<%BANNER%>

Interactions of Atomic Oxygen with Pt(111) and Nitrided Si(100)


PAGE 1

INTERACTIONS OF ATOMIC OXYGEN WI TH Pt(111) AND NITRIDED Si(100) By ALEX GERRARD 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 2005

PAGE 2

Copyright 2005 by Alex Gerrard

PAGE 3

This dissertation is dedicated to my late grandfather, Delbert Barrow.

PAGE 4

iv ACKNOWLEDGMENTS I would first like to acknowledge my supervisory committee chair Dr. Jason Weaver for his guidance in the field of surf ace science. Working with him and building his laboratory has truly been a rich experience. I would also like to thank Dr. Gar Hoflund for his advice since I first arrived at th e University of Florida. He has always been generous with his time, expertise and opt imism. I am grateful for all of his support. During my tenure at UF, I had the privilege or working under Dr Crisalle and Dr. Chauhan as a teaching assistant. They taught me a great deal about teaching and effectively communicating complicated ideas to students. I am also indebted to James Hinant, Dennis Vince, Dr. Bosworth, Dr. Re n, Dr. Butler, Dr. Orazem, Dr. Anderson, and the rest of the faculty and staff of the Chem ical Engineering Department for their all of their assistance and support. I would also like to show my gratitude to my spiritual director, Fr. Ed. Murphy and my family, Alex, Mary, Anne, Allanah, and Aaron Gerrard for their guidance, kindness and encouragement. I am grateful to have made many friends and colleagues in Gainesville who have assisted me in virtually every aspect of my life, to many of whom to mention here. One particular soul who does deserve a great amount of recognition is Jau-Jiun Chen. I greatly apprec iate all of her dedicated ha rd work and her company. I truly treasure the relatio nship I have with her.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... xi CHAPTER 1 INTRODUCTION...................................................................................................1 1.1 Research Objective......................................................................................1 1.2 Literature Search..........................................................................................4 1.2.1 Reaction Mechanisms......................................................................4 1.2.2 Molecular and Atomic Oxygen Interactions with Si(100)...............7 1.2.3 Molecular and Atomic Oxyge n Interactions with Pt(111)...............8 1.3 Experimental System...................................................................................9 1.3.1 The UHV Chamber..........................................................................9 1.3.2 Sample Manipulator.......................................................................10 1.3.3 Sample Temperature Control.........................................................11 1.3.4 Beam Chamber...............................................................................12 1.3.5 Plasma Characterization.................................................................13 1.3.6 Gas Handling.................................................................................17 1.3.7 Calibrated Molecular Beam...........................................................17 1.3.8 Calibration of O2 Beam Flow from Plasma Source.......................22 1.4 Detection Techniques.................................................................................22 1.4.1 Reaction Product Monitoring.........................................................22 1.4.2 Temperature Programmed Desorption and Reaction.....................24 1.4.3 X-ray Photoelectron Spectroscopy................................................25 2 OXIDATION OF NITRIDED Si( 100) BY GASEOUS ATOMIC AND MOLECULAR OXYGEN.....................................................................................42 2.1 Introduction................................................................................................43 2.2 Experimental Methods...............................................................................47

PAGE 6

vi 2.3 Results........................................................................................................50 2.3.1 NH3 Decomposition on Si(100).....................................................50 2.3.2 Oxidation of Nitrided Si(100) at 300 K.........................................54 2.3.3 Oxidation of Nitrided Si(100) at 800 K.........................................59 2.4 Discussion..................................................................................................64 2.5 Conclusions......................................................................................................70 3 DYNAMIC DISPLACEMENT AND DISSOCIATION OF O2 ON Pt(111) BY ATOMIC OXYGEN..............................................................................................86 3.1 Introduction................................................................................................86 3.2 Experimental Methods...............................................................................88 3.3 Results........................................................................................................89 3.4 Discussion..................................................................................................98 3.5 Conclusions..............................................................................................104 4 CARBON MONOXIDE OXIDATION FR OM HIGH OXYGEN COVERAGE PHASES ON Pt(111)...........................................................................................109 4.1 Introduction..............................................................................................110 4.2 Experimental Methods.............................................................................113 4.3 Results......................................................................................................115 4.3.1 Carbon Monoxide Adsorption and TPR on O Precovered Pt(111) Surfaces...........................................................................115 4.3.2 CO Oxidation on High-Coverages of Chemisorbed Oxygen Atoms Under Isothermal Conditions...........................................123 4.3.3 Carbon Monoxide Oxidation from Platinum Oxide....................129 4.4 Discussion................................................................................................135 4.5 Conclusions..............................................................................................138 5 CONCLUSIONS AND SUGGESTI ONS FOR FUTURE WORK.....................150 5.1 Synopsis...................................................................................................150 5.2 Future Work.............................................................................................151 5.2.1 Nonthermal Reaction Mechanisms..............................................151 5.2.2 Scanning Tunneling Microscopy Measurements.........................152 5.2.3 Mechanistic Catalytic Studies......................................................152 LIST OF REFERENCES.................................................................................................154 BIOGRAPHICAL SKETCH...........................................................................................163

PAGE 7

vii LIST OF TABLES Table page 2-1 Oxygen coverages on clean and nitrided Si(100).....................................................73 3-1 Displacement and desorption yields of 18O2 when exposed to 16O-atoms while holding the temperature at 90, 100 and 110 K.......................................................105 3-2 Initial evolution rate, the estimated in itial dissociation rate, and the sum of the two rates when exposing an 18O2 saturated on Pt(111) to a beam with a flux of 0.005 ML/sec of 16O-atoms....................................................................................105 4-1 Maximum CO coverage achie ved as a function of initial O.................................140

PAGE 8

viii LIST OF FIGURES Figure page 1-1 Mechanisms describing surface chemical reactions..............................................26 1-2 UHV system constructed for our study..................................................................26 1-3 Top view of the UHV system................................................................................27 1-4 Specimen mount on the sample manipulator.........................................................28 1-5 Sample temperature control scheme used for these experiments..........................29 1-6 Typical linear temperature ramp with a heating rate of 1 K/sec............................30 1-7 Beam chamber housing the plasma source............................................................31 1-8 Appearance potential measurement, monitoring O+ (m/e 16) with the plasma activated and deactivated with a firs t stage beam chamber pressure of 3.0*10-5 Torr........................................................................................................................32 1-9 Mass spectrometer beam trace experiment............................................................33 1-10 Oxygen uptake on Pt(111) with a surface temperature of 450 K..........................34 1-11 Gas flow from the beam onto the sample..............................................................35 1-12 Hypothetical beam trace depicting th e partial pressure of a gas using a calibrated molecular beam, where A is proportional to the quantity of gas adsorbed on the sample..........................................................................................36 1-13 Molecular beam doser used for our experiments...................................................37 1-14 Pressure of carbon monoxi de in the gas reservoir pl otted as a function of time while pumping out through the laser drilled VCR Gasket.....................................38 1-15 Calibration of the molecular oxygen fl ow rate through the beam chamber is compared with the pressure in th e first differential pumping stage.......................39 1-16 Dynamic displacement of 18O2 taken upon exposing a Pt(111) surface saturated with 18O2 to 16O-atoms, while holding the sample temperature constant at 90 K.....................................................................................................40

PAGE 9

ix 1-17 Temperature programmed desorption performed on Pt(111) after exposing to a beam of oxygen atoms........................................................................................41 2-1 Si2p spectrum obtained from a Si(100) surface after a 160 ML NH3 exposure at a surface temperature of 900 K..........................................................................73 2-2 N1s spectrum obtained from a Si(100) surface after a 160 ML NH3 exposure at a surface temperature of 900 K..........................................................................74 2-3 N1s spectra obtained from Si(100)........................................................................75 2-4 Si2p spectra obtained after e xposing clean Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed lin e), and after depositing 1.2 ML of oxygen on the nitrided Si(100) surface held at 300 K using the plasmaactivated beam (solid line).....................................................................................76 2-5 N1s spectra obtained after exposing Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed line), and after depositing 1.2 ML of oxygen on the nitrided Si(100) surface held at 3 00 K using the plasma-activated beam (solid line)..............................................................................................................77 2-6 O1s spectra obtained after incorpora ting 1 ML of oxygen atoms on Si(100) at a surface temperature of 300 K (dashed line), and after depositing 1.2 ML of oxygen on the nitrided Si(100) surface held at 300 K using the plasmaactivated beam (solid line).....................................................................................78 2-7 N1s and O1s spectra obtained at electr on collection angles of 0 (dashed line) and 60 (solid line) after depositing 1.2 ML of oxygen on nitrided Si(100) at a surface temperature of 300 K.................................................................................79 2-8 Si2p spectra obtained after e xposing clean Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed lin e), and after depositing 1.5 ML of oxygen on the nitrided Si(100) surface held at 800 K using the plasmaactivated beam (solid line).....................................................................................80 2-9 N1s spectra obtained after exposing Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed line), and after depositing 1.5 ML of oxygen on the nitrided Si(100) surface held at 8 00 K using the plasma-activated beam (solid line)..............................................................................................................81 2-10 O1s spectra obtained after incorpor ating 2.4 ML of oxygen atoms on Si(100) at a surface temperature of 800 K (das hed line), and after depositing 1.5 ML of oxygen on the nitrided Si(100) surface held at 800 K using the plasmaactivated beam (solid line).....................................................................................82 2-11 N1s and O1s spectra obtained at electr on collection angles of 0 (dashed line) and 60 (solid line) after depositing 1.5 ML of oxygen on nitrided Si(100) at a surface temperature of 800 K.................................................................................83

PAGE 10

x 2-12 Model for O2 dissociation and incorp oration into Si(100).....................................84 2-13 Model for O-atom adsorption a nd incorporation into Si(100) ...............................85 3-1 Partial pressure traces of 18O2 desorbing from Pt(111) when exposed to 16Oatoms....................................................................................................................106 3-2 Temperature programmed desorption spectra taken after precovering the Pt(111) surface with 18O2 and exposing this surface to 16O-atoms......................107 3-3 Normalized dynamic displacement beam traces of 18O2 when exposed to a 16O-atom beam interrupted every thirty seconds.................................................108 4-1 Temperature programmed desorption spectra taken from the Pt(111) surface with a ramp rate of 3 K/sec..................................................................................140 4-2 Temperature programmed reaction spectra taken with a ramp rate of 1 K/sec, showing O2, CO and CO2 desorption after preparing a saturation coverage of CO on Pt(111) with 0.39 ML of O-atoms at Ts = 100 K.....................................141 4-3 Temperature programmed desorption and reaction spectra taken from the Pt(111) surface with a ramp rate of 3 K/sec........................................................142 4-4 Carbon dioxide rate cu rves with various initia l oxygen coverages at 300 K plotted against CO exposure................................................................................143 4-5 Carbon dioxide rate cu rves with various initia l oxygen coverages at 400 K plotted against CO exposure................................................................................144 4-6 Carbon dioxide rate curv es with various initial oxygen coverages at 500 K, plotted against CO exposure................................................................................145 4-7 Carbon dioxide formation rate plotted as a function of CO exposure with an initial O-atom coverage of 1.7 ML between 400 and 550 K...............................146 4-8 Carbon dioxide formation rate plotte d as a function of CO exposure with various initial O-atom coverages at 400 K..........................................................147 4-9 TPD traces taken with an initial coverage of 1.7 ML and with the same coverage after a 20 ML CO exposure (~0.5 ML) at 500 K.................................148 4-10 LEED images taken after reduc ing platinum oxide with CO..............................149

PAGE 11

xi 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 INTERACTIONS OF ATOMIC OXYGEN WI TH Pt(111) AND NITRIDED Si(100) By Alex Gerrard May 2005 Chair: Jason Weaver Major Department: Chemical Engineering Gas-phase oxygen atoms play a critical ro le in several applications including degradation of materials in low Eart h orbits, semiconductor processing, and heterogeneous catalysis. In each of these a pplications, the highly reactive nature of the oxygen atom dictates the details of the oxygen atom-surface reactions. Whether the goal is to develop new materials resistant to oxi dation or to use the oxygen atoms to generate surfaces with unique properties, a fundamental understanding is needed of the chemistry governing the interactions of O-atoms with surfaces. Silicon nitride is used in applications b ecause it resists oxidation. We conducted an X-ray photoelectron spectroscopy (XPS) st udy to examine the surface chemistry of nitrided Si(100) toward mo lecular and atomic oxygen. A decrease was observed in the Si(100) surface dangling bond density due to ni tridation, and this was accompanied by a subsequent decrease in surface reactivity for both molecular and atomic oxygen. This

PAGE 12

xii indicates that oxygen atoms preferentially re act at these dangling bond sites, and do not insert directly into silicon-silicon bonds. Using mass spectrometry, we explor ed the reactivity of gaseous 16O-atoms toward chemisorbed 18O2 on the Pt(111) surface. The gas-phase oxygen atoms stimulate both displacement and dissociation of adsorbed 18O2. As the surface temperature increased, the desorption yield decreased, with the bala nce remaining on the surface in the form of chemisorbed atoms. Molecular oxygen is known to adsorb in superoxo and peroxo configurations on Pt(111). The strong te mperature dependence on the dissociation yield is attributed to an increas e in the population of the peroxo chemisorbed species with temperature, which is more prone to dissociate. Platinum is catalytically active towa rd the oxidation of CO. Using O2 under ultrahigh vacuum conditions, a maximum surf ace coverage of 0.25 m onolayers (ML) of O-atoms can be generated. Surfaces with higher oxygen coverages can be generated by exposing Pt(111) to oxygen atoms. Using mass spectrometric methods, an in situ CO oxidation study was conducted on these surfaces. The CO oxidation kinetic behavior was found to be consistent with a CO adsorption precursor model. Carbon monoxide oxidation on platinum oxide occurs predomin ately at the interface between the metallic regions and the oxide.

PAGE 13

1 CHAPTER 1 INTRODUCTION 1.1 Research Objective Our main objective was to develop a funda mental understanding of interactions between oxygen atoms and technologically re levant surfaces. Oxygen atoms are present in semiconductor plasma processes and in low Earth orbits. These radicals are known to be significantly more reactive than their molecula r counterpart. It is believed that O-atom chemistry plays an important role in semi conductor processes, polymer erosion, and low pressure metal oxide formation. Conduc ting well characterized O-atom beam experiments on metal and semiconductor surf aces provides insight into the underlying chemistry in these areas. Silicon nitride and oxynitride films have been studied extensively in recent years because of the advantages afforded by inco rporating these materials in the dielectric layers used in metal-oxide-semiconducto r (MOS) devices. Adding small amounts of nitrogen to the SiO2-Si interface is known to improve the structural quality of the interface, and results in lower leakage current across the gate of a MOS device as well as enhanced resistance to boron diffusion into the SiO2 [1]. The low activity of silicon nitride toward oxidation has also proved bene ficial to the growth of alternative gate oxides such as Ta2O5 and ZrO2 that have higher dielectr ic constants (k) than SiO2. Recent studies show that depositing high k oxi des directly onto silicon can result in formation of an SiO2 layer that dominates the capacitance of the gate stack [2-6]. Incorporating nitrogen in the near-surface region of silicon alleviates this problem by

PAGE 14

2 inhibiting SiO2 formation during the deposition of Ta2O5, thereby maximizing the benefits of the high k oxide f ilm [3,7,8]. Another important application of silicon nitride is as a protective coating on ceramic component s such as bearings and turbine blades for which tolerance to high temperature oxidizing environments is critical. Despite these important applications, however, the oxida tion of silicon nitride films is not well understood at the molecular level. We studied the oxidation of a silicon nitride film by both gaseous atomic and molecular oxygen in ultrahigh vacuum (UHV), focusing our efforts on elucidating the fundamental origin of the oxidation resistance of nitrided silicon. X-ray photoelectron spectrosc opy (XPS) was conducted before and after thermally decomposing nitrogen on the Si(100) surface. The nitrided surface was then exposed to a beam of molecular or atomic oxygen. Changes in the surface due to the oxidant beam were characterized by XPS. Platinum is considered an important oxidation catalyst in several oxidation reactions. One of the first st eps in understanding heterogene ous catalysis on platinum is to learn about the molecular level oxygen inte ractions with this metal. The Pt(111) crystal face was selected for these studies beca use of the vast literature characterizing the interactions of molecular oxygen with this surface [9-15]. Molecular oxygen is known to physisorb onto Pt(111) at 20 K. When the surface is heated to 38 K, the molecular oxygen becomes chemisorbed, and generates a saturation coverage of 0.44 monolayers (ML) [11,14]. On heating the sample to 130 K, a competitive process occurs between dissociation and desorption of molecular oxygen, yielding a surface saturated with Oatoms at a coverage of 0.25 ML. The oxygen atoms become mobile on the surface at about 200 K, forming islands with a p(2x2) st ructure [14]. On continued heating, the

PAGE 15

3 atoms recombine and desorb at about 710 K. The atomic oxygen coverage does not increase beyond 0.25 ML due to the limiting O2 flux imposed under vacuum conditions. Our study used a beam consisting of atomic and molecular oxygen, and it is useful to characterize the surface behaviors of O-atom radicals with molecular oxygen on Pt(111). Oxygen atom impingement onto a Pt(111) surface saturated with molecular oxygen has been investigated previously [16,17]. Using isotopically labeled 18O2 adsorbed on the surface and gas-phase 16O-atoms, the displacement of 18O2 from the surface and the formation of 18O16O was observed at a surface temperat ure of 80 K. The displacement phenomenon was also observed when dire cting nitrogen and hydrogen atoms at the 18O2 covered surface. Employing time-of-flight mass spectrometry, the desorbing molecules were observed to have a bimodal energy di stribution [17]. This was found to be independent of the adsorbing atom. The lower energy desorption component is consistent with thermal evolution of 18O2 from the surface. The high-energy desorption feature is direct evidence that some energy from the adsorbing atom is transferred directly into the 18O2 before it desorbs. By probing the role of surface temperature, deeper insights into the nature of these phenomena may be found. Intuitively, one would anticipate an increase in th e surface temperature would amplify the thermal desorption rate. In our study, we measured molecula r oxygen displacement by impingent oxygen atoms as a function of the surface temperatur e. We also characterized oxygen uptake caused by impinging oxygen atoms. The adsorbed 18O2 was not found to be displaced by gas-phase 16O2, and hence no 16O2 adsorbed onto the surface. Any 16O-atoms found on the surface must, therefore, originate from the 16O-atoms in the beam.

PAGE 16

4 Platinum is considered an active oxidati on catalyst in many areas [18-20]. Along with probing oxygen atom-adsorbate r eactivity, oxygen atoms can generate high coverages of adsorbed oxygen atoms on Pt (111), previously unattainable using molecular oxygen in UHV (0.25 ML). Car bon monoxide oxidation on Pt(111) has been studied in detail with oxygen coverages ( O) on the order of 0.25 ML [8,21-40]. This makes CO an excellent probe molecule for ch aracterizing the relative reactivity of the high-coverage phases of oxygen at oms on Pt(111). Interestingl y, this also presents an opportunity to perform molecula r-level experiments on cataly tic surfaces that can be generated under high-pressure r eaction systems. In the experiments presented in chapter four, surfaces with high oxygen atom cove rages were generated by exposing clean Pt(111) to an oxygen atom beam. This surf ace was then exposed to a CO beam while holding the surface temperature constant and monitoring the background partial pressure of CO2 using a mass spectrometer. Temperat ure programmed desorption spectroscopy (TPD) was performed after each beam experi ment to quantify the products remaining on the surface. Next, I discuss gas-surface reaction mech anisms and the nature of O-atom interactions with Si(100) and Pt(111). To conduct such experiments, a UHV chamber was constructed and equipped with all of the equipment necessary to conduct the spectroscopic measurements described above The system and its capabilities are described in Section 1.3. 1.2 Literature Search 1.2.1 Reaction Mechanisms The studies presented here entail character izing the reactivity of oxygen atoms with surfaces and characterizing the surface cha nges caused by the gas phase O-atom. The

PAGE 17

5 free radical nature of the oxygen atom s uggests that any oxygen atom-adsorbate interactions will rapidly occur and should not be dependent on surface temperature. Characterizing these surfaces w ith a probe molecule (such as CO), TPD, or temperature programmed reaction (TPR) requires a sound understanding of thermal reaction chemistry occurring on solid su rfaces. Three distinct me chanisms describing surface reactions are discussed here. A classical description of a thermal he terogeneous catalytic reaction is the Langmuir-Hinshelwood (LH) mechanism. To sh ow how this mechanism works, consider the surface reaction A+B AB. First, both species A and B thermally accommodate to the surface. At sufficiently high surface temp eratures, these species become mobile and diffuse across the surface. Once the reactant s find each other and if they have enough energy to overcome the activation energy fo r reaction, they will r eact to form AB. Finally, AB desorbs from th e surface (Figure 1-1). Gas-phase oxygen atoms are highly reactiv e, and any interaction between this species and adsorbates or surfaces should not depend on surface temperature. It is then useful to understand fundamental nonthermal reaction mechanisms; the simplest is the Eley-Rideal (ER) mechanism. This mechanism is shown in the following reaction steps. First, one of the componen ts (A) thermally accommodates with the surface. An oncoming B particle collides with A in a sing le collision and they react to form AB. A portion of the reaction energy be tween A and B is transferred into the kinetic energy of the product molecule, allowing AB to immedi ately desorb (Figure 1-1). This energy could be transferred into the nuc lear motion of the products (the translational, vibrational, and rotational modes), which allows the reactio n products to desorb in a directed manner

PAGE 18

6 [41]. Note that this reaction has a very small cross-se ction, because B must strike A in the first collision, otherwise the B would eith er back scatter or stick to the surface. Alternatively, a nonthermal reaction may o ccur via the hot atom (HA) mechanism. The HA mechanism is very similar to the ER mechanism, and proceeds in the following manner. First, A thermally accommodates to th e surface. Particle B then collides with the surface and becomes trapped, but not ther mally accommodated to the surface. After several B-surface collisions, the B then collid es with A and reacts to form AB. The product AB then immediately desorbs from the surface (Figure 1-1). Since the B particle is trapped at the surface, it may collide fr equently with the surface, increasing the probability that B will react with A. This causes the cross-section for the reaction by the HA mechanism to be higher than that of the ER mechanism. The ER and HA mechanisms have distinct ki netic features, that are observed as the reaction products are monitored. Using oxygen atoms as the highly reactive “B” species, the ER mechanism should have a reaction cro ss-section similar to that of the atomic dimensions (10-16 cm2). The rate for the ER mechanism is given by the expression R= where is the flux of gaseous oxygen atoms at the surface, is the reaction cross-section, and is the adsorbate surface concentration. In this mechanism, the product formation rate would fi rst begin at the maximum, and then decay exponentially to the baseline. Through quantitative analysis of the rate data, the HA mechanism can be distinguished from the ER mechanism in the following ways. The HA mechanism may have a substantially larger cross-section b ecause of the multiple atom-surface collisions that occur before the reaction [41,42]. This also generates a surface concentration of these hot atoms on the substrate, which can infl uence the kinetics. For example, if sites

PAGE 19

7 are available on the surface befo re the reaction, the probability of an O-atom sticking in an available site will compete with the probabi lity of reacting with an adsorbed product A. This will yield a time delay before the rate maximum, which has been observed when H-atoms were directed toward surfaces co ated with D-atoms [42,43]. Additionally, unanticipated products may form when th e HA mechanism dominates the nonthermal chemistry. While directing hydrogen atoms onto deuterium-covered metal surfaces, Kammler et al. [42,43] observed the formation of HD and D2. The homo-isotopic product was attributed to the generation of secondary hot D-atoms. Secondary hot atom formation has been observed in a number of studies for gaseous H-atoms reacting with adsorbates [41,44-47] and can be detected by isotopically labeling the adsorbate molecule. 1.2.2 Molecular and Atomic Oxyg en Interactions with Si(100) Although UHV studies on the oxidation of silicon nitride films are sparse, oxidation of single crystal silicon surfaces has been studied extensively. Of particular relevance to our study are detailed UHV studies by Engstrom et al. [48] on the oxidation Si(100) and Si(111) by both gaseous atomic and molecular oxygen. These and other results were discussed by Engel [49] in a review of Si oxidation. Briefly, under UHV conditions, the dissociative adsorption of O2 on Si(100) yields effective saturation coverage of only about 1 ML of oxygen atoms when the surf ace is held at 300 K during oxidation. The saturation coverage can be increased by oxidizing in UHV at elevated surface temperature, but the oxygen uptake is still rather limited. For example, the saturation oxygen coverage increases to 2 ML when Si(100) is exposed to O2 at a surface temperature of 800 K. Not surprisingly, Engs trom et al. [48] found that gaseous oxygen atoms adsorb on Si(100) with much higher probability than does O2, and that oxygen

PAGE 20

8 coverages greater than 10 ML can be obtai ned by oxidizing Si(100) held at 300 K using an atomic oxygen beam. Unlike results obtained using O2, the uptake of oxygen atoms was insensitive to the surface temperature for oxygen coverages up to about 5 ML, which indicates non-activated adsorp tion and possibly direct inser tion of gaseous O-atoms into surface Si-Si bonds. 1.2.3 Molecular and Atomic Oxygen Interactions with Pt(111) Molecular oxygen interactions with Pt(111) have been studied extensively, and at 100 K O2 is known to chemisorb onto the Pt (111) surface, generating a saturation coverage of 0.44 ML [11,14]. As the surface is heated to 140 K, a competitive process between desorption and dissociation occurs. What remains after this process is a platinum surface saturated with oxygen atom s (0.25 ML). These oxygen atoms become mobile at about 200 K, and organize into is lands with a p(2x2) st ructure, with the Oatoms residing in the FCC hollow site. When the sample temperature reaches ~710 K, the adsorbed oxygen atoms recombine and deso rb [9-15]. Higher surface coverages of oxygen atoms have been formed by electron dissociation of O2/Pt(111) [9,50], and exposing Pt(111) to NO2 [51-53], O3 [54,55], and gaseous O-atom s [56]. Our research group successfully generated oxygen-atom c overages up to 2.9 ML on Pt(111) with a beam consisting of molecular and atomic oxygen. Oxygen atom uptake and the resulting high O-atom coverage phases on Pt(111) were examined using X-ray photoelectron spectroscopy (XPS), electron energy loss spectroscopy ( ELS), low energy electron diffraction (LEED), and TPD. As the surface coverage of oxygen on the Pt(111) surface exceeds 0.25 ML, the onset of O2 desorption shifts to lower temperatur es during TPD [51,55]. Two distinct

PAGE 21

9 low-temperature desorption features were obser ved at about 550 and 640 K. Parker et al. [51] attributed this to strong repulsive interactions among neighboring oxygen atoms as well as O-atoms adsorbed in weakly bound hcp hollow sites. As the coverage increases beyond 0.75 ML, the intensity of the low temp erature desorption features decrease. Accompanying this observation is the formati on of a single sharp desorption peak that shifts to higher temperatures with increasing O. Using XPS and ELS measurements in conjunction with TPD in our labor atory, this sharp desorption f eature is attributed to the presence of oxide islands forming on the surface. 1.3 Experimental System Probing the oxygen atom-surface reactivity at the molecular level required the construction of an ultrahigh vacuum chamber with the capability of monitoring gas phase species and surface properties. A calibrated be am system was designed to create uniform adlayers on the surface and to conduct isothermal reaction rate measurements. To generate the atomic oxygen beams used in th ese studies a microwave plasma source was employed. The plasma source was inserted into a chamber with two stages of differential pumping to maintain UHV during the plasma operation. Oxygen atoms can react with adsorbed species both on the sample and the chamber walls, which could influence the measurements during the beam experiments. This was addressed by designing a set of collimating apertures for the differentially pumped beam chamber, which minimized the pressure rise in the UHV chamber and focused the beam onto the sample. 1.3.1 The UHV Chamber The experiments were conducted in a thre e-level UHV chamber (Figure 1-2) that reaches a base pressure less than 2 x 10-10 Torr. The chamber is evacuated by an ion pump (400 L/sec), a turbo molecular pump (210 L/sec), and a titanium sublimation pump

PAGE 22

10 that is inserted into a liquid nitrogen cooled cryoshield. The upper level of the chamber (Figure 1-3) houses a hemisphe rical analyzer (Specs EA10 plus), a dual Al/Mg anode Xray source (Specs XR-50), a variable-energy electron source (Specs EQ 22/35) and an ion sputter source (Specs IQE 11), for performi ng XPS, Auger electron spectroscopy (AES), ELS and low energy ion scattering spectroscopy (LEISS) as well as surface cleaning by ion sputtering. The middle level of the ch amber is designed primarily for gas-dosing, and contains a directed doser connected to a leak valve as well as a calibrated molecular beam doser, the design of which closely follows that described by Yates [57]. The lower level of the chamber (Figure 1-3) houses low-ener gy electron diffraction (LEED) optics, a twostage differentially-pumped plasma beam source, and a quadrupole mass spectrometer (QMS). 1.3.2 Sample Manipulator A custom-designed sample manipulator (McAllister Technical Services) was mounted to the top of the UHV chamber. Samp le motion in all three Cartesian directions and rotation by 360o about the vertical axis were accomplished using an XY translator, a single-axis translator and a rotary platform. The specimen holder mounts to a copper reservoir that is brazed to the bottom of a st ainless steel tube. The top end of the tube mounts onto the rotary platform via a flange with multiple feedthroughs for attaching thermocouples and power wires to the sample A flat copper plate protrudes from the bottom of the copper reservoir and L-shaped co pper legs are bolted on each side of the plate, with sapphire spacers pr oviding electrical isolation between the reservoir and the copper legs (Figure 1-4). The copper legs are each 0.25” wide, extend 2” below the bottom of the copper plate, and are separated from one another by a distance of 1.12”. A specimen is mounted to the sample holder e ither by directly cli pping each side of the

PAGE 23

11 specimen to Ta plates that are bolted to the front of the flat copper legs, or by spotwelding wires to both the specimen a nd the Ta plates. The Ta plates are in thermal contact with the copper reservoir and can be c ooled to temperatures as low as 80 K when the reservoir is filled with liquid nitrogen. 1.3.3 Sample Temperature Control Accurate temperature control is requir ed during the experiments and to conduct TPD measurements. The following PID control scheme was used for this purpose and is depicted in Figure 1-5. Sample heating is performed resistively by passing current through the sample and the surface temperature is measured with a K-type thermocouple attached to the back of the specimen. The heating current is generated by a Sorenson DCS 33-33 DC power supply, and the sample te mperature is regulated using an Omega cn 1166 PID controller. This particular cont roller has a recorder and controller output. The controller output is wire d into the analog control i nput on the Sorenson power supply. A computer records the output uti lizing the analog input port located on the mass spectrometer. One requirement for analyzing TPD data accurately is the ability to generate a linear temperature ramp. Typically, the heat ing ramp consists of two components: an initial nonlinear transient component, fo llowed by a linear ramp. The temperature controller must be tuned to maximize the li near region of the ramp and minimize the duration of the transient featur es. Figure 1-6 shows a typica l heating curve for a 1 K/sec ramp rate. A concern with this heating configura tion is that the specimen mount may be damaged when applying current to the sample. This can be minimized if the following

PAGE 24

12 suggestions are followed. Occasionally, th e power supply or cont roller will malfunction and send a power spike across the sample, when initially activated. To avoid this, open the switch in the high current circuit (Figure 1-5) before activating the power supply or controller. When manually changing the set-point, it is safe to rapidly decrease the temperature. However, if the set-point is rapidly increased, even if the actual temperature is 100 C higher than the setpoint, the controller may de liver enough current across the sample to damage the mount. If a temperature ramp is desired, it is imperative to initiate the ramp while the set-point is within 10 C of the measured temperature. 1.3.4 Beam Chamber A two-stage differentially-pumped cham ber containing a commercial plasma source (Oxford Scientific Instruments) is at tached to the UHV chamber and was used to maintain ultrahigh vacuum conditions in the main chamber while operating the plasma source (Figure 1-7). Gaseous oxygen atoms are produced in this system by dissociating O2 in a plasma that is confined to a small re servoir at the end of the plasma source. The plasma source operates at a microwave frequency (2.45 GHz) and employs electron cyclotron resonance to increase the plasma de nsity. The plasma reservoir is fabricated from high purity alumina to minimize atom recombination and is terminated by a 2 mm thick alumina plate that has five thru holes, each of 0.4 mm diameter, that are arranged in a centered-(2x2) pattern within a 2 mm ar ea. Species exit the plasma volume through these holes, and form a beam that is direct ed toward the sample surface held in the UHV chamber. In the first pumping stage, the beam passes between oppositely charged parallel plates (10 kV/cm) that deflect ions and electrons from th e beam path. After flowing through a conical skimmer ( = 3 mm) separating the firs t and second pumping stages,

PAGE 25

13 the gas travels down a quartz tube before en tering the UHV chamber. The quartz tube is 60 mm long and has an inner diameter of 6 mm. The quartz tube provides a lower conductance between the source and UHV chambers than would be obtained with a thinwalled orifice of the same diameter. The quart z tube also provides sufficient collimation to confine the beam to the sample surf ace, which facilitate s reactive scattering measurements using mass spectrometry. In addi tion, collisions at th e inner walls of the tube are expected to reduce the fraction of atoms and molecules in vibrationally and electronically excited states, which should result in beams containing primarily ground state species, specifically O(3P) and O2(3g -). The first pumping stage of the beam chamber is evacuated with a 1200 L/sec diffu sion pump (Varian VHS 4) and the second stage is evacuated with a 70 L/sec turbo molecular pump and a titanium sublimation pump mounted inside a liquid-nitrogen cooled cryoshield. A mechanical shutter is located in the first pumping stage, which en ables control over beam introduction into the main UHV chamber. 1.3.5 Plasma Characterization The oxygen beam generated using the plasma source is comprised primarily of molecular and atomic oxygen. Before inves tigating the surface chemistry of gas-phase oxygen atoms, it is imperative to quantify the O-atom concentration in the beam. Several plasma radical quantification tools including ultraviolet, visible, vibrational, and ionization spectroscopies have successfully b een employed to quantify the radical density in plasmas. The ultrahigh vacuum chamber, constructed for these studies, is equipped with a mass spectrometer, allowing the use of mass spectrometric techniques to quantify the O-atom beam concentration.

PAGE 26

14 The first method used to detect the presen ce of O-atom radicals in the beam was line-of-sight threshold ionization mass sp ectrometry otherwise known as appearance potential spectroscopy, as disc ussed by Agarwal et al. [58]. This technique is based on the following principle. Electron impact ionization of oxygen atoms and molecules can yield the same ion. Consider generating O+ ions for detection with the mass spectrometer. They can be generated by eith er direct ionization of O-atoms or through O2 dissociation, as depicted by reaction Equations 1-1 and 1-2. O + eO+ + 2e(E1 = 13.8 eV) (1-1) O2 + eO+ + O +2e(E2 = 18.0 eV) (1-2) The parameters E1 and E2 denote the threshold electr on energies to generate O+ ions through the processes sh own in Equations 1-1 and 1-2 re spectively, and are taken from reference 58. The direct ionization process (1-1) always has a lower threshold than the dissociative process (1-2) due to the additional energy requ ired to break a molecular bond. This difference provides a means of det ecting oxygen radicals in the plasma beam by monitoring the 16 amu signal intensity as a function of electron energy. Figure 1-8 shows appearance potential measurements take n with and without the plasma activated after a background subtraction ta ken at the lowest electron energy probed (12 eV). The beam with the plasma not activ ated shows the formation of O+ ions before the threshold energy of 18.0 eV. This is attribut ed to thermal dissociation of O2 on the hot filament in the ionizer, which is subsequently ionized and detected. Comparing the two traces shown, it may be seen that the O+ ion signal at E < 18 eV incr eases by about an order of magnitude after activating the plasma. This increase provides di rect evidence that Oatoms are present in the plasma-activated beam.

PAGE 27

15 Alternatively, beam experiments may be performed to estimate the O-atom cracking fraction and at the same time search for contaminants. This allows the changes of the beam components to be quantified wh en activating or deac tivating the plasma beam. This method is complicated in that introducing a beam into the chamber may also displace other gases from the chamber wa lls, which may be interpreted as beam components. The predominant components in the oxygen beam with the plasma not activated are 32 and 16 amu, with trace cont aminates of 2 amu (hydrogen) and 28 amu (nitrogen). The measured 16 amu component ar ises from electron impact dissociation of O2 in the mass spectrometer ionizer. The i on impact dissociation fraction was measured at a value of 10%, which is close to the li terature value of 12.3 % for 70 eV electrons [59]. When the plasma was activated, the predominant co mponent was still 32 amu, but an increase in the signals of masses 16, 18 and 30 amu was observed concurrent with a decrease in 28 and 32 amu. The rise in 16 am u is attributed to oxygen atom formation by the dissociation of O2, the rise in mass 18 is attributed to oxygen atom reactions with hydrogen to form water and plasma source out gassing, and mass 30 and is attributed to the conversion of nitrogen (mass 28) in to NO (mass 30). The mass spectrometer sensitivity was increased to enhance the 16O atom signal, and masses 16, 18, 28 and 30 amu are shown in Figure 1-9 to compare the re lative intensities of each beam component. At time equal to zero, the plasma has already been activated, and the shutter is closed. After 85 seconds, the shutter was opened and a rise in the mass 16 signal was observed. Two hundred seconds elapse and then the plas ma is deactivated without changing the O2 feed rate to the plasma discharge tube. The 16 amu signal was observed to drop. This

PAGE 28

16 technique shows clearly the presence of oxygen radicals in the beam. Masses 18, 28, and 30 are shown in Figure 1-9 to illustrate th e low contaminant levels in the beam. The dissociation fraction may be estimate d in the following manner. When the plasma is deactivated, it is reasonable to assu me that the sole source of the m/e 16 signal is due to the dissociation of molecular oxygen in the ionizer. With an electron energy of 70 eV, this fraction is 10%, which provides a measure of the molecular oxygen flow rate with the plasma off. Since the pressure in the beam chamber remains constant when the plasma is deactivated, the beam flow rate is also assumed to remain constant. With this information, a mass balance can be performe d on molecular and atomic oxygen and the dissociation fraction may be estimated using the following relationships. 10 0 ) (0 2P P Ooff Tot (1-3) ) )( 1 ( 10 0 ) ( 20 0 0P P f P P f P Poff off on (1-4) Here, Pon, Poff, P0, O2tot, and f are the 16O partial pressure when the plasma is on, the 16O partial pressure when the plasma is off, the baseline pressure, the total 16O2 entering the ionizer and the dissociation fraction, respectiv ely. Equation 1-3 provides a measure of the total flow rate of O2 into the ionizer and the sec ond expression may be solved to determine the dissociation fraction of O-atoms (f). Solving this expression for f yields an estimation of the O2 dissociation fraction of about 3% (flux of about 0.02 ML/sec). It should be noted that the flux of O2 into the ionizer consists of a background and a direct component. The background component can be measured by obstructing the beam flow path with the sample manipul ator, preventing any direct O2 flux into the mass

PAGE 29

17 spectrometer. Once this value was obtained, it was subtracted off of the mass 16 partial pressure trace, providing a measure of the direct O2 flux into the ionizer. The final method used to determine the O-atom flux was to measure the oxygen uptake on Pt(111). Figure 1-10 shows the O-atom uptake curve as a f unction of exposure time while the sample was held at a constant temperature of 450 K. It is known that molecular oxygen will dissociatively adsorb ont o the Pt(111) surface at this temperature and saturate at 0.25 ML. Therefore, any a dditional amount of oxygen deposited onto the surface is attributed to the adsorption of oxyge n atoms. Within the first 180 seconds of the atomic oxygen exposure, the uptake curve is approximately linear. Assuming a unity sticking probability in this uptake region, th e slope provides an approximation of the Oatom flux on the sample surface, which is 0.03 ML/sec. This is in reasonable agreement with the quantity determined by mass spectrometric methods es pecially considering that the sample is closer in proximity to the b eam source than the mass spectrometer, yielding a higher atomic oxygen impingement rate onto the sample. 1.3.6 Gas Handling A inch VCR manifold is used to direct the gases to various ports on the system. To minimize the introduction of contaminants in the gas, the manifold is evacuated with a 70 l/s turbo pump. The pressure inside the manifold is monitored using a thermocouple gauge tube. Gas lines from the manifold are connected to gas cylinde rs and all gas entry ports into the chamber. 1.3.7 Calibrated Molecular Beam A typical means of exposing a sample to a gas in vacuum is to fill backfill the chamber with a gas for a measured period of time. The product of the pressure and the time provides a measure of the gas exposure on the surface. For example, an exposure of

PAGE 30

18 about one monolayer of gas onto the surface corresponds to about one Langmuir, which is 10-6 Torr*sec. This poses a problem when ga s dosing with “sticky” molecules such as water or ammonia, which could lead to l ong pump out times. Molecular beams provide a solution to this problem by providing an enha nced exposure to the sample surface while minimizing the gas load on the system. Seve ral advantages are afforded by employing a calibrated molecular beam. If the gas flow rate is known, it becomes possible to directly measure gas uptake onto the surface and the sticki ng probability of the gas with coverage. The beam operates under the following principles and is depicted by Figure 1-11. An example background partial pres sure trace is shown in Figure 1-12. A mass balance on the gas-phase beam species is shown in Equation 1-5. net ad gR dt dN out (1-5) The quantities Ng, Rad net, and out are the number of gas molecules in the system, the net adsorption rate, the total molecular flow ra te into the system and the total molecular flow rate out of the system, respectively. For large 'pS 0dt dNg (1-6) Equation 1-6 will be justified later. Substituting P'pS for out and rearranging Equation 1-6 yields: ) ( R 1 ) (adt S t Pnet p (1-7) And

PAGE 31

19 ) ( ) S( d net adR F R (1-8) where is the relative adsorbate coverage, S() is the sticking proba bility as a function of coverage, F is the intercepted fraction, and Rd() is the desorption rate from the surface. If the exposure is conducted below the deso rption temperature of the adsorbate, Rd() can be neglected. Substituting Equation 1-8 into 1-7 yields ) ( 1 S ) (pFS t P (1-9) At t = Rad net = 0, then P = / 'pS and P(t)=[1-PFS()], where P and P(t) are the steady state pressure after adso rption has ceased and the partia l pressure as a function of time. Substituting this information into Equation 1-9 and rearranging yields the following expression: ) (p net adS t P P R (1-10) 0 0 d R d Rnet ad t net ad (1-11) Winkler et al. performed calculations to determine the intercepted fraction (F) as a function of doser-sample distances and geomet ries [60], which allows F to be read directly from a chart provided in the reference. This de rivation shows that adsorption rates and sticking probabilities can be de termined directly from the background gas partial pressure traces. The relative coverage can be determined by Equation 1-11. The next portion of this derivation will justif y the approximation given by Equation 1-6.

