Citation
Nitrogen Incorporated Hafnia Gate Dielectric Thin Film and Titanium Based Metal Gate Electrode for Dual Gate Application

Material Information

Title:
Nitrogen Incorporated Hafnia Gate Dielectric Thin Film and Titanium Based Metal Gate Electrode for Dual Gate Application
Creator:
Son, Seung-Young
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (165 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Materials Science and Engineering
Committee Chair:
Singh, Rajiv K.
Committee Members:
Norton, David P.
Pearton, Stephen J.
Gila, Brent P.
Ren, Fan
Graduation Date:
8/9/2008

Subjects

Subjects / Keywords:
Annealing ( jstor )
Dielectric materials ( jstor )
Electric potential ( jstor )
Electrodes ( jstor )
Electrons ( jstor )
Nitrogen ( jstor )
Oxidation ( jstor )
Oxides ( jstor )
Oxygen ( jstor )
Permittivity ( jstor )
Materials Science and Engineering -- Dissertations, Academic -- UF
dielectric, hafniumoxide, metalgate, mos, thermalstability, uniaxialstess, workfunction
Genre:
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Materials Science and Engineering thesis, Ph.D.

Notes

Abstract:
We evaluated the effect of nitridation temperature on interface layer (IL) quality of Hf-silicate gate dielectric has been reported. An increase in IL density and IL roughness was observed as the nitridation temperature was increased. Preferential interface reaction at dielectric-Si interface at higher temperatures was analyzed. The progressive increase in IL roughness finally led to degradation of breakdown voltage, shift in flat band voltage (0.54V) and deterioration of electron channel mobility by 20 % in samples nitrided at 1123 K. To reduce the interface degradation of thermal nitridation process, a low temperature process (623 K) for nitrogen incorporation in hafnia gate dielectric has been proposed. This method is based on post-deposition nitridation under ultraviolet light illuminated NH3 ambience. Uniformity of nitrogen distribution in the film was measured. Moreover, the amount of nitrogen incorporated by this process was comparable to that of high temperature thermal nitridation, maintaining low interface roughness (0.3 nm). An evaluation of Ti based gate metals has been reported. The effective metal work function were 4.27, 4.56 and 5.08 eV for Ti, TiN and TiB2, respectively. Regardless of gate electrodes, the conduction mechanism of the samples fitted with Poole-Frenkel model which is related to oxygen vacancies in the film. Ti gate electrode was found to be more favorable for NMOS device and TiB2 gate electrode can be used for PMOS. Further research for thermal stability of TiB2 gate electrodes was conducted. The extracted effective metal work function for TiB2 gate was about 5.08 eV. The work function showed almost identical values and sharp interface between metal and dielectric was confirmed after post deposition annealing by 1273 K. The work function lowering (4.91 eV) at 1373 K was caused by metal-dielectric intermixing and oxygen vacancy formation. TiB2 gate electrode was found to be suitable for use in PMOS device. Finally, Uniaxial?mechanical?strain altered gate leakage current and dielectric constant of MOS device are measured. Uniaxial stress is applied using four?point wafer bending along one one zero plane direction. The gate leakage current and dielectric constant are found to increase by up to 2 % under tensile and compressive stress direction. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2008.
Local:
Adviser: Singh, Rajiv K.
Statement of Responsibility:
by Seung-Young Son.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Son, Seung-Young. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Classification:
LD1780 2008 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

NITROGEN INCORPORATED HAFNIA GATE DIELECTRIC THIN FILM AND TITANIUM BASED METAL GATE ELECTRODE FO R DUAL GATE APPLICATION By SEUNG-YOUNG SON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008 1

PAGE 2

2008 Seung-Young Son 2

PAGE 3

To my mom, wife and my friends 3

PAGE 4

ACKNOWLEDGMENTS My sincere thanks go to my advisor, Dr. Rajiv K. Singh fo r having given me a chance to work in his Laser group and for supporting me academically and financially all through my graduate study. I truly appreciate his feedback in research, life, t echnical writing and for inculcating the qualities of persev erance and creativity in scientific research. I would like to thank Dr. Stephen Pearton for exposing me on having a professional outlook to life, for his research feedback. Thanks go to Dr. David Nort on and Dr. Fan Ren for their interest in my research and for serving on my dissertation co mmittee. I truly apprecia te his friendship and concern. Dr. Brent Gila deserves special thanks fo r his friendship, for his patience in listening to all my complaints about research and life and for his suggestions on thin film processing. My sincere thanks go to Jennifer Wrighton for patie ntly helping me again with the paperwork and travel reimbursements. I would like to express my special thanks to Korean students in UF MSE (Wantae, Junghun, Jaewon,Chanwoom and so on) fo r beautiful times with alcohol and golf. Thanks go to all past and current members of Singh group (far too many to mention) for their friendship and assistance with my research (especially Kumar, Taekon, Jaeseok, Sushant, Myunghwan and Sejin). Finally I wa nt to extend my thanks and gratitude to my mentors at Samsung, Korea Dr. Kang, Mr. Cho and Jung for e xposing me to the world of microprocessors by giving a chance to work under them filled with wonderful discussions. Of course, all of this would not have been possible ha d it not been for my wonderful family. I love them and I am eternally indebted to them for giving me a hea lthy upbringing and teaching me the values of life and the importance of education. De dicating my Ph.D. to them is just one of my life-long list of ways of expressing my love and gratitude to them. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ..........8LIST OF FIGURES.........................................................................................................................9ABSTRACT...................................................................................................................................13 CHAPTER 1 INTRODUCTION................................................................................................................. .15Fundamentals of Dielectric..................................................................................................... 15Basic Transistor......................................................................................................................16Metal Oxide Semiconductor Field Effect Transistor..............................................................17Metal Insulator Semiconductor Diode....................................................................................18Demand of Industry................................................................................................................21Advantageous Properties Hafnium Based Gate Dielectric.....................................................23Dielectric Constant and Band Offset...............................................................................23Thermodynamic Stability................................................................................................25Drawbacks of Highk Gate Dielectric.....................................................................................26Interface Quality..............................................................................................................26Low Crystallization Temperature....................................................................................27Channel Mobility Degradation........................................................................................28Incompatibility with Poly Si Gate...................................................................................292 LITERATURE REVIEW.......................................................................................................41Hafnium oxide and Hf-Silicate Dielectric..............................................................................41Bonding and Electrical Properties...................................................................................41HfO2 Gate Dielectrics......................................................................................................42Hf-Silicate Dielectrics.....................................................................................................44Nitrogen Incorporation in Hafnia Film............................................................................45Ultra-Violet (UV) Assisted Thin film Processing...........................................................46Metal Gate Process for CMOS Device...................................................................................49Candidate Metals for Metal Gate Application.................................................................49Fermi Energy Level Pinning...........................................................................................53Strain Engineering for High Channel Mobility MOSFET.....................................................54Band Modification Model...............................................................................................54Application Scheme for Future Device...........................................................................553 OUTLINE OF RESEARCH...................................................................................................624 EQUIPMENT SET-UP AND CHARACTERIZATION........................................................63 5

PAGE 6

Laser System...........................................................................................................................63Pulsed Laser Ablation via Excimer Lasers.............................................................................65Vacuum Chamber, Com ponent and Operation.......................................................................66Atomic Layer Deposition.......................................................................................................6 7Atomic Layer Deposition Process...................................................................................67Wafer Bending Experiment....................................................................................................69Material and Chemical Characterization Techniques.............................................................70X-ray Reflectivity............................................................................................................70X-ray Photoelectron Spectroscopy (XPS).......................................................................71High Resolution Transmission El ectron Microscopy (HR-TEM)...................................72Auger Electron Spectroscopy (AES)...............................................................................72Secondary Ion Mass Spectroscopy (SIMS).....................................................................73Electrical Characterization.................................................................................................... ..74Current-Voltage Measurements.......................................................................................75Capacitance-Voltage Measurements...............................................................................765 STUDY OF INTERFACE DEGRAD ATION OF HAFNIUM-SILICATE DIELECTRICS DURING THERMAL NITRIDATION PROCESS.....................................85Introduction................................................................................................................... ..........85Experimental Detail............................................................................................................ ....86Results and Discussion......................................................................................................... ..87Conclusions.............................................................................................................................926 HIGH EFFICIENCY NITROGEN INCORPORATION TECHNIQUE USING UV ASSISTED LOW TEMPERATURE PR OCESS FOR HAFNI A DIELECTRICS................99Introduction................................................................................................................... ..........99Experimental Detail............................................................................................................ ..100Results and Discussion......................................................................................................... 101Conclusion............................................................................................................................1047 UV ASSISTED LOW TEMPERATURE NI TRIDATION AND POST DEPOSITION OXIDATION TECHNIQUE FOR HfO2 GATE DIELECTRIC..........................................110Introduction................................................................................................................... ........110Experimental Detail............................................................................................................ ..111Results and Discussion......................................................................................................... 112UV Assisted Low Temperature Growth of Hafnia Film...............................................112UV Assisted Oxidation Resistance of Hafnia Film.......................................................115Conclusions...........................................................................................................................1178 AN EVALUATION OF COMPATIBILIT Y FOR TITANIM BASED METAL GATE ELECTRODE ON HAFNIUM-SILICATE DIELECTRICS FOR DUAL METAL GATE APPLICATIONS......................................................................................................124Introduction................................................................................................................... ........124Experimental Detail............................................................................................................ ..125 6

PAGE 7

Results and Discussions........................................................................................................ 126Conclusions...........................................................................................................................1299 AN EVALUATION OF THERMAL STABIL ITY OF TITANIUM BORIDE METAL GATE ON HF-SILICATE FOR P-CHANN EL METAL OXIDE SEMICONDUCTOR APPLICATION....................................................................................................................134Introduction................................................................................................................... ........134Experimental Detail............................................................................................................ ..135Results and Discussion......................................................................................................... 136Conclusions...........................................................................................................................13910 STRAIN INDUCED CHANGES IN GATE LEAKAGE CURRENT AND DIELECTRIC CONSTANT OF NITRIDED HF-SILICATE DIELECTRIC SILICON MOS CAPACITORS............................................................................................................144Introduction................................................................................................................... ........144Experimental Detail............................................................................................................ ..145Result and Discussion.......................................................................................................... .146Conclusion............................................................................................................................14711 CONCLUSIONS................................................................................................................. .153LIST OF REFERENCES.............................................................................................................155BIOGRAPHICAL SKETCH.......................................................................................................164 7

PAGE 8

LIST OF TABLES Table page 1-1 The 2003 ITRS technology roadmap for memory device.................................................231-2 Gibbs free energy of formation for po tential alternative high-k dielectrics......................265-1 Summary of x-ray reflec tivity data on roughness, density, and thickness of IL and HfSiNO films.....................................................................................................................936-1 Summary of x-ray reflec tivity data on roughness, density, and thickness of IL and Hf-silicate films.............................................................................................................. .107 8

PAGE 9

LIST OF FIGURES Figure page 1-1 Schematics of a typical MOSFET device..........................................................................301-2 Schematics of MIS device................................................................................................301-3 Energy band diagrams of ideal MIS diodes at V=0 (a) n-type and (b) p-type..................31 semiconductor................................................................................................................. ...311-4 Energy band diagrams for ideal MIS diode structure when the gate is biased. (a) Accumulation (b) Depleti on and (c) Inversion..................................................................321-5 C-V curve shift based on the gate bias due to the positive or negative charges (a) for p-type semiconductor.........................................................................................................331-6 Distortion or stretch out of the C-V cu rve due to the presence of interface trapped charges...............................................................................................................................331-7 Various electron tunneling modes. In films with thicknesses greater ~ 50 FN tunneling dominates and the mechanism shifts to direct tunneling in films thinner than 50 ............................................................................................................................341-8 Expected leakage current density and leak age current operation limit with respect to the equivalent oxide thickness (EOT)................................................................................351-9 Band gap and dielectric constant for gate oxide candidate materials................................361-10 Conduction and valence band offset of va rious gate oxide candidate materials...............371-11 Comparison of GIXD of hafnium oxide film deposited at 200 oC and 600 oC.................371-12 Various scattering centers of MOSFET.............................................................................381-13 Comparison of channel mobility of SiO2 and higk-k dielectric MOSFET.......................391-14 Defect formation at the poly-Si and highk dielectric interface is most likely the cause of the Fermi level pinning which causes high threshold voltages in MOSFET (M = Zr or Hf)................................................................................................................. ...402-1 Plot of work function versus atomic nu mber. Work functions are generally observed to increasing moving across a row of transition metal elements.......................................582-2 Process flow for the stacked dual metal gate integration. The difficulty in this approach is that two gate stacks of different heights need to be etched simultaneously. ............................................................................................................................................59 9

PAGE 10

2-3 Process flow for the metallic integra tion showing how two metals are formed by depositing a metal stack, selectively patte rning and etching the top metal, and annealing to alloy the bi-metal stack.................................................................................592-4 Process flow for the ion implantation and solid-state diffusion of nitrogen to form a nitrogen-rich (NMOS) and a nitroge n-deficient electrode (PMOS)..................................602-5 The six-fold degeneracy of electron, before and after biaxially tensile strain...................602-6 Schematic features of the various process strain: (a) silicon nitride capping layer to create a tensile channel (b) embedded Si Ge S/D process strain to create a compressive strain............................................................................................................. .614-1 Sketch of the pulsed lase r deposition (PLD) system.........................................................784-2 Photograph of an UV lamp array in th e pulsed laser deposition system. The lamps were located at a distance of ~ 5-7 cm from the substrate.................................................784-3 Atomic layer deposition of ZrO2........................................................................................794-4 ALD acceptable temperature window...............................................................................794-5 Schematic illustration of ALD HfSiOx deposition cycle.................................................804-6 A fixture to simulate uni axially-strained MOS devices. For a uniaxial stress, two pairs of cylindrical rods ar e used and a sample is inserted between the pairs...................814-7 Illustration of a uniaxial wafer bending ji g. (a) an unstressed sample (b) a stressed sample. (c) is the practical de flection and induced stress..................................................824-8 Typical X-ray reflectivity (XRR) spectru m from a thin film deposited on Si...................834-9 Typical X-ray photoelectron spectroscopy spectrum of graded hafnium silicate film deposited on silicon. The binding energy shif ts of Hf 4f photoelectron peak can be used to determine the nature of bonding present in the films. The different angles represent the photoelect ron take off angles or escape angles............................................834-10 Metal oxide semiconductor capacitance-vol tage curves (a) low frequency (b) high frequency and (c) deep depletion.......................................................................................845-1 The comparison of XRR patterns with re spect to the nitrid ation temperature..................935-2 Dependence of (a) Hf 4 f peak region and (b) N 1 s core level of XPS scan on nitridation temperature. The percentages in N 1 s deconvoluted spectra represent area fraction of the respective peaks..........................................................................................945-3 Electrical measurement results of Pt/Hf-silicate/Si MOS device fabricated on films with different nitridation treatments (a) Flat band voltage and (b) current density gate voltage characteristic..................................................................................................95 10

PAGE 11

5-4 Cumulative probability of breakdown volta ge for 7 nm thick Hf-silicate on Pt electrode with respect to the nitridation temperature. Lines are fits to a Weibull distribution (Eq. 5-1)......................................................................................................... .965-5 Interface state density ( Dit) changes with respect to th e nitridation temperature.............975-6 (a ) The change of peak electron mo bility with respect to the nitridation temperatures (b) Electron mobility versus electric field in MOSFET device fabricated for HfSiON films of 1.2 nm EO T (poly Si gate) nitrided at 650 oC and 850 o C................................................................................................................................986-1 XPS Si 2p spectra comparison of Hf-silic ate films with respec t to the methodological approaches (350 C thermally nitrided sample 1, 650 C thermally nitrided sample 2, and 350 C UV nitrided sample 3)...................................................................................1066-2 X-ray reflectivity (XRR) scans of ther mally nitrided (samples 1 and 2) and UV nitrided (sample 3) films..................................................................................................1076-3 SIMS depth profile of th ermally nitrided film (650 C) and UV nitrided film...............108 (350 C) after 400 C UV oxidati on (a) oxygen and (b) nitrogen...................................1086-4 C V comparison plot of three set of samples with 400 C UV oxidation........................1097-1 XPS Si 2p spectra comparison of nitrided samples (sample 1 and 2) and non-nitrided sample 3...........................................................................................................................1197-2 Cross-sectional HR-TEM images. (a) non-ni trided sample 3 (b) UV assisted nitrided sample 1...........................................................................................................................1207-3 Comparison of electrical properties of nitrided samples and non-nitrided sample (a) C-V comparison (b) J-V comparison................................................................................1207-4 The interfacial layer thic kness changes with respect to the PDA times for sample 1 and 3.................................................................................................................................1217-5 The interfacial layer density changes with respect to the PDA times for sample 1 and 3........................................................................................................................................1227-6 XPS Si 2p spectra comparison with re spect to the PDA times (a) UV assisted nitrided sample 1 (b) non-nitrided sample 3....................................................................1227-7 Arrhenius plots for interfacial growth for UV assisted nitrogen incorporated films and vacuum deposited films(a). Compar ison of activation energies for oxygen diffusion in various oxidation tim es for sample 1 and 3(b).............................................1238-1 C V characteristics of the MOS de vice with Ti, TiN, and TiB2 gate on Hf-silicate film at 1 MHz.................................................................................................................. .130 11

PAGE 12

8-2 The (qVfb + sub) vs EOT plots for the Ti, TiN and TiB2 gates. The EOTi and dielectric constant were determined by EOT vs Thighk plot inset in figure......................1318-3 J-V plot for Ti, TiN and TiB2 gates on Hf-silicate dielectric. Inset in figure shows Poole-Frenkel (ln(J/E ) vs E1/2) plot of three kinds of Ti based electrodes on Hfsilicate at 25 oC.................................................................................................................1328-4 Extracted trap depth of MOS devices with three kinds of gate electrode using slope of ln( J/E ) vs 1/T plot at different gate voltage which were inset in the figure................1339-1 Electrical measurement results (a) C-V characteristics at 1MHz of the MOS device with TiB2 gate on Hf-silicate film with at different annealing temperatures, (b) The qVfb + sub vs EOT plots for the TiB2 gate after different annealing. meff with consideration of interfac e is inset in figure......................................................................1409-2 AES depth profiles of Au/Ti/TiB2/Hf-sil icate/Si MOS structure as a function of annealing temperature......................................................................................................1419-3 HRTEM images of TiB2/ Hf-silicate (7 nm)/Si stack at as-deposited state and after annealing at 1100 C under N2 atmosphere.....................................................................1429-4 Electrical measurement results (a) EOT vs leakage current density plot at 1MV/cm for TiB2 gates on Hf-silicate dielectric of di fferent annealing temperatures. Inset figure is hysteresis ( VH) of TiB2/Hf-silicate/Si MOS capaci tors as a function of annealing temperature......................................................................................................14310-1 C-V characteristics at 1MHz of the MOS devi ce with Pt and Al gate on nitrided Hfsilicate film. The inset show s current density-voltage ( J-V ) measurements of both devices..............................................................................................................................14910-2 Poole-Frenkel (ln( J/E ) vs E1/2) plot of Pt and Al gate on nitrided Hf-silicate film at 25 oC. Inset in figure shows a schematic band diagram for MOS capacitors with HfSiON dielectric and interlayer and metal gates (Pt and Al) under negative gate bias. ..........................................................................................................................................15010-3 Changes in gate leakage current of Si MOS capacitors with HfSiON dielectric as a function of applied stress.................................................................................................15110-4 Changes in dielectric consta nt of HfSiON, HfSiOx and HfO2, measured from C-V and PF slope change.........................................................................................................152 12

PAGE 13

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy NITROGEN INCORPORATED HAFNIA GATE DIELECTRIC THIN FILM AND TITANIUM BASED METAL GATE ELECTRODE FO R DUAL GATE APPLICATION By Seung-young Son August 2008 Chair: Rajiv K. Singh Major: Materials Science and Engineering We evaluated the effect of nitridation temp erature on interface laye r (IL) quality of Hfsilicate gate dielectric has b een reported. An increase in IL density and IL roughness was observed as the nitridation temperature was increased. Preferential interface reaction at dielectric-Si interface at higher temperatures was analyzed. The progressive increase in IL roughness finally led to degradati on of breakdown voltage shift in flat ba nd voltage (0.54V) and deterioration of electron channel mobility by 20 % in samples nitrided at 1123 K. To reduce the interface degradation of thermal nitridation process, a low te mperature process (623 K) for nitrogen incorporation in hafnia gate dielectri c has been proposed. This method is based on postdeposition nitridation under ul traviolet light illuminated NH3 ambience. Uniformity of nitrogen distribution in the film was measured. Moreover, the amount of nitrogen incorporated by this process was comparable to that of high temp erature thermal nitridation, maintaining low interface roughness (0.3 nm). An evaluation of Ti based gate metals ha s been reported. The effective metal work function were 4.27, 4.56 and 5.08 eV for Ti, TiN and TiB2, respectively. Regardless of gate electrodes, the conduction mechanism of the samp les fitted with Poole-Frenkel model which is related to oxygen vacancies in the film. Ti gate electrode was found to be more favorable for 13

PAGE 14

14 NMOS device and TiB2 gate electrode can be used for PM OS. Further research for thermal stability of TiB2 gate electrodes was conducted. The extr acted effective metal work function for TiB2 gate was about 5.08 eV. The work function showed almost identical values and sharp interface between metal and diel ectric was confirmed after post deposition annealing by 1273 K. The work function lowering (4.91 eV) at 1373 K was caused by metal-dielectric intermixing and oxygen vacancy formation. TiB2 gate electrode was found to be su itable for use in PMOS device. Finally, Uniaxialmechanicalstrain altered gate leakage current and dielectric constant of MOS device are measured. Unia xial stress is applied using fourpoint wafer bending along one one zero plane direction. The gate leakage curren t and dielectric constant are found to increase by up to 2 % under tensile and co mpressive stress direction.

PAGE 15

CHAPTER 1 INTRODUCTION Fundamentals of Dielectric Dielectrics are a class of materials, which have no free carriers for electrical conduction. They are also known as insulators (non-metallic) and exhibit or can be made to exhibit a dipole structure. A dipole refers to the separation of positively and negatively charged entities on an atomic level. These charged en tities are bound to the atom or mo lecule and are therefore not available for conduction. When diel ectric materials are placed in an electric field, a phenomenon called polarization occurs whereby there is a sh ift in the charge distribution that leads to the dielectric behavior of the material system.1 Polarization induces electric dipoles, which are aligned with the applied field. The resistance of a material to polarization is measured by a parameter called the dielectric constant ( k ). Dielectric materials are widely used in modern electronics. Passivation of high voltage junctions, isolation of de vices and interconnects, gate insulation in field effect transistors are a few of the applications which involve dielectric materials. Silicon dioxide (SiO2), silicon nitride (Si3N4) and aluminum oxide (Al2O3) are a few of the commonly used dielectric materials in different areas of elec tronics and technology. The dielectric constant is the ratio of the permittivity of the material ( ) to the permittivity of vacuum or free space ( o). = / o (1-1) When two metal plates (paralle l plate capacitor structure) ar e placed under an electric field, one plate becomes positively charged and the other becomes negatively charged. The capacitance C is related to the magnitude of charge stored on either plate Q by C = Q / V (1-2) 15

PAGE 16

where V = voltage applied across the plates. The units of capacitance are farads (F) or coulombs/volt. When vacuum is present in between the parallel plates, the capacitance is computed by C = 0 [A / L] (1-3) where A = area of the plates L = distance between the plates 0 = permittivity of free space or vacuum (8.85 10-12 F/m) If a dielectric material is inserted in be tween the two plates, then the capacitance is computed by C = [A / L] (1-4) Where = permittivity of the dielectric material. is always greater than unity The ratio of the permittivity of the dielectric material to the permittivity of free space or vacuum is termed as the relative permittivity. It is denoted by r = / o (1-5) Relative permittivity is unitless and is also called as the dielectric constant of the material. As it is greater than unity, insertion of a dielectric material between the two metal plates represents an increase in charge stori ng capacity of the parallel plate capacitor. Basic Transistor The transistor is arguably one of the most im portant inventions of the twentieth century. The invention of this device has created countless technological advancements in the field of modern science and technology. Tremendous advancements based on the basic transistor have now enabled transfer of data, info rmation and ideas from one corner of the world to the other in a matter a few seconds. In the current generation devices, the microprocessors (heart of a computer) contain millions of transistors, which are meticulously and completely integrated onto a small microchip. The basis of the current co mputer technology is the use of complementary 16

PAGE 17

metal oxide semiconductor (CMOS) schemes as th e logic components in integrated circuits (ICs). Vacuum tubes were initially used (bef ore the invention of transistors) for signal amplification purposes in the long distance co mmunication devices. Though the state of art technology then, the vacuum tubes came with a number of disadvant ages. They were bulky, required extensive cooling, and consumed e normous power for operation. Around the mid 1940s research was being conducted at Bell La boratories (popularly known as Bell Labs) towards a new variety of materials called se miconductors. These are non-metals (Si and Ge), which could be made conducting by the addition of minute amounts of im purities (dopants). The first transistor was created on the 16th December 1947 and it consisted of strips of gold foil on a plastic triangle in intimate contact with a slab of germanium (Ge). This design soon underwent improvements with the creation of the junction sandwich transistor. The use of semiconductors presented the scientific community with adva ntages such as no warm up time, less power consumption and higher efficiency. Towards the end of the 1950s, the transistor design was modified (individually by Jack Kilby and Robert Noyce) into a simplif ied integrated circuit (device which contained more than one transistor). Metal Oxide Semiconductor Field Effect Transistor The metal oxide field effect transistor (MO SFET) is one of the key components in an electronic device, particul arly in digital integrated circuits. In this device a voltage applied on a contact called gate, which is isolated from a conducting channe l by a dielectric, controls the current flow through a conducting channel in the device. The transistor basically works as a switching device based on the field effect theor y. A schematic of a MOSFET is shown in Figure 1-1.2 The conducting channel could be n or p type. Th e substrate polarity is opposite of that of the channel. For example, an n type channel is created in a p-type Si substrate. Holes and electrons are the majority carriers responsible for current conduction in p-t ype and n-type silicon 17

