<%BANNER%>

Ultraviolet-Assisted oxidation and nitridation of hafnium and hafnium aluminum alloys as potential gate dielectrics for ...

University of Florida Institutional Repository

PAGE 1

ULTRAVIOLET ASSISTED OXIDATION A ND NITRIDATION OF HA FNIUM AND HAFNIUM ALUMINUM ALL OYS AS POTENTIAL GATE DIEL ECTRIC S FOR METAL OXIDE SEMICOND UCTOR APPLICATIONS By CHAD ROBERT ESSARY A DISSERTATION PRESENTED TO THE GRADUATE SCHOO L OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Chad Robert Essary

PAGE 3

This document is dedicated to my loving family.

PAGE 4

iv ACKNOWLEDGMENTS I would like to first thank my advisor, Dr. Rajiv Singh, for giving me the opportunity to study this exciting field of electronic materials. Also I want to thank Dr. Valentin Craciun and Dr. Joshua Howard for being good friends as well as mentors to me during my time here at the University of Florida. To all the group members I have had the pleasure to work with and to our wonderful secretary Margaret, I would like to offer my sincere gratitude for all your help and friendship all of you have provided my four years here at the University of Florida. Furthermore, I would like to thank Dr. Cammy Abernathy, Dr. Stephen Pearton, Dr. David Norton, and Dr. Fan Ren for taking the time to serve on my committee and for providing direction and guid ance. Finally, I would like to thank the electronics group at the University of Texas Austin for helping with the electrical measurements on a portion of this dissertation.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. iv LIST OF TABLES ................................ ................................ ................................ ............ viii LIST OF FIGURES ................................ ................................ ................................ ............. x ABSTRACT ................................ ................................ ................................ ...................... xiii CHAPTER 1 LITERATURE REVIEW ................................ ................................ ............................. 1 Introduction ................................ ................................ ................................ ................... 1 CMOS Transitors ................................ ................................ ................................ .......... 3 Properties and Limitations of SiO 2 ................................ ................................ ............... 5 Materials Selection Guideline ................................ ................................ ....................... 7 Band Gap and Alignment ................................ ................................ ...................... 8 Thermodynamic Stability ................................ ................................ ...................... 8 Film Morphology and Inter face Quality ................................ .............................. 10 Implementation and Process Compatibility ................................ ......................... 10 Reliability ................................ ................................ ................................ ............ 11 Hafnium Oxide Properties ................................ ................................ .......................... 11 Hafnium Oxide Research ................................ ................................ ............................ 12 Nitride Barrier Layers ................................ ................................ .......................... 13 Other Novel Approaches ................................ ................................ ..................... 14 UV Assisted Oxidation ................................ ................................ ............................... 15 Motivation ................................ ................................ ................................ ................... 18 2 EXPERIMENTAL PROCEDURES ................................ ................................ ........... 19 Deposition Setup and Equipment ................................ ................................ ............... 19 UVPLD System ................................ ................................ ................................ ... 19 Laser Ablation ................................ ................................ ................................ ..... 21 Laser System ................................ ................................ ................................ ....... 23 Experimental Setups and Samples ................................ ................................ .............. 25 Interfacial Layer Formation and Growth ................................ ............................. 25 UV Assisted Oxidation and Nitridation of Hafnium Films ................................ 26 UV Assist ed Oxidation and Nitridation of Hafnium/Aluminum Alloy Films .... 27

PAGE 6

vi Experimental Characterization Techniques ................................ ................................ 28 Variable Angle Spectrosc opic Ellipsometry ................................ ....................... 28 X ray Reflectivity ................................ ................................ ................................ 29 X ray Diffraction and Glancing Incidence X ray Diffraction ............................. 31 X ray Photoelectron Spectroscopy ................................ ................................ ...... 33 Atomic Force Microscopy ................................ ................................ ................... 34 Electrical Characterization ................................ ................................ .................. 34 Current Voltage Measurements ................................ ................................ ........... 36 Capacitance Voltage Measurement ................................ ................................ ..... 37 3 KINENTICS OF INTERFACIAL LAYER FORMATION DURING DEPOSITION OF HAFNIA ON SILICON ................................ ................................ ........................ 43 Deposition Conditions ................................ ................................ ................................ 43 X ray Results ................................ ................................ ................................ ....... 45 XPS Analysis ................................ ................................ ................................ ....... 49 Ellipsometry Results ................................ ................................ ............................ 53 Summary ................................ ................................ ................................ ..................... 54 4 EFFECT OF UV ASSISTED OXIDATION AND NITRICATION OF HAFNIUM METAL FILMS ................................ ................................ ................................ .......... 56 Experiment ................................ ................................ ................................ .................. 56 Results and Discussion for Low Temperature Deposited Samples ............................ 60 Surface Morphology ................................ ................................ ............................ 60 X Ray Analysis ................................ ................................ ................................ ... 61 X Ray Reflectivity ................................ ................................ ............................... 64 XPS Analysis ................................ ................................ ................................ ....... 64 Electrical Properties Analysis ................................ ................................ .............. 72 Capacitance Voltage Results ................................ ................................ ............... 72 Repeatability ................................ ................................ ................................ ........ 74 Interfacial Trap Density Measurements ................................ .............................. 76 Current Voltage Results ................................ ................................ ...................... 77 TEM Analysis ................................ ................................ ................................ ...... 78 Proposed Model ................................ ................................ ................................ ... 79 Results and Discussion of the High Temperature Depositions ................................ .. 81 XPS ................................ ................................ ................................ ...................... 82 X Ray Analysis ................................ ................................ ................................ ... 83 Electrical Characterization Results ................................ ................................ ...... 85 Discussion ................................ ................................ ................................ ............ 87 Summary ................................ ................................ ................................ ..................... 89 5 UV ASSISTED OXIDATION AND NITRIDATION OF HAFNIUM ALUMINUM METAL ALLOY FILMS ................................ ................................ ........................... 90 Experiment ................................ ................................ ................................ .................. 90 X Ray Analysis ................................ ................................ ................................ ... 92

PAGE 7

vii Compositional Comparisons ................................ ................................ ............... 92 Electrical Characterization Results ................................ ................................ ...... 94 High Temperature Anneal ................................ ................................ ................... 96 Summary ................................ ................................ ................................ .............. 98 6 CONCLUSIONS ................................ ................................ ................................ ......... 99 A PPENDIX A ELECTRICAL DATA EXTRACTION METHOD ................................ ................ 102 B XPS DATA ................................ ................................ ................................ .............. 107 LIST OF REFERENCES ................................ ................................ ................................ 111 BIOGRAPHICAL SKETCH ................................ ................................ ........................... 116

PAGE 8

viii LIST OF TABLES Table page 1 1: Industrial roadmap for minimum feature size ................................ ............................. 5 1 2: Various properties of high k oxide ma terials. ................................ ............................. 8 1 3: Gibbs free energy of formation for potential alternative high k dielectrics. ............. 10 1 4: Various properties of HfO 2 ................................ ................................ ....................... 12 2 1: Conditions for growth of ultra thin HfO 2 films. ................................ ........................ 26 2 2: Growth and postdeposition conditions for UV oxidat ion of Hf ............................... 27 2 3: High temperature Hf metal deposition and oxidation conditions. ............................. 27 2 4: Growth and annealing conditions for Hf/Al metal films. ................................ .......... 28 2 5: Conversion factors for series parallel electrical equivalent circuits. ......................... 41 3 1: Conditions for growth of ultra thin HfO 2 film s. ................................ ........................ 44 3 2: XRR modeling parameters used to fit the spect ra ................................ ..................... 48 3 3: Ellipsometry modeling results ................................ ................................ ................... 54 3 4: Comparison of the thickness values (in ) for HfO 2 thin films deposited on Si ...... 55 4 1: Growth and postdeposition conditions for UV oxidat ion of Hf ............................... 59 4 2: Z range and RMS roughness for oxidized Hf metal films. ................................ ........ 61 4 3: d/ l values for samples H2 H5 ................................ ................................ .................... 67 4 4: Dielectric constants and EOT for samples H2 H4. ................................ ................... 74 4 5: Repeatability study results for k and EOT of oxidized hafnium metal films. ........... 76 4 6: Defect density results for samples analyzed at UT Austin. ................................ ...... 77 4 7: EOT and k for the high temperature deposited s amples ................................ ............ 87

PAGE 9

ix 5 1: Sample deposition conditions ................................ ................................ .................... 91 5 2: Sample composition from elemental analysis using XPS. ................................ ........ 93 5 3: Predicted film density and bulk dielectric constant. ................................ .................. 94 5 4: Dielectric constants and EOT for the aluminum study samples. ............................... 96

PAGE 10

x LIST OF FIGURES Figure page 1 1: Typical CMOS inverter ................................ ................................ .............................. 4 1 2: Energy band diagram of an n+ poly Si/SiO 2 /n Si structur e ................................ ........ 7 1 3: Bandgap and band offset comparison for var ious dielectric materials ........................ 9 2 1: UVPLD system with excimer laser and optics set up. ................................ ............... 20 2 2: Photograph of UV lamp placement in the chamber. ................................ ................. 21 2 3: Laser energy distribution during target ablation. ................................ ...................... 22 2 4: Typical XRR plot showing which film properties most influence various features. 30 2 5: Typical XR D equipment geometry setup ................................ ................................ .. 32 2 6: AFM operational setup showing laser detection of the cantilever deflection ........... 35 2 7: MOS capacitor setup used for CV/IV measurements. ................................ ............... 36 2 8: Kiethley Instruments Inc. Win 82 measurement system ................................ .......... 38 2 9: Typical capacitance/voltage response curve. ................................ ............................. 40 2 10: Equivalent circuits for estimatio n of capacitance ................................ .................... 42 3 1: Comparison of GIXD ( W =1) for the 200 C and 600 C samples ............................ 46 3 2: Acquired XRR spectrum and its simulation for a HfO 2 film ................................ ... 47 3 3: Schematic of gate oxide stack model used for XRR simulations. ............................. 47 3 4: Interfacial layer thickness dependence on O 2 pressure for the 600 C samples. ....... 49 3 5: High resolution XPS scans of the Hf 4f region ................................ ......................... 50 3 6: Raw data and fit for the Si 2p region for the 600 C, 1x10 2 Torr O 2 sample. .......... 51 3 7: Variations of d/ l values for a 5 nm pure SiO 2 on Si and a deposited sample ........... 52

PAGE 11

xi 3 8: Interfacial layer thickness as a function of temperature and pressure. ...................... 53 4 1: CV response of an UV grown Si x N y film. ................................ ................................ 5 7 4 2: 2D AFM scans of samples a) H2 and b) H3 taken using contact mode .................... 62 4 3: 2D AFM scans of samples a) H4 and b) H5 taken using contact mode ..................... 63 4 4: 3D AFM micrographs of both a) UV oxidized and b) non irradiated ...................... 65 4 5: GIXD spectra ( W =1) of Hf films deposited on Si and oxidized .............................. 66 4 6: XRR plot and fit for as deposited Hf metal film. ................................ ...................... 66 4 7: Angle resolved XPS Si 2p spectra for system calibration. ................................ ........ 68 4 8: Angle resolved XPS Si 2p spectra for H2 and H3 (UV and no UV films). .............. 68 4 9: O 1s spectra for H2 and H3 (UV and no UV) films at 45 and 90. ........................... 69 4 10: Si 2p spectra for UV oxidized films H4 and H2 (with and without nitrogen). ....... 70 4 11: N 1s peak of as deposited nitrided Hf metal film. ................................ ................... 71 4 12: N 1s peak after PDA for samples H4 and H5. ................................ ......................... 71 4 13: Comparison of high and low frequency capacitance responses for sample H4. ..... 75 4 14: High frequency capacitance voltage responses for samples H2 H4. ...................... 76 4 15: CV response and model fit data for samples H2, H4, and H5. ................................ 77 4 16: Current voltage results from oxidized hafnium metal films. ................................ ... 78 4 17: XTEM of the UV annealed nitrided sample H4. ................................ ..................... 79 4 18: XTEM of non UV annealed hafnium film H3. ................................ ........................ 81 4 19: Model of the effect of UV and nitrogen. ................................ ................................ 82 4 20: O ls Peaks for the high temperature deposited sample s et. ................................ ..... 83 4 21: Hf 4f Peaks for the h ig h temperature deposited sample s et. ................................ ... 84 4 22: GIXD for the high temperature deposited s amples taken at W = 0.5. .................... 85 4 23: XRR plots and models for the high t emperature deposited samples. ...................... 86 4 24: CV plots for the high temperature deposited s amples ................................ ............. 88

PAGE 12

xii 4 25: IV plots for the high temperature deposited s amples ................................ .............. 88 5 1: GIXD scans taken at W =1 comparing the effects of aluminum additions ............... 92 5 2: Compilation of t he capacitance voltage results ................................ ........................ 95 5 3: Compilatio n of the current voltage results ................................ ............................... 97 5 4: GIXD for sample AH3 annealed at 900 C ................................ ............................... 97

PAGE 13

xiii Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ULTRAVIOLET ASSISTED OXIDATION A ND NITRIDATION OF HA FNIUM AND HAFNIUM ALUMINUM ALLOYS AS POTENTIAL GATE DIEL ECTRIC S FOR METAL OXIDE SEMICOND UCTOR APPLICATIONS By Chad Robert Essary December, 2004 Chair: Rajiv K. Singh Major Department: Materials Science and Engineering The continued miniaturization of silicon based compli mentary metal oxide semiconductor (CMOS) devices is pushing the limits of the silicon dioxide (SiO 2 ) gate dielectric. As the channel widths are decreased to increase packing densities and functionality of new chips, proportional vertical scaling of the di electric must be maintained to keep constant capacitances. Silicon dioxide is approaching its fundamental limit in which it can be used as the gate dielectric due to high leakage currents resulting from direct tunneling through the layer. In order for th e continued use of current CMOS gate design, an alternative material with a higher dielectric constant must be found. Several materials have been proposed but are still not providing the electrical characteristics favorable for use in the devices due to p roblems with excessive leakage and hysteresis resulting from the quality of the film and oxygen defects. The goal of this study is to create higher quality films at lower processing temperatures with low leakage and less hysteresis than has been achieved with hafnium

PAGE 14

xiv oxide films. This study first examines the formation of the interfacial layer in pulsed laser deposited hafnium oxide films to understand the kinetics behind its formation. The second section focuses on the oxidation of pulsed laser deposit ed (PLD) hafnium metal thin films using ultraviolet (UV) assisted post deposition annealing. Another set of samples was deposited in an ammonia atmosphere in order to incorporate nitrogen into the films. Comparisons of microstructure and stoichiometry of oxidized hafnium and oxy nitride films were made using x ray photospectroscopy, variable angle spectroscopic ellipsometry, glancing angle x ray spectroscopy, x ray reflectivity, and atomic force microscopy. Analysis of the interface between the films and the silicon substrate was carried out using x ray reflectivity. The electrical characteristics of the films were characterized using capacitance voltage and current voltage measurements in order to compare the quality of the films. Ultimately a model of the effect of UV and nitrogen additions was presented. In the final portion of this work, additions of aluminum were used to increase the crystallization temperature of the films. The effect of aluminum both with and without nitrogen incorporation on th e dielectric constant and leakage for the films was studied.

PAGE 15

1 CHAPTER 1 LITERATURE REVIEW Introduction Over the past 30 years, the integrated circuit industry has strived to maintain constantly shrinking device geometries and increased chip density. 1 The decrease in device size has allowed for greater packing densit ies, thus increasing the functionality and speed of the chips. The industry trend has followed what is known as Moores Law. In 1965, Gordon Moore from Intel first noted that every two to three years a new technology generation with approximately doubled logic circuit density and a 40% performance increase is realized as design and processing methods are improved. 2 3 Since then, through continued miniaturization of gate widths of the metal oxide semiconductor field effect transistors (MOSFET) and proport ional vertical scaling, Moores Law has been maintained. In order to continue miniaturization using current MOSFET structure designs, the gate oxide, which is currently silicon dioxide, thickness must be reduced to less than 15 4 This represents five atomic layers of oxide or less. Problems not only arise in processing to keep the thickness from varying across the gate, but also in the amount of leakage current that results due to quantum mechanical tunneling. Tunneling current increases exponentiall y as the thickness of the layer is decreased. 5 Losses due to tunneling would exceed the power consumption standards the industry wishes to use. 1 In order for continued reduction in gate widths, alternative materials with higher dielectric constants (high k) than silicon dioxide must be implemented.

PAGE 16

2 Silicon dioxide (SiO 2 ) has been vital to the industry due to its abundance and many useful properties. First of all, SiO 2 is readily formed by the oxidation of the surface of the silicon layer and can be preci sely controlled through both chemical and thermal means. This provides an easier process than a deposition technique and eliminates interfacial defects experienced with deposition because SiO 2 forms an atomically abrupt interface with silicon limiting the number of carrier traps that degrade the performance of the transistor. Furthermore, post annealing in a hydrogen containing ambient can easily passivate dangling bonds. This excellent interface allows for high mobility of the carriers in the MOSFET cha nnel and high device performance. 5 Finding an alternative gate dielectric that will act as well as SiO 2 is not an easy task. Research groups are working on a wide range of single metal oxides and nitrides, 6 9 binary compounds, 10 13 and some ternary oxid es. 9 The replacement oxide must be thermodynamically stable on silicon up to current processing temperatures for the chips (~900 1000C) 1, 14 15 to avoid reactions with the silicon that create compounds with lower dielectric constants that parasitically h inder the benefits of the higher k material. The dielectric also needs to resist leaching of the dopants from the silicon channel and the gate contact, which is currently heavily doped poly silicon. The layer must form an atomically abrupt or near atomic ally abrupt interface with silicon to minimize carrier traps at the surface of the channel and prevent mixing and formation of a lower k silicate interface layer. Ideally, the layer would be single crystalline or amorphous to inhibit leakage due to grain boundaries. Growth of single crystals is a slow process and thus does not lend itself well to implementation into industry. With amorphous layers, some

PAGE 17

3 materials will crystallize when heated, so careful selection of the material must be made based on the wafer processing conditions it will be subjected to after deposition. 16 CMOS Transitors The basis of the computer revolution of the past 30 years has been the use of complementary metal oxide semiconductor (CMOS) transistors as the logic components of the integrated circuits. Research on a new class of materials known as semiconductors conducted at Bell Laboratories began in 1945 in an attempt to replace vacuum tubes for signal amplification in long distance telecommunication applications. Vacuum tubes w ere bulky, unreliable, consumed too much power, and produced too much heat. The first design to utilize this new class of materials was the point contact transistor, created on December 16 th 1947. The point contact transistor consisted of gold foil on t op of a plastic triangular piece in intimate contact with a chunk of germanium. This first design was modified within a year into the junction sandwich transistor and proved to be more practical, rugged, and easier to manufacture. Twelve years later, the field effect transistor was created based on the original theories predicted in the early 1900s and has remained the key design that has been continuously used and adjusted for the past forty years. Figure 1 1 shows a typical CMOS inverter structure c ontaining both a nMOS and a pMOS device working joinly 17 The gates are made of poly Si and are heavily doped to reduce the sheet resistance along the lines and to tailor the work function of the gate to the type of channel being used The poly Si is ins ulated from the MOS channels by a thin gate oxide dielectric. When a voltage is applied to the gate, a potential is set up across the dielectric. Depending on the type of the channel, the potential will either decrease or increase the

PAGE 18

4 Figure 1 1: Typical CMOS inverter [17]. current from the source to the drain. For the inverter structure, when a voltage is applied (represented by a 1) typically the nMOS device will turn on and the pMOS device will turn off. The source for the nMOS is connected to ground (represented by a 0) and so passes the 0 to the drain. And vice versa, when a 0 is applied to the gates, the pMOS structure, whose source is wired to line voltage, will turn on and pass the 1 signal to the drain while t he nMOS will be turned off. By packing more of such logic components on the area of a chip, faster processing speeds can be obtained. There are design rules that must be followed as to the spacing between each device; however the rules, for the most part scale with the dimensions of the gates. 17 The current technology on the market now is 0.13 m m gate widths used in the Intel Pentium 4 processors. 18 This corresponds to SiO 2 gate dielectric thickness of ~20. For further gate width reduction, proportio nal scaling of the gate dielectric must be maintained to keep a constant capacitance across the gate channel area. By replacing the SiO 2 with a material having a greater dielectric constant, the capacitance can be maintained while affording a thicker diel ectric layer and thus reduce the leakage inherent with thinner films. This relationship is shown in equation (1 1). ( ) k high ox k high eq t t = k k (1 1)

PAGE 19

5 t eq = thickness of the equivalent SiO 2 layer t ox = thickness of the SiO 2 layer k ox = dielectric constant of SiO 2 k high k = dielectric constant of the alternative material The interface between the silicon and the gate dielectric is very important because formation of a lower k layer will result in thicker equivalent thickness. T his is because the capacitances of the stacked layers add in series as shown in equation ( 1 2). 2 1 1 1 1 C C C tot + = (1 2) C tot = total capacitance C 1,2 = respective capacitances of the layers Ultimately by using higher k materials if the interfa ce can be controlled, t eq thicknesses of <10, which are needed for further advancement of the silicon based CMOS devices, can be realized. Table 1 1 shows the predicted roadmap for equivalent gate oxide thickness required for the minimum feature size. 1 T able 1 1: Industrial roadmap for minimum feature size and equivalent dielectric thickness in MOSFET devices. Year Minimum Feature Size ( m m) Equivalent Dielectric Thicknes () 1997 0.25 40 50 1999 0.18 30 40 2001 0.15 20 30 2003 0.13 20 25 2006 0.10 1 5 20 2009 0.07 10 15 2012 0.05 <10 Properties and Limitations of SiO 2 Amorphous SiO 2 has served as the gate dielectric for silicon based CMOS devices due to its excellent properties. SiO 2 has a bandgap of 9 eV and a dielectric constant of 3.9 and can readily be formed on silicon with low defect concentrations at the interface. Using modern processing techniques, defect charge densities on the order of 10 10 /cm 2

PAGE 20

6 midgap interface state densities of ~10 10 /cm 2 eV, and hard breakdown fields of 15 MV/cm are routinely obtained. 1 The industry has come to rely on these standards when designing new chips. This presents a major challenge when trying to replace SiO 2 However, SiO 2 is approaching its limit of usefulness as the gate dielectric thickness requiremen ts are pushed below 15 SiO 2 has a large conduction band offset (~3.5 eV) compared with silicon and therefore provides a stable electrical insulation layer. However, this offset decreases as the thickness of the layer is decreased. Ultimately, the SiO 2 layer loses its bulk properties at a thickness of around 7 8 19 Below this thickness, the Si rich interfacial regions from the channel and poly Si gate interfaces overlap, causing an effective short through the dielectric, rendering it useless a s an insulator. Experimental evidence has shown that ultrathin (13 15 ) SiO 2 films can be processed which operate satisfactorily. However, high leakage current densities of 1 10 A/cm 2 have been measured for such devices. 2 0 In order to meet the power gu idelines set forth by the industry, current leakage densities must be kept below 1x10 1 A/cm 2 for gate biases of 1.2 V, corresponding to a contribution of only a few mW to the overall chip dissipation. 2 1 For SiO 2 layers thicker than 6 nm, the leakage curr ent mechanism is explained by Fowler Nordheim electron tunnelling. As the layer thickness decreases, the mechanism switches to direct tunnelling. The difference between Fowler Nordheim tunnelling (FNT) and direct tunnelling (DT) of electrons is determine d by the shape of the tunnel barrier. If the oxide tunnel barrier is triangular, FNT occurs while DT takes place through a trapezoidal barrier as shown in Figure 1 2. 2 2 This occurs because of the band offset reduction as the layer becomes thinner. As a result, the FNT leakage current is exponentially dependent

PAGE 21

7 on the thickness of the oxide. To illustrate the drastic effect this relationship has, Buchanan in his review article gave the example: for silicon dioxide, at a gate bias of 1 V, the leakage cur rent changes from 1x10 12 A/cm 2 at 35 to 1x10 A/cm 2 at 15 a change of 12 orders of magnitude. 23 Figure 1 2: Energy band diagram of an n+ poly Si/SiO 2 /n Si structure in the case of (a) Fowler Nordheim tunnelling and (b) direct tunnelling of electrons from the Si degenerate accumulation layer into the poly Si gate conduction band. [22] Materials Selection Guideline Many materials are being studied for use as replacement gate dielectrics. Over the past decade, Al 2 O 3 Zr O 2 Y 2 O 3 TiO 2 Ge 2 O 3 La 2 O 3 Ta 2 O 3 HfO 2 SrTiO 3 and Ba x Sr 1 x TiO 3 have all been considered. Various properties of these materials are shown in Table 1 2. 5 However, the following key properties suggested by Wilk et al 1 must be considered when choosing a material: Permittivity, band gap, and band alignment to silicon Thermodynamic stability Film morphology Interface quality Compatibility with current or expected materials used in processing

PAGE 22

8 Process compatibility Reliability Table 1 2: Various properties of high k oxide materials. Oxide Material Dielectric Constant Conduction Band Offset (eV) Density (g/cm 3 ) Melting Point (C) Stable in Contact with Silicon Al 2 O 3 9.3 2.8 3.97 2054 YES BaSr 0.5 Ti 0.5 O 3 80 3600 0.1 0.1 6.02 1625 NO CeO 2 7 --7.65 24 00 YES HfO 2 22 25 1.5 9.68 2774 YES Ta 2 O 5 24 65 0.3 8.20 1785 NO TiO 2 80 170 --4.23 1843 NO Y 2 O 3 10 2.3 5.03 2439 YES ZrO 2 ~25 1.4 5.68 2710 YES Band Gap and Alignment The material should have a band gap offset of greater than 1 eV to that o f the conduction band of silicon. This requirement rules out some of the higher dielectric materials such as SrTiO 3 Figure 1 3 graphically compares the band gaps and offsets of some potential alternative dielectrics to that of Si and SiO 2 24, 25 Typica lly, materials with higher dielectric constants will have smaller band gaps, so an optimization must be made to balance the dielectric constant and band gap to get the best overall results. It was first believed that materials with very high dielectric co nstants such as SrTiO 3 and BaTiO 3 would be ideal so thick dielectric layers could be implemented. However due to their small band gap offsets to Si, they had shortcomings for use in gate dielectric applications but have found uses in RAM applications. The rmodynamic Stability The dielectric must be thermodynamically stable in contact with Si to prevent formation of SiO 2 or lower k silicate structures. Single component oxides can be examined by comparing the Gibbs free energy of formation with respect to o xygen

PAGE 23

9 Figure 1 3: Bandgap and band offset comparison for various dielectric materials. [24] bonding. Elements having greater magnitude free energies than silicon will not readily give up oxygen to form SiO 2 when in conta ct with Si. 26 Table 1 3 lists the Gibbs free energies for various metals that form potential alternative high k oxides. 26 Ta, Mo, and W have been ruled as unstable in contact with silicon because their energies of formation are less in magnitude than sil icon and thus will give up oxygen to the silicon, and also tend to form silicates that will hinder the dielectric constant and equivalent thickness of the stack. Further considerations must be taken into account because this table only shows the values fo r systems in equilibrium. However, during deposition or formation of devices, rarely is the system at a true equilibrium. Therefore, other reactions could occur, so a

PAGE 24

10 good understanding of the kinetics behind the deposition or growth and processing steps is key in predicting non equilibrium products. 9 Table 1 3: Gibbs free energy of formation for potential alternative high k dielectrics. Metal D G f (10 22 kcal/O atom) of metal oxide Expected oxidation product of metal silicide on silicon Si 1.70 Ta 1 .52 SiO 2 /TaSi 2 /Si Mo 0.88 SiO 2 /MoSi 2 /Si W 1.01 SiO 2 /WSi 2 /Si Y 2.40 Y 2 O 3 SiO 2 /Si La 2.26 La 2 O 3 SiO 2 /Si Zr 2.06 ZrO 2 SiO 2 /Si Hf 2.16 HfO 2 SiO 2 /Si Film Morphology and Interface Quality To avoid high leakage paths due to grain boundaries, single cryst alline or amorphous films are desired. Grain boundaries also promote the diffusion and leaching of dopants from the channel when thermally processed. However, Al 2 O 3 is the only single metal oxide that does not have a preferable crystalline phase under no rmal processing conditions. Single crystal oxides with excellent interface structure can be grown using molecular beam epitaxy. However, this process is extremely slow and requires ultra high vacuum conditions resulting in very low throughput. 1 Ultimate ly, amorphous barrier layers may have to be used at the interfaces of the dielectric even if it will hinder the benefits of using that high k dielectric. Implementation and Process Compatibility Ultimately whatever material that is chosen as the replacemen t for silicon dioxide as the gate dielectric must be able to be fit into the large scale production scheme the micro electronics industry runs on. Thus, additional processing steps or a hindrance to

PAGE 25

11 throughput are unacceptable. The new material must be a ble to be implemented into the fabrication line with as little equipment and process flow changes as possible. Reliability Finally, the reliability of devices using the new gate material must be proven. The devices must meet stringent ten year lifetime standards set by industry. 1 These tests use higher voltages than normal operating voltage to stress the devices until breakdown conditions occur. This allows for lifetime estimates under actual working conditions to be extrapolated. To date, few reports have been generated of this nature and additional work must still be conducted. Further issues with threshold voltage shifts during operation must also be considered. Stability of the threshold voltage will ensure repeatable response of the device. If the threshold shifts too far, the device will cease to function as the operating voltage being used is not sufficient to operate the gate. Hafnium Oxide Properties One of the most promising materials currently being studied is HfO 2 Table 1 4 lists some of the properties of this dielectric. 24, 27 28 Full ternary phase diagrams for the Hf Si O system have not yet been reported. However, similarities to the well studied Zr Si O system have been shown. 9 Both metals are from the same family of the periodic table and exhibit the same crystal structure and preferred equilibrium compounds. Furthermore, due to the large negative free energy of formation for Hf O bonds, hafnium metal has the ability to pull oxygen from SiO 2 thus limiting interfacial formation. For these reasons, Hf and its subsequent oxide and oxy nitride will be used for this study.

PAGE 26

12 Table 1 4: Various properties of HfO 2 Property Value Dielectric constant 20 30 Band Gap 6 eV Conduction Band Offset 2.8 eV Crystallization temperature 450 500C Crystal Structure Monoclinic Thermal Stability w.r.t. Silcon 950C Hafnium Oxide Research Over the past three years, many groups have been focusing their research efforts on HfO 2 and trying to use various methods for improving its performance as a gate dielectric. The interface quality between silicon and the dielectric is highly dependent on the deposition and annealing processes used. B. Y. Park et al. 29 grew HfO 2 thin films with thicknesses from 27 55 on hydrofluoric acid treated Si wafers at a temperature of 200C by chemical vapor deposition. They noted that the initial film growth was controlled by fast oxidation of the Si wafer. The Hf concentration in the film increased with deposition time. The films were post annealed in N 2 at 800 C. Upon annealing the film separated into two distinct layers having approximate dielectric constants of 5.6 and 9.3 respectively. Equivalent thicknesses of 19 were reported for the thinnest layers. The incorporation of Si into the depositing HfO 2 fil m must be controlled if equivalent thicknesses of ~10 or less are to be realized. Using reactive DC magnetron sputtering, L. Kang et al 30 formed HfO 2 films with physical thicknesses of 45 The sputtering was done at room temperature in a modulated O 2 flow to control the interface qualities and to suppress the additional growth of an interfacial layer. The deposition was followed by an ex situ anneal at 500C in N 2 ambient for 5 min. Pt was used as the gate electrode. They reported an equivalent

PAGE 27

13 thi ckness of ~13.5 The leakage currents for the films were about 1x10 4 A/cm 2 which is four orders of magnitude less than that for an equivalent SiO 2 film. Hysteresis values of <100 mV were also obtained. This study shows that HfO 2 can be successfully used as an alternative gate dielectric under the right processing conditions. However, thinner equivalent thicknesses will be required for future technologies. S. W. Nam et al. 31 studied the effect of annealing ambient on reactive DC magnetron sputtered HfO 2 films. First, Hf metal was pre deposited as an oxidation barrier, and then a thin HfO 2 layer was deposited at room temperature in Ar and O 2 ambient from an Hf metal target. The Hf metal layer helps to form HfO 2 by reducing the native oxide on the s ubstrate. Samples were annealed at temperatures ranging from 500 900C in both N 2 and O 2 The as deposited films were amorphous in structure. Formation of a monoclinic phase was reported for annealing temperatures above 650C and at greater than 800C a n orthorhombic HfO 2 structure emerges. The equivalent thickness values obtained in this study were calculated to ~19 for N 2 annealed HfO 2 films and ~28 for O 2 annealed films. The increase in equivalent thickness is believed to be due to the increase of the thickness in the interfacial oxide layer. Nitride Barrier Layers Attempts have been made to control the interface by using barrier layers. P. D. Kirsch et al 32 investigated the electrical properties of HfO 2 films on nitrided and un nitrided Si substrates. The samples were prepared by magnetron sputter deposition of Hf metal and subsequent oxidation on both bare silicon and on wafers which had been exposed to NH 3 for 30 s at a temperature of 700C. The samples were annealed in N 2 ambient for 10 s at 600C. After annealing, MOS capacitors with HfO 2 thicknesses of ~50 were fabricated using TaN as the gate electrode. The nitrided samples

PAGE 28

14 demonstrated leakage current densities two orders of magnitude less (4x10 5 A/cm 2 ) than the non nitrided sam ples (4x10 3 A/cm 2 ) and an increase in capacitance of 15%. Equivalent oxide thickness for the 50 films was estimated to be 10.6 and 11.7 with and without SiN x respectively. Therefore, use of a nitride barrier layer can improve the leakage and capaci tance characteristics of HfO 2 gate dielectrics. However, the nitride interface layer degrades mobility and capacitive response due to additional interface traps. Other Novel Approaches Other novel approaches have been used to tailor the electrical charac teristics of HfO 2 H. Lee et al. 33 examined using Dy doping to control oxygen vacancies in the film. Since the electronegativity of Dy is lower than that of Hf, the concentration of oxygen vacancies in the Dy doped HfO 2 was lower than that of their contr ol HfO 2 and thus yielded leakage current densities over three orders of magnitude smaller than the control. Furthermore, the large atomic radius of Dy was predicted to reduce the leakage current through the dielectric due to a packing density effect. In a study by H. Yu et al ., 34 Al 2 O 3 was added to HfO 2 and a range of compositions were compared in an attempt to control and model the energy gap and band alignment for the dielectric on Si. Band gap and offset values for various compositions ranging from p ure HfO 2 to pure Al 2 O 3 were carried out using high resolution XPS. The Al 2p, Hf 4f, O 1s core levels spectra, and O 1s energy loss spectra all showed continuous changes with the variation of the HfO 2 mole fraction. From this data, they were able to mode l the band gap and band offsets for the (HfO 2 ) x (Al 2 O 3 ) 1 x system. This system is important because of the higher temperatures at which the film will remain amorphous. It has been shown that Al concentrations of greater than 31.7% will result in

PAGE 29

15 crystalli zation temperatures between 850 and 900C, which is about 400C higher than that for HfO 2 Various studies have also investigated the interface layer of the HfO 2 dielectric to the gate electrode. One of the approaches used barrier layer formation by nitri ding the top surface of the HfO 2 before electrode deposition creating a thin Hf x N y layer so B doped poly Si gates could be used. 35 A similar study was conducted using Al 2 O 3 as the capping layer. 36 Both types of barrier layers were shown to reduce leakage and improve the stability of the doping levels in the gate. UV Assisted Oxidation Ultraviolet (UV) radiation has been shown to assist in the thermal oxidation of silicon. 37 39 For temperatures around 850 C, thermal oxidation of silicon in dry O 2 is on t he order of 2/min even though the supply of O 2 from the gas phase to the SiO 2 surface is plentiful. This is due mainly to the high activation energies of >1.5 eV required for O 2 diffusion through the oxide to react with the underlying silicon. By using a UV source producing photons with wavelengths of less than 200 nm, the energy of the photons is great enough to split O 2 bonds, thus forming more reactive species. The following equations (1 3..1 6) show the reactions that take place in the gas phase. 40 M is a third body. O 2 + h n 2O (1 3) O + O 2 + M O 3 + M (1 4) O 3 + O 2O 2 (1 5) O 3 + O 2 O 2 + O 2 + O (1 6) Furthermore interaction of photons with wavelengths less than 254 nm will cause the ozone species to break apart as s hown in equation (1 7).

