrochem. Soc. v149 (12), (2002), p. G651. Pai88 P. L. Pai, C. H. Ting, and W. G. Oldham, 5th VMIC IEEE Santa Clara, Ca., Cat. No. 88CH2624-5, IEEE, NY (1988), p. 108. Par04 S. Park, IEEE Transactions on Semiconductor Manufacturing, v17 (3), (2004), p. 362. Par04b J.-G. Park, T. Katok, and U. Paik, J. of Mater. Res. v19 (6), (2004), p. 1783. Pour66 M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions Pergamon Press Ltd. (1966), p. 194. Pre27 F. Preston, J. Soc. Glass Technol. v11 (1927) p. 247. Qin04 K. Qin, C. Park, and B. Moudgil, Thin Solid Films, v446, (2), (2004), p.277. Ren94 P. Renteln and J. Coniff, MRS Symp. Proc. v337 (1994) 104.

PAGE 146

132 Sab00 R. Sabia, and H. Stevens, Machining Science and Technology v4 (2), (2000), p. 235. Sam73 The Oxide Handbook edited by G. V. Samsonov, Translated from Russian by C. N. Turton and T. I. Turton, Plenum, New York, (1973). Sch95 J. Schlueter, SemiCon Southwest 95, (1995), p. 1. Sch99 J. Schlueter, Semiconductor International, October (1999). Sik01 A. K. Sikder, et al. J. of Electronic Mat., v30, (12), (2001), p. 1520. Sim01 A. Simpson, L. Economi kos, F-F. Jamin, A. Ticknor, Mat. Res. Soc. Symp. Proc. v671 (2001), p. M4.1.1. Sin98 P. Singer, Semiconductor International June, 1990. Sin02 R. K. Singh, et. al., MRS Bulletin v27 (2002), p. 752. Sin03 R. K. Singh, Slurries for Chemical M echanical Planarization (CMP) Technology Short Course Particle Engi neering Research Center, Summer School in Winter, (2003). Siv92 S. Sivaram, et al., Proc SRC, Topical Research Conference on Chemical Mechanical Polishing for Planarization, SRC, Research Triangle Park, NC, (1992), p. 8. Sko02 N. V. Skorodumova et al, Physical Review Letters v89 (16), (2002), p. 166601-1. Ste92 F. A. Stevie, in Encyclopedia of Materials Characterization edited by C. R. Brundle, C. A. Evans Jr., and S. Wilson, Butterworth-Heinemann, Boston, (1992), p. 699. Ste97 J. M. Steigerwald, S. P. Murarka, and R. J. Gutmann, Chemical Mechanical Planarization of Microe lectronic Materials Sti97 G. Stix, Scientific American v276 (1997), p. 45 Stu92 W. Stumm, Chemistry of the Sold–Water Interface Tro94 J. A. Trogolo, K. Rajan, J. of Materials Science, v29 (1994), p. 4554. Tse01 W-T, Tseng, et al., Electrochem. and Solid-State Letters, v4 (5), (2001), p. G42. Tye04 H. Tye, Drug Discovery Today v9 (11), (2004), p. 485.

PAGE 147

133 Ver48 J. W. Verwey and J. T. Overbeek, Theory of the Stability of Lyophobic Colloids Elsevier, Amsterdam, (1948). Vie01 N. H. Viet, Ph.D. Dissertation, Un iversiteit Twente, Netherlands, (2001). Wal96 W. Wallace, W. L. Wu, and R. A. Carpio, Thin Solid Films, v280 (1996), p. 37. Wil 93 S. R. Wilson, C. J. Tracy, and J. L. Freeman, Handbook of Multilevel Metallization for Integrated Circuits,” Noyes Publications, New Jersey, 1993, Ch. 1. Zam03 A. A. Zaman, F. Demir, and E. Finch, Applied Clay Science v22 (2003), p. 237. Zeta Colloidal fundamentals, User’s Ma nual Zeta Reader by Brookhaven, p. IV-1. Zha99 B. Zhao and F. G. Shi, Electrochem. and Solid-State Letters v2 (3), (1999), p. 145. Zha01 E. Zhao, and C. Shan, Semiconductor International June (2001).

