GROWTH A N D CHARACTERIZATION OF THIN OXIDE FILMS O N SiGe by L A N Z H E N G B . S c , Fudan University, 1989 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF M A S T E R OF S C I E N C E in T H E F A C U L T Y OF G R A D U A T E STUDIES Department of Chemistry We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y OF B R I T I S H C O L U M B I A September 1997 © L a n Zheng, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract Atomic oxygen from a remote plasma oxidation was used to grow a high quality gate oxide on SiGe at low temperatures. Atomic oxygen from a remote O2 plasma was also used to form thin oxide films on SiGeo.025 that is capped with 12.5 A of Si at ~200°C. The characteristics and thicknesses of the oxide were determined by X-ray Photoelectron Spectroscopy (XPS) . It was observed that the oxidation that resulted from exposure to O-atoms, produced a thin oxide film with both Si and Ge oxidized. The interfacial trap densities were continuously monitored with an RF-probe. The changes in trap density were quite rapid during the process of oxidation and subsequent exposure to hydrogen atoms. However, the disappearance of the carrier traps, when the H-atoms were shut off, was found to be slower and could be followed. This is true at all temperatures ranging from 20°C to 200°C. The kinetic analysis of this trap removal process reveals that, this is either a second order reaction with an activation energy of 0.35 eV, or there are two concurrent first order processes with activation energies of 0.21 eV and 0.16 eV. B y comparing the behavior of Si capped SiGeo.025 with SiGeo.025, SiGeo.3 and intrinsic Si at ~200°C, it was found that all interfaces except SiGe 0.3 benefit from exposure to atomic hydrogen after oxidation, and that after such a treatment the interfacial trap density is not significantly different on Si capped SiGeo.025, SiGeo.025 and intrinsic S i . The passivation level achieved by low temperature H-atoms treatment was also compared to a high temperature (~450°C) H2 annealing process which was conducted according to the standard industrial practice. Similar passivation levels were achieved with both techniques. i i Table of Contents Abstract i i Table of Contents i i i List of Tables v List of Figures v i Acknowledgments Chapter 1 Introduction 1 .1.1 Semiconductor Materials and Physics 2 1.1.1 Sil icon and Germanium 2 1.1.2 SiGe 11 1.2 Semiconductor Devices 12 1.2.1 Bipolar 12 1.2.2 M O S F E T 14 1.3 Process Technology 16 1.3.1 Fabrication of a Bipolar Transistor 16 1.3.2 Fabrication of a M O S F E T 18 1.3.3 Oxidation, Plasma vs. Thermal Oxidation 20 1.4 Oxide/Semiconductor Interface Properties 23 1.5 Objectives of Present Research 26 Chapter 2 Experimental 29 2.1 Materials 29 2.1.1 Silicon and Silicon Germanium Samples 29 2.1.2 Chemicals 31 2.2 Apparatus 31 2.2.1 Reaction Chamber 33 2.2.2 RF-Probe 34 i i i 2.2.3 Mass Spectrometer 37 2.3 Experiment Procedure 39 2.3.1 Sample Cleaning 39 2.3.2 Generation of O-atoms and Plasma Oxidation 39 2.3.3 H-atoms Generation and Its Exposure to Samples 41 2.4 X-ray Photoelectron Spectroscopy (XPS) 42 2.4.1 Fundamental Principles 42 2.4.2 Instrumentation 45 2.4.3 Spectral Analysis 47 2.4.4 Application 52 Chapter 3 Results and Discussions 53 3.1 Oxidation of Si Capped SiGe with O-atoms at Different Temperatures 53 3.2 X P S Analysis of Si Capped SiGe 55 3.2.1 X P S of Clean and Oxidized Samples 55 3.2.2 Angle Dependent X P S 58 3.2.3 Determination of the Atomic Sensitivity for Ge 2p and Ge 3d 62 3.2.4 Quantitative Analysis 64 3.3 Effect of Exposure to H-atoms on the Charge Carriers 70 3.3.1 Unoxidized Samples 70 3.3.2 Oxidized Samples 72 3.3.3 Kinetic Analysis and Activation Energy Calculation 75 3.3.4 Mechanism of the Trap Decay 80 3.4 Comparison of Si Capped SiGe Substrate with Other Substrates 85 3.4.1 SiGeo.3 85 3.4.2 Intrinsic Si and Uncapped SiGe0.o25 88 3.4.3 Effect of H 2 annealing 93 Chapter 4 Conclusion 95 References 97 iv List of Tables Table 1.1 Selected important properties of Si and Ge at 300°K 10 Table 3.1 Peak intensities for the Si 2p and Ge 2p peaks as a function of plasma oxidation time (take-off angle = 90°) 57 Table 3.2 Intensity ratios as a function of plasma oxidation time (take-offangle = 90°) 57 Table 3.3 Peak intensities from peak fitting of Si 2p and Ge 2p as a function of take-off angle (oxidation time = 50 min) 61 Table 3.4 Intensity ratio as a function of take-off angle (oxidation time = 50 min) 61 Table 3.5 Integrated Ge 2p and Ge 3d peak intensities for pure Ge and mean path paths 62 Table 3.6 Calculated atomic intensity ratio and sensitivity ratio of Ge 2p/ Ge 3d 64 Table 3.7 Peak intensities and mean free paths of Ge 2p and Ge 3d for an H F washed sample 66 Table 3.8 Mean free paths (X) used in the calculation 69 Table 3.9 Oxide thickness for both Si and Ge as a function of oxidation time. The last column lists the calculated values for a 45° take-off angle with a 50 min oxidation time 69 Table 3.10 Tabulated data of second order rate constants 78 Table 3.11 Trap decay constants ki and k2 at different temperatures when H-atoms were turned off 83 List of Figures Figure 1.1 Sil icon crystal structure 3 Figure 1.2 Mi l l e r index 5 Figure 1.3 Band structure diagram of Si and Ge 7 Figure 1.4 (a) A bipolar transistor fabricated by the methods of planar technology (b) Schematic of active elements of an npn bipolar transistor 12 Figure 1.5 The physical structure of an n M O S F E T 14 Figure 1.6 A n npn transistor fabrication 17 Figure 1.7 A Simplified n M O S F E T fabrication 2 19 Figure 2.1 Schematic of reactor and associated equipment 32 Figure 2.2 A n RF-probe and RF-circuit 35 Figure 2.3 Mass Spectrometer 38 Figure 2.4 Schematic of the X P S process 44 Figure 2.5 X-ray photoelectron spectroscopy 46 Figure 2.6 X P S spectra of (a) Si 2p and (b) Ge 2p 48 Figure 2.7 Surface sensitivity enhancement by variation of the electron 'take-off angle 51 Figure 3.1 Sketch of Si capped SiGeo.025 sample 53 Figure 3.2 Change of carrier concentration for the Si capped SiGeo.025 sample during oxidation at 25°C, 205°C and 440°C 54 Figure 3.3 Si 2p and Ge 2p spectra from Si capped SiGeo.025 sample after an H F wash and after a 50 min oxidation 56 v i Figure 3.4 Plot of relative intensity ratios of Si / Si and G e 4 I G e u with oxidation time 59 Figure 3.5 Comparison of Si 2p and Ge 2p spectra obtained at two take-off angles (90° and 45°) for a Si capped SiGe sample after 50 min oxidation, showing the effect of increasing the take-off angle on the relative oxide/element intensities 60 Figure 3.6 Ge 2p and Ge 3d spectra from a pure Ge sample showing the relative peak intensities 63 Figure 3.7 Structure of the H F washed Si capped SiGeo.025 sample 65 Figure 3.8 X P S spectra of Ge 2p and Ge 3d from an H F washed sample (take off angle = 45°) 66 Figure 3.9 Model used to estimate oxide thicknesses from X P S peak intensity ratios...67 Figure 3.10 Effect of H-atoms on an H F washed Si capped SiGeo.025 sample at 198°C 71 Figure 3.11 Change of carrier concentration for an oxidized and annealed Si capped SiGeo.025 sample during long exposure to H-atoms at 20°C, 83°C, and 173°C 73 Figure 3.12 Change of trap density for an oxidized and annealed Si capped SiGeo.025 sample during long exposure to H-atoms at 20°C, 83°C, and 173°C 74 Figure 3.13 Plot of the logarithm of trap density versus time for 20°C, 83°C, 148°C, 173°C, and 192°C 76 Figure 3.14 Plot of the inverse of trap density versus time and the "best" straight line through the points for 20°C, 83°C, 148°C, 173°C, and 192°C 77 Figure 3.15 Arrhenius plot of the rate constants for the disappearance of traps when H-atoms are shut off. This process was analyzed by second order kinetics 79 Figure 3.16 Trap density decay when H-atoms are turned off at different temperatures. The dashed lines are fitting lines from a double exponential decay function y = y 0 + yi exp(-kit) + y 2 exp(-k 2t) 82 Figure 3.17 Arrhenius plot of trap decay rate constants (ki for a fast process and k 2 for a slow process) when H-atoms are turned off. These processes were analyzed by a two-concurrent first order kinetics 84 Figure 3.18 Change of carrier concentration for SiGeo.3 during O-atom and H-atom treatments at 209°C 86 Figure 3.19 Change of trap density for SiGeo.3 during O-atom and H-atom treatments at 209°C 87 Figure 3.20 Change of trap density for an intrinsic Si surface during O-atom and H-atom treatments at 209°C 89 Figure 3.21 Change of trap density for a SiGeo.025 surface during O-atom and H-atom treatments at 209°C 90 Figure 3.22 Change of trap density for a Si capped SiGeo.025 surface during O-atom and H-atom treatments at 209°C 91 Figure 3.23 Comparison of H-atom treatment at 202°C on intrinsic Si samples that (a) have been annealed in H 2 for 20 minutes at 450°C, and (b) that have not been annealed in H 2 94 Acknowledgments I would like to express my sincere gratitude to my supervisor, Professor Elmer Ogryzlo for his guidance and sound advice throughout this work. He has always been a source of inspiration and knowledge. I am especially grateful to Dr. Hongjun L i for sharing me with his experimental experience and ideas. I would also like to express my thanks to other members in our group: Dr. L ig ia Gheorghita, Professor Jun-gill Kang and Dr. Bagher Bahavar for their assistance and thoughts. M y thanks also go to Professor B . Heinrich in the Department of Physics at Simon Fraser University, for X P S measurements and for helping me with the interpretation of the X P S results. It is a pleasure to acknowledge many people in the Department of Chemistry and at the Advanced Materials and Process Engineering Laboratory ( A M P E L ) who provided their support and kindness that have made my academic studies a valuable learning experience. Finally, I would like to express my deepest appreciation to my mother, my father and my sister for their love and encouragement throughout these years of my life. ix Chapter 1 Introduction Semiconductors have made possible one of the greatest scientific and technological breakthroughs of this century. The road to semiconductor devices began with the invention of the first germanium transistor in 1947. Germanium is directly below silicon in the periodic table and has the same number of bonding electrons and the same crystal structure as silicon. The attractive features of germanium from the point of view of device theory are its carrier mobilities (3900 cm 2 /Vs for electrons and 1500 2 2 cm /Vs for holes). These are more than twice the values found for silicon (1500 cm /Vs for electrons and 450 cm 2 /Vs for holes). However, the oxide of germanium does not have the low interfacial state density, good dielectric properties, and the great chemical stability of Si02. Because of its smaller temperature range and inferior oxide, germanium was soon replaced by less costly silicon in the 1960's. Sil icon is the second most abundant element in the Earth's crust, where it is present principally as the oxide. In elemental form, it is nontoxic and is an excellent conductor of heat. It can be grown into ultra-pure, very large diameter crystals and readily forms a stable insulating oxide of high quality. A large measure of the success of silicon as a material for electronics is due to its oxide, Si02, which is easily grown on any Si surface. This oxide is an excellent electrical insulator and is chemically inert, protecting the silicon from attack by other substances. Especially important is the low density of states at the interface between the silicon and the Si02. Such interfacial states trap electrons, bending the valence and conduction bands so that the Fermi level at the interface is pinned at a position between the bands that is determined by the partially 1 filled traps rather than being controllable by doping or by the application of a "gate voltage" in Metal-Oxide-Silicon-Field-Effect-Transistor devices. Even in the case of the very good Si-Si02 interface extreme cleanliness and care are needed to keep the surface state density low enough to allow electronic devices to work. Such a favorable condition has not been consistently produced in any other system and the field effect transistor has, consequently, not been successful with any other semiconductor. The properties of silicon make it a natural for integrated circuit (IC) manufacturing. Yet, from a transistor designer's perspective, silicon is hardly the perfect semiconductor. Compared with some of the other semiconductors, it is quite poor in terms of how fast charge carriers can travel through the crystal lattice. This sluggishness limits the speed at which silicon devices can operate. 1.1 Semiconductor Materials and Physics 1.1.1 Silicon and Germanium Crystal Structure Silicon and Germanium are group IV elements, which therefore have 4 valence electrons. Those are the electrons which participate in chemical bonding when the atoms form compounds. Upon solidifying from the liquid phase both Si and Ge crystallize with covalent bonding in the diamond structure, as illustrated in Figure 1.1. 2 Figure 1.1 Silicon crystal structure . A diamond lattice unit cell, in which each atom is surrounded by four nearest neighbors as shown by the black atoms. 3 There are 3 principal crystal planes or directions of interest. These planes exhibit different properties and are defined in terms of the unit cell, which in this case is a cube with side dimension a. The planes (100), (110), and (111) are shown shaded in Figure 1.2. The three ordered digits, "1,0,0" for example, are called Miller indices and are commonly used to identify the planes and directions in cubic crystal structures. The numbers are the reciprocals, normalized with respect to a, of the intersection of the shaded plane and the x-axis, v-axis and z-axis, respectively. The three indices are normally placed in round parentheses (e.g. (100)) to designate the plane. The direction normal to the plane is indicated by the same Mi l le r indices, but they are enclosed in square brackets, viz.,[100] as shown in Figure 1.2. The family of equivalent directions is indicated by <100>. Doping The number of free electrons, n, in a pure semiconductor is exactly equal to the number of holes, p, since the production of both is due to Si -> S i + + e" where S i + is designated by "p" (a hole) and e" is designated by "n" (an electron). Such pure semiconductors are called intrinsic and an index ' i ' is appended to n and p in this case: ni = pi (1.1) If a silicon atom in a Si crystal lattice is replaced by an impurity atom from Group V of the periodic table (phosphorus or arsenic, for example), four of the valence electrons of the arsenic w i l l take part in the covalent bonding with the neighboring Si atoms while the fifth electron w i l l be only weakly attached to the arsenic atom. A t room temperature, 4 a 5 all such dopant centered electrons are promoted into orbitals in the conduction band which encompass the entire crystal. Such a semiconductor is called "n-type", to describe the presence of an excess of negative charge carriers. The corresponding doping impurity is called a donor, having 'donated' an electron to the conduction band. If a Si crystal is doped with an atom from Group III, having only 3 valence electrons, at the location of that impurity one of the covalent bonding electrons is missing. Wi th the expenditure of a relatively small amount of energy, one of the valence electrons in Si can contribute an electron to the Group III atom. A hole is created at the position vacated by that valence electron. The impurity atom, having accepted an additional electron, is called an acceptor. A semiconductor doped with acceptor is rich in holes, i.e. positive charge carriers. Such a semiconductor is called p-type. Electrons in n-type material and holes in p-type are called majority carriers, while holes in n-type and electrons in p-type are called minority carriers. The concentrations of n and p in Si at 25°C are governed by the relationship, np = 2 . 1 0 x l 0 2 0 cm" 6 (1.2) Band Gap Engineering A n energy band structure, shown in Figure 1.3 for Si and Ge, is a plot of the allowed electron energies E as a function of the momentum vector k. The Y axis represents energy and the x axis shows the magnitude of the momentum vector along some important crystal directions. The curves above the band gap (EG) are conduction bands (C.B.) , while the curves below EG describe the valence bands (V.B. ) . The 6 difference between the minimum in the conduction bands and maximum in the valence-bands is called the energy band gap and noted as EQ. Electrons in a crystal are not completely free, but instead interact with the periodic potential of the lattice. A s a result, their 'wave-particle' motion cannot be expected to be the same as for electrons in free space. The technique commonly used for correcting for this when equations of electrodynamics are applied to charge carriers in a solid, is to alter the electron mass, which is then called the effective mass. The effective charge carrier mass, m* is then related to the rate of curvature of the band in which the electron (or hole) finds itself. h2 (1.3) m* = — -d2EI 3k 2 here % = h / 2n, where h is Planck's constant. Table 1.1 lists the effective masses and other important properties of Si and Ge at 300°K. From Figure 1.3 and Table 1.1, one can see that Ge has a smaller band gap and lower m* for both electrons and holes (mo is the free electron rest mass). In thermal equilibrium, the electrons in the conduction band and holes in the valence band move in random fashion with a velocity that depends on the temperature. The disturbance of thermal equilibrium by the application of an electric field can cause carriers to acquire a drift velocity in a direction parallel to the electric field. The drift velocity (v2 on Si , most of the interface trapped charge can be neutralized by low-temperature (450°C) hydrogen annealing. The value of Q j t can be reduced to as small as 10 1 0 cm" 2. A t this level, the interface trapped charge has a negligible effect on device performance3. Surface states are also frequently referred to as interface traps, since they effectively trap free carriers at the Si-Si02 interface. They are electrically active defects and can degrade the electrical properties of the M O S F E T device. Degraded characteristics include changes in threshold voltage, reduced inversion layer mobility, increased minority carrier generation, and increased noise. Electron spin resonance (ESR) experiments have demonstrated that these interfacial defects have one dangling bond, and are called Pb centers when they occur at the S i 0 2 / S i (111) interface ' . Carriers in the silicon semiconductor interact with these defects, leading to degraded electrical characteristics. It was found that molecular hydrogen can passivate these dangling bonds at about 450°C where the following reaction occurs 4 0 ' 4 1 , H 2 + P b • P b H + H (1.8) where PbH is the hydrogen-passivated silicon dangling bond. This is a first order kinetic process 4 2 with an activation energy of 1.66 eV. The dissociation reaction, P b H • P b + H (1.9) was also postulated 4 3 to happen under vacuum thermal annealing with an activation energy of 2.56 eV. 24 It has also been suggested that atomic hydrogen produces interface defects via the reaction: P b H + H • H 2 + P b (1.10) which is the reverse passivation reaction of (1.8), while the reaction (1.11) (which is the reverse of (1.9)) removes interfacial defects. P b + H • P b H (1.11) Both reaction (1.10) and (1.11) are exothermic chemical reactions, probably requiring very little thermal activations. On the (100) face of silicon, two related defects called Pbo and P b i are observed. Pbi behaves in a manner analogous to the Pb center on (111), but the Pbo center behaves entirely differently. The Pbo center is present in the "as-grown" oxide, and it cannot be passivated by the same thermal treatment that passivates Pb(l 11) and Pbi(100) centers. A continuous slow growth of the Pbo center is observed during the thermal cycle. It is likely that the relevant ingredient in the annealing of Pbi(100) and Pb(l 11) is water vapor. The different behavior of Pbo vs the other Pb center may be a matter of kinetics rather than energetic. For example, the Pbo center may lie deeper in the silicon, two or three atomic layers away from the interface. In this position it would be less susceptible to passivation by water vapor, since water w i l l diffuse into the silicon less readily than through S i 0 2 4 4 . The density of interface traps (Dj t) can be reduced by hydrogen annealing at low temperatures (< 500 °Q 4 1 ' 4 2 ' 4 5 ' 4 6 . A very detailed study on the chemistry of the S i - S i 0 2 interfacial trap annealing was carried out by M . L . Reed, et al.47. 