Article pubs.acs.org/JPCC
Synchrotron Radiation Photoelectron Spectroscopy Study on Oxide Evolution during Oxidation of a Si(111)‑7 × 7 Surface at 300 K: Comparison of Thermal Equilibrium Gas and Supersonic Molecular Beams for Oxygen Adsorption Akitaka Yoshigoe* and Yuden Teraoka Quantum Beam Science Directorate, Japan Atomic Energy Agency, 1-1-1 Kouto, Sayo-gun, Sayo-cho, Hyogo 679-5148, Japan ABSTRACT: The translational energy of gas molecules is a fundamental physical quantity which dominates adsorption processes in gas−solid reactions, and for this reason molecular beam experiments have been widely conducted. Although gas in thermal equilibrium (thermal gas) plays an important role in many beneficial chemical reactions supporting the most advanced technologies, understanding the adsorption processes of thermal gas in terms of translational energy remains an unsolved issue. Here, for thermal-O2 and supersonic O2 molecular beams, we present a comparative study of oxide evolution at early stages of oxidation of a Si(111)-7 × 7 surface at room temperature. Real-time observations of oxides at the surface were conducted by employing high energy resolution photoelectron spectroscopy using high brilliance synchrotron radiation. It was found that ins structure which has one oxygen atom at the backbond of a Si adatom is the common first product for oxidation by thermal-O2 and supersonic O2 molecular beams. Similarities between thermal-O2 and supersonic O2 molecular beams were found for oxide evolution and illustrate the notion that translational energy of thermal O2 can be ascribed to the average molecular kinetic energy defined based on the most probable speed of the Maxwell−Boltzmann distribution. This experimental result suggests that translational energy is the unifying reaction parameter that rationalizes adsorption mechanisms for both thermal gas and supersonic molecular beams.
1. INTRODUCTION
Experimentally, molecular beams have been widely used to control the translational energy of gas molecules in the research of adsorption mechanisms.8−12 For example, Rettner and Mullins reported systematic and fruitful experiments using a molecular beams scattering method on oxygen adsorption at a Pt(111) surface, which is an important part of catalytic reactions.13 They showed typical features of initial sticking probabilities as a function of incident translational energy and suggested two types of oxygen adsorption processes. One is a precursor-mediated mechanism, and the other is a direct dissociation process without passing through precursor states. This systematic and fruitful experiment using molecular beam scattering method has become a fundamental approach to clarify dynamical aspects of adsorption mechanisms on many solid surfaces in subsequent work. Thereafter, Nolan et al. succeeded in clarifying the translational energy selection of precursor states to oxygen adsorption on a Pt(111) surface.14,15 They confirmed the presence of the molecular precursor state and its dependence on translational energy by in situ
The oxidation of silicon surfaces is attracting considerable importance in the formation of ultrathin oxide films because atomically controlled interfaces between the oxides and the substrate are required in the fabrication of metal-oxidesemiconductor field-effect transistors (MOS-FETs) due to ceaseless shrinkage of microelectronics. Apart from its technological interest, this reaction is scientifically interesting as a model system to investigate the oxygen adsorption at solid surfaces.1−3 Generally, it is well-known that adsorption is the first step in the chemical reaction between gas molecules and solid surfaces. This fundamental phenomenon is widely observed in many profitable chemical reactions supporting the most advanced technologies, such as those in not only the semiconductor industry but also catalytic chemical engineering.4 Because adsorption is essential in the interactions of gas molecules with solid surfaces, the translational energy associated with gas molecules is commonly accepted as a fundamental physical quantity to investigate dynamic aspects of adsorption processes.5−7 Therefore, there is considerable interest in the dependence of the initial sticking probability of gas molecules on translational energy for a number of years. © 2014 American Chemical Society
Received: November 3, 2013 Revised: March 27, 2014 Published: April 28, 2014 9436
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
The Journal of Physical Chemistry C
Article
