USF Libraries
USF Digital Collections

Chemical mechanical planarization

MISSING IMAGE

Material Information

Title:
Chemical mechanical planarization study of conditioner abrasives and synthesis of nano-zirconia for potential slurry applications
Physical Description:
Book
Language:
English
Creator:
Manocha, Chhavi
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Chemical mechanical polishing
Conditioning
Pad wear
Contact area
Plasma synthesis
Dissertations, Academic -- Chemical and Biomedical Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Chemical Mechanical Planarization (CMP) has emerged as the central technology for polishing wafers in the semiconductor manufacturing industry to make integrated multi-level devices. As the name suggests, both chemical and mechanical processes work simultaneously to achieve local and global planarization. In spite of extensive work done to understand the various components and parameters affecting the performance of this process, many aspects of CMP remain poorly understood. Among these aspects of CMP is the role of abrasives in the processes of conditioning and polishing. These abrasives are present in the chemical slurry between the wafer and the pad for polishing and play an important role during the conditioning to regenerate the clogged polishing pads. This thesis has focused on the study of abrasives, both in conditioning and polishing. The first part of the thesis concentrates on the effect of abrasive size for conditioning purposes.Diamond is being widely used as an abrasive for conditioning the polishing pad. Five different sizes of diamonds ranging from 0.25μm to 100μm were selected to condition the commercially available IC 1000 polishing pad. Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM) analysis were carried out on the pad to study the effect of the abrasive size on the pad morphology. In-situ 'coefficient of friction' was also monitored on the CETR bench top Tester. The final impact was seen in the form of surface defects on the polished copper wafers. As pad morphologies resulting from different conditioning affect contact areas, the second part of the thesis focuses on developing a simple method to quantify the area of contact between the wafer and pad using optical microscopy. Optical images that were obtained were analyzed for the change in contact area with the change in operating conditions.Finally, the third part of the thesis details the synthesis and characterization of nano-zirconia for potential slurry applications. Nano-zirconia was synthesized using the plasma route and then characterized using different analytical techniques like TEM and XRD. These nanoparticles were then used to make abrasive slurry for oxide CMP and the polished wafers were analyzed for surface defects.
Thesis:
Thesis (M.S.Ch.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Chhavi Manocha.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 96 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002007020
oclc - 401294663
usfldc doi - E14-SFE0002741
usfldc handle - e14.2741
System ID:
SFS0027058:00001


This item is only available as the following downloads:


Full Text

PAGE 1

I Chemical Mechanical Planarization: Study of Conditioner Abrasives and Synthesis of Nano-Zirconia for Potentia l Slurry Applications by Chhavi Manocha A thesis submitted in partial fulfillment of the requirements for the degree of Master of Chemical Engineering Department of Chemical a nd Biomedical Engineering College of Engineering University of South Florida Co-Major Professor: Vinay K. Gupta, Ph.D. Co-Major Professor: Ashok Kumar, Ph.D. Andrew M. Hoff, Ph.D. John T. Wolan, Ph.D. Date of Approval: October 31, 2008 Keywords: chemical mechanical polishing, c onditioning, pad wear, c ontact area, plasma synthesis Copyright 2008 Chhavi Manocha

PAGE 2

II Dedication I dedicate my thesis to my parents, Dr L. M. Manocha and Dr. S. Manocha. Their constant support, encouragement and guidan ce always made it possible for me to sail through the toughest times and believe in myself

PAGE 3

III Acknowledgements I take this opportunity to thank my adviso rs, Dr. Vinay K. Gupt a and Dr. Ashok Kumar, for their help and guidance throughout my resear ch work. I would also like to thank Dr. Anandan for his constant support and en couragement. Some of the work presented here was done during my summer internship at 3M. I would like to thank Rajesh Katare for giving me an opportunity to work on th e Plasma synthesis, Dr. Gundu Sabde and Gary Palmgren for the work done on optical microscopy. I’d like to express my gratitude towards Dr. Raghu Mudhivarthi Dr. Makoto Hirai and Cecil Coutinho for introducing me to the subject and helping me with my experiments

PAGE 4

I I Table of Contents List of Tables ................................................................................................................ IV List of Figures ............................................................................................................... .. V ABSTRACT .................................................................................................................. XII Chapter 1 Introduction and Background .......................................................................... 1 1.1. Planarization techniques ................................................................................... 1 1.2. Chemical Mechanical Planarization ................................................................. 3 1.2.1. Working of the CMP process ................................................................ 3 1.2.2. CMP mechanism ................................................................................... 4 1.2.3. Input and output va riables in CMP ....................................................... 5 1.2.4. Relation between the input and the output variables ............................ 7 1.2.5. Advantages and disadvantages of CMP ................................................ 8 Chapter 2 Effect of Diamond Size on Conditioning and Polishing in CMP ................. 13 2.1. Polishing pad ................................................................................................... 13 2.1.1. Pad surface and structure .................................................................... 13 2.1.2. Pad properties and polishing performance .......................................... 14 2.2. Pad conditioning ............................................................................................. 15 2.2.1. Conditioners ........................................................................................ 17 2.3. Research focus ................................................................................................ 18 2.4. Experimental setup .......................................................................................... 18

PAGE 5

II II 2.5. Conditioning experiments ............................................................................... 19 2.6. Results of conditioning experiments ............................................................... 20 2.6.1. Pad morphology .................................................................................. 20 2.6.2. Pad wear .............................................................................................. 22 2.6.3. Coefficient of friction ......................................................................... 23 2.6.4. Pad roughness ..................................................................................... 24 2.6.5. Dynamic Mechanical Analysis ........................................................... 26 2.7. Polishing experiments ..................................................................................... 27 2.8. Results of the polishing experiments .............................................................. 28 2.8.1. Optical images of the polished wafer .................................................. 28 2.8.2. Evaluation of topography .................................................................... 29 2.8.3. Pad morphology .................................................................................. 30 2.8.4. Coefficient of friction ......................................................................... 31 Chapter 3 Optical Measurement of the C ontact Area between the Pad and Wafer ....... 55 3.1. Concept of contact ar ea and its importance .................................................... 55 3.2. Measurement of contact area .......................................................................... 58 3.3. Experimental details ........................................................................................ 59 3.4. Results and discussion .................................................................................... 60 Chapter 4 Synthesis of Zirconia Nanopa rticles through Plasma Route for CMP Slurry Applications ...................................................................................... 74 4.1. Synthesis of zirconia through various methods .............................................. 74 4.2. Plasma synthesis of nanopowders ................................................................... 77 4.3. Zirconia nanoparticles through the plasma route ............................................ 78

PAGE 6

III III 4.4. Role of slurry in Chemical Mechanical Polishing .......................................... 79 4.4.1. Slurry composition .............................................................................. 80 4.4.2. Abrasives in slurry .............................................................................. 80 4.5. Research focus: synthesis, char acterization and a pplications ......................... 82 4.6. Results and discussion .................................................................................... 83 Chapter 5 Conclusion ..................................................................................................... 88 References ................................................................................................................... .. 92

PAGE 7

IV IV List of Tables Table 2.1: Experimental conditi ons for wafer polishing ...................................... 48 Table 3.1: Statistics showing the cha nge in contact area for a dry pad ................ 72 Table 3.2: Statistics showing the cha nge in contact area for a wet pad ................ 73 Table 4.1: Experime ntal details ............................................................................ 87

PAGE 8

V V List of Figures Figure 1.1: Multilevel metallization structure a) Non-planarized; b) Planarized ............... 9 Figure 1.2: Surface smoothing ............................................................................................ 9 Figure 1.3: Local planarization ........................................................................................... 9 Figure 1.4: Global planarization ......................................................................................... 9 Figure 1.5: Schematic of CMP process ............................................................................. 10 Figure 1.6: Removal rate as a function of ab rasive diameter at (a) low and (b) high values of PxV. ..................................................................................................11 Figure 1.7: Removal rate as a function of Px V showing the effect of slurry flow. .......... 12 Figure 1.8: Correlation between removal rate and pad r oughness. The data in the figure is from literature ( 14 ). ...........................................................................12 Figure 2.1: CETR bench top tester ................................................................................... 32 Figure 2.2: Photograph of a 0.25 m abrasive conditioner ............................................... 33 Figure 2.3: SEM image of a 0.25m abrasive conditioner ............................................... 33 Figure 2.4: Photograph of a 2 m abrasive conditioner .................................................... 33 Figure 2.5: SEM image of a 2 m abrasive conditioner .................................................... 33 Figure 2.6: Photograph of a 8 m abrasive conditioner .................................................... 33 Figure 2.7: SEM image of a 8 m abrasive conditioner .................................................... 33 Figure 2.8: Photograph of a 68 m abrasive conditioner .................................................. 34 Figure 2.9: SEM image of a 68 m abrasive conditioner .................................................. 34 Figure 2.10: Photograph of a 100 m abrasive conditioner .............................................. 34

PAGE 9

VI VI Figure 2.11: SEM image of a 100m abrasive conditioner .............................................. 34 Figure 2.12: SEM image of a fresh pad ............................................................................ 35 Figure 2.13: SEM image of pad conditioned using 0.25m abrasive .............................. 35 Figure 2.14: SEM image of pad conditioned using 100m abrasive ................................ 35 Figure 2.15: SEM images of pad conditione d with 0.25m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X .................................36 Figure 2.16: SEM image of pad conditioned with 2m abrasive for 5 min at 150 rpm at two different magnifica tions (a)150X (b)300X ........................................36 Figure 2.17: SEM image of pad conditioned with 8m abrasive for 5 min at 150 rpm at two different magnifica tions (a)150X (b)300X ........................................36 Figure 2.18: SEM image of pad conditioned w ith 68m abrasive for 5 min at 150 rpm at two different magnifica tions (a)150X (b)300X ........................................37 Figure 2.19: SEM image of pad conditioned with 100m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X .................................37 Figure 2.20: SEM image of pad conditione d with 0.25m abrasive for 20 min at 150 rpm at two different magni fications(a)150X (b)300X ...........................38 Figure 2.21: SEM image of pad conditioned w ith 2m abrasive for 20min at 150 rpm at two different magnifica tions (a)150X (b)300X ........................................38 Figure 2.22: SEM image of pad conditioned w ith 8m abrasive for 20min at 150 rpm at two different magnifica tions (a)150X (b)300X ........................................38 Figure 2.23: SEM image of pad conditioned with 68m abrasive for 20 min at 150 rpm at two different magnifications (a)150X (b)300X .................................39

PAGE 10

VII VII Figure 2.24: SEM image of pad conditioned with 100m abrasive for 20 min at 150 rpm at two different magnifications (a)150X (b)300X .................................39 Figure 2.25: SEM image of pad conditioned with 0.25m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X .................................40 Figure 2.26: SEM image of pad conditioned with 2m abrasive for 5 min at 200 rpm at two different magnifica tions (a)150X (b)300X ........................................40 Figure 2.27: SEM image of pad conditioned with 8m abrasive for 5 min at 200 rpm at two different magnifica tions (a)150X (b)300X ........................................40 Figure 2.28: SEM image of pad conditioned w ith 68m abrasive for 5 min at 200 rpm at two different magnifica tions (a)150X (b)300X ........................................41 Figure 2.29: SEM image of pad conditioned with 100m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X ................................41 Figure 2.30: SEM image of pad conditioned with 0.25m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X .................................42 Figure 2.31: SEM image of pad conditioned with 2m abrasive for 20 min at 200 rpm at two different magnifica tions (a)150X (b)300X ........................................42 Figure 2.32: SEM image of pad conditioned with 8m abrasive for 20 min at 200 rpm at two different magnifica tions (a)150X (b)300X ........................................42 Figure 2.33: SEM image of pad conditioned with 68m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X .................................43 Figure 2.34: SEM image of pad conditioned with 100m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X .................................43

PAGE 11

VIII VIII Figure 2.35: Plot of pad wear at different conditioning times for different abrasive sizes ( m) at 150 rpm ...................................................................................44 Figure 2.36: Plot of pad wear at different conditioning times for different abrasive sizes ( m) at 200 rpm ...................................................................................44 Figure 2.37: Plot of COF at di fferent conditioning times for different abrasive sizes (m) at 150 rpm ............................................................................................45 Figure 2.38: Plot of COF at di fferent conditioning times for different abrasive sizes (m) at 200 rpm ............................................................................................45 Figure 2.39: Plot of roughness at different conditioning times for different abrasive sizes ( m) at 150 rpm. ..................................................................................46 Figure 2.40: Plot of roughness at different conditioning times for different abrasive sizes ( m) at 200 rpm ...................................................................................46 Figure 2.41: Plot of storage modulus at a fr equency of 30 Hz for pads conditioned for 20 min with different abrasives.....................................................................47 Figure 2.42: Plot of storage modulus at a frequency of 100Hz for pads conditioned for 20 min with different abrasives.....................................................................47 Figure 2.43: Optical image of first wafer polished on the pad conditioned with 8m abrasive at 20X ............................................................................................ 49 Figure 2.44: Optical image of third wafer polished on the pad conditioned with 8m abrasive at 20X ............................................................................................ 49 Figure 2.45: Optical image of first wafe r polished on the pad conditioned with 68m abrasive at 20X ............................................................................................ 49