PAGE 32

20 Assuming the sample is already saturated with the gas, and no adsorption on a sample is occurring, Equation 1-5 can be written as Equation 1-12. PS pdt dNg (1-12) Substitute Ng=PV/RT and 'pS =Sp/RT, where Sp is the pumping speed in L/sec, R is the gas constant, V is the chamber volume in L, and P is the pressure in Torr. V S P V dt dPpRT (1-13) Let p equal the pumping time constant, where p= V/Sp. For our system with only the turbo pump evacuating the chamber, sec 25 0 sec / 200 50 L Lp (1-14) Equation 1-13 is a first order differential equation, when solved yields Equation 1-15. ) ( ) (0P P e P t Ppt (1-15) With such a small p, the system pressure will reach P with shorter time scales than adsorption or reaction. Figure1-13 shows a schematic of the beam doser that has been mounted to the vacuum chamber via a single-axis translator The calibrated molecular beam doser was assembled in the following manner. The primary components required for the beam doser were machined by the A&N Corporation, and consist of a inch stainless steel tube passing through a CF 275 blank flange a nd a head piece that mates with the tube through a VCR connection. A 10 micron diam eter inch VCR gasket was obtained from Lenox Laser, and is used as the flow orifice. A glass micr o-capillary array was obtained from Buhrl electro-optics and was mounted in the doser head to provide a

PAGE 33

21 directed flux to the sample. The array is he ld place by a stainless steel cap. Outside of the vacuum chamber, a 4-way CF 275 cross is us ed as the gas reservoir. Mounted to the cross is a Baratron gauge capable of reading pressures from 0.01 to 10 Torr, inlet and evacuation valves, and a line running up to th e molecular beam doser. The flow rate through the beam was controlled by regulating the pressu re in the cross. To calibrate the beam flow, the conductance across the pi n hole was determined by the following procedure. First the volume of the cross and all lines l eading up to the gasket was measured (255 ml). The cross was then char ged with a gas to an initial pressure P0. The gas was allowed to flow out of the reservoi r and through the pin-hole, while recording the pressure inside the cross for an excess of 10 hours. By performing a mass balance of the gas inside the cross, the pump out time c onstant was related to the conductance of the orifice and the volume of the cross, in the following way. The gas flow rate through an orific e of a known conductance is given by the following expression, 2 1 1dt d P P C P (1-16) where C is the conductance, and P1 and P2 correspond to the pressure in the reservoir and in the vacuum chamber, respectively. Since the pressure in the vacuum chamber is on the order of 10-10 Torr, P2 can be neglected. The conduc tance of the pin hole may be calibrated by measuring the pressure of the reservoir as time progresses. The pump out rate follows the expression shown in Equa tion 1-17. Upon integration it yields an exponential decay shown in Equation 1-18. A pl ot of the pressure de cay in the reservoir as a function of time is shown in Figur e 1-14 and yields a time constant of 31758

PAGE 34

22 seconds. In this example, the conductance is 8.03*10-6 L/sec using carbon monoxide as the beam gas. Cp dt dp V (1-17) V Cte p p/ 0 (1-18) 1.3.8 Calibration of O2 Beam Flow from Plasma Source The addition of the calibrated molecular beam to the system also provided the means for calibrating the flow rate of mo lecular oxygen through the plasma beam. Initiating the beam caused a rise in the pressure in the first stage of the beam chamber. Correlating the pressure rise in the first ch amber with the oxygen flow rate provides a simple means of checking the beam flow rate Using the mass spectrometer, the partial pressure of O2 can be correlated with the beam chamber pressure. The pressure curve is then converted to number flow by comparing the measurements with those obtained by the calibrated molecular beam. A calibration curve is then obtained and is shown in Figure 1-15. 1.4 Detection Techniques 1.4.1 Reaction Product Monitoring In surface adsorption and chemical reacti on systems, real time reaction kinetic data can be obtained by monitoring the tem poral evolution of the background partial pressures of both reactants and products duri ng the reaction. For a given experiment, a sample is prepared and exposed to a reactant beam. A derivati on is given to illustrate the relationship of the monitored pa rtial pressures to the reaction rate and is similar to the derivation shown in Section 1.3.7. For ex ample, consider the dynamic displacement of

PAGE 35

23 18O2 from Pt(111) due to the impingement of 16O-atoms as shown in Figure 1-16. In this experiment, the surface was initially covered with 0.44 ML of 18O2 and then subsequently translated into the plasma beam path, while the shutter was closed. At time zero, the shutter was opened. The partial pressure of 18O2 initially jumped, and went through a maximum and then decayed to the baseline. First, assume that the system is initially at steady state (t < 0 sec.), with a constant background of 36 amu. This background is at tributed to a consta nt leak rate of 18O2 and described by Equation 1-19, eq PP KS L (1-19) where L is the leak rate, K is the co nversion constant to molecular flow, Sp is the pumping speed in L/sec, and Peq is the equilibrium pressure in Torr. The product K*S*Peq is the rate at which gas is pumped out of the system. The leak rate could originate from species displaced from the walls or from the beam. Assume also that the gas desorbing from the surface does not re-ads orb during the course of the experiment. This is reasonable under the conditions examined in this study. A mass balance around the desorbing species is shown in Equation 1-20. dt dp KV KSP L t A (1-20) Here, A is the sample area, V is the volume of the chamber, and is the desorption rate. Since the flux of the reagen ts remains constant during the experiment, the desorption rate can either be calculated as a function of reactant fluence or time. The second term on the right hand side of Equati on 1-20 depicts the particle accumulation in the system due to a rise in the partial pr essure. Combining Equations 1-19 and 1-20 yields the following expression.

PAGE 36

24 dt dp KV KSP KSP t Aeq (1-21) Substitute a = A/KV, P*=P-Peq, =V/S, where is the characteristic pump out time constant. dt dP P t aN* (1-22) In Section 1.3.7, I showed that the derivative term coul d be neglected, leaving the desorption rate directly proportional to the pa rtial pressure traces obtained from the mass spectrometer. The area under the curve shown in Figure 116 is then directly proportional to the desorption yield of 18O2. If the desorption yield is know n, the data shown in Figure 1-16 can be directly converted to the desorption ra te in ML/s using the following expression, where s is a dummy integration variable, P is the pressure of the system minus the baseline, and 0 is the initial coverage. 0 0) ( ) ( ds s P t P rate (1-23) 1.4.2 Temperature Programmed Desorption and Reaction Temperature programmed desorption and reac tion techniques were used to quantify the amount of each species residing on the surface. These spectra also contain information regarding the deso rption activation energy for each species [61,62]. The derivation relating the part ial pressure to the desorption rate shown in 1.3.7, also pertains to TPD analysis. Figure 1-17 shows an example TPD spectrum of atomic oxygen recombining on Pt(111) and desorbing as O2 with O equal to 0.25, 0.38, and 0.59 ML. Integrating the signal intensity of each sp ectrum gives the total amount of products

PAGE 37

25 desorbing. Desorption features are observed at 550, 640, and 710 K. Each of these features is attributed to oxygen atoms resi ding in environments with different surfaceadsorbate binding strengths. In this particular example, one can see the lower temperature desorption features increasing with O, indicating that the oxygen atoms are experiencing strong lateral repulsions from the neighboring atoms and have a weaker oxygen atom-surface bond. 1.4.3 X-ray Photoelectron Spectroscopy X-ray photoelectron spectroscopy is a t echnique which entails illuminating the sample with X-ray radiation, and examini ng the energy of the photoelectrons evolving from the surface. The kinetic energy of th ese electrons is given by the terms of the Equation 1-24, BE h KE where h is the photon energy, BE is the bi nding energy of th e electron, and is the work function difference between the sample and th e electron analyzer. Each element has a unique electronic structure, and hence has its own unique XPS spectrum. The binding energy is a function of the lo cal element bonding environment, giving this technique the means to probe the chemical state of each el ement. The photo-electrons are generated in the near surface region of the sample. The depth probed with this technique is characterized by the following expression, cos d (1-25) where d, and represents the distance from the surface, electron mean free path, and the electron take off angle respectively. By varying the angle at which the photo-

PAGE 38

26 electrons are taken, XPS will probe the concentr ations of each species at different depths within the surface. This forms th e basis for angle-resolved XPS. Figure 1-1. Mechanisms describing surf ace chemical reactions. Left) The LH mechanism. Middle) The ER mech anism. Right) The HA mechanism. Figure 1-2. UHV system c onstructed for our study. A B A B A B A B A B A B A A A B A B A B A A A B A B

PAGE 39

27 A) B) Figure 1-3. Top view of the UHV system. A) Upper level. B) Lower level.

PAGE 40

28 Figure 1-4. Specimen mount on the sample manipulator.

PAGE 41

29 Figure 1-5. Sample temperature control scheme used for these experiments. Current is driven across the sample by the DC power supply, which is regulated by the controller. Sample TC Power supply Power interrupt switch

PAGE 42

30 Figure 1-6. Typical linear temperature ra mp with a heating rate of 1 K/sec.

PAGE 43

31 A) B) Figure 1-7. Beam chamber housing the plasma source. A) A cross-sectional top view. B) A side view.

PAGE 44

32 Figure 1-8. Appearance potenti al measurement, monitoring O+ (m/e 16) with the plasma activated and deactivated with a firs t stage beam chamber pressure of 3.0*10-5 Torr.

PAGE 45

33 Figure 1-9. Mass spectrometer beam trace experiment taken with the plasma initially activated and the shutter closed with an electron energy of 70 eV. Masses 16, 18, 28 and 30 were monitored. Afte r 85 seconds, the shutter was opened. Once 200 seconds have elapsed, the plasma was powered down, while maintaining the same O2 flow rate.

PAGE 46

34 Figure 1-10. Oxygen uptake on Pt(111) with a surface temperature of 450 K. The oxygen uptake is approximately linear in the range of 0-180 seconds, with a slope of 0.03 ML/sec.

PAGE 47

35 Figure 1-11. Gas flow from the beam onto the sample, where , F, and out represent the total molecular flow rate into the chamber (#/sec), the intercepted fraction by the sample and the total molecular flow rate out of the system. The system pressure (Torr) and the pumping speed (#(sec)-1(Torr)-1) are denoted as P and 'pS respectively. out= P*'pS Sample F

PAGE 48

36 Figure 1-12. Hypothetical beam trace depic ting the partial pressure of a gas using a calibrated molecular beam, where A is proportional to the quantity of gas adsorbed on the sample. Pb P0 P Beam activated Beam deactivated Time A

PAGE 49

37 A B Figure 1-13. Molecular beam doser used for our experiments. A) Overall beam doser view. B) Beam doser head piece.

PAGE 50

38 Figure 1-14. Pressure of carbon monoxide in the gas reservoir plot ted as a function of time while pumping out through the lase r drilled VCR Gasket. The first order decay time constant is 31758 seconds.

PAGE 51

39 Figure 1-15. Calibration of the molecular oxyg en flow rate through the beam chamber is compared with the pressure in the first differential pumping stage.

PAGE 52

40 Figure 1-16. Dynamic displacement of 18O2 taken upon exposing a Pt(111) surface saturated with 18O2 to 16O-atoms, while holding the sample temperature constant at 90 K.

PAGE 53

41 Figure 1-17. Temperature programmed deso rption performed on Pt(111) after exposing to a beam of oxygen atoms. The surface coverage of oxygen atoms from lowest to highest coverage is 0.25, 0.38 and 0.59 ML.

PAGE 54

42 CHAPTER 2 OXIDATION OF NITRIDED Si(100) BY GASEOUS ATOMIC AND MOLECULAR OXYGEN The nitridation of Si(100) by ammonia and th e subsequent oxidation of the nitrided surface by both gaseous atomic and molecula r oxygen was investigated under ultrahigh vacuum (UHV) conditions using X-ray photoele ctron spectroscopy (XPS). Nitridation of Si(100) by the thermal decomposition of NH3 results in the formation of a subsurface nitride and a decrease in the concentra tion of surface dangling bond sites. Based on changes in the N1s spectra obtained after NH3 adsorption and decomposition, we estimate that the nitride resides about four to five layers below the vacuum-solid interface and that the concentration of surface dangling bonds after nitr idation is only 59% of its value on Si(100)-(2x1). Oxidation of the nitrided surface is found to produce an oxide phase that remains in the outer layers of the solid, and interacts only weakly w ith the underlying nitride for oxygen coverages up to 2.5 ML. Slight change s in the N1s spectra caused by oxidizing at 300 K are suggested to arise prim arily from the introduction of strain within the nitride, and by the formation of a small amount of Si2-N-O species at the nitr ide-oxide interface. The nitrogen bonding environment changes neg ligibly after oxidizing at 800 K, which is indicative of greater phase separation at elevat ed surface temperature. Nitridation is also found to significantly reduce the reactivity of the Si(100) surface toward both atomic and molecular oxygen. A comparison of the oxyge n uptake on the clean and nitrided surfaces shows quantitatively that the decrease in da ngling bond concentration is responsible for

PAGE 55

43 the reduced activity of the nitrided surface to ward oxidation, and therefore that dangling bonds are the initial adsorpti on site of both gaseous oxyge n atoms and molecules. Increasing the surface temperat ure is found to promote the uptake of oxygen when O2 is used as the oxidant, but bri ngs about only a small enhancement in the uptake of gaseous O-atoms. The different effects of surf ace temperature on the uptake of O versus O2 are interpreted in terms of the efficiency at which dangling bond pairs are regenerated on the surface. In particular, it is suggested th at elevated surface temperatures promote subsurface oxygen migration and concomitant regeneration of empty surface dimers, which are required for O2 activation. In contrast, the av ailability of si ngle dangling bonds needed for the adsorption of a gaseous oxygen atom is suggested to be relatively unaffected by the surface temperature. 2.1 Introduction Silicon nitride and oxynitride films have b een extensively investigated in recent years due to the advantages afforded by incor porating these materials into the dielectric layers used in metal-oxide-semiconducto r (MOS) devices. The addition of small amounts of nitrogen to the SiO2-Si interface is known to improve the structural quality of the interface, and results in lower leakage cu rrent across the gate of a MOS device as well as enhanced resistance to boron diffusion into the SiO2 [1]. The low activity of silicon nitride toward oxidation has also proved bene ficial to the growth of alternative gate oxides such as Ta2O5 and ZrO2 that have higher dielectr ic constants (k) than SiO2. Recent investigations have shown that the deposition of high k oxides directly onto silicon can result in the formation of an SiO2 layer that dominates the capacitance of the gate stack [2-6]. Incorporation of nitrogen in the near-surface region of silicon alleviates this problem by inhibiting SiO2 formation during the deposition of Ta2O5, thereby

PAGE 56

44 enabling the benefits of the high k oxide film to be more fully realized [7,8,63]. Another important application of sili con nitride is as a protectiv e coating on ceramic components such as bearings and turbine blades for wh ich tolerance to high temperature, oxidizing environments is critical. Despite these im portant applications, how ever, the oxidation of silicon nitride films is not we ll understood at the molecular leve l. In this article, we discuss results of an ultrahigh vacuum (UHV) investigation of the oxidation of a silicon nitride film by both gaseous atomic and molecular oxygen in which we focused our efforts on elucidating the fundamental origin for the oxidation resi stance of nitrided silicon. Several early studies have characterized th e oxidation of thick silicon nitride films under conditions of high oxidant pressure [64-67]. For example, Kuiper et al. investigated the oxidation of a thick silicon nitrid e film by exposure to O2 and H2O in an atmospheric furnace. They report that the ra te of oxidation of the nitride film is two orders of magnitude lower than the oxidation rate of Si(100), and th at the presence of hydrogen (H2 and H2O) was necessary to oxidize the surface under the conditions examined. They assert that hydrogen reacts with the nitride to form gaseous ammonia and elemental silicon and that the surface act ivity toward oxidation is enhanced as a result since elemental silicon is more easily oxidized than the nitride. Similar results have been reported in studies of the dry oxidation of silicon nitride films [64,65,67]. While these investigations have characterized the oxidation resistance of silicon nitride at high pressure, experiments conducted unde r more well-defined and controllable conditions are needed to determine the underlyi ng cause for the oxidation resistance of silicon nitride films. Experiments of this type have been reported recently by Wallace et

PAGE 57

45 al. [68]. In this work, the investigators thermally decomposed ammonia on Si(111) to generate silicon nitride films in UHV a nd then oxidized the nitrided surface with molecular oxygen without breaking vacuum. From in situ analysis of the surface using X-ray photoelectron spectroscopy (XPS), the authors observed negligible oxygen uptake at surface temperatures below 873 K, and onl y a small amount of uptake above 873 K. The slow oxygen uptake was suggested to arise from a decrease in the concentration of surface dangling bonds after nitridation, t hough this effect was not quantified. Prior investigations of the nitridation of Si(100) by NH3 provide important insights for understanding how nitridation alters the prop erties of the Si(100) surface. At room temperature, ammonia adsorbs dissociatively on Si(100)-(2x1) to produce an adsorbed hydrogen atom and an NH2 moiety [69-72]. Heating the ammonia-saturated surface to about 700 K then leads to the decomposition of adsorbed NH2 and the complete desorption of hydrogen. Early investigations of this system also showed that the nitrogen atoms occupy subsurface sites after the NH2 groups decompose [72,73]. For example, Dresser et al. [72] observed significant attenua tion of the N(KLL) AES peak after sample heating, but only observed small amounts of NH3 desorption (< 10%). From these observations, Dresser et al. c oncluded that nitrogen migrat es to the sub-surface region after the adsorbed NH2 species thermally decompose on Si(100). Subsequent studies have confirmed that nitrogen migrates to th e subsurface of Si(100) during nitridation at elevated temperature (> 600 K) [74-78] For example, using high resolution photoemission, Peden et al. [74] obtained comp elling evidence that silicon nitridation by NH3 occurs by a mechanism in which nitrogen atoms diffuse into the subsurface region and leave a thin layer of elemental silicon adjacent to the vacuum-solid interface that

PAGE 58

46 persists as the underlying n itride film thickens. Experi ments using low energy electron diffraction (LEED) also reveal that anneali ng the ammonia-covered surface results in a decrease in the intensity of the fractional order diffraction spots, signifying that NH2 decomposition and nitrogen penetration to the subsurface disrupts the long-range order of the surface [79,80]. Considering that nitrogen resides below the vacuum-solid interface, direct interactions between an oxidant molecu le and nitrogen may be expected to have only a minor influence on the oxidation behavior of nitrided Si(100). A change in the structure of the surface, as indicated by LEED experiments, may therefore be the predominant cause for the change in the reac tivity of the surface toward oxidation. Although few UHV investigations of the oxidation of silicon n itride films have been reported, the oxidation of single cr ystal silicon surfaces has been studied extensively. Of particular relevance to the present work are detailed UHV studies by Engstrom et al. [48] on the oxidation Si( 100) and Si(111) by both gaseous atomic and molecular oxygen. These and other results ma y also be found in a review of Si oxidation written by Engel [49]. Briefly, under UHV c onditions the dissocia tive adsorption of O2 on Si(100) results in an effective satura tion coverage of only about 1 ML of oxygen atoms when the surface is held at 300 K dur ing oxidation. The saturation coverage can be increased by oxidizing at elevated surf ace temperature, but the oxygen uptake is still rather limited. For example, the satura tion oxygen coverage increases to 2 ML when Si(100) is exposed to O2 at a surface temperature of 800 K. Not surprisingly, Engstrom et al. [48] found that gaseous oxygen atom s adsorb on Si(100) with much higher probability than does O2, and that oxygen coverages grea ter than 10 ML can be obtained by oxidizing Si(100) held at 300 K using an atomic oxygen beam. In contrast to the

PAGE 59

47 results obtained using O2, the uptake of oxygen atoms was fo und to be insensitive to the surface temperature for oxygen coverages up to about 5 ML, which is indicative of nonactivated adsorption and possibly direct inse rtion of gaseous O-atoms into surface Si-Si bonds. In the present study, we used X-ra y photoelectron spect roscopy (XPS) to investigate the nitridation of Si(100) by the thermal decomposition of ammonia as well as the oxidation of the resulting ni tride film by both gaseous atomic and molecular oxygen. The key objectives of this study were to dete rmine the surface prope rties responsible for the oxidation resistance of silicon nitride films, and to characterize the mechanisms for oxidation with these oxidants and the properties of oxidized nitride films. We find that surface dangling bonds play a critical role in the adsorption of both O-atoms and O2 and provide quantitative evidence that a decrease in the surface dangling bond concentration is the primary cause for the decrease in oxyge n uptake by Si(100) after nitridation. 2.2 Experimental Methods All experiments were conducted in an u ltrahigh vacuum chamber described in Section 1.3. Briefly, this apparatus was e quipped with a variable energy electron source, dual anode X-ray source capable of generating Al and Mg K radiation, ion source, and a hemispherical charged particle analyzer, gi ving the system the capability of performing Auger electron spectroscopy (A ES) and X-ray photoelect ron spectroscopy (XPS). Surface structural measurements were perf ormed using low energy electron diffraction (LEED) optics. A commercial microwave pl asma source was employed to generate an oxygen atom beam. The atom source was mount ed in a stainless steel reaction chamber with two stages of differential pumping. The communication between the plasma source

PAGE 60

48 and the vacuum chamber was through a quartz tube with a diameter of 6 mm and a length of 60 mm. The oxygen feeding the plasma source was supplied by BOC gases with a purity of 99.999% without any further purification. The Si(100) sample used in this stud y was taken from an arsenic-doped (0.005 ohm-cm) silicon wafer that was cut and polished to within + 0.5 of the (100) plane. An approximately 2 x 1 cm rectangular section wa s cut from the wafer, and tightly fastened to the Ta plates that contact the cooling reservoir of the manipulator. Only tantalum parts were used to fasten the Si sample to the hol der since the use of stainless steel parts was found to introduce small amounts of nickel in to the sample. To measure the sample temperature, a K type thermocouple was spot-w elded to a thin strip of Ta foil that was bent into a U-shape and clipped to the back of the Si(100) sample. The sample was cleaned by sputtering with 2 keV Ar+ ions followed by annealing for several minutes at 1000 K. The sample was considered to be cl ean when no contaminants could be detected with XPS, and a sharp (2x1) LEED pattern was observed. Ammonia was dosed onto the sample as a beam generated with the calibrated doser. Typical NH3 fluxes used in these experiments were ~5 x 1013 cm-2 sec-1, which is estimated from the known NH3 flow rate from the doser and the angular emission characteristics of micro-capillary arrays [57]. Exposures of NH3 are reported in units of ML, where 1 ML is defined as the surface atom density of 6.8 x 1014 cm-2 of the Si(100)(2x1) surface. Pure O2 beams were dosed onto the sample by flowing oxygen through the plasma source with the microwave power disabled. Th e size of the beam spot on the sample was about 9 mm in diameter. We typically employed an O2 beam flux of 0.26 ML/sec, which

PAGE 61

49 was determined using the spot size estimate and by comparing the O2 pressure rise in the UHV chamber due to the beam with that resulting from a known flow rate of O2 admitted through the calibrate d beam doser. Beams containing oxygen atoms were ge nerated by activating the microwave plasma with O2 flowing through the source. With an initial O2 flux of 0.26 ML/sec, a measurable change in the 16 amu intens ity was not observed by line-of-sight mass spectrometry when the plasma was activated To estimate the O-atom flux, the oxygen uptake was measured on clean Si(100)-(2x1) he ld at 300 K as a function of exposure to oxygen beams with and without the microwave power enabled. From a comparison with a previous investigation of the adsorption of gaseous O and O2 on Si(100) [48], we estimate that O-atom fluxes of ~0.001 ML /sec impinged on the sample surface for the beam conditions employed. Subsequent to th ese experiments, it was determined that the beam source was slightly misaligned with the collimating apertures which caused substantial O-atom recombination prior to the beam entering the UHV chamber. Although the O-atom fluxes that we em ployed were relatively low, significant enhancements in the rate of oxidation were observed when the surfaces under study were oxidized at a given flue nce by a plasma-activated beam versus a pure O2 beam All XPS spectra reported in this st udy were obtained using Al K X-rays (h = 1486.6 eV) with the analyzer operating in a retarding mode at a pass energy of 27 eV. The electron takeoff angle was varied by rotating the sample with respect to the analyzer axis. An angular resolution of 5o is estimated from the geometry of the analyzer, and electron takeoff angles are specified with respect to the surface normal. Unless stated otherwise, the spectra presented here were obtained by measuring photoelectrons emitted at an angle

PAGE 62

50 of 60o from the surface normal. Even at this glancing takeoff angle, the area of the sample from which photoelectrons were collect ed was smaller than the spot size of the oxygen beam, thus ensuring that the XPS measurements probe d only the regions of the surface that were dosed with gases. The X PS spectra presented here were each processed using 21-point Savitzky-Golay smoothing, followed by background subtraction using the Shirley method [81]. Oxygen coverages were de termined from the ratio of O1s to Si2p integrated intensities, and assuming that exposure of clean Si(100)-(2x1) at 300 K to O2 produces a saturation coverage of 1 ML [48,49]. For the low oxygen coverages investigated here, we found it unnecessary to account for O1s and Si2p signal attenuation due to inelastic photoelectron scattering si nce the oxygen atoms remain in the outer surface layers. Nitrogen coverages were comp uted by a similar procedure, but inelastic scattering corrections were necessary in this case, as described in Section 2.3. 2.3 Results 2.3.1 NH3 Decomposition on Si(100) An ultrathin nitride film was grown on the Si(100) substrate prior to each oxygen exposure by thermally decomposing 160 ML of ammonia on the surface at 900 K. Several layers of nitrogen in corporate into the solid duri ng this exposure since ammonia decomposition and hydrogen desorption are rapid at 900 K [71,78]. The Si2p and N1s spectra obtained from the Si surface after this treatment are shown in Figures 2-1 and 2-2. The Si2p spectrum exhibits a main component at 99.2 eV due to elemental Si, and a smaller feature centered at about 101.8 eV. The small Si2p feature appears at a binding energy that is less than that of Si3N4 [74,77,78,82-85], which sugge sts the presence of a sub-stoichiometric nitride. For films prepared under similar conditions, previous

PAGE 63

51 investigations suggest that silicon is present in a Si3 N configuration in which each N atom is bonded with three silicon atoms. The N1s spectrum obtaine d after nitridation exhibits a single peak centered at a binding en ergy of 397.4 eV, which is also consistent with previous reports [77,78,82,86]. Experiments were conducted to probe the interaction of NH3 with the Si(100)-(2x1) surface so that the properties of the nitride f ilm could be characteri zed in more detail. XPS spectra were first obtained after exposi ng clean Si(100) held at 300 K to a saturation dose of 160 ML of ammonia. This exposure pr oduces a nominal coverage of 0.5 ML of adsorbed NH2 groups, with the balance of the su rface sites occupied by hydrogen atoms, and these species do not undergo furthe r reaction at 300 K under UHV conditions [72,75]. The N1s spectrum obtained after th e 300 K exposure exhibits a single peak centered at a binding energy of 398.1 eV (Figure 2-3A), which is consistent with previous reports of the N1s binding energy of adsorbed NH2 on Si(100) [73]. The N1s spectrum shown in Figure 2-3B was then obtained af ter annealing the amino-covered surface for 5 minutes at 900 K, which results in the complete desorption of hydrogen from the surface. It is noted that the spectrum did not change from that shown in Figure 2-3B when the sample was annealed for longer times. After annealing, the N1s peak position is shifted to a binding energy of 397.5 eV, which is consistent with the formation of Si3 N species in the near-surface region. Annealing the am ino-saturated surface also causes the N1s to Si2p intensity ratio to decrease to about 60% of its initial value. The magnitude of this decrease is in excellent agreement with that observed by Dresser et al. [72] after heating amino-saturated Si(100) to temperatures gr eater than about 700 K and examining the surface with AES. Dresser et al. estimated that less than 10% of the nitrogen desorbs

PAGE 64

52 from the surface during heating, and therefore attributed the reducti on in the N KLL peak intensity to nitrogen penetrati on into the sub-surface [72]. From the intensity changes in the N1s and Si2p spectra, we estimated the distance at which the nitrogen atoms reside below the surface after annealing the amino layer. For this calculation, nitrogen desorption is neglec ted and it is assumed that all of the nitrogen atoms initially present in the amino layer resi de in a single layer below the surface after the sample is heated to 900 K. It is furt her assumed that the probability of generating an N1s photoelectron is the same for an adsorbed NH2 group as for the nitride. Under these assumptions, the attenuated N1s to Si2p inte nsity ratio of 60% can be approximated by exp(-x/cos), where x is the distance of th e nitrogen atoms beneath the surface, is the inelastic mean free path of an N1s photoelectron through elemental Si and is the photoelectron takeoff angle measured from the surface normal. Assuming an inelastic mean free path of 22.3 [87], the 40% decreas e in N1s peak intensity suggests that the nitrogen atoms diffuse 5.8 below the Si surface layer. Base d on the spacing between the Si layers closest to the (100)-(2x1) surf ace, this calculation suggests that nitrogen atoms reside between the fourth and fifth layers after the amino-covered surface is annealed at 900 K. Considering the simplicity of the analysis, our estimate is in good agreement with recent electroni c structure calculations which predict that a nitrogen atom has a energetic preference to be located betw een the third and fourth layers of Si(100)(2x1) and to bond with three Si at oms in the third layer [76]. After annealing the initial amino layer, the surface was held at 300 K and again exposed to 160 ML of ammonia. As may be seen in Figure 2-3C, this exposure causes the N1s peak to grow in intensity and to shift toward higher binding energy, which

PAGE 65

53 indicates the presence of adsorbed NH2 groups. Similar observations have been made by Avouris, Bozso and Hamers [73], and indicate that annealing the amino layer partially restores the reactivity of the surface towa rd ammonia dissociation. According to quantum chemical calculations [88,89], the dissociation of amm onia on Si(100)-(2x1) involves the adsorption of NH3 on a single atom of a surface dimer, followed by N-H bond cleavage and hydrogen transfer to the opposing silicon atom of the dimer. Considering this prediction, it may be c oncluded that nitrogen diffusion into the subsurface of Si(100) regenerates pairs of surface dangling bonds, and thereby partially reactivates the surface toward ammonia dissociat ion. Interestingly, however, the increase in the N1s to Si2p intensity ratio after saturating the annealed surface with amino groups is only 54% of that obtained after saturati ng clean Si(100) with amino groups (Figure 23A). This difference suggests that nitrogen incorporation into the sub-surface of Si(100) is accompanied by a structural rearrangement of the surface that reduces the density of dangling bond pairs by nearly a f actor of two from its value on the clean surface. Indeed, in a prior study, LEED images taken afte r ammonia decomposition on Si(100) show a diffuse background that eclipses the fracti onal order spots [79, 80], indicating that nitrogen incorporation does alter the stru cture of the surface. The total dangling bond coverage on the nitrided surface may be esti mated as 0.59 ML when all of the surface dangling bonds are assumed to exist in pair s, and taking into acc ount attenuation of the N1s signal from subsurface nitrogen due to inelastic electron scattering from the NH2 adsorbed at the surface. As stated above, the procedure we em ployed for growing a nitride film for subsequent oxidation studies was to expose the clean surface held at 900 K to 160 ML of

PAGE 66

54 ammonia. The intensity of the N1s peak obt ained after this proce dure (Figure 2-2) is about three times greater than that obtained from the amino-saturated Si(100) surface. To estimate the thickness of the nitride film, we assume that a layered structure is formed and that each layer contains 0.5 ML of nitrogen atoms. Furthermore, based on our analysis of the amino-saturated surface before and after annealing, it is assumed that the nitride layer closest to the vacuum-solid inte rface resides four to five layers below the surface; this assumption is supported by the oxidation results discussed below. With these assumptions, and invoking a simple mode l to account for signal attenuation due to inelastic electron scattering, we estimate that a nitride film of 5 to 6 atomic layers in thickness is generated by the 160 ML ammoni a exposure at 900 K. The findings from these experiments that have a key impact on the understanding of oxidation of the nitrided surface are 1) that the nitride films produced by ammonia decomposition on Si(100) reside in the sub-surface region and 2) that nitridation reduces the density of surface dangling bond sites. 2.3.2 Oxidation of Nitrided Si(100) at 300 K Atomic versus molecular oxygen. Oxidation of the nitrided surface by both molecular and atomic oxygen was investigated at a surface temperature of 300 K. After nitridation the surface was exposed to the oxygen beam for 60 minut es, and the surface was then analyzed with XPS. A 60 min beam exposure corresponds to ~930 ML of O2 for the fluxes employed. Exposing the nitrided surface to the pure O2 beam for 60 minutes results in an oxygen coverage of 0.27 ML, which was found to be the limiting coverage for oxidation of the nitrided surface at 300 K by O2. The oxygen coverage increased to 1.2 ML when the nitrided surf ace was exposed to the same beam fluence but

PAGE 67

55 with the plasma power enabled. This re sult shows that gas-phase oxygen atoms are significantly more reactive toward nitrided Si(100) than is O2, particularly considering that only about 3 ML of oxygen atoms are estimated to have impinged on the surface during the 60 minute exposure. The O1s feature obtained after oxidizing with the plasma-activated beam is very similar in sh ape and peak location to that obtained after oxidizing with O2 at 300 K (not shown). The O1s peak is shifted to a higher binding energy by 0.1 eV after oxidizing with the plasma-activated beam, compared with oxidation with pure O2, but this shift is consistent with the higher oxygen coverage that is achieved with the atomic oxygen beam. It is well known that as the coverage of oxygen is increased, the oxygen atoms on the Si( 100) surface experience changes in their bonding environment that alters the O1s binding energy [48,49]. The similarities in the O1s spectra indicate that ga seous oxygen atoms and molecules produce similar chemical states of oxygen on the nitrided surface, wh ich suggests that after adsorption (or O2 dissociation) the processes by which oxygen at oms incorporate into the nitrided surface are independent of the identity of the gaseous oxidant. It therefore follows that the enhanced uptake achieved with the plasma-activ ated beam is due to the higher adsorption probability of oxygen atoms compared with O2 on the nitrided surface. Clean versus nitrided Si(100). A comparison of the oxygen uptake on the clean and nitrided Si(100) surfaces reveals that nitridation significantly lowers the surface reactivity toward oxidation. For example, exposing the clean surface to 930 ML of O2 at 300 K results in an oxygen coverage of about 1 ML, whereas a coverage of only 0.27 ML is reached on the nitrided surface by oxidizi ng under the same conditions. The difference in the reactivity of these surfaces toward gaseous oxygen atoms is less pronounced, but is

PAGE 68

56 still quite significant. Specifically, an oxygen coverage of 1.2 ML is obtained by exposing the nitrided surface at 300 K to th e plasma-activated beam for 60 minutes, whereas a coverage of 2.1 ML is obtained on the clean surface for the same exposure and surface temperature. Since the nitride film is shown to reside several layers below the vacuum-solid interface, the lower reactivity of the nitrided surface compared with the clean surface does not arise from a direct interaction between nitrogen and oxygen ( vide infra ) but is attributed to th e lower concentration of da ngling bonds on the nitrided surface. Chemical state changes induced by oxidation at 300 K Shown in Figures 2-4 to 2-6 are the Si2p, N1s and O1s spectra obtaine d from the nitrided surface after depositing 1.2 ML of atomic oxygen at a surface temper ature of 300 K. Also shown are the Si2p and N1s spectra obtained from the nitrid ed surface before oxidation, and the O1s spectrum obtained after exposi ng clean Si(100)-(2x1) held at 300 K to a saturation dose of O2, which results in about 1 ML of atomic oxygen on the surface. Each spectrum has been normalized by its peak height to facilita te comparison. Among all these spectra, the most distinct change caused by oxidation is an increase in the intensity of a feature centered at about 102 eV in th e Si2p spectrum, see Figure 2-4. This spectral change indicates an increase in the amount of Si at oms present in partially oxidized states as oxygen atoms are incorporated within the nitr ided surface. As discussed below, this feature is assigned specifically to Si2+ and Si3+ species that are dire ctly bonded to oxygen atoms in the outermost surface layers. The N1s spectra obtained before and af ter oxidation at 300 K are shown in Figure 2-5. Oxidation causes small changes in th e N1s feature, but th e changes observed are

PAGE 69

57 reproducible and distinct. In particular, after oxidation the center of the N1s peak is shifted to lower binding energy by about 0.2 eV, and a small f eature appears at a binding energy of 399.2 eV. Previous studies have reported that Si2=N-O structures give rise to an N1s feature at a binding energy of about 399 eV. The appearance of the high binding energy shoulder is therefore consistent with a small quant ity of N-O bonds being formed upon oxidation at 300 K. The incorporation of both nitrogen and oxygen atoms in the near-surface region of Si(100) has been found in most cases to cause the N1s peak to shift to a higher binding energy (BE) relative to the N1s BE obtained from pure nitride surfaces [76,90,91]. A positive binding energy shift of the N1s peak may result from a core-hole screening effect that arises from the formation of a dielectric film near the vacuum-solid interface. However, such a sc reening effect should be negligible for the films we have investigated since no more th an 2 ML of oxygen atom s are present at the outer surface. Positive BE shif ts of the N1s feature have also been attributed to a second nearest neighbor (NN) interac tion in which oxygen atoms withdraw charge from Si atoms that are directly bonded to nitrogen in the f ilm [76,91]. The negative binding energy shift of the main N1s feature observed in the pres ent study suggests that at 300 K the majority of oxygen atoms do not penetrate far enough below the vacuum-solid interface to occupy second NN positions with respect to the nitrogen atoms. The small shift of the N1s peak to lower binding energy could be caused by th e introduction of strain at the nitride-Si interface when oxygen atoms are incorporated into the top surface layers. Since these spectra show that the majority of the N and O atoms in the film do not directly interact, the growth of the 102 eV f eature in the Si2p spectrum following oxidation may be

PAGE 70

58 attributed to the formation of Si2+ and Si3+ species that are directly bonded to oxygen atoms in the outermost surface layers. The O1s spectra obtained af ter depositing 1 and 1.2 ML of oxygen on the clean and nitrided surfaces, respectively, at 300 K are shown in Figure 2-6. The O1s peak obtained from the nitrided surface after oxidation is similar in shape to that obtained from the pure oxide layer, but is shifted to higher binding energy by about 0.3 eV. This difference in binding energies indicates that oxygen at oms at a concentration of about 1 ML experience slightly different chemical environments when adsorbed on clean versus nitrided Si(100) at 300 K. Based on the sm all N1s feature observed at 399 eV, a small fraction of the oxygen atoms appear to be di rectly bonded with nitr ogen atoms in the film. This bonding interaction could alte r the O1s binding energy, and produce a shift from the binding energy obtained from the pure oxide film. In addition, the oxygen atoms near the vacuum-solid interface, wh ich do not directly bond with nitrogen, may experience a different chemical environmen t than a similar quantity of oxygen atoms incorporated into the clean surface. Such an effect could arise if the structures in the near-surface layers of the solid differ for the clean and nitrided surfaces. This latter interpretation is consistent with the a mmonia uptake experiments which show that nitridation lowers the surface dangling bond density, probably by inducing a structural change at the surface. Angle-resolved XPS data. The chemical changes suggested by the XPS spectra provide the general picture that oxidation of the nitrided surf ace at 300 K results in nearly segregated oxide and nitride layers, with the oxide layer being closer to the vacuum-solid interface. To further examine this possibili ty, XPS spectra were obtained at different

PAGE 71

59 photoelectron takeoff angles to vary the dept h resolution of the measurements. Shown in Figure 2-7 are the O1s and N1s spectra collected at 0 and 60 takeoff angles with respect to the surface normal after adsorbing 1.2 ML of oxygen atoms on the nitrided surface at 300 K. In order to illustrate the variation in the O1s/N1s intensity ratio with average sampling depth, the spectra obtained at a give n takeoff angle are scaled by the integrated intensity of the N1s spectrum measured at th at angle. Qualitative differences in the spectra obtained at different collection angles are slight, suggesting that the chemical states of nitrogen and oxygen remain fairly uniform throughout the film. The slight broadening of the N1s feature toward high binding energy at glancing takeoff angle may arise from a small amount of N-O bonding at the nitride-oxide in terface closest to the surface. The most distinct difference between these spectra is cl early the increase of approximately 15% in the O1s/N1s intensity ratio for the measurements performed at a 60 takeoff angle, i.e. the more surface sensitiv e configuration. This result confirms that the oxygen atoms reside closer to the outer surface of the so lid than do the nitrogen atoms when as much as 1.2 ML of oxygen atoms are adsorbed on nitrided Si(100) at 300 K. 2.3.3 Oxidation of Nitrided Si(100) at 800 K Atomic versus molecular oxygen Oxidation of the nitrided surface by both atomic and molecular oxygen was also inves tigated at a surface temperature of 800 K to compare with the oxidation behavior observe d at 300 K, and to therefore assess the influence of surface temperature on oxidation. Increasing the surface temperature from 300 to 800 K significantly enhances the react ivity of the nitrided surface toward O2. In particular, exposing the nitrid ed surface to 930 ML of pure O2 produces an atomic oxygen coverage of 0.85 ML when the surface is held at 800 K, which appears to be a

PAGE 72

60 saturation coverage for these oxidation conditi ons. In contrast, only 0.27 ML of oxygen is deposited by exposing the nitrided surface to O2 at 300 K. The enhancement in reactivity with increasing su rface temperature is less pronou nced when oxidizing with oxygen atoms. For example, exposing the nitr ided surface to the plasma-activated beam for 60 minutes produces oxygen coverages of 1.2 and 1.5 ML when the surface is held at 300 and 800 K, respectively. This is only a 25% increase in oxygen uptake, which is much lower than the 215% increase that is brought about by increasing the surface temperature when O2 is used as the oxidant. Despite the small enhancement in oxygen uptake with surface temperature, a higher oxyge n coverage is still obtained by oxidizing with gaseous O-atoms at 800 K compared with O2. Similarities in the O1s spectra (not shown) indicate that similar chemical states of oxygen are generated on the nitrided surface when oxidation is conducted using either gaseous oxygen atoms or molecules at a su rface temperature of 800 K. Thus, after adsorption (or dissociation) the processes by which oxygen atoms incorporate into the solid again appear to be indepe ndent of the identity of the gaseous oxidant, at least for the low coverages considered. The O1s peak does shift to higher binding energy by about 0.2 eV as the oxygen coverage on the nitrided surface is increased from 0.85 to 1.5 ML. However, this is a small difference in binding energy, considering that the oxygen coverage is nearly doubled, so it may be c oncluded that the increas e in oxygen coverage from 0.85 to 1.5 ML causes only minor alterations to the chemical bonding environment of oxygen atoms incorporated into the nitrided surface at 800 K. Clean versus nitrided Si(100). Table 2-1 summarizes the oxygen coverages obtained by oxidizing the nitrided and clean surfaces at the conditions indicated.