observation using electron energy-loss spectroscopy (EELS). These results clearly demonstrated the importance of molecular beam experiments to study gas adsorption dynamics at solid surfaces. Concerning oxygen adsorption onto silicon surfaces, many experimental works using molecular beam have been reported.16−18 Generally, the same trends of initial sticking probability as a function of translational energy have been observed for oxygen adsorption onto Si(100) and Si(111) faces. It is widely accepted that oxygen adsorption mechanism on silicon surfaces has been categorized into two types. Oxygen molecules with translational energies higher than ca. 0.06 eV react with the surface via a direct adsorption process: a dissociated chemisorption of oxygen molecule directly occurs without passing through precursor states on the surface. In contrast, low-energy oxygen below ca. 0.06 eV adsorbs via the precursor-mediated adsorption process: oxygen gas is initially trapped into a physisorption state and then the physisorbed oxygen molecule converts to molecularly chemisorbed oxygen followed by a dissociated chemisorption state. Contrary to the foregoing advantages of molecular beam experiments to clarify adsorption mechanisms, it is generally recognized that a gas in thermal equilibrium (thermal gas) described by the Maxwell−Boltzmann (MB) distribution plays a key role in many beneficial reaction systems.19,20 Despite the importance of thermal-gas adsorption processes, there has been little discussion on the relationship between thermal gas and molecular beams in terms of translational energy. In our previous study on the initial sticking probability of oxygen molecule at the Si(111)-7 × 7 surface, we found that thermalO2 adsorption takes place via the trapping-mediated process as well as the O2 beams below 0.06 eV.18 This result strongly suggests that there has been some common physical description regarding the adsorption mechanism between thermal-O2 and low-energy oxygen molecular beams. Although initial sticking probability data have established consensus for O2 adsorption mechanisms depending on incident oxygen conditions, such adsorption models raise a fundamental question about the relationship between the initial sticking probability and the initial oxides. Most notably, concern about the similarity or distinctiveness for initial oxides with regard to adsorption mechanisms remains unresolved. It is well-known that core-level photoelectron spectroscopy is a powerful tool to obtain detailed information on the chemical environment of oxides at solid surfaces.21−25 Figure 1 depicts the models of oxide structures discussed in this work. There are several positions corresponding to oxygen bonding configurations named as ins, ad, tri, and paul oxygen. Here, oxide structures are described by using italic notation in this work. For instance, the oxide is named ins×2 structure, when two ins oxygen atoms exit at the silicon atom. The coordination number of the silicon atom is denoted by the term n (n, integer), referring to the Si oxidation states (Sin+). Previously, we have demonstrated the usefulness of real-time X-ray photoelectron spectroscopy using synchrotron radiation (SR-XPS) to reveal oxide evolution at the early stages of thermal-O2 oxidation at Si(111)-7 × 7 surfaces.21 The O 1s and Si 2p core-level spectra as a function of oxygen exposures have elucidated adsorbed oxygen amounts, oxygen bonding configurations, and Si oxidation states. This experimental finding is particularly interesting as the result of oxide evolution, clarifying the first product corresponding to the initial sticking probability data. Additionally, we believe that the oxide
Figure 1. Ball and stick illustrations of the oxide structures discussed in this work. An oxygen atom between a Si−Si bond is called an ins oxygen. An oxygen atom at the on-top site is called an ad oxygen. An interstitial oxygen atom is called a tri oxygen. The paul oxygen is a molecular oxygen. The term n of n+ is a coordination number of the silicon atom.
evolution may be influenced according to O2 adsorption mechanisms as revealed by molecular beam experiments.16−18 However, there is a lack of comparison of oxide evolution in the oxidation by thermal-O2 and O2 molecular beams. Therefore, a unified physical picture for oxygen adsorption processes with regard to thermal-O2 and molecular beam oxidation still lacks proper development. In this work, we present a comparative analysis of thermal-O2 and supersonic O2 molecular beams, focusing on the oxide evolution in the early stages of oxidation of a Si(111)-7 × 7 surface at room temperature. To this end, we applied real-time photoelectron spectroscopy using high brilliance and high energy resolution synchrotron radiation to obtain measurements of the oxygen bonding configurations and Si oxidation states as a function of exposure dose. We observed a significant relationship between initial sticking probability and oxide structures for each oxidation condition, implying that it might be possible to rationalize the classification of the adsorption mechanisms for thermal-O2 and supersonic O2 molecular beams in terms of the translational energy of O2.
