Papers

AVS1996 Session SS-WeA: Oxidation and Etching of Semiconductor Surfaces

Wednesday, October 16, 1996 2:00 PM in Room 203B

Wednesday Afternoon

Start	Invited?	Item
2:00 PM		SS-WeA-1 Surface Band Structure and Electron Affinity on Cesiated C(100) Surfaces T. Mercer, P. Pehrsson (Naval Research Laboratory) The surface electronic properties of C(100):H, C(100), C(100):O and C(100):O:Cs were examined with High Resolution Electron Energy Loss Spectroscopy (HREELS), Secondary Electron Emission Spectroscopy (SEES) and a Kelvin Probe. With these techniques we correlated variations in surface band structures with band movement, Fermi level position, work function, and electron affinity, including the development of negative electron affinity (NEA). The cesiated surface electron affinity and work function were lower than those of the hydrogenated surfaces, which were in turn lower than those of the reconstructed and oxidized surfaces. HREELS electronic excitation spectra confirmed two midgap surface states on the reconstructed C(100) surface, consistent with the downward band-bending, higher work function, and positive electron affinity on this surface observed using SEES and the Kelvin probe. The midgap states disappeared after surface oxidation and cesiation. The cesiation results conformed generally to theoretical predictions by other workers.
2:20 PM		SS-WeA-2 Silicon Oxycarbide Formation on SiC and at the SiC/SiO\sub 2\ Interface C. \um O\nneby (Pennylvania State University); C. Pantano (Pennsylvania State University) The SiC/SiO\sub 2\ interface has received considerable attention because of its significance in microelectronics, ceramics and composites. The presence of a transition oxide at the interface could alter the electronic properties of a multilayer SiC-based electronic device, or change the high-temperature mechanical properties of SiC ceramics and composites. In previous work, a transition oxide was proposed to form during the initial stages of SiC oxidation. Recently, the transition oxide was attributed to a possible silicon oxycarbide but the exact configuration was not reported.In this work, the XPS fingerprints for the various silicon oxycarbide species have been established using standards. This information has been coupled with an extensive XPS-analysis in order to determine the specific bonding configuration of the silicon oxycarbides formed at the initial stages of oxide growth on SiC. Both single crystal SiC and sputter deposited SiC were exposed to the ambient environment as well as to controlled amounts of oxygen from various sources such as O\sub2\ gas, ozone and an oxygen plasma. It has been concluded that silicon oxycarbide forms the native oxide on SiC. Moreover, it can exist at the SiC/ SiO\sub 2\ interface similar to the suboxides that are known to be present at the Si/ SiO\sub 2\ interface.
2:40 PM		SS-WeA-3 Scanning Tunneling Microscopy Study of Oxygen Adsorption on 6H-SiC(0001) (3x3) and (=883x=883) Surfaces L. Li, Y. Hasegawa, T. Sakurai (Tohoku University, Japan) Silicon carbide (SiC) is an attractive semiconductor for applications requiring a wide bandgap, high temperature/high power and chemical inertness. An investigation of the initial stages of oxidation of SiC, therefore, is important for optimizing the growth of thin, insulating oxide layers in SiC devices. In the present work, scanning tunneling microscopy (STM) has been applied to the study of the initial stages of oxygen adsorption at room temperature on the Si-rich (3x3) and Si-deficient (=883x=883) surfaces of 6H-SiC(0001). On the (3x3) surface, only one reaction product has been identified in the STM topographs, which appears as bright sites in the empty state image mode. While on the (=883x=883) surface, adsorption of oxygen produced a disordered surface. The bright sites on the (3x3) surface are attributed to the oxygen atom inserted in one of the Si adatom's back bonds.
3:00 PM		SS-WeA-4 Study of Oxygen Chemisorption on GaN(0001)-(1x1) V. Bermudez (Naval Research Laboratory) Clean, ordered GaN(0001)-(1x1) surfaces are prepared by sputtering with nitrogen ions and annealing in UHV. The surfaces are subsequently exposed at room temperature to O\sub 2\ and the chemisorption process studied. Auger, valence and core-level pho- toemission and electron energy loss spectroscopies, LEED and work function measurements have been used to examine the chemisorbed layer. Saturation occurs at a coverage of about 0.4 monolayers and is accompanied by removal of surface states near the band edges. The continued presence of a clear (1x1) LEED pattern, together with other data, indicates a well-defined adsorption site, but the rela- tive importance of Ga-O and N-O bonding remains undetermined. The realization that surface states exist near the valence band maximum has led to a more accurate determination of the surface Fermi-level position, and of dependent quantities, than given previously [1]. The clean-surface data are also compared with those for surfaces prepared [1] by in-situ deposition of Ga metal followed by thermal desorption. No significant differences are seen, which suggests that nitrogen-ion sputtering and annealing is suitable for prepar- ing clean, ordered GaN(0001) surfaces. The results for O chemi- sorption on atomically-clean surfaces have been applied to evaluat- ing the passivation of surfaces prepared by ex-situ treatment in aqueous NH/sub 4/OH. The band bending is found to be about 0.5 eV less than on atomically-clean surfaces, indicating a nearly passi- vated surface. [1] V.M. Bermudez et al., J. Appl. Phys. 79, 110 (1996).