PAGE 73

61 Examination of the coverages given in the table shows that the influence of surface temperature on oxygen uptake is similar for the nitrided and clean Si(100) surfaces. For example, a 930 ML O2 exposure to clean Si(100)-(2x1) generates oxygen coverages of 1.0 and 2.0 ML at surface temperatures of 300 and 800 K, respectively. These coverages are in good agreement with previous studies [48,49], and demonstrate that oxygen uptake by the clean surface is enhanced considerably by increasing the surface temperature when O2 is used as the oxidant, which is similar to the behavior found for th e nitrided surface. As also observed for the nitrided surface, increasing the surface temperature produces a smaller increase in oxygen uptake by the clean surface when oxidizing with gaseous oxygen atoms. Table 2-1 shows that oxygen co verages of 2.1 and 2.4 ML result after exposing the clean surface to the plasma-activated beam fo r 60 minutes with the surface temperature maintained at 300 and 800 K, resp ectively. In the work by Engstrom et al. [48], it was found that atomic oxygen adsorp tion on the clean Si(100) surface remains independent of the surface temperature up to oxygen coverages of 4 to 5 ML. Thus, the relative insensitivity to surface temperature that we observed when oxidizing with the plasma-activated beam is expected for the oxygen coverages examined (< 3 ML). We do observe a small enhancement in uptake from the plasma-activated beam with increasing surface temperature. However, since O2 is by far the majority beam component, this small enhancement in uptake most likely refl ects the influence of surface temperature on O2 incorporation. Chemical state changes induced by oxidation at 800 K. Figures 2-8 to 2-10 display the Si2p, N1s and O1s spectra obtain ed after oxidizing the nitrided surface with oxygen atoms at a surface temperature of 800 K, which produces an oxygen coverage of

PAGE 74

62 1.5 ML. Figures 2-8 and 2-9 also show th e Si2p and N1s spectra obtained from the nitrided surface before oxidation. Figure 210 contains an O1s spectrum after depositing 2.4 ML of oxygen onto the clean Si(100) surf ace. Each spectrum has been normalized to its respective peak height to augment the contrasting features. The most pronounced spectral change following oxidation at 800 K is an increase in the intensity of the high binding energy feature in the Si 2p spectrum (Fig. 2-9) that is centered at about 102 eV and extends to about 104 eV. The appear ance of this feature indicates that Si+n (n > 0) states are generated during oxidation of the n itrided surface. While similar results were obtained following oxidation at 300 K (Figure 2-5), the intensity of the high BE Si2p feature is clearly greater and the feature extends to higher BE when oxidation is conducted at 800 K versus 300 K. The formation of a higher concentration of Si+n species not only arises from the higher oxygen coverages that are obtained during high temperature oxidation, but also from temp erature dependent cha nges in the oxidation process. For example, for nearly the sa me oxygen coverage, we find that oxidation at 800 K versus 300 K results in a greater amount of Si+2, Si+3 and Si+4 states. This observation is consistent with the oxidation be havior of clean Si(100) -(2x1) [49]. At 300 K, oxidation occurs more uniformly across the surface, with the average Si oxidation state increasing in proportion to the oxygen c overage. Increasing the surface temperature enhances surface atom mobility and results in the formation of more highly oxidized clusters at relatively low oxygen coverage. The incorporation of oxygen into oxidized areas of the surface likely alle viates strain in the surface layers during oxidation. Shown in Figure 2-9 are the N1s spectra ob tained before and after oxidizing the nitrided surface at 800 K to reach an oxygen c overage of 1.5 ML. After oxidation, the

PAGE 75

63 N1s peak is slightly narrower and the peak maximum is shifted by only about 0.1 eV to lower binding energy. Since these spectral changes are slight, it may be concluded that the nitrogen bonding environment is altered negl igibly during oxidation at 800 K, at least when the oxygen coverage is increased up to 1.5 ML. Furthermore, the BE shift is in the opposite direction to that observed when O and N atoms occupy second NN sites [76,90,91], which suggests that the oxygen and nitrogen at oms in the film remain segregated, even though oxidation at the el evated temperature enhances surface atom mobility, as clearly evidenced by the formation Si+3 and Si+4 oxidation states at low oxygen coverage (Fig. 2-8). Interestingly, the N1s BE shift is smalle r than that observed after oxidation at 300 K and the small feature at 399.2 eV is not evident in the N1s spectrum. These observations suggest that segregation of the nitride and oxide phases occurs to a greater extent at elevated surface temperature, with immeasurable formation of Si2=N-O structures. Since the growth of highly oxidized SiOx clusters involves substantial surface restructuri ng, it is conceivable that oxidation at the elevated temperature enables the sub-surface nitride to adopt a more favorable structure in which nitrogen atoms experience a more uniform bonding environment. Such a change may explain the slight narrowing of the N1s peak observed after oxidation of the nitrided surface at 800 K. The O1s spectra obtained after oxidizing th e clean and nitrided Si(100) surfaces at 800 K to oxygen coverages of 2.4 and 1.5 ML are remarkably similar (Fig. 2-10). Indeed, the similarity between the O1s spectra indicates that the presence of a nitride in the sub-surface of Si(100) has a negligible in fluence on the chemical state(s) of oxygen that form during oxidation at 800 K. Thus, it appears that SiOx regions with similar

PAGE 76

64 properties grow on clean and nitr ided Si(100) at 800 K, despit e the structural differences of these surfaces. The similarity in the O1s spectra (Fig. 2-10) is also consistent with enhanced segregation of the oxide and n itride phases when oxidation is conducted at elevated surface temperature. Angle resolved XPS data. Analysis of the XPS spect ra obtained after oxidizing the nitrided surface at 800 K provides evidence that the oxidized and nitrided regions remain segregated. Angle-resolved XPS spectra provide additional support for this interpretation. Figure 2-11 shows the O1s a nd N1s spectra obtained at electron take-off angles of 0 and 60 with respect to th e surface normal, after adsorbing 1.5 ML of oxygen on the nitrided surface at 800 K. To compare the O1s/N1s intensity ratio as a function of sampling depth, the spectra obtai ned at a given angle have been normalized with respect to the N1s intensity at that a ngle. Only minor qualitative changes in the spectra obtained at different take-off angles are observed. For example, the O1s peak obtained at a 60 take off angle is shifted by only 0.1 eV to lower binding energy relative to the O1s peak obtained at an emission angle of 0. Differences between the N1s spectra obtained at these take-off angles are slight. Similar to the angle resolved data obtained after oxidizing at 300 K, the O1s/N1s ratio increases by about 10% when the collection angle is adjusted to the more surface sensi tive configuration. This observation confirms that oxygen resides closer to the vacuum-so lid interface than does the nitride region. 2.4 Discussion The present results show that nitridation of Si(100) at elevated temperature reduces the concentration of surface dangling bonds by nearly a factor of two and that the reactivity of the surface toward both at omic and molecular oxygen decreases significantly. Since the XPS resu lts also reveal that the majo rity of oxygen and nitrogen

PAGE 77

65 atoms do not directly interact within the film s studied, but remain in nearly segregated layers, the decrease in surface dangling bond concentration appears to be the primary cause for the diminished activity of the nitr ided surface. Physically, this conclusion implies that both gaseous oxygen atoms and molecules adsorb predominantly, if not exclusively, at dangling bond sites or pairs on the surface, and that the uptake of oxygen by the nitrided surface is limited by the availabi lity of such sites. That dangling bonds are the active sites for O2 adsorption is not at all surprisi ng. In fact, quantum chemical calculations predict that the lowest energy pathway for O2 activation on Si(100)-(2x1) involves the formation of a peroxy sp ecies across a surf ace dimer [89]. It is perhaps more surprising that ga seous oxygen atoms have such a strong tendency to adsorb at dangling bond sites since this im plies that insertion directly into SiSi bonds occurs to a negligible extent. While it is possible that an O atom incident from the gas-phase must overcome an activation barrier to directly insert into a Si-Si bond, we consider this possibility to be unlikely since formation of a Si-O-Si linkage is exothermic by at least 6 eV. A propensity for oxygen atom s to adsorb at dangling bonds, rather than to directly insert into Si-Si bonds, may be e xplained if we assume that the majority of oxygen atoms in the beam exist in the ground 3P electronic state and then consider electron spin effects. Because the 3P state is a triplet, direct O-atom insertion into a Si-Si bond is spin-forbidden whereas adsorption at a da ngling bond site is not. In this case, the rate at which O-atoms from the beam directly insert into Si-Si bonds would be limited by the rate of non-adiabatic curve crossing events that transform the el ectronic configuration of the incident oxygen atom to a state such as the singlet 1D state for which direct insertion is allowed. Such events are likely to be rare in a single gas-surface collision at

PAGE 78

66 thermal impact energy. Thus, the observati on of selective O-atom adsorption at surface dangling bond sites suggests that the initial adsorption event tends to be electronically adiabatic for the beam conditions employed. A quantitative comparison of the uptake of oxygen by the clean and nitrided Si(100) surfaces provides additional insight fo r understanding the role of dangling bonds in the oxidation of these surfaces. Th e bottom row of Table 2-1 shows the oxygen coverages obtained on the nitrided surface rela tive to that obtained on the clean surface for various oxidizing conditions. As may be seen in the table, the oxygen coverages obtained by exposing the nitrided surface to the atomic oxygen beam at surface temperatures of 300 and 800 K are 57 and 62.5% lower than that obtained on the clean surface. These values are remarkably close to the ratio of dangling bond concentrations on the nitrided and clean surfaces (59%), a nd provide quantitative evidence that gaseous O-atoms adsorb preferentially on surf ace dangling bonds on both surfaces. This comparison is even more favorable when considering that the contribution of O2 to the uptake achieved during the plasma-activated beam exposure is more significant at a surface temperature of 800 K. Inte restingly, for oxidation with O2, the maximum oxygen coverage obtained on the nitrided surface at 300 K is only 27% of that obtained on the clean surface (Table 2-1). This value is less than half of the ratio of dangling bond concentrations on the nitrided versus clean surfaces. Assuming that an O2 molecule does dissociate across a single dimer on the clean Si(100)-(2x1) surface, as predicted by electronic structure calc ulations [89], this comparison s uggests that at least two dimers are consumed when a single oxygen mol ecule dissociates and the oxygen atoms incorporate into the nitrided surface at 300 K. While it is difficult to envision four

PAGE 79

67 dangling bonds being requi red to activate one O2 molecule, it is conceivable that dangling bond pairs could be arranged on the nitr ided surface in such a way that O2 activation on one pair could render a nei ghboring pair unable to readily activate a second O2 molecule. The uptake of oxygen on the nitrided surface in creases to 44% of that on the clean surface when oxidation is conducted at 800 K using O2, which may indicate that fewer dangling bond pairs are consumed or are more efficiently regenerated on the nitrided compared with the clean surface during oxidation at elevated temperature. Overall, these observations suggest that oxidation with gaseous oxygen atoms occurs by a similar mechanism on the clean and nitr ided surfaces, with the main difference being that fewer adsorption sites are available on the nitrided surface. In contrast, the mechanism for O2 dissociative chemisorption and oxygen incorpor ation appears to be more sensitive to structural differences between the nitrided and clean surfaces. Increasing the surface temperatur e enhances the uptake of O2 on both the clean and nitrided surfaces, but produces only a small in crease in the uptake of gaseous O-atoms. High surface temperatures are thought to f acilitate the oxidation of clean Si(100) by O2 by promoting oxygen penetration into the subsu rface layers [49]. Su ch penetration is likely to regenerate dangling bond sites at the surface that are needed to activate O2 molecules, thereby restoring the surface activity toward O2 dissociation. It is noted that a molecular beam study by Ferguson et al. [92] shows that the dissoci ation probability of O2 on Si(100) is only weakly dependent on the surface temperature at low gastemperatures. Thus, more facile regenerati on of active surface site s is the more likely explanation for the enhancement in oxygen upt ake with surface temperature than would be promotion of O2 bond cleavage at higher surface temperature.

PAGE 80

68 The relative insensitivity to surface temperature in the uptake of gaseous O-atoms was first observed by Engstrom et al. [48] a nd was quite reasonably interpreted by those authors as evidence that oxygen at oms incident from the gas-pha se insert readily into SiSi bonds. The uptake of gaseous O-atoms was found to increase with surface temperature only at oxygen coverages greater than about 4-5 ML, which corresponds to oxygen atoms inserted into all of the Si-Si bonds that are directly accessible from the gasphase. However, the findings from the curren t study indicate that surface dangling bonds are the preferred adsorption site for a gaseous O-atom, and that direct insertion into a SiSi bond occurs to a negligible extent. C onsidering this finding, it is difficult to understand why an increase in the surface temp erature effects only a small enhancement in the uptake of gaseous oxygen atoms. In pa rticular, if more eff ective regeneration of surface dangling bond sites is the primary reason that an increase in surface temperature enhances O2 uptake, then it is reasonable to e xpect that the uptake of gaseous oxygen atoms would also be promoted by raising the surface temperature since O-atoms also adsorb selectively at dangling bond sites and mo re of these sites should be available at high surface temperature. A recent computational inve stigation by Widjaja and Mu sgrave [89] may offer a plausible explanation for understanding the diffe rent effects of surf ace temperature in the oxidation of Si(100) with gaseous O atoms versus O2. The top panel of Figure 2-12 shows a schematic of key structures and the associated energy changes that were predicted to occur by those authors when O2 adsorbs and then dissociates on the Si(100)(2x1) surface [89]. It is im portant to note that the molecu lar representations shown in Figures 2-12 and 2-13 are only intended to de pict the steps in the proposed model, and do

PAGE 81

69 not precisely illustrate the bond lengths and angl es for these structures as predicted by density functional theory. Following adsorption, the O2 molecule is predicted to span the dimer to form a peroxy-like species that then dissociates to produce an oxygen atom inserted across the dimer and a siloxy radical. The oxygen atom of the siloxy radical then inserts into a Si-Si backbond, resulting in the final structur e shown in the top panel. Clearly, the formation of a second peroxy species on the final structure would be significantly hindered by the pres ence of the O-atom bridging the dimer. Thus, if the oxygen atoms in this final structure have limited mobility, then effectively only one O2 molecule can dissociate for each dangling bond pair on the surface. Notice that this situation would result in an oxygen coverage of 1.0 ML on the Si(100)-(2x1) surface, and may help to explain the subs tantial reduction in oxygen uptak e that occurs at 1.0 ML when the clean surface is exposed to O2 at 300 K. The bottom panel of Figure 2-12 illustrates elementary steps by which the bridging oxygen atom could migrate to a backbond site. These reactions have not been explored computationally as far as we know. The first step in the scheme shows the formation of a siloxy radical by cleavage of an Si-O bond of the bridging oxygen species, and the second step involves oxygen insertion into a Si-Si backbond. This migration process regenerates an empty dimer, and would thereby enable a second O2 molecule to bind in the peroxy configuration. Although energy barriers for these steps have not been explicitly predicted, the results of Widjaj a and Musgrave suggest that the first step, production of the siloxy radical, should have th e larger energy barrier This barrier may be comparable to the 1.38 eV barrier require d for the reverse of the final reaction shown in the top panel of Figure 2-12. Considering the large energy barrier, the migration of the

PAGE 82

70 bridging oxygen atom to a backbond site shou ld be promoted significantly by raising the surface temperature. Thus, according to this mechanism, oxygen uptake by O2 dissociation is enhanced at elevated surface temperature since the population of empty dimers increases with increasing surface temperature. Shown in Figure 2-13 are pathways proposed for the incorporation of a gaseous oxygen atom into the Si(100) surface. Based on the present results, the O-atom is assumed to adsorb initially on a dangling bond s ite to form a siloxy radical. From this site, the oxygen atom can insert either acro ss the surface dimer or into a Si-Si backbond to form the structures shown in the figure. The energy changes illustrated in this figure were also taken from the work of Widj aja and Musgrave, and show only slight differences in the energetics of these inserti on pathways. Since an O-atom adsorbs at a single dangling bond site, and therefore does not have the strict steric requirements for adsorption as does O2, we speculate that a second O-atom will adsorb with roughly equal probability on each of the one O-atom structur es shown in Figure 2-13. Thus, according to this interpretation, an in crease in surface temperature has only a minor influence on the uptake of gaseous oxygen atoms because enha nced oxygen migration to sub-surface sites does not significantly affect the availability of single dangling bond sites at the surface. 2.5 Conclusions We have investigated the n itridation of Si(100) and the subsequent oxidation of this surface by both gaseous atomic and molecular oxygen under UHV conditions. Nitridation of Si(100) by the thermal decom position of ammonia at 900 K results in the formation of a subsurface nitride and a decr ease in the concentration of surface dangling bond sites. Based on changes in N1s spectra after NH3 adsorption and decomposition, we estimate that the nitride resides four to five layers below the vacuum-solid interface and

PAGE 83

71 that the concentration of da ngling bonds on the nitrided surface is about 0.59 ML or 59% of that on the clean surface. Oxidation of the nitrided surfac e at surface temperatures of 300 and 800 K produces an oxide phase that resides in the outer surface layers and remains largely segregated from the subsur face nitride for oxygen coverages up to about 2.5 ML. At a surface temperature of 300 K, the incorporation of about 1 ML of oxygen into the near surface layers alters the nitrogen bonding environment only slightly, most likely by introducing strain in the subsurface ni tride, and the N1s spect ra indicate that a small quantity of Si2=N-O also forms. At 800 K, the nitride bonding environment changes negligibly for oxygen coverages as hi gh as 2.5 ML, which is consistent with greater segregation of the nitride and oxide phases and enhanced stru ctural relaxation in these phases. In addition, at a gi ven oxygen coverage, th e quantity of Si3+ and Si4+ states that are detected increases when oxidation is conducted with the surface held at 800 K versus 300 K, indicating a tendency for regi ons of high local oxyge n concentration to form at elevated temperature. The reactivity of Si(100) toward both atomic and molecular oxygen decreases significantly after nitridation of the subsur face region due to the decrease in surface dangling bond concentration that accompanies nitride growth. Quantitative support for this conclusion is given by the observation th at, for the same exposure to gaseous oxygen atoms, the oxygen coverage obtained on the ni trided surface relativ e to that on clean Si(100) is within 5% of the ratio of dangli ng bond concentrations on these surfaces. This finding also provides strong evidence that gaseous O(3P) atoms adsorb initially at dangling bond sites on these surfaces, and that direct insertion into Si-Si bonds occurs to a negligible extent. An increase in surface temperature is found to significantly enhance

PAGE 84

72 oxygen uptake by the nitrided surface when O2 is used as the oxida nt, but brings about only a slight increase in uptake when gaseous oxygen atoms are employed. It is proposed that an increase in surface temperature promotes oxygen migr ation to the subsurface, and thereby results in more effective regeneration of empty dimers. Since the activation of an O2 molecule on the Si(100) surface has more stringent steric requirements than does Oatom adsorption, the facile penetration of oxygen to the subsurface at high temperature has a greater influence on the adsorption of O2 than O.

PAGE 85

73 Table 2-1. Oxygen coverages on clean and nitrided Si(100) Atomic Oxygen Molecular Oxygen 300 K 800 K 300 K 800 K [O] (nitrided)a 1.2 1.5 0.27 0.85 [O] (clean) a 2.1 2.4 1.0 2.0 nitrided/clean 0.57 0.62 0.27 0.44 a The oxygen coverages in these rows are given in units of ML, as defined in the text, and were obtained after exposing the surface to the oxidant beam for 60 minutes at the surface temperatures indicated. A 60 min exposure corresponds to an O2 fluence of 930 ML when the plasma power is disabled, and to ~928 ML O2 and ~3 ML O when the plasma is activated. Figure 2-1. Si2p spectrum obtained fr om a Si(100) surface after a 160 ML NH3 exposure at a surface temperature of 900 K.

PAGE 86

74 Figure 2-2. N1s spectrum obtained fr om a Si(100) surface after a 160 ML NH3 exposure at a surface temperature of 900 K.

PAGE 87

75 Figure 2-3. N1s spectra obtained from Si(100). A) After a 160 ML NH3 exposure at 300 K. B) A subsequent anneal to 900 K for 5 minutes. C) A 160 ML exposure at a surface temperature of 300 K to the surface generated in B. The relative integrated areas of each N 1s spect rum is shown in each panel.

PAGE 88

76 Figure 2-4. Si2p spectra obtained afte r exposing clean Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed lin e), and after depositing 1.2 ML of oxygen on the nitrided Si(100) surface held at 300 K using the plasmaactivated beam (solid line).

PAGE 89

77 Figure 2-5. N1s spectra obtained after exposing Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed line), a nd after depositing 1.2 ML of oxygen on the nitrided Si(100) surface held at 300 K using the plasma-activated beam (solid line).

PAGE 90

78 Figure 2-6. O1s spectra obtained after inco rporating 1 ML of oxyge n atoms on Si(100) at a surface temperature of 300 K (dashed lin e), and after depositing 1.2 ML of oxygen on the nitrided Si(100) surface held at 300 K using the plasmaactivated beam (solid line).

PAGE 91

79 Figure 2-7. N1s and O1s spectra obtained at el ectron collection angles of 0 (dashed line) and 60 (solid line) after depositing 1.2 ML of oxyge n on nitrided Si(100) at a surface temperature of 300 K. The O1s and N1s peak heights have been normalized to the N1s peak height at the respective angle

PAGE 92

80 Figure 2-8. Si2p spectra obtained afte r exposing clean Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed lin e), and after depositing 1.5 ML of oxygen on the nitrided Si(100) surface held at 800 K using the plasmaactivated beam (solid line).

PAGE 93

81 Figure 2-9. N1s spectra obtained after exposing Si(100) to 160 ML NH3 at a surface temperature of 900 K (dashed line), a nd after depositing 1.5 ML of oxygen on the nitrided Si(100) surface held at 800 K using the plasma-activated beam (solid line).

PAGE 94

82 Figure 2-10. O1s spectra obtained after incorporating 2.4 ML of oxygen atoms on Si(100) at a surface temperature of 800 K (dashed line), and after depositing 1.5 ML of oxygen on the n itrided Si(100) surface held at 800 K using the plasma-activated beam (solid line).

PAGE 95

83 Figure 2-11. N1s and O1s spectra obtained at electron collection angles of 0 (dashed line) and 60 (solid lin e) after depositing 1.5 ML of oxygen on nitrided Si(100) at a surface temperature of 800 K. The O1s and N1s peak heights have been normalized to the N1s peak height at the respective angle.

PAGE 96

84 Adsorption and Insertion Incorporation and Site RegenerationHigh T + O2-2.84 eVE = -2.81 eV-0.83 eVE* = 1.5 eVE* = 0.55 eV Figure 2-12. Model for O2 dissociation and incorporation into Si(100). The top panel shows the structures and energetics fo r the dissociative chemisorption of O2 on Si(100) as predicted by DFT calculations [89]. The bottom panel shows possible elementary steps for oxygen migration to the subsurface that results in the regeneration of an empty dimer.

PAGE 97

85 -4.49 eV -2.18 eV -1.91 eV High T + O + OE = E = 0 3 3 e VE = 0 2 8 e V Figure 2-13. Model for O-atom adsorption a nd incorporation into Si(100). The energy changes, where indicated, were predic ted using DFT calculations as reported in reference 89.

PAGE 98

86 CHAPTER 3 DYNAMIC DISPLACEMENT AN D DISSOCIATION OF O2 ON Pt(111) BY ATOMIC OXYGEN The role of surface temperature on the dynamic displacement of 18O2 from Pt(111) stimulated by the adsorption of 16O atoms was investigated. The maximum displacement rate increased with surface temperature, while the desorp tion yield decreased with temperature. Heteronucl ear product evolution (18O16O) from the surface was below the experimental detection limits (< 10% of the 18O2 product evolution). The adsorption of 16O -atoms also induced the dissociation of pre-adsorbed 18O2 molecules at low surface temperatures. The initial displacement rates with a 16O-atom flux of 0.005 ML/sec were 0.0024, 0.0025 and 0.0027 ML/sec at 90, 100 and 110 K, respectively. Dissociation of 18O2 was found to only occur during the first 0.2 ML fluence of 16O-atoms at which the total atomic oxygen coverage was about 0.44 ML Estimates for the initial dissociation rate for 18O2 at 90, 100 and 110 K and with the same 16O-atom flux was estimated of 0.0015, 0.0020 and 0.0030 ML/sec, respectively. 3.1 Introduction Molecular oxygen displacement from Pt( 111) by incident oxygen, nitrogen, and hydrogen atoms at 80 K was first reported by Rettner and Lee [17]. By adsorbing 18O2 onto Pt(111) and subsequently exposing this surface to a beam of 16O-atoms at 80 K, they observed 18O2 displacement from the surface. When exposing this surface to 16O-atoms, they also observed the formation of 16O18O products. The displacement phenomenon was found to be independent of the adsorbing spec ies, indicating that the desorption is not

PAGE 99

87 collisionally induced. Using time-of-flight mass spectrometry, Rettner and Lee observed that 18O2 desorbed with a bimodal energy distribu tion. This indicates that desorption occurs through two distinct channels. The lower energy co mponent is consistent with thermal desorption, whereas the high energy feature indicates that a fraction of the adsorption energy of the incoming 16O-atom is transferred into the 18O2 stimulating desorption. Similar experiments were performed by Wh eeler et al. [16]. Using a supersonic 16O-atom beam directed towards a Pt(111) surface saturated with 18O2 at 77 K, they investigated the role of translational kine tic energy and incident angle on the initial molecular oxygen displacement rate. They observed both 18O2 and 18O16O evolving from the surface, and report that the initial probability of forming the mixed isotope species is ~16.5 % of the total O2 (18O2 and 18O16O) displacement probability These investigations also found that the displacement probability of 18O2 increased with the initial translational energy and decreased at glanci ng incident angles of the 16O-atom beam. The probability of forming the mixed isotope product appeared to weakly depend on the 16O-atom energy and incidence angle. From Rettner and Lee’s work it seems appa rent that both thermal and nonthermal mechanisms govern the displacement of 18O2 during the 16O-atom adsorption on Pt(111). During the 16O-atom exposure, the surface concentration of 16O-atoms is increasing. It is known that the presence of adso rbed oxygen atoms weakens the O2-Pt bond [12,16], which could be responsible for the observed thermal desorption of 18O2. Dynamic displacement is not fully understood, so it is reasonable that varying the surface temperature would provide additional info rmation into the dynamic displacement

PAGE 100

88 process. In this study, re al-time reaction product monitori ng and subsequent temperature programmed desorption (TPD) measurements were employed to examine the role of surface temperature on 18O2 dynamic displacement stimulated by gaseous 16O-atoms. 3.2 Experimental Methods These experiments were conducted in a three-level UHV chamber described in Section 1.3, with a brief desc ription provided here. This ch amber is equipped to perform X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES), low energy ion scattering spectros copy (LEISS), reaction product m onitoring, and TPD. The first plane contains a dual Al and Mg a node X-ray source, a va riable energy electron source, a variable energy ion source, and a he mispherical charged particle analyzer. The second plane includes a calib rated molecular beam doser and a leak valve controlled molecular beam doser. The final plane contains a quadrupole mass spectrometer, a microwave plasma source mounted in a beam chamber with two stages of differential pumping, and a LEED optics. A microwave plasma source was employed to decompose molecular oxygen using 2.45 GHz microwave ra diation. The end of the plasma source cavity is capped by an alumina faceplate with 5 laser drilled 0.4 mm through holes, which collimated the atom beam. A pair of oppositely charged plates (+ 10 kV/cm) is located at either side of the beam to remove charged pa rticles from the beam path. The beam flows from the first stage to the second through a 3 mm skimmer, spaced 15 mm from the alumina collimating plate. This second st age is pumped via a 66 l/s turbo molecular pump and a liquid nitrogen cooled titanium sublimation pump. Co mmunication from the second stage to the UHV chamber occurs thr ough a quartz tube with a diameter and length of 6 mm and 60 mm, respectively.

PAGE 101

89 The sample was mounted to a custom-built, liquid nitrogen cooled manipulator fabricated by McAllister Technical services. It is capable of 360 rotation about the vertical axis, and movement in all three Cart esian coordinates. The sample is cooled through thermal contact with a liquid nitrogen reservoir. Sample heating was conducted resistively using a DC power supply and th e temperature was re gulated using a PID controller. The oxygen atom flux from the beam was estimated through measuring the 16Oatom uptake as a function of atom beam e xposure. Assuming a sticking probability of unity, the flux was estimated at 0.005 ML/sec. The molecular oxygen flow rate into the UHV chamber was determined by monitoring the 16O2 partial pressure rise with the mass spectrometer. This partial pressure was then compared to that of a rise with a known flow rate generated by the cal ibrated molecular beam doser and was measured at 0.1 ML/sec. 3.3 Results Before each experiment, the sample was exposed to 20 Langmuirs (L) of 18O2 which was sufficient to reach the saturation coverage (0.44 ML of 18O2) at 100 K. The sample was then heated or cooled to th e desired temperature. Immediately after saturating the sample with 18O2, the 16O2 plasma was ignited with the beam shutter in the closed position. Since the shutter is located in the first stage of differential pumping, no 16O-atoms flowed into the chamber containing the sample. The mass spectrometer was then activated, with the instrument sensitivity set to maximize the 34 amu (16O18O) and 36 amu (18O2) signal intensities. At this sensitivity, the count rate for 16O2 was above the toleration limits for the electron multiplier, so the 32 amu signal was not monitored

PAGE 102

90 during the beam exposures. Upon opening the shutter, the sample was exposed to a mixture of 0.1 ML/sec of 16O2 and 0.005 ML/sec of 16O-atoms. Under this flux condition, it is assumed that the total O2 coverage (16O2 and 18O2) remains saturated, although the saturation O2 coverage decreases du ring the course of the 16O-atom exposure. Shown in Figure 3-1 is the desorption rate of 18O2 measured as a function of time as an 16O-atom beam impinges on the Pt(111) surface initially pr e-saturated with 18O2 at surface temperatures Ts of 90, 100 and 110 K. In each of the traces, the 18O2 desorption rate increases abruptly when the beam exposur e is initiated. After a short time delay of about 30 seconds, the desorption rate reaches a maximum, and then decays to the initial baseline, suggesting that all of the 18O2 desorbed from the surface within 250 seconds or equivalently after only about 1.25 ML of 16O-atoms collided with the surface. In similar experiments, Wheeler et al [16] also observed the 18O2 desorption rate to pass through a maximum as an 16O-atom beam was directed toward an 18O2-presaturated Pt(111) surface, and suggested that a decrease in the O2-Pt binding strength as the 16O -atom coverage increases may cause the 18O2 desorption rate to incr ease producing the rate maximum. In the present study, the time de lay before the desorption rate maximum is also found to be independent of the surface temperature, but the value of the maximum desorption rate increases with temperature and the desorption rate decreases to the baseline after the maximum more quickly at higher surface temperatures. The partial pressure of 18O16O was also found to increase when the 16O-atom impinged on the 18O2-covered surface, but the 18O16O molecules do not appear to evolve from the Pt(111) surface. The 18O16O partial pressure increased to about twice the level

PAGE 103

91 as the initial 18O2 desorption rate, but remained appr oximately constant during the beam exposure, even after the 18O2 desorption rate had returned to its baseline value, and the 18O16O partial pressure dropped back to its initial level only after the flow of 16O-atoms into the chamber was term inated. The quantity of 18O16O in the beam did not change when the plasma was deactivated, indicating that the 34 amu signal is due to the natural abundance of 18O16O in the beam. Wheeler et al. [16] also observed an increase in the 18O16O partial pressure during their beam expe riments. Recall that they observed an initial 18O16O formation rate of 16.5 % [16]. In the present study, an increase in the 18O16O partial pressure arising from such an exch ange reaction is difficult to detect due to the relatively high 34 amu background in tensity that was observed. The 18O2 gas used in the present study consists of about 0.4% as measured by the mass spectrometer. With the high 34 amu background in these experiments, the 18O16O products were less than 10% of the 18O2 displacement rate. Another possibility is that the exchange reaction is more probable for the beam conditions employed by Wheeler et al. [16]. These author s generated supersonic beams by expanding mixtures of He/O2 and Ar/O2 from RF plasma operating pressures near 1 bar, and produced O-atoms with kinetic energies in the range of 0.20 to 0.47 eV. In the current study, gaseous O-atoms with kineti c energies of ~0.05 eV were generated by flowing pure O2 through a microwave plasma operating at about 10-2 Torr from which the flow is effusive. The differences in these plasma conditions could certainly affect the composition of the plasma, and hence the surfac e reactivity that is observed. Finally, 18O2 and 18O16O were not observed to desorb at low Ts when the 18O2 covered surface

PAGE 104

92 was exposed to the beam with the plasma -power disabled, which shows that the 18O2 desorption is stimulated by the impingent 16O-atoms. Integration of the 18O2 desorption traces reveals that the total amount of 18O2 that desorbs during the 16O-atom beam exposure decreases when the experiments are conducted at increasing surface temperatures. Specifically, the desorption yields at Ts = 100 and 110 K are 9% and 20% lower, respectiv ely, than the desorption yield measured at 90 K. This result is surprising since the lack of measurable 18O16O evolution from the surface was initially believed to indicate that desorption is the only process by which the coverage of adsorbed 18O2 molecules is depleted during th e experiments. Moreover, the decay of the 18O2 desorption rate back to its initial baseline value also suggests that all of the 18O2 molecules desorb. Clearly, however, a significant fraction of the 18O2 molecules initially adsorbed on the surface do not desorb as 16O-atoms impinge on the surface. Since 18O was only observed to desorb as 18O2 during the 16O-atom beam exposure, the decrease in the 18O2 desorption yield with increas ing surface temperature suggests that 18O remains on the surface at the end of the beam experiments. To examine this possibility, a temperature programmed desorp tion measurement was performed at the completion of each 16O-atom beam exposure to monitor for the evolution of 16O2, 18O16O and 18O2 from the surface. Indeed, as shown in Figure 3-2, each of these isotopomers was observed to desorb during TPD, confirming that 18O remains on the surface after the dynamic displacement of 18O2 by gaseous 16O-atoms is completed at low Ts. Notice that the TPD trace for each isotopomer is similar in shape, and that desorption occurs at high rates over a narrow range of temperature cent ered at about 620 K. The recombinative desorption of oxygen atoms gives rise to O2 desorption at these hi gh temperatures, while

PAGE 105

93 desorption of molecularly adsorbed O2 occurs at temperatures near 150 K. Thus, the TPD results indicate that the vast majority of 16O and 18O species are present in atomic form on the surface after the dynamic displ acement experiments are complete, though a small quantity of molecularly adsorbed 16O2 is also present on the surface, as indicated by the small desorption feature observed at low te mperature in the TPD spectra. Integration of these TPD spectra rev eals that ~1.4 ML of 16O atoms adsorb on the surface during a 6 minute (1.80 ML 16O-atom fluence) beam exposure. To determine if 18O could exist on the surface in mo lecular form after the beam exposure, and then dissociated during TPD, we adsorbed about 1.4 ML of 16O-atoms on the initially clean Pt(111) su rface and then attempted to saturate the surface with 18O2 at 90 K. Subsequent TPD analysis reve aled that only small quantities of 18O2 (<0.03 ML) could be adsorbed on the surface at this high 16O-atom coverage, and that the 18O2 that adsorbs does not dissociate during TPD. Hen ce, it may be concluded that only atomic 18O remains on the surface after the 16O-atom beam exposure, and therefore that 18O2 dissociates as 16O-atoms impinge upon the surface at Ts < 110 K. The similar shapes of the desorption traces indicate that the 16O and 18O species populate the same adsorption states, which is also consiste nt with each isotope being adsorbed in atomic form after 18O2 dissociation is completed at low surface te mperature. We recently investigated the oxidation of Pt(111) by gaseous O-atoms and fou nd that Pt oxide forms on this surface at atomic oxygen coverages greater than about 0. 75 ML. Thus, the la ck of appreciable O2 adsorption on Pt(111) covered by about 1.4 ML of O-atoms indicates that O2 binds weakly on Pt oxide, which is consistent with recent findings for CO adsorption on this surface as discussed in Chapter 4.

PAGE 106

94 A series of 18O16O TPD spectra obtained after completing the dynamic displacement experiments at surface temperat ures of 90, 100, and 110 K are shown in the right panel of Figure 3-2. These spectra show an increase in the 18O16O desorption yield measured after 16O-atom beam exposures performed at increasing surface temperatures from 90 to 110 K. In fact, the amount of 18O atoms residing on the surface, as determined from the post-exposure 18O16O and 18O2 TPD spectra, is within 5% of the decrease in the 18O2 desorption yield, determined by integrating the direct 18O2 displacement traces (Fig. 3-1). The agreem ent between these analyses supports the interpretation that the concentration of adsorbed 18O2 is reduced to zero during the 16Oatom beam exposure at low Ts by only two reaction channels, namely, direct 18O2 desorption and dissociation for which the resulting 18O atoms remain on the surface. The absolute 18O2 dissociation yield as a function of the temperature of the surface during 16O-atom bombardment can also be determined from the integrated 18O16O and 18O2 TPD spectra obtained after the beam experiments. To determine the 18O-atom coverages from this TPD data, the area s under the TPD traces were scaled by the 18O2 TPD area obtained after thermally dissociating 18O2 on Pt(111) until saturation, which is well known to result in an 18O-atom coverage of 0.25 ML. The displacement yield can then be determined by applyi ng the mass balance shown in Equation 3-1, in which it is assumed that the pre-adsorbed 18O2 is depleted from the surface only through direct displacement or dissociation; th is assumption is in good agreement with the data analysis discussed above. As shown in Table 3-1, the amount of 18O2 that dissociates as 16Oatoms impinge on the surface is quite sensitiv e to the surface temperature, increasing from 0.06 to 0.11 ML with a relatively sma ll increase in the surface temperature from 90

PAGE 107

95 to 110 K. Also, the dissociation yields at 90 and 110 K represent 13.6 and 25 % of the initial 18O2 coverage, indicating that a significant fraction of the adsorbed molecules dissociate during the atomic oxygen beam exposure. Recall that the formation rate of the mixed isotope product was below the detecti on limits (~10%). The dissociation yields shown are greater than this quantity, indicating that the dissociation rates are greater than the exchange product formation rate under the conditions examined. dis gO Y ML O2 18 ) ( 2 1844 0 (3-1) The analysis discussed above predicts the total amount of 18O2 molecules that dissociate during the 1.8 ML 16O-atom beam exposure to the surface, but it does not provide information about the variat ion of the dissociation yield with 16O-atom fluence. To determine how the dissoci ation yield changes with 16O-atom fluence, or more specifically, with the total atomic oxygen coverage, the 18O2 saturated Pt(111) surface was exposed to 16O-atom beam for durations of 20, 40, 60, 80 and 200 seconds, which corresponds to 16O-atom fluences of 0.1, 0.2, 0.3, 0.4 and 1 ML, and TPD was performed after each exposure. The TPD spectra were th en integrated to determine the amount of 18O2 molecules that dissociate at different surface temperatures as a function of the 16Oatom exposure. Recall that chemisorbed O2 thermally dissociates during TPD if the total atomic oxygen coverage is less than 0.25 ML. Therefore, this analysis can provide a measure of the coverage of 18O-atoms present on the surface on ly when the total O-atom coverage (16O + 18O) is greater than 0.25 ML, otherwis e it is not possibl e to distinguish 18O present in atomic form at Ts < 110 K from 18O atoms generated by the thermal dissociation of 18O2 during the TPD measurement. From this analysis, 18O2 dissociation was found to occur only during the first 40 seconds (~0.2 ML -atom fluence) of the

PAGE 108

96 beam exposure, producing 18O atom coverages of 0.12 and 0.22 ML at surface temperatures of 90 and 110 K, respectively. At Ts = 110 K, the 18O-atom coverage nearly doubles, increasing from 0.14 to 0.22 ML, as the beam exposure continues from 20 to 40 seconds. Analysis of the TPD spectra also reveals that the total coverage of O-atoms increases from 0.26 to 0.44 ML during the 20 to 40 second time interval over which 18O2 molecules continue to dissociate efficiently. This is an interesting result since the thermal dissociation of O2 on Pt(111) under UHV conditions ceas es once the total atomic oxygen coverage reaches 0.25 ML. Thus, the observation of 18O2 dissociation above this limiting O-atom coverage, and at low surface temper ature, supports the in terpretation that the dissociation is stimulated by inte ractions between the adsorbed O2 molecules and the Oatoms incident from the gas-phase. Since the O2-Pt(111) binding energy decreases as the atomic oxygen coverage increases during the beam exposure, it is impor tant to estimate the contribution of thermal desorption to the total rate of 18O2 desorption observed during the 16O-atom beam exposures. Shutter interruption experiment s were performed to determine how the thermal desorption rate varies during the b eam exposures. In these experiments, the mechanical shutter was interposed in th e beam path to abruptly discontinue 16O-atom impingement onto the sample surface at different times, and hence atomic oxygen coverages. Shown in Figure 3-3 are the 18O2 desorption traces obtained at surface temperatures of 90 and 110 K when the beam exposure was continued and discontinued in 30 second intervals. At the start of the experiment, th e surface is initially saturated with only 18O2 molecules. If the thermal desorption of 18O2 is negligible, then the 18O2 desorption rate should drop to its ini tial baseline value upon discontinuing the 16O-atom

PAGE 109

97 beam exposure, and should return to the le vel observed before terminating the exposure when the shutter is opened again. At Ts = 90 K the 18O2 partial pressure drops rapidly when the beam exposure is interrupted at 30 seconds and nearly re turns to the initial baseline (Figure 3-3). Upon reopening the shu tter, the desorption rate rose to a value nearly equal to that observed at the end of the first 30 sec ond exposure. These changes show that the thermal desorption of 18O2 occurs very slowly at 90 K as the O-atom coverage increases. The thermal desorption of 18O2 occurs at a measurable rate during the beam exposure at Ts = 110 K. Figure 3-3 shows that the 18O2 desorption rate drops abruptly when the beam exposure is terminated at 30 seconds, but reaches a value well above the initial baseline, and then decays sign ificantly during the 30 seconds that 16O-atom beam is isolated from the chamber. Specifica lly, the desorption rate observed immediately after terminating the beam exposure for the fi rst time is about 24% of the maximum rate obtained during the first 30 second beam exposur e, but after 30 seconds without exposing the sample to the beam, the desorption rate is only about 6% of th e highest rate observed during the initial exposure period. The decay of 18O2 thermal desorption rate indicates that the thermal desorption channel is slow ly deactivating, desp ite the abundance of 18O2 remaining on the surface. A likely interpretation is that 18O2 residing in regions near the O-adatoms are desorbing. When the shutter is reopened at 60 s econds, the desorption rate jumped to a value that is lower than that observed at 30 seconds and then rises to a maximum and decreases, until the beam is re -interrupted. Upon reopening the shutter, the surface concentration of Oatoms begins to increase agai n. The observation of a new maximum indicates that thermal desorption component becomes reactivated as the O-

PAGE 110

98 adatom coverage grows. These changes show that 18O2 thermally desorbs at appreciable rates at Ts = 110 K as the coverage of O-atoms increases during the beam exposure, but that the thermal desorption rate is nearly neg ligible during the beam experiments at 90 K. Using the thermal desorption rates determined from the beam interruption experiments, it is estimated that about 0.04 ML and 0.004 ML of the 18O2 thermally desorbs during the 6 minute beam exposures performed at Ts = 90 and 110 K, respectivel y. Thus, the thermal desorption of O2 accounts for only a small frac tion of the total amount of 18O2 molecules that desorb while the 16O-atom beam impinges on the su rface, at the temperatures examined. 3.4 Discussion Dynamic displacement and dissociation of 18O2 was observed to occur when directing a beam of 16O-atoms toward Pt(111) saturated with 18O2 at temperatures between 90 and 110 K. Molecular oxygen disso ciation at low surface temperatures have also been observed when exposing O2-covered Pt(111) to gaseous hydrogen atoms [14] or an electron beam [9]. The experiments conducted here show that the probability for 18O2 dissociation increases significantly with a m odest change in the surface temperature. At Ts = 110 K, the adsorbed 18O2 continues to dissociate under the 16O-atom flux when the total O-atom coverage, O (16O+18O), increases above 0.25 ML. The thermal dissociation of O2 on Pt(111) ceases under UHV conditi ons once the O-atom coverage reaches 0.25 ML so the observation of 18O2 dissociation above this coverage is convincing evidence for a dissociation mechan ism in which a substantial amount of the 16O adsorption energy is transferred to the 18O2 molecules, efficiently stimulating desorption or dissociation. The dissociation of 18O2 occurred only during the first 40

PAGE 111

99 seconds (0.2 ML O-atom exposure) of the b eam experiment at each surface temperature, indicating that the dissociati on probability depends on the total O-atom coverage, and decreases to zero once a critical O is reached. This observation is consistent with reports that the energy barrier for O2 dissociation on Pt(111) incr eases significantly with increasing O-atom coverage [51]. To gain insights for understanding the surface temperature dependence of the O2 dissociation probability induced by incident O-atoms, it is useful to consider the behavior of O2 on Pt(111) in the absence of impi nging oxygen atoms. Molecular oxygen chemisorbs associatively on Pt(111) at 100 K. Two distinct types of adsorbed O2 species have been identified on Pt(111) using high re solution electron energy loss spectroscopy, ultraviolet photoemission sp ectroscopy and near-edge X-ray absorption spectroscopy measurements [10,93-96]. At low coverages, O2 adsorbs in so-called peroxo and superoxo configurations [9,95], which corre spond to adsorption onto threefold hollow and bridge sites, respectively [97,98]. Molecular oxygen bound at lo w coverages in the peroxo state has been found to dissociate on Pt(111) even at 96 K [9,95] although dissociation is slow at this temp erature. As the coverage of O2 increases, the a majority of the O2 resides in the superoxo state., and at the saturation coverage of 0.44 ML the dissociation of O2 on Pt(111) is no longer observed at surface temperatures below about 92 K [9]. The increase in the dissociation probability of O2 observed when increasing the surface temperature during the 16O-atom exposure may simply indicate that O2 thermally dissociates during these experiments. As di scussed above, the thermal dissociation of O2 on Pt(111) has been observed at surface temperatures below 110 K when the O2 coverage

PAGE 112

100 is below saturation [9,95]. Thus, therma l dissociation could occur during the beam experiments if the displacement of 18O2 from the surface generated enough unoccupied sites to accommodate the 18O atoms resulting from dissociation of adsorbed 18O2 molecules. Arguments against this are as follows. Firstly, the flux of 16O2 from the beam is approximately 20 times the 16O-atom incident flux so the adsorption of 16O2 could compete effectively with 18O2 dissociation for empty surface sites, depending on the characteristic timescales for these processes. Additional evidence that thermal dissociation of 18O2 is negligible in these experiments is the observation that the dissociation yield ne arly doubles at Ts = 110 K when the 16O-atom fluence is increased from 0.1 to 0.2 ML (i.e. 20 to 40 seconds ). If empty site s are created during 18O2 displacement, and the observed dissociation of 18O2 is due to thermal activation of molecules adjacent to these sites, then dissociation should continue during the approximately 10 minute time period that pa sses between termination of the beam exposure and the start of the subsequent TPD measurement. Thus, a factor of about two increase in the 18O2 dissociation yield observed when the 16O-atom exposure is prolonged by only 20 seconds cannot be easily explained by thermally activated dissociation since prior experiments indicate that this reaction is quite slow at surface temperatures of 110 K and below. Finally, the observation that 18O2 dissociation continues to occur up to total O-atom coverages of ~0.40 ML is perhap s the most compelling evidence to support a nonthermal or dynamic mechanism for 18O2 dissociation during 16O-atom adsorption. It is well known that O2 does not dissociate thermall y on Pt(111) in UHV at O-atom coverages greater than 0.25 ML.

