Papers

AVS1999 Session EM-TuM: Si Surface Chemistry and Etching, Passivation, and Oxidation

Tuesday, October 26, 1999 8:20 AM in Room 608

Tuesday Morning

Start	Invited?	Item
8:20 AM		EM-TuM-1 Structural Transition Layers at the Interface of SiO₂/Si(100) Fabricated by Ozone K. Nakamura, H. Itoh, A. Kurokawa, S. Ichimura (Electrotechnical Laboratory, Japan); K. Koike, G. Inoue, T. Fukuda (Iwatani International Corporation, Japan) A novel processing technique for oxidation with rapid growth kinetics at lower substrate temperature is strongly required to fabricate a much thinner silicon dioxide film <4nm for gate oxides in MOSFET in the near future. However, properties of the synthesized oxide must be maintained or upgraded simultaneously in comparison with those of a conventionally used thermally grown oxide. Especially, the interface of oxide and substrate is critical because the oxide reliability is expected to depend on the characteristics of the interface affected by strain, defects, roughness, P_b centers, etc. Using ozone as an alternative oxidant, we have found a distinguished feature in the structural transition layers at the interface. An ultrathin silicon dioxide film was fabricated on a Si(100) surface by concentrated ozone: The substrate temperature was between room temperature and 700 °C and the ozone pressure was at 8x10^-4Pa or atmospheric pressure. Structural transition layers in the SiO₂ ultrathin film by ozone was limited to a much thinner region than that of thermally grown oxide with > 1nm thickness. This was confirmed by a change of the etching rate of SiO₂ film with dilute hydrofluoric acid solution. Such thinner region of transition layers in the ozone-oxide was implemented either on clean Si(100)2x1 or on Si(100) with an already existing native oxide film at 300°C or more. However, exposure of ozone to Si(100) with an already existing thermally grown oxide film, for example at 350°C, caused no change in the distribution of transition layers in the oxide. This contrast indicates that the oxide growth by ozone or the further oxidation of lower oxidized silicon atoms in the native oxide by ozone formed transition layers with much less thickness, while highly oxidized silicon atoms in the thermally grown oxide film remained unreacted even by reactive ozone. Structural transition layers on the opposite side of the interface, i.e. in the substrate, will also be discussed.
8:40 AM		EM-TuM-2 Infrared Spectroscopy as a Probe of Semiconductor/Dielectric Interfaces: Growth and Structure of SiO₂ on Si K.T. Queeney, M.K. Weldon, Y.J. Chabal, K. Raghavachari (Bell Laboratories, Lucent Technologies) The structure and quality of the Si/SiO₂ interface are crucial to the performance of transistors with gate oxide thicknesses < 20 Å. We have exploited the intrinsic sensitivity of infrared absorption spectroscopy to microscopic chemical environment in order to elucidate structural details of this interface between crystalline Si and amorphous SiO₂. Infrared spectra of thermally grown SiO₂ are acquired as a function of film thickness by etchback of device-quality films. Modeling the mechanical and optical properties of these films reveals that substoichiometry at the Si/SiO₂ interface dominates the spectra of ultrathin (< 10 Å) SiO₂; different thermal histories are shown to affect the quality of this "transition region." To understand the microscopic structure of this interfacial substoichiometry, we have grown and characterized a model Si/SiO_x interface via controlled H₂O reaction of Si(100)-(2x1) in ultrahigh vacuum. Coalescence of dimer-based silicon epoxide species (capped by triangular Si-O-Si linkages) into an extended silicon-oxygen network results in the birth of SiO_x phonon modes (975 and 1130 cm^-1) whose microscopic structural origins are for the first time well understood. This epoxided interface is trnasformed at room temperature into high-quality SiO₂, and the mechanism for room-temperature H₂O-induced oxidation is compared to that observed for technologically relevant surface terminations.
9:20 AM		EM-TuM-4 Real Time Observation on Si(001) Surface Oxidation by Scanning Tunneling Microscopy K. Miki (Electrotechnical Laboratory, Japan); Y. Kudo, M. Murata, K. Yamabe (Tsukuba University, Japan) We have succeeded in real time observation on oxidation of Si(001) surface by scanning tunneling microscopy at elevated temperatures up to 1100 K. First we made clear the boundary between etching and oxidation regions. At the 900 K, etching both from step edges and in the terrace was dominant at low oxygen partial pressure under $1 \times 10^(-5) Pa$ while we observed oxidation island nucleation over this critical pressure. Under low rate oxidation, we found that etching is allowed together and it stops in the vicinity of the oxidation island. Our map whether etching or oxidation occurs is consistent of the previous reports. At the high temperature region the boundary is good agreement with the previous report by Gelain and et al. [Oxidation of metals 3 (1971) 139]. In the low temperature region under 870 K the oxidation speed of the first 1 ML is independent of temperature and this results is consistent with reflection electron microscopy experiment by Watanabe and et al. [Phys. Rev. Lett. 80 (1998) 345] We found three types of absorbant. Although the one type is still a misery, we could identify that one is atomic oxygen adsorbant in the center of a Si dimer and the other is back bond oxidation. First one was well seen in the initial stage, as oxidation proceeded the latter became more dominant. This observation suggests that oxidation of Si(001) surface has dual species at least, which is previously reported by Engstrom and et al. [Surf. Sci. 256 (1991) 317]. Backbond oxidation extended normal to dimer rows as ordered spots. The ordering eventually came to have disordering around 1ML oxidation. This suggests that stress during is very important even in the initial stage.