25 In view of this annealing behavior of SiC>2, annealing of SiGe oxide was also studied in an attempt to improve the interface properties. It was found by D. Tchikatilow, 48 et. al. that low temperature H 2 O vapor annealing of the SiGe oxide (oxidized from Sio.85Geo.15) decreases the interface state density. It is proposed that changes, which occur as a result of annealing in forming gas and H 2 O vapor, result from the ability of hydrogen species to react with unsaturated silicon atoms at the oxide/semiconductor interface forming S i - H bonds. These workers considered that atomic hydrogen diffuses to the interface and reacts with unsaturated silicon to eliminate interfacial states. The atomic hydrogen diffusion constant value is D ~ 2x10"7 cm 2/s at 300°C for S i C V The diffusion length is therefore l=2(Dt) 1 / 2 ~ 1CJ3 cm (at 300°C, t=ls), which is three orders of magnitude larger than the oxide thickness. The post-oxidation and post-metal annealing cycle was also found to result in a low interface trap density 3 7. 1.5 Objectives of Present Research It is essential to use low temperature processing techniques in fabricating SiGe M O S F E T devices in order to ensure the structural integrity of the strained SiGe layer is maintained. Conventional gate oxide growth is a major burden in this respect. A n alternative strategy is to use plasma enhanced oxide growth which can involve lower temperatures. The objective of this work is to grow a thin gate oxide on SiGe with a minimum density of interfacial carrier traps. The approaches we have used are: (1) Obtain 4 types of samples (i) SiGeo.025 epitaxial layer capped with an intrinsic Si . 26 (ii) SiGeo.3 epitaxial layer grown on a Si wafer. (iii) intrinsic Si epitaxial layer on a Si wafer. (iv) SiGeo.025 epitaxial layer grown on a Si wafer. A Si capped SiGe is an alternative substrate for growing high-quality SiGe oxide. Our initial work, therefore, was on this sample. The oxidation of uncapped SiGe samples, with low and high Ge%, was also considered worth studying to see i f the Si cap is really necessary. (2) Oxidize these samples in a remote O2 plasma system. L o w temperature is preferred in SiGe oxidation. The oxidation system used in this work is a remote plasma system to take advantage of its low temperatures which can be as low as 25°C. (3) Expose surfaces to H 2 and/or H-atoms as a post-oxidation treatment. H 2 annealing has been shown to reduce the interface trap density of S i / S i 0 2 at ~ 450°C. It has been suggested that this effect involves atomic hydrogen. The question that arises is what w i l l happen in the case of the SiGe oxide during H 2 annealing and H-atoms exposure? W i l l these processes also improve the SiGe oxide? These measurements should provide answers to these questions. (4) Monitor the carrier traps with a remote RF-probe during the oxidation and subsequent hydrogen treatment. Attempt a kinetic analysis of these processes. 27 It is noted that the loss of photo-generated charge carriers occurs principally on dangling bonds (carrier traps) at the surface. The steady-state carrier concentration, therefore, can be used as a technique for following the formation and loss of dangling bonds. Since Si and Ge are indirect band gap semiconductors, the radiative recombination of carriers, i.e. photoluminescence, cannot be used for monitoring their concentrations. A novel RF-probe was developed in this lab. With this unique probe, we can continuously monitor the carriers and thus monitor any changes in the carrier trap density. (5) Characterize the oxide surface and interface structure with X P S . X-ray photoelectron spectroscopy (XPS) is a powerful surface analysis technique. The thin SiGe oxide grown by plasma oxidation can be characterized with X P S . With this technique, the nature and thicknesses of the oxide can be determined. 28 Chapter 2 Experimental 2.1 Materials 2.1.1 Silicon and Silicon Germanium Samples Four kinds of semiconductor samples were used in this work. A l l of them were grown in a "Sirius" C V D reactor at the Institute for Micro structural Sciences ( N R C , Ottawa). The "Sirius" C V D hot wall reactor from Leybold A G with a quartz reactor was operated at 525°C and a base pressure of 10"9 Torr. Ultra-Large-Scale-Integration grade silane and germane from Matheson were used as precursors at a deposition pressure of ~ 10"3 Torr. Details of the growth conditions can be found in the reference4 9. Sample 1. SiGe, Si-capped: 12.5 A intrinsic Si (i-Si) on top of 1000 A SiGeo.025- P-type, boron-doped float-zone Si substrate with a resistivity of 30-60 Qcm. i-Si, 12.5 A SiGe, '0.025 IOOO A Si Wafer 30-60 Qcm 29 Sample 2. SiGe, uncapped, high Ge%: 70 A SiGeo.3 on top of 150 A intrinsic Si . SiGe, ILL i^Si Si Wafer 70 A 150A Sample 3. Intrinsic Si: 1200 A intrinsic Si grown on Czochralski (CZ) Si(100) P-type, boron-doped substrate with a resistivity of 13-18 Qcm. 1200 A 13 -18 Qcm Sample 4. SiGe, uncapped, low Ge%: 1050 A SiGe0.o25 on top of a 150 A intrinsic Si buffer. P-type, boron-doped float-zone Si substrate with a resistivity of 30-60 Qcm. SiGeo . 0 25 1050 A ,«feSfcv..'. ; " 150 A Si Wafer 30-60 Qcm 30 2.1.2 Chemicals The oxygen, hydrogen and argon used are all ultra high purity standard from Praxair. The quoted purity of oxygen is 99.993%, hydrogen 99.999% and argon 99.999%. The maximum moisture in oxygen, hydrogen and argon are all less than 3 ppm. Nitrogen dioxide was supplied by Matheson with a quoted purity of 99.5% in liquid phase. The hydrofluoric acid used was 48% H F , A . C . S . reagent grade from Aldrich with an assay of 48.0-51.0%. De-ionized water with a resistivity of 18 M Q c m was supplied by the Advanced Materials and Process Engineering Laboratory ( A M P E L ) at the University of British Columbia. 2.2 Apparatus A diagram of the reactor and associated equipment is shown in Figure 2.1. It mainly consists of a remote microwave plasma, a heated reaction chamber, an RP-probe to monitor the substrate during the reaction, and a mass spectrometer chamber to monitor the gaseous species. 31 He-Ne laser Light chopper Microwave cavity Thermocouple Mass spectrometer Molecular drag pump Mechanical pump 2.2.1 Reaction Chamber The reactor was constructed mostly of Pyrex. Only the discharge region was made of quartz which can withstand the high plasma temperature. The quartz discharge region had a 10.5 mm inner diameter and was 100 mm in length. The tube in the reaction region had an inner diameter of 24 mm and was 160 mm long. Two light traps were added between the discharge region and the reaction region to prevent the radiation from the plasma from impinging on the sample. The plasma was created by a quarter wave cavity attached to an E . M . S . Microtron 200 microwave power generator (incident power 0-200 W). Compressed air around the cavity was used to cool down the plasma region when the plasma was on. The sample was placed flat on the bottom of the Pyrex reactor tube. A 6 mm diameter Pyrex sample holder was bent to touch and hold the sample. A heating wire was ' wrapped around the sample region to heat the sample when necessary. A Variac transformer, connected to the heating wire, was used to control the temperature of the sample by adjusting the current "flowing through the heating wire. A 1 mm diameter flexible temperature probe (Cole-Parmer H-08514-96, K-type) was inserted into the sample holder for measuring the temperature, which was displayed on a digital readout (Omega, model 115 K C ) . The sample holder was connected to the loading cap by an O-ring. A Cajon fitting was used to attach the cap to the reaction chamber. Gases were introduced into the system through % inch copper tubing. The pressure of gases was set by flow control valves and monitored by a capacitance manometer (Edwards Barocel Series 600) with a pressure range of 10" -10 Torr. A cold 33 cathode pressure gauge (HPS Model 421) was used to measure the pressure in the range of 10" 2 - 10"1 0 Torr. In the case of low pressure (10"6 - 10"1 Torr) operation, the molecular drag pump (Alcatel M D P 5010, 7.5 liters/second nitrogen of pumping speed) backed by a rotary pump (Sargent Welch Model No . 1375, 300 liters/minute of pumping speed) was used. When a pressure higher than 100 mTorr was needed, the molecular drag pump was closed, leaving only the rotary pump running. Periodic maintenance was necessary for the reaction chamber in order to maintain the high vacuum, to obtain high concentration of atomic species and keep the experiments repeatable. For this purpose, the reaction chamber could be disconnected from the system. The Pyrex and quartz parts were cleaned with dilute H F solution, rinsed with DI water and dried with air. Gaskets and O-rings were checked periodically and replaced when necessary. 2.2.2 RF-Probe A n RP-probe, shown in Figure 2.2, provides a "contactless" technique for monitoring the steady-state carrier concentration of the sample. The probe was pressed against the outside wall of the reactor where the sample was placed to maximize the sensitivity of the probe to the changes in the conductivity of the sample which is determined by the steady-state carrier concentration. 34 He-Ne laser RF generator Splitter -± x . M ixer Lock-in amplifier Computer Figure 2.2 A n RF-probe and RF-circuit. Light chopper Sample Coupler RF-probe The operation of the RF-probe is described in detail by L i and Ogryz lo 5 0 ' 5 1 . The RF-probe consists of a helical resonator made of a copper coil , which is inductively coupled to the sample. The RF-circuit contains a splitter, coupler and mixer connected as shown in Figure 2.2. A n R F signal from an R F generator (HP model 3200B V H F Oscillator, frequency range 10 to 500 M H z , R F output > 150 m W in band 130 to 260 M H z ) passed into a splitter (Min i Circuits ZFSC-2-1) . The splitter produces two signals. One goes to a coupler (Min i circuits ZFDC-10-2) , and then into the C u coil from which it is reflected with an altered magnitude and phase into the mixer (Min i circuits Z F M - 2 H ) . The other part of the signal from the splitter goes directly into the same mixer to act as a reference. The phase shift and magnitude of the reflected signal are affected by the conductivity of the sample. The signal from the mixer is then detected by a lock-in amplifier ( E G & G Princeton Applied Research model 5102). A 10 m W , 633 nm He-Ne laser beam, chopped at a reference frequency by a chopping unit (Grubb Parsons), hits directly on the sample. Free electron and hole pairs are formed since the photon energy is larger than the band gap of the substrate. The change in sample conductivity due to the injected electron and hole concentrations is proportional to the photo-generated minority carrier concentration. The chopping frequency of 200 H z was chosen so that it was (a) rapid enough to measure exposure times of less than 1 s, but (b) at least two orders of magnitude slower than the longest carrier lifetime. The lock-in amplifier measures only the carriers generated by the chopped laser beam at 200 Hz . The output of the lock-in amplifier was connected to both a D C voltage meter and a computer so that the data could be collected for further analysis. 