PAGE 113

101 The dissociation of O2 observed in this study appear s to be a nonthermal process that competes with the nonthermal displacement of O2 that is stimulated by the adsorption of oxygen atoms incident from the gas-phase. These reactions may be represented by Equations 3-2 and 3-3 shown below. 16O(gas) + 18O2(ad) 16O(ad) +18O2(gas) (3-2) 16O(gas) + 18O2(ad) 16O(ad) + 218O(ad) (3-3) Here, Equation 3-2 depicts the dynamic displacement (not including thermal O2 desorption) of 18O2 and Equation 3-3 represen ts dynamic dissociation of 18O2 caused by a gas-phase 16O-atom. One would intuitively expect that the rate of the dynamic displacement and dissociation could be expressed in the following manner: 2 18O ERate (3-4) c O O Dk Rate 12 18 for O < c (3-5) where RateE and RateD denote the nonthermal 18O2 evolution and dissociation rates. The cross-section for nonthermal displacement is denoted as A simple rate expression is assumed in Equation 3-5, however, the dependence on O is not known. The temperature dependence of the dynamic displacement rate is assumed to be separable and in the form of a rate constant represented by k in Equa tion 3-5. Other parameters are denoted as follows: is the 16O-atom flux onto the surface, 18O2 is the fractional coverage of 18O2, O represents the total fractional co verage of adsorbed oxygen atoms (16O + 18O), and c is the critical O-atom coverage at which th e dissociation process ceases. An analysis of the rate data can be performed to determ ine the relative rates of desorption and dissociation. This method may be complicated when O > 0 due to the thermal desorption

PAGE 114

102 of 18O2. The analysis will then be focused around the initial rate data. It is not possible to determine the initial dissociation rate e xperimentally due to the limitations imposed upon using TPD. The dissociation yield data derived from TPD measurements can be approximated by calculating an average disso ciation rate, by divi ding the dissociation yield by the duration of time ove r which dissociation occurs (40 seconds). The initial evolution and dissociation rates are recorded along with a sum of these rates in Table 3-2. The initial displacement rates shown are consistent with the findings of Wheeler et al. [16] where they observed the initia l desorption probabilit y (desorption rate/16O-atom flux) on the order of 0.4-0.5. As the surface temperature increased, marginal increases in the initial evolution rate were observed. The dissociation rate increase is more pronounced with increasing surface temperature. Experiments perfor med with a surface temperature > 100 K showed that for every 16O-atom impinging on the surface, at least one 18O2 molecule was displaced or dissociat ed. Assuming that the reaction rate Equations 3-2 and 3-3 capture this chem istry, one would anticipate that the 18O2 evolution rate would decrease as the dissociation yield increases. However, the initial displacement probability remained approximately constant. An explanation may be attributed to the two adsorption geometries of adsorbed O2 onto Pt(111). With an 18O2 coverage of 0.44 ML and a su rface temperature of 77 K, the molecule is known to reside predominately at the bridge site. Mo lecular oxygen can also reside at a threefold hollow s ite, where the O-O bond is weaker than that of the superoxo oxygen [95]. As the surface is heated to about 140 K, the superoxo adsorbed molecular oxygen will desorb, and simultaneously pass through the peroxo state and decompose into oxygen atoms. It is possible that the gaseous 16O-atoms may be interacting with two

PAGE 115

103 different types of adsorbed O2 species, and that the relative populations of these species vary with temperature. Since the peroxo species has a lower O-O bond strength, this species may tend to decompose when exposed to oxygen atoms, whereas the superoxo species may be more prone to desorb. The temperature dependence of the dissoci ation process may provide information regarding the activation energy required to convert the supe roxo species into the peroxo state. Temperature programmed desorption provides the total dissociation yield of molecular oxygen under the reaction conditions examined. By definition, the total dissociation yield is the integr al of the rate expression in Equation 3-5 over time yields Equation 3-6. dt T k yieldc O O 1 ) (2 (3-6) Equation 3-6 makes the assumption that th e temperature dependent component is separable and in the form of a rate constant. Also, with the knowledge that the thermal desorption of 18O2 is on the order of 10%, then the ratio of the dissociation yields between two temperatures would be approximately equal to the ratio of the rate constants. This would provide a means for measuring th e average activation energy of dissociation via the following expression. 2 / 1 / 2 1kT Ea kT Ea T Te e yield yield (3-6) Using the information in Table 3-1, an apparent activation energy for dynamic dissociation was found in the rang e of 2.8-3.3 kJ/mol. Recent ab initio calculations suggest that this barrier s hould be about 14 kJ/mol [98]. The activation barrier would most certainly be influenced by the local atom and molecule coverages, which could

PAGE 116

104 explain the differences between these valu es. Recall that the dissociation channel deactivates after the fi rst 40 seconds (0.2 ML 16O-atoms) of the exposure, showing that the activation energy is a function of O. The limitations imposed using TPD makes it difficult to resolve the activation energy for molecular oxygen dissociation as a function of O, in coverage ranges where molecular oxygen decomposes thermally. The calculated activation energy is then an averag e activation energy over the entire range of 0 < O < 0.44 ML. The calculated activation energy obtained is very small when compared with heats of adsorption for O/Pt (111) (350 kJ/mol) [14]. The incoming Oatoms have more than enough energy to overcome this barrier. A more likely interpretation of this data is that the appare nt activation energy is the barrier required for the superoxo species to be converted into th e peroxo species on the su rface, with the zero point taken to be the Pt(111) surface covered with 18O2 in the superoxo state. The oncoming 16O-atom then effectively interacts w ith two different forms of molecular oxygen, one with high probability of displaci ng (superoxo), and the other more likely to dissociate (peroxo). 3.5 Conclusions Dynamic displacement and dissociation of 18O2 adsorbed on Pt(111) was observed when exposed to beam of gaseous 16O-atoms. The displacement rate maximum increased with surface temperature, while the total desorption yield decreased with increasing surface temperature. Subsequent temper ature programmed desorption measurements revealed that the balance of 18O2 decomposed on the surface through a dynamic dissociation event. The dissociation increa sed with surface temper ature and the initial dissociation probabilities (dissociation rate / 16O-atom flux) are estimated as 0.3, 0.4 and

PAGE 117

105 0.6 for surface temperatures of 90, 100 and 110 K respectively. For the dissociation yields an apparent activation energy for dissociation is esti mated as ~3 kJ/mol, which is only about 1% of the 16O-atom adsorption energy (350 KJ/mol) on a clean Pt(111) surface. The temperature dependence of th e dissociation probability and the low activation energy for dissociation were interpreted as evidence for shifts in the relative populations of the O2 from superoxo to peroxo states wi th increasing temperature, with the superoxo species favoring desorption a nd the peroxo species te nding to dissociate. Table 3-1. Displacement and desorption yields of 18O2 when exposed to 16O-atoms while holding the temperature at 90, 100 and 110 K. Temperature (K) 18O2 dissociated (ML) 18O2 displaced (ML) 90 0.06 0.38 100 0.08 0.36 110 0.11 0.33 Table 3-2. Initial evolution rate, the estimated initial dissociation rate, and the sum of the two rates when exposing an 18O2 saturated on Pt(111) to a beam with a flux of 0.005 ML/sec of 16O-atoms. Ts (K) Initial evolution rate (ML/s) Estimated dissociation rate (ML/s) Sum of evolution and dissociation rate (ML/s) 90 0.0024 0.0015 0.0039 100 0.0025 0.0020 0.0045 110 0.0027 0.0030 0.0057

PAGE 118

106 Figure 3-1. Partial pressure traces of 18O2 desorbing from Pt(111) when exposed to 16Oatoms with a surface temperature of 90, 100 and 110 K.

PAGE 119

107 Figure 3-2. Temperature programmed desorp tion spectra taken after precovering the Pt(111) surface with 18O2 and exposing this surface to 16O-atoms. The left panel shows the desorption traces of mass 32, 34 and 36 when the exposure was conducted at 90 K. The right panel shows a comparison of the mass 34 spectra taken after exposing this surface to 1.8 ML of 16O-atoms at 90, 100 and 110 K.

PAGE 120

108 Figure 3-3. Normalized dynamic displacement beam traces of 18O2 when exposed to a 16O-atom beam interrupted every thirty seconds, taken while holding surface temperatures at 90 and 110 K.

PAGE 121

109 CHAPTER 4 CARBON MONOXIDE OXIDA TION FROM HIGH OXYGEN COVERAGE PHASES ON PT(111) The oxidation of carbon monoxide was inves tigated from high coverage phases of oxygen on Pt(111) generated through O-atom beam exposures. Carbon monoxide was found to adsorb onto oxygen covered Pt(111) held at 100 K, with oxygen coverages in the range of 0.25-0.6 monolayer (ML). The total CO uptake was found to be independent of O and is on the order of 0.30 ML. Carbon monoxide adsorption onto platinum oxide was not observed. Temperature progra mmed reaction measurements conducted on surfaces with CO coadsorbed with oxygen atoms show that CO reacts with all oxygen states on Pt(111), but prefer entially reacts with the oxygen atoms having the weakest oxygen atom-platinum binding strength. Isot hermal CO oxidation experiments were conducted on Pt(111) with initial oxygen covera ges in the range of 0.25-0.6 ML and from platinum oxide. The partial pressure CO2 traces from high oxygen coverage phases taken at surface temperatures of 300, 400 and 500 K each had similar features. Each rate curve was characterized by an immediate jump in the CO2 partial pressure, an increase to a rate maximum and finally a decay to the baseline which is consistent with a precursor adsorption mechanism. The overa ll reaction rate increased w ith surface temperature. The oxidation of carbon monoxide on platinum oxide exhibited inherently different kinetic behavior from that observed for coverages of chemisorbed oxygen atoms. Initially, the CO2 formation rate was very low, after a delay, the rate gradually increased to a maximum, then finally decayed to the baseline.

PAGE 122

110 4.1 Introduction Platinum is known for its catalytic activ ity towards oxidizing carbon monoxide [2140,99]. Despite numerous studies performe d on CO oxidation on platinum surfaces, the development of a mechanistic understanding of this reaction remains a challenge. One contribution to this problem is the difficu lty in generating atomic oxygen coverages on Pt(111) greater than 0.25 monolayers (ML) under UHV conditions, where 1 ML is defined as the Pt(111) surface atom density of 1.505 x 1015 cm-2, so that detailed investigations of the reactivity of different forms of atomic oxygen can be performed. Recently, higher oxygen atom coverages on Pt(1 11) have been generated when exposing Pt(111) to active oxidants such as ozone, NO2 and oxygen atoms [51,52,55]. This capability affords an opportunity to both advance the fundamental understanding of the Pt-catalyzed CO oxidation reaction and to ch aracterize the reactivity of oxygen atoms in high-coverage phases on Pt surfaces. The interactions of O2 with the Pt(111) surface have b een studied in detail. It is known that O2 chemisorbs on Pt(111) and generate s a saturation coverage of 0.44 ML [15]. Heating O2-saturated Pt(111) results in O2 desorption at about 140 K, as well as dissociation to produce 0.25 ML of oxygen atoms chemisorbed on the surface, which is the maximum O-atom coverage that can be generated on Pt(111) using O2 in UHV. These oxygen atoms become mobile at a temper ature of about 200 K, and organize into islands with a p(2x2) structure in which th e O-atoms bind at fcc hollow sites [52,100]. Raising the sample temperature to ~7 50 K causes the adsorbed oxygen atoms to recombine and desorb [9,12]. High surface coverages of oxygen atoms have been formed by electron dissociation of O2/Pt(111) [9,50], and by exposing Pt(111) to NO2 [51-53], O3 [54,55], and O-atoms [56]. Recently, we successfully generated atomic oxygen

PAGE 123

111 coverages as high as 2.9 ML on Pt(111) using a beam of oxygen atoms. The development and properties of the resulting surface oxygen phases were investigated in detail using several analytical techniques. As the coverage of oxygen atoms on the Pt(111) surface exceeds 0.25 ML, the onset of O2 desorption is observed to shift to lower temperatures in temperature programmed desorption (TPD) experiments. Two distinct desorption features are observed at temperatures (~550 and 640 K) below that for O2 desorption at O-atom coverages less than 0.25 ML, indicating that the O-Pt(111) binding energy decreases significantly as the O-atom coverage increa ses in this range. Increasing the O-atom coverage beyond about 0.75 ML is accompanied by a decrease in the O2 desorption rate at low temperature, and the development of a single sharp desorption feature that shifts toward higher temperature as the O-atom coverage is increased. Using X-ray photoelectron spectroscopy (XPS) and electron energy loss spectroscopy (ELS) measurements in conjunction with TPD, the sh arp desorption feature could be attributed to the growth of Pt oxide domains on Pt(111) at coverages above 0.75 ML. The adsorption of CO on Pt(111) has b een studied extensively [26,94,101-104]. On clean Pt(111), CO adsorbs on both atop and bridge sites of Pt(111), and reaches saturation coverages of 0.5 and 0.66 ML at surface temperatures of 300 K and 100 K, respectively. At room temperature, CO is distributed equally between atop and bridge sites. The mechanism for CO oxidation on Pt(111) with less than 0.25 ML involves several steps, including the molecular adsorpti on of CO and the dissociative adsorption of O2, the diffusion of CO and O on the surface, and reaction between CO and O to produce CO2. Recently, a CO2 precursor has also been identif ied on Pt(111) [32]. Using single

PAGE 124

112 crystal adsorption calorimetry, Yeo et al. found that the initia l sticking probability of CO onto Pt(111) was enhanced by the presence of adsorbed oxygen atoms, consistent with a precursor-mediated mechanism for CO adso rption [22]. High re solution XPS and ELS studies revealed that the oxygen atoms block the bridge sites, leaving only atop sites for CO adsorption [31,105], which is consistent with density functiona l theory calculations (DFT) [27,34]. At 120 K, CO oxidation does not occur on Pt(111) and CO accumulates on the O-covered surface, reaching a coverage of approximately half of that achieved on the clean surface [31,34]. In the temperature range of 270-300 K, the presence of CO on the surface enhances O-atom migration into islands [25]. Ba sed on results of high resolution XPS, CO oxidation was postulated to occur at the boundari es of p(2x2) O-atom islands [31], which is consistent with real-time scanning tunne ling microscopy (STM) measurements [23]. Above 400 K, the CO2 formation rate is primarily determined by the CO impingement rate on the surface [21]. With the ability to prepare high-covera ge oxygen phases, including Pt oxide, on Pt(111) under UHV conditions, th e basic understanding of Pt-c atalyzed CO oxidation can now be extended to include atomic oxygen cove rages and phases that are certain to be important under conditions relevant to many i ndustrial applications, such as CO oxidation in oxygen-rich environments. In this st udy, the oxidation of CO on oxygen-covered Pt(111) was investigated in UHV over a wide range of initial atomic oxygen coverages using temperature programmed reaction (TPR), isothermal kinetic measurements, and low energy electron diffraction (LEED) to ga in insights into the mechanism for CO

PAGE 125

113 oxidation on different phases of oxygen on Pt( 111) and to characteri ze the reactivity of different states of oxygen on this surface. 4.2 Experimental Methods The experiments were conducted in a thre e-level UHV chamber that reaches a base pressure less than 2 x 10-10 Torr, as described in Section 1.3. Briefly, the chamber is evacuated by an ion pump (400 l/sec), a tu rbo molecular pump (210 l/sec) and a titanium sublimation pump that is inserted into a liquid nitrogen cooled cryoshield. The upper level of the chamber houses a hemispherical analyzer (Specs EA10 plus), a dual Al/Mg anode X-ray source, a variable-energy electr on source and an ion sputter source, which provides capabilities for performing X-ra y photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES), electron en ergy loss spectroscopy (ELS) and low energy ion scattering spectroscopy (LEISS) as well as surface cleaning by ion sputtering. The middle level of the chamber is designed primar ily for gas-dosing, and contains a directed doser connected to a leak valve as well as a calibrated molecular beam doser. The lower level of the chamber houses a low energy el ectron diffraction (LEED) optics, a two-stage differentially-pumped plasma beam source a nd a quadrupole mass spectrometer (QMS). The Pt(111) crystal that was used in this study is a circular disk (10 mm x 1.5 mm) that was cut and polished to within 0.1 of the (111) plane. The crystal was spotwelded to Ta wires and attached to a copper sample holder that is in thermal contact with a liquid nitrogen cooled reservoir. A type K thermoc ouple was spotwelded to the back side of the crystal to measure the sample temperature. The sample was heated resistively and the sample temperature was controlled using a PI D controller to vary the output of a programmable DC supply that delivers power to the sample for heating. The surface temperature could be varied from about 90 to 1200 K with this configuration. The

PAGE 126

114 sample was initially cleaned by sputtering with 600 eV Ar+ ions at a surface temperature of 500 K, followed by annealing for several minutes at 1000 K. Subsequent cleaning involved exposing the Pt sample to O2 for several minutes at 800 K, followed by flashing the sample to about 1000 K to desorb oxygen and carbon oxides. The sample was considered to be clean when no contaminants could be detected with AES, and a sharp, hexagonal (1x1) LEED pattern was observed. Beams containing oxygen atoms were generated using a microwave plasma source that is mounted in a two-stage differentially -pumped chamber attached to the main UHV chamber. Pure O2 (BOC gases, 99.999%) is conti nuously supplied to the discharge chamber of the plasma source and partially di ssociated in a microwave plasma. Species exit the discharge chamber through small holes, a nd form a beam that is directed into the UHV chamber. In the first pumping stage, the beam passes between oppositely charged parallel plates (10 kV/cm) that deflect ions and electrons from the beam. After flowing through a conical skimmer ( = 3 mm) separating the firs t and second pumping stages, the species travel down a quartz tube before entering the UHV chamber. The quartz tube is 60 mm long and has an inner diameter of 6 mm. A mechanical shutte r is located in the first pumping stage to enable control over beam introduction into the main UHV chamber. Mass spectrometric analysis in dicates that beams containing about 6% Oatoms in a balance of O2 are generated under the typical plasma conditions employed, and that O-atom fluxes of ~3 x 1013 cm-2 s-1 reach the Pt(111) surface when the sample is rotated 45 off normal to the beam and locat ed about 50 mm from the end of the quartz tube.

PAGE 127

115 A calibrated molecular beam doser was employed to generate a CO (Praxair 99.99%), beam the design of which closely fo llows the design given by Yates [57]. The distance between the sample and the doser was set to 1.5 cm to ensure a uniform beam profile across the crystal f ace. The mass spectrometer was placed away from the beam region to ensure that the partial pressure traces did not contain any artifacts due to reflection of molecules from the surf ace. Unless otherwise specified, the CO2 pressure traces shown are the original data. The calcu lated reaction rate data was obtained first by smoothing the beam data with a 5 point adjacent averaging method, followed by a background subtraction. 4.3 Results 4.3.1 Carbon Monoxide Adsorption and TPR on O Precovered Pt(111) Surfaces High-coverage oxygen phases were genera ted by exposing the Pt(111) surface to the oxygen atom beam while holding the sample temperature fixed at 450 K. Shown in Figure 4-1 are O2 temperature programmed desorpti on (TPD) traces obtained after adsorbing different amounts of atomic oxygen using the beam. The TPD trace corresponding to an initial O-atom coverage of 0.25 ML exhibits a single, broad feature, labeled as the desorption feature. It is well known that the feature arises from the recombinative desorption of oxygen atoms in itially adsorbed in ordered p(2x2) domains on Pt(111) [9,12,14]. As the atomic oxyge n coverage increase s beyond 0.25 ML, two distinct features, labeled as and appear at desorption temperatures below that observed for the feature. The and features continue to gr ow in intensity as the oxygen coverage increases to about 0.75 ML. The development of the and features indicates that the average O-Pt(111) binding energy decr eases with increasing oxygen

PAGE 128

116 coverage up to 0.75 ML. From leading edge analysis of the TPD traces, our group found that the desorption activation energy is estimated to decrease from ~190 to 120 kJ/mol as the oxygen coverage increases from 0.25 to 0. 75 ML. Prior work indicates that oxygen atoms begin to adsorb in hcp hollow sites at coverages greater than 0.25 ML [52], at which the binding energy is about 50 kJ/mol lo wer than at the fcc hollow site [34] – the preferred O-atom binding site at < 0.25 ML. Closer packing in the chemisorbed layer also results in stronger latera l repulsive interactions betw een adsorbed oxygen atoms as the coverage increases. Both of these factors lower the average O-Pt(111) binding energy. The order in which the desorption features appear with increasing coverage shows that oxygen atoms first arrange into a p(2x2) stru cture, and that the su rface saturates with these domains before more repulsive states ar e generated on the surf ace. In contrast, as the coverage increases beyond 0.25 ML, the lowest temperature state appears before the state saturates, suggesting that oxygen atoms adsorb in distinct repulsive environments in this coverage range, rather than sequentially populating each environment as coverage increase s. We note, however, that the feature grows in intensity much more sharply as the oxygen coverage increases beyond 0.50 ML, than over the 0.25 to 0.50 ML range (Figure 4-1). As discussed previously, the feature appears to arise predominantly from the deso rption of oxygen atoms in itially adsorbed in hcp hollow sites, which form (2x2) domains with a “honeycomb” structure and a local coverage of 0.50 ML [51]. We attributed the feature to desorpti on of oxygen initially adsorbed in disordered domains of local conc entration greater than 0.50 ML, in which the O-surface binding energy is apparently weaker than in the honeycomb domains. If this

PAGE 129

117 interpretation is correct, the TPD results in dicate that the disordered domains form in small quantities at total coverages below 0. 50 ML, and grow more rapidly once the total coverage exceeds 0.50 ML. Islands of platinum oxide grow on the Pt(111) surface when the atomic oxygen coverage increases above about 0.75 ML [55]. As the in itial amount of oxide on the surface increases, the O2 desorption rate maximum shifts toward higher temperature and sharpens dramatically, gi ving rise to explosive O2 desorption. A TPD spectrum obtained from an oxidized Pt(111) surf ace, initially covere d with 1.7 ML of oxygen atoms, is shown in the right panel of Figure 4-1 along with a desorption spectrum obtained from 0.25 ML of O-atoms chemisorbed on Pt(111). The maximum desorption rate is more than 50 times greater from the oxide-covered surface than from the surface with 0.25 ML of O-atoms, and the desorption peak is cen tered at a higher temp erature than both the and features. Notice also that the trailing e dges of the desorption features are virtually identical in Figure 4-1. Th is suggests that 0.25 ML of chemisorbed oxygen atoms on Pt(111) are in similar arrangements on the surface, and therefore experience similar bonding environments, regardless of whether the 0.25 ML coverage is generated by adsorbing oxygen on the initially clean surface or by desorbing oxygen from highcoverage oxygen phases. Temperature programmed reaction (TPR) experiments were performed to characterize the reactivity of the surface oxygen toward CO. In these experiments, atomic oxygen coverages of 0.25, 0.39, 0.59 and 1.7 ML were generated on the Pt(111) sample using the atomic oxygen beam, and the surface was then held at 100 K and saturated with CO supplied from the calibra ted beam doser. TPR measurements were

PAGE 130

118 then conducted using a heating ra te of 3 K/sec, and monitoring the partial pressures of CO, CO2 and O2. No other species were observed to desorb in these experiments. Carbon monoxide did not adsorb on the Pt( 111) surface covered with 1.7 ML of oxygen atoms at a surface temperature of 100 K, indi cating that CO binds very weakly on Pt oxide. The TPR spectra obtained from the surface initially covered with 0.25 ML of Oatoms reveal that the oxygen atoms are completely converted into CO2 and that excess CO desorbs from the surface. For initial atomic oxygen coverages between 0.25 and 0.60 ML, CO, CO2 and O2 were all observed to desorb fr om the surface, indicating that a fraction of the adsorbed CO and O-atoms do not react at these initial coverages. Figure 4-2 shows a TPR spectrum obtained from Pt( 111) initially covered with 0.39 ML of Oatoms and a saturation amount of CO. Carbon di oxide is found to desorb in a single peak centered at 300 K, and the background CO2 pressure was observed to continuously increase during the temperature ramp due to CO2 desorption from the sample holder and the Ta support wires. Two CO desorption feat ures are also observed at temperatures of ~415 and 500 K. The high temperature featur e is known to arise from CO molecules desorbing from surface defect sites, whic h are apparently unreactive toward oxygen adsorbed on Pt(111) [21]. The observati on of CO desorption at 415 K during these experiments indicates that a fr action of the CO adsorbed on Pt(111) terrace sites does not react with the chemisorbed oxyge n atoms either. Finally, O2 desorbs in a single feature centered near 680 K, consistent with oxygen present on the surface in p(2x2) domains once CO is completely removed from the surface.

PAGE 131

119 These results show that a fraction of th e adsorbed CO and O-atoms do not react, even though CO is present on the surface in a limiting concentration, as shown below. The strong binding of CO at defect sites may render these molecules relatively unreactive toward the adsorbed oxygen atoms; however, ab out half of the CO that does not react resides in a more weakly bound state on the te rraces. Previous workers have suggested that CO binding to Pt(111) is destab ilized by chemisorbed O-atoms, causing the oxidation reaction to occur readily at initial oxygen coverages near 0.25 ML [21]. As the oxygen coverage decreases and isolated oxygen atoms are generated during CO oxidation, the CO binding to the surface strengt hens and the CO apparently becomes less reactive toward the surface oxygen. The TPR spectra were analyzed to determine the coverage of CO adsorbed on the surface before and after reaction as a function of the initi al oxygen coverage as well as the amount of oxygen remaining on the surface afte r reaction. To convert the integrated TPD areas into surface coverages, we obtaine d and integrated TPD spectra from wellestablished saturation covera ges of 0.25 and 0.65 ML of O [9,14,51] and CO [26] generated on Pt(111), respectively, by the dissociative chemisorption of O2, and separately from CO adsorption at a surface temp erature of 100 K. The analysis neglects variation in the sensitivity of the mass spectrometer thr oughout the day, which was found to be a reasonable approximation based on the reproducibility of integr ated intensities of the TPD obtained from the saturation O and CO coverages stated above. Finally, the analysis assumes that the CO and O are removed from the surface only by desorption and reaction to generate CO2. For the analysis, the oxyge n coverages before and after reaction are determined directly from the O2 TPD data, and the CO coverage remaining

PAGE 132

120 on the surface after reac tion is also calculated from the CO desorption trace. Since the CO2 background pressure increased continuous ly during the TPR measurements, we chose to calculate the amount of CO molecules that are oxidized to CO2 by computing the difference between the oxygen coverages ob tained before and after reaction. Table 4-1 shows the initial CO and O c overages as well as the amounts of O and CO that do not react as a function of the in itial O-atom coverage. The analysis shows that the saturation CO cove rage at 100 K decreases from 0.65 to 0.31 ML when O increases from 0 to 0.25 ML, which is in ag reement with previous findings [31,100]. The saturation CO coverage decreases remains cons tant, within the experimental uncertainty, as the initial oxygen coverage increases to 0.59 ML. The decrease in the CO saturation coverage to about half its value on the clean surface for an oxygen coverage of 0.25 ML is known to be caused by O-atoms blocking CO adsorption on bridge sites, limiting CO adsorption to only atop sites [31,105]. The i nvariance of the CO saturation coverage as the O-atom coverage increases from 0.25 to about 0.60 ML indicates that oxygen atoms chemisorbed at these higher coverages do not appreciably impede CO adsorption on atop sites. Although we did not investigate CO adsorption at oxygen coverages other than those listed in the table, the inability of CO to adsorb on the oxidized surface at Ts = 100 K suggests that the CO saturation coverage decreases as the fraction of the surface covered by oxide particles increases. The TPR data also provides insights into the relative reactivity of the different surface oxygen phases. Table 4-1 shows that 0.33 ML of the adsorbed O-atoms do not react with CO during the TPR experiment pe rformed with 0.30 ML of CO initially coadsorbed with 0.59 ML of oxygen. Shown in the left panel of Figure 4-3 are O2 TPD

PAGE 133

121 spectra obtained during the TPR experiment and from the surface initially covered with 0.59 ML of oxygen atoms. As may be seen in the figure, the conversion of CO to CO2 is accompanied by a decrease in each of the O2 desorption features, with the largest decrease in the low temperature feature. The trailing edge of the desorption trace is also lower in the TPD spectrum obtained af ter CO oxidation. For an initial oxygen coverage of 0.59 ML, the surface is believe d to be covered with oxygen atoms in both disordered domains and high-coverage ordered domains, and the amount of O2 desorbing in the and features is assumed to be proportion al to the initial concentrations of oxygen atoms in these phases. The latter a ssumption considers the conversion of oxygen atoms from the disordered phase to the highcoverage ordered phase to be negligibly slow during the TPD experiment. In th is case, the larger decrease in the feature suggests that O-atoms associated with the disordered domains are more reactive than those present in the high-coverage ordered domains. This conclusion is reasonable in light of the lower binding energy of oxygen in the disordered phase as suggested by the TPD results (Figure 4-1). However, if the rate of oxygen intercha nge between phases is appreciable during TPD, then the relative decreases in the O2 desorption features observed after reaction with CO will not provide a simple measure of the relative reactivity of the different oxygen phases. For example, each phase ma y be equally reactive toward CO, but a fraction of the oxygen in the more repulsive di sordered phase may be converted to a more stable phase during TPD. This would cau se the amount of oxyge n desorbing in the feature to be less than that remaining on the surface after reaction with CO at lower temperature (~300 K).

PAGE 134

122 A comparison of the O2 TPD spectra obtained from su rfaces with similar initial oxygen coverages, but prepared in different ways, suggests that oxyge n atoms are fairly immobile at temperatures near those that produce O2 desorption. As shown in Table 4-1, the surface is covered with 0.33 ML of oxygen at oms at the end of the TPR experiment in which a saturation coverage of CO was ini tially co-adsorbed with 0.59 ML of O-atoms on Pt(111). Shown in the right panel of Figure 4-3 are the O2 TPD spectra obtained during this TPR experiment and that obt ained from a surface on which 0.25 ML of oxygen atoms were adsorbed by dissociating O2 on the surface. The spectrum obtained in the TPR experiment shows that the 0. 33 ML of oxygen atoms remaining after CO oxidation desorb in varying amounts in the , and desorption states, whereas only the desorption feature is observed in the spect rum obtained from an initial oxygen coverage of 0.25 ML prepared by molecular oxygen adsorption. This shows clearly that th e distribution of oxygen phases or states is not only a function of the oxygen coverage, but is also dependent upon the manner in which that coverage is obtained. Since the feature is not saturated in the TPD spectrum obtained after CO oxidation, the migra tion of oxygen atoms to more en ergetically favorable states must be slow at the temperatures for recombinative O2 desorption. Jerdev et al. [52] have also reported evidence that O-atoms adsorbed at high coverages on Pt(111) have limited mobility at these surface temperatures. Si nce the adsorbed O-atoms are found to be relatively immobile on the surface, the decrease in the desorption feature after reaction with CO is consistent with oxygen atoms in the high-coverage disordered phase being more reactive toward CO than are th e more strongly-bound oxygen species.

PAGE 135

123 When considering all of the observations in this section, the following points can be made. First, CO does not appreciably adso rb onto the platinum oxide, even at 100 K. The total CO uptake is a strong function of O-atom coverage in the range of 0
PAGE 136

124 a function of the surface oxygen coverage si nce oxygen is continuously removed from the surface during reaction. Additionally, sin ce the surface is maintained at a constant temperature, the direct rate measurements al so aid in separating the effects of surface coverage and temperature on th e CO oxidation kinetics. The temporal evolution of the CO2 production rate exhibits similar characteristics for all the surface temperatures investigated and initial oxygen coverages less than 0.60 ML (Figures 4-4 to 4-6). In each measurement, the CO2 production rate increases abruptly when the CO beam exposure to the su rface is initiated. Th e reaction rate then increases toward a maximum during an induc tion period, and decays thereafter to a steady level. Oxidation of CO at the chamber walls causes the final CO2 partial pressures to remain greater than the values obtained before introducing CO into the UHV chamber. Although not shown in the figures, the CO2 pressure rapidly returns to its initial value when the CO flow into the chamber is terminated. Additionally, the final CO2 pressures vary among the measurements shown mainly be cause the rate curves were obtained using slightly different CO fluxes, as indicate d in the figures. After each direct rate measurement was completed, TPD was performed to characterize the species remaining on the surface. Neither O2 nor CO2 was detected by TPD after any of the direct rate experiments, which indicates that all of the adsorbed oxyge n atoms are converted to CO2 by reaction with CO under the conditions investigated. The CO2 production rate at a surface temperat ure of 300 K is shown in the left panel of Figure 4-4 as a f unction of monolayers (ML) of CO exposed to the surface for initial atomic oxygen coverages o i of 0.25, 0.36 and 0.44 ML. The right panel of the figure shows the CO2 production rate as a function of the oxygen atom coverage, which

PAGE 137

125 was determined by integrating the rate versus time data and setting the o scale using the known value of o i. The oxygen coverage o(t) at any time during reaction is determined from Equation 4-1, 0 0) ( ) ( 1 ds s P ds s P tt iO (4-1) where P is the partial pressure rise of CO2 from its initial level, or equivalently the CO2 production rate, and s is an integration variable. Distinct differences are observed in the rate behavior at 300 K measured at the different initial oxygen coverages. Firstly, the maximum reaction ra te increases and is reached at larger CO e xposures with increasing o i. Notice that the maximum CO oxidation rate is uncorrelated with the CO fluxes employed, wh ich suggests that the rate of CO oxidation at 300 K is not controlled by the rate of CO adsorption for CO incident fluxes near 0.1 ML/sec. The difference between the maximum and initial reaction rates, (rmax-ri) also increases with increasing o i. After passing the maxi mum, the reaction rate decreases rapidly but shortly th ereafter the rate curve exhibi ts an inflection point after which the decline in the rate slows down. Th e reaction rate then d ecreases more rapidly and reaches its terminal value after no mo re than 20 ML of CO has impinged on the surface. Note that the CO desorption rate is negligible at a surface temperature of 300 K relative to the adsorpti on rate for CO incident fluxes near 0.1 ML/sec. Thus, the surface coverage of CO will reach high values at 300 K if CO oxidation is also slower than adsorption, which appears to be the case.

PAGE 138

126 An induction period prior to the CO2 production rate maximum has been previously observed in the oxidation of CO over oxide-s upported Pd particles [106], and could be explained in that case by invoking a pr ecursor-mediated desc ription of the CO chemisorption probability. In this model, CO first adsorbs into a weakly bound precursor state from which the molecule can either de sorb or chemisorb into a more strongly bound state. Adsorption into the precursor stat e was assumed to occur only on oxygen-covered domains of the surface and the parameters of the model were chosen such that the CO chemisorption probability sCO decreases with increasing su rface temperature and atomic oxygen coverage according to Equation 4-2, ) ) exp( 1 (sat O O s B app app sat CO CO i CO COT k E s s (4-2) In agreement with experimental observations of CO oxidation on Pd particles [106], the model predicts that the induc tion period prior to the rate maximum and the difference between the initial and maximum reacti on rates increase with increasing surface temperature. This prediction originates fr om the inverse relationship between the CO chemisorption probability and the surface temper ature that is built into the model. Since the chemisorption probability also decreases with increasi ng atomic oxygen coverage, the model will correctly predict th e general trends in the react ion rate with initial oxygen coverage observed in the present study, na mely, the increase in the induction period and the increase in (rmax-ri ) with increasing o i. The change in the slope of the rate curve after the rate maximum is indicative of a sudden increase in the surface coverage of CO due to depletion of oxygen from the surface below a coverage of 0.25 ML. Shown in the right panel of Figure 4-4 are the rate curves plotted as a function of the O-atom coverage, which were obtained by integrating

PAGE 139

127 the rate versus time cu rves and setting the o scale with the known value of o i. Notice that in each case the inflection in the rate curve occurs below an oxygen coverage of 0.25 ML, though not always at the same oxygen c overage. As discussed in Section 4.3.1, the maximum CO coverage CO that can be generated on Pt(111) is relatively insensitive to the O-atom coverage for o = 0.25 to 0.60 ML, at least at 100 K, but CO doubles as o decreases from 0.25 ML to zero. Thus, as oxygen-free regions are generated on the surface during reaction, th e rate of CO adsorption and he nce the CO coverage increases more rapidly and offsets the steep decline in the CO2 production rate due to oxygen removal from the surface. Finally, the in crease in the maximum reaction rate with increasing o i could result from higher oxygen or CO coverages on the surface once the rate maximum is reached. However, as seen in the right pa nel of Figure 4-4, the values of o at the rate maximum (~0.30 ML) do not differ appreciably with increasing o i. Furthermore, it is unlikely that higher CO cove rages are generated since CO is relatively insensitive to a change in oxygen coverage between 0.25 and 0.60 ML. Thus, the increase in rmax with o i appears to be most consistent with a s light increase in the intrinsic reactivity of oxygen atoms populating high-coverage domains rather than the low-coverage p(2x2) phase. This conclusion is consistent with the TPR results discussed in Section 4.1. The production rate of CO2 at a surface temperature of 400 K is shown in Figure 45 as a function of CO exposure in the left panel and O-atom coverage in the right panel for initial atomic oxygen coverage s of 0.25, 0.37 and 0.45 ML. At Ts =400 K, the induction period prior to the rate maximum and the difference between the initial and maximum reaction rates both increase with increasing o i as observed at 300 K.