9:40 AM		EM-TuM-5 Scanning Tunneling Microscopy Study of Surface Morphology of Si(111) after Synchrotron Radiation Stimulated Desorption of SiO₂ Y. Gao, T. Miyamae, H. Mekaru, T. Urisu (Institute for Molecular Science, Japan) We have used scanning tunneling microscopy to investigate the surface morphology of Si(111) after the native SiO₂ layer was removed by synchrotron radiation stimulated desorption at 650 °C. The surface shows large regions of atomically flat Si(111)-7x7 structure. An interesting feature of the surface is the formation of atomic steps nicely registered to the crystal structure, and the pinning of the steps by nanometer scale dust is evident. This is in sharp contrast to Si(111) surfaces after thermal desorption of SiO₂ at temperatures 880°C and above, where the surface steps are much more irregular. The registration of the surface steps to the underlying crystal structure indicates that the surface atomic layer reaches thermodynamic equilibrium under synchrotron radiation at temperatures much lower than that necessary for thermal desorption of SiO₂.
10:00 AM		EM-TuM-6 How Important are Second Nearest Neighbor Effects in Silicon 2p Photoemission Spectroscopy of Si/SiO₂ Interfaces? J. Eng, Jr., K. Raghavachari (Bell Labs, Lucent Technologies) The proper interpretation of Si 2p photoemission spectra of Si/SiO₂ interfaces has been a controversial topic since 1993, when Banaszak-Holl and McFeely proposed that second nearest neighbor effects can cause significant chemical shifts in Si 2p photoemission features.¹ Their claims were based upon model Si/SiO₂ surfaces produced by the adsorption of H₈Si₈O₁₂ clusters on Si(100) at room temperature. Arguing that the clusters are bonded to Si(100) dimers through a single vertex (due to Si-H bond cleavage), they proceeded to correlate the relative peak positions and peak intensities with different Si species at the interface. This correlation led them to conclude that the entire formal oxidation state framework is inadequate for interpreting Si 2p photoemission spectra of Si/SiO₂ interfaces, and that second nearest neighbor effects are important. The key issue in this controversy is understanding how the H₈Si₈O₁₂ clusters bond to the Si(100) surface. Using transition state calculations, we present detailed mechanistic arguments which show that the clusters do not react with the Si(100) surface through Si-H bond cleavage, but rather through Si-O bond cleavage. The resulting "cracked" cluster allows us to predict the Si 2p photoemission features of the clusters on Si(100) using the conventional formal oxidation state model, without invoking second nearest neighbor effects. Finally, the normal mode frequencies of the "cracked" cluster are in excellent agreement with infrared studies of the clusters on Si(100). ¹ M. M. Banaszak-Holl and F. R. McFeely, Phys. Rev. Lett., 71(15) (1993) p.2441.
10:20 AM		EM-TuM-7 FTIR Studies of the Nitridation of Si(100)-(2x1) and Oxidized Silicon K.T. Queeney, Y.J. Chabal, J. Eng, Jr., K. Raghavachari (Bell Laboratories, Lucent Technologies); X. Zhang, E. Garfunkel (Rutgers University); S.B. Christman, E.E. Chaban (Bell Laboratories, Lucent Technologies) We have investigated the incorporation of N into Si and SiO₂ with IR absorption spectroscopy, in order to elucidate the mechanisms by which nitridation and oxynitridation influence the structure and thus the electrical characteristics of Si/SiO₂ interfaces. Studies of the adsorption and decomposition of NH₃ on Si(100)-(2x1) demonstrate that the chemistry of this nitriding agent is governed by a combination of dissociative and molecular adsorption, with metastable NH_3(a) characterized by a dramatically redshifted NH₃ deformation mode. Higher coverages and temperatures favor NH₃ dissociation into H and NH₂, followed by competing desorption and decomposition. The mechanism for N insertion into the Si substrate is studied via repeated cycles of NH₃ dosing and annealing with post-dosing of atomic H to identify discrete N-containing structures by the perturbation in ν(Si-H) frequencies. Oxynitride growth by reaction of NO on Si(100)-(2x1) reveals the interplay between oxygen and nitrogen, as both Si-O (~900 cm^-1) and Si-N (~775 cm^-1) modes exhibit frequencies distinct from those observed for pure oxide and nitride films. These fundamental growth studies are used to interpret structural details contained in the complex spectra of device-quality nitride and oxynitride films grown on Si.
