AVS1997 Session EM-TuP: Issues in Silicon Processing and in Microelectronics
Tuesday, October 21, 1997 5:30 PM in Room Exhibit Hall 1
Tuesday Afternoon
Time Period TuP Sessions | Topic EM Sessions | Time Periods | Topics | AVS1997 Schedule
|
EM-TuP-1 Integratibility of PVD TiN for sub-0.25 micron Interconnect
S.S. Sengupta, D. Baker, S. Bothra, S. Sethi (VLSI Technology, Inc.) For the past few device generations, conventionally-sputtered TiN has proven adequate for use as barrier layers, contact or via liners, and as anti-reflective coatings (ARC). However, as plug aspect ratios in advanced interconnect systems climb to >3:1, and contact/via hole diameters shrink below 0.3 micron, alternative TiN processes, not limited by step coverage, need to be considered. While CVD TiN deposition gives conformal step coverage, the use of precursors can result in the incorporation of impurities in the film and higher process temperatures compared with PVD. In this paper, we demonstrate the applicability of an ionized metal plasma (IMP) process for sputter-deposition of TiN. IMP TiN gave superior bottom coverage at narrow, high aspect ratio contacts and vias, compared with standard TiN. Electrical results were obtained for W-filled vias and salicided contacts containing IMP versus standard TiN liner. The improvement in bottom coverage with IMP TiN enabled the use of thinner glue layers and liners, which effectively reduced the via resistance for smaller vias. In the field of lithography, TiN is widely used as an anti-reflective coating on top metal films and in this regard the optical properties of anti-reflective coatings are extremely important. In the case of IMP and standard PVD TiN, the optical properties are very similar. Refractive index and extinction coefficient of the PVD TiN film at an imaging wavelength of 365 nm are 2.221 and 1.035 respectively. The corresponding numbers for the IMP TiN film are 2.263 and 0.96 respectively. Due to the similarity of the optical properties, the linewidth response with these two films is also similar. In this paper we also show the imaging performance with these two films at exposure wavelengths of 365 nm and 248 nm. The final part of this paper discusses the impact of these two films on lithography. Empirical data on linewidth response is compared with optical lithography simulations. |
|
EM-TuP-2 Analytic Representations of the Dielectric Functions of Materials for Device and Structural Modeling
J. Leng, J. Opsal, H. Chu, M. Senko (ThermaWave); D.E. Aspnes (Aspnes Associates, Inc.) Analytic representations of the dielectric function are needed for the analysis of optical data of complex materials and structures. Here, we examine various harmonic-oscillator-based representations of the dielectric functions of the silicon-related materials crystalline Si (c-Si), amorphous (a-Si), and silicon nitride. For crystalline semiconductors we develop a new representation with a prefactor proportional to the inverse second power of the energy, the expected response for materials with wavefunctions that are eigenfunctions of the momentum operator. We determine a 4-oscillator model for c-Si that represents the data from 1.5 to 6.0 eV with a maximum discrepancy of less than 2%. Data for a-Si and silicon nitride are best fit over the same energy range with a model recently proposed by Jellison and Modine. |
|
EM-TuP-3 Spectroscopic and Thermal Studies of Methyl and Hydrogen Incorporation and Reactivity in Thin Amorphous Silicon Carbide Films
M.-S. Lee, S.F. Bent (New York University) In group IV alloys such as a-SiC:H and a-SiGe:H, the different nature and strength of bonding between alloy atoms and hydrogen play an important role in the growth of these materials and in voids formation. To investigate the reactivity and bonding of hydrogen and methyl incorporated into the films, highly hydrogenated thin a-SiC:H films were grown by different methods, including hot-wire CVD and ECR plasma-enhanced CVD at low temperature on Si(100) with mono-, tri-, and tetramethylsilane as precursors. The different growth methods were compared using in situ multiple internal reflection FTIR and temperature programmed reaction/desorption(TPR/D). The results indicate that while growth using pure methylsilane precursors leads to films with SiHx(x=1-3) and mainly intact methyl groups, diluting with atomic hydrogen can lead to CH2 incorporation. Thermal studies suggest that terminal SiH3 and (CH3)SiH2 groups are the least stable in the film because (methyl)silanes evolve at low temperature (below 350°C). Hydrogen and methane evolution at higher temperatures is correlated with decomposition of silicon hydrides and methyl groups. By comparison with methyl/Si(100) studies and related polymer systems, possible thermal decomposition mechanisms are proposed. |