PAGE 148

134 BIOGRAPHICAL SKETCH Jeremiah Terrell Abiade was born Jeremi ah Terrell Grey to Candy and Renee Grey, September 22, 1976, in Chicago, Illinois. Jere miah attended Southern University and A&M College in Baton Rouge, Louisiana where he received a B.S. in physics. There he was a member of the Timbuktu Academy a nd recipient of the NASA undergraduate student award for research. Jeremiah began his graduate study at the University of Florida in 1999 under the tutelage of Dr. Rajiv K. Singh and obtained his Ph.D. in 2004. Jeremiah’s graduate study was supported by fellowships from the National Science Foundation’s Alliance for Graduate Educa tion and the Professoriate Program (NSFAGEP), UF College of Engineering Graduate Minority Fellowship Program, UF Office of Graduate Minority Programs-Supple mental Retention Award, and the NSF Dissertation Fellowship.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20110115_AAAABJ INGEST_TIME 2011-01-15T11:13:58Z PACKAGE UFE0008394_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 598 DFID F20110115_AAAZXC ORIGIN DEPOSITOR PATH abiade_j_Page_128.txt GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
40a2f6b4c9d517df94d4a1feb8eaef15
SHA-1
91f3dc7c8fbcb8e4eddf98a3592338ca8acb65c2
2280 F20110115_AAAZWN abiade_j_Page_109.txt
dd8a379fde8e7112d9e8bfd3f349a43f
bd2b6b387d9e2eeb4de5b37eb9fc8766c6df7aa0
166166 F20110115_AAAYSX abiade_j_Page_060.jpg
ef18c675d2c4522eebc8e89b105daf37
cbc4f64df4c95f8d356c0e3ffa5c20a61efb1aa8
1809 F20110115_AAAZVZ abiade_j_Page_090.txt
3f6907b6c4d8d263feecb41dfda1da18
edd8fa50bfb563013e70a3e3bbc3dfa85b153793
69715 F20110115_AAAYUA abiade_j_Page_114.QC.jpg
88f8a6ca5b8d6027b41aa27db890620f
e949d618245796f24ab33a7da0069147ea866169
1512 F20110115_AAAYTL abiade_j_Page_120.txt
9aab3b3f61556f3cae8a6d53e3c64cbc
c17bc40ef005bb2de8a58653404a9fece71fd1a7
1270 F20110115_AAAZXD abiade_j_Page_129.txt
5d4fe521a3823c8a3fe359458f0a7445
a1cd4748767d7e3593c460e8e025453cacaef973
1994 F20110115_AAAZWO abiade_j_Page_110.txt
33fb9a779330f7583649434ba48728ee
a8e7b343273c2f32ae054f080550dc5f78fb6cce
85064 F20110115_AAAYSY abiade_j_Page_108.QC.jpg
44353620a985a58894569340d48c8e9f
549592104140eff8f2665874ea03c4abc05bf303
69523 F20110115_AAAYUB abiade_j_Page_082.QC.jpg
915f8a7aa3061a1f19472ae3b0f1a012
852ff23ef19042fbbffda8e7241aac10baef0a01
1053954 F20110115_AAAYTM abiade_j_Page_070.tif
562c03142f7fdd089d7287c50e8f6d7a
25769196726da08eba02c3c576065472ab68874c
1262 F20110115_AAAZXE abiade_j_Page_130.txt
78ac1ce870fbbd895f2bea0210a77e46
8db1e4a2387435420cc73b2df5ab291c4afeaa64
2094 F20110115_AAAZWP abiade_j_Page_111.txt
1359ad0aa5fc8342722a29f5138f18b1
cb6131a5196cb2885eb8abbc2f319f656ab6d5e1
48873 F20110115_AAAYSZ abiade_j_Page_107thm.jpg
105f7dd49787056a70ec2a2e96e0c953
a7e6aae02c72e3345fbdd13d51f5151b7def3d21
1964 F20110115_AAAYUC abiade_j_Page_056.txt
9becdff20a8de4d6bafa136488779ba9
a35ae880becb44c19775e0e8d74b216d8a1cbf81
35316 F20110115_AAAYTN abiade_j_Page_125.pro
434ce1645f154f7e7f20ccb8bb2cf957
488d8102f5495b583955286a509b5b7aa3d1741a
1773 F20110115_AAAZXF abiade_j_Page_131.txt
ff431821eacb4340038bd35dda3ac758
2c7b04517c0d3a3aae2912166f89d5b3eb1f9f3d
704 F20110115_AAAZWQ abiade_j_Page_112.txt
82c14d1e5db1c568eca5dfb6d10d87cb
79608f5845cbeb42600461aa0c1fa62597253878
16644 F20110115_AAAYUD abiade_j_Page_012.pro
e3b2c948cb0c1c643e220beb717db621
0a780c13be9a7395eaa6d058b59e148f542660f7
81614 F20110115_AAAYTO abiade_j_Page_096.QC.jpg
45a2fd293095a395c56d2d2f0217d3f4
7666c927e426abfc7078ae7f0f2e45967f41f048
1901 F20110115_AAAZXG abiade_j_Page_132.txt
f5a13828bb05a9e9272628292645b47c
5486b79475e41e42d0796e8e4a2c9cbd2c4fbd8c
1691 F20110115_AAAZWR abiade_j_Page_113.txt
6d7303ab75d7619425ba81ca71ca7152
00cdabcb2044ca6c8ca995ef756774844208555b
120455 F20110115_AAAYUE abiade_j_Page_007.pro
4d6c5e8944b1fa5f05dc27ed7995db32
26b582aa01ff1f9f43f64a072945cc707defc2eb
87927 F20110115_AAAYTP abiade_j_Page_141.QC.jpg
37e462454678b9038c8b7b2459c1bc78
4381f3d50ebc082d7c5c3c967dc5775bd27bffe4
2061 F20110115_AAAZXH abiade_j_Page_133.txt
b3d195f0d85909d3e5708220d21739bd
2e9b4ae33c845ce4860d8ea4423b284f16e16b67
1872 F20110115_AAAZWS abiade_j_Page_114.txt
093a204832239bd15a20ba64f943de11
7911eb05f3278604fdb8a5b62ff698cbac117315
23125 F20110115_AAAYUF abiade_j_Page_020.pro
fe72e33171bfaf562e42055806e54f0a
a70649113de3d05f8cca265c4b454e720cca6ae1
517735 F20110115_AAAYTQ abiade_j_Page_103.jp2
fb88330cc4ec3311be08843eef213a67
fac60d43dec2a3be3a1d78a56c51534736145828
1342 F20110115_AAAZWT abiade_j_Page_117.txt
fed9a442b1943e49bb61bb715d509413
bb9b3a96a75afb6ce2fc4ae14bedc34d934905b5
F20110115_AAAYTR abiade_j_Page_119.tif
4b8f746677f00f6d6686e73aea3cea89
aeeeb15640f001bf80548687ed15bf62058ba321
41245 F20110115_AAAZAA abiade_j_Page_037.pro
0bea111161d69f4f467d30309c661fd1
788bb70e88424a4ab866be530c424b1c9ed51986
2044 F20110115_AAAZXI abiade_j_Page_135.txt
218849a241f028149975c0414ea79040
293716c5ab9e7bcb92c41513642c888f837d1f4a
2053 F20110115_AAAZWU abiade_j_Page_118.txt
3438993e98ffd3fa30b9a7da2a963db2
7ab1a87b4c6495604295dd7a6d7ed5729a1a1942
111248 F20110115_AAAYUG abiade_j_Page_048.jp2
f5aadfc5fcd2afdfa069c522984257a1
4a4042f9b8bf0240e72abefa61ccd8e44ff2102e
F20110115_AAAYTS abiade_j_Page_125.tif
0343315069f8994bd733e8e1183ee79c
2ac0dd769aef1bb42a77e47d3a491f2f44d254da
63263 F20110115_AAAZAB abiade_j_Page_139.jpg
71cf63ef0b84af58706ec9d0309b91d8
8d67e11b01ca266be5a2c3e98d678fbce43fe47c
1814 F20110115_AAAZXJ abiade_j_Page_138.txt
a8a4e25760c0f46b2174889ed75770e9
b9fd324b8fe72a6262c0fe1b5540e0aa78fad045
2008 F20110115_AAAZWV abiade_j_Page_119.txt
46eda1b4e5844b86b9065a59c510e335
a2659002e23d7ce36347bc8e669975d91d8dc6b8
44952 F20110115_AAAYUH abiade_j_Page_004.pro
61f4459e412727a2de16c18053f2a345
08fb3bd0d0613a088741ee5197111cb1bab0b495
F20110115_AAAYTT abiade_j_Page_078.tif
ddc3325098f74f9f4ee8842aaaf9b54d
c314e0d1e895c152aa00cbef644ef6a9b074000f
20425 F20110115_AAAZAC abiade_j_Page_113thm.jpg
4fec9d3fe1992a3eb241e2c670841c7b
171cd477e34118a385a4fb278f40360b62022c6e
1314 F20110115_AAAZXK abiade_j_Page_140.txt
ba4298e86d965030da29355bac27b168
a56c791b2ffcc97f03e882645d4647249150b957
2107 F20110115_AAAZWW abiade_j_Page_121.txt
649166c20cbf1cd1e58ac0013edfa848
d3fcc75d2e8b7af538dae49978fb8c4da1888a85
F20110115_AAAYUI abiade_j_Page_042.tif
9ec2cc3089215be28f9500b9d63bfa19
207505a5ce0ff9106db105bd94dad35225cf67c1
241279 F20110115_AAAYTU abiade_j_Page_008.jpg
d844d75c9a61c77b42e1de11be75af1a
01ad0ef108bdbcd7982c01f1427a89f2aabab4f0
1041055 F20110115_AAAZAD abiade_j_Page_141.jp2
58529750ef91b47a2655a7da83d77755
381e614ff78450a99816eb193374ad4be9c67aa8
25261 F20110115_AAAZYA abiade_j_Page_028thm.jpg
a45594921db6e21d723ec55840ac1b7b
52ceb3358840f8caad830ce26955de9aa3d381c0
1849 F20110115_AAAZXL abiade_j_Page_141.txt
938893d4b20a34a883663093eb2e3dd8
7132839dd40ae4b3f58900992d072e693e5c1624
1570 F20110115_AAAZWX abiade_j_Page_122.txt
3c6e44759f3ce1f2ae3f484a4e2b2754
64c127f6911cb89af39bef603ffc074f229c937e
25271604 F20110115_AAAYUJ abiade_j_Page_123.tif
81bbd34a397b996d43f460954badb5ac
105917d151cafac756fb83d3ce3869e36ca3b293
801313 F20110115_AAAYTV abiade_j_Page_117.jp2
473ba6516dd2ec98f1dd83c931afa877
b65707c9820f3d17f09a8fa6fc782922d5237a57
631357 F20110115_AAAZAE abiade_j_Page_068.jp2
151f455df84ce55c328458f400a4f7a9
71592e50fed6013ab816f4c5341a22b7e6d4b05c
13593 F20110115_AAAZYB abiade_j_Page_148thm.jpg
614aff5415bc99c58d147be9df6c377f
f8f69de2da03843de32d09a76c255707e6122d30
2142 F20110115_AAAZXM abiade_j_Page_143.txt
40dc2ee56f027cf96833e25fe471924a
97cf2736eb5e059152972329b9ca8e7bbf6a1f45
1895 F20110115_AAAZWY abiade_j_Page_123.txt
34af9daffd84df95185a7f7ad4ece514
d42037035c713672ccfcdd62d6e1e16185a202f0
22350 F20110115_AAAYUK abiade_j_Page_082thm.jpg
7fecf528592795bce2ad63663679c791
2f965bed191436c0c06804aba6e2ee546b179a5a
108936 F20110115_AAAYTW abiade_j_Page_079.jp2
f77bcaaed79b9ca98c8db843eea60d19
50681bcfba6292f291c366431914c90486ed35d5
484 F20110115_AAAZAF abiade_j_Page_001.txt
1593c7c09de894ebc956205f4bc37bfe
7df2ef7e7c35546190123dc1fbaaa76e3377ff2c
81288 F20110115_AAAZYC abiade_j_Page_092.QC.jpg
8b735c4469c6f3c140a29f883d5a2978
edea4ac09a491e422d78619719ee5c284b235f30
2070 F20110115_AAAZXN abiade_j_Page_144.txt
cc1c6f0a4ae2810490df3b2ed7f1bf8f
7ee9ce5a9f36aced5322615ed94c85cc2c5f935a
984 F20110115_AAAZWZ abiade_j_Page_124.txt
0a3edbbd43f874d16c0092bfd857cb26
a65782fbaaf6ee4f46e54aa53d36c79c8bedd517
48244 F20110115_AAAYVA abiade_j_Page_043.pro
d512d72d27a0b5feb29a9b8a57c1e099
1ed226af0af4a726d1cf9a0636fdc8572f9eac36
2261 F20110115_AAAYUL abiade_j_Page_084.txt
7c8610da1dabe35745f060f3a27daa2f
1dfac2098ff7138ca590549580e6aba5dfd7e7b3
189110 F20110115_AAAYTX abiade_j_Page_081.jpg
05e90da227a0db75127b4d0d7921463a
97baa0174dcd000e98f58e20c2c6f54edc7a1e3d
F20110115_AAAZAG abiade_j_Page_025.tif
9d25b8fedea7f4c9584a153849994aba
32cb2d6aa5cbd95496d6c296cc1278057e77b25f
50763 F20110115_AAAZYD abiade_j_Page_120.QC.jpg
6ad2b31473e47b1ad623e07fe983f5b1
c44fd03e909b6c672743ecbfd40d25f2f3d0fbc2
2263 F20110115_AAAZXO abiade_j_Page_145.txt
06fa8cd38127a964f23ffc605dd75375
b6ff57a363464d513234a8b74d26bb521a18f6e4
22709 F20110115_AAAYVB abiade_j_Page_148.pro
347e607bdf248326993e1c019c9f5c7c
06ad8d7b199b9fd7a17d64e06f40e02253607cce
1562 F20110115_AAAYUM abiade_j_Page_107.txt
ab81c22ce9d70e92525a1e05dc0ed8e5
5f3f5696deb15f71eabda222cfa57c8353e7d20f
213885 F20110115_AAAYTY abiade_j_Page_136.jpg
fdf93ae30cc7b3731ab10bcdbfa91164
a18252941a5c9d4e967b56ba9300318b0c51fdb8
24992 F20110115_AAAZAH abiade_j_Page_016thm.jpg
937afef79aac8869e44ae6fb6399702c
310568a602fa856af80e03efd5ca7cd6b13bd541
5038 F20110115_AAAZYE abiade_j_Page_003thm.jpg
4074fc0a6392c70476299c1b0dee3734
2fef8a9eee79c4180259c05c76ec109c5835ed26
1982 F20110115_AAAZXP abiade_j_Page_146.txt
b43ec80aae8eab3aa19ee408dcecc06b
3e2f8923308c3f6f11022a9c235d2e7671288c7f
25621 F20110115_AAAYVC abiade_j_Page_035thm.jpg
0726bdc20ab4884400310c0dead77dbf
b3f99fec45c501eac4c4abf5ff275fc4b2530cbb
187911 F20110115_AAAYUN abiade_j_Page_075.jpg
b13ce1ef422f647890b10554058d7043
2d84bd09bbda2e4397f6e99733c606034aa2e898
48078 F20110115_AAAYTZ abiade_j_Page_025thm.jpg
145152a29f645856e34fbde1df710529
84a07c1eb113fd59e112413d0493467b8618b3a1
F20110115_AAAZAI abiade_j_Page_106.tif
df8495c7833192ceaacb90ee130708ad
68632ae4d4dd407f72d82203e15b0ace74ffea15
80702 F20110115_AAAZYF abiade_j_Page_118.QC.jpg
194a8a22e4dbaa9712ca53a3c2e4449c
36476a8ebefe52164ce4fe615984ae4281561b5f
950 F20110115_AAAZXQ abiade_j_Page_148.txt
bb0605776920fb933c1bcb4d06fcbc2b
9b075b00c7620b3e25b31449256ad0e0357bf2d4
69740 F20110115_AAAYVD abiade_j_Page_130.jp2
5b334a5cb38186e14758fb799db223fc
b1508a6e1b3837a8e33506bf7270d7efc1c1056f
111837 F20110115_AAAYUO abiade_j_Page_101.jp2
976db6a191b1287286f4a3a50ca9570c
dca7c7f7e1c87ef10e919f31dc51a046977db2b4
83893 F20110115_AAAZAJ abiade_j_Page_038.QC.jpg
1a1219914e2f887cf078603f94c09b90
d6ca154d4d2557606909ec25d918db8f22a7b13d
48107 F20110115_AAAZYG abiade_j_Page_043thm.jpg
66da1adb428d5d244a513fabb2aad0b4
a6fe5ea64587f1452ef9d0833e06a87991d2ccd6
7541 F20110115_AAAZXR abiade_j_Page_001thm.jpg
a5ae617266adf100aec7ebdefd4dc0bc
a15dbdf8cc34ac1d93680aff99ca6d6eb08abea0
25114 F20110115_AAAYVE abiade_j_Page_033thm.jpg
c85db1aba3d4c86287c22581d1e10903
3cf60f80bfa321633f0a3ce8dac2c0a98a1eb587
94756 F20110115_AAAYUP abiade_j_Page_135.QC.jpg
9d8d06e2bcff4511a4b198c4d52139c9
fbfe1732387c3d0b7da000d9c09fcf7b9a94b6b3
578 F20110115_AAAZAK abiade_j_Page_139.txt
dde156afb3017bdaaacaaa3d94feb02f
12b2b4bb4fce50919359982745a4d6d3785bd70c
77611 F20110115_AAAZYH abiade_j_Page_049.QC.jpg
d265f92ecabde6739369880b0e5bcba7
dcca8be1d82fed8f37492b66c468a87f9a2a01f0
79580 F20110115_AAAZXS abiade_j_Page_093.QC.jpg
8867381743d30b6cfda3f33795388288
baf23f229f803ba3da112edbd6d59872d61bfc96
1970 F20110115_AAAYUQ abiade_j_Page_077.txt
ce5fd2b25461c96f5c2e5e33d8edcf3f
9f9a81ec9e34b651d607d1ac0135e9296ac8f2aa
3582 F20110115_AAAZAL abiade_j_Page_006.txt
91f858ddf0c39367c1f0a0556e7f691a
c5aeaa6a218f5be0db175248523a107b110a8c65
25067 F20110115_AAAYVF abiade_j_Page_080thm.jpg
516d3ede9a1ecd0f9ab50f34e6b61e57
7ca32d984b66267cba34c04bd6428ffca839e980
25521 F20110115_AAAZYI abiade_j_Page_044thm.jpg
41264c2b4bc53c8bca1661e0715fcb42
25843797dc36695859aa6e5fba869bfa122d5487
40553 F20110115_AAAZXT abiade_j_Page_103thm.jpg
7a0b1d929ee506d7a625b515ec234bb8
87d7ade5f3e570b674cef69632cb30fb59b08b6b
1903 F20110115_AAAYUR abiade_j_Page_034.txt
3c4f24a27f0b83d23e3585171bb48242
65c0dfa1d29a818c49bad4ffd895c005d7cacc3c
99764 F20110115_AAAZBA abiade_j_Page_065.jp2
5bd98c3d9aee219096334bbc7fb2fb76
c7a4e20296aa266592fd8f63de4cfbb7f23672d5
F20110115_AAAZAM abiade_j_Page_019.tif
24bdc53566ba4e9a266511e62a83d771
c5b3a79ac44fa2ae6807dae93688cacb14f35eda
F20110115_AAAYVG abiade_j_Page_039.tif
5a29c6807c3658288b04f38ee430d9c9
862518d1365efec28d22cb5e9f018d2bd1f19fb1
23773 F20110115_AAAZXU abiade_j_Page_056thm.jpg
22ad0be668453a0e68cc6da99332fc7b
4035e97404f4dd4fbb202f39788b83dfdc7aa875
1611 F20110115_AAAYUS abiade_j_Page_085.txt
b5585e65c63840e6c1a199ed3eabfc82
3293aeee520106a79716056a2a614c7a72f81308
1051982 F20110115_AAAZBB abiade_j_Page_007.jp2
5de248f920f95e5c432bb2f6250939f5
7b65338a359f39dfeeade5b268bca5481a285a21
51240 F20110115_AAAZAN abiade_j_Page_097.pro
9583709b4e31ea4d74c234508ecc9214
edfbf08940796c092864e3e6f446fbe665c2b577
45298 F20110115_AAAZYJ abiade_j_Page_098thm.jpg
859138425812b90b807a98bc07f95793
bd1cd6529de4c81b5c603d8ebfeb1d6a39e30457
48622 F20110115_AAAZXV abiade_j_Page_094thm.jpg
a40470ed46cdd0abf739498e037d180a
7f9e110f4aea5d19787b64bff9ebec4bd021fdd9
1527 F20110115_AAAYUT abiade_j_Page_083.txt
72ffd26c5a0a7e0086f58548c1f97e6f
61641a5fb21eb79660c531efb0a3e5d20a6021ab
F20110115_AAAZBC abiade_j_Page_142.tif
abdd8c30984acbc7113d97aaaafa4f4e
01448fe2fa63d79cab5b4f08648576e5d51b6100
50819 F20110115_AAAZAO abiade_j_Page_143.pro
d05bc140e6e6dcdc3ed2d39fdf1ded05
bd7dd8b219f0301ad6a714e1c5ed0b853c648f99
217729 F20110115_AAAYVH abiade_j_Page_101.jpg
c867a611c38123fdd15d8d6eedb84c5a
e7a68f03f8f2577e10739ab0c879977810662388
33733 F20110115_AAAZYK abiade_j_Page_147.QC.jpg
0ee576c60eeb274d43761e6de17eeddc
8cca5675e875d033bc508ee772791a8d48956d41
47546 F20110115_AAAZXW abiade_j_Page_058.QC.jpg
49ec0d3f57885928314cf44c335dd1de
df56362b15f3032cc67495d8f965b1bb8271ed51
24250 F20110115_AAAYUU abiade_j_Page_143thm.jpg
6e3c5d1e25fc816c075279d6715dcf15
2ebf3775410a60f4aafc3b450425db672382f1f5
81704 F20110115_AAAZBD abiade_j_Page_016.QC.jpg
aa41b8ab445f5e785de6a09499ec7bb6
45a6cc41778bcc542196b2ec11cde9bb9fa9646c
24445 F20110115_AAAZAP abiade_j_Page_115thm.jpg
33299526731598ac706c5497d81266cb
10b7131cb109bb03e55c0d36f74fe07952e11017
24133 F20110115_AAAYVI abiade_j_Page_046thm.jpg
d819ad9fff6ddd49f3b88dc7673e3325
4b613682db21df079894970ac14193df6820f225
80305 F20110115_AAAZZA abiade_j_Page_017.QC.jpg
c4e5e90f8eae006076de78be1cf9a144
47d65eee51800c584b40a6d28110a2647a71f381
24887 F20110115_AAAZYL abiade_j_Page_051thm.jpg
3914d6642b7dcb6ef8a94f4403beefde
f8c2a546592ef96753b8654e198d9030ecd558e9
81449 F20110115_AAAZXX abiade_j_Page_032.QC.jpg
92189cae7d4c27c891bebab4fdb9aaf7
d3407e59e454bbd9b4b866d33bc599eebb6cc5f9
F20110115_AAAYUV abiade_j_Page_009.tif
2ac781bb2a9838817e993e336e10ad8f
7c71cda7950fc6474bfa0df7f74adc9a0bfd8a75
43171 F20110115_AAAZBE abiade_j_Page_084.pro
708ccd9b046d14b8bf0312d41c35f27e
f64307ab7b505e4667e05b48b91b9ecaa3cfec9f
117590 F20110115_AAAZAQ abiade_j_Page_011.QC.jpg
1aae3db43a34b26b1d3449796e7f0284
10aa03b9d319c2400bed9df93d92e135a64cd7e1
917719 F20110115_AAAYVJ abiade_j_Page_100.jp2
a554b78e4e856463a6a954f3ee9fc5e2
8b8d6f06ad62e5bd4b2aa2647c74e90f7e0951ba
44053 F20110115_AAAZZB abiade_j_Page_128thm.jpg
6f8686474da69f92eb723aea3d11e278
467f79769b419ec01aa5601e4ecead3edad83957
46163 F20110115_AAAZYM abiade_j_Page_021thm.jpg
7ac37c11cf50728ab289253fc6f9b86f
62d908f772d2a8da6f7c961a4ee65933aa2d68ad
24533 F20110115_AAAZXY abiade_j_Page_132thm.jpg
fe5474ceb9f862633548734423701eab
e5b7ee655db783b93d265a2d2a72c017ea89620d
97951 F20110115_AAAYUW abiade_j_Page_116.jp2
b70de9e44991450cf3cf153c2367e64a
a40f93d2b18889f485ed87bb69efb23b0992cfe2
1521 F20110115_AAAZBF abiade_j_Page_015.txt
b05e4a1e6a15845bcf94f8790ad3d106
7f56930b5d2439094f6f5d2c01109a5bf872913d
23853 F20110115_AAAYVK abiade_j_Page_134thm.jpg
cd2894f314b5cc8600c11bd3a960b032
7dd6da07365bde17b85501b7a330e018bd7a2e20
16103 F20110115_AAAZZC abiade_j_Page_130thm.jpg
656653d15c4f74324614ee87e2fc9e4e
ac58cbf66e5b1503883be5905094bbd166d3972e
23010 F20110115_AAAZYN abiade_j_Page_144thm.jpg
b6893d5e7b2d151bc632cc59121b62c0
9e5c8cd02e95d84226380e5640f7196cf7272d66
81384 F20110115_AAAZXZ abiade_j_Page_015.QC.jpg
8db26a1f4d335b796bcada55bb5023c7
ea3665b035a0d963b66302425adfbf9531ddf1c7
199734 F20110115_AAAYUX abiade_j_Page_123.jpg
7721066b5f93115bd5fcddf42582c989
c33429fb9b2f11375606edac7c86e9697bc1e2c7
6074 F20110115_AAAZBG abiade_j_Page_002.jp2
5f99d041dbe59483a1711951df2464a9
e0d358a93c63e900822ae1f3bcd7634987aaabd3
66062 F20110115_AAAYWA abiade_j_Page_103.QC.jpg
8fbe30002808cd25ba9d6a3696a81e85
3931e226358e94b0e824f640e2af39499953fe77
84149 F20110115_AAAZAR abiade_j_Page_100.QC.jpg
c59e196d5c669c6a97d500a004fada6e
68b3649d2448c3593d59365e00a69c3264819bdb
F20110115_AAAYVL abiade_j_Page_148.tif
b487eaf984c689de6672c2757401c5ff
8cdf6fdc1db1ed37a1b9734b104e2807d411df26
12997 F20110115_AAAZZD abiade_j_Page_020thm.jpg
daa9cf14606e3e7e756d2373676d6383
4d6d2b53dc38fb7f76af73895fbcee3ac4800055
9997 F20110115_AAAZYO abiade_j_Page_112thm.jpg
f93951dc3a968c1dea37f03f6846ad7c
0ff2693cfd85746ee3d288c69fea967d9d291ff3
17777 F20110115_AAAYUY abiade_j_Page_005thm.jpg
a6b9c3b3e04d2bb0bf9364fc75f38755
481abcb469106e6df9714b41d246efb7a945dcc5
F20110115_AAAZBH abiade_j_Page_035.tif
cf14c388d8d072c37953d872d6114888
b46f4b39e3858f56386d1e396ab17c706a21a4db
72113 F20110115_AAAYWB abiade_j_Page_066.QC.jpg
1412cd648ba54d42ec5efbeda0d9c2bc
ac2030c9138498d18544921963919f1d111875b9
94099 F20110115_AAAZAS abiade_j_Page_082.jp2
d438c6d3d3d911dbf45b77a11c964cfd
6cc98ccbad75e9c05b83a8b6c32bdbbc3a131638
658894 F20110115_AAAYVM abiade_j_Page_088.jp2
dad08c9de915c3483ecce8e8eb764f6b
eb20df98a74f9baf9dab822705d025aab962f411
25829 F20110115_AAAZZE abiade_j_Page_133thm.jpg
ba3c2a0a6f1ee8d49c1bf80b9f9b0e58
4c5c3e91e0c3b8ec8972a9bfbebaed6a6aa30699
24864 F20110115_AAAZYP abiade_j_Page_048thm.jpg
a4f740d9a786c6d71a6fbeb9feb9d760
1efcb7080a30991bd2f9a9c9e0f1feee19134224
42722 F20110115_AAAYUZ abiade_j_Page_131.pro
9b5d2818be74fdd87de717d69e57a567
845d8cffbcb3bb9394a1fcf15469357461030fb2
25443 F20110115_AAAZBI abiade_j_Page_096thm.jpg
d15e9908d7b60fdee5333e2dbc298056
416567f73920444fe3e84205d4d6697a3d30bf6d
825804 F20110115_AAAYWC abiade_j_Page_087.jp2
2d6daccc624ec1af26afdd8fe8db32a9
36deabceda8924adfff00caf41fd98f67dfea18d
81156 F20110115_AAAZAT abiade_j_Page_137.jp2
5d7fc5b31bc3cef153c407dc7b6b10a8
882b2413d4db391a8264d8325a14a60920c8a1fd
46490 F20110115_AAAYVN abiade_j_Page_022.pro
59d47c6e3c40b494e1ae2a44dd0038c2
85dcbe5220eed1dbaf588c5b5c00442782bf7275
45054 F20110115_AAAZZF abiade_j_Page_099thm.jpg
f15026efc130410ec1fd00a986bbf557
fbe69891965533406ba553645851957a38cbe570
90796 F20110115_AAAZYQ abiade_j_Page_123.QC.jpg
583446efc41030d63a530c7d912decb0
35a3c4bdb0f23da12b5aacb1eafff5393baa8cfb
18701 F20110115_AAAZBJ abiade_j_Page_120thm.jpg
cd2d61db2c7f1483f5eaf63f9169af67
4476a4a98746d996bc7f12646e770c14c3bcc867
23252 F20110115_AAAYWD abiade_j_Page_078thm.jpg
5e299909affda7bead10ca9a51851cb2
5451effa6dc037bf13fc6ed4f021a34790cdc646
94870 F20110115_AAAZAU abiade_j_Page_090.jp2
c3c4444f2b22b7b7d8c64ac804d40231
ce3d07e3cffa8241bc589ed2f26d8851f52de4ae
188685 F20110115_AAAYVO abiade_j_Page_090.jpg
2ddbf84de10366987686747b0cbb78f7
fd6bc3f34b255a1d3cbc10a09ee70804f242a814
70213 F20110115_AAAZZG abiade_j_Page_075.QC.jpg
78922cfe079bfe479415e4a9a21aa330
d07e9f8ea45db0ea22c05095919dd41389a4dd53
46468 F20110115_AAAZYR abiade_j_Page_108thm.jpg
96930d83315b0522708c5ef1351b4f81
4dc5b0f8161cb82bab9cd539bb3bfa9c3802ca99
46969 F20110115_AAAZBK abiade_j_Page_062thm.jpg
e8f061a1df2ef5ad88e91524d764ffbe
023e27e99b582e940f86bca5d0fbf812ee4f5972
48544 F20110115_AAAYWE abiade_j_Page_042.pro
83553b9ab7243e0a79149bcdf27f7c57
b217052a6cee7e3a740b68f6bbfa2111042b84db
81658 F20110115_AAAZAV abiade_j_Page_072.QC.jpg
230efaecfb0c7c98f5b3445d14d6680d
eb987182de046285ad8ffac399628b30fe463e78
21956 F20110115_AAAYVP abiade_j_Page_116thm.jpg
4f0333610cca993b3afe3339840a71f6
e03132bc7ff683166f3910c7ae8b010e36698ef3
15525 F20110115_AAAZZH abiade_j_Page_058thm.jpg
4d9aacc89dd63987d44e5d364a5c04ca
b300d297b4de00940876102081e66fb953c83098
45877 F20110115_AAAZYS abiade_j_Page_122.QC.jpg
0990a0710bf8deb8815fa86c3199a379
a08bd82b58d8f67ac8eb2184164b807d74a9fbeb
74733 F20110115_AAAZBL abiade_j_Page_132.QC.jpg
ecd3127f9dd8c38f052f9781a6fa6592
5df4cee837f8ae79e301a6606b210381d9f7f7f8
99187 F20110115_AAAYWF abiade_j_Page_109.QC.jpg
807ca29b4f97e0cc4186ccc05b1fef0d
c4525c1c4e2bc7eaba87af2bfadd5f433f318a12
1797 F20110115_AAAZAW abiade_j_Page_116.txt
7b696bed5c07eef44c389362c395f8af
24d8473a4e58b5561cf19f2fcc11428afca01d7f
951654 F20110115_AAAYVQ abiade_j_Page_127.jp2
316c25a5cb0d79e87a8a1eae17c784be
c276add21ea28bf8e999e16095848516649c58d1
103109 F20110115_AAAZZI abiade_j_Page_010.QC.jpg
473c3fb5695ec1cc28a15ae5c9918e68
d6c77964ac69730a72c3df7ed678bcca2b3e48ed
23999 F20110115_AAAZYT abiade_j_Page_102thm.jpg
74c185a073813c6132ae022dcc129ed1
c4c44cc0a010a47037908304ccef29efcfa119f9
1455 F20110115_AAAZCA abiade_j_Page_137.txt
b932b765c749e4a1a229413dfa154e5e
af852a44074237a0dee35d5a75100f8a0060e6ec
192192 F20110115_AAAZBM abiade_j_Page_070.jpg
b9fa491c6b818543892404a5d18f8da3
f970b956617a5f845f38cc76f21a0386c4c9aa4f
25186 F20110115_AAAYWG abiade_j_Page_101thm.jpg
cf06a545773d17ca901b1cbc41f68cb7
eb70f5dea9331d058eadae88447e6a6970a0ee88
1974 F20110115_AAAZAX abiade_j_Page_030.txt
1508e632a1d072501e7e691458f9667b
5d92c94954024a7e4c72ec8ddf52823df7edbdf2
109905 F20110115_AAAYVR abiade_j_Page_143.jp2
6a497f7ef33f96782fdc0464332d36a1
6483964c604614cafe0bb2adf234a1970063f35c
61778 F20110115_AAAZZJ abiade_j_Page_061.QC.jpg
696cf351e471c3bb3ad45383c966fd44
39d97860a725807c2c930062591d958b76f87237
83826 F20110115_AAAZYU abiade_j_Page_127.QC.jpg
442abbd52da84b1b881717ab8d71bb6d
c8deaa3ba8f797a973a3db58965c11c6683cb4fe
215334 F20110115_AAAZCB abiade_j_Page_017.jpg
9080afb4010a222ee23d7fcbf23b67fb
af150a9469a66e4f545606191f9117569951f492
88658 F20110115_AAAZBN abiade_j_Page_126.QC.jpg
a9a3fb3dd6ff7f040074f8bb031fea61
419f0cbc49507ec9fea4547b4cdabedb41e80405
99628 F20110115_AAAYWH abiade_j_Page_094.QC.jpg
ea9fabcfb1ea8a2fe3b82183e89c10c4
477be0a33c028459eb51b852662adb3c84218505
23171 F20110115_AAAZAY abiade_j_Page_119thm.jpg
d0e034bd81acec70037ed3273b9bcdd6
82ed0866d73c9667b3f8d1036d9289563a2d5424
112267 F20110115_AAAYVS abiade_j_Page_080.jp2
a3906e3f2753bc35f929ee850589a00a
b377c50d7350f81abe0cddc08da00862414cf239
45390 F20110115_AAAZYV abiade_j_Page_104thm.jpg
816706aa1602f504aa819af6a651fd57
ea228c8eda32ca2160cab58017f97db04c2ec6d2
95444 F20110115_AAAZBO abiade_j_Page_037.QC.jpg
fa64556e1266f6a2f4d2d6071cf56ce2
f6ce7afe6f7b6cc94af0f7ce7ad125d136a30e4a
8423998 F20110115_AAAZAZ abiade_j_Page_057.tif
250a286b3e3b64889cf6097db0ad8c0e
02a8e0abdf6d3b09e95dddb00d72d750293428b1
221143 F20110115_AAAYVT abiade_j_Page_024.jpg
8aca1ed6760c82980710b794552412ee
292bfbd67076becca31163c30fa4040068c6015c
89346 F20110115_AAAZCC abiade_j_Page_084.QC.jpg
4eb0592277f7973ec8333eb2b2a1ced5
424a2781094d876c5e972e8a1109043e282fc266
15958 F20110115_AAAZZK abiade_j_Page_122thm.jpg
ae39cb52e8c010ea9ae46bbf3ee3c844
88ebc7771b48aaaac9d0f6f11e9a52bfa3a54712
74349 F20110115_AAAZYW abiade_j_Page_019.QC.jpg
67bdfe690a7876ec5f72b3ee724bd404
ca3e6439597953e3be98c10778e31fa023842869
24255 F20110115_AAAZBP abiade_j_Page_072thm.jpg
e2d031ca0aa03509244ea328dd9a794c
e88ea9eb00835ee8ba78b963d33cbc30418215ef
218787 F20110115_AAAYWI abiade_j_Page_033.jpg
36b4d5c45a4591d26be45b253b408034
3b288047e1b7b1af015b4f7e7a5d43a633eb4b7b
202190 F20110115_AAAYVU abiade_j_Page_146.jpg
f23e21eb67c6aa07a04f02d8f90fe539
678968529c3e5192ae8ba3b3e863a87d92fcb95e
62468 F20110115_AAAZCD abiade_j_Page_055.QC.jpg
405f49961f8895427187ef5b2a730bec
99125bfa41a61ed6be3c32b087f912f900f28749
24328 F20110115_AAAZZL abiade_j_Page_136thm.jpg
6e39b3c47c87d5bff053b759b57f0aa0
f0393d9e4d00466c69457d7b5661060073940bb4
49198 F20110115_AAAZYX abiade_j_Page_053thm.jpg
05d42c58540aee70b0610d9afb20afd6
feb97a117549f869dde27dd9fbd0516fa8c1e31f
50174 F20110115_AAAZBQ abiade_j_Page_078.pro
f98ee0b8e70206d3a1886786c55ccd6c
845ccd1b039ae11db10918e94d15c3b4ff72566e
8902 F20110115_AAAYWJ abiade_j_Page_001.pro
e0fd06fd7ec395989de0ea39d7b399c1
28f6dc391d2ddfb64674c7591acb99ee9e95fc96
47978 F20110115_AAAYVV abiade_j_Page_019.pro
c56aaf8cd2b1701b0dbf70f618998ff7
bfc7e2f47d132a3df118a84ef183ac1bb6e64dc0
86520 F20110115_AAAZCE abiade_j_Page_121.QC.jpg
b804b892c2eb6b2ac618b1bd9f61f086
dd00b4fbf340f66c5bc325cf60f32e4a885a24fe
65540 F20110115_AAAZZM abiade_j_Page_138.QC.jpg
979366bc5539e9bcb3aae27a04040151
7d9cd04a57a26c0471ef841f7822468ea4f45d9e
59156 F20110115_AAAZYY abiade_j_Page_005.QC.jpg
528c0e9278daeaf3c437c82657cfd498
e98981ac5b0643b47049b7a7b665814d06e135f8
2072 F20110115_AAAZBR abiade_j_Page_136.txt
c41bc809cf9e1ada9e39e42f8c3e25a5
7432ece0da5171f2fea2d0757db0d9f017661811
142452 F20110115_AAAYWK abiade_j_Page_028.jpg
8e54fa57a91d6bcd05fe26639d9f897f
fbb3eb117a4f4b23415c635495bd3c36b4fc08a1
17552 F20110115_AAAYVW abiade_j_Page_112.pro
afd77c76961a61ff4147e3a338d45d05
98d865b46f435b1b94dc74f73a988001d2f92c07
105671 F20110115_AAAZCF abiade_j_Page_054.jp2
34f9976eb12d028b8494677ed92f3697
4cd2d8e4d643e3491a6e448a464e267f599b6e5b
23357 F20110115_AAAZZN abiade_j_Page_065thm.jpg
17100da7ec287eb5b8ecf039f844805b
be76c295c24c7a5a7c0e09e81ea193ce2e2f447c
21059 F20110115_AAAZYZ abiade_j_Page_131thm.jpg
8c72a003c47e34e90abbd241a0820ff6
0fd4e6c9088da8ef340cef11d215071e0c394ffa
46287 F20110115_AAAYXA abiade_j_Page_012.QC.jpg
36081cf3e2328bd0859309852bf35a0c
552d98b56a7ac6da59ea5233ae914cc0d0f8a750
204314 F20110115_AAAYWL abiade_j_Page_086.jpg
97fd5be9e31e31c19ccec71742121c21
983e03762c33cb3f9017b190d8d0d8d702c82d61
86181 F20110115_AAAYVX abiade_j_Page_099.QC.jpg
99dee4092ac58673302060429832282b
b4da61eab6c78643844dc2e1e6831dc91feaf9a7
29590 F20110115_AAAZCG abiade_j_Page_045thm.jpg
3a9fff924242dd9839acbf663ca3cab3
c8eb40b36ac1c0531a5360331ba78cdeaebbe94a
79291 F20110115_AAAZZO abiade_j_Page_134.QC.jpg
407caaa6cb5922c7687b5bd5b0f31be7
92c24bc8f5176cb8b5c9a95cea990d1b3dba6bc5
175262 F20110115_AAAYXB abiade_j_Page_098.jpg
185c81ac5e1a8492230a1bec2da5b3a1
1ca81789c58e9cabf7b9ffa900fd56f442dce14a
71327 F20110115_AAAZBS abiade_j_Page_120.jp2
b56e061ef0f702be440df63cbe2b41dd
45a8d1c6613e494f7f8795fdac87aa3a10250c3a
46720 F20110115_AAAYWM abiade_j_Page_064.pro
a9ed13e231a9b29148129b9445f34f5b
e068ed1652522605d628d661dab564aced5bd38f
F20110115_AAAYVY abiade_j_Page_044.tif
14d41ed292e2e1f7dba65938d073be76
1be38e7d5e1c3f203d80a718f085c929b520c144
23118 F20110115_AAAZCH abiade_j_Page_142thm.jpg
d434c5d46451556201d14ffbbb09dd66
489296d31fc447f23b003561647482b0cb4b1c2b
49648 F20110115_AAAZZP abiade_j_Page_109thm.jpg
e59356c29631f5b2989f6e601fefdd0a
bc2a46a0181fa5ad093a809e1c8944e25c239142
30425 F20110115_AAAYXC abiade_j_Page_139.jp2
af00fd0ef4ef447f41097b0591003e3b
a4ab1a4264af77fc81d19b00a2beec258dd4ef00
24971 F20110115_AAAZBT abiade_j_Page_017thm.jpg
968f1d79680b80c633bf53d53a7750b1
0daf49c0f97438a09ce35d949343caba243ad0f1
113769 F20110115_AAAYWN abiade_j_Page_024.jp2
127c0daab15c2185fce7508fe948ce8a
11a27b350840ea0d07685348a81db90370dc4972
116085 F20110115_AAAYVZ abiade_j_Page_121.jp2
675cba647c6043df556dc33e835ec5e6
53a73ea25a939d9c49f777508f877b5dadeead76
92305 F20110115_AAAZCI abiade_j_Page_083.QC.jpg
29b9915e4f719b09a0af904bd9830786
2ba3c6f43a7a3718d33e3ad099d7d9257d687e17
82025 F20110115_AAAZZQ abiade_j_Page_097.QC.jpg
e6fcfdfa73d6f764ac550c891022ed85
8c81d102c66e388f2ecbe1f78097951b2f58ab16
52293 F20110115_AAAYXD abiade_j_Page_101.pro
194f8080a50fb4142aa70fd536999cde
3e896703e2d97ecf53b6745c7c0016c70a3d2a54
F20110115_AAAZBU abiade_j_Page_116.tif
dfb2226eda1a8b6980ffd3900f9c13f7
d3ae9198fa8575cfa1e3ae559993de75eca40ce0
79778 F20110115_AAAYWO abiade_j_Page_048.QC.jpg
19f185f521f847e58fa8da2e901a1d60
aac0e4609a5ca88f2317ddc8649c1dbcad82a436
45366 F20110115_AAAZCJ abiade_j_Page_014.QC.jpg
59b40ca79f3f64234899889bb2df6591
696dbfa250abd7f22988824be20afa8e7b69c639
21011 F20110115_AAAZZR abiade_j_Page_090thm.jpg
7fcb6b4793febf55e29213a6379cf4bf
45941efaf4259cd04ba6b6e438aa51f92268eaaf
7850382 F20110115_AAAYXE abiade_j.pdf
4561a26f6b57b13a7da9135163c08a62
1131074fde93fb623b555d4ae02dd2d9c4150580
1051974 F20110115_AAAZBV abiade_j_Page_021.jp2
8a3aa25cc325922cc691d8e464b19749
05fd4bc6c86cd4043cde37b194944a713c986195
40279 F20110115_AAAYWP abiade_j_Page_140.jp2
1701b602de72c300636236baaa2213c5
ba6d8ed88502c6abb044a6a025980e300a8a26a4
46775 F20110115_AAAZCK abiade_j_Page_083thm.jpg
b30c23f4b61a1637b8b9449a7c8989ab
7c5b3125fc89d8e33833ce0cf1e38e7fb77b47dc
74139 F20110115_AAAZZS abiade_j_Page_143.QC.jpg
d7bf124335219a305464b1f7af1c4401
7ca9c8facc1a0743c100654ba3632cea601ee532
190571 F20110115_AAAYXF abiade_j_Page_073.jpg
897df56abcc08800c27e7dcfa1f7388e
f203e974bf3ca761e20af2990091577d8bc1fab1
31249 F20110115_AAAZBW abiade_j_Page_055.pro
871aa021aef91d81d5b42643b80ae625
9a06a1ee5a3837eed7ab97c4e2a7b173bbd258af
80393 F20110115_AAAYWQ abiade_j_Page_080.QC.jpg
2383756eb496a08f1ca6b041ae4f9780
db64b4f9503120a8734fa94e25530eb0098b8705
171068 F20110115_AAAZCL UFE0008394_00001.mets FULL
0628b71a2cba245949a55f07748d53b8
6b55467b46daf5fe37643fa50077a37fdf1908c4
77384 F20110115_AAAZZT abiade_j_Page_079.QC.jpg
649aabb299db4bba61b4aa6196dac6fd
1189f4c8c120ceb1d1aa24144620cfcf57d1136d
477385 F20110115_AAAYXG abiade_j_Page_012.jp2
3618026cf4d2621e6f421274895547ab
55d3efcc9e37b6eb251754d1e558655ec647da3c
750308 F20110115_AAAZBX abiade_j_Page_098.jp2
1b06ce545ea0ffea5e4f92e3654f0587
bff87ff573cc314e0577ef8df2c094012118f37f
24349 F20110115_AAAYWR abiade_j_Page_029thm.jpg
52e76623ac02cdb60a46dd52926ddfc3
7c638fe78a35ad777e80274cff6255dc06632067
220420 F20110115_AAAZDA abiade_j_Page_018.jpg
ae1d53f03751d4b3e12e6e6d3533b5fa
d46a4ee93ce145fd721d7bb811cc42ac1c2f8a98
71081 F20110115_AAAZZU abiade_j_Page_073.QC.jpg
5226df0984fe958091ee6c8cc7cea33d
fb8a23371c10d59e9960e14cc3965554e104e7a4
113875 F20110115_AAAYXH abiade_j_Page_016.jp2
ca06955bf1b8d28180dbc26bed053aab
36a4626ab2446353189da07cc1156efd6afe00d3
51725 F20110115_AAAZBY abiade_j_Page_096.pro
591e8133285ea81965c3f84671bf0cb4
1e750bc981c8aa12394b3d6295dd60c22108ff06
944 F20110115_AAAYWS abiade_j_Page_147.txt
2000604f894356c33af1044e4409948f
d06e15f1fc9d6fd9fca2020cb32570ce40daf0bb
203132 F20110115_AAAZDB abiade_j_Page_019.jpg
f84995cf10f167923e17ac28ed62e014
c3ba617bb2169b8cb0e084fe501c44e757a29d95
46944 F20110115_AAAZZV abiade_j_Page_084thm.jpg
81d84b6cc8323dded4ff461b2540ef03
aed8fe9f4c84617e6db2167759eec661d7969649
214264 F20110115_AAAYXI abiade_j_Page_051.jpg
3c78474fb66fe420310d6763c136062c
75bfbfdc980a9b2c6509d254d53cc62caf993568
212750 F20110115_AAAZBZ abiade_j_Page_091.jpg
6e5e71b6e6005917bb4b4415c47acc2f
b2372c295eeea34380a826255ea5784b8ae104ca
1452 F20110115_AAAYWT abiade_j_Page_026.txt
38ff3686a7aeabe025c7a28623e1e83f
8d066e53f596b3dd361a9f9761a1d2c0c479d57b
222507 F20110115_AAAZDC abiade_j_Page_021.jpg
836516ed929892932011109853035024
642bc1df188444488c2a0d9e54e3c1631be655eb
59225 F20110115_AAAZCO abiade_j_Page_001.jpg
3061cdba5ec5238a6c8b6a8ecb3da13f
e688bd0a93e4630aff575d098dcb3d285d504201
75610 F20110115_AAAZZW abiade_j_Page_054.QC.jpg
864bb79f50923930f20f5ccbad4a3fe5
b164976e24638c77ddb346c406d9def44d29f654
1724 F20110115_AAAYWU abiade_j_Page_106.txt
e283af332af6b59ab18d9b85743718cc
d6dba4fc57c490e8310b0d8419f6fd2395a6b220
189805 F20110115_AAAZDD abiade_j_Page_022.jpg
8abd0bd033d9dd25790899bcbe69cefc
10e875665e53fb57e9eb2e7e26e90574fcdb265d
15597 F20110115_AAAZCP abiade_j_Page_002.jpg
f0971c04aafdebba25c63ba664b3cf55
15d6ea40899f9fc44aa8e246b4f13470a2d3d463
23060 F20110115_AAAZZX abiade_j_Page_042thm.jpg
13fdbdc0a51e537521ed5b472b8dfeb8
e8c668854e29a8d157b78a1de223fb7b27a3244a
F20110115_AAAYXJ abiade_j_Page_138.tif
a2e20560784beff9d8e9466f9b686481
8ff17007190fae979108103d1f583b0a50f46607
61555 F20110115_AAAYWV abiade_j_Page_014.jp2
6b47370a36f683f0604f0d92e675e985
ed8bdccbebb97bb8948610174bfe0c2f61e056d2
217690 F20110115_AAAZDE abiade_j_Page_023.jpg
1ab17f6a08be7528e402cc13bd38164a
95a44a2211adff1b42e66218fee0fe7f1181bcb2
44517 F20110115_AAAZCQ abiade_j_Page_003.jpg
5d3900f76c732fa32f40cbd3cd3019fb
8e40723d7f52967975a4a512c3c64feeeecc5805
44613 F20110115_AAAZZY abiade_j_Page_106thm.jpg
992d35466714616e576ac79ba5304f48
f0f5e5607a8897e681c313d9b4f2919772c4fb4d
2182 F20110115_AAAYXK abiade_j_Page_142.txt
3e64a39d3e58cca007ea667667530fca
d2121e548050a5e2779472c73f8efd173b777fbb
1540 F20110115_AAAYWW abiade_j_Page_061.txt
2b951698750ad9f96b0b990ebe22b59f
495a9dee6f58b39d2b361f01b20b945015584eac
243301 F20110115_AAAZDF abiade_j_Page_025.jpg
cef494190c73b4375bf3b2c911624bab
5a016b7f628454661d4902a727ffde5adb8895ce
196303 F20110115_AAAZCR abiade_j_Page_004.jpg
8b9d4b4aefa2a0c674035cfba6a30bf5
a05e51bc5ceec5cee8fca68e5783f1288f37c1c8
44607 F20110115_AAAZZZ abiade_j_Page_129thm.jpg
6cf70ab63c7708c8caafc7eae684ef75
bbf0bda061989f5df208168a4d0d88d6a7e78777
F20110115_AAAYXL abiade_j_Page_051.tif
3f9fdd1437a3912fda3a8b7a92016aad
481b334fea9b62402376d650d508d4961b3304da
84393 F20110115_AAAYWX abiade_j_Page_106.QC.jpg
fb836b2496f17ce99597dd680b9f61f1
f7408d035739d7b8f3a170a8970aa155bd34ab96
210006 F20110115_AAAZDG abiade_j_Page_026.jpg
9557872067fc009fc457ac1dfeca4c68
3315e39972bc7bbe0403ccd2393c8ad30b8e5fed
F20110115_AAAYYA abiade_j_Page_143.tif
d8476dc9e6273ee68cb3141126c40296
4c3909982823da09639f08021a4e2774ea981580
157356 F20110115_AAAZCS abiade_j_Page_005.jpg
488f4e0143305a949c41ab7058dfa4d7
ad9c0a595640bab5f385fca6b86ee111ac0e62d2
49599 F20110115_AAAYXM abiade_j_Page_071.pro
a20f79ab5f057b6e98debd231878d535
52d977d0d604c6edd9ba955df95080297c20b7d4
16323 F20110115_AAAYWY abiade_j_Page_074.jp2
41516cb9c1ab2d62be4615c2ff50e189
c0af480f22fe9ef3d11b6152d8194611a1f23f5f
222887 F20110115_AAAZDH abiade_j_Page_027.jpg
1fb3c6d1b930b3e5b5326440dcdd78ca
6ba4a9cfebf68430518e91bf880886ea843adba1
F20110115_AAAYYB abiade_j_Page_055.tif
8f566d96206284a5ff2242c85d6995fc
114ebcc8cc0a9a0bbe2802419a323fa8f8b4efcf
F20110115_AAAYXN abiade_j_Page_064.tif
a6841784d104fc527fa8af61e431a681
b6c52fddfb263aadb721cf2b18c5d3b22c8aade3
71189 F20110115_AAAYWZ abiade_j_Page_022.QC.jpg
7d95c9060b40bed6b2891339bdf559d9
f2cb0d15f9b620c88e0d5217a81fe3691c31b0e8
211259 F20110115_AAAZDI abiade_j_Page_029.jpg
51b8d999f945bd1f583827058529f3aa
557ce53723477ff1acddf84bc11f8fb8d6d401a5
75080 F20110115_AAAYYC abiade_j_Page_144.QC.jpg
4701abf7e5739adbe68a27980ada7ca3
bc71794f2c4bbce0570e21c9230c056fe9e9fbbc
246433 F20110115_AAAZCT abiade_j_Page_006.jpg
4d43734f27c38eff59948724bbeed790
b6d2b927f3fd3b5b13a89cf43171902c5c7135e5
104021 F20110115_AAAYXO abiade_j_Page_020.jpg
580bf1b8bece8ffa77adff7f62acc405
cb40dd98aaff6dc643bb5d74fc64b30e66974fac
210035 F20110115_AAAZDJ abiade_j_Page_030.jpg
5199882908e24480190a0d0a00e01b01
cd9c314bb166014ff26dfe143c43184f8f80fb2b
1991 F20110115_AAAYYD abiade_j_Page_095.txt
febb0d5d504126d8ac8f95420a9e8dc2
8d3528a89229bd0077bc441207428f9b286d5be5
59557 F20110115_AAAZCU abiade_j_Page_009.jpg
ac90688a91321ef3d5dd4687e5191188
a7e5a9fd6dddf577e8e2577dcb20f6c2899bccf3
170159 F20110115_AAAYXP abiade_j_Page_067.jpg
ac83c7da2f13e74f971e00a78814c3f2
022d3489fe52076a31f3321808b3f56449b425f5
198829 F20110115_AAAZDK abiade_j_Page_031.jpg
77aee4d3ea5f95a6b7b3e57f6addd67d
d136e29afe02314f7ed187e81b77c6f1f10f67e9
F20110115_AAAYYE abiade_j_Page_053.tif
70bb145e4f0dff348c880eeffec8c2df
dc030407f135edf7430ceb98efbf0ac71634dccd
241919 F20110115_AAAZCV abiade_j_Page_010.jpg
dd057beb6ea6d444fa274164b4b3ba38
f4798bf14d0294aabd67121a15dc8e76f1370edb
F20110115_AAAYXQ abiade_j_Page_063.tif
bed4ecb595f14b60a85bd70cecb058e1
03962a9a14e18d3951ec35f3ffa8eaba58f3e5ab
214209 F20110115_AAAZDL abiade_j_Page_032.jpg
3495ce2fd32378a2ac2c29ab8d801743
9d928879e90e6a0c813fad1febdba8dc35552f20
78976 F20110115_AAAYYF abiade_j_Page_027.QC.jpg
92de25486f6b01f736ab5df560f92550
e469569b8651179be31bd84e71e036c7ed85d4ba
90818 F20110115_AAAZCW abiade_j_Page_012.jpg
cd950ad388466ec38cb0c03184385bcc
a910b444ec22667961b5f8edca71496a8180a3e8
29116 F20110115_AAAYXR abiade_j_Page_108.pro
5b7ed38964043dbd9f3b2a099eb0dea1
f12b3c80c55078c8fbd8d07b543f4b0306e27458
207760 F20110115_AAAZEA abiade_j_Page_049.jpg
803f4f5d0b6d2cbecc561cfc9851fd19
367b91a873d775444af0eaa89578e537cc0f0a52
195335 F20110115_AAAZDM abiade_j_Page_034.jpg
96ff6c0215e802d4f50951caf94657fd
5a00cf076af83bef00610687acb963327540c7c7
212437 F20110115_AAAYYG abiade_j_Page_111.jpg
f1696bbe3a1cc87e4566c3f34f6eee81
9becee1d1e44b9c496855ad0bad4fbc820203b26
172884 F20110115_AAAZCX abiade_j_Page_013.jpg
d56222c60a225b8e3f03f008a2ee712a
a8a61674243907534b5a43e4a003e16febddfbb3
208323 F20110115_AAAYXS abiade_j_Page_077.jpg
a8c08d490297cb1bf651e3c26cb333b2
17195676d7d6169134c6e3805a968fb5f12860e1
199993 F20110115_AAAZEB abiade_j_Page_050.jpg
83f53833e8166ec1335305b5877e0a6a
70ba2be93c424dcef6c253353a8d84c568206d1e
218628 F20110115_AAAZDN abiade_j_Page_035.jpg
42d1c615cd87446529c363429f2ae73a
b093094e3f32957fc74d88b98b9dcee67c971257
108672 F20110115_AAAYYH abiade_j_Page_032.jp2
cc54f3a9e78e0fbf0b8febb969b6b953
5a35cff41f5866c06a5c8d40f77cef926a7cec47
118670 F20110115_AAAZCY abiade_j_Page_014.jpg
5ae369715f41ccb23b6a44920ce14013
50fd977cfa313b82e19afcfbaa5a9b758ac97079
94403 F20110115_AAAYXT abiade_j_Page_075.jp2
0ff4568bc11dfbedabacc4671006f946
34a00747bfef6f8c17ba2c2293f6dea331076819
208708 F20110115_AAAZEC abiade_j_Page_052.jpg
8742ad6bd914297ad8d7d66b12e62cdd
25f0b13d68e06164b4f28b7b933acbe70a5d8f21
213448 F20110115_AAAZDO abiade_j_Page_036.jpg
b79f11e3d1e06d8ed1b3f565e5608e29
a3ceae7ffe1485af5c62f068910f9be921ffaeee
F20110115_AAAYYI abiade_j_Page_061.tif
58f2d9737f25bf64cce7afd202dd1f9a
a9170244126ac8a89de870481c7487858c6ec548
220686 F20110115_AAAZCZ abiade_j_Page_016.jpg
f5610b698d55c3d6a07e1e4ff026eeff
fde71e527dd796f1f8d10a601b848e307bf632b5
188239 F20110115_AAAYXU abiade_j_Page_104.jpg
34249e49f459f5a52e0a8de93a1b8123
7fafeaf590dfedbe0bb256544004c41fb36ab85f
228450 F20110115_AAAZED abiade_j_Page_053.jpg
e66b54e81a651a01c9b6bd6f490a3e13
5c737e5f6a39349ae34515dc315a46e85900e0ed
212330 F20110115_AAAZDP abiade_j_Page_037.jpg
34b7538121112536751be6741fefae59
684f31dc27ca3a83225c5d5ebd82a4f6ab81a374
F20110115_AAAYYJ abiade_j_Page_074.tif
0bb9ce88622f5fcf54477d92f88508c3
b8b566300d87db609298325617f309b8bca24f6d
955676 F20110115_AAAYXV abiade_j_Page_135.jp2
17dee562428c079cc9d8f0163e3574d8
88a9edb8aa5718c4054a97b0a6f0b2b1abe530c2
201723 F20110115_AAAZEE abiade_j_Page_054.jpg
b66e0340216fdf46486033b32575589e
29f0b80dbe0be6235e3a69b35ab3313b85cc00be
183873 F20110115_AAAZDQ abiade_j_Page_038.jpg
ed7e91f68a0da11e850c4a901063acd0
95fa44e5640ecf46071865c8ba89606e5b4ed636
15032 F20110115_AAAYXW abiade_j_Page_140thm.jpg
e9d32cf69e36f11c18bc0dd57bc53740
a0320baee236b8229251d4af7a161961d5f2e31e
151984 F20110115_AAAZEF abiade_j_Page_055.jpg
650f516ca2d6d812a110dac4f53a4aae
e187f08a88be7ac9872c828b693b8611593d49b1
218973 F20110115_AAAZDR abiade_j_Page_039.jpg
414c053a9a9413457f270fb217c9a630
0081dcdf2dd6faa68366ab082d69e7507ca1725e
F20110115_AAAYYK abiade_j_Page_098.tif
c231dff582c810ec75faefdb24497b5c
84e6cad9ae717f6af476f2edb75acdcd45f76496
F20110115_AAAYXX abiade_j_Page_033.tif
ec08bf39ecee34ef01568d5d777ca6b2
aa6662046d7ac78fd32a5110b1c8ca483b355cea
204707 F20110115_AAAZEG abiade_j_Page_056.jpg
30d5b1a96f100b19adb066e32d9bb4af
c7438726de4ac6cdb99b6ef68508e27bd2724759
1948 F20110115_AAAYZA abiade_j_Page_115.txt
7836b0dd2bcb5e6bb0d3c8bec9473c17
e5ba32b0f8341caaf9da4bc2a75bc2a01856040d
202946 F20110115_AAAZDS abiade_j_Page_040.jpg
db969c7fc8e5226ba1e2f5d7ffdc08d8
c80b6a839fec50ac81f2973766df8a4dbbf75c0a
F20110115_AAAYYL abiade_j_Page_112.tif
dbffd00246c139b1bf4d9e3f86264a7e
fc3c9b7694f400007e216e59b7c771c9552c7216
216637 F20110115_AAAYXY abiade_j_Page_080.jpg
f343e88cedcfe596e735e4e10fb4c836
be12cd78cf99543e419c604038cb87f2a42eeacf
155740 F20110115_AAAZEH abiade_j_Page_057.jpg
a81b55a48b24366157bbabfb5fdf596b
6982a9ba4c11258a7a327892f1f9772f5a8252b6
100823 F20110115_AAAYZB abiade_j_Page_070.jp2
0d2de2cf29f9e826962b0cbb18138012
e97397fe564459c4875753a583dcf15088fdd83d
209155 F20110115_AAAZDT abiade_j_Page_041.jpg
e4e708147e2cbe34ab11b291e9779814
902bceb45a6c4b53c2f72a3df65313d0327d1fde
5463 F20110115_AAAYYM abiade_j_Page_074thm.jpg
7e5aca845855586fa1c98ba42feb7ab1
e47fe7e98d5ef52a184931642d671b84e37a6096
105413 F20110115_AAAYXZ abiade_j_Page_046.jp2
0f55706a7caebd957db1cf8a2abd0434
d4f1b283446a5f900fed0f1e5401d94b8c7ad4a5
127080 F20110115_AAAZEI abiade_j_Page_058.jpg
49e2715c7e890ef29d7f16fa32cb02f6
9e137f35e193609393f174564872afbb77ab404b
2703 F20110115_AAAYZC abiade_j_Page_011.txt
175bd3405044d6f37b3caf69c5557f16
f7d3565968ec55c54ed083de4366f72ae5655591
29440 F20110115_AAAYYN abiade_j_Page_025.pro
3602b4a5a76d4756aef7ac4698fb7d94
ab532dacbe2729df4701a6ef8f07e4932de6ba7a
165293 F20110115_AAAZEJ abiade_j_Page_061.jpg
f5d1ef0eed172534fe50b107d6478787
31f6580f8e6002cfe5016d6e82c38d3ec48049f9
104667 F20110115_AAAYZD abiade_j_Page_146.jp2
e025b2cb0ca30cdd537d0c8e903d0600
f639ae0befd02538b7ac53579e5d8e4fe41d51ad
199446 F20110115_AAAZDU abiade_j_Page_042.jpg
8c71d719dca9120efed98733f9ca25e1
743cb933653d7417ee55c2af157f24ff45693c2a
1239 F20110115_AAAYYO abiade_j_Page_100.txt
11a3d50b37fa2cdeac14c7a3e0f6857b
aca0aa49d0ee71fd73acadc843ed2ee05ae5db44