PAGE 18

respectively. Alternatively, electr ons and holes are referred as minority carriers in p-Si and n-Si respectively. The schematic in Figure 1-1 shows a p-type Si substrate ont o which n-type regions called source and drain are di ffused or implanted. These re gions contain n-type dopants (elements which offer an electron for conduction purposes), which are introduced via a process called ion implantation. An n-Si can also be implanted with p-t ype dopants (elements with offer a hole for conduction purposes). The source and dr ain regions are subsequently annealed at a high temperature (~ 1050 oC) to electrically ac tivate the dopants (make them contribute carriers for electrical conduction). A thin layer of dielectric (conventionally SiO2) separates the gate metal from the Si substrate. This dielectric is usua lly referred to as the gate dielectric. No current flow can occur between the source and drain wi thout a conducting channel between them. As a voltage is applied to the gate charged carriers from extrinsic dopants in silicon are either attracted or repelled from the silicon / silicon diox ide interface. Hence, this gate metal / dielectric / silicon stack is a parallel pl ate capacitor in a nutshell. Under certain operating conditions, the minority carriers in Si (similar in polarity to the dopants in source and drain) can be employed to create a conducting channel from the source to drain. This channel allows current flow from the source to drain. Hence, a 1 or 0 interpretation can be derived based on the status of the channel ( ON conducting or OFF -non conducting). Metal Insulator Semiconductor Diode The metal oxide insulator diode is the most im portant device in the study of semiconductor surfaces. This device has been extensively explored and studied because it is related to most of the planar devices and integrated circuits. A typical MIS diode structure is shown in Figure 1-2 where d is the thickness of the insulator and V is the applied voltage on the metal gate contact. The energy band diagram of an ideal MIS diode st ructure at V = 0 is sh own in Figure 1-3. In Figure 1-3, (a) and (b) are for n-t ype and p-type semiconductors. In the case of an idea diode, at 18

PAGE 19

V = 0, the energy difference between the metal work function m and the semiconductor work function is zero. In short, the metal-semiconductor work function difference, ms is zero. The relationship is expressed by th e following equations for n-type and p-type semiconductors3 ms = m [ + Eg / 2q B ] = 0 for n-type (1-6) ms = m [ + Eg /2q + B ] = 0 for p-type (1-7) where m is the metal work function, is the semiconductor electron affinity, Eg is the band gap of the semiconductor, B is the potential barrier between the metal and the semiconductor, and B is the potential difference between the Fermi level EF and the intrinsic Fermi level Ei. The band is flat (flat band condi tion) when no bias is applied. When an ideal MIS diode is biased there ar e three cases that exis t at the semiconductor surface. The three cases are shown in Figure 1-4. Let us consider the case of a p-type semiconductor. When the device is in accumulation (a negative gate bias), the capacitance, in Farads, measured is the insulator capacitance, Ci and it is given the equation Ci [F/cm] = i / d (1-8) Where I is the insulating layer dielectric constant and d is the thickness of the insulator. In depletion, the overall capacita nce of the MIS stack decreases with increasing gate voltage. In the depletion layer, the overa ll capacitance of the MIS structure can be represented by two capacitors, Ci and Cs in series. Cs is the capacitance across the deple tion layer due to the separation of charges. The overall depletion layer capacitance is given by the equation2,3 si totalCCC 111 (1-9) 19

PAGE 20

When a larger positive voltage is applied on the gate, the bands bend even more downward, as denoted by Figure 1-4, so that the intrinsic level, Ei at the surface cross over the Fermi level, EF. At this point, the minority carriers in the semiconductor at the surface are larger than that of the majority (holes). Thus the surface is inverted and results in the thir d case called Inversion. Similar results are obtained in n-type semiconducto r. The polarity of th e gate bias, however, should be changed for the n-type semiconductor. When the metal-semiconductor work function difference is not equal to zero, charges are create d on the surface leading to a finite difference in the work function difference. Figure 1-5 shows the shift along the voltage axis of a high frequency C-V curve when positive or negative charges are present at the f ilm-Si interface. These charges are called fixed charges, Qf. They are fixed and are cannot be charged or discharged depending on the variation of s. These charges are generally located within a 25-30 distance of the Si-SiO2 interface. In electrical m easurements, the fixed charge, Qf can be regarded as a sheet of charge located at the Si-SiO2 interface. It has been suggeste d that excess Si (trivalent Si) or the loss of an electron from excess oxygen centers (non-bridging oxyge n) neart the Si-SiO2 interface is the origin of fixed ch arges. With the bands bent, a bias needs to be applied to first bring the bands to a flat band condition. This voltage is called the flat band voltage ( Vfb ). The voltage is determined by the metal-semiconducto r work function difference and the total fixed charge. The Vfb is given by the equation2,3 Vfb = ms +/Qf / Ci (1-10) Almost all the alternate dielec trics exhibit fixed charges at the film-Si interface leading to voltage shifts in the C-V response. An other form of charge located at the interface is called the interface trapped charge, Qit. This charge exists within the forbidden gap of due to the interruption of the pe riodic structure of the lattice at the surface of the crysta lline Si. Most of 20

PAGE 21

these interface trapped charged can be neutralized by a low temperature (~ 450 oC) annealing process in high purity H2 ambient. When a voltage is applied, the interface trapped charge moves up and down within the valence a nd the conduction bands while the Fermi level remains fixed. A change of charge in the interface trapped charge occurs when it crosses the Fermi level. This change contributes to the MIS capacitance and alters the MIS C-V curve. The charge in capacitance gives rise at a distorti on in the depletion region of the C-V curve as shown in Figure 1-6. SiO2 generally has a Qit of 1010 cm-2 or lower. The alternate ga te dielectrics have a higher Qit (> 1012 cm-2). This is a serious issue, which hinder s the implementation of the alternate gate dielectrics in conventiona l CMOS processing. High Qit values are deleterious for a reliable device operation. Steps must me taken to reduce the levels to lower than 1011 cm-2 for a successful implementation of high-k gate dielectrics in next generation CMOS devices. The difference between interface trap charge ( Qit) and interface state density ( Dit) is that the charge which occupy the interface trap is called interf ace trap charge. However, in most case of the MOSFET researches, the Dit and Qit is considered as similar parameters which determine the interface quality. Demand of Industry Since 1960, which saw the inception of th e metal-oxide-semiconductor field effect transistor (MOSFET), the most important device for modern integrated circuits, thermally grown silicon oxide (SiO2) has been used as gate dielectric s because of its advantages: 1) the electrically stable Si-SiO2 interface (i.e. Dit ~ 2 1010 eV-1cm-2), 2) the high dielectric breakdown strength ( 10 MV/cm) and 3) the thermal stabil ity at high temperature (remaining in amorphous state after the integration processes)4. For the last four decades, the improvement of speed and shrinkage of chip area of integrat ed circuits were achieved by scaling down of 21

PAGE 22

physical thickness of the SiO2 gate dielectrics and gate length (L). However, beyond the 100 nm node technology, SiO2 has reached its physical limitations: hi gher leakage current and reliability concerns. As shown in Table 1-1,5 continuing scaling down of the MOSFET device with the minimum feature size of 90 nm and below require s EOT (Equivalent Oxide Thickness) less than 15 A 10-15-thick SiO2 layer corresponds to only ar ound 3-4 mono-layers of SiO2. In this thinner EOT range, SiO2 suffers from leakage current too high to be used (particularly) for low power operation due to the direct tunne ling of electrons as shown Fig. 1-7. 2,6 Leakage current of SiO2 is governed by Fouler-Nordhe im tunneling where conduction occurs by field assisted electron tunneling at the field range of Vi Ei d > B while direct tunneling of electrons at the lower field Vi Ei d < B where Vi Ei and B represent voltage applied to dielectric, electric field across dielectric and barriers height between gate electrode and dielectric, respectively.2 In other words, as the thickness of SiO2 becomes thinner, leakage current is more likely to be governed by direct tunneling current which increases significantly as thickness becomes thinner by equation (1-11). 2 exp 2 2 exp 22V Bd V V Bd V d A JB B B B (1-11) Where h q A22, h qm B*24 and B is barrier height, respec tively. In equation (1-11) q, m*, and h represent electron charge, effec tive mass of electron, and Planks constant, respectively. In addition, SiO2 thickness uniformity across a 12 inch wafer imposes even more crucial difficulty in the growth of such a thin film, since even a mono-layer difference in thickness represents a large percentage difference and thus can result in the variatio n of threshold voltage 22

PAGE 23

( Vt) across the wafer.7 Reliability also becomes a huge concern for a SiO2 film only 10-15 thick.8,9 Therefore, by using physically thicker highk dielectrics for the same EOT, leakage current can be reduced by several orders Figure 18 shows the simulated result of leakage current for future device which said that expected l eakage current will exceed the operation limit from the SiO2 thickness 18 The highk dielectrics provide the same capacitance with a thicker film thickness (t) as the capacitance achieved using physically thinner SiO2 by equation (1-12) where C represents capacitance, k is dielectric constant, o is the permittivity of free space a constant, A is capacitor area, and t is dielectric thickness. C = k o[A/d] (1-12) Consequently, a research about dielectrics with higher dielectric consta nt than that of SiO2 or highk dielectrics have been a highlighted as a solution of future advanced CMOS technology. In next section high-k dielectrics as candidates for MOSFET application will be reviewed. Table 1-1. The 2003 ITRS technology roadmap for memory device 2003 2005 2007 2009 2012 2015 2018 Gate length (nm) 107 80 65 50 30 25 18 high speed 1.3 1.2 0.9 0.8 0.7 0.6 0.5 EOT (nm) Low power 1.6 1.4 1.2 1.0 0.9 0.8 0.7 Advantageous Properties Hafnium Based Gate Dielectric Dielectric Constant and Band Offset Highk gate dielectrics have been studied as alternative gate dielectrics for the 70 nm technology node and beyond to replace conventional SiO2 or silicon oxynitrides (SiOxNy). Principal requirements for highk dielectric applications are 1) hi gh dielectric constant 2) high 23

PAGE 24

band offset with electrodes (i.e. barrier height) to suppress l eakage current 3) thermally and chemically stable in contact with Si substrate. In Figure 1-9, dielectr ic constants of highk candidates were summarized. A TiO2 showing profoundly higher k in Figure 1-9, was reported not to be thermally stable with silicon substrates.10 Moreover, it is worth mentioning that highk dielectrics such as BST with a too high dielectric constant ( >100) does not seem to be appropriate since the excessive high dielectric constant causes field induced ba rrier lowering (FIBL) which degr ade short channel effects of MOSFETs.11 For electrons traveling from the Si substrat e to the gate, the conduction band offset ( Ec) is the barrier and for electrons traveling from the gate to the Si substrate. A favorable band alignment is very essential because the leakag e current increases expone ntially with decreasing barrier height and thickness for direct electron tunneling tr ansport mechanism. Figure 1-10 shows the relative band alignment of various pr ospective material systems with Si. If the experimental Ec values of these systems were much less than 1.0 eV, electron transport would lead to unacceptable leakage current precluding the implementation of a particular material system. Since most of the new systems do not have reported values of Ec, the band gap (Eg) is commonly used as an indicati on of the possible values of Ec. However, the valence and conduction band offsets need not always be symmetri c as in some material systems, the valence band offsets constitute most of the band gap. Hence, the prospective material of choice should posses a favorable band offset as well as a higher band gap. A Ta2O5, which has been studied widely for the application in DRAM storage capacitors appears to be inappropriate for the ga te electrode application. Also, the Al2O3 and Si3N4 which show reasonable band offsets, can not satisfy th e dielectric constant criteria as shown in 24

PAGE 25

Figure.1-9. When considering band o ffset and dielectric constant, HfO2 and ZrO2 are the most promising candidates for SiO2 replacement. However, ZrO2 has been reported that it was not compatible with poly-Si gate due to the reaction of Zr with Si and metal gates.12 In contrast, hafnia gate dielectric has been shown to be compatible with various gate metals including polysilicon.13-16 Especially, the MOSFETs with HfO2 dielectrics and TaN gate showed very low EOT (~10-12) and low leakage current even af ter the conventional CMOS process flow.16 Thermodynamic Stability The gate dielectric must be thermodynamically stable on Si, with respect to the formation of both SiO2 and MSi x At the appropriate process temperatures to which the wafer will be exposed, Si should not be react with the gate dielectric to form SiO2 or silicide. If SiO2 were formed, the capacitance budget w ould be consumed by a low dielec tric constant material. If the silicide were formed, a conductiv e path across the channel might be created. Stability criteria for all simple, as well as some multi-component metal oxides, have been thoroughly explored.17 The imposition of the thermodynamic stability cr iteria substantially reduces the field of acceptable alternate gate dielectrics. Single component oxides can be examined by comparing the Gibbs free energy of formation with respec t to oxygen. Elements having greater magnitude free energies than silicon will not readily give up oxygen to form SiO2 when in contact with Si.18 Table 1-2 lists the Gibbs free ener gies for various metals that form potential alternative highk oxides. 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 oxy gen to the silicon, and also tend to form silicates. Further considerations must be taken into account because this table only shows the values for systems in equilibrium. Therefore, other reactio ns could occur, so a good understanding of the kinetics behind the deposition or growth and processing steps is the key in predicting non-equilibrium products.19 25

PAGE 26

Therefore, considering dielectric consta nt, band gap, band offset and thermodynamic stability, HfO2 is the most promising candidate for future advanced CMOS application. However, there are several problems in hafnia dielectric in order for implement in integrated circuit manufacturing. In next section, the technical barriers of hafnium oxide dielectric are to be mentioned. Table 1-2. Gibbs free energy of formation fo r potential alternative high-k dielectrics Drawbacks of Highk Gate Dielectric Hafnia gate dielectric has been shown very promising possibility to apply to the next generation device integration. As described previous section, hafnium oxide have reasonable band gap, dielectric constant, conduction/va lence band offset and thermodynamic stability. However, hafnia film also has drawbacks to be considered. Interface Quality The interface property with Si is one of the critical issues for the application because channel of MOSFET is located at the interface between Si substr ate and gate oxide. The electron mobility at channel region will be affected by inte rface state. Most of the alternate materials of choice have an interfacial state density ( Dit) of 1011 1012 / cm2. The Dit values for a Si-SiO2 interface are around 2 1010 / cm2.1-5 This increase in Dit is mainly attributed to the interfacial reactions at the film-Si junction. Also, to minimize electrical l eakage and mass transport ( dopant, 26

PAGE 27

Si etc), the interfacial layer should be am orphous phase because the grain boundary of polycrystalline interface could be an inappropriate path. Hence, interface engineering schemes have been developed to form oxynitrides and oxide /nitride reaction barri ers between these high-k metal oxide materials and Si in an attempt to prevent or at least minimize reaction with the underlying Si.20-25 These barrier layers have been shown to reduce the extent of reaction between the highk dielectric and Si, as well as to help main tain a high channel carrier mobility. It is important to note, however, that us ing an interfacial layer of SiO2 or another low permittivity material, will limit the highest possible gate stack capacitance, or equivalently, the lowest achievable EOT value. Moreover, several simple oxides including HfO2 have been reported as having high oxygen diffusivity. Any annealing treatments which have an excess of oxygen present (either from the ambient or from a sidewall oxide, for exampl e) will lead to rapid oxygen diffusion through the oxides, resulting in SiOx or SiO2-containing interface layers. Since this oxide layer is completely uncontrollable and unexpected layer, the quality of interface can not be same as intentionally thermal grown SiO2 film. It is important to note that using an interfacial layer of SiO2 or another low permittivity material, will limit the highest po ssible gate stack capacitance, or equivalently, the lowest achievable EOT value Low Crystallization Temperature The most serious problem of hafnium oxide is known as low crystallization temperature. Generally, the device integration pro cess includes high temperature ( < 1000 oC) process for example carrier activation annealing after ion implantation, metal filling on high aspect ratio contact, RTA and etc. Therefore, the gate oxi de heat budget can not be avoidable. During the integration process, amorphous hafnium oxi de films are easily crystallized under 600 oC and then changed to a partially crystallized poly crystall ine hafnium oxide film. As the grain boundary of 27

PAGE 28

poly crystalline oxide will be a large current path and mass transportation source, the low crystallization temperature is a critical barrier to be overcome. Figure 1-11 shows the comparison of grazing incidence x-ray diffrac tion (GIXD) of hafnium oxide films which were deposited at 200 oC and 600 oC. The sample in the 200C deposition remained amorphous which is to be expected as the crystalli zation temperature for HfO2 is 450-500 C. However, the 600 oC deposited film at exhibited broad peaks, a sign of rather poor and randomly oriented crystallites. Identification of the main peak s and relative intensities was f ound to match the monoclinic HfO2 phase The peak at 55.5 is from the (311) plane of the silicon substrate. Channel Mobility Degradation The third problem which should be consider ed is channel mobility degradation in MOSFET. The electron mobility in the channel region will be affected by several factors. One of most important factors is interface state which was mentioned before. Because of deposition process, the interfaces be tween high-k oxide and Si have larg er amount of defects than interface between conventional SiO2 film and Si substrate. The increa sed interface state can be a scattering center with electron within the channel region. Hence, the possi bility of channel mobility degradation has to be increased. The other critic al reason for mobility degradation is arise from the polarization properties of highk materials. At inversion st ate, carriers flow along with channel from source to drain. Howe ver, due to the positive bias at the gate, the dipole moment is generated in the highk materials and the polarization disrupts electron movement. The various scattering center of MOSFET is illustrated in Figure 1-12. Among the various scattering center, remote phonon which is caused by dipole moment of high k is the most serious source for mobility degradation. Therefore, the channel mobility of high k MOSFET is always lower than the conventional silicon oxide MOSFET. Figur e 1-13 shows the mobility degradation of 28

PAGE 29

MOSFET using SiO2 gate and highk dielectric. The channel mobility of high-k device is around 40% lower than SiO2 gate due to the increased pola rization and remote phonon scattering.26-29 Incompatibility with Poly Si Gate The final technological barrier of highk dielectric application is incompatibility with polySi gate material. The conventional CMOS technologi es have used poly-Si gate materials due to the relatively easy process a nd good compatibility with SiO2/SiON gate dielectrics. Thus, the huge amounts of integration knowledge have be en accumulated based on the poly-Si gate. The main advantages of poly-Si gate are listed below. Compatible with high temperature process : self-aligned source & drain (SAC process) Work function can be tuned via doping contro l : easy approach to dual gate scheme Easy to deposit : LPCVD minimal damage and good step coverage Easy to etch and good removal by CMP : poly-Si can be etched (having high selectivity with SiO2 and nitride) using SF6, NF3, Cl2 chemistry However, high-k dielectrics and poly-Si are inco mpatible due to the Fermi level pinning at the poly-Si/high-k interface,30 which causes high threshold voltages in MOSFET transistors. The Fermi level pinning is most likely caused by def ect formation at the polySi/high-k dielectric interface, as illustrated in Fig. 1-14. High leve l of threshold voltage means that high leakage current and high power cons umption can be expected. Chapter 2 consists of a concise summary of the background research performed on HfO2 and alloying of hafnium oxide by groups around the world. A brief overview of metal gate processing of highk gate dielectrics is also included in chapter 2.. 29

PAGE 30

Figure 1-1. Schematics of a typical MOSFET device. Figure 1-2. Schematics of MIS device. 30

PAGE 31

Figure 1-3. Energy band diagrams of ideal MIS diodes at V=0 (a) n-type and (b) p-type semiconductor 31

PAGE 32

Figure 1-4. Energy band diagrams for ideal MIS diode structure when the gate is biased. (a) Accumulation (b) Deple tion and (c) Inversion 32

PAGE 33

Figure 1-5. C-V curve shift based on the gate bias due to the positive or negative charges (a) for p-type semiconductor Figure 1-6. Distortion or stretc h out of the C-V curve due to the presence of interface trapped charges 33

PAGE 34

Figure 1-7. Various modes of el ectron tunneling. In films with thicknesses greater ~ 50 FN tunneling dominates and the mechanism shifts to direct tunneling in films thinner than 50 34

PAGE 35

Figure 1-8. Expected leakage current density and leakage current operation limit with respect to the equivalent oxide thickness (EOT). 35

PAGE 36

Figure 1-9. The band gap and dielectric cons tant for gate oxide candidate materials. 36

PAGE 37

Figure 1-10. The conduction and valence band offset of various gate oxide candidate materials. Figure 1-11. The comparison of GIXD of hafnium oxide film deposited at 200 oC and 600 oC. 37

PAGE 38

Figure 1-12. The various scat tering centers of MOSFET. 38

PAGE 39

Figure 1-13. The comparison of channel mobility of SiO2 and higkk dielectric MOSFET. 39

PAGE 40

Figure 1-14. Defect formati on at the poly-Si and highk dielectric interface is most likely the cause of the Fermi level pinning whic h causes high threshold voltages in MOSFET (M = Zr or Hf). 40

PAGE 41

CHAPTER 2 LITERATURE REVIEW HfO2 and Hf-Silicate Dielectric A significant amount of research and monetary funding has gone into finding alternate gate dielectric materials to re place the current gate oxide (SiO2) in the past decade. Several variations in conventional thin film processing have been devel oped to tailor the material and electrical properties of the gate dielectric to attain the desired objectives. The material systems investigated as prospect ive candidates include Al2O3, HfO2, ZrO2, Ta2O5 and alloyed oxides of Hf, Ti and Al. Processes such as nitridation of th e dielectric films, surface nitridation of Si prior to dielectric growth, plasma treatme nts, ultra-violet (UV) radiation assisted thin film processing and UV assisted metal oxidation are among the few of the variations in conventional thin film processing. Though there is a huge disparity in the obtained results on a particular material system, several new constraints are now in pl ace for selection of the next generation gate dielectric. Factors such as minimal threshold volt age instability, microstructure stability at high temperatures (~ 1000 oC), minimal charge trapping and carrier mobility closer to SiO2 (> 90%) have become the key norms nowadays. Bonding and Electrical Properties Among all the above-mentioned material systems, HfO2 and Hf silicate like structures have shown the most promise as future gate dielectri cs. The key advantages of hafnium oxide include a relatively high dielectric cons tant (~ 21-25) depending on pr ocessing conditions, reasonable band offsets for acceptable leakage propertie s and thermodynamic stability with Si.4,31 Hf silicate exhibits an interface, which is the closest to Si-SiO2 interface due to similar bonding chemistry and coordination. Hf silicates also have a larger bandgap (~ 6 eV) with favorable conduction band alignments with Si and have also been repo rted to be thermally stable with Si at high 41

PAGE 42

temperatures. Hf silicates exhibit dielectric constants ranging from 8 14 depending on the amount of Hf present in the films.4,11,12 Bluementhal et al. have reported that the structure of ZrSiO is tetragonal and is composed of parallel chains of atoms consisting of Zr 2O Si 2O Zr 2O Zr and so on where each Zr and Si atom shares bonds with four oxygen atoms within the chains.32 In addition, each Zr and Si atom also shares two oxygen atoms with the neighbouring chains providing a three dimensional st ability to the material. The structural units are ZrO2 and SiO2. A dielectric constant of 12.6 was reported for ZrSiO4 by Bluementhal et al. which is reasonable considering that this material system is comprised of SiO2 ( ~ 3.9) and ZrO2 ( ~ 25). Analogously, HfSiO4 is also expected to have a simi lar structure due to its chemical similarity with ZrO2. The HfSiO4 is also expected to exhibit a dielectric constant of ~ 12-15 based on the processing conditions, microstructure of the film a nd on the relative Hf content. Considering all the above-mentioned properties of Hf and Zr silicates, they are promising candidates for alternate gate dielectric applications. HfO2 Gate Dielectrics L. Kang et al. have investigated the electrica l and material properties of R.F. sputtered HfO2 films on Si.33 The films (~ 45 thick) were deposited at room temperature in a controlled O2 ambient to suppress excessive growth of the low k interfacial layer at the dielectric-Si interface. They reported an equiva lent oxide thickness (EOT) of ~ 13.5 Pt was used as the gate electrode. The estimated leakage current densities were ~ 10-4 A/cm2. For comparison, a 13 SiO2 film is expected to exhibit a leakage curre nt density of 100 A/cm2 or higher. In addition to an EOT of 13.5 hysteresis values lower than 100 mV were also obtained. This study shows that the HfO2 can be implemented as the next genera tion gate dielectric when processed under the right processing conditions. The effects of high temperature treatment of HfO2 dielectrics were studied by S. W. Nam et al.34 The films were deposited by DC magnetron sputtering 42

PAGE 43

process at room temperature. First, a layer of Hf metal was deposited as an oxidation barrier and then a thin layer of HfO2 was deposited on the Hf metal in an ambient of Ar and O2. The Hf metal layer helps to form HfO2 by reducing the native oxide on Si The as-deposited films were found to be amorphous. However, the monoclinic and orthorhombic phases of polycrystalline HfO2 were reported to emerge after annealing at 650 oC and 900 oC respectively in O2 and N2 ambient. The measured EOT values were ~ 19 for N2 annealed samples and ~ 28 for O2 annealed samples. The increase in EOT is believe d to be due to the oxid ation of the underlying Si substrate. Lee et al. reported EOT values around 9 with reasonabl y low leakage in HfO2 MOS capacitors with Pt gate electrodes.35 They also reported a slight growth of an interfacial layer at the HfO2 Si interface. However, the main concerns of implementing a metal gate are lack of thermal stability and unfavorable band alignments. Pt is considered to be inadequate for NMOS applications due to its large work function (~ 5.3 eV). Nevertheless, Pt, due to its chemical inertness, is still an effective gate electrode for preliminary studi es of gate dielectrics through MOS capacitors. The main issues with HfO2 are the growth of the low k interfacial layer at the hafnia-Si interface and the amorphous to polycrystalline transformation at hi gh temperatures. The interfacial layer severely reduces the overall dielectric constant of HfO2 and the formation of grain boundaries is deleterious for a reliable device operation as they act as leakage paths. Hence, from an EOT point of view, HfO2 is reaching its scalability limits due to the above-mentioned issues. Other methods such as alloying HfO2 with high permittivity materials such as TiO2 are being actively considered to obtain EOT values of 15 and lower with acceptable leakage current densities.36 43