2. EXPERIMENTAL DETAILS Experiments on in situ observations during oxidation at a Si(111)-7 × 7 surface were conducted with the surface reaction analysis apparatus (SUREAC2000) constructed at the soft Xray beamline (BL23SU) of the synchrotron radiation facility, SPring-8, Hyogo, Japan. This experimental apparatus has the potential to measure real-time photoelectron spectroscopy using high brilliance synchrotron radiation under the exposure of not only thermal-O2 but also supersonic O2 molecular beams.18,21,26 All experiments were conducted using an ultrahigh-vacuum apparatus for which the base pressure is constantly kept below 2 × 10−8 Pa. 2.1. Surface Preparation and Characterization. We prepared a clean and well-defined Si(111)-7 × 7 surface according to our well-established method: wet chemical treatments, followed by annealing of the substrate under an ultrahigh-vacuum condition.18,21,26 An n-type Si(111) substrate (0.5−10 Ω cm) with thickness less than 500 μm was initially cleaned by a chemical solution. The sample was then annealed in an ultrahigh-vacuum chamber (1.0 × 10−8 Pa) at ca. 600 K, 9437
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
The Journal of Physical Chemistry C
Article
followed by flash-annealing at a temperature over 1420 K for a few seconds under a pressure lower than 1.0 × 10−6 Pa. The sample was cooled to 300 K within 30 min using a liquid nitrogen cooling system to prevent oxidation from residual water in the chamber. The surface temperature was measured with a W/Re thermocouple placed behind the sample. All samples were thoroughly checked for cleanliness and surface structures prior to oxidation. After following these sample preparations carefully, the effects from both contamination and structure morphology of the sample surface on oxidation are completely eliminated. We confirmed the surface characteristics from the sharpness in the low-energy electron diffraction (LEED) pattern of the 7 × 7 surface structure. In addition to this characterization, we also checked the absence of contamination by highly sensitive photoelectron spectroscopy using synchrotron radiation. Furthermore, we also confirmed the surface components, such as the rest of the atoms 7 × 7 structure,21,26 in the Si 2p photoelectron spectrum. 2.2. Oxidation Methods. In this study, we used two methods for room-temperature oxidation at a Si(111)-7 × 7 surface: (i) the surface is exposed to pure O2 gas (ca. 99.9995%) at 5.2 × 10−7 Pa (1.4 × 1012 molecules cm−2 s−1) through a variable leak valve, and (ii) the surface is irradiated by supersonic O2 molecular beams. Supersonic O2 molecular beams were generated using a temperature-controlled pyrolitic boron nitride (PBN) nozzle of diameter 100 μm. In addition to nozzle-temperature control (300 or 1400 K),27 a seeding technique using rarefied O2 gas with He gas, and further mixing with and without Ar, was employed to change the translational energy of O2 from 0.06 to 2.33 eV.28 A typical O2 flux density was of the order of 1014 molecules cm−2 s−1, which was evaluated from the partial pressure measurements of O2, He, and Ar gases, respectively. Each partial pressure is measured using the quadrupole mass analyzer equipped along the supersonic O2 molecular beam axis. The angle between the molecular beam axis and surface normal is 10°. We used a liquid nitrogen cryopump during the experiments to prevent oxidation by residual water in the chamber. 2.3. Time-Resolved Photoelectron Spectroscopy Measurement. We performed continuous photoelectron measurements during oxidation using synchrotron radiation beams with a photon energy of 669.1 eV. This photon energy makes it possible to carry out simultaneous measurements of both O 1s and Si 2p photoelectrons with high energy resolutions and surface-sensitive measurement for O 1s at BL23SU. The beam size of the synchrotron radiation was lower than 0.02 mm2 at the sample position. The total energy resolution through the electron energy analyzer was below 0.25 eV. The takeoff angle of the photoelectrons was 60°. High energy resolution photoelectron spectra for the Si 2p and O 1s orbitals during oxidation were alternately recorded with accumulation times of 18 and 22 s, respectively. Here, we note that no difference between oxidation with and without continuous illumination of synchrotron radiation light was observed. Thus, we are convinced that the influence of high-flux photon irradiation for oxygen dissociation is negligibly small for the growth of oxides. Furthermore, the oxide decomposition by irradiation with synchrotron radiation is almost negligible because no changes in the profiles for O 1s and Si 2p spectra were observed after synchrotron radiation light illumination in the absence of O2 gas.
3. RESULTS AND DISCUSSION Figure 2 describes the photoelectron spectra of O 1s and Si 2p recorded during the oxidation by supersonic O2 molecular
Figure 2. Real-time photoelectron spectra for O 1s and Si 2p during oxidation by a 0.06 eV molecular beam. The photon energy of synchrotron radiation is 669.1 eV.