|
EM-TuP-4 Adsorption of High Purity Ozone on Si(111)7x7 and Si(100)2x1: Surface Insensitiveness of Initial Sticking Probability
K. Nakamura, A. Kurokawa, S. Ichimura (Electrotechnical Laboratory, Japan) Oxidation of silicon surfaces at lower temperatures requires more active oxidizing reagents. Ozone is expected to be one of such oxidants with high reactivity so that analysis of its adsorption process is important as an initial step of oxidation. We have so far investigated an initial adsorption process of ozone molecules on Si(111)7x7 by second harmonic generation (SHG) and x-ray photoelectron spectroscopy (XPS) 1,2. In this study, we compared adsorption of ozone on Si(111)7x7 and Si(100)2x1 by SHG and XPS, and clarified that ozone molecules can adsorb on Si(111)7x7 and Si(100)2x1 with the same sticking probability close to unity. Second harmonics (SH) of 532nm generated by 1.064µm Nd:YAG laser beam were measured in-situ during exposure of Si(111)7x7 and Si(100)2x1 to high purity ozone in an ultra-high vacuum. The incidence and scattering angles of laser beam were 55° from a normal axis to the surface, respectively. Sticking probability of ozone adsorption on one dangling bond as an active site, estimated from the rate of SH intensity decay corresponding to the termination rate of the dangling bonds, is insensitive to surface structures of room temperature (RT) Si(111)7x7 and Si(100)2x1. This is different characteristics from that of molecular oxygen. Temperature dependence of rates of SH intensity decay between RT and 400°C was reverse between that of ozone and oxygen adsorption: higher temperature decreases oxygen adsorption rate but increases that of ozone. However, O1s XPS intensity showed no temperature dependence of the adsorption rate of ozone. This suggests that initial adsorption proceeds with direct adsorption mechanism with higher sticking probability, although it forms different species such as bridged Si-O-Si at higher temperatures to further decrease SH intensity.
|
|
EM-TuP-5 Polysiloxane Thin Films with Low Dielectric Constant Prepared by Polymerization of Hexamethyldisiloxane using Inductively Coupled Plasma Source
T. Fujii, M. Hiramatsu, M. Nawata (Meijo University, Japan); M. Hori, T. Goto (Nagoya University, Japan); S. Hattori (Nagoya Industrial Science Research Institute, Japan) As the device dimensions of ultralarge scale integration circuits (ULSI) are getting smaller, the need for a low dielectric constant of interlayer dielectric films is required to realize high-performance ULSI devices. In this work, we propose a simple deposition system using a 13.56 MHz inductively coupled plasma (ICP) source for preparation of polysiloxane thin films as a new insulator with a dielectric constant lower than 3.0. Hexamethyldisiloxane (HMDSO) monomer was used as a source gas, and an oxygen radical source was employed for the polymerization of HMDSO monomer. The monomer gas was poured onto Si substrate which was located in the downstream of oxygen plasma. The total pressure of HMDSO/O2 gases in the reaction chamber was kept at 100 mTorr. The films were analyzed by x-ray photoelectron spectroscopy (XPS) and Fourier transform infrared spectroscopy (FT-IR). The effects of HMDSO and oxygen pressures on the film properties were investigated. From FT-IR analysis, the films had polysiloxane structure which contained Si-O and Si-CH3 bonds. From XPS analysis, the carbon content of the films increased with the increase of HMDSO partial pressure. The dielectric constant of the films decreased with the increase of HMDSO partial pressure. The dielectric constant of the film formed at rf power of 300 W, HMDSO/O2 pressures of 50/50 mTorr, and room temperature was 2.5. |
|