PAGE 44

Hf-Silicate Dielectrics Hafnium silicate based chemistries were also / are still being heavily investigated for their attractive properties such as a moderate k, la rge bandgap (~ 6 eV) and excellent interfacial stability with Si. Wilk et al. have reported on the electrical an d material properties of Hf-Si-O material system.37 The films were sputter deposited at various substrate te mperatures ranging from 25 600 oC. No evidence of Hf-Si formation wa s detected by XPS analysis and MOS capacitors were fabricated on the Hf-Si-O dielectri c films using Au as the gate electrode for its relatively large work function (~ 5.3 eV). The wo rk function of the Au electrode is helpful in creating a zero flatband voltage c ondition. The films deposited at 500 oC on Si exhibited a value of 11 and an EOT of 17.8 No quantum mechanical corrections were included in the extraction of EOT values. The films exhibited leakage current densities around 10-6 to 10-5 A/cm2 at 1 to 1.5 V bias range. In addition to excelle nt preliminary electrical results, the Hf-Si-O films were also found to be thermally stable after an 800 oC anneal in N2 ambient. The Hf-Si-O and Si interface was found to be atomistically sh arp after the high temperature annealing. De Gendt et al. have reported on Hf silicates processed by atomic vapor deposition (AVD) at 550 oC.38 The films exhibited dielec tric constants ranging form 6 14 depending on the film composition. EOT values of 13 were extracted with leakage current densities lower than 10-2 A/cm2. The flat band voltage was found to vary with the film composition from -0.1 V to 0.45 V. These studies establish the prom ising aspects of Hf-Si-O materi al systems as alternate gate dielectrics. However, a sub 13 w ith leakage current densities of 10-4 A/cm2 and lower are more desirable for high performance applications a nd for reliable device operation. A composition window for Hf content is extremely important to tailor the electrical pr operties to suit the objectives. 44

PAGE 45

Nitrogen Incorporation in Hafnia Film Nitrogen is a well-known diffusi on barrier to oxygen and is regularly used in CMOS processing. Si3N4 is an extension of the current gate dielec tric. It has a dielectr ic constant of ~ 8 depending on the processing conditions. Given the di ffusion barrier properties and the fact that the nitrides tend to have a better electrical response in comparis on to oxides, extensive research has been done on incorporating nitrogen in HfO2. Kang et al. have reported on th e electrical and interfacial pr operties of hafnium oxynitride gate dielectrics proce ssed by reactive sputtering.39 The films were postannealed in N2 ambient at 650 oC and the nitrogen incorporation was confirmed by XPS analysis of Hf 4f and Si 2p peak regions. In comparison to the control HfO2 films, Hf-ON films exhibited lower EOT values and leakage current densities. Even af ter a post metal anneal at 950 oC in N2, an EOT of 9.6 was obtained indicating excellent therma l stability. This enhanced electrical response was attributed to the presence of Si-N bonds at the film-S i interface and due to the blocking of oxygen by nitrogen bonding. Kirsch et al. investigated the inte rfacial properties of nitrogen doped HfO2 films and have reported on the impact of nitrog en on the electrical properties.40 A Hf target was used to initially sputter Hf onto HF terminated Si and NH3 annealed Si substrates. The nitridation of Si was achieved by pre-annealing the Si substrates in an NH3 ambient for 30 seconds at 700 oC. The asdeposited Hf films were later annealed in N2 ambient for 10 seconds at 600 oC. Given the affinity of Hf atoms towards oxygen, oxidation was mere ly accomplished by exposing the Hf films to the oxygen present in the processing ambient. The HfO2 films on nitrided Si exhibited lower leakage current densities (~ 10-5A/cm2 at 1 V) in comparison to the control HfO2 sample. A dielectric constant of 19 was obtained from the nitrided samp le in comparison to a dielectric constant of 17 from the un-nitrided sample. 45

PAGE 46

Recently, there have been several reports41-44 on the incorporation of both Si and N into HfO2 (i.e. HfSiON, hafnium silicon oxynitride) wh ich was found to improve thermal stability further compared to HfON. Especially, M.R. Vi sokay et al. performed comparative study on the effects of nitrogen incorporation for hafnium oxide and hafnium silicate films. According to M.R. Visokays report 30,43, incorporation of both Si and N into HfO2 is inevitable to increase crystallization temperature and avoid phase separation at CMOS processing temperature. However, dielectric constants are reduced in HfSiON due to the presence of silicon oxide bonds with much lower dielectric constant than HfO2. According to a report,44 HfSiON with optimized composition remained amorphous state up to 1100 oC whereas dielectric constant decreased down to ~10. In terms of application, the HfSiON appears to be very promising materials for the low power devices rather than high speed device requiring further scali ng-down of EOTs < 10 in the near future. From the above studies, it is clearly evident that nitrogen incorporation can play an important in enhancing the electrical properties of the HfO2 stack. However, the main drawback with nitrogen incorporation is the increase in Dit values. Fedorenko et al. have performed studies on the impact of nitrogen addition on the density of interface traps. They have reported an increase in the interface trap dens ity in the upper part of the Si band gap and have concluded that in this energy range, nitrogen prevents the passivation of interface traps by forming gas annealing or by hydrogen.45 Ultra-violet (UV) Assisted Thin film Processing Ultra-violet radiation as sisted thin film processing and thin film oxidation is an interesting alternative to achieve an enha nced degree of oxidation in Si.46,47 At temperatures as high as 850 oC, the oxidation rate during the thermal oxidatio n process of Si is around 2/minute despite a surplus oxygen supply. This low rate of oxidation is mainly attribut ed to a large energy barrier or 46

PAGE 47

activation energy (> 1.5 eV) required for oxygen to diffuse through the oxide to reach the underlying Si. When an UV source is employed in thin film processing, the energy of the photons with wavelengths around 183 nm and lo wer is sufficient enough to convert the molecular oxygen, which is regularly used fo r oxidation purposes, into atomic oxygen and ozone (O3). This newly generated ozone further dissociates into O2 and atomic oxygen O (1D) ranging from several percent to greater than 10%. The above-mentioned steps can be represented by the following equations.48 O2 + h 2O (2-1) O + O2 +M O3 + M (2-2) O3 + O 2O2 (2-3) O3 + O2 O2 + O2 + O (2-4) Further interaction of ozone with the short wavelength (254 nm and lower) photons will lead to the dissociation of ozone as shown by the equation 2-5. O3 + h O2 + O (2-5) Photo-induced transitions occur from the conduc tion band of Si to the conduction band of SiO2 (3.15 eV) and from the valence band of Si to the valence band of SiO2 (4.25 eV) when the sample is irradiated. Hence, with the influence of UV illumination, the Si is supplied with atomic oxygen from the gas interface and electrons from Si/SiO2 thus the probability of ionization of oxygen species is enhanced. Boyd et al. claim that the atomic oxygen is more reactive than molecular oxygen and has the ability to move more easily through the SiO2 matrix enabling enhanced oxidation rates. This effect has also been reported to be more pronounced at lower and moderate temperatures (~ 450 oC). Ramanathan et al. have investigated ultraviolet radiation assisted oxidation of high k dielectric films.49-51 Metal oxides (Zr and Hf) were sputter deposited 47

PAGE 48

at room temperature on thin layers of SiO2 or nitrided SiO2. The films were oxidized for various times (0.5 60 minutes) and at various oxygen partial pressures (80 mTorr 600 Torr). The oxide films were found to be polycrystalline wi th no preferred orientation. Normally, the oxidation of a metal occurs by (i) physisorption and (ii) dissociative chemisorption. The oxygen molecules initially form weak Van der walls fo rces with the metal. Subsequently the oxygen molecules dissociate into atomic oxygen and fo rm metal-oxide bonds. Th is dissociation of molecular oxygen requires an activation energy and is the rate limiting step. With the addition of an UV radiation source, the chemisorption step pr oceeds without any barrier as the energy of the 183 nm wavelength radiation is high enough to creat e atomic oxygen. The initial oxidation is rate is increased when compared to natural oxidation. Moreover, th e films after UV oxidation were found to be completely oxygenated (free of oxygen deficiencies) and exhibite d superior electrical performance. Hence the benefits of UV oxidation are two fold (i) en hanced oxidation rate an (ii) enables the formation of a stoichiometric oxide Zirconia films up to 5 nm in thickness were grown by the UV oxidation at room temperature whereas the natu ral oxidation was found to selflimiting at 1.5 nm. Punchaipetch et al. reported on room temper ature ultraviolet (UV) oxidation of hafnium silicide films for highk gate dielectric applications.52 The Hf silicide films were deposited on hydrogen terminated Si substrates at room temp erature by magnetron sputtering using Ar as the carrier gas. The Hf silicide films were subsequently oxidized under UV/O3 exposure for conversion of Hf silicide into Hf silicate. Angle resolved xray photoelectron spectroscopy (ARXPS) analysis confirmed the oxidation of Hf silicide films into Hf silicate. The films with ~ 12% Hf exhibited a dielectric constant of 89 and a leakage current density of ~ 4 10-5 A/cm2 at a bias of Vfb + 1 (volts). A 4.7 nm thick Hf silicate film had minimal interfacial layer formation 48

PAGE 49

and an equivalent oxide thickness of 26 (w ithout quantum mechanical corrections) was measured by C-V measurements at 100 KHz. The above -mentioned studies on UV assisted processing of gate dielectric films establish the efficacy of UV illumination towards achieving enhanced rates of oxidation and eliminating oxygen deficiencies in the films. With the current drive towards low temperature device proces sing, UV assisted oxida tion technique is an interesting alternative to fabricat e stoichiometric metal oxide or metal silicate films for alternate gate dielectric applications. Metal Gate Process for CMOS Device To meet the scaling trends and overcome degraded inversion capacitance due to polydepletion, metal gate electrodes are being evaluated as a replacement to poly-silicon electrodes.53 Metal gate electrodes have the a dded benefit of reducing the gate resistance and eliminating the threat of boron penetration from p+ poly-silicon into the channe l region. For bulk CMOS devices, simulations indicate that the desired work f unction for NMOS (PMOS) electrodes is near the conduction (valence) band edge of silicon.54,55 Candidate Metals for Metal Gate Application A quick review of the work functions of the elements3 shows a periodicity between work function and atomic number. Work functions increase from left to right across a row on the periodic table (Figure 2-1). This indicates that metals with work functions above the conduction band edge of Si (<4.1 eV) are typically in the first three columns of the periodic table. Alternatively, metals with work functions below the valence band edge of Si (>5.2 eV) tend to be late transition metals. In partic ular, the platinum group metals such as Pt, Ir, Os, Au, Ni, Ru, Pd, and Rh have high work functions. The vast majority of the transition metals have work functions that exist within the band edges of silicon (4.1 eV < ms < 5.2 eV). The work functions of metals 49

PAGE 50

have been studied using a variety of techni ques. These include the photoelectric effect, thermionic emission, field emission, contact potential difference, and extracted from MOS capacitors. These techniques do not necessarily produce the same value which complicates the search for candidate materials. To further narrow the list of viable metal gate electrodes, candidates must also possess thermal stability up to the dopant activation temperatures of approximately 900-1000 C. Thermal stability includes the absence of gr oss reactions between the dielectric and other surrounding materials, no inter-diffusions with surrounding materials, sufficient bulk phase stability, and smooth interfaces with the dielectr ic. Due to these stringent thermal requirements refractory metals such as W, Re, Ta and Mo have attracted interest. Metal nitrides and carbides have also gather ed attention as metal gate electrodes. These materials are renowned for their use as diffu sion barriers in the semiconductor industry. Transition metal nitrides and carbides are comp rised of a face-centered-cubic metal structure with nitrogen or carbon atoms occupying the octahedr al interstices. This re sults in the rock-salt structure (NaCl). Since N and C occupy interstices the lattice parameter is typically only ~5% larger than that of the pure me tal compound. Stuffing the interstices with N or C helps give these materials their excellent diffusion barrier prop erties. These materials exhibit metal-like conduction, but at the same time are highly refractory compared to the pure metal constituent. It is believed that a mixture of metallic, covale nt, and ionic bonding charact er is responsible for this behavior.56 There have been numerous studies on the wo rk function of interstitial nitrides for electrode and emitter applications. This includes MoN,57 WN,58 NbN,59 TaN,58,60 TiN,57,59-61 ZrN,59 and HfN63 among others. TiAlN has also been wide ly studied as a metal gate electrode. 64 Although it is a ternary metal nitrid e, it is more closely related to the binary nitrides than the 50

PAGE 51

amorphous ternary metals because this compound exists in the rock salt structure for Al:Ti ratios <0.40 and in a mixture of rock salt and wu rtzite structures for >0.40 Al:Ti ratio.64 TiAlN loses its conductive properties with increasing presence of the AlN wutzite phase. Ternary alloys of a transition metal (TM), si licon, and nitrogen have also been widely investigated for their properties as diffusion barri ers. These films are found to exist in a highly metastable amorphous structure.65 This property makes them excellent diffusion barriers to elevated temperatures. Ta-Si-N is currently be ing investigated as a stable NMOS electrode candidate. Another problem is the need to integrate two me tals with different work functions to build both NMOS and PMOS devices. This is easily done with poly-silicon electrodes by ion implanting n-type or p-type dopants into NM OS or PMOS electrodes, respectively. The implantation is performed selectively by patte rning the NMOS region with photoresist while implanting dopants into the PMOS regions and vice-versa. The implants shift the Fermi energy of n-type poly-silicon toward s the conduction band and shift th e Fermi energy of p-type polysilicon towards the valence band to achieve the appropriate work functions and low device threshold voltage. Different integration approaches have been proposed for metal gates. Both conventional integrations and replacement gate67 integrations have been proposed. Replacement gate integrations dramatically reduce the temperatures that the metal gate must withstand because the source/drain activation anneals ar e performed before the metal gate is deposited. The other integration approaches can be cla ssified into one of four types. These include stacked dual metal gate integration, metal alloy integration, nitr ogen modulation integrati on, and fully-silicided metal gate (FUSI) integration. 51

PAGE 52

The integration approach that most easily accommodates vastly different metals for NMOS and PMOS electrodes is the stacked dual metal gate integration (Figure 2-2). This integration involves depositing one metal over both the NMOS and the PMOS regions. Then patterning one of the two regions and removing th e electrode using a wet etch chem istry that is selective to the underlying gate dielectric. A second metal can th en deposited over the first metal. The most difficult aspect of this integration is the plasma etching two different ga te stacks of different heights. This integration has been demonstrated on Si3N4 gate dielectric and on HfO2 gate dielectrics.68 In the metal alloy approach two metals are in itially deposited on top of each other. Either the NMOS or PMOS region is then patterned so th at the top metal can be selectively removed. A high temperature annealing process is then used to inter-diffuse the two metals. This leaves an inter-diffused metal for one of the electrodes an d an elemental metal for the other electrode (Figure 2-3). This approach ha s been performed using Ru-Ta69,70, Ti-Ni71, and Pt-Ta72 alloys. The alloyed electrode approach has two foreseeable problems. First, it may be difficult to get one alloy system to span the entire 1.1 eV range to meet the NMOS and PMOS work function requirements. Second, some of the elements used in the alloy may not be thermally stable in contact with a high permittivity dielectric. The nitrogen modulation integr ation has been reported in two forms (Figure. 2-4). The first method is to ion implant nitrogen into a metal to shift its work function. This has been demonstrated for Mo-based electrodes,73 and for TiN-based electrodes where the N concentration is modulated via implantation of N.74 Adjusting the nitrogen concentration has also been accomplishe d by solid-state diffusion of nitrogen from a nitrogen-rich film into a nitrogen-deficient film. Like the alloy approach, it may be difficult to 52

PAGE 53

span the entire work function range to meet the NMOS and PMOS electrode requirements by implanting nitrogen. Fermi Energy Level Pinning A roadblock to successful implementation of metal gate electrodes with the proper work functions is Fermi level pinning. Fermi pinni ng is a consequence of forming an interface between a metal and a dielectric (or semiconducto r). When an interface is formed, the effective metal work function becomes pinned at a differe nt energy than its vacuum work function. This results from interfacial charge exchange between the metal Fermi level and gap states at the metal-dielectric interface causi ng it to shift with respect to its unpinned location. A recent comprehensive review of Schottky ba rrier concepts has been published.75 Fermi level pinning of poly-Si76 and metal electrodes77 on HfO2 has been investigated, and Fermi pinning models have been extensively tested on a broad class of interfaces.78.79 The most widely accepted of these models is the metal indu ced gap states (MIGS) model. The origin of the gap states in the MIGS model is from the da ngling bonds of under-coordin ated surface atoms. These dangling bonds produce surface states that are dispersed in c ontinuum at energies throughout the band gap of the dielect ric. In the MIGS model, when a metal is placed in contact with a dielectric, the metal surface states induce the gap states in the dielectric. This results in interfacial charge exchange be tween the metal and the dielectric gap states causing the metal Fermi level to shift with respec t to its unpinned location towards a characteristic energy level in the semiconductor Other models have been proposed where the states responsible for Fermi level pinning have an extrinsic origin. These include the unified defect theory where a specific defect, with a high density of states at a given energy is responsible for the pinning behavior ; disorder-induced 53

PAGE 54

gap states, where variations in bond length and bond angle at an interface result in states dispersed across the band gap; and finally the e ffective work function theory where the effective work function is determined by the work function of a metallic species that precipitates at the interface. Strain Engineering for High Channel Mobility MOSFET Aggressive scaling of complementary meta l-oxide-semiconductor (CMOS) technology requires a high drive current to increase circuit speed. Both the gate capacitance and carrier mobility can improve the drive current. The highk dielectrics and metal gate can increase gate capacitance. However, mobility degradation, and interfacial layer scalability delay the introduction of highk dielectrics into industria l applications. In order to enhance the mobility, mobility enhancement by strain, new materials such as Ge or SiGe channels, and new substrate orientation such as (110) and (111) offers alternative ways to increase drive current.80 Applying stress to induce appropriate strain in the channel region of metal-oxide-semiconductor field effect transistors (MOSFETs) increases both electron and hole mobilities in the strained channel.81,82 The mobility enhancement mechanism is explained by band modification. The easiest way to check the strain on the Si cha nnel is stress inducement by bending the Si wafer directly (mechanical strain). Th e mechanical strain is called Package-strain due to strain inducement at package process. Package strain can improve the NMOS and PMOS device performance after the fabrication of VLSI and the cost is relatively low. Band Modification Model The enhanced electron mobility can be e xplained by the conduc tion band modification induced by the biaxial tensile strain, which lifts the six-fold degeneracy in the conduction band and lowers the energy of the two valleys along the growth direction ( 001) (Figure 2-5). The electrons occupy preferentially th e two lower energy valleys, which has the low effective in54

PAGE 55

plane transport mass. Energy splitting also suppre sses intervalley s cattering. The mobility at the low field ( <0.4 MV/cm) is dominated by impurity s cattering (Coulomb scattering), the high field mobility ( >1MV/cm) is determined by the roughness sc attering, and the intermediate field mobility is dominated by the phonon scattering. No te that the effective electrical field is perpendicular to the channel and is controll ed by the gate voltage and device structure. Conventionally, the roughness scattering is orig inated from the oxide/Si interface. Electron mobility enhancement of substrate-strained Si devices at high channel doping (up to 6 1018/cm3 ) is reported. If the inversion layer carrier concentration is low, the strain-induced mobility enhancement decreases due to the ionized impurity scattering, but the enhancement recovers at higher inversion ch arge concentrations, where the screening is more efficient. Application Scheme for Future Device Substrate-strain results in biaxial strain to channel and enhances electron and hole mobilities. However, cost, scalability, and complexities such as low-defect Si1 xGex buffers, low thermal budget, and integration, remain issues fo r production. Moreover, the hole mobility of the Si channel at high fields is not improved significantly except at the Ge or dual channel because the energy separation between light and heavy ho le bands is not enough due to the quantization effect. In contrast, uniaxial strain offers sim ilar electron mobility enhancement compared with biaxial strain, while the hole mobility impr ovement is more significant at high field.83 For <110> channel orientation on (100) substrates, the uniaxi al compressive strain parallel to the channel and the uniaxial tensile stress transverse to the channel result in larger hole mobility enhancement than biaxial tensile strain at the sa me stress level. Hole mobility enhancement with these uniaxial stresses is mainly due to band warping, resulting in the decrease of the transport effective mass along the <110> direction.84 Furthermore, these uniaxial strains may have lower 55

PAGE 56

surface roughness scattering due to the large out -of-plane effective mass (less wave function penetration into dielectrics).83 Process induced strain has been used to produce the desired uniaxial strain.85 Several approaches (Figure 2-6) such as silicon nitride (Si3N4 ) cap layer, shallow-trench isolation (STI), silicidation processes, and embedde d SiGe S/D have been utilized to realize the local strain.86 Silicon nitride film is well-known to produce a high level of stress [Figure 2-6(a)]. Si3N4 film can have either tensile or compressive strain depending on the deposition conditions, and the drive currents of both nand p-channel MOSFETs can be improved by controlling the stress of the Si3N4 layer selectively. The tensile cap la yer deposited by thermal chemical vapor deposition (CVD) can improve the performance of NM OS due to the induced tensile strain in the channel region, while the compressive cap la yer deposited by plasma enhanced CVD can improve the PMOS due to the induced co mpressive strain in the channel region.86 Shimizu et al. reported that a highly tensile strained Si3N4 cap layer can improve NMOS performance but degrade PMOS performance.86 Capping a highly-tensile Si3N4 layer shows 25% NMOS drain current improvement,87 while selective Ge implanted into the capping layer can recover the degradation of PMOS devices. The strained PMOS transistor features an ep itaxially grown strained SiGe film embedded in the S/D regions using a selective epitaxial growth was reported by Intel [Figure 2-6 (b)].88 The source and drain regions, ma de of an alloy of Si1 xGex, are deposited in the recessed source and drain region. The lattice constant of the SiGe alloy is larger as compared to the bulk Si due to the inclusion of Ge. The larger lattice constant of Si1 xGex creates a compressive stress in the channel of a 45-nm gate length transistor between the source and drain regions, thereby resulting in significant hole mobility improvement (50%) with 17% Ge incorporation. It is interesting to 56

PAGE 57

note that hole mobility is also enhanced at the high field region for the uniaxial compressive strain parallel to the channel. By integrati ng and taking advantage of aforementioned strain engineering techniques, CMOS performance can be further improved. As a final note for process induced strain, the process strain is effec tive for the 90-nm node and beyond, but its enhancement effect diminishes for large channel devices. 57

PAGE 58

Figure 2-1. Plot of work function versus atomic number. Work functions are generally observed to increasing moving across a row of transition metal elements. 58

PAGE 59

Figure 2-2. Process flow for the stacked dual metal gate integration. The difficulty in this approach is that two gate stacks of different heights need to be etched simultaneously. Figure 2-3. Process flow for the metallic inte gration showing how two metals are formed by depositing a metal stack, selectively patterning and etching the top metal, and annealing to alloy the bi-metal stack. 59

PAGE 60

Figure 2-4. Process flow for the ion implantati on and solid-state diffusion of nitrogen to form a nitrogen-rich (NMOS) and a nitr ogen-deficient electrode (PMOS). Figure 2-5. The six-fold degene racy of electron, before and after biaxially tensile strain. 60

PAGE 61

(a) (b) Figure 2-6. Schematic features of the various pr ocess strain: (a) silicon nitride capping layer to create a tensile channel (b) embedde d SiGe S/D process strain to create a compressive strain. 61

PAGE 62

CHAPTER 3 OUTLINE OF RESEARCH Keeping all the above mentione d constraints and requirements in mind, the search for an alternate dielectric requires unders tanding of material properties of the prospective candidates and the effects of chemical bonding on the electri cal response of the diel ectric. With all its advantages hafnia (HfO2 and Hf-silicate) is still the most promising candidate and this dissertation is based on the tailoring of Hf-silicat e to achieve specific goals such as a medium dielectric constant (15 20), leak age current densities lower than 10-3 A/cm2 at 1 MV/cm and EOT values around 20 and lower. Chapter 4 fo cuses on the experimental part of this dissertation. The processing condi tions, vacuum science, equipm ent, sample characterization techniques and the excimer laser set up along with the UV illumination source are described in detail. Chapters 5, 6 and 7 describe the results of nitridation process me thodologies used to tailor the properties of hafnia film. Chapter 5 describes the interface degradation within the thermal nitridation process. In chapter 6 and 7, UV assisted low temper ature method for nitridation is introduced. Chapter 8 and 9 explain the result of Ti based gate el ectrode properties in order for dual metal gate application. In chapter 10, sinc e future CMOS device wi ll adopt the strained MOSFET structure as described above, an eff ect of mechanical stre ss on MOS capacitor with nitrided Hf-silicate dielectric has been studied Finally, chapter 11 summarizes the research performed in this dissertation with suggestions fo r future work to further possibly finetune the experimentation and data anal ysis to achieve the goals. 62

PAGE 63

CHAPTER 4 EQUIPMENT SET-UP AND CHARACTERIZATION All the dielectric film samples, studied in this dissertation, were processed via pulsed laser deposition (PLD). The PLD system was purchased from Neocera Inc, Gaithersburg, MD. The deposition chamber is equipped with a multi-targ et carousel, which can hold a maximum of 6 targets. The metal films were initially deposited on p-Si substrates by pulsed laser ablation of high purity targets (Hf, Ti and Si) from SCI Engineered Materials. The metal films were subsequently oxidized in high purity oxygen ambient under the pres ence of ultra-violet (UV) illumination from a low pressure-vacuum compatible Hg lamp array. The lamps were warmed up for two minutes prior to the oxidation anneal. The substrate temperature was monitored by a thermocouple and an optical pyrometer. The ga ses used in the experiments were high purity oxygen and ammonia. Laser System A Lambda Physik LPX 305 i (KrF) excimer la ser (s/n 9412-E-4188) was used for all laser ablation experiments. The laser works on a pulsed mode delivering 25 nanosecond duration square wave shaped pulses at frequencies ra nging from 1-50 Hz and output energies from 101100 mJ (fluence 0-3 J/cm2). The laser operation is controlled and triggered by a computer and the mode of trigger can be internal or from a remote external computer. Excimer lasers are nowadays commonly used in semiconductor manufacturing and in eye surgery. Excimer actually refers to an Excited Dimer. Most "excimer" lasers are of the noble gas halide type, for which the term excimer is strictly speaking a misnomer (since a dimer refers to a molecule of two identical or similar parts). The final output wavelength of the laser is dependent on the type of gases used. The Kr F excimer laser shoots pulses around 254 nm, which is in the ultra-violet regime. Laser action in an excimer molecule occurs because it has a bound 63