237190 F20110115_AAAZEK abiade_j_Page_062.jpg
621dc8811a7fcde25aca1ce0f2cf4e6d
4ab82508dfb774b722302c955e3ab968eaf9bf2e
F20110115_AAAYZE abiade_j_Page_102.tif
12a878c82499e7cef2f9977f9e2c54da
c4c364a7fcfaf54f85202c30dc72431f89a8b666
238884 F20110115_AAAZDV abiade_j_Page_043.jpg
10cf063b9755e026ecc7cc5a384d3a9c
71558912eb98c430661145d30e81a85641e5c26b
105652 F20110115_AAAYYP abiade_j_Page_056.jp2
73aab5a760f975c2c4b59f1ea497cd3e
f050b26c564df632c6610fa75dd79f15cfb57ea2
186173 F20110115_AAAZEL abiade_j_Page_063.jpg
f5fd70e10e38fc63a8faff8c117ad0f0
c9cd7a49512588e8d5eaf52d5e1f061a5619d17c
23932 F20110115_AAAYZF abiade_j_Page_071thm.jpg
6bab8b885a7195ad4119b3784aab4d1c
846caa1182a6d28fa59f3b7cf0ca3b2fa038313f
184063 F20110115_AAAZDW abiade_j_Page_045.jpg
623d01922d7c5e952929934e9f5de8a2
ead93ecf52d6f061614311f84f395c790eefe6a8
F20110115_AAAYYQ abiade_j_Page_136.tif
f6cc9eff47051834e31ec49713c0bf33
17cc2b2884264f691700d190d20c09585d3959cb
190705 F20110115_AAAZEM abiade_j_Page_064.jpg
4df5ec685814984a27afd9a3eee7ee7e
0a022fd845a9009b42c6d76e288efb49a0a962ff
55862 F20110115_AAAYZG abiade_j_Page_125.QC.jpg
6719db69ff8c4d30d63b03bffed4d2fe
e9f9c9d0c6d650d2fe197199366b737cff3bd257
204096 F20110115_AAAZDX abiade_j_Page_046.jpg
18ce74738257e6f1503b16d1e44ff0b2
3be9a93dc68bc86ee1e6b4834ff0e28171ee494e
2110 F20110115_AAAYYR abiade_j_Page_080.txt
46c6862e1b83ecca5dde292688da79a5
6d62c7121a5f2883e85edf4316d1e6381b4ff596
157623 F20110115_AAAZFA abiade_j_Page_085.jpg
e2cab0a598eabf0be840079e3789072a
af09b64629fd68f4cf471f2b4ba0344cc8ef3267
194155 F20110115_AAAZEN abiade_j_Page_065.jpg
3ffc72e9c55cd49e825b737d1fc6b7c2
dde367ba9fc2737576f94087aaef3bc04b7d54d8
24833 F20110115_AAAYZH abiade_j_Page_036thm.jpg
7c03d6f5e370edb64e365d0755fc11b3
0e3e5007807eb5fb7cda61c3266c7621dd337aaf
136876 F20110115_AAAZDY abiade_j_Page_047.jpg
7fdd890631535bd8f7177226adf75545
f5ebf38a04bef6ade62a5b88d94fc46f0ba8e33d
50895 F20110115_AAAYYS abiade_j_Page_017.pro
79dd75e0fdfc587c4a79735cd3f6d1ed
6b986fef0af27d99c358c1b83f21af2f31ccc836
184272 F20110115_AAAZFB abiade_j_Page_087.jpg
7a0347702a158d607a6ae07250133cab
77fa01482074f52bfdd1b7b5d6a33a4164be350b
196776 F20110115_AAAZEO abiade_j_Page_066.jpg
297a1b362051edb7f4844ad02c892363
2a68692871bad4236d5a88a1d5d796646676911c
1409 F20110115_AAAYZI abiade_j_Page_125.txt
58dab4a76ea8e58c5bd5676af4133776
0a70766ae47d800bb71432201db1db00ed53f3b9
213285 F20110115_AAAZDZ abiade_j_Page_048.jpg
2ed61199428840a058711ae377913630
ec63dbc107e6e7007ed4d3024feca38bc01a40a7
1988 F20110115_AAAYYT abiade_j_Page_076.txt
491f4fc228c71cf44cf23fd82bc7ebce
4c852fa1717dcce38c2cc84f8f4b97b1d1ea3a0c
218151 F20110115_AAAZFC abiade_j_Page_092.jpg
0ce29fb80f2a7ab17b236162d2b61ef4
bd89d8f5c6f71776e14581b8884a1b3316535133
156650 F20110115_AAAZEP abiade_j_Page_068.jpg
74a864f7a0ee2e928100ce1f11dc3c86
6b5893592020616f28530545de6b77dd71217ac3
75884 F20110115_AAAYZJ abiade_j_Page_065.QC.jpg
17aa43f402187af0d07765e3c75f7828
36914fa12bb0a358d0060d7f70a7effe08486c7e
292790 F20110115_AAAYYU abiade_j_Page_011.jpg
2e1c9a6a4d2a884fdab0d6c8ab3b6a3a
4a933c26adb4bd99c3dfad5fcfedd99d85e9d40d
213547 F20110115_AAAZFD abiade_j_Page_093.jpg
c3d4475a5892054ce0f37cc0cbc8ab07
d2ffc172739f8dd6fc4405145a60de2659635b52
216059 F20110115_AAAZEQ abiade_j_Page_069.jpg
ce3bca86edf512ba2421d4142a4a6c33
b47a8f43be3e6cf468bc5312691a0fd01285e455
29502 F20110115_AAAYZK abiade_j_Page_085.pro
223b309c6a29759a59688951fc73f963
abcc49606c183ee4a597225439b5a9eaf384aab3
87104 F20110115_AAAYYV abiade_j_Page_089.jpg
55c92cd8a9c088ba28af2ae24a386e55
5cde3ae467452cb8d6f659fdf9908da7e865edc1
231740 F20110115_AAAZFE abiade_j_Page_094.jpg
f5888b96d2f1283f83dfe03aa657c019
b31f191762157ed377fcc66de24df9bba9eb52bb
208473 F20110115_AAAZER abiade_j_Page_071.jpg
726b71065db865fe7c004e5c31b2d1f6
8d2a46d09536085a9a097fc4418be6c206f7c911
48267 F20110115_AAAYYW abiade_j_Page_040.pro
865e70668deafea96b4473e963d8aa3e
9241780ba6a016f51204991cf7dd13635dcff713
215028 F20110115_AAAZFF abiade_j_Page_095.jpg
7a759e91e267661081ae3d0dbe7f4fc9
1df416b5279224b9daeb897febe5513ba89cfa9a
215391 F20110115_AAAZES abiade_j_Page_072.jpg
7adcb9bf86a95ef26f415edbe3b9a71e
4f23887e10035825c4babfb5cb70c83a6c8f4bf0
101194 F20110115_AAAYZL abiade_j_Page_053.QC.jpg
487c0fd606f5a0262ffe1229d3c7235d
81c42e5271240dae1c440ba12012a9b0fe10290a
1883 F20110115_AAAYYX abiade_j_Page_099.txt
18831c18a0fb827c72b3e9dc078bb259
ddeb17ba4c1fe0e5d72405e34e841072cf0dbdfb
215538 F20110115_AAAZFG abiade_j_Page_096.jpg
44672cfe7655a3aa519bb98870fa8c06
c41e43b3e4ab9ae84524c80cde09d812a81d5908
32033 F20110115_AAAZET abiade_j_Page_074.jpg
9da074a6f4a13a51b912058584b4886e
2e701769007701a927eb85f735861f61c3c95e14
81348 F20110115_AAAYZM abiade_j_Page_044.QC.jpg
a7ccac28593db44252ae5668b09d33f0
bdaa7eae35a30b3aa3f566c84237c3d651726e48
F20110115_AAAYYY abiade_j_Page_075.tif
7dec1ec1586c83af0bd9de39e1725bfb
d7d8efa06b36232374868d205cb4a62ce018776d
217071 F20110115_AAAZFH abiade_j_Page_097.jpg
62d855e8113fa214547db1cc18739d8a
d7bafb36b1ff0a2819d9fcab3b9029ff67f1a38d
209041 F20110115_AAAZEU abiade_j_Page_076.jpg
db9b66411ece0578ace9897b5589d1fa
bd7a8db404493734154f78e3236b394253bbb5d3
39246 F20110115_AAAYZN abiade_j_Page_020.QC.jpg
5c03fa54083ac5f1ca271c6db7f06414
b85a5bca5e6c08559495f669007ec5bb201d44e6
F20110115_AAAYYZ abiade_j_Page_093.tif
0b509496ca98b61f7d24761dc6dffcce
6a897c2ca6e53f0c70a61e5340dfa3d603f619aa
187804 F20110115_AAAZFI abiade_j_Page_099.jpg
b2dc9b801e34961f95651e7a85dd602a
014debeda9a1c8b4486f311a10eadcb1e43187fe
351919 F20110115_AAAYZO abiade_j_Page_007.jpg
b966fb973d4cd34490ee01773d8b8b0d
a41210bf4d853c1914ff8009a676ac94f9f729b3
184997 F20110115_AAAZFJ abiade_j_Page_100.jpg
35e0112ecf4097ee9bab83493e845e6e
7a8d357e2118ec34d6f81eb42f23bdcc97398daf
209744 F20110115_AAAZEV abiade_j_Page_078.jpg
b9746f6d34d9e47f1aba84f1112d591e
a2c67f027107bdd191731411c75033333f6cc811
1411 F20110115_AAAYZP abiade_j_Page_005.txt
b8aff07d3cddb3d6dc778efe4699bf64
24ed7d75a730abbb2cec47076ce2e238a0333752
200165 F20110115_AAAZFK abiade_j_Page_102.jpg
5a0835e0d9dc25d65591bf44887468e5
11dfb6517795517ca15e4c9ac1113782f1436ed0
207197 F20110115_AAAZEW abiade_j_Page_079.jpg
43caf1f7bb65de0c5627ba0a5cd35a0d
d7a50555d0dc7fc10e1df65cd2185ffa61036c0e
42377 F20110115_AAAYZQ abiade_j_Page_088thm.jpg
f7ac8b6a6427606cdb2a60d5417d1bd9
b16d2abdbe418c3bbd4af915c858103a1bf7b91e
135428 F20110115_AAAZFL abiade_j_Page_103.jpg
ec58ac22ff42b02efac2a0f477ed708a
9e751f03879e8cb9ff6064674b38c37980033355
180448 F20110115_AAAZEX abiade_j_Page_082.jpg
07e12833d5736f4d403adbce61e9108a
b70fae232743c18efa8cc46171863ba945a3df3a
219921 F20110115_AAAYZR abiade_j_Page_044.jpg
b8e775f545e5e5554fdd3c59a973da8b
f5161ff168669e282e9fde5a5c835841c434b4fd
144258 F20110115_AAAZGA abiade_j_Page_120.jpg
3bb2d0230e7502694ab6a521edd5fdfe
58752bb9d1f45d5be63c26ca69524dc1d207f58f
229899 F20110115_AAAZFM abiade_j_Page_105.jpg
af9461e42e07c6f713b64738e0e9831b
bc280745c985b31511416df4ddea7dce65be172c
212620 F20110115_AAAZEY abiade_j_Page_083.jpg
937d69efcb4e59d58e7197836ec4e037
be5b9459acb06cd95f78b9504dfd8315a6dcd97d
107972 F20110115_AAAYZS abiade_j_Page_071.jp2
23389f0349273d57b3a7ae8c82326129
1f0152c2049df9ab69785b9b1af1cbbc45c8ba60
230442 F20110115_AAAZGB abiade_j_Page_121.jpg
89e2a448bbb005e86175e5fcfa6b252c
a8f2045bdd7c78a902fe2528a72f56804d2b0598
175419 F20110115_AAAZFN abiade_j_Page_106.jpg
9f2eb6ae2d97efbb81f39c7756ec0a03
29754319251c154504df130c797240324931a14d
195250 F20110115_AAAZEZ abiade_j_Page_084.jpg
ce9dab6e8413d722a1a4612a9b9c3e98
d44eb3b6a216709ba99bdd1e9c0fdfc141080765
80041 F20110115_AAAYZT abiade_j_Page_111.QC.jpg
a35903c6f24590539e5625486ec56bc8
69c6895f84ff8bc9391aebd69ac5d49a3cc5f45c
125235 F20110115_AAAZGC abiade_j_Page_122.jpg
628cda6250ee89be23aac6513976d1f7
5e9fdaa33fd52826d90e080badd5f1cdcafacc34
231536 F20110115_AAAZFO abiade_j_Page_107.jpg
8178a9755b08d49ecf2d827eb1289b8f
9513d710373a864e53dd25a8961618cf7d9e1c7e
27924 F20110115_AAAYZU abiade_j_Page_088.pro
9b84d7f1c6d7b3ee5ce00aec51ada94d
52e15a624729dbf7f7a547d06c68b5c78d0f960b
183922 F20110115_AAAZGD abiade_j_Page_124.jpg
346e17f9dcf4b3e4ae94386139b90646
d079d16b3df3d9d754fd231534850854c1ad500f
176928 F20110115_AAAZFP abiade_j_Page_108.jpg
67258937ec2824762f494c2249fc245c
4f5be36a77e189a5cffbefa2eab054b17fd16b84
F20110115_AAAYZV abiade_j_Page_040.tif
1693a37da74ed6e35270fee84bda824d
398b4c0fd2df1e361b2cc0e48b42128a58a4b417
154394 F20110115_AAAZGE abiade_j_Page_125.jpg
13f568ce1781f3352959578f90aecf36
62b599dca8b4e7432a83bc761637b31e9a9a66bf
220867 F20110115_AAAZFQ abiade_j_Page_109.jpg
e29031da3ef342386ad34e235afb3d05
c89fb0285b67e192f5115b1c7ca42937e8db938e
48101 F20110115_AAAYZW abiade_j_Page_102.pro
e6c142cd54a627fc605fab056f075fce
dc3b1555c4ab55b1cd589f29e7b40b5e8b6403bb
205779 F20110115_AAAZGF abiade_j_Page_126.jpg
17841d54df5865c6eb3bb1a7646d6660
aabada404469c3e863e1160e9bf4ae3a5e231520
207574 F20110115_AAAZFR abiade_j_Page_110.jpg
7992d1edd29c941853e3fa21019b6f96
cc31bbc096adae6a837f60dd5fdca6c2f62ed374
154920 F20110115_AAAYZX abiade_j_Page_088.jpg
4d00d0bca610f4d31214953b69440a8c
9e6f0a65d151371f03a0607898ecaccc86c302cd
189474 F20110115_AAAZGG abiade_j_Page_127.jpg
9d0ebd60a86017fda183afd7c294be5d
8d84e9160522c864395e5e3fb6f2a43b1a6a3eaa
77132 F20110115_AAAZFS abiade_j_Page_112.jpg
1cf9e3deffa8af279880ff2bf7aab9f7
fdb7464cfdd5c715ea22522a702a76a19cd4ce46
F20110115_AAAYZY abiade_j_Page_046.tif
ce12e94803ec65026c30453a99996a00
8b0d2172e5888a44db241536df879f389ac034a1
169138 F20110115_AAAZGH abiade_j_Page_128.jpg
61d5e74933ef0984ba9c2f2837447ffc
269c961a00b21a86fa3750430dc549ec25290978
176636 F20110115_AAAZFT abiade_j_Page_113.jpg
3714d49d3338b858d012292c843a863b
e7c299b9d7cfb9b233410b11d737b703d75e023a
1625 F20110115_AAAYZZ abiade_j_Page_028.txt
90c4055c32724cf847a5f03eee43694e
68bb5025cbdc983c5fa00b23faeb2212dfd99557
190079 F20110115_AAAZGI abiade_j_Page_129.jpg
e578335656d6fb7376c7d6c2d58f3451
3fc3071606003ae70ffb338f7163cba7ec2d06b5
191870 F20110115_AAAZFU abiade_j_Page_114.jpg
29ecfe119b55ae247a72d7ce2455437b
452b180f39137e72f03f38774d2686981ac60a5b
133091 F20110115_AAAZGJ abiade_j_Page_130.jpg
87638e44f093d6c58c816e9e5f854e06
fa1b1ae9caedee0d7e14ce2a9033eb3b791909a2
208555 F20110115_AAAZFV abiade_j_Page_115.jpg
0e0c3a9e9ba960eb87d1b9aa82700e5b
e44b5181a83792b21be0eff1b13850a97a4ce390
187894 F20110115_AAAZGK abiade_j_Page_131.jpg
518a7ee4d8743e80fc14bb25bc4ab392
19f8c089731ed2a51659744710540a2d8e19dd2d
203596 F20110115_AAAZGL abiade_j_Page_132.jpg
c248558b8795bf0dcdf6f64984ec07eb
af3a0390014aeccb78e65204003954a3afdfd28c
192559 F20110115_AAAZFW abiade_j_Page_116.jpg
4c769c9334d1efbe6688385bb01fb7d5
285cfd25b32a11c6ad39eedfdadc27c4c69f113c
218658 F20110115_AAAZGM abiade_j_Page_133.jpg
20e69d9cb5f96ce921a649c7fadddcbb
62f6cd0e1ba5a2012e9426089350fb51a68cb180
177293 F20110115_AAAZFX abiade_j_Page_117.jpg
d19b99c365d9f1a0ae563bfdd1f390fa
51cf2d696f646347349f957242c017b1453197f6
80640 F20110115_AAAZHA abiade_j_Page_005.jp2
11560c77373da6c02699c03956b3069e
e8e507e32911aaa88eb43e4c7e81b3e5cceabaeb
210650 F20110115_AAAZGN abiade_j_Page_135.jpg
015299cd83281eba2d37637233372491
f92e4c6a8d3296b4cb1c1bee92bccdc9a5e47142
214172 F20110115_AAAZFY abiade_j_Page_118.jpg
f6469161c2a871a54a7a922d25add0f3
16b9ea8e60453b8e1fa75731835f551c9c0826fe
1051985 F20110115_AAAZHB abiade_j_Page_006.jp2
d3983a00cf29646e77f3a79032d6406c
b45344f89d280a8163b82139f355c81e78e49251
156141 F20110115_AAAZGO abiade_j_Page_137.jpg
68d63094eb9e5724f5a8b6faa5a56536
b54177dcb5d55059ba51c6e3d6a5399d71397fc9
199542 F20110115_AAAZFZ abiade_j_Page_119.jpg
962b3902f17e617de93c31b039d2e82b
a2911fdc51e99017ce00defe8f957a01f54d0a8d
F20110115_AAAZHC abiade_j_Page_008.jp2
fc89c10a95df2f6be0bb9d11c98a7ffc
60a30ac75d7ae3eb74c956dadee48685b8615b4e
98912 F20110115_AAAZGP abiade_j_Page_140.jpg
81e6b3a2b701a6dc5f7af02576f8683f
05be84b55ec523375a504170e8de5d197f539ec0
237514 F20110115_AAAZHD abiade_j_Page_009.jp2
82bee1c3530a19e2ef33dd94074cb376
0de921764424460f6bab378b778b208a3865c102
210387 F20110115_AAAZGQ abiade_j_Page_141.jpg
8c4641525fd3ead1902454bee5824989
053df34f70390f5bf465afc4fa33319926c1ed59
1051964 F20110115_AAAZHE abiade_j_Page_010.jp2
22c93641152f1dd0c2f5c1d5386ae91b
5fa2b78b8f3ec142c6a71160f2aa2a655d7a7f2e
219494 F20110115_AAAZGR abiade_j_Page_142.jpg
751dcfb755eab3b81416055160313a86
e894e3cb9086068b32fc0285d26e314515b073f3
1051983 F20110115_AAAZHF abiade_j_Page_011.jp2
989fe94022107deef6c02ff284e4e7ee
0f7b33d2de28ca630917164dc4f779cd87ea21c4
219585 F20110115_AAAZGS abiade_j_Page_143.jpg
2a052df87c79143068c9b52ecc23cc7c
2c9bff4ead65d041edb46709fc99316b6acc02f2
89419 F20110115_AAAZHG abiade_j_Page_013.jp2
06e0b279c484d14bf766dbe71ef16670
850ecdc1222b18a940cfa4578e8e3137be2cc3af
214726 F20110115_AAAZGT abiade_j_Page_144.jpg
7d90b59718733481529d4d3faeb2413b
e97a0bebf406cd129e70d42612dcd76e309a8fda
912347 F20110115_AAAZHH abiade_j_Page_015.jp2
2fe8c4ee933419f69eb67f41e725f27d
c80168d091641b16ad7d222342eb19244569c47d
223493 F20110115_AAAZGU abiade_j_Page_145.jpg
376520e5850c1104f7f9d78eb1d36fe1
013ab1fb1ea6e1c393cdf3d774e3bc26298963fe
109567 F20110115_AAAZHI abiade_j_Page_017.jp2
5252836a53ce41bf498a4d0667fe3e9e
1e2dbb54ced6fa0d6e969ca4e56f3c7b87dd6524
94959 F20110115_AAAZGV abiade_j_Page_147.jpg
bad9396fc5c268a5dd8d710ab4921f3d
b220efc3e8827b19b6903d798439079e4a7fb2d9
F20110115_AAAZHJ abiade_j_Page_018.jp2
0a3bd25b89ec36721c92f781c56f5e9c
a47c0d7b34d6d0e17460e42a3b6d03f883fd548f
108345 F20110115_AAAZGW abiade_j_Page_148.jpg
f7d86075e46c541ea41355b4f4a306a3
3e3db6fd471d983cfd896feee11554b5330df79b
104247 F20110115_AAAZHK abiade_j_Page_019.jp2
4e55c855709625c98e90ff2f9262ddd0
cd26eff08695f0f75c95a0c90fab3503ae8ee2a6
53714 F20110115_AAAZHL abiade_j_Page_020.jp2
d23296f90c8ff556dc1bd33d844e08ba
e6669aec250d589d785bbab251663546ba3fdb93
26010 F20110115_AAAZGX abiade_j_Page_001.jp2
26614d3eaeec582092cd1a50af282835
32304d59044439fcfa1262d21c3d9ab173f92d4c
106004 F20110115_AAAZIA abiade_j_Page_040.jp2
a9a8105507fab31ce0e95fdf63c0d8c1
69d51df3595691dd39a06a99c4757b38f898092f
110074 F20110115_AAAZHM abiade_j_Page_023.jp2
211b915685e63c26342873339c15f9a7
7b74167b1b11be575af9184c8164e6fc85267b7d
21140 F20110115_AAAZGY abiade_j_Page_003.jp2
d8cf767bf053919ac81bfe53b62c803b
32a4f47940504b7f29207f32d2e44662000137a8
107347 F20110115_AAAZIB abiade_j_Page_041.jp2
9e977c430eeb0854e16b1c36111e2709
325b470a0791fea3f7c392e2285af5f8d7b574f8
F20110115_AAAZHN abiade_j_Page_026.jp2
587f1437ed86afef3770fab0966d7b0e
75d67b2630381b610b8a508f68a05c6f22f590ab
98635 F20110115_AAAZGZ abiade_j_Page_004.jp2
bc4627da1c8ff4e49627f4e51c24fff0
21ce60e76364543d41491c7be7704b6a8bd1486a
104378 F20110115_AAAZIC abiade_j_Page_042.jp2
bd5d770335af0823fdd2c946ad25fff8
15054dee0da57ac8ab0d4d938891096f7755a8fe
114229 F20110115_AAAZHO abiade_j_Page_027.jp2
7159696ca093f3b7778b90fcc9f99dbc
2d36f558e9ddff0a6fee65bb2340fa9d4b051452
1051960 F20110115_AAAZID abiade_j_Page_043.jp2
aa9dd1eea3f5cf838f17341cfb2caac1
c483fece0b9886d8d3b06597665570a3d58c0b98
690485 F20110115_AAAZHP abiade_j_Page_028.jp2
2b14f21dc1f04ce23e9d26010632e24b
090d50a5df243c1e6c008e6638647b29e95b20ab
113101 F20110115_AAAZIE abiade_j_Page_044.jp2
4898b3b6bff47e4c6a11466474fa4559
c1a0e07bae5ddef2cfbf5eb8f108051c4af88583
111075 F20110115_AAAZHQ abiade_j_Page_029.jp2
2d33f377f21d269ff2d4bae49702391c
e4172df9bf7ec3393a0ea35418558801b47f801d
928001 F20110115_AAAZIF abiade_j_Page_045.jp2
8fbababf3401ee5b62528b07c3f95f89
26a4d51ff6e485638d0ca9794b2600154ae7ddbe
108838 F20110115_AAAZHR abiade_j_Page_030.jp2
75da8dac2050a4bb291b6eea77888834
21023dae0d9ad074983f635d7fcb47e20ae3c8cc
650609 F20110115_AAAZIG abiade_j_Page_047.jp2
c6d9ae49b249ddac2662049b40ec64c3
0299bb89bb479c7379a64699e3acee6096ee8022
905191 F20110115_AAAZHS abiade_j_Page_031.jp2
559110017a8e12d37fcaab8dd46a018b
6e026e62756916b5ec9b5459d7d2d0f3054c6e50
105911 F20110115_AAAZIH abiade_j_Page_049.jp2
8c8a945454fab60191d88a7a24d4f0dc
63aa5f159ce610cb04048b846710ed0a361c46c2
111915 F20110115_AAAZHT abiade_j_Page_033.jp2
0f0a6ba4430ba80acb3304b3b6341360
b565a885d5bf1bc80a8b4e3343605eb1a179126e
103717 F20110115_AAAZII abiade_j_Page_050.jp2
3f4cef14022d291903cc9cee26555341
37a4646a7cdd036912a413b1a6fe739039089957
102306 F20110115_AAAZHU abiade_j_Page_034.jp2
6c8e1a6f1d58ec4edcf9d8ab6c5a30cd
ff791fe689d7af0ea6d2d26a90317d857c547ec3
111032 F20110115_AAAZIJ abiade_j_Page_051.jp2
325f3bf4432bc115f6f1560ab409e326
e28f28004fa403cc5558d1b925610977adc6101e
111772 F20110115_AAAZHV abiade_j_Page_035.jp2
797b223400a5c20830a49d34f744c3fd
d44057113559062bc1fff88455b084be2bfbd926
107674 F20110115_AAAZIK abiade_j_Page_052.jp2
b7245ef8dee9acc7fd9859be34a2209d
3ced5f0456b7c431c702efb00fa6ff0e7a11c6a2
109924 F20110115_AAAZHW abiade_j_Page_036.jp2
4122c773957badd1ab1a50e60a317da0
244ccecc4d53ea49afd210456880a47e4b933f50
1051984 F20110115_AAAZIL abiade_j_Page_053.jp2
2f14eb6cb08edd20484c28aed3749745
e2f855abcb9a104954102c79a85c1f114d5589c6
1030919 F20110115_AAAZHX abiade_j_Page_037.jp2
3fff9b95d81dc32290209e9ca9eac9bd
9782c3da13395fd5a597bf42e6b998c5bf2784ff
107653 F20110115_AAAZJA abiade_j_Page_076.jp2
f7c00f28c42968768350f9161cfcf2f0
b4c31faba45af366fd32590ee93afc9fce67ac53
736435 F20110115_AAAZIM abiade_j_Page_055.jp2
04d2595a9e553628659d3c6e83843dea
e11237e1aece338ce0933302e42244f80856778b
108985 F20110115_AAAZJB abiade_j_Page_077.jp2
eb617ca57df13732c67af42591b0c0c3
31b4525a6cdf6f0b29fc9311a04295c0002f138e
766898 F20110115_AAAZIN abiade_j_Page_057.jp2
0836a7b7d13050b97dc25dbd9ffbce0c
82e95065d3dea10df2d1546e358565cfbf3a07d8
817371 F20110115_AAAZHY abiade_j_Page_038.jp2
b55244c2c627de589921c3af276c9003
a8bee61e53e02ef67b98ec8c27947f0d57a4f0b4
107361 F20110115_AAAZJC abiade_j_Page_078.jp2
32efdc573f8a449262db97b0f5d68b5f
1001ae9bf0abed2d12b39a63b5096c5b3fa66e2d
66949 F20110115_AAAZIO abiade_j_Page_058.jp2
417bb571d541d51dab3793fa582761c6
2ace255726a6c8c284fa5731add7c2d4f671996c
114500 F20110115_AAAZHZ abiade_j_Page_039.jp2
431823d7afd458bbe8de8310f2850530
64e4e9610f5cc584fe09a86ab097158888a73338
849048 F20110115_AAAZJD abiade_j_Page_081.jp2
b2785cdf5e1aa891de904b910b364a29
906b3662f56106546fc93d2b02fd6fcb478b853d
78992 F20110115_AAAZIP abiade_j_Page_059.jp2
fa354d676bba930628e7bfc65b13d4f6
0bb4f56b9756f1b5cdcaf4c1b5c015682eba500b
975463 F20110115_AAAZJE abiade_j_Page_083.jp2
6f3749eb41cad9e02700f1f168b38346
f6c9b29e01f6bac90af7bee368545df32d996b54
85165 F20110115_AAAZIQ abiade_j_Page_060.jp2
1dae2127750f85321e5235bf129c2f12
7f713a3b877afef1e48345cb53fefea4ecee4c99
875679 F20110115_AAAZJF abiade_j_Page_084.jp2
9e894cc83aa815938b007fcdd1df9bd6
9846259b8df7000a8946dc616f2708c1f21db21c
83269 F20110115_AAAZIR abiade_j_Page_061.jp2
360405c7142e16d94522b73f4a821896
1ff9b54aa756d84b3b66a06a3cbad2e36300119c
654500 F20110115_AAAZJG abiade_j_Page_085.jp2
5981c227308e0eca07922fc88f157365
87e36fd6c4a316a91b63e61aa145e6aefb084de4
1051979 F20110115_AAAZIS abiade_j_Page_062.jp2
dca3efe9071b0dada06b0a7b888ecd87
4741784d1fb96c5bd9461ba0cb1beb7125aa916d
978574 F20110115_AAAZJH abiade_j_Page_086.jp2
b98897a97630c293b39d9b495de7cfd6
5092f3deaaa0166655589d6f2b759e70997ddbb5
841390 F20110115_AAAZIT abiade_j_Page_063.jp2
e071f223933a0b69f2da6dd3146ec4a4
e0284a1ab554fdcce5cee30776c54fd5259749c0
108521 F20110115_AAAZJI abiade_j_Page_091.jp2
c31b9dc4dc532d565381f2104482772e
ce9ca62ea0f124a37a3e527ca90829352c8126be
99448 F20110115_AAAZIU abiade_j_Page_064.jp2
506f5dda0074eb26d17503574f328ac7
70cb5f4bbd5e429fe8f128f30e03ceeb5abb0d4f
112348 F20110115_AAAZJJ abiade_j_Page_092.jp2
8e84544c9e68ebef4081321d14e7d25b
037448ba99a6ee32d66211dc0039e458953a0343
101150 F20110115_AAAZIV abiade_j_Page_066.jp2
313eb70922a505524b45bfb539c2c827
14364c6b11116fa50679ea62d3cc972f60e95fac
109507 F20110115_AAAZJK abiade_j_Page_093.jp2
27749ef8e99a33de9dc948379e895eea
88a4d09d1d454b8c7a56e643dd667799a183502b
87413 F20110115_AAAZIW abiade_j_Page_067.jp2
21206a1ca1a42c6ebbdc4ac7b5ba5e51
835661f597deee4d1c5106de7be1adf09f2dfd33
1051966 F20110115_AAAZJL abiade_j_Page_094.jp2
fed3b918d63740c8defba7fd757e8907
12b296363b86eacee8219ae6504e8323bd27c4c6
111185 F20110115_AAAZIX abiade_j_Page_069.jp2
b0fc01080d915dc824f483d703248644
7f99bf569658ac4304d34100fc2cc1ce40ae00af
109768 F20110115_AAAZJM abiade_j_Page_095.jp2
3261d6cd67ddfeb08cf88e98c57f6b8e
bf2cdf98e937e7183aec122885e7a002af28a29e
111237 F20110115_AAAZIY abiade_j_Page_072.jp2
1d50bc5682a78e59346283496aeedcf5
dfb8efd2abf402e3cb4aff32a9a9956170e79a39
108100 F20110115_AAAZKA abiade_j_Page_115.jp2
48c53021f027861f98bf6d8f037af6ae
d0bdd779a1125e3f15bab5de0f6e329e9a549992
110779 F20110115_AAAZJN abiade_j_Page_096.jp2
96cb773f70c6730bc22410bcda5a1444
e7e2348c5edc7b64bfd3c46e3de0da59f665a19f
110791 F20110115_AAAZKB abiade_j_Page_118.jp2
5417dac9e9328a862e3374daebd6df1d
41d2c6700352587a9c8b7b69d40c41a4a6991a70
111127 F20110115_AAAZJO abiade_j_Page_097.jp2
0976ce764cc8128872f006607db27712
2ab6e19375e1f2bfa092ba0af521566bd531dd86
99206 F20110115_AAAZIZ abiade_j_Page_073.jp2
2c14b12bfa33f2088dd02d65d21c6562
ef1acb69d2c3910c787962dcdb74547965ee62ba
104403 F20110115_AAAZKC abiade_j_Page_119.jp2
29a7ecd737ed10195429ac232d94c710
d7d0213da14687e04a732e8f7f37867e029f1ff9
101819 F20110115_AAAZJP abiade_j_Page_102.jp2
3fbcd2d091b045e7f887ec0720678f0f
d3286dfc5e5b6203d8cbfecf59305f8f09f81ce0
60052 F20110115_AAAZKD abiade_j_Page_122.jp2
23dceeb60f40714db8289456718198ab
e46748d230c4736265ddf587ba453b2ef9941e56
814351 F20110115_AAAZJQ abiade_j_Page_104.jp2
f6ee6e3f39653b426360f132c794e377
7cd95a15325e1b120d7016fa5da653dddb916d59
876423 F20110115_AAAZKE abiade_j_Page_123.jp2
95806e96a23846a0adcc6800eeb4a43e
41754b60f57f8caa9896d7994510d27fa2ba597b
1051980 F20110115_AAAZJR abiade_j_Page_105.jp2
e30732333df15b565e9a14667d450afb
4c45ae3ac0abc4177ee00ac1048a1b4ed9013527
1014930 F20110115_AAAZKF abiade_j_Page_124.jp2
82b18b7721b039a0e5f7ff5f4e016e94
4e0371bc8dcb461e37b38487a574005b9a0f6675
1051986 F20110115_AAAZJS abiade_j_Page_107.jp2
ca3b34ce36add6dc93c1c78c2596a92a
440977519eadd4fc84154a00003c7ba2bfafae3a
F20110115_AAAZKG abiade_j_Page_125.jp2
6f1ebefe7bd59f3f3030ef143df578d3
88d35fc6ade8ad3eda05367e2e8f9967571a4a1c
763849 F20110115_AAAZJT abiade_j_Page_108.jp2
d4474926ee8c2ff3bbb992970b54b667
a2ca14c0c5a2366c3b505136696ce11857621005
1018126 F20110115_AAAZKH abiade_j_Page_126.jp2
9e6ede2896cb1f989067e0e536b68618
9941b5ebb180c48b175d11d541fc023fafadee59
1044362 F20110115_AAAZJU abiade_j_Page_109.jp2
87e07d1ac25c01efe2cb60c1c42dfb1d
a85232ff92f8cb06117e918a9b9a1e7ad798a90f
836294 F20110115_AAAZKI abiade_j_Page_128.jp2
ec917a1cd97fc82e6dd3d9e57b9f99c5
2a0956daeb079f571680296ae5fc431470b279ed
105612 F20110115_AAAZJV abiade_j_Page_110.jp2
a0a7e7038a788aacf05ca04ebe382fc4
8b35c1323fcc3de2aaa47a8d46080dd02ce5acb3
862779 F20110115_AAAZKJ abiade_j_Page_129.jp2
6c4e77d232e2106ae460984c0bd5655d
32324b2ad2c7cb229dbf05788371aa0a4f97e563
110881 F20110115_AAAZJW abiade_j_Page_111.jp2
cbb99bea1f805d7440e9b496b4457773
0bacc30adef31100d5ef73cc0b46abc54ec3bf8c
93725 F20110115_AAAZKK abiade_j_Page_131.jp2
d963840dd52acb5e278075c0cf9893cb
d993b65f2c81eae4976967876df1f68d8035b9e0
40362 F20110115_AAAZJX abiade_j_Page_112.jp2
ba356a5a2255bb084ff34c59234c9677
6506bf709b3eeebb3d458de3e26e326572f1f235
103929 F20110115_AAAZKL abiade_j_Page_132.jp2
c5b2f56e135dc2338ca42f9499b525f1
30461c22a72612e2c094710910eacb4c1a05800e
F20110115_AAAZLA abiade_j_Page_006.tif
ef07cff01afe121c4913469e486b6907
7378cd14d50f03f60c90953b11186c4acec3cb68
112379 F20110115_AAAZKM abiade_j_Page_133.jp2
cdbe53fe2c95db92d04abcf24fe9fb99
6696ccb73e3be348496d4049fef42a7f2b50e09f
88610 F20110115_AAAZJY abiade_j_Page_113.jp2
1794b48593a0470f1068ba63426fb7c4
4432e22f927d7e4adb8af9240b6ba4c0c8a5146b
F20110115_AAAZLB abiade_j_Page_007.tif
f12252eb560b4429418cc3e94afb923f
21d55d44c57e4f629bbc9346239513dd915ef1f0
107229 F20110115_AAAZKN abiade_j_Page_134.jp2
09eabd49058a2771cbbe49cf76841791
dc49d1f8623645bfea4ed45747c7b6326b86a525
99879 F20110115_AAAZJZ abiade_j_Page_114.jp2
3ba4d305540d8034d509de6189693c13
62ff94476a142ec8a30f7e1f385f3ca4aef709ba
F20110115_AAAZLC abiade_j_Page_008.tif
28782679c7e369643d062d01580d821c
3564ce127e3d47ef32a1c5971083f10cff8486bd
109931 F20110115_AAAZKO abiade_j_Page_136.jp2
41ef7298d1eeac25e26e353d9579fe28
24193c68225968fea1738b7633d5b9c01b69d9b0
F20110115_AAAZLD abiade_j_Page_010.tif
ca5d19700a56537b029d777ad17186b6
cf88a500086e0ea5d2f3e748312541fb628fb98c
93525 F20110115_AAAZKP abiade_j_Page_138.jp2
274921fc64914e38376bc9b7292ed4e5
e8e9c35f061cbc928eee23f4a69fb5c3fafd7c03
F20110115_AAAZLE abiade_j_Page_011.tif
e8718b5528f03b903e4ea81947b52a22
5fce782dd6ae8aca61616ea593f9d682b9a679b2
113071 F20110115_AAAZKQ abiade_j_Page_142.jp2
dc6b055cf61da5405b9ec6ad7d933a17
9ec2d4d6b2323b2659d459d2d0b075c99befd9ef
F20110115_AAAZLF abiade_j_Page_012.tif
0ba1277749737a389bc983e8d8128e80
38278061a07ca748c80ce2d71c9ee30fa26d351f
107778 F20110115_AAAZKR abiade_j_Page_144.jp2
2776a050fe86d7c2c41d8927b59dec2f
6284e7427b28289d1c856fa4c6f2fc1d758ddfef
F20110115_AAAZLG abiade_j_Page_013.tif
563207bbf2d6aface7fbf0ba4d3d5055
d67a62d6ee83831c11989d7ac7b8c6cc81cd9783
115799 F20110115_AAAZKS abiade_j_Page_145.jp2
d7f2f2d36da65c8943ef1295de6c64c2
6ff028a7efcfcb267e89ac9f8fcd07d3d729b2af
F20110115_AAAZLH abiade_j_Page_014.tif
11ff4a22f0aee872b56ed4c81bc8d78b
12566a6c36a2bdb0275ad2d08f16ef94032a569d
48522 F20110115_AAAZKT abiade_j_Page_147.jp2
e38eca4be826e0b79907d8af0bb61853
ef1b7980339902c0a3fd0e98ae17e90646536a83
F20110115_AAAZLI abiade_j_Page_015.tif
faf926ec179c7cacaabd286a0a3a3a22
c9e68ce3f44666969eb2d0ab4992c729644f951d
54438 F20110115_AAAZKU abiade_j_Page_148.jp2
6ad0166c5311badceb9807507ad4643b
a80c917bb7a529d6124f82c512a51de79fffe0ad
F20110115_AAAZLJ abiade_j_Page_016.tif
770de93486913c94d6e064ce47c3c0fa
bb2b6a607b92833d035e0a8f657b3ee2d106358e
F20110115_AAAZKV abiade_j_Page_001.tif
9507805016fba1bdff500092518c8943
2e1a3d6c3c58acebb3124051729f44c6d5115679
F20110115_AAAZLK abiade_j_Page_017.tif
9bb6a350d4a4458163e900545e957c27
8990b3e39bf4f6dcc3ab99a74b60f9b23ca0a765
F20110115_AAAZKW abiade_j_Page_002.tif
ed1094f4ccf204bf10756e93591eb412
f1f463d8abcc626ecb8824b9863463df9466750f
F20110115_AAAZLL abiade_j_Page_018.tif
f943024c3d4533aed2b195a92af4da5d
082fd7cd05c271c579314b45b474e0d461877436
F20110115_AAAZKX abiade_j_Page_003.tif
dc5672caad8270b2b61ed41f1dd95556
c4bb7f64cd9ae54509d1d3812e8d5cbe99751587
F20110115_AAAZLM abiade_j_Page_020.tif
45d3056ed1c2705b2cdb0a919c06cf59
8399874f14f082894cede9478f174c981f0e6e4f
F20110115_AAAZKY abiade_j_Page_004.tif
4088fb088904a935bd75c2b944c3235f
5f642702ce4d6f4b2eefd2cc4174be3704acd11e
F20110115_AAAZMA abiade_j_Page_038.tif
7483cb9280844387f3d3a0b3a2f884d8
30600bfc90d4093fbbb65bf2e7225b85e59da91b
F20110115_AAAZLN abiade_j_Page_021.tif
aaf009f4f18810df98baeb275d926584
c7daac1dc0d6bc6ee3735dc234fb69aabd26883a
F20110115_AAAZKZ abiade_j_Page_005.tif
ca231417febba4516c85b991e43959c3
f319e75900b7d2c42c127f231e1803de25873b76
F20110115_AAAZMB abiade_j_Page_041.tif
5687ad8cb3c5ba372cc7760583dd3c66
dd3655287b07815473aa23f63e0dd817e2ba3a82
F20110115_AAAZLO abiade_j_Page_022.tif
0002ba4263b6171de857d863f4112704
8a847e1af5e415f0c6ec124fad24c55738d7b9e8
F20110115_AAAZMC abiade_j_Page_043.tif
47ad7d4c937a5d981bdbca23d934af46
488b17beec8d1353673477f13a7d1b08ee866317
F20110115_AAAZLP abiade_j_Page_023.tif
a75aabae2b5ffe1f1625fa6d30ae7d1f
5088d1987c7bbd19455a1beffb5ed8f8bd5ae3ca
F20110115_AAAZMD abiade_j_Page_045.tif
13e940484ba237ca07ed5c4e87c5fd72
94461da99a6f986de3c0d627252fc2ac3d58a5e3
F20110115_AAAZLQ abiade_j_Page_024.tif
e9408d731ad231b3730ee0ea7aea311b
99c480fde573925becc587262cab4e663810f98c
F20110115_AAAZME abiade_j_Page_047.tif
f6ffa3a11310ac8494c43dd161252532
55d32eed5c9e3e37f4ceac29586b9801099a3c83
F20110115_AAAZLR abiade_j_Page_026.tif
730a4f9939e1f58fa0c30409de606dfb
cd7f12e66cefad02a204977ca2fdbc4a3da75b6e
F20110115_AAAZMF abiade_j_Page_048.tif
bf0256efa613a67fcc5d6be293a62c5c
8ee82758e378caae42950deb053ba12283c0d47b
F20110115_AAAZLS abiade_j_Page_027.tif
67aa8b60cd5060b267ab9bb83dfa33d1
5a3e2cd0f86dad6907a4e104d780a778e98235d8
F20110115_AAAZMG abiade_j_Page_049.tif
5f1d810eaf7068eabdc54dc8861d3be5
2c46ebf85ceefa9a13baa8073be448c2074c613a
F20110115_AAAZLT abiade_j_Page_028.tif
defa64efaf2b88f57aa17b18a4b448f3
6b3e0b6ff69c70d7173f4746fa4af3b5fce3b9a8
F20110115_AAAZMH abiade_j_Page_050.tif
588546d41b0262d8a6851929fbaf53f2
8e98d01348ad3fefec5d579984e96e30441778d4
F20110115_AAAZLU abiade_j_Page_029.tif
ebbcefcce8d461c1b9d9a69f02e6d4d4
74f4ed4f78772d98ac29d79623320a78c8f6ee20
F20110115_AAAZMI abiade_j_Page_052.tif
6c87d57b727b2dcdf49f59238a6520d0
1a6afa5c25e096b8578a00f934a8288f2d3f806b
F20110115_AAAZLV abiade_j_Page_031.tif
b5111a874eb7c67bc3bfd0bc06ca06bf
22f830e5cf0e913d803efd6837adc0c71f934bd5
F20110115_AAAZMJ abiade_j_Page_054.tif
f54ccf9557ce3cbd55b32c597de1a03d
a125da08831913d4c2aebf31d7494d326a912997
F20110115_AAAZLW abiade_j_Page_032.tif
bfb5923fb46e319d7eaaaf2e946295f0
e7b298ce33214abb713362506d794e0808f4f01d
F20110115_AAAZMK abiade_j_Page_056.tif
c02ed559e40c901bf64fd29d36014f27
16948d4799a9c44d9e09290e382fa9e968c23453
F20110115_AAAZLX abiade_j_Page_034.tif
7d7556214f8b15e203ffc2e6cd0d037b
b31d14f455f7e9337c4df5241e6bda1da00db113
F20110115_AAAZML abiade_j_Page_058.tif
b9a6ba547357e4789a59db9db747b27a
d69c61654d36136f07629e386c4b8eeb77e1ee5a
F20110115_AAAZLY abiade_j_Page_036.tif
11e9e6ef7adc08111b879f51c6f5a991
be9015bdf55cb7dd3be3f70e8b40d823f756fe38
F20110115_AAAZNA abiade_j_Page_080.tif
e4a7e50c1ea0fa956d9ebf5eac6f3b8e
5f85ebb53d38fb8195edf734490024cc6528ee67
F20110115_AAAZMM abiade_j_Page_059.tif
cf8910c3902464f504ed3fe57528b38e
a941e5283ab09a52fe1b6c0406b9364fc39a8840
F20110115_AAAZLZ abiade_j_Page_037.tif
122ac7b206813eb24068ebd018da1536
56f5573d8148593715dcbf6884b50b9b8386cc5d
F20110115_AAAZNB abiade_j_Page_081.tif
bca26d9af0a81f4e7c7d87321b9106f1
419a2f3cc9bd422a8d34920aaa2bd99aa38e0827
F20110115_AAAZMN abiade_j_Page_060.tif
d2a10d09c053bab31a2c36e88723e59f
6a7b05470c6ae1e3841f16d9aeb1b77428f45dde
F20110115_AAAZNC abiade_j_Page_082.tif
69cef28cea75c3a68f4cc3cb59339e8b
06e1e74be6120855f318fef60c83ad7b0b09e86b
F20110115_AAAZMO abiade_j_Page_062.tif
ffc7e6dbb2133f4d273e401cced43979
f62d453f460b106463b2c78ccea211eb4739686e
F20110115_AAAZND abiade_j_Page_083.tif
2f50e83c5d19b9e114042682f0028e4b
73b1fc5f2e5330df3ec1a2b52922a7ff891323bd
F20110115_AAAZMP abiade_j_Page_065.tif
c9c7a9ae5de1248c4d0e13b9755c57db
c96573ebeb97ed269ba2ce86d362dfaef1b96894
F20110115_AAAZNE abiade_j_Page_084.tif
4f794c2d59b32898556e38469b2136a5
a325466f05786a5a43cf536e9498a2c3000a3f1b
F20110115_AAAZMQ abiade_j_Page_066.tif
5c30b9105cde6916c7ac3f95caa4ead7
f90efaf568e10645256d8c70b01c18cfd3ea1037
F20110115_AAAZNF abiade_j_Page_085.tif
640e1b644e53cbdc3b8ba0b6a400166c
9f7730769a153040b26a735a63b42fc25cdcf00a
F20110115_AAAZMR abiade_j_Page_067.tif
69e5d769452f10f6177e71679e03bcf8
52bf2c362b58a8b390eec9bd97da1d037898bd4e
F20110115_AAAZNG abiade_j_Page_086.tif
45bcf232be44fb3fe9dacc72591afc7b
cce7f97ce49b14277c340df2c60af036be7d9a5c
F20110115_AAAZMS abiade_j_Page_068.tif
161bcb6ffc3339b179f7fe1db4350b23
50111c3dabb789b6baa1689f10bb1e60cd8f808b
F20110115_AAAZNH abiade_j_Page_087.tif
43f8ea15035935fd9cd8e8812becc885
f12868d48798baa6ea604d7c1fd417363ef8588e
F20110115_AAAZMT abiade_j_Page_069.tif
a07d1482a4e95ce9d765803123b74be0
80bd6cba1572888bc4722f7e47c3c2ed32b493c5
F20110115_AAAZNI abiade_j_Page_088.tif
5ac7be227236f8452fd6486f8207d709
0d6bf1da0eaef6d54c0590b436a76480895f589c
F20110115_AAAZMU abiade_j_Page_071.tif
d82c077c9dfdc803c6d7c3e62a4b5e36
fe296b771401c128040cc6216caab05729b0052b
F20110115_AAAZNJ abiade_j_Page_089.tif
466d879dd552ca7f2a84f6d2f197b413
058ca30dd42dd77e2a543ca11866d66820e24505
F20110115_AAAZMV abiade_j_Page_072.tif
638537e81bd8dce0329dce7a68ba5ab0
dbf9e49d44bcb0637d12a9f465d05fc0ffff67ab
F20110115_AAAZNK abiade_j_Page_090.tif
76855832a18a5ad579232544664ac65d
2fbfcd92ee59a7422527402071581c34cd8974f8
F20110115_AAAZMW abiade_j_Page_073.tif
aa9908041ad2838da02257749eff98e1
5d12472cb164c66ce19f7813c3212ac4d2545c92
74322 F20110115_AABAAA abiade_j_Page_004.QC.jpg
94ba487d5ce1863b5e0aa4379cd5d0c7
f09cfd7d855d6cd9bf5e52bd03c39989f9258190
F20110115_AAAZNL abiade_j_Page_091.tif
e92afcbaca3324209eadb1f3d2a8b4f5
45dfc2440334f206b9df18d4ee457bfd1b1da866
F20110115_AAAZMX abiade_j_Page_076.tif
0478a8ffbcb319649fac83339cdca3eb
5213ea74e567e2c37bcae0fa8545e03525b6b3e6
39340 F20110115_AABAAB abiade_j_Page_140.QC.jpg
8ae340fee48cba4f2e54c9a4b7ec03fd
3ab1710aab412c00acf765993f22d146c68b76e4
F20110115_AAAZOA abiade_j_Page_113.tif
d0d521cdf0e2f84e1a5b80c077046575
731585d16a7d8faa2eb8f3e90d22ce7aadede6b3
F20110115_AAAZNM abiade_j_Page_094.tif
e9fb5873bd9318cdaab5a3f1bc0468b0
99b3b954adcaf7a8d0ab6438c5f692b8a4306981
F20110115_AAAZMY abiade_j_Page_077.tif
6ccfede8872fa846b1af61282f237fbb
9ce83a2b20122b533c8d76cf4590d63dd0aa6ac7
43150 F20110115_AABAAC abiade_j_Page_006thm.jpg
beb4b5a045809fc28f7c3fca5b60aa8c
d8cc4d96ddd9265d8acce6105e15b8b0e7cff532
F20110115_AAAZOB abiade_j_Page_114.tif
96e314673cd32b67ce5a1eeaffbc8a3c
dc0b09918cae5b115643546fcfb72e9369156067
F20110115_AAAZNN abiade_j_Page_095.tif
1f1c63bbaab8fabda8ea6916c67075b2
e9e9175e51f73cd19c68feb52b7b7093570cdeef
F20110115_AAAZMZ abiade_j_Page_079.tif
e8c5bd9aa89bb277441ddef257b6c6b3
122b4be7b2e65cbb3cecce3f45c1ce181b084b8a
20622 F20110115_AABAAD abiade_j_Page_138thm.jpg
c5b64f94f57b8ff3a03d8b8293488524
0df2b569b0eb921579aacd052dc1e25dc1ba4f3a
F20110115_AAAZOC abiade_j_Page_115.tif
34fb5e48fae331a33561b7b7f07b88f7
bab9bd36f030aef913f633863aec0d2b45a20853
F20110115_AAAZNO abiade_j_Page_096.tif
336fab55dd72853bb5bddf65d8d4278e
8802c6fc94c717f7f09617bab290380811ac4acd
25036 F20110115_AABAAE abiade_j_Page_027thm.jpg
2e009c6745e3b4bab56df454b0c95c49
878173276a570a48cf7640f14c53efb414bfdd25
F20110115_AAAZOD abiade_j_Page_117.tif
7cff409908f0b9811dab2d9c661e75e7
87a8a42ea951be49f33062d65767a2db2afa6c76
F20110115_AAAZNP abiade_j_Page_097.tif
ad3ac19bd8f42a6b6b2cbbcb8d02e748
0c401b7266edf1d0e456a8fe4e0c1f24c162498b
53066 F20110115_AABAAF abiade_j_Page_047.QC.jpg
faf75a1f439d37d61557b33932513bb7
b65f6c11cbf24cf04d974e05dd316eafa1c0e04b
F20110115_AAAZOE abiade_j_Page_118.tif
1e8f5e050248a8f20b63c497c1f31fc9
f759e8cc115927780203c6345cfe1d407110e5b0
F20110115_AAAZNQ abiade_j_Page_099.tif
b7e9685a3fe0aa7f7cc159b3fc5bf085
651251b34b6001e73f14d8c54b5b8d80823c3216
67046 F20110115_AABAAG abiade_j_Page_131.QC.jpg
eb4c16dbd06f513162369daa07743a59
9ad21e21d9818162323b772da4574847f89c71de
F20110115_AAAZOF abiade_j_Page_120.tif
72b73be9417ec51bff857056a5b50cbb
e97ad159e97bd78ff35794db7e817e952d8ce1d6
F20110115_AAAZNR abiade_j_Page_100.tif
87ce8154b1c5780adc2e4b9c7cabf68e
b9f246a5560580c159bf20f12b01bb00ab1cc777
92713 F20110115_AABAAH abiade_j_Page_021.QC.jpg
64a5f015bac23a033891a37c1cc26db0
58dc3d27b1f392ab15a02a29f6b2794993da308d
F20110115_AAAZOG abiade_j_Page_121.tif
b5461d375c0815ffb5b3c08c516385ae
4ef8d49ad6e312e6dc3023b9738ef22f92ad4ab5
F20110115_AAAZNS abiade_j_Page_101.tif
581c433215b6dc1ca2c41c27707989e8
4bd8dd828b06ee22855850589d14a5e71ee7dc80
81983 F20110115_AABAAI abiade_j_Page_033.QC.jpg
b3a32da31d2fb7b9a98cf3276caa1306
a70e5d51013b75df48afeb3ddadd587a01b70126
F20110115_AAAZOH abiade_j_Page_122.tif
0a16e0641eb7c8e7882556604df70f2b
4158e17095e7134646758f6778a0a4095fb1b4cd
F20110115_AAAZNT abiade_j_Page_103.tif
29a4fad08276a491a5b75619d86a5492
1c57df5e2612146788c22ba45357912922edd130
33017 F20110115_AABAAJ abiade_j_Page_089.QC.jpg
6321bcb1843d6c2907c1c8498a65135d
7333f6b6d8841fefbc508ad30813d99a560b3b42
F20110115_AAAZOI abiade_j_Page_124.tif
1af00df49a4c784ccefc905cf75fe5c8
b071e8704f0c5bbb07c2757569c4edfe9c30facd
F20110115_AAAZNU abiade_j_Page_105.tif
8d3571d2b9d10a017b86d82e39cf05c6
597e2e3fc99f42b05413209c1e2c97c12cd7e63c
73940 F20110115_AABAAK abiade_j_Page_070.QC.jpg
40de4e08277c9b8df11c7ff89a671998
6bd02a8d03e89b43979160f3fd375087067d3391
F20110115_AAAZOJ abiade_j_Page_126.tif
9dd82577eee5d3a9b37000ac5fbdbb40
34367b9b9e5051799f359b7f6a592c02e325a2e9
F20110115_AAAZNV abiade_j_Page_107.tif
3a2560aaf6aa7cedcf184e134e7dd4e7
d3f2e865ad76ec1d924dd044b3eebf5193c51b8f
F20110115_AAAZOK abiade_j_Page_127.tif
b389086f5f80fd54279adc852e20f330
c1c2bf652565381fe045a9a5a7d52c7edab08555
F20110115_AAAZNW abiade_j_Page_108.tif
c95c712b271ce2fd454cd772d4e24133
89bc1539b20be5cd2ff26a1d38af6aea89485b8a
91051 F20110115_AABABA abiade_j_Page_031.QC.jpg
da58141f45a5fe4a9610037a34b18373
a18fc45e902c7de94d707785066dc738a7cb42f6
47317 F20110115_AABAAL abiade_j_Page_126thm.jpg
5a000916848ea85993bee0a85d38f7a9
60e663a0a23bca947e8b72bf437bab48c1f381ec
F20110115_AAAZOL abiade_j_Page_128.tif
774273c3198dd8016a5f50c9ebdca8ac
6d501b4a49317ec5c7cbb342e8e76eea8afa89c0
F20110115_AAAZNX abiade_j_Page_109.tif
87b58bb69d6f2bf5998e47d386867b17
8e4beddc6f9a5fd7e93715455a8076c640e438d8
78703 F20110115_AABABB abiade_j_Page_076.QC.jpg
79c867e76d313d6496d498d075fe8b3f
a42123971c532efa3d5046035368a6382d173933
8516 F20110115_AABAAM abiade_j_Page_139thm.jpg
4113859702185c8e17d1e49a3f5f779d
6e757a19aadbe5b0f15616a4e7e875975b11c35a
F20110115_AAAZOM abiade_j_Page_129.tif
8e27942014e4e39b35e9a93fdcd9ec06
e1d553950694c3207eb57d5fad35864f29a8572e
F20110115_AAAZNY abiade_j_Page_110.tif
e9d02e8fd22e055b4ca2c76d56273582
4309fc696703c5be026678d78375a7c0e1aa664b
9042 F20110115_AAAZPA abiade_j_Page_003.pro
bb92611d3912748191411ea399c661c1
7b734ca3e4747b11a966c72e197c75af368d92a7
46884 F20110115_AABABC abiade_j_Page_031thm.jpg
f43ab1705e345dd585b5895a0878d960
17d7fe20b3d6641cb842c394c4f2d785c3ed00d2
43476 F20110115_AABAAN abiade_j_Page_015thm.jpg
1fc50f50ef87f85ab9f11ba8699b000f
38530e67caaaa35215620c16151b7dcb111fd242
F20110115_AAAZON abiade_j_Page_130.tif
9fe94b6bfa3e6667555310f79b0c554c
452ab75f277c187c4af8dcbebcbefe3f315f44cd
F20110115_AAAZNZ abiade_j_Page_111.tif
ace3985d7303e423d6894f683840b639
2e68b8b55fa2d37348b18b8059b85ecca176d52f
35201 F20110115_AAAZPB abiade_j_Page_005.pro
f55da89e612628aa10370f2a2d38d6e1
af922b2bd5d8e94a89bfc3babf531656c76b8a6b
101077 F20110115_AABABD abiade_j_Page_025.QC.jpg
04040da38f8bdabd7d150aa8a6e15be1
c78e886e59ce69eccdf2dc4d08e6a4cdba9b804a
F20110115_AABAAO abiade_j_Page_008.QC.jpg
465b07fd0a39d63a953b38cbc7a3247d
8011a4fe064a78f79e77535e06a7c26c1cda565c
F20110115_AAAZOO abiade_j_Page_131.tif
826f8eb948fed392162b0181a3e0f0c8
91c9c8f31835932ad502a9d7eeba6716406a87fc
88745 F20110115_AAAZPC abiade_j_Page_006.pro
5592acfd42756d4648a48e7458301857
181d34120500dacfc562990b2c45c4b9ec02089a
48552 F20110115_AABABE abiade_j_Page_010thm.jpg
8dbe31ae2aab01766d6ae0f3f917c3b7
6dfafd29b92751a20e4a2473c3ab80bb38ab2297
120699 F20110115_AABAAP abiade_j_Page_007.QC.jpg
d882ea6098387b39bb0ffa94e1f92217
bd92068676718d2b28990b677bc35ee5599a4f1d
F20110115_AAAZOP abiade_j_Page_132.tif
d4af9f673e7b18a008a289315d74876b
a496ccdaa8e114bab68035bfbeda29bd881e6592
69662 F20110115_AAAZPD abiade_j_Page_008.pro
44dc2843cec5686600b8cc82522cdf85
75da87aaf02b261376332826f9dabea273fcbd38
81818 F20110115_AABABF abiade_j_Page_098.QC.jpg
6d5ac93728f8cfa38fb832f49e29da0f
916860bb2668efd36c37adb9c0b6e80c770c0619
45765 F20110115_AABAAQ abiade_j_Page_141thm.jpg
bc475684f01c876cccb2dcba7db56a63
28aa83c37a223ebfb1b50a869681fa9a9ac0cdcb
F20110115_AAAZOQ abiade_j_Page_133.tif
4aa4d7f2a8510715728ae396a1a84078
2a4baa5c63ca0b34b8c4d25ca5f6dff2485a03fc
7143 F20110115_AAAZPE abiade_j_Page_009.pro
524dcb82029f1076740d0a3e05e7eeea
aa4e40a4fd220ad28d34ac958364c79734d435a2
86592 F20110115_AABABG abiade_j_Page_063.QC.jpg
814a8933f35bdf5074c76fdfc5e12308
ef5990b42854faf0f526ed3aabf53de6f43588b1
78139 F20110115_AABAAR abiade_j_Page_030.QC.jpg
08a349251ea3d17da6e33171a4aa1cab
786745202277c8c9013695438472e659c1847b35
F20110115_AAAZOR abiade_j_Page_134.tif
952bd5218eaee783116f86f5ca122032
57132f600ffc9973fe4eb36602530582704830ba
63543 F20110115_AAAZPF abiade_j_Page_010.pro
55305434710e36e8f29f5a54e64431df
9c680ccdb2d3b1ec2e8394dc691cef4f63f06b61
75768 F20110115_AABABH abiade_j_Page_115.QC.jpg
fe29a1d9a703a36c5c519831959b728c
ab993b079840e5912941aaa38521cfdfb6ace2b8
78896 F20110115_AABAAS abiade_j_Page_142.QC.jpg
86cbee26cb3fab77c106b6d995d9ea0b
b45d73899effdd599fad14f83d2d61ad8b6ac433
F20110115_AAAZOS abiade_j_Page_135.tif
12d1aeb32419b83a38411dc9f4fa5013
7a99934ce095967b0228f1cc2d06dfbd8a062184
69041 F20110115_AAAZPG abiade_j_Page_011.pro
8dcdaabcd9336a64bfd7999d295edde6
881ca9b0ed3850684b8e42c97caba833961a4a83
79072 F20110115_AABABI abiade_j_Page_023.QC.jpg
94dc35fd7442ab6b420c1af408f95f13
23c13c9f4bef80ea00364a11d58749ec9f389111
47824 F20110115_AABAAT abiade_j_Page_018thm.jpg
a6d64a5d5b5c109afdb12779c627b431
ad97e96002bd44071954a197617179ed815d65f9
F20110115_AAAZOT abiade_j_Page_139.tif
a214032b372422f423243110003c00a7
955b5d34dc7975927705e23e8d5a29fb2858b0fc
39850 F20110115_AAAZPH abiade_j_Page_013.pro
915081a4ba09db2cfc92e459e848bd5f
31c2ea6486d1c133648b9d87634e349d68dd3a47
21752 F20110115_AABABJ abiade_j_Page_146thm.jpg
83fc4f333505d93ec25052baae03bbe4
912be50d0e6e1ea24bd298d26e0fa6c0e561d39b
23907 F20110115_AABAAU abiade_j_Page_052thm.jpg
09a8889feb279375f47a7438a3eb3b28
61fab155e417e578b820db1798200bc13a13011c
F20110115_AAAZOU abiade_j_Page_141.tif
86f277eccc343310e7514507ccb29a29
11b9bb05ed19a45caa36cbe1ad72baa92c48dea4
27195 F20110115_AAAZPI abiade_j_Page_014.pro
622f9cc20532a3a43bc24167be78198c
e78e8c6296c8bc3029f1c14e220e9f74d83ede52
25028 F20110115_AABABK abiade_j_Page_041thm.jpg
3f7a68fc918ed7c3bc40bc70930cc048
398f3759d7dbffd2a63758a84b42444e1ae47bbf
6202 F20110115_AABAAV abiade_j_Page_002.QC.jpg
81360bdf356ad760c0761bedddac5736
a987e42d1b023ab41994fdf4517ff3dba3332483
32600 F20110115_AAAZPJ abiade_j_Page_015.pro
b991391c11936255e931a45639cfd369
af6b70ff99255e9fb8007944af6125cb28fde5f2
F20110115_AAAZOV abiade_j_Page_144.tif
7d630f27c3974fabeea087e29af30df7
7c859e8af1e4652888fb4970091ac5238141b7a6