PAGE 140

128 Characteristics of the rate behavior do ch ange, however, when CO oxidation is conducted at the higher temperature. Interestingl y, the induction period obtained at a given o i remains largely invariant to an increase in the surface temperature is increased from 300 to 400 K, but the value of (r max-ri) does increases with surface te mperature in this range. The model for precursor-mediated adsorption pred icts that both quantities increase with increasing Ts. Also, the maximum reaction rate no longer increases with increasing o i at Ts = 400 K, but appears to scale with the incident CO fl uxes. This behavior is consistent with the CO adsorption rate beginni ng to limit the overall CO2 production rate at 400 K. Since the intrinsic rate constant for CO oxidation on the surface increases with increasing surface temperature, while the net CO adsorption rate decreases, the surface reaction will be limited by the CO coverage as the surf ace temperature is increased. The CO2 production rates are also higher when the surface temperature is increa sed to 400 K. This may be seen by comparing the values of the maxi mum reaction rates obtai ned at 300 and 400 K, and also by noticing that the CO2 production rate falls to its final value after about 15 ML CO is exposed to the surface at 400 K, which is faster than observed at 300 K. The onset of O2 thermal desorption from high-coverage oxygen phases on Pt(111) occurs at about 500 K so this temperature represents an upper bound at which to investigate the CO oxidation reaction. As may be seen in Figure 4-6, the induction period as well as the value of (r max -ri ) both increase with increasing o i. The induction period for a given o i is also approximately the same as that observed at lower surface temperatures. However, the values of (r max -ri ) are now lower for a given o i than were observed at 400 K. Hence, the difference in the initial and maximum reaction rates at

PAGE 141

129 fixed o i do not decrease monotonically with su rface temperature as predicted by the simple CO adsorption model given in Equation 4-2. This observation indicates th at other factors influence the reaction rate at early times. One key difference is that the desorp tion rate of chemisorbed CO molecules, in addition to that of the weakly bound precursors, is significantly high er at 500 K than at the lower temperatures investigated. Cons equently, much lower CO surface coverages are obtained at 500 K and the overall CO2 production rate becomes limited by the rate of CO adsorption. This has been shown to be th e case by Zaera et al. [21] for CO oxidation at Ts > 400 K and o i < 0.25 ML. Although the CO coverages are much lower, the overall CO2 production rate is greater at Ts = 500 K than at lower surface temperature. Again, this is evident from the rate curves since the rate maxima are higher at 500 K than at lower temperatures, and the reaction is co mpleted at lower total CO exposures. 4.3.3 Carbon Monoxide Oxidation from Platinum Oxide The reactivity of CO toward platinum oxide grown on Pt(111) was also investigated using direct rate m easurements. Rate curves for CO2 production as a function of the CO exposure are shown in Figure 4-7 for measurements conducted at surface temperatures from 400 to 550 K and with an initial atomic oxygen coverage of 1.7 ML generated using the atomic oxygen beam. At o i = 1.7 ML, the surface is covered with islands or particles of Pt oxide surrounded by domains of chemisorbed oxygen atoms. In each measurement, the CO2 production rate increases abruptly upon initiating the CO exposure, and then rises more slow ly to a local maximum followed by decay. The behavior observed at early times is attr ibuted to CO reacting with chemisorbed Oatoms on the unoxidized portions of the surf ace. This conclusion was verified by

PAGE 142

130 performing similar experiments in which only about half of the sa mple was located in line-of-sight of the atomic oxygen beam dur ing oxidation of the surface, which resulted in a larger fraction of the surface being covered by chemisorbed O-atoms, rather than oxide, and an increase in th e initial feature in the CO2 production rate curve. After passing through the first maximum, the CO2 production rate begins increasing again and gradually reaches a second, higher maximum at CO exposures greater than 20 ML. The rate maximum sh ifts to longer exposures as the surface temperature is increased from 400 to 550 K. Interestingly, the delay before the rate maximum is significantly longer at Ts = 550 K compared with the lower temperatures. However, once the rate begins increasing, th e CO exposure required to reach the second maximum is relatively insensitive to the surf ace temperature. The value of the maximum CO2 production rate does not differ significantly with surface temperature, nor does it exhibit an obvious trend. The observed varia tions in the delay before the maximum rate may be caused by differences in the incide nt CO fluxes that were employed in the measurements. Finally, TPD measurements showed neither the desorption of O2 nor CO2 after completion of the isothermal CO oxidati on experiments, indica ting that all of the oxygen atoms are converted to CO2 during the beam experiments. It should also be noted that CO2 evolution was not observed when exposing the oxidized surface to the CO beam at Ts = 300 K. The CO oxidation kinetics observed on the oxidized Pt(111) surface is similar to that reported previously for CO oxidation on high-coverage oxygen phases adsorbed on Pd(100) [107]. In that study, STM images pr ovide evidence that CO oxidation actually occurs on the low-coverage (2x2) phase, but that the high-coverage phases rapidly supply

PAGE 143

131 oxygen to the low-coverage phase. As a result the (2x2) domains gr ow on the surface as the high-coverage domains shrink. Although this behavior sugge sts that the highcoverage phases are more active toward CO oxidation, the weak binding of CO on the high-coverage phases prevents CO from adsorb ing on these phases and gives rise to the delay in the overall rate of CO oxidation. A similar mechanism is likely to dictate the oxidation kinetics on oxidized Pt(111). Figure 4-8 shows the CO2 formation rate as a function of CO exposure measured at a surface temperature of 400 K and for init ial atomic oxygen coverages of 1.7, 2.3, and 2.6 ML. As o i increases, the time delay prior to the CO2 rate maximum grows longer, and the rate feature also broadens. The maximum reaction rate is also relatively insensitive to the initial atomic oxygen coverage for o i from 1.7 to 2.6 ML, and, in fact, is nearly the same magnitude for these ini tial coverages and incident CO fluxes. An increase in the induction period as well as the rate broadening with increasing o i were also observed by Zheng and Altman for CO oxidation on Pd(100) and is therefore consistent with the reaction mechanism propos ed by those authors [ 107]. As the initial oxygen coverage increases in the range where Pt oxide forms, the fr action of the surface on which CO can adsorb decreases, causing a lo nger delay before reaction initiates. Once the reaction initiates, the reac tion rate increases since the ar ea of surface available for CO to adsorb and react increases as the oxide pa rticles are consumed. This mechanism as it applies to CO oxidation on oxidized Pt(111) will be elaborated in Section 4.4. To probe the distribution of oxygen phase s on Pt(111) during the course of the reaction, the CO beam exposure was termin ated before and shortly after the CO2 production rate maximum and O2 TPD spectra were collected. Staring with o i = 1.7 ML,

PAGE 144

132 reaction was carried out at Ts = 500 K until the CO exposure reached ~5 ML, which is during the second induction period (F igure 4-7). The subsequent O2 TPD trace is very similar to that obtained from the surface before reaction, and indeed we find from integrating the desorption spectra that le ss than 5% of the initial oxygen atoms are removed during the initial 5 ML CO expos ure to the surface, and is within the experimental uncertainty. Figure 4-9 shows the O2 TPD spectrum obtained after exposing the surface to ~20 ML of CO, which is sufficient to pass the maximum in the CO2 production rate (Fig. 47). The atomic oxygen coverage remaini ng on the surface is 0.47 ML after the 20 ML CO exposure. Also shown in the figure are O2 TPD spectra obtained after adsorbing 0.59 and 1.7 ML of oxygen atoms using the atomic oxygen beam with the surface held at 450 K. A rather sharp O2 desorption feature at about 660 K is observed after the long CO exposure, which is consistent with the majority of surface oxygen atoms residing in oxide domains. Interestingly, the trailing edge of the desorption feature (~ 750 to 800 K) is lower than that obtained from the surfaces with initial O-atom coverages of 0.59 and 1.7 ML prepared by O-atom adsorption. Indeed, the trailing edges near ly overlap in the O2 TPD spectra obtained from Pt(111) after ad sorbing greater than 0.25 ML of atomic oxygen (see Figures 4-1 and 4-9). Notice also that the leadin g edge of the desorption trace obtained after CO oxidation also lies below the leading edge obtained from the surface with o i = 1.7 ML, whereas the opposite behavi or is observed when an atomic oxygen coverage of 0.59 ML is obtained by oxyge n adsorption. Thus, as discussed in Section 4.3.1, the distribution of oxygen phases on the surface depends quite sensitively on how the oxygen coverage is generated.

PAGE 145

133 Dramatic evidence for the dependence of the oxygen phase distribution on the sample history may be seen in the TPD spect ra obtained from Pt(111) covered with 0.47 ML of O-atoms, generated by removal of oxygen atoms from the surface with o i = 1.7 ML versus O-atom adsorption (Figure 4-9). As shown originally by Koel and coworkers [51,52,55], oxygen atoms populate chemisorbed states at o ~ 0.50 ML when the coverage is reached by the adsorption of activ e oxidants. At this coverage, the surface is believed to be covered by disordered and ordered domains of chemisorbed oxygen atoms, which give rise to the and desorption features discussed in Section 4.3.1 and shown in Figures 4-1 and 4-9. The TPD results th erefore firstly indicate that oxide domains remain on the surface when the coverage is decreased to at least 0.50 ML by reaction with CO at Ts = 500 K. The lower leading and trai ling edges of the desorption trace, compared with those obtained after adso rbing O-atoms, indicates further that chemisorbed O-atoms are more reactive toward CO than the oxide phase. Hence, the picture that emerges is that CO adsorb s and reacts selectiv ely on surface domains containing chemisorbed O-atoms. Since CO adsorption on the oxide particles is inefficient, at best, the ox ide particles decompose by suppl ying oxygen to the regions of the surface covered by chemisorbed oxygen atom s where reaction with CO is relatively rapid. The migration of oxygen atoms from the oxide to the low-coverage domains appears to be slow at 500 K, with the resu lt being that most of the surface oxygen atoms remain in oxide domains at about 0.50 ML. 4.3.4 Low Energy Electron Diffraction Results Forming platinum oxide by exposing Pt(111) to an oxygen atom beam disrupts the long range order of the surface. A series of LEED experiments were performed to

PAGE 146

134 determine qualitatively how the structure of the oxidized Pt(111) surface changes during reaction with CO. At each step, a digi tal photograph was taken of the LEED image. Before initiating the CO reaction, the surface was initially covered with 1.7 ML of Oatoms by exposing the surface to the oxygen atom beam while holding the surface temperature at 450 K. In this coverage re gime, TPD shows that the majority of the oxygen atoms have agglomerated into oxide is lands. A LEED image was taken from this surface and is shown in Figure 4-10A. The su rface was then exposed to the CO beam with the surface held at 400 K, until CO had removed all of the oxygen from the surface. A LEED image was taken from this surface a nd is shown in Figure 4-10B. The reduced surface was then flashed to 600 K and 950 K a nd LEED images were recorded as shown in Figures 4-10C and D respectively. The diffuse substrate (1x1) pattern, along with a high intensity background may be seen in the LEED image obtained from the oxidized surface (Figure 4-10A), and reveals that oxide island formation roughens the surf ace significantly. After completely reducing the surface with CO at Ts = 400 K, the (1x1) pattern of the Pt(111) substrate sharpens and the diffuse background diminishes significan tly (Figure 4-10B), indicating that the removal of oxygen by CO restores the long-or der of the surface considerably, even at Ts as low as 400 K. The faint background that may be seen in the LEED image could arise from small quantities of CO remaining on the surface after reaction, or may indicate that the long-range order of the surface was not re stored to its condition before oxidation. Heating the surface to 600 K desorbs all of the CO from the surface, but does not noticeably reduce the background intensity in the LEED image (Figure 4-10C), thereby suggesting that the Pt surface does maintain a degree of disorder after the removal of

PAGE 147

135 oxygen by CO at 400 K. Heating the surface to 950 K causes the diffuse background to diminish (Figure 4-10D), indicating that elev ated temperatures are required to improve the crystallinity of the surface after oxidation and then reduction by CO. 4.4 Discussion The isothermal CO oxidation experiments conducted with O < 0.5 ML, at each temperature show that the CO2 formation rate is characterized by an initial jump, which proceeded through a rate maximum, and fina lly decayed back to the baseline. The reaction rates increased with su rface temperature, indicating th at an increase in the rate constant compensates for the decrease in CO with increasing temperature. Qualitative features such as the O dependent variations in the quantity (r max -ri ) were captured quite well by the precursor mechanism described by Pi ccolo et al. [106]. However, this model does not capture all of the physics, one such example being the non-monotonic variation of the quantity (r max -ri ) with Ts. Possible differences between our reaction system and the model, given by Equation 4-2, may be responsible for these discrepancie s. For example, the model assumes that CO sat remains constant. The TPR data obtai ned in the present study shows that the maximum amount of CO deposited on th e surface remains constant for O in the range of 0.25-0.6 ML. However, CO sat varies significantly below a 0.25 ML coverage of adsorbed oxygen atoms. Additionally, an implicit assumption of model is an average O dependence on the CO sticking probability. Temperature progr ammed desorption shows three distinct desorption features in this coverage range. Temperature programmed reaction conducted after saturating CO on the high coverage pha ses of oxygen revealed a reduction in all

PAGE 148

136 three oxygen desorption states. This indica tes that multiple oxygen phases reside on the surface and react with CO. It is possibl e that the CO sticking probability may be different on each phase. Finally, the activation energy for desorption of oxygen is different in each state. Temperature programmed reaction measurements show that the CO will preferentially react with oxygen atoms in the state. This strongly implies that the rate constant for the CO oxidation reaction may be dependent on O. Despite the rich CO and O-atom behavior on Pt(111), the mode l given by Piccolo et al. [106] does reproduce the main kinetic features of this system quite well in the oxygen atom coverage range of 0.25-0.6 ML. Carbon monoxide oxidation on platinum oxide follows a reaction kinetic behavior that differs from that observed for high c overage phases of chemisorbed oxygen atoms on Pt(111). Indeed, the CO oxidation kinetics a ppears similar to that observed on palladium surfaces covered with high density oxygen pha ses [106,108-110]. Of pa rticular relevance is an investigation performed by Zheng a nd Altman [107], wher e they studied the reactivity of various oxygen phases on Pd(100) toward CO. Surfaces with different atomic oxygen phases were exposed to CO wh ile monitoring the par tial pressure of CO2. As the atomic oxygen coverage increase d, the delay before the onset of CO2 formation increased and the CO2 reaction feature broadened. The temperature dependence of the rate behavior was explained through a limited lifetime of a precursor CO molecule on the surf ace. As the surface temperature increases, the CO lifetime on the surface decreases, reducing the reaction rate. The relative reactivity of carbon monoxide was highest on the lowest oxygen coverage phase, and

PAGE 149

137 decreased as the O-atom density increased. This decrease in the reaction rate with increasing oxygen coverage was explained by a decrease in the CO sticking probability on the high density O-atom phases. To explai n their rate curves, they propose that there is a rapid exchange between these phases, with the oxide (surface or bulk) rapidly supplying O-atoms to the lower coverage phase s, which is then consumed in the CO oxidation reaction. Thus, the phase with the hi ghest O-atom density feeds the phase with the next highest O-atom density. This was confirmed by STM measurements, which show that only areas diminished during the reaction was the portion covered by the highest O-atom density phase. Once the highest density phase was removed, the surface area covered by the phase with the next highest density began to shrink. As these phases are removed, domains form on which CO has a higher sticking probability, thereby promoting the CO oxidation reaction. The CO oxidation rate behavior observed by Zheng et al. appears very similar to the rate curves obtained from the oxidized Pt (111) surface. Specifically, they observed a delay before the reaction initiated, followed by a gradual rise toward the rate maximum. In our study, we observed a delay before the ra te initiation, which is attributed to a low sticking probability of CO onto the oxidized surface. Additional supporting evidence for this is given in the Section 4.3.1, where CO was found not to appreciably adsorb onto the oxide even at 100 K. Zheng et al. observed a rapid equilibri um between the oxygen phases on Pd(100). In our study, the equilibrium between the phases with different O-atom density does not appear to be rapid. First, c onsider the TPR results. The desorption state correspond to chemisorbed oxygen atoms with the highest surface-binding energy. After TPR was

PAGE 150

138 conducted with 0.59 ML of O-atoms and about 0. 30 ML of CO molecules, a decrease in the quantity of oxygen desorbing the feature was observed. Th is suggests firstly that some oxygen atoms in the state are converted to CO2 and secondly that the migration of oxygen atoms into these states is slow. Also the TPD spectrum taken after partially reducing the oxidized surface with CO show s that oxygen residing on the surface exists primarily as oxide islands. Further evidence for the slow process comes from contrasting the rate data with 0 < O i < 0.5 ML with that of the oxide The reaction rate increased with temperature for the O-atom covered su rface, whereas the overall reaction rate appears to remain independent of temperatur e for reaction with the oxide. This suggests that the oxide does not efficiently supply the adatom phase with O-atoms, limiting the reaction region to the interface between the platinum metal a nd the oxide interface. Since this process is slow, it is assumed to be rate limiting. The slow initial CO oxidati on rates from the oxidized su rface can be interpreted as CO reacting at the island in terfaces. Once metallic domai ns are nucleated, the oxide slowly supplies oxygen to the reaction at the metal sites near the island boundaries. Additional supporting evidence comes from the CO oxidation curves with increasing amounts of oxide initially depos ited on the surface. As th e oxide coverage increases, a corresponding increase was observed in the lag before the onset of CO2 formation. This is consistent with the idea that the reacti on with the oxide occurs primarily at island boundaries, where the quantity of platinum metal sites decreases as the oxide grows. 4.5 Conclusions Carbon monoxide oxidation was investigat ed on Pt(111) surf aces with initial oxygen coverages greater than 0.25 ML. At 100 K, CO was not observed to adsorb onto

PAGE 151

139 platinum oxide. However, CO readily adsorbed onto Pt(111) with a surface concentration of oxygen atoms in the range of 0.25-0.6 ML, with the saturation CO coverage remaining constant at about 0.30 ML. Temperature programmed reaction from these surfaces showed the formation of carbon dioxide as well as desorbing carbon monoxide and oxygen. An examination of the O2 desorption features clearly shows a preference for CO reaction with oxygen atoms that generate the desorption state and that the oxygen migration between th e desorption states is slow. The isothermal reaction data shows that the reaction kineti cs for CO oxidation platinum oxide differs significantly from that observed for chemisorbed O-atoms on Pt(111). High coverage oxyge n phases (0.25-0.6 ML) reacting w ith CO were studied as a function of surface temperature, and the re action kinetics closely follows a precursormediated mechanism for CO adsorption. Si milar studies were performed on platinum oxide, and each curve has the following characte ristics. A delay is observed before the onset of the reaction, followed by an increase in the reaction rate towards the maximum and a subsequent decay to the baseline. This indicates that the oxidation reaction occurs primarily at the interface between the oxide and the metal surface. The reaction rate initiation was most efficient at the lowest temperature examined, and is attributed to longer CO precursor lifetimes on the surface. A large amount of disorder is observe d upon oxidizing Pt(111). LEED results show a restoration of the (1x1) pattern w ith the presence of a background when reducing the surface with CO at 400 K, indicating a signi ficant amount of order regeneration. The background was only removed after flashing the sa mple to annealing temperatures. This

PAGE 152

140 shows that the surface order is not complete ly restored by reduci ng the surface with CO at 400 K. Table 4-1. Maximum CO coverage achieved as a function of initial O. Initial O Total CO O desorbing CO desorbing 0.00 0.65 0.00 0.65 0.25 0.31 0.00 0.06 0.38 0.32 0.12 0.06 0.59 0.30 0.33 0.04 1.70 0.00 1.70 0.00 Figure 4-1. Temperature programmed desorpti on spectra taken from the Pt(111) surface with a ramp rate of 3 K/sec Left panel: coated with 0.25, 0.39 and 0.59 ML oxygen atoms. Three desorption st ates are observed and labeled and Right panel: TPD spectra taken from an oxidized Pt(111) surface with a coverage of equal to 1.7 ML compared with TPD taken with a coverage of 0.25 ML.

PAGE 153

141 Figure 4-2. Temperature program med reaction spectra taken with a ramp rate of 1 K/sec, showing O2, CO and CO2 desorption after preparing a saturation coverage of CO on Pt(111) with 0.39 ML of O-atoms at Ts = 100 K.

PAGE 154

142 Figure 4-3. Temperature programmed desorpti on and reaction spectra taken from the Pt(111) surface with a ramp rate of 3 K/sec. Left Panel: An oxygen TPD spectrum from Pt(111) with an oxygen atom coverage of 0.59 ML compared with a TPR spectrum taken when this surface was saturated with CO. Right panel: A TPR spectrum yielding an oxygen atom coverage of 0.33 ML compared with a TPD spectrum corre sponding to a 0.25 ML coverage of Oatoms on Pt(111).

PAGE 155

143 Figure 4-4. Left panel: Carbon dioxide pressure tr aces with initial oxygen coverages of 0.25 ML ( CO = 0.12 ML/sec), 0.33 ML ( CO = 0.09 ML/sec) and 0.44 ML ( CO = 0.10 ML/sec) taken while the Pt (111) sample was held at 300 K plotted against CO exposure. Ri ght panel: Calculated rate of CO2 formation plotted as a function of O.

PAGE 156

144 Figure 4-5. Left panel: Carbon dioxide pressure tr aces with initial oxygen coverages of 0.25 ( CO = 0.12 ML/sec), 0.37 ( CO = 0.09 ML/sec) and 0.45 ML ( CO = 0.10 ML/sec) taken while the sample was held at 400 K, plotted against CO exposure. Right panel: Calculated rate of CO2 formation plotted as a function of O.

PAGE 157

145 Figure 4-6. Left panel: Carbon dioxide pressure tr aces with initial oxygen coverages of 0.25 ML ( CO = 0.12 ML/sec), 0.35 ML ( CO = 0.10 ML/sec) and 0.48 ML ( CO = 0.11 ML/sec) taken while the Pt( 111) sample was held at 500 K, plotted against CO exposure. Ri ght panel: Calculated rate of CO2 formation plotted as a function of O.

PAGE 158

146 Figure 4-7. Carbon dioxide form ation rate plotted as a func tion of CO exposure with an initial O-atom coverage of 1.7 ML at surface temperatures of 400 ( CO = 0.11 ML/sec), 450 ( CO = 0.09 ML/sec), 500 ( CO = 0.12 ML/sec) and 550 K ( CO = 0.10 ML/sec).

PAGE 159

147 Figure 4-8. Carbon dioxide formation rate pl otted as a function of CO exposure with initial O-atom coverages of 1.7, 2.3 a nd 2.6 ML with surface temperature held constant at 400 K.

PAGE 160

148 Figure 4-9. TPD traces taken with an initia l coverage of 1.7 ML and with the same coverage after a 20 ML CO exposure (0.47 ML) at 500 K. The dashed trace corresponds to a surface with an initial O-atom coverage of about 0.5 ML.

PAGE 161

149 A) B) C) D) Figure 4-10. LEED images taken with an el ectron energy of 163 eV, L1/3 = 518 V, L2 = 32 V, Screen = 4 kV, Wehnelt = 0, and suppressor = 9V on A) Pt(111) covered with 1.7 ML of oxygen atoms, B) after removing all of the oxygen from part a with CO at 400 K, C) flas hing the sample prepared in B to 600 K to remove all adsorbed CO, and D) flashing the surface to 950 K. Each image has been inverted and the contrast has been slightly changed by the same amount to enhance the background.

PAGE 162

150 CHAPTER 5 CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK 5.1 Synopsis Investigations of the interactions of oxygen atoms with solid surfaces provided deeper insight into semiconductor oxidation. By performing XPS before and after exposing nitrided Si(100) to a plasma activated O2 beam, conclusive evidence was found showing that the oxygen atoms primarily adsorb at dangling bond sites with a negligible amount of direct insertion in to silicon-silicon bonds. Th e primary reason that silicon nitride is resistant to oxida tion is its lower dangling bond de nsity compared with pure Si surfaces. Through quantitative analysis of the N1s and O1s spectra, it was determined that the nitrogen species remained phase separated from the oxidic phase. Molecular oxygen is known to chemisorb ont o Pt(111) at cryoge nic temperatures. Exposure of adsorbed 18O2 to gaseous 16O-atoms yields both displacement and dissociation of 18O2 initially adsorbed on Pt(111). The desorption rate increased with surface temperature, whereas the desorption yi eld decreased with su rface temperature. Upon performing TPD, it was determined that the balance of 18O2 dissociated on the surface. The initial displacement probabilities (displacement rate/16O-atom impingement rate) are 0.48, 0.50 and 0.54 at 90, 100 and 110 K, respectively. Dissociation occurred only during the first 40 sec onds (0.2 ML fluence of 16O-atoms) of the exposure and the initial dissociation probabilities (dissociation rate/16O-atom impingement) are estimated to be 0.3, 0.48 and 0.66 at 90, 100 and 110 K, respectively. An apparent average activation energy of ~3 kJ/mol was f ound for dissociation, which can be easily

PAGE 163

151 surmounted by the energy dissipated as the 16O-atoms adsorb onto Pt(111) (350 kJ/mol). The temperature dependence of the dissoci ation probability, and the low activation energy for dissociation were interpreted as evidence of an incr easing peroxo population with surface temperature, with the peroxo species more susceptible to dissociate. By exposing platinum surfaces to active oxidants such as atomic oxygen, high oxygen coverages on Pt(111) can now be prepared under UHV conditions. This provides the means to conduct molecular beam studies on this important catalyst. The reactivity of high oxygen coverage phases generated on Pt(111) toward CO oxidation was probed through isothermal kinetic measurements, TPD, TPR, and LEED. CO uptake in the Oatom coverage range of 0.25 to 0.6 ML was i ndependent of O-atom coverage. However, no CO uptake was observed on platinum oxide. The oxygen atoms with the lowest surface binding energy preferen tially reacted to form CO2. High coverage oxygen phases (0.25-0.6 ML) reacting with CO were st udied as a function of surface temperature, and the reaction kinetics closely follows a precursor-mediated mechanism for CO adsorption. CO oxidation on platinum oxide oc curs at the interface between the metallic regions and the oxide islands. LEED resu lts show a significant amount of order restoration when reducing the oxidized surface with CO at 400 K. The presence of a background indicates that some degree of di sorder remains on the surface at 400 K. 5.2 Future Work 5.2.1 Nonthermal Reaction Mechanisms The ability to generate gas phase at omic oxygen presents an opportunity to contribute in the areas of space vehicle degr adation, plasma processing, plasma catalysis, combustion reactions, and fundamental reaction mechanisms. A means of contributing to each area would be to perform molecular leve l studies on these systems. Experiments

PAGE 164

152 can be designed to investigate these system s by preparing well defined surface-adsorbate systems and subsequently exposing these surfaces to the O-atom beam. One could then tailor the materials in these volatile reaction environments to achieve the desired results. 5.2.2 Scanning Tunneling Microscopy Measurements LEED images have provided some information regarding surface structural changes during the oxidation of Pt(111). Pa rticularly relevant data regarding the long range order of the surface was obtained on th e O-atom/Pt(111) system. More detailed information can be obtained using the atomic resolution capabilities of STM. This would serve to verify the information obtaine d through the LEED measurements, and also determine the structural charac teristics of the oxide islands. Scanning tunneling microscopy would also provide valuable information as the adsorbed oxygen species reacts with CO. The kinetic data and TPR measurements suggest an inhomogeneous O-atom removal me chanism during the course of the reaction. Using STM, it would be possible to determine the structures of the oxygen phases as they are removed during the reaction. This would pr ovide deeper insight into the surface sites that are reactive toward CO oxidation. 5.2.3 Mechanistic Catalytic Studies Catalytically active metal oxides can now be formed under UHV conditions, by exposing metallic surfaces to an active oxidant such as O-atoms. This enables molecular level surface induced reactions to be st udied on these new oxygen phases on Pt(111). Immediate suggestions for future studies w ould be to grow platinum oxide on Pt(111) and conduct reactive scattering measurements w ith relevant reagents in the molecular beam such as hydrocarbons or NOx gases.

PAGE 165

153 Alternatively, similar experi ments can be performed on di fferent platinum crystal faces. This would provide another dimensi on in these reaction studies revealing both how the oxide structure varies on these faces, and the reactiv ity of these different oxide structures towards the reactant in question. A comparison of this data would contribute to the structural design of model catalysts, which would better simu late actual reactive surface particles. These experiments by no means should be limited to platinum. Several catalytically active metals su ch as palladium, gold, iridiu m, and ruthenium should be examined in this manner as well. As a de tailed understand ing of the reactions on these surfaces and oxides are developed at the molecular level, improved catalytic materials and processes can be engineered.

PAGE 166

154 LIST OF REFERENCES 1. M. Takahashi, M. Tamura, Asuha, T. Kobayashi, and H. Kobayashi, "Ultrathin silicon oxynitride formed by low-energy el ectron impact plasma nitridation and chemical oxidation methods," J. Appl. Phys. 94, 726 (2003). 2. G. D. Wilk, R. M. Wallace, and J. M. Anthony, "High-kappa gate dielectrics: Current status and materials properties considerations," J. Appl. Phys. 89, 5243 (2001). 3. S. W. Park, Y. K. Baek, J. Y. Lee, C. O. Park, and H. B. Im, "Effects of annealing conditions on the properties of tantal um oxide-films on silicon substrates," J. Electron Mater. 21, 635 (1992). 4. W. R. Hitchens, W. C. Krusell, and D. M. Dobkin, "Tantalum oxide thin-films for dielectric applications by low-pressure chemical-vapor-dep osition physical and electrical-properties," J. Electrochem. Soc. 140, 2615 (1993). 5. R. A. B. Devine, "Nondestructiv e measurement of interfacial SiO2 films formed during deposition and annealing of Ta2O5," Appl. Phys. Lett. 68, 1924 (1996). 6. E. Atanassova and D. Spassov, "X-ray photoelectron spectroscopy of thermal thin Ta2O5 films on Si," Appl. Surf. Sci. 135, 71 (1998). 7. H. F. Luan, B. Z. Wu, L. G. Kang, B. Y. Kim, R. Vrtis I, D. Roberts I, and D. L. Kwong, "Ultra thin high quality Ta2O5 gate dielectric prepared by in situ rapid thermal processing," in C.C. Proc. IEDM Tech. Dig. San Francisco, CA, 385, 5-8 April (1999). 8. M. Passacantando, F. Jolly, L. Lozzi, V. Salerni, P. Picozzi, S. Santucci, C. Corsi, and D. Zintu, "The effects of silicon n itride and silicon oxynitride intermediate layers on the properties of tantalum pent oxide films on silicon: X-ray photoelectron spectroscopy, X-ray refl ectivity and capacitance-voltage studies," J. Non-Cryst. Solids. 322, 225 (2003). 9. H. Steininger, S. Lehwald, and H. Ibach, "Adsorption of oxygen on Pt(111)," Surf. Sci. 123, 1 (1982). 10. C. Puglia, A. Nilsson, B. Hernnas, O. Karis, P. Bennich, and N. Martensson, "Physisorbed, chemisorbed and dissociated O2 on Pt(111) studied by different corelevel spectroscopy methods," Surf. Sci. 342, 119 (1995).

PAGE 167

155 11. A. C. Luntz, M. D. Williams, and D. S. Bethune, "The sticking of O2 on a Pt(111) surface," J. Chem. Phys. 89, 4381 (1988). 12. J. L. Gland, B. A. Sexton, and G. B. Fish er, "Oxygen interacti ons with the Pt(111) surface," Surf. Sci. 95, 587 (1980). 13. C. T. Campbell, G. Ertl, H. Kuipers, and J. Segner, "A molecular-beam study of the adsorption and desorption of oxygen from a Pt(111) surface," Surf. Sci. 107, 220 (1981). 14. J. Biener, E. Lang, C. Lutterloh, and J. Kuppers, "Reactions of gas-phase H-atoms with atomically and molecularly adsorbed oxygen on Pt(111)," J. Chem. Phys. 116, 3063 (2002). 15. A. N. Artsyukhovich, V. A. Ukraintsev, and I. Harrison, "Low temperature sticking and desorption dynamics of oxygen on Pt(111)," Surf. Sci. 347, 303 (1996). 16. M. C. Wheeler, D. C. Seets, and C. B. Mullins, "Angular dependence of the dynamic displacement of O2 from Pt(111) by atomic oxygen," J. Chem. Phys. 107, 1672 (1997). 17. C. T. Rettner and J. Lee, "Dynamic displacement of O2 from Pt(111) a new desorption mechanism," J. Chem. Phys. 101, 10185 (1994). 18. Y. Yazawa, N. Kagi, S. Komai, A. Satsuma, Y. Murakami, and T. Hattori, "Kinetic study of support effect in the propane combustion over pl atinum cat alyst," Catal. Lett. 72, 157 (2001). 19. R. F. Klie, M. M. Disko, and N. D. Browning, "Atomic scale observations of the chemistry at the metal-oxide interface in heterogeneous catalysts," J. Catal. 205, 1 (2002). 20. P. A. Carlsson, L. Osterlund, P. Thormahle n, A. Palmqvist, E. Fridell, J. Jansson, and M. Skoglundh, "A transient in situ FTIR and XANES study of CO oxidation over Pt/Al2O3 catalysts," J. Catal. 226, 422 (2004). 21. F. Zaera, J. Y. Liu, and M. D. Xu, "Isothermal study of the kinetics of carbon monoxide oxidation on Pt(111): Rate dependence on surface coverages," J. Chem. Phys. 106, 4204 (1997). 22. Y. Y. Yeo, L. Vattuone, and D. A. Ki ng, "Calorimetric heats for CO and oxygen adsorption and for the catalytic CO oxidation reaction on Pt(111)," J. Chem. Phys. 106, 392 (1997). 23. S. Volkening and J. Wintterlin, "CO oxidation on Pt(111) scanning tunneling microscopy experiments and Monte Carlo simulations," J. Chem. Phys. 114, 6382 (2001).

PAGE 168

156 24. J. M. Orts, E. Louis, L. M. Sander, J. M. Feliu, A. Aldaz, and J. Clavilier, "Monte Carlo simulation of CO adlayers electrooxidation on Pt(111)," Surf. Sci. 416, 371 (1998). 25. I. Nakai, H. Kondoh, K. Amemiya, M. Nagasaka, A. Nambu, T. Shimada, and T. Ohta, "Reaction-path switching induced by spatial-distribution change of reactants: CO oxidation on Pt(111)," J. Chem. Phys. 121, 5035 (2004). 26. J. S. McEwen, S. H. Payne, H. J. Kreuzer, M. Kinne, R. Denecke, and H. P. Steinruck, "Adsorption and desorption of CO on Pt(111): a comprehensive analysis," Surf. Sci. 545, 47 (2003). 27. M. Lynch and P. Hu, "A density functional theory study of CO and atomic oxygen chemisorption on Pt(111)," Surf. Sci. 458, 1 (2000). 28. E. S. Kurkina and N. L. Semendyaeva, "Fluctuation-induced transitions and oscillations in catalytic CO oxi dation: Monte Carlo simulations," Surf. Sci. 558, 122 (2004). 29. K. L. Kostov, P. Jakob, and D. Me nzel, "A new high density CO/oxygen coadsorbate layer on Pt(111) a nd its role in CO oxidation," Surf. Sci. 377, 802 (1997). 30. O. Kortluke and V. N. Kuzovkov, "Simula tion of kinetic oscillations in surface reactions on reconstructing surfaces," J. Chem. Phys. 110, 11523 (1999). 31. M. Kinne, T. Fuhrmann, J. F. Zhu, C. M. Whelan, R. Denecke, and H. P. Steinruck, "Kinetics of the CO oxidati on reaction on Pt(111) studied by in situ high-resolution X-ray photoelectron spectroscopy," J. Chem. Phys. 120, 7113 (2004). 32. S. J. Han, C. W. Lee, H. Y oon, and H. Kang, "Discovery of CO2 precursor in the reaction of CO and O on Pt(111)," J. Chem. Phys. 116, 2684 (2002). 33. R. J. Finlay, T. H. Her, C. Wu, and E. Mazur, "Reaction pathways in surface femtochemistry: routes to de sorption and reaction in CO/O2/Pt(111)," Chem. Phys. Lett. 274, 499 (1997). 34. A. Eichler and J. Hafner, "Reaction cha nnels for the catalytic oxidation of CO on Pt(111)," Surf. Sci. 435, 58 (1999). 35. S. Deliwala, R. J. Finlay, J. R. Goldman, T. H. Her, W. D. Mieher, and E. Mazur, "Surface Femtochemistry of O2 and CO on Pt(111)," Chem. Phys. Lett. 242, 617 (1995). 36. R. E. R. Colen, J. Christoph, F. Pena, and H. H. Rotermund, "Oxidation of CO on Pt(111) at intermediate pressures," Surf. Sci. 408, 310 (1998).

PAGE 169

157 37. D. J. Burnett, A. T. Capitano, A. M. Gabelnick, A. L. Marsh, D. A. Fischer, and J. L. Gland, "In situ soft X-ray studies of CO oxidation on the Pt(111) surface," Surf. Sci. 564, 29 (2004). 38. K. H. Allers, H. Pfnur, P. Feulner, and D. Menzel, "Fast reac tion-products from the oxidation of CO on Pt(111) angular a nd velocity distributions of the CO2 product molecules," J. Chem. Phys. 100, 3985 (1994). 39. A. Alavi, P. J. Hu, T. Deutsch, P. L. Silvestrelli, and J. Hutter, "CO oxidation on Pt(111): An ab initio density functional theory study," Phys. Rev. Lett. 80, 3650 (1998). 40. C. Akerlund, I. Zoric, and B. Kase mo, "A collision induced reaction: CO2 production on O2 and CO covered Pt(111)," J. Chem. Phys. 104, 7359 (1996). 41. D. V. Shalashilin, B. Jackson, and M. Pe rsson, "Eley-Rideal and hot-atom reactions of H(D) atoms with D(H)-covered Cu( 111) surfaces; quasiclassical studies," J. Chem. Phys. 110, 11038 (1999). 42. T. Kammler and J. Kuppers, "Interacti on of H-atoms with Cu(111) surfaces: Adsorption, absorption, and abstraction," J. Chem. Phys. 111, 8115 (1999). 43. T. H. Kammler, J. Lee, and J. Kuppers "A kinetic study of the interaction of gaseous H(D) atoms with D(H) adsorbed on Ni(100) surfaces," J. Chem. Phys. 106, 7362 (1997). 44. C. T. Rettner, D. J. Auerbach, and J. Lee, "Dynamics of the formation of CD4 from the direct reaction of in cident D atoms with CD3/Cu(111)," J. Chem. Phys. 105, 10115 (1996). 45. J. Y. Kim and J. Lee, "Kinetics, mech anism, and dynamics of the gas-phase H(D) atom reaction with adsorbed D(H) atom on Pt(111)," J. Chem. Phys. 113, 2856 (2000). 46. J. Y. Kim and J. Lee, "Mechanism and cross sections for HD and CH4-xDx (x=1-4) formation in D(g)+CH3/Cu(111) reaction at 100 K," J. Chem. Phys. 112, 6015 (2000). 47. T. A. Jachimowski and W. H. Weinber g, "Direct reaction of gas-phase atomichydrogen with chemisorbed hydrogen on Ru(001)," J. Chem. Phys. 101, 10997 (1994). 48. J. R. Engstrom, D. J. Bonser, and T. Engel, "The reaction of atomic oxygen with Si(100) and Si(111) .2. adsorp tion, passive oxidation and th e effect of coincident ion-bombardment," Surf. Sci. 268, 238 (1992). 49. T. Engel, "The interac tion of molecular and atom ic oxygen with Si(100) and Si(111)," Surf. Sci. Rep. 18, 91 (1993).

PAGE 170

158 50. J. L. Gland, "Molecular and atomic ad sorption of oxygen on the Pt(111) and Pt(S)12(111)X(111) surfaces," Surf. Sci. 93, 487 (1980). 51. D. H. Parker, M. E. Bartram, and B. E. Koel, "Study of high coverages of atomic oxygen on the Pt(111) surface," Surf. Sci. 217, 489 (1989). 52. D. I. Jerdev, J. Kim, M. Batzill, and B. E. Koel, "Evidence for slow oxygen exchange between multiple adsorption s ites at high oxygen coverages on Pt(111)," Surf. Sci. 498, L91 (2002). 53. D. Dahlgren and J. C. Hemminger, "Decomposition of NO2 to NO and O on Pt(111)," Surf. Sci. 123, L739 (1982). 54. N. A. Saliba, Y. L. Tsai, and B. E. Koel, "Oxidation of Pt(111) under UHV conditions using ozone.," Abstr. Pap. Am. Chem. Soc. 213, 393 (1997). 55. N. Saliba, Y. L. Tsai, C. Panja, and B. E. Koel, "Oxidation of Pt(111) by ozone (O3) under UHV conditions," Surf. Sci. 419, 79 (1999). 56. C. R. Parkinson, A. Walker, and C. F. McConville, "Reac tion of atomic oxygen with a Pt(111) surface: chemical and stru ctural determination using XPS, CAICISS and LEED," Surf. Sci. 545, 19 (2003). 57. J. T. Yates, Experimental innovations in surfa ce science: a guide to practical laboratory methods and instruments New York: AIP Press: Springer. xv. (1998). 58. S. Agarwal, G. W. W. Quax, M. C. M. van de Sanden, D. Maroudas, and E. S. Aydil, "Measurement of absolute radical densities in a plasma using modulatedbeam line-of-sight threshold ionization mass spectrometry," J. Vac. Sci. Technol. A 22, 71 (2004). 59. J. F. O'Hanlon, A user's guide to vacuum technology. 2nd ed. New York: Wiley. xvii. (1989). 60. A. Winkler and J. T. Yates, "Capilla ry array dosing and angular desorption distribution measurements a general formalism," J. Vac. Sci. Technol. A 6, 2929 (1988). 61. R. I. Masel, Principles of adsorption and reaction on solid surfaces Wiley series in chemical engineering. New York: Wiley. xiv. (1996). 62. P. A. Redhead, "Thermal desorption of gases," Vacuum 12, 203 (1962). 63. D. Park, Q. Lu, T. J. King, C. Hu, A. Kalnitsky, S. P. Tay, and C. C. Cheng "SiON/Ta2O5/TiN gate-stack transistor with 1.8 nm equivalent SiO2 thickness," C.C. Proc. IEDM Tech. Dig. San Francisco, CA, 6-9 December, 381 (1998).