PAGE 12

IX IX Figure 2.46: Optical image of third wafer polished on the pad conditioned with 68m abrasive at 20X .............................................................................................49 Figure 2.47: Optical image of first wafer polished on the pad conditioned with 100m abrasive at 20X .............................................................................................49 Figure 2.48: Optical image of third wafe r polished on the pad conditioned with 100m abrasive at 20X .................................................................................49 Figure 2.49: AFM image of the first wafe r polished on a pad conditioned with 8m abrasive .........................................................................................................50 Figure 2.50: AFM image of the third wa fer polished on a pad conditioned with 8mabrasive ..................................................................................................50 Figure 2.51: AFM image of the first wafe r polished on a pad conditioned with 68m abrasive .........................................................................................................50 Figure 2.52: AFM image of the third wafe r polished on a pad conditioned with 68m abrasive .........................................................................................................50 Figure 2.53: AFM image of the first wafe r polished on a pad conditioned with 100m abrasive .........................................................................................................50 Figure 2.54: AFM image of the third wa fer polished on a pad conditioned with 100m abrasive .............................................................................................50 Figure 2.55: AFM image of pad conditioned with 8m abrasive ..................................... 51 Figure 2.56: AFM image of pad conditioned with 68m abrasive ................................... 51 Figure 2.57: AFM image of pad conditioned with 100m abrasive ................................. 51 Figure 2.58: SEM image of pad conditioned with 8m abrasive ..................................... 52 Figure 2.59: SEM image of pad conditioned with 68m abrasive ................................... 52

PAGE 13

X X Figure 2.60: SEM image of pad cond itioned with 100m abrasive ................................. 52 Figure 2.61: Plot of COF for abrasive size of 8 m .......................................................... 53 Figure 2.62: Plot of COF for abrasive size of 68 m ........................................................ 53 Figure 2.63: Plot of COF for abrasive size of 100 m ...................................................... 54 Figure 3.1: A schematic diagram depicti ng contact points between pad and wafer during polishing ............................................................................................64 Figure 3.2: Plot of % pad contact vs. head pr essure. The data in the figure is from the literature ( 37 ) ............................................................................................... 64 Figure 3.3: A schematic diagram of the experime ntal setup for measuring contact area 65 Figure 3.4: Nikon measuresc ope UM-2 tester .................................................................. 65 Figure 3.5: Optical image of dry pad at 20X .................................................................... 66 Figure 3.6: Optical image of wet pad at 20X .................................................................... 66 Figure 3.7: Optical image of wet pad afte r introduction of bl ue ink at 20X ..................... 66 Figure 3.8: SEM image of a pad ....................................................................................... 66 Figure 3.9: Optical image of wet pad with a focused laser at 20X ................................... 67 Figure 3.10: Optical image of a pad at 20X showing laser points that represent the debris on the pad ..............................................................................................67 Figure 3.11: Optical image of wet pad at 20X .................................................................. 68 Figure 3.12: Optical image of wet pa d with laser points at 20X ...................................... 68 Figure 3.13: Optical image of pad at 30 torr ..................................................................... 69 Figure 3.14: Optical image of pad at 50 torr ..................................................................... 69 Figure 3.15: Optical image of pad at 70 torr ..................................................................... 69 Figure 3.16: Optical image of pad show ing contact points at 30 torr ............................... 70

PAGE 14

XI XI Figure 3.17: Optical image of pad show ing contact points at 50 torr ............................... 70 Figure 3.18: Image showing analysis performed by ImagePro ........................................ 70 Figure 3.19: Optical image of wet pad at 30 torr .............................................................. 71 Figure 3.20: Optical image of wet pad at 50 torr .............................................................. 71 Figure 3.21: Image showing analysis performed by ImagePro ........................................ 71 Figure 4.1: TEM images of zirconia nanopart icles at different magnifications ............... 85 Figure 4.2: XRD analysis of zirconia nanoparticles ......................................................... 85 Figure 4.3: Optical images s howing different regions of the wafer polished with zirconia slurry ...............................................................................................86

PAGE 15

XII XII Chemical Mechanical Planarization: Study of Conditioner Abrasives and Synthesis of Nano-Zirconia for Potential Slurry Applications Chhavi Manocha ABSTRACT Chemical Mechanical Planarization (CMP ) has emerged as the central technology for polishing wafers in the semiconductor manufacturing industry to make integrated multi-level devices. As the name suggests, bot h chemical and mechanical processes work simultaneously to achieve local and global plan arization. In spite of extensive work done to understand the various components and para meters affecting the performance of this process, many aspects of CMP remain poorly understood. Among these aspects of CMP is the role of abrasives in the processes of conditioning and polishing. These abrasives are present in the chemical slurry between th e wafer and the pad for polishing and play an important role during the conditioning to regenerate the clogged polishing pads. This thesis has focused on the study of abrasives, both in conditioning and polishing. The first part of the thesis con centrates on the effect of abrasive size for conditioning purposes. Diamond is being widely used as an abrasive for conditioning the polishing pad. Five different si zes of diamonds ranging from 0.25 m to 100 m were selected to condition the commercially available IC 1000 polishing pad. Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM) anal ysis were carried

PAGE 16

XIII XIII out on the pad to study the effect of the abrasive size on the pad morphology. In-situ ‘coefficient of friction’ was also monitored on the CETR bench top Tester. The final impact was seen in the form of surface de fects on the polished copper wafers. As pad morphologies resulting from diffe rent conditioning affect cont act areas, the second part of the thesis focuses on developing a simp le method to quantify the area of contact between the wafer and pad using optical micr oscopy. Optical images that were obtained were analyzed for the change in contact area with the change in operating conditions. Finally, the third part of the thesis details the synthesis and ch aracterization of nanozirconia for potential slurry applications. Nano-zirconia was synthesized using the plasma route and then characterized using different analytical techniques like TEM and XRD. These nanoparticles were then used to ma ke abrasive slurry for oxide CMP and the polished wafers were analyzed for surface defects.

PAGE 17

1 1 Chapter 1 Introduction and Background To enable superior IC perf ormance, it is critical to reduce device dimensions. As the devices become smaller seve ral challenges arise in thei r design and fabrication. After fabrication of the devices on the silicon subs trate, metallic interconnects are used to interface these devices with each other and to the outside world. At the silicon level, the devices are interconnected using metallization, wh ich is referred to as the first level of metallization. The connections between the device network and the outside world are referred to as the second level of metallizat ion. In order to successfully build a multilevel metallization structure, the surfaces of each prev ious metal layer has to be perfectly flat. In the absence of planarity, the irregular surfaces and uneven thicknesses lead to an inefficient pattern transfer (Figure 1.1) A planarized surface promotes homogeneous thickness of metal layers and results in lowering the level defects and the electro migration effects. Thus, planarization play s an important role in the semiconductor industry. 1.1. Planarization techniques Wafer planarization may be classified into three categories of planarity. These are summarized below ( 1, 2 ):

PAGE 18

2 2 Surface Smoothing: Feature corners are smoothed and high aspect ratio holes are filled as in Figure 1.2. 1. Local planarity: Surfaces ar e flat locally, but the surface height may vary across the die as in Figure 1.3. 2. Global planarity: The surface is flat across the entire stepper field as in Figure 1.4. The requirement for surface smoothing and lo cal planarity comes from metal step coverage, which is defined by the ratio of thinne st point in metal film to the thickest point in metal film ( 3 ). The requirement for global planarity increases when the circuit dimensions reach sub-0.5. Several tec hniques exist to achieve local and global planarization. Different planarization proce sses are seen to achiev e different degrees of planarity. Some of these techni ques are discussed below. Doped Gas Reflow: One of the first plan arization techniques used in the IC industry involved the synthe sis of low pressure chemical vapor deposited (LPCVD) boron and phosphorous doped silicon oxide ( 4, 5 ). This was used for the planarization of the first layer of dielectric Better smoothing of step corn ers could be achieved by doping SiO2 with boron and phosphorous. Spin Etch Planarization: Spin Etch Planarization is based on the principles of controlled chemical etching of metals ( 6 ). In this process, the wafer is suspended horizontally on a nitrogen cushion above a ro tating chuck. Locking pins are used on the edge of the wafer to locate the substrate laterally. Wet etch chemistries are dispensed into the wafer, as it is being rotated by the chuc k. Using an appropriate etching solution, the planar surface is achieved while the wafer is being rotated and excess Cu is removed.

PAGE 19

3 3 Spin on Deposition : In this process, the base catalyst and necessary organic additives are mixed at room temperature to pr epare the precursor solution for the material to be deposited. Pretreating of the wafer surface is conducted in order to promote sol spreading. The sol is dripped on the wafer while it is spinning. The wa fer is then rinsed, spun dried, baked, and then cured. It results in local planarization ( 7 ). 1.2. Chemical Mechanical Planarization As discussed earlier, the process of sm oothening and planing a surface is referred to as planarization. When the smoothening a nd planning of surfaces is aided by chemical and mechanical forces, it is referred to as Chemical Mechanical Planarization (CMP). Chemical Mechanical Planarizat ion was introduced by IBM to facilitate planarization of inter-level dielectrics ( 8 ). In addition, CMP is also used to polish multiple materials in a Damascene process. Chemical Mechanical Pl anarization or CMP is the method of choice for achieving local and global planarity in th e wafers. As the name suggests, CMP is a combination of chemical and mechanical processes working at the same time to achieve planarization. Further, the ongoi ng chemical reaction is seen to enhance the mechanical removal rate. The key to achieving planarity is maintaining high removal rate at high surface features and low removal rates at low surface features ( 9 ). 1.2.1. Working of the CMP process Figure 1.5 shows the schematic of the CMP process. The CMP polisher consists of a bottom rotating platen on which a polishi ng pad is fixed. The top platen holds the wafer to be polished, face down on the pad. During the polishing run, the wafer comes in contact with the pad surface w ith a set downforce. Both the platens rotate during the run.

PAGE 20

4 4 Polishing slurry consisting of abrasive particle s and other necessary chemical constituents are fed to the pad surface to carry out the polishing where the pad co mes in contact with the wafer being polished. The abrasive slur ry has several functions. It provides the necessary chemical input for the planarization process, acts as a lubricant to maintain the temperature caused due to the friction and also transports the particles across the polishing surface. The pad used for polishing is seen to glaze under the effect of polishing. To get higher removal rates, the pad needs to be regenerated. This is done by conditioning, which is the second stage of the CMP process. The rotating polishing pad is abraded using a conditioner which is an abrasi ve disk that opens up the clogged pores of the pad. The polishing mechanism can be explained on the basis of the contacts occurring in the system. The chemicals in the slurry, on contact with the wafer, soften the wafer surface, while the abrasives in the slurry re move this layer. In the absence of the mechanical fraction, the chemical effect is limited. 1.2.2. CMP mechanism The polishing takes place at points where the wafer surface comes in contact with the surface of the pad, which in turn holds the slurry particles. This mechanism is best described by the Preston’s law, which states that the material removal rate is directly proportional to the applied pressure a nd relative velocity of the particles( 10 ). The mathematical expression for the law is given by the equation shown below. RR = KP.P.V RR = Removal rate; P = Applied pressure; V = Velocity; KP = Preston’s constant;

PAGE 21

5 5 Preston’s equation is an empirical law. It shows a linear dependence of removal rate on pressure and relative velocity. Pres ton’s constant accounts for the other chemical and physical parameters involved in polishi ng. The Polishing pad surface gets more compressed by the raised wafer surfaces, which results in higher pressure giving higher removal rates. 1.2.3. Input and output variables in CMP The CMP process is quite complicated a nd involves a large number of input and output variables. The different output variab les include polish rate, planarization rate, surface quality, surface damage, and feature size dependence. 1. Polish Rate: The polish rate is defined as the ratio of the thickness of the film removed to the polish time. Its units are either mm/min or m/min. Higher polish rates imply shorter polish times and are he nce, desirable. However, controlling the process is difficult when th e polish rates go higher. 2. Planarization Rate: Planarization rate is defined in terms of the time taken to reduce the topography of a surface to the desired level. 3. Surface Quality: The expected yield and re liability of the interconnections are dependent on the surface quality. The surface quality can be improved and roughness minimized by properly balanci ng the chemical and mechanical components of the CMP process. 4. Surface Damage: Damage may occur to meta l films, which directly affects the yield and reliability of the interconnections. Structural damage may include scratches, delamination of the film in terfaces, introduction of impurities in the film, etc.

PAGE 22

6 6 In case of input variables, four broad categ ories can be used: Wafer, slurry, polishing pad and operating conditions. 1. Wafer: The wafer to be polished determin es the other associated input variables for polishing. The wafer surface can be ma de up of one material or can be a combination of more than one material. Polis hing of the top film is affected by the mechanical properties of the films below it. Other factors such as film stress, microstructure and hardness also play an important role in achieving planarization ( 11 ). Physical parameters such as wa fer curvature or size and mounting techniques also affect the wafer polis hing. These factors work by the way of influencing the slurry transport and th e distribution of th e applied load. 2. Slurry: The slurry is a combination of a chemicals and abrasives. These components are selected on the basis of th e material being polished as different materials show different chemical interactions. The chemical part of the slurry consists of buffering agents to maintain the pH, oxidizers to increase the oxidation rate, and complexing agents to increase the solubility of the film being polished. The nature and concentration of these ch emicals contributes to the etch rate during polishing. The slurry abrasives cons titute the mechanical aspect of the slurry by causing abrasion. Chemical and physical properties of these abrasive particles such as nature, size, hardness, and concentration aff ect the removal rate and the surface damage during polishing. 3. Polishing Pad: The polishing pad suppor ts the wafer being polished while transporting the slurry across its surface. Some of the important pad properties such as fiber structure, compressibility, hardness, m odulus, thickness, pore size

PAGE 23

7 7 and distribution, and chemical durability determine the polishing efficiency as the pad is constantly influenced by the other input parameters. 4. Operating Conditions: These refer to th e experimental conditions such as temperature, pressure, platen velocity for the pad and wafer, and slurry flow rate. These parameters again, are controlled in coordination with the other input parameters, mainly for optimizing the process. 1.2.4. Relation between the input and the output variables As stated earlier, the output variables ar e the end-representation of the chosen input variables. Extensive re search has been done to bett er understand the relationship between these two sets of variables. To give an example, some of the interaction results are discussed here keeping removal rate as the output para meter of interest. 1. Removal rate, surface defects and abrasive size ( 12 ): At low pressures and velocities, the removal rate is seen to be independent of abrasive size while at higher values of pressure and velocity, larger abrasive size shows higher removal rates as illustrated in Figure 1.6. Surface defects increase with an increase in the abrasive size. These surface defects are se en in terms of micro-scratches on the wafer surface. 2. Removal rate and slurry flow rate ( 13 ): At constant values of P x V, with an increase in the slurry flow rate, the remova l rate is seen to decrease as shown in Figure 1.7. This has been explained on the basis of a higher flow rate producing more cooling. 3. Removal rate and pad roughness ( 14 ): Conditioning is an important regeneration process in CMP. In the absence of conditioning, the roughness of the pad is seen

PAGE 24

8 8 to decrease. As a consequence, the rem oval rate, which is related to the pad roughness, is also seen to decrease. Th e results can be seen in Figure 1.8. 1.2.5. Advantages and disadvantages of CMP CMP is a suitable process to achieve global planarizati on. Different materials can be planarized using CMP. This implies that a wide range of wafer surfaces may be planarized using CMP. A major advantage of using CMP is that multiple materials can be polished in the same cycle. Use of CMP helps in reducing severe topography, which facilitates the fabrication of IC components w ith tighter tolerances and design rules. CMP also provides an alternate method to etch metals. This helps in overcoming difficulties associated with etching some metals and allo ys and also eliminates the need to plasma etch. More importantly, CMP is a subtrac tive process and helps in removing surface defects. No hazardous gases are used in CMP process unlike dry etching. One of the major limitations of the CMP pr ocess is that it is a relatively new technology for wafer planarization. Hence, th ere is not much control over the process variables. The introduction of a new tec hnology implies introduction of new defects, which can affect the die yield. These defects may prove critical for features that are smaller than 0.25m in size. Since CMP is a relatively new technology, additional development is required to control the proce ss and metrology. For exam ple, at this point it is difficult to control the end point of th e CMP process in order to achieve the desired thickness. The CMP process is also expensive due to high cost of equipment and the high maintenance cost from frequent replacement of parts.