3:20 PM		SS-WeA-5 Fluorine Exposure to Hydrogen-terminated Si(111) Surface and Subsequent Oxidation Y. Horiike, Y. Morikawa, T. Ichiki (Toyo University, Japan); S. Fujimura, H. Ogawa (Fujitsu Ltd., Japan) To realize the self-limiting reaction in the atomic layer-by-layer etching, the initial reaction of fluorine(F) atoms with a hydrogen(H)-terminated Si(111) was studied employing a combined system of FTIR-ATR and XPS. For XeF\sub 2\ exposure less than 2000 L, XPS take-off angle and ATR measurements revealed that F atoms penetrated into the subsurface, and a Si-H bond appeared at 2083 cm\super -1\(B\sub 1\) shifted to 2086 cm\super -1\(B\sub 2\) and 2090 cm\super -1\(B\sub 3\). Since the sum of peak areas of B\sub 1\, B\sub 2\ and B\sub 3\ was roughly the same, B\sub 2\ and B\sub 3\ peaks were considered to result from Si-H bonds newly appeared instead of decrease in the Si-H peak of B\sub 1\. The latter result demonstrates that F atoms penetrate into the Si without removal of terminated H bonds. It was speculated from Si 2p\sub 3/2\ and F\sub 1s\ spectra that about half amount of fluorine formed Si-F bond and residual fluorine existed as atomic state. Further exposure of F atoms, however, caused production of SiF\sub 2\ and SiF\sub 3\ bonds. Hence, F\sub 2\ gas which was less reactive than atomic fluorine was investigated. The FTIR-ATR and XPS measurements clarified following results: With increase in exposure dose of 5%F\sub 2\/He to H-terminated Si(111) surface, F atoms penetrated into the Si surface, and the peak of Si-H bonds began to decrease from 250 L and disappeared completely at 3x10\super 5\ L, while penetration depth of fluorine was limited to 0.2-0.38 nm even for more exposure of F\sub 2\ to 10\super 10\ L. The fluorinated layer consisted of only Si-F bonds and the depth corresponded to the subsurface region of Si(111). This indicates that the self-limiting process has been realized. To investigate inclusion of fluorine in the oxide grown after HF wet cleaning of the Si wafer, which may cause reliability, distribution of fluorine in oxide thermally grown at 700 degrees C, 60 min after XeF\sub 2\ controlled exposure of 500 L-5000 L to the Si(111) wafer was measured. It was found that oxide film thickness was kept constant at about 4 nm in these wide range exposure and most amount of fluorine atoms stayed on the oxide film surface.
3:40 PM		SS-WeA-6 HREELS Investigation of the Oxidation of the Ammonium Fluoride-Treated Si(100) Surface G. Kluth, R. Maboudian (University of California, Berkeley) High resolution electron energy loss spectroscopy has been employed to examine the oxidation behavior of ammonium fluoride-treated Si(100) surfaces exposed to air. The hydrogen terminated surfaces that result from NH\sub 4\F treatment show an initial, slow oxidation period which is followed by a period of more rapid oxidation. In the slow uptake region, oxidation takes place by insertion of oxygen into bridge bonding sites, while the hydrogen termination remains intact. At this stage, oxidation is not restricted to only the surface layer, but is observed in bulk bridge bonding sites also, suggesting that oxide growth does not occur layer by layer initially. Hydroxyl groups, which are present on hydrogen fluoride-treated surfaces, are not observed, suggesting that they do not play a major role in the initial oxidation phase of NH\sub 4\F-treated surfaces. The initial oxidation behavior is similar to that observed for in situ oxygen adsorption at elevated temperatures on clean silicon.