PAGE 30

16 O 3 + h n O 2 + O (1 7) Within the bulk, photoinduced transitions occur from the conduction band of Si to the conduction band of SiO 2 (3.15 eV) and from the valence band of Si to the conduction band of SiO 2 (4.25 eV) when the sample is irradiated. 41 42 Therefore during irradiation of UV, the SiO 2 layer is supplied with both atomic oxygen from the gas interface and electrons from the Si/SiO 2 and thus the probability of ionization of the oxygen species is enhanced. Th e oxidation rate of light enhanced thermal oxidation of silicon at temperatures above 740 C can be explained using the photo enhanced Deal and Grove model. 38,43 Below this temperature, the Carera and Mott model demonstrated by Ishikawa et al better expla ins the initial 70 of film growth and thereafter the Deal and Grove model seems to fit. 44 In this model, the oxidants are assumed to absorb on the surface creating vacancies and thereby encouraging electrons to tunnel through the oxide to fill them. Th e charged species then move under the influence of the space charge through the oxide to the oxide/silicon surface. Current work is being conducted on the oxidation of zirconium metal to form dielectric films. 45 49 Ramanathan et al deposited thin 20 30 films of Zr at room temperature onto peroxide washed Si substrates. The washing process leaves a thin amorphous SiO 2 /SiON layer. The films were then exposed to Hg vapor lamps emitting primarily at wavelengths of 185 and 254 nm. They have shown that th e light interacts with O 2 to form atomic O and ozone enhancing the oxidation rate of Zr enabling formation of a stoichiometric metal oxide up to about 55 thick after 90 minutes at room

PAGE 31

17 temperature. 46 They compared these films to ones oxidized without th e use of UV assisted annealing. The films oxidized without UV demonstrated severe frequency dispersion in both the accumulation and depletion regions of the capacitance voltage plots. Using electron energy loss spectroscopy and chemical mapping of the ox ygen concentration, they were able to explain that the cause of the poor electrical characteristics resulted from a high concentration of oxygen vacancy defects. The films oxidized using UV showed very little frequency dispersion and had a more constant o xygen distribution through the thickness of the layer. In another study Ramanathan et al 47 proposed that the oxidation of Zr films by the UV ozone method is expected to be self limiting at room temperature. To verify this, capacitors were fabricated b y oxidizing ~20 Zr films for 60 min and 90 min. Both sets of samples were subjected to a forming gas anneal at 400C for 30 min. Then capacitors were made from the films by using Pt electrodes. Capacitance voltage curves were measured at 100 KHz from both samples and yielded nearly identical results. This implies that the oxidation is self limiting for the thickness range investigated and also shows that lengthy room temperature oxidations do not result in significant growth of interfacial SiO 2 Furt her analysis in this study compared the hysteresis of reported values for ZrO 2 from other groups to the films they formed by the UV ozone oxidation of Zr. They were able to obtain films with ~100 mV hysteresis for physical thicknesses of 50 Their valu e was markedly better than the 200 mV hysteresis reported by Wang et al 50 Electrical traps in the zirconia films are proposed to be the cause of the hysteresis. 51 UV ozone oxidation eliminates many of the oxygen defects which contribute to the

PAGE 32

1 8 hysteresi s. Ultimately films with <15 mV are desired for future implementation into device structures. 52 Motivation Even after years of research in an attempt to find a suitable replacement dielectric for SiO 2 more research is still needed to understand the role of the interface, to tailor the chosen dielectric, and to insure reliability. This three part work first examines the interfacial layer formation of pulse laser deposited (PLD) HfO 2 ultra thin films under different processing conditions. The role of tem perature and oxygen deposition pressures was studied. By understanding the nature of the interface and how it is formed we can better control it to tailor the ultimate properties of the film. The second portion of the work combines the use of UV assisted oxidation and the incorporation of nitrogen into the films during deposition to produce higher quality films and interfaces. The use of UV radiation lowers the temperature required to fully oxidize the thin metal films and limits the oxygen vacancy and no n stoichiometry issues typically encountered in PLD deposition of oxides. The nitrogen incorporation during deposition limits the diffusion of oxygen to the interface thus slowing interfacial oxide growth. Finally, aluminum was added to the films during deposition to increase the crystallization temperature in an attempt to maximize the dielectric constant while minimizing the leakage current. Initial studies of the composition and structural properties of the films, as well as, the electrical characteristics were conducted after deposition and UV oxidation processes were concluded. The samples with the lowest concentration of aluminum were subsequently heat treated at 500, 700 and 900 C to determine the c rystallization temperature of the films.

PAGE 33

19 CHAPTER 2 E XPERIMENTAL PROCEDUR ES Deposition Setup and Equipment All the thin film samples in this work were deposited using a pulsed laser deposition (PLD) technique in an ultra high vacuum compatible chamber utilizing an excimer laser system. The chambe r has been modified for use with an ultra violet (UV) lamp array for irradiation during deposition and for in situ post deposition (PD) anneals. UVPLD System A Neocera brand single chamber vacuum system was used for all the sample depositions. Figure 2 1 shows a schematic of the system along with laser and the optics setup used for the pulsed laser depositions. Due to the lack of a loadlock for the system, the chamber was backfilled with nitrogen to bring it to atmosphere each time a sample was loaded/unl oaded. The chamber pumping system consists of a Pfeiffer MD 4T oil free diaphragm roughing pump backing a Pfeiffer TMU 230 turbo pump. Vacuum levels of 1x10 6 Torr can be reached within an hour, 1x10 7 Torr within twenty four hours, and 1x10 8 Torr with in 3 days. Addition of a liquid nitrogen cooling to the system allows for 1x10 8 Torr within 24 hours. Higher vacuum levels are possible if a loadlock system is added and thus eliminating the viton seals on three of the ports. Samples were mounted on a 2 Neocera brand stainless steel resistive heater capable of 850C and positioned vertically in the chamber. The system has a variable controller to regulate the substrate temperature and has been calibrated within C. This system includes a compute r controlled multi target carousel available for

PAGE 34

20 Figure 2 1: UVPLD system with excimer laser and optics setup. depositions of up to six different materials for multilayer/mixture or superlattice experiments. This carousel was used for the third portion of the experiments. An array of 4 11 u bend low pressure Hg lamps from Atlantic UV was added to the conventional

PAGE 35

21 PLD setup to convert it into a UVPLD apparatus. The lamps were placed on a rack in a plane parallel to the surface of the substrate halfway between the substrate and the target. Figure 2 2 shows the placement of the lamps and target. Highly sensitive Varian brand leak valves were used to control input of ultrapure gases into the chamber allowing for a wide range of deposition ambients and p ressures. Figure 2 2: Photograph of UV lamp placement in the chamber. Laser Ablation M any techniques are available f or the deposition of thin films onto a substrate, however pulsed laser deposition has several key advantages that make it an ide al small batch research tool. PLD provides near s toichiometric transfer of molecules from the tar get to the substrate, rapid testing of many different materials, and offers a wide range of applications, though it does not lend itself well for industrial scale pro duction. The ablation process itself is characterized by an input of energy from the laser to a given target material. Figure 2 3 shows a schematic of the energy distribution at the

PAGE 36

22 target surface during ablation. The total energy for the system is repre sented in equation 2 1. E = E r + E p + E d + E c (2 1) E = incident energy E r = reflected energy E p = plasma plume energy E d = energy of disintegration E c = energy absorbed by the cavity wall Figure 2 3: Laser energy distribution during target ablation. There are additional factors related to the laser energy that play an important role in surface response to pulse energy. The incident pulse energy must exceed the ablation threshold energy for the material for ablation to occur. If the pulse energy is less than the ablation threshold, material will not be physically removed from the surface (i.e., E p and E d tend to zero), but that energy will be absorbed by the cavity wall (E c ) as heat, which

PAGE 37

23 can be useful for laser annealing appl ications. The degree to which the target material is ablated depends on the extent the energy is exceeded. If excessive energy is used, large particulates from the target can be blown off. This is unwanted for high quality films. All experimentation in this study was carried out at an energy value greater than the ablation threshold but low enough to minimize particulate ablation. Laser System A Lambda Physik LPX 305 I KrF excimer laser (s/n 9412 E 4188) was used to deposit all the samples in the follo wing experiments. This model laser uses a pulsed mode delivering 25 nanosecond duration square shaped wave pulses at frequencies from 1 50 Hz and output energies from 10 1100 mJ yielding ultimate fluences from 0.1 4 J/cm 2 depending on optics used. The la ser is fired or triggered by either an internal computer module near the laser or from a remote external computer placed near the deposition chamber, which also controls the multi target carousel. Excimer lasers function on the principle of stimulated phot oemission of gases within a set cavity. This laser uses a mixture of Kr, F, and a buffer gas, typically Ne or He. High voltages ranging from 16 23 kV elevate electrons in the Kr and F atoms into excited states causing excited KrF* complexes to form. Upo n their decay, 248 nm single wavelength radiation is given off. The cavity is designed such that conditions of stimulated emission of coherent radiation occurs and thus subsequent amplification of the radiation is achieved resulting in high energy laser o utput in a pulsed mode. A series of lenses and an aperture are used to select the highest energy portion of the beam, direct, collimate, and focus the pulse until it impinges the sample. For the PLD system, the laser beam first passed through a 2 cm x 1 c m aperture to select the center portion of the spot from the laser. This eliminates the lower energy diffuse side portions

PAGE 38

24 of the spot and gives the initial rectangular spot shape. Due to positioning of the chamber, the beam is then reflected using a UV mirror optic placed at 45 to the beam path. Once reflected, the incident beam is passed through a collimating lens to keep the radiation from diverging due to scattering as it passes through the ambient air. One more mirror is used to reflect the beam t o the chamber. The beam then is passed through a focusing optic with a 30 cm focal length placed just outside of the chamber. The beam enters the chamber through a fused silica window port rated for use with excimer radiation. This window passes 90% of 248 nm radiation. A Gentec Sun Series EM1 energy meter (s/n 86052) was used to determine the fluence reaching the sample. The laser was operated at the lowest energy possible, the focusing optic was removed, and the meter head was placed inside the cham ber. Lower energies had to be used as the meter head has a low maximum energy tolerance before damage to the surface will occur. Averages of 20 pulses were used to record the energy of the spot impinging the meter head. An additional 10% of the energy wa s subtracted to take into account the focusing optic. By taking the energy value obtained and dividing by the area of the spot on the target, precise calculation of the laser fluence was possible and then scaled with the laser energy settings used for the experiments. The laser ablation spot impinging the target had dimensions of 2 x 5 mm, maintaining the shape of the rectangular aperture used at the start of the beam path. By moving the focusing optic to a position of 35 cm from the target, the target is past the focal length of 30 cm of the optic and thus is in the imaging plane region allowing for conservation of the spot shape. By placement of the lens such that the focal point coincides with the target surface, smaller spot sizes and greater fluence can be obtained.

PAGE 39

25 However, it was unnecessary because the threshold value for laser ablation of the oxide materials and metals used in our experimentation were sufficiently exceeded with the current optics setup. Having a larger spot size also gives more uniform ablation of the target so trenching does not occur as rapidly. Another argument for not using the focal point for ablation arises from the irregular shape of the spot which results. The spot geometry is shifted to an elongated oval shape with two additional satellite peaks to either side of the main ablation spot. The energy of the main spot was characterized by a Gaussian type distribution while the satellite peaks were of lower energies based on burn paper tests. All of the samples in this wor k were deposited using the previously mentioned laser and optics setup. Experimental Setups and Samples There are three main areas that comprise the work discussed in this dissertation. This section delineates the differences in the samples and gives the motivation behind each portion of the experiments. Interfacial Layer Formation and Growth In this initial portion of the experimentation, the formation and characteristics of the interfacial layer were investigated. Ultrathin HfO 2 samples were deposited on silicon by ablating a high purity HfO 2 target in a conventional PLD chamber under different oxygen ambients and substrate temperatures. The main goal for this investigation was to determine the nature of the interfacial layer and the factors that contr ol its formation and growth. The conditions for deposition are shown in Table 2 1. The samples were investigated by variable angle spectroscopic ellipsometry, x ray reflectivity, and angle resolved x ray photoelectron spectroscopy. These techniques will be discussed later in this chapter.

PAGE 40

26 Table 2 1: Conditions for growth of ultra thin HfO 2 films. Sample ID# Temp. C Pressure Torr Hf11 1x10 4 Hf12 200 1x10 3 Hf13 1x10 2 Hf14 1x10 4 Hf15 400 1x10 3 Hf16 1x10 2 Hf17 1x10 4 Hf18 600 1x10 3 Hf1 9 1x10 2 UV Assisted Oxidation and Nitridation of Hafnium Films The second portion of this dissertation examines both the effect of UV oxidation of hafnium metal thin films and the effect of nitrogen incorporation into the films. Four conditions were u sed and are denoted in Table 2 2. Half of the samples were deposited in ambient vacuum levels of 1x10 6 Torr while half were deposited in 1x10 2 Torr of ammonia. Half the samples were oxidized in 300 Torr of oxygen at 650 C without the aid of UV radiati on while the other half were oxidized for 30 minutes with the lamps on under the same temperature and pressure used for the non irradiated samples. Target film thicknesses of 40 50 were obtained after oxidation for all the samples. This experiment was r epeated three times to verify reliability and repeatability of the process. The goal of this portion of the experiments was to improve the electrical characteristics and film stoichiometry while using lower processing temperatures. Lower deposition tempe ratures usually result in poorer quality films when using PLD to deposit them, however by depositing a metal then oxidizing it with a post deposition (PD) annealing step higher quality films can be realized. In an attempt to incorporate a significant amo unt of nitrogen into the films, higher temperature depositions carried out at 650 C. A set of four samples were produced, all

PAGE 41

27 of which were deposited in 1x10 2 Torr of ammonia. Half were deposited with the use of UV radiation during deposition and half were deposited with the lamps off. Then half of the samples were annealed in UV during the oxidation step. Table 2 3 shows the sample identification and conditions for this set of samples. Table 2 2: Growth and postdeposition conditions for UV oxidation of Hf metal ultra thin films. Sample ID # Deposition Pressure Torr Anneal Conditions H2 1x10 6 UV H3 1x10 6 No UV H4 1x10 2 Ammonia UV H5 1x10 2 Ammonia No UV Table 2 3: High temperature Hf metal deposition and oxidation conditions. Sample ID # Dep osition Conditions Anneal Conditions HN1 UV UV HN2 No UV UV HN3 No UV No UV HN4 UV No UV UV Assisted Oxidation and Nitridation of Hafnium/Aluminum Alloy Films Finally, to increase the crystallization temperature of the films aluminum additions were s tudied. This set of experiments used the carousel with two high purity targets of Hf and Al. The pulse ratio between targets was varied to obtain different film stoichiometry. The sample conditions are listed in Table 2 4. Half of the samples were depo sited in ammonia to incorporate nitrogen into the structure as was done in the second set of experiments. After electrical characterization, the samples were annealed at 900 C in N 2 to study the detrimental effect of high temperature anneals and to deter mine if the aluminum additions will suppress crystallization.

PAGE 42

28 Table 2 4: Growth and annealing conditions for Hf/Al metal films. Sample ID # Pulse Ratio 1x10 2 Torr Ammonia AH1 1 Al: 1 Hf No AH2 1 Al: 2 Hf No AH3 1 Al: 3 Hf No AHN1 1 Al: 1 Hf Yes AHN 2 1 Al: 2 Hf Yes AHN3 1 Al: 3 Hf Yes Experimental Characterization Techniques Many techniques are available for characterizing the structure, morphology, stoichiometry, and electrical properties of thin films. This section reviews the techniques used t o compare the samples in this dissertation to delineate the effect of temperature, pressure, UV oxidation, nitrogen incorporation, and aluminum additions. Variable Angle Spectroscopic Ellipsometry Variable angle spectroscopic ellipsometry (VASE) is a simpl e, yet very powerful, nondestructive analysis tool for determination of thickness in thin film multilayered structures. Optical properties such as index of refraction and the extinction coefficient can also be obtained. This technique is best suited for flat planar materials with thicknesses ranging from 1 1000 nm having low surface roughness, which makes it an invaluable tool for analysis of the thin film samples in this work. A J. A. Woollam brand M 88 variable angle ellipsometer was used for sample a nalysis. VASE operates by reflecting a collimated beam of light of known polarization from a Xenon lamp off of the surface of the sample to be analyzed. A second polarizer is used to determine the polarization shift resulting from interaction with the sa mple. The angle of incidence and wavelength of light used are extremely important factors

PAGE 43

29 dependent on the thickness and type of sample being analyzed. For dielectric films, an incidence angle of 75 is optimal. All electromagnetic phenomena are govern ed by Maxwells equations and thus certain mathematical relationships can be determined when light encounters boundaries between different materials. Ellipsometry uses Snells Law and Fresnel reflection relationships to determine delta and psi for the fil m. The software package that runs the equipment plots the psi and delta values obtained. A model is then needed to interpret the data. It is best if information such as estimated thickness and density are known. The software has a database of material properties for use in the models. Various mixture routines are available for multi component films. The software takes the initial information the user inputs and iterates through a routine until the model has the lowest least square fit. X ray Reflecti vity X ray reflectivity (XRR) measurements were made with the assistance of a Panalytic MRD XPert system. XRR is another non destructive optical characterization technique that yields similar data to that of VASE. XRR will verify the VASE result and can be used to better fit the models used to interpret the raw data obtained from both analysis techniques. X ray reflectivity plots provide information on the thickness, roughness, and density of a given material. Figure 2 4 demonstrates an example of a ty pical plot showing how the features of an acquired spectrum relate to the various data that can be extracted from the spectrum. Similar to VASE, this technique is well suited to smooth flat samples with low surface roughness and is most sensitive to sampl es in the 10 400 nm thickness range.

PAGE 44

30 This particular x ray based analysis technique works in the traditional q 2 q geometry and scans angles from subcritical to a few degrees. The detector records the intensity of the radiation reflected from the surface of the sample. At angles of less than the Brewsters angle, also know as the critical angle, total reflection occurs. At angles greater than the Brewsters angle, the radiation begins to interact with the surface and interfaces within the sample. This interaction is governed by Snells law and Fresnals relationships similar to VASE. The constructive and destructive nature of the x rays at a particular angle generates the fringe pattern in the detected signal. The raw data are modeled using a user inp ut system whereby the density, roughness, thickness, and absorption coefficients can be changed until a suitable fit is obtained. Figure 2 4: Typical XRR plot showing which film properties most influence various features.

PAGE 45

31 X ray Diffraction and Glanc ing Incidence X ray Diffraction X ray diffraction (XRD) is an invaluable tool that provides phase analysis as well as strain, grain size, epitaxial quality, phase composition, preferred orientation, and defect structure. This technique is nondestructive a nd noncontact and can be used for films greater than 50 The basic equipment setup is shown in Figure 2 5. X rays (typically Cu K a ) impinge the sample at a specific angle. The scattering of the X rays by the sample atomic planes produces constructive and destructive interference. The condition for constructive interference from planes with a given spacing is given by Braggs Law shown in Equation 2 2. When this condition is met, peaks in the intensity detected will occur. ? = 2d hkl sin? hkl (2 2) ? = wavelength of the x ray radiation d hkl = d spacing between (hkl) planes ? hkl = angle between the atomic plane and the incidence direction For single crystal films, there is only one specimen orientation that will satisfy the conditions for Bragg diffraction for each family of planes, so careful alignment procedures must be taken. However, with t hin films that are polycrystalline, fiber textured, or exhibit preferred orientations several families of planes may contribute to a diffraction system. X ray diffraction data for this disserta tion was obtained with a Panalytic XPert MRD diffraction system. Once an x ray diffraction pattern is generated, positive phase identification can be achieved by comparing measured d spacing from the diffraction p attern (and their integrated intensities) to a known JCPDS powder diffraction standard. Additional information such as grain size can be generated from the full width at half max of the peaks.

PAGE 46

32 Normal XRD mode of operation could not be used for the thin fi lms produced in this dissertation. The high intensity peaks from the underlying silicon substrate masked the weak signal from the film. G lancing in cidence x ray diffraction had to be used in this case In this setup, the incident angle of the x ray syst em is fixed at a small value (typically from 0.250 1.000) so only the surface of the sample is probed and the receiving slit is allowed to scan through a typical range for a conventional XRD 2? scan. As a result, only planes that satisfy the Bragg condit ion with the additional constraint in place will produce peaks. In a polycrystalline sample with many orientations, this will still generate a representative x ray diffraction plot, but with much higher surface sensitivi ty without any masking peaks from t he silicon substrate. This method proved to be beneficial for several samples in this dissertation. Figure 2 5: Typical XRD equipment geometry setup [53]

PAGE 47

33 X ray Photoelectron Spectroscopy X ray photoelectron spectroscopy (XPS) is one of the most commo nly used techniques for thin film chemical analysis along with Auger electron spectroscopy (AES). This surface sensitive technique provides elemental as well as bonding environment information. Elemental analysis can be obtained for all elements except h ydrogen and helium. XPS spectra were collected using a Perkin Elmer 5100 installation using Mg K a radiation (1486.6 eV) at takeoff angles from 45 90. This technique works by impinging high energy photons at a specific angle to the surface. The photons cause electrons from the atoms in the sample to ionize and form photoelectrons. The kinetic e nergy of the electron is measured and used to calculate the binding energy based on Einsteins Photoelectron law given in Equation 2 3. KE = h? BE (2 3) KE = Kinetic energy h? = Energy of incident photons BE = Binding energy Each a tom can be identified by the binding energies of its electrons. By comparing the ratios of photoelectrons from the different atoms along with the sensitivities for each a quantitative chemical analysis can be obtained. Compositional changes across the th ickness of the thin film sample can be generated by one of two ways. The first is by using an argon sputtering gun which removes a portion of the sample from the surface at a set rate. The other, which was used in these experiments, is angle resolved XPS whereby different take off angles are used and the data generated is compared. This allows for different interaction volumes of the sample to be probed to compare any differences that might occur through the thickness.

PAGE 48

34 Atomic Force Microscopy Atomic for ce microscopy measurements were made with a Digital Instruments brand Nanoscope III operating in both contact and tapping mode. This technique was chosen for its exceptional ability to produce topographic images of surfaces in three dimensions at high res olutions. Atomic resolution is attainable, if proper conditions are set. This technique is also perfectly suited to the insulating, low roughness samples created for this dissertation. In an atomic force microscope, a sharp tip is mounted on a flexible cantilever. A piezoelectric scanner moves the cantilever/tip assembly along the surface of the sample, while a laser reflecting off the end of the cantilever maps the topographical changes the cantilever senses. When the tip comes into close proximity of a sample surface, van der Waal forces repel the tip causing the cantilever to deflect, in contact mode the tip stays at a constant distance from the surface while in tapping mode a potential is applied at a specific frequency moving the tip near the sampl e surface at a certain cyclical rate. The typical operational setup for the AFM is shown schematically in Figure 2 6. Electrical Characterization A primary tool for indication of quality of thin film is the use of electrical characterization techniques. After deposition or processing of thin films, metal oxide semiconductor (MOS) devices were fabricated and measured. Typical preparation of the film included the deposition of platinum (Pt) contacts via DC magnetron sputtering. A shadow mask with an arra y of circular dots ranging from 25 500 m m was used to create dot arrays on the samples for the gate contacts. The backside contact was silver (Ag).

PAGE 49

35 Figure 2 6: AFM operational setup showing laser detection of the cantilever deflection. [53] The backs ide of the wafer was first abraded and swabbed with 1% HF so that clean silicon, without any native silicon dioxide, was present. It is of paramount importance to choose the proper metals for the dielectric film being measured and the type of silicon subs trate that was used. If improper metals are chosen for a given setup, band alignment conditions may exist whereby non ohmic contacts are created within the device. As an example, the backside metal contact on a p type silicon wafer should have a work fun ction greater than that of the silicon. If these conditions are met, there should be an ohmic contact, if not, the contact will be a rectifying Schottky contact. A schematic of a MOS capacitor fabricated for this dissertation is shown in Figure 2 7 demon strating the testing setup used for these experiments. After the front side gate metallization was complete, samples underwent heat treatments in a conventional tube

PAGE 50

36 furnace under a flowing forming gas atmosphere (4% H 2 balance N 2 ). Both current voltage and capacitance voltage measurements were taken once the MOS capacitors were formed. Figure 2 7: MOS capacitor setup used for CV/IV measurements. Current Voltage Measurements Determination of the leakage current is an essential first step in analyzing an MOS device. Once MOS devices were fabricated, a Keithley Instruments Inc., KI236 source measurement unit (SMU) was used to measure the current flow through a device. The 236 SMU was attached to a pair of Signatone Inc, micromanipulators mounted on a probe station. Tungsten probes that were milled to produce a fine ~5 m m tip were fitted in the micromanipulators. The output of the SMU was connected to the micromanipulator in contact with the Pt gate while the input was connected to the backside Ag con tact. This configuration is optimal for the determination of current leakage pathways directly below the device being measured (i.e., it avoids stray leakage paths). The current compliance threshold, a value that may not be exceeded by the machine, of 10 0 nA was input into the measurement parameters. Then a direct current bias sweep was conducted over a voltage range (from negative to positive) and the amount of current that passed through the MOS

PAGE 51

37 structure was monitored. To reduce creation of bias indu ced defects, typically it is best to start at a small voltage sweep range so that it is possible to determine if the device is leaky before a wider voltage sweep range is implemented. The main goal of the leakage current measurement is to identify a high quality device that can be used for capacitance voltage measurements and to determine if excessive leakage is present that needs further investigation. Capacitance Voltage Measurement Capacitance voltage measurements serve as one of the most versatile and sensitive of all electrical characterization techniques. It is the ultimate tool for determining discreet differences in a MOS device that may serve as the final word in whether a given processing condition has resulted in a high enough quality device to apply to MOSFET applications. These measurements were carried out with a Kiethley Instruments Inc. Win 82 measurement system. This system, as seen in Figure 2 8, is comprised of four main components working in unison. The Keithley 595 capacitor is used for simultaneous quasistatic low frequency measurements. The Keithly 590 capacitance meter is used for high frequency capacitance measurements at 100 kHz and 1 MHz. The Keithley 230 voltage source is used for static bias condition measurements. These thr ee devices are wired into the Keithley 5951 remote input coupler, which serves to filter and relay the device data to and from the various pieces of equipment. Due to its ability to give detailed information about the quality of the MOS device, the Win 82 system will be discussed in Appendix A. The output of a typical CV curve can be seen in Figure 2 9. There are several important features that should be noted. First, there are three main regions with respect

PAGE 52

38 Figure 2 8: Kiethley Instruments Inc. Win 82 measurement system. Modified from the Win 82 manual. to gate bias voltage to take into consideration, known as accumulation, depletion, and inversion. The presence of the different regions is a result of the majority charge carriers (e.g., holes in a p type silicon wafer) in the semiconductor. When a negative bias is applied to the gate electrode, positively charged holes are attracted from the semiconductor bulk region to the oxide/semiconductor interface where they accumulate (the accumulation regi on) in order to maintain charge balance in the system. The depletion region is generated when a gate is made less negative and the reduced field across the oxide allows the charge at the interface to diminish. As the sign on the voltage changes from a ne gative to a positive, majority carriers are repelled from the interface creating an area depleted of majority carriers (the depletion region). Finally, the inversion region is generated when the voltage becomes very positive and the depletion width has in creased to a point where other mechanisms become important. For example,

PAGE 53

39 in the depletion region, the product of the concentration of electrons and hole (np) is much less than the square of the intrinsic carrier concentration (n i 2 ) and in this case, pair generation may occur and the subsequent minority carriers can migrate to the interface. The result of this is prevention of further depletion and a constant value for the capacitance is maintained. An additional possible scenario may occur whereby insuff icient time is allowed for pair generation in which case a model called deep depletion will occur. The high frequency measurement system has the ability to measure two different frequencies and using Metrics ICS software with two different equivalent ci rcuit models, seen in Figure 2 10, to generate capacitance values. This system does not directly measure a capacitance value for the films, but instead measures resistance and conductance and uses the models to generate the capacitance values. The paral lel and series models represent two different physical structures. The series model addresses a capacitor in series with a resistance, possibly from the semiconductor. The equation that describes this scenario is: C + = i R Z (2 3) where Z is the impedance, R is the resistance, and X is the reactance. Alternatively, the parallel model addresses a capacitor in parallel with some sort of conductance, possibly resulting from leakage through an ultrathin film. The equation that describes th is scenario is: iB G Y + = (2 4) where Y is the admittance, G is the conductance, and B is susceptance. Note that the

PAGE 54

40 admittance, conductance, and susceptance are each the reciprocals of the impedance, resistance, and reactance, resp ectively. Figure 2 9: Typical capacitance/voltage response curve demonstrating major areas of interest modified from the Win 82 manual. The reactance and susceptance can then be further described in a capacitive sense as: S C w 1 = C (2 5) P C Y w = (2 6) where ? is the frequency of the setup and C S and C P are the capacitance for the series model and parallel model respectively. The net impedance of the equivalent series and parallel circuits at a given freq uency are equal, but the individual components are not: iB G i R + = C + 1 (2 7)

PAGE 55

41 If there is a lossless circuit (i.e. R = 0 and G = 0) the C S and C P are equal. However, since circuits do have losses, a dissipation factor is added to the system R C D S w = (2 8) P C G D w = (2 9) Conversion from one equivalent circuit model to the other is obtainable through further manipulation of the previous equations yielding the conversion factors shown in Table 2 5 Table 2 5: Conversion factors for series parallel electrical equivalent circuits. Model Dissipation Factor Capacitance Resistance or Conductance Parallel C P G P C G Q D w = = 1 P s C D C ) 1 ( 2 + = G D D R ) 1 ( 2 2 + = Series C S R R C Q D S w = = 1 2 1 D C C S P + = R D D G ) 1 ( 2 2 + = This treatment has been applied to illustrate how capacitance values are generated from the impedance values and subsequent conversion to series and parallel cases. As mentioned earlier, the series represents a physical scenario where a capacitor is in line with some type of resistance, while the parallel mode represents a physical scenario where a capacitor is in parallel with some type of conductance. This however does not take into co nsideration the physical possibility of an ultrathin film that is leaky, but also encounters resistance from the substrate. In this case, a three element electrical circuit, also seen in Figure 2 10, could be analyzed in a similar manner as above to devel op equations for converting from either the parallel or series mode to the three element equivalent circuit that more appropriately represents the physical structure.