PAGE 25

9 9 Figure 1.1: Multilevel metallization struct ure a) Non-planarized; b) Planarized Figure 1.2: Surface smoothing Figure 1.3: Local planarization Figure 1.4: Global planarization

PAGE 26

10 10 Figure 1.5: Schematic of CMP process

PAGE 27

11 11 (a) 1960 1980 2000 2020 2040 2060 2080 2100 2120 2140 0510152025303540size (nm)Removal rate(A/min) (b) 0 1000 2000 3000 4000 5000 6000 7000 0510152025303540size (nm)Removal rate (A/min) Figure 1.6: Removal rate as a function of abrasive diameter at (a) low and (b) high values of PxV. The data in these figures is from literature ( 12 ).

PAGE 28

12 12 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 05000100001500020000Pressure X Velociy (Pa/ms)Removal rate (A/min) 80 cc/min 140 cc/min Figure 1.7: Removal rate as a function of PxV showing the effect of slurry flow. The data in the figure is from literature ( 13 ). 0 500 1000 1500 2000 2500 3000 22.12.22.32.42.52.62.72.82.9Rpk(micron)Removal rate (A/min) Figure 1.8: Correlation between removal rate and pad roughness. The data in the figure is from literature ( 14 ).

PAGE 29

13 13 Chapter 2 Effect of Diamond Size on Co nditioning and Polishing in CMP As discussed in chapter 1, polishing pad is one of the consumables of the CMP process with a direct effect on the output. It acts as the plat form on which the polishing is carried out. The pad is made up of a matrix of polyurethane foam with a filler material such as polymer felt ( 15 ). Mechanical reliability and chem ical resistance are essential for the pad to extend the th oroughness of polishing. 2.1. Polishing pad 2.1.1. Pad surface and structure Polishing pads are generally made up of polyurethane as the chemical properties of this polymer allow modifications to the pa d in order to comply with the mechanical needs for the process. Polyurethanes ha ve interesting morphologies. They show contrasting molecular structure of linear elastomers as well as of rigid cross linked polymers. Based on the microstructure, pads can be divided into four categories ( 16 ). Type 1 pads typically are polymer impregnated felts. The microstructure of a pad of Type 1 is characterized by non-woven polyester fibers in fused with polyurethane. Poromerics form the Type 2 pads and pads of this type displa y a porous layer on a substrate. Type 3 pads are filled polymer sheet and it is observed th at a closed foam structure with macro pores

PAGE 30

14 14 from mechanical machining occur in these pa ds. The microstructure of pad of Type 4, which is an unfilled textured polymer sheet, ha s a non-porous structure. All four types of pads are polyurethane based but are manufactured differen tly and hence exhibit varied properties. SEM images of these various pads are available in literature ( 16 ) and clearly show the different microstructure characteristics described above. 2.1.2. Pad properties and polishing performance The pad properties influence the polishi ng of the wafer during the CMP process. The material of the pad should be chemica lly stable and not re act with the slurry chemicals. The pad material should be such th at it does not get affected by the slurry pH, which may be acidic or alkaline based on th e process. The pad should also possess high surface tension in order to promote hydrophi licity for wetting purposes. The hydrophilic nature of the pads aids in the formation of thin film of slurry between the wafer and the pad surface enabling the entire su rface of the wafer to be equally subjected to the slurry chemistry. Mechanically, the pad is required to have high strength, acceptable levels of hardness, and good abrasion resistance. These prop erties can be related to the structure of the pad at the macro and micro levels. At a macro-level, the grooves on the pad surface assist the slurry flow across the surface of the pad and also reduce the potential of hydroplaning. At the micro-scal e, the pad roughness and porosity are also seen to influence the removal rate. With water acti ng as a plasticizer, pad modulus and hardness decrease in wet pads, while there is an increase in the flexibility and ductility ( 16 ). The removal rate depends on the pad m acroand micro-structures. The macrostructures refer to the groove s in the pad to prevent hydropl aning. In the absence of the macro-structure, type 4 pads show no removal rates ( 17 ) while type 3 perforated pads

PAGE 31

15 15 have shown higher and stable removal rates. Removal rate is also seen to depend on the pad microstructure, which relates to the pa d porosity. Type 4 pads, without any porosity show no removal( 18 ). Along with the removal rate, the wa fer planarity is also affected by the pad properties such as hardness and stiffness ( 16, 19-22 ). While the stiffness of the pad promotes planarity, the pad hardness is more related to defects in the wafers in the form of scratches. The planarity of the wafer is also seen to increase with the decrease in pad roughness ( 23 ). Though the pad properties aff ect the polishing, it is difficult to completely understand the relationships as there are many variables involved in the process. In order to maintain the pad structure to sustain removal rates, pad conditioning is a basic requirement. Studies have shown that in the absence of conditioning, the removal rates go down. In the absence of a regenerati on process, the properties of the polishing pad decline with a marked influence on the material removal rate and polishing uniformities. During polishing, the pad surfaces get glazed. This glazing reduces the pad porosity and roughness, which affects the slurry transport and the c ontact area between the pad and the wafer. All these factors re sult in reducing the e fficiency of the CMP process. In order to achieve higher removal rates and planarit y, regeneration of the pad is a necessity. This is done by a process known as conditioning where the pad surface is abraded to open up the clogged pores and break the glazed areas. 2.2. Pad conditioning As discussed earlier, pad conditioning is an important process to restore the pad properties that deteriorate over time during polishing. The surface interactions involved

PAGE 32

16 16 in the process of polishing are influenced by the pad texture resulting from conditioning. The process of conditioning is used to: 1. maintain the roughness of the pad and pr omote effective slurry distribution 2. remove unwanted products after polishing. Various pad conditioning methods have been used to improve the pad properties and stabilize the removal ra tes. The most effective method found was using diamond as the abrasive material. The properties of the conditioner such as diamond density and diamond mounting play a major role. The ot her input variables for the conditioning process include parameters such as conditio ning downforce and relative speed of rotation (rpm) of the pad platen and the conditioner. The process of conditioning can be quantified in terms of material removal rate (MRR), pad roughness, and wear of pad. It has been found that a pad conditioned before the first polished wafer doubled the removal rate compared to the unconditioned pads. A repeated conditioning also showed a stable removal rate ( 16 ). The conditioning maintains the removal rate by maintaining the asperity height and density on the pad surface ( 24 ). Though the properties of the glazed pad impr ove with conditioning, there exists a side effect of pad wear. An uneven pad wear resu lts in uneven distribution of the pressure affecting the planarization uniformity and removal rate. During the process of conditioning, the conditi oner disk rotates about its axis and simultaneously moves linearly towards and away from the center of the pad for a uniform conditioning of the pad surface. The conditioni ng time has to be adjusted so as to maintain the pad thickness. Kinematic equations have been used to model the conditioning processes ( 25 ).

PAGE 33

17 17 2.2.1. Conditioners Conditioners or conditioning disks are used as a medium to present the conditioning abrasive to the pad surface in a way that the desired conditioning effects are achieved. The abrasives present in the slurry ar e extremely hard. In order to prevent wear of the conditioners due to the presence of such abrasives on the pad, it is necessary to use diamonds as abrasives on the conditioner. The conditioning disks are available in a variety of configurations ra nging from solid disks to open patterns. A variety of ways exist to bind diamonds to the surface of conditioning disks. The manner in which diamonds are bound to the surface of the conditioning disks has a bearing on the disk lifetime and the extent of damage caused to wafers due to scratches. The shape of the abrasive as it protrudes from the surface ma y be controlled by the choice of the diamond. All these factors affect either the conditioni ng process or the life of the conditioning disk. The conditioning effect is influenced by the profile of the diamond, density of the diamond, the downforce, and the shear velocity. The diamond may either be placed in an orderly fashion across the c onditioner disk or they may be randomly placed on the surface of the conditioning disk. Besides the abra sive shape, the effect of the conditioning profile is dependent on the degree of protrusion of the diamond from the conditioning disk. If the degree of protrusion is high, a large torque is experienced by the adhesive mounting surface bond which may lead to higher instability.

PAGE 34

18 18 2.3. Research focus Extensive work ( 26-29 ) has been done in order to understand the effect of conditioning on the polishing pad and polishing rates. As discussed in the previous section, the diamond abrasives present on th e conditioning disks play a major role. Therefore, effect of diamond grit size and conditioning force on the polishing pads has been investigated in literature. For example, interferometry analys is was conducted on IC 1000 pads that were conditioned by 100-gr it and 325-grit diam ond conditioners under varying force and incremental loading. Larger diamonds were seen to create a rougher pad surface under greater force. The other impor tant observation that was made was with regards to the decay lengths. It was conclude d that wet pads showed higher decay lengths than the dry pads ( 30 ). In contrast to the literature studies, the objective of this research is to investigate the effect of diamond abrasive size under th e same loading conditions. The goal is to understand the impact on the pad morphology and subsequently the effect of conditioning on the polished wafer. In addition, in this thes is the effect of durat ion of conditioning is explored in a systematic manner. 2.4. Experimental setup The results presented in this chapter focus on the effect of different abrasive sizes on the pad morphology. The experiments were conducted on a bench top CMP tester manufactured by CETR, Inc. (Figure 2.1). Th e CMP process on this bench top polisher closely imitates the large wafer fabrication pr oduction processes in i ndustries. As can be seen from Figure 2.1, the bench top CMP tester consists of an upper platen, which can

PAGE 35

19 19 hold a two inch wafer. The upper platen itself is held in place by a carriage. The carriage rotates about its own axis at a set rpm (typically 100-150) during polishing. A strain gauge force sensor (0-200 N) is present in the carriage that records both vertical and friction forces. This helps in monitoring the co efficient of friction dur ing the process. An acoustic emission (AE) sensor is also present in the carriage that ha s a frequency range of 0.5 to 5 KHz. An AE event occurs whenever there is a rapid release of elastic energy within any material. This rapi d release of energy propagates an elastic wave that can be detected and analyzed using appropriate se nsors. In the case of the bench top CMP process, the AE sensor helps in capturing the acoustic emission from the contact of rubbing surfaces which is an important parameter of friction and wear. The CMP tester also consists of a lower platen on which a six inch polishing pad is mounted. The lower platen rotates at a set rpm during the process and the set rpm ranges from 100 to 200. The slurry flows over th e pad during the polishing process. Pipes are provided for the inflow and outflow of th e slurry. The bench-top tester is controlled through a computer. 2.5. Conditioning experiments Towards the objective of evaluating the effect of abrasive size on the pad morphology, three parameters were varied. Thes e parameters include abrasive size, time of conditioning, and rotational sp eed of the polishing pad. The polishing pad used was IC 1000 and sub-pad used was SUBA IV. Deionize d (DI) water was used as fluid in these experiments.