EM-TuP-6 Study of Hydrogen Adsorption and Desorption on Ge/Si Surfaces Using Multiple Internal Reflection Infrared Spectroscopy
D. Wang, J.E. Crowell (University of California, San Diego) The ability to manipulate the bandgap of SiGe alloys coupled with their strained-layer epitaxial heterostructure makes these alloys an important material in semiconductor device fabrication. Understanding the surface reaction processes involved is critical for controlled epitaxial growth of these alloys. It has been shown that the addition of Ge to Si surfaces can enhance the SiGe alloy CVD growth rate significantly, even when the Ge content is low. This indicates an accelerated removal of surface hydrides to liberate more reactive sites, namely surface dangling bonds. Weakened Si-H bonding and strengthened Ge-H bonding have been observed when Ge is introduced to Si surfaces. In the studies presented here, various Ge coverages are prepared onto Si(111) and Si(100) single crystal surfaces using digermane adsorption. Following digermane exposure, the Si surface is heated to fully decompose all germanium hydride species and to desorb all hydrogen. The annealed Ge/Si surface is subsequently exposed to H atoms using H2 and a hot tungsten filament. Auger Electron Spectroscopy is used to determine Ge surface coverage. Multiple Internal Reflection Infrared Spectroscopy (MIRIRS) is then used to study the effect of Ge on the adsorption and desorption of hydrogen from these Ge/Si surfaces as a function of surface structure, Ge coverage, hydrogen coverage, and surface temperature. During hydrogen adsorption, with increasing Ge coverage, the total Si-H IR peak area decreases as the Ge-H IR peak area increases. Upon heating the Ge/Si surface, the Ge-H IR peak area decreases until totally removed by 600K, while the Si-H IR peak area increases initially between 300-500K, then decreases until totally removed by 800K. These changes indicate the diffusion of hydrogen from Ge atoms to Si atoms during hydrogen desorption. The effect of surface structure and electronic interactions between Ge and Si on the observed hydrogen vibrational frequencies and the hydrogen desorption behavior will be discussed. |
|
EM-TuP-7 The Effect of UV Irradiation on the Reaction of Disilane and Digermane with Single Crystal Silicon Surfaces
G.J. Batinica, J.E. Crowell (University of California, San Diego) The photo-induced reaction of disilane and digermane with Si(111) and Si(100) using ultra-violet irradiation has been studied using temperature programmed desorption (TPD) and Auger electron spectroscopy (AES). Hydrogen and silane desorption yields were used to determine the reactivity of disilane on the Si surfaces with and without UV irradiation. Hydrogen and germane desorption yields and relative Ge/Si AES signals were used to determine the reactivity of digermane. UV irradiation during or after dosing of the Si crystal surfaces at 110K greatly enhanced the reactivity of disilane and digermane compared to similar exposures without UV irradiation. UV irradiation during disilane exposure of Si at low temperatures results in greater Si deposition than does UV irradiation following disilane exposure. Although adsorption temperature dependent, we find that UV irradiation after dosing the Si(111) surface with digermane enhances the reactivity more than simultaneous UV irradiation at 110K. Comparison of the reactivity of disilane with Si(100) versus Si(111) show that Si(111) is thermally less reactive than the Si(100) surface. However, both surfaces show similar photo-reactivity to disilane. The photo-induced mechanism of disilane and digermane reactions has been studied using clean, partially deuterated and fully deuterated Si(111) and Si(100) surfaces. Thermally, the reactivity is controlled by the surface dangling bonds; whereas the photo- induced deposition chemistry appears to primarily involve insertion by the photo-generated diradicals, silylene and germylene. |
|
EM-TuP-8 Effects of Surface Topography in Plasma-Enhanced Silicon Dioxide Deposition from TEOS