PAGE 56

42 Figure 2 10: Equivalent circuits for estimation of capacitance in a) parallel, b) seri es, and c) combination models

PAGE 57

43 CHAPTER 3 KINENTICS OF INTERFA CIAL LAYER FORMATION DURING DEPOSITION O F HAFNIA ON SILICON Oxide thin films are essential components of many advanced structures and devices. One of the most commonly used substrates for thin film deposition, both for appli cations and for properties characterization, is Si wafers. In many instances, the oxide films are directly deposited on silicon substrates. As long as the oxide films are thick, the chemical and physical phenomena occurring at the interface with the Si s ubstrate do not usually play a major role in determining the properties of the deposited oxide films. However, a renewed interest in the growth of very thin (<10 nm) high dielectric constant (high k) oxide films directly on Si has raised the question of t he interfacial layer formation and its properties. 54 56 In this portion of the work, a pulsed laser deposition (PLD) technique was used to deposit HfO 2 high k dielectrics on Si. New results regarding the chemical composition and the kinetics of th e inter facial layer growth during deposition of HfO 2 thin films on Si are presented here in this portion of the work. Deposition Conditions For this study, a 1 diameter x thick hafnium oxide target 99.99% pure from SCI Engineered Materials was ablated using a KrF excimer laser system in a typical PLD setup which was described in detail previously. 57 59 The deposition parameters used were laser fluence of 2 J/cm 2 5 Hz repetition rate, 430 pulses (yield thickness from 3 to 4 nm), substrate temperatures from 200 to 600 o C, and oxygen pressure during deposition from

PAGE 58

44 1x10 4 up to 1x10 2 Torr. Table 3 1 reviews the sample matrix used for the nine samples in this portion of the experiments. The films were deposited on (100) Si p substrates from MEMC cut to 2.5 x2.5 cm squares then washed with acetone and methanol to remove any surface contamination. The samples were then dipped in a 1% HF solution for 10 minutes to remove the native oxide and were mounted onto the resistive heater using silver paste. This clea ning process passivates the dangling bonds of the Si with hydrogen giving between 10 30 minutes to pump down the sample before adsorption of water species and other contaminates to the surface occur. 59 62 Table 3 1: Conditions for growth of ultra thin HfO 2 films. Sample ID# Temp. C Pressure Torr Hf11 1 x10 4 Hf12 1 x10 3 Hf13 200 1x10 2 Hf14 1 x10 4 Hf15 1 x10 3 Hf16 400 1x10 2 Hf17 1 x10 4 Hf18 1 x10 3 Hf19 600 1x10 2 The surface morphology, thickness, and roughness of the deposited la yers were investigated by x ray reflectivity (XRR) with a Panalytical XPert MRD system. The XRR spectra were simulated using the software package Wingixa The same system was used for crystalline structure investigations by grazing incidence and symmet ric q-2q x ray diffraction (GIXD and XRD). The chemical composition and bonding of the films were investigated by x ray photoelectron spectroscopy (XPS) in a Perkin Elmer 5100 installation using Mg K a radiation (1253.6 eV) at various take off angles rangi ng from 45 90. The thickness and optical properties of the films were investigated by

PAGE 59

45 variable angle spectroscopic ellipsometry (VASE) with a J. A. Woollam M 88 instrument using an incidence angle of 75. The thickness, refractive index and extinction c oefficient values were obtained by fitting the acquired data with a Cauchy model using the equipment software package. X ray Results Upon sample fabrication, samples were cut into 1 x 1 cm squares for analysis. Due to the cosine distribution type of plume dynamics the film is not of uniform thickness across the whole sample thus it is necessary to use only the central 1 x 1 cm portion which is consistently the most uniform with a variation of Furthermore this sample size is ideal for all the analysi s techniques. GIXD investigations were conducted at an incidence angle of W =1 for all nine samples. The results from the two extreme temperature conditions of the samples are plotted in Figure 3 1. All samples in the 200 C and 400 C sets remained amo rphous regardless of the oxygen pressure, which is to be expected as the crystallization temperature for HfO 2 is 450 500 C. All films deposited at 600 C exhibited broad peaks, a sign of rather poor and randomly oriented crystallites. Identification of the main peaks and relative intensities was found to match the monoclinic HfO 2 phase according to JCPDS reference card #34 0104. The peak at 55.5 is from the (311) plane of the silicon substrate. By using a glancing angle of 1 the proper diffraction co nditions for this peak is present when the sample is mounted with the <010> or <001> planes in line with the beam. The samples were cut along these directions; typically the samples are mounted into the XPert system so that the beam impinges the sample p erpendicular to the edge and centered on the sample. The (311) peak can be eliminated by using a phi rotation of the sample of 10 45, however no peaks

PAGE 60

46 from HfO 2 are in this region so this alignment step was not used. Symmetric XRD investigations did not reveal any particular texture in these films, the crystals are randomly oriented. After each GIXD and XRD scan was made, an XRR spectrum was taken. The XPert system allows for all three measurements to be made sequentially once sample alignment has be en completed with just a change in the optic slits. Angles from 0.125 to 5 were scanned. A typical XRR spectrum of the deposited HfO 2 films is displayed in Figure 3 2. Each spectrum was modeled using the WINGIXA software package. Also displayed in Figu re 3 2 is the simulation of this spectrum obtained using a three layer structure: interfacial layer (IL), oxide layer (OL), and contamination layer (CL) located on the top surface which takes into account the carbon and water containing species. This thre e layer structure is represented in schematic form in Figure 3 3. Figure 3 1: Comparison of GIXD ( W =1) for the 200 C and 600 C samples deposited in 1x10 4 Torr O 2

PAGE 61

47 Figure 3 2: Acquired XRR spectrum and its simulat ion for a HfO 2 film deposited on Si at 200 o C and 1x10 4 Torr of O 2 Figure 3 3: Schematic of gate oxide stack model used for XRR simulations The parameters used for simulations of all nine samples, i.e. layer thickness, roughness and density, togeth er with the deposition conditions are displayed in Table 3 2. The mean square error of the fit c is also included in the table. For reference, a good fit has c values of less than 1x10 2 To begin the iterative modeling process, the mass density of the hafnium oxide layer was first assumed to be the bulk tabulated value of 9.68 g/ cm 3 One can note that the region of the critical angle (~0.22) was well simulated by using this standard value thus indicating that very dense compact films were produced.

PAGE 62

48 T he hafnium oxide layer density was held constant for all the sample models while the density of the IL and CL were allowed to change along with the thickness and roughness values for all the layers. The contamination layer affects the initial signal fallo ff and is necessary to bring the model in line with the raw data. This layer was initially assumed to have a density of 7.26 g/cm 3 which is 75% of the HfO 2 layer density to account for voids and carbon contamination. From the fit data, the density of the interfacial layer was consistently higher than 2.19 g/cm 3 the density of amorphous SiO 2 This is a clear indication that the interfacial layer is not pure SiO 2 but a mixture containing HfO 2 The interface layer roughness values, both towards the Si su bstrate and towards the deposited oxide layer are rather high, sometimes even higher than the surface roughness values. This is another indication that there was mixing between the SiO 2 and the deposited layer. Table 3 2: XRR modeling parameters used t o fit the spectra. Note values for thickness and roughness are giving in and the densities are given in g/cm 3 Sample ID Thk. IL Thk. OL Thk. CL Rgh. IL Rgh. OL Rgh. CL Dens. IL Dens. OL Dens. CL c Hf11 24.6 41.3 7.8 5.3 3.4 1.8 4.19 9.68 7.07 1.02E 02 Hf12 31.8 34.3 9.2 9.9 2.7 1.8 3.61 9.68 7.42 3.30E 02 Hf13 16.3 31.1 7.9 6.1 3.9 2.1 3.38 9.68 6.32 1.27E 02 Hf14 24.4 37.7 6.7 4.1 7.2 3 .0 4.09 9.68 7.22 1.65E 02 Hf15 22.9 37.9 1.9 3 .0 6.1 1. 1 3.88 9.68 4.35 3.11E 02 Hf16 25.7 30.9 7.9 5.9 3.4 2.4 3.45 9.68 7 .00 7.75E 03 Hf17 25 .0 39.3 3.1 4.8 5.5 2.9 3.12 9.68 6.17 3.27E 02 Hf18 31.3 39 .0 6.8 4.2 2.1 3 .0 3.72 9.68 3.87 2.54E 02 Hf19 33.5 34.4 5 .0 5.3 1.7 5.7 3.46 9.68 5.97 9.55E 03 Incr eased temperature was expected to promote oxygen diffusion and reaction with the silicon at the interface of the film. Average interfacial layer thickness did increase with increasing temperature, however all the data is not sufficient to statistically de rive this conclusion as the 200 C samples Hf11 and Hf12 exceed the thickness of the 400 C

PAGE 63

49 samples. The 200 C samples should be further examined to rule out any experimental errors that might have been encountered which promoted the growth of the interf acial layer. Another trend that was expected is the increase in interfacial layer formation with the increase in oxygen pressure. The trend is sufficiently exemplified by the 600 C sample set H17, H18, and H19 deposited at 1x10 4 1x10 3 and 1x10 2 Tor r O 2 respectively. This trend is shown in Figure 3 4 indicating a nearly linear relationship between the interfacial layer thickness and the log of oxygen pressure. y = 1.8458Ln(x) + 42.683 20 22 24 26 28 30 32 34 36 0.00001 0.0001 0.001 0.01 0.1 Pressure (Torr) IL Thickness (A) 600 deg. C Log. (600 deg. C) Figure 3 4: Interfacial layer thickness dependence on O 2 pressure for the 600 C sample s XPS Analysis XPS was used to determine the composition, bonding environmental changes, and to estimate the thickness of the interfacial layer. Surveys from 0 1000 eV were taken for all the samples to first determine the elemental makeup and to detect a ny contaminates.

PAGE 64

50 Then a multi plex consisting of the C 1s, O Auger, O 1s, Si 2p, and Hf 4d peak regions were collected. The C 1s was used to determine calibration shifts resulting from charging in the samples. All materials analyzed using XPS have suffi cient adventitious carbon contamination on the surface to produce the C 1s signal with known peak position of 284.6 eV. The plots were then fit using software that estimates the peaks with a 90% Gaussian curve. Fit information and analysis tables have be en included in Appendix B. XPS revealed symmetric peaks for the Hf 4d for all the samples, the two extreme conditions are compared in Figure 3 5, indicating one single oxidation state, which was identified as HfO 2 from their binding energies of 19.5 and 21.2 for the d 7/2 and d 5 /2 orbitals respectively. Figure 3 5: High resolution XPS scans of the Hf 4f region acquired from two HfO 2 films deposited on Si under very different conditions. High resolution scans of the Si 2p region were acquired at takeo ff angles of 45 o 65 o and 90 o for the deposited samples. Figure 3 6 shows a typical scan of the Si 2p

PAGE 65

51 region along with a three peak model taken using binding energies ranging from 95 to 108 eV. Peak A centered on 100.88 eV, before the C 1s shift was ta ken into account, represents the signal from Si 0 Si bonded with Si, and peak B located at 104.45 eV is the signal from Si +4 Si bonded with O. The third peak, C which was present in the 600 C samples, is attributed to substoichometric silicon dioxide re presented by silicon bonded to 3 oxygen atoms with a dangling bond. Figure 3 6: Raw data and fit for the Si 2p region for the 600 C, 1x10 2 Torr O 2 sample. The ratio of the intensities of the Si 0 to Si 4+ peaks, represented by peak A and B respectively were then used for estimations of the thickness of the SiO 2 layer according to: ] 1 ) /( ln[ sin + = Si oxy oxy oxy I I d b a l (3 1)

PAGE 66

52 where l oxy =36 is the photoelectron inelastic mean free path in the oxide film, a is the photoelectron take off angle, I oxy is the peak area for the Si O bonding, I Si is the integrated peak area for the signal from the substrate, and b =0.8 is a calibration cons tant for our XPS system which will be further discussed in the next chapter 63 Although it has been claimed that Equation 3 1 could only work for takeoff angles greater than 60 we have found that it can give rather accurate predictions down to 35 The d/ l values for both a 5 nm pure SiO 2 standard and for a sample deposited at 1x10 2 Torr of O 2 and 600 C are plotted in Figure 3 7. The variation between the 65 and 90 measurements for the deposited sample exceeds the error of our system based on the 5 nm SiO 2 standard. This suggests that the interfacial layer is a physical mixture of HfO 2 /SiO 2 If the interfacial layer was uniform throughout, the d / l would have yielded the same result for both the 65 and 90 degree scans. The estimated IL thickness us ing Equation 3 1 and a take off angle of 90 o is displayed in the summary Table 3 4. Figure 3 7: Variations of d/ l values for a 5 nm pure SiO 2 on Si and a deposited sample with changes in the take off angle. -----------------------------------------------------------------------------------------------------

PAGE 67

53 The interfacial layer thickness values obt ained using XPS are plotted in Figure 3 8. The 400 C and 600 C sample sets show a nearly linear increase with the log of oxygen deposition pressures used. The slope of the relationship is the same for both sets. Furthermore, higher temperature increas ed the interfacial layer thickness formation. The 200 C sample set again did not follow the trends seen with the other two sets. Problems such as incomplete removal of the native oxide during the cleaning process might be the cause of this error. Fi gure 3 8: Interfacial layer thickness as a function of temperature and pressure. Ellipsometry Results Finally, the thickness of the deposited structures was estimated using VASE. The acquired spectra were simulated using the three layer model used for X RR simulation. An effective medium approximation Cauchy model consisting of 75% SiO 2 and 25% HfO 2 for the IL and 50% HfO 2 and 50% voids for the CL was used to simulate the acquired data. The simulation data were quite close to the acquired data without a ny

PAGE 68

54 need to modify of the HfO 2 refractive index or add absorption, suggesting high optical quality of the deposited films. The layer thickness values are shown in Table 3 3. Table 3 3: Ellipsometry modeling results Sample ID SiO 2 ( 50%HfO 2 )[ ?] HfO 2 [ ?] HfO 2 +40%void[ ?] Hf 11 25 .0 36.8 9.1 Hf12 31.7 34.9 9.4 Hf13 16.2 39.5 16.2 Hf 14 24.4 36.3 7 .0 Hf15 22.9 36.1 2.4 Hf16 25.6 28.1 8.1 Hf 17 25 .0 42.9 3.3 Hf18 31.6 32.2 5.1 Hf19 33.2 29.8 4.52 Summary XPS XRR and SE investigations were used to deter mine the structure of the HfO 2 thin films deposited on Si. It was found that samples deposited with substrate temperatures of 600 C exhibit small randomly oriented crystals but samples deposited up to 400 C remained amorphous. The interfacial layer was shown to consist of SiO 2 physically mixed with the deposited oxide which always formed, even when the deposition temperature was only 200 C and at low oxygen pressures. The thick ness of this layer is around 1 4 nm, depending on the d eposition conditions used. I ts density is greater than that of pure bulk SiO 2 resulting from a physical mixture of silicon suboxides with the grown oxide layer. By inspecting the results in Table 3 4, one can clearly note that there is a good agreement between the layer thic kness values obtained using these three methods. The higher the deposition temperature and/or oxygen pressure used during deposition, the thicker the IL. Also, the SE and XRR results support the XPS indication that the IL contains a physical mixture betw een the SiO 2 layer and the deposited HfO 2 oxide, similar

PAGE 69

55 Table 3 4: Comparison of the thickness values (in ) for HfO 2 thin films deposited on Si XRR Ellipsometry XPS Sample ID# Temp. o C Pressure Torr IL ? OL ? CL ? IL ? OL ? CL ? IL ? Hf11 1 x 10 4 24.6 41.3 7.8 25.0 36.8 9.1 19.4 Hf12 1 x10 3 31.8 34.3 9.2 31.7 34.9 9.4 27.5 Hf13 200 1x10 2 16.3 31.1 7.9 16.2 39.5 8.1 16.4 Hf14 1 x10 4 24.4 37.7 6.7 24.4 36.3 7.0 19.5 Hf15 1 x10 3 22.9 37.9 1.9 22.9 36.1 2.4 21.3 Hf16 400 1x10 2 25.7 30.9 7 .9 25.6 28.1 8.1 23.5 Hf17 1 x10 4 25.0 39.3 3.1 25.0 42.9 3.3 30.5 Hf18 1 x10 3 31.3 39.0 6.8 31.6 32.2 5.1 31.0 Hf19 600 1x10 2 33.5 34.4 5.0 33.2 29.8 4.5 33.8 to our previous findings done by our group for Y 2 O 3 60 and ZrO 2 61 films deposited on Si By understanding the nature of the interfacial layer and its formation, better control of the layer can be realized to tailor the properties of the gate stack in future samples.

PAGE 70

56 CHAPTER 4 E FFECT OF UV ASSISTED O XIDATION AND NITRICA TION OF HAFNIUM METAL FILMS Hafnium oxide has been shown to be a promising candidate for an alternative high k dielectric due to its high dielectric constant, large bandgap, and stability on silicon. 50 52 However processing temperatures required for standard deposition techniques are too high and exceed the crystallization temperature for the films thus resulting in current leakage and diffusion paths at the grain boundaries. Work has been done using u ltraviolet assisted oxidation of zirconium metal films deposited using a radio frequency direct current sputtering method which was shown to produce higher quality films at lower processing temperatures. 45 49 Furthermore, nitridation of various metal oxid e films has been used to retard the growth of interfacial layers by acting as a diffusion barrier to oxygen. 6 4 68 This portion of the work extends these investigations to the hafnium metal system in an attempt to understand the effect of UV oxidation and the incorporation of nitrogen into the films on their electrical and structural properties under lower processing temperatures. Experiment Initially, it was believed a thin Si x N y layer grown at the interface using UV radiation would serve as a superior bas e for the deposition of the films in this portion of the work. The silicon wafers were cut into 2.5 x 2.5 cm squares and mounted on the resistive heater and placed into the vacuum chamber. After pumping to 1x10 6 Torr, the chamber was backfilled with 500 Torr of ammonia. The substrate was then heated to 650

PAGE 71

57 C and irradiated with the UV lamp array for 30 minutes. This formed a 20 layer of Si x N y verified by XPS and ellipsometry. Capacitors were formed out of the films and capacitance/voltage measureme nts were taken. Figure 4 1 is the CV response of one of the nitrided silicon films that was tested. As can be seen from the plot, the response is not sharp but instead is elongated over 2 volts. This is a result of extreme amounts of interface traps cre ated at the silicon/silicon nitride interface. Furthermore, at higher negative bias the capacitance is not constant but instead starts to drop off. This is suggestive of a poor quality film which is leaky. From this result in order to eliminate the char ge trapping defects inherent when using silicon nitride, it was decided to deposit the hafnium films onto bare silicon. Figure 4 1: CV response of an UV grown Si x N y film.

PAGE 72

58 A 248 nm KrF excimer laser system in a typical pulsed laser deposition (PLD) set up was used to ablate a hafnium metal target 99.95% pure from SCI Engineered Materials. The deposition parameters used were laser fluence of 3.0 J/cm 2 at a 5 Hz pulse rate with a total pulse count of 1500 for metal film thickness of 35 and oxidized film thicknesses of 45 50 The films were deposited at 350 C on p (100) Si substrates from MEMC (9 18 W cm) that were cleaned using an SC1 solution. This solution contains 80 vol% distilled water, 16 vol% hydrogen peroxide diluted to 30 vol% in distille d H 2 O, and 4 vol% ammonium hydroxide. The samples were then dipped in a 1% HF solution for ten minutes to remove the native oxide. Depositions were performed either under residual vacuum at 1x10 6 Torr or under 1x10 2 Torr of ammonia. Post deposition an neals were carried out in situ in 300 Torr of O 2 at 400 C both with and without a Hg lamp array for 30 minutes. Each lamp emits 254 and 185 nm wavelength radiation at a total power density of 200 mW/cm 2 The power density from the 185 nm wavelength port ion of the emission spectra is 10 15 mW/cm 2 Previous experiments determined that the use of UV during deposition in ammonia had no effect on the films at the deposition temperature of 300 C. At 300 C even with the activation of the ammonia by the remo val of one or two hydrogen atoms due to UV interaction, there is not sufficient energy to promote bonding with the hafnium metal being deposited or with the silicon substrate. Therefore all depositions were conducted with the UV lamps off. A summary of t he sample matrix is given in Table 4 1. This data set was repeated 3 times to study repeatability and to estimate error. The surface morphology was studied using atomic force microscopy (AFM, Digital Instruments Nanoscope III) both in contact and tapping modes. Crystallinity

PAGE 73

59 Table 4 1: Growth and postdeposition conditions for UV oxidation of Hf metal ultra thin films. Sample ID # Deposition Pressure Torr Anneal Conditions H2 1x10 6 UV H3 1x10 6 No UV H4 1x10 2 Ammonia UV H5 1x10 2 Ammonia No UV c omparisons and phase identification was done using grazing incidence x ray diffraction (GIXD, Panalytical XPert system). The chemical bonding and composition of the films was investigated by x ray photoelectron spectroscopy (XPS, Perkin Elmer 5100) using Mg K a radiation (1253.6 eV). A calibration study using a 50 SiO 2 film on Si standard was also carried out with the XPS system to fit the results of the angle resolved measurements. High resolution spectra of the Si 2p region from the standard film were acq uired at takeoff angles from 15 to 90 and the area ratios of the Si 0 and Si 4+ peak were compared to obtain the calibration constant for our system. Finally MOS capacitors were fabricated by sputter deposition of 1000 thick platinum electrodes using a shadow mask having circular dot patterns of various sizes. Capacitance voltage and current voltage measurements were taken using a Keithley Instruments Inc., 590 capacitance meter, KI236 source measurement unit (SMU), and the Kiethley Instruments Inc. Win 82 measurement system. Confirmation of results and further analysis was done at the University of Texas Austin. Cross sectional transmission electron microscopy (XTEM, JEOL 2010) was used to verify thickness and layer uniformity after electrical measure ments were concluded for two of the samples. An additional set of four samples was deposited using substrate temperatures of 650 C. At this temperature, it was believed that a more significant amount of nitrogen could be incorporated into the hafnium met al thin film. Previous work done by N.

PAGE 74

60 Bassim et al showed that UV radiation from our lamp array promotes the activation of ammonia into reactive species which then form a nitride layer on the surface of silicon substrates. Significant nitride layer gro wth occurred at substrate temperatures of 650 C. Nitrogen does not readily react with metals and requires large activation energies. Thus for this set of experiments substrate temperatures of 650 C were used to afford greater thermal energy for the rea ction. Furthermore effects of UV during deposition and the oxidation annealing steps were compared. Finally, the electrical properties of the high temperature deposited samples were compared to the first set of samples deposited at 350 C. Results and Di scussion for Low Temperature Deposited Samples Surface Morphology Atomic force microscopy was first conducted using contact mode for all four samples. An area of 1 m m x 1 m m was scanned using a tip capable of near atomic resolution on relatively smooth sa mples when properly calibrated. The 2D scans obtained are shown in Figures 4 2 and 4 3. Root mean square roughness was calculated for each sample based on the scans shown. These results are given in Table 4 2. The UV oxidized samples, H2 and H4, exhibi t greater Z ranges and higher RMS roughness values. There also is a slight but noticeable decrease in roughness with the addition of nitrogen. The quality of the scans obtained using contact mode was poor, possibly due to washout from tip effects, so samp les were reanalyzed using tapping mode, which affords higher resolution as the tip does not make contact with the sample but instead rides above the sample at a predetermined distance based on van der Waals forces and then is

PAGE 75

61 Table 4 2: Z range and RMS r oughness for oxidized Hf metal films. Sample ID Z range RMS Roughness H2 58.7 3.7 H3 24.6 2.3 H4 33.6 2.5 H5 26.5 2.0 cyclically flexed toward the sample during the measurement. The 3D scans taken from samples H2 and H3 are plotted in Figure 4 4 From the tapping mode results, the surface of the UV irradiated samples again showed a greater degree of roughness. The UV irradiated sample exhibited a z range of 7.8 nm and a RMS value of 6.4 while the non irradiated sample had a z range of 1.7 nm and a RMS value of 1.2 It is believed that the UV radiation promotes crystallization of the films. This effect has been seen before with BST films that our group has previously reported. 69 The increase in roughness is due to crystallization resulting from the UV radiation acting as a non thermal energy source thus affording sufficient energy for promoting nucleation and growth of randomly oriented crystals at temperatures below the crystallization temperature. The 185 nm radiation is absorbed by the oxygen within 5 cm of the lamps based on the measured falloff of the intensity done at atmosphere with a detector head specifically designed to measure 185 nm radiation. Each sample is 8.5 cm from the nearest lamps, so the crystallization enhancement must be from the 254 nm radiation (4.88 eV). X Ray Analysis Upon examination of the x ray data collected, it was shown that the UV radiation did in fact promote crystallization in the films. GIXD spectra taken at W =1, plotted in Figure 4 5, demonstrate the e mergence of a monoclinic HfO 2 phase (JCPDS 34 0104)

PAGE 76

62 Figure 4 2: 2D AFM scans of samples a) H2 and b) H3 taken using contact mode. The scale is in m m. a) b)

PAGE 77

63 Figure 4 3: 2D AFM scans of samples a) H4 and b) H5 taken using contact mode. The scale is in m m. a) b)

PAGE 78

64 with the UV assisted oxidation. The radiation not only acts to break the oxygen into more reactive species, but also acts as a non thermal energy source providing the activation energy for adatoms diffusion, crystal nucleation, and growth. Similar eme rgence of the monoclinic phase of HfO 2 was noted in the nitrided samples with no noticeable change in the lattice parameters. X Ray Reflectivity As deposited metal films were analyzed using XRR in the Panalytic XPert system. A typical scan and fit is sh own in Figure 4 6. The films were all 40 with a density of 12.9 g/cm 3 and a mixed interface of ~8 having a density of 6.3 g/cm 3 The density of the metal layer suggests that 1 3% porosity was present as the bulk density of Hf metal is 13.3 g/cm 3 XPS Analysis XPS was used at take off angles of 45 and 90 for all of the scans in order to compare changes in composition with depth in the film and to better understand the nature of the silicon/film interface. Prediction of SiO 2 layer thickness and uniformity can be obtained using the previously mentioned Equation 3 1: ] 1 ) /( ln[ sin + = Si oxy oxy oxy I I d b a l (3 1) where l oxy is the photoelectron effective attenuation length in the oxide film, a is the photoelectron take off angle, I oxy is the peak area for the Si O bonding, I Si is the peak area for the signal from the substrate, and b is the calibration constant for the s ystem. 63 Our system was calibrated using the high resolution Si 2p scans, shown in Figure 4 7, taken at angles from 15 90. From these results, values of b = 0.8 and l = 36 were obtained and d/ l was calculated to be 1.40.05 for angles 45 90.

PAGE 79

65 Figure 4 4: 3D AFM micrographs of both a) UV oxidized and b) non irradiated hafnium films. a) b)

PAGE 80

66 Figure 4 5: GIXD spectra ( W =1) of Hf films deposited on Si and oxidized with and without UV. Figure 4 6: XRR plot and fit for as deposited Hf metal film.

PAGE 81

67 The Si 2p scans plotted in Figure 4 8 were fit with two 90% Gaussian curves representing the Si Si and Si O bonding. The areas were used to calculate the d/ l values for both the 45 and 90 scans. The results of these calculations are given in Table 4 3. For all but one of the samples the d/ l values were nearly identical. Therefore the interfacial layer is primarily SiO 2 /SiON x with very little degree of mix ing occurring between the silicon dioxide and hafnium oxide layers. Sample H3 (no nitrogen, no UV) was the only sample with a mixed interface. The UV radiation is believed to promote the separation of the silicon dioxide and hafnium oxide into two distin ct layers, while the nitrogen prevents the initial mixing normally encountered with PLD deposition. Table 4 3: d/ l values for samples H2 H5 Sample ID Angle Si Si Area Si O Area d/ l H2 45 400 1370 1.176 90 986 1572 1.096 H3 45 586 833 0.722 90 817 1532 1.207 H4 45 419 1239 1.094 90 1207 1660 1.000 H5 45 3506 2042 0.547 90 1676 1865 0.616 The Si 2p scans also show a larger percentage of Si O peak area in the UV irradiated samples and a peak shift of 0.5 eV to higher binding energies appr oaching the reported value of 103.6 eV for silcon bonded to oxygen in a pure thick SiO 2 layer. 60 Therefore, the UV radiation is promoting oxidation of the interfacial silicon. The O 1s data for the same samples, plotted in Figure 4 9, shows a similar tre nd whereby the irradiated samples have a higher energy shoulder located at binding energies of 535 eV representing greater concentrations of Si O bonding. Peak fit analysis using two peaks, one at 531.5 eV representing the oxygen atoms bonded to silicon a nd the other peak at

PAGE 82

68 533 eV indicative of oxygen bonded with Hf, indicates an increase of 30% Si O bonding in the UV irradiated samples. 0 5000 10000 15000 20000 25000 30000 35000 40000 97 99 101 103 105 107 Energy (eV) Counts Figure 4 7: Angle resolved XPS Si 2p spectra for system calibration. Fi gure 4 8: Angle resolved XPS Si 2p spectra for H2 and H3 (UV and no UV films). Si2p Si O Si Si 15 30 45 60 75 90 UV 45 UV 90 no UV 90 no UV 45 Si2p

PAGE 83

69 16000 26000 36000 46000 56000 66000 76000 86000 96000 529 531 533 535 537 Energy (eV) Counts Figure 4 9: O 1s spectra for H2 and H3 (UV and no UV) films at 45 and 90. According to XPS chemical composition analysis, depositing in ammonia only allowed for less than 2 atomic% incorporation of nitrogen into the films. Further trials using UV radiation during deposition were tested to attempt to increase the concentration of nitrogen, but yielded the same result. It can be concluded that the energy of the radiation is not sufficient to promote highly significant amounts of reaction of the ammonia with the depositing hafnium metal or with the silicon substrate at the substrate temperature of 350 C. Even with just 2% incorporation of nitrogen into the films, the interfacial layer of the nitrided films have a larger ratio of Si Si peak area suggesting less interfacial growth as shown in Figure 4 10. The Si O peak also undergoes a shift of 1 eV to lower binding energies due to the chang e the bonding environment of the silicon atoms at the interface. The shift could result from the formation of a thin Si 3 N 4 layer. Binding energies for Si N type species lie within the intermediate range between the Si Si and Si O peaks. For fully nitrid ed silicon this peak is located at 101.8 eV. UV 90 no UV 45 no UV 90 O 1s UV 45

PAGE 84

70 Figure 4 10: Si 2p spectra for UV oxidized films H4 and H2 (with and without nitrogen). In order to confirm the nature of the nitrogen in the film, high resolution XPS scans were taken for the N 1s peak range. Due to an overlap of the N 1s peaks and a higher energy satellite peak of the Hf 4p3 peak, an oxidized Hf metal film was used as a standard to subtract out the background resulting from the hafnium peak. The result of the as deposited Hf metal fi lm is shown in Figure 4 11. The majority of the peak occurs at a binding energy of 397.5 eV which is near the reported value of 396.3 eV for Hf N. Granted the substrate temperature of 350 C is not sufficient to promote a significant degree of bonding. Upon annealing the peak broadens out and shifts to higher binding energies, as shown in Figure 4 12. The broad peak is indicative of multiple bonding environs. The shift to higher binding energies during the oxidation anneals results from the nitrogen mo ving to the silicon interface. Si N bonds have been reported to have values in the range of 398 to 399 eV. Furthermore upon annealing, the intensity of the signal diminishes. It is believed that a majority of the nitrogen within the metal film is replac ed by oxygen when annealed. Comparing the UV to the non UV annealed sample Si 2p with nitrogen without nitrogen

PAGE 85

71 (i.e. H4 and H5 respectively) shows a similar peak profile but a loss of intensity in the nitrogen signal. The UV radiation promotes the replacement of the nitrogen within the fil m and at the interface. Figure 4 11: N 1s peak of as deposited nitrided Hf metal film. Figure 4 12: N 1s peak after PDA for samples H4 and H5. 390 393 396 399 402 405 408 411 414 Binding Energy (eV) Counts (a. u.) N 1s Hf N Si N 391 394 397 400 403 406 409 Binding Energy (eV) Counts (a. u.) Hf N Si N N 1s H5 H4

PAGE 86

72 Electrical Properties Analysis Upon completion of the chemical and structural analysis, the sample surfaces were washed with acetone then methanol and blow dried with dry nitrogen. Then 1000 platinum electrodes were sputtered on the surface using a radio frequency (RF) sputtering system. The shadow mask used for this step had repeated sets of circular dots of various sizes. The size dot used for this portion of the experiments had an area of 4.66x10 4 cm 2 A forming gas anneal was then carried out at 450 C for 30 minutes in a flowing gas environment us ing a gas mixture of 4% H 2 and 96% N 2 The forming gas anneal step is important because the hydrogen passivates the defects between the surface of the metal and the film. The backside of the samples were then abraded and swabbed with a 1 % HF to remove a ny native oxide. The samples were then mounted on glass microscope slides using a silver paste which was then cured at 150 C on a hotplate to remove the organic. The silver forms an ohmic contact with the p type silicon for proper electrical data collec tion. Capacitance Voltage Results Capacitance voltage measurements were taken for a series of seven dots on each sample for both the 100 kHz and 1 MHz input frequencies to compare the frequency dependence on the capacitance response. Furthermore, scans we re taken using both the parallel and series models. Figure 4 13 compares these results for sample H4, the UV oxidized sample deposited in ammonia which overall had the best electrical characteristics of the samples. There is a large dependence of the cap acitance on the frequency used for the measurement. This sample exhibited a 40% decrease in capacitance between the parallel models when going from 100 kHz to 1 MHz. The change in the response is mainly due to the choice of substrates. The low doping le vel of

PAGE 87

73 ~2x10 14 /cm 3 provides very few charge carriers. Thus when in the accumulation region where a negative bias is being applied to the gate, the depletion region in the channel beneath the gate becomes large. As a result, the RC time constant for the carriers to move to respond to the frequency at which the voltage is being applied becomes significant when 1 MHz measurements are carried out. Thus the 1 MHz capacitance response is less than that of the 100 kHz. Ultimately films with no frequency dispe rsion are desired and the films should be tested on standard p type wafers instead of the p ones that were used. This plot also demonstrates the need to choose the proper model for the sample type being studied. For ultra thin film oxides (<100 ), like the samples in this work, the parallel model is best suited due to increased leakage resulting in a conductance term which must be taken into account. The high frequency (1 MHz) parallel model scans were used for comparison of each of the samples. Figure 4 14 shows a plot of these results. From the graph it can be seen that the small incorporation of nitrogen into the films by depositing in ammonia greatly increases the capacitance of the film. This is attributed to limited interfacial layer formation d uring oxidation and a resulting high dielectric constant for the interface. Furthermore the use of UV oxidation during post deposition annealing further improves the capacitance of the stack by increasing the quality of the hafnium oxide layer. The thres hold voltage V T for the samples also underwent a shift from ~ 1.5 V for sample H3, the non irradiated, non nitrided sample, to 0 V for sample H4. The curve response shape is also improved by nitrogen and UV treatment suggesting minimization of interfacial and bulk defects such as oxygen vacancies. Thus combination of both nitrogen incorporation and UV irradiation results in higher quality electrical characteristics.