PAGE 171

159 64. S. I. Raider, R. Flitsch, J. A. Aboaf, and W. A. Pliskin, "Surface oxidation of silicon-nitride films," J. Electrochem. Soc. 123, 560 (1976). 65. S. P. Murarka, C. C. Chang, and A. C. Adams, "Thermal nitridation of silicon in ammonia gas composition and oxidation resistance of the resulting films," J. Electrochem. Soc. 126, 996 (1979). 66. A. E. T. Kuiper, M. F. C. Willemsen, J. M. L. Mulder, J. B. O. Elferink, F. H. P. M. Habraken, and W. F. Vanderweg, "Thermal -oxidation of siliconnitride and silicon oxynitride films," J. Vac. Sci. Technol. B 7, 455 (1989). 67. Y. Hayafuji and K. Kajiwara, "Nitrida tion of silicon and oxidized-silicon," J. Electrochem. Soc. 129, 2102 (1982). 68. R. M. Wallace and Y. Wei, "Dry oxidati on resistance of ultrat hin nitride films: Ordered and amorphous silicon nitride on Si(111)," J. Vac. Sci. Technol. B 17, 970 (1999). 69. X. L. Zhou, C. R. Flores, and J. M. White, "Decomposition of NH3 on Si(100) a SSIMS Study," Surf. Sci. 268, L267 (1992). 70. P. A. Taylor, R. M. Wallace, W. J. Choyke M. J. Dresser, and J. T. Yates, "The dissociative adsorption of ammonia on Si(100)," Surf. Sci. 215, L286 (1989). 71. C. U. S. Larsson and A. S. Flodstrom, "Dissociative NH3 adsorption on the Si(100) 2x1 surface at 300 K," Surf. Sci. 241, 353 (1991). 72. M. J. Dresser, P. A. Taylor, R. M. Wa llace, W. J. Choyke, and J. T. Yates, "The adsorption and decomposition of NH3 on Si(100) detection of the NH2(a) species," Surf. Sci. 218, 75 (1989). 73. P. Avouris, F. Bozso, and R. J. Hamers, "The reaction of Si(1 00) 2x1 with NO and NH3 the role of surf ace dangling bonds," J. Vac. Sci. Technol. B 5, 1387 (1987). 74. C. H. F. Peden, J. W. Rogers, N. D. Shinn, K. B. Kidd, and K. L. Tsang, "Thermally grown Si3N4 thin-films on Si(100) surface and interfacial composition," Phys. Rev. B 47, 15622 (1993). 75. C. U. S. Larsson, C. B. M. Andersson, N. P. Prince, and A. S. Flodstrom, "Si(100) 2x1 the clean and ammonia exposed surface studied with high-resolution corelevel spectroscopy," Surf. Sci. 271, 349 (1992). 76. J. W. Kim, H. W. Yeom, K. J. Kong, B. D. Yu, D. Y. Ahn, Y. D. Chung, C. N. Whang, H. Yi, Y. H. Ha, and D. W. Moon, "Spontaneous N incorporation onto a Si(100) surface," Phys. Rev. Lett. 90, 106101 (2003). 77. J. W. Kim and H. W. Yeom "Thermal decomposition of NH3 on the Si(100) surface," Surf. Sci. 546, L820 (2003).

PAGE 172

160 78. S. Ishidzuka, Y. Igari, T. Takaoka, and I. Kusunoki, "Nitridation of Si(100) surface with NH3," Appl. Surf. Sci. 132, 107 (1998). 79. A. Glachant, D. Saidi, and J. F. Delo rd, "A LEED, AES and TDS study of very thin nitride film growth on Si(100) by direct thermal nitridation in NH3," Surf. Sci. 168, 672 (1986). 80. S. M. Cherif, J. P. Lacharme, and C. A. Sebenne, "Surface-properties of Si(100) 2x1 upon NH3 adsorption and vacuum annealing," Surf. Sci. 262, 33 (1992). 81. D. A. Shirley, "High-resolution X-ray phot oemission spectrum of valence bands of gold," Phys. Rev. B 5, 4709 (1972). 82. T. Seino, D. Muto, T. Matsuura, and J. Murota, "Thermal effects in atomic-order nitridation of Si by a nitrogen plasma," J. Vac. Sci. Technol. B 20, 1431 (2002). 83. S. Miyazaki, M. Narasaki, A. Suyama, M. Yamaoka, and H. Murakami, "Electronic structure and energy band offsets for ultrathin silicon nitride on Si(100)," Appl. Surf. Sci. 216, 252 (2003). 84. V. A. Gritsenko, S. N. Svitasheva, I. P. Petrenko, Y. N. Novikov, Y. N. Morokov, H. Wong, R. W. M. Kwok, a nd R. W. M. Chan, "Characterization of the silicon nitride-thermal oxide inte rface in oxide-nitride-oxid e structures by ELS, XPS, ellipsometry, and numerical simulation," Microelectron. Reliab. 38, 745 (1998). 85. B. Brijs, J. Deleu, T. Conard, H. De Witte, W. Vandervorst, K. Nakajima, K. Kimura, I. Genchev, A. Bergmaier, L. Goergens, P. Neumaier, G. Dollinger, and M. Dobeli, "Characterization of ultra thin oxynitrides: A general approach," Nucl. Instrum. Meth. B 161, 429 (2000). 86. E. K. Hlil, L. Kubler, J. L. Bischo ff, and D. Bolmont, "Photoemission-study of ammonia dissociation on Si(100) below 700 K," Phys. Rev. B 35, 5913 (1987). 87. L. Kubler, J. L. Bischoff, and D.Bo lmont, "General comparison of the surface processes involved in nitridation of Si(100) 2x1 by NH3 and in SiNx film deposition a photoemission study," Phys. Rev. B 38, 13113 (1988). 88. X. Xu, S. Y. Kang, and T. Yamabe, "N itridation of Si(100) 2 x 1 surface by NH3: A quantum chemical cluster model study," Phys. Rev. Lett. 88, 076106 (2002). 89. Y. Widjaja and C. B. Musgrave, "Atomis tic mechanism of the initial oxidation of the clean Si(100) 2x1 surface by O2 and SiO2 decomposition," J. Chem. Phys. 116, 5774 (2002). 90. J. W. Kim, H. W. Yeom, Y. D. Chung, K. Jeong, C. N. Whang, M. K. Lee, and H. J. Shin, "Chemical configuration of nitrog en in ultrathin Si oxynitride on Si(100)," Phys. Rev. B 66, 035312 (2002).

PAGE 173

161 91. M. Bhat, G. W. Yoon, J. Kim, D. L. Kwong, M. Arendt, and J. M. White, "Effects of NH3 nitridation on oxides grown in pure N2O ambient," Appl. Phys. Lett. 64, 2116 (1994). 92. B. A. Ferguson, C. T. Reeves, and C. B. Mullins, "Oxygen adsorption on Si(100) 2x1 via trapping-mediated and direct mechanisms," J. Chem. Phys. 110, 11574 (1999). 93. I. Panas and P. Siegbahn, "A theoretical-s tudy of the peroxo and superoxo forms of molecular oxygen on metal-surfaces," Chem. Phys. Lett. 153, 458 (1988). 94. N. R. Avery, "Electron-energy loss spectroscopic study of CO on Pt(111)," J. Chem. Phys. 74, 4202 (1981). 95. P. D. Nolan, B. R. Lutz, P. L. Tanaka, J. E. Davis, and C. B. Mullins, "Molecularly chemisorbed intermediates to oxygen adso rption on Pt(111): A molecular beam and electron energy-loss spectroscopy study," J. Chem. Phys. 111, 3696 (1999). 96. W. Wurth, J. Stohr, P. Feulner, X. Pan, K. R. Bauchspiess, Y. Baba, E. Hudel, G. Rocker, and D. Menzel, "Bonding, struct ure, and magnetism of physisorbed and chemisorbed O2 on Pt(111)," Phys. Rev. Lett. 65, 2426 (1990). 97. A. Eichler and J. Hafner, "Molecular precu rsors in the dissocia tive adsorption of O2 on Pt(111)," Phys. Rev. Lett. 79, 4481 (1997). 98. A. Eichler, F. Mittendorfer, and J. Hafner, "Precursor-mediated adsorption of oxygen on the (111) surfaces of platinum-group metals," Phys. Rev. B 62, 4744 (2000). 99. K. Yoshimi, M. B. Song, and M. Ito, "Carbon monoxide oxidation on a Pt(111) electrode studied by in situ IRAS and STM: Coadsorption of CO with water on Pt(111)," Surf. Sci. 368, 389 (1996). 100. A. Eichler and J. Hafner, "Reaction cha nnels for the catalytic oxidation of CO on Pt(111)," Phys. Rev. B 59, 5960 (1999). 101. H. Steininger, S. Lehwald, and H. Ibac h, "On the adsorption of CO on Pt(111)," Surf. Sci. 123, 264 (1982). 102. J. C. Tracy and P. W. Palmberg, "Struc tural influences on adsorbate binding energy .I. Carbon monoxide on (100) palladium," J. Chem. Phys. 51, 4852 (1969). 103. W. D. Mieher, L. J. Whitman, and W. Ho, "A time resolved electron energy loss spectroscopy study of CO on Pt (111) adsorption site occupations versus coverage and temperature," J. Chem. Phys. 91, 3228 (1989).

PAGE 174

162 104. M. Kinne, T. Fuhrmann, C. M. Whelan, J. F. Zhu, J. Pantforder, M. Probst, G. Held, R. Denecke, and H. P. Steinruck, "Kinetic parameters of CO adsorbed on Pt(111) studied by in situ high resolu tion X-ray photoelect ron spectroscopy," J. Chem. Phys. 117, 10852 (2002). 105. J. L. Gland and E. B. Kollin, "Vibra tional characterization of carbon monoxide oxidation on the Pt(111) surface," Surf. Sci. 151, 260 (1985). 106. L. Piccolo, C. Becker, and C. R. Henr y, "Reaction between CO and a pre-adsorbed oxygen layer on supported palladium clusters," Appl. Surf. Sci. 164, 156 (2000). 107. G. Zheng and E. I. Altman, "The reactivity of surface oxygen phases on Pd(100) toward reduction by CO," J. Phys. Chem. B 106, 1048 (2002). 108. T. Dellwig, J. Hartmann, J. Libuda, I. Meus el, G. Rupprechter, H. Unterhalt, and H. J. Freund, "Complex model catalysts unde r UHV and high pressure conditions: CO adsorption and oxidation on alumina-supported Pd particles," J. Mol. Catal. A 162, 51 (2000). 109. I. Stara, V. Nehasil, and V. Matolin, "I nfluence of substrate st ructure on activity of alumina supported Pd particles: CO adsorption and oxidation," Surf. Sci. 365, 69 (1996). 110. V. Matolin and I. Stara, "CO diffusi on over the alumina suppor t of Pd particle model catalysts," Surf. Sci. 398, 117 (1998).

PAGE 175

163 BIOGRAPHICAL SKETCH Alex Gerrard was born on February 27, 1977 in Portsmouth, NH to Alex and Mary Gerrard. When he was 4 years old, his fath er was assigned to work at Elmendorf Air Force Base in Anchorage, AK. This is wher e Alex grew into adu lthood with his brother Aaron and two sisters Anne and Allanah. While attending Service High School, Alex had the opportunity to participate as a fore ign exchange student to Magadan, Russia and Shari-cho, Japan. His college level studies began after gr aduating from high school in May 1995. In May 1999, he earned a B.S. degree in Chemical Engineering from the University of Missouri-Rolla. Alex enrolled in the Ph.D. program at th e University of Florida in August 1999, where he joined Assistant Professo r Jason Weaver’s group. As one of Dr. Weaver’s first students, Alex’s first task was to set up a Surface Science laboratory in which to conduct his research. Along with his academic studies, Alex has spent a great deal of time working on self development and studying human interact ions. These studies have comprised of abstract areas such as theology and philosophy to practical subjects including leadership and conflict resolution. One of Alex’s strongest personal co nvictions is to support the community. While residing in Gainesville, he has volunteered his time visiting the elderly at nursing homes and assisted at a lo cal church. In his spare time, Alex enjoys spending his time traveli ng, reading, and baking.


Permanent Link: http://ufdc.ufl.edu/UFE0010104/00001

Material Information

Title: Interactions of Atomic Oxygen with Pt(111) and Nitrided Si(100)
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0010104:00001

Permanent Link: http://ufdc.ufl.edu/UFE0010104/00001

Material Information

Title: Interactions of Atomic Oxygen with Pt(111) and Nitrided Si(100)
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0010104:00001


This item has the following downloads:


Full Text












INTERACTIONS OF ATOMIC OXYGEN WITH Pt(1 11) AND NITRIDED Si(100)


By

ALEX GERRARD

















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

UNIVERSITY OF FLORIDA


2005

































Copyright 2005

by

Alex Gerrard

































This dissertation is dedicated to my late grandfather, Delbert Barrow.















ACKNOWLEDGMENTS

I would first like to acknowledge my supervisory committee chair Dr. Jason

Weaver for his guidance in the field of surface science. Working with him and building

his laboratory has truly been a rich experience. I would also like to thank Dr. Gar

Hoflund for his advice since I first arrived at the University of Florida. He has always

been generous with his time, expertise and optimism. I am grateful for all of his support.

During my tenure at UF, I had the privilege or working under Dr. Crisalle and Dr.

Chauhan as a teaching assistant. They taught me a great deal about teaching and

effectively communicating complicated ideas to students. I am also indebted to James

Hinant, Dennis Vince, Dr. Bosworth, Dr. Ren, Dr. Butler, Dr. Orazem, Dr. Anderson, and

the rest of the faculty and staff of the Chemical Engineering Department for their all of

their assistance and support.

I would also like to show my gratitude to my spiritual director, Fr. Ed. Murphy and

my family, Alex, Mary, Anne, Allanah, and Aaron Gerrard for their guidance, kindness

and encouragement. I am grateful to have made many friends and colleagues in

Gainesville who have assisted me in virtually every aspect of my life, to many of whom

to mention here. One particular soul who does deserve a great amount of recognition is

Jau-Jiun Chen. I greatly appreciate all of her dedicated hard work and her company. I

truly treasure the relationship I have with her.
















TABLE OF CONTENTS

page

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

LIST OF TABLES ....................................................... ............ .............. .. vii

LIST OF FIGURES ..................................................... .......... ................ viii

ABSTRACT ........ .............. ............. ...... ...................... xi

CHAPTER

1 IN TR O D U C TIO N ........................ .... ........................ ........ ..... ................

1.1 R research Objective .................................. ...................................... 1
1.2 Literature Search .............................................. .... ..... ........ .. ..
1.2.1 R action M echanism s .................. ......... ............................. ... 4
1.2.2 Molecular and Atomic Oxygen Interactions with Si(100).............7
1.2.3 Molecular and Atomic Oxygen Interactions with Pt(111) ...........8
1.3 Experim ental System ............................................................................9
1.3.1 The U H V C ham ber...................................... ........................ 9
1.3.2 Sam ple M anipulator.................................... ........ ............... 10
1.3.3 Sample Temperature Control............... ....................................... 11
1.3.4 B eam C ham ber......................................... .......................... 12
1.3.5 Plasma Characterization................... ..... ...................... 13
1.3.6 G as H handling ............................. ................... .. .......... 17
1.3.7 Calibrated M olecular B eam ................................... ... ..................17
1.3.8 Calibration of 02 Beam Flow from Plasma Source....................22
1.4 Detection Techniques................ ............................... 22
1.4.1 Reaction Product M onitoring................................. ... ................ 22
1.4.2 Temperature Programmed Desorption and Reaction...................24
1.4.3 X-ray Photoelectron Spectroscopy .........................................25

2 OXIDATION OF NITRIDED Si(100) BY GASEOUS ATOMIC AND
M OLECULAR OX Y GEN .......................................................... ............... 42

2.1 Introduction .. ............................................................. ............... 43
2.2 Experim ental M ethods ............................................................ ........ 47









2 .3 R results .................................................................................... 50
2.3.1 NH3 Decomposition on Si(100) ...............................................50
2.3.2 Oxidation of Nitrided Si(100) at 300 K ............................ ......... 54
2.3.3 Oxidation of Nitrided Si(100) at 800 K ............................ .........59
2.4 D discussion .......... ......... ... ......... ................ ... ............. 64
2.5 Conclusions .............. ...... ............................... 70

3 DYNAMIC DISPLACEMENT AND DISSOCIATION OF 02 ON Pt(1 11) BY
A T O M IC O X Y G E N ...................................................................... .................. 86

3.1 Introduction ........... ...................................... ................ .... ......... 86
3.2 Experim ental M ethods .................................................... ...... ......... 88
3.3 Results ............... .......... ................................. ......... 89
3.4 D discussion ................................................................ .. ........ .... 98
3.5 C onclu sions ............................................................. ........... .. 104

4 CARBON MONOXIDE OXIDATION FROM HIGH OXYGEN COVERAGE
PH A SES ON Pt(111).............................. ....................... ....... ............... 109

4.1 Introduction ................. .............. ............... 110
4.2 Experim mental M methods ................................................... .......... 113
4.3 Results ................ ......... ........... ....... ............... 115
4.3.1 Carbon Monoxide Adsorption and TPR on O Precovered
Pt( 111) Surfaces ................... ... ................. .... .... .. .......... ...... 115
4.3.2 CO Oxidation on High-Coverages of Chemisorbed Oxygen
Atoms Under Isothermal Conditions........................................ 123
4.3.3 Carbon Monoxide Oxidation from Platinum Oxide ..................129
4.4 D discussion ................................................. ............................. 135
4.5 Conclusions .......................................... ............. ............... 138

5 CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK.....................150

5 .1 S y n o p sis ................................150.............................
5.2 Future W ork ............... ...... .. ........ .. ............. ........ ... .. .......... 151
5.2.1 Nonthermal Reaction Mechanisms .....................................151
5.2.2 Scanning Tunneling Microscopy Measurements......................152
5.2.3 M echanistic Catalytic Studies................................. ... ......152

L IST O F R EFER EN CE S ......... ................... ........................................ ............... 154

BIOGRAPH ICAL SKETCH .............. ........................ .................... ............... 163















LIST OF TABLES


Table page

2-1 Oxygen coverages on clean and nitrided Si(100)............................... ...............73

3-1 Displacement and desorption yields of 102 when exposed to 160-atoms while
holding the temperature at 90, 100 and 110 K .......................................................105

3-2 Initial evolution rate, the estimated initial dissociation rate, and the sum of the
two rates when exposing an 102 saturated on Pt(1 11) to a beam with a flux of
0.005 M L/sec of 160-atom s. ....... ................. ............... ......... ........ .. 105

4-1 Maximum CO coverage achieved as a function of initial o. ...........................140
















LIST OF FIGURES


Figure page

1-1 Mechanisms describing surface chemical reactions. ...........................................26

1-2 UHV system constructed for our study....................................... ............... 26

1-3 Top view of the U H V system .............................. ...................... ............... 27

1-4 Specimen mount on the sample manipulator. ................. .............................. 28

1-5 Sample temperature control scheme used for these experiments ........................29

1-6 Typical linear temperature ramp with a heating rate of 1 K/sec..........................30

1-7 Beam chamber housing the plasma source ...............................................31

1-8 Appearance potential measurement, monitoring O+ (m/e 16) with the plasma
activated and deactivated with a first stage beam chamber pressure of 3.0*10-5
T o rr .............................................................................. 3 2

1-9 Mass spectrometer beam trace experiment............................... ................33

1-10 Oxygen uptake on Pt( 11) with a surface temperature of 450 K ..........................34

1-11 Gas flow from the beam onto the sample ................................... .................35

1-12 Hypothetical beam trace depicting the partial pressure of a gas using a
calibrated molecular beam, where A is proportional to the quantity of gas
adsorbed on the sam ple ........... ............................. ................ ............... 36

1-13 M olecular beam doser used for our experiments................................................ 37

1-14 Pressure of carbon monoxide in the gas reservoir plotted as a function of time
while pumping out through the laser drilled VCR Gasket.................................38

1-15 Calibration of the molecular oxygen flow rate through the beam chamber is
compared with the pressure in the first differential pumping stage....................39

1-16 Dynamic displacement of 1802 taken upon exposing a Pt(1 11) surface
saturated with 102 to 160-atoms, while holding the sample temperature
constant at 90 K .....................................................................40









1-17 Temperature programmed desorption performed on Pt(1 11) after exposing to
a beam of oxygen atom s .............................................. .............................. 41

2-1 Si2p spectrum obtained from a Si(100) surface after a 160 ML NH3 exposure
at a surface temperature of 900 K. ............................................... ............... 73

2-2 Nis spectrum obtained from a Si(100) surface after a 160 ML NH3 exposure
at a surface temperature of 900 K. ............................................... ............... 74

2-3 NIs spectra obtained from Si(100) ........................................................75

2-4 Si2p spectra obtained after exposing clean Si(100) to 160 ML NH3 at a
surface temperature of 900 K (dashed line), and after depositing 1.2 ML of
oxygen on the nitrided Si(100) surface held at 300 K using the plasma-
activated beam (solid line). ........................................ .......................................76

2-5 NIs spectra obtained after exposing Si(100) to 160 ML NH3 at a surface
temperature of 900 K (dashed line), and after depositing 1.2 ML of oxygen on
the nitrided Si(100) surface held at 300 K using the plasma-activated beam
(so lid lin e). ....................................................... ................ 7 7

2-6 Ols spectra obtained after incorporating 1 ML of oxygen atoms on Si(100) at
a surface temperature of 300 K (dashed line), and after depositing 1.2 ML of
oxygen on the nitrided Si(100) surface held at 300 K using the plasma-
activated beam (solid line) ......................................................... ............... 78

2-7 NIs and Ols spectra obtained at electron collection angles of 00 (dashed line)
and 600 (solid line) after depositing 1.2 ML of oxygen on nitrided Si(100) at a
surface tem perature of 300 K .......................................... ........................... 79

2-8 Si2p spectra obtained after exposing clean Si(100) to 160 ML NH3 at a
surface temperature of 900 K (dashed line), and after depositing 1.5 ML of
oxygen on the nitrided Si(100) surface held at 800 K using the plasma-
activated beam (solid line). ........................................ .......................................80

2-9 NIs spectra obtained after exposing Si(100) to 160 ML NH3 at a surface
temperature of 900 K (dashed line), and after depositing 1.5 ML of oxygen on
the nitrided Si(100) surface held at 800 K using the plasma-activated beam
(so lid lin e ) .............................................................................................................8 1

2-10 Ols spectra obtained after incorporating 2.4 ML of oxygen atoms on Si(100)
at a surface temperature of 800 K (dashed line), and after depositing 1.5 ML
of oxygen on the nitrided Si(100) surface held at 800 K using the plasma-
activated beam (solid line). ........................................ .......................................82

2-11 NIs and Ols spectra obtained at electron collection angles of 00 (dashed line)
and 600 (solid line) after depositing 1.5 ML of oxygen on nitrided Si(100) at a
surface tem perature of 800 K ....... ....... ..... ..................... ..... .......... 83









2-12 Model for 02 dissociation and incorporation into Si(100)............................... 84

2-13 Model for O-atom adsorption and incorporation into Si(100) ...............................85

3-1 Partial pressure traces of 1802 desorbing from Pt( 11) when exposed to 160-
atom s ..................................... .................. ................ ..........106

3-2 Temperature programmed desorption spectra taken after recovering the
Pt(1 11) surface with 1O2 and exposing this surface to 160-atoms......................107

3-3 Normalized dynamic displacement beam traces of 1802 when exposed to a
160-atom beam interrupted every thirty seconds .............................................. 108

4-1 Temperature programmed desorption spectra taken from the Pt(1 11) surface
w ith a ram p rate of 3 K /sec ......... ............. ................................ ............... 140

4-2 Temperature programmed reaction spectra taken with a ramp rate of 1 K/sec,
showing 02, CO and CO2 desorption after preparing a saturation coverage of
CO on Pt( 11) with 0.39 ML of O-atoms at Ts = 100 K ...................................141

4-3 Temperature programmed desorption and reaction spectra taken from the
Pt( 111) surface with a ramp rate of 3 K/sec ............................. ............... 142

4-4 Carbon dioxide rate curves with various initial oxygen coverages at 300 K
plotted against C O exposure .......................................................................... 143

4-5 Carbon dioxide rate curves with various initial oxygen coverages at 400 K
plotted against C O exposure .......................................................................... 144

4-6 Carbon dioxide rate curves with various initial oxygen coverages at 500 K,
plotted against CO exposure. ......................................................................... 145

4-7 Carbon dioxide formation rate plotted as a function of CO exposure with an
initial O-atom coverage of 1.7 ML between 400 and 550 K ..............................146

4-8 Carbon dioxide formation rate plotted as a function of CO exposure with
various initial O-atom coverages at 400 K. ................................. ............... 147

4-9 TPD traces taken with an initial coverage of 1.7 ML and with the same
coverage after a 20 ML CO exposure (-0.5 ML) at 500 K ..............................148

4-10 LEED images taken after reducing platinum oxide with CO ...........................149















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

INTERACTIONS OF ATOMIC OXYGEN WITH Pt(1 11) AND NITRIDED Si(100)


By

Alex Gerrard

May 2005

Chair: Jason Weaver
Major Department: Chemical Engineering

Gas-phase oxygen atoms play a critical role in several applications including

degradation of materials in low Earth orbits, semiconductor processing, and

heterogeneous catalysis. In each of these applications, the highly reactive nature of the

oxygen atom dictates the details of the oxygen atom-surface reactions. Whether the goal

is to develop new materials resistant to oxidation or to use the oxygen atoms to generate

surfaces with unique properties, a fundamental understanding is needed of the chemistry

governing the interactions of O-atoms with surfaces.

Silicon nitride is used in applications because it resists oxidation. We conducted an

X-ray photoelectron spectroscopy (XPS) study to examine the surface chemistry of

nitrided Si(100) toward molecular and atomic oxygen. A decrease was observed in the

Si(100) surface dangling bond density due to nitridation, and this was accompanied by a

subsequent decrease in surface reactivity for both molecular and atomic oxygen. This









indicates that oxygen atoms preferentially react at these dangling bond sites, and do not

insert directly into silicon-silicon bonds.

Using mass spectrometry, we explored the reactivity of gaseous 160-atoms toward

chemisorbed 1802 on the Pt(1 11) surface. The gas-phase oxygen atoms stimulate both

displacement and dissociation of adsorbed 1802. As the surface temperature increased,

the desorption yield decreased, with the balance remaining on the surface in the form of

chemisorbed atoms. Molecular oxygen is known to adsorb in superoxo and peroxo

configurations on Pt( 11). The strong temperature dependence on the dissociation yield

is attributed to an increase in the population of the peroxo chemisorbed species with

temperature, which is more prone to dissociate.

Platinum is catalytically active toward the oxidation of CO. Using 02 under

ultrahigh vacuum conditions, a maximum surface coverage of 0.25 monolayers (ML) of

O-atoms can be generated. Surfaces with higher oxygen coverages can be generated by

exposing Pt( 11) to oxygen atoms. Using mass spectrometric methods, an in situ CO

oxidation study was conducted on these surfaces. The CO oxidation kinetic behavior was

found to be consistent with a CO adsorption precursor model. Carbon monoxide

oxidation on platinum oxide occurs predominately at the interface between the metallic

regions and the oxide.














CHAPTER 1
INTRODUCTION

1.1 Research Objective

Our main objective was to develop a fundamental understanding of interactions

between oxygen atoms and technologically relevant surfaces. Oxygen atoms are present

in semiconductor plasma processes and in low Earth orbits. These radicals are known to

be significantly more reactive than their molecular counterpart. It is believed that O-atom

chemistry plays an important role in semiconductor processes, polymer erosion, and low

pressure metal oxide formation. Conducting well characterized O-atom beam

experiments on metal and semiconductor surfaces provides insight into the underlying

chemistry in these areas.

Silicon nitride and oxynitride films have been studied extensively in recent years

because of the advantages afforded by incorporating these materials in the dielectric

layers used in metal-oxide-semiconductor (MOS) devices. Adding small amounts of

nitrogen to the SiO2-Si interface is known to improve the structural quality of the

interface, and results in lower leakage current across the gate of a MOS device as well as

enhanced resistance to boron diffusion into the SiO2 [1]. The low activity of silicon

nitride toward oxidation has also proved beneficial to the growth of alternative gate

oxides such as Ta205 and ZrO2 that have higher dielectric constants (k) than SiO2.

Recent studies show that depositing high k oxides directly onto silicon can result in

formation of an SiO2 layer that dominates the capacitance of the gate stack [2-6].

Incorporating nitrogen in the near-surface region of silicon alleviates this problem by









inhibiting SiO2 formation during the deposition of Ta20O, thereby maximizing the

benefits of the high k oxide film [3,7,8]. Another important application of silicon nitride

is as a protective coating on ceramic components such as bearings and turbine blades for

which tolerance to high temperature oxidizing environments is critical. Despite these

important applications, however, the oxidation of silicon nitride films is not well

understood at the molecular level. We studied the oxidation of a silicon nitride film by

both gaseous atomic and molecular oxygen in ultrahigh vacuum (UHV), focusing our

efforts on elucidating the fundamental origin of the oxidation resistance of nitrided

silicon. X-ray photoelectron spectroscopy (XPS) was conducted before and after

thermally decomposing nitrogen on the Si(100) surface. The nitrided surface was then

exposed to a beam of molecular or atomic oxygen. Changes in the surface due to the

oxidant beam were characterized by XPS.

Platinum is considered an important oxidation catalyst in several oxidation

reactions. One of the first steps in understanding heterogeneous catalysis on platinum is

to learn about the molecular level oxygen interactions with this metal. The Pt(1 11)

crystal face was selected for these studies because of the vast literature characterizing the

interactions of molecular oxygen with this surface [9-15]. Molecular oxygen is known to

physisorb onto Pt( 11) at 20 K. When the surface is heated to 38 K, the molecular

oxygen becomes chemisorbed, and generates a saturation coverage of 0.44 monolayers

(ML) [11,14]. On heating the sample to 130 K, a competitive process occurs between

dissociation and desorption of molecular oxygen, yielding a surface saturated with O-

atoms at a coverage of 0.25 ML. The oxygen atoms become mobile on the surface at

about 200 K, forming islands with a p(2x2) structure [14]. On continued heating, the









atoms recombine and desorb at about 710 K. The atomic oxygen coverage does not

increase beyond 0.25 ML due to the limiting 02 flux imposed under vacuum conditions.

Our study used a beam consisting of atomic and molecular oxygen, and it is useful to

characterize the surface behaviors of O-atom radicals with molecular oxygen on Pt(l 11).

Oxygen atom impingement onto a Pt(1 11) surface saturated with molecular oxygen

has been investigated previously [16,17]. Using isotopically labeled 1802 adsorbed on the

surface and gas-phase 160-atoms, the displacement of 1802 from the surface and the

formation of 180160 was observed at a surface temperature of 80 K. The displacement

phenomenon was also observed when directing nitrogen and hydrogen atoms at the 1802

covered surface. Employing time-of-flight mass spectrometry, the desorbing molecules

were observed to have a bimodal energy distribution [17]. This was found to be

independent of the adsorbing atom. The lower energy desorption component is

consistent with thermal evolution of 1802 from the surface. The high-energy desorption

feature is direct evidence that some energy from the adsorbing atom is transferred

directly into the 1802 before it desorbs. By probing the role of surface temperature,

deeper insights into the nature of these phenomena may be found. Intuitively, one would

anticipate an increase in the surface temperature would amplify the thermal desorption

rate. In our study, we measured molecular oxygen displacement by impingent oxygen

atoms as a function of the surface temperature. We also characterized oxygen uptake

caused by impinging oxygen atoms. The adsorbed 1802 was not found to be displaced by

gas-phase 1602, and hence no 1602 adsorbed onto the surface. Any 160-atoms found on

the surface must, therefore, originate from the 160-atoms in the beam.









Platinum is considered an active oxidation catalyst in many areas [18-20]. Along

with probing oxygen atom-adsorbate reactivity, oxygen atoms can generate high

coverages of adsorbed oxygen atoms on Pt( 11), previously unattainable using

molecular oxygen in UHV (0.25 ML). Carbon monoxide oxidation on Pt(1 11) has been

studied in detail with oxygen coverages (0o) on the order of 0.25 ML [8,21-40]. This

makes CO an excellent probe molecule for characterizing the relative reactivity of the

high-coverage phases of oxygen atoms on Pt(l 11). Interestingly, this also presents an

opportunity to perform molecular-level experiments on catalytic surfaces that can be

generated under high-pressure reaction systems. In the experiments presented in chapter

four, surfaces with high oxygen atom coverages were generated by exposing clean

Pt( 11) to an oxygen atom beam. This surface was then exposed to a CO beam while

holding the surface temperature constant and monitoring the background partial pressure

of CO2 using a mass spectrometer. Temperature programmed desorption spectroscopy

(TPD) was performed after each beam experiment to quantify the products remaining on

the surface.

Next, I discuss gas-surface reaction mechanisms and the nature of O-atom

interactions with Si(100) and Pt( 11). To conduct such experiments, a UHV chamber

was constructed and equipped with all of the equipment necessary to conduct the

spectroscopic measurements described above. The system and its capabilities are

described in Section 1.3.

1.2 Literature Search

1.2.1 Reaction Mechanisms

The studies presented here entail characterizing the reactivity of oxygen atoms with

surfaces and characterizing the surface changes caused by the gas phase O-atom. The









free radical nature of the oxygen atom suggests that any oxygen atom-adsorbate

interactions will rapidly occur and should not be dependent on surface temperature.

Characterizing these surfaces with a probe molecule (such as CO), TPD, or temperature

programmed reaction (TPR) requires a sound understanding of thermal reaction

chemistry occurring on solid surfaces. Three distinct mechanisms describing surface

reactions are discussed here.

A classical description of a thermal heterogeneous catalytic reaction is the

Langmuir-Hinshelwood (LH) mechanism. To show how this mechanism works, consider

the surface reaction A+B--AB. First, both species A and B thermally accommodate to

the surface. At sufficiently high surface temperatures, these species become mobile and

diffuse across the surface. Once the reactants find each other and if they have enough

energy to overcome the activation energy for reaction, they will react to form AB.

Finally, AB desorbs from the surface (Figure 1-1).

Gas-phase oxygen atoms are highly reactive, and any interaction between this

species and adsorbates or surfaces should not depend on surface temperature. It is then

useful to understand fundamental nonthermal reaction mechanisms; the simplest is the

Eley-Rideal (ER) mechanism. This mechanism is shown in the following reaction steps.

First, one of the components (A) thermally accommodates with the surface. An

oncoming B particle collides with A in a single collision and they react to form AB. A

portion of the reaction energy between A and B is transferred into the kinetic energy of

the product molecule, allowing AB to immediately desorb (Figure 1-1). This energy

could be transferred into the nuclear motion of the products (the translational, vibrational,

and rotational modes), which allows the reaction products to desorb in a directed manner









[41]. Note that this reaction has a very small cross-section, because B must strike A in

the first collision, otherwise the B would either back scatter or stick to the surface.

Alternatively, a nonthermal reaction may occur via the hot atom (HA) mechanism.

The HA mechanism is very similar to the ER mechanism, and proceeds in the following

manner. First, A thermally accommodates to the surface. Particle B then collides with

the surface and becomes trapped, but not thermally accommodated to the surface. After

several B-surface collisions, the B then collides with A and reacts to form AB. The

product AB then immediately desorbs from the surface (Figure 1-1). Since the B particle

is trapped at the surface, it may collide frequently with the surface, increasing the

probability that B will react with A. This causes the cross-section for the reaction by the

HA mechanism to be higher than that of the ER mechanism.

The ER and HA mechanisms have distinct kinetic features, that are observed as the

reaction products are monitored. Using oxygen atoms as the highly reactive "B" species,

the ER mechanism should have a reaction cross-section similar to that of the atomic

dimensions (10-16 cm2). The rate for the ER mechanism is given by the expression

R=0o(Y, where D is the flux of gaseous oxygen atoms at the surface, C is the reaction

cross-section, and 0 is the adsorbate surface concentration. In this mechanism, the

product formation rate would first begin at the maximum, and then decay exponentially

to the baseline. Through quantitative analysis of the rate data, the HA mechanism can be

distinguished from the ER mechanism in the following ways. The HA mechanism may

have a substantially larger cross-section because of the multiple atom-surface collisions

that occur before the reaction [41,42]. This also generates a surface concentration of

these hot atoms on the substrate, which can influence the kinetics. For example, if sites









are available on the surface before the reaction, the probability of an O-atom sticking in

an available site will compete with the probability of reacting with an adsorbed product

A. This will yield a time delay before the rate maximum, which has been observed when

H-atoms were directed toward surfaces coated with D-atoms [42,43]. Additionally,

unanticipated products may form when the HA mechanism dominates the nonthermal

chemistry. While directing hydrogen atoms onto deuterium-covered metal surfaces,

Kammler et al. [42,43] observed the formation of HD and D2. The homo-isotopic

product was attributed to the generation of secondary hot D-atoms. Secondary hot atom

formation has been observed in a number of studies for gaseous H-atoms reacting with

adsorbates [41,44-47] and can be detected by isotopically labeling the adsorbate

molecule.

1.2.2 Molecular and Atomic Oxygen Interactions with Si(100)

Although UHV studies on the oxidation of silicon nitride films are sparse,

oxidation of single crystal silicon surfaces has been studied extensively. Of particular

relevance to our study are detailed UHV studies by Engstrom et al. [48] on the oxidation

Si(100) and Si(11) by both gaseous atomic and molecular oxygen. These and other

results were discussed by Engel [49] in a review of Si oxidation. Briefly, under UHV

conditions, the dissociative adsorption of 02 on Si(100) yields effective saturation

coverage of only about 1 ML of oxygen atoms when the surface is held at 300 K during

oxidation. The saturation coverage can be increased by oxidizing in UHV at elevated

surface temperature, but the oxygen uptake is still rather limited. For example, the

saturation oxygen coverage increases to 2 ML when Si(100) is exposed to 02 at a surface

temperature of 800 K. Not surprisingly, Engstrom et al. [48] found that gaseous oxygen

atoms adsorb on Si(100) with much higher probability than does 02, and that oxygen









coverages greater than 10 ML can be obtained by oxidizing Si(100) held at 300 K using

an atomic oxygen beam. Unlike results obtained using 02, the uptake of oxygen atoms

was insensitive to the surface temperature for oxygen coverages up to about 5 ML, which

indicates non-activated adsorption and possibly direct insertion of gaseous O-atoms into

surface Si-Si bonds.

1.2.3 Molecular and Atomic Oxygen Interactions with Pt(lll)

Molecular oxygen interactions with Pt(1 11) have been studied extensively, and at

100 K 02 is known to chemisorb onto the Pt(1 11) surface, generating a saturation

coverage of 0.44 ML [11,14]. As the surface is heated to 140 K, a competitive process

between desorption and dissociation occurs. What remains after this process is a

platinum surface saturated with oxygen atoms (0.25 ML). These oxygen atoms become

mobile at about 200 K, and organize into islands with a p(2x2) structure, with the O-

atoms residing in the FCC hollow site. When the sample temperature reaches -710 K,

the adsorbed oxygen atoms recombine and desorb [9-15]. Higher surface coverages of

oxygen atoms have been formed by electron dissociation of 02/Pt(l 11) [9,50], and

exposing Pt(1 11) to NO2 [51-53], 03 [54,55], and gaseous O-atoms [56]. Our research

group successfully generated oxygen-atom coverages up to 2.9 ML on Pt(1 11) with a

beam consisting of molecular and atomic oxygen. Oxygen atom uptake and the resulting

high O-atom coverage phases on Pt( 11) were examined using X-ray photoelectron

spectroscopy (XPS), electron energy loss spectroscopy (ELS), low energy electron

diffraction (LEED), and TPD.

As the surface coverage of oxygen on the Pt(1 11) surface exceeds 0.25 ML, the

onset of 02 desorption shifts to lower temperatures during TPD [51,55]. Two distinct









low-temperature desorption features were observed at about 550 and 640 K. Parker et al.

[51] attributed this to strong repulsive interactions among neighboring oxygen atoms as

well as O-atoms adsorbed in weakly bound hcp hollow sites. As the coverage increases

beyond 0.75 ML, the intensity of the low temperature desorption features decrease.

Accompanying this observation is the formation of a single sharp desorption peak that

shifts to higher temperatures with increasing 0o. Using XPS and ELS measurements in

conjunction with TPD in our laboratory, this sharp desorption feature is attributed to the

presence of oxide islands forming on the surface.

1.3 Experimental System

Probing the oxygen atom-surface reactivity at the molecular level required the

construction of an ultrahigh vacuum chamber with the capability of monitoring gas phase

species and surface properties. A calibrated beam system was designed to create uniform

adlayers on the surface and to conduct isothermal reaction rate measurements. To

generate the atomic oxygen beams used in these studies a microwave plasma source was

employed. The plasma source was inserted into a chamber with two stages of differential

pumping to maintain UHV during the plasma operation. Oxygen atoms can react with

adsorbed species both on the sample and the chamber walls, which could influence the

measurements during the beam experiments. This was addressed by designing a set of

collimating apertures for the differentially pumped beam chamber, which minimized the

pressure rise in the UHV chamber and focused the beam onto the sample.

1.3.1 The UHV Chamber

The experiments were conducted in a three-level UHV chamber (Figure 1-2) that

reaches a base pressure less than 2 x 10-10 Torr. The chamber is evacuated by an ion

pump (400 L/sec), a turbo molecular pump (210 L/sec), and a titanium sublimation pump









that is inserted into a liquid nitrogen cooled cryoshield. The upper level of the chamber

(Figure 1-3) houses a hemispherical analyzer (Specs EA10 plus), a dual Al/Mg anode X-

ray source (Specs XR-50), a variable-energy electron source (Specs EQ 22/35) and an ion

sputter source (Specs IQE 11), for performing XPS, Auger electron spectroscopy (AES),

ELS and low energy ion scattering spectroscopy (LEISS) as well as surface cleaning by

ion sputtering. The middle level of the chamber is designed primarily for gas-dosing, and

contains a directed doser connected to a leak valve as well as a calibrated molecular beam

doser, the design of which closely follows that described by Yates [57]. The lower level

of the chamber (Figure 1-3) houses low-energy electron diffraction (LEED) optics, a two-

stage differentially-pumped plasma beam source, and a quadrupole mass spectrometer

(QMS).

1.3.2 Sample Manipulator

A custom-designed sample manipulator (McAllister Technical Services) was

mounted to the top of the UHV chamber. Sample motion in all three Cartesian directions

and rotation by 360 about the vertical axis were accomplished using an XY translator, a

single-axis translator and a rotary platform. The specimen holder mounts to a copper

reservoir that is brazed to the bottom of a stainless steel tube. The top end of the tube

mounts onto the rotary platform via a flange with multiple feedthroughs for attaching

thermocouples and power wires to the sample. A flat copper plate protrudes from the

bottom of the copper reservoir and L-shaped copper legs are bolted on each side of the

plate, with sapphire spacers providing electrical isolation between the reservoir and the

copper legs (Figure 1-4). The copper legs are each 0.25" wide, extend 2" below the

bottom of the copper plate, and are separated from one another by a distance of 1.12". A

specimen is mounted to the sample holder either by directly clipping each side of the









specimen to Ta plates that are bolted to the front of the flat copper legs, or by spot-

welding wires to both the specimen and the Ta plates. The Ta plates are in thermal

contact with the copper reservoir and can be cooled to temperatures as low as 80 K when

the reservoir is filled with liquid nitrogen.