G.J. Feng (Cypress Semiconductor Corp); J.R. Doyle (Macalester College) The effect of surface topography on TEOS-based oxide deposition is investigated experimentally and theoretically. Deposition rates and step coverage were measured on metal line structures used in SRAM devices as a function of trench width and aspect ratio. It is found that the oxide growth rate is significantly less on the top surface of the lines compared to flat wafers for fixed conditions of power, flow, and pressure. This effect can be understood from the perturbation on film precusor densities near the surface caused by radical trapping effects of the topography. Deposition profiles and effective reflection coefficients are calculated using Monte Carlo simulations. Solutions to the diffusion-reaction equation are fit to the data and under conditions appropriate for device applications an effective microscopic surface reflection coefficient of approximately 0.75 is deduced for the film precursors. The implications of this result for film quality and step coverage will be discussed. |
|
EM-TuP-9 Infrared Spectroscopy of Si-H(D) Bending Modes: A New Probe of Hydrogenated Silicon Structures
A.B. Gurevitch (Columbia University); M.K. Weldon, Y.J. Chabal, Y. Caudano, M. Collot (Bell Laboratories, Lucent Technologies) Hydrogen-terminated silicon surfaces have been extensively studied because of their ubiquity in silicon processing. Despite this interest, all spectral information to date has been obtained by analysis of the Si-H stretching modes, with the lower frequency bending modes being largely unexplored due to practical limitations. Importantly, additional information should be contained in the spectral signatures of the Si-H bending modes, given the near-degeneracy of the these modes with the Si substrate phonons, so that the observed vibrations should be sensitive to the precise details of the local environment. Indeed, changes in the strength of this coupling to the substrate modes upon deuteration have been invoked to explain the manifold increase in the reliability of silicon devices when they are subject to a deuterium post-anneal. In this work, we report the use of novel optical geometries that allow probing of the relevant Si-H(D) bending and stretching modes with high resolution and sensitivity. We have studied these modes in a wide variety of different systems, such as; i) the water-oxidized Si(100)-2x1 surface; ii) NH4F etched, H-terminated stepped Si(111) surfaces and; iii) H-implanted silicon substrates. We find that there is indeed a pronounced increase in sensitivity of the bending modes (over that observed for the related Si-H stretching modes) to the oxidation state of the surface and the surface stress. Furthermore, a detailed understanding of the behavior of the Si-H and Si-D bending modes of a wealth of structures is beginning to emerge that affords additional insight into technologically relevant processes. |
|
EM-TuP-10 Theoretical Calculations of Hydrogen Molecule Trapped in Silicon Crystal
K.G. Nakamura, M. Kitajima (National Research Institute for Metals, Japan); A. Endo, M. Kubo, A. Miyamoto (Tohoku University, Japan); K. Murakami (University of Tsukuba, Japan) The states of hydrogen in semiconductors have been extensively investigated because hydrogen is the simplest impurity and plays important role in improving the electrical properties of semiconductors. Recently we have found that hydrogen molecules are formed in crystalline silicon by a hydrogen-atom remote treatment using Raman spectroscopy (a vibrational line at 4158 cm-1). In the present sudy we performed the ab initio calculations in order to study a trapping site and a motion of the hydrogen molecule in silicon. We modeled the Si crystal lattice by the silicon cluster (Si10H16). The Hartree-Fock self-consistent field calculations were performed with the 6-31G basis set. Two hydrogen atoms were put into the cluster and the geometry optimization was performed. The geometry optimization showed that the trapping site of H2 with the minimum energy was the tetrahedral site. The H2 molecular axis was oriented almost in the <100> direction. The optimized H-H bond length was 0.706Å which is 4.6% shorter than that obtained by experiments for gaseous H2 (0.74Å). The Mulliken charge of H2 was calculated to be -0.059 and then the H2 in the cluster is almost neutral. The H2 frequency was calculated to be 4393 cm-1 with a scaling factor of 0.89. The calculated frequency is in reasonable agreement with the recent Raman experiment. The rotational barrier was also obtained to be 13 meV. H2 can rotate almost freely at room temperature. The diffusion barrier of H2 in the silicon crystal is was also calculated to be 1.5 eV. A comparison between Hartree-Fock and density functional theory calculations will be also discussed. |