PAGE 64

(associative) excited state, but a repulsive (disassociative) gro und state. This is because noble gases such as xenon and krypton are highly iner t and do not usually form chemical compounds. However, when in an excited state (induced by an electrical discharge), they can form temporarily bound molecules with themselves (dim ers) or with halides (complexes) such as fluorine and chlorine. The ex cited compound can give up its excess energy by undergoing spontaneous or stimulated emission, resulting in a strongly-repulsive ground state molecule which very quickly (on the order of a picos econd) disassociates back into two unbound atoms. This forms a population inversion between the two st ates. In the case of ou r laser, the Kr and F species are elevated to excited states by applic ation of very high voltages (16 21 kV) so that excited KrF* species are formed. The excited sp ecies owing to their extremely short residence time in the excited state, decay to ground stat e and in the process photons corresponding to a wavelength of 254 nm are emitted. As a result of specific cavity design, conditions exist whereby stimulated emission of coherent radiation occurs and subsequent amplified high-energy laser output in a pulsed mode can be obtained. The laser radiation emitted fr om the cavity is directed towards the deposition chamber through a variety of optics an aperture, beam re flectors and final focus lens, before it finally ablates the target of choice in the chamber, which is pumped to a medium-high vacuum. The ablation spot on the target had the dimensions of 2 5 mm in a nearly perfect representation of the rectangular aperture used earlier in the beam path. This type of spot was achieved by adjusting the focusing lens position to a local grea ter than the focal point of 25 cm to a position of ~35 cm, which coincided with the image plane of the lens. Owing to the use of an aperture, imaging of the beam was possible and a spot with a highly uniform energy density was obtained. 64

PAGE 65

Pulsed Laser Ablation via Excimer Lasers In the last two decades, pulsed laser deposit ion (PLD) has emerged as one of the most commonly used research tools fo r oxide thin film processing. The technique has a lot of advantages with low capital investment, stoichio metric transfer of material from target to substrate, easy operation, lower vacuum requi rements when compared to MBE, effective prototyping of many different ma terials being main highlights of this technique. The ablation process can be explained by the inpu t of energy (incident) from the la ser to the target material of choice. The total incident energy is a summation of E = Er + Ed + Ep + Ec (3-1) E = incident energy Er = reflected energy Ep = plasma plume energy Ec = energy absorbed by the cavity wall Ed = energy of disintegration For ablation to occur, the pulse or inpu t laser energy should be higher than the bond strength or energy of the material to be ablate d. In short, the combination of pulse energy and ablation threshold energy determines whether or not material ablation is possible and along with the degree of ablation. If the pulse energy is le sser than the threshold en ergy, there will be no ablation but the energy is still absorbed and this is helpful for laser annealing experiments. However, if the energy of incidence is too high, particulate ablation can take place. This is deleterious for high quality film growth. All the experiments were carried out with an incident laser energy greater than the ablation threshold but low enough to preven t excessive particulate ablation. 65

PAGE 66

It is well known than metals have a different electronic structure than oxides. Owing to their high reflectivity, higher energies (~ 9001000 mJ) are typically employed to achieve ablation. Vacuum Chamber, Component and Operation The pulsed laser depositions performed for this dissertation were done in a Neocera brand vacuum system. A schematic of our Neocera PLD sy stem is shown in Figure 4-1. The system is a single chamber design that is routinely backfille d with nitrogen gas to atmospheric pressure so that samples may be mounted or removed. Vacuum levels of 1 10-5 Torr and 1 10-6 Torr were easily achieved within 3 and 6-7 hours respec tively. The chamber is connected to a Pfeiffer MD-4T oil free diaphragm roughing pump and a Pfeiffer TMU 230 turbo pump. A calibrated Neocera brand stainless steel resistive heater capable of reaching temperatures around 850 oC is mounted vertically in the chamber and used to co ntrollably heat and cool the substrate to and from a desired temperature. The studies performe d in this dissertation were done under a base pressure of 1 10-5 Torr or under 1 10-2 Torr of high purity NH3 during nitridation to incorporate nitrogen. The chamber has a computer controlled multi-target carousel. Up to 6 targets can be loaded for multi-layer or super latt ice depositions. An array of Hg lamps has been added to the chamber to perform ultraviolet as sisted PLD experiments. Four Hg lamps emit a majority of UV radiation (~240 nm) and a small portion (~ 10 %) of in the 183 nm range. The Hg lamps are composed of a fused silica envelope with allow more than 85 % of the emitted 183 nm radiation to be transmitted. This 183 nm is mainly responsible for conversion of oxygen into ozone and other reactive species. Figure 4-2 sh ows a picture of the UV lamp set up in our Neocera PLD system. Ultra-pure gases may be a dded to the system through a highly sensitive Varian brand leak valve for a wide range of deposition ambients and pressures. 66

PAGE 67

Atomic Layer Deposition Traditionally gate silicon dioxide has been grown and then ntireded, DRAM capacitor dielectrics have been deposited by Chemical Vapor Deposition (CVD). Problem with thickness control and uniformity generally limit CVD process to films thicker than 10nm. As addressed in previous section, the gate oxi de thickness of future CMOS de vice should be less than 10 nm. Therefore, the need for new depos ition technologies is apparent. Deposition Mechanism In a standard CVD process a wafer or a group of wafers are placed in a vacuum chamber where chemical vapors are thermally reacted at lo w pressure to deposit a film on the wafer. The deposition process is continuous the vapors flow continually into the chamber during the deposition cycle. The deposited film thickness de pends on the temperature, pressure, gas flows volumes and uniformity., chemical depletion effect s and time ( for state of the art processes batch systems generally no longer used due to the difficulty of achieving temperature and gas flow uniformity over multiple substrates at the same time). Controlling all of these parameters to the level required for good thickness control of thin films is very difficult. Atomic Layer deposition (ALD) deposits film s using pulses of gas producing one atomic layer at a time. Within fairly wide process windows the deposited film thickness is only dependent on the number of deposition cycles providing extremely high uniformity and thickness control. The basic deposition process is illustrated in Figure 4-3. Figure 4-3 illustrates the deposition of zirconium dioxide which deposition sequence is very similar with hafnium oxide. The deposition steps are: 1) ZrCl4 vapor is introduced into the process chamber. 67

PAGE 68

2) The ZrCl4 vapor forms and an adsorbed monolayer on the surface of the wafer. Although not shown, following monolayer fo rmation the chamber would be purged of ZrCl4 vapor by an inert gas pr ior to the next step. 3) H2O vapor is introduced into the chamber. 4) The H2O vapor reacts with the ZrCl4 surface monolayer to produce one monolayer of ZrO2. Because only a monolayer of ZCl4 exists on the wafer surface, only one monolayer of ZrO2 is produced making the process self limited. Following ZrO2 formation the chamber would be purged again and additional cycles would be performed as necessary to produce the desired film thickness. ALD reactions are typically carried out in the 200 oC to 400 oC temperature range. If the deposition temperature is too high, chemical bonding cannot be sustaine d or the density of chemically reactive sites is reducedreducing de position rates. If the deposition temperature is too low thermally activated chmisorption and fi lm forming reaction rates decrease reducing deposition rates. As the deposition temperature is increased form low to high the deposition rate increase reaches a peak and then decreases. The temperatur e window for maximum deposition rate is relatively wide compared to CVD processes that are much more temperature sensitive. Figure 4-4 schematically illustrates the a llowable temperature window for ALD. Precursors must be volatile and thermally stable to ensure efficient transportation so that reactions will not be precursors transportation controlled The vapor pressure of precursors must be high enough to completely fill the deposition chamber so that the monolayer deposition takes place within a reasonable length of time. Precurs ors must chemisorb onto the surface or rapidly react with surface groups and react aggressively w ith each other to keep deposition times short. 68

PAGE 69

The precursor cannot self-decomposes or the se lf limiting property of ALD is lost and the precursors should not etch or dissolute into the film or substrate. Figure 4-5 shows a schematic diagra m of deposition cycle of ALD HfSiOx. As shown in Figure. 4-5, one cycle of ALD HfSiOx (m+n) is composed of m cycles of HfO2 and n cycles of SiO2. It is possible that compositional control of HfSiOx is achieved by c ontrolling the ratio of the deposition cycle of HfO2 (m) and SiO2 (n). Very slow ALD SiO2 (m=0) deposition (< 0.1 /cycle) was observed with sequentia l exposure of Si precursor and H2O. However, ALD HfSiOx films were successfully deposited over 0.5 /cycle where the m and n have the intermediate values between 1 and 10 (1 m 10 and 1 n 10). Wafer Bending Experiment The uniaxial jig used in applying stress is a f our point bending fixture. Figure 4-6 shows a uniaxial jig. Such a bending structure has been well studied and a relation between the applied force and stress under uniform stress is given by 89 2)(3 DLF wt (3-2) where F is the applied force, D and L are the inner and outer support distances respectively, and w and t are the samples width and thickness as shown in Figure 4-7. This formula is accurate when the sample is not severely bent to the applied forces and the dimensions w and t are small enough compared with D and L .89 Under these conditions, the stress directions applied on the both surfaces of the sample can be approxim ated to be tangential, and the magnitude of stress applied everywhere between the inner supp orts can be treated as a constant. A detailed diagram is shown in figure 4-7. Eq. (3-2) is a useful formula in calibrating stress sensors.89,90 However, we can not use it to directly relate the jig parameters with the measured physical quantities. 69

PAGE 70

Another form of Eq. (3-2) fit for our experiments is found in some literatures 91: 3 2 2 2 aL a dt YY (3-3) Here, s and e are the stress and strain values at the center of the sample respectively, Y is Youngs modulus of Si along the stress direction, 2 DL a and the deflection d is the vertical displacement between the upper and lower plates of the uniaxial ji g when we apply stress. In figure 4-7, d is defined as d = di df, and actually measured by th e change in micrometer graduations. Material and Chemical Ch aracterization Techniques The as-deposited and annealed films were subject ed to a variety of material and chemical analysis techniques. Information about densit y, chemical bonding, chemical composition and microstructure were extracted from the results ob tained by the techniques. A brief description of the material and chemical techniques is given below. X-ray Reflectivity X-ray reflectivity (XRR) measurements were made using a PANalytic al XPert system. Data generated from an x-ray reflectivity plot include film thickness, roughness and density of the material under inspection. This technique is very well suited for smooth thin films and is very sensitive to samples in the 20 400 nm range. This technique works by impinging the sample with x-rays over an array of a ngles ranging from slightly sub-cr itical angles to the first few degrees after the critical angle. The critical angle also known as Brewsters angle corresponds to the point at which x-rays change from total re flection off the sample surface to absorption and interaction with the sample as defined by Snells law. The retrieved data results form monitoring the intensity of the x-ray beam reflected from various interfaces relative to the incident beam as a 70

PAGE 71

function of the scattering transfer vector. Fresnel equations will then describe the interaction of the x-rays with one another and with the in terfaces encountered in the film stack. The constructive or destructive interf erence nature of the x-rays at a given angle results in the generation of a fringe pattern. Figure 4-8 s hows a typical XRR plot. The generated XRR spectrum is modeled with a theoretical fit with a model comprising of the di fferent layers in the sample. For all the samples in this dissertati on, a 4-layer model was employed. The layers include (from substrate and going up ) 1) Substrate silicon in this case 2) Interfacial layer (IL) formed at the film Si interface 3) the deposit ed film 4) Surface contamination layer comprising of sub-stiochiometric organics as all the sample s contain This technique can generate important information when employed to characterize high k as almost all of the high k films when deposited on Si form an interfacial layer. The film and interfacial layer thickness & density can be obtained from the fit and hence information about the nature of the inte rface of the film stack can be deduced from those parameters. However, the better the theore tical model, the more accurate will the results match the real physical structure. X-ray Photoelectron Spectroscopy (XPS) XPS is a photon in-electron out technique based on the ionization of inner shell (core) level electrons by a monochromatic photon source. The sa mples in this dissertation were analyzed using a Perkin Elmer 5100 equipped with Mg K radiation source (1253.5 eV). When a sample is illuminated by a monoenergetic soft x-rays (Mg K ), the process of photoelectric effect causes inner shell electrons with kinetic energies ~ 0 1500 eV to be emitted. The emitted photoelectrons have a short mean free path (top few atomic layers) and hence XPS is a highly surface sensitive technique. The number of photoelectrons emitted are measured based on their kinetic energy by an electron analyzer. This resu lts in the characteristic photoelectron spectrum as shown in Figure 4-9. The photoelectron yield in the Y-axis is normally plotted against the 71

PAGE 72

binding energy (BE). The kinetic energy is th e energy possessed by the photoelectron after subtracting the energy required to overcome the work function. Each orbital is associated with a signature binding energy. Thus th e photoelectron spectrum can be us ed to analyze the type of bonding and elemental composition in the films base d on the binding energy of the characteristic orbital. To analyze the layers at a depth, Angle Resolved X-ray Photoelectron Spectroscopy (AR-XPS) is normally used where the photoelect rons emanating at various escape angles are analyzed. Hence, scans at various take off a ngles (150, 300 and so on) are performed. The 150 scan is surface sensitive while the 900 scan can be used to analyze the chemical environment in the bulk of the film (in depth). XPS uses photoionization and energy dispersive analysis for chemical composition analysis and study of electronic surface states. High Resolution Transmission Electron Microscopy (HR-TEM) Transmission electron microscopy is one of the most powerfu l microstrucural analysis techniques available. By this technique we can determine defects in films and achieve atomic imaging of the interfaces in addition to the regula r microstructure analysis. The interfacial layer abruptness, microstructure and thickness confir mations were performed using a JEOL 2100. One of the main drawbacks of HR-TEM measurements is the intensive sample preparation required to obtain high quality images. The sample has to be made extremely thin (~ 80-100 nm) for electron transparency purposes. A focused ion be am was used to fabricate TEM samples to the required thicknesses instead of the conventional i on milling process. Figure 9-3 shows a typical HR-TEM micrograph showing the film microstructure and interfacial layers. Auger Electron Spectroscopy (AES) AES is a popular technique for determining the composition of the top few layers of a surface. It cannot detect hydrogen or helium, but is sensitive to all other elements, being most sensitive to the low atomic number elements. Th e electron beam can be focused over a large or 72

PAGE 73

small surface area, or it can be directly focused on a small surface feature. This ability to focus the electron beam to diameters of 10 nm and less makes AES an extr emely useful tool for elemental analysis of small surface features. More over, the ability to ra ster the electron beam over an adjustable surface area provides control ove r the size of the analytical area. When used in combination with ion sputter sources, AES can perform largeand small-area compositional depth profiling. Despite the adva ntages of high spatial resolution and precise chemical sensitivity attributed to AES, there are se veral factors that can limit the a pplicability of this technique, especially when evaluating solid specimens. One of the most common li mitations encountered with Auger spectroscopy are charging effects in non-conducting samples. Charging results when the numbers of secondary electrons leaving the sa mple are greater or less than the number of incident electrons, giving rise to a net polarity at the surface. Both positive and negative surface charges severely alter the yield of electrons emitted from the sample and hence distort the measured Auger peaks. To complicate matters, neutralization methods em ployed in other surface analysis techniques, such as secondary ion mass spectrometry (SIMS), are not applicable to AES, as these methods usually involve surface bombardme nt with either electr ons or ions (i.e. flood gun). In this work, to confirm the chemical intermixing between metal and dielectric film with respect to the PDA temperature, AES depth pr ofile analysis was conducted using a Physical Electronics 660 Scanning Auger Microprobe. The electron beam conditions were 10 keV, 1 mA beam current at 45o from sample normal. Secondary Ion Mass Spectroscopy (SIMS) SIMS depth profiles were acquired with a Pe rkin-Elmer PHI 6600 SIMS system using a 1keV cesium primary ion beam with negative and positive secondary acquisition. The current 73

PAGE 74

intensity was set at 32 nA. The raster size was 450 450m2 with 70% gating during negative secondary acquisition and 55 % gating w ith positive secondary acquisition. Under Cs bombardment, nitrogen has to be de tected combined with an impurity ion or another species in the matrix. With negative secondary acquisition, N was acquired as mass 43SiN-. Characterization of these f ilms with negative secondary acquisition has one drawback: mass interferences (i.e. secondary ions that occur at the same nominal mass). Mass interferences occur whenever an isotope of one element has the same nominal mass as an isotope of another. Isotopes of one element, dimers, and molecular combinations may also have the same nominal mass as others. Positive secondary acquisition using CsX+ cluster ions, where X is the impurity element of interest has less matrix effects but the intensity of the signals is much lower than with negative acquisition. Nitrogen was monitored using 147CsN+. Oxygen, Si, and Hf were monitored using 149CsN+, 161CsSi+, and 180CsHf+, respectively. This SIMS data is only qualitative. SIMS is used for detection of minor impurity elements (less than 1%in the matrix) and quantification of impurities requires the analysis of standards under the same instrumental conditions as the un knowns. This is because SIMS is a very matrix dependent technique, that is, the sensitivity of the same element varies with matrix composition. Standards can be either implants or bulkdoped samples with known concentrations of the species of interest in exactly the same matrix as the unknown samples. When standards are not available, the levels of the impurities of interest can be compared between samples after ratioing the intensity of the impurity of interest to a constant matrix ion. Electrical Characterization The quality of the dielectric films can be an alyzed and confirmed by the use of electrical characterization techniques such as capacitance-voltage ( C-V ) and current-voltage ( I-V ) methods. 74

PAGE 75

After the dielectric film depos ition and material analysis (XRR, XPS etc), metal oxide semiconductor (MOS) devices were fabricated by depositing Pt dots of known areas on the dielectric by R.F sputtering. The typical thickne ss of the dots was around 900 Silver was used as the backside contact. The Pt dots were deposited via a shadow ma sk with an array of circular dots ranging from 25 500 m diameter were used to create th e dot arrays on the samples. The dot size used for analysis in this dissertation was 3.16 10-4 A/cm2. It is important to choose the appropriate metal for the contacts. If improper metals are chosen, band alignments may exist whereby the nonohmic contacts are created within th e device itself. As an example, the backside contact on a p-Si wafer should have a workfunction greater than that of Si. If not, the contact will be a Schottky contact (rectifying) After the deposition of contact s was completed, the samples were annealed in forming gas (95 % N2 and 5 % H2) at 450 oC for 30 minutes in a conventional furnace. This is mainly done to passivate the dangling bonds that might exist at the gatedielectric and dielectric-Si interfaces. Both current-voltage and capacitance-voltage measurements were conducted after the forming gas annealing process. Current-Voltage Measurements Once the MOS capacitors structures were fabr icated, a Keithley Instruments Inc., K1236 source measurement unit (SMU) was used to meas ure the current flow through the device. The 236 SMU was attached to a black box probe stat ion equipped with a pair of Signatone Inc, micromanipulators. The manipulators were fitted with tungsten probes that were milled to produce a fine ~ 5 m tip. The output of the SMU was conn ected to the micromanipulator in contact with the gate (Pt) while the input was connected to th e backside contact (Ag). This configuration is optimal for the determination of current leakage pathways directly below the device being measured (i.e, it avoids stray leakage paths). Determination of the leakage current is an important step in analyzing the quality of the MOS device. Typically the current compliance 75

PAGE 76

threshold (a value that may not be exceeded) of 100 nA is fed into the measurement parameters. Then a direct current bias sweep is conducted over a voltage range and the amount of current that passes through the MOS structure is monitored. Typically, it is be st to start at a small voltage sweep range so that it is possible to determin e if the device is lea ky without causing a large amount of bias induced defects. The main goal of the leakage current measur ements is to identify a high quality MOS device that can be used for capacitance-voltage measurements to extract parameters such as capacitance, dielectric c onstant and equivalent oxide thickness (EOT). Capacitance-Voltage Measurements Capacitance-voltage measurements serve as one of the most versatile and sensitive of all electrical techniques. It is the ultimate tool for determining disc reet differences in a MOS device that may serve as a final word in whether a given processing condition has resulted in a high quality device. The measurements in this di ssertation were carried out with a Keithley Instruments Inc. Win-82 measurement set-up. This system is comprised of four main components that work in unison. The Keithley 59 0 capacitance meter is used for high frequency capacitance measurements at 100 KHz and 1 MH z. The Keithley 595 capacitor meter was used for simultaneous quasistatic low frequency m easurements. The Keithley 230 voltage source is used for static bias condition measurements. Th ese three devices were wired into the Keithley 5951 remote input coupler, which serves to filter the device data to and from the various pieces of equipment. The output of a typical C-V curve is shown in figure 4-10. There are several important features that should be noted. First, there are th ree important regions with respect to gate bias voltage to take into consideration, known as accumulation, depletion and inversion. The presence of the different regions is a result of majority charge carriers in the semiconductor. When a negative bias is applied to the gate electrode, pos itively charged holes are attracted from the bulk 76

PAGE 77

of the semiconductor to the oxide semiconduc tor interface region where they accumulate (accumulation). The depletion region is generated when a gate is made less negative and the reduced field across the oxide causes the charge at the interface to diminish. As the sign on the voltage changes from negative to positive, ma jority carriers are repelled from the interface creating an area depleted of majo rity carriers (depletion region). Finally, the inversion region is generated when the voltage becomes very positiv e and the depletion width has increased to a point where the other mechanisms may become im portant. For example, in the depletion region, the product of the concentration of electrons and holes (np) is mu ch less than the square of the intrinsic carrier concentration (ni 2) and in this case, pair genera tion may occur and the subsequent minority carriers may migrate to the oxide-s emiconductor interface. Further depletion is prevented due to screening eff ect of the minority carriers. 77

PAGE 78

Figure 4-1. Sketch of the pulse d laser deposition (PLD) system. Figure 4-2. Photograph of an UV lamp array in the pulsed laser deposit ion system. The lamps were located at a distance of ~ 5-7 cm from the substrate. 78

PAGE 79

Figure 4-3. Atomic layer deposition of ZrO2. Figure 4-4. ALD acceptable temperature window. 79

PAGE 80

Figure 4-5. Schematic illustration of ALD HfSiOx deposition cycle. 80

PAGE 81

Figure 4-6. A fixture to simula te uniaxially-strained MOS devi ces. For a uniaxia l stress, two pairs of cylindrical rods ar e used and a sample is inserted between the pairs. 81

PAGE 82

(a) (b) (c) Figure 4-7. Illustration of a uniaxial wafer bending jig. (a) an unstressed sample (b) a stressed sample. (c) is the practical deflection and induced stress. 82

PAGE 83

Figure 4-8. Typical X-ray re flectivity (XRR) spectrum from a thin film deposited on Si. Figure 4-9. Typical X-ray photoe lectron spectroscopy spectrum of graded hafnium silicate film deposited on silicon. The binding ener gy shifts of Hf 4f photoelectron peak can be used to determine the nature of bonding present in the films. The different angles represent the photoelectron take off angles or escape angles. 83

PAGE 84

Figure 4-10. Metal oxide semiconductor capacitan ce-voltage curves (a) low frequency (b) high frequency and (c) deep depletion. 84

PAGE 85

CHAPTER 5 STUDY OF INTERFACE DEGRADATION OF HAFNIUM-SILICATE DIELECTRICS DURING THERMAL NITRIDATION PROCESS An evaluation of the effect of nitridation te mperature on interface layer (IL) quality of Hfsilicate gate dielectric prepared by atomic layer deposition method has been reported. An increase in IL density and IL roughness wa s observed by X-ray reflectivity (XRR) as the nitridation temperature was increased. X-ray photoelectron spectroscopy showed preferential interface reaction at dielectric-Si interface at high er temperatures. The progressive increase in IL roughness finally led to degrada tion of breakdown voltage, shift in flat band voltage (~0.54V) and deterioration of electron channel mob ility by ~20% in samples nitrided at 850 oC. Introduction Hafnium silicate has been extensively studied as a new gate dielectric material in advanced metal oxide field effect tran sistors (MOSFETs) due to high crystallization temperature, thermodynamic stability with Si, high permittivity and relatively large band gap (5.68 eV).12,30,9296 However, hafnium silicate film still has is sues such as hafnium inter diffusion, lowk interface layer formation and phase separation between hafnium oxide and silicon oxide. Nitrogen incorporation in the hafnium silicate film was explored to overcome some of these issues. Nitrogen incorporation minimized the IL form ation and further redu ced the crystallization tendency of the HfSiOx films.30,49,97-99 In addition, it also reduced boron diffusion from poly-Si gate through hafnium silicate.100 Thermal nitridation or plasma nitridation has been studied to incorporate nitrogen into the dielectric film. The high temperature treatments involved in the nitridation process particular ly post deposition annealing pr ocess is bound to alter the dielectric/Si interface.41,51,99-107 Since electrical properties of MOSFET such as flat band voltage, electron/hole mobility and leakage current density are affected by interface quality, the study of interface properties is very important to unde rstanding the electrical behavior of MOSFET 85

PAGE 86

device. Previous reports have al ready revealed that engineerin g IL by nitrogen incorporation improved the electrical pr operties of MOS device.95 However, all the above mentioned results were qualitative studies, there are only a fe w quantitative reports for interface properties especially, interface roughness change with respect to the nitridati on process. In this chapter, I focused on quantitative analysis of the interf ace layer roughness change with respect to the nitridation temperature by XRR studies and rev eal the origin of IL roughness changes by the investigation of chemical state variation using XPS. Finally elec trical characteri stics of HfSiOx films corresponding to the physical properties of IL are examined. Experimental Detail HfSiOx films were deposited dire ctly on MEMC p-type (100) pre-cleaned (SC1, 1% HF solution and DI water rinse) Si substrates with a resistivity of 3-25 cm, by atomic layer deposition (ALD) using GENUS Lynx2TM at 300oC. SiH[N(CH3)2]3 was used as Si precursor whereas Hf(NEtMe)4 (TEMAH) was employed as Hf pr ecursor. The oxidizing agents for TEMAH and Si were O3 and H2O, respectively.18 In order to minimize the dielectric constant loss of Hf-silicate film, Hf and Si ratio (Hf/Hf+Si) in the film was tuned to 0.9. HfSiOx film deposition was followed with nitrogen inco rporation by rapid thermal anneal in NH3 ambience for 60 s. Nitridation temperature was varied from 300oC to 850oC. The change in chemical state was investigated using x-ray photoel ectron spectroscopy (XPS) with Mg K radiation. Interface layer (IL) thickness and interface roughness was determined using x-ray reflectivity (XRR, PANalytical Xpert system). For XRR spectra, an X-ray mirror and parallel plate collimator were used as the primary and secondary optics, respec tively. In order to inve stigate the electrical characteristics of films, metal-oxide-semiconducto r (MOS) capacitor with Pt gate electrode and poly Si gate MOSFET was fabricated using oxide mask technique. The C V characteristics were analyzed at high frequency (1 MHz) using an HP 4284A LCR meter. NCSU program was used to 86