PAGE 88

74 From the capacitance response data, the oxide gate stack dielectric constant was calculate d using equation 4 1. Thickness values used for the calculations for the films were taken from the XRR and VASE modeling results. The equivalent oxide thickness was calculated using equation 4 2. 0 e e = A t C (4 1) C A EOT O = e 9 3 (4 2) e = dielectric constant for the film C = capacitance A = area of the MOS capacitor e 0 = permittivity of free space EOT = equivalent oxide thickness Compilation from these calculations is presented in Table 4 4. The use of UV irrdation and nitr ogen incorporation into the films has been shown to produce film stacks with EOT values of less than 10 Table 4 4: Dielectric constants and EOT for samples H2 H4. Sample ID k EOT H2 10.6 18.8 H3 9.4 21.3 H4 21.0 9.1 H5 13.8 14.5 Repeatability The sample set was repeated twice using the same experimental setup to confirm the results. Errors were estimated using the root mean square error method. From the results it is shown that the data is highly repeatable and yields nearly identical values between each set.

PAGE 89

75 Figure 4 13: Comparison of high and low frequency capacitance responses for sample H4. 0.0E+00 5.0E+02 1.0E+03 1.5E+03 2.0E+03 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 Voltage (V) Capacitance (pF) H2 H3 H4 H5

PAGE 90

76 Figure 4 14: High frequency capacitance voltage responses for samples H2 H4. Table 4 5: Repeatability study results for k and EOT of oxidized hafnium metal films. Set 1 Set 2 Set 3 Sample ID k EOT k EOT k EOT Avg. k Avg. EOT H2 10.6 18.8 10.5 19 .0 10.8 18.6 10.6.1 18.8.1 H3 9.4 21.3 9.5 21.2 9.6 21 .0 9.5 0.1 21.2 0.1 H4 21 .0 9.1 20.9 9.3 21.2 8.9 21 .0 0.1 9.1 0.1 H5 1 3.8 14.5 13.6 14.7 13.5 15 .0 13.6 0.1 14.7 0.2 Interfacial Trap Density Measurements The samples were sent to the University of Texas Austin for further analysis. Using a similar system setup to ours, capacitance voltage measurements were taken at 1 MH z in a parallel model mode. These results were then fit by the Terman method, discussed further in Appendix A, using an in house Mathcad program which simulates the CV data based on the metals used for contacts and the known film thickness and contact are a. The plots generated from the fits of the models give the quasi static representation of the MOS stack. By comparing the fit to a curve generated based on an ideal film with no defects, an estimate of interfacial trap defect density, D it can be yielde d. The results of the defect density calculations are given in Table 4 6. Figure 4 15 plots the data and the fits obtained from the three samples which could be modeled. Sample H3 was too poor of quality for any information to be properly fit by this me thod. All the samples had average defect densities ranging in order of magnitude from 10 11 10 12 /cm 2 The UV oxidized samples H4 and H5 had the highest defect densities. In order for an alternative high k oxide to be used it must be able to compete with the low defect densities of 10 10 1/cm 2 obtained using silicon dioxide.

PAGE 91

77 Table 4 6: Defect density results for samples analyzed at UT Austin. Sample ID D it (1/cm 2 ) H2 3.07E+11 H4 2.18E+12 H5 1.73E+12 Figure 4 15: CV response and model fit data for s amples H2, H4, and H5. Current Voltage Results The current voltage (IV) characteristics of the films were obtained by scanning MOS capacitors using a voltage range of 2 to 2 volts with the Keithly Win 82 measurement system. Figure 4 16 shows the IV resul ts for the oxidized hafnium metal films. All samples exhibited leakage current densities of less than 5x10 4 A/cm 2 at voltages of 1.2 V. This is the standard operating voltage that industry uses to compare leakage. For a film to be used it must have le akages less than 1x10 2 A/cm 2 at this voltage to stay within the power consumption guidelines. Sample H4, which had the highest capacitance, exhibited the lowest leakage for both the negative and positive bias regions. The use of nitrogen incorporation a nd UV assisted oxidation together provide a

PAGE 92

78 synergistic effect creating higher quality films with less defects which promote leakage paths even though this film had the highest degree of crystallinity. Figure 4 16: Current voltage results from oxidized hafnium metal films. TEM Analysis Cross sectional TEM samples were made from samples H4 and H5 to observe the film uniformity, crystallinity, and interface quality. Figures 4 17 and 4 18 depict a portion of samples H4 and H3 respectively. The thickness of the film and the interfacial thicknesses were measured from the micrograph. Sample H4 clearly shows signs of polycrystalline growth, evidenced by grains of roughly the size of the thickness of the film. EDS line scans were attempted but the drift was too great to get accurate readings across the thickness of the samples. The degree of sample tilt required (~10 ) to acquire the measurement further hinders accurate elemental analysis. Scans taken at specific points in the interface region for both sam ples did show mostly Si and O with a trace

PAGE 93

79 amount of Hf. From both the EDS and XPS results previously, the interface for all the samples is a distinct uniform SiO 2 layer (nitrogen is unable to be detected using this analysis technique). Ultraviolet radia tion was shown to increase the growth of the interfacial layer. However by incorporating nitrogen into the film during deposition, the growth of the layer is hindered. Figure 4 17: XTEM of the UV annealed nitrided sample H4. Proposed Model Based on th e data from this set of four samples, UV radiation during the oxidation anneal promotes crystallization which, in and of itself, is a detrimental quality for the gate oxide. A polycrystalline film has grain boundaries which act as leakage paths for curren t

PAGE 94

80 through the film and also serve as diffusion pathways for oxygen and other contaminants. Furthermore, UV radiation promotes oxidation of the interface, thus parasitically reducing the benefits of using a high k dielectric. The benefits of UV radiation, however, are improved oxygenation of the film and reduction of physiaborbed and physically trapped oxygen. Lower temperatures can be used to fully oxidize the film and eliminate sub oxygen species. Thus UV radiation promotes the formation of higher qual ity oxide films. In order to counter some of the detrimental effects of UV assisted processing, nitrogen was incorporated into the film during deposition. The nitrogen was shown to reduce the formation of the interfacial silicon dioxide layer by both a cting as a diffusion barrier to the oxygen and by the oxygen preferentially replacing the nitrogen within the film structure. The XPS results suggest a thin silicon nitride layer was formed at the interface. Nitride barrier layers have been used as oxyge n diffusion barrier layers for years in the semi conductor industry. An ultra thin nitride layer will slow the reaction of the highly active oxygen species created by UV radiation. Furthermore, hafnium has a greater affinity for oxygen. The thermodynami cs behind the reactions favor the formation of hafnium oxide over hafnium nitride. Therefore with an overpressure of oxygen species and no nitrogen during the oxidation anneal, the oxygen replaces the nitrogen within the structure and also drives out the physically trapped nitrogen. Ultimately nitrogen incorporation of the films controls the interface quality of the films, thus countering the detrimental effect of UV assisted processing. Figure 4 19 shows a schematic of the effects of UV and nitrogen on the films.

PAGE 95

81 Figure 4 18: XTEM of non UV annealed hafnium film H3. Results and Discussion of the High Temperature Depositions The motivation behind this set of samples was to first attempt to increase the degree of nitridation of the films during deposit ion by increasing the substrate temperature during processing. Then comparisons in the use of UV during high temperature deposition and annealing after were made to judge the effect of UV on the nitridation and film structure. Ultimately electrical measu rements were conducted on these samples to compare the low temperature and high temperature processing methods

PAGE 96

82 Figure 4 19: Model of the effect of UV and nitrogen. XPS Elemental analysis carried out by comparing the area of the scans for the O 1s, N 1s, Hf 4f, and Si 2p regions and then taking into account for the sensitivity factors yielded nitrogen percentages of 4 5 atomic % for all four samples. Therefore by depositing using substrate temperatures of 650 C an additional 1 2% of nitrogen can be inco rporated into the films. Figure 4 20 shows the comparison of the O 1s peaks for these samples. Figure 4 21 compares the Hf 4f peaks for these samples. These films were thicker than previous samples, ranging from 80 120 so detailed Si 2p peaks were di fficult to obtain due to the weak signal resulting from the large escape depth that the electrons from the Si atoms near the interface have to travel. The O 1s peaks for the UV deposited samples (HN1 and HN4) exhibit a higher energy shoulder suggesting ~10 15% more Si O type bonding environs. This is believed to be a result of initial crystallization of the films during deposition resulting in a higher

PAGE 97

83 percentage of grain boundaries for the diffusion of oxygen to the interface. Thus upon annealing the oxygen can readily move to the interface to react with the silicon. The Hf 4f peaks were all identical in shape, position, and ratio of the 4f 5/2 and 4f 7/2 peaks. Therefore UV does not affect the main bonding in the bulk of the hafnium oxide layer. Figure 4 20: O ls peaks for the high temperature deposited sample set. X Ray Analysis Glancing incidence angle measurements were made at W =0.5 for all the samples and are plotted in Figure 4 22. All of the samples exhibited a high degree of crystalliz ation. The FWHM for the (111) peak located at ~28 of the UV deposited and UV annealed sample (HN1) was 0.1 narrower than the other samples suggesting a slightly greater degree of crystallinity possibly due to crystal growth. This result follows

PAGE 98

84 the pre vious observance of the promotion of crystallization resulting from using UV during processing. Figure 4 21: Hf 4f peaks for the high temperature deposited sample set. X ray reflectivity scans were also taken for all of the samples and are plotted alon g with the models in Figure 4 23. The XRR scans show that UV during deposition increases the growth of the interface layer as was predicted from the XPS O 1s scans. Furthermore, the film processed without UV for both the deposition and anneal has the lea st amount of interfacial oxide growth. The interface density is larger than SiO 2 for all the samples suggesting either there is mixing of SiO 2 and HfO 2 The higher density also could result from nitridation of the interface as the density of silicon nitr ide is ~4 5 g/cm 3 Low roughness values for the interface and the film layer suggest abrupt interfaces and relatively smooth uniform films.

PAGE 99

85 Figure 4 22: GIXD for the high temperature deposited samples taken at W = 0.5. Electrical Characterization Re sults Capacitance/voltage and current/voltage measurements were taken for the high temperature deposited sample in a similar manner to the previous samples. The results of the capacitance/voltage measurements are shown in Figure 4 24. The data set HN1 wa s offset to avoid confusing overlap with HN3. From the results it is shown that UV during the deposition is detrimental to the electrical properties of the film especially when the sample is not UV annealed during the oxidation step. Sample HN4 (deposite d with UV and annealed without UV) exhibits similar characteristics to the Si x N y samples that were first produced. The response has a large stair step due to electrical traps and then falls off at large negative bias. Similarly, sample HN3 (deposited and annealed without UV) exhibits a stair step in the depletion region. Upon UV oxidation annealing, the defects

PAGE 100

86 Figure 4 23: XRR plots and models for the high h emperature deposited samples

PAGE 101

87 are eliminated. The estimated k and EOT values for these films were calculated and are tabulated in Table 4 7. The highest k obtained was 13.8 for sample HN3 (no UV during deposition or annealing) with an EOT of 27.5 The higher temperature during processing is hindering the capabilities of the material to reach k values of the expected 25 for HfO 2 Leakage currents were measured using a bias scan from 4 to +4 V. Larger scan ranges were not possible for two of the samples because they were leaky and exceeded the compliance current preset in the Keithley equipment before the full bias range could be completed. The IV results plotted in Figure 4 25. The samples which were UV annealed exhibited the lowest leakage currents. Furthermore, UV during deposition causes a 1 2 order of magnitude increase in leakage. This could be a result of UV causing crystallization during deposition which creates a poorer initial film with many grain boundary leakage paths. By UV annealing the samples during oxidation, the oxygen defects and many of the interface traps are eliminated and thus the leakage is reduced. Table 4 7: EOT and k for the high temperature deposited samples Sample ID EOT k HN1 UV/UV 41.4 11.5 HN2 N/UV 33.1 12.2 HN3 N/N 27.5 13.8 HN4 UV/N 82.5 5.7 Discussion From XPS, UV radiation during deposition did no t improve the quantity of nitrogen incorporated into the films but does increase the subsequent formation of interfacial oxide

PAGE 102

88 Figure 4 24: CV plots for the high temperature deposited samples Figure 4 25: IV plots for the high temperature deposited s amples

PAGE 103

89 upon annealing. This result is supported by the XRR results. From the electrical data, the UV radiation during deposition increased the leakage in the films by 1 2 orders of magnitude. Furthermore degradation of the capacitive response occurs if the samples were not ann ealed in UV after deposition. Use of UV during the oxidation anneal is key to eliminating the oxygen defects and charge trapping defects present within the films. This is clearly shown by the elimination of the stair step within the depletion region of t he CV curve and the larger slope of the response curve. Summary Hafnium metal films were deposited using PLD in both residual vacuum and UV excited ammonia and were then annealed in situ in O 2 both with and without UV. It was shown that the UV radiation caused greater crystallinity and roughness in the films due to the non thermal energy source that the radiation provides promoting nucleation and growth of small randomly oriented crystallites below the crystallization temperature of the film. XPS analysi s showed that the UV radiation promoted the formation of a mixed HfO 2 /SiO 2 interface layer. Nitrogen incorporation was shown to reduce the formation of the SiO 2 layer, counteracting the detrimental effect of the UV oxidation process. By combining both th e nitrogen incorporation and UV oxidation step films with EOT of less than 10 can be obtained. Leakages values which meet the industry standard were also realized using this process. The D it for the films was two orders of magnitude higher than that fo r silicon dioxide gate oxide layers, so more work must be done to bring this level down into an acceptable range. By understanding the effect of nitrogen and UV assisted oxidation of metal films, higher quality films with low interfacial formation can be produced.

PAGE 104

90 CHAPTER 5 UV ASSISTED OXIDATION A ND NITRIDATION OF HA FNIUM ALUMINUM METAL ALLOY FILMS Aluminum additions have been shown to increase the crystallization temperature and stabilize high k oxides so they can withstand the higher temperature processing steps which are encountered after deposition. 34 70 72 This portion of the work extends these investigations to the UV and nitrogen treated sample sets from the previous section to understand the role aluminum has on the electrical and structural properties of the film. Experiment A 248 nm KrF excimer laser system in a typical pulsed laser deposition (PLD) setup with the multi target carousel was used to alternatively ablate a hafnium metal target, 99.95% pure, and an Al metal target, 99.95% pure, from SCI Engin eered Materials. Pulse ratios of 1:1, 1:2, and 1:3 for the aluminum and hafnium targets were used and the total pulse count was 1500 for target thicknesses after oxidation of 80 100. The deposition parameters used were laser fluence of 3.0 J/cm 2 at a 5 Hz pulse rate. The films were deposited at 350 C on p (100) Si substrates from MEMC that were cleaned using an SC1 solution. This solution contains 80 vol% distilled water, 16 vol% hydrogen peroxide diluted to 30 vol% in distilled H 2 O, and 4 vol% ammon ium hydroxide. The samples were then dipped in a 1% HF solution for ten minutes to remove the native oxide. Depositions were performed either under residual vacuum at 1x10 6 Torr or under 300 Torr of ammonia. Post deposition anneals were carried out in situ in 300 Torr of O 2

PAGE 105

91 at 400 C with the Hg lamp array on for 30 minutes. A summary of the sample matrix is given in Table 5 1. Table 5 1: Sample deposition conditions Sample ID # Pulse Ratio 1x10 2 Torr Ammonia AH1 1 Al: 1 Hf No AH2 1 Al: 2 Hf No AH3 1 Al: 3 Hf No AHN1 1 Al: 1 Hf Yes AHN2 1 Al: 2 Hf Yes AHN3 1 Al: 3 Hf Yes Crystallinity comparisons and phase identification was done using grazing incidence x ray diffraction (GIXD, Panalytical XPert system). The chemical bonding and compositio n of the films was investigated by x ray photoelectron spectroscopy (XPS, Perkin Elmer 5100) using Mg K a radiation (1253.6 eV). Finally MOS capacitors were fabricated by sputter deposition of 1000 thick platinum electrodes using a shadow mask having cir cular dot patterns of various sizes. Capacitance voltage and current voltage measurements were taken using a Keithley Instruments Inc., 590 capacitance meter, KI236 source measurement unit (SMU), and the Kiethley Instruments Inc. Win 82 measurement system Samples AH3 and AHN3, which contained the least amount of aluminum were then annealed in a conventional tube furnace at 900 C for 1 minute under a flow of N 2 to simulate an extreme rapid thermal annealing condition which might be encountered in an indus trial process. GIXD scans were taken for these samples to observe any change in crystallization temperature with the addition of aluminum.

PAGE 106

92 X Ray Analysis Upon examination of the x ray data collected, it was shown that aluminum additions eliminated the cry stallization caused by UV radiation at the oxidation temperature of 400 C. GIXD spectra taken at W =1, plotted in Figure 5 1, for samples H2, AH3, and AHN3 show a marked decrease in crystallization after the UV assisted oxidation treatment with samples AH3 and AHN3. Thus the aluminum addition increases the crystallization temperature of the film at l east to an extent to overcome the irradiative effect of the UV radiation. Figure 5 1: GIXD scans taken at W =1 comparing the effects of aluminum additions in samples AH3 and AHN3 to sample H2. Compositional Comparisons XPS was used to determine the comp osition and bonding environmental changes in the films. Surveys from 0 1000 eV were taken for all the samples to first determine the

PAGE 107

93 elemental makeup and to detect any contaminates. Then a multi plex consisting of the C 1s, O Auger, O 1s, N 1s, and Hf 4d peak regions were collected. The elemental percentages were obtained from the surveys and converted to molar percentages for the HfO 2 and Al 2 O 3 compounds. Results from this analysis are given in Table 5 2. Table 5 2: Sample composition from elemental a nalysis using XPS. Sample ID# HfO 2 Al 2 O 3 N AH1 53.8% 46.2% --AH2 77.7% 22.3% --AH3 89.4% 10.6% --AHN1 67.0% 29.0% 4% AHN2 82.2% 15.8% 2% AHN3 92.4% 5.6% 2% The hafnium metal target was shown to have a greater deposition rate than the aluminum target, thus for the samples with 1:1 pulse ratios the hafnium oxide content after oxidation was greater than the aluminum oxide. This effect is even greater when the samples were deposited in ammonia. The marked increase in hafnium oxide in these films is attributed to the higher mass density of the atoms. Because of the greater mass of the particles being ablated, less scattering due to the ammonia gas atoms occurs and a more uniform plume dynamic results. The molar fraction results were then conver ted to volume fractions by taking the relative mass of each present and dividing by the tabulated densities of 9.68 g/cm 3 for HfO 2 and 3.98 g/cm 3 for Al 2 O 3 The bulk dielectric constant of the film can be estimated from this result. In a paper by D. Youn g, et al. 72 it was shown that finely dispersed mixtures of two dielectric materials yield a bulk dielectric constant that follows a logarithmic mixing rule. This relationship is given by: log k bulk = S i v i log k i (5 1)

PAGE 108

94 where k bulk is the dielectri c constant of the mixture, v i is the volume fraction of each component, and k i is the respective dielectric constant for each component. Table 5 3 shows the predicted dielectric constant for each of the films. Table 5 3: Predicted film density and bulk d ielectric constant. Sample ID Mole Fraction HfO 2 Mole Fraction Al 2 O 3 Density g/cm 3 Volume Fraction HfO 2 Volume Fraction Al 2 O 3 Bulk k AH1 0.54 0.46 7.04 0.74 0.26 19.2 AH2 0.78 0.22 8.40 0.89 0.11 22.5 AH3 0.89 0.11 9.07 0.95 0.05 23.9 AHN1 0.69 0.31 7. 91 0.84 0.16 21.3 AHN2 0.83 0.17 8.72 0.92 0.08 23.1 AHN3 0.93 0.07 9.30 0.97 0.03 24.4 Electrical Characterization Results Upon completion of the chemical and structural analysis, the sample surfaces were washed with methanol and blow dried with dry n itrogen. Then 1000 platinum electrodes were sputtered on the surface using a radio frequency (RF) sputtering system. The shadow mask used for this step had repeated sets of circular dots of various sizes. The size dot used for this portion of the expe riments had an area of 4.66x10 4 cm 2 A forming gas anneal was then carried out at 450 C for 30 minutes in a flowing gas environment using a gas mixture of 4% H 2 and 96% N 2 After MOS devices were fabricated capacitance voltage and current voltage measu rements were taken with the Keithly Instruments Win 82 system. For the capacitance voltage measurements, voltage was scanned from 3 to +2 V using the 1 MHz, parallel model setup. Compilation of the results from the six samples is presented in Figure 5 2. Additions of aluminum into the films were shown to decrease the capacitance as was expected. Bulk aluminum has a dielectric constant of 9 compared

PAGE 109

95 to 25 for HfO 2 The samples without nitrogen exhibit a large flat band voltage shift of ~2V. The shift is due to an increase in the amount fixed charge within the layer caused by the aluminum addition. 34 Furthermore, the slope of the response decreases as aluminum is added. The slope of the curve is highly dependent on the concentration of interfacial tr aps. It is concluded that the aluminum creates higher concentrations of these traps. By depositing the samples in ammonia, it can be seen that a more stable capacitance response is obtained. The nitrided samples exhibit almost no shift in the flatband voltage and have similar slopes. Therefore the nitrogen helps to passivate the effect of fixed charges from the aluminum and also eliminates a majority of the interfacial traps. Figure 5 2: Compilation of the capacitance voltage results from the alumin um addition samples.

PAGE 110

96 Table 5 4 gives the bulk dielectric constants and the EOT for each based on 80 100 film thicknesses from VASE modeling. The resulting dielectric constants were lower than the predicted ideal values due to the formation of a lower di electric constant interfacial layer which has been seen in all previous experiments. Table 5 4: Dielectric constants and EOT for the aluminum study samples. Sample ID Dielectric Constant EOT () AH1 10.1 39.6 AH2 11.6 34.5 AH3 17.2 23.3 AHN1 11.3 35.4 AHN2 15.8 25.3 AHN3 20.4 19.6 Current voltage measurements were then scanned from 3 to 3 V for all the samples. The average leakage current density for each sample is presented in Figure 5 3. The samples deposited without nitrogen have an average o f two orders of magnitude lower leakage current densities. This has been attributed to the increased growth of the interfacial layer. All the samples exhibited leakage values which fit the industry standard requirements; however, this is somewhat mislead ing as the samples are thicker than would be used if implemented into a device, as the EOT needs to be less than 10 Further analysis using thinner samples must be carried out before these films would be deemed worthy as replacement gate oxide materials High Temperature Anneal All the samples were subjected to a 900 C RTA in flowing N 2 for 1 minute. After the anneal GIXD measurements were taken for the films. Figure 5 4 shows the result from sample AH3 which contained only ~11% Al 2 O 3 No crystalliza tion was detected for these films. Previous reports have suggested that ~30% Al 2 O 3 is necessary to achieve

PAGE 111

97 crystallization temperatures above 900 C. Figure 5 3: Compilation of the current voltage results from the aluminum addition samples. Figure 5 4: GIXD for sample AH3 annealed at 900 C

PAGE 112

98 Summary The experiments from the second portion of this dissertation were extended in this chapter to study the effect o f aluminum additions. It was shown that aluminum increased the crystallization temperature of the films sufficiently to counteract the crystallization which was resulting from the UV assisted oxidation process. Furthermore aluminum was shown to drastical ly hinder the capacitance response of the films with respect to bulk and interface traps, as well as, reduction of the dielectric constant. Incorporation of nitrogen into the aluminum/hafnium metal alloys during deposition increased the stability of the f ilms by eliminating fixed charge from the aluminum and by reducing the amount of interfacial layer formation.

PAGE 113

99 CHAPTER 6 CONCLUSIONS Since SiO 2 /SiON x is vast approaching its physical limits for use in MOSFET device structures, there exists a driving force to replace the existing SiO 2 layer with an alternative layer with a higher dielectric constant. Though many re search teams have investigated numerous different materials in an attempt to accomplish this goal, to date, none have succeeded due to new problems associated with the formation an unwanted low k interfacial layer which forms between the silicon substrate and the dielectric material. The studies associated with this dissertation addressed the issues related to the kinetics of the interfacial layer formation of pulsed laser deposited HfO 2 films, the use of nitrogen incorporation during deposition of Hf meta l films, post deposition oxidation using UV radiation, and additions of aluminum. First, experimentation was conducted whereby ultrathin HfO 2 samples were deposited using varying temperatures and oxygen pressures. From the various characterization techniq ues, it was determined that the films deposited by pulsed laser deposition exhibited an interfacial layer even for samples that were deposited at 200 C in very low oxygen pressures. The nature of the interface in all the samples was shown to be a physica l mixture of HfO 2 and SiO 2 No intermediate silicate structures were found to form. Increased deposition temperature caused increases in the interfacial layer thickness. The log of the pressure was shown to linearly increase the layer thickness formatio n. By understanding the kinetics, better control of the unwanted interfacial layer can be realized.

PAGE 114

100 UV assisted oxidation of metal films provides a lower temperature process whereby high k films can be produced. However, UV radiation hinders the film pro perties by promoting crystallization and interfacial layer formation. It was found that small additions of nitrogen into the film during deposition by depositing in ammonia help to counteract the detrimental effects of the UV assisted oxidation step. The nitrogen prevents diffusion of the highly reactive oxygen species to the interface and eliminates a majority of the bulk oxygen defects usually seen with PLD samples. To control the crystallization temperature of the films, aluminum additions were studied The as deposited samples all were amorphous even after the UV assisted oxidation step. Aluminum increases the crystallization temperature of the films at least to an extent at which the energy from the UV radiation does not cause crystallization. The electrical characteristics of the films show that aluminum degrades the capacitance response by first decreasing the dielectric constant as it is a lower k material. Furthermore, aluminum was shown to greatly shift the flatband voltage due to a greater am ount of negative fixed charge as the concentration was increased. Finally the samples with higher concentrations of aluminum had increased interfacial defect densities shown by the decrease in the capacitance response slope. By adding nitrogen into the a lloy during deposition, the flatband voltage was stabilized and the concentration of interface traps was reduced. Studies in this dissertation indicated that there is development of an interfacial layer during deposition which can be controlled by using a combination of nitrogen and UV treatment to yield samples with acceptable electrical characteristics for use as an alternative gate dielectric material. The addition of UV radiation was shown to provide

PAGE 115

101 beneficial removal of oxygen defects in the film but also detrimentally increased the interfacial layer formation and crystallization in the film. Depositing the Hf metal films in ammonia incorporates a small amount of nitrogen into the film. The nitrogen acts as a getter for the oxygen and slows the d iffusion of oxygen species to the interface. Thus in effect the nitrogen counters the detrimental effects associated with UV assisted oxidation. Further improvements of the films were made by depositing alloys of Hf and Al to increase the crystallization temperature of the films.

PAGE 116

102 APPENDIX A ELECTRICAL DATA EXTR ACTION METHOD When conducting electrical characterization, once the proper model has been chosen for the with the Metrics ICS software, measurements may be acquired and a significant amount of information may be extracted fro m this data. For example, by calculating the total fixed charge in a MOS stack, numerous other important data are generated from the process. The derivation of the fixed charge will be done as an example. It is important to first get a current voltage me asurement from one of the MOS devices to determine an appropriate diode to make capacitance voltage measurements on. That is, a diode with a high breakdown voltage and turn on characteristics. Once this is determined, a high frequency/quasistatic output from the Win 82 system may be generated. A typical result for a p type wafer is shown in Figure 2 11. While both curves are important, for this process, the calculations will be based off the high frequency curve. The first step is to use the CV curve t o calculate doping concentration in the substrate. This is done by plotting the reciprocal of the square of the capacitance versus the gate voltage. The new plot should exhibit a linear region. By taking the slope of the linear region, it is possible to calculate the doping in the semiconductor. It is important to note that curvature in this region may represent nonuniform doping in the semiconductor. The equation for determining the doping is shown below: ) ( 2 2 slope A q N S e = (A 1)

PAGE 117

103 where q is the electric charge, S e is the semiconductor dielectric constant, A 2 is the gate area, and N is the doping (acceptor or donor). With the value, it is now possible to calculate the Debye length (L D ). N q kT L S D 2 e = (A 2) where k is the Boltzmann constant and T is temperature in Kelvin. With this value the semiconductor flatband capacitance can be calculated. D S FBS L A C e = (A 3) The next step is to calculate or experimentally det ermine the capacitance of the dielectric layer (C i ). In the scenario described here, the capacitance has been measured and represents the value of capacitance of the high frequency curve in the accumulation region. This typically is taken at a value of m inimal leakage current, as ascertained by the previous IV measurements. If, one was unable to measure the capacitance, but had knowledge such as the thickness of the dielectric layer, d, area of the MOS stack, A, and the dielectric constant, i e of the dielectric layer, then it is possible to calculate the corresponding capacitance of the insulator. The relationship is shown as follows: d A C i i e = (A 4) The next important parameter in the process is calculation of t he overall flatband capacitance and the ratio: i FBS i FBS FB C C C C C + = (A 5)

PAGE 118

104 i FB C C (A 6) The next step is to calculate the V shift this is done by plotting C/C i versus gate voltage. Then use the ratio of C FB /C i to to visually determ ine the flatband voltage shift (?V shift ). In an ideal MOS capacitor, the flatband voltage condition corresponds to a ?V shift = 0 thereby allowing the conclusion that any voltage shift should be related to total fixed charge. This charge may come from man y different sources: Q m > Mobile Charges > location dependent on bias Q ot > Oxide Trapped > distributed in the oxide Q f > Fixed Oxide > near surface charge Q it > Interfacial Traps > at the interface The first three of these charges causes shifts of the ov erall CV curve to the left or right while the fourth charge manifests itself as changes in slope (stretching) of the overall CV curve. The next step is to determine the value f ms, which is the work function of the metal minus the work function of the semiconductor. Depending on the wafer type used, this would be done as follows p type Y + + C = = B g M M MS E 2 f f f f (A 7) n type Y + C = = B g M M MS E 2 f f f f (A 8) where E g /2 is the intr insic Fermi level, f M is the metal work function, f S is the semiconductor work function, ? is the electron affinity of the semiconductor, and ? B is the Fermi level difference between the Fermi level and the intrinsic Fermi level.

PAGE 119

105 Finally, the total oxide c harge may be calculated with the following equation by collating all of the previously determined values as: n type i MS f C q V N D = f (A 9) Another important electrical value that is commonly cited in literature is the value of the interface traps Qit. There are several possibilities for determining this value: 1) comparison of the high frequency capacitance to a theoretical capacitance without any interface traps, 2) comparison of a low frequency capacitance to a theoretical capacitance without any interface traps, or 3) comparison of a high frequency capacitance to a low frequency capacitance. The third option, first developed by Castagn and Vapaille consisted of combining the low and high frequency curves (i.e., the output of the Win 82 syste m) to determine the silicon surface capacitance per unit area, C S Unlike the Terman method where only the high frequency capacitance (C HF ) is measured, or the Berglund method where only low frequency capacitance (C LF ) is measured, this method measures bo th C LF and C HF This eliminates the need for the generation of any of the theoretical computations needed for the other methods. As outlined in more detail by Nicollan and Brews, 7 3 the basic result is that interface trap level can be determined at a giv en voltage as a function of position in the bandgap by simple input of measured parameters into the following equation: 1 1 1 1 D + D = ox HF ox HF it C C C C C q C D (A 10) where D it is the interface trap density, C ox is the capacitance per unit area, and LF HF C C C = D (A 11) This equation is used to make a plot if D ito versus C HF /C ox where

PAGE 120

106 o ito it x D D 1000 = (A 12) and x o is the thickness of the oxide in question. Combination of the directly measured CV data and the above equation it is therefore possible to determine the Dito. It is important to realize, however, that accurate values of D it cannot be determined over the entire range of gate bias due to a variety of reasons including round off errors, errors due to use of a 1 Mhz CV curve, and error from C LF Each of these errors is addressed in detail in Nicollian and Brews 7 3

PAGE 121

107 APPENDIX B XPS DATA The following tables are the compilation from the XPS data collected for analysis in Chapter 3. Extensive peak position, area, intensity, and calculation of layer thickness are presented in the tables for each of the samples at the thr ee take off angles used.

PAGE 122

108

PAGE 123

109

PAGE 124

110

PAGE 125

111 LIST OF REFERENCES 1 G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys 89 5243 (2001). 2 G. E. Moore, Electron 19 114 (1965). 3 J. D. Plummer and P. B. Griffin, Proceedings of the IEEE 89 240 (2001). 4 International Technology Roadmap fo r Semiconductors 2002 edition, ITRS Home Page, http://public.itrs.net/Files/2002Update/2001ITRS/ExecSum.pdf (02/16/2004) 5 R. Degraeve, E. Cartier, T. Kauerauf, R. Carter, L. Pantisano, A. Kerber, and G. Groseseneken, Mat. Res. Soc. Bulletin 27 222 (2002). 6 J. Kwo, M. Hong, A. R. Kortan, K. L. Queeney, Y. J. Chabal, R. L. Opila Jr., D. A. Muller, S. N. Chu, B. J. Sapjeta, T. S. Lay, J. P. Mannaerts, T. Boone, H. W. Krautter, J. J. Krajewski, A. M. Sergent, and J. M. Rosamilia, J. Appl. Phys 89 3920 (2001). 7 K. J. Hubbard and D. G. Schlom J. Mater. Res 11 2757 (1996). 8 J. P. Liu, P. Zaumseil, E. Bugiel, and H. J. Osten, Appl Phys. Lett 79 671 (2001). 9 D. G. Schlom and J. H. Haeni, Mat. Res. Soc. Bulletin 27 198 (2002). 10 R. D. Shannon, J. Appl. Phys 73 398 (1993). 11 A. Shanware, J. McPherson, M. R. Visokay, J. J. Chambers, A. L. Rotondaro, H. Bu, M. J. Bevan, R. Khamankar, and L. Colombo, IEEE Electron Devic es Meeting, 2001. IEDM Technical Digest International, 2001 p. 6.6.1 6.6.4. 12 W. Zhu, T. P. Na, H. Tamagawa, Y. Di, J. Kim, R. Carruthers, M. Gibson, and T. Furukawa, IEEE Electron Devices Meeting, 2001. IEDM Technical Digest International 2001 p. 20.4. 1 20.4.4. 13 C. W. Yang, Y. K. Fang, C. H. Chen, W. D. Wang, T. Y. Lin, M. F. Wang, T. H. Hou, J. Y. Cheng, L. G. Yao, S. C. Chen, C. H. Yu and M. S. Liang, Electronics Lett 38 1223 (2002). 14 National Technology Roadmap for Semiconductors. Semiconduc tor Industry Association, San Jose, CA, 1997 p. 72.