PAGE 36

20 20 Five different sizes of diamond abrasives ranging from 0.25 m to 100 m were chosen to carry out the conditioning on the commercially available polishing pad. The different sizes of diamond abrasives used were 0.25 m, 2 m, 8 m, 68 m and 100 m. Though all these conditioners appe ar alike to the naked eye, the different diamond sizes can be clearly distinguished in corresponding SEM images as seen in Figure 2.2 through Figure 2.11. In order to evaluate the infl uence of abrasive size on pad morphology, conditioning was carried out fo r four different times: five, ten, fifteen, and twenty minutes. Thus, for each size of abrasive, f our different experime nts were conducted. In addition, each experiment was conducted at two rotation speeds (150 rpm and 200 rpm) of the pad. The results of these experime nts are discussed in the next section. 2.6. Results of conditioning experiments In these experiments, in-situ coeffici ent of friction was monitored. The pad surfaces after conditioning were characterized using SEM, AFM, Dynamic Mechanical Analysis (DMA) and profilometry. The pads were evaluated with respect to their morphology, roughness, and storage modulus. The hypothesis in these experiments is that the size of the diamond abrasives affects the degree of abrading as long as the conditioning time remains the same. 2.6.1. Pad morphology Figure 2.13 shows the morphology of the pad conditioned with an abrasive size of 0.25m. In this case the pad has been conditi oned for five minutes at 150 rpm. From this image, it can be observed that the pad mo rphology remains relatively the same as compared to the pad morphology prior to conditioning as in Figure 2.12. Similarly,

PAGE 37

21 21 Figure 2.14 shows the morphology of the pad having been conditioned with an abrasive size of 100m. In this case, the pad has been conditioned for five minutes at 150 rpm. On comparing the morphology of the conditione d pad to the morphology of the pad prior to conditioning, it can be seen that signi ficant abrasion has occurred. On comparing Figure 2.13 and Figure 2.14, it can be inferred that for the same conditioning time and the same rotation speed, the degree of abrasion is significantly more for a higher abrasive size. Figure 2.15 to Figure 2.19 show the mor phology of the pads that have been conditioned for five minutes us ing abrasives with sizes of 0.25m, 2m, 8m, 68m, and 100m at 150 rpm. These images are an indicatio n that as the abrasive size increases, the degree of abrading increases. Figure 2.20 to Figure 2.24 show the pads conditioned for five minutes at 200 rpm. These results show the same trend as shown by pad morphologies conditioned at a lower rpm. On comparing the two sets of conditioning results, it can be concluded that for same conditioning time, the rpm does make a significant difference on the pad morphologies. However, since five minutes may not be enough to draw conclusions, the conditioning process was carried out for the same abrasive sizes mentioned a bove but for different times. Having conducted the conditioning experiment for five minutes, the second step is to conduct the conditioning e xperiment for times of 10, 15, and 20 minutes at 150 and 200 rpm. Figure 2.25 to Figure 2.29 show the morphology of the pad that has been conditioned for twenty minutes for abrasive sizes of 0.25m, 2m, 8, 68m, and 100m at 150 rpm. On comparing Figure 2.15 and Figur e 2.25, it can be seen that as the time of conditioning increases from five to twenty minutes, the degree of abrading is

PAGE 38

22 22 significantly more. The same is the case with the 100 abrasive size. This result can be inferred by comparing Figure 2.19 and Figure 2.29. Figure 2.30 through Figure 2.34 show the pad conditioned for twenty minutes at 200 rpm. On comparing these results to the previous conditioning results, the maximum pad abrasion is seen in this set of results because of the combination of higher platen speed and longer abrasion time. The pad surfaces conditioned for twenty minutes at 200 rpm show a more uniform abrasion. The smaller abrasive do not show much change in the pad morphology for a short duration but with longer condi tioning time, pad abrasion is seen clearly whereas the effect of bigger abrasives becomes clear even with a short conditioning time. Again, a higher rpm shows more pad abrasion than a lower rpm. Though, ther e does not exist a wide gap in the rpm values chosen here, the difference becomes prominent for longer conditioning duration. 2.6.2. Pad wear The reduction in thickness of the pad as the conditioning proceeds is defined as pad wear. The conditioner has to maintain constant contact with the pad throughout the conditioning process. In order to maintain constant contact, the carriage automatically comes down by the same amount by which th e thickness of the pa d decreases. The distance by which the carriage travels downward is a measure of the amount of wear in the pad. Figure 2.35 and Figure 2.36 show plots of the amount of pad wear against time for different abrasive sizes at a pad rota tion speed of 150 and 200 rpm. From Figure 2.35 at 150 rpm, it is observed that for larger abrasi ve, the amount of wear in the pad is higher

PAGE 39

23 23 than at smaller abrasive. For example, for an abrasive size of 100m the amount of wear at five minutes is 0.018m whereas for an abra sive size of 0.25m, the amount of wear is 0.006m. As the time for conditioning increases, the amount of wear also increases. For example, when the abrasive size is 10 0m, the amount of wear is 0.026m for a conditioning time of 10 minutes, and 0.03m for a conditioning time of 20 minutes. Similarly, when the abrasive size is 0 .25m, the amount of wear is 0.008m for a conditioning time of 10 minutes, and 0.01m fo r a conditioning time of 20 minutes. The same trend is observed in Figure 2.36, when the pad rotation speed is set at 200 rpm. However, the amount of wear is more at 200 rp m as compared to the amount of wear at a rotation speed of 150 rpm. For example, at 150 rpm, the amount of wear when using 8m abrasive size for a conditioning time of 20 minutes is 0.015m, whereas at 200 rpm it is 0.016m. As bigger abrasives abrade the pa d to a higher degree than the smaller abrasives, the pad wear caused due to this abrasion is also more in the case of bigger abrasives. Higher the pad wear, shorter is the pad life. An uneven pad wear also results in uneven slurry transport and non-uniform polishing. 2.6.3. Coefficient of friction Having looked at the pad morphology and pad wear, the next step is to evaluate the coefficient of friction resulting from the different abrasive sizes. During the conditioning process, the lower platen exerts a normal for ce equivalent to the downward force experienced by the pad. There is a shear force also present between the pad and conditioner. The coefficient of friction betw een the pad and the conditioner is computed as a ratio of the shear fo rce to the normal force.

PAGE 40

24 24 Figure 2.37 and Figure 2.38 show the plots of coefficient of friction against time for the conditioning runs at 150 and 200 rpm re spectively. From the plots it is observed that as the abrasive size increases, the coe fficient of friction increases. For example, in Figure 2.37, for an abrasive size of 100 m, the COF is 0.633 at 150 rpm when the time for conditioning is five minutes, whereas, the COF is 0.32 for an abrasive size of 0.25 m. Similarly, for 100 m abrasive size, when the time for conditioning is 20 minutes, the COF is 0.645, whereas for a conditioning size of 0.25m the COF is 0.47. Figure 2.38 shows a plot of coeffici ent of friction against time for the conditioning run at a pad rotation speed of 200 rpm. In this case as well, the coefficient of friction for larger abrasive si ze is generally higher than th e coefficient of friction for a smaller abrasive size. As discussed above, pad wear is more at higher pad rotation speeds than at lower pad rotation speeds and so there is a decrea se of shear force between the pad and the conditioner. This further implies that the coe fficient of friction should decrease as the pad rotation speed increases as seen in the plots. For an abrasi ve size of 100m, the average coefficient of friction at 150 rpm is 0.624 whereas at 200 rpm the average coefficient of friction is 0.589. Similarly, when the abrasive size is 2m, the average COF at 150 rpm is 0.44 whereas at 200 rpm, the COF is 0.40. 2.6.4. Pad roughness The roughness of the pad affects the rem oval rate as the wafer being polished comes in direct contact with the rough su rface of the pad (31). Having evaluated the coefficient of friction, the next step is to evaluate the roughness of the pad. This was done using a contact mode profilometer with a stylus of radius 12 m. A standard scan was

PAGE 41

25 25 conducted over a 50 m scan length. There were five measurements for each pad. The average roughness measurements of each pad ar e plotted against the abrasive size of the conditioner used for that pad. Figure 2.39 shows the plot of pad roughness with respect to time for different abrasive sizes at 150 rpm. It can be seen fr om the plot that the roughness increases with time. For the smaller abrasive sizes of 0.25m, 2m and 8m the increase in pad roughness with time is not much. For a condi tioning time of 5 minutes, the pad roughness values were 0.775, 1.03 and 1.26 m for abrasi ve sizes of 0.25m, 2m and 8m, respectively. For a conditioning time of 20 minutes, the roughness values for 0.25m, 2m and 8m abrasives are 1.23, 1.28 and 2.05 m, respectively. The conditioning with bigger abrasives results in higher roughness co mpared to the smaller abrasives. For 15 minute conditioning, the pad roughness value for 68m and 100m are 2.26 and 2.38m, respectively. As can be seen from the plot, for a conditioning run of 20 minutes, the pad roughness for 68m is still increasing while for 100m the pad roughness value seems to decrease. For 100m abrasive for 20 minutes, as the pad abrasion is relatively high, this could result in rupturing of the pad surface rather than just restoring the roughness. Figure 2.40 shows the plot of pad roughness with respect to time for different sizes at 200 rpm. On comparing the roughness va lues obtained at 200 rpm with those at 150 rpm, it is observed that the pad roughness va lues at 150 rpm are hi gher than those at 200 rpm. This can be because that at a hi gher rotational speed of 200 rpm, the pad does not get abraded uniformly.

PAGE 42

26 26 2.6.5. Dynamic Mechanical Analysis Having investigated the roughness of the pads, dynamic mechanical analysis was conducted to evaluate the rela tion between structure and prop erties of the pad. DMA is used to study the mechanical properties of pol ymers and analysis is based on the principle that viscoelastic materials dissipate stor ed mechanical energy when deformed ( 32 ). DMA studies on CMP pads have been perf ormed to understand the structure property relations. The storage and loss modulus give an insight into th e elastic and viscous nature respectively of the viscoelastic materials. The ratio of these two moduli is given by tan This ratio is useful during the characterization of polymers since it is related to the material’s ability to dissipate energy in the form of heat ( 33 ). For this experiment, the pad that wa s conditioned with 100m abrasive size conditioner for 20 minutes at a pad rota tion speed at 200 rpm was used. Samples (35 mm X 15 mm) were cut from each pad. In or der to facilitate efficient clamping of the pad coupon to the holder, the bottom glue la yer was scrubbed off of the coupon. The data was obtained on a TA Instruments DMA 2980 (New Castle, DE) at 4oC with an isothermal time of one mi nute per increment starting from room temperature (30oC) and going up to 80oC. A single cantilever clamp was used in the flexural mode with amplitude of 3m. The data was collected at six different fre quencies; 0.6, 6, 12, 30, 60, and 100 Hz. As the pads conditioned for twenty minutes at 200 rpm showed the maximum change in terms of pad mor phology, pad wear and pad roughness, DMA was carried out for this set of conditioned pads. Figure 2.41 and Figure 2.42 show the plots of storage modulus of the pad against temperature at 30Hz and 100Hz respectively. The plot shows the storage modulus of pads

PAGE 43

27 27 conditioned with conditioners of abrasive si zes 0.25m, 2m, 68m, and 100m. As can be seen from the plot, the storage modulus doe s not vary much for the different pads. For example, the storage modulus of the pad conditioned with 2m abrasive size conditioner is higher than that of the pad conditioned using a 0.25m conditioner. However, for the pad conditioned using conditioner of 68m abra sive size, the storage modulus was higher compared to all other pads. Excessive wear of the pad conditioned using 100m abrasive conditioner may have resulted in a lower storage modulus. The pad conditioned with 8m seems to show abnormally lower storage modulus compared to the other conditioned pads. Th e number of punched holes in this sample were about 18 while in the other samples, th e number of holes were about 12-14. So, the sample used for testing was less stiff compared to the others. 2.7. Polishing experiments In this experiment, the same CETR be nch-top tester was used as in the conditioning experiments. Polis hing differs from conditioning in that the upper platen holds a wafer instead of a conditioner. The obj ect of interest is the wafer being polished rather than the pad as in conditioning. Th e conditioned pad acts as a polisher in the presence of slurry and smoothens the wafer. He nce, the pad plays an important role in the outcome of the polishing process. Since th e pad morphology is affected by the abrasive size, the abrasive size indirectly affects th e surface of the wafer. Hence, the set of experiments described here inve stigate the affect of using pa ds of different morphologies that resulted from different abrasive si ze conditioners on the surface of the wafer. As mentioned earlier, the objective of these ex periments is to evaluate the effect of abrasive size on the surface of the polished wafer. In order to do so, three different

PAGE 44

28 28 abrasive sizes were used; 8m, 68m, a nd 100m. The polishing pad used was IC 1000 and sub-pad used was SUBA IV. The sl urry used during polishing was Cabot 5001, which is commercially available. The results obtained for these expe riments are discussed in the next section. 2.8. Results of the polishing experiments After conducting the experiments, an optic al microscope was used in order to obtain optical images of the polished wafers to evaluate surface defects such as microscratches. The topography of the wafer was ev aluated using an AFM. In a manner similar to the conditioning experiments, the in-situ coefficient of fric tion was monitored. SEM images of the pad were taken after polishi ng in order to evalua te the pad morphology. 2.8.1. Optical images of the polished wafer For the polishing experiment, having po lished a wafer using a pad, the pad was conditioned again before reusing it for the next wafer. This process was repeated for three wafers. Figure 2.43 and Figure 2.44 show the optical images of the first and third wafers that were polished using pads conditione d by an 8m abrasive size conditioner. Similarly, Figure 2.45 and Figure 2.46 show the optical images of the first and third wafers that were polished using pads condi tioned by a 68m abrasive size conditioner. Optical images of the first and third wafe rs polished using a pad conditioned by 100m abrasive size conditioner are s hown in Figure 2.47 and Figure 2.48. On comparing images of the first wafe r polished by the pads conditioned using the three different abrasive size conditioners, it is observed that the pad conditioned using the 100m abrasive size conditioner produces a large number of scratches on the wafer.