PAGE 87

analyze the electrical measurements and extract the flat band voltage and mobility. The gate current density-voltage ( J-V ) and dielectric breakdown (BD) measurements were performed using HP4156B. The statistical approach for BD behavior was analyzed by using of Weibull distribution function. The in terfacial state density (Dit) at flat band voltage was determined by the method of Lehovec.104-106 Results and Discussion X-ray reflectivity (XRR) spectra of Hf-silicate films are presented in figure 5-1. Since XRR spectra were obtained using the symmetric /2 configuration with ve ry small angles, this technique is suitable for the investigation and characterization of thin films. From the curve fitting of periodic pattern width and height, we can find the interface roughness and thickness. For the best curve fitting (Wingixa software, Philips), a multi-layer model, which consisted of the interfacial, Hf-silicate, and contamination layers, was applied. Film density was determined by the critical angle (c) in which the experimental deviation is about 3% in our study. The non-specular diffuse scattering as well as the specular scattering was considered in order to obtain the surface and interfacial roughness values which we are focusing on in this report. The physical structure of Hf-silicate dielectric such as thickness, density and roughne ss of the film and the IL as obtained from XRR measurement afte r thermal nitridation is tabulated in Table 1. It can be observed from table 5-1 that both the fi lm and IL density is increasing with nitridation temperature. Increase in nitrogen incorporation with increase in the nitridation temperature has been reported earlier.95 The densification of film and IL can be attributed to the presence of larger amount of nitrogen. The increase in thickn ess of IL is due to highe r diffusivity of nitrogen at higher temperatures. It should also be noted that the density of the film is more than twice that of the IL indicating IL to be mostly Si bonded to O and N. 87

PAGE 88

Roughness of the IL is an important property wh ich can affect the interface charge density and cause scattering of elect rons, surprisingly it has not been studied in HfO2, Hf-silicate system to the best of our knowledge. It is especially important in the ca se of post deposition nitridation process (thermal nitridation, plas ma nitridation) where growth of IL occurs at high temperatures and is diffusion controlled. As can be observed from table 1, roughness of the IL is increasing with nitridation temperature. The increase in nitroge n content in the film is coming at the cost of degradation of IL/Si inte rface. This degradation is particularly rapid at nitridation temperatures of 750 and 850oC as is the case with other properties. Figure 5-2 (a) shows the effect of nitridati on temperature on high resolution XPS scan of Hf 4 f peak region. The Hf 4 f7/2 peak for as deposited sample which is predominantly HfSixOy is at 18.1 eV. The Hf 4 f7/2 peak undergoes a shift towards lowe r binding energy (BE) with increase in nitridation temperature partic ularly at temperatures above 650oC. The lowering of BE indicates an increase in the screening of Hf 4 f level, which results from increased electron density around Hf atom. In case of nitridation, increase in el ectron density comes from the incorporation of N atoms having lower electronegativity i.e. by fo rmation of Hf-N-Hf or Hf-N-Si bonding states.102,103 The continuous shift towards lower BE implies continuously increasing nitrogen amount in the film w ith higher nitridation temperature. The broadening of Hf 4 f peaks with the increase in nitridation temperature can also be seen in the figure. Peak broadening is probably the result of randomiza tion of Hf environment by incorpor ation of higher percentage of N atoms.104 Shown in figure 5-2 (b), are the N 1 s core level XPS scan and the deconvoluted spectra of the dielectric films nitrided at different temperatures. A drastic change in the shape of the peaks with increase in nitrid ation temperature can be observed, indicating a change in the nature of N atom bonding. The N 1s spectra of nitrided Hf-silicate films was shown by Cho et al. 88

PAGE 89

to be a fusion of numerous peaks 397.0.3 eV (N1), 398.1.2 eV (N2), 398.7.2 eV (N3), 400.0.1 eV (N4), and 403.3 eV (N5), each origin ating from different bonding environment of N atoms.101,102 The low BE peaks (N1, N2, N3) comes fr om N atoms bonded mostly to Si atoms, N1 from plannar structured N Si3, N2 differing from N1 with respect to the second nearest neighbor and N3 peak from O-N=Si2. High BE N4 peak is attributed to N atoms bonded to Hf in HfO2 matrix.101,102 The N5 peak was assigned to N2 and NOx molecular states which can easily out diffused at high temperature.101,102 The 300oC nitrided sample showed a major portion of N4 state and existence of N5 peak with minor N1 state. The prominence of Hf bonded N (N4) and N5 at 300oC can be explained by surface saturation of N atoms due to limited diffusion of N inside the film. The increase in nitridation temperature to 650oC is marked by (i) increase in the intensity of N1 peak, (ii) a ppearance of another Si bonded N3 state and (iii) disappearance of molecular N2, NOx peaks(N5) from the spectrum. The increase in nitridation temperature will increase the N diffusion inside the film and to the IL. This led to increase in the thickness of IL as shown in Table 1. But the disproportionate increase in the intensity of N1 and N2 peaks compared to the IL growth at 750 and 850 oC nitridation shows a preferential buildup of N at the interface. The higher affinity of nitrogen to Si compared to Hf is well reported, it seems like Si interface is acting as a sink for nitrogen during high temperature nitridation. Therefore, the degraded IL roughness which was observed by XRR measurements, can be understood by the result of the N inter diffusi on and preferential accumulation in the IL. The IL roughness degradation without film surface roughness change (table 5-1) also ascertains that the IL roughness degradation was a result of inter diffusion process. I investigated the electrical properties of th e Hf silicate films to determine the structureproperty correlation. The variati ons in the flat band voltage w ith nitridation temperature are 89

PAGE 90

shown in figure 5-3 (a). Flat ba nd voltage moved to the negative potential region in proportion to the nitridation temperature. Oxygen vacancies in the dielectric can be occupied by nitrogen atoms. The nitrogen bonded oxygen vacancies are a major source of fixed charge in the film causing increased interface state.108-110 Therefore the flat band voltage movement is due to the nitrogen accumulation at the IL and film. Current voltage ( J-V ) characteristics for Pt/Hfsilicate/Si MOS device are shown in figure 5-3 (b ). Leakage current density of all samples was lower than 10-4 A/cm2 at VGVFB = -1 V. But, higher density of film and IL from increased incorporation of nitrogen led to decrease in leakage current a nd improvement in the breakdown voltage ( VBD). The leakage current density of nitrided Hf-silicate film at VGVFB = -1 V was lower than that of as-grown hafnia film. Figure 5-4 shows a comparison of the cumulative BD probability versus voltage for 7 nm Hf-silicate on Pt electrode with respect to the nitridation temperatures. The cumulative probability ( P ) variation with VBD can be fitted with a Weibull distribution.106-108 The considered Weibull distribution was: oo BDS S V V Pexp1 (5-1) where, S is the capacitor surface ar ea (in this case 4.66 10-4 cm2), S0 is a reference surface area, and and V0 are the Weibull parameters. The mean VBDvalue ( VBD at 50% probability) improved 87 % (from -3.3 V for as grown film to -6.2 V for 750 oC sample). However, it is worth noting that the VBDof 850 oC nitrided sample displayed lower mean VBD value than that of 750 oC nitrided sample. Also, from the linearizing Eq. (1), Weibull slope () can be extracted. Weibull slope () of the films nitrided at 850 oC and 750 oC were 2.33 and 5.13, respectively. Higher indicates a lower dispersion of the VBD for 750 oC nitrided 90

PAGE 91

film.107 The lowering of mean VBDand of the film nitrided at 850oC is due to the roughness effect. As seen from XRR measurements, IL ro ughness continuously increas ed with nitridation temperature, rougher IL led to formation of elect ric field concentration points in the film along with increase of specific contact area, this probably led to drop in VBD of the film nitrided at 850oC despite the increase in film and IL density. Furthermore, excessive nitrogen accumulation at IL resulted in higher dispersion of the VBD for 850oC nitrided film. Figure 5-5 shows the change of the interface stat e density at flat band voltage with respect to the nitridation temperatures. The Dit values were determined by Lehovec method from C-V plots. The Dit of nitrided sample show higher values than that of as-deposited sample. This increase in Dit level in the nitrided sample due to nitrogen incorporation is consistent with the reports by Fedorenko et al.45 However, 850 oC nitrided film displayed significant increase in Dit compared to other samples. The highest Dit of 850 oC nitrided film can be understood as a result of the thick IL thickness and surface potenti al fluctuation corresponding to the excessive nitrogen incorporation causing roughness degr adation and nonuniformity of oxide charge distribution. Figure 5-6 show the change of peak electr on mobility extracted from split C-V method with respect to the nitridation temper atures. The electron mobility decreased with proportion to the nitridation temperature. The degradation of electron mobility at nitrided samples is consistent with previous reports. It is believed that the mobility degradation is due to the presence of coulomb scattering sites at the in terface of the HfSi ON. These scattering sites are likely due to the Si-N bonds at the HfSiON/Si interface.111,112 The electron mobility vs electric field curves of 650 and 850 oC nitrided Hf-silicate on poly-Si gate is inset in figure 5-6. The peak (0.4 MV/cm) and high field electron mobility (1.0 MV/cm) of 650oC nitrided film is higher than that of film nitrided at 850oC. The degradation of peak m obility in film nitrided at 850oC is 91

PAGE 92

correlated with the result of Dit degradation which can be a coulomb scattering center.111,112 However, it should be noted that the difference in electron mobility of film nitrided at 650 and 850oC is larger at high electric field than at the field of peak mobility. The mobility degradation at high field can be understood as a result of the IL roughness degradation caused by excessive nitrogen piling up effect. Conclusions The effect of temperature in thermal nitridati on process on physical structure of the IL was investigated using XRR studies. At high nitridat ion temperatures, the nitridation efficiency increased leading to increase in IL roughness apart from increase in film and IL density and IL thickness. There was a general improvement in the electrical properties of films at higher nitridation temperature till the degradation of interface quality due to interface roughening led to degradation of some of the elec trical properties. An excessive nitrogen piling up at IL of 850 oC nitrided Hf-silicate resulted in nonuniform distribution of ni trogen and oxide charge, high roughness IL and degradation of Dit, electron channel mobility and breakdown resistance. Therefore, the quantitative results of IL roughness by XRR study showed reliable correspondence with electrical resu lts and can have a potential as an analysis tool of thin film structure. 92

PAGE 93

Table 5-1. Summary of x-ray re flectivity data on roughness, de nsity, and thickness of IL and HfSiNO films. Figure 5-1. The comparison of XRR patterns wi th respect to the nitridation temperature. 93

PAGE 94

(a) (b) Figure 5-2. Dependence of (a) Hf 4f peak region and (b) N 1 s core level of XPS scan on nitridation temperature. The percentages in N 1 s deconvoluted spectra represent area fraction of the respective peaks. 94

PAGE 95

Figure 5-3. (a) Flat band voltage and (b) current density gate voltage characteristic of Pt/Hfsilicate/Si MOS device fabricated on films with different nitridation treatments. 95

PAGE 96

Figure 5-4. Cumulative probability of breakdown voltage for 7 nm thick Hf-silicate on Pt electrode with respect to the nitridation temperature. Lines are fits to a Weibull distribution (Eq. 5-1). 96

PAGE 97

Figure 5-5. Interface state density ( Dit) changes with respect to the nitridation temperature. 97

PAGE 98

Figure 5-6. (a ) The change of peak electron mobility with respect to the nitridation temperatures (b) Electron mobility ve rsus electric field in MOSFET device fabricated for HfSiON films of 1.2 nm EOT (poly Si gate) nitrided at 650 C and 850 C. 98

PAGE 99

CHAPTER 6 HIGH EFFICIENCY NITROGEN INCORPORATION TECHNIQUE USING UV ASSISTED LOW TEMPERATURE PROCESS FOR HAFNIA DIELECTRICS An evaluation of a low temperature process (~350 oC) for nitrogen incorporation in hafnia gate dielectric has been reported. This method is based on post-depos ition nitridation under ultraviolet light illuminated NH3 ambience. X-ray photoelectron spectroscopy confirmed the amount of nitrogen incorporated by this process was comparable to that of high temperature (~650 oC) thermal nitridation (~7%). Uniformity of nitrogen distribution in the film was analyzed by secondary ion mass spectroscopy. A capacitance density of ~ 3.96 F/cm2 with 9.4 equivalent oxide thickness and 10 thick inte rface layer were obtained by ultraviolet assisted nitridation process Introduction Hafnium silicate has been extensively studied as an alternative gate dielectric material in advanced metal oxide field effect transistors (M OSFETs) due to high crysta llization temperature, thermodynamic stability with Si, high permittivity and relatively large band gap (5.68 eV). 30,9296,113. However, hafnium silicate film still has is sues such as hafnium inter diffusion, lowk interface layer (IL) formation and phase sepa ration between hafnium and silicon oxides. Nitrogen incorporation in hafnium silicate was e xplored to enhance the resistance to thermal degradation. Nitrogen incorporation minimized the IL fo rmation and further reduced the crystallization tendency of the HfSiOx films.30,114-117 In addition, it also reduced boron diffusion from poly-Si gate through hafnium silicate.48 However, most of the nitrogen incorporation methods employ post deposition annealing process such as rapid thermal annealing ( RTA ) in NH3 or N2O ambience at high temperatures ( > 650 oC). As a result of high temperature nitridation process, additional in terface degradation such as lowk oxide layer growth and boron diffusion have been observed.116,117 Hence, to reduce the tota l heat budget during device 99

PAGE 100

integration, there is a critical need to explore a low temper ature nitrogen doping method which has comparable efficiency as high temperature methods in order to minimize interface degradation. As ultraviolet (UV) light irradiation has been reported to create highly reactive species by photochemical dissociation process,48,118-121 UV assisted nitridation method was thought to have the potential to lower the incorporation process temperature. In this chapter, I introduce low temperature nitrogen incorporation method us ing UV illumination and thereafter compare the interface properties, incorporation efficiency and electrical performance w ith thermal nitridation process. Experimental Detail HfSiOx films were deposited direct ly on p-type (100) pre-clean ed (1% HF solution and DI water rinse) Si substrates, by atomic layer deposition (ALD) using GENUS Lynx2TM at 300 oC. SiH[N(CH3)2]3 was used as Si precursor whereas Hf(NEtMe)4 (TEMAH) was employed as Hf precursor. The oxidizing agents for TEMAH and Si were O3 and H2O, respectively. After film deposition, three different nitrogen doping met hods were employed; sample 1 was prepared using low temperature therma l incorporation method (1-2 Torr NH3, 350 oC, 30 min), sample 2 was prepared under high temperature NH3 ambient (1-2 Torr NH3, 650 oC, 30 min) and sample 3 employed UV (184 nm radiation fr om UV lamp) assisted low temperature incorporation method (1-2 Torr NH3, 350 oC, 30 min). All the samp les underwent an in-situ oxidation at 400 oC in high purity oxygen at 300 Torr for 30 min with ultraviolet illumination from a low pressure Hg lamp array. This step en sured complete oxidation of the film and also its evaluation of resistance against oxidation. MOS capaci tors with Pt gate electrode were fabricated using RF magnetron sputtering. For all MOS de vices, post metallizati on annealing (PMA) was carried out in a tube furnace at 400 oC in forming gas (10% H2 / 90% N2) ambience for 30 min. 100

PAGE 101

The C V characteristics were measured at hi gh frequency (1 MHz) using an HP 4294A LCR meter. North Carolina State University (NCSU) program which consider quantum mechanical effects of thin gate oxide combined with non-lin ear fitting techniques, was used to analyze the electrical measurements and extraction of flat band voltage.121 Results and Discussion The chemical bonding characteristics of IL were analyzed by high resolution x-ray photoelectron spectrosc opy (XPS, Perkins Elmer 5100) using Mg K radiation (1253.6 eV). Figure 6-1. shows the XPS Si 2 p scan for the three sets of samples. The Si4+ peak of high temperature nitrided sample 2 and UV nitrided sample 3 undergoes a shift of ~ 0.7 eV to lower binding energy in comparison to the low temperatur e nitrided sample 1, suggesting a difference in bonding environment between the samples at the filmSi interface. The nitrogen amount which was analyzed by sensitivity factor (not shown) in the 350 oC UV nitrided film (sample 3) and 650 oC thermally nitrided film (sample 2) were 6.7 and 7.3 % respectively, whereas nitrogen in the low temperature nitridation process (sample 1) was low (~1 %). The Si4+ peaks of stoichiometric SiO2 and Si3N4 films are located at higher ener gies (103.6 eV for oxide and 101.7 eV for nitride) than the Si0 peak originating from the Si subs trate (99.3 eV in this experiment).122125 Low nitrogen incorporation efficiency of the low temperature thermal nitridation caused the increase of Si interaction with oxygen at film-Si interface, hence the peak shift of sample 1, away from Si-N and towards Si-O bonding, was observed. From the nitrogen amount and Si4+ peak position, it can be concluded that the nitrogen incorporation effici ency of UV assisted process at 350 oC was comparable to high temperatur e thermal nitridation process (650 oC without UV illumination). 101

PAGE 102

X-ray reflectivity (XRR) scan of Hf-silicate films are presented in figure 6-2. Since XRR scans were obtained using symmetric /2 configuration with very sma ll angles, this technique is suitable for investigation and characterization of thin films.105,106,125 For curve fitting (Wingixa software, Philips), a multi-layer model, which co nsists of the interfacial, Hf-silicate, and contamination layers, was applied. Film dens ity was determined by the critical angle ( c) in which the experimental deviation was about 3% in our study. Non-sp ecular diffuse scattering as well as specular scattering was considered to obtain the interface roughness values. The physical characteristic of Hf-silic ate dielectric such as thickness, density and roughness of the film and the IL as obtained from XRR measuremen t is tabulated in Table 6-1. The higher film and IL density of sample 2 and 3 is the result of higher efficiency of nitridation process. Similar changes in density were observed in earlier reports because of nitrogen incorporation and outgasing of hydrocarbons.105,106 The IL roughness of sample 2 (4 .54 ) is higher than that of UV assisted low temperature nitrided sample 3 (2.98 ). The IL thickness from XRR curve fitting of sample 2 and 3 were 16 and 10 resp ectively. It should be noted that even though the nitrogen contents and film density of sample 2 and 3 were similar, the interface thickness and roughness were different. The larger IL thicknes s and higher IL roughness of high temperature nitrided sample could be understood by the NH3 dissociation model and mass transport phenomenon. Nitrogen incorporation depends on i) su rface concentration (adsorption) of atomic nitrogen and ii) diffusion of nitrogen into the film. Surface concentrati on of nitrogen depends on dissociation of NH3 which can be enhanced by both temperature and UV illumination. The low level of nitrogen at low temperature and in th e absence of UV illumination (sample 1) is the result of lower nitrogen adsorption because of poor efficiency of NH3 dissociation. Increase in the nitridation temperature (sample 2) enhances NH3 dissociation and hence increases atomic 102

PAGE 103

nitrogen concentration but it also increases the diffusion of nitrogen and other species like oxygen in the film. The increased diffusivity of these species is seen in the increase of IL thickness and roughness. J. L. Gavartin et al reported atomic N, NH2 -, NH0 and hydrogen to be the final incorporated products of ammonia PDA process in the film.109 Of these products, NH2 and NH0 species can occupy oxygen vacancies and does not trap interstiti al oxygen, so during nitridation and oxidation process, oxygen species are able to diffuse to the substrate and form SiO-N interface. In contrast, UV illumination has a possibility to gene rate reactive species including atomic nitrogen more than thermally ni trided process because of excitations from UV photon energy.109,127 Atomic nitrogen can react with inte rstitial oxygen with an energy gain of 1~1.5 eV. 109 In addition, low thermal energy for oxygen diffusion and increase in possibility for oxidation state formation during UV assisted ni tridation process assists immobilizing fast diffusing oxygen. In order to investigate the depth uniformity of incorporat ed nitrogen, SIMS depth profiling method was used. SIMS depth profiles were acquired with a Perkin-Elmer PHI 6600 SIMS system using a 1keV cesium primary ion beam w ith negative secondary acquisition. Under Cs bombardment, nitrogen is detected as an ionized molecular species in combination with another atom from the matrix. With negative secondary acquisition, N was acquired as mass 43SiN-. Figure 6-3 shows the SiNand O depth profiles of sample 2 and 3. In the UV nitrided sample 3, nitrogen and oxygen were distributed uniformly in the film with nitrogen and oxygen slightly piling up at the IL as a result of diffusion process. However, in the case of thermally nitrided (650 oC) sample 2, due to the limited NH0 and NH2 states and high thermal energy for oxygen diffusion, the concentration of nitrogen fluctuat ed rather than being uniform whereas the oxygen concentration decreased inside the film compared to the surface region. As a result of non103

PAGE 104

uniform components distribution, there was a severe pile up of nitrogen and oxygen at the IL. Figure 6-4 shows the C-V comparison plot for the three sets of samples. As seen from the figure, the UV nitrided sample 3 and thermally (650 oC) nitrided sample 2 showed a substantial increase in overall capacitance in comparison to thermal nitrided (350 oC) sample 1. A maximum capacitance density of ~3.69 F/cm2 corresponding to an EOT of 9. 4 and a dielectric constant of ~22 was extracted for sample 3. The low temp erature nitrided sample 1 exhibited the lowest capacitance of ~1.37 F/cm2 corresponding to an EOT larger th an 20 The larger EOT in sample 1 can be attributed to the thick low perm ittivity IL due to the lo w nitridation efficiency. The IL thickness of sample 1 was around 20 (not shown) with Si-O predominant bonding character (figure 1). The high temperature (650 oC) nitrided sample 2 showed a capacitance value between that of sample 1 and 3. The flat band voltage shift (~ 0.4 V) observed in sample 2 in comparison to sample 1 and 3 can be unders tood by the result of ammonia incorporation. A large amount of NH2 and NH0 compared to sample 3 react with oxygen vacancies whose defect location in energy band is near or inside valence band of Si and is a major source of fixed charge. It has been reported that the fl at band voltage shift could be cured by post deposition annealing process in oxidation an d/or reduction process.128 However, additional IL degradation (IL growth and high interface state ; Dit) may not be avoidable in case of high temperature nitridation method.95,128 Conclusion In conclusion, the UV assisted low temperatur e nitrogen doping technique for hafnia gate dielectric film was investigated. The incorporat ion efficiency of UV assisted low temperature method was comparable to the high temperature (~650 oC) thermal nitridati on process. Moreover, due to the low process temperature and multiplicity of dissociation species, uniform distribution of nitrogen and oxygen, thinner interface layer (10 ), higher capacitance value (3.69 F/cm2) 104

PAGE 105

and smaller EOT (9.4 ) were observed in the case of UV assisted low temperature nitrided sample. 105

PAGE 106

Figure. 6-1. XPS Si 2p spectra comparison of Hf-silicate films with respect to the methodological approaches (350 C thermally nitrided sample 1, 650 C thermally nitrided sample 2, and 350 C UV nitrided sample 3). 106

PAGE 107

Figure 6-2. X-ray reflectivity (XRR) scans of thermally nitrided (samples 1 and 2) and UV nitrided (sample 3) films. Table 6-1 Summary of x-ray refl ectivity data on roughness, density and thickness of IL and Hfsilicate films. 107

PAGE 108

Figure 6-3. SIMS depth profile of thermally nitrided film (650 C) and UV nitrided film (350 C) after 400 C UV oxidati on (a) oxygen and (b) nitrogen. 108

PAGE 109

Figure 6-4. C V comparison plot of three set of samples with 400 C UV oxidation. 109

PAGE 110

CHAPTER 7 UV ASSISTED LOW TEMPERATURE NITRIDATION & POST DEPOSITION OXIDATION TECHNIQUE FOR HfO2 GATE DIELECTRIC An evaluation of a low temperature method (~400 oC) for synthesis of nitrogenincorporated hafnia gate dielectr ic has been reported. This met hod is based on metal film growth in ammonia ambient and subsequent oxidati on under ultraviolet irradiation. X-ray photoelectronic spectroscopy confirmed the presence of nitrided interface layer with a thickness of ~12 Equivalent oxide thickness values of around 11.5 and leakage current densities lower than 1-4A/cm2 at an operation voltage (-1 V) were achieved. The post deposition UV oxidation process was performed to check th e interface oxidation resistance. The interface growth rate showed that as the interface bonding characteristics changed from Si-N to Si-O predominant bonding system of nitrogen incorpor ated films, the activation energy for oxygen diffusion changed from 18.0 kJ/mol to 9.8 kJ/m ole and the activation energy of un-doped hafnia films was 2.3 kJ/mol in every growth region. Introduction Hafnium oxide has been receiving intense attent ion as a new gate dielectric material in advanced metal oxide field effect transistors (MOSFETs) due to its reasonable permittivity (~ 25), relatively large band gap (5.68 eV ) and thermodynamic stability with Si.12,92,93 However, a high temperature annealing of around 1000 oC, used in the conven tional dopant activation process, causes HfO2 film crystallization ( ~ 500 oC) and a lowk interface layer (IL) formation between the film and substrate.30,39,94,95 To suppress crystallization a nd the IL formation, nitrogen incorporation in the film has been studied by different resear chers using various deposition techniques. Previous studies reported that nitrogen incorporation into highk gate dielectric help to improve the thermal stability and forms a thinner interface layer.30,39,114,115 The most common method for nitrogen incorporation is a de position technique usi ng carbon based gaseous 110

PAGE 111

precursors such as Hf[N(CH3)2]4 and Hf[N(CH3)(C2H5)]4 followed by post deposition nitridation (rapid thermal annealing, NH3 ambient) under high temperatures ( > 650 oC). In addition, due to the large amount of oxygen related defects in the deposited films,116,129 high temperature ( > 700 oC) post deposition annealing (PDA) in oxidation/reduction ambient must be employed for defect curing.116,129 As a result of high temperature nitrid ation and oxidation processes, unexpected interface degradation was observed, which can po tentially hinder scaling the device down in size.116,117 As ultraviolet (UV) light ir radiation has been reported to create highly reactive species by photochemical dissociation process,48,118-120 UV assisted nitridation techniques could potentially lower the incorporation process temperat ure. Therefore, in this work, low temperature nitrogen doping during the metal film deposition and in-situ oxidation under UV illumination is introduced in order to prevent degradation caused by high temperature post deposition nitridation and oxidation processes. Based on the nitridati on method, the enhanced IL qualities (composition, electrical properties and thickness) are discussed. In addition, the nitrogen inco rporation effects on the UV assisted low temperatur e oxidation process (IL growth rate, compositional change and diffusion activation energy) are investigated. Experimental Hafnium metal films were deposited on MEMC p-type (100) pre-cl eaned (SC1, 1% HF solution and DI water rinse) Si substrates having a resistivity of 9-18 cm by pulsed laser deposition (PLD) at a temperature of 350 oC. A 254 nm wavelength KrF excimer laser was used to ablate a high purity hafnium target (>99.99% ). To compare the nitrogen doping efficiency under the low temperature condition, three sets of samples were prepared at the same process temperature of 350 oC; sample 1 was deposited in NH3 ambient (1-2 Torr) with UV irradiation, sample 2 was deposited in NH3 ambient without UV irradiation and sample 3 was deposited in 111