PAGE 126

112 15 A. I. Kingon, J. P. Maria, and S. K. Streiffer, Nature 406 1032 (2000). 16 G. Lucovsky, J. Vac. Sci. Technol. A 19 1553 (2001). 17 N. H. Weste and K. Eshraghian, Principles of CMOS VLSI Design: A Systems Perspective, Second Edition Addison Wesley Publishing Company, New York, 1994 p 123. 18 Intel Home Page, http://www.intel.com/home/glossary/body.htm (06/24/2004) 19 S. Tang, R. M. Wallace, A. Seabaugh, and D. King Smith, Appl. Surf. Sci 135 137 (1998). 20 B. E. Weir, P. J. Silverman, M. A. Alam, F. Baumann, D. Monroe, A. Ghetti, J. D. Bude, G. L. Timp, A. Hamad, and T. M. Oberdick, Tech. Dig. Int. Electron Devices Meet. 1999 p. 437. 21 D. J. Frank, R. H. Dennard, E. Nowak, P. M. Solomon, Y. Taur, and H. P. Wong, Proceedings of the IEEE 89 259 (2001). 22 M. Depas, B. Vermeire, P. W. Mertens, R. L. Van Meirhaeghe, and M.W. Heynes, Solid State Electron 38 1465 (1995). 23 D. A. Buchanan, IBM J. Res. Develop 43 245 (1999). 24 H. P. Won g, IBM J. Res. Develop 46 133 (2002). 25 J. Robertson, Mat. Res. Soc. Bulletin 27 217 (2002). 26 J. J. Chambers and G. N. Parsons, J. Appl. Phys 90 918 (2001). 27 B. H. Lee, R. Choi, L. Kang, S. Gopalan, R. Nieh, K. Onishi, Y. Jeon, W.J. Qi, C. Kan g, and J. C. Lee, Electron Devices Meeting, 2000. IEDM Technical Digest. International 2000 p. 39. 28 L. Manchanda, M. L. Green, R. B. van Dover, M. D. Morris, A. Kerber, Y. Hu, J. P. Han, P. J. Silverman, T. W. Sorsch, G. Weber, V. Donnelly, K. Pelhos, F. Klemens, D. Jacobson, J. Eng, B. Busch, and H. Schulte, Electron Devices Meeting, 2000. IEDM Technical Digest. International 2000 p. 23. 29 B. Y. Park, J. Park, M. Cho, C. S. Hwang, K. Oh, Y. Han, and D. Y. Yang, Appl. Phys. Lett 80 2368 (2002). 30 L. Kang, B. H. Lee, W. J. Qi, Y. Jeon, R. Nieh, S. Gopalan, K. Onishi, and J. C. Lee, IEEE Electron Dev. Lett 21 181 (2000).

PAGE 127

113 31 S. W. Nam, J. H. Yoo, S. Nam, H. J. Choi, D. Lee, D. H. Ko, J. H. Moon, J. H. Ku, and S. Choi, J. Non Cryst. Solids 303 139 (2002). 32 P. D. Kirsch, C. S. Kang, J. Lozano, J. C. Lee, and J. G. Ekerdt, J. Appl. Phys 91 4353 (2002). 33 H. Lee, S. Jeon, and H Hwang, Appl. Phys. Lett 79 2615 (2001). 34 H. Yu, M. F. Li, B. J. Cho, C. C. Yeo, M. S. Joo, D. L. Kwong, J. S. P an, C. H. Ang, J. Z. Zheng, and S. Ramanathan, Appl. Phys. Lett 81 376 (2002). 35 H. Cho, C. S. Kang, K. Onishi, S. Gopalan, R. Nieh, R. Choi, S. Krishnan, and J. C. Lee, IEEE Electron Device Lett 23 249 (2002). 36 C. W. Yang, Y. K. Fang, C. H. Chen, W. D. Wang, T. Y. Lin, M. F. Wang, T. H. Hou, J. Y. Cheng, L. G. Yao, S. C. Chen, C. H. Yu, and M. S. Liang, Electron. Lett 38 1223 (2002). 37 A. Kazor and I. W. Boyd, J. Appl. Phys 75 227 (1994). 38 E. M. Young and W. A. Tiller, J. Appl. Phys 62 2086 (1987). 39 Z. Cui, J. M. Madsen, and C. G. Takoudis, J. Appl. Phys 87 8181 (2000). 40 D. L. Blaluch, R. A. Cox, R. F. Hampson, J. A. Herr, J. Tore, and R. T. Watson, Phys. Chem. Ref. Data 9 295 (1980). 41 A. M. Goodman, Phys. Rev. B 152 785 (19 66). 42 R. Williams, Phys. Rev. B 140 A 569 (1965). 43 B. E. Deal and A. S. Grove, J. Appl. Phys 36 3770 (1965). 44 Ishikawa, T. Shibamoto, and I. Nakamichi, Japn. J. Appl. Phys 31 1148 (1992). 45 S. Ramanathan, P. C. McIntyre, J. Luning, P. Pian etta, and D. A. Muller, Phil. Mag. Lett 82 519 (2002). 46 S. Ramanathan, D. A. Muller, G. D. Wilk, C. M. Park, and P. C. McIntyre, Appl. Phys. Lett 79 3311 (2001). 47 S. Ramanathan, G. D. Wilk, D. A. Muller, C.M. Park, and P. McIntyre, Appl. Phys. Le tt. 79 2621 (2001). 48 S. Ramanathan, C.M. Park, and P. C. McIntyre J. Appl. Phys 91 4521 (2002).

PAGE 128

114 49 S. Ramanathan and P. C. McIntyre, Appl. Phys. Lett. 80 3793 (2002). 50 S.J. Wang, Semicond. Sci. Tech 16 L13 (2001). 51 V. Misra, G. Lucovsky, and G. Parsons, Mat. Res. Soc. Bulletin 27 212 (2002). 52 H. Zhong, G. Heuss, and V. Misra, IEEE Electron Device Lett 21 593 (2000). 53 C.R. Brundle, C. A. Evans, and S. Wilson, Encyclopedia of Materials Characterization, Butterworth Heinemann, Boston 1992 54 R. K. Singh and D. Kumar, Mat. Sci. Engr. Reports R22 113 (1998). 55 V. Craciun and R. K. Singh, Electrochem. Solid State. Lett 2, 446 (1999). 56 V. Craciun, E. S. Lambers, N. D. Bassim, R.K. Singh, and D. Craciun, J. Mater. Res 15, 488 ( 2000). 57 V. Craciun, R.K. Singh, J. Perriere, J. Spear, and D. Craciun, J. Electrochem. Soc 147 1077 (2000). 58 A. Srivastava, V. Craciun J. M. Howard, and R. K. Singh, Appl. Phys. Lett 75 3002 (1999). 59 V. Craciun and R. K. Singh, Appl. Phys. Le tt 76 1932 (2000). 60 V. Craciun, J. M. Howard N. D. Bassim, and R. K. Singh, Transport and Microstructural Phenomena in Oxide Electronics, Editors: Dave H. Blank, David S. Ginley, Marilyn E. Hawley, Steph en K. Streiffer, David C. Paine, MRS Proceeding s Volume 666 F11.4, 2001 61 V. Craciun, N. D. Bassim, J. M. Howard, J. Spear, S. Bates, and R. K. Singh MRS Spring Meeting Transport and Microstructural Phenomena in Oxide Electronics Editors: Dave H. Blank, David S. Ginley, Marilyn E. Hawley, Stephe n K. Streiffer, David C. Paine, MRS Proceedings Volume 666, F8.11.1, 2001 62 V. Craciun, C. Chiritescu, F. Kel ly, and R.K. Singh, J. Optoelectronics and Adv. Mat. 4 21 26 (2002). 63 F. Iacon a, R. Kelly, and G. Marletta, J. Vac. Sci. Technol. A 17 2771 (1999). 64 L. A. Knauss, J. M. Pond, J. S. Horwitz, D. B. Chrisey, C. H. Mueller, and R. Treece, Appl. Phys. Lett 69 25 (1996). 65 H. Ono, Y. Hosokawa, T. Ikarashi, K. Shinoda, N. Ikarashi, K. Koyanagi, and H. Yamaguchi, J. Appl. Phys 89 (2), 995 (20 01).

PAGE 129

115 66 T. J. Vink, R. G. F. A. Verbeek, J. H. M. Snijders, and Y. Tamminga, J. Appl. Phys 87 (10), 7252 (2000). 67 V. Craciun, J. M. Howard, N. D. Bassim, and R. K. Singh, Appl. Surf. Sci 168 (1 4), 123 (2000). 68 C. H. Choi, T. S. Jeon, R. Clark, a nd D. L. Kwong, IEEE Electron Device Letters 24 215 (2003). 69 J. M. Howard, Ultraviolet assisted Processing of Dielectric Thin Films for Metal Oxide Semiconductor Applications, Ph.D. Dissertation, University of Florida, 2002 p.111. 70 Z. H. Lu, J. P. M cCaffrey, B. Brar, G. D. Wilk, R. M. Wallace, L. C. Feldman, and S. P. Tay, Appl. Phys. Lett 71 2764 (1997). 71 H. Sim, H. Chang, and H. Hwang, Jpn. J. Appl. Phys Part 1, No. 4A 42 1596 (2003). 72 D. Young, H. D. Wu, R. C. Y. Auyeung, and R. Modi, J Mater. Res No. 6, 16 1720 (2001). 73 E. H. Nicollian and J. R. Brews, MOS (Metal Oxide Semiconductor) Physics and Technology Wiley, New York, 1982

PAGE 130

116 BIOGRAPHICAL SKETCH Chad Robert Essary was born in Springfield, Missouri, on March 3 rd 1976. He grew up in Southwest Missouri and attended Forsyth High School in Forsyth, Missouri. Upon graduation as co valedictorian of his class in 1994, he began an u ndergraduate curriculum at the University of Missouri Rolla. Chad graduated in the spring of 1998 with a Bachelor of Science in ceramic engineering and minors in mathematics and history. He decided to stay at Rolla for his masters work under the guidance of Dr. Wayne Huebner. In the summer of 2000, Chad graduated with a Master of Science degree in ceramic engineering for his thesis work entitled Processing and Characterization of Ceramic Tape Layers for Ultra high Gradient Insulators. In the fall of 200 0, Chad moved to Gainesville, Florida, to pursue a doctorate degree in the Department of Materials Science and Engineering at the University of Florida.


Permanent Link: http://ufdc.ufl.edu/UFE0006612/00001

Material Information

Title: Ultraviolet-Assisted oxidation and nitridation of hafnium and hafnium aluminum alloys as potential gate dielectrics for metal oxide semiconductor applications
Physical Description: xiv, 116 p.
Language: English
Creator: Essary, Chad Robert ( Dissertant )
Singh, Rajiv K. ( Thesis advisor )
Abernathy, Cammy ( Reviewer )
Pearton, Stephen ( Reviewer )
Norton, David ( Reviewer )
Ren, Fan ( Reviewer )
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2004
Copyright Date: 2004

Subjects

Subjects / Keywords: Materials Science and Engineering thesis, Ph.D   ( local )
Dissertations, Academic -- UF -- Materials Science and Engineering   ( local )

Notes

Abstract: The continued miniaturization of silicon-based complimentary metal oxide semiconductor (CMOS) devices is pushing the limits of the silicon dioxide (SiO₂) gate dielectric. As the channel widths are decreased to increase packing densities and functionality of new chips, proportional vertical scaling of the dielectric must be maintained to keep constant capacitances. Silicon dioxide is approaching its fundamental limit in which it can be used as the gate dielectric due to high leakage currents resulting from direct tunneling through the layer. In order for the continued use of current CMOS gate design, an alternative material with a higher dielectric constant must be found. Several materials have been proposed but are still not providing the electrical characteristics favorable for use in the devices due to problems with excessive leakage and hysteresis resulting from the quality of the film and oxygen defects. The goal of this study is to create higher quality films at lower processing temperatures with low leakage and less hysteresis than has been achieved with hafnium oxide films. This study first examines the formation of the interfacial layer in pulsed laser deposited hafnium oxide films to understand the kinetics behind its formation. The second section focuses on the oxidation of pulsed laser deposited (PLD) hafnium metal thin films using ultraviolet (UV) assisted post-deposition annealing. Another set of samples was deposited in an ammonia atmosphere in order to incorporate nitrogen into the films. Comparisons of microstructure and stoichiometry of oxidized hafnium and oxy-nitride films were made using x-ray photospectroscopy, variable angle spectroscopic ellipsometry, glancing angle x-ray spectroscopy, x-ray reflectivity, and atomic force microscopy. Analysis of the interface between the films and the silicon substrate was carried out using x-ray reflectivity. The electrical characteristics of the films were characterized using capacitance-voltage and current-voltage measurements in order to compare the quality of the films. Ultimately a model of the effect of UV and nitrogen additions was presented. In the final portion of this work, additions of aluminum were used to increase the crystallization temperature of the films. The effect of aluminum both with and without nitrogen incorporation on the dielectric constant and leakage for the films was studied.
Subject: dielectric, gate, hafnium, high, nitridation
General Note: Title from title page of source document.
General Note: Document formatted into pages; contains 130 pages.
General Note: Includes vita.
Thesis: Thesis (Ph.D.)--University of Florida, 2004.
Bibliography: Includes bibliographical references.
General Note: Text (Electronic thesis) in PDF format.

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0006612:00001

Permanent Link: http://ufdc.ufl.edu/UFE0006612/00001

Material Information

Title: Ultraviolet-Assisted oxidation and nitridation of hafnium and hafnium aluminum alloys as potential gate dielectrics for metal oxide semiconductor applications
Physical Description: xiv, 116 p.
Language: English
Creator: Essary, Chad Robert ( Dissertant )
Singh, Rajiv K. ( Thesis advisor )
Abernathy, Cammy ( Reviewer )
Pearton, Stephen ( Reviewer )
Norton, David ( Reviewer )
Ren, Fan ( Reviewer )
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2004
Copyright Date: 2004

Subjects

Subjects / Keywords: Materials Science and Engineering thesis, Ph.D   ( local )
Dissertations, Academic -- UF -- Materials Science and Engineering   ( local )

Notes

Abstract: The continued miniaturization of silicon-based complimentary metal oxide semiconductor (CMOS) devices is pushing the limits of the silicon dioxide (SiO₂) gate dielectric. As the channel widths are decreased to increase packing densities and functionality of new chips, proportional vertical scaling of the dielectric must be maintained to keep constant capacitances. Silicon dioxide is approaching its fundamental limit in which it can be used as the gate dielectric due to high leakage currents resulting from direct tunneling through the layer. In order for the continued use of current CMOS gate design, an alternative material with a higher dielectric constant must be found. Several materials have been proposed but are still not providing the electrical characteristics favorable for use in the devices due to problems with excessive leakage and hysteresis resulting from the quality of the film and oxygen defects. The goal of this study is to create higher quality films at lower processing temperatures with low leakage and less hysteresis than has been achieved with hafnium oxide films. This study first examines the formation of the interfacial layer in pulsed laser deposited hafnium oxide films to understand the kinetics behind its formation. The second section focuses on the oxidation of pulsed laser deposited (PLD) hafnium metal thin films using ultraviolet (UV) assisted post-deposition annealing. Another set of samples was deposited in an ammonia atmosphere in order to incorporate nitrogen into the films. Comparisons of microstructure and stoichiometry of oxidized hafnium and oxy-nitride films were made using x-ray photospectroscopy, variable angle spectroscopic ellipsometry, glancing angle x-ray spectroscopy, x-ray reflectivity, and atomic force microscopy. Analysis of the interface between the films and the silicon substrate was carried out using x-ray reflectivity. The electrical characteristics of the films were characterized using capacitance-voltage and current-voltage measurements in order to compare the quality of the films. Ultimately a model of the effect of UV and nitrogen additions was presented. In the final portion of this work, additions of aluminum were used to increase the crystallization temperature of the films. The effect of aluminum both with and without nitrogen incorporation on the dielectric constant and leakage for the films was studied.
Subject: dielectric, gate, hafnium, high, nitridation
General Note: Title from title page of source document.
General Note: Document formatted into pages; contains 130 pages.
General Note: Includes vita.
Thesis: Thesis (Ph.D.)--University of Florida, 2004.
Bibliography: Includes bibliographical references.
General Note: Text (Electronic thesis) in PDF format.

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0006612:00001


This item has the following downloads:


Full Text












ULTRAVIOLET-ASSISTED OXIDATION AND NITRIDATION OF HAFNIUM AND
HAFNIUM ALUMINUM ALLOYS AS POTENTIAL GATE DIELECTRIC S FOR
METAL OXIDE SEMICONDUCTOR APPLICATIONS















By

CHAD ROBERT ESSARY


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2004

































Copyright 2004

by

Chad Robert Essary

































This document is dedicated to my loving family.















ACKNOWLEDGMENTS

I would like to first thank my advisor, Dr. Rajiv Singh, for giving me the

opportunity to study this exciting field of electronic materials. Also I want to thank Dr.

Valentin Craciun and Dr. Joshua Howard for being good friends as well as mentors to me

during my time here at the University of Florida. To all the group members I have had

the pleasure to work with and to our wonderful secretary Margaret, I would like to offer

my sincere gratitude for all your help and friendship all of you have provided my four

years here at the University of Florida. Furthermore, I would like to thank Dr. Cammy

Abernathy, Dr. Stephen Pearton, Dr. David Norton, and Dr. Fan Ren for taking the time

to serve on my committee and for providing direction and guidance. Finally, I would like

to thank the electronics group at the University of Texas-Austin for helping with the

electrical measurements on a portion of this dissertation.
















TABLE OF CONTENTS

Page

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

LIST OF TABLES .............. .................................viii

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

ABSTRACT.........................................................xiii

CHAPTER

1 LITERA TU RE REV IEW .................................................. ...............1

Introduction.................... ...................................... .......... ........ ...............
CM O S Transitors ............... .. ................... .................. ............. ......... .3
Properties and Lim stations of SiO 2 ....................................................................... 5
M materials Selection Guideline......................................................... 7
Band Gap and Alignment ................. .................................8
Thermodynamic Stability..........................................................8
Film Morphology and Interface Quality............... ........................................10
Implementation and Process Compatibility................ .................10
Reliability ................ .... ..... .. ............................ ...... ...............1
Hafnium Oxide Properties .......................................................... ... .. ............. 1
H afnium Oxide R esearch.............................................................................. 12
Nitride Barrier Layers.............................................................. .... ....... 13
Other Novel Approaches ................. ............. ...................14
UV Assisted Oxidation ............... .................. ........ ...... .... ..............15
Motivation............................ .................. ..................18

2 EXPERIMENTAL PROCEDURES ............................... ...............19

D position Setup and Equipm ent ........................................................ 19
UVPLD System ...................... ........ .... .................... 19
Laser Ablation............................ ............ ........21
Laser System .......................................... ........23
Experimental Setups and Samples ............................ ....................... 25
Interfacial Layer Formation and Growth............... ..... ...................25
UV Assisted Oxidation and Nitridation of Hafnium Films...............................26
UV Assisted Oxidation and Nitridation of Hafnium/Aluminum Alloy Films ....27










Experimental Characterization Techniques........................................................28
Variable Angle Spectroscopic Ellipsometry ..................................................28
X-ray Reflectivity ........................ ... ..... ............. ......... .......... 29
X-ray Diffraction and Glancing Incidence X-ray Diffraction.........................31
X-ray Photoelectron Spectroscopy............................................33
A tom ic Force M icroscopy.................................................. 34
Electrical Characterization ............................................................. ............ 34
Current-V oltage M easurem ents...................................................................... 36
Capacitance-Voltage M measurement ........................... ...... ............... 37

3 KINENTICS OF INTERFACIAL LAYER FORMATION DURING DEPOSITION
O F H A FN IA ON SILIC ON ............................................................................................43

Deposition Conditions ....................................... ........ ........ 43
X-ray Results .................................................45
XPS Analysis .................................................49
Ellipsom etry R results ................... ............................. ....... .. .. .. .. .. ............ 53
Summary..................... ..... .. ..................... 54

4 EFFECT OF UV-ASSISTED OXIDATION AND NITRICATION OF HAFNIUM
M ETAL FILM S .......... ....... ............................ 56

Experiment..................... ............................. ........... .... ............56
Results and Discussion for Low Temperature Deposited Samples .........................60
Surface M orphology ................... .................................... ........ .............. 60
X-Ray Analysis ...................................................61
X-Ray Reflectivity................. .................................................. 64
XPS Analysis .......................... ...................... 64
Electrical Properties A nalysis................................ .................... 72
Capacitance-V oltage R results ...................................................................... 72
Repeatability........................... .. ......... .........74
Interfacial Trap Density Measurements .................................... ....76
Current-V oltage R results .............................................. ............... 77
TEM Analysis..................... ....................... 78
Proposed M odel.................. .................. ..... ... ... ................................ 79
Results and Discussion of the High Temperature Depositions ...............................81
X PS ............................ .................................................... 82
X-Ray Analysis .................................................. .........83
Electrical Characterization Results.............................. ............... 85
Discussion...................................... ................... ............ ........87
Summary......................................... ..........89

5 UV-ASSISTED OXIDATION AND NITRIDATION OF HAFNIUM ALUMINUM
M ETAL ALLOY FILM S ................................................. ............... 90

Experiment ................... ............................................. ........ ..............90
X-Ray Analysis ........................................ .........92










Com positional Com prisons ........................................................ 92
Electrical Characterization Results.............................. ............... 94
High Temperature Anneal ...................................... ...............96
Summary...................................... ................... ............. ........98

6 CONCLUSIONS............... ........ ......................... 99

APPENDIX

A ELECTRICAL DATA EXTRACTION METHOD............... ................102

B X P S D A T A ....................................................... 107

LIST OF REFEREN CES ............................................. ................ .... ...............111

BIOGRAPHICAL SKETCH .................................................................. .....116
















LIST OF TABLES


Table page

1-1: Industrial roadmap for minimum feature size ................. .......................5

1-2: Various properties of high-k oxide materials. .......................................8

1-3: Gibbs free energy of formation for potential alternative high-k dielectrics...........10

1-4: Various properties of HfO 2. ............................................................. ............. 12

2-1: Conditions for growth of ultra-thin Hf02 films............ ....................26

2-2: Growth and postdeposition conditions for UV oxidation of Hf ...............................27

2-3: High temperature Hf metal deposition and oxidation conditions ..............................27

2-4: Growth and annealing conditions for Hf/Al metal films ................................28

2-5: Conversion factors for series-parallel electrical equivalent circuits.....................41

3-1: Conditions for growth of ultra-thin Hf02 films............ .....................44

3-2: XRR modeling parameters used to fit the spectra ..................................... 48

3-3: Ellipsom etry m odeling results ............................................ ............... 54

3-4: Comparison of the thickness values (in A) for HfO2 thin films deposited on Si......55

4-1: Growth and postdeposition conditions for UV oxidation of Hf ...............................59

4-2: Z range and RMS roughness for oxidized Hf metal films..................... 61

4-3: d/X values for samples H2-H5 ................. ...............................67

4-4: Dielectric constants and EOT for samples H2-H4. ...................................... 74

4-5: Repeatability study results for k and EOT of oxidized hafnium metal films.........76

4-6: Defect density results for samples analyzed at UT-Austin. ...................................77

4-7: EOT and k for the high temperature deposited samples.................................... 87









5-1: Sample deposition conditions ................. ...............................91

5-2: Sample composition from elemental analysis using XPS. .......................................93

5-3: Predicted film density and bulk dielectric constant .............................. ............... 94

5-4: Dielectric constants and EOT for the aluminum study samples...........................96
















LIST OF FIGURES


Figure page

1-1: Typical CM O S inverter. ................................................................4

1-2: Energy band diagram of an n+ poly-Si/SiO2/n-Si structure ..............................7

1-3: Bandgap and band offset comparison for various dielectric materials.....................9

2-1: UVPLD system with excimer laser and optics setup. ............................................20

2-2: Photograph of UV lamp placement in the chamber. ...........................................21

2-3: Laser energy distribution during target ablation. .......................................... 22

2-4: Typical XRR plot showing which film properties most influence various features. 30

2-5: Typical XRD equipment geometry setup ..........................................32

2-6: AFM operational setup showing laser detection of the cantilever deflection .......35

2-7: MOS capacitor setup used for CV/IV measurements.................. ...............36

2-8: Kiethley Instruments Inc. Win-82 measurement system............... ..............38

2-9: Typical capacitance/voltage response curve .............. ...... .... ....... ........ 40

2-10: Equivalent circuits for estimation of capacitance..............................................42

3-1: Comparison of GIXD (f=10) for the 200 'C and 600 'C samples.........................46

3-2: Acquired XRR spectrum and its simulation for a HfO2 film. ................................47

3-3: Schematic of gate oxide stack model used for XRR simulations.............................47

3-4: Interfacial layer thickness dependence on 02 pressure for the 600 'C samples. .......49

3-5: High resolution XPS scans of the Hf 4f region....... ......... .........................50

3-6: Raw data and fit for the Si 2p region for the 600 'C, Ix10-2 Torr 02 sample. ..........51

3-7: Variations of d/k values for a 5 nm pure SiO2 on Si and a deposited sample ...........52









3-8: Interfacial layer thickness as a function of temperature and pressure.....................53

4-1: CV response of an UV grown SNy film. .......................................57

4-2: 2D AFM scans of samples a) H2 and b) H3 taken using contact mode....................62

4-3: 2D AFM scans of samples a) H4 and b) H5 taken using contact mode...................63

4-4: 3D AFM micrographs of both a) UV-oxidized and b) non- irradiated ....................65

4-5: GIXD spectra ( Q=10) of Hf films deposited on Si and oxidized.............................66

4-6: XRR plot and fit for as deposited Hf metal film. .......................................... 66

4-7: Angle-resolved XPS Si 2p spectra for system calibration.......................................68

4-8: Angle-resolved XPS Si 2p spectra for H2 and H3 (UV and no-UV films)............68

4-9: 0 Is spectra for H2 and H3 (UV and no-UV) films at 45 and 90 .........................69

4-10: Si 2p spectra for UV-oxidized films H4 and H2 (with and without nitrogen). .......70

4-11: N Is peak of as deposited nitrided Hf metal film. ...................................... 71

4-12: N Is peak after PDA for samples H4 and H5..............................................71

4-13: Comparison of high and low frequency capacitance responses for sample H4. .....75

4-14: High frequency capacitance voltage responses for samples H2-H4. ...................76

4-15: CV response and model fit data for samples H2, H4, and H5.............................77

4-16: Current voltage results from oxidized hafnium metal films.................................78

4-17: XTEM of the UV annealed nitrided sample H4...............................................79

4-18: XTEM of non UV annealed hafnium film H3................. .................. .........81

4-19: Model of the effect of UV and nitrogen. ................................82

4-20: 0 Is Peaks for the high temperature deposited sample set. ....................................83

4-21: Hf 4f Peaks for the high temperature deposited sample set. ..................................84

4-22: GIXD for the high temperature deposited samples taken at Q = 0.5. ....................85

4-23: XRR plots and models for the high temperature deposited samples. ......................86

4-24: CV plots for the high temperature deposited samples ................ ........ ............88









4-25: IV plots for the high temperature deposited samples ...........................................88

5-1: GIXD scans taken at Q=10 comparing the effects of aluminum additions ............92

5-2: Compilation of the capacitance-voltage results.................................................95

5-3: Compilation of the current-voltage results. ............................................................97

5-4: GIXD for sample AH3 annealed at 900 C .................................... .....97















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ULTRAVIOLET-ASSISTED OXIDATION AND NITRIDATION OF HAFNIUM AND
HAFNIUM ALUMINUM ALLOYS AS POTENTIAL GATE DIELECTRIC S FOR
METAL OXIDE SEMICONDUCTOR APPLICATIONS

By

Chad Robert Essary

December, 2004

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

The continued miniaturization of silicon-based complimentary metal oxide

semiconductor (CMOS) devices is pushing the limits of the silicon dioxide (SiO2) gate

dielectric. As the channel widths are decreased to increase packing densities and

functionality of new chips, proportional vertical scaling of the dielectric must be

maintained to keep constant capacitances. Silicon dioxide is approaching its fundamental

limit in which it can be used as the gate dielectric due to high leakage currents resulting

from direct tunneling through the layer. In order for the continued use of current CMOS

gate design, an alternative material with a higher dielectric constant must be found.

Several materials have been proposed but are still not providing the electrical

characteristics favorable for use in the devices due to problems with excessive leakage

and hysteresis resulting from the quality of the film and oxygen defects.

The goal of this study is to create higher quality films at lower processing

temperatures with low leakage and less hysteresis than has been achieved with hafnium









oxide films. This study first examines the formation of the interfacial layer in pulsed

laser deposited hafnium oxide films to understand the kinetics behind its formation. The

second section focuses on the oxidation of pulsed laser deposited (PLD) hafnium metal

thin films using ultraviolet (UV) assisted post-deposition annealing. Another set of

samples was deposited in an ammonia atmosphere in order to incorporate nitrogen into

the films. Comparisons of microstructure and stoichiometry of oxidized hafnium and

oxy-nitride films were made using x-ray photospectroscopy, variable angle spectroscopic

ellipsometry, glancing angle x-ray spectroscopy, x-ray reflectivity, and atomic force

microscopy. Analysis of the interface between the films and the silicon substrate was

carried out using x-ray reflectivity. The electrical characteristics of the films were

characterized using capacitance-voltage and current-voltage measurements in order to

compare the quality of the films. Ultimately a model of the effect of UV and nitrogen

additions was presented. In the final portion of this work, additions of aluminum were

used to increase the crystallization temperature of the films. The effect of aluminum both

with and without nitrogen incorporation on the dielectric constant and leakage for the

films was studied.














CHAPTER 1
LITERATURE REVIEW

Introduction

Over the past 30 years, the integrated circuit industry has strived to maintain

constantly shrinking device geometries and increased chip density.' The decrease in

device size has allowed for greater packing densities, thus increasing the functionality

and speed of the chips. The industry trend has followed what is known as Moore's Law.

In 1965, Gordon Moore from Intel first noted that every two to three years a new

technology generation with approximately doubled logic circuit density and a 40%

performance increase is realized as design and processing methods are improved.23

Since then, through continued miniaturization of gate widths of the metal oxide

semiconductor field effect transistors (MOSFET) and proportional vertical scaling,

Moore's Law has been maintained.

In order to continue miniaturization using current MOSFET structure designs, the

gate oxide, which is currently silicon dioxide, thickness must be reduced to less than 15

A.4 This represents five atomic layers of oxide or less. Problems not only arise in

processing to keep the thickness from varying across the gate, but also in the amount of

leakage current that results due to quantum mechanical tunneling. Tunneling current

increases exponentially as the thickness of the layer is decreased.5 Losses due to

tunneling would exceed the power consumption standards the industry wishes to use.1 In

order for continued reduction in gate widths, alternative materials with higher dielectric

constants (high-k) than silicon dioxide must be implemented.









Silicon dioxide (SiO 2) has been vital to the industry due to its abundance and many

useful properties. First of all, SiO2 is readily formed by the oxidation of the surface of

the silicon layer and can be precisely controlled through both chemical and thermal

means. This provides an easier process than a deposition technique and eliminates

interfacial defects experienced with deposition because SiO 2 forms an atomically abrupt

interface with silicon limiting the number of carrier traps that degrade the performance of

the transistor. Furthermore, post-annealing in a hydrogen containing ambient can easily

passivate dangling bonds. This excellent interface allows for high mobility of the carriers

in the MOSFET channel and high device performance.5

Finding an alternative gate dielectric that will act as well as Si02 is not an easy

task. Research groups are working on a wide range of single metal oxides and nitrides,6-9

binary compounds,10-13 and some ternary oxides.9 The replacement oxide must be

thermodynamically stable on silicon up to current processing temperatures for the chips

(~900-1000oC)1 14-15 to avoid reactions with the silicon that create compounds with lower

dielectric constants that parasitically hinder the benefits of the higher-k material. The

dielectric also needs to resist leaching of the dopants from the silicon channel and the

gate contact, which is currently heavily doped poly-silicon. The layer must form an

atomically abrupt or near-atomically abrupt interface with silicon to minimize carrier

traps at the surface of the channel and prevent mixing and formation of a lower-k silicate

interface layer. Ideally, the layer would be single crystalline or amorphous to inhibit

leakage due to grain boundaries. Growth of single crystals is a slow process and thus

does not lend itself well to implementation into industry. With amorphous layers, some









materials will crystallize when heated, so careful selection of the material must be made

based on the wafer processing conditions it will be subjected to after deposition.16

CMOS Transitors

The basis of the computer revolution of the past 30 years has been the use of

complementary metal oxide semiconductor (CMOS) transistors as the logic components

of the integrated circuits. Research on a new class of materials known as semiconductors

conducted at Bell Laboratories began in 1945 in an attempt to replace vacuum tubes for

signal amplification in long distance telecommunication applications. Vacuum tubes

were bulky, unreliable, consumed too much power, and produced too much heat. The

first design to utilize this new class of materials was the point contact transistor, created

on December 16th, 1947. The point contact transistor consisted of gold foil on top of a

plastic triangular piece in intimate contact with a chunk of germanium. This first design

was modified within a year into the junction sandwich transistor and proved to be more

practical, rugged, and easier to manufacture. Twelve years later, the field effect transistor

was created based on the original theories predicted in the early 1900s and has remained

the key design that has been continuously used and adjusted for the past forty years.