36 B y varying the laser intensity we have determined that, at least for our samples, the output from the lock-in amplifier is proportional to the change in carrier density. Under constant laser intensity conditions the steady state carrier concentration is a direct measure of the surface or interface defect density, since carrier losses at the interfaces are several orders of magnitude greater than in the bulk of the semiconductor. Consequently under all operating conditions used in this work the probe output was a direct measure of the steady state carrier concentration and the reciprocal of this quantity was proportional to the interfacial trap density. A l l the data collected were plotted and fitted with a fitting program called Microcal Origin on a Pentium P C . 2.2.3 Mass Spectrometer The mass spectrometer vacuum chamber, shown schematically in Figure 2.3, was constructed of stainless steel. The system could be pumped down to 10"7 Torr with a turbomolecular pump (Edwards E X T 7 0 , pumping speed of 52 liters/second nitrogen) backed up by a rotary pump (Sargent-Welch Model No. 1402, pumping speed of 160 liters/minute). A gate valve connected the reaction chamber to the mass spectrometer chamber. A "differential gasket" was employed between the gate valve and flange on the side of the reaction chamber with a 100 um pin hole in it to leak a small amount of gases from the reaction chamber into the mass spectrometer chamber. 37 Pressure gauge Chamber —r L Sensor Turbo pump Computer Mechanical pump Figure 2.3 Mass Spectrometer. 38 The mass spectrometer was a quadruple mass analyzer ( M K S partial pressure transducer (PPT) residual gas analyzer) with a mass range of 1 to 200 atomic mass unit. It consists of a compact ion-source quadrupole sensor, an electronic control unit (ECU) , and interactive PPT software connected to a P C computer to control and monitor the mass spectrometry. The maximum operating pressure of the sensor is l x l 0 " 4 Torr. A total system pressure reading was continuously available in the range of l x l O " 4 to 2xl0" 9 Torr. The pressure of the M S chamber was also measured with an ionization vacuum gauge (Leybold-Heraeus Combitron C M 330). 2.3 Experiment Procedure 2.3.1 Sample Cleaning A 20x10 mm sample was cut from the wafer supplied by N R C . It was then cleaned in 2% H F for 60 s to remove the native oxide, followed by rinsing in DI water and drying with N 2 . The HF-treated Si surface is less reactive and more stable against oxidation in room air since the surface is passivated by H-termination of silicon dangling bonds, forming S i - H bonds and protecting the surface from chemical at tack 5 2 ' 5 3 ' 5 4 . 2.3.2 Generation of O-atoms and Plasma Oxidation Generation Oxygen atoms were generated in microwave plasma located 20 cm upstream from the reaction chamber. 500 mTorr of 0 2 with a flow of ~ 100 seem was selected to maximize the concentration of O-atoms and minimize their recombination in the 39 connecting tubing. The microwave generator produced a maximum output of 100 W. A titration experiment, using reactions (2.1) and (2.2) 5 5, performed by H . L i in our group, indicating that about 4% of the O2 was dissociated at the sample position, meaning that the atomic oxygen concentration is about 8%. The substrate, after an H F wash, was put into the flow reactor system and fixed with a sample holder. The system was then closed and pumped down to vacuum. The system was purged with O2 for 30 min before igniting the plasma to stabilize the flow and flush out any other gases. During this purging with O2, the system was heated to the desired temperature by the heating wire. The reactor walls, gas, and the substrate were all at the same temperature. The discharge was ignited with a Tesla coil . Then the sample was exposed to atomic oxygen from an upstream microwave discharge located 20 cm away from the substrate and separated by two light traps to eliminate all U V radiation, electrons and ions from the stream before it impinged on the substrate. After the oxidation was complete, the discharge was shut off by turning off the microwave generator power. N 0 2 + O -> N O + 0 2 0 + N O -> N 0 2 * N 0 2 + hv (2.1) (2.2) Oxidation 40 2.3.3 H-atoms Generation and Its Exposure to Samples Generation The same microwave discharge method was used to produce H-atoms. 5 mTorr of H 2 diluted in 30 mTorr of A r was passed through the microwave plasma which was operated at 40 W . The atomic hydrogen concentration was deliberately kept low to avoid heating of the sample by recombination of H-atoms on the surface. Exposure of H-atoms The H-atoms exposure experiments were carried out in a manner similar to that used for the oxidation experiments discussed in section 2.3.2. After the sample was loaded into the reactor, the system was closed and evacuated below 10"5 Torr. The desired H 2 / A r flow was then introduced into the system and the sample was heated up to the desired temperature. When the system was stable, the microwave plasma was then ignited to produce H-atoms. The sample now was exposed to atomic hydrogen. After this process was finished, the H-atoms were terminated by turning off the plasma. 41 2.4 X-ray Photoelectron Spectroscopy (XPS) 2.4.1 Fundamental Principles X-ray Photoelectron Spectroscopy (XPS), also known as Electron Spectroscopy for Chemical Analysis ( E S C A ) , is a widely accepted and powerful technique for studying surfaces and interfaces to determine the chemical and elemental properties. In X P S analysis, a monochromatic X-ray strikes at a sample with an energy hv > 1 keV. A s a result of the absorption of an X-ray photon, core-level photoelectrons are emitted from the sample with kinetic energies characteristic of the target elemental composition and its chemical state. This energy of the emitted photoelectrons is analyzed by the electron spectrometer. The data are presented as a graph of intensity (or counts per second) versus electron energy - the X-ray induced photoelectron spectrum. 5 6 The kinetic energy (Ek) of the electron is the experimental quantity measured by the spectrometer, but this is dependent on the energy of the X-ray source employed and is therefore not an intrinsic material property. The binding energy of the electron (ER) is the parameter which identifies the electron specifically, both in terms of its parent element and atomic energy level. The relationship between the parameters involved in X P S experiments is as fol lows: 5 7 Ek = hv - ER - W (2.3) where Ek is the measured kinetic energy of the ejected electron, hv is the X-ray photon energy, ER is the binding energy of the ejected electron, and W is the spectrometer work function. 42 Figure 2.4 illustrates the process of photoemission . A n electron from the K shell is ejected from the atom. The photoelectron spectrum wi l l reproduce the electronic structure of an element quite accurately as all electrons with a binding energy less than the photon energy wi l l feature in the spectrum. Those electrons which are excited and escape without energy loss contribute to the characteristic peaks in the spectrum. Each photoelectron has a discrete energy representative of the element from which it was emitted, thus allowing one to identify the atomic species present, chemical state of the atoms and the elemental composition in a given sample. Although the X-ray photons can penetrate 1-10 um deep into the sample, the electrons generated at that depth simply cannot make it out into the vacuum to be detected. Relatively low kinetic energy electrons (50 to 2000 eV), the type generally excited by X P S , have inelastic mean free paths (IMFP or X ) which range from about 0.5 to 3 nm. X is defined 5 9 as the distance that an electron w i l l travel before it suffers an inelastic collision with the nucleus of an atom in a solid. These collisions w i l l change the direction and the kinetic energy of electrons, decreasing the peak intensity. This process follows a standard exponential decay 5 6 (2.4), I (d) = I 0 exp (-d/ X) (2.4) where 1(d) is the number of electrons that are ejected without having undergone inelastic collisions from a depth, d, in the sample. I 0 is the number of electrons emitted from an infinitely thick substrate. Only 37%, i.e. 1/e, of the electrons that are ejected from an atom at IX, w i l l reach the surface with their characteristic kinetic energy. Therefore, relatively few electrons from depths of 2X to 3X w i l l leave the bulk with their initial 43 Photoelectron Figure 2.4 Schematic of the X P S process. kinetic energy. Thus, they w i l l not be detected. This accounts for the surface sensitivity of X P S . X P S can therefore provide a total elemental analysis of about the first 30 A of any solid surface which is vacuum stable 6 0. 2.4.2 Instrumentation A n X-ray photoelectron spectrometer, schematically shown in Figure 2.5, consists of an X-ray source, a sample support system, an electron energy analyzer and a detection system, all contained within a vacuum chamber. A data-system is used to convert the detected current into a readable spectrum. The spectrometer is based on a vacuum system designed to operate in the ultra-high vacuum range of 10"8 to 10"1 0 Torr. There are two reasons for this: (1) the low energy electrons are easily scattered by the residual gas molecules and unless their concentrations are kept at an acceptable level the total spectral intensity w i l l decrease whilst the noise present within the spectrum w i l l increase, and more importantly, (2) the high surface sensitivity of the technique requires careful control of surface composition, which ambient gases can change. X-rays are generated by bombarding the anode materials with electrons. A n ideal X-ray source must be sufficiently energetic to access core levels, intense enough to produce a detectable electron flux, have a narrow line width and be simple to use and maintain. The most common anode materials used are A l and M g which provide A l K a and M g K a photons of energy 1486.6 eV and 1253.6 eV, respectively. 