78532 F20110115_AABABL abiade_j_Page_091.QC.jpg
845b28043103ec927be409750ef02c30
5ae87422a7b4aa16c23dffc7c559a8ff8d48707a
25231 F20110115_AABAAW abiade_j_Page_057thm.jpg
f123d686da1ea69c7b3fd209251ff26c
2fca12aad237b682b4e5cc8801fedecfd73c203b
52673 F20110115_AAAZPK abiade_j_Page_016.pro
0f8280b208561f047a12cae9bdd223f6
5ec571f189ff5be9a342aa223bf3bcecf4049136
F20110115_AAAZOW abiade_j_Page_145.tif
11162ead4db65a7c88f29ed25c3bc043
7c69c41c681f34d5893eb0028406515886fe4c2f
24366 F20110115_AABACA abiade_j_Page_111thm.jpg
8814b51a93495fad758a2e5004c79714
e3cd767d1d4cd2fb71262dccdb43ccc1902844ba
45827 F20110115_AABAAX abiade_j_Page_127thm.jpg
625703fc52ef7d7f068275c177069ad2
fbcf5429ad2081f747df221d39dd456cca369867
52587 F20110115_AAAZQA abiade_j_Page_039.pro
1b4aeabde2900dc67dc1043c5b1b5fb4
dc241b61df1abdad392e3e3d1d541c8a345a022e
52915 F20110115_AAAZPL abiade_j_Page_018.pro
8d28a2f7f34ff45d641d05d036e6db17
6e50907739c4040695687070e40350ada8eec525
F20110115_AAAZOX abiade_j_Page_146.tif
fcc674ab122f46e3c5cd491f991c670b
3f96eadc385889c0092652b6cf5789084b26db80
26794 F20110115_AABACB abiade_j_Page_121thm.jpg
c154943687686d492516605f52bc482d
ad7846549827ab2836ce1d9319eaa9e661d03d2f
68250 F20110115_AABABM abiade_j_Page_090.QC.jpg
b31ae753b4de254044d577d098b844ff
3bf82a8d56c788b56ba8312df03ff3e11f4b6667
24621 F20110115_AABAAY abiade_j_Page_095thm.jpg
da6bbc5850a27f8ce87ad79e5b7ae43b
8a06c8026ea1d9bad75a03ba6ac98ff27bd504a5
24586 F20110115_AAAZPM abiade_j_Page_021.pro
8d819f5e350640ab75e4106290fddc96
1339f68516ff0630ef0401112890a4ff53cef999
F20110115_AAAZOY abiade_j_Page_147.tif
c69d70d8b260e422ebdfbaddb1df1e16
cc0f67150a9f88f78a918fc137430152c2636ecd
79603 F20110115_AABACC abiade_j_Page_052.QC.jpg
7625b5419d7a6cc13f467005e72e2681
e7dabb9308b14abbf30f2016607673418a529053
76066 F20110115_AABABN abiade_j_Page_050.QC.jpg
9be01f1b5b46b20ce3f05c515c5ea913
e390bdbf6fefc7832034f8961ef83ffdcca19522
82756 F20110115_AABAAZ abiade_j_Page_117.QC.jpg
6fcad31521cd04c5f5df2f3f3c100820
5eb4818c022edf2ace7b1854085cabd55c7db0fb
49247 F20110115_AAAZQB abiade_j_Page_041.pro
90a9345c65323037885de4c62bfc0e90
7148e2e5763fc6fbf91f3d29b1fc5fe2c32179f4
51851 F20110115_AAAZPN abiade_j_Page_023.pro
13fa44f74f88c1a83d21507143f39e01
75005831f7ae06d916c6a9f690970a89c7853ec0
1383 F20110115_AAAZOZ abiade_j_Page_002.pro
3ce07b1bd6e8de6807d7571cc99c3a19
04ef72b39dd18ceb747590373868dc5ea22e57cb
79105 F20110115_AABACD abiade_j_Page_069.QC.jpg
d473a209cd760dcb54b8948186d8adaa
b07635af7847c8378d1e5c2efbe2163798ada9da
42861 F20110115_AABABO abiade_j_Page_085thm.jpg
3aa1f76bf42a8345806ad65e1d37bf37
c8c317010f608925145b06351b5e3e07229d9a8c
52598 F20110115_AAAZQC abiade_j_Page_044.pro
cc7a1cda11477330e39ceb554d037b84
1d2f3a9ed242d5286c3176045096e37bef768b4d
53664 F20110115_AAAZPO abiade_j_Page_024.pro
4d06f079a1ac7cc9d8757e3ef6fea03c
02ffffa90418270dbb77094f0f6947a3607ed3f4
22640 F20110115_AABACE abiade_j_Page_070thm.jpg
0544dda9c08aeacc730dfbe339d7772d
e8803b271b61ed9f0042604a43a86ab611b5a4c5
19269 F20110115_AABABP abiade_j_Page_059thm.jpg
7c7a41ba4c8620fd3cb960fac5ec70b5
016aa9c24889f8f93567741551ba8db194f75273
43313 F20110115_AAAZQD abiade_j_Page_045.pro
b3f40690b22eeb03c34979cc742853c7
582e25319de2548d1e1e4a263b2f9bcacbdde972
35834 F20110115_AAAZPP abiade_j_Page_026.pro
4d62cee15828979152b638cf7775d976
86b02c9a7fa3cd4b445df93f675c08003d630f2b
77854 F20110115_AABACF abiade_j_Page_110.QC.jpg
917ab4493cf46f6a18a0d0fe1ee09b82
2aa9858d6974251b7fc5372c6a70b0eae4d778d8
11507 F20110115_AABABQ abiade_j_Page_089thm.jpg
5ddf466deafccb9a70825387b76e7749
b2c4a7384b133230dcb4617987fce665422b70b5
49096 F20110115_AAAZQE abiade_j_Page_046.pro
2303cf89e93105e4ce13bc941983c745
a5ac08e8159ef561eff99af4046e6d7585c44036
52489 F20110115_AAAZPQ abiade_j_Page_027.pro
30d43ee5490fd53716321b9f04e52d85
4295e6446b6b0951e3e00d98d0d540f0cc0b0200
23297 F20110115_AABACG abiade_j_Page_066thm.jpg
e91c594a18dbcfefb615e25e65bfad12
c6a3080aa4bf030172f84bed37d031937ef87d81
81802 F20110115_AABABR abiade_j_Page_039.QC.jpg
4079f41760dbd63d3d1b077ed1fa3c33
cc4045a401c2d1b8f95c5afb05e294b3273d71d5
28843 F20110115_AAAZQF abiade_j_Page_047.pro
57186c94666406d660c40aac8d6e2900
8799a66c049ed85526ba9767dc9f2ad7fe1dc3a8
50557 F20110115_AAAZPR abiade_j_Page_029.pro
70b4d9d37f7a01cf6a8327bed69ba826
90324565ce0e0beb1cf338335e614c8f9f4d9ff9
48571 F20110115_AABACH abiade_j_Page_135thm.jpg
44f7071b2578bdede8f32e4e78bbcab1
5f9d194235733c2aced98933fea6ce177c3993d4
79452 F20110115_AABABS abiade_j_Page_051.QC.jpg
8127cb23392fb5a38cbb131a81e95bce
cee4bb6ab02c871854dc9ac751ac0479f358f038
51773 F20110115_AAAZQG abiade_j_Page_048.pro
b93c3835d48ff78f28a552a398152d7b
5fe9309b864a40a377d907639902705154402767
50275 F20110115_AAAZPS abiade_j_Page_030.pro
7240cb96c8e0a87c37923a190ff25026
13d81b19519ab8f367aa7be8e1c39b089c6d6263
24834 F20110115_AABACI abiade_j_Page_079thm.jpg
23cf0e0b4f707805ba9c1abae59b3f19
aeec56da7f5159d558af119b3ca9de7ddd955c47
18509 F20110115_AABABT abiade_j_Page_125thm.jpg
6c90f1fcebd90286e18f8bca75703d4c
bddf9ab1e7a04d0a95c1019e86eef4ec6d00b65b
49553 F20110115_AAAZQH abiade_j_Page_049.pro
a5ba6b40330f8f6172fe2ab2cf56b849
bcba213fc8fc52dc0cab93e56667c49fe7833554
38650 F20110115_AAAZPT abiade_j_Page_031.pro
428364e86cc15b1d249144759effb4ff
21edae73e41fc448eed6a633c661fd7bf689ab6d
77163 F20110115_AABACJ abiade_j_Page_102.QC.jpg
e30d5f7e0d0d9235665de93506dfa22b
425bf45d25c1ccca2006825bb033cb1f1cee0ff9
44929 F20110115_AABABU abiade_j_Page_124thm.jpg
d60e339bc597a40ffad57875e41019b9
a9260db5423c786898cf30d69ed95aaa06bd0312
48837 F20110115_AAAZQI abiade_j_Page_050.pro
62d8df0cc7c2c63d295bc1db014e8fc3
7f7a41d0d6c33b0ca9406ddf91097a57359ff55e
50872 F20110115_AAAZPU abiade_j_Page_032.pro
d2794e7af02213f781a63654a604eeba
a44ff29ac2b8ab20e986d46f3b0acf64d147ef64
14627 F20110115_AABACK abiade_j_Page_003.QC.jpg
cbfd365fb4f8c728b62f294706bccf62
04fd8e7e08ef53e5d0be10e02bbd3ba5df2b572e
19526 F20110115_AABABV abiade_j_Page_013thm.jpg
35bcc931e11f8fceca73e823ad48f34a
b9de6b3f04e01caab9573909822c32f3cc34d90b
52058 F20110115_AAAZQJ abiade_j_Page_051.pro
c7554e86a632306bc9951594b65b3e5a
43558c381005b4c33a099b220481cba314502aed
51888 F20110115_AAAZPV abiade_j_Page_033.pro
c6667e0b596214684f5c2dab5751fb73
845f3abe8a8556a14395b7d627dd8c476609b3e4
221623 F20110115_AABACL UFE0008394_00001.xml
3c5f02de646f826c81f70a53ba0f2f03
22c4f1e71320fe65e2697198f2bce38d76476d43
24409 F20110115_AABABW abiade_j_Page_077thm.jpg
254bbb913e14a60a09f33ac959ed759b
14a91f6fe38342037478a90ffa9302589334a4d7
50562 F20110115_AAAZQK abiade_j_Page_052.pro
1b8626520f1bcb881f5bcd8a26bc996f
021a6f84c653f816ddfda9dcca4e39f8c7dafcfb
47315 F20110115_AAAZPW abiade_j_Page_034.pro
17fdc347cebf70f0181e417323e986a6
0ecf9a72f51777483c4fcb20facbf02aadd46199
24831 F20110115_AABADA abiade_j_Page_024thm.jpg
33575756cbd4e56fb4d4655d6720927b
8d875e9399b7bef2ff43936a80ab0b4d8cc09d2e
21590 F20110115_AABACM abiade_j_Page_001.QC.jpg
fb463403f5f02c58a98d465c8cd011a3
ae8c778fd6b6b692ca822e6a64f23011998706aa
23608 F20110115_AABABX abiade_j_Page_073thm.jpg
48fdf58875566625ca83bd93953c0753
c00ed4d0632e1887040d6e541e7c58e7f94dbed7
41792 F20110115_AAAZQL abiade_j_Page_053.pro
9cf7d587241aa10fd1061281be41dab0
068766de9b1ed459732cc63c461b8e298c1077cb
52172 F20110115_AAAZPX abiade_j_Page_035.pro
e456b70360974022e98b41e04c8521f6
a11d8bb00a1c1968ce243e9bc7407dad832314d2
51424 F20110115_AAAZRA abiade_j_Page_072.pro
dce0b14984d33c306f7973375a15e320
ac414d19bd9a5b94e67d2be90c3d535eed303a0f
82525 F20110115_AABADB abiade_j_Page_024.QC.jpg
0a2a931dbb4491da509904da7d4b45a8
734eabc1b59f54613f2657530f298605c8685595
72912 F20110115_AABABY abiade_j_Page_116.QC.jpg
bcc36d720e6a54e3498c673992bf4edd
ebe3d7f194af3d31c83ba22d28155e9a6aad4ef0
48985 F20110115_AAAZQM abiade_j_Page_054.pro
caf96944cf2ed91779b713a922bdec51
9a777cf85cf92d8e094a22360224020c65f64e44
50829 F20110115_AAAZPY abiade_j_Page_036.pro
ba299cca3dbafc2769d5596998f2c880
eb05b61980f61f4bd97d8566ba5fe1a81c63f9b1
5993 F20110115_AAAZRB abiade_j_Page_074.pro
e8c395311894e461b76444cad06e56b5
e196ae4699d95df6196b719883a7e7fd557774c8
46427 F20110115_AABADC abiade_j_Page_026thm.jpg
b6b41faecda5441dea8c3c75ca6e514d
719b2862082443ce91b89ff18fa0ed445ba54401
3046 F20110115_AABACN abiade_j_Page_002thm.jpg
88b3b1f92e7dc14599b058baa30fdf95
8445d10704eb02efbf0b17aa6b78ee2b70b833c9
23282 F20110115_AABABZ abiade_j_Page_019thm.jpg
6591d6fec280528b5d3d1fa8e7b4d84e
da6a2a1160e792377697d70488274a29233dc069
48732 F20110115_AAAZQN abiade_j_Page_056.pro
6b27ef0690bab4d9597108241710f2c3
555588e913bddf2193ce60f922cd228e24d5bb50
34949 F20110115_AAAZPZ abiade_j_Page_038.pro
d62588dae8b088132e31ad793c241050
56e366fdf24062de0713605c29e55fc631d84f6d
90970 F20110115_AABADD abiade_j_Page_026.QC.jpg
7c0bdfc8315839e40efc5753d54d08ad
27aa09e2032ff229232d0ec6e1903d2a43169448
23094 F20110115_AABACO abiade_j_Page_004thm.jpg
ba63ef8284537702a259fbbdc034b750
de74cd9dcc3ff066e1eeadb1f442401c597f5d92
33045 F20110115_AAAZQO abiade_j_Page_057.pro
752b1caaff622fccaf7ab65aadb66929
fb55723b272468b128e46b37e056be2734f2d86e
50535 F20110115_AAAZRC abiade_j_Page_076.pro
8551ffca32ede32793296804bf052275
075c05d966b1c4050473641dca08482a4de9129a
80001 F20110115_AABADE abiade_j_Page_029.QC.jpg
7707e210b5bb516eb615474483d60225
faba7a34f7c51580220e7e78d4ef2d552a52bcec
90035 F20110115_AABACP abiade_j_Page_006.QC.jpg
252e5da5ef9d117e2cf06c52a8944c53
2f003f232b976b9d3b2b3ccab2111bcd5c48517d
30164 F20110115_AAAZQP abiade_j_Page_058.pro
34ad5bc1edd107c963b45c816ffa2dd8
abbc3ba08799d6fec9d9311e9817d355af9f708d
50181 F20110115_AAAZRD abiade_j_Page_077.pro
92021b7285540a4c9630d006f27addef
d1fc17e12e886645719093a2b489c4f62b734313
24140 F20110115_AABADF abiade_j_Page_032thm.jpg
f2a33aa3710d85848e7632c51a2db5e7
811d6edc5e9d5db1442fe0b2463d34a6c908aab7
50776 F20110115_AABACQ abiade_j_Page_007thm.jpg
6e999046eadef51951f7e95754558881
1ee1ce9ae578a0f8ddc084b2d9806c87932d988f
35251 F20110115_AAAZQQ abiade_j_Page_059.pro
36e472468d9f620fe285d6d1f87165d0
449949736ff3ad588a46652438999aa6f4bfea41
50509 F20110115_AAAZRE abiade_j_Page_079.pro
24c12d16cecdfb47bf67ef85d1b983b9
54bdc3965fba20998a92b893081899a58ebfddca
24067 F20110115_AABADG abiade_j_Page_034thm.jpg
a7932c13407466928de0d58c490534ef
073381b635d3e7a0b8256d833f2cc1221a28f91d
45195 F20110115_AABACR abiade_j_Page_008thm.jpg
31da77f7531f413a4ae44cf3bb733e4c
091a33a0498aa09b301eec3f12cec6e4629e07fc
38788 F20110115_AAAZQR abiade_j_Page_060.pro
9a2b0b41241981b3adb854ad2c968cf5
e78d88f889682fb35226e9f4c1fa48d1bfb00b7d
52843 F20110115_AAAZRF abiade_j_Page_080.pro
10604fbfa437c7a97dd1e66a6be7a999
26370cc55a7f3eb295eed5399f9d0a3a557909a0
75653 F20110115_AABADH abiade_j_Page_034.QC.jpg
d3aee08a15f54c822f47dc23dfa187f5
57d3a55846b080169a1b3fcc6c1999de883937e5
30608 F20110115_AABACS abiade_j_Page_009thm.jpg
a3ac0c91dc1d25a011076074b3ab4edb
daeb54372a689a8356740cb07a4d043413845a10
37701 F20110115_AAAZQS abiade_j_Page_061.pro
424a286b6d8ff1152983c1c0fb4656df
7db23072d97104c98c3f7315762b1a82a97713c2
23697 F20110115_AAAZRG abiade_j_Page_081.pro
a7dd0f9501bc321b7b04f0a34e3f589e
2449b9aa8452898c85d6c5bd4e4a1629ff75e78b
81929 F20110115_AABADI abiade_j_Page_035.QC.jpg
924233d82ec452f8c096eaa44d4a26bf
6a811ba474ce5a26aca5bcfdc29d1c5896fc14a6
38719 F20110115_AABACT abiade_j_Page_009.QC.jpg
4ef78665904996ff0fbb8fdf27ba93b7
8d01ce920e4cb74db3e307ce03c1a857cb54eb45
33598 F20110115_AAAZQT abiade_j_Page_062.pro
b48dad4fa04209a313ec7c8f0535c746
8ee482b9f5a61d9b146ee1d3a252e2dc689b0146
43521 F20110115_AAAZRH abiade_j_Page_082.pro
c343a727d51179c7574e82b8cd6a19fb
41f60febcc230c2aaa52879f0d82f941f9375f0a
82197 F20110115_AABADJ abiade_j_Page_036.QC.jpg
5f0452c20ce3363008df9418b1555280
4fa25a88e519fb292c5119616a5640fbcf414dc5
53178 F20110115_AABACU abiade_j_Page_011thm.jpg
08c2154dcf52b6c3255666b4db83e838
2af0687d8117f90186ec4a79f991c08d44af9a72
34844 F20110115_AAAZQU abiade_j_Page_063.pro
de8bda04e6bcb8e4f70961eb40133bff
c7123473ab05299c1bc1e2a6303ed6e07672b72f
35589 F20110115_AAAZRI abiade_j_Page_083.pro
b3e3278be709fb339f88e763a9690032
f6141c0e2a5ddf1bcce9c7ecf874f90bbf115275
47163 F20110115_AABADK abiade_j_Page_037thm.jpg
54e44b49db28f0bc0e8845cc1a595a83
44fd9467df6cd5da64b59b98aadfff59ebbfd00e
33542 F20110115_AABACV abiade_j_Page_012thm.jpg
7d9444ba302d48b54ba7a215de7ba210
c3f46673730543a59e17be9fd5efcc5b6fd6172f
47714 F20110115_AAAZQV abiade_j_Page_065.pro
0f435952ad375175a8c21036ffe4caf0
67c64521b62b4d1cdfa8e6cc05587106e0bf8b2a
35220 F20110115_AAAZRJ abiade_j_Page_086.pro
cd20b96670a4c8bb26d5826ac75c0ab5
3634d0554d10286fb35653d4ddf39659cd9eee73
25489 F20110115_AABADL abiade_j_Page_039thm.jpg
08f8e1c9a58b909f7ad541c3f3b3b1a4
2fb8c4492a1aecf1de6da98a90c35466ca2aa758
14278 F20110115_AABACW abiade_j_Page_014thm.jpg
1651d291d62e92a2f301a2dd87c5886b
0bbc5c411ecbf40d1c15b54f10a5138f3a9ed0e4
47109 F20110115_AAAZQW abiade_j_Page_066.pro
87821e30b054cc0637465aa747793436
2c6f7d825f0df17054101caf0e8f88a41a1976b6
40926 F20110115_AAAZRK abiade_j_Page_087.pro
d1067d5db55889c38c2271ae165303bf
62a742a3608fe546e05268b32c23beb6d92f3aef
19107 F20110115_AABAEA abiade_j_Page_061thm.jpg
529bf8aa36fd26891d51249627924b88
91b2d6f5db3beb67f018489f06e58540cc88f19d
23963 F20110115_AABADM abiade_j_Page_040thm.jpg
3aa58b7bf28a22bd973a53bc413688e1
46369b0db8cc37a2711eb8534adc6bddea47b522
98284 F20110115_AABACX abiade_j_Page_018.QC.jpg
5310e5b96f270f76869af16438155f62
64082bad75a996eb49ae4316df3171e07e08146d
38769 F20110115_AAAZQX abiade_j_Page_067.pro
ff9e03be2e3d0bd8d3922ecb93da4b28
2f28e14372630585a6a63a26e09cb9a9b64fe6e4
50534 F20110115_AAAZSA abiade_j_Page_110.pro
174b098de9245e39b2712c2b08ab0680
1fd64313505cef6b616aedbb9f530fb8fce6fe41
19777 F20110115_AAAZRL abiade_j_Page_089.pro
9807f80972dfafeb0b9378dba755702b
8b1b46e2569e0a4079e0c948ec475d05456c9266
96931 F20110115_AABAEB abiade_j_Page_062.QC.jpg
192f757a5f3329e009928f0de66d513b
18784f533f57ecdf29280f0411de334f2fdcf9b4
75424 F20110115_AABADN abiade_j_Page_040.QC.jpg
d0dccfb6f27fbf9f60213970289c6f31
721d933b567110fabfdd09dcfcca0312d73a36e9
23082 F20110115_AABACY abiade_j_Page_022thm.jpg
f674497dce7f09a31729c26df77df0ed
34efe2f2d5e5df894d99dd5c63ba2c84f87dcfba
27224 F20110115_AAAZQY abiade_j_Page_068.pro
6a751d99352b53d858ca0a7dfb867e05
46bc3534ea720942befe32ff107fa95ce4057a5d
52407 F20110115_AAAZSB abiade_j_Page_111.pro
0b6ede19bd3d5b6d630d2f5876daadf1
0a3513ac415b9beca18126b66419401474eb7559
51566 F20110115_AAAZRM abiade_j_Page_091.pro
dbccc1d600f673b3a19c8c86913acdae
caddb36d56da0c69511ef90cf6e2d220e39e1eb3
46748 F20110115_AABAEC abiade_j_Page_063thm.jpg
0a23e30236336e447e57aa2a76e421f7
75bc7380ea45a0021dbf1418f6b99a96c05f6297
F20110115_AABACZ abiade_j_Page_023thm.jpg
2fee39b7606cca8c0e2737952fcd03e4
306500b1b5c8b8e8d0b1ab8eee576a23a74a8def
51438 F20110115_AAAZQZ abiade_j_Page_069.pro
3cc71fb43b6883da01cb001fa046f0f4
3243ca7a04cde829189cc2fb50077a87b1058e43
39867 F20110115_AAAZSC abiade_j_Page_113.pro
1bfdaf08f92d73cd267f4beaec9ffe48
81cf755eb0ae528de1b7498c01b3fc822b7ab78c
52975 F20110115_AAAZRN abiade_j_Page_092.pro
0786a434502f7f4ff4cd46e1f6b98578
92b40e572b04a16cedb6f4610f85d28045032db6
22081 F20110115_AABAED abiade_j_Page_064thm.jpg
3ae0734f3130656fbfd43b5b855de4eb
39e07bd1bc8f5a8c0a8065ee82dc2749ca4b9415
76761 F20110115_AABADO abiade_j_Page_041.QC.jpg
62fc65774d27000ea3ce7b6d1c5f7b0f
9ea8fe1af69755f8539367977af0e1c288e073b7
51760 F20110115_AAAZRO abiade_j_Page_093.pro
67bbdd53a4fac91f1b03831e9f8d8510
53ab757694c393d4fffa9a03b5879cdbcf12b188
71077 F20110115_AABAEE abiade_j_Page_064.QC.jpg
5f46a9cc2d43ccf19e0773a87b3d3283
4e1a3849c706a352c74ea890119cf95af35d268a
75380 F20110115_AABADP abiade_j_Page_042.QC.jpg
46ea37b5d6932e249da5c7642aff401d
4d554af52e625070b847311e376b2ad81071d8aa
46830 F20110115_AAAZSD abiade_j_Page_114.pro
fb2d69c7de720f6f55988e4699c79a5a
682f0ac9913f7757d763f29c730871e397e439f6
34329 F20110115_AAAZRP abiade_j_Page_094.pro
6650e99a56dd182b9bcdbf4893331232
93018548b5b221678588815303f625cecdd2b4b9
20514 F20110115_AABAEF abiade_j_Page_067thm.jpg
5905c8d85d717ef85b83696ee00506c6
78772c4d6ca0b0ae6dc2e37af239844006051ca5
98220 F20110115_AABADQ abiade_j_Page_043.QC.jpg
b6020c6455a41bb1c97cfa08e3d2018e
f602507e828a12d3b7821af8371edd40185b7325
49488 F20110115_AAAZSE abiade_j_Page_115.pro
fe056d988e4ddb06fe3277a799c2a999
bb5e368d65eea1a1f3647eefd87c88f56603128e
50575 F20110115_AAAZRQ abiade_j_Page_095.pro
fcba98296402cc0cdc6debaf7243f01c
b0577b94dfee47012dbef6e2013c90fec43d721b
64769 F20110115_AABAEG abiade_j_Page_067.QC.jpg
6677cec75f27ea2e53e73e5431e1dd0e
83e226b3225f838457c648c5f81f18ad5371e1f6
74843 F20110115_AABADR abiade_j_Page_045.QC.jpg
a169b5b2101d2534f35e2939a3733d98
360ba327913193b5294e217ed0de7dc818d93512
45307 F20110115_AAAZSF abiade_j_Page_116.pro
a98cb45df7108da87022e04530e9ac4d
7c803e592b41919bf129fd48b9a9cee2fcd86c54
37002 F20110115_AAAZRR abiade_j_Page_098.pro
5c8f3c81ea671023aa59975b9009ffa1
14ae452ab213f76fbfd40435856b4dde63a38d86
41631 F20110115_AABAEH abiade_j_Page_068thm.jpg
ea0bbe8d6faf91235fcb32ce5904695b
837164f3739794f9ad21d5c83f1d9dfd30932e00
77378 F20110115_AABADS abiade_j_Page_046.QC.jpg
6b71f6671bd8fd147d4448b9d695cf8d
cb4dcdbff8860fee4c6885730004f4663dd87ee7
32448 F20110115_AAAZSG abiade_j_Page_117.pro
9fb2db3498f51fb8d7a8fc9b30d04a02
ef7c664c203f2ea61418816f81df3fe8e506deca
39144 F20110115_AAAZRS abiade_j_Page_099.pro
a56422c87e77066742adaee32a5d1f31
44e717bddfddb9eb0441a999c91f21cec918f836
75399 F20110115_AABAEI abiade_j_Page_068.QC.jpg
5ba7dc8add9639b20270b265948ae87a
f50fd7b6c0725888e4216cd3bf042bf075ea930a
22609 F20110115_AABADT abiade_j_Page_047thm.jpg
958fb74ff6c47a60e624c8aff02199c5
96f46e54c6171c105aaa318ec01a4bf20f4faec1
51396 F20110115_AAAZSH abiade_j_Page_118.pro
338b09e49f21c6889f100715763f21bb
4fd64a8649339a5ed2ae39dbb07e0b2ce27eff64
29915 F20110115_AAAZRT abiade_j_Page_100.pro
78b72a98191c0aca3337ee862be66a2c
f465130a00caca548ee9c3dce7388151c224d1a9
25052 F20110115_AABAEJ abiade_j_Page_069thm.jpg
66f21ad4b99fbffee0d22d90bfa018ea
05f2832bc032f3ddebf38469b27cc839b788d427
23384 F20110115_AABADU abiade_j_Page_050thm.jpg
fd05c5c7b8a5b442f0c37c790cb0bc51
aa1f2f8e96f69c0802fd1ea6cd02a394377cbd74
49538 F20110115_AAAZSI abiade_j_Page_119.pro
028432b59f2a43549369aa337b7abf18
e275e1e73cb90390a31a031abc258a940a2d02b2
21390 F20110115_AAAZRU abiade_j_Page_103.pro
95b9df99b44697685f4d23b585205ccd
3a7ca5cb629d8b07bd7be4b363cae0c5d58434f8
78151 F20110115_AABAEK abiade_j_Page_071.QC.jpg
1cac10e944a0a3545dbca9dbb8a3929c
a9781092aa3f3a490f7bc67134badd0fb399a560
23666 F20110115_AABADV abiade_j_Page_054thm.jpg
65bee5b333049261fdb83de3d5e32913
2d61cffe621d7f81437b79c481faa5f188c2d7ea
28834 F20110115_AAAZSJ abiade_j_Page_120.pro
7e7eb2b02707f0681c6bbf65ae05a747
5847f76b0cb9082b090c7b3a33b15b15218e8668
33643 F20110115_AAAZRV abiade_j_Page_104.pro
3b3c3d4baf1a4c6f7254e39c6cb4619d
607b2a7a4799eb53985a8a887034e9de5e3d5ea4
13669 F20110115_AABAEL abiade_j_Page_074.QC.jpg
140030ed205410ff276e6da523f314c6
92c83b3edc4b35758b73dbb92124f508a7c677ae
25364 F20110115_AABADW abiade_j_Page_055thm.jpg
22430c9288efd1aba9a6cdcf47f7c986
cc9227e3621111ba3506d6088a324a4d02f5324b
53820 F20110115_AAAZSK abiade_j_Page_121.pro
dffb799c68b9eb753c8afdacb30f941c
bcb7aaf58e48dbab56615415adf4bc5b2d9e1116
35460 F20110115_AAAZRW abiade_j_Page_105.pro
ade892ed4cdf9fd7164f12c9cd7c3133
0bfcb0a5014868de007d03c8014fac392a445531
F20110115_AABAFA abiade_j_Page_095.QC.jpg
af6d21d2a2ea31ee49462f32a10f6fa3
a0c57dd526e55cde757f85f3808a5c3777142c51
23051 F20110115_AABAEM abiade_j_Page_075thm.jpg
674adc667b0356ab8d1ce158bcd7c18a
c0199c77d77dc32a5eea7256070e7340ba92482b
76731 F20110115_AABADX abiade_j_Page_056.QC.jpg
35c6dcd03a736fb74b194eaffb5e3c19
03a5ca04fd8be0875a8673401dc883ffee6b03e1
21302 F20110115_AAAZTA abiade_j_Page_140.pro
5f8ac5c618bd106da82406f81df02bc6
2f1de92ebf37657a11215a8569dcf548f70fa026
26191 F20110115_AAAZSL abiade_j_Page_122.pro
00c04899177ba14216a6398af13c29e0
92e544c970789ef70276c2eb8816d3cf209f3907
34619 F20110115_AAAZRX abiade_j_Page_106.pro
e0baaac809c28c2401ed6e5771a588f5
9f125ad79ddf073fe1de60db79598fac93b01d62
25474 F20110115_AABAFB abiade_j_Page_097thm.jpg
8b27088bfb4d686df716629cdd9c56aa
f20c45c6b596d082761f657b57ef2a0cf0122e63
24918 F20110115_AABAEN abiade_j_Page_076thm.jpg
d734d34b3d52a79553997b6375c31280
f28506bfb4a24b875d0dea07da80b55a39ecb88c
63317 F20110115_AABADY abiade_j_Page_057.QC.jpg
459933ebc0db2751433cd973533f4458
807d54583df81f432e29050e7befd7932fd62017
43108 F20110115_AAAZTB abiade_j_Page_141.pro
fbce06b6509ddbb40e612fe89eb52bd9
40d8af9cf9ff6328f1ae88878f4706e7ffe1cab1
39761 F20110115_AAAZSM abiade_j_Page_123.pro
04fdefb0b85905358300cc552eb15c6e
21e6c504c02cbafc49cc6142fdaa153ab19e19ce
38610 F20110115_AAAZRY abiade_j_Page_107.pro
28420f054b88c9ecfeb80ce9912583a8
7a9e12b7328d3fb819e540f6df85b6809d6648e3
44229 F20110115_AABAFC abiade_j_Page_100thm.jpg
1c0ae8bf6348ecea06e840899a236c62
4a10448934cb3ec0ae31595420e44dd6c2b47f81
76371 F20110115_AABAEO abiade_j_Page_077.QC.jpg
7a68a685b56ee9fe6a05677619fd7b07
f43e5ae2b165fc67d5a57e2476a8c205fc2977fe
55323 F20110115_AABADZ abiade_j_Page_059.QC.jpg
eeace75e846504168e7e831ffe4ebc8e
7bcd960a3a61d584be7709c7338a0b597a402590
51723 F20110115_AAAZTC abiade_j_Page_142.pro
d35f660fc7b4164b49bd42b521209df8
7c0c732d078fcfb4e3d305acfe602d8d6da8bee3
21999 F20110115_AAAZSN abiade_j_Page_124.pro
a2e522fe2afd70d9bb38ab788e47b5c8
5c990f3193a8f649e9999506e9e4b186cb567a86
48835 F20110115_AAAZRZ abiade_j_Page_109.pro
6ff62bfa290e48a0fda0f30db09b4184
a199f280f28705a4f3078df1bab672871653377c
88448 F20110115_AABAFD abiade_j_Page_104.QC.jpg
d1a6ffe5a671d48e6ec64bdf3d96e78a
3ee4249f90153544e94b716b8560f869da30caff
49731 F20110115_AAAZTD abiade_j_Page_144.pro
ac87ac110b2aea84c2af7335e98372de
b8be5d585fd326d9e8f51d3d70e8cabd6dddc39f
30266 F20110115_AAAZSO abiade_j_Page_126.pro
55bbcdfca0423e9c564650c60f89dc02
8f83a53e15fb38c8276192cd8dfb2ccc956fe5d2
46341 F20110115_AABAFE abiade_j_Page_105thm.jpg
6b0e955c0434e603ad0857117af88224
0b7e23b275ea4b50b881d8a64f7133b019cd38cb
78030 F20110115_AABAEP abiade_j_Page_078.QC.jpg
e837831e2853a3e72cabe6846e789d47
700dcf7666491e38cb7ea3f98138f1684add0925
28132 F20110115_AAAZSP abiade_j_Page_127.pro
49c09191014e9b248dcf24bb7583ff7f
bbfa70915e9bfc49abd306c32c894479466a867f
96717 F20110115_AABAFF abiade_j_Page_105.QC.jpg
206fd67991f1215dc52a33386ecbc0de
c049170a2f138124f4e1c044552ee5eb61b4b546
44679 F20110115_AABAEQ abiade_j_Page_081thm.jpg
04523184030e2b8b274590486c7762cf
7d1511055773a4c138d01940bf611f72abb80fff
52402 F20110115_AAAZTE abiade_j_Page_145.pro
2f0a366af74f2b7b254b90593e8fa437
9d3f4423563efe70d4ec2bb94ccb94114d5db325
15836 F20110115_AAAZSQ abiade_j_Page_128.pro
a5ef2ae7c34e4eeba64eb25a6fb196ea
880f78dc4c695d8a13e933f2334147c6b3b7f1fa
30432 F20110115_AABAFG abiade_j_Page_112.QC.jpg
612e44760943b843bd1ddddb28a18762
aa19c388d78e0dbf3b1374975a34f4338a03c000
85854 F20110115_AABAER abiade_j_Page_081.QC.jpg
5f3d6c0db46860170a75ead9211df180
76116db7df4b4c882d60379c43cbc29515821646
46399 F20110115_AAAZTF abiade_j_Page_146.pro
0f99477aea4ebbaa66978954d6b7d9c7
6275a1f74e19c6bfe54962f5e46b903d2dd077fc
30214 F20110115_AAAZSR abiade_j_Page_129.pro
883307d04c8e866164a4813412c3ae29
02a8ebcc38d0c3a431d7f9475435571dcf507484
64256 F20110115_AABAFH abiade_j_Page_113.QC.jpg
07a230fd37351f56c577ae568a4f1ac8
1c545f34572d30e2bcb0d24ab370890c1673a096
76350 F20110115_AABAES abiade_j_Page_085.QC.jpg
b3109f841020c043ec9c84d1a508912d
6291e5579c57d81af9bd391da22c7b30ac1a5d8f
20913 F20110115_AAAZTG abiade_j_Page_147.pro
12e30dff5af784330ea4d0ff728e3c91
488cc402de7cf573b24118e481ec474e230cfc83
31401 F20110115_AAAZSS abiade_j_Page_130.pro
cd9d49d5ba9c4d61ebf19db2b0978108
0bb8cca93c22832a2416397a8270b366f04161d0
23102 F20110115_AABAFI abiade_j_Page_114thm.jpg
f9c8218d0d2a36fb4a4943c585c7cbd0
8289e9fdc9f8de20c621a66c82aba27487e2b2a6
46040 F20110115_AABAET abiade_j_Page_086thm.jpg
2c4cc19fb5208c3f6c6969b10a00c46b
55b8d97db4767ec24d59414ebad9abdee07962d1
114 F20110115_AAAZTH abiade_j_Page_002.txt
76f051a572428ddc78e97ff1b148ad3e
fa2bd7bed16609003671d4c3649619475958c28b
48094 F20110115_AAAZST abiade_j_Page_132.pro
c7797e8f69a262417b32cdeb9034681b
ab89390a847c666ea710e1866d2fc508a614e0ff
43806 F20110115_AABAFJ abiade_j_Page_117thm.jpg
8b10db7f11ba1e89286956fec50928d1
73b4b528ea4eda0b641f36f38e07388e785b8913
45203 F20110115_AABAEU abiade_j_Page_087thm.jpg
cdb7cf28b19e779b25afea3acd42af1b
b02de846d3682116665c1b73a643444d13711481
391 F20110115_AAAZTI abiade_j_Page_003.txt
838a593abd9691374921838cd3dde8f2
d0cad3432dc035ea692b3fd55103bb598c889b1b
49999 F20110115_AAAZSU abiade_j_Page_134.pro
5370dede988a0a2eaceb4591a98f0080
647b332ec38e837244eff6356427cd835c688f65
76552 F20110115_AABAFK abiade_j_Page_119.QC.jpg
c2df3b7bff62b4e734499594104e5c81
058a8b792bf66cdb6682c6a6fa5c446413b6bf59
88365 F20110115_AABAEV abiade_j_Page_087.QC.jpg
bef206434bff75e947b172c45b149e26
88b5f91664d9c5532260055012e04f6f801bf073
1804 F20110115_AAAZTJ abiade_j_Page_004.txt
946f1d4d07639f5236eb5d545c06f50d
7be031bf9eeef9e0b2097ed05b51a6dbb8c30107
45116 F20110115_AAAZSV abiade_j_Page_135.pro
17b61753fa6c635acd0d6097fee39821
9c57e3ac90dc25c30b5ea4f180f4f720ddefc126
48272 F20110115_AABAFL abiade_j_Page_123thm.jpg
2c66fd8e833639e18a1b0a9b5fc2bf64
0eeb9bcd0a74a6ec5de90f0356e1341393772b26
78235 F20110115_AABAEW abiade_j_Page_088.QC.jpg
3d1b9eb8c6ecd07df4f7ad09d6b12aff
8413abdaff3a1168c8379821ab7cda06fc640bf9
4866 F20110115_AAAZTK abiade_j_Page_007.txt
73f8706a68494f7ed166c8ba7d8f3a1b
68a10b47998baf93a44826494ebe55ef86d2be28
52200 F20110115_AAAZSW abiade_j_Page_136.pro
012dba389aa1993c92e5dabfb2550ab8
98e21a96d2765c6f8430159096138d05cb301f42
83558 F20110115_AABAFM abiade_j_Page_124.QC.jpg
93b022e978a47b1976c47322faaa4e1d
4e54b9ffa758400df8b0851c456a208f33a08f70
25278 F20110115_AABAEX abiade_j_Page_091thm.jpg
d4e5d93eff215857de17ec4ac6b3cd82
d1089324b324040b3464b33134518e8b80216532
330 F20110115_AAAZTL abiade_j_Page_009.txt
d64f23cb39dfeec64b482ea827a8630e
5d0a81e37824fea5e601d8bc8d071df348ad6ca0
36356 F20110115_AAAZSX abiade_j_Page_137.pro
9ac21feeed089729bca8586ef7f90231
c0c689f7a77880d036c780aee6e8bf4c55fc6ede
2060 F20110115_AAAZUA abiade_j_Page_027.txt
76aecbf83a3d8b265de94f323d188216
47ce8297cf3915d4506648ccd141e2268b0a0ccc
77763 F20110115_AABAFN abiade_j_Page_128.QC.jpg
caee111221880b385f66d2b54ff2341b
8b857f6dfe9c22b91c72b6aec4605a731bd4bf78
24359 F20110115_AABAEY abiade_j_Page_092thm.jpg
6aa3de9dcec7ea7a03ae0b8363380764
029c7802117b87535ca3c108a9eb447fcef85d71
2555 F20110115_AAAZTM abiade_j_Page_010.txt
0ac097f4b3792cc44d53857a3a5efc0c
925f5c245181b060881a7b22fa853161db602e27
43544 F20110115_AAAZSY abiade_j_Page_138.pro
b70a49c070f4b94e0a6ed6583e915f46
8b3a85d1b38403325d1c2caf3793fb6a7586d36e
1986 F20110115_AAAZUB abiade_j_Page_029.txt
c92c4e95ff40746484aa906adb8e8417
418fbdd0b42c8fd0162634ebe7195fb3e66cdf0d
84046 F20110115_AABAFO abiade_j_Page_129.QC.jpg
1dd9782564edd1848b8c8366086dc38c
43db3b7a824b85fa65f5e476cd2689acc055d55a
25065 F20110115_AABAEZ abiade_j_Page_093thm.jpg
a77a92486fe54191c854a90e157f5c72
74d174fcfa35369ea020689283c61dcb42980671
671 F20110115_AAAZTN abiade_j_Page_012.txt
2e5d7be5917e39a676a484d30286a460
c7651f6953490dad3bf60ce2e8aca240bbd5af19
12375 F20110115_AAAZSZ abiade_j_Page_139.pro
7828081c67f156dd7e07d2b52de15e39
d8c8b01ef9a684201e13f7d420bb867462325beb
1770 F20110115_AAAZUC abiade_j_Page_031.txt
09f5c0a3df754ab28959f6c4dca1321f
c897e9b38a92d4991fafb4f64bdbfca8dd43df7d
51288 F20110115_AABAFP abiade_j_Page_130.QC.jpg
c8854b500e7bd745437f2bc1bfef6589
8f5f398fd0588223987bbeacc0a2bc726396cc5d
1743 F20110115_AAAZTO abiade_j_Page_013.txt
ff3e4febe89a4218d290cd75f22c74b7
2b88913fd95a5ed432d1d16a60f0701e674c8da5
2031 F20110115_AAAZUD abiade_j_Page_032.txt
7c33409facabd3ed32d9f5730dafec19
44722ef018b73f7debc9947e30ae0137da21715c
1085 F20110115_AAAZTP abiade_j_Page_014.txt
aebf568205ab2546484b26ae80dfba47
af8f6a80c5560bec497a10ac5f61c6d45a69efd9
F20110115_AAAZUE abiade_j_Page_033.txt
0a1fb3db497e0ceeef6f37f318760438
c46c473772ffe60290d2a78c8ca1baefaa2546d2
81387 F20110115_AABAFQ abiade_j_Page_133.QC.jpg
a8b8efd0ee21e6297889bdb98ea04095
b3920e73c36276879ea4dc5f94a0554fce35175e
F20110115_AAAZTQ abiade_j_Page_016.txt
fc70ffdffb32f1d989739c6c9b90bda5
25f62c1e47b89c72a41d183f809846f5a1384c16
79504 F20110115_AABAFR abiade_j_Page_136.QC.jpg
73bc7357fed52e78fdef87e90dd46059
965875c65a8df23314403e2ee04176c150491273
1996 F20110115_AAAZTR abiade_j_Page_017.txt
7443ad965ea3080587f5663f229f1ea0
d7b3ef2465f0e7ff9661dcab8d103d50f963c039
F20110115_AAAZUF abiade_j_Page_035.txt
3e0eeb366299f297d7618f427c80cdd2
f484d4960ea75b025eb9315ae4dd5a25a26ff24c
18666 F20110115_AABAFS abiade_j_Page_137thm.jpg
9689a81ecc1258ae8f3122ccd20d4e21
84b79461d95510022ae44a38ba411d2f22d47d0d
1995 F20110115_AAAZUG abiade_j_Page_036.txt
ee47a2b07e0d8c96d6ca5e4168b19e72
50d511bd31541a3c37100db2f4d84ba625fb9f92
2837 F20110115_AAAZTS abiade_j_Page_018.txt
d27da06a550db9a82323ff47789e43aa
d48eda533aec32377506e5a43a8e4bd13c5aacf3
58182 F20110115_AABAFT abiade_j_Page_137.QC.jpg
b11ddfb77ad2d43b024807205e9fec36
2ad164197a5c500e8c0fcd2197de3be0dc71c79f
1852 F20110115_AAAZUH abiade_j_Page_037.txt
e40ac9561e55ddabda01f2d417f20ae2
530ff1efdeb3c6626ba5799cb44af681a101d3d5
1935 F20110115_AAAZTT abiade_j_Page_019.txt
78ea86b6e8f162a2ffdf40dc4e236fcb
e512c8024cf320e380e6af204301e781ab1369e3
25112 F20110115_AABAFU abiade_j_Page_139.QC.jpg
a357afe0e9e0bde16f0d8e685b711357
0c88577e7150a8c42b70606679c58523c99a3d3c
1506 F20110115_AAAZUI abiade_j_Page_038.txt
156d2f58467363fa2687002aa0b694d3
dcb708f8873a821ac7a2db4904a54b97f7dc088d
940 F20110115_AAAZTU abiade_j_Page_020.txt
dc12fb59253f671db6d625d9a376cd38
81a23bc93ab83c81eb3304350f477c17f2f4821f
24289 F20110115_AABAFV abiade_j_Page_145thm.jpg
f1b32fde266b62322199453f83948830
1ee975cbe41078d2e6b59888647f0f54f382f134
2063 F20110115_AAAZUJ abiade_j_Page_039.txt
977eee5b7f5eb3c2d97a06dbf375313d
96d98bebca9a50f013e261a713ec790657189ab4
1167 F20110115_AAAZTV abiade_j_Page_021.txt
f42e5c0d19a33a9223b95309db509eda
b1a6676d7cc71dff14566e80fb0da9796bada97f
79067 F20110115_AABAFW abiade_j_Page_145.QC.jpg
43423f957c1e40ab17a6949e146abb36
2f22afef4a5bb65ab720b1bca4fb07357d5b1e13
1905 F20110115_AAAZUK abiade_j_Page_040.txt
59406d8f54c4dd6237dba40407d3697d
a3036df107d327355ec6e4dcf4fdd7934baf150d
1953 F20110115_AAAZTW abiade_j_Page_022.txt
89f8aecd1b01feb0bb38e2a2e5b763cc
3d944b7f4886055cdafd4b079f6fb7b073b97ead
70290 F20110115_AABAFX abiade_j_Page_146.QC.jpg
c89f10a3521058af7788c48c18018c93
4da7598de5e8bf9f74a37a4b50fd3a8ea421185c
1465 F20110115_AAAZVA abiade_j_Page_057.txt
1c3dbf7fdadc08ff2aff9c90d6532bbc
ca435f6b2c671d7435bc25d113f8388d39b56b00
1951 F20110115_AAAZUL abiade_j_Page_041.txt
e5c85a3ea1f32dd8ceb530946fb9a099
5c3bf1565ebe213e5904768803e153cdebdd16c2
2029 F20110115_AAAZTX abiade_j_Page_023.txt
f311e5bdcf57c889c6748fdfc66363f4
ecf43ef21c82079091073bbf2c6dd6527dc98765
12486 F20110115_AABAFY abiade_j_Page_147thm.jpg
7c6a25a3ad59a7621f7c9b0e86aa1cba
79ac0c9bf4c20f429d9333cda184371989842d0d
1212 F20110115_AAAZVB abiade_j_Page_058.txt
1d4394c4c497c3f0cb23bd4773377293
a6ff6e75c8b20948aa9c50871c7cb6a7015dad5c
F20110115_AAAZUM abiade_j_Page_042.txt
c920bd50bbdd8342cf984e6a63821e0c
8bb144962e977ae5c7411922dd8c2f6caaeb84ab
2126 F20110115_AAAZTY abiade_j_Page_024.txt
589909923870380a9791ee0a998a9087
6c4add688743ed8be29f6319337d484726736f5e
41280 F20110115_AABAFZ abiade_j_Page_148.QC.jpg
b396d0d3992bceab21f3b2a01b3a9d87
a97085bc95e7debbf59cb063283d7afe1a2a7d43
2005 F20110115_AAAYSA abiade_j_Page_134.txt
9f6f84fbdb4ebdc475139ba6a6928b7d
ea0254992d685bb273880c70f6562fc169653cb6
1555 F20110115_AAAZVC abiade_j_Page_059.txt
2d371dc2fa0cd9ecfe2915b0347c58ae
73ac6e7f2fd6e92894e46f970f4530dad52a5a28
2233 F20110115_AAAZUN abiade_j_Page_043.txt
c3b4326020f6b28d60c0f6aaeea6d6c9
4c03cf4e75dbb679c09fb13fe329c99fb4d92346
1202 F20110115_AAAZTZ abiade_j_Page_025.txt
7e2c4e64f27e2766f20beeaadbd5445f
cf2afa3000b16153bd12580fe0b16491da1ede5c
F20110115_AAAYSB abiade_j_Page_092.tif
1bf5744073d6a72ae2b9ef65931334fd
381461f71a156a7e5f10c433b839866c165d8c9a
1557 F20110115_AAAZVD abiade_j_Page_060.txt
126d243faacb860e632b37de25105a99
441b4da2a8d12d0e2b3aa2388df2cb810ef44d53
2083 F20110115_AAAZUO abiade_j_Page_044.txt
5f634ebae118d87fb334caec1532cf35
6963a981dc9361514ad46f1d2187c36f425804eb
25479 F20110115_AAAYSC abiade_j_Page_030thm.jpg
1a60d7141d288eb532984bfa50c15ad5
e597d972c1afd8447469a5e1ad4ab6a1c6097051
1496 F20110115_AAAZVE abiade_j_Page_062.txt
72cda3dd9c4d3d6b529aa2e5c314213e
61ea624ef78f11f85222e5a9cd7f3f7a111e591d
2013 F20110115_AAAZUP abiade_j_Page_045.txt
b3278fc2c4b84101dbe8c5fce2b6edac
72ccaa2c9ea372a0dfda241979d78e4ba87c65e5
57170 F20110115_AAAYSD abiade_j_Page_028.QC.jpg
81e89f81bb3979f035c1243d84019e9f
ea3183b8f3fc8c7279668ce75f955d23de996484
1927 F20110115_AAAZVF abiade_j_Page_064.txt
0ad6170fff3707019e18b4e758b86e8f
3588ca882d1c6ca29135990fac4e9b39a14ded9d
1961 F20110115_AAAZUQ abiade_j_Page_046.txt
a3e81601de83d28b71f8f7bb2af25d40
3a5c0620095bdbc196f04cd0606ccc26db9af820
1377 F20110115_AAAZUR abiade_j_Page_047.txt
b8334717d0def2fa0872ea9ffd3ed139
afb32652d5ef222ce9b857d40c9bed0d55dee7f0
24808 F20110115_AAAYSE abiade_j_Page_049thm.jpg
7275d07311f8560c526212b456226013
df5f03388617aaf422f259d6f0c541bb51a373e9
1940 F20110115_AAAZVG abiade_j_Page_065.txt
466206ea686a954b378ce7cade31e704
e8103ffd7aefa158e828dcb818baa2e1d0107d5c
2036 F20110115_AAAZUS abiade_j_Page_048.txt
b713f74079066eea9f345dac68124d0a
2b089bf623e7099c216eafed2f510829241a0f0d
F20110115_AAAYSF abiade_j_Page_137.tif
f272e899ce2c4418e68fcf364adde570
3df675da98138df055e2c7b2cb7697b0b57ac54e
1868 F20110115_AAAZVH abiade_j_Page_066.txt
2410e2dde3107b002e34b6314c067a0a
b8d9bd1c2e02fd5b7eda10581be29902dd5b9067
1980 F20110115_AAAZUT abiade_j_Page_049.txt
72c0c754405affc309ce6ab3f9bff9a6
7137fd056315c04e1e354b732acdbc1e7f22b5b9
63151 F20110115_AAAYSG abiade_j_Page_013.QC.jpg
80ac906b0db9664c891fbfb2b8abd6d3
5c893cdb01a82bb10c2caf00b77b7b24c40821b0
1573 F20110115_AAAZVI abiade_j_Page_067.txt
8b95facadc47de252ae0ae59ba231180
d4f5d1c2c92112dab1f82ba42bd9ce9cdcc59af1
1928 F20110115_AAAZUU abiade_j_Page_050.txt
2f9e600c55c1915d9e127adda64d27e7
c2c41202beecc7890d7e3c087d9acecb653c4af9
840943 F20110115_AAAYSH abiade_j_Page_099.jp2
7bc4b959151cf2d292fe60e899435689
71029ddbd6b1e1c0200163b560902f37274633f8
1433 F20110115_AAAZVJ abiade_j_Page_068.txt
bed27613a52c7ca4fe516ef8bdf41b4a
0a195a27f7bdc287ad1dd0bf6c1a0e404dbed0b2
2038 F20110115_AAAZUV abiade_j_Page_051.txt
e65e8e92f4741f729b9bf4fc5405e3ff
62bf4a6f8f0bde3bf2a4e34ce3910e435b7c8711
52403 F20110115_AAAYSI abiade_j_Page_133.pro
1db6307c5a79dd62c7c53e93b1c6a949
64c46c81a81074a1d0ac74e9661145adb0d1d9df
2026 F20110115_AAAZVK abiade_j_Page_069.txt
734a16c2a07af694806663ff1ddc8d0b
ab820db32d09ec55ea4e143fb749e9c6ffd4a422
2022 F20110115_AAAZUW abiade_j_Page_052.txt
0780942def5a9e658dc7b5fe1e94a140
7641f5df0c03e7bb3e3360d1d68d791af9d9b1e4
30114 F20110115_AAAYSJ abiade_j_Page_028.pro
4e8d40cf5db0b66c591ada4ac05437be
7f6a639398f9e7690f886feb73ba341a9d68ec39
1864 F20110115_AAAZVL abiade_j_Page_070.txt
501e5a5ebf746d31468336f089663842
95170a51d6c3d4f58f4b92f8715025965f244db8
1681 F20110115_AAAZUX abiade_j_Page_053.txt
adea3e1c53352375f175abf10c2887a8
8bcd514969db940165e1d45ab2c92d01814f6786
2028 F20110115_AAAZWA abiade_j_Page_091.txt
7083e49273dbaaf0e8140d8fac19c520
52018ef338579b9a59fb51bd743f1bbae5f7a304
90273 F20110115_AAAYSK abiade_j_Page_086.QC.jpg
30c7dff2fd4d90b6719592bb896f6361
dbd7b84a2926cf19b16d66abc7f54a5dfb084fcc
1957 F20110115_AAAZVM abiade_j_Page_071.txt
5dd3f23e2cf1a828c9a7e3ace4f8f83d
b4804bff7ccd80ce7b25c846b31d4797a749f0ac
1930 F20110115_AAAZUY abiade_j_Page_054.txt
aeeddb5a385d5fb2d92b6fe9290d2179
3722173d0f66a9a305d4c1d8aeb3905ecf40447e
2078 F20110115_AAAZWB abiade_j_Page_092.txt
5a95d0c6003add91f40d6ca9ca949c1a
db59fa245361c695822b7421cc01bfe0520e7a98
42644 F20110115_AAAYSL abiade_j_Page_075.pro
f28984cc66bacb2558ff4e3eac9005a9
1215c3e68c5b604aee27aa7fb37d882db164f517
F20110115_AAAZVN abiade_j_Page_072.txt
1e86715512d1b939055a331be17ed4da
825800f2a477510e52b8bc5ec17551dff9e8f776
1336 F20110115_AAAZUZ abiade_j_Page_055.txt
bcff0372078c9d50c90c62f7385b2c6d
f4cc9068b93de855636a542f8ce361223d3e88b3
97686 F20110115_AAAYTA abiade_j_Page_022.jp2
b04a6833fdc99bd81f6c4b80f56d7b72
ddac7632ee5a482976680604e88a9931220fbf1e
2030 F20110115_AAAZWC abiade_j_Page_093.txt
ccc4cff3e59d1d2006b0322abf6cca16
090d801404232731b6a26c0c0d00bc55c7efd7f2
185080 F20110115_AAAYSM abiade_j_Page_138.jpg
43a62fabdb82cd89749f724865ededff
0ed90d80cfe536b08bf82d5e2e62e0170294fcb5
1875 F20110115_AAAZVO abiade_j_Page_073.txt
b663ceffdeac8c28018a719bd5c23e01
d30c4d9860f11b060474cace5c8c5958b6536a27
47240 F20110115_AAAYTB abiade_j_Page_038thm.jpg
c9198a7139f56c04c543ad4a63c9b6a9
aaf3b375db361efec480deec640eb2bd94b84d52
1366 F20110115_AAAZWD abiade_j_Page_094.txt
2cae439b998bb1493df57c983e019075
650831bbff75cba6629c9d180350767b70c16d8f
82304 F20110115_AAAYSN abiade_j_Page_101.QC.jpg
9057ab39040c3339adf2014eb510e99c
73bdc8f5ce7da5903b105e89e3f38c59b3331c07
281 F20110115_AAAZVP abiade_j_Page_074.txt
79c136de70c461acd3e769ef046a0572
02aeccfb75d4511d80e437adfc1376de997a014c
42844 F20110115_AAAYRZ abiade_j_Page_090.pro
cff88c221ea030f956c3508bccd1d3f9
5c74b8e41fafc0c02f879ee79c790ce1c37a5306
2794 F20110115_AAAYTC abiade_j_Page_008.txt
daa1008a22b44667b2fc291fdf19b33b
9b82e34ed48e4973465850fd983fb8d35a95598a
2034 F20110115_AAAZWE abiade_j_Page_096.txt
9ffee409ad7749d08c85f86100eaff16
7d305179ea7d2a595455be0ba2aa696d682cd431
2050 F20110115_AAAZWF abiade_j_Page_097.txt
84fe0a398d0fdbcbbb597fb272640198
19293bae805fd9b501038c74ba1ae230419b9767
20156 F20110115_AAAYSO abiade_j_Page_060thm.jpg
795f3438e9c08f0f4cc79c2b68413b14
1d7666b287535e11b59c4c76e7006bb58e1d9c15
1758 F20110115_AAAZVQ abiade_j_Page_075.txt
e92170c15d101db1783aee064cef90a4
b31f50293bcc4b7429afcfbde3c4e6a787b2d4c2
25319 F20110115_AAAYTD abiade_j_Page_118thm.jpg
0a5ac2447490da0a6a9346a258c87245
b2b0f6fc20b293ae6b8772b24f3a31fbb7ac353d
F20110115_AAAZWG abiade_j_Page_098.txt
2f85215085e2d2fea4d6c64f4aa78eef
ba56c04dc61992c64bbb9127db7857cd4301d0b1
45606 F20110115_AAAYSP abiade_j_Page_070.pro
e36b0c485b72dcd72e29f0f1184479c0
2c30491916c92f35a8480232004f408c86ef6b66
2020 F20110115_AAAZVR abiade_j_Page_078.txt
cca21916a86c63b0272d6c5890cc53bb
f27c2044f667ddfeb5696dd3ea4a766ae7f30d61
F20110115_AAAYTE abiade_j_Page_104.tif
ecb5610f7349b20e97ac99e460a1aca0
f4d50fc64b97acd2eed864fc00cdc9c616825ad8
46306 F20110115_AAAYSQ abiade_j_Page_073.pro
759308c028c828ae1db3832cae2854d1
dbdbc913a92d1f97cbce1c08df2c6ee127e0d642
F20110115_AAAZVS abiade_j_Page_079.txt
5fd148ac2eb799048360f5b2a0cdfccb
2008981c16aebda14339dd0c021287af920ff334
2057 F20110115_AAAZWH abiade_j_Page_101.txt
5f6f967964012d55848df864b78d6012
267f6f8e9c733edba441ef313b9996e64c0da643
764867 F20110115_AAAYSR abiade_j_Page_106.jp2
a87512f7dcea874e81329acc2d565bf4
7be9abdfc9fd87beb58d5719266522c7657a2b7d
1012 F20110115_AAAZVT abiade_j_Page_081.txt
f2003589b4a77144996edc3066ae4f12
e23052be9b9fdfea543029b4a763433958f41473
177444 F20110115_AAAYTF abiade_j_Page_015.jpg
4e062d30ba04af9031407c40555568e8
a6aadf8c6f9770512531715e5fb07f4d0a63eb73
1896 F20110115_AAAZWI abiade_j_Page_102.txt
d6e5613a4746e19f7d6190d837b4f342
0ad189448637f5a7718f804fd9b74fc6bb6c5977
F20110115_AAAYSS abiade_j_Page_030.tif
255c939c7ae6d0c9c39307dca92b66bc
c3248a681a7e5a3a4fe60ec6c79232f99f2ae480
1746 F20110115_AAAZVU abiade_j_Page_082.txt
6c9e2f2477a866837c8c5e8511ff2622
ff649e9400eb61a59e25c36b8e62942cdb844d39
24224 F20110115_AAAYTG abiade_j_Page_110thm.jpg
e3a8cf386616fb6037b35b4cc6e26677
7ca4dd86d69400365b1fa8044aea2e101b5c63f3
972 F20110115_AAAZWJ abiade_j_Page_103.txt
a353d51cb56a1e886e9e8e7020f6248f
7dc55e750372a64e0030aa1b79ab755b63c197a3
207777 F20110115_AAAYST abiade_j_Page_134.jpg
307d9e8a5b4be8fe03bd0c8617f5db16
3715d42f47ffa363ead0e0e587680180af470d97
1714 F20110115_AAAZVV abiade_j_Page_086.txt
712a65cc75c761a02fb3df6f20c0a7ec
fe00f3d8cd8487875db0162b7fce3734bc67c7b9
F20110115_AAAYTH abiade_j_Page_025.jp2
78bd151f592d18289b2c0357493996cd
c9edf80f553eb4672b2d49672c6771fb15646428
1629 F20110115_AAAZWK abiade_j_Page_104.txt
0990519753d3236ff2ba2376cd671ddc
7cb3cdee92f73d7363d89716ec7bc1e59b9b5bdb
58536 F20110115_AAAYSU abiade_j_Page_060.QC.jpg
1dd8a9ff594e20119e1ec7b445cbabfb
89bc82a31704626d5ebbdd42f83e1fe95e3d1d7b
2173 F20110115_AAAZVW abiade_j_Page_087.txt
2ddfd5abda59c600d47ff26835eb45f5
d5e463f2646285433e98a1d7b771fab951e7d81d
96668 F20110115_AAAYTI abiade_j_Page_107.QC.jpg
78f2d0dc8cec03ec0169ddd819993b5f
034116d594f5bb4c9b4ba9559f573f5524fc9099
1296 F20110115_AAAZXA abiade_j_Page_126.txt
92582261265e5465ae39667753bf2291
fc92652e23ab45253ec14143984565e5de982456
1690 F20110115_AAAZWL abiade_j_Page_105.txt
ef76189ca6f4e1d74b62187ee4d01aed
27bf2c122e2e527e95c8b94e699ad8674b142d96
45419 F20110115_AAAYSV abiade_j_Page_089.jp2
066e56dffe593834443e626340772e41
421043233530513e953008d6103eec73c87c148e
1198 F20110115_AAAZVX abiade_j_Page_088.txt
544d40b0096e9981fdccc792880588b3
a2e4bd146812f205fae7155ac7958ecf36094aa7
1732 F20110115_AAAYTJ abiade_j_Page_063.txt
a465616d5d1b1673dc24de6056c050b1
1a4a0d76387545f0150e8f78789224f4d826baac
1183 F20110115_AAAZXB abiade_j_Page_127.txt
3c30070a336cee5abf66ffbddd836714
ff5bf0bb611dd4bb00b322ddececb00b37189281
1255 F20110115_AAAZWM abiade_j_Page_108.txt
c8217158d4fdce2086d9c945b96a9ce8
7f5ae522797216fd26bd6b8d230714cc41121acf
F20110115_AAAYSW abiade_j_Page_140.tif
b8f8fd57fc30144bf24da41665c20f70
ef66239a63b8188dde3dfba710d944557dcbdf0e
794 F20110115_AAAZVY abiade_j_Page_089.txt
675d6430d43cc10ffac224de4a28f8fe
c5d91653a9c5098bfe1627b749cf9674f7fb074b
152593 F20110115_AAAYTK abiade_j_Page_059.jpg
6019cc96203cb4d484f64fb40832bb50
9972909af40e27babf8fbb2477cd4c06e604f6fa