PAGE 112

high vacuum conditions (1-6 Torr). Next, in order to oxidize th e metal film, all of the samples then underwent an in-situ oxidation at 400 oC in high purity oxygen at 300 Torr for 30 min with UV illumination from a low pressure Hg lamp arra y. To compare the IL growth rate of each experimental sets, the PDA time period ( 30 to 120 min) and temperature (400 to 550 oC)were varied. The film and IL thickness was crosschecked by x-ray reflectivity (XRR) and high resolution transmission electron microscopy (H RTEM JEOL 2100). XRR (PANalytical Xpert system) was also used to determine the IL dens ity. The chemical bonding characteristics of the films and IL were confirmed by x-ray photoe lectron spectroscopy (XPS, Perkins Elmer 5100) using Mg K radiation (1253.6 eV). A ll the peaks were calibrate d with respect to C 1 s at 284.6 eV. In order to investigate n ear the interface bonding state, 90o take off angle were applied. In order to investigate the electrical characteristics of the deposited or annealed dielectric films, a metal-oxide-semiconductor (MOS) ca pacitor with Pt gate electr ode was fabricated using the shadow mask technique. The el ectrical properties of the film s were characterized by current density-voltage and capacitance-voltage measur ements using a Keithley instruments Win-82 system. The interface trap density Dit was calculated by th e conductance method.130 Results and Discussion UV Assisted Low Temperature Gr owth of Hafnia Film Figure 7-1 shows the Si 2 p scans for three sets of sample s after 30 minutes of oxidation processing. The Si4+ region of the peak in the UV assisted nitrided film (sample 1) undergoes a shift of approximately 0.5 eV towards a lower binding energy in comparison to the thermally nitrided film (sample 2) and the vacuum deposit ed film (sample 3), sugge sting a change in the bonding environment at the f ilmSi interface Because 90o take off angle can detect the interaction near the interface, it is believed that the Si peak posit ion variations originated from 112

PAGE 113

the interface bonding environmental changes. Previous analyses have reported that in the case of stoichiometric SiO2 and Si3N4 films, the Si4+ peaks are located at higher energy regions (103.6eV for oxide and 101.7 eV for nitride) than the Si0 peak (99.3eV in this experiment) which comes from the Si substrate.121-124 Hence, the shift in peak of the UV assisted nitrided sample, away from Si-N and towards Si-O bonding, is due to the interaction between Si-N atoms with oxygen at film-Si interface. Since the magnitude of the shift is not very large, the influence of oxygen involved in this interaction is assumed to be minimal. The nitrogen amount, analyzed using appropriate sensitivity factors (not shown) of the UV assisted sa mple 1, was around 7 % which is comparable to the high temperature process,[7] whereas for that of thermally nitrided sample 2 showed a lower content (~1 %). Since the nitridation efficiency is reduced at low temperatures, other energy sources e.g. UV ra diation which could assist the generation of reactive species should be used, in order to inco rporate nitrogen into the films at low temperatures ( below 350 oC). Cross-sectional HRTEM imaging was used to measure the interfacial layer thickness. Figure 7-2 shows the HRTEM micrographs of sample 1 a nd 3. The film thickness including the IL was approximately same for the samples, whereas th e IL thickness of the nitrogen incorporated sample (1.2 nm) was thinner than that of the vacuum deposited sample (2.1 nm). It should be noted that the reduced thickness of the IL in the UV assisted nitrided sample is comparable to the values obtained by high temperatur e assisted techniques such as atomic layer deposition by with rapid thermal annealing.114,115,125 In general, a reaction in the IL depends on the energetic species of the deposition system because the conditions affect the mixing of atoms and chemical reactions occurring near the substrate by influencing mome ntum transfer and diffusion.125,131 Therefore, the similar IL thickness with results of high temperature process indicate that the oxidation efficiency of UV oxidatio n process is comparable with high temperature process. Also, 113

PAGE 114

even though UV assisted oxidation has been shown to increase the growth of the IL due to the presence of the highl y active oxygen species,118,119 IL thickness of the UV nitrided sample 1 being less than that of non-ni trided vacuum deposited sample 3 can be understood by the Si-N bonding predominant IL causing high resistance to oxidation. Figure 7-3 (a) shows the C-V comparison plot for the three samples after 400 UV oxidation. The UV assisted nitrided sample 1 showed a substantial increase in overall capacitance in comparison with other samples. A maximum capacitance density of ~3.0 F/cm2 corresponding to an EOT of 11.5 and a diel ectric constant of approximately 21.7 were determined for sample 1. The non-nitrided samp le 3 exhibited the lowe st capacitance of ~1.15 F/cm2 corresponding to an EOT greater than 25 The larger EOT in non-nitrided sample can be attributed to the thick IL and to a low permittivity of undoped IL. The thermally nitrided sample 2 showed a capacitance and EOT value similar to sample 3 due to the low nitridation efficiency as described above. The estimated Dit values in the samples were 212, 811 and 311 eV-1cm-2 for UV assisted nitrided sample 1, thermally nitrided sample 2 and undoped sample 3 respectively. The high interface trap dens ity in the UV assisted nitrided sample is due to the positively charged nitrogen incorporation and high efficien cy of UV assisted nitridation processes. The increase in trap de nsity with nitrogen incorporation is consistent with the report by Fedorenko et al .49 The leakage current density for nitrid ed and non-nitrided samples is shown in Figure 3b. All the samples exhibited leakag e current densities less than 10-4 A/cm2 at -1 V gate bias. The low leakage density levels mean that th e leakage current of the nitrided sample was not seriously affected by the high inte rface trap density. From the f act that similar IL thickness (figure 7-2) and Dit value of sample 3 with reported high temperature HfO2/Si film grown using 114

PAGE 115

ALD and treated to a 1000 oC PDA process,49,132 it is believed that the oxidation efficiency of the UV assisted low temperature process was comp arable to the thermal oxidation process. UV Assisted Oxidation Resistance of Hafnia Film To study the effect of nitrogen on the PD A characteristics in UV assisted oxidizing ambient conditions, post oxidation temp eratures were varied from 400 oC to 550 oC. Figure 7-4 shows the IL thickness dependen ces of sample 1 and 3 on the oxidation time and temperature. The IL thickness was cross-checked by HRTEM and x-ray reflectivity measurements (XRR). The vacuum deposited samples showed relatively th icker interfacial layers and faster growth rates than the nitried samples. The oxide growth rate in the v acuum deposited film was 2.5~2.8 /min, and the growth rate in the nitrogen in corporated films was 0.5~1 /min in the same temperature range (400~550 oC). In the case of UV assisted nitrogen incorporated samples oxygen diffusion to the Si substrate was hindered. This is due to the Si-N bonded interfacial layer and the multiplicity of the NO oxidation state formation.109 It has been reported that diffusion was more suppressed in th e SiON network in the N-incorporated hafnia film than in the SiO network in the hafnia film without N incorporation.133 Therefore, there was a drastic increase in the IL thickness wi th oxidation time in non-nitrided sa mples due to the absence of the Si-N bonded IL. In order to investigate the IL density variation with respec t to the oxidation process, the IL densities of sample 1 and 3 from XRR analysis is shown in Figure 7-5. XRR has been reported to be an effective method to characterize highk materials since most highk materials have a density greater than Si and any inte rfacial reaction can be detected.134 The IL densities were 2.82 g/cm3 for the UV assisted nitrided film and 1.95 g/cm3 for non-nitrided film without post heat treatment. Since the density of amorphous SiO2 and Si3N4 film are known to be ~ 2.2 and 3.1 g/cm3, respectively, the higher IL density of as deposited nitrided film as compared to non115

PAGE 116

nitrided film can be attributed to nitrogen in corporation. As the UV assisted PDA time is increased, the IL density of both samp les exhibited similar values ~2.3 g/cm3 which are close to the amorphous SiO2 film density. It should be noted th at the film and IL experienced the densification process during PDA as in the case of sample 3 where the film and interface density increased with PDA time.105,106 However, in the case of UV assisted nitrided sample, the interface density decr eased with PDA process. The reduced interface density could be explained by change in the bonding environment with interface oxidation. In order to confirm the bonding environments of the IL after the UV assisted PDA process, the Si 2 p peak positions were analyzed by XPS. The Si 2 p peak position is shown as a function of oxidation time in Figure 7-6. The binding energy (BE) of the Si4+ nitrided sample which underwent 30 minutes oxidation was found to be 102.5 eV, further increas e in oxidation time to 90 and 120 minutes shifts the peak to 102.9 eV and 103.3 eV respectively(figure 6a). Since a large amount of oxygen species are consumed by the metal film oxidation, oxygen diffusion to the substrate can be assumed to be minimal dur ing the 30 minutes oxidation time. Hence, the Si4+ peak location represented a predominantly Si-N bonding system. The 102.5 eV peak position of 30 min oxidized sample is very similar with the peak position of ALD grown HfSiON film, indicating the Si-N predominant interface bonding characteristics.114,115,131 However, with increasing oxidation time, the oxygen diffusion to Si increased leading to the formation of Si-ON bonding thus causing the peak position to shift to higher energies. Finally, for the 120 minutes oxidation of sample 1, the Si4+ peak was spread and the peak position moved to high BE region indicating the existence of various Si related bonding with a dominant Si-O bonding system. In contrast, the non-nitrided samples did not exhibit peak shifts ( 103.5eV) with annealing periods, as shown in figure 6b, implying that the interfacial bonding charac teristic showed a predominant 116

PAGE 117

Si-O bonding environment from the initial stage of the PDA process. The separation between Si0 and Si4+ peaks after 120 minutes of oxidation was a bout 4.1-4.3 eV which is similar to the energy separation of SiO2 film grown by thermal oxidation.136 It also ascertains that predominant Si-O bonding environment after 120 min oxidation. Since the UV illumination process generates highly energetic oxygen species, the IL oxidation process can be assumed to be an oxygen diffusion limited process and the oxygen diffusivity would be affected by film composition and interface layer properties.116 Because incorporated nitrogen has multiple state with oxygen, nitrogen incorporated hafnia film In order to obtain the activation energy for oxygen diffusi on with interface bonding characteristics under UV oxidation ambient conditions, Arrhenius plots for interfacial growth with different time ranges were plotted. Linear fits of the data were then used to ob tain the activation energies(figure 7-7 (a)). Figure 7-7 (b) shows the comparison of activation energies of sample 1 and 3 with respect to oxidation times. Region 1 refers to the oxidation time range between 30 and 60 min, the activation energy of the UV assisted nitr ogen-doped sample was 18 kJ/mol, whereas the activation energy of vacuum deposited sample wa s 2.3 kJ/mol. The UV assisted nitrided samples had a higher activation energy because of the Si -O-N bonding character of IL and the tendency of nitrided hafnia to trap oxygen.117 Region 2 refers to the oxid ation period between 90 and 120 min, since the IL characteristics changed to th e predominantly lower density Si-O bonding phase, the activation energy of nitrogen doped sample was decreased to 9.8 kJ/mol. However, due to the consistent bonding characteristics of the IL, the activation energy of the non-nitrided samples did not vary with oxidation time. Conclusions The UV assisted low temperature metal oxidati on technique for nitroge n doped hafnia gate dielectric films was investigated. At low temper atures, thermal nitridation was not an effective 117

PAGE 118

method for nitrogen incorporation due to the deficiency of energeti c nitrogen species. The amount of nitrogen in the film showed around a 6 % difference between UV assisted nitrogen incorporation and the thermal inco rporation process. In the case of UV assisted nitrided hafnia film, relatively dense IL comp rising of predominantly Si-O-N bonding was formed which caused enhanced electrical and IL oxida tion resistant propert ies. The non-nitrided samples had higher interface layer growth rates a nd lower activation energies fo r oxygen diffusion due to the predominately low density Si-O interface layer. Finally, the activation energy of the oxygen diffusion of UV nitrided sample was not constant within the ox idation process, but decreased with increasing amount of low density Si-O bonding in the interface layer. 118

PAGE 119

Figure 7-1. XPS Si 2p spectra comparison of nitrided samples (sample 1 and 2) and nonnitrided sample 3. 119

PAGE 120

Figure 7-2. Cross-sectional HR-TEM images. (a) non-nitrided sample 3 (b) UV assisted nitrided sample 1. Figure 7-3. Comparison of electrical properties of nitrided samples and non-nitrided sample (a) C-V comparison (b) J-V comparison 120

PAGE 121

Figure 7-4. The interfacial la yer thickness changes with respect to the PDA times for sample 1 and 3. 121

PAGE 122

Figure 7-5. The interfacial layer density change s with respect to the PDA times for sample 1 and 3. Figure 7-6. XPS Si 2p spectra comparison with respect to the PDA times (a) UV assisted nitrided sample 1 (b) non-nitrided sample 3. 122

PAGE 123

Figure 7-7. Arrhenius plots for interfacial gr owth for UV assisted nitr ogen incorporated films and vacuum deposited films(a). Compar ison of activation energies for oxygen diffusion in various oxidation times for sample 1 and 3(b). 123

PAGE 124

CHAPTER 8 AN EVALUATION OF COMPATIBILITY FO R TI BASED METAL GATE ELECTRODE ON HF-SILICATE DIELECTRICS FOR DUAL METAL GATE APPLICATIONS An evaluation of Ti based gate metals (Ti, TiN and TiB2) on Hf-silicate gate dielectric prepared by atomic layer deposition method ha s been reported. The effective metal work function, calculated by taking an interface layer an d interface charge into consideration, were 4.27, 4.56 and 5.08 eV for Ti, TiN and TiB2, respectively. Regardless of gate electrodes, the conduction mechanism of the samples fitted with Poole-Frenkel model which is related to oxygen vacancies in the film. Ti gate electrode was found to be more favorable for NMOS device and TiB2 gate electrode can be used for PMOS with Hf-silicate dielectrics. Introduction Many highk materials are currently being consid ered as potential replacement of SiO2 based dielectrics in future compleme ntary metal oxide semiconductor (CMOS) technology.12,30,92-95 Among the many candidate materials, nitrogen incorporated hafnium silicate has been receiving intense atten tion as a new gate dielectric material due to its reasonable permittivity (15~25), relatively large band gap (5.68 eV) and thermodynamic stability with Si. 4,12,30,92-99Nitrided Hf-silicate is also more resistant to boron diffusion from poly-Si gate through the dielectric.100 However, the incompatibility between poly-Si gate electrode and Hf based dielectric such as poly Si depl etion effect, Fermi ener gy pinning and sheet resistance constraint, limit its usefulness in advanced CMOS de vice especially beyond 45 nm technology node.5,136 Therefore, advanced high-performance device s require high dielec tric constant (high-k) gate dielectrics and dual work function metal gate electrodes.136 Dual metal gate CMOS integration requires two different metals with work functi ons near the Si band edges, around 4.0 eV for NMOS and 5.0 eV for PMOS.136 Consequently, the evaluation of various single metal and/or 124

PAGE 125

metal alloys on nitrogen incorporated Hf-silicat e dielectric could be highlight for future advanced CMOS technology development. Because of good thermal stability and minimal interaction with highk films, titanium based materials could be an attractive candidate for metal gate application.137,138 Ti based refractory materials such as TiN and TiB2 have been previously cons idered as a diffusion barrier in metal contact formation due to its high electri cal conductivity and excelle nt chemical inertness at high temperature.139 In this chapter, the electrical ch aracteristics of MOS devices with Ti based metal gate (Ti, TiN and TiB2) on nitrogen incorporated Hf-silicate dielectric are investigated. The effective metal work functi ons are extracted in order to examine its compatibility with NMOS or PMOS devices. In addition, the conduction mechanism of Ti based metal/Hf-silicate dielectric/ p-type Si MOS device is investigated. Experimental Detail HfSiOx films were deposited directly on MEMC p-type (100) pre-cleaned (SC1, 1% HF solution and DI water rinse) 8 inch Si substrates with a resistivity of 3-25 cm, by atomic layer deposition (ALD) using GENUS Lynx2TM tool at 300oC. SiH[N(CH3)2]3 was used as Si precursor where as Hf(NEtMe)4 (TEMAH) was employed as Hf precursor.19 The oxidizing agents for TEMAH and Si were O3 and H2O, respectively. HfSiOx film deposition was followed with nitrogen incorporation by rapid thermal anneal at 650 oC in NH3 ambience for 60 s. MOS capacitors with Ti, TiN and TiB2 gate electrode were fabricated using RF magnetron sputtering. All of the metal/compounds were deposited by Ar plasma assist ed rf sputtering at a pressure of 15 mTorr and 150W rf (13.56 MHz) power. For all MOS devices, post metallization annealing (PMA) was carried out in tube furnace at 400oC in forming gas (10% H2 / 90% N2) ambience for 30 min. The C V characteristics were analyzed at hi gh frequency 1 MHz using an HP 4294A LCR meter and NCSU program was used to analyze the electrical measurements and extract the 125

PAGE 126

flat band voltage. The gate current density-voltage (J-V ) measurements were performed using HP4156B. Results and Discussions Figure 8-1 shows the C-V characteristics of the MOS device with Ti, TiN and TiB2 gate electrode on 7 nm Hf-silicate dielectric. Th e accumulation capacitance values are almost identical (~2.0 10-9 F and EOT = 2.7 nm) regardless of ga te electrodes. An excellent Ti/HfO2 interface with little effect on flat band voltage ( Vfb) has been previously reported.19 The identical capacitance values and parallel shift of C-V curves for TiB2 and TiN gate electrodes indicates a low interface state density similar to Ti gate.138,140 The Vfb for Ti, TiN and TiB2 extracted by NCSU simulator, were -0.7, -0.25 and +0.24 V, resp ectively. It is expected that the difference in Vfb between the Ti based gate electrodes is cause d by the difference in effective metal work function (meff).138,140,141 The meff on high-k dielectric can be ex tracted using the relationship between Vfb and EOT, expressed as 140-142 (8-1) EOT QQ EOT Q qVSiO i O ff2 20 0 subiihk Si ihk me sub fb// / 0 and where 2SiO are permittivity of vacuum and dielectric constant of SiO2, respectively sub is the work function of substrate (in this work 5.01 eV) and Q is interface charge. Qhk/i and Qi/sub are defined as interface charges between highk and interfacial layer and between interfacial layer and substrate, respectively. EOTi represents the EOT of interfacial la yer. Two underlying assumptions are the existence of an interfacial layer between highk and substrate, and the charges at interface being significantly higher th an bulk charge of Hf-silicate and interfacial layer.140 Since forming gas anneal passi vates the dangling bonds of SiOx and Hf-silicate and also reduces interface charge between interface layer and substrate ( Qi/sub < 1 1012 cm-2)140,143, it can 126

PAGE 127

be assumed that This assumption makes the extraction of Qhk/i/0SiO2 from the slope of ( qVfb + sub) vs EOT plot possible. Furt her, by using EOTi value, meff. can be determined. ihk subiihkQQQ/ // Figure 8-2 shows the plot of ( qVfb + sub) vs EOT along with linear fitting analysis for the different gate electrodes. Inse t figures shows thickness of high-k ( thighk) plotted against EOT for determination of EOTi. Thickness of the highk film was cross checke d by X-ray reflectivity (XRR) measurement and high resolution transm ission electron microscopy (HRTEM). The intercept and slope of linear fit to the EOT vs thighk plot was used to obtain EOTi (1.1nm) and dielectric constant (16.6) respectively. meff values calculated using eq (1) for Ti, TiN and TiB2 gate were 4.27, 4.56 and 5.08 eV, respec tively. It has been reported that Qi/sub shows a weak dependence on meff ; the meff values changed by onl y 0.7 % within the Qi/sub variation range of 0 to 1012 cm-2.140,142 Therefore, our extracted meff of Ti, TiN and TiB2 gate on Hf-silicate film with consideration of an interface layer (EOTi) and interface charge ( Qhk/i) could be believed as acceptable. The meff value of Ti and TiB2 is consistent with repor ted bulk work function (Ti : 4.32 eV, TiB2 : 5.01 eV).138,140 Therefore we believe th at in the Ti and TiB2 gate on Hf-silicate dielectric, Fermi energy pinning effect is not serious. It is reported that meff of TiN vary with respect to Ti/N ratio from 4.2 to 4.8 eV.144 The meff of TiN gate in our de vice is believed within reasonable range. From the extracted meff values, it is worth to note that the Ti and TiB2 gate electrode with Hf-silicate di electrics can be potential candidates for NMOS and PMOS, respectively. However, to addre ss the possibility of Ti based electrodes for device application, further study is necessary to understand the nature of the meff after high temperature annealing. 127

PAGE 128

The J-V characteristics of the three different Ti based electrodes on Hf-silicate film are illustrated in figure 8-3. All samples exhibi ted leakage current densities lower than 10-4A/cm2 at a gate bias of -1 V. It should be noted that the leakage current densi ties were not seriously affected by electrodes at normal operation range s (-0.5 ~ -2 V). To verify the conduction mechanism of Ti based electrodes on Hf-silicate film, J-V analysis at various temperature (25 ~120 oC) was conducted. Inset in figu re 3 shows Poole-Frenkel (ln( J/E ) vs E1/2) plot 145,146 of the three kinds of Ti base electrodes on Hf-silicate at 25 oC. The linearity of Poole-Frenkel (PF) plot for all samples in the high field region indi cated a PF conduction behavi or for leakage current. Also, the high frequency dielectric constant extracted from the slope of the PF plot was 4.3. Refractive index (n) is given by the square root of optical dielectric constant.145,146 Refractive index of our dielectric film was found to be 2.07 from extracted high frequency dielectric constant which is within the range of refractive index (1.8~2.4) of Hf-O-N films.36 It also ascertains that the conduction of Ti based electrodes on Hf-silicate dielectric is governed by the PF conduction mechanism. Figure 8-4 shows the trap depth of three Ti based gates on Hf-silicate dielectric. From the extrapolation of the slope of ln( J/E ) vs 1/T plot at di fferent gate voltage (inset in figure 4), activation energy of conduction can be obtained.145,146 Since the PF conduction takes place by thermionic emission of electrons from trap site in the film, activati on energy of conduction is directly related to the trap depth in the film. Trap depth of Ti, TiN and TiB2 gate electrodes on Hf-silicate film shows almost similar values of 1.64, 1.7 and 1.74 eV, respectively. Trap depths can be correlated to the oxygen vacancy defects in the Hf-silicate film which were located at ~1.6 eV below the conduction edge and within the Si band gap region.110,128 These result mean 128

PAGE 129

conduction in these sample are mainly affected by Hf-silicate film property and not by the type of electrodes. Conclusions MOS device with Ti, TiN and TiB2 gate electrode on Hf-silicate were examined. Considering EOTi and Qhk/i, the extracted meff values for Ti, TiN and TiB2 gate were 4.27, 4.56 and 5.08 eV, respectively. Regardless of gate electrodes, the conduction mechanism of the samples well fitted with Poole-Frenkel conduc tion model and conduction activation energies showed almost identical values (~ 1.7 eV). Ther efore, Ti gate electrode is more favorable for NMOS device and TiB2 gate electrode can be used for PMOS with Hf-silicate dielectrics 129

PAGE 130

Figure 8-1. C V characteristics of the MOS device with Ti, TiN, and TiB2 gate on Hf-silicate film at 1 MHz. 130

PAGE 131

Figure 8-2. The (qVfb + sub) vs EOT plots for the Ti, TiN and TiB2 gates. The EOTi and dielectric constant were determined by EOT vs Thighk plot inset in figure. 131

PAGE 132

Figure 8-3. J-V plot for Ti, TiN and TiB2 gates on Hf-silicate dielectric. Inset in figure shows Poole-Frenkel (ln(J/E ) vs E1/2) plot of three kinds of Ti based electrodes on Hfsilicate at 25 oC. 132

PAGE 133

Figure 8-4. The extracted trap de pth of MOS devices with three kinds of gate electrode using slope of ln( J/E ) vs 1/T plot at different gate volta ge which were inset in the figure. 133

PAGE 134

CHAPTER 9 AN EVALUATION OF THERMAL STABILITY OF TIB2 METAL GATE ON HF-SILICATE FOR P-CHANNEL METAL OXIDE SEMICONDUCTOR APPLICATION An evaluation of TiB2 gate metal on Hf-silicate dielectric prepared by atomic layer deposition method has been reported. The extr acted effective metal work function for TiB2 gate was about 5.08 eV. The work function showed almost identical values and sharp interface between metal and dielectric was confir med after post deposition annealing by 1000 oC. The work function lowering (4.91 eV) at 1100 oC was caused by metal-dielectric intermixing and oxygen vacancy formation. TiB2 gate electrode was found to be suitable for use in p-channel metal oxide semiconductor device. Introduction Nitrogen incorporated hafnium silicate has b een receiving intense attention as a gate dielectric material for complementary metal oxide semiconductor (CMOS) devices.4,12,92-100 However, the incompatibility between poly-Si gate electrode and Hf based dielectric such as poly Si depletion effect, Fermi energy pinning and sheet resistance constraint, limit its usefulness in advanced CMOS device especially beyond 45 nm technology node.5,76 Therefore, advanced high-performance devices require high dielectric constant (highk) gate dielectrics and metal gate electrodes. Since p-type MOS in tegration requires metal with wo rk function near the Si band edge, close to 5.0 eV, PMOS metal candidates are limited. Potential candidates such as Ru, RuO2, and Mo have been studied for PMOS application. However, Ru electrode on SiO2 film showed complete intermixing at high temperature (>950 oC). Physical and electrical instability of Mo and RuO2 gate on highk material after high temperature annealing (>850 oC) has also been reported. Consequently, the evaluation of various single metal or metal alloys on nitrogen incorporated Hfsilicate dielectrics after high temperature post deposition annealing (P DA) process could be highlight for future advanced CMOS technology development. 134