Figure 1-1 shows a typical CMOS inverter structure containing both a nMOS and a

pMOS device working joinly.17 The gates are made of poly-Si and are heavily doped to

reduce the sheet resistance along the lines and to tailor the work function of the gate to

the type of channel being used. The poly-Si is insulated from the MOS channels by a

thin gate oxide dielectric.

When a voltage is applied to the gate, a potential is set up across the dielectric.

Depending on the type of the channel, the potential will either decrease or increase the










contact cut oxideoxide



p n. f V -
n-well gate width p-substrate


Figure 1-1: Typical CMOS inverter [17].

current from the source to the drain. For the inverter structure, when a voltage is applied

(represented by a 1) typically the nMOS device will turn on and the pMOS device will

turn off. The source for the nMOS is connected to ground (represented by a 0) and so

passes the 0 to the drain. And vice-versa, when a 0 is applied to the gates, the pMOS

structure, whose source is wired to line voltage, will turn on and pass the 1 signal to the

drain while the nMOS will be turned off. By packing more of such logic components on

the area of a chip, faster processing speeds can be obtained. There are design rules that

must be followed as to the spacing between each device; however the rules, for the most

part, scale with the dimensions of the gates.17 The current technology on the market now

is 0.13 |tm gate widths used in the Intel Pentium 4 processors.18 This corresponds to

Si02 gate dielectric thickness of -20A. For further gate width reduction, proportional

scaling of the gate dielectric must be maintained to keep a constant capacitance across the

gate channel area.

By replacing the SiO2 with a material having a greater dielectric constant, the

capacitance can be maintained while affording a thicker dielectric layer and thus reduce

the leakage inherent with thinner films. This relationship is shown in equation (1-1).

t high-k ox' C
iChigh-k









teq = thickness of the equivalent SiO 2 layer
tox = thickness of the Si02 layer
Kox = dielectric constant of SiO 2
Khigh-k= dielectric constant of the alternative material

The interface between the silicon and the gate dielectric is very important because

formation of a lower-k layer will result in thicker equivalent thickness. This is because

the capacitances of the stacked layers add in series as shown in equation (1-2).

1 1 1 (1-2)
-=-+- (1-2)
Ctot C C1 2

Ctot = total capacitance
C1,2 = respective capacitances of the layers
Ultimately by using higher-k materials if the interface can be controlled, teq

thicknesses of <10A, which are needed for further advancement of the silicon based

CMOS devices, can be realized. Table 1-1 shows the predicted roadmap for equivalent

gate oxide thickness required for the minimum feature size.1

Table 1-1: Industrial roadmap for minimum feature size and equivalent dielectric
thickness in MOSFET devices.
Year Minimum Feature Size (|jm) Equivalent Dielectric Thicknes (A)
1997 0.25 40-50
1999 0.18 30-40
2001 0.15 20-30
2003 0.13 20-25
2006 0.10 15-20
2009 0.07 10-15
2012 0.05 <10


Properties and Limitations of SiO2

Amorphous Si02 has served as the gate dielectric for silicon-based CMOS devices

due to its excellent properties. Si02 has a bandgap of 9 eV and a dielectric constant of

3.9 and can readily be formed on silicon with low defect concentrations at the interface.

Using modem processing techniques, defect charge densities on the order of 1010/cm2,









midgap interface state densities of-1010/cm2eV, and hard breakdown fields of 15 MV/cm

are routinely obtained.' The industry has come to rely on these standards when designing

new chips. This presents a major challenge when trying to replace Si02. However, Si02

is approaching its limit of usefulness as the gate dielectric thickness requirements are

pushed below 15 A.

Si02 has a large conduction band offset (-3.5 eV) compared with silicon and

therefore provides a stable electrical insulation layer. However, this offset decreases as

the thickness of the layer is decreased. Ultimately, the Si02 layer loses its bulk properties

at a thickness of around 7-8 A.19 Below this thickness, the Si-rich interfacial regions

from the channel and poly-Si gate interfaces overlap, causing an effective "short" through

the dielectric, rendering it useless as an insulator. Experimental evidence has shown that

ultrathin (13-15 A) Si02 films can be processed which operate satisfactorily. However,

high leakage current densities of 1-10 A/cm2 have been measured for such devices.20

In order to meet the power guidelines set forth by the industry, current leakage

densities must be kept below 1x10'1 A/cm2 for gate biases of 1.2 V, corresponding to a

contribution of only a few mW to the overall chip dissipation.21 For Si02 layers thicker

than 6 nm, the leakage current mechanism is explained by Fowler-Nordheim electron

tunnelling. As the layer thickness decreases, the mechanism switches to direct

tunnelling. The difference between Fowler-Nordheim tunnelling (FNT) and direct

tunnelling (DT) of electrons is determined by the shape of the tunnel barrier. If the oxide

tunnel barrier is triangular, FNT occurs while DT takes place through a trapezoidal

barrier as shown in Figure 1-2.22 This occurs because of the band offset reduction as the

layer becomes thinner. As a result, the FNT leakage current is exponentially dependent









on the thickness of the oxide. To illustrate the drastic effect this relationship has,

Buchanan in his review article gave the example: for silicon dioxide, at a gate bias of 1

V, the leakage current changes from 1x10'12 A/cm2 at 35A to 1x10 A/cm2 at 15 A, a

change of 12 orders of magnitude.23

(a) (b






i h _____-__ -^' ---

^0 IV-; -qV


n Si n-Si nJI-A-Si n-Si

Figure 1-2: Energy band diagram of an n+ poly-Si/SiO2/n-Si structure in the case of (a)
Fowler-Nordheim tunnelling and (b) direct tunnelling of electrons from the Si
degenerate accumulation layer into the poly-Si gate conduction band. [22]

Materials Selection Guideline

Many materials are being studied for use as replacement gate dielectrics. Over

the past decade, A1203, ZrO2, Y203, TiO2, Ge203, La203, Ta203, Hf02, SrTiO3 and

BaxSri-xTi03 have all been considered. Various properties of these materials are shown

in Table 1-2.5 However, the following key properties suggested by Wilk et al.1 must be

considered when choosing a material:

Permittivity, band gap, and band alignment to silicon
Thermodynamic stability
Film morphology
Interface quality
Compatibility with current or expected materials used in processing









Process compatibility
Reliability
Table 1-2: Various properties of high-k oxide materials.
Oxide Dielectric Conduction Density Melting Stable in
Material Constant Band Offset (g/cm3) Point (C) Contact
(eV) with Silicon
A1203 9.3 2.8 3.97 2054 YES
BaSro.5Tio.503 80 -3600 -0.1 -0.1 6.02 1625 NO
CeO2 7 --- 7.65 2400 YES
HfO 2 22-25 1.5 9.68 2774 YES
Ta205 24 -65 0.3 8.20 1785 NO
TiO2 80 -170 --- 4.23 1843 NO
Y203 10 2.3 5.03 2439 YES
Zr02 ~25 1.4 5.68 2710 YES

Band Gap and Alignment

The material should have a band gap offset of greater than 1 eV to that of the

conduction band of silicon. This requirement rules out some of the higher dielectric

materials such as SrTiO3. Figure 1-3 graphically compares the band gaps and offsets of

some potential alternative dielectrics to that of Si and SiO 2.2425 Typically, materials with

higher dielectric constants will have smaller band gaps, so an optimization must be made

to balance the dielectric constant and band gap to get the best overall results. It was first

believed that materials with very high dielectric constants such as SrTiO3 and BaTiO3

would be ideal so thick dielectric layers could be implemented. However due to their

small band gap offsets to Si, they had shortcomings for use in gate dielectric applications

but have found uses in RAM applications.

Thermodynamic Stability

The dielectric must be thermodynamically stable in contact with Si to prevent

formation of SiO2 or lower-k silicate structures. Single component oxides can be

examined by comparing the Gibbs free energy of formation with respect to oxygen












4. -



2 -











-4- 4 _-.







I I II III I I I


Figure 1-3: Bandgap and band offset comparison for various dielectric materials. [24]

bonding. Elements having greater magnitude free energies than silicon will not readily

give up oxygen to form Si02 when in contact with Si.26 Table 1-3 lists the Gibbs free

energies for various metals that form potential alternative high-k oxides.26 Ta, Mo, and

W have been ruled as unstable in contact with silicon because their energies of formation

are less in magnitude than silicon and thus will give up oxygen to the silicon, and also

tend to form silicates that will hinder the dielectric constant and equivalent thickness of

the stack.

Further considerations must be taken into account because this table only shows the

values for systems in equilibrium. However, during deposition or formation of devices,

rarely is the system at a true equilibrium. Therefore, other reactions could occur, so a









good understanding of the kinetics behind the deposition or growth and processing steps

is key in predicting non-equilibrium products.9

Table 1-3: Gibbs free energy of formation for potential alternative high-k dielectrics.
-AGf (10-22 kcal/O atom) of metal Expected oxidation product of metal
Metal oxide silicide on silicon
Si 1.70
Ta 1.52 Si02/TaSi2/Si
Mo 0.88 Si02/MoSi2/Si
W 1.01 Si02/WSi2/Si
Y 2.40 Y203-Si02/Si
La 2.26 La203-SiO2/Si
Zr 2.06 ZrO2-Si02/Si
Hf 2.16 HfO2-SiO2/Si


Film Morphology and Interface Quality

To avoid high-leakage paths due to grain boundaries, single crystalline or

amorphous films are desired. Grain boundaries also promote the diffusion and leaching

of dopants from the channel when thermally processed. However, A1203 is the only

single-metal oxide that does not have a preferable crystalline phase under normal

processing conditions. Single-crystal oxides with excellent interface structure can be

grown using molecular beam epitaxy. However, this process is extremely slow and

requires ultra-high vacuum conditions resulting in very low throughput.' Ultimately,

amorphous barrier layers may have to be used at the interfaces of the dielectric even if it

will hinder the benefits of using that high-k dielectric.

Implementation and Process Compatibility

Ultimately whatever material that is chosen as the replacement for silicon dioxide

as the gate dielectric must be able to be fit into the large scale production scheme the

micro-electronics industry runs on. Thus, additional processing steps or a hindrance to









throughput are unacceptable. The new material must be able to be implemented into the

fabrication line with as little equipment and process flow changes as possible.

Reliability

Finally, the reliability of devices using the new gate material must be proven. The

devices must meet stringent ten year lifetime standards set by industry.1 These tests use

higher voltages than normal operating voltage to stress the devices until breakdown

conditions occur. This allows for lifetime estimates under actual working conditions to

be extrapolated. To date, few reports have been generated of this nature and additional

work must still be conducted. Further issues with threshold voltage shifts during

operation must also be considered. Stability of the threshold voltage will ensure

repeatable response of the device. If the threshold shifts too far, the device will cease to

function as the operating voltage being used is not sufficient to operate the gate.

Hafnium Oxide Properties

One of the most promising materials currently being studied is HfO2. Table 1-4

lists some of the properties of this dielectric.24' 27-28 Full ternary phase diagrams for the

Hf-Si-O system have not yet been reported. However, similarities to the well studied Zr-

Si-O system have been shown.9 Both metals are from the same family of the periodic

table and exhibit the same crystal structure and preferred equilibrium compounds.

Furthermore, due to the large negative free energy of formation for Hf-O bonds, hafnium

metal has the ability to pull oxygen from SiO2 thus limiting interfacial formation. For

these reasons, Hf and its subsequent oxide and oxy-nitride will be used for this study.









Table 1-4: Various properties of HfO2.
Property Value
Dielectric constant 20-30
Band Gap 6 eV
Conduction Band Offset 2.8 eV
Crystallization temperature 450-5000C
Crystal Structure Monoclinic
Thermal Stability w.r.t. Silcon 9500C


Hafnium Oxide Research

Over the past three years, many groups have been focusing their research efforts

on Hf02 and trying to use various methods for improving its performance as a gate

dielectric. The interface quality between silicon and the dielectric is highly dependent on

the deposition and annealing processes used. B. Y. Park et al.29 grew HfO2 thin films

with thicknesses from 27-55 A on hydrofluoric acid-treated Si wafers at a temperature of

2000C by chemical vapor deposition. They noted that the initial film growth was

controlled by fast oxidation of the Si wafer. The Hf concentration in the film increased

with deposition time. The films were post-annealed in N2 at 8000C. Upon annealing the

film separated into two distinct layers having approximate dielectric constants of 5.6 and

9.3 respectively. Equivalent thicknesses of 19 A were reported for the thinnest layers.

The incorporation of Si into the depositing HfO2 film must be controlled if equivalent

thicknesses of -10 A or less are to be realized.

Using reactive DC magnetron sputtering, L. Kang et al.30 formed HfO2 films with

physical thicknesses of 45 A. The sputtering was done at room temperature in a

modulated 02 flow to control the interface qualities and to suppress the additional growth

of an interfacial layer. The deposition was followed by an ex-situ anneal at 5000C in N2

ambient for 5 min. Pt was used as the gate electrode. They reported an equivalent









thickness of -13.5 A. The leakage currents for the films were about Ix104 A/cm2, which

is four orders of magnitude less than that for an equivalent SiO 2 film. Hysteresis values

of <100 mV were also obtained. This study shows that HfO2 can be successfully used as

an alternative gate dielectric under the right processing conditions. However, thinner

equivalent thicknesses will be required for future technologies.

S. W. Nam et al.31 studied the effect of annealing ambient on reactive DC

magnetron sputtered Hf02 films. First, Hf metal was pre-deposited as an oxidation

barrier, and then a thin HfO2 layer was deposited at room temperature in Ar and 02

ambient from an Hf metal target. The Hf metal layer helps to form HfO 2 by reducing the

native oxide on the substrate. Samples were annealed at temperatures ranging from 500-

900'C in both N2 and 02. The as-deposited films were amorphous in structure.

Formation of a monoclinic phase was reported for annealing temperatures above 6500C

and at greater than 8000C an orthorhombic HfO2 structure emerges. The equivalent

thickness values obtained in this study were calculated to -19 A for N2 annealed HfO2

films and -28 A for 02-annealed films. The increase in equivalent thickness is believed

to be due to the increase of the thickness in the interfacial oxide layer.

Nitride Barrier Layers

Attempts have been made to control the interface by using barrier layers. P. D.

Kirsch et al.32 investigated the electrical properties of HfO2 films on nitrided and un-

nitrided Si substrates. The samples were prepared by magnetron sputter deposition of Hf

metal and subsequent oxidation on both bare silicon and on wafers which had been

exposed to NH3 for 30 s at a temperature of 7000C. The samples were annealed in N2

ambient for 10 s at 6000C. After annealing, MOS capacitors with HfO2 thicknesses of

-50 A were fabricated using TaN as the gate electrode. The nitrided samples









demonstrated leakage current densities two orders of magnitude less (4x10-5 A/cm2) than

the non-nitrided samples (4x10-3 A/cm2) and an increase in capacitance of 15%.

Equivalent oxide thickness for the 50 A films was estimated to be 10.6 and 11.7 A with

and without SiNx, respectively. Therefore, use of a nitride barrier layer can improve the

leakage and capacitance characteristics of HfO2 gate dielectrics. However, the nitride

interface layer degrades mobility and capacitive response due to additional interface

traps.

Other Novel Approaches

Other novel approaches have been used to tailor the electrical characteristics of

HfO2. H. Lee et al.33 examined using Dy doping to control oxygen vacancies in the film.

Since the electronegativity of Dy is lower than that of Hf, the concentration of oxygen

vacancies in the Dy-doped HfO2 was lower than that of their control HfO 2 and thus

yielded leakage current densities over three orders of magnitude smaller than the control.

Furthermore, the large atomic radius of Dy was predicted to reduce the leakage current

through the dielectric due to a packing density effect.

In a study by H. Yu et al.,34 A1203 was added to HfO2, and a range of

compositions were compared in an attempt to control and model the energy gap and band

alignment for the dielectric on Si. Band gap and offset values for various compositions

ranging from pure Hf02 to pure A1203 were carried out using high-resolution XPS. The

Al 2p, Hf 4f, O is core levels spectra, and O is energy loss spectra all showed continuous

changes with the variation of the HfO2 mole fraction. From this data, they were able to

model the band gap and band offsets for the (HfO2)x(Al203)1-x system. This system is

important because of the higher temperatures at which the film will remain amorphous.

It has been shown that Al concentrations of greater than 31.7% will result in









crystallization temperatures between 850 and 900'C, which is about 400'C higher than

that for HfO 2.

Various studies have also investigated the interface layer of the Hf02 dielectric to

the gate electrode. One of the approaches used barrier layer formation by nitriding the

top surface of the HfO2 before electrode deposition creating a thin HfxNy layer so B

doped poly-Si gates could be used.35 A similar study was conducted using A1203 as the

capping layer.36 Both types of barrier layers were shown to reduce leakage and improve

the stability of the doping levels in the gate.

UV Assisted Oxidation

Ultraviolet (UV) radiation has been shown to assist in the thermal oxidation of

silicon.37-39 For temperatures around 8500C, thermal oxidation of silicon in dry 02 is on

the order of 2A/min even though the supply of 02 from the gas phase to the SiO 2 surface

is plentiful. This is due mainly to the high activation energies of >1.5 eV required for 02

diffusion through the oxide to react with the underlying silicon. By using a UV source

producing photons with wavelengths of less than 200 nm, the energy of the photons is

great enough to split 02 bonds, thus forming more reactive species. The following

equations (1-3..1-6) show the reactions that take place in the gas phase.40 M is a third

body.

02 + hv 20 (1-3)

0+02+M->03+M (1-4)
03 + 0 202 (1-5)

03+ 02 -> 02 +02 + 0 (1-6)
Furthermore interaction of photons with wavelengths less than 254 nm will cause the

ozone species to break apart as shown in equation (1-7).












03 + hv 02 + 0 (1-7)

Within the bulk, photoinduced transitions occur from the conduction band of Si to

the conduction band of SiO2 (3.15 eV) and from the valence band of Si to the conduction

band of SiO2 (4.25 eV) when the sample is irradiated.4142 Therefore during irradiation of

UV, the Si02 layer is supplied with both atomic oxygen from the gas interface and

electrons from the Si/SiO2 and thus the probability of ionization of the oxygen species is

enhanced.

The oxidation rate of light-enhanced thermal oxidation of silicon at temperatures

above 7400C can be explained using the photo-enhanced Deal and Grove model.38'43

Below this temperature, the Carera and Mott model demonstrated by Ishikawa et al.

better explains the initial 70 A of film growth and thereafter the Deal and Grove model

seems to fit.44 In this model, the oxidants are assumed to absorb on the surface creating

vacancies and thereby encouraging electrons to tunnel through the oxide to fill them. The

charged species then move under the influence of the space charge through the oxide to

the oxide/silicon surface.

Current work is being conducted on the oxidation of zirconium metal to form

dielectric films.45-49 Ramanathan et al. deposited thin 20-30 A films of Zr at room

temperature onto peroxide washed Si substrates. The washing process leaves a thin

amorphous Si02/SiON layer. The films were then exposed to Hg vapor lamps emitting

primarily at wavelengths of 185 and 254 nm. They have shown that the light interacts

with 02 to form atomic 0 and ozone enhancing the oxidation rate of Zr enabling

formation of a stoichiometric metal oxide up to about 55 A thick after 90 minutes at room









temperature.46 They compared these films to ones oxidized without the use ofUV

assisted annealing. The films oxidized without UV demonstrated severe frequency

dispersion in both the accumulation and depletion regions of the capacitance-voltage

plots. Using electron energy loss spectroscopy and chemical mapping of the oxygen

concentration, they were able to explain that the cause of the poor electrical

characteristics resulted from a high concentration of oxygen vacancy defects. The films

oxidized using UV showed very little frequency dispersion and had a more constant

oxygen distribution through the thickness of the layer.

In another study Ramanathan et al.47 proposed that the oxidation of Zr films by

the UV-ozone method is expected to be self-limiting at room temperature. To verify this,

capacitors were fabricated by oxidizing -20 A Zr films for 60 min and 90 min. Both sets

of samples were subjected to a forming gas anneal at 4000C for 30 min. Then capacitors

were made from the films by using Pt electrodes. Capacitance-voltage curves were

measured at 100 KHz from both samples and yielded nearly identical results. This

implies that the oxidation is self-limiting for the thickness range investigated and also

shows that lengthy room temperature oxidations do not result in significant growth of

interfacial SiO 2. Further analysis in this study compared the hysteresis of reported values

for Zr02 from other groups to the films they formed by the UV-ozone oxidation of Zr.

They were able to obtain films with -100 mV hysteresis for physical thicknesses of 50 A.

Their value was markedly better than the 200 mV hysteresis reported by Wang et al.50

Electrical traps in the zirconia films are proposed to be the cause of the hysteresis.51 UV-

ozone oxidation eliminates many of the oxygen defects which contribute to the









hysteresis. Ultimately films with <15 mV are desired for future implementation into

device structures.52

Motivation

Even after years of research in an attempt to find a suitable replacement dielectric

for SiO2, more research is still needed to understand the role of the interface, to tailor the

chosen dielectric, and to insure reliability.

This three part work first examines the interfacial layer formation of pulse laser

deposited (PLD) HfO2 ultra-thin films under different processing conditions. The role of

temperature and oxygen deposition pressures was studied. By understanding the nature

of the interface and how it is formed we can better control it to tailor the ultimate

properties of the film.

The second portion of the work combines the use of UV-assisted oxidation and the

incorporation of nitrogen into the films during deposition to produce higher quality films

and interfaces. The use of UV radiation lowers the temperature required to fully oxidize

the thin metal films and limits the oxygen vacancy and non-stoichiometry issues typically

encountered in PLD deposition of oxides. The nitrogen incorporation during deposition

limits the diffusion of oxygen to the interface thus slowing interfacial oxide growth.

Finally, aluminum was added to the films during deposition to increase the

crystallization temperature in an attempt to maximize the dielectric constant while

minimizing the leakage current. Initial studies of the composition and structural

properties of the films, as well as, the electrical characteristics were conducted after

deposition and UV-oxidation processes were concluded. The samples with the lowest

concentration of aluminum were subsequently heat treated at 500, 700 and 900 'C to

determine the crystallization temperature of the films.














CHAPTER 2
EXPERIMENTAL PROCEDURES

Deposition Setup and Equipment

All the thin film samples in this work were deposited using a pulsed laser

deposition (PLD) technique in an ultra-high vacuum compatible chamber utilizing an

excimer laser system. The chamber has been modified for use with an ultra-violet (UV)

lamp array for irradiation during deposition and for in-situ post-deposition (PD) anneals.

UVPLD System

A Neocera brand single chamber vacuum system was used for all the sample

depositions. Figure 2-1 shows a schematic of the system along with laser and the optics

setup used for the pulsed laser depositions. Due to the lack of a loadlock for the system,

the chamber was backfilled with nitrogen to bring it to atmosphere each time a sample

was loaded/unloaded. The chamber pumping system consists of a Pfeiffer MD-4T oil

free diaphragm roughing pump backing a Pfeiffer TMU 230 turbo pump. Vacuum levels

of 1x10-6 Torr can be reached within an hour, 1x10-7 Torr within twenty four hours, and

1x10-' Torr within 3 days. Addition of a liquid nitrogen cooling to the system allows for

Ixl10' Torr within 24 hours. Higher vacuum levels are possible if a loadlock system is

added and thus eliminating the viton seals on three of the ports.

Samples were mounted on a 2" Neocera brand stainless steel resistive heater

capable of 8500C and positioned vertically in the chamber. The system has a variable

controller to regulate the substrate temperature and has been calibrated within 5 'C.

This system includes a computer controlled multi-target carousel available for












KrF
Excimer

Laser


248 nm


Lens


Reflecting


2" Stainle-s
Srtel HWatier


Reflecting Mirror





) Focal Lens


Load Lock
Doorway Plume Hg Lamp Variable Leak
Array /Valve

Load Lock
Doorway NH,

Multi-Target
Carousel
Figure 2-1: UVPLD system with excimer laser and optics setup.

depositions of up to six different materials for multilayer/mixture or superlattice

experiments. This carousel was used for the third portion of the experiments. An array

of 4-11" u-bend low pressure Hg lamps from Atlantic UV was added to the conventional









PLD setup to convert it into a UVPLD apparatus. The lamps were placed on a rack in a

plane parallel to the surface of the substrate halfway between the substrate and the target.

Figure 2-2 shows the placement of the lamps and target. Highly sensitive Varian brand

leak valves were used to control input of ultrapure gases into the chamber allowing for a

wide range of deposition ambients and pressures.























Figure 2-2: Photograph of UV lamp placement in the chamber.

Laser Ablation

Many techniques are available for the deposition of thin films onto a substrate,

however pulsed laser deposition has several key advantages that make it an ideal small

batch research tool. PLD provides near-stoichiometric transfer of molecules from the

target to the substrate, rapid testing of many different materials, and offers a wide range

of applications, though it does not lend itself well for industrial scale production.

The ablation process itself is characterized by an input of energy from the laser to a

given target material. Figure 2-3 shows a schematic of the energy distribution at the









target surface during ablation. The total energy for the system is represented in equation

2-1.


E = E + E+ Ed+ Ec

E = incident energy
Er reflected energy
E, = plasma plume energy
Ed = energy of disintegration
Ec = energy absorbed by the cavity wall


(2-1)


E=E +E +Ed+EC


JEWL


irradiated
material


.- -a -


ablation
depth


Figure 2-3: Laser energy distribution during target ablation.

There are additional factors related to the laser energy that play an important role in

surface response to pulse energy. The incident pulse energy must exceed the ablation

threshold energy for the material for ablation to occur. If the pulse energy is less than the

ablation threshold, material will not be physically removed from the surface (i.e., Ep and

Ed tend to zero), but that energy will be absorbed by the cavity wall (Ec) as heat, which









can be useful for laser annealing applications. The degree to which the target material is

ablated depends on the extent the energy is exceeded. If excessive energy is used, large

particulates from the target can be blown off. This is unwanted for high-quality films.

All experimentation in this study was carried out at an energy value greater than the

ablation threshold but low enough to minimize particulate ablation.

Laser System

A Lambda Physik LPX 305 I KrF excimer laser (s/n 9412 E 4188) was used to

deposit all the samples in the following experiments. This model laser uses a pulsed

mode delivering 25 nanosecond duration square shaped wave pulses at frequencies from

1-50 Hz and output energies from 10-1100 mJ yielding ultimate fluences from 0.1-4

J/cm2 depending on optics used. The laser is fired or triggered by either an internal

computer module near the laser or from a remote external computer placed near the

deposition chamber, which also controls the multi-target carousel.

Excimer lasers function on the principle of stimulated photoemission of gases

within a set cavity. This laser uses a mixture of Kr, F, and a buffer gas, typically Ne or

He. High voltages ranging from 16-23 kV elevate electrons in the Kr and F atoms into

excited states causing excited KrF* complexes to form. Upon their decay, 248 nm single

wavelength radiation is given off. The cavity is designed such that conditions of

stimulated emission of coherent radiation occurs and thus subsequent amplification of the

radiation is achieved resulting in high energy laser output in a pulsed mode.

A series of lenses and an aperture are used to select the highest energy portion of

the beam, direct, collimate, and focus the pulse until it impinges the sample. For the PLD

system, the laser beam first passed through a 2 cm x 1 cm aperture to select the center

portion of the spot from the laser. This eliminates the lower energy diffuse side portions









of the spot and gives the initial rectangular spot shape. Due to positioning of the

chamber, the beam is then reflected using a UV mirror optic placed at 450 to the beam

path. Once reflected, the incident beam is passed through a collimating lens to keep the

radiation from diverging due to scattering as it passes through the ambient air. One more

mirror is used to reflect the beam to the chamber. The beam then is passed through a

focusing optic with a 30 cm focal length placed just outside of the chamber. The beam

enters the chamber through a fused silica window port rated for use with excimer

radiation. This window passes 90% of 248 nm radiation.

A Gentec Sun Series EMI energy meter (s/n 86052) was used to determine the

fluence reaching the sample. The laser was operated at the lowest energy possible, the

focusing optic was removed, and the meter head was placed inside the chamber. Lower

energies had to be used as the meter head has a low maximum energy tolerance before

damage to the surface will occur. Averages of 20 pulses were used to record the energy

of the spot impinging the meter head. An additional 10% of the energy was subtracted to

take into account the focusing optic. By taking the energy value obtained and dividing by

the area of the spot on the target, precise calculation of the laser fluence was possible and

then scaled with the laser energy settings used for the experiments.

The laser ablation spot impinging the target had dimensions of 2 x 5 mm,

maintaining the shape of the rectangular aperture used at the start of the beam path. By

moving the focusing optic to a position of 35 cm from the target, the target is past the

focal length of 30 cm of the optic and thus is in the imaging plane region allowing for

conservation of the spot shape. By placement of the lens such that the focal point

coincides with the target surface, smaller spot sizes and greater fluence can be obtained.









However, it was unnecessary because the threshold value for laser ablation of the oxide

materials and metals used in our experimentation were sufficiently exceeded with the

current optics setup. Having a larger spot size also gives more uniform ablation of the

target so trenching does not occur as rapidly. Another argument for not using the focal

point for ablation arises from the irregular shape of the spot which results. The spot

geometry is shifted to an elongated oval shape with two additional satellite peaks to either

side of the main ablation spot. The energy of the main spot was characterized by a

Gaussian type distribution while the satellite peaks were of lower energies based on burn

paper tests. All of the samples in this work were deposited using the previously

mentioned laser and optics setup.

Experimental Setups and Samples

There are three main areas that comprise the work discussed in this dissertation.

This section delineates the differences in the samples and gives the motivation behind

each portion of the experiments.

Interfacial Layer Formation and Growth

In this initial portion of the experimentation, the formation and characteristics of

the interfacial layer were investigated. Ultrathin HfO2 samples were deposited on silicon

by ablating a high purity HfO2 target in a conventional PLD chamber under different

oxygen ambients and substrate temperatures. The main goal for this investigation was to

determine the nature of the interfacial layer and the factors that control its formation and

growth. The conditions for deposition are shown in Table 2-1. The samples were

investigated by variable angle spectroscopic ellipsometry, x-ray reflectivity, and angle

resolved x-ray photoelectron spectroscopy. These techniques will be discussed later in

this chapter.









Table 2-1: Conditions for growth of ultra-thin Hf02 films.
Sample ID# Temp. 'C Pressure Torr
Hfll 1x10-4
Hfl2 200 1x10-3
Hf13 1x10-2
Hfl4 Ix10-4
Hfl5 400 1x10-3
Hfl6 1x10-2
Hfl 7 Ix10-4
Hfl8 600 1x10-3
Hfl 9 Ix10-2


UV Assisted Oxidation and Nitridation of Hafnium Films

The second portion of this dissertation examines both the effect of UV oxidation of

hafnium metal thin films and the effect of nitrogen incorporation into the films. Four

conditions were used and are denoted in Table 2-2. Half of the samples were deposited

in ambient vacuum levels of IxlO-6 Torr while half were deposited in IxlO-2 Torr of

ammonia. Half the samples were oxidized in 300 Torr of oxygen at 650 'C without the

aid of UV radiation while the other half were oxidized for 30 minutes with the lamps on

under the same temperature and pressure used for the non-irradiated samples. Target film

thicknesses of 40-50 A were obtained after oxidation for all the samples. This

experiment was repeated three times to verify reliability and repeatability of the process.

The goal of this portion of the experiments was to improve the electrical characteristics

and film stoichiometry while using lower processing temperatures. Lower deposition

temperatures usually result in poorer quality films when using PLD to deposit them,

however by depositing a metal then oxidizing it with a post-deposition (PD) annealing

step higher quality films can be realized.

In an attempt to incorporate a significant amount of nitrogen into the films, higher

temperature depositions carried out at 650 'C. A set of four samples were produced, all









of which were deposited in 1x10-2 Torr of ammonia. Half were deposited with the use of

UV radiation during deposition and half were deposited with the lamps off. Then half of

the samples were annealed in UV during the oxidation step. Table 2-3 shows the sample

identification and conditions for this set of samples.

Table 2-2: Growth and postdeposition conditions for UV oxidation of Hf metal ultra-thin
films.
Sample ID # Deposition Pressure Torr Anneal Conditions
H2 1x10-6 UV
H3 1x10-6 No UV
H4 1x10-2 Ammonia UV
H5 1x10-2 Ammonia No UV


Table 2-3: High temperature Hf metal deposition and oxidation conditions.
Sample ID # Deposition Conditions Anneal Conditions
HN1 UV UV
HN2 No UV UV
HN3 No UV No UV
HN4 UV No UV


UV Assisted Oxidation and Nitridation of Hafnium/Aluminum Alloy Films

Finally, to increase the crystallization temperature of the films aluminum additions

were studied. This set of experiments used the carousel with two high purity targets of

Hf and Al. The pulse ratio between targets was varied to obtain different film

stoichiometry. The sample conditions are listed in Table 2-4. Half of the samples were

deposited in ammonia to incorporate nitrogen into the structure as was done in the second

set of experiments. After electrical characterization, the samples were annealed at 900 oC

in N2 to study the detrimental effect of high temperature anneals and to determine if the

aluminum additions will suppress crystallization.