45 High vacuum system Filament Hemispherical electron analyzer Figure 2.5 X-ray photoelectron spectroscopy. 46 Sample for analysis is first mounted on a sample holder and brought into the fore-chamber. It is then transferred into the analysis chamber. Once the sample is in the analysis chamber, it needs to be positioned accurately. Angle dependent measurements can be conducted by rotating the sample around the axis to vary the take-off angle of electrons accepted into the analyzer. The kinetic energy of the ejected electron is measured using a concentric hemispherical electron analyzer which gives the best energy resolution. A s shown in Figure 2.5, two hemispheres are placed concentrically. The incoming electrons, with a pass energy E , go in between these two hemispheres. A potential difference is applied between the inner and outer hemispheres so only electrons in a small energy range (E + AE) w i l l be transmitted to the other end of the analyzer to the detector. A spectrum can be produced when the voltage between the inner and outer hemispheres is ramped. The analyzer basically measures the number of electrons with different kinetic energies. The information is processed by a computer to produce a spectrum of photoelectron intensity as a function of binding energy. 2.4.3 Spectral Analysis Qualitative A n example of an X P S spectrum obtained from a Si capped SiGeo.025 sample after a 10 min oxidation at 193°C is illustrated in Figure 2.6. The spectrum is made on a Physical Electronics Instrument. It is a Si 2p and Ge 2p spectrum with an A l K a source and a "pass energy" of 50 eV. The spectrum includes several peaks. B y comparing 47 ioooo u 78000 76000 Z- 74000 >< ->—* c 72000 (1) 70000 U 68000 U Binding Energy (eV) 1224 1222 1220 1218 1216 1214 Binding Energy (eV) Figure 2.6 X P S spectra of (a) Si 2p and (b) Ge 2p. binding energies with reference data 6 1, individual peaks can be identified and labeled on the spectrum. X P S spectra can provide information on the chemical valence state of an atom from the "chemical shift" in binding energy when the binding changes. This chemical shift can arise in several ways including changes in the formal oxidation state and changes in the lattice site. In general it is due to the change in the environment of an atom. In Figure 2.6a, by comparing X P S peaks from the Si atom in a silicon crystal and in SiCh, a shift towards higher binding energy is observed for S i C V The same kind of chemical shift was observed in the Ge 2p spectrum shown in Figure 2.6b. It is generally found that the core electron binding energy increases with increasing positive oxidation state. The full width at half-maximum ( F W H M ) is the peak width at half the signal height. The measured F W H M , E M , is a convolution of contributions from the photon source Ep, the electron energy analyzer EA, and the natural line width of the atomic level E N . The sum of the squares of these factors is the square of the measured F W H M 5 8 . A E M 2 = AEp 2 + A E A 2 + A E N 2 (2.5) Quantitative Quantitative analysis is performed by determining the area under the peaks in question and applying a previously-determined sensitivity factor. For a homogeneous sample, the number of photoelectrons per second, I (peak intensity), in a given peak, assuming constant photon flux and fixed geometry, is given b y 6 2 49 I = K N a A, A T (2.6) where K = constant, N = number of atoms of the element per cm 3 , a = photoionization cross section for the element, X = inelastic mean-free path length for photoelectrons, A = area of the sample from which the photoelectrons emanated and T = analyzer transmission function. If we define the sensitivity factor for element x as S x = K C T A A T , then I x = N X S X . Angle Dependent XPS The surface sensitivity of X P S can be further enhanced by decreasing the take-off angle from its customary 90°. This effect is demonstrated in Figure 2.7. If X is the attenuation length of the emerging electron then 95% of the signal intensity is derived from a distance 3X within the solid. However, the vertical depth sampled is clearly given and this is a maximum when oc= 90°. In the case of a substrate (s) with a uniform thin overlayer (o) the angular variation of intensities is given by: d = 3 X sin a (2.7) I s d = I s exp (-d/ X sin a) (2.8) and I 0 d = I 0 (1- exp (-d/ X sin a)) (2.9) 50 2.4.4 Application One very important area of X P S application is in the semiconductor industry. 6 4 X P S studies on Si , especially thin oxide/Si interfaces, have been widely carried out because of the importance of thin films in electronic devices. The initial stage of oxidation in the clean-room air 6 5 at room temperature and in dry oxygen 6 6 at 300 °C were studied with X P S to determine the growing mechanism and uniformity of such oxidation. Interfacial states in the Si band gap present at oxide/Si interfaces were investigated 6 7 by measurements of X-ray photoelectron spectra on "biased" samples (i.e. with applied voltages). This method was also employed to determine the effect of various wet cleaning processes. It was found that the density of interface states was affected by the interface roughness 6 8. K . B . C la rk 6 9 et al. optimized the growth and annealing conditions with the help of X P S . From a quantitative point of view, the X P S methodology can provide a precise determination of the thickness of thin fi lm silicon oxides 7 0 ' 7 1 on Si and silicon nitride 7 2 on S i , etc. Recently, X P S has been used to explore SiGe and its oxidation process. By identifying the various elements and their chemical shifts in the oxidized SiGe, the structure of the oxide can be clearly determined 7 3 ' 7 4 ' 7 5 . 52 Chapter 3 Results and Discussions 3.1 Oxidation of Si Capped SiGe with O-atoms at Different Temperatures The Si capped SiGeo.025 sample was oxidized at three different temperatures: 25°C, 205°C and 440°C. A s shown schematically in Figure 3.1, 1000 A of SiGe0.o25 was grown on a p-type Si substrate wafer, and then a 12.5 A intrinsic Si cap was grown on top. i -Si SiGeo.025 p-Si wafer Figure 3.1 Sketch of Si capped SiGeo.025 sample. During this oxidation process, the laser-generated steady-state carrier concentration was continuously monitored by the remote RF-probe. It was found that at any temperature in this range, the carrier concentration drops to a very low level as soon as the surface is exposed to oxygen atoms. This is illustrated in Figure 3.2. Further exposure to O-atoms does not affect the passivation level significantly. Since the probe can not make measurements at temperatures higher than 250°C, the data at 440°C was recorded by lowering the temperature to 25°C before measuring the steady-state carrier concentration with the probe. 12.5 A 1000 A 53 300 400 Time (s) Figure 3.2 Change of carrier concentration for the Si capped SiGeo.025 sample during oxidation at 25°C, 205°C and 440°C. 54 3.2 XPS Analysis of Si Capped SiGe 3.2.1 XPS of Clean and Oxidized Samples Figure 3.3 shows the Si 2p and Ge 2p X P S spectra of a Si capped SiGeo.025 sample after an H F wash (top half) and after a 50 min oxidation at 193°C (the lower spectra). Take-off angles are all 90° (perpendicular to the surface). In the Si 2p spectra, both the H F washed and the oxidized sample showed the S i 0 peak at 99.5 eV due to the underlying Si substrate. However, for the oxidized sample, a broad S i 4 + peak, due to Si02 appeared 3.9 eV above the S i 0 peak and a small S i + peak 7 6 appeared ~1.0 eV above the S i 0 . In the Ge 2p spectra , both the H F washed and the oxidized sample show a Ge° peak at 1217.1 eV. In addition, in the case of the oxidized sample, a G e 4 + peak 3 3 appeared 3.8 eV above the Ge° peak. These results suggest that after 50 min of exposure to O-atoms the oxidation has extended beyond the Si cap as shown in the next sketch. This implied that we should attempt a shorter oxidation time. We therefore exposed new samples to O-atoms for 10 min and also for 30 s. Very similar spectra were obtained for 10 min and 30s oxidation. Oxidized Si cap partially oxidized SiGe p-Si wafer 55 S2p Ge2p 3 HFwash 50 rrin oxidation Si° 108 106 104 102 100 98 96 Binding energy (eV) HFwash G2> \ 50 nin oxidation Ge*+ I . I . I , 1226 1224 1222 1220 1218 1216 1214 1212 Binding energy (eV) Figure 3.3 Si 2p and Ge 2p spectra from Si capped SiGeo.025 sample after an H F wash and after a 50 min oxidation. Peak intensities, obtained from an X P S peak fitting program, and the relative intensity ratios for peaks are given in Table 3.1 and Table 3.2. R S JO and Ro e o in Table 3.2 are relative intensity ratios of Si x + /S i° and Ge 4 + /Ge° , respectively. 56 Table 3.1 Peak intensities for the Si 2p and Ge 2p peaks as a function of plasma oxidation time (take-off angle = 90°) N . Time 0 s 30 s 10 min 10 min 50 min Sample with native oxide Intensity Si° 3 .7x l0 3 2.55xl0 3 2 .2x l0 3 3.0xl0 3 2 .3x l0 3 3 .1x l0 3 S i x + ( x = l & 4 ) < 3 . 8 x l 0 2 l .OxlO 3 7 .1x l0 2 1.7xl0 3 1.2xl0 3 8.2xl0 2 Ge° 3 .6x l0 2 2.35xl0 2 1.2xl0 2 2 .3x l0 2 6 .4x l0 2 1.8xl0 2 G e 4 + < 1.6x10' 1.5xl0 2 6.6x10' 1.75xl0 2 4 . 5 x l 0 2 1.4xl0 2 Table 3.2 Intensity ratios as a function of plasma oxidation time (take-off angle = 90°) Time R a t i o ^ \ ^ ^ 0 s 30 s 10 min 50 min Sample with native oxide R s ,o = S i x + / S i ° < l.OxlO" 1 4.0x10"' 4.5x10"' 5.1x10"' 2.65x10"' RGeo = G e 4 + / G e u <4.4xl0" 2 6.4x10-' 6.6x10"' 6.9x10"' 7.9x10"' RGeO / R-SiO 4.3x10-' 1.6 1.5 1.35 3.0 57 Table 3.1 shows absolute peak intensities obtained after an H F wash (labeled 0 s), after 30 s, 10 min, and 50 min oxidation, and with the initial native oxide. The absolute peak intensities are not important, but the relative values are significant. The relative peak intensities, shown in Table 3.2, are the more meaningful data. They revealed that the longer the oxidation time, the higher the S i 4 + intensity is relative to the S i 0 . The same is true for the G e 4 + /Ge° ratio. This indicates that more oxide was grown. For the native oxide, Rsio lies in between those of the H F washed sample and the 30 s oxidized sample. But RGeo for samples with a native oxide is surprisingly large. This could be due to some Ge contamination in the capping Si layer. Figure 3.4 shows a plot of intensity ratios of Si x + /S i° and Ge 4 + /Ge° for oxidation times of 0 s, 30 s, 10 min and 50 min. It can be seen from this figure that there is a rapid initial increase of the oxide layer which reaches saturation in about 10 min. 3.2.2 Angle Dependent XPS Detailed angle dependent X P S spectra (Figure 3.5) were recorded at take-off angles of 90° and 45° (relative to the surface). The Si 2p and Ge 2p bands are shown for the Si capped SiGeo.025 sample oxidized for 50 min. The relative intensity ratios of S i x + / S i 0 and G e 4 + / Ge° are larger at higher take-off angle, i.e. at the shallower sampling depth, implying that the oxide region is near the surface as expected. This behavior is also clearly shown in Table 3.3 and Table 3.4 for the fitted peak intensities and their relative ratios, respectively. 