PAGE 45

29 29 A lot of scratches developed ar e in clusters and the scratche s are small in size. Similarly, when the pad conditioned using a 68m cond itioner is used, scratches are produced in clusters. However, the number of scratches in each cluster is relatively less than that when using the 100m conditioner. The scratc hes themselves are relatively longer when compared to those produced by the 100m c onditioner. The pad conditioned using the 8m conditioner produces relatively longer sc ratches when compared to the scratches produced by the pads conditioned using the 68m and 100m conditioners and unlike the other wafers, the scratches are isolated. From the optical images obtained fo r the third wafer polished by the pad conditioned using the 100m ab rasive size conditioner, it is observed that the pad produces less number of scratches on the thir d wafer as compared to the first wafer. Similarly, the pad conditioned using the 68 m abrasive size conditioner produces less number of scratches on the third wafer as comp ared to the first wafer. However, the pad conditioned using the 8m abrasive size cond itioner produces relatively the same number of scratches on the firs t and third wafers. 2.8.2. Evaluation of topography Atomic Force Microscope was used to evaluate the wafer and pad topography. For this purpose, a Digital Instruments DimensionTM 3100 AFM was used and operated at 256 Hz frequency of the cantilever. Wafer Topography: For this experiment, the roughness measurements were taken at different points across the su rface of the wafer from the center of the wafer to its edge. Figure 2.49 and Figure 2.50 show the AFM images of the first and third wafers that were polished using pads conditioned by an 8m abra sive size conditioner. Similarly, Figure

PAGE 46

30 30 2.51 and Figure 2.52 show the AFM images of the first and third wafers that were polished using pads conditioned by a 68m abrasive size conditioner. AFM images of the first and third wafers polished using a pad conditioned by 100m abrasive size conditioner is shown in Figure 2.53 and Figure 2.54. On comparing the AFM images of the fi rst and third wafer polished using a pad conditioned with a 100m abrasive size conditio ner, it can be observed that the first wafer is rougher. The roughness value of the third wafer is 34 nm whereas the roughness of the first wafer is 38 nm. Similarly, when using a pad conditioned with an 8m abrasive size conditioner, the first wafer has a roughness of 83 nm while the third wafer has a roughness of 70 nm. On comparing the roughness values of the wafers polished using each of the three pads, it is observed that the roughness value of the wafer polished by the pad conditioned using the 100m conditioner is significantly le ss compared to the wafers polished using the other two pads. This implies that it is po ssible to achieve more planarization using the pad conditioned with the 100m abrasive size conditioner. Pad Topography: The AFM images of the pa ds used in polishing the wafers are shown in Figure 2.55 to Figure 2.57 for the abrasive size of 8m, 68m and 100m respectively. On comparing the images of the different pads used in the polishing process, it is observed that the pad conditioned using the 100m abrasive size conditioner has higher roughness values as comp ared to the other two pads. 2.8.3. Pad morphology Figure 2.58 shows the morphology of the pad conditioned with an abrasive size of 8m. As can be seen, this pad does not exhibit uniform conditioning. Figure 2.59 shows

PAGE 47

31 31 the morphology of the pad conditioned with an abrasive size of 68m and Figure 2.60 show the pad conditioned with a 100m conditioner. From these images, it can be observed that the higher abrasive size conditions the pad more uniformly. 2.8.4. Coefficient of friction Having looked at the pad morphology, the next step was to evaluate the COF resulting from using different abrasive si zes. Figure 2.61 to Figure 2.63 show plot of coefficient of friction against time for the polished wafer conditioned on pad with abrasive size of 8m, 68m and 100m. Fo r an abrasive size of 8m, the COF is 0.6045. COF for an abrasive size of 68m is 0.2690 as seen in Figure 2.62. For an abrasive size of 100m, Figure 2.63 shows th at the COF is 0.407. As can be seen, the COF plots for 68m and 100m seem to follow the same patte rn with high variation in the beginning and then the variations ar e reduced with time. For 8 m abrasive, COF plot shows increasing variations with time. This implie s that a polishing run of three minutes may not be sufficient for this abrasive size. If th e process were allowed to run longer, the COF value may decrease.

PAGE 48

32 32 Figure 2.1: CETR bench top tester Carriage Slurry out Lower platen holding the polishing pad Wafer Carrier Force Sensor Slurry feed

PAGE 49

33 33 Figure 2.2: Photograph of a 0.25m abrasive conditioner Figure 2.3: SEM image of a 0.25m abrasive conditioner Figure 2.4: Photograph of a 2m abrasive conditioner Figure 2.5: SEM image of a 2m abrasive conditioner Figure 2.6: Photograph of a 8m abrasive conditioner Figure 2.7: SEM image of a 8m abrasive conditioner

PAGE 50

34 34 Figure 2.8: Photograph of a 68m abrasive conditioner Figure 2.9: SEM image of a 68m abrasive conditioner Figure 2.10: Photograph of a 100m abrasive conditioner Figure 2.11: SEM image of a 100m abrasive conditioner

PAGE 51

35 35 Figure 2.12: SEM image of a fresh pad Figure 2.13: SEM image of pad conditioned using 0.25m abrasive Figure 2.14: SEM image of pad conditioned using 100m abrasive

PAGE 52

36 36 (a) (b) Figure 2.15: SEM images of pad conditioned with 0.25m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.16: SEM image of pad conditioned with 2m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.17: SEM image of pad conditioned with 8m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X

PAGE 53

37 37 (a) (b) Figure 2.18: SEM image of pad conditioned with 68m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.19: SEM image of pad conditioned with 100m abrasive for 5 min at 150 rpm at two different magnifications (a)150X (b)300X

PAGE 54

38 38 (a) (b) Figure 2.20: SEM image of pad conditioned with 0.25m abrasive for 20 min at 150 rpm at two different magnifications(a)150X (b)300X (a) (b) Figure 2.21: SEM image of pad conditioned with 2m abrasive for 20min at 150 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.22: SEM image of pad conditioned with 8m abrasive for 20min at 150 rpm at two different magnifications (a)150X (b)300X

PAGE 55

39 39 (a) (b) Figure 2.23: SEM image of pad conditioned with 68m abrasive for 20 min at 150 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.24: SEM image of pad conditioned with 100m abrasive for 20 min at 150 rpm at two different magnifications (a)150X (b)300X

PAGE 56

40 40 (a) (b) Figure 2.25: SEM image of pad conditioned with 0.25m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.26: SEM image of pad conditioned with 2m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.27: SEM image of pad conditioned with 8m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X

PAGE 57

41 41 (a) (b) Figure 2.28: SEM image of pad conditioned with 68m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.29: SEM image of pad conditioned with 100m abrasive for 5 min at 200 rpm at two different magnifications (a)150X (b)300X

PAGE 58

42 42 (a) (b) Figure 2.30: SEM image of pad conditioned with 0.25m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.31: SEM image of pad conditioned with 2m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.32: SEM image of pad conditioned with 8m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X

PAGE 59

43 43 (a) (b) Figure 2.33: SEM image of pad conditioned with 68m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X (a) (b) Figure 2.34: SEM image of pad conditioned with 100m abrasive for 20 min at 200 rpm at two different magnifications (a)150X (b)300X

PAGE 60

44 44 Figure 2.35: Plot of pad wear at different conditioning times for different abrasive sizes ( m) at 150 rpm Figure 2.36: Plot of pad wear at different conditioning times for different abrasive sizes ( m) at 200 rpm

PAGE 61

45 45 Figure 2.37: Plot of COF at different conditioning times for different abrasive sizes (m) at 150 rpm Figure 2.38: Plot of COF at different conditioning times for different abrasive sizes (m) at 200 rpm

PAGE 62

46 46 Figure 2.39: Plot of roughness at different conditioning times for different abrasive sizes ( m) at 150 rpm. Figure 2.40: Plot of roughness at different conditioning times for different abrasive sizes ( m) at 200 rpm

PAGE 63

47 47 Figure 2.41: Plot of storage modulus at a frequency of 30 Hz for pads conditioned for 20 min with different abrasives Figure 2.42: Plot of storage modulus at a frequency of 100Hz for pads conditioned for 20 min with different abrasives

PAGE 64

48 48Table 2.1: Experimental conditions for wafer polishing Parameter Value Slurry flow rate 75 ml/min Polishing Pad Rodel IC 1000 SUBA IV Wafer polished 2 inch blank copper wafers Polishing time 3min Platen speed 200 rpm

PAGE 65

49 49 Figure 2.43: Optical image of first wafer polished on the pad conditioned with 8m abrasive at 20X Figure 2.44: Optical image of third wafer polished on the pad conditioned with 8m abrasive at 20X Figure 2.45: Optical image of first wafer polished on the pad conditioned with 68m abrasive at 20X Figure 2.46: Optical image of third wafer polished on the pad conditioned with 68m abrasive at 20X Figure 2.47: Optical image of first wafer polished on the pad conditioned with 100m abrasive at 20X Figure 2.48: Optical image of third wafer polished on the pad conditioned with 100m abrasive at 20X

PAGE 66

50 50 Figure 2.49: AFM image of the first wafer polished on a pad conditioned with 8m abrasive Figure 2.50: AFM image of the third wafer polished on a pad conditioned with 8mabrasive Figure 2.51: AFM image of the first wafer polished on a pad conditioned with 68m abrasive Figure 2.52: AFM image of the third wafer polished on a pad conditioned with 68m abrasive Figure 2.53: AFM image of the first wafer polished on a pad conditioned with 100m abrasive Figure 2.54: AFM image of the third wafer polished on a pad conditioned with 100m abrasive

PAGE 67

51 51 Figure 2.55: AFM image of pad conditioned with 8m abrasive Figure 2.56: AFM image of pad conditioned with 68m abrasive Figure 2.57: AFM image of pad conditioned with 100m abrasive

PAGE 68

52 52 Figure 2.58: SEM image of pad conditioned with 8m abrasive Figure 2.59: SEM image of pad conditioned with 68m abrasive Figure 2.60: SEM image of pad conditioned with 100m abrasive

PAGE 69

53 53 020406080100120140160180200 Time sec 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 COF-Ff COF-Ff 0.6045 Figure 2.61: Plot of COF for abrasive size of 8 m 020406080100120140160180200 Time sec 0 0.1 0.2 0.3 0.4 0.5 0.6 COF-Ff COF-Ff 0.2690 Figure 2.62: Plot of COF for abrasive size of 68 m

PAGE 70

54 54 020406080100120140160180200 Time sec 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 COF-Ff COF-Ff 0.4077 Figure 2.63: Plot of COF for abrasive size of 100 m

PAGE 71

55 55 Chapter 3 Optical Measurement of the Contac t Area between the Pad and Wafer 3.1. Concept of contact area and its importance As described in chapter 2, the polishing pa d is the platform on which the process of polishing is carried out. The wafer to be polished is held face down on the pad while the abrasive slurry flows between these 2 surfaces. As can be seen from Figure 3.1, the pad and the wafer come in contact at few points. The total area under these points of contact is known as the contact area. Since the pad is a polymer material, it does not have a very smooth or planar structure (Fi gure 3.1). The micro-hills and valleys on the pad material determine the contact area between the pad and the wafer. Some of the other factors affecting the contact area include dow n pressure, slurry-particle size, and waferpad rotating speed ( 34 ). The wafer-pad-slurry interacti ons can be understood better on the basis of microand nano-scale effects. At the micro-scale, particle ba sed slurries interact with the pad and the wafer. The active partic les in the slurry, tra pped between the wafer surface and the pad cause the removal of materi al. At the nano-scale, the formation of a surface layer by the slurry chemicals and its re moval as a result of abrasion, leads to the polishing effect( 35 )

PAGE 72

56 56 Various techniques have been developed li terature for measur ing the roughness of the pad. For example, the Greenwood-Williamson elastic model ( 36 ) has been used to measure the roughness of the pad. This model is based on two assumptions. First, it is assumed that asperities have spherical shap e. Second, it is assumed that the height and radius of asperities follow a Gaussian distribution. It was found that the total ar ea can be computed using the equation shown below. z d0Aaddz where, is the density of the asperities; z is the height of the asperities; is the radius of the asperities; and z are Gaussian distribution functions. The use of Fourier Transform Infrared Spectroscopy / Attenuat ed Total internal Reflection spectroscopy (FTIR/ATR) for the measurement of contact area has been explored in literature ( 37 ). In this technique, the area unde r the peaks of the spectra is measured to evaluate the intensities of th e peaks. The intensity values were then compared with the intensities resulting fr om a 100% pad contact. A 100% pad contact implies that the pad used is non-porous, defect free and has a flat surface. This helps in determining the percentage of area contact. The results obta ined using this technique were found to be similar to the results calcu lated by using the Green wood-Elastic Model. Figure 3.2 shows a plot of %contact area of the pad ag ainst applied load. The down pressure on the pad is evaluated by divi ding the load applied to the wafer by the area of the wafer. The results obtained were found to be in agreement with the results obtained from the calculated resu lts of Yu et al. From the gr aph it can be inferred that the contact area of the pad increases as th e down pressure experienced by the wafer

PAGE 73

57 57 increases. It may be inferred that as the down pressure increases, the area of contact increases which results in a higher de gree of polishing of the wafer. Evaluating the total area of contact be tween the pad and the wafer can give valuable information about the polishing run. Th e contact area information can be used to draw inferences about the effectiveness of the conditioning process. Amongst other factors, contact area depends upon the pad mo rphology or structur e resulting from the process of conditioning. As was seen in Chap ter 2, diamond abrasive size affects the pad morphology. Hence it can be inferred that, the diamond size on the conditioner affects the contact area between the pad a nd the wafer during polishing. In order to evaluate the contact area, it is necessary to evaluate the 3D topography of the pad. Profilometry and laser interferomet ry are two techniques that can be used for evaluating the 3D topography of the pad. T hough profilometry gives great images, it is limited by the fact that is a contact type method. On the other hand, interferometry, an optical technique requires the pad to be reflective in natu re for measurement of the contact area, which again limits the use of this technique for the measurement purpose. Different factors affecting the contact area between the pad and the wafer have been investigated in literature. For example, confocal reflectance interference contrast microscopy (C-RICM) was used to evaluate the contact area ( 38 ). This technique uses light interference to recognize and measure the contact areas. A sapphire cover slip was used on a hard pad and then a soft pad made of porous polyurethane pad material. It was concluded that the contact area increases with the increase in the applied load. It was further observed that the conditi oner type also affected the co ntact area and that different pads respond differently to conditioning. This is an ex-situ measurement.