Permanent Link: http://ufdc.ufl.edu/UFE0008394/00001

Material Information

Title: Characterization of the Chemical Effects of Ceria Slurries for Dielectric Chemical Mechanical Polishing
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0008394:00001

Permanent Link: http://ufdc.ufl.edu/UFE0008394/00001

Material Information

Title: Characterization of the Chemical Effects of Ceria Slurries for Dielectric Chemical Mechanical Polishing
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0008394:00001


This item has the following downloads:


Full Text












CHARACTERIZATION OF THE CHEMICAL EFFECTS OF CERIA SLURRIES FOR
DIELECTRIC CHEMICAL MECHANICAL POLISHING














By

JEREMIAH TERRELL ABIADE


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


2004



























Copyright 2004

by

Jeremiah Terrell Abiade



























This dissertation is dedicated to my wife Kalia and daughter Asha; my parents Candy
Grey and Renee McGee; my siblings, Marguerita, Paulette, Jonathan, Candon, Pilar, and
Joshua. I also would like to dedicate this work to all of the elders of my family who were
denied the educational experiences that I have been so fortunate to enjoy.















ACKNOWLEDGMENTS

First and foremost I would like to thank God for giving me the ability to complete

this dissertation and surrounding me with such delightful people that have made this such

an enlightening experience. I would also like to thank my advisor, Dr. Rajiv K. Singh,

whose emphasis on sound reasoning, the fundamentals of materials science, and creative

thinking has contributed immeasurably to my graduate experience. My committee, Dr.

Brij Moudgil, Dr. Wolfgang Sigmund, Dr. Stephen Pearton, and Dr. Vaneica Young,

have been extremely helpful. I would also like to thank Dr. Chang-won Park for his many

insightful comments and for dutifully maintaining the CMP group, and Dr. Valentin

Craciun and Dr. Hassan El-Shall for lending their expertise to particular areas of my

research. I am also extremely grateful to Dr. Diola Bagayoko of Southern University for

convincing me that the Ph.D. was a realistic goal to work toward.

I am extremely grateful to the National Science Foundation Alliance for Graduate

Education and the Professoriate Program (NSF-AGEP), the UF College of Engineering

Minority Fellowship Program, NSF-AGEP Dissertation Fellowship Program, and the

Office of Graduate Minority Programs, the Particle Engineering Research Center

(PERC), and the industrial partners of the PERC for financial support throughout my

graduate program. Special thanks go to Dr. Anne Donnelly, Dr. Dovie Gamble, Dr. Terry

Mills, Dr. Jonathan Earle, and Martha McDonald for encouragement, moral support and

assistance with any problems I encountered. I would also like to thank the PERC and

MAIC staff members Gary Schieffle, Gil Brubaker, John Henderson, Mike Beasley, Dr.









Emilia Hodge, Dr. Kevin Powers, Sophie Leone, Eric Lambers, Kerry Seiben, and Mike

Tollon for their assistance throughout. I am extremely grateful for the assistance of Mr.

James Albury of the ETD Office at the University of Florida. I would also like to extend

my gratitude to Professor Brij Moudgil, Professor Paul Holloway, Professor Rolf

Hummel, and Professor Robert Dehoff who have set great examples of lifetime

excellence in both teaching and research.

I thank my colleagues, past and present, Uday Mahajan, Zhan Chen, Seung-Mahn

Lee, Bahar Basim, Kyo-se Choi, Wonseop Choi, Wonseok Kim, Madhavan Esayanur,

Suresh Yeruva, Kyoung-ho Bu, Su-Ho Jung, Sang Hyun Yoon, Han Ho Choi, Karthik

Ramani, Anthony Stewart, Robert Crosby, Eboni Westbrooke, Lizandra Williams,

Jeremiah Smith, Nakato Kibuyaga, and Samesha Barnes, for their contributions to this

research with valuable discussions and friendship.

Of course I would like to thank my family and friends. My wife Kalia and daughter

Asha have provided support, comfort, and love throughout this process. My parents,

Candy Grey and Renee McGee; my siblings, Marguerita, Paulette, Jonathan, Candon,

Pilar, and Joshua; and the rest of my family that have supported me throughout. Finally,

thank you to the brothers of the Omega Psi Phi Fraternity, Beta Pi and Omicron Zeta

Chapters.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ..................................................... ix

LIST OF FIGURES .............. .................................................... ...... .x

A B S T R A C T .................................................................................................... ........ .. x iii

CHAPTER

1 IN TR O D U C T IO N ........ .. ......................................... ..........................................1.

B a c k g ro u n d ................................................................................................................... 1
M ultilevel M etallization ................. ...........................................................2......
C hem ical M echanical Polishing ....................................................... ...............2...
Scope of the D issertation ...................................................................... ............... 5

2 L IT E R A TU R E R E V IE W .................................................................. ..................... 7

Introduction to Multilevel Metallization (MLM) ...................................................7...
M motivation ..........................................................................................7..........
P process F low ................................................... .......................................... . 9
Chemical Mechanical Polishing ....................................................... 10
CM P C om ponents ................ .............. ............................................ 12
P olishing pad .................................................................... ............. 13
W a fe r ...................................................................................................... . 1 5
S lu rry ...................................................................................................... . 1 7
Silica-Silica C M P M echanism s ..................................... ..................... ............... 18
G lass P o lish in g .................................................................................................... 19
Silica H ydroly sis ........................................................................................ 20
Silica CMP................................................................ ........................... 22
Introduction to Shallow Trench Isolation (STI) ....................................................30
C eria-Silica C M P M echanism s.............................................................. ................ 32
C hem ical T ooth ............................................................................................. 33
S u rface D effect ..................................................................................................... 3 6
P articulate G generation ......................................... ........................ ................ 37
C eria Surface C hem istry .................................................................. ................ 38









3 M A TERIALS AND M ETH OD S .......................................................... ................ 45

Introduction .................................................................................. ....................... 45
Samples ............................................. ............................... 45
Silica W afers ................................................................................................. 45
C eria Slurries .................................................................................................. 45
C eria T hin F ilm s ...............................................................................................46
Functionalized Cantilevers ...............................................................................46
C antilever oxidation ..................................................................................46
Particle attachm ent ....................................................................................46
CM P Equipm ent .......................................................................................... . 47
CM P Tabletop Polishers................................................................. ............... 47
Friction Force M easurement Apparatus...........................................................48
Particle C haracterization .......................................................................................49
Zeta Potential M easurem ent.............................................................................49
P article Size A analysis .......................................................................................51
Slurry Stability..................................................................................................52
Characterization of Silica Surface M orphology ......................................................53
Atomic Force M icroscopy (AFM ) ....................................................................53
Scanning Electron M icroscopy.........................................................................55
X-ray Photoelectron Spectroscopy .................... ....................................56
Spectroscopic Ellipsometry (VASE) ................................................................58

4 THE EFFECT OF PRESSURE, PH, AND VELOCITY ON THE SILICA
REMOVAL RATE AND SURFACE FINISH: A STATISTICAL DESIGN OF
E X P E R IM E N T S .....................................................................................6 1

In tro d u c tio n ................. ...............................................................................................6 1
Selecting a Statistical D esign ...........................................................................63
Role of Pressure, Velocity, and pH in Silica CMP ............................ ...............64
E x p e rim e n ta l ............................................................................................................... 6 6
R results and D iscu ssion .......................................................................... ...............67
R em oval R ate R response ................................................................................67
Surface Finish R response ...................................................................................72
Sum m ary ......................................................................................... ............. ....... 74

5 THE EFFECT OF CERIA SLURRY pH ON THE SILICA REMOVAL RATE,
FRICTION FORCE, AND SURFACE MORHOLOGY DURING CHEMICAL
M E CH A N IC A L PO LISH IN G .................................................................................... 76

In tro d u ctio n .................................................................................................................7 6
Surfactant Based STI CM P ..................... ..........................................78
Fixed Abrasive Pads for STI CMP.................... ....................................80
E xperim mental ................................................................................ ....................... 83
R results and D iscu ssion .......................................................................... ...............84
Ceria Slurry Characterization...........................................................................84
Silica R em oval R ate .........................................................................................88









Friction at the Ceria-Silica Interface .................................................. 91
Surface M orphology ....................................................................... ................ 93
Comparison of pH Effects of Ceria-Silica to Silica-Silica CMP .....................94
S u m m ary ......................................................... ................................................ 9 7

6 NEAR SURFACE ANALYSIS OF POLISHED SILICA: CHARACTERIZATION
OF THE CERIA-SILICA CHEMICAL REACTION...........................................99

In tro d u ctio n ................................................................................................................ 9 9
Ceria CM P M odels ..................................................................... ............. 99
R recent Progress ............. ................. .............................................. 100
E x p erim en tal ............................................................................................................. 10 4
R results and D discussion ................ .............. ............................................ 105
Interaction Force Measurements ....... ... ..........................105
X P S A n aly sis ..................................................................................................... 10 6
FE -SE M A analysis ........................................................ ......... ...... ... ............ .. 109
Wear with Silica-Coated AFM Cantilevers.................................................. 111
S u m m a ry ............................................................................................................... ... 1 1 5

7 SUMMARY, CONCLUSIONS, AND SUGGESTIONS FOR FUTURE WORK.. 117

S u m m a ry ................................................................................................................... 1 1 7
C o n c lu sio n s.............................................................................................................. 12 0
Suggestions for Future W ork................................... ...................... ............... 122

APPENDIX

A DEFINITION OF COMMON STATISTICAL TERMS ............... ..................... 124

B BOX BEHNKEN DESIGN AND RESPONSES .......................... ..................... 125

L IST O F R E F E R E N C E S ................................................................................................. 127

BIOGRAPH ICAL SKETCH .................. .............................................................. 134
















LIST OF TABLES


Table page

6-1 XPS binding energies for silica samples before and after polishing................... 108

B-1 Box Behnken design for three factor experiment........................ ...................126
















LIST OF FIGURES


Figure page

1-1 Schem atic illustration of a typical CM P tool...................................... ...............1...