beams with 0.06 eV. Energy values indicated in this work represent the surface-normal component of the translational energy of the incident O2.29 Dose values were obtained by multiplying the duration of O2 exposure by the O2 flux density. The sequence O 1s spectra clearly show the increment of the amount of adsorbed oxygen with increasing O2 exposure dose and suggest the growth of several components assigned to oxygen bonding configurations. In contrast, it is found that the components at the higher binding energy side of the bulk component in Si 2p spectra appear with the growth of O 1s spectra. These results indicate the typical spectral changes depending on the exposure dose as already observed for thermal-O2 oxidation.21 To obtain the detailed information on oxygen bonding configurations and Si oxidation states, we carefully conducted curve-fitting analyses for the Si 2p and O 1s photoelectron spectra according to our methods.21,26 Figure 3 shows the comparison of curve-fitting results on O 1s and Si 2p photoelectron spectra depending on the oxidation conditions. Prior to the curve-fitting analysis, the background of all spectra was initially subtracted using the Shirley method. The Si 2p spectra were deconvoluted to decompose into Si 2p1/2 and Si 2p3/2 lines prior to the curve-fitting analysis. The spin−orbit splitting value between Si 2p1/2 and Si 2p3/2 is set to 0.61 eV, and the intensity ratio of Si 2p1/2 to Si 2p3/2 is set to 1/ 1.98. We used the Voigt function, a Lorentzian function with full width at half-maximum (fwhm) set at 0.1 eV. From these precise curve-fitting analyses, we resolved the oxide-related components located at 1.00 (Si1+), 1.80 (Si2+), 2.70 (Si3+), and 3.75 eV (Si4+) relative to the bulk component. The fwhm of the Gaussian function for each component was determined to be 9438
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
The Journal of Physical Chemistry C
Article
located at −4.1, −1.3, and +3.2 eV are attributed to the paul oxygen, and the sum intensity of these three components is taken as the intensity for the paul oxygen in this work. Furthermore, we assume that each oxygen atom, ins, ad, tri, and paul, has the same initial- and final-state effects on the chemical shift for the Si 2p core-level photoelectron spectra. In this study, we ascribed peak areas obtained by curve fitting procedures to signal intensities for each component. In practice, we quantitatively evaluated the adsorbed oxygen amount: density and coverage at the oxidized surface. Details of the evaluation method have already been described in our previous papers.18,21,26 The above-mentioned combination of in situ photoelectron spectroscopy measurements with precise curve-fitting analysis enabled accurate and reliable data depending on oxidation time to be assessed for oxygen bonding configurations and Si oxidation states, as already reported.21 Figure 4 shows the plots of oxygen uptake curves (top), the evolution of oxygen bonding configurations (middle), and the evolution of Si oxidation states (bottom) of oxides as a function of O2 exposure dose for room-temperature oxidation of a Si(111)-7 × 7 surface by thermal-O2 and supersonic O2 molecular beams. Energy values presented at the top of panels b−d represent the surface-normal component of the translational energy of the incident O2. The horizontal axes of all panels describe dose values. The uptake curves were obtained by evaluating the O 1s peak area. Here, we note that the dose range for each molecular beam datum is 102 larger than that for thermal O2. It is found that the curve feature for not only oxygen uptake but also each component strongly varies depending on oxidation conditions. As suggested in our previous study, the difference of uptake curves, especially the initial slope, reflects the results for different initial sticking probabilities.18 The data in Figure 4 obtained by real-time observation of Si 2p and O 1s
Figure 3. Comparison of curve-fitting results on O 1s and Si 2p photoelectron spectra depending on oxidation conditions.
0.41 ± 0.02 (bulk), 0.55 ± 0.02 (Si1+), 0.79 ± 0.02 (Si2+), 1.05 ± 0.1 (Si3+), and 1.05 ± 0.1 (Si4+). For this work, we illustrated the evolution of these four kinds of Si oxidation states. Here, we note that the signal located at the Si1+ energy position at zero dose is not assigned to the Si oxidation state because no signal for O 1s was completely detected.30 Thus, all Si1+ intensities are subtracted by this intensity at zero dose. We also resolved the O 1s line profiles into six components using the Voigt functiona Lorentzian function with fwhm of 0.2 eVand a Gaussian function with fwhm of about 0.95 ± 0.05 eV. The binding energy of the ins oxygen component which can be observed from the initial stages of oxidation is assigned to 531.3 eV. The ad oxygen atom and the tri oxygen atom are resolved at +1.5 and +0.6 eV, respectively, with respect to the ins oxygen component. The three components
Figure 4. Plots of oxygen uptake curves (top), evolution of oxygen bonding configurations (middle), and Si oxidation states (bottom) as a function of O2 dose for room-temperature oxidation of a Si(111)-7 × 7 surface by thermal-O2 and supersonic O2 molecular beams. The energies appearing above panels b−d are values of the surface-normal component of the translational energy of the incident O2 beams. 9439
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
The Journal of Physical Chemistry C
Article
increasing translational energy.16 Thus, the paul oxygen signal detected presumably suggests that small contributions from trapping into the physisorption state still remain even for oxidation with a 0.15 eV molecular beam. In contrast, the paul oxygen signal was not detected for 0.39 eV. In our previous study, we suggested that the trapping state of oxygen gas onto the surface seems to play an important role in generating the paul oxygen.21 Considering that the trapping probability might be almost negligible at 0.39 eV, the results obtained in this study are well-consistent with our previously proposed adsorption model including mobile oxygen on the surface. Therefore, it seems likely that direct adsorption without passing through the trapping state becomes dominate at 0.39 eV. That is, the fact that the first oxide is the ins structure indicates that the same processes take place after the molecularly chemisorbed state such as peroxo-like states, followed by dissociative chemisorption states.15,18 In addition to this confirmation, we found that the Si2+ signal only appears after the formation of the ins structure at 0.39 eV, indicating that the ins×2 structure is a plausible secondary product in this instance. It provides compelling evidence that this oxide may be formed by the reaction of O2 gas with the ins structure via the direct dissociative chemisorption