58 Figure 3.4 Plot of relative intensity ratios of S i x / Si and Ge / Ge with oxidation time. 59 Figure 3.5 Comparison of Si 2p and Ge 2p spectra obtained at two take-off angles (90° and 45°) for a Si capped SiGe sample after 50 min oxidation, showing the effect of increasing the take-off angle on the relative oxide/element intensities. 60 Table 3.3 Peak intensities from peak fitting of Si 2p and Ge 2p as a function of take-off angle (oxidation time = 50 min) Angle 90° 45° Intensity ( a . u . j " - \ ^ S i 0 9.4x10 3 4 .9x l0 3 S i x + 4 .8x l0 3 4 .4x l0 3 G e u 6 .4x l0 2 1.2xl0 2 G e 4 + 4 .5x l0 2 2 .1x l0 2 Table 3.4 Intensity ratio as a function of take-off angle (oxidation time = 50 min) Angle 90° 45° R S io = Si x + /S i° 5.1x10"' 9.0x10-' Roeo = G e 4 + / G e ° 6.9X10"1 1.7 RGeO / RsiO 1.35 1.9 61 3.2.3 Determination of the Atomic Sensitivity for Ge 2p and Ge 3d In order to determine the relative atomic sensitivities of the Ge 2p and Ge 3d bands, a pure Ge sample was analyzed by X P S . Figure 3.6 shows the X P S spectra for Ge 2p and Ge 3d from this pure Ge sample. Peak intensities obtained by integrating over the Ge 2p and Ge 3d peaks, and mean free paths (X) from C. J. Powell 's computer calculations ("National Bureau of Standards", U.S.) are given in Table 3.5. Table 3.5 Integrated Ge 2p and Ge 3d peak intensities for pure Ge and mean free paths Peak Intensity (I) (a.u.) MA) Ge 2p l .Ox lO 4 6.9 G e 3 d 1 . 25x l0 3 22.5 For pure Ge, the following expression holds . I G e 2 p _ P G e * A G e 2 p * ^ G e 2 p (3-1) ^Ge3d PGe* A G e 3 d * ^ G e 3 d where I c e 2p, l G e 3 d are the peak intensities for Ge 3p and Ge 3d from the peak fitting program. po e is the Ge density. A o e 2 P and A o e 3 d are the atomic sensitivities for the Ge 2p and Ge 3d X P S peaks. Xoap and A,Q e3d are the electron mean free paths for Ge 2p and Ge 3d photoelectrons, respectively. 62 Figure 3.6 Ge 2p and Ge 3d spectra from a pure Ge sample showing the relative peak intensities. 63 Rearranging the above expression gives the atomic sensitivity ratio of Ge 2p to Ge 3d (3.2). The results are given in Table 3.6. A G e 2 p = P G e * I G e 2 p * ^ G e 3 d (3-2) A G e 3 d PGe* ^Ge3d * ^Ge2p Table 3.6 Calculated atomic intensity ratios and sensitivity ratios of Ge 2p/Ge 3d l G e 2 p / l G e 3 d 8.0 A G e 2 p / A G e 3 d 26 3.2.4 Quantitative Analysis The wafer (sample 1) used in this study has a Si cap of -12 A thick covering -1000 A SiGeo.025 layer. However, since some of the Si cap may have been oxidized on standing, and hence some of this oxidized Si was removed when the surface was washed with H F , it was considered important to establish how thick the Si layer was before the oxidation experiment was initiated, and then how thick the oxidized layer was after the O-atom oxidation. Figure 3.7 illustrated schematically a structure of Si capped SiGe sample after an H F wash, " x " is the thickness of Si cap after an H F wash, a is the take-off angle in X P S measurements. 64 Si SiGeo.025 p-Si wafer Figure 3.7 Structure of the H F washed Si capped SiGeo.025 sample. From a Ge 2p/Ge 3d intensity ratio, the Si layer thickness (x) can be calculated with the following equation 7 6, (3 3) I G « 2 p = P G e * A G e 2 p * ^ S i ( E G e 2 p ) SMI PL * e X p ( - X / \ s ( E G e 2 p ) Sin P i ) ^ G e 3 d P G e * A G e 3 d * ^ - s i ( E G e 3 d ) s i n a * exp(-x / X s j ( E G e 3 d ) sin a) where Ioe2p and Ioe3d are the Ge 2p and Ge 3d peak intensities respectively. po e is the Ge density. Ac e2p and A o e 3 d are the atomic sensitivities for Ge 2p and Ge 3d lines. Xsi(Eo e2P) , A,si(EGe3d) are the Ge 2p and Ge 3d electron mean free paths in Si . a is the take-off angle for the spectrometer. Figure 3.8 gives the Ge 2p and Ge 3d spectra at a take-off angle of 45° for an H F washed sample. Wi th an X P S fitting program for these two spectra, the Ge 2p and Ge 3d peak intensities were obtained and are listed in Table 3.7, together with their electron mean free paths in Si . I* 65 G62p G e 3 d 1224 1222 1220 1218 1216 1214 BrdrgEnercy(e,v) 30 32 34 36 38 Brief ng Energy (eV) Figure 3.8 X P S spectra of Ge 2p and Ge 3d from an H F washed sample (take off angle 45°). Table 3.7 Peak intensities and mean free paths of Ge 2p and Ge 3d for an H F washed sample Peak Intensity (a.u.) * si (A) Ge 2p 7 .2x l0 2 9.3 G e 3 d 2.4x10 2 32.6 66 From the data listed in Table 3.7 and the equation above, the Si layer thickness was calculated to be x= 8.3 A . After the H F washed sample was treated with O-atoms, X P S spectra revealed traces of Ge02. Therefore, we assume that the Si cap is completely oxidized into Si02 (in order to simplify the calculation, S i + and S i 4 + are all considered as Si0 2 ) . The underlying SiGeo.025 was partially oxidized to form a thin layer of mixed Si and Ge oxides. This assumed structure is illustrated schematically in Figure 3.9. D Figure 3.9 Model used to estimate oxide thicknesses from X P S peak intensity ratios. 67 Oxide thickness can be determined from X P S spectra by the following technique 7 0. 76 Equation (3.4) relates the measured intensity ratio Isi02/Isi to the total oxide thickness (D) in the above diagram. Values for other parameters in this equation are listed in Table 3.8. R = _ PSJO2 * ^s,o2 s i n a * (1 - e x p ( - D / s ing)) ( 3 - 4 ) I si P si * 1 si s i n a * e x p ( - D / A. S l 0 ; sin a ) Similarly, to obtain the Si02 cap thickness (D-d), we can use equation (3.5) where the parameters are also listed in Table 3.8. ^ G e O ~ y = Psio, * Ko2 s i n a * exp ( - (D - d) / X'Si0i s i ng ) * (1 - exp( -d / X'SiDi s ina)) Psi * ^ si s m a * e x p ( - D / X'si0^ sin g ) _ Psio ; * Ko2 * e x P ( d / ^sio 2 s i n a ) * d ~ exp( -d / X's^ s ing)) ( 3 - 5 ) Psi * K hio, hi, iGeo and Io e are the peak intensities of S i 4 + (+Si +), S i 0 (from Si 2p), G e 4 + , Ge° (from Ge 2p), respectively. psi02 and psi are the Si02 and Si densities. A-si02, A-si are the electron mean free paths in Si02 and Si . A,'si02, X's\ are the electron mean free paths for Ge 2p electrons in Si02 and Si . Mean free paths are listed in Table 3.8. a is the take-off angle for the spectrometer. D and d are the thicknesses of the Si02 and Si02/Ge02 layers, respectively. 68 B y solving equations (3.4) and (3.5), using X from Table 3.8, the values of D and d were calculated. The values listed in Table 3.9 are for 3 different oxidation times. Table 3.8 Mean free paths (X) used in the calculation 7 7 Elements X(A) X\A) S i 0 2 37 11.5 Si 31 9.3 Table 3.9 Oxide thickness for both Si and Ge as a function of oxidation time. The last column lists the calculated values for a 45° take-off angle with a 50 min oxidation time. Oxidation time Substrate 30 sec 10 min 50 min 50 min S i 0 2 (D) 1 8 A 20 A 22 A 23 A S i 0 2 + G e 0 2 (d) 7.9 A 8.0 A 8.3 A 10 A Although the oxide layers become slightly thicker with increasing oxidation time, the variation is small, and it is clear that much shorter reaction times are necessary to stop 69 the oxidation before it reaches the Si/SiGe interface. The layer thicknesses calculated from 0° and 45° take-off angles for 50 min oxidation are reasonably consistent. 3.3 Effect of Exposure to H-atoms on the Charge Carriers 3.3.1 Unoxidized Samples A s shown in Figure 3.10, hydrogen atoms, when exposed directly to the H F washed Si capped SiGeo.025 sample, lowers the steady state carrier concentration significantly, i.e. the passivation level is adversely affected. In this experiment, the sample was first treated with hydrogen atoms for 7 s at 198°C. This caused a dramatic drop of the steady-state carrier concentration (to almost 0). When the atoms were turned of f , there was some recovery of the passivation level. However, longer exposure to H -atoms dropped the passivation level even more, from which there was no significant recovery when the H-atoms were turned off. 70 Hen "nrre(s) Figure 3.10 Effect of H-atoms on an H F washed Si capped SiGeo.025 sample at 198°C. 3.3.2 Oxidized Samples When an oxide layer is grown on the surface first, hydrogen atom exposure produces a different effect i f the temperature is high enough. Figure 3.11 and Figure 3.12 illustrate this effect in terms of carrier concentration and trap density, respectively. Three samples were initially oxidized for 10 minutes at 207°C and then annealed in H-atoms for 40 min at the same temperature. They were then separately exposed to H -atoms at 20°C, 83°C and 173°C for 10 min and monitored before, during, and after the exposures. The carrier concentrations at all these 3 temperatures were normalized to the same initial value. It can be seen that at all three temperatures, exposure to H-atoms dropped the steady-state carrier concentration as it did for the unoxidized sample. At room temperature, there was very little recovery of the signal when the H-atoms were removed. However, as the temperature was raised to 173°C, in contrast to the unoxidized sample, there was a large recovery in carrier concentration when the H-atoms were shut off. 72 1.2 0.01 0 500 173°C 83 °C ****** 20 °C 1000 1500 2000 2500 3000 "[irre(s) Figure 3.11 Change of carrier concentration for an oxidized and annealed Si capped SiGeo.025 sample during long exposure to H-atoms at 20°C, 83°C, and 173°C. 73 Figure 3.12 Change of trap density for an oxidized and annealed Si capped SiGeo.025 sample during long exposure to H-atoms at 20°C, 83°C, and 173°C. 74 3.3.3 Kinetic Analysis and Activation Energy Calculation Although the rise in trap density that occurs when the sample is exposed to H -atoms is too rapid to be followed with our instrumentation, the disappearance of the traps, when the H-atoms are shut off can be subjected to a kinetic analysis. ' Figure 3.13 shows a plot of ln(trap density) versus time. It is not a straight line and the process therefore is not first order. If the process is second order then a plot of l/(trap density) versus time should be a straight line. Figure 3.14 plots the inverse of trap density versus time for 20°C, 83°C, 148°C, 173°C, and 192°C. It can be seen that the fit to second order kinetics is not too bad at the three lower temperatures. However, for the two higher temperature the lines are distinctly curved in the initial region, i.e. the reaction is initially even faster than predicted by the second order rate constant. 