10:40 AM		EM-TuM-8 Silicon Passivation Chemistry for MEMS Technology R. Maboudian (University of California, Berkeley) Adhesion, friction , and wear are prevalent problems in a majority of micro-electro-mechanical systems (MEMS) devices. Since gravity is negligible at the dimension of most microstructures, understanding of surface interactions in MEMS is of paramount importance for controlling stiction phenomena. After a brief introduction to Si micromachining, I will discuss the use of various micromachined testing devices, such as cantilever beam arrays, in conjunction with other surface characterization techniques, such as X-ray photoelectron spectroscopy and atomic force microscopy, to measure the surface forces present between polycrystalline silicon surfaces and to manipulate them by utilizing various surface passivation treatments.
11:20 AM		EM-TuM-10 STM Studies of the Site-specific Reactivity of Isopropanol in Aqueous Silicon Etching: Controlling Morphology with Surface Chemistry M.A. Hines, T.A. Newton, Y.-C. Huang, L.A. Lepak (Cornell University) Aqueous silicon etchants play an important role in the fabrication of microelectromechanical systems (MEMS). To improve performance, a reputedly inactive chemical agent -- isopropanol [(CH₃)₂CHOH] -- is often added to the etchant. Not only does this simple additive reduce undercutting by up to 75%, it also produces much smoother etched surfaces, in part by suppressing the formation of pyramidal etch hillocks. Using a combination of STM measurements and kinetic Monte Carlo simulations, we will show that these morphological changes have a simple, chemical origin. The site-specific rates of isopropanol reaction on vicinal Si(111) surfaces were studied by kinetic competition with an etchant of known anisotropy and quantified using concentration-dependent changes in the etched surface morphology. The unique properties of isopropanol-enhanced etchants are explained by the relatively high reactivity of the isopropoxide ion. Once formed, the silicon isopropoxy species transiently suppresses etching at specific surface sites and modifies the anisotropy of the etchant.
11:40 AM		EM-TuM-11 Removal of Native Oxide Employing Heated NH₃/NF₃ Mixture H. Ogawa (The University of Tokyo, Japan); T. Arai, T. Ichiki (Toyo University, Japan); Y. Takamura, Y. Horiike (The University of Tokyo, Japan) The removal of the native oxide employing a heated NH₃/NF₃ mixture was studied using in-situ XPS and FTIR-ATR/RAS. The mixture was heated in an 13 mm diameter Al₂O₃ tube whose surface was wound by a resistive heater, then being exposed to a sample set on a stage cooled by a N₂ gas with room temperature. The chemical oxide was grown in a H₂SO₄/H₂O₂ solution. The NH₃/NF₃ mixture with partial pressure ratio of unity at a pressure of 1 Torr started to remove the oxide from a Al₂O₃ tube temperature of 500 °C. For the Si surface after removal of the oxide, new XPS peaks appeared at 103.8 eV in Si2p and 402.2 eV in N1s, respectively. Absorption spectra of Si-H(2100 cm^-1), N-H(stretching; 3330 cm^-1, bending;1454cmcm^-1) and Si-F (783 cm^-1) were also observed in the IR measurement. These results exhibit the presence of a (NH₄)₂SiF₆ film deposited on the Si surface after removal of the oxide.¹The film desorbed readily at 100 °C in a vacuum and then was terminated by hydrogen. The higher the partial pressure, the faster the oxide removal rate, whereas NH₃ or NF₃ alone did not demonstrate any etching reaction. The result that the Al₂O₃ tube heated at higher temperature led to the higher etch rate implies generation of the oxide etchant within the tube. Thus, NH₃ and NF₃ were introduced separately to two tubes, and when NF₃ alone was heated, the present reaction was confirmed. The result implies thermally decomposed NF₃ reacts with NH₃, generating the oxide etchant. The removal rate ratio of thermal grown SiO₂ to BPSG (boron phosphorus silicate glass) films was almost unity. It is well known that the usual HF solution produces about ten times higher etch rate for BPSG than for the thermal SiO₂. Accordingly, this technology allows us to offer a new dry cleaning method of the contact hole surface in ULSI multi-level interconnection process. ¹M. Hirose, S. Yokoyama and Y. Yamakage: J. Vac. Sci. Technol., B3 (1985)1445.