process. This beam-assisted reaction is interesting and should be explored with theoretical approaches. Finally, we turn to a discussion of the initial sticking probability difference between thermal O2 and 0.06 eV. As reported in our previous molecular beam study,18 we found that the initial sticking probability for the three oxidation processes at room temperature has the following order: thermal O2 ≫ 0.15 eV > 0.06 eV. Taking into account that the fraction of oxygen gas above 0.15 eV in the MB distribution is a negligibly small value (0.8%), the amount of O2 gas of energy above 0.15 eV falls short of the increment in the oxygen amount, as seen in Figure 4c. Consequently, our experimental results rule out the contribution of O2 gas with larger translational energies for the oxidation using thermal-O2 gas. Here, providing that the translational energy for thermal O2 is ascribed to the average kinetic energy kBT (kB, Boltzmann constant; T, gas temperature), which is the kinetic energy defined on the basis of the most probable speed in MB distribution, our findings would seem to explain why the initial sticking probability for thermal O2 is much larger than that for 0.06 eV. Specifically, the difference in the initial sticking probability between thermal O2 and 0.06 eV is owing to the trapping probability variation, which depends on the translational energy; thus a larger trapping probability for thermal O2 results in a larger initial sticking probability. Consequently, we can establish the categorization of the oxygen adsorption processes depending on oxidation methods as depicted in Figure 5. For low-energy conditions such as thermal-O2 and supersonic O2 molecular beams with 0.06 eV, trapping-mediated dissociative adsorption (path A) where the physisorption state plays a role as a trapping state takes place. The trapped oxygen subsequently converts into molecularly chemisorbed oxygen which may be a peroxo-like state.31,32 Here, we note that this chemisorbed oxygen is different from the paul oxygen as seen in Figure 3. This conversion process is competitive for the thermal desorption process from the trapping state to the gas phase.18 The oxygen at the molecularly chemisorbed state instantaneously dissociates to form ins structure as the first product because the height of the energy barrier is negligible small.33 In fact, to the knowledge of the authors, there have been no
spectra suggests a valuable way to study time correlation between oxygen uptake and oxygen adsorption sites. In order to confirm the origin of this difference in detail, we focus on hereafter the time correlation between oxygen bonding configurations and Si oxidation states. We first note that the signals for both Si1+ and ins oxygen simultaneously appear immediately after oxidation starts for all oxidation conditions. This remarkable result reveals that the ins structure, which has one oxygen atom bonding between the Si− Si back-bonds of the Si adatom, is formed ahead of the other oxides as reported in ref 21. From this result, we confirm that the initial sticking probability for not only thermal-O2 but also supersonic O2 molecular beams (0.06, 0.15, and 0.39 eV) corresponds to the probability to create the ins structure. This indicates that our procedure is superior to the conventional methods based on the oxygen uptake measurements. Interestingly, Figure 4 highlights that there is no significant difference in the evolution of the oxygen bonding configurations and the Si oxidation states just after formation of the ins structure for thermal-O2 and the supersonic O2 molecular beams with 0.06 and 0.15 eV. We note that the signals for both Si2+ and paul oxygen were simultaneously observed immediately after the formation of the ins structure. This time correlation between the two signals indicates that the paul oxygen is attributed to a molecular oxygen chemisorbed on the ins structure: the ins−paul structure forms as previously reported in ref 21. Our experimental results strongly suggest that the paul oxygen observed in the oxidation at the Si(111)-7 × 7 surface even at room temperature is not the same kind of molecularly chemisorbed oxygen such as peroxo-like oxygen as observed in the O2/Pt(111) system.14,15 Indeed, the paul oxygen cannot be observed as a first product, and there is some duration until the appearance of the paul oxygen signal. Therefore, we believe that a mobile O2 trapped in a physisorption state might also engage in oxidation for both 0.06 and 0.15 eV beams as discussed in our previous study on thermal-O2 oxidation.21,26 It is probably concluded that the ins×2 structure is not formed through the reaction of O2 with clean Si adatoms because ins oxygen and Si2+ signals are not observed immediately after the start of oxidation. In other words, ins×2 structure is likely to be the secondary product after the reaction between ins structure and atomic oxygen. Here, the oxygen atom is the O2 dissociation product when other ins structure is formed through the reaction of O2 with nonoxidized Si atoms. This atomic oxygen may be so reactive that it reacts with the neighboring ins structure. This implies that the increment of ins structure density is needed to produce the ins×2 structure. Thus, we implicitly explain that the interval is needed to observe the Si2+ signal. In the following discussion, we focus on the dose dependence of the formation of the paul oxygen depending on the oxidation conditions because of its importance as a precursor state in oxygen adsorption processes. Our previous molecular beam study18 showed that the initial sticking probability for 0.15 eV is larger than that for 0.06 eV, suggesting that the direct adsorption might become dominant for 0.15 eV. Contrary to such fact, the paul oxygen signal was also observed even for 0.15 eV in this study and there was some interval before the appearance of the paul oxygen signal as well as thermal O2 and 0.06 eV. This implies that a process to form the paul oxygen for 0.15 eV takes place similarly to that for the case of thermal O2 and 0.06 eV. It is generally known that the trapping probability into a physisorption state decreases with 9440
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
The Journal of Physical Chemistry C
Article
energy. Our experimental data potentially imply that the relation between the initial sticking probability and the adsorption mechanisms can be categorized in terms of the translational energy of O2, where the translational energy of thermal O2 is ascribed to the average molecular kinetic energy defined by the most probable speed in the MB distribution. Although similar experiments have been reported for many gas−solid systems, this study might be useful in rationalizing the adsorption processes for beneficial chemical reactions, such as catalytic and semiconductor processes.