1.3.3 Sample Temperature Control

Accurate temperature control is required during the experiments and to conduct

TPD measurements. The following PID control scheme was used for this purpose and is

depicted in Figure 1-5. Sample heating is performed resistively by passing current

through the sample and the surface temperature is measured with a K-type thermocouple

attached to the back of the specimen. The heating current is generated by a Sorenson

DCS 33-33 DC power supply, and the sample temperature is regulated using an Omega

cn 1166 PID controller. This particular controller has a recorder and controller output.

The controller output is wired into the analog control input on the Sorenson power

supply. A computer records the output utilizing the analog input port located on the mass

spectrometer.

One requirement for analyzing TPD data accurately is the ability to generate a

linear temperature ramp. Typically, the heating ramp consists of two components: an

initial nonlinear transient component, followed by a linear ramp. The temperature

controller must be tuned to maximize the linear region of the ramp and minimize the

duration of the transient features. Figure 1-6 shows a typical heating curve for a 1 K/sec

ramp rate.

A concern with this heating configuration is that the specimen mount may be

damaged when applying current to the sample. This can be minimized if the following









suggestions are followed. Occasionally, the power supply or controller will malfunction

and send a power spike across the sample, when initially activated. To avoid this, open

the switch in the high current circuit (Figure 1-5) before activating the power supply or

controller. When manually changing the set-point, it is safe to rapidly decrease the

temperature. However, if the set-point is rapidly increased, even if the actual temperature

is 100 C higher than the set-point, the controller may deliver enough current across the

sample to damage the mount. If a temperature ramp is desired, it is imperative to initiate

the ramp while the set-point is within 10 C of the measured temperature.

1.3.4 Beam Chamber

A two-stage differentially-pumped chamber containing a commercial plasma

source (Oxford Scientific Instruments) is attached to the UHV chamber and was used to

maintain ultrahigh vacuum conditions in the main chamber while operating the plasma

source (Figure 1-7). Gaseous oxygen atoms are produced in this system by dissociating

02 in a plasma that is confined to a small reservoir at the end of the plasma source. The

plasma source operates at a microwave frequency (2.45 GHz) and employs electron

cyclotron resonance to increase the plasma density. The plasma reservoir is fabricated

from high purity alumina to minimize atom recombination and is terminated by a 2 mm

thick alumina plate that has five thru holes, each of 0.4 mm diameter, that are arranged in

a centered-(2x2) pattern within a 2 mm area. Species exit the plasma volume through

these holes, and form a beam that is directed toward the sample surface held in the UHV

chamber.

In the first pumping stage, the beam passes between oppositely charged parallel

plates (+10 kV/cm) that deflect ions and electrons from the beam path. After flowing

through a conical skimmer (4 = 3 mm) separating the first and second pumping stages,









the gas travels down a quartz tube before entering the UHV chamber. The quartz tube is

60 mm long and has an inner diameter of 6 mm. The quartz tube provides a lower

conductance between the source and UHV chambers than would be obtained with a thin-

walled orifice of the same diameter. The quartz tube also provides sufficient collimation

to confine the beam to the sample surface, which facilitates reactive scattering

measurements using mass spectrometry. In addition, collisions at the inner walls of the

tube are expected to reduce the fraction of atoms and molecules in vibrationally and

electronically excited states, which should result in beams containing primarily ground

state species, specifically 0(3P) and 02(3g-). The first pumping stage of the beam

chamber is evacuated with a 1200 L/sec diffusion pump (Varian VHS 4) and the second

stage is evacuated with a 70 L/sec turbo molecular pump and a titanium sublimation

pump mounted inside a liquid-nitrogen cooled cryoshield. A mechanical shutter is

located in the first pumping stage, which enables control over beam introduction into the

main UHV chamber.

1.3.5 Plasma Characterization

The oxygen beam generated using the plasma source is comprised primarily of

molecular and atomic oxygen. Before investigating the surface chemistry of gas-phase

oxygen atoms, it is imperative to quantify the O-atom concentration in the beam. Several

plasma radical quantification tools including ultraviolet, visible, vibrational, and

ionization spectroscopies have successfully been employed to quantify the radical density

in plasmas. The ultrahigh vacuum chamber, constructed for these studies, is equipped

with a mass spectrometer, allowing the use of mass spectrometric techniques to quantify

the O-atom beam concentration.









The first method used to detect the presence of O-atom radicals in the beam was

line-of-sight threshold ionization mass spectrometry otherwise known as appearance

potential spectroscopy, as discussed by Agarwal et al. [58]. This technique is based on

the following principle. Electron impact ionization of oxygen atoms and molecules can

yield the same ion. Consider generating O+ ions for detection with the mass

spectrometer. They can be generated by either direct ionization of O-atoms or through

02 dissociation, as depicted by reaction Equations 1-1 and 1-2.

O + e--O + 2e- (El = 13.8 eV) (1-1)

02+ e--O+ + O +2e- (E2 =18.0 eV) (1-2)

The parameters El and E2 denote the threshold electron energies to generate O+ ions

through the processes shown in Equations 1-1 and 1-2 respectively, and are taken from

reference 58. The direct ionization process (1-1) always has a lower threshold than the

dissociative process (1-2) due to the additional energy required to break a molecular

bond. This difference provides a means of detecting oxygen radicals in the plasma beam

by monitoring the 16 amu signal intensity as a function of electron energy. Figure 1-8

shows appearance potential measurements taken with and without the plasma activated

after a background subtraction taken at the lowest electron energy probed (12 eV). The

beam with the plasma not activated shows the formation of O+ ions before the threshold

energy of 18.0 eV. This is attributed to thermal dissociation of 02 on the hot filament in

the ionizer, which is subsequently ionized and detected. Comparing the two traces

shown, it may be seen that the 0+ ion signal at E < 18 eV increases by about an order of

magnitude after activating the plasma. This increase provides direct evidence that O-

atoms are present in the plasma-activated beam.









Alternatively, beam experiments may be performed to estimate the O-atom

cracking fraction and at the same time search for contaminants. This allows the changes

of the beam components to be quantified when activating or deactivating the plasma

beam. This method is complicated in that introducing a beam into the chamber may also

displace other gases from the chamber walls, which may be interpreted as beam

components. The predominant components in the oxygen beam with the plasma not

activated are 32 and 16 amu, with trace contaminates of 2 amu (hydrogen) and 28 amu

(nitrogen). The measured 16 amu component arises from electron impact dissociation of

02 in the mass spectrometer ionizer. The ion impact dissociation fraction was measured

at a value of 10%, which is close to the literature value of 12.3 % for 70 eV electrons

[59]. When the plasma was activated, the predominant component was still 32 amu, but

an increase in the signals of masses 16, 18 and 30 amu was observed concurrent with a

decrease in 28 and 32 amu. The rise in 16 amu is attributed to oxygen atom formation by

the dissociation of 02, the rise in mass 18 is attributed to oxygen atom reactions with

hydrogen to form water and plasma source outgassing, and mass 30 and is attributed to

the conversion of nitrogen (mass 28) into NO (mass 30). The mass spectrometer

sensitivity was increased to enhance the 160 atom signal, and masses 16, 18, 28 and 30

amu are shown in Figure 1-9 to compare the relative intensities of each beam component.

At time equal to zero, the plasma has already been activated, and the shutter is closed.

After 85 seconds, the shutter was opened and a rise in the mass 16 signal was observed.

Two hundred seconds elapse and then the plasma is deactivated without changing the 02

feed rate to the plasma discharge tube. The 16 amu signal was observed to drop. This









technique shows clearly the presence of oxygen radicals in the beam. Masses 18, 28, and

30 are shown in Figure 1-9 to illustrate the low contaminant levels in the beam.

The dissociation fraction may be estimated in the following manner. When the

plasma is deactivated, it is reasonable to assume that the sole source of the m/e 16 signal

is due to the dissociation of molecular oxygen in the ionizer. With an electron energy of

70 eV, this fraction is 10%, which provides a measure of the molecular oxygen flow rate

with the plasma off. Since the pressure in the beam chamber remains constant when the

plasma is deactivated, the beam flow rate is also assumed to remain constant. With this

information, a mass balance can be performed on molecular and atomic oxygen and the

dissociation fraction may be estimated using the following relationships.


(P -Po)
02Tot (1-3)
0.10

2 f(Po Po )
Pon P = +(1- f)(Po -Po) (1-4)
0.10

Here, Pon, Poff, Po, O2tot, and fare the 160 partial pressure when the plasma is on, the 160

partial pressure when the plasma is off, the baseline pressure, the total 1602 entering the

ionizer and the dissociation fraction, respectively. Equation 1-3 provides a measure of

the total flow rate of 02 into the ionizer and the second expression may be solved to

determine the dissociation fraction of O-atoms (f). Solving this expression for f yields an

estimation of the 02 dissociation fraction of about 3% (flux of about 0.02 ML/sec). It

should be noted that the flux of 02 into the ionizer consists of a background and a direct

component. The background component can be measured by obstructing the beam flow

path with the sample manipulator, preventing any direct 02 flux into the mass









spectrometer. Once this value was obtained, it was subtracted off of the mass 16 partial

pressure trace, providing a measure of the direct 02 flux into the ionizer.

The final method used to determine the O-atom flux was to measure the oxygen

uptake on Pt(l 11). Figure 1-10 shows the O-atom uptake curve as a function of exposure

time while the sample was held at a constant temperature of 450 K. It is known that

molecular oxygen will dissociatively adsorb onto the Pt(1 11) surface at this temperature

and saturate at 0.25 ML. Therefore, any additional amount of oxygen deposited onto the

surface is attributed to the adsorption of oxygen atoms. Within the first 180 seconds of

the atomic oxygen exposure, the uptake curve is approximately linear. Assuming a unity

sticking probability in this uptake region, the slope provides an approximation of the O-

atom flux on the sample surface, which is 0.03 ML/sec. This is in reasonable agreement

with the quantity determined by mass spectrometric methods especially considering that

the sample is closer in proximity to the beam source than the mass spectrometer, yielding

a higher atomic oxygen impingement rate onto the sample.

1.3.6 Gas Handling

A 14 inch VCR manifold is used to direct the gases to various ports on the system.

To minimize the introduction of contaminants in the gas, the manifold is evacuated with a

70 1/s turbo pump. The pressure inside the manifold is monitored using a thermocouple

gauge tube. Gas lines from the manifold are connected to gas cylinders and all gas entry

ports into the chamber.

1.3.7 Calibrated Molecular Beam

A typical means of exposing a sample to a gas in vacuum is to fill backfill the

chamber with a gas for a measured period of time. The product of the pressure and the

time provides a measure of the gas exposure on the surface. For example, an exposure of









about one monolayer of gas onto the surface corresponds to about one Langmuir, which

is 10-6 Torr*sec. This poses a problem when gas dosing with "sticky" molecules such as

water or ammonia, which could lead to long pump out times. Molecular beams provide a

solution to this problem by providing an enhanced exposure to the sample surface while

minimizing the gas load on the system. Several advantages are afforded by employing a

calibrated molecular beam. If the gas flow rate is known, it becomes possible to directly

measure gas uptake onto the surface and the sticking probability of the gas with coverage.

The beam operates under the following principles and is depicted by Figure 1-11. An

example background partial pressure trace is shown in Figure 1-12. A mass balance on

the gas-phase beam species is shown in Equation 1-5.

dN
N = N ot Radnet (1-5)
dt

The quantities Ng, Radnet, N and &out are the number of gas molecules in the system, the

net adsorption rate, the total molecular flow rate into the system and the total molecular

flow rate out of the system, respectively.

For large S',


dNg 0 (1-6)
dt

Equation 1-6 will be justified later. Substituting P S,' for N out and rearranging Equation

1-6 yields:

P(t) 1 adnet(t) (1-7)
S
P


And









Radnet = F1S(0) Rd (0) (1-8)

where 0 is the relative adsorbate coverage, S(0) is the sticking probability as a function of

coverage, F is the intercepted fraction, and Rd(0) is the desorption rate from the surface.

If the exposure is conducted below the desorption temperature of the adsorbate, Rd(0) can

be neglected. Substituting Equation 1-8 into 1-7 yields


P(t) = [1- FS(O)] (1-9)
S
p

At t = o, Radnet = 0, then P = 1N/ / S,' and P(t)=[1-PJFS(0)], where P. and P(t) are the

steady state pressure after adsorption has ceased and the partial pressure as a function of

time. Substituting this information into Equation 1-9 and rearranging yields the

following expression:

Rodne, = [Pt -P(t)]S (1-10)

t
f Rd netd
0 = 0 (1-11)
J Rad netd
0

Winkler et al. performed calculations to determine the intercepted fraction (F) as a

function of doser-sample distances and geometries [60], which allows F to be read

directly from a chart provided in the reference. This derivation shows that adsorption

rates and sticking probabilities can be determined directly from the background gas

partial pressure traces. The relative coverage 0 can be determined by Equation 1-11. The

next portion of this derivation will justify the approximation given by Equation 1-6.









Assuming the sample is already saturated with the gas, and no adsorption on a sample is

occurring, Equation 1-5 can be written as Equation 1-12.

dNg = PS (1-12)
dt

Substitute Ng=PV/RT and Sp'=Sp/RT, where Sp is the pumping speed in L/sec, R is

the gas constant, V is the chamber volume in L, and P is the pressure in Torr.

dP frRT (P (1-13)
dt V [V)

Let T, equal the pumping time constant, where zp= V/Sp. For our system with only

the turbo pump evacuating the chamber,

50L
T 0L = 0.25 sec (1-14)
P 200L/sec

Equation 1-13 is a first order differential equation, when solved yields Equation 1-15.


P(t) = Pm e (P -Po) (1-15)

With such a small zp, the system pressure will reach Po with shorter time scales than

adsorption or reaction.

Figure 1-13 shows a schematic of the beam doser that has been mounted to the

vacuum chamber via a single-axis translator. The calibrated molecular beam doser was

assembled in the following manner. The primary components required for the beam

doser were machined by the A&N Corporation, and consist of a 1/4 inch stainless steel

tube passing through a CF 275 blank flange and a head piece that mates with the tube

through a VCR connection. A 10 micron diameter 14 inch VCR gasket was obtained

from Lenox Laser, and is used as the flow orifice. A glass micro-capillary array was

obtained from Buhrl electro-optics and was mounted in the doser head to provide a









directed flux to the sample. The array is held place by a stainless steel cap. Outside of

the vacuum chamber, a 4-way CF 275 cross is used as the gas reservoir. Mounted to the

cross is a Baratron gauge capable of reading pressures from 0.01 to 10 Torr, inlet and

evacuation valves, and a line running up to the molecular beam doser. The flow rate

through the beam was controlled by regulating the pressure in the cross. To calibrate the

beam flow, the conductance across the pin hole was determined by the following

procedure. First the volume of the cross and all lines leading up to the gasket was

measured (255 ml). The cross was then charged with a gas to an initial pressure Po. The

gas was allowed to flow out of the reservoir and through the pin-hole, while recording the

pressure inside the cross for an excess of 10 hours. By performing a mass balance of the

gas inside the cross, the pump out time constant was related to the conductance of the

orifice and the volume of the cross, in the following way.

The gas flow rate through an orifice of a known conductance is given by the

following expression,

dP
S=C(P -P2) (1-16)


where C is the conductance, and P1 and P2 correspond to the pressure in the reservoir and

in the vacuum chamber, respectively. Since the pressure in the vacuum chamber is on the

order of 10-10 Torr, P2 can be neglected. The conductance of the pin hole may be

calibrated by measuring the pressure of the reservoir as time progresses. The pump out

rate follows the expression shown in Equation 1-17. Upon integration it yields an

exponential decay shown in Equation 1-18. A plot of the pressure decay in the reservoir

as a function of time is shown in Figure 1-14 and yields a time constant of 31758









seconds. In this example, the conductance is 8.03*10-6 L/sec using carbon monoxide as

the beam gas.


dp
V =Cp (1-17)
dt


P = Poe ctv (1-18)

1.3.8 Calibration of 02 Beam Flow from Plasma Source

The addition of the calibrated molecular beam to the system also provided the

means for calibrating the flow rate of molecular oxygen through the plasma beam.

Initiating the beam caused a rise in the pressure in the first stage of the beam chamber.

Correlating the pressure rise in the first chamber with the oxygen flow rate provides a

simple means of checking the beam flow rate. Using the mass spectrometer, the partial

pressure of 02 can be correlated with the beam chamber pressure. The pressure curve is

then converted to number flow by comparing the measurements with those obtained by

the calibrated molecular beam. A calibration curve is then obtained and is shown in

Figure 1-15.

1.4 Detection Techniques

1.4.1 Reaction Product Monitoring

In surface adsorption and chemical reaction systems, real time reaction kinetic

data can be obtained by monitoring the temporal evolution of the background partial

pressures of both reactants and products during the reaction. For a given experiment, a

sample is prepared and exposed to a reactant beam. A derivation is given to illustrate the

relationship of the monitored partial pressures to the reaction rate and is similar to the

derivation shown in Section 1.3.7. For example, consider the dynamic displacement of









1802 from Pt(1 11) due to the impingement of 160-atoms as shown in Figure 1-16. In this

experiment, the surface was initially covered with 0.44 ML of 1802 and then subsequently

translated into the plasma beam path, while the shutter was closed. At time zero, the

shutter was opened. The partial pressure of 1802 initially jumped, and went through a

maximum and then decayed to the baseline.

First, assume that the system is initially at steady state (t < 0 sec.), with a constant

background of 36 amu. This background is attributed to a constant leak rate of 1802 and

described by Equation 1-19,

L = KSpPeq (1-19)

where L is the leak rate, K is the conversion constant to molecular flow, Sp is the

pumping speed in L/sec, and Peq is the equilibrium pressure in Torr. The product

K*S*Peq is the rate at which gas is pumped out of the system. The leak rate could

originate from species displaced from the walls or from the beam. Assume also that the

gas desorbing from the surface does not re-adsorb during the course of the experiment.

This is reasonable under the conditions examined in this study. A mass balance around

the desorbing species is shown in Equation 1-20.


AN(t)+L =KSP + KVd (1-20)
dt

Here, A is the sample area, V is the volume of the chamber, and N is the

desorption rate. Since the flux of the reagents remains constant during the experiment,

the desorption rate can either be calculated as a function of reactant fluence or time. The

second term on the right hand side of Equation 1-20 depicts the particle accumulation in

the system due to a rise in the partial pressure. Combining Equations 1-19 and 1-20

yields the following expression.










AN(t)+ KSP = KSP + KV p (1-21)
dt

Substitute a = A/KV, P*=P-Peq, T=V/S, where c is the characteristic pump out time

constant.

dP*
azN(t) = P*+ -- (1-22)
dt

In Section 1.3.7, I showed that the derivative term could be neglected, leaving the

desorption rate directly proportional to the partial pressure traces obtained from the mass

spectrometer.

The area under the curve shown in Figure 1-16 is then directly proportional to the

desorption yield of 18O2. If the desorption yield is known, the data shown in Figure 1-16

can be directly converted to the desorption rate in ML/s using the following expression,

where s is a dummy integration variable, AP is the pressure of the system minus the

baseline, and 90 is the initial coverage.

AP(t)
rate = 00 t) (1-23)
J AP(s)ds
0

1.4.2 Temperature Programmed Desorption and Reaction

Temperature programmed desorption and reaction techniques were used to quantify

the amount of each species residing on the surface. These spectra also contain

information regarding the desorption activation energy for each species [61,62]. The

derivation relating the partial pressure to the desorption rate, shown in 1.3.7, also pertains

to TPD analysis. Figure 1-17 shows an example TPD spectrum of atomic oxygen

recombining on Pt(1 11) and desorbing as 02 with 0o equal to 0.25, 0.38, and 0.59 ML.

Integrating the signal intensity of each spectrum gives the total amount of products









desorbing. Desorption features are observed at 550, 640, and 710 K. Each of these

features is attributed to oxygen atoms residing in environments with different surface-

adsorbate binding strengths. In this particular example, one can see the lower

temperature desorption features increasing with 00, indicating that the oxygen atoms are

experiencing strong lateral repulsions from the neighboring atoms and have a weaker

oxygen atom-surface bond.

1.4.3 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy is a technique which entails illuminating the

sample with X-ray radiation, and examining the energy of the photoelectrons evolving

from the surface. The kinetic energy of these electrons is given by the terms of the

Equation 1-24,

KE= h v BE (1-24)

where hv is the photon energy, BE is the binding energy of the electron, and 4 is the work

function difference between the sample and the electron analyzer. Each element has a

unique electronic structure, and hence has its own unique XPS spectrum. The binding

energy is a function of the local element bonding environment, giving this technique the

means to probe the chemical state of each element. The photo-electrons are generated in

the near surface region of the sample. The depth probed with this technique is

characterized by the following expression,

d =A cos(0) (1-25)

where d, k, and 0 represents the distance from the surface, electron mean free path, and

the electron take off angle respectively. By varying the angle at which the photo-









electrons are taken, XPS will probe the concentrations of each species at different depths

within the surface. This forms the basis for angle-resolved XPS.


AB A B

A-B A-B


A-B A-B
M*M .4


B

A \A



A-B B-A
2M*zz^zzzz^


A A





A-B


Figure 1-1. Mechanisms describing surface chemical reactions. Left) The LH
mechanism. Middle) The ER mechanism. Right) The HA mechanism.


Figure 1-2. UHV system constructed for our study.


A \B

















Vmrpcat


Tiwpcrt


ion sorxce


viewpmpt









11rbo mp


LEED


Figure 1-3. Top view of the UHV system. A) Upper level. B) Lower level.


































Figure 1-4. Specimen mount on the sample manipulator.









Power interrupt switch
Power supply


Sample


TC -


Figure 1-5. Sample temperature control scheme used for these experiments. Current is
driven across the sample by the DC power supply, which is regulated by the
controller.


Ir



















1000



800 Linear Heating ramp
I K/s


600
E
I-
400



200-



0- I *I
0 200 400 600 800 1000
Time (sec)




Figure 1-6. Typical linear temperature ramp with a heating rate of 1 K/sec.
































A)


















B)

Figure 1-7. Beam chamber housing the plasma source. A) A cross-sectional top view.
B) A side view.


















1000000 "-



0 100000

SI Plasma oft
.0
10000



1000 I Direct and dissociative
ionization


100
10 12 14 16 18 20 22 24 26 28 30 32
Electron Energy eV



Figure 1-8. Appearance potential measurement, monitoring O+ (m/e 16) with the plasma
activated and deactivated with a first stage beam chamber pressure of
3.0*105 Torr.





















-i

C-
0O



Masses 16
P4
Masses 18. 28. and 30



0 100 200
Time (sec)



Figure 1-9. Mass spectrometer beam trace experiment taken with the plasma initially
activated and the shutter closed with an electron energy of 70 eV. Masses
16, 18, 28 and 30 were monitored. After 85 seconds, the shutter was opened.
Once 200 seconds have elapsed, the plasma was powered down, while
maintaining the same 02 flow rate.













3.0


2.5


2.0


S1.5 S
(C)(2) + ) Pt(l II)
S02o ~ 0.25 ML/sec
o 10
Ts=450 K
linear slope 0.03 ML/sec
0.5


0.0
0 1000
time (sec)


Figure 1-10. Oxygen uptake on Pt(l 11) with a surface temperature of 450 K. The
oxygen uptake is approximately linear in the range of 0-180 seconds, with a
slope of 0.03 ML/sec.





























Figure 1-11. Gas flow from the beam onto the sample, where N, F, and Nout represent
the total molecular flow rate into the chamber (#/sec), the intercepted
fraction by the sample and the total molecular flow rate out of the system.
The system pressure (Torr) and the pumping speed (#(sec)-(Torr)-) are
denoted as P and S ', respectively.


NPout = P* S P










Beam deactivated



A


Po ---



Pb
Pb Beam activated

Time

Figure 1-12. Hypothetical beam trace depicting the partial pressure of a gas using a
calibrated molecular beam, where A is proportional to the quantity of gas
adsorbed on the sample.






















A


#2-56

0.0119 017
060178


2,00
S0.15



B
Figure 1-13. Molecular beam doser used for our experiments. A) Overall beam doser
view. B) Beam doser head piece.















2.5



2.0



S1,5



1.0-



0.5


0 1x104 2x104 3x10' 4x104 5x10 6x104

time (sec)




Figure 1-14. Pressure of carbon monoxide in the gas reservoir plotted as a function of
time while pumping out through the laser drilled VCR Gasket. The first
order decay time constant is 31758 seconds.













4 5x10"' I

4 0x10'"

3.5x10' ,
(D
S3 10xo104 RawData
S3.X ---Linear Fit
0
E 25x10"

0 20x10"
ii
1.5x14 y= 7.3E13*x + 3.51E13
1.x104 -

1.0x10" -

50x10 -

0 1 2 3 4 5

First Stage Pressure (Torr 10)



Figure 1-15. Calibration of the molecular oxygen flow rate through the beam chamber is
compared with the pressure in the first differential pumping stage.






















C-
z









0 100 200
Time (sec)



Figure 1-16. Dynamic displacement of 1802 taken upon exposing a Pt(1 11) surface
saturated with 802 to 160-atoms, while holding the sample temperature
constant at 90 K.






















C,











500 600 700 800 900
Temp(K)



Figure 1-17. Temperature programmed desorption performed on Pt(1 11) after exposing
to a beam of oxygen atoms. The surface coverage of oxygen atoms from
lowest to highest coverage is 0.25, 0.38 and 0.59 ML.














CHAPTER 2
OXIDATION OF NITRIDED Si(100) BY GASEOUS ATOMIC AND MOLECULAR
OXYGEN

The nitridation of Si(100) by ammonia and the subsequent oxidation of the nitrided

surface by both gaseous atomic and molecular oxygen was investigated under ultrahigh

vacuum (UHV) conditions using X-ray photoelectron spectroscopy (XPS). Nitridation of

Si(100) by the thermal decomposition of NH3 results in the formation of a subsurface

nitride and a decrease in the concentration of surface dangling bond sites. Based on

changes in the Nis spectra obtained after NH3 adsorption and decomposition, we

estimate that the nitride resides about four to five layers below the vacuum-solid interface

and that the concentration of surface dangling bonds after nitridation is only 59% of its

value on Si(100)-(2xl).

Oxidation of the nitrided surface is found to produce an oxide phase that remains in

the outer layers of the solid, and interacts only weakly with the underlying nitride for

oxygen coverages up to 2.5 ML. Slight changes in the NIs spectra caused by oxidizing at

300 K are suggested to arise primarily from the introduction of strain within the nitride,

and by the formation of a small amount of Si2-N-O species at the nitride-oxide interface.

The nitrogen bonding environment changes negligibly after oxidizing at 800 K, which is

indicative of greater phase separation at elevated surface temperature. Nitridation is also

found to significantly reduce the reactivity of the Si(100) surface toward both atomic and

molecular oxygen. A comparison of the oxygen uptake on the clean and nitrided surfaces

shows quantitatively that the decrease in dangling bond concentration is responsible for









the reduced activity of the nitrided surface toward oxidation, and therefore that dangling

bonds are the initial adsorption site of both gaseous oxygen atoms and molecules.

Increasing the surface temperature is found to promote the uptake of oxygen when 02 is

used as the oxidant, but brings about only a small enhancement in the uptake of gaseous

O-atoms. The different effects of surface temperature on the uptake of O versus 02 are

interpreted in terms of the efficiency at which dangling bond pairs are regenerated on the

surface. In particular, it is suggested that elevated surface temperatures promote

subsurface oxygen migration and concomitant regeneration of empty surface dimers,

which are required for 02 activation. In contrast, the availability of single dangling bonds

needed for the adsorption of a gaseous oxygen atom is suggested to be relatively

unaffected by the surface temperature.

2.1 Introduction

Silicon nitride and oxynitride films have been extensively investigated in recent

years due to the advantages afforded by incorporating these materials into the dielectric

layers used in metal-oxide-semiconductor (MOS) devices. The addition of small

amounts of nitrogen to the SiO2-Si interface is known to improve the structural quality of

the interface, and results in lower leakage current across the gate of a MOS device as well

as enhanced resistance to boron diffusion into the SiO2 [1]. The low activity of silicon

nitride toward oxidation has also proved beneficial to the growth of alternative gate

oxides such as Ta205 and ZrO2 that have higher dielectric constants (k) than SiO2.

Recent investigations have shown that the deposition of high k oxides directly onto

silicon can result in the formation of an SiO2 layer that dominates the capacitance of the

gate stack [2-6]. Incorporation of nitrogen in the near-surface region of silicon alleviates

this problem by inhibiting SiO2 formation during the deposition of Ta205, thereby









enabling the benefits of the high k oxide film to be more fully realized [7,8,63]. Another

important application of silicon nitride is as a protective coating on ceramic components

such as bearings and turbine blades for which tolerance to high temperature, oxidizing

environments is critical. Despite these important applications, however, the oxidation of

silicon nitride films is not well understood at the molecular level. In this article, we

discuss results of an ultrahigh vacuum (UHV) investigation of the oxidation of a silicon

nitride film by both gaseous atomic and molecular oxygen in which we focused our

efforts on elucidating the fundamental origin for the oxidation resistance of nitrided

silicon.

Several early studies have characterized the oxidation of thick silicon nitride films

under conditions of high oxidant pressure [64-67]. For example, Kuiper et al.

investigated the oxidation of a thick silicon nitride film by exposure to 02 and H20 in an

atmospheric furnace. They report that the rate of oxidation of the nitride film is two

orders of magnitude lower than the oxidation rate of Si(100), and that the presence of

hydrogen (H2 and H20) was necessary to oxidize the surface under the conditions

examined. They assert that hydrogen reacts with the nitride to form gaseous ammonia

and elemental silicon and that the surface activity toward oxidation is enhanced as a

result since elemental silicon is more easily oxidized than the nitride. Similar results

have been reported in studies of the dry oxidation of silicon nitride films [64,65,67].

While these investigations have characterized the oxidation resistance of silicon nitride at

high pressure, experiments conducted under more well-defined and controllable

conditions are needed to determine the underlying cause for the oxidation resistance of

silicon nitride films. Experiments of this type have been reported recently by Wallace et









al. [68]. In this work, the investigators thermally decomposed ammonia on Si( 11) to

generate silicon nitride films in UHV and then oxidized the nitrided surface with

molecular oxygen without breaking vacuum. From in situ analysis of the surface using

X-ray photoelectron spectroscopy (XPS), the authors observed negligible oxygen uptake

at surface temperatures below 873 K, and only a small amount of uptake above 873 K.

The slow oxygen uptake was suggested to arise from a decrease in the concentration of

surface dangling bonds after nitridation, though this effect was not quantified.

Prior investigations of the nitridation of Si(100) by NH3 provide important insights

for understanding how nitridation alters the properties of the Si(100) surface. At room

temperature, ammonia adsorbs dissociatively on Si(100)-(2xl) to produce an adsorbed

hydrogen atom and an NH2 moiety [69-72]. Heating the ammonia-saturated surface to

about 700 K then leads to the decomposition of adsorbed NH2 and the complete

desorption of hydrogen. Early investigations of this system also showed that the nitrogen

atoms occupy subsurface sites after the NH2 groups decompose [72,73]. For example,

Dresser et al. [72] observed significant attenuation of the N(KLL) AES peak after sample

heating, but only observed small amounts ofNH3 desorption (< 10%). From these

observations, Dresser et al. concluded that nitrogen migrates to the sub-surface region

after the adsorbed NH2 species thermally decompose on Si(100). Subsequent studies

have confirmed that nitrogen migrates to the subsurface of Si(100) during nitridation at

elevated temperature (> 600 K) [74-78]. For example, using high resolution

photoemission, Peden et al. [74] obtained compelling evidence that silicon nitridation by

NH3 occurs by a mechanism in which nitrogen atoms diffuse into the subsurface region

and leave a thin layer of elemental silicon adjacent to the vacuum-solid interface that









persists as the underlying nitride film thickens. Experiments using low energy electron

diffraction (LEED) also reveal that annealing the ammonia-covered surface results in a

decrease in the intensity of the fractional order diffraction spots, signifying that NH2

decomposition and nitrogen penetration to the subsurface disrupts the long-range order of

the surface [79,80]. Considering that nitrogen resides below the vacuum-solid interface,

direct interactions between an oxidant molecule and nitrogen may be expected to have

only a minor influence on the oxidation behavior of nitrided Si(100). A change in the

structure of the surface, as indicated by LEED experiments, may therefore be the

predominant cause for the change in the reactivity of the surface toward oxidation.

Although few UHV investigations of the oxidation of silicon nitride films have

been reported, the oxidation of single crystal silicon surfaces has been studied

extensively. Of particular relevance to the present work are detailed UHV studies by

Engstrom et al. [48] on the oxidation Si(100) and Si( 11) by both gaseous atomic and

molecular oxygen. These and other results may also be found in a review of Si oxidation

written by Engel [49]. Briefly, under UHV conditions the dissociative adsorption of 02

on Si(100) results in an effective saturation coverage of only about 1 ML of oxygen

atoms when the surface is held at 300 K during oxidation. The saturation coverage can

be increased by oxidizing at elevated surface temperature, but the oxygen uptake is still

rather limited. For example, the saturation oxygen coverage increases to 2 ML when

Si(100) is exposed to 02 at a surface temperature of 800 K. Not surprisingly, Engstrom

et al. [48] found that gaseous oxygen atoms adsorb on Si(100) with much higher

probability than does 02, and that oxygen coverages greater than 10 ML can be obtained

by oxidizing Si(100) held at 300 K using an atomic oxygen beam. In contrast to the









results obtained using 02, the uptake of oxygen atoms was found to be insensitive to the

surface temperature for oxygen coverages up to about 5 ML, which is indicative of non-

activated adsorption and possibly direct insertion of gaseous O-atoms into surface Si-Si

bonds.

In the present study, we used X-ray photoelectron spectroscopy (XPS) to

investigate the nitridation of Si(100) by the thermal decomposition of ammonia as well as

the oxidation of the resulting nitride film by both gaseous atomic and molecular oxygen.

The key objectives of this study were to determine the surface properties responsible for

the oxidation resistance of silicon nitride films, and to characterize the mechanisms for

oxidation with these oxidants and the properties of oxidized nitride films. We find that

surface dangling bonds play a critical role in the adsorption of both O-atoms and 02 and

provide quantitative evidence that a decrease in the surface dangling bond concentration

is the primary cause for the decrease in oxygen uptake by Si(100) after nitridation.

2.2 Experimental Methods

All experiments were conducted in an ultrahigh vacuum chamber described in

Section 1.3. Briefly, this apparatus was equipped with a variable energy electron source,

dual anode X-ray source capable of generating Al and Mg Ka radiation, ion source, and a

hemispherical charged particle analyzer, giving the system the capability of performing

Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS).

Surface structural measurements were performed using low energy electron diffraction

(LEED) optics. A commercial microwave plasma source was employed to generate an

oxygen atom beam. The atom source was mounted in a stainless steel reaction chamber

with two stages of differential pumping. The communication between the plasma source









and the vacuum chamber was through a quartz tube with a diameter of 6 mm and a length

of 60 mm. The oxygen feeding the plasma source was supplied by BOC gases with a

purity of 99.999% without any further purification.

The Si(100) sample used in this study was taken from an arsenic-doped (0.005

ohm-cm) silicon wafer that was cut and polished to within + 0.50 of the (100) plane. An

approximately 2 x 1 cm rectangular section was cut from the wafer, and tightly fastened

to the Ta plates that contact the cooling reservoir of the manipulator. Only tantalum parts

were used to fasten the Si sample to the holder since the use of stainless steel parts was

found to introduce small amounts of nickel into the sample. To measure the sample

temperature, a K type thermocouple was spot-welded to a thin strip of Ta foil that was

bent into a U-shape and clipped to the back of the Si(100) sample. The sample was

cleaned by sputtering with 2 keV Ar+ ions followed by annealing for several minutes at

1000 K. The sample was considered to be clean when no contaminants could be detected

with XPS, and a sharp (2x1) LEED pattern was observed.

Ammonia was dosed onto the sample as a beam generated with the calibrated

doser. Typical NH3 fluxes used in these experiments were -5 x 1013 cm-2 sec-1, which is

estimated from the known NH3 flow rate from the doser and the angular emission

characteristics of micro-capillary arrays [57]. Exposures of NH3 are reported in units of

ML, where 1 ML is defined as the surface atom density of 6.8 x 1014 cm-2 of the Si(100)-

(2x1) surface.

Pure 02 beams were dosed onto the sample by flowing oxygen through the plasma

source with the microwave power disabled. The size of the beam spot on the sample was

about 9 mm in diameter. We typically employed an 02 beam flux of 0.26 ML/sec, which









was determined using the spot size estimate and by comparing the 02 pressure rise in the

UHV chamber due to the beam with that resulting from a known flow rate of 02 admitted

through the calibrated beam doser.

Beams containing oxygen atoms were generated by activating the microwave

plasma with 02 flowing through the source. With an initial 02 flux of 0.26 ML/sec, a

measurable change in the 16 amu intensity was not observed by line-of-sight mass

spectrometry when the plasma was activated. To estimate the O-atom flux, the oxygen

uptake was measured on clean Si(100)-(2xl) held at 300 K as a function of exposure to

oxygen beams with and without the microwave power enabled. From a comparison with

a previous investigation of the adsorption of gaseous O and 02 on Si(100) [48], we

estimate that O-atom fluxes of -0.001 ML/sec impinged on the sample surface for the

beam conditions employed. Subsequent to these experiments, it was determined that the

beam source was slightly misaligned with the collimating apertures which caused

substantial O-atom recombination prior to the beam entering the UHV chamber.

Although the O-atom fluxes that we employed were relatively low, significant

enhancements in the rate of oxidation were observed when the surfaces under study were

oxidized at a given fluence by a plasma-activated beam versus a pure 02 beam All XPS

spectra reported in this study were obtained using Al Kac X-rays (hv = 1486.6 eV) with

the analyzer operating in a retarding mode at a pass energy of 27 eV. The electron

takeoff angle was varied by rotating the sample with respect to the analyzer axis. An

angular resolution of + 5 is estimated from the geometry of the analyzer, and electron

takeoff angles are specified with respect to the surface normal. Unless stated otherwise,

the spectra presented here were obtained by measuring photoelectrons emitted at an angle









of 600 from the surface normal. Even at this glancing takeoff angle, the area of the

sample from which photoelectrons were collected was smaller than the spot size of the

oxygen beam, thus ensuring that the XPS measurements probed only the regions of the

surface that were dosed with gases. The XPS spectra presented here were each processed

using 21-point Savitzky-Golay smoothing, followed by background subtraction using the

Shirley method [81]. Oxygen coverages were determined from the ratio of Ols to Si2p

integrated intensities, and assuming that exposure of clean Si(100)-(2xl) at 300 K to 02

produces a saturation coverage of 1 ML [48,49]. For the low oxygen coverages

investigated here, we found it unnecessary to account for Ols and Si2p signal attenuation

due to inelastic photoelectron scattering since the oxygen atoms remain in the outer

surface layers. Nitrogen coverages were computed by a similar procedure, but inelastic

scattering corrections were necessary in this case, as described in Section 2.3.

2.3 Results

2.3.1 NH3 Decomposition on Si(100)

An ultrathin nitride film was grown on the Si(100) substrate prior to each oxygen

exposure by thermally decomposing 160 ML of ammonia on the surface at 900 K.

Several layers of nitrogen incorporate into the solid during this exposure since ammonia

decomposition and hydrogen desorption are rapid at 900 K [71,78]. The Si2p and NIs

spectra obtained from the Si surface after this treatment are shown in Figures 2-1 and 2-2.

The Si2p spectrum exhibits a main component at 99.2 eV due to elemental Si, and a

smaller feature centered at about 101.8 eV. The small Si2p feature appears at a binding

energy that is less than that of Si3N4 [74,77,78,82-85], which suggests the presence of a

sub-stoichiometric nitride. For films prepared under similar conditions, previous









investigations suggest that silicon is present in a Si3=N configuration in which each N

atom is bonded with three silicon atoms. The Nis spectrum obtained after nitridation

exhibits a single peak centered at a binding energy of 397.4 eV, which is also consistent

with previous reports [77,78,82,86].

Experiments were conducted to probe the interaction of NH3 with the Si(100)-(2xl)

surface so that the properties of the nitride film could be characterized in more detail.

XPS spectra were first obtained after exposing clean Si(100) held at 300 K to a saturation

dose of 160 ML of ammonia. This exposure produces a nominal coverage of 0.5 ML of

adsorbed NH2 groups, with the balance of the surface sites occupied by hydrogen atoms,

and these species do not undergo further reaction at 300 K under UHV conditions

[72,75]. The NIs spectrum obtained after the 300 K exposure exhibits a single peak

centered at a binding energy of 398.1 eV (Figure 2-3A), which is consistent with previous

reports of the Nls binding energy of adsorbed NH2 on Si(100) [73]. The NIs spectrum

shown in Figure 2-3B was then obtained after annealing the amino-covered surface for 5

minutes at 900 K, which results in the complete desorption of hydrogen from the surface.

It is noted that the spectrum did not change from that shown in Figure 2-3B when the

sample was annealed for longer times. After annealing, the NIs peak position is shifted

to a binding energy of 397.5 eV, which is consistent with the formation of Si3=N species

in the near-surface region. Annealing the amino-saturated surface also causes the Nls to

Si2p intensity ratio to decrease to about 60% of its initial value. The magnitude of this

decrease is in excellent agreement with that observed by Dresser et al. [72] after heating

amino-saturated Si(100) to temperatures greater than about 700 K and examining the

surface with AES. Dresser et al. estimated that less than 10% of the nitrogen desorbs









from the surface during heating, and therefore attributed the reduction in the N KLL peak

intensity to nitrogen penetration into the sub-surface [72].

From the intensity changes in the Nis and Si2p spectra, we estimated the distance

at which the nitrogen atoms reside below the surface after annealing the amino layer. For

this calculation, nitrogen desorption is neglected and it is assumed that all of the nitrogen

atoms initially present in the amino layer reside in a single layer below the surface after

the sample is heated to 900 K. It is further assumed that the probability of generating an

N1s photoelectron is the same for an adsorbed NH2 group as for the nitride. Under these

assumptions, the attenuated NIs to Si2p intensity ratio of 60% can be approximated by

exp(-x/kcosO), where x is the distance of the nitrogen atoms beneath the surface, X is the

inelastic mean free path of an NIs photoelectron through elemental Si and 0 is the

photoelectron takeoff angle measured from the surface normal. Assuming an inelastic

mean free path of 22.3 A [87], the 40% decrease in NIs peak intensity suggests that the

nitrogen atoms diffuse 5.8 A below the Si surface layer. Based on the spacing between

the Si layers closest to the (100)-(2xl) surface, this calculation suggests that nitrogen

atoms reside between the fourth and fifth layers after the amino-covered surface is

annealed at 900 K. Considering the simplicity of the analysis, our estimate is in good

agreement with recent electronic structure calculations which predict that a nitrogen atom

has a energetic preference to be located between the third and fourth layers of Si(100)-

(2xl) and to bond with three Si atoms in the third layer [76].