75 Figure 3.13 Plot of the logarithm of trap density versus time for 20°C, 83°C, 148°C, 173°C,and 192°C. 76 50 I 1 1 1 1 I I . I I I I I ! I -500 0 500 1000 1500 2000 2500 3000 Time(s) Figure 3.14 Plot of the inverse of trap density versus time and the "best" straight line through the points for 20°C, 83°C, 148°C, 173°C, and 192°C. 77 The second order rate constants (k) extracted from the fitted straight lines at times longer than 250 s (Figure 3.14) are listed in Table 3.10. According to the Arrhenius equation, k = A e E / k T (3-6) Therefore, the activation energy can be calculated from the slope of Arrhenius plot shown in Figure 3.15. The activation energy calculated is 0.35 eV. Table 3.10 Tabulated data of second order rate constants T ( ° C ) 1/T (1/°K) k(s- ') Ink (a.u.) 20 3.4xl0" 3 9.2x10"5 -9.3 83 2.8xl0" 3 l . l x l O " 3 -6.8 148 2.4xl0" 3 6.1xl0" 3 -5.1 173 2.2x10"3 9.2xl0" 3 -4.7 192 2.15xl0" 3 1.6xl0" 2 -4.1 78 Figure 3.15 Arrhenius plot of the rate constants for the disappearance of traps when H -atoms are shut off. This process was analyzed by second order kinetics. 79 3.3.4 Mechanism of the Trap Decay To explain a second order process, we would have to postulate the following model in which the carrier trap is a bound hydrogen atom (H-Site): k i H-Site • H + Site (3.7,3.8) H + H-Site • H 2 + Site (3.9) where reaction (3.9) is very much slower than reactions (3.7) and (3.8). When the H -atoms are turned on, the trap concentration is determined largely by reactions (3.8) and (3.9) (governed by k 2 and k3). When the discharge-produced H-atoms are removed, the loss of traps is governed by the reaction (3.7) which produces a small steady-state H-atom concentration at the interface. The rate of loss of H-site is given by, d [ H - S i t e ] _ d [ H 2 ] 2dt dt k 3 [H][H-Si te ] (3.10) If the concentration of H is determined principally by reaction (3.7) and (3.8), k j H - S i t e ] (3.11) [H] which is derived from, k, [H-Site] = k 2 [H] [Site] (3.12) Substituting (3.11) in (3.10) (3.13) 80 If we assume [Site] ~ constant, i.e. there are a large number of sites, then the reaction (3.7) becomes second order. The experimental activation energy which we measure is then identified with the composite rate constant k 3ki/k2. k 2 and kj would be predicted to have very small activation energies, and therefore ki w i l l have an activation energy of about 0.35 eV, which is not unreasonable. It is also possible to analyze the trap removal process by a multiple-exponential decay. This can be explained into a two-concurrent first order kinetics. Figure 3.16 plots the trap density versus time when the H-atoms were turned off at 20°C, 83°C, 148°C, 173°C, and 192°C. The trap density at each temperature is fitted with a double-exponential in the form: T=T 0 + T i exp(-kit) + T 2 exp(-k 2t) (3.14) where T is the trap density. This double exponential decay fits the experimental data reasonably well . It is therefore also possible that H-atom exposure generated 3 kinds of traps. After the H-atoms are turned off, one To remained unchanged, the other two (Ti and T 2 ) decayed with rate constants k i and k 2 . The decay constants k i and k 2 are listed in Table 3.11. Their change with temperature is presented as Arrhenius plots in Figure 3.17. The experimental value obtained for k i at 20°C was ignored when the straight line was drawn through the points in Figure 3.17 because it lies several standard deviations from the value predicted by the other 4 points. The Arrhenius plots yield an activation energy of 0.21 eV for the rapidly decaying traps, and of 0.16 eV for the slowly decaying traps. 81 Figure 3.16 Trap density decay when H-atoms are turned off at different temperatures. The dashed lines are fitting lines from a double exponential decay function , y = y 0 + yi exp(-kit) + y 2 exp(-k 2t). 82 Table 3.11 Trap decay constants ki and k 2 at different temperatures when H-atoms were turned off T ( ° C ) 1/T (K" 1) k, (s"1) k 2 (s- ' ) lnki (a.u.) lnk 2 (a.u.) 20 3.4xl0" 3 4.5x10"4 4.8xl0" 4 -7.7 -7.6 83 2.8xl0" 3 1.75xl0" 2 1.3xl0" 3 -4.05 -6.65 148 2.4x10"3 4.75xl0" 2 3.4xl0" 3 -3.05 -5.7 173 2.2xl0" 3 6.8xl0" 2 4.0xl0" 3 -2.7 -5.5 192 2.15xl0~ 3 8.0x10"2 4.1xl0" 3 -2.5 -5.5 83 Figure 3.17 Arrhenius plot of trap decay rate constants (ki for a fast process and k 2 for a slow process) when H-atoms are turned off. These processes were analyzed by a two-concurrent first order kinetics. 84 3.4 Comparison of Si Capped SiGe Substrate with Other Substrates 3.4.1 SiGeo.3 In order to identify the effects that may be due to Ge during oxidation and hydrogen passivation, several experiments were performed on a layer of SiGeo.3 that is very much richer in Ge than the samples used earlier. SiGeo.3 was exposed to oxygen atoms and then treated with H-atoms at 209°C. The change of steady-state carrier concentration during the processes is shown in Figure 3.18. The change of trap density is shown in Figure 3.19. It can be seen that the exposure to H-atoms after a 10 minutes exposure to O-atoms resulted in very little improvement. It would appear that the oxidation of silicon with a 30% Ge content produces an oxide that has a poorly passivated interface with the SiGe crystal, and no treatment that we could apply improved the passivation level significantly. 85 0.75 r-3 I 8 0.50 h-0.25 h 0.00 h Time (s) Figure 3.18 Change of carrier concentration for SiGeo.3 during O-atom and H-atom treatments at 209°C. 86 Tirre'(s) Figure 3.19 Change of trap density for SiGeo.3 during O-atom and H-atom treatments at 209°C. 87 3.4.2 Intrinsic Si and Uncapped SiGeo.025 A n intrinsic Si surface, an uncapped SiGeo.025 surface, and the capped sample treated earlier were subjected to identical treatments. Figure 3.20, Figure 3.21, and Figure 3.22 records the changing trap density during the treatment of these three substrates. The following numbers wi l l be used in these figures to identify the conditions under which the measurements are being made. The samples were kept at a constant T= 209°C. (1) H F washed surface with O2 passing over the surface (2) Surface exposed to O-atoms for 10 min (3) H-atom treatment (4) Discharge shut off but H 2 flow retained 88 Time (s) Figure 3.20 Change of trap density for an intrinsic Si surface during O-atom and H-atom treatments at 209°C. 89 2 6h-0 Oaff,Hai Hon f (Tf^Octi Hoi 1Q50A 150A km Hon 2000 4000 6000 Time (s) Figure 3.21 Change of trap density for a SiGeo.025 surface during O-atom and H-atom treatments at 209°C. 90 Figure 3.22 Change of trap density for a Si capped SiGeo.025 surface during O-atom and H-atom treatments at 209°C. 91 Consider first the changes that occurred at the Si surface as recorded in Figure 3.20. The H F wash produced a surface with a very low carrier trap density (region (1)). The oxidation that resulted from exposure to O-atoms for as little as 30 s produced a layer of Si02 of about 20 A (see Section 3.2). This corresponds to the oxidation of about 8 atomic layers of Si . A s soon as the oxygen atoms were introduced, the carrier trap density increased by more than an order of magnitude, and within a second it settled to a value that remained unchanged during the 10 minute exposure (region (2) in Fig. 3.20). When the hydrogen atoms were passed over this oxidized surface the trap density dropped to about 40% and then did not change a great deal while it was being exposed to H-atoms for a long period (region (3)). However, when H-atoms were removed, i.e. the microwave discharge was shut off (region (4)), the carrier trap density dropped slowly to a level that, with longer exposures to H-atoms, approached the initial trap density of the H F washed sample. Comparing Figures 3.20, 3.21 and 3.22, it is clear that the behavior of the three samples is almost indistinguishable. We therefore conclude that the presence of 2.5% Ge in the Si crystal insignificantly affects the response of the material to oxidation and subsequent H-atom treatment. It is also clear from the data in Figure 3.19 that this is not true when the concentration of Ge in the Si is 30 mole%. 92 3.4.3 Effect of H 2 annealing Since standard industrial practice ( R C A patent 1972) involves the annealing of "gate oxides" in H 2 at about 450°C in order to reduce the density of interfacial carrier traps, the important question is whether the treatment with H-atoms that we have used in this work has the same effect as H 2 annealing. To test this we annealed one intrinsic silicon sample in H 2 at 450°C after oxidation. A comparison of the response of this sample and an unannealed sample to H -atoms is shown is Figure 3.23. It can be seen that annealing the oxidized sample in H 2 at 450°C for 20 minutes removed about 80% of the carrier traps. When this sample or one that has not been annealed in H 2 was now exposed to H-atoms the carrier trap density changed to the same value (Figure 3.23). The subsequent behavior of the two samples was very similar and the eventual trap density after the H-atoms were shut off was very similar. Both are better than the original H 2 annealed sample. 93 Time® Figure 3.23 Comparison of H-atom treatment at 202°C on intrinsic Si samples that (a) have been annealed in H2 for 20 minutes at 450°C, and (b) that have not been annealed in H2. 