|
EM-TuP-11 Defect Evolution under Thermal Anneal in ECR Hydrogen Plasma Exposed Epitaxial Si
D.Z. Chi, S. Ashok (The Pennsylvania State University); D. Theodore (Motorola Inc.) Hydrogen in Si has been a subject of intense interest because of its possible penetration in almost every processing step and its capability to alter the physical properties of Si. Previously, we reported generation of pronounced deep levels under thermal anneal in electron cyclotron resonance (ECR) hydrogen plasma exposed CZ-Si samples. However, the nature of these defects and their generation mechanism were not quite understood. To further explore the nature of these defects and their generation mechanism, recently we have studied the effects of ECR hydrogen plasma exposure on Epi-Si samples. The selection of oxygen-lean Epi-Si samples was to ascertain any possible roles of oxygen in creating the H-related defects. As in CZ-Si samples, we have observed several DLTS peaks, i.e., deep levels, in both p- and n-types of Epi-Si samples after ECR hydrogen plasma exposure and subsequent thermal anneal at 450 °C or above, while no DLTS peak was observed in as-hydrogenated samples. These deep levels disappear after high temperature anneal (>700 °C) or re-hydrogenation. The absence of of DLTS peaks after re-hydrogenation indicates that these defects can be self-passivated by hydrogen. In monitoring the defect generation process using RTA anneal, we found that these levels begin to evolve during the very initial stage of the annealing process ( less than 10 seconds at 500 °C) with different generation rates for different levels. Also, by measuring depth profiles of these deep levels, it was found that the depth profiles of these deep levels became broader with increasing annealing time, indicating some kind of diffusion of these defects during annealing process. This finding implies that the dimensional scales of the observed electrically active defects, i.e., deep levels, must be very limited, since, otherwise, one would not have observed the diffusion-like behavior of these defects during annealing process. Finally, some microscopic extended defects in the region very close to surface were observed both in p- and n-type samples using TEM measurement. A very interesting finding here is that the observed extended defects exhibit quite different annealing behavior for p- and n-type samples. For p-type samples, these extended defects were observed only in ECR hydrogenated and subsequently annealed annealed samples, while no extended defect was detected in as-hydrogenated samples. In contrast, for n-type samples, these extended defects were observed in as-hydrogenated samples with highest population, while no extended defects was observed in samples which were RTA annealed for 30 sec or longer at 500 °C. The distribution depth of these extended defects (about 0.4 um) is found much less than that of those electrically active defects (about 2.5 um), suggesting that they are not directly associated with those electrically active defects, i.e., deep levels. A detailed discussion will be carried out to explain these experimenetal results. |
|
EM-TuP-12 Gate Leakage Current: A Sensitive Characterization Parameter for Plasma-induced Damage Detection in Ultra-Thin Oxide Submicron Transistors