After annealing the initial amino layer, the surface was held at 300 K and again

exposed to 160 ML of ammonia. As may be seen in Figure 2-3C, this exposure causes

the NIs peak to grow in intensity and to shift toward higher binding energy, which









indicates the presence of adsorbed NH2 groups. Similar observations have been made by

Avouris, Bozso and Hamers [73], and indicate that annealing the amino layer partially

restores the reactivity of the surface toward ammonia dissociation. According to

quantum chemical calculations [88,89], the dissociation of ammonia on Si(100)-(2x1)

involves the adsorption of NH3 on a single atom of a surface dimer, followed by N-H

bond cleavage and hydrogen transfer to the opposing silicon atom of the dimer.

Considering this prediction, it may be concluded that nitrogen diffusion into the

subsurface of Si(100) regenerates pairs of surface dangling bonds, and thereby partially

reactivates the surface toward ammonia dissociation. Interestingly, however, the increase

in the NIs to Si2p intensity ratio after saturating the annealed surface with amino groups

is only 54% of that obtained after saturating clean Si(100) with amino groups (Figure 2-

3A). This difference suggests that nitrogen incorporation into the sub-surface of Si(100)

is accompanied by a structural rearrangement of the surface that reduces the density of

dangling bond pairs by nearly a factor of two from its value on the clean surface. Indeed,

in a prior study, LEED images taken after ammonia decomposition on Si(100) show a

diffuse background that eclipses the fractional order spots [79,80], indicating that

nitrogen incorporation does alter the structure of the surface. The total dangling bond

coverage on the nitrided surface may be estimated as 0.59 ML when all of the surface

dangling bonds are assumed to exist in pairs, and taking into account attenuation of the

N1s signal from subsurface nitrogen due to inelastic electron scattering from the NH2

adsorbed at the surface.

As stated above, the procedure we employed for growing a nitride film for

subsequent oxidation studies was to expose the clean surface held at 900 K to 160 ML of









ammonia. The intensity of the Nis peak obtained after this procedure (Figure 2-2) is

about three times greater than that obtained from the amino-saturated Si(100) surface. To

estimate the thickness of the nitride film, we assume that a layered structure is formed

and that each layer contains 0.5 ML of nitrogen atoms. Furthermore, based on our

analysis of the amino-saturated surface before and after annealing, it is assumed that the

nitride layer closest to the vacuum-solid interface resides four to five layers below the

surface; this assumption is supported by the oxidation results discussed below. With

these assumptions, and invoking a simple model to account for signal attenuation due to

inelastic electron scattering, we estimate that a nitride film of 5 to 6 atomic layers in

thickness is generated by the 160 ML ammonia exposure at 900 K. The findings from

these experiments that have a key impact on the understanding of oxidation of the

nitrided surface are 1) that the nitride films produced by ammonia decomposition on

Si(100) reside in the sub-surface region and 2) that nitridation reduces the density of

surface dangling bond sites.

2.3.2 Oxidation of Nitrided Si(100) at 300 K

Atomic versus molecular oxygen. Oxidation of the nitrided surface by both

molecular and atomic oxygen was investigated at a surface temperature of 300 K. After

nitridation the surface was exposed to the oxygen beam for 60 minutes, and the surface

was then analyzed with XPS. A 60 min beam exposure corresponds to -930 ML of 02

for the fluxes employed. Exposing the nitrided surface to the pure 02 beam for 60

minutes results in an oxygen coverage of 0.27 ML, which was found to be the limiting

coverage for oxidation of the nitrided surface at 300 K by 02. The oxygen coverage

increased to 1.2 ML when the nitrided surface was exposed to the same beam fluence but









with the plasma power enabled. This result shows that gas-phase oxygen atoms are

significantly more reactive toward nitrided Si(100) than is 02, particularly considering

that only about 3 ML of oxygen atoms are estimated to have impinged on the surface

during the 60 minute exposure. The Ols feature obtained after oxidizing with the

plasma-activated beam is very similar in shape and peak location to that obtained after

oxidizing with 02 at 300 K (not shown). The Ols peak is shifted to a higher binding

energy by 0.1 eV after oxidizing with the plasma-activated beam, compared with

oxidation with pure 02, but this shift is consistent with the higher oxygen coverage that is

achieved with the atomic oxygen beam. It is well known that as the coverage of oxygen

is increased, the oxygen atoms on the Si(100) surface experience changes in their

bonding environment that alters the Ols binding energy [48,49]. The similarities in the

Ols spectra indicate that gaseous oxygen atoms and molecules produce similar chemical

states of oxygen on the nitrided surface, which suggests that after adsorption (or 02

dissociation) the processes by which oxygen atoms incorporate into the nitrided surface

are independent of the identity of the gaseous oxidant. It therefore follows that the

enhanced uptake achieved with the plasma-activated beam is due to the higher adsorption

probability of oxygen atoms compared with 02 on the nitrided surface.

Clean versus nitrided Si(100). A comparison of the oxygen uptake on the clean

and nitrided Si(100) surfaces reveals that nitridation significantly lowers the surface

reactivity toward oxidation. For example, exposing the clean surface to 930 ML of 02 at

300 K results in an oxygen coverage of about 1 ML, whereas a coverage of only 0.27 ML

is reached on the nitrided surface by oxidizing under the same conditions. The difference

in the reactivity of these surfaces toward gaseous oxygen atoms is less pronounced, but is









still quite significant. Specifically, an oxygen coverage of 1.2 ML is obtained by

exposing the nitrided surface at 300 K to the plasma-activated beam for 60 minutes,

whereas a coverage of 2.1 ML is obtained on the clean surface for the same exposure and

surface temperature. Since the nitride film is shown to reside several layers below the

vacuum-solid interface, the lower reactivity of the nitrided surface compared with the

clean surface does not arise from a direct interaction between nitrogen and oxygen videe

infra) but is attributed to the lower concentration of dangling bonds on the nitrided

surface.

Chemical state changes induced by oxidation at 300 K Shown in Figures 2-4 to

2-6 are the Si2p, Nis and Ols spectra obtained from the nitrided surface after depositing

1.2 ML of atomic oxygen at a surface temperature of 300 K. Also shown are the Si2p

and NIs spectra obtained from the nitrided surface before oxidation, and the Ols

spectrum obtained after exposing clean Si(100)-(2xl) held at 300 K to a saturation dose

of 02, which results in about 1 ML of atomic oxygen on the surface. Each spectrum has

been normalized by its peak height to facilitate comparison. Among all these spectra, the

most distinct change caused by oxidation is an increase in the intensity of a feature

centered at about 102 eV in the Si2p spectrum, see Figure 2-4. This spectral change

indicates an increase in the amount of Si atoms present in partially oxidized states as

oxygen atoms are incorporated within the nitrided surface. As discussed below, this

feature is assigned specifically to Si2+ and Si3+ species that are directly bonded to oxygen

atoms in the outermost surface layers.

The NIs spectra obtained before and after oxidation at 300 K are shown in Figure

2-5. Oxidation causes small changes in the NIs feature, but the changes observed are









reproducible and distinct. In particular, after oxidation the center of the Nis peak is

shifted to lower binding energy by about 0.2 eV, and a small feature appears at a binding

energy of 399.2 eV. Previous studies have reported that Si2=N-O structures give rise to

an NIs feature at a binding energy of about 399 eV. The appearance of the high binding

energy shoulder is therefore consistent with a small quantity of N-O bonds being formed

upon oxidation at 300 K. The incorporation of both nitrogen and oxygen atoms in the

near-surface region of Si(100) has been found in most cases to cause the NIs peak to shift

to a higher binding energy (BE) relative to the NIs BE obtained from pure nitride

surfaces [76,90,91]. A positive binding energy shift of the NIs peak may result from a

core-hole screening effect that arises from the formation of a dielectric film near the

vacuum-solid interface. However, such a screening effect should be negligible for the

films we have investigated since no more than 2 ML of oxygen atoms are present at the

outer surface. Positive BE shifts of the Nls feature have also been attributed to a second

nearest neighbor (NN) interaction in which oxygen atoms withdraw charge from Si atoms

that are directly bonded to nitrogen in the film [76,91]. The negative binding energy shift

of the main NIs feature observed in the present study suggests that at 300 K the majority

of oxygen atoms do not penetrate far enough below the vacuum-solid interface to occupy

second NN positions with respect to the nitrogen atoms. The small shift of the NIs peak

to lower binding energy could be caused by the introduction of strain at the nitride-Si

interface when oxygen atoms are incorporated into the top surface layers. Since these

spectra show that the majority of the N and O atoms in the film do not directly interact,

the growth of the 102 eV feature in the Si2p spectrum following oxidation may be









attributed to the formation of Si2+ and Si3+ species that are directly bonded to oxygen

atoms in the outermost surface layers.

The Ols spectra obtained after depositing 1 and 1.2 ML of oxygen on the clean and

nitrided surfaces, respectively, at 300 K are shown in Figure 2-6. The Ols peak obtained

from the nitrided surface after oxidation is similar in shape to that obtained from the pure

oxide layer, but is shifted to higher binding energy by about 0.3 eV. This difference in

binding energies indicates that oxygen atoms at a concentration of about 1 ML

experience slightly different chemical environments when adsorbed on clean versus

nitrided Si(100) at 300 K. Based on the small NIs feature observed at 399 eV, a small

fraction of the oxygen atoms appear to be directly bonded with nitrogen atoms in the

film. This bonding interaction could alter the Ols binding energy, and produce a shift

from the binding energy obtained from the pure oxide film. In addition, the oxygen

atoms near the vacuum-solid interface, which do not directly bond with nitrogen, may

experience a different chemical environment than a similar quantity of oxygen atoms

incorporated into the clean surface. Such an effect could arise if the structures in the

near-surface layers of the solid differ for the clean and nitrided surfaces. This latter

interpretation is consistent with the ammonia uptake experiments which show that

nitridation lowers the surface dangling bond density, probably by inducing a structural

change at the surface.

Angle-resolved XPS data. The chemical changes suggested by the XPS spectra

provide the general picture that oxidation of the nitrided surface at 300 K results in nearly

segregated oxide and nitride layers, with the oxide layer being closer to the vacuum-solid

interface. To further examine this possibility, XPS spectra were obtained at different









photoelectron takeoff angles to vary the depth resolution of the measurements. Shown in

Figure 2-7 are the Ols and NIs spectra collected at 0 and 600 takeoff angles with respect

to the surface normal after adsorbing 1.2 ML of oxygen atoms on the nitrided surface at

300 K. In order to illustrate the variation in the Ols/N1s intensity ratio with average

sampling depth, the spectra obtained at a given takeoff angle are scaled by the integrated

intensity of the NIs spectrum measured at that angle. Qualitative differences in the

spectra obtained at different collection angles are slight, suggesting that the chemical

states of nitrogen and oxygen remain fairly uniform throughout the film. The slight

broadening of the N1 s feature toward high binding energy at glancing takeoff angle may

arise from a small amount of N-O bonding at the nitride-oxide interface closest to the

surface. The most distinct difference between these spectra is clearly the increase of

approximately 15% in the Ols/N s intensity ratio for the measurements performed at a

600 takeoff angle, i.e. the more surface sensitive configuration. This result confirms that

the oxygen atoms reside closer to the outer surface of the solid than do the nitrogen atoms

when as much as 1.2 ML of oxygen atoms are adsorbed on nitrided Si(100) at 300 K.

2.3.3 Oxidation of Nitrided Si(100) at 800 K

Atomic versus molecular oxygen. Oxidation of the nitrided surface by both

atomic and molecular oxygen was also investigated at a surface temperature of 800 K to

compare with the oxidation behavior observed at 300 K, and to therefore assess the

influence of surface temperature on oxidation. Increasing the surface temperature from

300 to 800 K significantly enhances the reactivity of the nitrided surface toward 02. In

particular, exposing the nitrided surface to 930 ML of pure 02 produces an atomic

oxygen coverage of 0.85 ML when the surface is held at 800 K, which appears to be a









saturation coverage for these oxidation conditions. In contrast, only 0.27 ML of oxygen

is deposited by exposing the nitrided surface to 02 at 300 K. The enhancement in

reactivity with increasing surface temperature is less pronounced when oxidizing with

oxygen atoms. For example, exposing the nitrided surface to the plasma-activated beam

for 60 minutes produces oxygen coverages of 1.2 and 1.5 ML when the surface is held at

300 and 800 K, respectively. This is only a 25% increase in oxygen uptake, which is

much lower than the 215% increase that is brought about by increasing the surface

temperature when 02 is used as the oxidant. Despite the small enhancement in oxygen

uptake with surface temperature, a higher oxygen coverage is still obtained by oxidizing

with gaseous O-atoms at 800 K compared with 02.

Similarities in the Ols spectra (not shown) indicate that similar chemical states of

oxygen are generated on the nitrided surface when oxidation is conducted using either

gaseous oxygen atoms or molecules at a surface temperature of 800 K. Thus, after

adsorption (or dissociation) the processes by which oxygen atoms incorporate into the

solid again appear to be independent of the identity of the gaseous oxidant, at least for the

low coverages considered. The Ols peak does shift to higher binding energy by about

0.2 eV as the oxygen coverage on the nitrided surface is increased from 0.85 to 1.5 ML.

However, this is a small difference in binding energy, considering that the oxygen

coverage is nearly doubled, so it may be concluded that the increase in oxygen coverage

from 0.85 to 1.5 ML causes only minor alterations to the chemical bonding environment

of oxygen atoms incorporated into the nitrided surface at 800 K.

Clean versus nitrided Si(100). Table 2-1 summarizes the oxygen coverages

obtained by oxidizing the nitrided and clean surfaces at the conditions indicated.









Examination of the coverages given in the table shows that the influence of surface

temperature on oxygen uptake is similar for the nitrided and clean Si(100) surfaces. For

example, a 930 ML 02 exposure to clean Si(100)-(2xl) generates oxygen coverages of

1.0 and 2.0 ML at surface temperatures of 300 and 800 K, respectively. These coverages

are in good agreement with previous studies [48,49], and demonstrate that oxygen uptake

by the clean surface is enhanced considerably by increasing the surface temperature when

02 is used as the oxidant, which is similar to the behavior found for the nitrided surface.

As also observed for the nitrided surface, increasing the surface temperature produces a

smaller increase in oxygen uptake by the clean surface when oxidizing with gaseous

oxygen atoms. Table 2-1 shows that oxygen coverages of 2.1 and 2.4 ML result after

exposing the clean surface to the plasma-activated beam for 60 minutes with the surface

temperature maintained at 300 and 800 K, respectively. In the work by Engstrom et al.

[48], it was found that atomic oxygen adsorption on the clean Si(100) surface remains

independent of the surface temperature up to oxygen coverages of 4 to 5 ML. Thus, the

relative insensitivity to surface temperature that we observed when oxidizing with the

plasma-activated beam is expected for the oxygen coverages examined (< 3 ML). We do

observe a small enhancement in uptake from the plasma-activated beam with increasing

surface temperature. However, since 02 is by far the majority beam component, this

small enhancement in uptake most likely reflects the influence of surface temperature on

02 incorporation.

Chemical state changes induced by oxidation at 800 K. Figures 2-8 to 2-10

display the Si2p, N1s and O1s spectra obtained after oxidizing the nitrided surface with

oxygen atoms at a surface temperature of 800 K, which produces an oxygen coverage of









1.5 ML. Figures 2-8 and 2-9 also show the Si2p and Nls spectra obtained from the

nitrided surface before oxidation. Figure 2-10 contains an Ols spectrum after depositing

2.4 ML of oxygen onto the clean Si(100) surface. Each spectrum has been normalized to

its respective peak height to augment the contrasting features. The most pronounced

spectral change following oxidation at 800 K is an increase in the intensity of the high

binding energy feature in the Si2p spectrum (Fig. 2-9) that is centered at about 102 eV

and extends to about 104 eV. The appearance of this feature indicates that Si+n (n > 0)

states are generated during oxidation of the nitrided surface. While similar results were

obtained following oxidation at 300 K (Figure 2-5), the intensity of the high BE Si2p

feature is clearly greater and the feature extends to higher BE when oxidation is

conducted at 800 K versus 300 K. The formation of a higher concentration of Si+n

species not only arises from the higher oxygen coverages that are obtained during high

temperature oxidation, but also from temperature dependent changes in the oxidation

process. For example, for nearly the same oxygen coverage, we find that oxidation at

800 K versus 300 K results in a greater amount of Si+2, Si+3 and Si+4 states. This

observation is consistent with the oxidation behavior of clean Si(100)-(2xl) [49]. At 300

K, oxidation occurs more uniformly across the surface, with the average Si oxidation

state increasing in proportion to the oxygen coverage. Increasing the surface temperature

enhances surface atom mobility and results in the formation of more highly oxidized

clusters at relatively low oxygen coverage. The incorporation of oxygen into oxidized

areas of the surface likely alleviates strain in the surface layers during oxidation.

Shown in Figure 2-9 are the NIs spectra obtained before and after oxidizing the

nitrided surface at 800 K to reach an oxygen coverage of 1.5 ML. After oxidation, the









Nls peak is slightly narrower and the peak maximum is shifted by only about 0.1 eV to

lower binding energy. Since these spectral changes are slight, it may be concluded that

the nitrogen bonding environment is altered negligibly during oxidation at 800 K, at least

when the oxygen coverage is increased up to 1.5 ML. Furthermore, the BE shift is in the

opposite direction to that observed when O and N atoms occupy second NN sites

[76,90,91], which suggests that the oxygen and nitrogen atoms in the film remain

segregated, even though oxidation at the elevated temperature enhances surface atom

mobility, as clearly evidenced by the formation Si+3 and Si+4 oxidation states at low

oxygen coverage (Fig. 2-8). Interestingly, the NIs BE shift is smaller than that observed

after oxidation at 300 K and the small feature at 399.2 eV is not evident in the Nls

spectrum. These observations suggest that segregation of the nitride and oxide phases

occurs to a greater extent at elevated surface temperature, with immeasurable formation

of Si2=N-O structures. Since the growth of highly oxidized SiOx clusters involves

substantial surface restructuring, it is conceivable that oxidation at the elevated

temperature enables the sub-surface nitride to adopt a more favorable structure in which

nitrogen atoms experience a more uniform bonding environment. Such a change may

explain the slight narrowing of the NIs peak observed after oxidation of the nitrided

surface at 800 K.

The Ols spectra obtained after oxidizing the clean and nitrided Si(100) surfaces at

800 K to oxygen coverages of 2.4 and 1.5 ML are remarkably similar (Fig. 2-10).

Indeed, the similarity between the Ols spectra indicates that the presence of a nitride in

the sub-surface of Si(100) has a negligible influence on the chemical states) of oxygen

that form during oxidation at 800 K. Thus, it appears that SiOx regions with similar









properties grow on clean and nitrided Si(100) at 800 K, despite the structural differences

of these surfaces. The similarity in the Ols spectra (Fig. 2-10) is also consistent with

enhanced segregation of the oxide and nitride phases when oxidation is conducted at

elevated surface temperature.

Angle resolved XPS data. Analysis of the XPS spectra obtained after oxidizing

the nitrided surface at 800 K provides evidence that the oxidized and nitrided regions

remain segregated. Angle-resolved XPS spectra provide additional support for this

interpretation. Figure 2-11 shows the Ols and NIs spectra obtained at electron take-off

angles of 00 and 600 with respect to the surface normal, after adsorbing 1.5 ML of

oxygen on the nitrided surface at 800 K. To compare the Ols/Nls intensity ratio as a

function of sampling depth, the spectra obtained at a given angle have been normalized

with respect to the NIs intensity at that angle. Only minor qualitative changes in the

spectra obtained at different take-off angles are observed. For example, the Ols peak

obtained at a 600 take off angle is shifted by only 0.1 eV to lower binding energy relative

to the Ols peak obtained at an emission angle of 00. Differences between the NIs spectra

obtained at these take-off angles are slight. Similar to the angle resolved data obtained

after oxidizing at 300 K, the Ols/N s ratio increases by about 10% when the collection

angle is adjusted to the more surface sensitive configuration. This observation confirms

that oxygen resides closer to the vacuum-solid interface than does the nitride region.

2.4 Discussion

The present results show that nitridation of Si(100) at elevated temperature reduces

the concentration of surface dangling bonds by nearly a factor of two and that the

reactivity of the surface toward both atomic and molecular oxygen decreases

significantly. Since the XPS results also reveal that the majority of oxygen and nitrogen









atoms do not directly interact within the films studied, but remain in nearly segregated

layers, the decrease in surface dangling bond concentration appears to be the primary

cause for the diminished activity of the nitrided surface. Physically, this conclusion

implies that both gaseous oxygen atoms and molecules adsorb predominantly, if not

exclusively, at dangling bond sites or pairs on the surface, and that the uptake of oxygen

by the nitrided surface is limited by the availability of such sites. That dangling bonds

are the active sites for 02 adsorption is not at all surprising. In fact, quantum chemical

calculations predict that the lowest energy pathway for 02 activation on Si(100)-(2x1)

involves the formation of a peroxy species across a surface dimer [89].

It is perhaps more surprising that gaseous oxygen atoms have such a strong

tendency to adsorb at dangling bond sites since this implies that insertion directly into Si-

Si bonds occurs to a negligible extent. While it is possible that an O atom incident from

the gas-phase must overcome an activation barrier to directly insert into a Si-Si bond, we

consider this possibility to be unlikely since formation of a Si-O-Si linkage is exothermic

by at least 6 eV. A propensity for oxygen atoms to adsorb at dangling bonds, rather than

to directly insert into Si-Si bonds, may be explained if we assume that the majority of

oxygen atoms in the beam exist in the ground 3P electronic state and then consider

electron spin effects. Because the 3P state is a triplet, direct O-atom insertion into a Si-Si

bond is spin-forbidden whereas adsorption at a dangling bond site is not. In this case, the

rate at which O-atoms from the beam directly insert into Si-Si bonds would be limited by

the rate of non-adiabatic curve crossing events that transform the electronic configuration

of the incident oxygen atom to a state such as the singlet 1D state for which direct

insertion is allowed. Such events are likely to be rare in a single gas-surface collision at









thermal impact energy. Thus, the observation of selective O-atom adsorption at surface

dangling bond sites suggests that the initial adsorption event tends to be electronically

adiabatic for the beam conditions employed.

A quantitative comparison of the uptake of oxygen by the clean and nitrided

Si(100) surfaces provides additional insight for understanding the role of dangling bonds

in the oxidation of these surfaces. The bottom row of Table 2-1 shows the oxygen

coverages obtained on the nitrided surface relative to that obtained on the clean surface

for various oxidizing conditions. As may be seen in the table, the oxygen coverages

obtained by exposing the nitrided surface to the atomic oxygen beam at surface

temperatures of 300 and 800 K are 57 and 62.5% lower than that obtained on the clean

surface. These values are remarkably close to the ratio of dangling bond concentrations

on the nitrided and clean surfaces (59%), and provide quantitative evidence that gaseous

O-atoms adsorb preferentially on surface dangling bonds on both surfaces. This

comparison is even more favorable when considering that the contribution of 02 to the

uptake achieved during the plasma-activated beam exposure is more significant at a

surface temperature of 800 K. Interestingly, for oxidation with 02, the maximum oxygen

coverage obtained on the nitrided surface at 300 K is only 27% of that obtained on the

clean surface (Table 2-1). This value is less than half of the ratio of dangling bond

concentrations on the nitrided versus clean surfaces. Assuming that an 02 molecule does

dissociate across a single dimer on the clean Si(100)-(2xl) surface, as predicted by

electronic structure calculations [89], this comparison suggests that at least two dimers

are consumed when a single oxygen molecule dissociates and the oxygen atoms

incorporate into the nitrided surface at 300 K. While it is difficult to envision four









dangling bonds being required to activate one 02 molecule, it is conceivable that dangling

bond pairs could be arranged on the nitrided surface in such a way that 02 activation on

one pair could render a neighboring pair unable to readily activate a second 02 molecule.

The uptake of oxygen on the nitrided surface increases to 44% of that on the clean

surface when oxidation is conducted at 800 K using 02, which may indicate that fewer

dangling bond pairs are consumed or are more efficiently regenerated on the nitrided

compared with the clean surface during oxidation at elevated temperature. Overall, these

observations suggest that oxidation with gaseous oxygen atoms occurs by a similar

mechanism on the clean and nitrided surfaces, with the main difference being that fewer

adsorption sites are available on the nitrided surface. In contrast, the mechanism for 02

dissociative chemisorption and oxygen incorporation appears to be more sensitive to

structural differences between the nitrided and clean surfaces.

Increasing the surface temperature enhances the uptake of 02 on both the clean and

nitrided surfaces, but produces only a small increase in the uptake of gaseous O-atoms.

High surface temperatures are thought to facilitate the oxidation of clean Si(100) by 02

by promoting oxygen penetration into the subsurface layers [49]. Such penetration is

likely to regenerate dangling bond sites at the surface that are needed to activate 02

molecules, thereby restoring the surface activity toward 02 dissociation. It is noted that a

molecular beam study by Ferguson et al. [92] shows that the dissociation probability of

02 on Si(100) is only weakly dependent on the surface temperature at low gas-

temperatures. Thus, more facile regeneration of active surface sites is the more likely

explanation for the enhancement in oxygen uptake with surface temperature than would

be promotion of 02 bond cleavage at higher surface temperature.









The relative insensitivity to surface temperature in the uptake of gaseous O-atoms

was first observed by Engstrom et al. [48] and was quite reasonably interpreted by those

authors as evidence that oxygen atoms incident from the gas-phase insert readily into Si-

Si bonds. The uptake of gaseous O-atoms was found to increase with surface

temperature only at oxygen coverages greater than about 4-5 ML, which corresponds to

oxygen atoms inserted into all of the Si-Si bonds that are directly accessible from the gas-

phase. However, the findings from the current study indicate that surface dangling bonds

are the preferred adsorption site for a gaseous O-atom, and that direct insertion into a Si-

Si bond occurs to a negligible extent. Considering this finding, it is difficult to

understand why an increase in the surface temperature effects only a small enhancement

in the uptake of gaseous oxygen atoms. In particular, if more effective regeneration of

surface dangling bond sites is the primary reason that an increase in surface temperature

enhances 02 uptake, then it is reasonable to expect that the uptake of gaseous oxygen

atoms would also be promoted by raising the surface temperature since O-atoms also

adsorb selectively at dangling bond sites and more of these sites should be available at

high surface temperature.

A recent computational investigation by Widjaja and Musgrave [89] may offer a

plausible explanation for understanding the different effects of surface temperature in the

oxidation of Si(100) with gaseous O atoms versus 02. The top panel of Figure 2-12

shows a schematic of key structures and the associated energy changes that were

predicted to occur by those authors when 02 adsorbs and then dissociates on the Si(100)-

(2x1) surface [89]. It is important to note that the molecular representations shown in

Figures 2-12 and 2-13 are only intended to depict the steps in the proposed model, and do









not precisely illustrate the bond lengths and angles for these structures as predicted by

density functional theory. Following adsorption, the 02 molecule is predicted to span the

dimer to form a peroxy-like species that then dissociates to produce an oxygen atom

inserted across the dimer and a siloxy radical. The oxygen atom of the siloxy radical then

inserts into a Si-Si backbond, resulting in the final structure shown in the top panel.

Clearly, the formation of a second peroxy species on the final structure would be

significantly hindered by the presence of the O-atom bridging the dimer. Thus, if the

oxygen atoms in this final structure have limited mobility, then effectively only one 02

molecule can dissociate for each dangling bond pair on the surface. Notice that this

situation would result in an oxygen coverage of 1.0 ML on the Si(100)-(2xl) surface, and

may help to explain the substantial reduction in oxygen uptake that occurs at 1.0 ML

when the clean surface is exposed to 02 at 300 K.

The bottom panel of Figure 2-12 illustrates elementary steps by which the bridging

oxygen atom could migrate to a backbond site. These reactions have not been explored

computationally as far as we know. The first step in the scheme shows the formation of a

siloxy radical by cleavage of an Si-O bond of the bridging oxygen species, and the

second step involves oxygen insertion into a Si-Si backbond. This migration process

regenerates an empty dimer, and would thereby enable a second 02 molecule to bind in

the peroxy configuration. Although energy barriers for these steps have not been

explicitly predicted, the results ofWidjaja and Musgrave suggest that the first step,

production of the siloxy radical, should have the larger energy barrier. This barrier may

be comparable to the 1.38 eV barrier required for the reverse of the final reaction shown

in the top panel of Figure 2-12. Considering the large energy barrier, the migration of the









bridging oxygen atom to a backbond site should be promoted significantly by raising the

surface temperature. Thus, according to this mechanism, oxygen uptake by 02

dissociation is enhanced at elevated surface temperature since the population of empty

dimers increases with increasing surface temperature.

Shown in Figure 2-13 are pathways proposed for the incorporation of a gaseous

oxygen atom into the Si(100) surface. Based on the present results, the O-atom is

assumed to adsorb initially on a dangling bond site to form a siloxy radical. From this

site, the oxygen atom can insert either across the surface dimer or into a Si-Si backbond

to form the structures shown in the figure. The energy changes illustrated in this figure

were also taken from the work of Widjaja and Musgrave, and show only slight

differences in the energetic of these insertion pathways. Since an O-atom adsorbs at a

single dangling bond site, and therefore does not have the strict steric requirements for

adsorption as does 02, we speculate that a second O-atom will adsorb with roughly equal

probability on each of the one O-atom structures shown in Figure 2-13. Thus, according

to this interpretation, an increase in surface temperature has only a minor influence on the

uptake of gaseous oxygen atoms because enhanced oxygen migration to sub-surface sites

does not significantly affect the availability of single dangling bond sites at the surface.

2.5 Conclusions

We have investigated the nitridation of Si(100) and the subsequent oxidation of this

surface by both gaseous atomic and molecular oxygen under UHV conditions.

Nitridation of Si(100) by the thermal decomposition of ammonia at 900 K results in the

formation of a subsurface nitride and a decrease in the concentration of surface dangling

bond sites. Based on changes in NIs spectra after NH3 adsorption and decomposition, we

estimate that the nitride resides four to five layers below the vacuum-solid interface and









that the concentration of dangling bonds on the nitrided surface is about 0.59 ML or 59%

of that on the clean surface. Oxidation of the nitrided surface at surface temperatures of

300 and 800 K produces an oxide phase that resides in the outer surface layers and

remains largely segregated from the subsurface nitride for oxygen coverages up to about

2.5 ML. At a surface temperature of 300 K, the incorporation of about 1 ML of oxygen

into the near surface layers alters the nitrogen bonding environment only slightly, most

likely by introducing strain in the subsurface nitride, and the Nis spectra indicate that a

small quantity of Si2=N-O also forms. At 800 K, the nitride bonding environment

changes negligibly for oxygen coverages as high as 2.5 ML, which is consistent with

greater segregation of the nitride and oxide phases and enhanced structural relaxation in

these phases. In addition, at a given oxygen coverage, the quantity of Si3+ and Si4+ states

that are detected increases when oxidation is conducted with the surface held at 800 K

versus 300 K, indicating a tendency for regions of high local oxygen concentration to

form at elevated temperature.

The reactivity of Si(100) toward both atomic and molecular oxygen decreases

significantly after nitridation of the subsurface region due to the decrease in surface

dangling bond concentration that accompanies nitride growth. Quantitative support for

this conclusion is given by the observation that, for the same exposure to gaseous oxygen

atoms, the oxygen coverage obtained on the nitrided surface relative to that on clean

Si(100) is within 5% of the ratio of dangling bond concentrations on these surfaces. This

finding also provides strong evidence that gaseous 0(3P) atoms adsorb initially at

dangling bond sites on these surfaces, and that direct insertion into Si-Si bonds occurs to

a negligible extent. An increase in surface temperature is found to significantly enhance






72


oxygen uptake by the nitrided surface when 02 is used as the oxidant, but brings about

only a slight increase in uptake when gaseous oxygen atoms are employed. It is proposed

that an increase in surface temperature promotes oxygen migration to the subsurface, and

thereby results in more effective regeneration of empty dimers. Since the activation of an

02 molecule on the Si(100) surface has more stringent steric requirements than does O-

atom adsorption, the facile penetration of oxygen to the subsurface at high temperature

has a greater influence on the adsorption of 02 than O.









Table 2-1. Oxygen coverages on clean and nitrided Si(100)
Atomic Oxygen


300 K


800 K


30


Molecular Oxygen
0 K 800 K


[O] (nitrided)a 1.2 1.5 0.27 0.85
[O] (clean) a 2.1 2.4 1.0 2.0
nitrided/clean 0.57 0.62 0.27 0.44
a The oxygen coverages in these rows are given in units of ML, as defined in the text, and
were obtained after exposing the surface to the oxidant beam for 60 minutes at the
surface temperatures indicated. A 60 min exposure corresponds to an 02 fluence of 930
ML when the plasma power is disabled, and to -928 ML 02 and -3 ML O when the
plasma is activated.


105 104 103 102 101 100 99 98 97 96 95
Binding Energy (eV)


Figure 2-1. Si2p spectrum obtained from a Si(100) surface after a 160 ML NH3 exposure
at a surface temperature of 900 K.






















Z








402 401 400 399 398 397 396 395 394 393
Binding Energy (eV)



Figure 2-2. Nls spectrum obtained from a Si(100) surface after a 160 ML NH3 exposure
at a surface temperature of 900 K.














































402 401 400


399 398 397 396 395 394 393


Binding Energy (eV)


Nls spectra obtained from Si(100). A) After a 160 ML NH3 exposure at 300
K. B) A subsequent anneal to 900 K for 5 minutes. C) A 160 ML exposure
at a surface temperature of 300 K to the surface generated in B. The relative
integrated areas of each N Is spectrum is shown in each panel.


Figure 2-3.































104 102 100 98 96
Binding Energy (eV)


Si2p spectra obtained after exposing clean Si(100) to 160 ML NH3 at a
surface temperature of 900 K (dashed line), and after depositing 1.2 ML of
oxygen on the nitrided Si(100) surface held at 300 K using the plasma-
activated beam (solid line).


Figure 2-4.































402 401 400 399 398 397 396 395 394 393
Binding Energy (eV)



Figure 2-5. Nls spectra obtained after exposing Si(100) to 160 ML NH3 at a surface
temperature of 900 K (dashed line), and after depositing 1.2 ML of oxygen
on the nitrided Si(100) surface held at 300 K using the plasma-activated
beam (solid line).

































536 535 534 533 532 531 530 528 527
Binding Energy (eV)



Figure 2-6. Ols spectra obtained after incorporating 1 ML of oxygen atoms on Si(100) at
a surface temperature of 300 K (dashed line), and after depositing 1.2 ML of
oxygen on the nitrided Si(100) surface held at 300 K using the plasma-
activated beam (solid line).
1
















activated beam (solid line).

















0 Is Al Ka


S60"
60"


40 I I I I I6 5 I 2 5 0 52' I
402 400 398 396 394 536 534 532 530 528


Binding Energy (eV)


Binding Energy (eV)


N1s and Ols spectra obtained at electron collection angles of 00 (dashed line)
and 600 (solid line) after depositing 1.2 ML of oxygen on nitrided Si(100) at
a surface temperature of 300 K. The Ols and NIs peak heights have been
normalized to the NIs peak height at the respective angle


Figure 2-7.
































106 104 102 100 98 96 94
Binding Energy (eV)


Figure 2-8. Si2p spectra obtained after exposing clean Si(100) to 160 ML NH3 at a
surface temperature of 900 K (dashed line), and after depositing 1.5 ML of
oxygen on the nitrided Si(100) surface held at 800 K using the plasma-
activated beam (solid line).

































402 401 400 399 398 397 396 395 394 393
Binding Energy (eV)


Nls spectra obtained after exposing Si(100) to 160 ML NH3 at a surface
temperature of 900 K (dashed line), and after depositing 1.5 ML of oxygen
on the nitrided Si(100) surface held at 800 K using the plasma-activated
beam (solid line).


Figure 2-9.
































536 535 534 533 532 531 530 529 528 527
Binding Energy (eV)



Figure 2-10. Ols spectra obtained after incorporating 2.4 ML of oxygen atoms on
Si(100) at a surface temperature of 800 K (dashed line), and after
depositing 1.5 ML of oxygen on the nitrided Si(100) surface held at 800 K
using the plasma-activated beam (solid line).















1.0 IN IS Al KI
O Is Al Ka

0.8 i


0.6 -----
S 600

0.4


0.2


0.0
l I ll 1I I I I I ,
402 400 398 396 394 536 534 532 530 528
Binding Energy (eV) Binding Energy (eV)



Figure 2-11. NIs and Ols spectra obtained at electron collection angles of 00 (dashed
line) and 600 (solid line) after depositing 1.5 ML of oxygen on nitrided
Si(100) at a surface temperature of 800 K. The Ols and NIs peak heights
have been normalized to the NIs peak height at the respective angle.












Adsorption and Insertion


E* = 1.5 eV


AE = -2.84 eV


Incorporation and Site Regeneration



SHighT + 2





Figure 2-12. Model for 02 dissociation and incorporation into Si(100). The top panel
shows the structures and energetic for the dissociative chemisorption of 02
on Si(100) as predicted by DFT calculations [89]. The bottom panel shows
possible elementary steps for oxygen migration to the subsurface that
results in the regeneration of an empty dimer.


E* = 0.55 eV
-----


-2.81 eV


-0.83 eV










+0


) 8
'0-^


-1.91 eV


IHigh T
S+0


AE = -4.49 eV


-2.18 eV

Figure 2-13. Model for O-atom adsorption and incorporation into Si(100). The energy
changes, where indicated, were predicted using DFT calculations as reported
in reference 89.














CHAPTER 3
DYNAMIC DISPLACEMENT AND DISSOCIATION OF 02 ON Pt(1 11) BY ATOMIC
OXYGEN

The role of surface temperature on the dynamic displacement of O102 from Pt(l 11)

stimulated by the adsorption of 160 atoms was investigated. The maximum displacement

rate increased with surface temperature, while the desorption yield decreased with

temperature. Heteronuclear product evolution (180160) from the surface was below the

experimental detection limits (< 10% of the 1802 product evolution). The adsorption of

160 -atoms also induced the dissociation of pre-adsorbed 1802 molecules at low surface

temperatures. The initial displacement rates with a 160-atom flux of 0.005 ML/sec were

0.0024, 0.0025 and 0.0027 ML/sec at 90, 100 and 110 K, respectively. Dissociation of

102 was found to only occur during the first 0.2 ML fluence of 160-atoms at which the

total atomic oxygen coverage was about 0.44 ML. Estimates for the initial dissociation

rate for O102 at 90, 100 and 110 K and with the same 160-atom flux was estimated of

0.0015, 0.0020 and 0.0030 ML/sec, respectively.

3.1 Introduction

Molecular oxygen displacement from Pt( 11) by incident oxygen, nitrogen, and

hydrogen atoms at 80 K was first reported by Rettner and Lee [17]. By adsorbing 1802

onto Pt(1 11) and subsequently exposing this surface to a beam of 160-atoms at 80 K, they

observed 1802 displacement from the surface. When exposing this surface to 160-atoms,

they also observed the formation of 160180 products. The displacement phenomenon was

found to be independent of the adsorbing species, indicating that the desorption is not









collisionally induced. Using time-of-flight mass spectrometry, Rettner and Lee observed

that 102 desorbed with a bimodal energy distribution. This indicates that desorption

occurs through two distinct channels. The lower energy component is consistent with

thermal desorption, whereas the high energy feature indicates that a fraction of the

adsorption energy of the incoming 160-atom is transferred into the 1802 stimulating

desorption.

Similar experiments were performed by Wheeler et al. [16]. Using a supersonic

160-atom beam directed towards a Pt(1 11) surface saturated with 1802 at 77 K, they

investigated the role of translational kinetic energy and incident angle on the initial

molecular oxygen displacement rate. They observed both 1802 and 10160 evolving from

the surface, and report that the initial probability of forming the mixed isotope species is

-16.5 % of the total 02 (1802 and 10160) displacement probability. These investigations

also found that the displacement probability of 1802 increased with the initial translational

energy and decreased at glancing incident angles of the 160-atom beam. The probability

of forming the mixed isotope product appeared to weakly depend on the 160-atom energy

and incidence angle.

From Rettner and Lee's work it seems apparent that both thermal and nonthermal

mechanisms govern the displacement of 1802 during the 160-atom adsorption on Pt(1 11).

During the 160-atom exposure, the surface concentration of 160-atoms is increasing. It is

known that the presence of adsorbed oxygen atoms weakens the 02-Pt bond [12,16],

which could be responsible for the observed thermal desorption of 1802. Dynamic

displacement is not fully understood, so it is reasonable that varying the surface

temperature would provide additional information into the dynamic displacement









process. In this study, real-time reaction product monitoring and subsequent temperature

programmed desorption (TPD) measurements were employed to examine the role of

surface temperature on 802 dynamic displacement stimulated by gaseous 160-atoms.

3.2 Experimental Methods

These experiments were conducted in a three-level UHV chamber described in

Section 1.3, with a brief description provided here. This chamber is equipped to perform

X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES), low

energy ion scattering spectroscopy (LEISS), reaction product monitoring, and TPD. The

first plane contains a dual Al and Mg anode X-ray source, a variable energy electron

source, a variable energy ion source, and a hemispherical charged particle analyzer. The

second plane includes a calibrated molecular beam doser and a leak valve controlled

molecular beam doser. The final plane contains a quadrupole mass spectrometer, a

microwave plasma source mounted in a beam chamber with two stages of differential

pumping, and a LEED optics. A microwave plasma source was employed to decompose

molecular oxygen using 2.45 GHz microwave radiation. The end of the plasma source

cavity is capped by an alumina faceplate with 5 laser drilled 0.4 mm through holes, which

collimated the atom beam. A pair of oppositely charged plates (+ 10 kV/cm) is located at

either side of the beam to remove charged particles from the beam path. The beam flows

from the first stage to the second through a 3 mm skimmer, spaced 15 mm from the

alumina collimating plate. This second stage is pumped via a 66 1/s turbo molecular

pump and a liquid nitrogen cooled titanium sublimation pump. Communication from the

second stage to the UHV chamber occurs through a quartz tube with a diameter and

length of 6 mm and 60 mm, respectively.