1-2 Correlation of system-level variations with micro- and nanoscale effects and
output parameters ................... .. ........... .........................................4

2-1 Cross-section schematic of MLM structure with silica dielectric and aluminum
m etalliz atio n ..................................................................................................... .

2-2 Schematic depiction of cumulative increase in topography for non-planar
M L M stru ctu res. .................................................................................................... 1 1

2-3 D degrees of planarization. ................ ............................................................ 12

2-4 The effect of water on the mechanical properties of polyurethane chemistry .......14

2-5 Current recorded for one stroke during tribological experiments on a tungsten
substrate .............. ...................................................... .................................... 17

2-6 Material removal rate of silica as a function of the concentration of silica
slurries for different particle sizes..................................................... ................ 23

2-7 Indentation depth as a function of particle size. ............................... ................ 24

2-8 Particle motion as a function of silica particle size and concentration...............29

2-9 Schematic depiction of device isolation: .................................................... 31

2-10 Glass polish rate versus IEP of common abrasives ..........................................33

2-11 C eria crystal structure ................................................................... ................ 39

2-12 Potential versus distance for a charged surface. ...............................................41

2-13 pH versus acid/base concentration ................................................... ................ 43

3-1 SE M analysis of C M P pad .......................................... ...................... ............... 48

3-2 Schem atic of in-situ friction force apparatus ................................... ................ 49









3-3 Schem atic diagram of A FM apparatus ................................................ ............... 54

4-1 Factor space for different statistical designs ............... .............. ..................... 67

4-2 Predictions of removal rate as a function of pressure and pH ..............................69

4-3 Interaction profiles for removal rate as a function of pressure and pH .............. 70

4-4 Predicted removal rate as a function of pressure, pH, and velocity....................71

4-5 Predictions of RMS roughness as a function of pressure and pH.......................72

4-6 Interaction profiles for RMS roughness as a function of pressure and pH............73

4-7 Predicted removal rate as a function of pressure, pH, and velocity....................74

5-1 SEM micrograph of fixed abrasive pad surface (200x) [Vel00]............................80

5-2 Ceria slurry characteristics as a function of pH ............................... ................ 84

5-3 Ceria slurry stability ........................................................................ 85

5-4 FE -SEM im age of ceria particles ......................................................... ............... 86

5-5 Silica removal rate as a function of ceria slurry pH..........................................89

5-6 Post-CMP slurry characteristics as a function of pH ........................................89

5-7 Zeta potential versus pH for ceria, silica, and co-dispersions...............................90

5-8 Friction force and rem oval rate......................................................... ................ 91

5-9 Qualitative comparison of electrostatic interactions.........................................92

5-10 AFM topographical analysis of silica wafers ............................. ..................... 93

5-11 RMS roughness of silica wafers as a function of slurry pH. ................................94

5-12 pH dependent ceria-silica and silica-silica interactions.................................95

6-1 Ratio of Ce Ms/M4 white lines versus position...... .................. ...................103

6-2 Interaction force .... .. .................................. .......................................... 106

6-3 X P S analysis of silica....................................... ......................... ............... 109

6-4 FE-SEM analysis of silica polished at pH 6.0. .................................. ............... 110

6-5 FE-SEM analysis of silica polished at pH 2.5 ........................110









6-6 C eria-silica w ear at pH 2.5............................... ......................... ............... 112

6-7 C eria-silica w ear at pH 4.5.................................. ...................... ............... 113

6-8 C eria-silica w ear at pH 6.0........................................................ ............... 114

6-9 Ceria-silica w ear at pH 11.0...... ............ ............ ..................... 114

6-10 Silica-silica w ear versus pH ...... ............ ............ ..................... 115

7-1 Schematic depiction of pH-dependent ceria-silica removal mechanism ...........121














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

CHARACTERIZATION OF THE CHEMICAL EFFECTS OF CERIA SLURRIES FOR
DIELECTRIC CHEMICAL MECHANICAL POLISHING
By

Jeremiah Terrell Abiade

December 2004

Chair: Rajiv K. Singh
Major Department: Materials Science and Engineering

In a little over a decade, chemical mechanical polishing (CMP) has grown from a

niche application to an enabling technology for global planarization of virtually any

material. Initially, CMP was limited to planarization of the silicon substrate during

integrated circuit (IC) manufacturing. However, CMP has been successfully implemented

in all phases of IC manufacturing. Since CMP is materials insensitive and the only

process capable of producing globally planar surfaces, it is being investigated for new

processing techniques like shallow trench isolation (STI), micro electromechanical

systems (MEMS) fabrication, and a variety of other applications where the surface

properties control device performance.

Despite the widespread acceptance of CMP, it remains the least understood step in

the microelectronic device fabrication process. The emergence of novel technologies like

STI that demand tighter control over the CMP process have stimulated increased interest

in the fundamentals of the CMP process. In this work, the fundamental properties of ceria

abrasives for STI CMP were studied.









Ceria slurries are promising for STI CMP because of their intrinsic preferential

polishing of oxides over nitrides or polysilicon. This advantageous behavior has been

attributed to a chemical reaction between the ceria abrasive and the oxide substrate. To

date, ceria-silica polishing mechanisms have been supported by little to no experimental

evidence. As a result of fundamental investigations, the silica polish rate and surface

finish has been found to be a strong function of the ceria slurry pH, with the maximum

polish rate occurring near the isoelectric point of ceria. An assortment of complementary

surface analytical techniques was used to study the proposed abrasive-surface reaction

that is widely attributed to ceria-silica CMP. The ceria-silica reaction has been confirmed

by x-ray photoelectron spectroscopy (XPS), field emission scanning electron microscopy

(FE-SEM), and atomic force microscopy (AFM) Finally, the surface chemistry of

ceria has been reviewed and correlated with ceria-silica CMP observations.














CHAPTER 1
INTRODUCTION

Background

In little more than a decade, chemical mechanical polishing or chemical mechanical

planarization (CMP) has grown from a niche application to an enabling technology for

producing globally planar surfaces. Initially, CMP was developed to minimize

topographical variations on Si substrates. The technology has since been applied to

dielectrics (Si02, Si3N4, polymers, etc.), metals (W, Al, Cu, Pt, Ni), and semiconductors

(SiC, GaN, silicon) for integrated circuit (IC) manufacturing and a variety of other

emerging technologies such as microelectromechanical systems (MEMS) and organic

light emitting diodes for flat panel displays. The use of CMP in such diverse technologies

has created interest in fundamental aspects of the CMP process and development of novel

designer slurries with tightly controlled polishing characteristics. Dielectric CMP

processes essential for manufacturing metal-oxide-semiconductor field effect transistors

(MOS-FETs) for microprocessor or dynamic random access memory (DRAM)

applications are the focus of this research.


RuSUM

V \, /*Cherric is

Figure 1-1 Schematic illustration of a typical CMP tool.



Figure 1-1 Schematic illustration of a typical CMP tool.









Multilevel Metallization

Generally, CMP systems consist of three main components: a polymeric polishing

pad, the slurry with abrasive particles, and, of course, the surface to be polished (wafer)

(figure 1.1). During CMP, the wafer is pressed against the compliant pad, which serves as

a conduit for the abrasive particles that are suspended in an aqueous solution. Surface

topography is reduced by simultaneous chemical and mechanical interactions that may be

controlled by judicious selection of particle chemistry and slurry additives. The

popularity of CMP in microelectronic device fabrication corresponds directly with device

shrinkage, increasing wafer diameter, and an exponential increase in the number of IC

components. For example, there were approximately 2300 transistors on a single chip in

Intel's first generation of microprocessors. This number is expected to reach one billion

by 2007 [ITRS03]. In fact, Intel recently manufactured chips with five hundred million

transistors [Biz04]. Due to the finite wafer area available to chip manufactures, multilevel

metallization (MLM) techniques were adopted. An IC fabricated via MLM techniques is

analogous to construction in crowded urban areas like New York City or Tokyo. Once

the available real estate has been consumed, the only option is to build vertical structures

of increasing height. In MLM, alternating layers of dielectric and metal are deposited on

the semiconducting substrate. Since topographical variations are additive, each layer must

be globally planar before the next is deposited. To date, the CMP process is the only

technique capable of producing globally planarized services [Ste97, Sin02, Eva02].

Chemical Mechanical Polishing

CMP is a deceptively complex process. Ideally, CMP should produce high removal

rates (up to 800 nm/min), global planarity, good surface finish, and high selectivity to

underlying barrier or isolation materials all with minimal defectivity, dishing, or erosion









[Sin02]. An additional complicating factor is the number of variables and their

interactions that influence the CMP process. Despite its importance, the CMP process

remains the least understood step during IC fabrication [Sin02]. In fact, early CMP

processes were based on little more than an empirical understanding of CMP phenomena

[Ste97, Sin02]. However, narrowing of the CMP process window because of the

increasing needs of microelectronic device manufacturers has driven the need to reach a

more mechanistic understanding of CMP. A major hurdle is the lack of understanding of

interactions at the pad-particles-wafer interface. Conventional CMP processes rely

heavily on end-point detection methods to terminate the CMP process before

overpolishing ensues [Sin02]. Although the CMP process has matured significantly since

its inception, efforts to understand wafer-slurry interactions still deserve considerable

attention. This is especially true for emerging applications such as shallow trench

isolation (STI), which requires alternative slurries with tunable properties for enhanced

selectivity and planarity [Sin02, Eva04].

Initially, silica slurries were used to planarize interlevel dielectric (ILD) layers.

However, the advent of new technologies coupled with increasingly stringent

requirements for the CMP process has driven development of new abrasive particles for

CMP slurries. Ceria slurries have stimulated widespread interest because of

enhancements in material removal rate, surface finish, and selectivity over slurries

formulated with silica abrasives. The replacement of local oxidation of silicon with STI

has increased the demand for high selectivity slurries capable of removing the overburden

of silica with minimal polishing of the underlying silicon nitride, which isolates active

areas of the device. This shift to novel slurries is difficult without an understanding of the











slurry-wafer interactions. To develop CMP processes for STI, a thorough understanding


of the variables that control the CMP process must be reached. In the early stages of

CMP, empirical cause-and-effect observations of CMP performance as a function of

applied pressure, carrier velocity, and pad type were utilized to explain CMP

mechanisms. Years later, fundamental investigations were launched into the effect of

process parameters such as slurry pH, abrasive size, and abrasive concentration to further

explain CMP phenomena. Two major obstacles to understanding of CMP mechanisms

are: the large number of process variables (figure 1.2) and their interactions and the lack

of sufficiently sensitive techniques for measuring nano-scale phenomena during the CMP

process. The CMP field has matured to the point of understanding that systematic

investigations of micro- to nano-scale interactions at the pad-particles-wafer interface

must be understood to extend the applicability of the CMP process to future generations

of IC manufacturing.

SINPL'I I ACRO-SCAL 4 I'NANO.SCALE L
VARIABLES mTEr.I0n I R.C-IIONaj
Particle
Chlrmcteristic
S ize Pd CHEMO.
SSize Distribution L .::n.l'Ar J MECHANICAL
Shape r n t.p fDynr; f Burfare
Mechanical Prop, Lyrr Formatimu
'Jl-i i ti Panrt Ir nn Ped l T'ia:rd:
I Lr. e"31t.LJity P r e Il j'if.,I '-i-'iL.
Concentration Coverage u.3ci
ChamiIt h*LAYER REMOVAL
SSlurry Chemicals Chmrnncal o
P pH ComrrnrntrUn &,
DnwnPressure& C W quCa ,,l
iurii I*y *mnrtcd &
FVdelCinct i l Cotact Mode ri .n -J'-
> Mechanical Prop, J i v, MECIL.N ICAL
SToporaphy H.Jr.:1rani y* El j -
Substrate I t'h~nJ :J E':]r "'vil
FMechanical PFopr
ik-rr 'Rsmovalrnte
"' ''T r rT L PlanunzaIaou
PARA1MET3& 1Surface Finish
Seleciity


Figure 1-2 Correlation of system-level variations with micro- and nanoscale effects and
output parameters in chemical. mechanical polishng [Sin02a].









Scope of the Dissertation

This study will focus specifically on dielectric CMP processes using ceria slurries.

Despite the promise of ceria for STI CMP, few studies exist on the fundamentals of

slurry-wafer interactions. Ceria-silica CMP mechanisms are assumed to be identical to

ceria-glass polishing mechanisms. Furthermore, ceria CMP efforts are toward the use of

surfactants to selectively adsorb onto the silicon nitride surface to enhance selectivity

with respect to the overlying silica. Material removal during CMP takes place because of

synergistic chemical and mechanical interactions. Ceria slurries are expected to have an

enhanced chemical contribution to polishing. Therefore, investigations in this study are

focused on chemical and surface chemical effects that control the CMP process. A

synopsis of the efforts in this study follows.

Chapter 2 reviews the state-of-the-art of the CMP process. Motivations for the

implementation of the CMP process are discussed. The roles of the pad, slurry, and wafer

are covered in greater detail. Further, the current understanding of silica-silica CMP

processes is summarized. A review of STI and STI CMP follows. Ceria behaves

differently than silica both in solution and at the pad-wafer interface. Therefore, the

surface chemistry of ceria will be reviewed.

Chapter 3 details the materials and methods used for this study. This work was

approached with the viewpoint that CMP experiments alone are insufficient to explain

polishing phenomena. Experiments and measurements were obtained using standard

laboratory CMP equipment and surface sensitive characterization techniques.

Chapter 4 features a statistical examination of the effect of pressure, pH, and

velocity on ceria-silica CMP. Studies that examine these crucial process parameters









neglect interactions between the variables. Using statistical methods the individual and

synergistic effects may be ascertained.

Chapter 5 examines the effect of pH on the silica removal rate, surface finish, and

in-situ lateral friction force. Particle characteristics (zeta potential and particle diameter):

before and after CMP will be presented.

Chapter 6 is a study of the nano-scale phenomena that affect the CMP process.

Atomic force microscopy (AFM), x-ray photoelectron spectroscopy (XPS), and field

emission scanning electron microscopy (FE-SEM) have been used to characterize the

extent of particle-wafer interactions during ceria-silica CMP.

Chapter 7 is a summary and offers conclusions of this work. A pH-dependent ceria-

silica polishing model is proposed. Chapter 7 ends with some suggestions for future work

for further enhancement of the STI CMP process.














CHAPTER 2
LITERATURE REVIEW

Introduction to Multilevel Metallization (MLM)

Motivation

It is impossible to provide a comprehensive review of CMP for IC fabrication

without discussing multilevel metallization (MLM), figure 2-1. Modem ICs contain



LAYERS 5,

















Figure 2-1 Cross-section schematic of MLM structure with silica dielectric and aluminum
metallization [Sti97].

several million components (transistors, resistors, and capacitors) on a single chip. To

form a circuit, these elements must be connected with some form of wiring. MLM

provides a more efficient means of routing metal lines. Historically, each new generation

of IC has involved a shrink of the device size [Wil93]. Device shrinkage results in

improved speed, increased packing density, and greater functionality. MLM also provides
4k6 t .


















improved speed, increased packing density, and greater functionality. MLM also provides









a means of separating long, winding metal lines resulting in increased yield and device

reliability. CMP is an enabling technology for the production of MLM structures.

Combined, these factors are responsible for the decline in prices and product dimensions

for seemingly omnipresent items like the desktop or laptop computer, mobile phone, and

calculator. As the device size decreases, and the number of components in the circuit

increases, the interconnect becomes a limiting factor in chip performance.

The most widely used measure of IC performance is the product of the resistance in

ohms (Equation 2.1) and capacitance in farads (Equation 2.2), commonly called the RC

time constant (Equation 2.3), measured in seconds, which is the time required for a

voltage at one end of a metal line to reach ~ 63% of its final value at the opposite end

[Ste97].

R =p /wd = p 1/A (2.1)

C = swl/t (2.2)

RC = p 12/td (2.3)

Where p is the resistivity of the metal line, 1, w, d, and A are the length, width, thickness,

and cross sectional area respectively and F is the permittivity of the dielectric, which

equals the insulator dielectric constant (er) times the permittivity of free space (So), and t

is the thickness of the insulator. According to (Equation 2.3), the time constant may be

lowered by using metal lines of lower resistivity (i.e. replacing Al with Cu), insulators

with lower dielectric constants (i.e., porous silica, or polymeric materials instead of

SiO2), or by decreasing the length of the metal lines. Equation 2.3 does not show a direct

dependence on line width, but as devices are continually miniaturized, the metal line

thickness (d) and interlevel (ILD) thickness (t) are also scaled to smaller dimensions,









causing an increase in the time constant. Scaling the device dimensions is one alternative

for improving device speed, since device delay decreases with decreasing feature size

[Ste97]. However, for 0.5 [tm features, interconnect delay accounts for ~ 25% of the total

delay, while for 0.25 [tm dimensions, ~ 50% of the delay is due to the interconnect

[Ste97]. With continued scaling, interconnect performance dominates device speed

resulting in diminishing returns as the dimensions are aggressively reduced. Any change

that decreases p, 1, F, or increases t and d will enhance device speed by decreasing the

interconnect delay. Since RC is proportional to 12, the most dramatic decreases in

interconnect delay can be achieved by avoiding long metal lines. This is possible using

MLM [Wil93, Ste97].

Process Flow

The process flow for MLM is as follows [Wil93]: silicides (metal-Si compounds)

are formed on the source, gate, and drain of MOS transistors as well as on doped

polysilicon films that are used as resistors or contacts in bipolar circuits. Next, the pre-

metal dielectric (ILDO) is deposited. The ILDO and subsequent levels of dielectric (ILD1,

2, 3, ...) are different materials. Boron phosphate silicate glass (BPSG) [Ols93] is

commonly chosen for the ILDO layer whereas silica (Si02), silicon nitride (Si3N4), or

silicon oxynitride (SiOxNy) are used for the ILD. Depending on the planarity

requirements and thermal budget, planarization ensues via thermal flowing [Wil93],

etchback [Che97], or CMP [Ste97]. The ILDO layer is then patterned and etched to form

contact holes through the oxide to the underlying structures or silicide. The holes or vias

(VI, 2, ...) are then filled by barrier film if none exists on silicide, then a conducting film

(Al or W) is deposited. Currently, W is the metal of choice because of its ability to

conformally fill high aspect ratio vias [Ire97]. The W is etched back to form a plug. The









interconnect metal (Ml, 2, ...) is then deposited. Cu has replaced Al as the metal of

choice for the metal interconnect because of its low resistivity and improved

electromigration resistance resulting in faster, more reliable devices [Lee03]. Vertical

structures are built by replicating the procedure of dielectric deposition, etching to form

vertical vias, patterning and interconnect deposition. The MLM process is completed by

depositing a final dielectric layer that hermetically protects the underlying layers from

corrosive ions, moisture, and solder chemicals. For feature sizes of -100 am, 12 metal

layers are required when using Al interconnects and SiO2 as the ILD. In contrast, only 6

levels are required using Cu interconnects and a low dielectric constant material [Sin98].

MLM is not a simple process. A variety of tools, materials, and techniques are involved.

Stringent control must be maintained over each sequence in the process.

Chemical Mechanical Polishing

Initially, it may seem counter-intuitive to remove any of the dielectric or metal

layers that have been so carefully deposited, especially by exposing the pristine

environment of microelectronic fabs to such a dirty process as CMP. Topographical

variations or thickness non-uniformities are deleterious to device performance [Ols93].

Also, in multilevel structures necessary for ICs, topographical variations are additive

(stack-up effect) [Vie01]. There are several sources of topography generation during

fabrication. Non-uniform deposition may occur if curved wafers are used. Wafer

curvature may be because of intrinsic stress in the substrate [Ste97]. The fabrication

process itself may contribute to non-planarity. Repetitive cycles of deposition, patterning,

and etching are known to generate steps at the surface [Ste97]. Auto-focus errors,

residual lens aberrations, and resist thickness variations may also produce non-planar

surfaces during lithography [Gry93, Ste97]. Planarization has its benefits at each level of









the MLM structure [Ols93]. Planarization of the pre-metal and interlevel dielectric

improves the patternability of metal interconnects, decreases line resistance, and

increases the reliability of the device. Also, as topography is minimized before ILD and

metal deposition, the stack-up effect is reduced, which reduces the complexity of

subsequent planarization steps. Along with the benefits of dielectric planarization, metal

planarization results in an increase in step coverage, device speed and device yield, and a

reduction in manufacturing cost.























Figure 2-2 Schematic depiction of cumulative increase in topography for non-planar
MLM structures [Vie01l].

Several techniques capable of achieving some level of planarization exist. For

example, thermal flow of BPSG films is used to planarize the pre-metal dielectric layer.

After chemical vapor deposition (CVD) of the BPSG film, an anneal step is conducted to

soften and flow the film over voids or thickness non-uniformities [Ols93]. This technique

is not universally applicable because of the presence of metals with low melting









temperatures like Al. Resist etchback (REB) or reactive ion etching (RIE) uses a

sacrificial spun-on material as a planarizing agent [Ols93, Ste97]. After dielectric

deposition onto the interconnect, photoresist is spun onto the entire wafer. The smooth

surface of the resist is transferred to the underlying dielectric layer by a non-selective

blanket etchback process. The photoresist etch rate is selected to match that of the oxide

etch rate so that the surface is removed uniformly. A second layer of dielectric is then

deposited to achieve the desired dielectric thickness. This technique is undesirable

because several attempts are necessary to achieve planarization and the plasma used for

etching is composed of hazardous gases [Mur93]. Thermal flowing, reactive etching, and

all other planarization techniques have one common flaw: none can produce universal

globally planarized surfaces.

CMP Components

CMP is the only technique capable of producing global planarization. Figure 5.3

shows schematically the difference in planarization length for local and global

planarization. Generally, CMP systems consist of only 3 main components, pad, slurry,

and wafers. Currently, intense research is being conducted to understand the

fundamentals of the process.










Global Planritio V


Figure 2-3 Schematic depiction of the degrees of planarization.









Polishing pad

Until recently, CMP polishing pads have received little attention. Although the

polishing pad structure determines the polish rate and planarity of a CMP process, pad

effects and properties are poorly understood. Generally, manufacturing processes for

CMP pads are empirically derived. Because of the ability to control pad characteristics,

polyurethane impregnated felts or cast polyurethane foams with filler materials are the

most desirable materials for commercial CMP pads [Ste97, Jai94]. A hydroxyl-

terminated resin known as polyol is reacted with di-isocyanate, in the presence of a

catalyst and foaming agent to form polyurethane, a condensation polymer [Jai94]. Filler

material may be added to achieve the desired mechanical properties.

During the CMP process, the pad has 3 functions [Ste97]. Pores in the bulk

transport slurry to the pad-wafer interface, the pad asperities provide points of contact to

form the abrasive-wafer interface, and the pad also transports abraded material away

from the pad-particles-wafer interface. Appreciable removal rates are expected only

when the abrasive particles are embedded in the CMP pad and slide against the wafer

surface [Zha99]. The difficulty in engineering the performance of CMP pads is related to

ambiguity on the effect of pad properties on the CMP performance. Specific gravity,

compressibility, hardness, shear modulus, dynamic shear modulus, pad asperity height,

width, and distribution, and pad roughness control planarization efficiency. To achieve

planar surfaces, the pressure at high features should exceed the pressure at low features.

Empirical results indicate hard pads with low compressibility are more effective in

achieving planar surfaces, because they do not conform to wafer topography. However,

considerable difficulty exists in determining what pad properties should be measured to

determine pad hardness. The dynamic nature of the CMP process further complicates the









measurement. Through diffusion, water molecules break hydrogen bonds in the

polyurethane structure, which leads to pad softening (figure 5.4) [Hog73]. The pad

performance, in terms of planarity and removal rate, is dependent on the its modulus of

elasticity (G) [Ste97]. To maintain the pressure differential at high and low features, high

G values are desirable for CMP pads.



H
Case A: low water content
N-C-O0
0
S --- Hydrogen Bond
0- C-N
0

Case B: high water content
N-C-0



H H
H 0

,HzT "-H Bond breakage by water molecules
O0-C-N
0

Figure 2-4 Schematic depicition of the effect of water on the mechanical properties of
polyurethane chemistry adapted from [Hog73].

Polyurethane responds elastically and viscoelastically to shear stresses during

CMP. CMP phenomena occur in short timeframes-approximately micro to milliseconds.

Further insight into planarization behavior is possible by measuring the dynamic shear

modulus (6) under periodic stress. G increases with measurement frequency. Planarity

has also been shown to increase with pad rotation speed. At higher frequencies, the pad









has less time to deform, decreasing the pad sensitivity to wafer topography. Early

measurements of pad properties were conducted in dry conditions. More recent

measurements have shown a dependence of pad properties on water content, temperature,

polishing cycles, and slurry chemistry [Ste97, Mur94, Jai94]. Water soaking and pad

heating are known to lower the mechanical properties of the pad. Alteration of the pad

properties leads to inconsistent CMP performance. To maintain reproducible conditions,

CMP pads are conditioned between runs or during the process. In our studies, a diamond

pad conditioner was used between CMP runs. The conditioner reproduces the original

rough surface and also removes slurry particles from the previous CMP run. Conditioning

extends pad lifetime and prevents glazing-plastic wear of the pad surface.

Fixed abrasive pads-wherein the pad is coated with slurry particles-are receiving

increased attention. With fixed abrasive pads, slurry handling is no longer an issue. Fixed

abrasive pads are also less sensitive to wafer topography, which may increase the

planarization capability of the CMP process. Since slurry stability is not a concern, more

diverse slurry chemistries may be investigated. Fixed abrasive pads are potentially

applicable to STI [Gor03], Cu CMP [Ngu01], and MEMS [Kul04] fabrication processes

where over-polishing, erosion, and pattern density effects complicate the conventional

CMP process.

Wafer

Virtually all materials are candidates for planarization via CMP. Knowledge of the

chemical, structural, and mechanical properties of the surfaces to be polished is key for

designing materials-specific CMP processes. The mechanisms for dielectric and metal

CMP differ significantly. Metal CMP may be considered a repetitive corrosion/erosion

process [Ste97], wherein the slurry chemistry is tailored to form a modified surface layer









or oxide scale. This scale is then rapidly removed by repetitive abrasive contact. To

control removal rate, planarity, defectivity, and selectivity considerable latitude exists in

metal CMP processes. Several chemical additives are used in metal CMP slurries to

control the kinetics of layer formation and the mechanical properties of the surface layer

[Ste97, Sin02, Lee03]. Metal CMP slurries are more chemically active than dielectric

slurries. Generally, dielectric CMP processes are based on dissolution of the substrate or

wafer via slurry additives or abrasive particle indentation. Chemical processes in

dielectric CMP are largely due to the presence of water. For example, little polishing is

observed in oil or alcohol based slurries [Coo90] even with diamond abrasives [Ste97,

Coo90].

Few techniques exist that are capable of measuring the modified surface layer

properties in-situ. This problem is particularly glaring in dielectric CMP. However, for

metal CMP, electrochemistry and surface sensitive thin-film techniques have been used

to determine the properties of the metal surface layer. Bielmann demonstrated the active-

passive behavior in W [Bie98] by measuring the current vs. time while rubbing a sphere

against the W surface and while motionless. Also, x-ray photoelectron spectroscopy

(XPS), chrono-amperometry, potentio-dynamic measurements, and x-ray reflectivity

techniques were applied to Cu CMP system. To achieve planarization without

contributing to surface topography, the CMP process must interact with only the topmost

atomic layers. For dielectric CMP, and silica in particular, attempts to characterize

surface layer or 'gel' layer formation are conducted ex-situ. Ellipsometry [Fran99],

infrared reflectivity [Wal96], and atomic force microscopy (AFM) [AdlOO] indicate the

presence of a softened surface layer with thickness in the nanometer range.










MIi otionless

Rubbing

0.1 -

0.14



0.08

0.04 -
002
0 00 005 0.10 0.1 5 20 0.25
Tiule [s]


Figure 2-5. Current recorded for one stroke during tribological experiments on a tungsten
substrate [Bie98].

Slurry

CMP slurries consist of abrasive particles 50-500nm in diameter suspended in de-

ionized (DI) water. Abrasive particles in the slurry are responsible for mechanical

abrasion, the M in CMP. Highly reactive particles like ceria (CeO2) and zirconia (ZrO2)

may also contribute to the chemical component via particle-substrate bonding. The

abrasive particles are also responsible for pad glazing during the actual CMP process.

Residual particles must be removed between each CMP run. The abrasive particles may

be characterized by measuring a variety of properties. The particle type, hardness, size,

shape, isoelectric pH-the pH at which the surface is neutral-chemical reactivity, and

solubility in the slurry environment all affect the planarization rate, surface finish, and

defectivity of the process. The trend in slurry formulation is toward more 'gentle' CMP

processes [Sin02]. Hard abrasives like alumina (A1203), which were originally used in W

and Cu CMP are being replaced by ceria (limited cases), silica, or abrasive-free slurries

that contain little to no particles. For dielectric CMP ceria is gaining widespread

attention. Selectivity, enhanced removal rate, and improved surface finish are the factors









that have driven interest in ceria-based slurries. Oxidizers, corrosion inhibitors,

completing agents, stabilizers, dissolution agents, and surfactants are among the

chemical additives found in CMP slurries [Sin02]. Dielectric CMP slurries tend to be less

complex. Dispersants-to inhibit particle agglomeration dissolution agents-to enhance

surface solubility and surfactants-to enhance polish selectivity tend to be the only

additives in CMP slurries. Traditionally, slurries for silica CMP consisted of silica

particles.

The most intense research efforts in CMP are in the areas of slurry formulation and

particle properties. A tremendous potential for commercialization exists in CMP slurry

manufacturing. In less than ten years the CMP slurry market has grown from non-existent

to a billion dollar enterprise [Sin03]. Currently, the slurry market accounts for ~ 40% of

the worldwide nano-particle market share. These facts are wonderful for industrial

companies or the researcher who develops marketable CMP slurry. However, for the

academic researcher, the tremendous profitability has led to reluctance to release

proprietary information, which complicates efforts to understand CMP fundamentals.

What is available in the open and patent literature does not sufficiently explain CMP

mechanisms. Fundamental investigations into slurry properties are necessary, especially

in a setting where a sustained effort can be maintained. Considerable room exists in the

academic arena to contribute to the knowledge of CMP slurry development, particularly

in areas like STI and MEMS that will be introduced in the near future.

Silica-Silica CMP Mechanisms

Aside from planarization of silicon wafers, the most fundamental studies in CMP

exist in oxide CMP for back-end planarization after pre-metal and interlevel dielectric

deposition [Ste97]. CMP of silica with silica abrasives is a mature process and may be









considered a model for other dielectric CMP processes. This section will cover the

current understanding of silica-silica CMP fundamentals. Much of the knowledge in

oxide CMP is based on over a century of glass finishing for optical lens applications

[Cook90, Ste97, Kel98]. A significant amount of polishing data exists for many abrasive

types diamond and oxides of iron, cerium, chromium, silicon, zirconium, titanium, zinc,

lanthanum, nickel, and aluminum [Cor63, Izu79, Izu84, Coo90, Ste97]. Therefore, it is

worthwhile to begin this section of the review with glass polishing mechanisms.

Although these glasses are not used in IC manufacturing, the variance in their physical

properties is useful for understanding polishing mechanisms in glass, including that of

silica for ILDs in IC manufacturing [Ste97]. Oxide CMP may be considered a specialized

form of glass polishing.

Glass Polishing

The mechanisms for glass polishing may be grouped into three categories: wear,

flow, and chemical [Hol64, Isu79]. The wear mechanism suggests surface smoothing is

wholly mechanical because of increased contact at high points resulting in minimal wear

at low points on the surface. The flow mechanism suggests the glass surface flows during

polishing from peak to valley regions, until a smooth surface is achieved. Finally, glass

polishing via a chemical mechanism is attributed to increased dissolution at high surface

features because of the high pressures exerted by the abrasive particles. Izumitani

determined the glass polishing mechanism by examining the polish rate as a function of

glass hardness (bulk), softening point-a measure of a glass' resistance to flow, and

chemical durability [Izu79]. Among the three, only chemical durability influenced the

polish rate. The chemical durability of a glass is measured by the weight loss in water or

dilute acids. Weight loss occurs because heavy modifier ions (Pb2+ and La ) are leached









from the glass and replaced by lighter hydrogen ions [Ste97]. The influence of chemical

durability on polish rate indicates that chemical reactions influence the glass polishing

mechanism. Chemical modification of the glass surface alters the surface properties,

including hardness. Izumitani measured the effect of leaching time on Vickers

microhardness and polish rate versus surface hardness of glasses. Here the hardness was

shown to decrease with leaching time and the polish rate correlated well with the surface

hardness of the glass [Izu79]. Glass polishing is crucially dependent on the presence of

water [Coo90]. Polishing rates are negligible in hydrocarbon liquids like kerosene

[Izu84], paraffin [Cor63], or oil [Izu79], and non-hydroxyl containing liquids like

formamide [Cor63]. Furthermore, the polish rate in dry conditions or oil solutions

correlated with the bulk hardness of the glass [Izu79]. Surface modification is not

expected in dry or oily conditions. In these conditions, wear dominates the glass

polishing mechanism. The surface quality of glass polished in solutions other than water

or in dry conditions was poor and polish rates were significantly lower. Therefore,

optimum glass polishing occurs due to synergistic interactions between chemical and

mechanical interactions.

Silica Hydrolysis

From the preceding discussion two points should be emphasized. First, water-based

slurries are essential, both for effective planarization and desired surface finish. Second,

during CMP the abrasive particles only interact with the topmost, chemically modified

layers of the surface to be polished. In the case of silica, this modified layer results

because of hydrolysis, which proceeds as follows [Ile79].

Si02 + 2H20 -- (SiO2)x-i + Si(OH)4 (2.4)









In the hydrolysis reaction, Equation 2.4, interactions between siloxane bonds (Si-

O-Si) and water determine the behavior of silica during CMP. Budd described this

reaction, in three stages, as the attack of the solvated hydrogen ion (H30 ) on the

negative sites of silica [Bas02a, Bud61]. First, a water molecule from the environment

attaches to the Si-O-Si) bond by forming a hydrogen bond between its hydrogen and

the oxygen atom of the silica surface; lone pair orbitals of the oxygen from water interact

with the Si atoms. Second, hydrogen bound to water transfers a proton to the oxygen of

silica, while an electron is simultaneously transferred from the oxygen of water to the Si

atom. Finally, two new bonds are formed (Si-O-H), by destroying the original

siloxane bond. The dissolution of silica occurs if all four bridging oxygen bonds are

hydrated, forming the solute species Si(OH)4 (silanol) [Ile79, Coo90, Tro94]. The rate of

this reaction is controlled by the diffusion of water into the silica surface, the pH of the

polishing slurry, and the indent depth of the abrasive particle into the silica surface [Ile79,

Coo90]. The dissolution rate of silica increases rapidly above pH 9.0. Therefore,

formulating slurries in the basic pH regime enhances the removal rate and the surface

finish [Cook90]. According to Budd, the OH- ion acts a catalyst for water attack on the

silica network [Bud61]. OH- ions create excess electrons, resulting in higher negative

surface potentials, which increases the affinity for attack by H30+.

The diffusion of water is known to affect the surface hardness [Izu79, Hir87],

fatigue [Han90, Han91], and structure of silica [Tro94, Dav96, Wal96, Adl01]. However,

attempts to measure the properties, specifically the thickness and hardness, of the

chemically modified layer on silica formed during CMP have been inconclusive. Trogolo

and Rajan have reported, using transmission electron microscopy (TEM) and Fourier









transform infrared reflectivity (FTIR), the presence of two near-surface regions, 3 and 15-

20 nm below the surface after CMP [Tro94]. The 3nm surface layer possessed density

lower than the bulk value, with the 15-20 nm layer displaying an increased density with a

gradual return to bulk values ~ 20nm below the surface. However, Wallace et al., using

x-ray reflectivity, reported the opposite [Wal96]. According to their measurements, the

near-surface region after CMP is compacted, resulting in an increased density. They

further dispute the validity of a 20 nm region of increased density, because the

accompanying increase in electron density due to water diffusion was not observed.

Adler observed a short-range repulsion at the silica surface by direct measurement of

surface forces with AFM [Adl01]. This behavior was attributed to the formation of a 'gel

layer' because of swelling of the silica surface as a result of water diffusion. Since in-situ

techniques capable of measuring these phenomena are not readily available, research

efforts in oxide CMP have centered largely on particle effects. Fundamental observations

have also been reported by measuring variations in the in-situ friction force during CMP

conditions [Mah99a, Mah99b, MahOO, Cho03, Cho04a, Cho04b, Cho04c].

Silica CMP

In the last five to six years, considerable progress has been made in CMP research.

Our group in the Particle Engineering Research Center (PERC) has been at the forefront,

particularly in oxide CMP. The properties of the abrasive particles in CMP slurries have a

significant impact on the polishing behavior. Alteration of the particle size, size

distribution, and/or solids loading changes the number of abrasives available for

polishing at the pad-wafer interface. This affects the polish rate and surface finish by

modulating the load per abrasive particle. Therefore, the polishing behavior as a function










of particle size, size distribution, solids loading, and the contact characteristics at the

pad-wafer interface should be understood.

The particle size effect has been studied by a number of researchers with

conflicting results. Brown and Jairath reported an increase in polish rate with an increase

in particle size [Bro81, Jai94]. Izumitani reported the opposite, a decrease in polish rate

with an increase in particle size [Izu79]. In total disagreement, Cook and Sivaram

concluded that the polish rate was not influenced by the particle size [Coo90, Siv92].

Despite the numerous studies on the effect of particle size in oxide CMP, considerable

confusion remained. By studying the effect of particle size (0.2 atm-1.5 ktm) and solids

loading (0.2-15 wt%) simultaneously, Mahajan, a former member of the CMP group in

the PERC, was able to clarify the particle size effects [MahOO]. According to his results,

the polish rate was highest for larger particles at the lowest solids loadings. Except for 0.2

tm particles, the polish rate decreased with increasing solids loading (figure 5.5).

Furthermore, silica surfaces polished with larger particles showed significant surface

deformation, whereas, 0.2 tm abrasive particles produced smoother surfaces.

7000
---- 0 2 m
6000 0 5 -m
2 5000 1 0 P
4000 -
P 3000


1000 -
0 1

0 2 4 6 8 10 12 14 16
Solids Loading (wt %)

Figure 2-6 Material removal rate of silica as a function of the concentration of silica
slurries for different particle sizes [MahOO].









To explain the CMP behavior as a function of particle size, two mechanisms were

proposed [Bie98]. The first is a contact or surface area based mechanism, in which the

polish rate depends on the total contact area between the abrasive particles and the

surface undergoing CMP. The following expression gives the dependence of contact area

on particle diameter:

A oc Co1/3 -1/3 (2.5)

where A is the contact area, Co is the abrasive concentration, and 4 is the particle

diameter. According to Equation 2.5, surface area controls the removal rate for smaller

size particles. Also, more particles are available for polishing due to tighter packing at the

pad-wafer interface. For the second, an indentation-based mechanism, polishing occurs

due to penetration of the surface by the abrasive particles. The indent volume may be

expressed as follows:

V oc Co-1/3 4/3 (2.6)

where V is the indent volume. Other results have also indicated that the indent depth

increases with increasing abrasive particle size [Coo90, Qin04], figure 2-6.













Figure 2-7 Schematic depiction of the indentation depth as a function of particle size.

CMP processes must produce planar surfaces with minimal surface deformation.