■
AUTHOR INFORMATION
Corresponding Author
*Tel.: +81-791-58-0802 ext 3913. Fax: +81-791-58-0311. Email:
[email protected]. Notes
The authors declare no competing financial interests.
■
Figure 5. Categorization of oxygen adsorption processes for thermalO2 and supersonic O2 molecular beams at the Si(111)-7 × 7 surface at room temperature. The horizontal line in the potential energy (PE) diagram for O2 adsorption represents the distance from the surface.
ACKNOWLEDGMENTS We thank Dr. Y. Saitoh and Dr. Y. Fukuda for their support of the synchrotron radiation operation; without their help this work would not have been possible. We are grateful to Prof. Y. Takakuwa and Drs. S. Ishidzuka, S. Ogawa, and A. Agui for their continuous help and discussions. We also thank Drs. M. Ohashi, T. Narushima, W. Yashiro, and K. Miki, who gave us much valuable advice in the early stages of this work. The synchrotron radiation experiments were performed at the BL23SU of SPring-8 with the approval of the Japan Atomic Energy Agency (JAEA) and the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal Nos. 2005B3803 and 2006A3803).
reports on the detection of O2 as a thermal desorption species from the oxygen-adsorbed silicon surfaces.1 This barrierless process may evoke that this molecularly chemisorbed state at the unoxidized Si(111)-7 × 7 surface does not undertake the role of the observable precursor state as well as that observed in the O2/Pt(111) system.15 For the supersonic O2 molecular beam with 0.15 eV, reaction path B without a trapping state can simultaneously occur in addition to reaction path A. Furthermore, with the molecular beam at 0.39 eV, reaction path B only occurs. As shown in Figure 5, the dissociation process after the molecularly chemisorbed state is the same and therefore we conclude that the ins structure can be observed inherently as the initial oxide. The current study was unable to resolve experimentally the translational energy distribution of thermal O2. Our work clearly has limitations in controlling the translational energy using molecular beams based on the gas-seeding method. Nevertheless, we believe that our work is promising and could be the framework for categorizing adsorption processes for thermal-O2 and supersonic O2 molecular beams in terms of the translational energy of O2. More work on comparing our data with theoretical studies based on computational simulations would help to establish a detailed physical picture of surface adsorption processes.34−39 This topic remains an important and interesting subject for the immediate future.