J. Jiang, O.O. Awadelkarim (The Pennsylvania State University); J. Werking (SEMATECH) We report the results of experiments performed to study plasma-induced damage in ultra-thin gate oxide of metal-oxide-semiconductor field-effect transistors (MOSFETs). The transistors used were 0.5 um n-channel MOSFETs with ≤65Å-thick gate oxides fabricated on 200 mm p/p+ boron-doped silicon substrate using a full complementary MOS (CMOS) flow up to and including metal-2 processes. Our results show that the three prominent transistor parameters, the threshold voltage Vth, the maximum transconductance Gm and the subthreshold swing S, become ineffective in discerning charging damage in these thin oxide MOSFETs. In contrast, the gate leakage current Ig, measured at a gate-to-substrate voltage of 2 V, is shown to emerge as the more sensitive damage monitor in this case. The sensitivity of Ig is attributed to trap-assisted direct tunneling which is enhanced by decreasing oxide thickness: in the oxide thickness range used in these experiments direct tunneling increases the magnitude of Ig (1pA/um2 ≤ Ig ≤ 1uA/um2)compared to Ig Ã1fA/um2 for thickness ≥ 90Å oxides making it more observable. The sensitivity of Ig in measuring charging damage is demonstrated using special MOSFET structures that utilize charge antennas at the polycrystalline silicon gate definition etch, contact etch, and metal etch steps.We also show that the large Ig in thin oxides can degrade the sensitivity of charge pumping measurements which are commonly used to examine the condition of oxide/silicon interface in MOSFETs. |
|
EM-TuP-13 Characterization of RuO2 Etch Using a High Density TCP Plasma Source
M.A. Kennard, Y.M. Melaku (Lam Research Corporation) Alternative capacitor materials will be an enabling technology in the production of 1 Gbit and 4 Gbit devices. These dielectric and ferroelectric perovskite materials require electrode materials, such as RuO2 and Pt, which are not currently used in IC manufacturing. The process of patterning RuO2 films to be used as lower electrodes is described. An inductively coupled rf source was used in the ion assisted etch of RuO2 films. In contrast to RuO2 films masked with silicon dioxide, etching of photoresist patterned RuO2 is dominated by physical sputtering, with only a minimal chemical contribution. The major challenge is to control the redeposition of etch product, which is crucial to achieve CD control, vertical profiles and no residues. We have evaluated several chemistries in combination with argon, and multiple frequency plasma generation. Effective pattern transfer was achieved with high etch rates, selectivity to photoresist better than 1:1, and greater than 70 deg profile. The etch results were strongly correlated to the frequency of the applied power. Mass spectroscopy and optical spectroscopy were used to characterize the etch products. The behavior and reactivity of the redeposited etch product was also studied. |
|
EM-TuP-14 Mass Spectroscopy of Recoiled Ions (MSRI) for In Situ and Ex Situ Analysis of Thin Films and for Real Time Control of Growth Parameters
J.A. Schultz (Ionwerks) MSRI is a newly emerging technique which is now being successfully used exsitu in conjunction with static and dynamic SIMS for analysis of materials. MSRI provides complementary information particularly when only elemental or isotopic identification is desired. A newly developed reflectron time of flight mass spectrometer has been shown useful for either detection of 10 eV secondary ions (SIMS) or of keV directly recoiled surface atoms ionized during direct binary collision sequences with the impinging primary ion beam. The resulting mass spectrum of recoiled ions (MSRI) comprise elemental ions and their isotopes and are devoid of isobaric molecular ions. Resolutions are typically around 200 which is sufficient for trace element analysis since no interferring molecular isobars survive at the keV energies of the recoils. Insitu and realtime process control are also possible with MSRI. Surface analysis can be performed at pressure-path lengths of 50 mTorr-cm which allows many thin film surfaces to be interrogated during growth. Examples of this aspect of the technique will be given. |
|
EM-TuP-15 Fluorination of Si(001)-[2x1] Surface near Step Edges via Molecular Dynamics Simulations
D. Srivastava, T. Halicioglu (NASA Ames Research Center); T. Schoolcraft (Shippensburg University) Fluorination of a dimerized Si(001)[2x1] surface near step edges has been studied with molecular dynamics simulations using Stillinger-Weber1 potential for Si-Si and Si-F interactions. The deposition, adsorption and distribution dynamics of F atoms and molecules on three types of step edges2 have been investigated in detail. Using these studies we explore the role of surface defects such as step edges in the initiation of F etching on Si(001)[2x1] surface3.