PAGE 74

58 58 Dual Emission Laser Induced Fluorescence (DELIF) has been used to study the interactions between th e pad and the wafer ( 39 ). It can also be used to measure the in-situ thickness of the slurry. DELIF has been used in the past to study the pad wafer interaction in terms of average fluid layer thickness, compressi bility of the asperity layer on the pad and measurement of surface roughness ( 40 ). This technique is based on the fluid layer profile formed at the interface of the pad and the wafer. At the contact points of the pad and wafer, the slur ry thickness is zero. DELIF uses this principle to determine the area of contact. The per centage of contact area f ound using the DELIF method has been found to be in agreement with that found using the C-RICM method discussed above. 3.2. Measurement of contact area This part of the thesis focuses on deve loping a simple method of measuring the contact area in contrast to the various techniques discu ssed above. The advantage of developing such a technique is that the pro cess can be understood better and faster in order to observe the changes in the contact area as the pol ishing process progresses. The information that is obtained as a result of evaluating the contact area between the pad and the wafer provides a motivation for the research here. As part of this research the contact area is measured at different pressures while comparing the change in contact area due to changes in the applied pressures.

PAGE 75

59 59 3.3. Experimental details The preliminary research involved l ooking at the pad morphology under the optical microscope. The pad used for this experiment was IC 1000 which was conditioned earlier using a diamond conditioner. The experimental setup, as shown in Figure 3.3, consisted of a pressure cell. The cell consisted of a circular di sc with a cylindrical blind hol e at the center. A cylindrical spacer was placed in the hole such that the t op face of the spacer was flush with the top surface of the cylindrical disc. A small rect angular piece of the IC 1000 pad was cut and placed on top of the spacer. The size of the pad was such that it was smaller than the top surface of the cylinder. The pad was then cove red by a cover slide. The cover slide acts as the wafer being polished. The idea behind building this apparatus was to simulate the down pressure experienced by the wafer in a CMP process a nd observing the changes in the contact area of the pad. The operating principle of the appa ratus is that as the vacuum between the cover slide and the spacer increases, the dow n pressure experienced by the cover slide increases. As the vacuum present in the spac e between the cover slid e and the spacer was varied using a vacuum gauge, changes were observed in the contact area of the pad. To begin with, a dry pad was placed on t op of the spacer and was covered with the cover slide. The dry pad was observed unde r the optical microscope to evaluate its morphology. However, the pad morphology was not visible under dry conditions. Since real-time measurements involve the use of a wet pad, the dry pad was made wet using water and examined under the optical micr oscope. In order to clearly see the pad structure, a drop of pigmentation was added on the pad surface. This was done using a

PAGE 76

60 60 commercially available blue ink. Once the we t pad surface was clea rly seen, a laser was used to view the contact points. Laser sources of different powe r were used to get the best results. The laser was focused on the cover slid e at an angle such th at on hitting the cover slide over the pad, th e laser beam undergoes total intern al reflection. Due to internal reflection, several bright laser points were seen on the pad when it was observed under the optical microscope. Optical images were taken for dry and we t pads at different pressures and analyzed using software called ImagePro developed by MediaCybernetics. 3.4. Results and discussion The optical microscope used for th e study was a Nikon Measurescope UM-2 Tester. This microscope, shown in Figure 3.4, op erates in reflected as well as transmitted light. It also operates in bo th, dark and bright field im aging modes. The contact area research was carried out in dark fiel d in order to obtain clear images. Figure 3.5 shows the optical images of the dry pad whereas Figure 3.6 shows the optical images of the wet pad. The images ar e captured at a magnification of 50X. As can be seen from Figure 3.5, it was difficult to observe the pad morphology when the dry pad was observed under the optical microscope. Be fore introducing the blue ink on the pad, the morphology could not be clearly observed for even for the wet pad as in Figure 3.6. It is not possible to evaluate th e configuration of the pad su rface though the pad surface can be seen in both the images. From these imag es, it is not possible to derive concrete inferences about the pad microstructure, which is an importa nt factor in determining the contact area. With the introduction of pigmentation (blue ink), the pad surface became much clearer as seen in Fi gure 3.7. Various hills and depr essions can be observed from Figure 3.7 which is not clearly seen in Fi gure 3.5 and Figure 3.6. The conditioning trail

PAGE 77

61 61 marks can be distinctly observed in Figur e 3.7. Though an optical image does not deliver as much information as a SEM image (Fi gure 3.8), the optical method is a simpler technique to understand the pad microstructure. Having observed the pad structure under th e optical microscope determining the contact area was the next step. When the laser beam was focused on the cover slide, certain bright spots could be seen under the microscope as seen in Figure 3.9. On observing closely, some of the bright spots showed debris deposited on the pad surface. This is evident from Figure 3.10. Figure 3.11 sh ows the surface of the pad while Figure 3.12 shows the optical image of the same s pot on the pad as in Figure 3.11 but in presence of the laser beam. The laser was most effective in absence of the reflected light as it minimized the interference. Hence, all measurements were taken in the absence of reflected light. As discussed above, the pad-wa fer contact area increases with an increase in down pressure. The same trend was followe d by the observed laser points. On varying the vacuum and thereby varying the down pr essure, the number of bright points also varied. With an increase in the down pressure, the number of bright points increased and with a decrease in the down pressure, the num ber of bright points decreased. This may be observed from Figure 3.13, Figure 3.14 and Fi gure 3.15. Apart from an increase in the number of points, the intensity of the point s also increased. The change in the number and intensity of the laser points wa s used for the image analysis. The software used for purposes of anal ysis was ImagePro. Analysis was first carried out on a dry pad and later on a wet pad using the laser and in absence of reflected light. For the analysis, the first optical im age was taken at a magnification of 20X on a dry pad at a low pressure (F igure 3.16). The second image was taken at a higher pressure

PAGE 78

62 62 (Figure 3.17) and at the same magnification. The number of points in Figure 3.16 is less than the number of points in Figure 3.17. Also, on comparison of the two images, the variation in intensity of the bright spots can be observed. Further, these optical images were analyzed using ImagePro to infer the change in the contact area resulting due to a change in down pressure. The analyzed image is shown in Figure 3.18. The change in intensity of a bright spot due to change in pressure can be observed in the analyzed image. For example, Figure 3.16 shows a point X. Figure 3.17 which is taken at a higher pressure contains the same point X but has a higher inte nsity. Figure 3.18 shows the image obtained by analyzing Figure 3.16 and Figure 3.17 using ImagePro. The point X can be observed in Figure 3.18 but the intensity of the point represents the difference in intensity as a result of the ch ange in pressure. On an averag e, based on the analysis done at different points on the same pad surface and for three different pad samples, the resultant bright spots covered about 0.67% of the entire area, which represents the change in contact area with the change in pressure. For the dry pad, the aver age number of points seen on the surface was twenty five. Average st atistics for the change in contact area are shown in Table 3.1. Similar analysis was conducted for a wet pad. Figure 3.19 shows the image taken at lower pressure and Figure 3.20 shows the image taken at a higher pressure. The resulting difference due to ch ange in pressure may be obs erved in Figure 3.21. In case of the wet pad, the bright spots cover about 0.85% of the entire area, which represents the change in contact area with the change in pressure on changing the pressure. Average statistics for the change in cont act area are shown in Table 3.2.

PAGE 79

63 63 Certain assumptions have been made duri ng this analysis. The flaws in the cover slide have not been taken into considerati on. The defects in the c over slide may reflect some light and cause some bright spots in th e image. But, it was assumed that when the images are being subtracted, these bright spot s tend to nullify. The other factor that has not been accounted for is the presence of de bris which showed up as bright spots. Since the position of the debris doe s not change on changing the pressure, it may be assumed that it cancels on analyzing the images. Howe ver, the intensity of the bright spots representing the debris may change due to change in the down pressure. Therefore, on analyzing the images using ImagePro, the debr is may show up as a bright spot in the analyzed image with a different intensity. Hence the debris may contribute towards the change in contact area. In othe r words, if debris is present, the actual change in contact area may be lesser than the calculated change.

PAGE 80

64 64 Figure 3.1: A schematic diagram depicting contact points between pad and wafer during polishing 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 02468101214 Head Pressure% Pad Contact Figure 3.2: Plot of % pad contact vs. head pressure. The data in the figure is from the literature ( 37 )

PAGE 81

65 65 Figure 3.3: A schematic diagram of the experimental setup for measuring contact area Figure 3.4: Nikon measurescope UM-2 tester

PAGE 82

66 66 Figure 3.5: Optical image of dry pad at 20X Figure 3.6: Optical image of wet pad at 20X Figure 3.7: Optical image of wet pad after introduction of blue ink at 20X Figure 3.8: SEM image of a pad

PAGE 83

67 67 Figure 3.9: Optical image of wet pad with a focused laser at 20X Figure 3.10: Optical image of a pad at 20X showing laser points that represent the debris on the pad

PAGE 84

68 68 Figure 3.11: Optical image of wet pad at 20X Figure 3.12: Optical image of wet pad with laser points at 20X

PAGE 85

69 69 Figure 3.13: Optical image of pad at 30 torr Figure 3.14: Optical image of pad at 50 torr Figure 3.15: Optical image of pad at 70 torr

PAGE 86

70 70 Figure 3.16: Optical image of pad showing contact points at 30 torr Figure 3.17: Optical image of pad showing contact points at 50 torr Figure 3.18: Image showing analysis performed by ImagePro

PAGE 87

71 71 Figure 3.21: Image showing analysis performed by ImagePro Figure 3.19: Optical image of wet pad at 30 torr Figure 3.20: Optical image of wet pad at 50 torr

PAGE 88

72 72Table 3.1: Statistics showing the ch ange in contact area for a dry pad Stats Area Diameter (mean) Min 3.77 2.23 Object Number 14 14 Max 408.74 22.07 Object Number 17 6 Range 404.97 19.84 Mean 121.02 10.05 Std.Dev 140.39 6.88 Sum 2420.53 201.04 Samples 20 20

PAGE 89

73 73Table 3.2: Statistics showing the ch ange in contact area for a wet pad Stats Area Diameter (mean) Min 2.438 1.767 Object Number 25 25 Max 648.80 28.23 Object Number 35 35 Range 646.36 26.46 Mean 106.73 9.62 Std.Dev 139.67 6.31 Sum 3057.94 423.27 Samples 44 44

PAGE 90

74 74 Chapter 4 Synthesis of Zirconia Nanoparticles th rough Plasma Route for CMP Slurry Applications 4.1. Synthesis of zirconia through various methods Aerosol technology has been considered advantageous over the other methods for using small liquid volumes and for being le ss expensive for the nanoparticle synthesis ( 41 ). Hartmann et al.( 42 ) used zirconium tetrachloride in oxygen flame to get fluffy tetragonal zirconia particles with diameter s ranging from 10nm to 50nm. The properties of these particles were controlled by the fl ame temperature. Another technique known as flame spray pyrolysis, (FSP) which is a singl e step method and a low cost process, was used by Nielson et al. ( 43 ) to make 5 m zirconia particles. They used zirconium sulfate as the precursor. Karthikeyan et al. ( 44 ), following the FSP approach, used zirconium nbutoxide in butanol as a precu rsor, and synthesized tetragona l zirconia nanopa rticles with diameters ranging from 12nm to 21nm. The sa me precursor, zirconium n-butoxide, was later used by Yuan et al.( 45 ) to make micrometer and smaller sized powders. These methods and results were later verified by ma ny groups that followed. It was seen that all the results obtained so far using FSP show ed a dominant tetragonal phase and a minor

PAGE 91

75 75 monoclinic phase. Further investigation show ed that these studies did not emphasize on the various factors affec ting the particle size. Work carried out by Mueller et al ( 41 ) showed the importance of precursor concentration, flow rate of dispersant gas on the morphology, size distribution and crystallinity of the zirconia nanoparticles following th e FSP method. They concluded that higher gas flow rate resulted in smaller flame heights, therefore, giving smaller particle size. Large aggregated particle s were seen with high precursor concentrations which is in disagreement with the findings of Killian et al.( 46 ). To explain the co-existence of the two phases, it was shown the fast quenchi ng resulted in the te tragonal phase which decreased with the longer resi dence time and appearance of monoclinic structures. The earlier work, using reverse emulsion tec hnique carried out by Kanai et al., and Ramamurthi et al.( 47 ) was done by bubbling ammonia gas through zirconium (IV) ion giving a non-uniform particle size distributi on of zirconia. This gave way to the twoemulsion technique. The techniqu e involves the use of two re verse emulsions with metal containing aqueous solutions or aqueous ammoni a that are suspended in oil. The size and shape of particles so produced is controlled by the size of the droplet s, which can in turn be controlled using a surfactan t. The two-emulsion precipita tion technique was used by Tai et al.( 48 ) for controlling the particle size of zirconia. They showed that the morphology and size distribution of particles was affected by the stability of the emulsion. The macro-emulsion, prepared using heptane for the oil phase, showed larger particle size, wider particle size distribution, and less aggl omeration with a high solute concentration. For the micro-emulsion system that was prepared using cyclohexane, the size was seen to increase with an increase in the zirconia (IV) ion. Hu et al. ( 49 ) worked

PAGE 92

76 76 on the synthesis of hydrous zirconia nanopa rticles by forced hydrolysis. This method comprises of homogeneous hydrolysis and co ntrolled polymerization of tetramers of zirconium and is performed in the aqueous solution of inorganic salts. The resulting powder was found to be porous monoclinic nanocr ystalline zirconia w ith an average size of 5nm. Another pape r by Srdic et al.( 50 ) talks about chemical vapor synthesis of alumina doped nano zirconia. The average part icle size was about 5 nm with 3 and 30% of surface alumina. They further concluded that the presence of alumina, even at very low concentrations, changed the prope rties of the zirconia. Zirconia powders have also been synthesized using the sol-gel process, i.e. by hydrolysis a nd condensation of alkoxides of zirconium. The powders formed at room te mperature, by this process are amorphous in nature ( 51 ). Another technique used for synthesi s is the low temper ature combustion. The product is voluminous in nature. On charac terization, nano-zirconia showed a compact distribution of partic les with the size 20 nm ( 52 ). The inert gas condensation (IGC) technique employs pure metal or monoxide ev aporation in presence of helium at low atmospheric conditions fo llowed by post-deposition ( 53-56 ) oxidation. This process resulted in non-agglomerating zirconia na nopowders with a size di stribution of 4-14 nm ( 57 ). Though there are various methods for the production of nanopowders as described above, these have some limitations. Some of them display a wide particle size distribution while others show agglomeration problems. Post-treatment needed in some cases for the end product results in sinteri ng problems. Cost and mass production is the limiting factor for many methods. To overcome these issues, the plasma route for the synthesis of nanoparticles has been of great interest ( 58 ).