4:00 PM		SS-WeA-7 Atomically-selective Chemisorption of I\sub 2\Cl\sub 6\ Molecules on the Si(111)-7x7 Surface A. Kummel, Y. Liu (University of California, San Diego) The chemisorption of monoenergetic I\sub 2\Cl\sub 6\ molecules (2.5 eV) on the Si(111)-7x7 surface at room temperature has been investigated using supersonic molecular beams, scanning tunnelling microscopy (STM) and Auger electron spectroscopy (AES). AES analysis demonstrated that the absolute ratio of chlorine to iodine chemisorbed on the Si(111)-7x7 is approximately 1 at low coverages and increases to 3 at high coverages. This suggests that, at low coverages, the Si(111)-7x7 surface selectively abstracts iodine atoms from I\sub 2\Cl\sub 6\ molecules, and that chlorine in I\sub 2\Cl\sub 6\ molecules are preferentially ejected back to the gas phase during adsorption. However, at high coverages, the adsorption of I\sub 2\Cl\sub 6\ proceeds in a different fashion so as to replenish the iodine-rich surface with chlorine. Empty-state STM topographic images show that the adsorption of planar, bridged, non- polar I\sub 2\Cl\sub 6\ (see insertion) onto the Si(111)-7x7 surface eliminates the adatom silicon dangling bonds to form silicon monohalide sites of four different sizes, namely isolated single, adjacent double, triple and quadruple sites. At 5.8% total coverage as defined by the percentage of the reacted silicon adatoms dangling bonds on the Si(111)-7x7 surface, 40% of the adsorbed halogen atoms form isolated single sites, 28% form adjacent double sites, and 32% form adjacent triple or quadruple sites. This adsorbate size distribution is significantly different from that of ICI, i.e. 58% isolated single, 28% adjacent double and 14% adjacent triple sites at the same total coverage. Since there are totally eight halogen atoms in an I\sub 2\Cl\sub 6\ molecule, the observation that the largest sights contain just four reacted adjacent silicon adatom dangling bonds implies that at least four of the eight atoms in an I\sub 2\Cl\sub 6\ molecules are ejected back into the gas phase. Possible reaction mechanisms are also proposed to interpret the observed experimental phenomena.
4:20 PM		SS-WeA-8 Room Temperature Etching of Si(111) 7x7 by ICI E. Lanzendorf, Y. Liu, K. Pettus, A. Kummel (University of California, San Diego) The room temperature etching of a Si(111) 7x7 surface by an interhalogen species, ICI, was detected using multiphoton ionization (MPI) spectroscopy and time- of-flight (TOF) mass spectrometry. Previously, room temperature etching of Si surfaces has only been observed for atomic halogens or xenon fluoride species; xenon fluoride bonds are unusually weak. In our studies, indirect evidence for the etching of SI by ICI is exhibited in the continuous uptake by the surface of ICI during dosing from a pulsed molecular beam. Direct evidence for the etching of Si by ICI is seen in MPI experiments by monitoring the species reflected from the surface during dosing by ICI. ICI was seeded in He and in a pulsed molecular beam. The kinetic energy of the incident ICI molecules was 1.1-1.2 eV. In the incident ICI beam, the ions observed in the 210 nm MPI spectrum were Ci, I, and a small amount of ICI and I\sub 2\. The spectrum of the molecules reflected or desorbing from the surface includes additional ion mass peaks: atomic Si, SiCl, and possibly SiCl\sub 3\. The SiCl\sub 3\ species has a mass of 133 amu and is difficult to distinguish from the atomic I signal observed in the incident beam (127 amu). The Si containing species are not coincident with either the incident beam or the unreacted scattered beam which is consistent with thermal desorption of etch products. The MPI spectrum of the species desorbed from the surface during flash desorption experiments yields the same products seen during etching. The room temperature etching of silicon by ICI is unexpected since the bond strength of ICI is similar to that of the homonuclear diatomic halogens.
4:40 PM		SS-WeA-9 Reflectance Anisotropy Spectroscopy - A New Technique for Determining Surface Kinetics A. Lees (Imperial College of Science, Technology and Medicine, United Kingdom); J. Zhang, A. Taylor, M. Xie (Imperial College, United Kingdom); Z. Sobiesierki (University of Wales College of Cardiff, United Kingdom); B. Joyce (Imperial College, United Kingdom) Most recent experimental work on H2 desorption from the monohydride Si(001) surface seems to point to a first order rate law and a dihyride transition state. However a wide range of desorption mechanisms and activation energies have been determined to date.In the work reported here, we present the first detailed application of in-situ isothermal desorption measurements using Reflectnce Anisotropy Spectroscopy (RAS). Desorption of chemisorbed surface hydrogen (produced by the adsorption of atomic hydrogen and the adsorption and pyrolysis of disilane or monosilane) from vicinal Si(001) surfaces was measured in real time. The results indicate that a combination of both zeroth and first order kinetics are obeyed. The dependence of the absolute rate on the degree of substrate misorientation suggests that the zeroth order desorption occurs via a precursor state associated with the steps on the surface. This zeroth order component dominates at high temperatures and high hydrogen coverages, indicating that under these conditions surface hydrogen diffusion is not the rate limiting step.This zeroth order kinetic pathway differs from previously reported results and the implications for gas source molecular beam epitaxy (GSMBE) of Si are discussed. Discrepancies in the activation energies reported in the literature maybe explained in terms of these competing desorption mechanisms.[1] U. Hofer, Leping Li, and T.F. Heinz, Physical Review B 45, 16, (1992), 9485. [2] Z. Jing, G. Lucovsky and J.L. Whitten, Surface Science Letters 296, (1993), L33-L37. [3] K.Sinnah, M.G. Sherman, L.B. Lewis et al, Physical Review Letters, 62, 5 (1989), 567.