|
|
EM-TuP-16 Low Energy Cathodoluminescence Spectroscopy of Etched 6H-SiC Surfaces
A.P. Young, L.J. Brillson, J. Jones (The Ohio State University) We have performed low energy cathodoluminescence spectroscopy (CLS) measurements of localized states near 6H-SiC (0001) and (000-1) surfaces under ultrahigh vacuum (UHV) conditions and with varying depth sensitivity. CLS reveals several new electronic states deep within the SiC band gap, at least two of which exhibits strong surface-dependent properties. C- and Si-surfaces grown by Cree Research with n = 1.6 x10e18 cm-3 were cleaned with a degrease / 10:1 DI water:HF clean sequence , then dried in nitrogen and mounted in vacuum. Electron beam excitation provides an effective means to promote high densities of electron- hole pairs across the relatively wide SiC band gap (3.03 eV) and to vary the maximum energy loss below the solid surface for near-surface vs. bulk sensitivity. We employed a Ge photodiode with a low-energy response cut-off of 0.7 eV to monitor the optical emission resulting from band-to-deep level and deep-level-to-band free carrier recombination generated by the incident electron beam. The combination of surface cleaning and UHV conditions yielded pronounced near-band-edge (3.03 eV) emission at 90 K, but also a set of deep levels with substantial emission intensity from only a few tens of nanometers from the free surface. CLS emission peaks at 1.6, 1.1, and 0.95 eV, with a shoulder at 0.8 eV, are evident over the energy range 1 -2.5 keV. More surface-sensitive spectra at the lower incident beam voltages produce a relative increase in the 1.6 and 1.1 eV peak intensities. Such near-surface deep level features have previously been unobtainable without UHV conditions to minimize surface as well as beam-induced contamination and associated recombination channels ("dead layers"). These measurements reveal the sensitivity of UHV-CLS techniques to localized states at SiC surfaces and offer a powerful tool to assess electronic structure as a function of chemical preparation at SiC interfaces. |
|
EM-TuP-17 Fermi Level Pinning on Chemically Etched and H-Terminated Silicon Surfaces Investigated by Photoelectron Spectroscopy
R. Schlaf (Colorado State University); M. Fujitani, R. Hinogami, S. Yae, Y. Nakato (Osaka University, Japan) A widely used approach to obtain smooth oxide-free (partially) H-terminated silicon (Si) surfaces is to immerse Si wafers into CP4A (a mixture of H20, HNO3, CH3COOH and HF in volume ratio of 22:5:3:3) and/or HF solutions of varying concentrations. It is usually assumed that such treatments result in a drastic reduction of the surface density of states (SDOS) and that, therefore, no surface band bending may occur. In our experiments we investigated the electronic surface structure of a number of CP4A/HF treated n- and p-Si wafers with varying doping densities by X-ray Photoelectron Spectroscopy (XPS). XPS allows the straight forward detection of surface stoichiometry as well as band bending and surface photovoltages (SPV) on semiconductor materials because the positions of the core level peaks directly depend on the position of the Fermi level within the band gap at the surface. Our experiments show that on all investigated surfaces still Fermi level pinning exists after the samples were immersed in the CP4A/HF solutions and that the pinning states are located close to the conduction band. Most of the samples also showed SPV when measured under illumination. The measurements also show that the surfaces contain a considerable amount of flourine atoms depending on the treatment and the doping density. From the amount of band bending we estimated the density of surface states present on the various samples. |
|
EM-TuP-18 Adsorption of Carbon Tetrachloride on GaAs(001) Surface: Etching and Carbon Incorporation
B.K. Han, L. Li, H. Qi, S. Gan, R.F. Hicks (University of California, Los Angeles) Adsorption of carbon tetrachloride (CCl4) on GaAs(001) c(2x8), (1x6) and c(8x2) reconstructions was studied by X-ray photoemission spectroscopy (XPS), temperature-programmed desorption (TPD), and scanning tunneling microscopy (STM). We have found that CCl4 dissociatively adsorbs on GaAs (001) surface at 30 to 200 C The uptake of CCl4 is proportioned to the Ga coverage and increases with temperature. Above 425 C, the adsorbed CCl4 decomposes by desorbing Cl2 and GaCl. No CClx species are detected by TPD, indicating that the carbon incorporates into the GaAs. Scanning tunneling microscophs obtained after CCl4 adsorption at 30 C, followed by annealing at 450 C reveal that the surface contains an interesting mixture of As-rich and Ga-rich phases. The As-rich regions and etch pits produced by GaCl desorption. They are c(2x8) terminated and are located at the step edges. Consequently, the Ga-rich regions consist of flat terraces which are (3x2)/c(6x4) reconstructed. This latter phase is stabilized by carbon doping. |