PAGE 93

77 77 4.2. Plasma synthesis of nanopowders Nanopowders are now being synthesized using the plasma route because of advantages such as providing high temperat ures zones, energy concentration, rapid quenching and above all, a clean environment. In the plasma processing of nano-powde rs, thermal plasma with high energy is used to generate precursor s in the vapor phase to s ynthesize nano-powders, when quenched. Synthesis of metals, ceramics, or composites has shown that thermal plasma route is one of the most promising met hods for producing nano-powders. There are several stages in the plasma synthesis. Thes e include, injection of the reactant in the plasma, reaction to form the product, quenchi ng leading to the nucle ation and growth of the particles ( 59 ). Parameters such as temperature, pressure, flow rate and power control the particle properties. The end powder can be characterized by sma ll particle size, high purity and high activity ( 60 ). Some of the nanopowders synthesized via the plasma synthesis are discussed below: 1. Oxides: Alumina nanopowder is synthesized in a DC arc plasma under atmospheric conditions using Al electrodes ( 61, 62 ). These particles show a spherical shape. Titania nanopowder ( 62 ) which is again spherical in shape is made from the reaction between TiCl4 and O2 It is also prepared from TiC in the Ar-O2 plasma ( 63 ). Other oxides prepared include CeO2, Y2O3, ZrO2 and Cr2O3. 2. Nitrides: Aluminium Nitrid e (AlN), is synthesized using an aluminium anode with an interaction between Al and N2/NH3( 64 ). Silicon nitride (Si3N4) is prepared from the reaction between silicon powder and am monia in an arc plasma

PAGE 94

78 78 ( 64 ). The other nitrides that are prepar ed by the plasma route include TiN and Mg3N2 ( 65-69 ). 3. Carbides: Silicon Carbide is synthesized using Si and CH4 in an induction plasma ( 70, 71 ).Tungsten Carbide is generally pr epared using a dual RF plasma. The other carbides that are being successfully prepared include those of titanium and boron. When carrying out the synthesis in gas phase, a balance has to be achieved in obtaining high production and fa ster rates and formation of agglomerates. The plasma route for synthesis helps in overcoming the i ssues of low temperatures and low reaction rates that are responsible for formation of chunks of nanoparticles. This is done by electrically charging the partic les. One of the major limitation of the plasma route is collection of the end product is generally a very small amount ( 72 ). 4.3. Zirconia nanoparticles through the plasma route Zirconia, crystalline oxide of the zirconium is categorized as the most promising ceramic materials of today. With a mola r mass of 123.22g/mol and a density of 5.89 g/cm3, this white crystalline solid finds use in many engineering applications. Zirconia is in monoclinic phase at room temperature and is seen to change its phase to tetragonal at relatively higher temp eratures. But small particles are seen to exhibit the tetragonal phase at room temperatur e as this is energetically favored while in the larger particles monoclinic phase is favore d. The stresses caused due to the change in phase result in cracking of the ceramic on c ooling. This calls for stabilization using different oxides. Some of the common oxides used for this purpose are magnesium oxide, calcium oxide, cerium oxide and yttrium oxide ( 41 ). Studies have shown that yttria

PAGE 95

79 79 stabilized zirconia (YSZ) has a higher resistance to the low temperature effects compared to the YSZ prepared by conventional methods( 60 ). Vollath and Szabo described the synthe sis of zirconia nanopowders by microwave plasma( 72 ). Here, the operating temperature of the gas is lower than DC or RF plasma. The zirconia nanoparticles obtained by this ro ute are about 5nm in size and are in the form of single crystals. Alumina coated zirconia nanoparticles synthesized using microwave plasma are crystalline in nature wi th a glassy coating. The size of zirconia can be controlled by varying the produc tion rates, concentration of precursor, and flow rate of gas dispersion( 41 ). Ultra-fine powders of yttria stabilized zi rconia (YSZ) have been synthesized using the plasma spray process for solid oxide fuel cell (SOFC) applications. These particles were prepared from zirconium oxynitrate hydrate and yttrium n itrate hexahydrate as precursors. The zirconia pa rticles were spherical a nd about 100-200 nm in size( 58 ). 4.4. Role of slurry in Chemical Mechanical Polishing As discussed in previous sections, the slurry is one of the important process variables in the process of ch emical mechanical planarizatio n (CMP). Slurry parameters such as solution chemistry, abrasive size and concentration, pH, temperature, flow rate, presence of oxidizers, surfactants and corrosion inhibitors etc., influence the planarization process to a great extent. The slurry formula tion is done in order to achieve high removal rates, low coefficient of friction, a nd low defectivity, among other targets( 12 ). Though the slurry is responsible in bringing about the chemical aspects of polishing, the interaction of the slurry with the wafer is a very complex mechanism. During polishing, frictional forces come into play when the slurry flows between the polishing pad and the

PAGE 96

80 80 wafer. These result in mechanical removal of material producing defect s. If the frictional forces are lowered, the removal rate also decr eases. So, the process has to be optimized to achieve high planarity with minimum defect s. Slurry choice has a great effect on the manufacturing, yield and reliability of the process.. Commercial CMP slurries are now being customized for specific process applications. 4.4.1. Slurry composition Slurries consist of dispersed abrasive particles along with certain chemical reagents to bring about the selectivity for metal CMP. The solution acts as a lubricant, controlling the temperature and transports chemical reagent for the abrasives and the chemical reagents to the polishing site. Th ese chemical reagents include oxidizers, surfactants, chelating agents, complexing agents buffer solutions etc. Slurries for metal CMP contain more additives than oxide CMP. The solution also provides electrostatic and steric balance for suspending the abrasi ves. For electrostatic stabilization, zeta potential is important for determining the stability of the slurry while for steric stabilization, the choice of the molecule is mo re important. This is because, electrostatic stabilization results from the repulsion betw een the like charges while steric stabilization is a result of physical interacti on of high-molecular weight species. 4.4.2. Abrasives in slurry The role of the abrasive is to provide the mechanical aspect of CMP ( 73 ). These particles act by abrading the dull and uneven surface. The first generation metal CMP used alumina abrasives and the first gene ration oxide CMP consisted of fumed silica

PAGE 97

81 81 particles. The abrasives are required to meet certain criteria of correct chemical properties, hardness and surface charge. 1. Alumina as a slurry abrasive: Alumina ha s been used in metal CMP for producing higher removal rates. Studies have show n that both, phase and hardness of the alumina particles affects th e polishing. Due to the hi gh hardness of the alumina particles, additional measures need to be taken to reduce the surface defects on wafers. 2. Silica as a slurry abrasive: Since, alumina particles showed high defectivity, silica abrasives have been the next method of choice. The silica abrasives used for the CMP process are fumed and colloidal pa rticles. Both forms are amorphous in nature. Polishing with colloidal silica ha s shown lower defectivity while slurries containing fumed silica have shown higher removal rates. 3. Ceria as a slurry abrasive: Ceria slurries are gaining importance in the field of oxide CMP. Owing to high removal rates, ce ria slurries are exte nsively being used for shallow trench isolation (STI) ( 73 ). Apart from the type of abrasive being us ed in the slurry, the particle properties such as size and hardness also play an important role. As the particle size and hardness of the abrasive particles in the slurry increases, the removal rate increases. The same effect is seen with the increase in the number of abra sive particles in the slurry. Bigger abrasive particles produce deeper surf ace defects such as micro-scratches. Research has shown that removal rates tend to decrease after a particular size threshold of the abrasive particles ( 35 )

PAGE 98

82 82 4.5. Research focus: synthesis, char acterization and applications As already discussed in the previous secti ons, abrasives play an important role in the CMP slurry. The objective of this research is to investigate th e effect of zirconia nanoparticles as abrasives in the CMP slurry. Zirconia nanopar ticles used in the slurry preparation were successfully synthe sized through the Plasma route. Zirconia nanoparticles were synthesized usi ng the plasma route. The set up can be divided in to three segments. The first se gment is the precursor delivery. Zirconium butoxide was used as the precursor and was fe d to the plasma torch in the vapor form. The second segment is the plasma chamber where the precursor gets decomposed. The third segment is the powder collection cham ber. The plasma torch is characterized by plasma tube surrounded by high velocity coolin g-water, to protect against the heat. The induction coil in the torch is connected to the RF power supply which generates the discharge. Different sets of gases are introduced in the chamber. The sheath gas reduces the heat flux and is directed n ear the ceramic tube. The central gas accounts for plasma stability and is introduced tangentially in th e chamber. The precursor, in the vapor form, is introduced in the chamber axially by m eans of stainless steel tubing. The powder evaporated in the induction plasma torch at high temperatures rapidly solidifies on quenching. The powder is then collected from the collection chamber. The properties of the nanopowders depend on the various paramete rs such as concentration of the solution and feed rate, plasma properties like composition and power, gas compositions and quenching rates ( 58 )

PAGE 99

83 83 The synthesized nanoparticles were used to make the abrasive slurry. The slurry was made by dispersing the zirconi a abrasive particles 3% by we ight in de-ionized water. 0.2 mmol of tetramethyl ammonium hydroxide (TMAH) was added. The slurry was then used to polish 2 inch oxide wafers against the 6 inch polishing pad on the CETR bench top tester. The in-situ coefficient of fr iction was monitored during the process. Zirconia nanoparticles synthe sized by the plasma route were characterized using various analytical techniques. XRD was done to analyze the crysta llographic structure and chemical composition. TEM was done to understand the shape, structure and size of these nanoparticles. After the polishing expe riments, optical microscopy and AFM was used to analyze the polished oxide wafers for defects. 4.6. Results and discussion Characterization of zirconia nanopartic les was done to understand the particle structure. Figure 4.1 shows the TEM images of the synthesized nanoparticles at various magnifications. The zirconia nanopa rticles were seen to occur in clusters. These particles show a near spherical shape with the part icle size ranging from 20 nm to 70 nm. Figure 4.2 shows X-ray diffraction analysis of th e zirconia particles. A major peak is seen at 2 which is a characteristic of the tetragona l zirconia. Tetragonal phase (101 orientation) was seen to be primarily present followed by traces of the monoclinic phase (-111 and 111 orientation). The widened peaks suggest a crystalline structure. The synthesized zirconia nanoparticles were used to make the abrasive slurry to polish the oxide wafers. The slurry was made by dispersing the zirc onia nanoparticles in DI water. TMAH was added as a surfactant to the slurry. Surfactants are organic

PAGE 100

84 84 compounds that are added to improve the stability of the slurry ( 74 ). The experiments were carried out on the CETR bench top test er. The working details of the tester are described in the earlier section. The experime ntal details of the polishing experiments are listed in Table 4.1. The zirconia particles show ed a high settling rate and were seen to deposit in the pipe and so the slurry flow rate was kept hi gh. The in-situ coefficient of friction was monitored and was found to be 0.6 which is unexpectedly high. Optical microscopy was used to analyze the polished wafer for surface defects that are caused primarily because of the continuous contact of the abrasives with the wafer. The optical images (Figure 4.3) show many scratches and pits on the wafer. The scratches were of varied sizes and were isolat ed as well as in clusters. The wafer surface appeared rough instead of smooth and planariz ed. This could be due to the high slurry flow rate that was maintained in order to prevent the particles from settling even before reaching the wafer surface.

PAGE 101

85 85 (a) (b) Figure 4.1: TEM images of zirconia nanoparticles at different magnifications Figure 4.2: XRD analysis of zirconia nanoparticles

PAGE 102

86 86 (a) (b) (c) (d) Figure 4.3: Optical images showing different regions of the wafer polished with zirconia slurry

PAGE 103

87 87Table 4.1: Experimental details Parameter Value Slurry 3 wt% zirconia dispersed in DI water Slurry flow rate 150 ml/min Pad Rodel IC 1000, SUBA IV Down Pressure 3 PSI Platen speed 150 rpm

PAGE 104

88 88 Chapter 5 Conclusion Chemical Mechanical Polishing is a pr ocess for achieving local and global planarization. The consumables involved in th e process are a subject of extensive studies to understand their effect on the output para meters such as material removal rate, polishing uniformities and surface defects. The polishing pad made of polyurethane ma terial is one of the major consumable. The pad surface degrades over time with polishi ng and needs to be regenerated to restore its properties in order to sustain good removal rates and polishing uniformity. This regeneration is done by using abrasive disks, to carry out a process known as conditioning. The abrasives abrade the pad su rface to unclog the pores. The size of the abrasives on these conditioning disks plays a major role in determining the conditioning efficiency. In order to understand the impact of these abrasives on the pad, abrasives of different sizes were used in this thesis to carry out the conditioni ng and the results were quantified in terms of pa d roughness, coefficient of friction, and pad wear. A systematic study of conditioner abrasives ranging in size from 0.25 m to 100 m was performed. The results showed that the conditioning with bigger abrasives gave higher pad roughness. For example, the pa d roughness upon conditioning with abrasives of size 100 m was nearly twice that obtained using 0.25 m abrasives. The roughness

PAGE 105

89 89 value was also found to increase with increase in conditioning time from 5 minutes to 20 minutes. However, the bigger 100 m abrasives, when used for a longer conditioning time, start to rupture the pad. Also, for thes e bigger abrasives, the longer conditioning times led to higher pad wear, which has importa nt consequences such as reduced pad life. The 68 m and 100 m conditioner abrasives also show ed a high value of COF (0.5-0.7) during conditioning, which was nearly twice the value found for the smaller abrasives. Though the smaller abrasives showed a stea dy increase in the CO F values with an increase in conditioning time, the change in COF value for bigger abrasive was nearly negligible. These results of the systematic study in this thesis i ndicate that smaller abrasives need longer conditioning times to achieve same degree of pad abrasion as the bigger conditioning abrasives. The bigger abrasives (8 m, 68 m, 100 m) were used to carry out the ex-situ conditioning for polishing the copper wafers and the effect of pad conditioning was observed on the polished wafers. More surf ace defects were found on the wafers that were polished using the pa ds conditioned with the 100 m abrasive while the pad conditioned with the smaller 8 m abrasive gave comparatively less scratches. It was also noted that for the same polishing time, the end-point for polishing was achieved when using the pad conditioned with the 100 m conditioning abrasive and not for the pad conditioned with the 8 m conditioning abrasive. This resu lt indicates that differences in conditioning can lead to st eady state at different ti mes during polishing. Thus, optimization has to be achieved between c onditioning times and size of the conditioning abrasive as it impacts the pad life, the pa d properties and finally the surface defects during polishing.