PAGE 135

Because of thermal stability and chemical in ertness, titanium based refractory materials could be an attractive candidate for metal gate application.137-139 In the Chapter 8, the possibility of TiB2 gate electrode as a PMOS application has been proposed.147 However, to address the possibility of TiB2 electrode for device application, the thermal stability of TiB2 electrode on Hfsilicate and nature of metal work function after high temperature annealing should be investigated. Therefore, in this chapter, the electrical characteristics of MOS devices with TiB2 metal gate on nitrogen incorporated Hf-silicate di electric for PMOS application are investigated. In addition, the stability of el ectrical and physical properties with respect to the post deposition annealing (PDA) temperature is examined. Experimental Detail HfSiOx films were deposited dire ctly on p-type (100) pre-cleaned (SC1, 1% HF solution and DI water rinse) 8 inch Si subs trates with a resistivity of 3-25 cm, by atomic layer deposition (ALD) using GENUS Lynx2TM at 300oC. SiH[N(CH3)2]3 was used as Si precursor where as Hf(NEtMe)4 (TEMAH) was employed as Hf precu rsor. The oxidizing agents for TEMAH and Si were O3 and H2O, respectively. HfSiOx film deposition was followed with nitrogen incorporation by rapid thermal anneal at 650 oC in NH3 ambience for 60 s. MOS capacitors with TiB2 gate electrode were fabricated us ing RF magnetron sputtering. Sintered TiB2 ceramic target was used for TiB2 metal deposition. To minimize the external resistance, a metallization scheme of Au(50nm)/Ti(10nm) was used on TiB2 films. All of metals and metal alloy were deposited by Ar plasma assisted rf sputtering at a pressure of 15 mTorr and 150W rf (13.56 MHz) power. To verify the thermal stability, post depos ition annealing (PDA) temperature was varied from 600 to 1100 oC in N2 ambience with 30 sec duration time. For all MOS devices, post metallization annealing (P MA) was carried out in tube furnace at 400oC in forming gas (10% H2 / 90% N2) ambience for 30 min. Auger el ectron spectroscopy (AES) depth 135

PAGE 136

profile analysis was conducted using a Physic al Electronics 660 Sca nning Auger Microprobe. The electron beam conditions were 10 keV, 1 mA beam current at 45o from sample normal. The C V characteristics were anal yzed at high frequency 1 MHz using an HP 4294A LCR meter and North Carolina State University (NCSU) program was used to analyze the electrical measurements and extract the flat band voltage ( Vfb). The gate current density-voltage ( J-V ) measurements were performed using HP4156B The effective metal work function (meff) was extracted using Vfb EOT method.140-142 The interface state densities ( Dit) were extracted by the method of Lehovec.105,106 Results and Discussion Figure 9-1 (a) shows the C-V characteristics of the MOS device with TiB2 gate electrode on 7 nm Hf-silicate dielectric after different PDA treatments temperatures. The accumulation capacitance and flat band voltage ( Vfb) are almost identical (~2.0 10-9 F, EOT = 2.7 nm and Vfb = 0.24 V) for samples annealed up to 1000 oC, whereas, the C-V curve of sample annealed at 1100 oC showed lower capacitance than others and di splayed large parallel shift to negative bias region ( C ~ 1.2 10-9 F and Vfb = 0.13 V). An excellent TiB2/Hf-Silicate interface of asdeposited sample with little effect on flat band voltage ( Vfb) has been reported in our previous work.19,20 The identical cap acitance for TiB2 gate electrodes after different PDA treatment indicates minimal reaction betw een electrode and dielectric.138,140,147 The stretched curve of 1100oC annealed sample indicates the in crease in interface state density ( Dit). The interface state densities at flat band voltage s howed almost similar value (~5.5 1011 eV-1cm-2) up to 1000oC annealed device. Whereas the increase in Dit (~11.3 1011 eV-1cm-2) of 1100oC annealed sample was measured. This lower capacitance and higher Dit at 1100oC annealed sample are attributed to interlayer growth which we can confir m in Fig. 3. The variations of ( qVfb + sub) with respect to 136

PAGE 137

the EOT after high temperature annea ling is presented in figure 1b (sub is the work function of substrate in this work 5.01 eV). The Vfb for TiB2 metal gate did not change up to 1000 oC anneal. However, for 1100 oC annealed TiB2 metal gate, a major shift of Vfb and an increase in slope of ( qVfb + sub) vs EOT plot compared with other samples indicated change of metal work function and increase in fixed charge at IL. From the ( qVfb + sub) vs EOT plot and EOT vs thickness of highk plot (not shown), we can extrac t effective metal work function (meff) with consideration of IL and fixed charge. These values ar e presented in inset of figure 9-1 (b). meff of as deposited, 700, 900 and 1000 oC annealed TiB2 gate metal showed almost identical values (~5.07 eV). However, a lower value of meff (~4.91 eV) for 1100 oC annealed TiB2 gate metal was found. A possible reason for decrease in work function after 1100 oC anneal could be instability of TiB2 alloy after 1100 oC exposure. Hence, it can be expected that degradation of capacitance values and meff was caused by intermixing of metal gate and dielectric at 1100 oC annealing. In order to check the interm ixing of metal and Hf-silicate, AES analysis was conducted. The AES depth profiles from the as deposited, 900 and 1100 oC annealed samples are shown in figure 9-2. The as-deposited sample shows shar p interfaces between the metal and Hf-silicate. Till 900 oC annealing, there is no significant movement of Ti or B through the underlying layers and metal/dielectric maintained shar p interface as in the cas e of as-deposited sample even if part of Ti diffused out to the surface. TiB2 appears to be a barrier fo r Ti diffusion, which probably excludes formation of TiN and TiSi phase s at the interface of the Hf-silicate.139 However, after 1100 oC annealing, metal and dielectric film interm ixed and there is an accompanying decrease in abruptness of the TiB2 interface with the Hf-sili cate film. The lowering of meff (4.91 eV) can 137

PAGE 138

be explained by intermixing causi ng formation of TiN and TiSi pha ses at the interface after 1100 oC annealing. HRTEM image of TiB2 gate electrode on Hf-silicate film of as-deposited and 1100 oC annealed samples is shown in figure 9-3. The sh arp interface between metal and Hf-silicate was observed in both case and no trace of serious intermixing can be found from the HRTEM images. However, there is one distinguished feature which should be noted. The IL between Hf-silicate and Si of 1100 oC annealed sample is thicker than IL of as-deposited sample. The thicker IL of sample annealed at 1100 oC was caused by oxygen diffusion to th e substrate, in the process generating oxygen vacancies in the dielectr ic film. The oxygen vacancy generation and subsequent electron transfer to the electrode could be a reason for the Fermi level pinning. The thick low-k IL also decreas ed the capacitance of 1100 oC annealed sample. Therefore, we believe that the Fermi level pinning of the TiB2 refractory gate electrode was caused by two different but simultaneous effects; i) material transfer causi ng intermixing between metal and Hf-silicate film confirmed by AES analysis, ii) th e transfer of electrons generated from the oxygen vacancies in the Hf-silicate film to the metal gate. The work function lowering after high temperature annealing was also reported in si ngle metal gate electrode case (4.7 eV for Ru and 4.63 eV for Ir).148 However, since the Fermi level pinning position of single metal gate were closer to Si midgap region, we believe TiB2 is more stable under high te mperature process condition. The J-V characteristics of the TiB2 gate electrodes on Hf-silicate film at electric field of 1MV/cm with various PDA temperatures and EOT is illustrated in figure 9-4. All samples exhibited leakage current densities lower than 10-4A/cm2 and abrupt degradation of gate leakage current is not observed in all temperature and EO T ranges. This fact indicates that the gate leakage current propertie s are not significantly affected by intermixed element and increased Dit 138

PAGE 139

value at 1100oC annealed device. On the contrary, the J-V characteristics of MOS device after 1100 oC annealing is about 1 order of magnitude lo wer than that obtained after low temperature annealing. The lower leakage current density of sample annealed at 1100 oC is possibly due to the reduction of hole current caused by Fermi level pinning 149 and thick IL between Hf-silicate and Si substrate. Inset figure shows C-V hysteresis (VH) of TiB2/Hf-silicate/Si MOS capacitors as a function of annealing temperature. As annealing temperature increased, the VH shows continuously decrease ra nged in 30~40 mV. The VH could be reduced (< 30 mV) with hightemperature anneals (1100C) due to the thicker interfacial layer limiting the electron injection in the Hf-silicate dielectric film.150 Conclusions The key points of our work may be summarized as follows: (i) The extracted meff values for TiB2 gate was almost identical (5.07 eV) till 1000 oC annealing treatments. (ii) The meff lowering phenomena (4.91 eV) was observed when annealed at 1100 oC, due to the metal and Hf-silicate intermixing by material transfer and electron transfer by oxygen vacancy generation. (iii) The leakage current density of TiB2 electrode MOS device were lower than ~ 10-4 A/cm2 and VH is reduced after 1100 oC annealing.. (iv) The possibility of TiB2 gate metal for the PMOS application was confirmed. The long term stability of boride and process co mpatibility remains to be established. 139

PAGE 140

Figure 9-1. (a) C-V characteristics at 1MHz of the MOS device with TiB2 gate on Hf-silicate film with at different annealing temperatures, (b) The qVfb + sub vs EOT plots for the TiB2 gate after different annealing. meff with consideration of interface is inset in figure. 140

PAGE 141

Figure 9-2. AES depth profiles of Au/Ti/TiB2/Hf -silicate/Si MOS struct ure as a function of annealing temperature. 141

PAGE 142

Figure 9-3. HRTEM images of TiB2/ Hf-silicate (7 nm)/Si stack at as-deposited state and after annealing at 1100 C under N2 atmosphere. 142

PAGE 143

Figure 9-4 (a) EOT vs leakage current density plot at 1MV/cm for TiB2 gates on Hf-silicate dielectric of different annealing temperatures. Inset figure is hysteresis ( VH) of TiB2/Hf-silicate/Si MOS capacitors as a function of annealing temperature. 143

PAGE 144

CHAPTER 10 STRAIN INDUCED CHANGES IN GATE LEAKAGE CURRENT AND DIELECTRIC CONSTANT OF NITRIDED HF-SILICATE DIELECTRIC SILICON MOS CAPACITORS Uniaxialmechanicalstrain altere d gate leakage current and diel ectric constant of silicon metaloxidesemiconductor (MOS) device with nitrid ed Hf-silicate gate dielectric are measured. Uniaxial stress is applied using fourpoint wafer bending along [110] direction. The gate leakage current and dielectric constant are found to increase by up to ~2 % under tensile and compressive stress direction. The decrease in interface trap activation ener gy is used to explain leakage current increment under mechanical stress. Doped n itrogen in Hf-silicate dielectric film occur N p band splitting resulting increase in electronic polarization and d ecrease in dielectric constant. Introduction Hafnium silicate has been extensively studied as a gate dielectric material in advanced metal oxide field effect tran sistors (MOSFETs) due to high crystallization temperature, thermodynamic stability with Si, high permittivity and relatively large band gap (5.68 eV). 4,12,92100 Nitrogen incorporated Hf-silicate (HfSiON) has received attention since HfSiON minimizes interface layer formation and reduces boron diffusion from the poly-Si gate.92-100 Also, recent reports have addressed additional advantages of Hf SiON dielectric such as increase in dielectric constant and improvement of reliability.151,152 Therefore, HfSiON is among the top candidates to be introduced as first generation highk gate dielectric beyond the 45 nm era.151,152However, due to the large remote phonon scattering, channe l mobility degradation of MOSFET device with highk is unavoidable.153 In order to compensate the mobility degradation, strain engineering has been introduced and explored as a promising technology.154 Recently, Thompson et al have shown that mechanical stress induced cha nnel mobility of MOSFET device with Hf-silicate dielectric is similar to intrinsic ch annel mobility of MOSFET device with SiO2 gate 144

PAGE 145

dielectric.155,156 However, to date, there is no systematic and quantitative result about strain effect on dielectric constant and gate leakage current of HfSiON dielectric film. In this work, we report, the effect of uniaxial stress on gate leakage current and dielectric constant of Si MOS capacitors with HfSiON ga te dielectric using cont rolled external applied mechanical stress by 4point bending method. Experimental Detail HfSiOx films were deposited direct ly on p-type (100) pre-clean ed (1% HF solution and DI water rinse) 8 inch Si substrates with a resistivity of 3-25 cm using atomic layer deposition (ALD) at 300oC. SiH[N(CH3)2]3 is used as Si precursor where as Hf(NEtMe)4 (TEMAH) is employed as Hf precursor. The oxidizing agents for TEMAH and Si are O3 and H2O, respectively. HfSiOx film deposition is followed by nitrogen incorporati on by rapid thermal anneal (RTA) at 650 oC in NH3 ambience for 60 s. MOS capacitors with Pt and Al gate electrode are fabricated using RF magnetron sputtering. All of the metals are deposited by Ar plasma assisted rf sputtering at a pr essure of 15 mTorr and 150W rf (13.56 MHz) power. For all MOS devices, post metallization annealing (PMA) was carried out in tube furnace at 400oC in forming gas (10% H2 / 90% N2) ambience for 30 min. Capacitance-voltage ( C-V ) and current-voltage ( JV ) measurements are performed with Agilent E4980A and Keithley 4200, respectively, using controlled external mechanical stress along [110] direction.157 To extract the frequencyindependent device capacitance valu e and eliminate the effect of both series and shunt parasitic resistances, two-frequency method is adopted.158 All C-V measurements are evaluated by NCSU program.159 To reduce the electrical instability from HfSiON charging 160, constant gate voltage (-1V) is applied for 160 s. The reduced instab ility allows monitoring of the strain-induced change in gate leakage current. 145

PAGE 146

Result and Discussion Figure 10-1 shows the C-V (a) and J-V (b) characteristics of Hf-silicate dielectric MOS capacitors. In order to verify the gate tunnel ing conduction mechanism, aluminum (Al) and platinum (Pt) electrodes are used due to their large bulk work function difference (Al 4.1 and Pt 5.8 eV). The accumulation capacitance values are almost identical (~2.0 10-9 F and EOT = 2.7 nm) regardless of the gate electrodes.161 The identical capacitance values and parallel shift of C-V curves for Pt and Al gate electrodes indicates minimal interaction between metal gate and dielectric film. Due to the difference in the wo rk function of Pt and Al flat band voltage ( VFB) shift and larger leakage current of Pt device are observed. Th e increase in leakage current density of sample with Pt electrode results from the dominant hole tunne ling from substrate149 and higher built-in oxide field of Pt device und er negative gate bias condition, as illustrated in the inset of figure 2. Figure 10-2 shows th e Pool-Frenkel (ln( J/E ) vs E1/2) plot of the two kinds of metal electrodes on Hf-silicate. The slope of PF plot can be expressed as follows; 0 r)( )/ln( )( q kT q E EJ Slope (10-1) where, k, T )( r0and are the Boltzmann constant, temper ature, high frequency dielectric constant of the insulator under external stress, and vacuum permittivity, respectively. The high frequency dielectric constant extracted from th e slope of the PF plot was 4.9. Refractive index (n) is given by the square root of optical dielectric constant.145,146 Refractive index of our dielectric film is found to be 2.23 from extract ed high frequency dielectric constant which is within the range of refractive index (1.8~2.4) of Hf-O-N films.76 146

PAGE 147

The strain altered gate leakag e currents for Al and Pt devices are shown in figure 10-3. Regardless of electrodes, leak age current density in 2.7 MV/c m gate field increase under both tensile and compressive stresses. Recently, Choi et al have suggested that interface trap activation energy of SiO2/Si interface decreased with both te nsile and compressive stress causing increase in trap-assisted gate leakage current.156 Since Si-O predominan t bonding characteristics at Hf-silicate/Si interface has been reported 114-118, similar explanation can be applied to Hfsilicate dielectric. Si-O predominant inte rface bonding mainly contribute trap-assisted conduction in gate leakage current. Therefore, the applied strain changes the interface trap activation energy, resulting in an increase in leakage current. The relative changes in strain altered dielectric constant of HfSiON, HfSiOx(nitrogen undoped Hf-silcate) and HfO2 film are shown in Figure 10-4. Unlike HfO2 and HfSiOx, the dielectric constant of HfSiON increases with both tensile and compre ssive stresses, crosschecked by C-V curve and slope extraction from PF plot We believe that the incorporated nitrogen in Hf-silicate film can be the origin of the strain-induced dielectric constant change. Even though an exact explanation is not provided yet, strain induced N p band splitting is a possible explanation. The N p band splitting by external strain increases the electronic transition from the N p band to conduction band, which lead s to band gap narrowing of HfSiON.160,161 The strain-induced band gap narrowing affects electronic polarization, which related with high frequency dielectric constant, resulting in an increased dielectric constant of HfSiON.161 Conclusion Effect of mechanical stress on silicon me tal-oxide-semiconductor capacitor with nitrided Hf-silicate dielectric has been studied. The gate leakage current and dielectric constant was found to be increase by up to ~2 % of both tensile and compressive stresses. Increase in leakage current came from decreased HfSiON/Si interface trap activation ener gy. The fact that dielectric 147

PAGE 148

constant variations under stress showed similar behavior by two different methods indicated the increase in dielectric constant could be cause d by increase in electroni c polarization from N p band splitting. Measurements of gate leakag e current and dielectri c constants of HfSiON dielectric film under two different mechanical stress conditions provides quantitative results and reveals that the incorporated nitrogen in the Hf-sil icate film is a origin of dielectric constant variation. 148

PAGE 149

Figure 10-1. C-V characteristics at 1MHz of the MOS de vice with Pt and Al gate on nitrided Hf-silicate film (a). The inset shows current density-voltage ( J-V ) measurements of both devices (b). 149

PAGE 150

Figure 10-2. Poole-Frenkel (ln( J/E ) vs E1/2) plot of Pt and Al gate on nitrided Hf-silicate film at 25 oC. Inset in figure shows a schematic band diagram for MOS capacitors with HfSiON dielectric and interlayer and meta l gates (Pt and Al) under negative gate bias. 150

PAGE 151

Figure 10-3. Changes in gate leakage current of Si MOS capacitors with HfSiON dielectric as a function of applied stress. 151

PAGE 152

Figure 10-4. Changes in dielectric co nstant of HfSiON, HfSiOx and HfO2, measured from C-V and PF slope change. 152

PAGE 153

CHAPTER 11 CONCLUSIONS The effect of temperature in thermal nitridati on process on physical structure of the IL was investigated using XRR studies. At high nitridat ion temperatures, the nitridation efficiency increased leading to increase in IL roughness apart from increase in film and IL density and IL thickness. There was a general improvement in the electrical properties of films at higher nitridation temperature till the degradation of interface quality due to interface roughening led to degradation of some of the elec trical properties. An excessive nitrogen piling up at IL of 850 oC nitrided Hf-silicate resulted in nonuniform distribution of ni trogen and oxide charge, high roughness IL and degradation of Dit, electron channel mobility and breakdown resistance. Therefore, the quantitative results of IL roughness by XRR study showed reliable correspondence with electrical resu lts and can have a potential as an analysis tool of thin film structure. To suppress the thermal nitridation induced interface degradation, the UV assisted low temperature nitrogen doping tech nique for hafnia gate dielectric film was investigated. The incorporation efficiency of UV assisted low temperature method was comparable to the high temperature (~650 oC) thermal nitridation process. Moreove r, due to the low process temperature and multiplicity of dissociation species, uniform distribution of nitrogen and oxygen, thinner interface layer (10 ), higher capacitance value (3.69 F/cm2) and smaller EOT (9.4 ) were observed in the case of UV assisted low temperature nitrided sample. By using UV assisted low temperature nitrided Hf-silicate film, in order to address the possibility of dual gate device applica tion, MOS device with Ti, TiN and TiB2 gate electrodes were examined. Considering EOTi and Qhk/i, the extracted meff values for Ti, TiN and TiB2 gate were 4.27, 4.56 and 5.08 eV, respectively. Re gardless of gate elec trodes, the conduction 153

PAGE 154

154 mechanism of the samples well fitted with Poole-Frenkel conduction model and conduction activation energies showed almost identical values (~ 1.7 eV). Th erefore, Ti gate electrode is more favorable for NMOS device and TiB2 gate electrode can be used for PMOS with Hf-silicate dielectrics Furthermore, in order for successful implements to future CMOS devices, thermal stability also should be confirmed. The extracted meff values for TiB2 gate was almost identical (5.07 eV) till 1000 oC annealing treatments. The meff lowering phenomena (4.91 eV) was observed when annealed at 1100 oC, due to the metal and Hf-silicate intermixing by material transfer and electron transfer by oxygen vacancy generation. The leakage current density of TiB2 electrode MOS device were lower than ~ 10-4 A/cm2 and VH is reduced after 1100 oC annealing. The possibility of TiB2 gate metal for the PMOS application was confirmed. The long term stability of boride and process compatibility remains to be established. Finally, since future CMOS device will adopt th e strained MOSFET structure, an effect of mechanical stress on silicon metal-oxide-semic onductor capacitor with nitrided Hf-silicate dielectric has been studied. The gate leakage current and dielectric constant was found to be increase by up to ~2 % of both tensile and compre ssive stresses. Increase in leakage current came from decreased HfSiON/Si interface trap activati on energy. The fact that dielectric constant variations under stress showed similar behavior by two different methods indicated the increase in dielectric constant could be caused by increa se in electronic polarization from N p band splitting. Measurements of gate leakage current and dielectric constants of HfSiON dielectric film under two different mechanical stress cond itions provides quantitative results and reveals that the incorporated nitrogen in the Hf-silicate film is a origin of dielectric constant variation.

PAGE 155

LIST OF REFERENCES 1. D. R. Askeland and P. P. Phule, The Science and Engineering of Materials, 4th ed. (Brooks Cole Publishing Company, St. Paul, Minnesota 2002). 2. T. Hori Gate Dielectrics and MOS ULSIs (Spr inger Series in Elect ronics and Photonics 34) 3. S. M. Sze, Physics of Semiconductor Devices, 2nd ed. (Wiley, New York, 1981). 4. G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89, 5243 (2001). 5. International Technology Roadmap for Semiconductors, 2002 edition 6. J. D. Plummer and P. B. Griffin, Proceedings of the IEEE 89, 240 (2001). 7. D. A. Buchanan, IBM, J. Res. Develop., vol. 43, pp. 245-264, (1999). 8. J. H. Stathis, IEEE Transact ions on Device and Materials Relia bility, 1, pp. 43 -59, (2004) 9. J. H. Stathis and D. J. Di Maria, IEEE International Elec tron Devices Meeting, pp. 167, (1998). 10. Y. J. Jeon, B. H. Lee, K. Zawadzki, W. J. Qi, A. Lucas, R. Nieh and J. Lee, IEDM Tech. Dig., pp.797, (1998). 11. B. Cheng, M. Cao, R. Rao, A. Inani, V. Voor de, W. M. Greene, J. M. C Stork, Y. Zhiping, P. M. Zeitzoff, J. C. S. Woo, Electron Devices, IEEE Tran sactions on Vol. 46 pp.1537 1544, (1999). 12. C. H. Lee, H. F. Luan, W. P. Bai, S. J. Lee, T. S. Jeon, Y. Senzaki, D. Roberts, and D. L. Kowng, IEDM Tech. Dig., pp. 27-30, (2000). 13. L. Kang, K. Onishi, Y. J. Jeon, B. H. Lee, C. S.Kang, W. Qi, R. Nieh, S. Gopalan, R. Choi, and J. C. Lee, IEDM Tech. Dig., pp. 35, (2000). 14. K. Onishi, L. Kang, R. Choi, E. Dharmarajan, S. Gopalan, Y.J. Jeon, C, S, Kang, B, H, Lee, R. Nieh, and J. C. Lee, VLSI Tech. Dig., p.131 (2001). 15. R. Choi, C. S. Kang, B. H. L ee, K. Onishi, R. Nieh, S. Gopalan, E. Dharmarajan and J. C. Lee, VLSI Tech. Dig., pp. 15, (2001). 16. Q. Lu, H. Takeuchi, X. Meng, T. J. King, C. Hu, K.Onishi, H.-J. Cho, J. Lee, VLSI Tech. Dig., pp. 86 -87, (2002). 17. K. J. Hubbard and D. G. Schl om, J. Mat. Res. 11, 2757 (1996). 18. J. J. Chambers and G. N. Parsons, J. Appl. Phys. 90, 918 (2001). 155

PAGE 156

19. D. G. Schlom and J. H. Haeni, Ma t. Res. Soc. Bulletin 27, 198 (2002). 20. A. Chatterjee, R. A. Chapman, K. Joyner, M. Otobe, S. Hattangady, M. Bevan, G. A. Brown, H. Yang, Q. He, D. R ogers, Tech. Dig. Int. Electron Devices Meet, pp. 777, (1998). 21. P. K. Roy and I. C. Kizilyalli, Appl. Phys. Lett. 72, 2835, (1998). 22. I. C. Kizilyalli, R. Y. S. Huang, and P. K. Roy, IEEE Electron Device Lett. 19, 423 (1998). 23. Q. Lu, D. Park, A. Kalnitsky, C. Chang, C. C. Cheng, S. P. Tay, Y.-C. King, T.-J. King, C. Hu, IEEE Electron Device Lett. 19, 341, (1998). 24. D. Park, Y.-C. King, Q. Lu, T.-J. King, C. Hu A. Kalnitsky, S.-P. Tay, and C.-C. Cheng, IEEE Electron Device Lett. 19, 441 (1998). 25. H. F. Luan, B. Z. Wu, L. G. Kang, B. Y. Kim, R. Vrtis, D. Roberts, and D. L. Kwong, Tech. Dig. Int. Electron Devices Meet, pp. 609 (1998). 26. C. W. Liu, S. Maikap and C. Y. Yu, I EEE circuit & Device mag azine pp 21-33, (2005) 27. M. H. Lee, P. S. Chen, W.-C. Hua, C.-Y. Yu, Y.-C. Lee, S. Maikap, Y.M. Hsu, C.W. Liu, S.C. Lu, W.-Y. Hsieh, and M.-J. Tsai, Pr oc. Int. SiGe Technol. And Device Meet. (ISTDM), Frankfurt, Germany, pp. 87, (2004). 28. M. L. Lee and E. A. Fitzgerald, IEDM Tech. Dig., pp. 429, (2003). 29. R. Oberhuber, G. Zandler, and P. V ogl, Phys. Rev.B, 58, pp. 9941, (1998). 30. M. R. Viscokay, J. J. Chambers, A. L. P. Rotondaro, A. Shanware and L. Colombo, Appl. Phys. Lett. 80 3138 (2002). 31. M. Balog, M. Schieber, M. Michman, and S. Patai, J.of Electrochem.Soc. 126, pp.12031207 (1979). 32. W. B. Bluementhal, The Chemical Behavior of Zirconium (Van Nostrand, Princeton, 1958). 33. L. Kang, B. H. Lee, W. J. Qi, Y. Jeon, R. Ni eh, S. Gopalan, K. Onishi, and J. C. Lee, IEEE Electron Device Letter s 21, pp.181-183 (2000). 34. 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-Crystallin e Solids 303, 139-143 (2002). 35. B. H. Lee, L. Kang, W. J. Qi, R. Nieh, Y. Je on, K. Onishi, and J. C. Lee, IEDM Tech. Dig., 133-136 (1999). 36. Q. Fang, J. Y. Zhang, Z. M. Wang, J. X. Wu, B. J. O'Sullivan, P. K. Hurley, T. L. Leedham, H. Davies, M. A. Audier, C. Jimenez, J. P. Senateur, and I. W. Boyd, Thin Solid Films 428, 263-268 (2003). 156