Surface deformation or defectivity may be classified as scratches, pits, film delamination,









and/or chemical or particulate residue [Ste97]. The presence of microscratches has a

deleterious effect on microprocessor performance [Sch95]. Earlier glass polishing

experiments indicated that the surface deformation was less when slurries were

formulated with monosized particles [Coo90]. A small percentage of oversized particles

may be responsible for pitting or scratches after CMP. To determine the role of oversized

particles during polishing, Basim formulated slurries with both hard and soft

agglomerates for silica CMP [Bas02a]. The effect of hard agglomerates was studied by

adding small amounts of oversized particles to the otherwise stable, monosized, and well-

dispersed slurry [Bas00, Bas02a]. Soft agglomerates were grouped into 3 categories: 1)

dry aggregates or partially dispersed slurries; 2) polymer flocculated slurries, which were

created by adding 0.5 mg/g of polyethylene oxide (PEO); 3) coagulated slurries, which

were produced by adding, 0.2M, 0.4M, and 0.6M NaCl [Bas02a, Bas02b].

In the case of hard agglomerates [Bas00, Bas02a], oversized particles were added

to the slurry until the particle size measurement tool was able to detect their presence

(-1.0 wt%). The baseline slurry consisted of 0.14 [tm sized silica particles at 12 wt%. 0.5

[tm, 1.0 [tm, and 1.5 [tm Geltech particles were used to represent hard agglomerates. For

all slurries except those spiked with 1.5 atm particles, the removal rate was less than that

obtained using the baseline slurry. In all cases, the surface finish worsened when

oversized particles were used. Scratches and pits were observed on the surface.

Furthermore, surface deformation worsened as the concentration of coarse particles was

increased. Basim explained these observations by considering the load distribution per

particle. According to her results, coarse particles acts as pillars at the pad-wafer

interface, decreasing the opportunity for smaller, primary slurry particles to participate in









the polishing process [Bas02a]. This assertion has two consequences: first, larger

particles experience an increased load in comparison to smaller particles, and second, the

total contact area at the pad-wafer interface decreases, which is because of a decrease in

the effective number of abrasives at the pad-wafer interface. The pitting depth after CMP

increased as the coarse particle size increased, because of the increased pressure and

greater penetration depth of the larger particles [Bas02].

Hard agglomerates may be filtered as the slurry is pumped to the pad-wafer

interface. However, even after filtering, defectivity may still be high [Ewa99]. Scratches

or pits in this case may be attributed to transient or soft agglomerates formed during the

CMP process [Bas02]. Van der Waals or weak physical bonds are responsible for

aggregation in the case of dry agglomerates. Post-CMP particle size measurements

indicate agglomerated particles are broken apart during the polishing process. During

polishing, these interconnected particles experienced the majority of the load. Higher

removal rates were obtained at the expense of surface quality. The effect of PEO addition

was also studied for dispersed slurries, in which the polymer molecules are believed to

completely surround each abrasive particle and flocculated slurries in which the mean

particle size shifted to 5.82 [tm from 0.20 atm. For flocculated slurries the polish rate

slightly decreased with a significant increase in defectivity. However, for well-dispersed

polymer containing slurries, surface deformation was comparable to the baseline values,

with diminished removal rate. AFM force measurements indicated that the polymer acts

as a lubricant at the pad-wafer interface, which is responsible for the removal rate

decrease [Bas02a, Bas02b].









The effect of coagulation on slurry performance can be simulated by adding ionic

salts to the slurry. In the studies of Basim, coagulated slurries were formulated by adding

different concentrations of NaCl. Salts enhance dissolution of the silica surface [Hou92,

Dov92] and increase electrostatic interactions at the pad-wafer interface by screening

charge on the opposing surfaces [Mah99b, Bas02a]. For 0.2 M, the particle size was

unchanged. However, the removal rate increased along with the maximum pitting depth.

For 0.4 and 0.6 M solutions, irreversible coagulation ensued. The particle diameter

increased in both cases along with an increase in root mean square roughness, and

maximum pitting depth. Furthermore, the particle size before and after CMP was

identical. Meaning, the particles remain coagulated during polishing, or re-coagulate after

polishing in the post-CMP slurry.

To this point, the effect of particulate properties has been covered. The research

reviewed, although beneficial, has not completely uncovered the oxide CMP mechanism.

A precise description of how the abrasive particle interacts with the surface layer from

the micro to the nano-scale is needed. In metal CMP, surface layer formation is easily

quantifiable. The layer properties may even be tuned for the desired CMP application. In

dielectric CMP, and oxide CMP in particular, it is extremely difficult to quantify the

extent of surface modification that occurs in the CMP environment. Mechanistic

investigations have instead focused on interfacial interactions. In-situ friction force

measurements are the most widely used diagnostic tool for oxide CMP interfacial

phenomena [Mah99a, Mah99b, Sik01l, Hom03, Cho03, Cho04a, Cho04b, Cho04c].

From a micro and nano-scale viewpoint, the surface removal rate (RR) can be

estimated by the following equation [Sin02]:









RR = Kc(AxScPc)V (2.7)

where Kc is a constant, Ax is the depth of indentation by a particle, So is the fraction of the

pad surface covered by abrasive particles, Pc is the contact area of the pad on the surface

(< 1% [Bas02a, Bas03]), and V is the linear velocity difference between the particle and

the wafer. So is dependent on particle size, concentration, and the state of dispersion. Pc is

a function of the applied pressure, relative velocity, and mechanical properties of the pad,

including pad degradation. In-situ friction force measurements may be used to estimate So

and Pc [Sin02].

Using a sapphire (A1203) substrate and silica abrasives Choi et al. attempted to

measure the contact area as a function of particle size, down pressure, and particle

concentration [Cho04b]. Sapphire, a hard, inert material, was used to determine the

specifics of mechanical contact in the absence of surface roughness effects. The

normalized fractional surface coverage (Sc) is a strong function of particle size. For

smaller sized particles, the pad was almost completely covered with particles [Bas02,

Cho03, Cho04b], which corresponds to a value for So near 1. So showed a weak

dependence on concentration and no pressure or down load dependence. These

measurements confirm Basim's assertion that oversized particles act as pillars at the pad-

wafer interface [BasOO, Bas03]. By varying the particle size and particle concentration,

Choi et al. confirmed early polishing results.

Friction force measurements showed analogous transitions from high to low

friction (figure 2-7) [Cho03, Cho04c] as a function of particle size as the removal rate

(figure 2-6). For 0.2 tm particles, the friction force steadily increased with particle

concentration. However, larger sized particles 0.5 am, 1.0 am, and 1.5 am showed a










transition from high to low friction as particle concentration increased. For larger

particles, the transition occurred at lower particle concentrations. Increasing the particle

concentration results in an increase in the number of abrasives available for polishing at

the pad-wafer interface. For 0.2 [tm particles, more efficient packing is achieved as

evidenced by an increase in the fraction of the polishing pad covered by abrasive particles

[Cho03, Cho04c]. As a result, pressure at the pad-wafer interface is more uniformly

distributed. This assertion is confirmed by the reduction in the root mean square (RMS)

surface roughness as a function of particle concentration after polishing. According to

these results, 0.2 [m particles are completely embedded into the polishing pad and

remove material by sliding on the wafer surface.


S(a) (b) .- -.
z 7 w :%'J ....30 z "'-
-7 .. ..... -.
o wo .-i"'-o': -" ~" a? 4 \ '...- t



LL 2 (aI A1
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time (sec) Time (sec)
5 7
(c) o..) [(d)


(d)u%) sii



0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70

Figure 2-8 Dynamic particle motion as a function of silica particle size and concentration.
As particle size increase, the sliding to rolling transition occurs at lower solids
concentration, a) 0.2 [tm sliding motion, b) 0.5 [m sliding to rolling
transition occurs at 8 wt%, c) 1.0 [tm sliding to rolling transition occurs at 5
wt%, d) 1.5 om sliding to rolling transition occurs at 4 wt% [Cho03, Cho04b].

In the particle size range (0.5-1.5 [m), the transition from high to low friction may

be considered the boundary between predominantly sliding particle motion to









predominantly rolling motion [Cho04c]. Again the surface or contact area of the particles

may be invoked to explain these phenomena. The decrease in surface area with increasing

particle diameter results in less interaction area between particles and wafer. According

to a CMP model proposed by Zhao and Shi [Zha99], the material removal mechanism is

dependent on a threshold pressure (Pth). Pth marks the onset of sliding to rolling or vice

versa. For P < Pth, material removal rates are expected to be negligible and polished

surfaces will have a 'shot blast appearance'. Conversely, for P > Pth, the particles are

embedded in the compliant CMP pad and surfaces with scratches are produced. Using in-

situ friction force measurements, theoretical predictions of the existence of a threshold

pressure were experimentally validated.

Introduction to Shallow Trench Isolation (STI)

The ability to produce surfaces with global planarity has resulted in the application

of CMP in emerging technologies. Applications where the surface dominates device

performance (i.e. optoelectronic devices, high k dielectrics, packaging, and flat panel

displays) are candidates for CMP [Ste97]. Shallow trench isolation (STI) is critically

dependent on the planarization capability of the CMP process.

The continued march down the International Technology Roadmap for

Semiconductors (ITRS) has profound implications on materials selection, processing, and

device architecture. Specifically, the demands for smaller geometry and higher circuit

density require more effective isolation of active regions of complementary metal oxide

semiconductor circuits [Lin99]. Traditionally, local oxidation of silicon (LOCOS) has

been the method of choice for device isolation. In the LOCOS process, silicon nitride

(Si3N4), a barrier to oxidation, is selectively deposited onto the sacrificial or pad oxide.

The silicon substrate is then thermally oxidized in regions without Si3N4. After removal










of the Si3N4 by chemical etching, the thermal oxide provides electrical isolation by

separating active regions of the device. LOCOS contributes to nonplanarity and reduces

device packing density, which is unacceptable for device dimensions less than 0.5 [m

[Ste97, Lin99, Sch99].

The STI process allows dense, more functional, high-speed devices to be fabricated

at lower costs [Zha01]. Furthermore, the introduction of nonplanarity because of 'bird's

beaks' associated with LOCOS is alleviated with STI (figure 2-9). In the STI process,

trenches are etched into the Si3N4 mask layer and the Si substrate. Chemical vapor

deposition (CVD) or spin on glass (SOG) techniques are used to fill the trenches with

Si02. CMP or a combination of RIE and CMP then remove the overlying oxide. The

planarized trench-fill oxide isolates the device after removal of the Si3N4 mask. To reduce

process complexity, single-step STI CMP processes are being developed [Zha01].

Local Oxidation of Silicon (LOCOS) Shallow Trench Isolation (STI)




SiO isolated Sl

/S /1S substrate
(a) (b)
Ideal result of STI CMP step STI structure after complete
nitride and oxide removal

Si N Si N isolated
Sls isolated S region

Sl substrate Sl substrate


(c) (d)

Figure 2-9 Schematic depiction of device isolation: a) LOCOS b) STI before CMP c) STI
after CMP d) STI structure after nitride and oxide removal.

STI is not possible without CMP. An ideal STI CMP process stops at the nitride

layer and requires uniform removal of the trench-fill oxide (figure 2-9c). Naturally, STI









CMP slurries must have high selectivity to oxide with respect to the nitride, and should

leave the surface planar and defect-free. Over polishing of the nitride or erosion of the

oxide causes damage to the active region and degrades isolation performance [Lin99].

Since STI CMP is a device level process, it must be more tightly controlled [Sch99].

Defectivity in the form of scratches, foreign contaminants, and particle residue should be

suppressed. Traditional ILD consumables are not sufficient for the STI CMP process

[Hos97, Sch99]. New slurry and pad chemistries with tunable properties are required to

meet the stringent planarity, removal rate, and surface finish requirements of the STI

process. For this application, ceria-based slurries outperform silica-based slurries. Ceria

slurries preferentially remove the oxide layer with minimal polishing of the underlying

nitride. To date, the ceria-silica polishing mechanism is unclear. Currently, the ceria CMP

literature is applications oriented. In this study, ceria-silica polishing is studied to

understand the fundamental interactions that lead to material removal. With this

knowledge, the STI CMP process can be more effectively controlled.

Ceria-Silica CMP Mechanisms

The ceria based CMP literature is analogous to that of the silica based literature

approximately 5 years ago. Rigorous fundamental investigations in the manner of silica-

silica CMP experiments have not been conducted. Despite the relative immaturity of the

ceria-silica CMP process, three competing mechanisms exist. The first two, 'chemical

tooth' and a surface defect model, are based on results from glass polishing. The third,

specifically for ceria-silica CMP, is more mechanical in nature. None of the models

conclusively explain the particle-surface interactions during CMP that lead to surface

removal.









Chemical Tooth

Why do ceria abrasives polish silica so much faster than silica-based slurries? This

is the fundamental question in ceria-silica CMP. For STI applications, understanding the

origins of preferential polishing of oxide to nitride surfaces is also important. The

chemical tooth model is the most widely accepted explanation of the ceria-silica

polishing mechanism. Much has been made of the enhanced chemical reactivity of ceria

abrasives, hence the name chemical tooth. However, little experimental validation of this

concept has been presented. According to the chemical tooth model, the material removal

rate is controlled by the relative rates of [Coo90]:

1. water penetration into the surface layer
2. surface dissolution under particle load
3. adsorption of solute onto the abrasive particle surface
4. re-deposition of solute products onto surface
5. surface dissolution between particle impacts




CeO2




/TiO2



Si ZnO
Si 2


2 6 10

Isoelectric pH

Figure 2-10 Glass polish rate versus IEP of common abrasive compounds [Coo90].









Ceria slurries, by far, are the most efficient in removing the glass or pure silica

surface (figure 2-10). In fact, Cook estimated that one SiO2 molecule is removed for

every twenty-four collisions between the ceria abrasive and silica substrate, compared to

five hundred million collisions for silica abrasives [Coo90]. Among the properties of

abrasive particles (Gibbs free energy of formation, bond dissociation energy, metal-

oxygen bond strength, aqueous solubility product, Z/Rionic, Z/Reff, pK1, and isoelectric

pH), only those related to surface charge, i.e. dissociation constant (pK), effective ionic

field strength (Z-cationic charge/ReT-separation distance between the cation and the

adsorbed proton), and charge density, which is related to the coordination number

correlated with polishing rates [Coo90]. Although a direct correlation was not found, the

product of the log of the single oxide bond strength and the difference between IEP and

slurry pH offers some predictive capability on polishing activity for prospective polishing

compounds [Coo90].

The chemical tooth model depicts CMP as a molecular scale process in which

material removal is achieved by displacing silica tetrahedra from the surface into the

abrasive slurry. Net removal occurs if the solute products are bound by the abrasive or

swept away by the turbulent motion of the flowing slurry. Otherwise, re-deposition may

occur. Due to particle impact at the silica surface, water penetrates the silica network,

equation 2.4 proceeds in the forward direction-creating solute species Si(OH)4 silicicc

acid), which is released into the solution, and hydroxylated surface species are created

with one, two, or three remaining linkages to the silica network. This process is repeated

as each abrasive particle traverses the pad-wafer interface. In the case of glass polishing

and STI CMP, polishing is done near the isoelectric pH (point of zero charge) of ceria,









which is ~ 6-8. According to Cook, the extent of particle-surface bonding is controlled by

a hydrolysis reaction of the form:

Si O- + Ce OH Ce O Si + OH- (2.8).

Silica, with an IEP ~ 2-3, remains negatively charged across the pH range. Because of its

median IEP, ceria is amphoteric-able to adsorb positive and negative charges. As the

particle moves away from the abrasive site, the Ce-O-Si bond, equation 2.8, is strained

and finally ruptured. If the strength of the Ce-O > Si-O, silica accretion at the particle

surface occurs and vice versa. This bond scission and release of surface species is

expected to happen only when one Si-O bond remains of the tetrahedrally coordinated

silica [Coo90]. The silicic acid molecule may re-bond to the glass surface, form a

complex with active sites on the abrasive particle via hydroxyl exchange, or form colloids

in the solution [Coo90, Oss02]. If particle-surface bonding indeed occurs, residual Ce-O

bonding should be present on the silica surface, an important point reserved for further

discussion. The chemical tooth model is based on interactions between surface functional

groups at the particle and silica surface. Therefore, the nature of the abrasive-silica

interactions should be strongly dependent on pH.

Osseo-Asare has examined the relationship between the silica removal rate and the

IEP of the abrasive material [Oss02]. In this work, silica removal is achieved by surface

complexation of the abrasive by silicic acid. Osseo-Asare stipulated that the slurry pH

should be greater than the IEP of the silica surface and near the IEP of the abrasive

particle. Furthermore, the IEP that demonstrates maximum silicate adsorption

corresponds with the IEP for maximum polishing (ceria). The central feature of Osseo-

Asare's theoretical formulation is the exchange of hydrogen and hydroxyl ions









(protonation and de-protonation) to form neutral hydroxyl groups at the abrasive surface.

In the case of an abrasive particle with IEP < slurry pH (M-OH -- M-0- + H ), there is a

lower concentration of neutral sites. Conversely, when the abrasive particle IEP > slurry

pH (M-OH + H+ M-OH2 ), there is also a lower concentration of neutral sites and the

main reaction is the transfer of protons to the silica surface [Oss02]. Therefore, the most

favorable conditions for maximum silicate removal by ceria particles occur at neutral pH,

which is near the IEP of the ceria abrasives [Oss02].

Surface Defect

Nearly 50 years ago, a general model for glass polishing was proposed based on

lattice defects (ionic vacancies) at the abrasive surface [Kal56a, Kal56b, Kal98]. Despite

its longstanding presence in the literature, the defect-based glass polishing mechanism

has been virtually disregarded in the literature [Kal98]. The central features of this theory

are:

1. The abrasive should be softer than the surface to be polished

2. Minute bits of the abrasive are rubbed off during polishing

3. The abrasive hardness should exceed some minimum value to ensure that the
particle is not pulverized during polishing.

According to Kaller, the abrasive particles are bound to the glass surface via a

tribochemical reaction and movement of the polishing disk results in breakage of the

solid bonds [Kal91]. Glass removal is expected to be a molecular-scale process. And

similar to the chemical tooth theory, the removed glass bonds to the abrasive particles at

high-energy defect sites [Kal91, Kal98]. At the time this theory was proposed, there was

little opportunity for experimental validation because of the lack of sufficiently sensitive

analytical techniques. Forty years later, Kaller presented milling and BET surface area









data as a function of time to prove the fractionization or creation of particle debris during

the polishing process [Kal98]. In the surface-defect theory, intrinsic defects are

considered to have a minimal effect on polishing. Defects or distortions may be created in

a variety of ways: 1) reduction/oxidation reactions, i.e., Ce4+ to Ce3+ incorporation of

foreign metal ions or molecules (i.e., alloying of iron oxide with Ni2+), 3) rapid

quenching from high temperature during synthesis, and 4) freezing in of thermal

vibrations after long dwell times at high temperature [Kal98]. These defects are expected

to be responsible for the gripping or bonding of the abrasive particle to the surface.

Furthermore, the defect density and hence the particle reactivity are expected to increase

as the particle is abraded due to the exposure of freshly created, more reactive sites

[Kal91, Kal98].

Particulate Generation

The first two ceria-silica mechanisms are general models for all metal oxide

abrasive particles for glass polishing. The accelerated removal rates for ceria with respect

to silica have been attributed to its median IEP [Cook90, Oss02] or multiple valence

states [Kal91, Kal98]. The final model is based on more mechanical interactions and is

proposed only for ceria-silica polishing as an alternative to the chemical tooth theory

[Hos01]. In this model, silica particulates are cleaved from the surface during wear and

bound by the abrasive particle. In support of this assertion, the particle size and

composition of the post-CMP slurry was analyzed. Hoshino observed an increase in

particle size from the original 80 nm to 600 nm lumps of a white paste-like substance.

Inductively coupled plasma atomic emission spectroscopy, scanning electron microscopy,

infrared spectroscopy, and x-ray microanalysis were conducted. According to their

analysis, the large, irregularly shaped, white paste was SiO2 cleaved from the surface. In









their opinion, if the chemical tooth mechanism was operative then silicic acid should be

isolated in the post-CMP slurry. Since lumps of SiO2 were present in the slurry they

assumed they must have been directly cleaved from the surface.

The interpretation of Hoshino et al. is complicated by the presence of an organic

surfactant, the slurry pH pH 8, and the dynamics of colloidal silica stability in this pH

regime. The presence of the surfactant likely complicates the analysis of the ceria-silica

polishing mechanism, since surfactants have been demonstrated to change the

characteristics of contact at the pad-particles-wafer interface [Bas02a, Sin02]. Also, at ~

pH 5 colloidal silica rapidly aggregates, with particle growth ensuing at ~ pH 7 [Ile79].

Recalling the formulation of Cook, in the post-CMP dispersion; the solute species,

Si(OH)4, is free to form a complex with the abrasive particle, which could change the

shape of the original ceria particle or coalesce and form colloids, which may explain the

growth of the silica particle.

To this point, three models with diverging mechanisms have been proposed.

However, each has a central feature ceria-silica bonding that is believed to occur upon

particle-wafer impact. Therefore, the surface chemistry of ceria should be considered in

all proposed mechanisms for ceria-silica CMP.

Ceria Surface Chemistry

Cerium, atomic number 58, is one of the rare earth elements in the lanthanide

series, which are characterized by the filling of 4f-orbitals [Sam73]. The chemistry of all

lanthanides is similar. With the exception of cerium, the lanthanides have a valence of 3.

However, cerium may also have a valence of 4, forming two oxides-Ce203 (cerrous

oxide) and CeO2 (ceric oxide). Ceria is an extremely efficient polisher of glass. A fact

that has been ascribed to enhanced chemical reactivity between ceria and silica even in









the absence of chemical additives to the liquid portion of the slurry. Ceria has a fluorite

crystal structure, with the oxygen atoms coordinated to four cerium atoms and the cerium

atoms coordinated to eight oxygen atoms. The cerium atoms assume face-centered cubic

(FCC) crystal symmetry with oxygen inserted into the eight tetrahedral spaces within the

FCC unit cell (figure 2-11) [Eva04].




*'- .1 o






Figure 2-11 Ceria crystal structure normal view (left), displaced by one-half a lattice
parameter (right) [Eva04].

The size of abrasive particles has continually decreased as more specialized CMP

processes are introduced. Because of the high surface area of smaller particles, surface

chemical effects are important for all CMP applications. This is especially true for such

reactive abrasives as ceria. The transition to nanometer scale particles from the sub-

micron regime has profound implications on the CMP process from slurry formulation,

handling, and transport, to actual polishing, and finally post-CMP cleans. A thorough

understanding of the abrasive surface chemistry allows tailoring of the surfaces for

enhanced dispersion before use, reactivity and desired polishing during use, and efficient

removal after use. Stable micron-sized dispersions may be obtained by simple

manipulation of the pH. However, in the sub-micron to nanometer scales, dispersion

agents like surfactants, polymers, or excess ions (electrolytes) are required [Nab93].

Nanosized particles, in particular, cannot be kept apart from each other through









electrostatic repulsions, because the interaction potentials are less than 15kBT

(Boltzmann's constant temperature) [Nab93, Ver48], ~8KBT for ceria [Ver48].

Therefore, in water nano- particles will experience multiple collisions, which should

result in aggregation due to particle-particle bonding [Nab93]. This is especially true for

ceria, which demonstrates three states of stability based on the chemical treatment

sequence [Nab93]. The loss of stability, i.e. agglomeration, has negative impacts on the

particle's polishing capacity because of the reduction in surface area and loss of active

sites.

Nabavi et al. have investigated the surface chemistry of ceria [Nab93]. Applying

Pauling's definition of formal valence charge, the multi-site complexation or 'MUSIC'

model was used to determine the acid-base behavior of surface hydroxyls [Nab93,

Hie89]. In this treatment, atoms in an ionic crystal give each of its neighbors a number of

electrons, v = Z/CN where z is the atomic number of the atom and CN equals the number

of nearest neighbors. For n metals, a charge of Z = nv results. Its nearest neighbors in a

bulk crystal compensate this charge. A hypothetical free surface may be created by

splitting the crystal along a plane of oxygen atoms, creating dangling bonds. Surface

passivation is achieved by adsorption of aqueous species (H OH-)-general protonation

or de-protonation equilibria. Through such reactions an electric charge is acquired at the

particle-liquid interface.

Additional complications arise when additional ions are present. This process is

driven by the surface energy of the oxide in water. Ion adsorption may be hindered by

steric repulsion or steric poisoning, which lowers the surface energy by binding of ions

from the solution. In addition to chemical reactions at the surface, charge may be








acquired by isomorphous substation within the lattice (i.e., Si4+ replaced by Al3+) and

adsorption of a hydrophobic species or surfactant ions [Stu92]. Thus the electrical state of

a surface depends on the spatial distribution of electronic or ionic charges in the near-

surface region. Using the Gouy-Chapman model this concept is idealized as the electrical

double layer. In this model, one layer is depicted as a region of fixed charge rigidly

bound to the surface while the other layer is distributed more diffusively in the liquid in

contact [Stu92]. The potential drop across the mobile portion of the double layer (figure

2-12), zeta potential (Q, is responsible for the electrokinetic phenomena, particle motion

in the presence of an electric field, which may be related to dispersion stability.




LP0

Shear plane













Distance (x)

Figure 2-12 Potential versus distance for a charged surface [Stu73].

The surface chemistry of ceria and its effect on dispersion stability is particularly

intriguing. In the work ofNabavi et al., nanosized ceria particles were obtained through

precipitation of Ce4+(NO3-)4 salt of the form CeO2(HNO3)o.5(H20)4 at very low pH









[Nab93]. Simply raising the pH caused irreversible aggregation and transformation of the

particle surfaces well below the predicted point of zero charge (pzc). Three dispersion

states were obtained based on the affinity for proton (H ) adsorption and also binding and

release of nitrates present from the nitric acid based precursor solution. The acid-base

equilibria at the ceria particle surfaces were investigated through titration of the surface

groups with a base (NaOH) or acid (HC104). The pH was measured and he concentration

of free nitrate ions (NO3-) at each step of the titration was also measured.

In Nabavi's work, the behavior of the dispersion was related to the difference in

hydroxyl and proton consumption in dispersions of acidic and basic pH and the presence

or lack of nitrates as counter-ions or covalently bound surface species [Nab93].

According to their results, titrating the dispersion above pH 5.0 results in a loss of active

sites that is reflected in an excess amount of hydroxyl consumption with respect to proton

consumption in the reverse cycle. The transition between different surface states occurred

in narrow pH ranges, generating a hysteresis-type curve for pH versus volume of acid or

base used (figure 2-13). The horizontal distance between the curves corresponds to an

excess consumption of hydroxyl during the first part of cycle [Nab03]. According to their

results, these phenomena may be explained as follows. The original dispersion contains a

high concentration of nitrate ions as either counter-ions or strongly bound surface

species. The counter-ions are released because charge is cancelled by a progressive

release of protons and adsorption of hydroxyls. Covalently bound nitrates were released

up to pH 6.0 in an exchange reaction of the form:

Ce-O-NO2 + OHf -- Ce-OH + NO3- (2.9).









Particle aggregation ensues, which results in a loss of surface. The remaining surface

sites determine the electrical charge characteristics of the dispersion. This is reflected by

a shift in the pzc from the theoretically predicted value of 7.9 to 10.0. Particle

aggregation results in flocculation of the dispersion. The dispersion stability of

nanoparticles ceria is thus related to the presence of strongly bound ionic surface species,

which influences the ability to adsorb aqueous species from the liquid environment.






7




5


3 --

0.5 1.5 2.5 3.5

V [Base/Acid] (ml)

Figure 2-13 pH versus acid/base concentration. Horizontal distance between curves is
related to excess consumption of OH- during acid to base cycle [Nab93].

These observations have wide-ranging implications for the various technological

applications of ceria particles. For CMP in particular, irreversible changes in the particle

dispersability, reactivity, and surface area should be reflected in the polishing behavior as

a function of pH. Three major questions form the basis of this research.

4. Why does ceria polish so much faster than other abrasives especially silica?

5. Why does ceria preferentially polish oxides over other materials?









6. How can the polishing characteristics of ceria slurries be controlled for optimum
STI polishing?

There are major gaps in the literature concerning ceria abrasives for silica CMP. As

previously mentioned, too few studies have focused on the fundamental interactions

between the ceria particle surface and the surface of the silica substrate. Most studies

complicate mechanistic studies by including surfactants into the ceria-based slurries.

However, recent publications indicate the focus is shifting towards more fundamental

investigations. The surface structure of individual ceria particles has been studied by

electron energy loss spectroscopy (EELS) [Gil04]. These measurements have confirmed

the presence of Ce3+ at the surface of bulk CeO2 particles. In this dissertation, ceria-silica

interactions will be investigated, specifically surface interactions. To confirm the

existence or absence of chemical reactions between the ceria particles and silica

substrates extensive surface analysis has also been conducted. The major application of

this research is in STI CMP. Therefore, general strategies for STI CMP are also

proposed.














CHAPTER 3
MATERIALS AND METHODS

Introduction

This research involves polishing experiments using ceria-based slurries and a silica

film. In-depth analysis of the effect of pH on ceria surface chemistry, silica removal rate,

ceria-silica contact characteristics, ceria particle morphology, and silica surface

morphology is discussed. The in-situ friction force is reported using a novel apparatus

that was fitted to our tabletop polisher. Slurry stability, particle size, and zeta potential

measurements are reported using a variety of light scattering instruments. Surface

analysis using x-ray photoelectron spectroscopy, scanning electron microscopy, atomic

force microscopy, and variable angle spectroscopic ellipsometry results are presented.

Samples

Silica Wafers

Silica (SiO2) samples for all experiments had a nominal thickness of 2 u[m. Plasma

enhanced chemical vapor deposition (PECVD) was used to the deposit the oxide layers

on 1 mm thick, p-type silicon substrates with (001) orientation. The silica wafers were

obtained from Silicon Quest International.

Ceria Slurries

Ceria particles were obtained from Samsung/Corning. The slurries were dispersed

in DI water at pH 4.5. KOH or HCL were used to adjust the pH to the desired value. The

concentration of the particles in suspension was between 2 and 3wt%.









Ceria Thin Films

Ceria thin films were grown on Si substrates using pulsed laser deposition (PLD).

The PLD ceria thin films were used to measure the force measurements with a 7.3 [im

silica particle. The deposition was conducted at 7000 C in 100 mtorr of 02. A KrF laser of

wavelength 248 nm was used. The laser energy was 500 mJ and the pulse rate was 5 Hz.

During PLD, a pulsed laser strikes a solid target. Target materials are removed in the

form of a plume. Substrates of choice may be coated by exposure to the plume, which is

created by laser bombardment. The major advantages of PLD include; film stoichiometry

close to that of the target, low contamination levels in deposited films, and high

deposition rates.

Functionalized Cantilevers

Cantilever oxidation

Silicon tapping mode atomic force microscopy (AFM) cantilevers were

functionalized by thermal oxidation at 1000 C for 2 hours in air. The cantilevers were

furnace cooled. After oxidation, the cantilevers were used to abrade the ceria thin films at

different pH values. The wear profile and cantilever composition was measured before

and after the experiment with a scanning electron microscope.

Particle attachment

AFM cantilevers can be further functionalized by attaching a particle for force

measurements. A micromanipulator fitted onto an optical microscope was used to attach

a 7.3 lim silica to a silicon nitride contact mode AFM tip. The spring constant of the

cantilever was 0.13.









CMP Equipment

CMP Tabletop Polishers

Polishing experiments were performed on a Struers Rotopol 31 and TegraPol 35

tabletop polishers. The TegraPol 35 polisher allows control of platen velocity and

pressure down to 2 psi. The ceria polishing slurry is delivered by a Struers Multidoser

unit that is connected to the polisher. The head and platen speed were 150 rpm and the

center of the sample to be polished was placed at a distance of 8.9 cm from the center of

the table. This setup leads to a linear velocity of 137.44cm/s. Perforated IC 1000/Suba IV

(figure 3-1) stacked pads by Rodel Inc. and Rohm and Haas were used for all polishing

tests. During polishing, the slurry is continuously stirred by a magnetic bar, and pumped

to the pad-wafer interface at a flow rate of 100 mL/min by the Multidoser unit. The pad

was conditioned before each polishing run with a grid-abrade diamond pad conditioner

manufactured by TBW ". The diamond conditioner minimizes pad glazing (figure 3-1d)

and ensures reproducible pad conditions for each polish run. The sample holder for the

polishing experiments is a stainless steel cylinder 2.25 inches in diameter and 1.125

inches in height. The flat surface of the cylinder has a recessed, machined area of 1 inch2

and 0.03 inches deep. The recessed area ensures the sample holder holds the wafer during

polishing experiments. A porous carrier film supplied by Rodel slightly raises the wafer










IC-1000 layer -


Bonding epoxy layer


SubaIV layer R 119 m
(a) 1.24 mm (b)- -








(c) (a)

Figure 3-1 SEM analysis of CMP pad: a) cross-section of IC 1000/Suba IV stacked
polishing pad (40 x) [Obe98] b) Plane view image with profilometer scan
showing pad roughness [Cho03] c) image of fresh pad [Ste97] d) image of
used pad [Ste97].

over the recessed area. Additionally, when wet, the film ensures the wafer remains

stationary during the experiment.

Friction Force Measurement Apparatus

The friction force studies were conducted using the tabletop Struers Rotopol 31

polisher. An apparatus, developed by us, was fitted to the polisher (figure 3-2). Glue was

used to mount the silica samples (0.8 in2) on a rectangular aluminum block. Circular

washers were then loaded onto the block to achieve the desired down force. The entire

assembly is then placed in a restraining fixture that ensures movement occurs only in one

direction. During the experiments, the lateral force applied by the aluminum block on the

restraining fixture is measured by a force transducer (Sensotec model 31 load cell). The

load cell is connected to a data acquisition system. A data point was taken every 250 ms.










Down Force
Force
Transii ucer
VO Sample Restrainin_ [eier
Holder Fixture

/ Wafer

Wafer Force/ i gPW
Transducer Polishing Pad
PROFILE PLAN VIEW


Figure 3-2 Schematic of in-situ friction force apparatus.

Particle Characterization

The particle size and surface charge are two properties that have a profound

influence on the slurry stability and resultant polishing characteristics. Multiple light

scattering techniques were used to measure the particle size and zeta potential, and to

determine the stability of the ceria slurries.

Zeta Potential Measurement

Almost all solids acquire a surface charge in polar liquids like water. The various

ways of charge accumulation have been covered in Chapter 2. The surface charge or

surface charge density plays a crucial role in slurry stability. This is especially true for

nano-sized particles. Because of dynamic motion and continuous particle-particle

collisions, attractive forces are always present. These forces originate from the

momentary, induced dipoles that arise when particles approach each other [Zeta].

London, van der Waals, and/or dispersion forces are common terms for these short-range

attractive forces. To maintain slurry stability, electrostatic and/or steric repulsive forces









can be applied. Because of their surface charge, particles move with an average velocity

(V) in an electric field (E), which is referred to as electrophoresis.

A Zeta potential analyzer manufactured by Brookhaven Instruments was used for

zeta potential measurements as a function of ceria slurry pH. There is not an instrument

that is capable of directly measuring the zeta potential. Instead, the velocity of the

moving particles is measured. The electrophoretic mobility ([te) is then calculated as a

function of velocity (V) and electric field (E).

S= V/E (3.1)

The particle, adsorbed ions, counter-ions, solvent molecules bound to ions on particle

and/or counterions, and surfactants or polymers if any exist, may be considered a single

unit. The velocity is measured experimentally as the charged particle is drawn to an

oppositely charged electrode. The zeta potential is then calculated from the mobility.

The concept of surface charge density leads to a simplified model referred to as the

electric double layer. In this approach, the particle surface is populated by counter-ions

with opposite charge of the particle and co-ions that possess the same charge as that on

the particle surface. Thermally driven collisions between particles and the solvent

molecules cause continuous agitation of the adsorbed ions. This results in an electrostatic

potential (Y) that decays exponentially for planes separated from the particle surface

(figure 3-2).

T = Toe-ex (3.2)

The electric double layer thickness (K-1), which is on the order of nanometers, determines

the distance from the particle surface over which various electrical potentials are

significant [Stu73]. This thickness is a function of temperature, fluid dielectric constant,









and the ionic strength or free ion concentration. For high ionic strength conditions, the

electrical double layer is compressed and the particle-particle interactions are more

attractive.

Thus the zeta potential is the potential between the bulk of the liquid and the

surface of shear or shear plane that extends out from the surface of the particle. This

potential is a direct measure of the repulsive force between two charged bodies.

However, the zeta potential and mobility depends on the model chosen. For a particle of

radius a, the Huckel limit is applied when Ka << 1. It is practically impossible to satisfy

this limit for most particle systems. Therefore, the zeta potential analyzer utilizes the

Smoluchowski limit for Ka >> 1, where zeta potential () is proportional to the product of

the fluid viscosity (f) and the mobility divided by the dielectric constant (F) of the fluid

[Hun01].

= lte/s (3.3)

Particles move in a direction perpendicular to the focused laser of wavelength 670 nm

(red). The Doppler shift due to the scattering of the incident beam is determined and

mobility and zeta potential are calculated as described above.

Particle Size Analysis

A Microtrac UPA 150 particle size analyzer manufactured by Honeywell was used

to determine the average particle diameter as a function of pH. Particles were sonicated in

solutions of pH 2.5-10.5. The background solution used in the particle size analyzer was

of the same pH as the ceria slurry. This instrument also utilizes dynamic light scattering

for measurement purposes. Again, the velocity and particle motion direction are

randomized by collisions with solvent molecules. The velocity distribution of a large

number of mono-sized particles will approach a well known functional form. After









compensating for thermal and viscosity effects, the velocity distribution becomes a

unique function of the particle diameter.

The scattered light by moving particles is proportional to the particle velocity. A

dispersion of particles will have a unique distribution of frequency shifts. Spectra from

small particles have more high frequency components than spectra from large particles.

Care must be taken to minimize thermal and fluid viscosity contributions. Both increased

temperatures and decreased fluid viscosities will increase the high frequency components

of the spectrum causing an underestimation of the particle size. The particle distribution

is computed directly from the measured frequency spectrum obtained from the Doppler

shifted scattered light.

Slurry Stability

Slurry stability is an anomalous property. CMP slurries must not only be stable at

the point of use, but also must maintain stability during shipping and while on the shelf.

In this work, a Hach model 2100AN laboratory turbidimeter is used to determine the

stability of the slurries at different pH values. This instrument measures turbidity from 0-

10,000 Nephelometric Turbidity Units (NTU). A tungsten filament lamp is used to

provide the incident beam. Lenses and apertures focus the light. Four detectors are used:

90 degree, forward-scatter, transmitted, and back-scattered light detectors. In the Ratio

On mode, mathematical calculations are used to average signals from each detector for

improved linearity, calibration stability, and wider measuring range, and to perform

measurements of colored solutions.

A nephelometer is an instrument that measures the amount of cloudiness in a

dispersion. The intensity of scattered light changes as the particles settle out of solution.









High NTU values correspond to dispersed or cloudy solutions. Measurements with this

instrument are presented as normalized NTU.

Characterization of Silica Surface Morphology

Atomic Force Microscopy (AFM)

Roughness is a surface property that affects device performance. Even well

polished surfaces have topographical variations that can be considered surface roughness.

Atomic force microscopy (AFM), a hybrid of scanning force microscopy allows imaging

of surface topographical variations with sub-nanometer resolution [Ste92]. In this study,

the topography after CMP was compared to as received samples using AFM. AFM,

although time consuming, is a robust technique that allows characterization of insulating

and conducting surfaces in liquids or in air.

AFM uses a sharp tip mounted on a flexible cantilever. Repulsive van der Waals

forces between atoms on the tip and substrate cause deflection of the cantilever. The

force between the tip and sample are a function of the separation distance. Therefore,

surface topography is obtained by monitoring the cantilever deflection. Raster scanning

during AFM is controlled by a piezoelectric tube, which maintains a constant separation

between the tip and sample. Cantilever deflection due to surface topography is measured

using light from a laser diode that reflects from the back of the cantilever to a position-

sensitive photodiode (figure 3-3). The AFM image is generated by monitoring the

position of the scanner.










Position Sensitive Diode Laser
Photo Diode (PSPD)






Probe tip Candtever
roe _- ._/ (18D [m long)
Sample (1VL nk)


PZT Piezoelectric
y / Tube


Figure 3-3 Schematic diagram of AFM apparatus.

In this study, a Digital Instruments Nanoscope III AFM apparatus was used. The

measurements were conducted in contact mode. Contact mode tips have four Si3N4

cantilevers with spring constants that vary from 0.06 to 0.58 N/m. For topographical

imaging, 10 ptm scan sizes were used to quantify the extent of scratching and surface

damage after CMP. The software bundled with the AFM calculates Root Mean Square

(RMS), average (Ra), and peak-to-valley (Rpv) roughness values. The RMS roughness is

used in this study.

RMS = ((Yi2/N))1/2 (3.4)

Ra= ZYi/N (3.5)

Rpv = Rmax Rmin (3.6)

In all three equations, Y is the height value and N corresponds to the total number

of measurements. For all scans, a plane fit and flatten data filter were applied to the AFM









images using the AFM software. With these filters, scanning artifacts can be removed

and the image quality is improved for display purposes.

Force measurements are also possible using the AFM. They may be conducted with

the standard AFM tip or with cantilevers that have been functionalized by attaching a

micron-sized particle. In this study, a 7.3 [m diameter silica particle was used to measure

the interaction force as a function of separation distance at various pH values using a

ceria thin film. Further functionalization is possible using specialty coatings. To further

study the ceria-silica mechanisms in the absence of particles, chemical vapor deposition

was used to coat silicon tapping mode cantilevers with a 100 nm thick silica layer. Tip

wear against a ceria surface was compared at different pH values.