94 Chapter 4 Conclusion L o w temperature remote microwave plasma oxidation was employed in this work to grow a gate oxide on SiGe with a minimum density of interfacial carrier traps. X P S was used to characterize the various layers and their thicknesses. Plasma oxidation of SiGe capped with 12.5 A of Si completely oxidized the Si cap and generated both S1O2 and GeC>2. The total oxide thickness was about 20 A after as little as 30 s oxidation. The oxide layer became slightly thicker with oxidation time. A novel RF-probe was used to continuously monitor the interfacial trap density. It was found that the trap density changes very rapidly during oxidization and the subsequent H-atom exposure. For Si capped SiGeo.025 the trap removal rate when H -atoms were shut off was subjected to a kinetic analysis. It was found that this process could be fitted either as a second order reaction or to 2 or more concurrent first order processes. To explain a second order rate law the following mechanism was proposed, H-Site , k l > H + Site (3.7,3.8) k 2 H + H-Site • H 2 + Site (3.9) Where the carrier trap is a bound hydrogen atom (H-Site). Reaction (3.9) must be very much slower than reactions (3.7) and (3.8). Temperature-dependence studies (20°C ~ 200°C) gave an activation energy of 0.35 eV for reaction (3.7). 95 Alternatively i f the trap removal process is analyzed in terms of two concurrent reactions of two independent species. These reactions are first order processes described by a double-exponential decay: T=T 0 + T, exp(-kit) + T 2 exp(-k2t) (3.14) where there are three types of traps created by the hydrogen atom exposure, one kind of trap (T 0 ) remained unchanged, the other two kinds of traps (Ti and T 2 ) are removed with rate constants k i and k 2 , that have activation energies of 0.21 eV and 0.16 eV respectively. B y monitoring the interfacial trap density during the oxidation of Si , SiGeo.3, Si capped SiGeo.025, and SiGeo.025 with atomic oxygen at ~200°C, we have observed that the presence of 2.5% Ge in the Si lattice does not result in the formation of larger concentrations of interfacial carrier traps than is formed in a pure Si lattice. Furthermore, just as exposure of the oxide layer to atomic hydrogen at ~200°C can result in the removal of a large portion of these traps in Si , it can cause an almost identical reduction in the SiGeo.025- However, in the case of SiGeo.3, this reduction was quite small. The H-atom treatment at 200°C results in a trap reduction that is comparable with that obtained with a high temperature H 2 anneal which is always used in the fabrication industry to reduce the interfacial traps and further improve the oxide interface quality. 96 References I George E . Anner, Planar Processing Primer, Van Nostrand Reinhold, N e w York, 1990. W . Scot Ruska, Microelectronic Processing, M c G r a w - H i l l , 1987. David L . Pulfrey and N.Garry Tarr, Introduction to Microelectronic Devices, Prentice Hal l , N e w Jersey, 1989. 4 Ad i r Bar-Lev, Semiconductors and Electronic Devices, Prentice Hal l , 1984. 5 James W. Mayer and S.S. Lau, Electronic Materials Science: For Integrated Circuits in Si and GaAs, Macmil lan Publishing Company, 1990. 6 Ben G . Streetman, Solid State Electronic Devices, Prentice Hal l , N e w Jersey, 1995. Robert W . Keyes, The Physics of VLSI Systems, Addison-Wesley, 1987. 8 J. D . Cressler, I E E E Spectrum, March, 49 (1995). 9 D . L . Harame, J. H . Comfort, J. D . Cressler, E . F. Crabbe, J. Y . - C . Sun, B . S. Meyerson, and T. Tice, I E E E Trans. Electron Devices, 42, 455 (1995). 1 0 G . L . Patton, J. H . Comfort, B . S. Meyerson, E . F. Crabbe, G . J. Scilla, E . D . Fresart, J. M . C. Stork, J. Y . - C . Sun, D . L . Harame, and J. N . Burghartz, I E E E Electron Device Lett. 11, 171 (1990). I I S. S. Iyer, G . L . Patton, J. M . C. Stork, B . S. Meyerson, and D . L . Harame, I E E E Trans. Electron Devices, 36, 2043, (1989). 12 T. E . Whal l and E . H . C. Parker, Journal of Materials Science: Materials in Electronics, 6, 249 (1995). Andres G . Fortino, Fundamentals of Integrated Circuit Technology, Prentice Hal l , Virginia, 1984. 1 4 S. M . Sze, Semiconductor Devices: Physics and Technology, John Wiley & Sons, 1985. 1 5 P. M . Garone, V . Venkataraman, and J. C. Sturm, I E E E Electron Device Letters, 13, 56 (1992). 1 6 S. C. Martin, L . M . Hitt, and J. J. Rosenberg, I E E E Electron Device Letters, 10, 325 (1989). 97 17 Sorab K . Ghandhi, VLSI Fabrication Principles: Silicon and Gallium Arsenide, John Wiley & Sons, 1983. 1 8 A . R. Srivatsa, S. Sharan, 0 . W . Holland, and J. Narayan, J. App l . Phys. 65, 4028 (1989). 1 9 G . L . Patton, S. S. Iyer, S. L . Delage, E . Ganin, and R. C. Mcintosh, Mat. Res. Soc. Symp. Proc. 102, 295 (1988). 2 0 J. P. Zhang, P. L . F. Hemment, S. M . Newstead, A . R. Powell , T. E . Whal l , and E . H . C. Parker, Thin solid films, 222, 141 (1992). 2 1 D . C. Paine, C. Caragianis, and A . F. Schwartzman, J. App l . Phys. 70, 5076 (1991). 2 2 H . K . L iou , P. M e i , U . Gennser, and E.S. Yang, App l . Phys. Lett. 59, 1200 (1991). 2 3 J. Eugene, F. K . LeGoues, V . P. Kesan, S. S. Lyer, and F. M . d'Heurte, App l . Phys. Lett. 59, 78 (1991). 2 4 D . K . Nayak, K . Kamjoo, J. S. Park, J. C. S. Woo, and K . L . Wang, App l . Phys. Lett. 57 ,369(1990). 2 5 D . Nayak, K . Kamjoo, J. C. S. Woo, J. S. Park, and K . L . Wang, App l . Phys. Lett. 56, 66 (1990). 26 F. K . LeGoues, R. Rosenberg, T. Nguyen, F. Himpsel, and B . S. Meyerson, J. Appl . Phys. 65, 1724(1989). 2 7 F. K . LeGoues, R. Rosenberg, and B . S. Meyerson, App l . Phys. Lett. 54, 644 (1989). 2 8 S. S. Iyer, P. M . Solomon, V . P. Kesan, A . A . Bright, J. L . Freeouf, T. N . Nguyen, and A . C. Warren, I E E E Electron Device Letters, 12, 246 (1991). 2 9 S. C. Jain and P. Balk, Thin Solid Films, 223, 348 (1993). 3 0 V . Q. Ho and T. Sugano, I E E E Trans. Electron Device, ED-27, 1436 (1980). 3 1 J. Kraichman, J. App l . Phys. 38, 4323 (1967). 3 2 D . L . Pulfrey, F. G . M . Hathorn, and L . Young, J. Electrochem. Soc. 120, 1529 (1973). 3 3 P. W . L i , H . K . L iou , and E . S. Yang, App l . Phys. Lett. 60, 3265 (1992). 98 3 4 M . Mukhopadhyay, S. K . Ray, C. K . Mai t i , D . K . Nayak, and Y . Shiraki, App l . Phys. Lett. 65, 895 (1994). 3 5 1 . S. Goh, J. F. Zhang, S. Ha l l , W . Eccleston, and K . Werner, IOP, 818 (1995). 3 6 M . Mukhopadhyay, S. K . Ray, T. B . Ghosh, M . Sreemany, and C. K . Mai t i , Semicond. Sci. Technol. 11, 360(1996). 3 7 M . Mukhopadhyay, S. K . Ray, C. K . Mai t i , D . K . Nayak, and Y . Shiraki, J. App l . Phys. 78 ,6135 (1995). 3 8 T. Hattori, Critical Reviews in Solid State and Materials Sciences, 20, 339 (1995). 3 9 M . J. Uren, J. H . Stathis, and E . Cartier, J. App l . Phys. 80, 3915 (1996). 4 0 E . Cartier and J. H . Stathis, App l . Phys. Lett. 69, 103 (1996). 4 1 W . L . Warren, K . Vanheusden, J. R. Schwank, D . M . Fleetwood, P. S. Winokur, and R. A . B . Devine, App l . Phys. Lett. 68, 2993 (1996). 4 2 K . L . Brower, Physical Review B 38, 9657 (1988). 4 3 K . L . Brower, Physical Review B 42, 3444 (1990). 4 4 J. H . Stathis in The Physics and Chemistry of S1O2 and the Si-Si02 Interface 2, (edited by C. R. Helms and B . E . Deal), Plenum, New York, 1993. 4 5 H . Fukuda, M . Yasuda, T. Iwabuchi, S. Kaneko, T. Ueno, and I. Ohdomari, J. App l . Phys. 72, 1906 (1992). 4 6 S. C. Vitkavage , E . A . Irene, and H . Z . Massoud, J. App l . Phys. 68, 5262 (1990). 4 7 M . L . Reed, and J. D . Plummer, J. App l . Phys. 63, 5776 (1988). 4 8 D . Tchikatilow, J. F. Yang, and E . S. Yang, App l . Phys. Lett. 69, 2578 (1996). 4 9 H . Lafontaine, D . C. Houghton, D . Elliot, N . L . Rowell , J. - M . Baribeau, S. Laframboise, G . I. Sproule, and S. J. Rolfe, J. Vac. Sci. Technol. B 14, 1675 (1996). 5 0 H . L i , E . A.Ogryzlo, T. Tiedje, and J. G . Cook, Proceedings of the 2 2 n d International Conference on the Physics of Semiconductors, 1, 541 (1995). 5 1 H . L i , and E . A . Ogryzlo, Can. J. Phys. 74, 685 (1996). 99 3 Z M Sakuraba, J. Murota, and S. Ono, J. App l . Phys. 75, 3701 (1994). 5 3 E . Yablonovitch, D . L M a r a , C. C. Chang, T. Gmitter, and T. B . Bright, Physical Review Letters 57, 249 (1986). 5 4 T. Takahagi, I. Nagai, A . Ishitani, H . Kuroda, and Y . Nagasawa, J. App l . Phys. 64, 3517(1988). 5 5 L . Elias, E . A . Ogryzlo, and H . I. Schiff, Can. J. Chem. 37, 1680 (1959). 5 6 A . J. Nelson in, Microanalysis of Solids, Edited by B . G . Yacobi, D . B . Holt and L . L . Kazmerski, Plemum Press. N . Y . 1994. 57 J. F. Watts, An Introduction to Surface Analysis by Electron Spectroscopy, Oxford University Press, N . Y . 1990. 58 H . Windawi and F. L . Ho, Applied Electron Spectroscopy for Chemical Analysis, John Wiley & Sons, U . S . A . 1982. 5 9 N . H . Turner in Investigations of Surfaces and Interfaces-Part B, Edited by Bryant W. Rossiter and Roger C. Baetzold, John Wiley & Sons, 1993. 6 0 L . C. Feldman and J. W . Mayer, Fundamentals of Surface and Thin Film Analysis, Elsevier Science, North-Holland, 1986. 6 1 J. F. Moulder, W . F. Stickle, P. E . Sobol, and K . D . Bornben, Handbook of X-ray Photoelectron Spectroscopy: a Reference Book of Standard Spectra for Identification and Intepretation of XPS Result. Pekin-Elmer Corporation, 1992. 62 M . H . K ibe l in Surface Analysis Methods in Materials Science, Edited by D. J. O'Connor, B . A . Sexton and R. St. C. Smart, Springer-Verlag, Germany, 1992. 63 Edited by D . Briggs and M . P. Seah, Practical Surface Analysis, Vol 1,(second edition), John Wiley & Sons, 1990. 6 4 P. H . Holloway and G . E . McGuire , Appl . Surface Sci, 4, 410 (1980). 6 5 F. Yano, A . Hiraoka, T. Itoga, H . Kojima, K . Kanehori, and Y . Mitsui , J. Vac. Sci. Technol. A 13, 2671 (1995). 6 6 T. Hattori, T. Aiba , E . Iijima, Y . Okube, H . Nohira, N . Tate, and M . Katayama, Applied Surface Science 104-105, 323 (1996). 100 Y . Yamashita, K . Namba, Y . Nakato, Y . Nishioka, and H . Kobayashi, J. Appl . Phys. 79, 7051 (1996). 68 Y . Yamashita, Y . Nakato, H . Kato, Y . Nishioka, and H . Kobayashi, Applied Surface Science, 117-118, 176 (1997). 6 9 K . B . Clark, J. A . Bardwell, and J. - M . Baribeau, J. App l . Phys. 76, 3114 (1994). 7 0 D . F. Mitchel l , K . B . Clark, J. A . Bardwell, W . N . Lennard, G . R. Massoumi, and I. V . Mitchell , Surface and Interface Analysis, 21, 44 (1994). 7 1 H . J. Mathieu, M . Datta, and D . Landolt, J. Vac Sci, Technol. A 3, 331 (1985). 7 2 A . Muto, T. Mine, and M . Nakazawa, Jpn. J. App l . Phys. 32, 3580 (1993). 7 3 Y . R. X i n g , J. A . W u and S. D . Y i n , Surface Science 334, 705, (1995). 7 4 O. Vancauwenberghe, O. C. Hellman, N . Herbots, and W . J. Tan, App l . Phys. Lett. 59, 2031 (1991). 7 5 V . Craciun, A . H . Reader, D . E . W . Vandenhoudt, S. P. Best, R. S. Hutton, A . Qndrei, and I. W . Boyd, Thin Solid Films 255, 290 (1995). 76 Joseph D . Andrade in Surface and Interfacial Aspects of Biomedical Polymers, Volume 1, Surface Chemistry and Physics, Edited by Joseph D . Andrade, Plenum Press, 1985. 77 S. Tanuma, C. J. Powell and D . R. Perm, Surface and Interface Analysis, 11, 577 (1988). 101