PAGE 157

37. G. D. Wilk and R. M. Wallace, Ap pl. Phys. Lett. 74, pp 2854-2856 (1999). 38. S. Van Elshocht, U. Weber, T. Conard, V. Kaus hik, M. Houssa, S. Hyun, B. Seitzinger, P. Lehnen, M. Schumacher, J. Lindner, M. Ca ymax, S. De Gendt, and M. Heyns, J.of Electrochem.Soc. 152, F185-F189 (2005). 39. G. Pant, A. Gnade, M. J. Kim, R. M. Wallace, B. E. Gnade, M. A. Quevedo-Lopez, P. D. Kirsch, and S. Krishnan, Appl Phys. Lett. 89, 032904, (2006). 40. P. D. Kirsch, C. S. Kang, J. Lozano, J. C. Lee, and J. G. Ekerdt, J. Appl. Phys. 91, pp43534363 (2002). 41. S. Inumiya, K. Sekine, S. Niwa, A. Kaneko, M. Sato, T. Watanabe, H.Fukui,Y. Kamata, M. Koyama, A. Nishiyama, M. Takayanagi, K. Eguchi and Y.Tsunashima, VLSI Tech. Dig., pp.17-18 (2003). 42. T. Watanabe, M. Takayanagi, R. Iijima, K. Ishimaru, H. Ishiuchi and Y. Tsunashima, VLSI Tech. Dig., pp. 19-20 (2003). 43. M. R.Visokay, J. J.Chamber, A. L .P.Rotondaro, R. Kuan, L.Tsung, M. Douglas, M. J. Bevan, H. Bu, A. Shanware, and L. Colombo, Proceedings of 3rd International Conference on Microelectronics and In terfaces, pp.127-129, (2002). 44. M. Koyama, A. Kaneko, T. Ino, M. Koike, Y. Kamata, R.Iijima, Y. Kamimuta, A. Takashima, M. Suzuki, C. Hongo, S. Inumiya, M. Takayanagi, and A. Nishiyama, IEDM Tech. Dig., pp. 849 -852 (2002). 45. Y. G. Fedorenko, L. Truong, V. V. Afanas'ev, A. Stesmans, Z. Zhang, and S. A. Campbell, J. Appl. Phys. 98, 123703 (2005). 46. E. M. Young and W. A. Tiller, J. Appl. Phys. 662, 2086 (1987). 47. I. W. Boyd, V. Craciun, and A. Kazor, Jap. J. of Appl. Phys. 32, 6141-6146 (1993). 48. D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr, J. Troe, and R. T. Watson, J. of Phys. and Chem. 9, 295-471 (1980). 49. S. Ramanathan, G. D. Wilk, D. A. Muller, C. M. Park, and P. C. McIntyre, Appl. Phys. Lett. 79, 2621 (2001). 50. S. Ramanathan, C. M. Park, and P. C. McIntyre, J. Appl. Phys. 91, 4521 (2002). 51. S. Ramanathan, D. Chi, P. C. McIntyre, C. J. Wetteland, and J. R. Tesmer, J. of the Electrochem. Soc. 150, F110 (2003). 52. P. Punchaipetch, G. Pant, M. J. Kim, R. M. Wa llace, and B. E. Gnade, J. Vac. Sci. Tech. A 22, 395 (2004). 157

PAGE 158

53. B. Maiti, P.J. Tobin, SPIE Conference on Mi croelectronic Device Technology III, SPIE vol. 3881 46 (1999). 54. I. De, D. Johri, A. Srivastava, C.M. Osburn, Solid State Electronics, 44, 1077 (2000) 55. B. Cheng, B. Maiti, S. Samavedam, J. Gran t, B. Taylor, P. Tobin, J. Mogab, 2001 IEEE Intl. SOI Conf., (2001). 56. H. Pierson, Handbook of Re fractory Carbides and Nitrides, (Noyes, 1996) 57. P. Ranade, H. Takeuchi, T-J. King, C. Hu, Electrochem. Sol. State Lett., 4 G85 (2001). 58. J. K. Schaeffer, S. B. Samavedam, D. C. Gilm er, V. Dhandapani, P. J. Tobin, J. Mogab, BY. Nguyen, B. E. White Jr., S. Dakshina-Murthy1, R. S. Rai, Z. -X. Jia ng, R. Martin, M. V. Raymond, M. Zavala, L. B. La, J. A. Smith, R. Garcia, D. Roan, M. Kottke2, and R. B. Gregory2., J. Vac. Sci. Tech. B 21 (2003) 59. Y. Saito, S. Kawata, H. Nakane, H. Adachi, Appl. Surf. Sci. 146 177 (1999). 60. C. S. Kang, H.-J. Cho, Y. H. Kim, R. Choi, K. Onishi, A. Shahriar, J. C. Lee J. Vac. Sci. Tech. B 21 5 (2003). 61. M. Moriwaki, T. Yamada, Jpn. J. Appl. Phys., 40 2679 (2001). 62. D-G. Park, H-J. Cho, K-Y. Lim, T-H. Cha, I-S Yeo, J-W. Park, J. Electrochem. Soc., 148 F189 (2001). 63. D-G. Park, K-Y. Lim, H-J. Cho, T-H. Cha, JJ. Kim, J-K. Ko, I-S. Yeo, J-W. Park, VLSI Tech. Dig., (2001). 64. H.Y. Yu, H.F. Lim, J.H. Chen, M.F. Li, C. Zhu, C.H. Tung, A.Y. Du, W.D. Wang, D.Z. Chi, D.-L. Kwong, IEEE Elec. Dev. Lett., 24, 4 (2003). 65. M.-A. Nicolet, P.H. Giauque, Microelectronic Engr., 55, 357 (2001). 66. Y-S. Suh, G. Heuss, V. Misra, J. Vac. Sci. Tech. B, 22, 1 (2004). 67. R. Li, Q. Xu, Trans. Elec tron Devices, 49, 11 (2002). 68. S. B. Samavedam, L. B. La, J. Smith, S. Daks ha-Murthy, E. Luckowski, J. Schaeffer, M. Zavala, R. Martin, V. Dhandapani, D. Triyoso, H. H. Tseng, P. J. Tobin, D. C. Gilmer, C. Hobbs, W. J. Taylor, J. M. Grant, R. Hegde, J. Mogab, C. Thomas, P. Abramowitz, M. Moosa, J. Conner, J. Jiang, V. Aruachalam M. Sadd, B-Y. Nguyen and B. White, IEDM Tech. Dig., 433-436, (2002). 69. J-H. Lee, H. Zhong, Y-S. Suh, G. Heuss, J. Gurganus, B. Chen, V. Misra, IEDM Tech. Dig. 359-362, (2002) 70. V. Misra, H. Zhong, H. Lazar, I EEE Electron Dev. Lett., 23 6 (2002). 158

PAGE 159

71. I. Polishchuk, P. Ranade, T-J. King, C. Hu, IEEE Electron Dev. Lett., 22 9 (2001). 72. B-Y. Tsui, C-F. Huang, IEEE Electron Dev. Lett., 24 3 (2003). 73. Q. Lu, R. Lin, P. Ranade, T-J. King, C. Hu, IEEE VLSI Symp., pp 45 (2004). 74. H. Wakbayashi, Y. Saito, K. Takeuchi, T. M ogami, T. Kunio, IEEE Trans. Electron Dev., 48 2363 (2001) 75. R.T. Tung, Mat. Sci. Engr. R, 35 1-138 (2001). 76. C. Hobbs, L. Fonseca, V. Dhandapani, S. Sama vedam, B. Taylor, J. Grant, L. Dip, D. Triyoso, R. Hegde, D. Gilmer, R. Garcia, D. Roan, L. Lovejoy, R. Rai, L. Hebert, H. Tseng, B. White and P. Tobin, VLSI Tech. Dig., (2003) 77. Y. C. Yeo, T. J. King, C. Hu, J. Appl. Phys. 92 12 7266 (2002). 78. H. Hasegawa, H. Ohno, J. V ac. Sci. Tech. B 4, 1130 (1986). 79. J. L. Freeouf, J.M. Woodall, Appl. Phys. Lett. 39, 727 (1981). 80. K. Rim, J. Chu, H. Chen, K.A. Jenkins, T. Kana rsky, K. Lee, A. Mocuta, H. Zhu, R. Roy, J. Newbury, J. Ott, K. Petrarca, P. Mooney, D. Lacey, S. Koester, K. Chan, D. Boyd, M. Ieong, and H.-S. Wong, VLSI Tech. Dig., pp. 98, (2002). 81. K.C. Liu, X. Wang, E. Quinones, X. Chen, X.D. Chen, D. Kencke, B. Anantharam, R.D. Chang, S.K. Ray, S.K. Oswal, and S.K. Banerjee, IEDM Tech. Dig., pp.63, (1999). 82. X. Wang, D.L. Kencke, K.C. Liu, A.F. Tasch, L.F. Register, and S. K. Banerjee, Proc. IEEE Int. Conf. Simulation of Semicond. Process. and Dev., pp.70 (2002). 83. K. Mistry, M. Armstrong, C. Auth, S. Cea, T. Coan, T. Ghani, T. Hoffmann, A. Murthy, J. Sandford, R. Shaheed, K. Zawadzki, K. Zha ng, S. Thompson, and M. Bohr, Symp. VLSI Tech. Dig., pp. 50 (2004). 84. M.D. Giles, M. Armstrong, C. Auth, S.M. C ea, T. Ghani, T. Hoffmann, R. Kotlyar, P. Matagne, K. Mistry, R. Nagisetty, B. Obradovic, R. Shaheed, L. Shifren, M. Stettler, S. Tyagi, X. Wang, C. Weber, and K. Zawadzki, Symp. VLSI tech. Dig., pp. 118 (2004). 85. G. Scott, J. Lutze, M. Rubin, F. Nour i, and M. Manley, IEDM Tech. Dig., pp. 827 (1999). 86. S. Pidin, T. Mori, R. Nakamura, T. Saiki, R. Tanabe, S. Satoh, M. Kase, K. Hashimoto, and T. Sugii, Symp. VLSI Tech. Dig., pp. 54 (2004). 87. T. Ghani, M. Armstrong, C. Auth, M. Bost, P. Charvat, G. Glass, T. Hoffmann, K. Johnson, C. Kenyon, J. Klaus, B. McIntyre, K. Mistry, A. Murthy, J. Sandford, M. Silberstein, S. Sivakumar, P. Smith, K. Zawadzki, S. Thompson, and M. Bohr, IEDM Tech Dig., pp. 978 (2003). 159

PAGE 160

88. R. E. Beaty, IEEE Trans. Components, H ybrids and Manufacturing Technology, (1992). 89. T. Chen and Y.-R. Huang, IEEE Trans. Co mponents and Packaging Technologies, (2002). 90. J. J. Wortman and R. A. Evans, J. Appl. Phys., 36, pp. 153-156, (1965). 91. K. Yamamoto, S. Hayashi, and M. Niwa, Appl. Phys. Lett. 83, 11 (2003). 92. J. F. Kang, H. Y. Yu, C. Ren, M.-F. Li, D. S. H. Chan, H. Hu, H. F. Lim, W. D. Wang, D. Gui, and D.-L. Kwong, Appl. Phys. Lett. 84, 1588 (2004). 93. P. K Park and S. W. Kang, Appl. Phys Lett 84, 19 (2006). 94. K. Ramani, C. R. Essary, S. Y. Son, V. Craciun and R. K. Singh, Appl. Phys. Lett 83, 242902 (2006). 95. R. C. Powell, N.-E. Lee, and J. E. Greene, Appl. Phys. Lett. 60, 2505 (1992). 96. M.-H. Cho, K. B. Chung and D.W. Moon, Appl. Phys. Lett. 89, 182908 (2006). 97. J. Kim and K. Yong. J. Appl. Phys. 100, 044106 (2006). 98. C. Kang, P. Lysaght, R. Choi, B. Lee, S. Rhee, C. Choi, M. Akbar, and J. Lee, Appl. Phys. Lett. 86, 222906 (2005). 99. M. A. Quevedo-Lopez, M. El-Bouanani, M. J. Kim, B. E. Gnade, R. M. Wallace, M. R. Visokay, A. LiFatou, J. J. Chambers and L. Colombo, Appl. Phys. Lett. 82, 4669 (2003). 100. M.-H. Cho, K. B. Chung, C. N. Whang, D.-H. Ko, J. H. Lee and N. I. Lee, Appl. Phys. Lett. 88, 202902 (2006). 101. M.-H. Cho, K. B. Chung, and D.-H. Ko, Appl. Phys. Lett. 89, 142908 (2006). 102. K. Ramani, P. Kumar, K. Siebein, V. Craciun, R. K. Singh, Electrochem. Solid-state Lett. 10, H66 (2007). 103. H. Kato, T. Nango, T. Miyagawa, T. Katagiri K. S. Seol, and Y. Ohki, J. Appl. Phys., 92, 1106 (2002). 104. R. Thomas, E. Rije, P. Ehrhart, A. Milanov, R. Bhakta, A. Bauneman, A. Devi, R. Fischer, and R. Waser, J. Electrochem. Soc. 154, G77 (2007). 105. A. R. Teren, R. Thomas, J. He and P. Ehrhart, Thin Solid Film, 478, 206 (2005). 106. C. Sire, S. Blonkowski, M. J. Gor don, and T. Baron Appl. Phys. Lett. 91, 242905 (2007). 107. P. Fiorenza, R. L. Nigro, V. Raineri, S. Lombardo, R. G. Toro, G. Malandrino, and I. L. Fragala, Appl. Phys. Lett. 87, 231913 (2005). 160

PAGE 161

108. J. L. Gavartin, A. L. Shluger, A. S. Fo ster, and G. I. Bersuker, J. Appl. Phys. 97, 053704 (2005). 109. K. Torii, K. Shirashi, S. Miyazaki, K. Ya mabe, M. Boero, T. Chikyow, K. Yamada, H. Kitajima, and T. Arikado, IEEE IEDM 04, 129 (2004). 110. M. A. Quevedo-Lopez, S. A. Krishnan, P. D. Kirsch, G. Pant, B. E. Gnade, and R. M. Wallace, Appl. Phys. Lett. 87, 262902 (2005). 111. M.-T. Wang, B. Y.-Y. Cheng, and J. Y.-M. Lee Appl. Phys. Lett. 88, 242905 (2006). 112. S. J. Lee, H. F. Luan, C. H. Lee, T. S. Jeon, W. P. Bai, Y. Senzaki, D. Roberts, and D. L. Kwong, VLSI Tech. Dig., 2001 133 113. A. Deshpande, R. Inamn, G. Jursich and C. G. Takoudis, J. Appl. Phys. 99, 9 (2006) 114. C. C. Yeo, M. S. Joo, B. J. Cho and S. J. Whang, Thin Solid Films 463, 90-95 (2004) 115. B. Sen, C. K. Sarkar, H. Wong, M. Chan and C.W. Kok, Solid-State Electronics 50, 237 (2006) 116. M. Lee, Z. H. Lu, W. T. Ng, D. Landheer, X. Wu and S. Moisa, Appl. Phys. Lett. 83, 2638 (2003) 117. V. Craciun and R. K. Singh, Appl. Phys. Lett. 76, 1932 (2000) 118. V. Craciun, R. K. Singh, J. Perriere, J. Spear and D. Craciun, J. Electrochem. Soc. 147, 1077 (2000) 119. A. Srivastava, V. Craciun, J. M. Howa rd and R. K. Singh, Appl. Phys. Lett. 75, 9002 (1999) 120. Z. Lencces, S. Guicciardi, C. Melandri, and A. Bellosi, Br. Ceram. Trans. 94, 38 (1995) 121. J. R. Hauser and K. Ahmed, AIP Conf. Proc. 449, 235 (1998). 122. C. D. Wagner, D. E. Passoja, H. F. Hillery, T. G. Kinisky, H. A. Six, W. T. Jansen, and J. A. Taylor, J. Vac. Sci. 21, 933 (1982) 123. J. M. Howard, V. Craciun, C. Essary, and R. K. Singh, Appl. Phys. Lett. 81, 3431 (2002) 124. J. H. Park, T. J. Park, M. J. Cho, S. K. Kim, S. H. Hong, J. H. Ki m, M. H. Seo, C. S. Hwang, J. Y. Won, R. J. Jeong and J. H. Choi, J. Appl. Phys 99, 094501 (2006) 125. J. Kim, S. Kim, H. Kang, J. Choi, H. Jeon, M. Cho, K. Chung, S. Back, K. Yoo, and C. Bae, J. Appl. Phys. 98, 094504 (2005). 126. J. Heitz, H. Niino and A. Yabe, Appl. Phys. Lett. 68, 2648 (1996). 161

PAGE 162

127. N. A .Chowdhury and D. Mi sra, J. Electrochem. Soc. 154, G30-G37 (2007). 128. R. K. Nahar, Vikram Singh and Aparna Sharma J. Mat. Sci Electron 18 615 (2007). 129. C. L. Cheng, K. C.-Liao, and T. Wang, Electrochem. Soc. Lett. 9 F80-F82, (2006). 130. Alfred Grill, Cold Plasma in Materials Fabrication Institute of Electri cal and Electronics Engineers, Inc., New York, pp. 74, (1994). 131. Y. K. Chiou, C. H. Chang, C. C. Wang, K. Y. Lee, T. B. Wu, R. Kwo, and M. Hong J. Electrochemical Society, 154 G99-G102 (2007). 132. N. Ikarashi, K. Watanabe, K. Masuzaki, T. Nakagawa, and M. Miyamura, J. Appl. Phys 100, 063507 (2006). 133. C. R. Essary, K. Ramani, V. Craciun and R. K. Singh Appl. Phys. Lett. 88, 182902 (2006). 134. S. Toyoda, J. Okabayashi, H. Takahashi, M. Oshima D. Lee, S. Sun, S. sun, P. A. Pianetta, T. Ando, and S. Fukuda, Appl. Phys. Lett 87, 182908 (2005). 135. F. J. Gruntharner, P. J. Gruntha rner, R. P. Vasquez, B. F. Lewis, J. Maserjian, and A. Madhukar, Phys. Rev. lett. 43, 1683 (1979). 136. M. A. Nicolet, Thin Solid Films, 52, 415, (1978). 137. C. S. Choi, Q.Wang, M. Carlton G. A. Ruggles, A.S. Shah, IEEE Transactions on Electron Devices, 39, 10, (1992). 138. R. Khanna, S. J. Pearton, F. Ren, I.I Kravchenko, Vac. Sci. Tech. B. 24(2006). 139. H. Yang, Y. Son, S. Baek, H. Hwang, H. Lim, H. S. Jung, Appl. Phys. Lett 86, 092107 (2005). 140. H. Yang, S. Heo, D. Lee, S. Choi and H. Hwang, Jpn. J. Appl. Phys. 45, 3 (2006). 141. R. Jha, J. Gurganos, Y. H. Kim, R. Choi and Jack Lee, IEEE Electron Devices Lett. 25, 6 (2004). 142. S. H. Bae, W. P. Bai, H. C. Wen, S. Mathew L. K. Bera, N. Balasubramanian, N. Yamada, M. F. Li, and D. L. Kwong, Symp. VLSI tech. Dig., 188 (2004). 143. J. Westlinder, G. Sjoblom and J. Olsson, Microelectronic Engineering 75 (2004). 144. D. S. Jeong, H. B.Park and C. S. Hwang, Appl. Phys. Lett 86, 072903 (2005). 145. D. Eom, S. Y. No, C. S. Hwang, a nd H. J. Kim, J. Electrochem. Soc. 153, 4 (2006). 146. S. Y. Son, P. Kumar, J. S. Lee, H. Cho, H. S. Jung, K. J. Min, C. J. Kang, and R. K. Singh, Electrochem. Soc. Lett.11 H81 (2008). 162

PAGE 163

147. Y. Akasaka, G. Nakamura, K. Shiraishi, N. Umezawa, K. Yamabe, O. Ogawa, M. Lee, T. Amiaka, T. Kasuya, H. Watanabe, T. Chikyow, F. Ootsuka, Y. Nara and K. Nakamura, Jpn. J. Appl. Phys. 45, L1289L1292, (2006). 148. S. J. Doh, J. H. Lee, J. P. Kim, J.-H. Lee, Y.-S. Kim, H.-J. Lim, H.-S. Jung, S. K. Han, M. J. Kim, N.-I. Lee, H.-K. Kang, S. G. Park, and S. B. Kang, IEEE International Reliability Physics Symp., (2005) 149. Y. S. Suh, H. Lazar, B. Chen, J. H. Lee, and V. Misra, J of Electrochem. Soc. 152, F138F141 (2005). 150. N. Umezawa, K. Shiraishi, T. Ohno, H. Wata nabe, T. Chikyow, K. Torii, K. Yamabe, K. Yamada, H. Kitajima, and T. Arika do, Appl. Phys. Lett., 86, 143507, (2005). 151. M. Koyama, H. Satake, M. Koike, T. Ino, M. Suzuki, R. Iijima, Y.Kamimuta, A. Takashima, C. Hongo, and A. Nishiyama, Tech. Dig. Int. Electron Device Meet, (2003). 152. S. Saito, K. Torii, M. Hiratani, and T. Onai, Appl. Phys. Lett., 81, 2391-2393, (2002). 153. C. W. Liu, S. Maikap and C. Y. Yu, IEEE Circuit and Device magazine May, pp. 21-36, (2005). 154. S. Suthram, P. Majhi, G. Sun, P. Kalra, P. Ha rris, K. J. Choi, D. Heh, J. Oh, D. Kelly, R. Choi, B. J. Cho, M. M. Hussain, C. Smith, S. Banerjee, W. Tsai, S. E. Thompson, H. H. Tseng, and R. Jammy, Tech. Dig. Int. Electron Device Meet, (2007). 155. Y. S. Choi, T. Nishida, and S. E. Thom pson, IEEE Electron Devi ce Lett., pp.154, (2008). 156. S. E. Thompson, G. Sun, K. Wu, J. Lim, and T. Nishida, Tech. Dig. Int. Electron Device Meet., pp. 94, (2004). 157. K. J. Yang and C. M. Hu, IEEE Transact ions on Electron Devices, 46, pp. 1500-1501, (1999). 158. J. R. Hauser and K. Ahmed, Characteri zation and metrology for ULSI technology, 1998. 159. Z. M. Rittersma, J. J. G. P. Loo, Y. V. Ponomarev, M. A. Verheijen, M. Kaiser, F. Roozeboom, S. Van Elshocht, and M. Caym ax, J. Electrochem. Soc., 151, G870-G877, (2004). 160. H. Momida, T. Hamada, T. Yamamoto, T. Uda, N. Umezawa, T. Chikyow, K. Shiraishi, and T. Ohno, Appl. Phys. Lett., 88, 112903, (2006). 161. N. Umezawa, K. Shiraishi, T. Ohno, H. Wata nabe, T. Chikyow, K. Torii, K. Yamabe, K. Yamada, H. Ktajima, and T. Arikado, Appl. Phys. Lett., 86, 143507, (2005). 163

PAGE 164

BIOGRAPHICAL SKETCH My mother was a chemistry professor at Ew ha Womens University, and I have been interested in science and app lied science since my childhood. On e day when I was a teenager I read an interesting article on semiconductors. Semi conductors were described as magic stones. Since then I have been fascinated with semi conductors, the magic materials. At university I studied inorganic materials engin eering and took a keen interest in the use of silicon. Learning semiconductor theory, such as th e conduction of electr icity in solids and the junction of doped semiconductors, I realized that re volutionary materials and designs can have vast potential for increasing the flexibility and efficiency of devi ces. And so I dreamed of becoming an engineer: driving progress in the field of microelectr onics by researching and developing innovative materials and analysis tools. In order to broaden my understanding of electronic materials, I chose thin film electronics materials as the resear ch topic of my master c ourse at the Thin Film Electronics Materials Lab in Hanyang University, one of the leading laboratories for semiconductor research in Korea. During my masters courses work, I was thrille d to study new materials for next-generation electronic devices and learn a va riety of theoretical and experimental concepts. My early research was focused on improvement of electrical properties between (Ba,Sr)TiO3 [BST] dielectric thin films and various materials for bottom electrode. My profes sor and I thought that the reason for the degradation might be a reducti on of interface propertie s caused by structural and chemical mismatch between dielectric film a nd electrodes. To solve this our group invented a new conducting oxide electrode (Ca,Sr)RuO3 [CSR] and (Ba,Sr)RuO3 [BSR] and proved that these materials show better electrical propert ies and less interface mism atch of which local epitaxial in BST/BSR and BST/CSR is directly obs erved by using a high resolution transmission electron microscope. This was my first success of tr eat magic materials. From this experience, I 164

PAGE 165

165 could get the idea of problem solving: how to define the proble m, how to design the experimental and how to overcome the barrier wh ich is initially likely to be unsolved. My academic experiences and efforts started to pay off even while I was still in school. In 1998 I received the silver prize in Humantech Thesis, one of the larges t thesis contests in semiconductor research sponsored by Samsung electronics. I also copublished several SCIpapers in international journals, including the Journal of Applied Phys ics and the Journal of Materials Science. As a process engineer at Sa msung Electronics, I was very happy that I could have the chance to use many kinds of high quality equipments for my research and to see the module process technology that help me to widen my understanding of ULSI process. For the last several years, I have been working on the device structure development that is the core of the fabrication technology development in Samsung. I participated in several cardinal projects such as 3-D gate design (STTM, FINFET, MRAM) and process deve lopment. To reduce the lattice and charge damage of extremely small size device, I tried to find fundamental solu tion and to approach many different ways. As a result of my works, I wrote several international patents and participated in several conferences. Also, I sh are an opinion that conventional device concepts may not work successfully without breakthrough a nd quantum jump of technology such as fine interface control. That is why I decided to study at University of Florida, which is one of leading research groups in the world.