PAGE 106

90 90 For future work, it will be interesting to look at the time needed for the smaller abrasives (0.25 m and 2 m) to show the same COF and pad roughness as shown by the bigger abrasives. Since smaller abrasives can reach the pores of the polishing pad and can prove more effective in unclogging them, it will be interesting to see if an increase in the density for the smaller abrasives gives bett er polishing results. Higher pad roughness has been reported to give high removal rates. So, another focus for future evaluation of abrasive conditioning would be to look at the material removal rate, which is an important CMP output parameter. High remova l rates indicate good pad properties and a better CMP performance. Another major contribution of this thesis has been in the development of a simple method to measure pad-wafer contact area. Since pad morphology was found to be affected by conditioning and the degree of change in the pad morphology depended on the conditioner abrasive size, the effect of pad morphology on contact area was investigated using a simple optical method. Optical microscopy was used to analyze the pad-wafer contact area. The contact area wa s seen to increase with an increase in pressure. The preliminary work presented here has demonstrated that optical microscopy can be a simple yet effective technique to m easure the contact area and allow future work on measurement of contact area as a func tion of the pad conditioning process. Finally, to understand the e ffect of abrasives in the slurries for CMP, zirconia nanoparticles (as abrasives) were synthesi zed through the plasma route. The wafer polished with these nanoparticle slurries ex hibited many surface defects such as pits and scratches. As many input variables work simu ltaneously in producing these results, these surface defects could be a result of high slurry flow rate, the relative platen speed, or the

PAGE 107

91 91 set down pressure. Further experiments need to be done to understand the effect of the zirconia based slurry and the in terplay with the other input va riables. Here, the first step would be to focus on the slurry chemistry as it plays a very important role in polishing the wafer. With a high settling rate of zi rconia particles in th e slurry, appropriate chemical additives need to be added to prevent the agglomeration and settling of the particles. Optimization of the input variables such as platen speeds and slurry flow rate will also improve the wafer polishing. An evaluation of the effectiveness of the zirconia slurry can then be performed by measuring the material removal rate after polishing.

PAGE 108

92 92 References 1. J. M. Steigerwald, S. P. Murarka, R. J. Gutmann, Chemical Mechanical Planarization of Microe lectronic Materials (Wiley-Interscience, 1997). 2. J. Olsen, F. Moghadam, Handbook of Multilevel Meta llization for Integrated Circuits: Materials, Te chnology, and Applications (1993). 3. S. Wolf, R. N. Tauber, Pattern Registration 473 (1986). 4. K. Kobayashi, Journal of Physics a nd Chemistry of Solids 59 1671 (1998). 5. K. Kobayashi, Materials Science & Engineering B 98 181 (2003). 6. J. Levert, S. Mukherjee, D. DeBear, in Electrochemical Society Conference (Hawaii, 1999). 7. J. H. Zhao et al. Applied Physics Letters 74 944 (1999). 8. D. Evans, Materials Research Society Bulletin 27 779 (2002). 9. J. W. Carr et al. (EP Patent 0,401,147, 1990). 10. F. W. Preston, Journal of the Society of Glass Technology 11 214 (1927). 11. C. W. Liu, B. T. Dai, C. F. Yeh, Journal of the Electrochemical Society 142 3098 (1995). 12. Z. Li, K. Ina, P. Lefevre, I. Koshiyama, A. Philipossian, Journal of the Electrochemical Society 152 G299 (2005). 13. Z. Li, L. Borucki, I. Koshiyama, A. Philipossian, Journal of the Electrochemical Society 151 G482 (2004). 14. K. H. Park, H. J. Kim, O. M. Chang, H. D. Jeong, Journal of Materials Processing Technology 187 73 (2007). 15. L. M. Cook, in Semiconductors and Semimetals 63 (2000). 16. M. R. Oliver, Chemical-Mechanical Planarizati on of Semiconductor Materials (Springer, 2004).

PAGE 109

93 93 17. L. M. Cook, J. V. H. Roberts, C. W. Jenkins, R. R. Pillai. (US Patent 5,489,233, 1996). 18. R. Bajaj, M. Desai, R. Ja irath, M. Stell, R. Tolles, Materials Research Society Proceedings 337 637 (1994). 19. R. Kolenkow, R. Nagahara, Solid State Technology 35 112 (1992). 20. I. Ali, S. R. Roy, G. Shinn, Solid State Technology 37 63 (1994). 21. R. Jairath, M. Desai, M. Stell, R. Tolles, D. Scherber-Brewer, Materials Research Society Symposium 337 121 (1994). 22. R. Dejule, Semiconductor international 20 54 (1997). 23. J. Grillaert et al. Materials Research Society Proceedings 45 (1999). 24. M. R. Oliver, R. E. Schmidt, M. Robinson, Chemical Mechanical Planarization IV: Proceedings of the International Symposium (2001). 25. C. Y. Chen, C. C. Yu, S. H. Shen, M. Ho, Journal of the Electrochemical Society 147 3922 (2000). 26. J. Z. Zhang et al. in CMP-MIC (1997), pp. 315-321. 27. B. J. Hooper, G. Byrne, S. Galligan, Journal of Materials Processing Technology 123 107 (2002). 28. T. S. Bullard, R. J. Lebel, R. Na deau, P. H. Smith Jr. (US Patent 6,022,266, 2000). 29. N. H. Kim, Y. J. Seo, W. S. Lee, Microelectronic Engineering 83 362 (2006). 30. T. Sun, L. Borucki, Y. Zhuang, A. Philipossian, Materials Research Society Symposium Proceedings 991 45 (2007). 31. K. H. Park, H. J. Kim, O. M. Chang, H. D. Jeong, Journal of Materials Processing Tech. 187 73 (2007). 32. H. Lu, Y. Obeng, K. A. Richardson, Materials Characterization 49 177 (2002). 33. T. Murayama, Dynamic Mechanical Analys is of Polymeric Material (Elsevier Scientific Pub. Co., 1978). 34. Y. R. Jeng, P. Y. Huang, W. C. Pan, Journal of The Elec trochemical Society 150 G630 (2003).

PAGE 110

94 94 35. R. K. Singh, R. Bajaj, Materials Research Society Bulletin 743 (2002). 36. J. A. Greenwood, J. B. P. Williamson, Proceedings of the Royal Society of London. Series A 295 300 (1966). 37. G. B. Basim, I. U. Vakarelski, B. M. Moudgil, Journal of Colloid and Interface Science 263 506 (2003). 38. C. L. Elmufdi, G. P. Muldowney, Materials Research Society Symposium Proceedings 991 15 (2007). 39. C. Gray et al. Materials Research Society Symposium Proceedings 991 27 (2007). 40. C. Gray et al. Materials Research Society Symposium Proceedings 867 247 (2005). 41. R. Mueller, R. Jossen, S. E. Pratsinis, M. Watson, M. K. Akhtar, Journal of the American Ceramic Society 87 197 (2004). 42. L. A. T. Hartmann W., Peuckert D. and Kleinschmit, Materials Science and Engineering, A. 109 243 (1989). 43. O. Flame, Ultrafine Particles (1963). 44. J. Karthikeyan et al. Nanostructured Materials 8 61 (1997). 45. F. L. Yuan, C. H. Chen, E. M. Kelder, J. Schoonman, Solid State Ionics 109 119 (1998). 46. K. A. a. M. T.F., Aerosol Sci. Technol. 34 227 (2001). 47. S. D. Ramamurthi, Z. Xu, D. A. Payne, Journal of the American Ceramic Society 73 2760 (1990). 48. C. Y. Tai, M. H. Lee, Y. C. Wu, Chemical Engineering Science 56 2389 (2001). 49. M. Z. C. Hu, M. T. Harris, C. H. Byers, Journal of Colloid And Interface Science 198 87 (1998). 50. V. V. Srdic, M. Winterer, A. Moller, G. Miehe, H. Hahn, Journal of the American Ceramic Society 84 2771 (2001). 51. J. Livage, F. Beteille, C. Roux, M. Chatry, P. Davidson, Acta Materialia 46 743 (1998).

PAGE 111

95 95 52. T. Mimani, K. C. Patil, Materials Physics and Mechanics(Russia) 4 134 (2001). 53. M. Yoshimura, S. Somiya, Materials Chemistry & Physics 61 1 (1999). 54. J. D. Lin, J. G. Duh, Journal of the American Ceramic Society 81 853 (1998). 55. S. Somiya, T. Akiba, Journal of the European Ceramic Society 19 81 (1999). 56. A. Cabanas, J. A. Darr, M. Poliakoff, E. Lester, Chemical Communications 2000 901 (2000). 57. G. Skandan, Nanostructured Materials 5 111 (1995). 58. Z. Wang et al. Journal of Power Sources 170 145 (2007). 59. L. Tong, R. G. Reddy, Journal of the Minerals, Metals and Materials Society 58 62 (2006). 60. G. Vissokov, I. Grancharov, T. Tsvetanov, Plasma Science and Technology 5 2039 (2003). 61. P. Madhu Kumar et al. Materials Chemistry and Physics 36 354 (1994). 62. S. M. Oh, D. W. Park, Thin Solid Films 386 233 (2001). 63. Y. L. a. T. Ishigaki, Thin Solid Films 407 79 (2002). 64. G. a. P. P. Vissokov, Ultradispersed powders Plasma -chemical Preparation and Properties 395 (1998). 65. S. J. Gubzica J, Mahai I, et al, Journal of Materials Science 35 3711 (2000). 66. K. P. a. P. E. Chang Y, Plasma Chemistry and Plasma Processes 9 73 (1989). 67. H. B. Liang S, Zheng Z, et al, Trans. Nanoferrous Metals Society, China 3 103 (1996). 68. C. Y. Young R M, Lecerf B, et al, 8th International Symposium on Plasma Chemistry, Tokyo 4 2040 (1987). 69. K. N. a. S. K. Futaki S, 8th International Symposium on Plasma Chemistry, Tokyo, 2040 (1987). 70. G. F. Guo J Y, Boulos M I, Plasma Chemistry and Plasma Processing 17 219 (1997).

PAGE 112

96 96 71. G. F. Guo J Y, Boulos M I, et al, Journal of Materials Science 32 5257 (1997). 72. D. Vollath, D. V. Szab, Journal of Aerosol Science 28 685 (1997). 73. L. M. Cook, Journal of Non-Crystalline Solids 120 152 (1990). 74. Y. Hong, U. B. Patri, S. Ramakrishnan, S. V. Babu, Chemical-Mechanical Planarization--Integration, Technology and Reliability 867 41 (2005).


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 002007020
003 fts
005 20091009142848.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 090616s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002741
035
(OCoLC)401294663
040
FHM
c FHM
049
FHMM
066
(S
090
TP145 (Online)
1 100
Manocha, Chhavi.
0 245
Chemical mechanical planarization :
b study of conditioner abrasives and synthesis of nano-zirconia for potential slurry applications
h [electronic resource] /
by Chhavi Manocha.
260
[Tampa, Fla] :
University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 96 pages.
502
Thesis (M.S.Ch.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Co-advisor: Vinay K. Gupta, Ph.D
Co-advisor: Ashok Kumar, Ph.D.
520
ABSTRACT: Chemical Mechanical Planarization (CMP) has emerged as the central technology for polishing wafers in the semiconductor manufacturing industry to make integrated multi-level devices. As the name suggests, both chemical and mechanical processes work simultaneously to achieve local and global planarization. In spite of extensive work done to understand the various components and parameters affecting the performance of this process, many aspects of CMP remain poorly understood. Among these aspects of CMP is the role of abrasives in the processes of conditioning and polishing. These abrasives are present in the chemical slurry between the wafer and the pad for polishing and play an important role during the conditioning to regenerate the clogged polishing pads. This thesis has focused on the study of abrasives, both in conditioning and polishing. The first part of the thesis concentrates on the effect of abrasive size for conditioning purposes.Diamond is being widely used as an abrasive for conditioning the polishing pad. Five different sizes of diamonds ranging from 0.25m to 100m were selected to condition the commercially available IC 1000 polishing pad. Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM) analysis were carried out on the pad to study the effect of the abrasive size on the pad morphology. In-situ 'coefficient of friction' was also monitored on the CETR bench top Tester. The final impact was seen in the form of surface defects on the polished copper wafers. As pad morphologies resulting from different conditioning affect contact areas, the second part of the thesis focuses on developing a simple method to quantify the area of contact between the wafer and pad using optical microscopy. Optical images that were obtained were analyzed for the change in contact area with the change in operating conditions.Finally, the third part of the thesis details the synthesis and characterization of nano-zirconia for potential slurry applications. Nano-zirconia was synthesized using the plasma route and then characterized using different analytical techniques like TEM and XRD. These nanoparticles were then used to make abrasive slurry for oxide CMP and the polished wafers were analyzed for surface defects.
653
Chemical mechanical polishing
Conditioning
Pad wear
Contact area
Plasma synthesis
690
Dissertations, Academic
z USF
x Chemical and Biomedical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2741