■
REFERENCES
(1) Engel, T. The Interaction of Molecular and Atomic Oxygen with Si(100) and Si(111). Surf. Sci. Rep. 1993, 18, 93−144. (2) Fundamental Aspects of Silicon Oxidation; Chabal, Y. J., Ed.; Springer Series in Materials Science; Springer-Verlag: Berlin, 2001. (3) Fundamental Aspects of Ultrathin Dielectrics on Si-Based Devices; Garfunkel, E., Gusev, E., Vul’, A., Eds.; Kluwer Academic: Dordrecht, The Netherlands, 1998. (4) Impact of Surface Science on Catalysis; Gates, B. C., Knozinger, H., Eds.; Academic Press: San Diego, CA, USA, 2000. (5) The Chemical Physics of Solid Surfaces; Woodruff, D., Ed. Surface Dynamics, Vol.11; Elsevier: Amsterdam, 2003. (6) Dynamics of Gas-Surface Interactions; Muiño, R. D., Busnengo, H. F., Eds.; Springer-Verlag: Berlin, Heidelberg, Germany, 2013. (7) Darling, G. R.; Holloway, S. The Dissociation of Diatomic Molecules at Surfaces. Rep. Prog. Phys. 1995, 58, 1595−1672. (8) Weaver, J. F.; Carlsson, A. F.; Madix, R. J. The Adsorption and Reaction of Low Molecular Weight Alkanes on Metallic Single Crystal Surfaces. Surf. Sci. Rep. 2003, 50, 107−199. (9) Yu, M. L.; DeLouise, L. A. Surface Chemistry on Semiconductors Studied by Molecular-Beam Reactive Scattering. Surf. Sci. Rep. 1994, 19, 285−380. (10) Libuda, J.; Freund, H.-J. Molecular Beam Experiments on Model Catalysts. Surf. Sci. Rep. 2005, 57, 157−298. (11) Arumainayagam, C. R.; Madiz, R. Molecular Beam Studies of Gas-Surface Collision Dynamics. Prog. Surf. Sci. 1991, 38, 1−102. (12) Davis, J. E.; Mullins, C. B. Adsorption Mechanisms of Translationally-Energetic O2 and N2: Direct Dissociation versus Direct Molecular Chemisorptions. Surf. Sci. 1997, 380, L513−L520. (13) Rettner, C. T.; Mullins, C. B. Dynamics of the Chemisorption of O2 on Pt(111): Dissociation via Direct Population of a Moleculary Chemisorbed Precursor at High Incident Kinetic Energy. J. Chem. Phys. 1991, 94, 1626−1635.
4. CONCLUSION We performed real-time observations of the oxidation on a clean and well-defined Si(111)-7 × 7 surface at room temperature using synchrotron radiation photoelectron spectroscopy. We succeeded in comparing the evolution of the oxides in the oxidation by thermal-O2 and supersonic O2 molecular beams with incident translational energies of 0.06, 0.15, and 0.39 eV. We revealed that the initial sticking probability corresponds to the probability of the formation of the ins structure, which has one oxygen atom at the back-bonds of a Si adatom. Our experimental result presumably suggests that the difference in the initial sticking probability between thermal-O2 and a 0.06 eV beam is owing to the trapping probability difference, which depends on the O2 translational 9441
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
The Journal of Physical Chemistry C
Article
(33) Lee, S. I.; Kang, M. I. Identification of the Initial-Stage Oxidation Products on Si(111)-(7 × 7). Phys. Rev. Lett. 1999, 82, 968−971. (34) Groß, A.; Eichler, A.; Hafner, J.; Mehl, M. J.; Papaconstantpoulos, D. A. Unified Picture of Molecular Adsorption Process: O2/Pt(111). Surf. Sci. 2003, 539, L542−L548. (35) Kato, K.; Uda, T.; Terakura, K. Backbond Oxidation of the Si(001) Surface: Narrow Channel of Barrierless Oxidation. Phys. Rev. Lett. 1998, 80, 2000−2003. (36) Behler, J.; Reuter, K.; Scheffler, M. Nonadiabatic Effects in the Dissociation of Oxygen Molecules at the Al(111) Surface. Phys. Rev. B 2008, 77, 115421 (16 pages).. (37) Pamungkas, M. A.; Joe, M.; Kim, B.-H.; Lee, K.-R. Reactive Molecular Dynamics Simulation of Early Stage of Dry Oxidation of Si(100) Surface. J. Appl. Phys. 2011, 110, 053513 (7 pages).. (38) Khalilov, U.; Neyts, E. C.; Pourtois, G.; van Duin, A. C. T. Can We Control the Thickness of Ultrathin Silica Layers by Hyperthermal Silicon Oxidation at Room Temperature? J. Phys. Chem. C 2011, 115, 24839−24848. (39) Khalilov, C.; Pourtois, U.; van Duin, A. C. T.; Neyts, E. G. Hyperthermal Oxidation of Si(100)2 × 1 Surfaces: Effect of Growth Temperature. J. Phys. Chem. C 2012, 116, 8649−8656.