Table 2-4: Growth and annealing conditions for Hf/Al metal films.
Sample ID # Pulse Ratio 1x102 Torr Ammonia
AH1 1 Al: 1 Hf No
AH2 1 Al: 2 Hf No
AH3 1 Al: 3 Hf No
AHNI 1 Al: 1 Hf Yes
AHN2 1 Al: 2 Hf Yes
AHN3 1 Al: 3 Hf Yes


Experimental Characterization Techniques

Many techniques are available for characterizing the structure, morphology,

stoichiometry, and electrical properties of thin films. This section reviews the techniques

used to compare the samples in this dissertation to delineate the effect of temperature,

pressure, UV oxidation, nitrogen incorporation, and aluminum additions.

Variable Angle Spectroscopic Ellipsometry

Variable angle spectroscopic ellipsometry (VASE) is a simple, yet very powerful,

nondestructive analysis tool for determination of thickness in thin film multilayered

structures. Optical properties such as index of refraction and the extinction coefficient

can also be obtained. This technique is best suited for flat planar materials with

thicknesses ranging from 1-1000 nm having low surface roughness, which makes it an

invaluable tool for analysis of the thin film samples in this work.

A J. A. Woollam brand M-88 variable angle ellipsometer was used for sample

analysis. VASE operates by reflecting a collimated beam of light of known polarization

from a Xenon lamp off of the surface of the sample to be analyzed. A second polarizer is

used to determine the polarization shift resulting from interaction with the sample. The

angle of incidence and wavelength of light used are extremely important factors









dependent on the thickness and type of sample being analyzed. For dielectric films, an

incidence angle of 750 is optimal.

All electromagnetic phenomena are governed by Maxwell's equations and thus

certain mathematical relationships can be determined when light encounters boundaries

between different materials. Ellipsometry uses Snell's Law and Fresnel reflection

relationships to determine delta and psi for the film. The software package that runs the

equipment plots the psi and delta values obtained.

A model is then needed to interpret the data. It is best if information such as

estimated thickness and density are known. The software has a database of material

properties for use in the models. Various mixture routines are available for multi-

component films. The software takes the initial information the user inputs and iterates

through a routine until the model has the lowest least square fit.

X-ray Reflectivity

X-ray reflectivity (XRR) measurements were made with the assistance of a

Panalytic MRD X'Pert system. XRR is another non-destructive optical characterization

technique that yields similar data to that of VASE. XRR will verify the VASE result and

can be used to better fit the models used to interpret the raw data obtained from both

analysis techniques. X-ray reflectivity plots provide information on the thickness,

roughness, and density of a given material. Figure 2-4 demonstrates an example of a

typical plot showing how the features of an acquired spectrum relate to the various data

that can be extracted from the spectrum. Similar to VASE, this technique is well suited

to smooth flat samples with low surface roughness and is most sensitive to samples in the

10-400 nm thickness range.










This particular x-ray based analysis technique works in the traditional 0-20

geometry and scans angles from subcritical to a few degrees. The detector records the

intensity of the radiation reflected from the surface of the sample. At angles of less than

the Brewster's angle, also know as the critical angle, total reflection occurs. At angles

greater than the Brewster's angle, the radiation begins to interact with the surface and

interfaces within the sample. This interaction is governed by Snell's law and Fresnal's

relationships similar to VASE. The constructive and destructive nature of the x-rays at a

particular angle generates the fringe pattern in the detected signal. The raw data are

modeled using a user-input system whereby the density, roughness, thickness, and

absorption coefficients can be changed until a suitable fit is obtained.


1 Qi- .i:-+ _: I___._._


1 rjCIEt:i: 1{-


Thickness i".:.irL Density
Silicon ao 6 A 2.33 g/cm3
iiin ill 16A 4A 2.73 gcm
HfO02 4I 4A 9.46 g/cm'


1 OOdE+02-4




1.00E+02 -


Den'.ity Thickness
1.00EI+00-----------------------
0 1 2 3 4 5
2-Theta (Degree)

Figure 2-4: Typical XRR plot showing which film properties most influence various
features.









X-ray Diffraction and Glancing Incidence X-ray Diffraction

X-ray diffraction (XRD) is an invaluable tool that provides phase analysis as well

as strain, grain size, epitaxial quality, phase composition, preferred orientation, and defect

structure. This technique is nondestructive and noncontact and can be used for films

greater than 50 A. The basic equipment setup is shown in Figure 2-5. X-rays (typically

Cu K a) impinge the sample at a specific angle. The scattering of the X-rays by the

sample atomic planes produces constructive and destructive interference. The condition

for constructive interference from planes with a given spacing is given by Bragg's Law

shown in Equation 2-2. When this condition is met, peaks in the intensity detected will

occur.

? = 2dhklsin?hkl (2-2)

? = wavelength of the x-ray radiation
dhkl = d-spacing between (hkl) planes
?hkl = angle between the atomic plane and the incidence direction


For single crystal films, there is only one specimen orientation that will satisfy the

conditions for Bragg diffraction for each family of planes, so careful alignment

procedures must be taken. However, with thin films that are polycrystalline, fiber

textured, or exhibit preferred orientations several families of planes may contribute to a

diffraction system. X-ray diffraction data for this dissertation was obtained with a

Panalytic X'Pert MRD diffraction system. Once an x-ray diffraction pattern is generated,

positive phase identification can be achieved by comparing measured d-spacing from the

diffraction pattern (and their integrated intensities) to a known JCPDS powder diffraction

standard. Additional information such as grain size can be generated from the full width

at half max of the peaks.










Normal XRD mode of operation could not be used for the thin films produced in

this dissertation. The high intensity peaks from the underlying silicon substrate masked

the weak signal from the film. Glancing incidence x-ray diffraction had to be used in this

case. In this setup, the incident angle of the x-ray system is fixed at a small value

(typically from 0.250-1.000) so only the surface of the sample is probed and the

receiving slit is allowed to scan through a typical range for a conventional XRD 2? scan.

As a result, only planes that satisfy the Bragg condition with the additional constraint in

place will produce peaks. In a polycrystalline sample with many orientations, this will

still generate a representative x-ray diffraction plot, but with much higher surface

sensitivity without any masking peaks from the silicon substrate. This method proved to

be beneficial for several samples in this dissertation.


incident


substrate
thin funrr


*1Ai! 3. Ird
y-r i,-


Figure 2-5: Typical XRD equipment geometry setup. [53]









X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) is one of the most commonly used

techniques for thin film chemical analysis along with Auger electron spectroscopy (AES).

This surface sensitive technique provides elemental as well as bonding environment

information. Elemental analysis can be obtained for all elements except hydrogen and

helium. XPS spectra were collected using a Perkin Elmer 5100 installation using Mg Ka

radiation (1486.6 eV) at takeoff angles from 45-90.

This technique works by impinging high energy photons at a specific angle to the

surface. The photons cause electrons from the atoms in the sample to ionize and form

photoelectrons. The kinetic energy of the electron is measured and used to calculate the

binding energy based on Einstein's Photoelectron law given in Equation 2-3.

KE = h? BE (2-3)

KE = Kinetic energy
h? = Energy of incident photons
BE = Binding energy


Each atom can be identified by the binding energies of its electrons. By comparing

the ratios of photoelectrons from the different atoms along with the sensitivities for each

a quantitative chemical analysis can be obtained. Compositional changes across the

thickness of the thin film sample can be generated by one of two ways. The first is by

using an argon sputtering gun which removes a portion of the sample from the surface at

a set rate. The other, which was used in these experiments, is angle-resolved XPS,

whereby different take-off angles are used and the data generated is compared. This

allows for different interaction volumes of the sample to be probed to compare any

differences that might occur through the thickness.









Atomic Force Microscopy

Atomic force microscopy measurements were made with a Digital Instruments

brand Nanoscope III operating in both contact and tapping mode. This technique was

chosen for its exceptional ability to produce topographic images of surfaces in three

dimensions at high resolutions. Atomic resolution is attainable, if proper conditions are

set. This technique is also perfectly suited to the insulating, low roughness samples

created for this dissertation. In an atomic force microscope, a sharp tip is mounted on a

flexible cantilever. A piezoelectric scanner moves the cantilever/tip assembly along the

surface of the sample, while a laser reflecting off the end of the cantilever maps the

topographical changes the cantilever senses. When the tip comes into close proximity of

a sample surface, van der Waal forces repel the tip causing the cantilever to deflect, in

contact mode the tip stays at a constant distance from the surface while in tapping mode a

potential is applied at a specific frequency moving the tip near the sample surface at a

certain cyclical rate. The typical operational setup for the AFM is shown schematically

in Figure 2-6.

Electrical Characterization

A primary tool for indication of quality of thin film is the use of electrical

characterization techniques. After deposition or processing of thin films, metal oxide

semiconductor (MOS) devices were fabricated and measured. Typical preparation of the

film included the deposition of platinum (Pt) contacts via DC magnetron sputtering. A

shadow mask with an array of circular dots ranging from 25-500 |tm was used to create

dot arrays on the samples for the gate contacts. The backside contact was silver (Ag).













mirror

N\'
\\\e.\ ""Y


L aser 60je

PSPF,



i
I
I
/
/
-I
I
1
I
I
(


ca n canilever'/ T
feedbacK loop





irnaao<---t -




PZT scanner


Figure 2-6: AFM operational setup showing laser detection of the cantilever deflection.
[53]

The backside of the wafer was first abraded and swabbed with 1% HF so that


clean silicon, without any native silicon dioxide, was present. It is of paramount


importance to choose the proper metals for the dielectric film being measured and the


type of silicon substrate that was used. If improper metals are chosen for a given setup,


band alignment conditions may exist whereby non-ohmic contacts are created within the


device. As an example, the backside metal contact on a p-type silicon wafer should have


a work function greater than that of the silicon. If these conditions are met, there should


be an ohmic contact, if not, the contact will be a rectifying Schottky contact. A


schematic of a MOS capacitor fabricated for this dissertation is shown in Figure 2-7


demonstrating the testing setup used for these experiments. After the front side gate


metallization was complete, samples underwent heat treatments in a conventional tube









furnace under a flowing forming gas atmosphere (4% H2, balance N2). Both current-

voltage and capacitance-voltage measurements were taken once the MOS capacitors were

formed.

Microprobe

Microprobe

SRcke CPt Gontact Gate Oxide
4As Bnckside Contnct '





Glass Slide-type Si


Figure 2-7: MOS capacitor setup used for CV/IV measurements.

Current-Voltage Measurements

Determination of the leakage current is an essential first step in analyzing an MOS

device. Once MOS devices were fabricated, a Keithley Instruments Inc., KI236 source

measurement unit (SMU) was used to measure the current flow through a device. The

236 SMU was attached to a pair of Signatone Inc, micromanipulators mounted on a probe

station. Tungsten probes that were milled to produce a fine ~5 tLm tip were fitted in the

micromanipulators. The output of the SMU was connected to the micromanipulator in

contact with the Pt gate while the input was connected to the backside Ag contact. This

configuration is optimal for the determination of current leakage pathways directly below

the device being measured (i.e., it avoids stray leakage paths). The current compliance

threshold, a value that may not be exceeded by the machine, of 100 nA was input into the

measurement parameters. Then a direct current bias sweep was conducted over a voltage

range (from negative to positive) and the amount of current that passed through the MOS









structure was monitored. To reduce creation of bias induced defects, typically it is best to

start at a small voltage sweep range so that it is possible to determine if the device is

leaky before a wider voltage sweep range is implemented. The main goal of the leakage

current measurement is to identify a high quality device that can be used for capacitance

voltage measurements and to determine if excessive leakage is present that needs further

investigation.

Capacitance-Voltage Measurement

Capacitance-voltage measurements serve as one of the most versatile and sensitive

of all electrical characterization techniques. It is the ultimate tool for determining

discreet differences in a MOS device that may serve as the final word in whether a given

processing condition has resulted in a high enough quality device to apply to MOSFET

applications. These measurements were carried out with a Kiethley Instruments Inc.

Win-82 measurement system. This system, as seen in Figure 2-8, is comprised of four

main components working in unison. The Keithley 595 capacitor is used for

simultaneous quasistatic low frequency measurements. The Keithly 590 capacitance

meter is used for high frequency capacitance measurements at 100 kHz and 1 MHz. The

Keithley 230 voltage source is used for static bias condition measurements. These three

devices are wired into the Keithley 5951 remote input coupler, which serves to filter and

relay the device data to and from the various pieces of equipment. Due to its ability to

give detailed information about the quality of the MOS device, the Win-82 system will be

discussed in Appendix A.

The output of a typical CV curve can be seen in Figure 2-9. There are several

important features that should be noted. First, there are three main regions with respect
































Figure 2-8: Kiethley Instruments Inc. Win-82 measurement system. Modified from the
Win-82 manual.

to gate bias voltage to take into consideration, known as accumulation, depletion, and

inversion. The presence of the different regions is a result of the majority charge carriers

(e.g., holes in a p-type silicon wafer) in the semiconductor. When a negative bias is

applied to the gate electrode, positively charged holes are attracted from the

semiconductor bulk region to the oxide/semiconductor interface where they accumulate

(the accumulation region) in order to maintain charge balance in the system. The

depletion region is generated when a gate is made less negative and the reduced field

across the oxide allows the charge at the interface to diminish. As the sign on the voltage

changes from a negative to a positive, majority carriers are repelled from the interface

creating an area depleted of majority carriers (the depletion region). Finally, the

inversion region is generated when the voltage becomes very positive and the depletion

width has increased to a point where other mechanisms become important. For example,









in the depletion region, the product of the concentration of electrons and hole (np) is

much less than the square of the intrinsic carrier concentration (ni2) and in this case, pair

generation may occur and the subsequent minority carriers can migrate to the interface.

The result of this is prevention of further depletion and a constant value for the

capacitance is maintained. An additional possible scenario may occur whereby

insufficient time is allowed for pair generation in which case a model called deep

depletion will occur.

The high frequency measurement system has the ability to measure two different

frequencies and using Metrics ICS software with two different equivalent circuit models,

seen in Figure 2-10, to generate capacitance values. This system does not directly

measure a capacitance value for the films, but instead measures resistance and

conductance and uses the models to generate the capacitance values. The parallel and

series models represent two different physical structures. The series model addresses a

capacitor in series with a resistance, possibly from the semiconductor. The equation that

describes this scenario is:

Z = R + iX (2-3)

where Z is the impedance, R is the resistance, and X is the reactance. Alternatively, the

parallel model addresses a capacitor in parallel with some sort of conductance, possibly

resulting from leakage through an ultrathin film. The equation that describes this

scenario is:

Y = G + iB (2-4)

where Yis the admittance, G is the conductance, and B is susceptance. Note that the







40


admittance, conductance, and susceptance are each the reciprocals of the impedance,

resistance, and reactance, respectively.






Cox
CQ
Depletion inversion
Accumulation


Capacitance


Onset of Strong Inversion


-V GS VF6 VTHRESHOLE, +VGS

GATE BIAS VOLTAGE V GS


Figure 2-9: Typical capacitance/voltage response curve demonstrating major areas of
interest modified from the Win-82 manual.

The reactance and susceptance can then be further described in a capacitive sense as:


X = -
c)C'


(2-5)


Y = oC,P (2-6)

where ? is the frequency of the setup and Cs and Cp are the capacitance for the series

model and parallel model respectively. The net impedance of the equivalent series and

parallel circuits at a given frequency are equal, but the individual components are not:

1


R + iX =
G +iB


(2-7)









If there is a lossless circuit (i.e. R = 0 and G = 0) the Cs and Cp are equal.

However, since circuits do have losses, a dissipation factor is added to the system.

D = wCsR (2-8)


D = (2-9)
coCP

Conversion from one equivalent circuit model to the other is obtainable through

further manipulation of the previous equations yielding the conversion factors shown in

Table 2-5.

Table 2-5: Conversion factors for series-parallel electrical equivalent circuits.
Resistance or
Model Dissipation Factor Capacitance Conductance
Parallel 1 G C, =(1+D2)C D2
CpD s PC R-\D
CP G Q CP (1+ D2)G
Series 1 CC D2
SeiR D I CsR CP S G=
Cs,R Q 1+D2 (1+D2)R



This treatment has been applied to illustrate how capacitance values are generated

from the impedance values and subsequent conversion to series and parallel cases. As

mentioned earlier, the series represents a physical scenario where a capacitor is in line

with some type of resistance, while the parallel mode represents a physical scenario

where a capacitor is in parallel with some type of conductance. This however does not

take into consideration the physical possibility of an ultrathin film that is leaky, but also

encounters resistance from the substrate. In this case, a three element electrical circuit,

also seen in Figure 2-10, could be analyzed in a similar manner as above to develop

equations for converting from either the parallel or series mode to the three element

equivalent circuit that more appropriately represents the physical structure.









(B)

R


(C)


Figure 2-10: Equivalent circuits for estimation of capacitance in a) parallel, b) series, and
c) combination models.


(A)














CHAPTER 3
KINENTICS OF INTERFACIAL LAYER FORMATION DURING DEPOSITION OF
HAFNIA ON SILICON

Oxide thin films are essential components of many advanced structures and

devices. One of the most commonly used substrates for thin film deposition, both for

applications and for properties characterization, is Si wafers. In many instances, the

oxide films are directly deposited on silicon substrates. As long as the oxide films are

thick, the chemical and physical phenomena occurring at the interface with the Si

substrate do not usually play a major role in determining the properties of the deposited

oxide films. However, a renewed interest in the growth of very thin (<10 nm) high

dielectric constant (high-k) oxide films directly on Si has raised the question of the

interfacial layer formation and its properties.54-56 In this portion of the work, a pulsed

laser deposition (PLD) technique was used to deposit HfO 2 high-k dielectrics on Si. New

results regarding the chemical composition and the kinetics of the interfacial layer growth

during deposition of HfO2 thin films on Si are presented here in this portion of the work.

Deposition Conditions

For this study, a 1" diameter x /4" thick hafnium oxide target 99.99% pure from

SCI Engineered Materials was ablated using a KrF excimer laser system in a typical PLD

setup which was described in detail previously.57-59 The deposition parameters used were

laser fluence of 2 J/cm2, 5 Hz repetition rate, 430 pulses (yield thickness from 3 to 4 nm),

substrate temperatures from 200 to 600 oC, and oxygen pressure during deposition from









1x10-4 up to Ix10-2 Torr. Table 3-1 reviews the sample matrix used for the nine samples

in this portion of the experiments.

The films were deposited on (100) Si p- substrates from MEMC cut to 2.5x2.5 cm

squares then washed with acetone and methanol to remove any surface contamination.

The samples were then dipped in a 1% HF solution for 10 minutes to remove the native

oxide and were mounted onto the resistive heater using silver paste. This cleaning

process passivates the dangling bonds of the Si with hydrogen giving between 10-30

minutes to pump down the sample before adsorption of water species and other

contaminates to the surface occur.59-62

Table 3-1: Conditions for growth of ultra-thin Hf02 films.
Sample ID# Temp. 'C Pressure Torr
Hf1 Ixl0-4
Hfl2 200 1x10-3
Hf13 1x10-2
Hfl4 Ix10-4
Hfl5 400 1x10-3
Hfl6 1x10-2
Hfl7 Ix10-4
Hfl8 600 1x10-3
Hfl9 Ix10-2


The surface morphology, thickness, and roughness of the deposited layers were

investigated by x-ray reflectivity (XRR) with a Panalytical X'Pert MRD system. The

XRR spectra were simulated using the software package WingixaTM. The same system

was used for crystalline structure investigations by grazing incidence and symmetric

o-20 x-ray diffraction (GIXD and XRD). The chemical composition and bonding of the

films were investigated by x-ray photoelectron spectroscopy (XPS) in a Perkin Elmer

5100 installation using Mg K, radiation (1253.6 eV) at various take off angles ranging

from 45-90'. The thickness and optical properties of the films were investigated by









variable angle spectroscopic ellipsometry (VASE) with a J. A. Woollam M-88 instrument

using an incidence angle of 750. The thickness, refractive index and extinction

coefficient values were obtained by fitting the acquired data with a Cauchy model using

the equipment software package.

X-ray Results

Upon sample fabrication, samples were cut into 1 x 1 cm squares for analysis. Due

to the cosine distribution type of plume dynamics the film is not of uniform thickness

across the whole sample thus it is necessary to use only the central 1 x 1 cm portion

which is consistently the most uniform with a variation of 5 A. Furthermore this sample

size is ideal for all the analysis techniques. GIXD investigations were conducted at an

incidence angle of Q=10 for all nine samples. The results from the two extreme

temperature conditions of the samples are plotted in Figure 3-1. All samples in the 200

'C and 400 'C sets remained amorphous regardless of the oxygen pressure, which is to be

expected as the crystallization temperature for HfO2 is 450-500 'C. All films deposited

at 600 'C exhibited broad peaks, a sign of rather poor and randomly oriented crystallites.

Identification of the main peaks and relative intensities was found to match the

monoclinic HfO2 phase according to JCPDS reference card #34-0104. The peak at 55.50

is from the (311) plane of the silicon substrate. By using a glancing angle of 10 the

proper diffraction conditions for this peak is present when the sample is mounted with the

<010> or <001> planes in line with the beam. The samples were cut along these

directions; typically the samples are mounted into the X'Pert system so that the beam

impinges the sample perpendicular to the edge and centered on the sample. The (311)

peak can be eliminated by using a phi rotation of the sample of 10-45', however no peaks









from HfO2 are in this region so this alignment step was not used. Symmetric XRD

investigations did not reveal any particular texture in these films, the crystals are

randomly oriented.

After each GIXD and XRD scan was made, an XRR spectrum was taken. The

X'Pert system allows for all three measurements to be made sequentially once sample

alignment has been completed with just a change in the optic slits. Angles from 0.125 to

50 were scanned. A typical XRR spectrum of the deposited HfO2 films is displayed in

Figure 3-2. Each spectrum was modeled using the WINGIXA software package. Also

displayed in Figure 3-2 is the simulation of this spectrum obtained using a three layer

structure: interfacial layer (IL), oxide layer (OL), and contamination layer (CL) located

on the top surface which takes into account the carbon and water containing species.

This three layer structure is represented in schematic form in Figure 3-3.




SIII,








4 21 1111 him






20 ip 40 4-P 711



Figure 3-1: Comparison of GIXD (f=10) for the 200 'C and 600 'C samples deposited in
Ix10-4 Torr 02.







47





counts/s
1E+7
1E+6 N
1E+5
1E+4
1E+3
1E+2

0.151 0.486 0.820 1.155 1.489 1.824 2.158 2.493 2.827 3.162 3.496
0-20, degrees


Figure 3-2: Acquired XRR spectrum and its simulation for a Hf02 film deposited on Si
at 200 oC and 1x10-4 Torr of 02.







CL









Figure 3-3: Schematic of gate oxide stack model used for XRR simulations.




The parameters used for simulations of all nine samples, i.e. layer thickness,

roughness and density, together with the deposition conditions are displayed in Table 3-2.

The mean square error of the fit X is also included in the table. For reference, a good fit


has x values of less than 1x10-2. To begin the iterative modeling process, the mass


density of the hafnium oxide layer was first assumed to be the bulk tabulated value of

9.68 g/cm3. One can note that the region of the critical angle (-0.220) was well simulated

by using this standard value thus indicating that very dense compact films were produced.









The hafnium oxide layer density was held constant for all the sample models while the

density of the IL and CL were allowed to change along with the thickness and roughness

values for all the layers. The contamination layer affects the initial signal falloff and is

necessary to bring the model in line with the raw data. This layer was initially assumed

to have a density of 7.26 g/cm3, which is 75% of the HfO2 layer density to account for

voids and carbon contamination.

From the fit data, the density of the interfacial layer was consistently higher than

2.19 g/cm3, the density of amorphous SiO 2. This is a clear indication that the interfacial

layer is not pure SiO 2, but a mixture containing HfO2. The interface layer roughness

values, both towards the Si substrate and towards the deposited oxide layer are rather

high, sometimes even higher than the surface roughness values. This is another

indication that there was mixing between the SiO 2 and the deposited layer.

Table 3-2: XRR modeling parameters used to fit the spectra. Note values for thickness
and roughness are giving in A and the densities are given in g/cm3.
Sample Thk. Thk. Thk. Rgh. Rgh. Rgh. Dens. Dens. Dens.
ID IL OL CL IL OL CL IL OL CL X
Hfll 1 24.6 41.3 7.8 5.3 3.4 1.8 4.19 9.68 7.07 1.02E-02
Hfl2 31.8 34.3 9.2 9.9 2.7 1.8 3.61 9.68 7.42 3.30E-02
Hfl3 16.3 31.1 7.9 6.1 3.9 2.1 3.38 9.68 6.32 1.27E-02
Hfl4 24.4 37.7 6.7 4.1 7.2 3.0 4.09 9.68 7.22 1.65E-02
Hfl5 22.9 37.9 1.9 3.0 6.1 1.1 3.88 9.68 4.35 3.11E-02
Hfl6 25.7 30.9 7.9 5.9 3.4 2.4 3.45 9.68 7.00 7.75E-03
Hfl7 25.0 39.3 3.1 4.8 5.5 2.9 3.12 9.68 6.17 3.27E-02
Hfl8 31.3 39.0 6.8 4.2 2.1 3.0 3.72 9.68 3.87 2.54E-02
Hfl9 33.5 34.4 5.0 5.3 1.7 5.7 3.46 9.68 5.97 9.55E-03


Increased temperature was expected to promote oxygen diffusion and reaction with

the silicon at the interface of the film. Average interfacial layer thickness did increase

with increasing temperature, however all the data is not sufficient to statistically derive

this conclusion as the 200 'C samples Hfl 1 and Hfl2 exceed the thickness of the 400 'C






49


samples. The 200 oC samples should be further examined to rule out any experimental

errors that might have been encountered which promoted the growth of the interfacial

layer. Another trend that was expected is the increase in interfacial layer formation with

the increase in oxygen pressure. The trend is sufficiently exemplified by the 600 'C

sample set H17, H18, and H19 deposited at Ix104, Ix10-3, and 1x102 Torr 02

respectively. This trend is shown in Figure 3-4 indicating a nearly linear relationship

between the interfacial layer thickness and the log of oxygen pressure.



36 -
y = 1.8458Ln(x)+ 42.683

32 -

S30 -

28 -

26 -
600 deg. C
24 Log. (600 deg. C)
22 -

20
0.00001 0.0001 0.001 0.01 0.1

Pressure (Torr)



Figure 3-4: Interfacial layer thickness dependence on 02 pressure for the 600 'C samples.

XPS Analysis

XPS was used to determine the composition, bonding environmental changes, and

to estimate the thickness of the interfacial layer. Surveys from 0-1000 eV were taken for

all the samples to first determine the elemental makeup and to detect any contaminates.









Then a multi-plex consisting of the C Is, 0 Auger, 0 Is, Si 2p, and Hf 4d peak regions

were collected. The C Is was used to determine calibration shifts resulting from charging

in the samples. All materials analyzed using XPS have sufficient adventitious carbon

contamination on the surface to produce the C Is signal with known peak position of

284.6 eV. The plots were then fit using software that estimates the peaks with a 90%

Gaussian curve. Fit information and analysis tables have been included in Appendix B.

XPS revealed symmetric peaks for the Hf 4d for all the samples, the two extreme

conditions are compared in Figure 3-5, indicating one single oxidation state, which was

identified as HfO2 from their binding energies of 19.5 and 21.2 for the d7/2 and d5/2

orbitals respectively.




lX1IiT Torr y.















25 23 19 -- 15
BuI-'.1ten L 'L

Figure 3-5: High resolution XPS scans of the Hf 4f region acquired from two HfO2 films
deposited on Si under very different conditions.

High resolution scans of the Si 2p region were acquired at takeoff angles of 450,

650, and 900 for the deposited samples. Figure 3-6 shows a typical scan of the Si 2p






51

region along with a three peak model taken using binding energies ranging from 95 to

108 eV. Peak A centered on 100.88 eV, before the C Is shift was taken into account,

represents the signal from SP, Si bonded with Si, and peak B located at 104.45 eV is the

signal from Si4, Si bonded with 0. The third peak, C which was present in the 600 'C

samples, is attributed to substoichometric silicon dioxide represented by silicon bonded to

3 oxygen atoms with a dangling bond.


Counts
A 100.88 eV 1.31 eV 6755.89 cts
4800 B 104.45 eV 2.02 eV 12119.1 cj
C 102.16 eV 2.01 eV 751.49 ctNs'
Baseline: 107.50 to 98.40 eV
47500 Chi square: 4.31393
47000-
46500-
46000-
45500-
45000-
44500-
44000-
43500-
43000-
42500r \J -/
42000-


N


0'


I i I


)vj\ \/

I I I


108 106 104 102 100 98 96
Bindir, rn crLc ,eV

Figure 3-6: Raw data and fit for the Si 2p region for the 600 'C, Ix102 Torr 02 sample.

The ratio of the intensities of the SP to Si4 peaks, represented by peak A and B

respectively, were then used for estimations of the thickness of the SiO2 layer according

to:


(3-1)


-L


do -, sin ahIn[ I, /(ps)+1]






52


where ,oxy=36 A is the photoelectron inelastic mean free path in the oxide film, ac is the

photoelectron take-off angle, oxy is the peak area for the Si-O bonding, Isi is the

integrated peak area for the signal from the substrate, and P=0.8 is a calibration constant

for our XPS system which will be further discussed in the next chapter.63 Although it has

been claimed that Equation 3-1 could only work for takeoff angles greater than 600, we

have found that it can give rather accurate predictions down to 350. The d/X values for

both a 5 nm pure SiO2 standard and for a sample deposited at Ix102 Torr of 02 and 600

'C are plotted in Figure 3-7. The variation between the 65' and 900 measurements for the

deposited sample exceeds the error of our system based on the 5 nm SiO2 standard. This

suggests that the interfacial layer is a physical mixture ofHf02/Si02. If the interfacial

layer was uniform throughout, the d/X would have yielded the same result for both the

65' and 90' degree scans. The estimated IL thickness using Equation 3-1 and a take off

angle of 900 is displayed in the summary Table 3-4.




I -- -I -

1.3
1.2--

-----.--- -------------------

0.9

: I I n *'' : I Im


0 20 -t o 100


Figure 3-7: Variations of d/k values for a 5 nm pure SiO2 on Si and a deposited sample
with changes in the take off angle.






53


The interfacial layer thickness values obtained using XPS are plotted in Figure 3-8.

The 400 'C and 600 'C sample sets show a nearly linear increase with the log of oxygen

deposition pressures used. The slope of the relationship is the same for both sets.

Furthermore, higher temperature increased the interfacial layer thickness formation. The

200 'C sample set again did not follow the trends seen with the other two sets. Problems

such as incomplete removal of the native oxide during the cleaning process might be the

cause of this error.







rL- r --" --





-4---


_' i-i-111111 i i -111111 i 1 11111
0 ,1' ': I 1-I1 14.'.




Figure 3-8: Interfacial layer thickness as a function of temperature and pressure.

Ellipsometry Results

Finally, the thickness of the deposited structures was estimated using VASE. The

acquired spectra were simulated using the three layer model used for XRR simulation.

An effective medium approximation Cauchy model consisting of 75% Si02 and 25%

HfO2 for the IL and 50% HfO2 and 50% voids for the CL was used to simulate the

acquired data. The simulation data were quite close to the acquired data without any









need to modify of the HfO2 refractive index or add absorption, suggesting high optical

quality of the deposited films. The layer thickness values are shown in Table 3-3.

Table 3-3: Ellipsometry modeling results
Sample ID Si02 (50%HfO2)r?1 Hf02r?] Hf02+40%voidr? ]
Hf 11 25.0 36.8 9.1
Hfl2 31.7 34.9 9.4
Hfl3 16.2 39.5 16.2
Hf 14 24.4 36.3 7.0
Hfl5 22.9 36.1 2.4
Hfl6 25.6 28.1 8.1
Hf 17 25.0 42.9 3.3
Hfl8 31.6 32.2 5.1
Hfl9 33.2 29.8 4.52

Summary

XPS, XRR and SE investigations were used to determine the structure of the HfO2

thin films deposited on Si. It was found that samples deposited with substrate

temperatures of 600 'C exhibit small randomly oriented crystals but samples deposited up

to 400 'C remained amorphous. The interfacial layer was shown to consist of Si02

physically mixed with the deposited oxide which always formed, even when the

deposition temperature was only 200 'C and at low oxygen pressures. The thickness of

this layer is around 1-4 nm, depending on the deposition conditions used. Its density is

greater than that of pure bulk SiO2 resulting from a physical mixture of silicon suboxides

with the grown oxide layer.

By inspecting the results in Table 3-4, one can clearly note that there is a good

agreement between the layer thickness values obtained using these three methods. The

higher the deposition temperature and/or oxygen pressure used during deposition, the

thicker the IL. Also, the SE and XRR results support the XPS indication that the IL

contains a physical mixture between the SiO2 layer and the deposited HfO2 oxide, similar









Table 3-4: Comparison of the thickness values (in A) for HfO2 thin films deposited on Si
Sample Temp. Pressure XRR Ellipsometry XPS
ID# 0C Torr IL? OL? CL? IL ? OL? CL? IL ?
Hfll 1x10-4 24.6 41.3 7.8 25.0 36.8 9.1 19.4
Hfl2 200 1x10-3 31.8 34.3 9.2 31.7 34.9 9.4 27.5
Hfl3 1x10-2 16.3 31.1 7.9 16.2 39.5 8.1 16.4
Hfl4 Ix10-4 24.4 37.7 6.7 24.4 36.3 7.0 19.5
Hfl5 400 1x10-3 22.9 37.9 1.9 22.9 36.1 2.4 21.3
Hfl6 1x10-2 25.7 30.9 7.9 25.6 28.1 8.1 23.5
Hfl7 Ix10-4 25.0 39.3 3.1 25.0 42.9 3.3 30.5
Hfl8 600 1x10-3 31.3 39.0 6.8 31.6 32.2 5.1 31.0
Hfl9 Ix10-2 33.5 34.4 5.0 33.2 29.8 4.5 33.8


to our previous findings done by our group for Y20360 and Zr0261 films deposited on Si.