Scanning Electron Microscopy

Scanning electron microscopy (SEM) provides a highly magnified image of the

surface of a material that is similar to what one would expect if one could actually see the

surface. SEM provides both topographical and elemental composition. Resolution down

to a few nanometers is possible with magnification of 300,OOOX.

During SEM analysis, electrons are flooded onto the specimen surface with a

variety of output signals. Electrons with energies of- 50 eV are called secondary

electrons [Bin92]. Secondary electrons result from inelastic scattering between the

primary electron beam and atomic electrons of the specimen [Bin92]. The surface

sensitivity of SEM is due to the shallow escape depth of secondary electrons. Inelastic

scattering in the specimen bulk results in electrons with insufficient energy to escape

from depths greater than a few nanometers. Backscattered electrons, including

characteristic Auger electrons, have energies in excess of 50 eV [Bin92]. Backscattered

electrons result from elastic scattering of the primary electron beam. A further utility of









the backscattered electron signal is composition mapping. The backscattered electron

intensity increases with atomic number, yielding brighter images [Bin92]. Characteristic

x-rays are also emitted as an excited atom decays to its ground state. The x-ray emission

single can be sorted by energy or wavelength in an energy dispersive or wavelength

dispersive x-ray detector. The distribution of energies or wavelengths is characteristic of

the elements that produce them.

There are three types of emission capable of producing beams for SEM analysis:

thermionic, LaB6, and field emission. Field emission provides the brightest emitted beam

with the sharpest electron energies, which limits chromatic aberrations of magnetic

defocusing lens [Bin92]. Field emission SEMs offer enhanced resolution at lower applied

voltages. In this study, a SEM equipped with a field emission gun (JEOL JSM 6335F)

will be utilized along with the energy dispersive x-ray detector (EDX) for compositional

analysis of the polished surfaces.

SEM analysis can be limited by the sample properties. To minimize charging,

insulating samples must be imaged at voltages below 2 keV or a thin (< 10 nm)

conductive coating must be applied. Also, the specimen should be stable at high vacuum

and be of appropriate size for the sample holder.

X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS)is based on the photoelectric effect

explained by Einstein in 1905 [Bru92]. High-energy x-rays ionize an atom producing free

photoelectrons.

KE = h) BE (3.7)

Since the energy of the incoming photons is known, by measuring the kinetic

energy (KE) of the ejected electron, the binding energy (BE) of the electron to the









particular atom can be determined. The binding energy of the electron is sensitive to the

chemical state of the atom. Therefore, XPS is a powerful surface analysis tool that

provides semi-quantitative elemental composition information. Since XPS peaks are due

to the ionization of core level electrons, they can be used to detect all elements except

hydrogen and helium. Liquids, gases, and solids can be analyzed. Generally, XPS spectra

give information from the topmost atomic layers (2-20) [Bru92].

The number of peaks in an XPS spectrum corresponds to the number of occupied

energy levels in the ionized atoms whose binding energy is lower than the energy (hu) of

the incident radiation. The peak position determines the respective binding energies of

the electrons in subatomic orbitals and provides a means of identifying atoms present in

the specimen. Additional features in the spectrum are due to the involvement of passive

electrons (non-photoelectrons) [Bru92].

Unlike valence electrons, core level electrons do not participate in chemical

bonding. Therefore, the elemental binding energies have characteristic values that

provide unique signatures of elements. Quantitative analysis (at%) of the relative

amounts of elements present is possible by comparing the relative peak intensities. This

requires knowledge of the photoionization cross section for each element and control of

experimental artifacts that affect peak intensities [Bru92]. XPS provides only semi-

quantitative elemental composition information because of: variations in peak widths and

peak line shapes, incorrect background subtraction, and instrumental artifacts. Coupled,

these factors introduce an uncertainty in the measurement that may be as high as 30%.

However, for well characterized XPS peaks with well known standards the uncertainty

may be reduced to only a few at. %.









In addition to quantitative analysis, the chemical environment of the ejected

photoelectrons may be determined as well. Slight shifts in the binding energy of XPS

peaks may be related to the respective chemical environment. In general, differences in

the binding energy are seldom more than a few electron volts. Other features of the XPS

spectrum may also be utilized to identify elements in the specimen. Shake-up satellites

are created because valence electrons may be excited to empty orbital levels during

relaxation. This results in photoelectrons of diminished energy, hence, a shift to higher

binding energy. Since valence electrons participate in chemical bonding, additional

chemical state information is obtained. Auger electrons are also generated when an

electron from a higher energy level fills the core hole left by the ejected photoelectron.

By changing the radiation source, overlapping XPS and Auger peaks may be resolved.

The primary use of XPS is for surface analysis. Electrons travel short distances

before losing energy in collisions with atoms of the specimen. Therefore, only near

surface photoelectrons contribute to XPS peaks. The background in a XPS spectrum may

be attributed to electrons from the bulk that have lost energy due to inelastic collisions.

Additional surface sensitivity is possible by utilizing a grazing incidence angle, which

further minimizes the escape depth of elastically scattered electrons. The XPS system

used in these experiments is a Perkin-Elmer PHI 5100 manufactured by Physical

Electronics.

Spectroscopic Ellipsometry (VASE)

Spectroscopic ellipsometry is a surface sensitive thin-film measurement technique

that measures changes in the polarization state of light after reflection or transmission

from a surface of interest. Large amounts of data may be obtained using probe beams of

various wavelengths and angles of incidence. A broad range of sample materials and









structures are suitable for this technique. As with all optical techniques the property of

interest (in this case film thickness) is not measured directly.

Ellipsometers measure the change in the polarization state (Y and A) of the light

reflected from the specimen surface. The parameters may be related to the Fresnel

reflection coefficients (Rp, Rs).

p = Rp/Rs = tan (Y) eiA (3.8)

Ellipsometry is a very precise and reproducible technique since it measures the ratio of

two values. The sensitivity of the measurement is enhanced by the complex component of

equation 3.6 that gives phase information.

Initially, the ellipsometer measures the reflected or transmitted intensity or the

polarization state of the reflected light. A model is developed based on the probe

wavelength, polarization state, angle of incidence, thickness, and optical constants. Next,

the model is fitted to the experimental data by varying physical parameters and

calculating data that closely matches the measured optical data. The ellipsometer in our

lab, manufactured by J. A. Woollam Co. Inc., uses the Levenberg-Marquardt multivariate

regression algorithm. Finally, the best-fit model should be evaluated to ensure the fit is

unique and is physically reasonable. This is achieved by monitoring the MSE value. For a

unique, reasonable fit the MSE value tends towards some unique minimum value that

approaches zero.

Although ellipsometry is a robust technique that requires very little sample

preparation, measurements should be made with caution. The roughness of the measured

spot of the specimen should be less than 10% of the beam's wavelength. Also, the

thickness of the film should vary no more than 10% over the width of the beam spot. The






60


presence of adsorbed contamination layers may slightly affect the measurement.

Angstrom level measurements should also be made with care since the optical constants

of thin films rarely are identical to bulk values.














CHAPTER 4
THE EFFECT OF PRESSURE, PH, AND VELOCITY ON THE SILICA REMOVAL
RATE AND SURFACE FINISH: A STATISTICAL DESIGN OF EXPERIMENTS

Introduction

Chemical Mechanical Polishing (CMP) began as a proprietary process at IBM

[BeyOO]. Initially used for planarization of silicon, the CMP process has been

successfully applied to all steps in integrated circuit manufacturing. In fact, CMP is an

enabling technology for multilevel, interconnected structures that have allowed the

continued shrinkage of microelectronic devices [Ste97]. Because of its ability to produce

globally planar surfaces, CMP is being implemented in a number of emerging

technologies. Although CMP is simple to perform, designing an optimum process for a

new set of materials is not a trivial task. Often, conventional CMP consumables (pads,

slurries, equipment) are not applicable to the new application. Furthermore, the use of

CMP in new applications is complicated by the lack of fundamental understanding of

general CMP processes that have been in use for well over a decade. Hence, conventional

CMP processes can not be strictly relied upon when developing strategies for novel CMP

applications.

Lacking a defined starting point for new CMP applications, fundamental

investigations of the nature of those reviewed in Chapter 2 for silica-silica CMP are often

necessary to develop an efficient, reliable CMP process. Singh et al have identified over

forty variables micro- and nano-scale effects, and output parameters that control the CMP

process [Sin02]. Common CMP experiments investigate the effects of only a few









variables at a time by varying one parameter while all others remain fixed. After an

optimum response is received at one level, another parameter is varied and so on until the

desired polish rate, surface finish, planarity, and/or selectivity are achieved. A thorough

investigation of the influence of each variable and its effect on the respective output

parameters is impossible using the one variable at a time experimental approach.

Statistically designed experiments are an attractive alternative to testing one variable at a

time.

Statistical methods are crucial tools for industrial CMP practitioners [Par04]. The

competitiveness of the microelectronics fabrication industry requires that resources be

used efficiently, that processes are reproducible with minimal lot-to-lot variation, and that

manufacturing delays be minimized to ensure maximum profit. Statistically designed

experiments based on sound scientific and engineering principles allow companies like

chip manufacturers to continually increase production at lower costs, resulting in higher

profits. In laboratory testing, statistical methods are usually applied only in data analysis

(i.e. measurement average and standard deviation from measurement average). CMP

research, with its many variables and inter-related interactions, is a prime area for

implementation of statistically designed experiments.

The basic principle of statistically designed experiments is that all measurements

are subject to variation [Mas89]. Variations in measurements may be because of

experimenter bias, changes in the experimental ambient, or measurement error. Failure to

minimize error can often result in the wrong interpretation of data. For example, if an

output parameter has a large variation about the measurement mean it is not clear that the

observations are significant, which results in the obscuring of the true effects of the









variables in question. Statistical experimental designs are effective in eliminating bias,

ensuring the measured output is gained from the experiment, and conservation of

resources. Each observation generally provides information on all the factors of interest

in the experiment. Conventional laboratory procedures limit the number of variables that

may be studied. Also, since only one variable is changed at a time, the evaluation of the

effect of interactions among experimental variables is not possible. Statistical designs

allow for the estimation of the magnitude of uncontrollable variation, and modeling of

relationships between measurements of interest and variables believed to influence these

measurements [Mas89].

Selecting a Statistical Design

Mason, Gunst, and Hess identified five statistical design criteria: 1) consideration

of objectives, 2) factor effects [Appendix A], 3) precision, 4) efficiency and 5)

randomization [Mas89]. While formulating the objectives of the experiment the nature of

the anticipated conclusions, the manner in which concepts are defined, and the manner in

which observations are measured and reported should be considered. For factor effects,

systematic errors should be eliminated, covariates should be measured [Appendix A],

variable relationships should be identified, and the entire experimental range should be

explored to ensure a meaningful experimental design is produced. Precise experiments

include an estimation of inherent uncertainty associated with the experiment, replication

of experimental runs, adjustment for covariates, and blocking [Appendix A]. Precise

experimental designs also ensure that the detail of the response and its relationship to the

factors is clear. Efficient experiments are characterized by the examination of multiple

factors with screening designs. Finally, the experiments should be randomized.









Randomization protects against unknown sources of bias like instrumental drift and helps

validate assumptions required to apply certain statistical techniques [Mas89].

Statistically based experimentation is widely practiced in various fields. Basim et al

investigated the effect of particle size, applied load, and particle concentration on the

silica-silica CMP removal rate and surface finish using a central composite design with

three factors at five levels [Bas03b]. Park used three designs: a nested design, a

completely randomized block design, and a factorial design to analyze gate polysilicon

critical dimensions for integrated circuit devices [Par03]. Martin et al used a fractional

factorial design to optimize cryo-preservation conditions for rat liver slices and to

evaluate the day-to-day variation in the levels of cryo-preservation [Mar03]. Tye has

reviewed the applications of statistical design of experiments in drug discovery [Tye04].

Finally, Zaman et al. used the Box-Behnken method to investigate the solubility of metal

ions dissolved from kaolin in the presence of different anionic dispersing agents

[Zam03].

Role of Pressure, Velocity, and pH in Silica CMP

Contact mechanics, wherein the polishing process is believed to occur because of

wear due to direct wafer-particle contact, are most commonly utilized to explain the CMP

process. In fact, most mathematical CMP models are based on an empirical relationship

(equation 4.1) initially proposed for glass polishing [Pre27]:

RR = KpPV (4.1)

where RR is the material removal rate, P is the applied pressure, V is the relative velocity

between the wafer and the pad, and Kp is a process dependent constant [Coo90, Bas02a,

Qin04]. Hence, the pressure and velocity are the most studied variables in CMP

experiments. As the applied pressure is increased, the contact area at the pad-wafer









interface increases. The increase in pad-wafer contact results in an increase in the number

of abrasives available for polishing. Particles not embedded in the polishing pad have a

negligible effect on the material removal rate [Zha99]. According to Basim, a decrease in

particle size or an increase in solids loading results in a significant increase in the

removal rate while an increase in the applied pressure does not significantly impact the

removal rate [Bas02a]. Due to the viscoelastic response of the polishing pad to applied

loads, the actual contact pressure at the particle-wafer interface is not changed

significantly. Improvements in surface quality were predicted for smaller particles, lower

solids loadings, and lower applied pressures.

The removal rate and surface finish are also affected by the relative velocity

between the pad and wafer. The pad-wafer contact area is influenced by the rotational

velocity. Using in-situ friction force measurements, Choi investigated the effect of

velocity on the pad-wafer contact area [Cho03]. According to his results, at low solids

loading (< 5 wt%) the pad-particle contact area is greater than the particle-wafer contact

area, which results in a decrease in friction force with increasing velocity. However, the

increase of friction force with platen velocity when the solids loading exceeded 5 wt%

was attributed to increased shear as the number of particles at the pad-wafer interface

increased. Choi's analysis did not consider the mechanical properties of the pad in his

analysis.

Generally, CMP pads are made of polyurethane or polyurethane impregnated felts

[Ste97]. Therefore, the response to an applied load is time-dependent. For maximum pad-

wafer contact -1% of the pad asperities are actually in contact with the wafer [Bas03]. As

the velocity is increased, the pad response time may affect the pad-wafer contact area.









This may be noticeable in particle-free and low solids loading conditions. There may not

be sufficient pad asperities in contact with the wafer to ensure sufficient abrasives are

available for polishing. However, as the number of particles at the pad-wafer interface

increases, the viscoelasticity of the pad may have less of an effect on the pad-wafer

contact area since sufficient particles may be present.

The pH also has an effect on the silica removal rate and surface finish. However, in

the pH regime of this study (2-6), the pH effects on the silica surface properties should be

negligible. As reviewed in Chapter 2, the pH has a significant effect on the ceria surface

chemistry, which controls the stability and reactivity of the abrasive particle. Above pH

6, rapid agglomeration ensues, which results in a reduction in the total ceria surface area.

This results in a decrease in the density of reactive sites on the ceria particle surface,

which lowers the chemical reactivity of the abrasive particles. Generally, shallow trench

isolation (STI) CMP is conducted below pH 7. Therefore, this study is restricted to the

pH regime in which the ceria slurry should be stable.

Experimental

My Design Expert software version 6.0.5 was used for statistical modeling of the

effect of pressure, pH, and velocity on the removal rate and surface finish during CMP of

silica with ceria abrasives. To formulate an effective statistical design, the general

outcome should be known. Therefore, an initial study of the effect of pH was conducted

at a fixed pressure and velocity. The Box Behnken method was used for this three factor

design (pressure, velocity, and pH). Box Behnken designs are alternatives to 3k factorials

(figure 4-1c, which are a combination of the central composite (figure 4-la)and Box

Behnken designs (figure 4-lb) [Mas89]. Box Behnken designs for 3 factors are efficient,

rotatable [Appendix A], and require fewer runs than central composite or fractional









factorial designs. For three factors, only fifteen experiments are required by the Box

Behnken method compared to 17 for central composite designs and 27 for a 33 factorial

design [Mas89]. In the Box Behnken design, all points may be envisioned as occupying a

sphere. Extreme edges of the cube created by two level factorial designs are not

considered [Box78]. The experimental design for this study, which uses the Box Behnken

method is shown in Appendix B.


x3


-1 0 +1
Factor A


(C) N o A
Figure 4-1 Factor space for different statistical designs: a) central composite b) Box-
Behnken c) fractional factorial [Mas89].

Results and Discussion

Removal Rate Response

Scientific problems are usually expressed as relations of one or more variables (i.e.

material removal rate = f(pressure, pH, velocity)). Using statistical design methodology,









this is accomplished by an empirically derived response function (Appendix A).

'Generally, knowledge of the form of the response function (f) is obtained by fitting

models to data obtained from the statistical design. Response functions are important for

multiple reasons [Mas89]:

1. The function is characterized in the region of interest to the experimenter.

2. They are used to model the relationship between factors and observations.

3. Gives concise summary of experiment.

4. Statistical inferences can be made on the sensitivity of responses to factors of
interest.

5. Factor levels can be determined for which response variable is optimum.

6. Factor levels can be determined that simultaneously optimize several responses.

Contour plots are a popular graphical representation of a response surface that can

be fitted to a statistical design. For simple response functions, contour plots are generated

by plotting one factor versus the other (two level experiment) and fixing the response.

When the response function is more complex, matrices of the response variables are

calculated and the numerical values of the response are plotted as a function of the

factors. The optimum response can then be located by fixing the experimental variables

to the desired values.

Figures 4-2 (a-c) shows the contour plots for pH versus pressure at 75, 112.5, and

150 RPM, respectively. At 75 RPM figure 4-2a, the region that predicts the highest

removal rate is at ~ 6 psi and pH values less than 3. The silica removal rate should be in

the range of 200-800 nm/min. This criterion is only met for low pH values and applied

pressures in

















2.00 4.00 6.00 8.00 10.00 2.00 4.00 6.00 8.00 10.00
(a) A; P.e.ure (b) .,,. P








(c) ,, .


Figure 4-2 Predictions of removal rate as a function of pressure and pH at: a) 75 RPM b)
112.5 RPM c) 150 RPM.

low to mid-pressure regimes < 6 psi at 75 RPM. At 112.5 RPM, the contour plot for pH

versus pressure changes, figure 4-2b. The removal rate is predicted to increase uniformly

as the pH and pressure increases. In contrast to the 75 RPM case, at 112.5 RPM the

minimum removal rate criteria is achievable at all values of pressure and pH. Figure 4-2c

(150 RPM) shows a similar trend to that predicted in Figure 4-2b. However, pH values in

excess of 3 are required to achieve the minimum removal rate criterion. If we compare

figures 4-2a-c, the optimum region for removal rate as function of pressure and pH is

clear only in figure 4-2a. However, in figures 4-2b and 4-2c, the removal rate increases in

the same direction (increasing pH and pressure). The lack of a contour with a maximum

response indicates the optimum removal rate is outside of the experimental region of

figures 4-2b and 4-2c for the respective experimental designs.

Figures 4-3 (a-c) are the interaction profiles for the variables under investigation.

Pressure increases at pH 6.0 have a more pronounced effect on the removal rate than at







70


Interaction Graph Interaction Graph
810.0- : P 959.6- : P"

560.3- 742.2-

0310.5- i 959.- -p2

60.8- 307.4 -

-189.0- 90.0-
2.00 4.00 6.00 8.00 10.00 2.00 4.00 6.00 8.00 10.00
(a) A: Pressure (b) A: Pressure

1369.0- B: P"p

975.4-



188.3--

-205.3 --
2.00 4.00 6.00 8.00 10.00
A: Pressure

(c)


Figure 4-3 Interaction profiles for removal rate as a function of pressure and pH at: a) 75
RPM b) 112.5 RPM c) 150 RPM.

pH 2.0 for the velocity range in this study. From 2-10 psi, the removal rate increases five-


fold at 75 RPM (figure 4-3a) and two-fold at 112.5 (figure 4-3b) and 150 RPM (figure 4-


3c) at pH 6.0. An increase in applied pressure only results in a slight enhancement of the


removal rate at pH 2.0 for all velocities. To understand this behavior, the actual contact


pressure at the pad-wafer interface should be considered. For slurry pH values << than


the IEP of ceria (~ 6.1-6.3 for these slurries), the particle size remains unchanged.


However, as the slurry pH approaches the IEP of the ceria abrasives, the effective particle


diameter increases because of agglomeration, which occurs as particle-particle repulsive


forces are reduced. Now, as the applied pressure increases, the number of particles at the


pad-wafer interface and the pad-wafer contact area increases. For smaller particles (low


pH case), the fraction of the pad that is covered with abrasives or fractional surface


coverage increases [Cho04b]. In this case, the actual contact pressure at the pad-particles-










wafer interface only slightly increases. However, in the case of pH 6.0 large agglomerates

are formed, the fractional surface coverage is less [Cho04b]. Hence the actual contact

pressure or load per particle is higher in the high pH case, which is evidenced by the

removal rate response as a function of applied pressure.

Figure 4-4 shows the predicted removal rate response for all three variables

(pressure (A), pH (B), and velocity (C)). A negative sign on the factor or variable

corresponds to the minimum variable in the range, with positive signs being used for the

maximum value. The numerical values are the removal rate responses for the respective

combinations of pressure, pH, and velocity. The maximum removal rate should be

reached for high pressure, pH, and velocity values (10 psi, pH 6.0, and 150 RPM).


DESIGN-EXPERT Plot Cube Graph
RR
RR
X = A: Pressure 567.25 1158.25
Y = B: pH
Z = C: Velocity


B+ 5.5 263.5




d 5.5 311.5 C+

SC: Velocity

B- 295.75 268.75 C-
A- A+
A: Pressure


Figure 4-4 Predicted removal rate as a function of pressure, pH, and velocity.










Surface Finish Response

A successful CMP process requires more than just a high removal rate. Among the

many parameters that may be considered, surface finish is perhaps second in importance

only to the removal rate. For integrated circuit manufacturing, the surface roughness or

topographical variations must be minimized to ensure conformal deposition of successive

materials and error-free lithography [Wil93]. In STI CMP for DRAM structures, the

surface roughness must be minimized to ensure the read-write heads of recording media

do not crash into the storage material.

063.04"0 ..... --7 0o.709427 09427
5.00- 0.450104
0.320443
**4.00- 4.00-
0.229096



(a) (b) 2.00 4.00 6.00 8.00 10.00
(a) (b) A Pressure









(c) .. .


Figure 4-5 Predictions of RMS roughness as a function of pressure and pH at: a) 75 RPM
b) 112.5 RPM c) 150 RPM

Figures 4-5 (a-c), show the predicted root mean square (RMS) roughness in

nanometers as a function of pressure and pH for 75, 112.5, and 150 RPM respectively.

For the 75 RPM case (figure 4-5a), the lowest roughness values are achieved for pH 3.0

and when the applied pressure ~ 4.0 psi. An increase in the velocity to 112.5 RPM (figure

4-5b) results in a further decrease in the lowest predicted roughness. Again this optimum












region is located at low pH (< 4.0) and low pressure (< 5 psi). Further improvements in


surface finish are predicted as the velocity is increased to 150 RPM, figure 4-5c. This


optimum region is also located at low pH (< 4.0) and low pressure (< 4 psi). As the


rotational velocity is increased, the regions of factor space that predict the lowest RMS


roughness values are found at increasingly lower pH and pressure values. The indentation


depth of a single particle is proportional to the applied pressure and the particle diameter


[Coo90]. An increase in pressure, pH, or both in tandem, results in deeper penetration


into the silica surface. The interaction profiles for the surface roughness predictions,


figures 4-6 (a-c), show that there is a strong interaction between the experimental


variables at low velocities (figure 4-6a). The mutual effects weaken as the rotational


velocity is increased (figure 4-6c and 4-6c). According to this design planarity should


increase with velocity.

nteration Graph Int ti ph
1.63396- B: PH 1.43362- U: PH

1.17781- 1.01637-

O.~1667 0. 9123-

0.26552- 0.181872-

-0.190626- 0.235378- -
2.00 4.00 6.00 8.00 10.00 2.00 4.00 6.00 8.00 10.00
(a) A: r...... (b) A: P.....u

Interalctin Graph
1.75896- B: PH



0.'7917-

1.0623954-

0.503126-
2.00 4.00 6.00 8.00 10.00
(c) A: Pressure


Figure 4-6 Interaction profiles for RMS roughness as a function of pressure and pH at: a)
75 RPM b) 112.5 RPM c) 150 RPM










Figure 4-7 shows the predicted surface finish as a function of pressure (A), pH

(B), and velocity (C). The optimum regions that produce minimal surface deformation

occur at low pressure and pH values and high rotational velocities. The statistical results

have been explained using actual CMP phenomena. Hence, this statistical model

accurately predicts the effect of pressure, pH, and velocity on the removal rate and

surface finish for ceria-silica CMP.


DESIGN-EXPERT Plot

surface finish
X = A: PRessure
Y = B: pH
Z = C: Velocity


B+




id


Cube Graph
surface finish


A: Pressure


Figure 4-7 Predicted removal rate as a function of pressure, pH, and velocity.

Summary

A statistical design based on the Box Behnken method was used to predict the

removal rate and RMS roughness after CMP as a function of pressure, pH, and velocity.

Both the removal rate and surface finish are predicted to be strongly pH and velocity

dependent. The applied pressure is expected to have only a slight effect on the material


C: Velocity









removal rate for most of the experimental conditions examined. However, the surface

roughness does depend strongly on the applied pressure, since the indentation or particle

penetration depth is proportional to the applied pressure. Using statistical methodology,

the isolated and joint effects of three variables were simultaneously investigated.

Statistical designs are a powerful means of process optimization. However, the

goal of this study is to understand the mechanisms of the ceria-silica interactions during

CMP that lead to material removal. Therefore, the remainder of this study will focus on

the fundamentals at the ceria particle silica wafer interface. In the next chapter, chemical

and surface chemical effects are investigated.














CHAPTER 5
THE EFFECT OF CERIA SLURRY PH ON THE SILICA REMOVAL RATE,
FRICTION FORCE, AND SURFACE MORHOLOGY DURING CHEMICAL
MECHANICAL POLISHING

Introduction

A statistical design based on the Box Behnken method was used in Chapter 4 to

predict the effect of pressure, pH, and velocity on the silica removal rate and surface

finish after CMP with ceria slurries. Statistical designs are important methods for process

optimization or materials selection. However, statistical methods are limited in usage by

the depth of knowledge of the system under investigation. Furthermore, mathematical

rather than statistical models are required for articulating the materialistic responses

based on the properties of the components of the system in question (i.e. surface layer

hardness, particle hardness, pad hardness, etc.). Hence, fundamental experiments are

required and the one-factor-at-a-time laboratory testing technique is unavoidable.

The rapid growth of the semiconductor industry is made possible by a combination

of innovative device integration schemes, technological advances, and judicious materials

selection. The continued scaling of microelectronic devices has resulted in unprecedented

improvements in microprocessor performance and an increased density of memory chips.

The advances in device integration and the introduction of new materials to integrated

circuit manufacturing have also driven the development of novel processing techniques.

Vertically stacked or hierarchal type structures that are critically dependent on global

planarity are utilized in a variety of manufacturing techniques.









Hence, CMP is increasingly relevant and has stimulated significant research interest in

the academic arena.

Shallow trench isolation (STI) is an example of a novel manufacturing technique

that is critically reliant upon CMP. Traditionally, silica-based slurries were used for the

interlevel dielectric CMP step during IC fabrication. The popularity of STI has driven the

interest in new CMP consumables (pads, slurries, and slurry additives), to meet the

stringent planarity, removal rate, selectivity, and surface finish requirements. Significant

improvements in device characteristics are realized by replacing conventional isolation

techniques with STI. However, STI CMP is much more complex. For a successful STI

CMP process, the oxide to nitride polish selectivity, oxide erosion or dishing, and nitride

loss must be controlled. Ideally, the STI CMP process will come to a halt immediately as

the oxide is cleared and the nitride is exposed. Ceria-based slurries are promising for high

selectivity polishing as they intrinsically polish oxides faster than nitrides or bare silicon

[Eva04]. The CMP process should also result in uniform removal of the trench oxide.

Deviations in oxide thickness in excess of 3%, trench oxide erosion greater than 50 nm,

and nitride overpolishing may cause damage to the underlying silicon substrate [Sch99,

Zha01]; all are deleterious to device performance.

The "holy grail" of sorts in STI CMP is an appropriately high selectivity, i.e. high

oxide polish rate with little to no nitride removal. If the selectivity is low, nitride loss may

occur. However, if the selectivity is too high, the oxide removal rate is uncontrollable and

oxide erosion results. To achieve the desired oxide to nitride selectivity (> 100:1), two

methods have been adopted. In the first method, surfactants (anionic, nonionic, or a

mixture of the two) are used to passivate or lubricate the nitride surface. In the case









where the slurry pH is less than the IEP of the silicon nitride surface, an anionic

surfactant in the ceria-based slurry is expected to selectively adsorb on the nitride surface.

The mechanisms of the adsorption are not understood, but electrostatic attraction because

of opposite charge at the nitride and surfactant surfaces is expected to play a role. Tseng,

S-D Kim, Katoh, and Park have studied the effect of surfactants on the silica removal rate

for high selectivity polishing [Tse01l, Kim02, Kat03, Par04b]. Fixed abrasive pads are

used in the second approach [SimO1, Gag02, Kul04].

Surfactant Based STI CMP

Tseng et al. studied the effect of two nonionic surfactants with similar molecular

weights: SHE, which has a siloxane head group, and CHE, which has a methyl

methacrylate head group [TseO 1]. Silica-based slurries were used for the studies. The

superior polish rate of CHE-based slurries was attributed to an acceleration of chemical

processes during polishing. The results of Tseng and co-workers are insufficient for the

STI CMP process. The maximum removal rate of the surfactant based slurries was less

than that for silica-based slurries with no additives. Since the oxide removal rate is not

accelerated, the oxide to nitride selectivity should remain low. Using atomic force

microscopy, Basim determined that the addition of surfactants lowers the removal rate

because of increased lubrication at the pad-wafer interface [Bas02a].

S-D Kim et al compared chip scale and wafer-to-wafer thickness variations for a

ceria-based high selectivity slurry (selectivity values were not given) and a silica-based

low selectivity slurry [Kim02]. According to their results, CMP with the conventional

silica slurry resulted in greater nitride overpolishing and oxide erosion. An in-situ filter

was required for the high selectivity slurry to minimize scratches on the polished

surfaces. Katoh et al. studied the effect of surfactant concentration and ceria particle size









on the oxide and nitride removal rate [Kat03, Par04b]. For larger particles (120 nm), the

polish rate without surfactant was twice that of the smaller particles (50 nm) for both

oxide and nitride polishing. However, as the acrylic acid (water-soluble surfactant)

concentration was increased to 0.10 wt%, the nitride polish rate rapidly saturated at ~ 5

nm/min, for the 120 nm and 50 nm (primary particle size) slurries. The oxide polish rate

was not affected until the surfactant concentration reached 0.40 wt% for the 120 nm

particles. The smaller particles showed similar polishing behavior for oxide and nitride

surfaces. These observations were attributed to the difference in particle velocity in the

viscous surfactant as a function of particle size. In their interpretation abrasive particles

freely move through the fluid film, impinge on the surface, and remove material.

According to their conclusions, the velocity of larger particles remains significant enough

to reach the surface until a critical fluid thickness is reached at higher surfactant

concentrations.

This assertion is problematic for several reasons. First, free-flowing particles that

are not embedded in the pad are not believed to contribute to the material removal rate

[Zha99]. Second, the suppression of the nitride removal rate for 120 nm at 0.10 wt%

acrylic acid does not agree with the proposed explanation. Finally, because of

agglomeration, the average particle size was nearly identical up to 0.4 wt% acrylic acid,

negating particle size effects. It is more likely that the acrylic acid passivates the nitride

surface resulting in a saturation of the nitride removal rate at very low surfactant

concentrations. For the oxide CMP case, the enhanced removal rate for larger particles

may occur because of the greater penetration depth of abrasive particles embedded into

the polishing pad. An interesting aspect of this study by Katoh and co-workers is the









possibility of nitride passivation using acrylic acid, which results in enhanced oxide to

nitride selectivity.

Fixed Abrasive Pads for STI CMP

STI CMP with high selectivity slurries is problematic because of increased oxide

erosion. Robust endpoint detection schemes are needed to stop polishing as the nitride is

exposed [Sim01]. Fixed abrasive (FA) pads are a revolutionary approach to STI CMP

[Sim01]. Using FA pads, problematic issues associated with particle-based slurries such

as handling, lifetime, and stability may be alleviated. Generally, FA CMP slurries consist

of pH-adjusted water and a separate chemical for enhanced selectivity. A further

complication of STI CMP with particle-based slurries is the requirement of an initial etch

step to minimize difficult-to-polish topography [Ste97]. CMP with FA pads allows direct

polishing and the additional advantage of reduced oxide erosion [Sim01]. The FA pad is

a composite of ceria abrasives and a resin binder supported by polymer backing (figure 5-

1). The composites have a well-defined shape and are regularly spaced to ensure transport

of the liquid slurry and polish debris. The planarization efficiency is a function of the

morphology of the composite.


Figure 5-1 SEM micrograph of fixed abrasive pad surface (200x) [VelOO].









Simpson et al. studied three different fixed abrasive pads supplied by 3M that were

classified as slow, medium, or fast according to their respective oxide removal rates

[Sim01l]. The slow pad consisted of cylindrically shaped ceria/polymer composite

structures. The oxide removal rate for the slow pad was less than 10 nm/min. Hence, this

pad was deemed suitable for CMP of thin oxide overfills of less than 20 nm [Sim01l].

Simpson and co-workers characterized this pad as tough, meaning micro-scale

topography was required to initiate polishing. The slow pad showed high selectivity to

topography, which means the removal rate at low areas was negligible. Oxide dishing

was eliminated with an overall improvement in oxide and nitride uniformity after CMP.

The polishing capability of the slow pad was critically dependent on the quality of the

oxide deposition [SimO 1]. The pad classified as medium was structurally similar to the

slow pad except that it releases more cerium during polishing [Sim01]. The medium pad

also produced very low oxide removal rates (-20 nm/min). Unlike the slow pad, the

medium pad did not provide a suitable overpolish process window (excessive oxide

dishing), but may be applicable to borophosphosilicate glass polishing [Sim01]. The fast

polishing pad was used for isolation trenches with thick overlying oxide layers. The oxide

polish rate was significantly higher (> 200 nm/min). The fast pad produced poor oxide to

nitride selectivity and excessive oxide dishing, and, was therefore deemed unsuitable for

STI CMP.

Gagliardi et al have also studied FA CMP for STI. L-Proline, an amino acid, was

used to tailor the oxide-to-nitride polish selectivity, eliminate nitride erosion, and oxide

dishing [Gag02]. In their study the polishing tool was specifically configured for the FA

study. A roll of fixed abrasive was loaded onto the CMP machine. After polishing a









wafer, the FA pad is moved ~ 3 mm. After a certain number of wafers are polished, each

subsequent wafer is exposed to an abrasive surface that has been used in an identical

manner [Gag02]. The abrasives were shaped like pyramids and were 63 [m tall and 125

[m wide. The combination of FA pad and L-Proline yielded a robust CMP process with

well-defined polish characteristics. The oxide and nitride removal rates were controlled

by the addition of L-Proline. An oxide to nitride selectivity in excess of 100 was obtained

for a variety of pH conditions and amino acid concentrations.

For precise process control of particle-based and fixed abrasive STI CMP solutions,

a fundamental understanding of the ceria-silica interactions is necessary. However, very

few investigators have focused on the mechanisms of the ceria-silica CMP process. In

chapter 2, the current understanding of the ceria-silica polishing process was reviewed.

Despite the lack of fundamental ceria-silica CMP investigations, three models can be

found in the glass polishing and CMP literature.

The chemical tooth model is most widely relied upon for explanations of the ceria-

silica CMP mechanism. However, nothing in the manner of experimental proof has been

extended to validate the proposed theory. Two other less-referenced models are also

available. One, which I have termed the particulate generation model, depicts the ceria-

silica polishing mechanism as more mechanical in nature. The other, which I have termed

surface defects model, stipulates that efficient polishing compounds must have surface

defects for desired polishing. Initial ceria-silica reactions are a central theme of all three

proposed models. In Chapter 2, the effect of pH on the ceria surface chemistry was

reviewed. Since the ceria polish rate is believed to be enhanced by its intrinsic chemical

reactivity and the solution pH affects the ceria surface chemistry and chemical reactivity;









the CMP performance should be strongly pH dependent. In this chapter, the effect of the

ceria slurry pH on the silica removal rate, friction force at the pad-ceria-silica interface,

and silica surface morphology is presented.

Experimental

All polishing slurries contained ceria and DI water only. To adjust the pH to

desired values (2-6.5), HCI and KOH were used. The slurry was delivered at a constant

rate of 100 ml/min. In-situ lateral friction force measurements and polishing experiments

were conducted on a Struers Rotopol 31 tabletop polisher. A more complete description

of the CMP polisher and friction force apparatus may be found in Chapter 3. The sample

size for polishing and friction tests was 1.0 x 1.0 in., and the samples were ~2 [tm thick

oxides deposited on p-type silicon (100) by chemical vapor deposition. The down

pressures for polishing experiments and friction force measurements were 8.8 and 3.5 psi,

respectively. All samples were polished for 30s. Slurry samples were collected at pH 2.0

and 6.0 during experimental runs. Rodel IC 1000/ Suba IV stacked pads were used for all

experiments. To ensure reproducible conditions, a Grid-Abrade diamond pad conditioner

was used to roughen the polishing pad before each run.

Dynamic light scattering techniques were used to measure the ceria particle size,

zeta potential, and stability. A scanning electron microscope (SEM) equipped with a field

emission gun was also used for particle size measurements. A Microtrac UPA 150

particle size analyzer was used to measure the particle size before and after CMP. A

turbidimeter Model 2100AN manufactured by Hach Company was used to assess the

slurry stability at pH 2.0, 6.0, 6.5, and 10.5. Zeta potential measurements on the slurries

before and after CMP were performed on an apparatus manufactured by Brookhaven

Instruments. Material removal rates were calculated by measuring the oxide thickness






84


before and after CMP with a variable angle spectroscopic ellipsometer. AFM was used to

characterize the morphology of the polished surfaces after CMP.

Results and Discussion

Ceria Slurry Characterization

In the pH range studied, the silica substrate possesses a predominantly negative

charge because of deprotonation of surface hydroxyls.

Si-OH-Si-O-+H+ (5.1)

Conversely, the ceria abrasives go from predominantly positive to mostly neutral as the

pH of the slurry approaches its isoelectric point.

Ce-OH2+-Ce-OH+H+ (5.2)


S i- l E 1400
E I 1200
-R 2D -1000
= iE 8M
600

40- <
0. 20 aw 400
*-200
a ^0 t(Sn


1 2 3 4 5 6 7 8
o 2 4 6 8 3 12 CeO2 slurry pH
PH

Figure 5-2 Ceria slurry characteristics as a function of pH a) zeta potential b) particle
diameter.

The IEP for silica is between pH 2-3 and pH 6-7 for ceria [Coo90]. The effect of pH on

the ceria slurry properties is shown in Figures 5-2a and 5-2b. In the absence of surfactants

or other dispersing agents, electrostatic means are solely responsible for the dispersion

state of the slurry. Figure 5-2a indicates the IEP of the ceria abrasives used is ~ pH 6.0.

As the pH is increased from 2-6.5, the electrostatic interactions become less repulsive. As









expected, slurry agglomeration ensues, resulting in an increase in abrasive size and a

decrease in slurry stability (Figures 5-2b and 5-3). Even at pH 10.5, the ceria slurry

remains unstable with the particles rapidly settling out of solution in less than 10 minutes.

Since direct imaging is possible, SEM measurements were conducted, (figure 5-4). The

primary particle size as determined by SEM is ~ 60 nm. The particle size values

determined by light scattering should be considered the secondary or agglomerated

particle size. The particle size measurements via light scattering were consistent even

after sonication.




1.1

D, 1.0
-' -I -- -
Z 0.9
V 0.8 pH 2.0
pH 6.0
.N -- pH 6.5
S0.7 pH 10.

0 0.6

0.5
0 20 40 60 80 100 120 140

time (minutes)


Figure 5-3 Stability of the ceria slurry as a function of time for different pH values.
Vertical axis is displayed as normalized nephelometric turbidity units (NTU).

The lack of ceria stability at pH 10.5 is unexpected, especially in light of the zeta

potential measurements. At pH 10.5, the zeta potential is ~ -40 mV, approximately equal

in magnitude to the low pH case. Therefore, it is not unreasonable to expect that ceria

dispersions at pH 10.5 should be as stable as dispersions at pH 2.5. The actual state of the

dispersions used for turbidity and zeta potential measurements and surface chemistry of









ceria as studied by Nabavi et al. [Nab93] may shed some light on these contradictory

observations.


Figure 5-4 FE-SEM image of ceria particles.

The dispersions for zeta potential measurements are much more dilute than for

turbidity measurements. For turbidity measurements, the ceria slurry is added to the

turbidity cell as is, since the concentration is less than 3 wt%. However, for zeta potential

measurements the specimen holder was initially filled with pH 10.5 DI water and then a

drop of the ceria slurry was added to the specimen. The inter-particle spacing is greater

for the dilute suspension and the likelihood of particle-particle collision diminishes as the

concentration of the dispersion decreases. The samples for turbidity measurements are

more representative of the actual slurries used for the CMP experiments. The zeta

potential is directly related to the density of charged sites at the abrasive particle surface.

Nabavi's investigation of the ceria surface chemistry as a function of pH is relevant to

this discussion.


29.9nm

62.Snm 56.2nm







MAIC SO 15.OkV X100.000 100nni WD 143akm