(14) Nolan, P. D.; Lutz, B. R.; Tanaka, P. L.; Davis, J. E.; Mullins, C. B. Translational Energy Selection of Molecular Precursors to Oxygen Adsorption on Pt(111). Phys. Rev. Lett. 1998, 81, 3179−3182. (15) Nolan, P. D.; Lutz, B. R.; Tanaka, P. L.; Davis, J. E.; Mullins, C. B. Molecularly Chemisorbed Intermediates to Oxygen Adsorption on Pt(111): A Molecular Beam and Electron Energy-Loss Spectroscopy Study. J. Chem. Phys. 1999, 111, 3696−3704. (16) Ferguson, B. A.; Reeves, C. T.; Mullins, C. B. Oxygen Adsorption on Si(100)-2 × 1 via Trapping-Mediated and Direct Mechanisms. J. Chem. Phys. 1999, 110, 11574−11584 and references therein. (17) Behringer, E. R.; Flaum, H. C.; Sullivan, D. J. D.; Masson, D. P.; Lanzendorf, E. J.; Kummel, A. C. Effect of Incident Translational Energy and Surface Temperature on the Sticking Probability of F2 and O2 on Si(100)-2 × 1 and Si(111)-7 × 7. J. Phys. Chem. 1995, 99, 12863−12874. (18) Yoshigoe, A.; Teraoka, Y. Adsorption Dynamics on Si(111)-7 × 7 Surface Induced by Supersonic O2 Beam Studied Using Real-Time Photoelectron Spectroscopy. J. Phys. Chem. C 2010, 114, 22539− 22545. (19) Cushing, G. W.; Navin, J. K.; Valadez, L.; Johanek, V.; Harrison, I. An Effusive Molecular Beam Technique for Studies of Polyatomic Gas-Surface Reactivity and Energy Transfer. Rev. Sci. Instrum. 2011, 82, 044102 (11 pages). (20) Kolasinski, K. W. Surface Science: Foundation of Catalysis and Nanoscience, 2nd ed.; John Wiley & Sons: Chichester, U.K., 2008. (21) Yoshigoe, A.; Teraoka, Y. Immediate Product after Exposing Si(111)-7 × 7 Surface to O2 at 300 K. Jpn. J. Appl. Phys. 2010, 49, 115704 (6 pages). (22) Himpsel, F. J.; McFeely, F. R.; Taleb-Ibrahimi, A.; Yarmoff, J. A.; Hollinger, G. Microscopic Structure of the SiO2/Si Interface. Phys. Rev. B 1988, 38, 6084−6096. (23) Yeom, H. W.; Uhrberg, R. High Resolution Photoemission Study of Low-Temperature Oxidation on the Si(001) Surface. Jpn. J. Appl. Phys. 2000, 39, 4460−4463. (24) Oh, J. H.; Yeom, H. W.; Hagimoto, Y.; Ono, K.; Oshima, M.; Hirashita, N.; Nywa, M.; Toriumi, A.; Kakizaki, A. Chemical Structure of the Ultrathin SiO2/Si(100) Interface: An Angle-Resolved Si 2p Photoemission Study. Phys. Rev. B 2001, 63, 205310−205315. (25) Ogawa, S.; Yoshigoe, A.; Ishidzuka, S.; Teraoka, Y.; Takakuwa, Y. Si(001) Surface Layer-by-Layer Oxidation Studied by Real-Time Photoelectron Spectroscopy using Synchrotron Radiation. Jpn. J. Appl. Phys. 2007, 46, 3244−3254. (26) Yoshigoe, A.; Teraoka, Y. Atomic Lineation of Products during Oxidation of Si(111)-7 × 7 Surface Using O2 at 300 K. J. Phys. Chem. C 2012, 116, 4039−4043. (27) No difference with changing the nozzle temperature was observed in the photoelectron spectra. Thus we consider that the influence of vibrationally excited O2 on oxidation is negligibly small. (28) Moritani, K.; Okada, M.; Teraoka, Y.; Yoshigoe, A.; Kasai, T. Kinetics of Oxygen Adsorption and Initial Oxidation on Cu(110) by Hyperthermal Oxygen Molecular Beams. J. Phys. Chem. A 2009, 113, 15217−15222. (29) We adapted so-called normal energy scaling in this study because no dependence on sample rotations was observed, implying that the effects of surface roughness or orientation of Si−Si interatomic bonding are negligible small. The angle of incidence is a mere 10°, which results in ca. 97% incident energy. We almost seem to apply a total energy scaling. (30) Le, G. L.; Göthelid, M.; Grehk, T. M.; Björkquist, M.; Karlsson, U. O.; Aristov, V. Y. Surface Core-Level Shifts of Si(111)7 × 7: Fundamental Reassessment. Phys. Rev. B 1994, 50, 14277−14282. (31) Haochang, P.; Horn, T. C. M.; Kleyn, A. W. Harpponing in Surface Scattering: O2− Formation in Collisions of O2+ from Ag(111). Phys. Rev. Lett. 1986, 57, 3035−3038. (32) Pechtien, J. H.; Imke, U.; Snowdon, K. J.; Reijnen, P. H. F.; Hoek, P. J.; Kleyn, A. W.; Namiki, A. O2− Formation at Si(001). Surf. Sci. 1990, 224, 35−42. 9442
dx.doi.org/10.1021/jp410821r | J. Phys. Chem. C 2014, 118, 9436−9442