By understanding the nature of the interfacial layer and its formation, better control of the

layer can be realized to tailor the properties of the gate stack in future samples.














CHAPTER 4
EFFECT OF UV-ASSISTED OXIDATION AND NITRICATION OF HAFNIUM
METAL FILMS

Hafnium oxide has been shown to be a promising candidate for an alternative high-

k dielectric due to its high dielectric constant, large bandgap, and stability on silicon. 50-52

However processing temperatures required for standard deposition techniques are too

high and exceed the crystallization temperature for the films thus resulting in current

leakage and diffusion paths at the grain boundaries. Work has been done using

ultraviolet-assisted oxidation of zirconium metal films deposited using a radio frequency

direct current sputtering method which was shown to produce higher quality films at

lower processing temperatures.45-49 Furthermore, nitridation of various metal oxide films

has been used to retard the growth of interfacial layers by acting as a diffusion barrier to

oxygen.64-68 This portion of the work extends these investigations to the hafnium metal

system in an attempt to understand the effect of UV oxidation and the incorporation of

nitrogen into the films on their electrical and structural properties under lower processing

temperatures.

Experiment

Initially, it was believed a thin SiNy layer grown at the interface using UV

radiation would serve as a superior base for the deposition of the films in this portion of

the work. The silicon wafers were cut into 2.5 x 2.5 cm squares and mounted on the

resistive heater and placed into the vacuum chamber. After pumping to Ix106 Torr, the

chamber was backfilled with 500 Torr of ammonia. The substrate was then heated to 650







57


'C and irradiated with the UV lamp array for 30 minutes. This formed a 20 A layer of

SixNy verified by XPS and ellipsometry. Capacitors were formed out of the films and

capacitance/voltage measurements were taken. Figure 4-1 is the CV response of one of

the nitrided silicon films that was tested. As can be seen from the plot, the response is

not sharp but instead is elongated over 2 volts. This is a result of extreme amounts of

interface traps created at the silicon/silicon nitride interface. Furthermore, at higher

negative bias the capacitance is not constant but instead starts to drop off. This is

suggestive of a poor quality film which is leaky. From this result in order to eliminate the

charge trapping defects inherent when using silicon nitride, it was decided to deposit the

hafnium films onto bare silicon.


-4 -3


-1 0

Bias :.' ".:. -- *".'}


Figure 4-1: CV response of an UV grown SkNy film.


.- 1,,







1





l~E- ii




0 rIlITE -1~1 ,









A 248 nm KrF excimer laser system in a typical pulsed laser deposition (PLD)

setup was used to ablate a hafnium metal target 99.95% pure from SCI Engineered

Materials. The deposition parameters used were laser fluence of 3.0 J/cm2 at a 5 Hz pulse

rate with a total pulse count of 1500 for metal film thickness of 35 A and oxidized film

thicknesses of 45-50 A. The films were deposited at 350 'C on p- (100) Si substrates

from MEMC (9-18 Q cm) that were cleaned using an SC1 solution. This solution

contains 80 vol% distilled water, 16 vol% hydrogen peroxide diluted to 30 vol% in

distilled H20, and 4 vol% ammonium hydroxide. The samples were then dipped in a 1%

HF solution for ten minutes to remove the native oxide. Depositions were performed

either under residual vacuum at Ix106 Torr or under Ix102 Torr of ammonia. Post

deposition anneals were carried out in-situ in 300 Torr of 02 at 400 oC both with and

without a Hg lamp array for 30 minutes. Each lamp emits 254 and 185 nm wavelength

radiation at a total power density of 200 mW/cm2. The power density from the 185 nm

wavelength portion of the emission spectra is 10-15 mW/cm2. Previous experiments

determined that the use of UV during deposition in ammonia had no effect on the films at

the deposition temperature of 300 'C. At 300 'C even with the activation of the ammonia

by the removal of one or two hydrogen atoms due to UV interaction, there is not

sufficient energy to promote bonding with the hafnium metal being deposited or with the

silicon substrate. Therefore all depositions were conducted with the UV lamps off. A

summary of the sample matrix is given in Table 4-1. This data set was repeated 3 times

to study repeatability and to estimate error.

The surface morphology was studied using atomic force microscopy (AFM, Digital

Instruments Nanoscope III) both in contact and tapping modes. Crystallinity









Table 4-1: Growth and postdeposition conditions for UV oxidation of Hf metal ultra-thin
films.
Sample ID # Deposition Pressure Torr Anneal Conditions
H2 Ix10-6 UV
H3 Ix10-6 No UV
H4 Ix102 Ammonia UV
H5 Ix102 Ammonia No UV


comparisons and phase identification was done using grazing incidence x-ray diffraction

(GIXD, Panalytical X'Pert system). The chemical bonding and composition of the films

was investigated by x-ray photoelectron spectroscopy (XPS, Perkin Elmer 5100) using

Mg K, radiation (1253.6 eV). A calibration study using a 50 A Si02 film on Si standard

was also carried out with the XPS system to fit the results of the angle-resolved

measurements. High resolution spectra of the Si 2p region from the standard film were

acquired at takeoff angles from 150 to 900 and the area ratios of the Si and Si+ peak were

compared to obtain the calibration constant for our system. Finally MOS capacitors were

fabricated by sputter deposition of 1000 A thick platinum electrodes using a shadow

mask having circular dot patterns of various sizes. Capacitance voltage and current

voltage measurements were taken using a Keithley Instruments Inc., 590 capacitance

meter, KI236 source measurement unit (SMU), and the Kiethley Instruments Inc. Win-82

measurement system. Confirmation of results and further analysis was done at the

University of Texas-Austin. Cross-sectional transmission electron microscopy (XTEM,

JEOL 2010) was used to verify thickness and layer uniformity after electrical

measurements were concluded for two of the samples.

An additional set of four samples was deposited using substrate temperatures of

650 'C. At this temperature, it was believed that a more significant amount of nitrogen

could be incorporated into the hafnium metal thin film. Previous work done by N.









Bassim et al. showed that UV radiation from our lamp array promotes the activation of

ammonia into reactive species which then form a nitride layer on the surface of silicon

substrates. Significant nitride layer growth occurred at substrate temperatures of 650 'C.

Nitrogen does not readily react with metals and requires large activation energies. Thus

for this set of experiments substrate temperatures of 650 'C were used to afford greater

thermal energy for the reaction. Furthermore effects of UV during deposition and the

oxidation annealing steps were compared. Finally, the electrical properties of the high

temperature deposited samples were compared to the first set of samples deposited at 350

oC.

Results and Discussion for Low Temperature Deposited Samples

Surface Morphology

Atomic force microscopy was first conducted using contact mode for all four

samples. An area of 1 |jm x 1 |jm was scanned using a tip capable of near atomic

resolution on relatively smooth samples when properly calibrated. The 2D scans

obtained are shown in Figures 4-2 and 4-3. Root mean square roughness was calculated

for each sample based on the scans shown. These results are given in Table 4-2. The UV

oxidized samples, H2 and H4, exhibit greater Z-ranges and higher RMS roughness

values. There also is a slight but noticeable decrease in roughness with the addition of

nitrogen.

The quality of the scans obtained using contact mode was poor, possibly due to

washout from tip effects, so samples were reanalyzed using tapping mode, which affords

higher resolution as the tip does not make contact with the sample but instead rides above

the sample at a predetermined distance based on van der Waals forces and then is









Table 4-2: Z range and RMS roughness for oxidized Hf metal films.
Sample ID Z range A RMS Roughness A
H2 58.7 3.7
H3 24.6 2.3
H4 33.6 2.5
H5 26.5 2.0


cyclically flexed toward the sample during the measurement. The 3D scans taken from

samples H2 and H3 are plotted in Figure 4-4.

From the tapping mode results, the surface of the UV- irradiated samples again

showed a greater degree of roughness. The UV-irradiated sample exhibited a z-range of

7.8 nm and a RMS value of 6.4 A while the non-irradiated sample had a z-range of 1.7

nm and a RMS value of 1.2 A. It is believed that the UV-radiation promotes

crystallization of the films. This effect has been seen before with BST films that our

group has previously reported.69 The increase in roughness is due to crystallization

resulting from the UV radiation acting as a non-thermal energy source thus affording

sufficient energy for promoting nucleation and growth of randomly oriented crystals at

temperatures below the crystallization temperature. The 185 nm radiation is absorbed by

the oxygen within 5 cm of the lamps based on the measured falloff of the intensity done

at atmosphere with a detector head specifically designed to measure 185 nm radiation.

Each sample is 8.5 cm from the nearest lamps, so the crystallization enhancement must

be from the 254 nm radiation (4.88 eV).

X-Ray Analysis

Upon examination of the x-ray data collected, it was shown that the UV-radiation

did in fact promote crystallization in the films. GIXD spectra taken at Q=1o, plotted in

Figure 4-5, demonstrate the emergence of a monoclinic HfO2 phase (JCPDS 34-0104)










-1.00




0.75




0.50





0.25



a)

0 0.25 0.50 0.75 1.00
-1.00




-0.75




0.50





0.25




b)

0 0.25 0.50 0.75 1.00


Figure 4-2: 2D AFM scans of samples a) H2 and b) H3 taken using contact mode. The
scale is in |tm.










-1.00





0.75





0.50





0.25



a)
-0
0 0.25 0.50 0.75 1.00

-1.00





0.75





0.50





0.25



b)

0 0.25 0.50 0.75 1.00

Figure 4-3: 2D AFM scans of samples a) H4 and b) H5 taken using contact mode. The
scale is in atm.









with the UV-assisted oxidation. The radiation not only acts to break the oxygen into

more reactive species, but also acts as a non-thermal energy source providing the

activation energy for adatoms diffusion, crystal nucleation, and growth. Similar

emergence of the monoclinic phase of HfO 2 was noted in the nitrided samples with no

noticeable change in the lattice parameters.

X-Ray Reflectivity

As deposited metal films were analyzed using XRR in the Panalytic X'Pert system.

A typical scan and fit is shown in Figure 4-6. The films were all 403 A with a density

of 12.9 g/cm3 and a mixed interface of ~82 A having a density of 6.3 g/cm3. The

density of the metal layer suggests that 1-3% porosity was present as the bulk density of

Hf metal is 13.3 g/cm3.

XPS Analysis

XPS was used at take off angles of 450 and 900 for all of the scans in order to

compare changes in composition with depth in the film and to better understand the

nature of the silicon/film interface. Prediction of SiO2 layer thickness and uniformity can

be obtained using the previously mentioned Equation 3-1:

do = o,, sin a In[ I, /(s(,) + 1] (3-1)

where koxy is the photoelectron effective attenuation length in the oxide film, a is the

photoelectron take-off angle, oxy is the peak area for the Si-O bonding, Isi is the peak area

for the signal from the substrate, and 3 is the calibration constant for the system.63 Our

system was calibrated using the high resolution Si 2p scans, shown in Figure 4-7, taken at

angles from 15-900. From these results, values of 3 = 0.8 and k = 36 A were obtained

and d/k was calculated to be 1.400.05 for angles 45-900.


















Tapping AFM

: :" ',


0.8 1
JM T

NanoScope Tapping AFN

U. rSI *I
S0i*i. -I ..


0.8
JM


Figure 4-4: 3D AFM micrographs of both a) UV-oxidized and b) non-irradiated hafnium

films.


NanoScope



Si. .... i. .., ,i
















1-I 1 -------




120

S100









20





20 (dlS.)



Figure 4-5: GIXD spectra ( Q= 1) of Hf films deposited on Si and oxidized
with and without UV.





1E+7

1E+6

1E+5

1E+4

1E+3

1E+2

1E+1

1E+O
0.126 0.363 0.600 0.837 1.074 1.311 1.548 1.785 2.022 2.259 2.496
Theta-2Theta Omega


Figure 4-6: XRR plot and fit for as deposited Hf metal film.









The Si 2p scans plotted in Figure 4-8 were fit with two 90% Gaussian curves

representing the Si-Si and Si-O bonding. The areas were used to calculate the d/k values

for both the 45 and 900 scans. The results of these calculations are given in Table 4-3.

For all but one of the samples the d/k values were nearly identical. Therefore the

interfacial layer is primarily Si02/SiONx with very little degree of mixing occurring

between the silicon dioxide and hafnium oxide layers. Sample H3 (no nitrogen, no UV)

was the only sample with a mixed interface. The UV radiation is believed to promote the

separation of the silicon dioxide and hafnium oxide into two distinct layers, while the

nitrogen prevents the initial mixing normally encountered with PLD deposition.

Table 4-3: d/X values for samples H2-H5
Sample ID Angle Si-Si Area Si-O Area d/k
H2 450 400 1370 1.176
900 986 1572 1.096
H3 450 586 833 0.722
900 817 1532 1.207
H4 450 419 1239 1.094
900 1207 1660 1.000
H5 450 3506 2042 0.547
900 1676 1865 0.616


The Si 2p scans also show a larger percentage of Si-O peak area in the UV-

irradiated samples and a peak shift of 0.5 eV to higher binding energies approaching the

reported value of 103.6 eV for silcon bonded to oxygen in a pure thick Si02 layer.60

Therefore, the UV-radiation is promoting oxidation of the interfacial silicon. The O Is

data for the same samples, plotted in Figure 4-9, shows a similar trend whereby the

irradiated samples have a higher energy shoulder located at binding energies of 535 eV

representing greater concentrations of Si-O bonding. Peak fit analysis using two peaks,

one at 531.5 eV representing the oxygen atoms bonded to silicon and the other peak at










533 eV indicative of oxygen bonded with Hf, indicates an increase of 30% Si-O bonding

in the UV irradiated samples.


40000 -

35000 -

30000 -

25000 -
-2-

S20000 -
o
15000 -

10000 -

5000-

0-


105


99 97


Energy (eV)

Figure 4-7: Angle-resolved XPS Si 2p spectra for system calibration.


13000

12000 no UV-900 Si2p

11000 -

10000
no UV-45'
S9000

8000 UV-900
UV-450

6000
106 105 104 103 102 101 100 99 98


Energy(eV)
spectra for H2 and H3


Si2p
Si-O




si-Si


Figure 4-8: Angle-resolved XPS Si 2p


(UV and no-UV films).










96000

86000 Ols
76000 -

66000 -
56000 no UV-90'

46000 noUV-45

36000 UV-900
26000 UV-45
16000
537 535 533 531 529
Energy (eV)
Figure 4-9: 0 Is spectra for H2 and H3 (UV and no-UV) films at 45 and 900.

According to XPS chemical composition analysis, depositing in ammonia only

allowed for less than 2 atomic% incorporation of nitrogen into the films. Further trials

using UV-radiation during deposition were tested to attempt to increase the concentration

of nitrogen, but yielded the same result. It can be concluded that the energy of the

radiation is not sufficient to promote highly significant amounts of reaction of the

ammonia with the depositing hafnium metal or with the silicon substrate at the substrate

temperature of 350 'C.

Even with just 2% incorporation of nitrogen into the films, the interfacial layer of

the nitrided films have a larger ratio of Si-Si peak area suggesting less interfacial growth

as shown in Figure 4-10. The Si-O peak also undergoes a shift of 1 eV to lower binding

energies due to the change the bonding environment of the silicon atoms at the interface.

The shift could result from the formation of a thin SiN4 layer. Binding energies for Si-N

type species lie within the intermediate range between the Si-Si and Si-O peaks. For

fully nitrided silicon this peak is located at 101.8 eV.












Si 2p
"I


-\\ltl llltogIenl
I


without nitrogen

I0 JIIIl I. I1| 1 IIIl[ I.



Figure 4-10: Si 2p spectra for UV-oxidized films H4 and H2 (with and without nitrogen).

In order to confirm the nature of the nitrogen in the film, high resolution XPS scans

were taken for the N Is peak range. Due to an overlap of the N Is peaks and a higher

energy satellite peak of the Hf 4p3 peak, an oxidized Hf metal film was used as a

standard to subtract out the background resulting from the hafnium peak. The result of

the as-deposited Hf metal film is shown in Figure 4-11. The majority of the peak occurs

at a binding energy of 397.5 eV which is near the reported value of 396.3 eV for Hf-N.

Granted the substrate temperature of 350 'C is not sufficient to promote a significant

degree of bonding. Upon annealing the peak broadens out and shifts to higher binding

energies, as shown in Figure 4-12. The broad peak is indicative of multiple bonding

environs. The shift to higher binding energies during the oxidation anneals results from

the nitrogen moving to the silicon interface. Si-N bonds have been reported to have

values in the range of 398 to 399 eV. Furthermore upon annealing, the intensity of the

signal diminishes. It is believed that a majority of the nitrogen within the metal film is

replaced by oxygen when annealed. Comparing the UV to the non-UV annealed sample









(i.e. H4 and H5 respectively) shows a similar peak profile but a loss of intensity in the

nitrogen signal. The UV radiation promotes the replacement of the nitrogen within the

film and at the interface.


414 411


408 405 402 399

Binding Energy (eV)


396 393


Figure 4-11: N is peak of as deposited nitrided Hf metal film.






eC H5


UA AAN^AI


VWVrW Si-N Hf-N W


409 406 403 400 397 394 391
Binding Energy (eV)

Figure 4-12: N Is peak after PDA for samples H4 and H5.


N Is


Si-N Hf-N


390


NIs









Electrical Properties Analysis

Upon completion of the chemical and structural analysis, the sample surfaces were

washed with acetone then methanol and blow dried with dry nitrogen. Then 1000 A

platinum electrodes were sputtered on the surface using a radio frequency (RF) sputtering

system. The shadow mask used for this step had repeated sets of circular dots of various

sizes. The size dot used for this portion of the experiments had an area of 4.66x10-4 cm2.

A forming gas anneal was then carried out at 450 'C for 30 minutes in a flowing gas

environment using a gas mixture of 4% H2 and 96% N2. The forming gas anneal step is

important because the hydrogen passivates the defects between the surface of the metal

and the film. The backside of the samples were then abraded and swabbed with a 1 %

HF to remove any native oxide. The samples were then mounted on glass microscope

slides using a silver paste which was then cured at 150 oC on a hotplate to remove the

organic. The silver forms an ohmic contact with the p-type silicon for proper electrical

data collection.

Capacitance-Voltage Results

Capacitance-voltage measurements were taken for a series of seven dots on each

sample for both the 100 kHz and 1 MHz input frequencies to compare the frequency

dependence on the capacitance response. Furthermore, scans were taken using both the

parallel and series models. Figure 4-13 compares these results for sample H4, the UV-

oxidized sample deposited in ammonia which overall had the best electrical

characteristics of the samples. There is a large dependence of the capacitance on the

frequency used for the measurement. This sample exhibited a 40% decrease in

capacitance between the parallel models when going from 100 kHz to 1 MHz. The

change in the response is mainly due to the choice of substrates. The low doping level of









~2x1014 /cm3 provides very few charge carriers. Thus when in the accumulation region

where a negative bias is being applied to the gate, the depletion region in the channel

beneath the gate becomes large. As a result, the RC time constant for the carriers to

move to respond to the frequency at which the voltage is being applied becomes

significant when 1 MHz measurements are carried out. Thus the 1 MHz capacitance

response is less than that of the 100 kHz. Ultimately films with no frequency dispersion

are desired and the films should be tested on standard p-type wafers instead of the p- ones

that were used. This plot also demonstrates the need to choose the proper model for the

sample type being studied. For ultra thin film oxides (<100 A), like the samples in this

work, the parallel model is best suited due to increased leakage resulting in a conductance

term which must be taken into account.

The high frequency (1 MHz) parallel model scans were used for comparison of

each of the samples. Figure 4-14 shows a plot of these results. From the graph it can be

seen that the small incorporation of nitrogen into the films by depositing in ammonia

greatly increases the capacitance of the film. This is attributed to limited interfacial layer

formation during oxidation and a resulting high dielectric constant for the interface.

Furthermore the use of UV-oxidation during post deposition annealing further improves

the capacitance of the stack by increasing the quality of the hafnium oxide layer. The

threshold voltage VT for the samples also underwent a shift from --1.5 V for sample H3,

the non-irradiated, non-nitrided sample, to 0 V for sample H4. The curve response shape

is also improved by nitrogen and UV-treatment suggesting minimization of interfacial

and bulk defects such as oxygen vacancies. Thus combination of both nitrogen

incorporation and UV-irradiation results in higher quality electrical characteristics.









From the capacitance response data, the oxide gate stack dielectric constant was

calculated using equation 4-1. Thickness values used for the calculations for the films

were taken from the XRR and VASE modeling results. The equivalent oxide thickness

was calculated using equation 4-2.

Cxt
E= X (4-1)
Ax ,

3.9x E xA
EOT = o (4-2)
C

s = dielectric constant for the film
C = capacitance
A area of the MOS capacitor
so = permittivity of free space
EOT = equivalent oxide thickness

Compilation from these calculations is presented in Table 4-4. The use of UV-

irrdation and nitrogen incorporation into the films has been shown to produce film stacks

with EOT values of less than 10 A.


Table 4-4: Dielectric constants and EOT for samples H2-H4.
Sample ID k EOT A
H2 10.6 18.8
H3 9.4 21.3
H4 21.0 9.1
H5 13.8 14.5


Repeatability

The sample set was repeated twice using the same experimental setup to confirm

the results. Errors were estimated using the root mean square error method. From the

results it is shown that the data is highly repeatable and yields nearly identical values

between each set.














3.. 7 .Tlll
2.50E-10 -100 1: Series




1 -0 --
?.ilO I 1I H -7 ..-J .I


1.50E-11 -






5.OOE-l00

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2



Figure 4-13: Comparison of high and low frequency capacitance responses for sample
H4.


2.0E+03
H2
1N 7H3
1.5E+03 -
s- H4
S-H5
1.OE+03

A A
O 5.OE+02



O.OE+OO


-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5
Voltage (V)









Figure 4-14: High frequency capacitance voltage responses for samples H2-H4.

Table 4-5: Repeatability study results for k and EOT of oxidized hafnium metal films.
Set 1 Set 2 Set 3
Sample Avg. k Avg. EOT
ID k EOTA k EOT A k EOT A JA
H2 10.6 18.8 10.5 19.0 10.8 18.6 10.60.1 18.80.1
H3 9.4 21.3 9.5 21.2 9.6 21.0 9.50.1 21.20.1
H4 21.0 9.1 20.9 9.3 21.2 8.9 21.00.1 9.1+0.1
H5 13.8 14.5 13.6 14.7 13.5 15.0 13.60.1 14.70.2


Interfacial Trap Density Measurements

The samples were sent to the University of Texas-Austin for further analysis.

Using a similar system setup to ours, capacitance voltage measurements were taken at 1

MHz in a parallel model mode. These results were then fit by the Terman method,

discussed further in Appendix A, using an in-house Mathcad program which simulates

the CV data based on the metals used for contacts and the known film thickness and

contact area. The plots generated from the fits of the models give the quasi-static

representation of the MOS stack. By comparing the fit to a curve generated based on an

ideal film with no defects, an estimate of interfacial trap defect density, D3t, can be

yielded. The results of the defect density calculations are given in Table 4-6. Figure 4-

15 plots the data and the fits obtained from the three samples which could be modeled.

Sample H3 was too poor of quality for any information to be properly fit by this method.

All the samples had average defect densities ranging in order of magnitude from 1011-

1012 /cm2. The UV-oxidized samples H4 and H5 had the highest defect densities. In

order for an alternative high-k oxide to be used it must be able to compete with the low

defect densities of 1010 I/cm2 obtained using silicon dioxide.









Table 4-6: Defect density results for samples analyzed at UT-Austin.
Sample ID Dit (1/cm2)
H2 3.07E+11
H4 2.18E+12
H5 1.73E+12
2500 -















-1.5 -I -I.S I .5 1 1.5 2



Figure 4-15: CV response and model fit data for samples H2, H4, and H115.

Current-Voltage Results

The current-voltage (IV) characteristics of the films were obtained by scanning

MOS capacitors using a voltage range of -2 to 2 volts with the Keithly Win-82

measurement system. Figure 4-16 shows the IV results for the oxidized hafnium metal

films. All samples exhibited leakage current densities of less than 5x10-4 A/cm2 at

voltages of -1.2 V. This is the standard operating voltage that industry uses to compare

leakage. For a film to be used it must have leakages less than 1x10-2 A/cm2 at this

voltage to stay within the power consumption guidelines. Sample H4, which had the

highest capacitance, exhibited the lowest leakage for both the negative and positive bias

regions. The use of nitrogen incorporation and UV-assisted oxidation together provide a






78


synergistic effect creating higher quality films with less defects which promote leakage

paths even though this film had the highest degree of crystallinity.

1.E-02























Figure 4-16: Current voltage results from oxidized hafnium metal films.

TEM Analysis

Cross sectional TEM samples were made from samples H4 and H5 to observe the

film uniformity, crystallinity, and interface quality. Figures 4-17 and 4-18 depict a

portion of samples H4 and H3 respectively. The thickness of the film and the interfacial

thicknesses were measured from the micrograph. Sample H4 clearly shows signs of

polycrystalline growth, evidenced by grains of roughly the size of the thickness of the

film. EDS line scans were attempted but the drift was too great to get accurate readings

across the thickness of the samples. The degree of sample tilt required (~10 0) to acquire

the measurement further hinders accurate elemental analysis. Scans taken at specific

points in the interface region for both samples did show mostly Si and 0 with a trace
points in the interface region for both samples did show mostly Si and 0 with a trace









amount of Hf. From both the EDS and XPS results previously, the interface for all the

samples is a distinct uniform SiO2 layer (nitrogen is unable to be detected using this

analysis technique). Ultraviolet radiation was shown to increase the growth of the

interfacial layer. However by incorporating nitrogen into the film during deposition, the

growth of the layer is hindered.

















Pr


1.23 mnil ': '







.. . . .

Figure 4-17: XTEM of the UV annealed nitrided sample H4.

Proposed Model

Based on the data from this set of four samples, UV radiation during the oxidation

anneal promotes crystallization which, in and of itself, is a detrimental quality for the gate

oxide. A polycrystalline film has grain boundaries which act as leakage paths for current









through the film and also serve as diffusion pathways for oxygen and other contaminants.

Furthermore, UV radiation promotes oxidation of the interface, thus parasitically

reducing the benefits of using a high-k dielectric. The benefits of UV radiation, however,

are improved oxygenation of the film and reduction of physiaborbed and physically

trapped oxygen. Lower temperatures can be used to fully oxidize the film and eliminate

sub oxygen species. Thus UV radiation promotes the formation of higher quality oxide

films.

In order to counter some of the detrimental effects of UV-assisted processing,

nitrogen was incorporated into the film during deposition. The nitrogen was shown to

reduce the formation of the interfacial silicon dioxide layer by both acting as a diffusion

barrier to the oxygen and by the oxygen preferentially replacing the nitrogen within the

film structure. The XPS results suggest a thin silicon nitride layer was formed at the

interface. Nitride barrier layers have been used as oxygen diffusion barrier layers for

years in the semi-conductor industry. An ultra-thin nitride layer will slow the reaction of

the highly active oxygen species created by UV radiation. Furthermore, hafnium has a

greater affinity for oxygen. The thermodynamics behind the reactions favor the

formation of hafnium oxide over hafnium nitride. Therefore with an overpressure of

oxygen species and no nitrogen during the oxidation anneal, the oxygen replaces the

nitrogen within the structure and also drives out the physically trapped nitrogen.

Ultimately nitrogen incorporation of the films controls the interface quality of the films,

thus countering the detrimental effect of UV-assisted processing. Figure 4-19 shows a

schematic of the effects of UV and nitrogen on the films.





81

f J .: ..............



...... .... .. .. ... ..



... .....
Figure .. ..... ... .... ......... .lm H 3
---- .. ...... III I ....... ... ...... o n
T h m ti at on be i ..... .. .. ... .frs a te ptto in re se th
d e g r e e o f n i t i d a t i n o f t e.. .. .............. .........su s t r a t
tempeaturedurin procssing Thencomprison...................n hig
......... ... ...... .......ae t jdgeth efec o UVonth
.. .. ad il trctr .. .. ... ... .. ... ..... .. .. ... .. ... .. .
these simple t comp........... ..... ..... .... .....ngmethod













P L1 Nlo liV

_Higih q|iilii Hf) Poo ;:,| 1 .:II ir 1-
Pouu qr;Jil\ lL --------------- ft ,


icm a ed EL |

lII)Iplote.:l lUal-k Rc(inced
fL hliickm.

HisJi (liri li. Hf. Fool .(iialir H10
Hili (liallrn IL Hiuli (pialih L
Sio) S(. Si 'N NH3






Figure 4-19: Model of the effect of UV and nitrogen.

XPS

Elemental analysis carried out by comparing the area of the scans for the 0 Is, N

is, Hf 4f, and Si 2p regions and then taking into account for the sensitivity factors yielded

nitrogen percentages of 4-5 atomic % for all four samples. Therefore by depositing using

substrate temperatures of 650 'C an additional 1-2% of nitrogen can be incorporated into

the films. Figure 4-20 shows the comparison of the O is peaks for these samples. Figure

4-21 compares the Hf 4f peaks for these samples. These films were thicker than previous

samples, ranging from 80-120 A, so detailed Si 2p peaks were difficult to obtain due to

the weak signal resulting from the large escape depth that the electrons from the Si atoms

near the interface have to travel.

The O is peaks for the UV deposited samples (HN1 and HN4) exhibit a higher

energy shoulder suggesting -10-15% more Si-O type bonding environs. This is believed

to be a result of initial crystallization of the films during deposition resulting in a higher









percentage of grain boundaries for the diffusion of oxygen to the interface. Thus upon

annealing the oxygen can readily move to the interface to react with the silicon.

The Hf 4f peaks were all identical in shape, position, and ratio of the 4f 5/2 and 4f

7/2 peaks. Therefore UV does not affect the main bonding in the bulk of the hafnium

oxide layer.


55000
50000 i- oHN1 UV/UV
45000 HN2 N/UV
7 40000 A, HN3 N/N
35000 HN4 UV/N
S350000-


25000
20000 + ...........
15000 ;'"
10000
536 534 532 530 528 526

Binding Energy (eV)


Figure 4-20: 0 Is peaks for the high temperature deposited sample set.

X-Ray Analysis

Glancing incidence angle measurements were made at 2=0.5 for all the samples

and are plotted in Figure 4-22. All of the samples exhibited a high degree of

crystallization. The FWHM for the (111) peak located at -280 of the UV deposited and

UV annealed sample (HN1) was 0.10 narrower than the other samples suggesting a

slightly greater degree of crystallinity possibly due to crystal growth. This result follows









the previous observance of the promotion of crystallization resulting from using UV

during processing.




35000
-0- HN1 UV/UV
30000 -B- HN2 NUV
-25000 HN3 N/N
25000
HN4 UV/N
S20000

I 15000

10000

5000

0
21 19 17 15
Binding Engergy (eV)


Figure 4-21: Hf 4f peaks for the high temperature deposited sample set.

X-ray reflectivity scans were also taken for all of the samples and are plotted along

with the models in Figure 4-23. The XRR scans show that UV during deposition

increases the growth of the interface layer as was predicted from the XPS O is scans.

Furthermore, the film processed without UV for both the deposition and anneal has the

least amount of interfacial oxide growth. The interface density is larger than SiO2 for all

the samples suggesting either there is mixing of Si02 and Hf02. The higher density also

could result from nitridation of the interface as the density of silicon nitride is -4-5

g/cm3. Low roughness values for the interface and the film layer suggest abrupt

interfaces and relatively smooth uniform films.










180 -
160 4- HN1 UV/UV
140 HN2 N/UV
120 HN3 N/N
HN4 UV/N
S100
80
a 60-
40 -
20 -
0 -

25 35 45 55 65

2 Theta (deg.)


Figure 4-22: GIXD for the high temperature deposited samples taken at Q = 0.50.

Electrical Characterization Results

Capacitance/voltage and current/voltage measurements were taken for the high

temperature deposited sample in a similar manner to the previous samples. The results of

the capacitance/voltage measurements are shown in Figure 4-24. The data set HN1 was

offset to avoid confusing overlap with HN3. From the results it is shown that UV during

the deposition is detrimental to the electrical properties of the film especially when the

sample is not UV annealed during the oxidation step. Sample HN4 (deposited with UV

and annealed without UV) exhibits similar characteristics to the SiNy samples that were

first produced. The response has a large stair step due to electrical traps and then falls off

at large negative bias. Similarly, sample HN3 (deposited and annealed without UV)

exhibits a stair step in the depletion region. Upon UV oxidation annealing, the defects






































IT- L\r I hiL kn ROis RnIhnies- Densi

Silicon 3.0 \ 2.33 g/cm
\ iO 8.5 A 5.2 A 3.01oLni
\ Hf0 724 A. 4.5 A 9.86 g/cm










HN2 N'UV


T.a\ er I hicknesrl i Rl1ii line Ill Dc'[it' i


S.ikon

1 C-i


3.1 .
84.1 A


3.0 A
6.9 \
3.5 A


2.33g/cm3
3.37 %Iti
9 i cni


1E I
1E-2


HN3 N/N


IALT er I hicknIL,-S RuOIL'IHILns,- DInsIi1\


- 3.0 A 2.33 Ai 111
i0 17.4 N 5.2 2.600 L rm
I HO 129.4 A 3.8 A 9.86 .' .i


HN4 UV/N


Figure 4-23: XRR plots and models for the high temperature deposited samples.


0 10 1 0391 0680 0970 1259 1549 1 38 212 2417 2707 2996
Omnsgo


1Clb


1C15


1L*b


1L*J


IE12


IEtl


3


3


0151 0436 0720 1005 12S9 1574 158 2143 2427 2712 2996

1Om-g7
1E 7


1E-4


-1


0151 0436 0720 1005 1289 1574 1858 2 143 2427 2712 2996

1E-7


0


1-.