AVS1996 Session EM-TuP: Electronic Materials and Processing Poster Session I

Tuesday, October 15, 1996 6:30 PM in Ballroom A

Tuesday Evening

Time Period TuP Sessions | Topic EM Sessions | Time Periods | Topics | AVS1996 Schedule

EM-TuP-1 Fluorine Atom Induced Decreases in Macroscopic Infrared Effective Charges of Si-O Bond-stretching Vibrations in Low-dielectric Constant Si-O-F Alloy Films G. Lucovsky (North Carolina State University) There is considerable interest in insulators with static dielectric constants, \epsilon\\sub o\, lower than SiO\sub 2\. A material that has attracted much attention is Si-O-F: fluorinated silicon dioxide. This paper discusses bonding of F atoms in the SiO\sub 2\ network using a pseudo-binary alloy description for compositions with up to one F/Si, and identifies the microcsopic mechanism responsible for observed reductions in \epsilon\sub o\ [1]. F atoms are incorporated into the SiO\sub 2\ network by replacing twofold coordinated O atoms by terminal Si-F bonds. Dielectrics with up to one F/Si are treated as pseudo-binary alloys: (SiO\sub 2\)\sub x\(Si\sub 2\O\sub 3\F\sub 2\)\sub 1-x\, so that Si-O and Si-F bond concentrations are functions of x alone. \epsilon\sub o\ is expressed in terms of a two-band model including separate contributions from electronic transitions and IR-active vibrations. By comparing these contributions in Si\sub 3\N\sub 4\, SiO\sub 2\ and Si-O-F alloys, we show that contributions from IR vibrations decrease as the partial (or ionic) charge on the Si atoms increases. Experiments show that replacement of F for O reduces the macroscopic IR effective charge, e\sub m\, and our bonding description correlates these reductions with increases in the Si partial charge. There are two contributions to e\sub m\: i) static contributions that scale with increasing partial (ionic) charge on the Si, and dynamic contributions to e*\sub m\ from charge redistribution which decrease with increasing Si partial charge. Using a mechanical model underpinned by quantum chemistry, we show that reductions in the dynamic contribution with increasing Si partial charge are greater than any increases in the static component so that substitution of a F for an O reduces IR contributions of the other three Si-O bonds at a gvien Si site accounting for large reductions in \epsilon\\sub o\ for relatively small F additions. [1] N. Hayasaka, et al., SSDM=B995, 157 (1995).
EM-TuP-2 Process Sensing and Metrology for RTCVD Oxide Deposition from SiH\sub 4\ and N\sub 2\O G. Lu, L. Tedder, G. Rubloff (North Carolina State University) We have applied active sampling mass spectrometry to investigate chemical mechanisms and thickness metrology in oxide rapid thermal CVD processes using 10%SiH/sub 4//Ar and N/sub 2/O at ~800C. Equipment and process behavior through the short process cycle are revealed in the detailed time-dependent changes of mass spec intensities. A H/sub 2/ reaction product is clearly observed (above the level of SiH/sub 4/ cracking fragments) when elevated wafer temperature and pressure are present, even for 0.5% SiH/sub 4/ concentration, indicative of the SiH/sub 4/ decomposition reaction to Si + 2H/sub 2/. In contrast, H/sub 2/O products are not observed, at least down to a level 10/super 3/X below that of the H/sub 2/ product. Thus it appears that the oxide deposition proceeds in two steps, SiH4 decomposition on the surface, followed by N/sub 2/O oxidation of the deposited Si. The integrated H/sub 2/ signal is found nearly proportional to the deposited oxide thickness, providing the basis for a real-time, noninvasive thickness metrology even at the low (0.5-2.0%) SiH/sub 4/ concentrations required for high oxide quality. Since the large oxidant excess (N/sub 2/O) is presumably needed to fully oxidize the deposited Si, it appears that oxide thickness can be monitored by the Si deposition extent itself. The N/sub 2/O excess represents environmental waste and contributes a background signal which may preclude observation of other N-O reaction products for environmental sensing. The results for the SiH/sub 4/-N/sub 2/O CVD oxide process may be indicative of more general behavior in bicomponent CVD reaction.
EM-TuP-3 Exploring the Mechanism for PECVD of Fluorinated Silicon Dioxide Films with Low \epsilon\ Dielectric Properties N. Salvo, K. Bogart, E. Fisher (Colorado State University) Miniaturization of electronic circuits with increased complexity and multilevel metal layers demand new materials with low dielectric constants. This need arises from the fact that chip speed is limited by both the resistance of the metal line and the surrounding insulator. At present, fluorinated oxides provide the easiest way to achieve low dielectric constant materials while maintaining good step coverage. We are exploring the mechanism for plasma enhanced chemical vapor deposition (PECVD) of fluorinated silicon dioxide in several different ways. First, we have used a variety of starting materials such as flurotriethoxysilane (FTES) and SiF\sub 4\/O\sub 2\ to explore film deposition. The effects of applied power, reactor pressure, substrate temperature and added oxidant are presented. In the FTES system, we find the substiutents on the starting material are retained in the deposited film and addition of O\sub 2\ decreases deposition rate. This system is compared to results from other novel alkoxysilane systems. Second, in order to study systems that may incorporate some SiH component with SiF, we add H\sub 2\ to our systems. Most importantly, we are using the IRIS (Imaging of Radicals Interacting with Surfaces) technique, a method that combines molecular beams and laser-induced fluorescence, to measure the surface reactivity of radicals of interest to the deposition process. Specifically, we have measured the reactivity of SiF and OH during deposition of F-SiO\sub 2\ from SiF\sub 4\/O\sub 2\ and FTES plasmas on SiO\sub 2\ and Si\sub 3\N\sub 4\ substrates. Comparison of OH reactivities from the FTES system to those from other alkoxysilane systems where we find a surface reactivity of >50% will be made. The substrate temperature dependence as well as implications for the deposition mechanism will be addressed. Preliminary results for other radicals such as SiH and SiF\sub 2\ may also be presented. These studies, made during the deposition process are vital to understanding the mechanism for formation of low \epsilon\ dielectric materials.
EM-TuP-4 Vapor Deposition Polymerization of Polyimide for Microelectronic Applications V. Malba, V. Liberman, A. Bernhardt (Lawrence Livermore National Laboratory) The electrical performance of the next generation of ULSI devices will depend on the continued development of low dielectric constant (k) materials . These low k materials must be compatible with all other semiconductor processes, and must provide conformal coverage of typical topographical features. Polyimide has been the organic dielectric of choice because it has a high thermal and chemical resistance, and has a lower k than silicon dioxide. However, spin-cast polyimide tends to shrink as much as 30% during cure, resulting in films which neither planarize, nor conform well to small features. Because most of the shrinkage results from solvent evaporation, vapor deposition polymerization (VDP) of polyimide solves this problem. Our results show that minimal shrinkage occurs during cure. Further, the cure time and temperature can be reduced with this solventless process. While most VDP research to date has concentrated on small sample sizes, we have developed a VDP system which has been used to deposit polyimide on 4" wafers with excellent uniformity and throughput. By adopting a modified evaporator chamber with rotating planetary, we have achieved acceptable conformality for both very large (500 micron) and very small (1 micron) features. We will present the details of the VDP system, the chemical and physical properties of films using FT-IR, AFM, internal reflectance spectroscopy, and prism wave coupler, and will give examples of VDP films used in ULSI and in electronics packaging. The properties of films from a number of different monomer systems will be compared with respect to their suitability for electronic applications.
EM-TuP-5 Remote Microwave Plasma Enhanced CVD of Low Dielectric Constant SiO\sub x\F\sub y\ Films from FASi-4 and Oxygen Z. Jin, M. Virmani (Arizona State University); G. Leusink (Materials Research Corporation); G. Raupp, T. Cale (Arizona State University) Reducing the dielectric constant is one way to reduce the RC delay. Studies show fluorine incorporation into ILD films reduces their dielectric constant from 4.0 (without fluorine) to 3.6 (with fluorine incorporation). FASi-4 has direct Si-F bond in the molecule. This may enhance the fluorine incorporation. We use mixtures of FASi-4 and oxygen to deposit SiO\sub x\F\sub y\ films, using a remote microwave plasma enhanced CVD system. For the deposition conditions studied, the deposition rate increases with increasing substrate temperature, increasing microwave plasma power, and decreasing reactor pressure, and is not sensitive of the O2:FASi-4 flow ratios. A maximum deposition rate of ~120 \Angstron\/min was obtained at a substrate temperature of 450 \degree\C, microwave plasma power of 250 W, and a reactor pressure of 0.25 Torr. RBS results indicate that decreasing the O2:FASi-4 ratio from 15:1 to 8:1 increases the fluorine content in the films from 2.0 to 8 at. %. FTIR results shows the exist of Si-F bonding at 935 cm-1, no measurable moisture absorption, and the peaks of the Si-O bond shifted to the higher wave number for the deposited SiO\sub x\F\sub y\ films which indicate the bonding is more stable. The dielectric constant of the films is found to decrease from 3.68 to 3.6 with an increase in the fluorine content in the films from 2 to 8 at. %.
EM-TuP-6 A Statistical Evaluation of the Precision of XPS Metrology of Gate Oxides for CMOS Devices J. Bernard, E. Adem (AMD) The shrinking dimensions of IC geometries lead to ever greater materials challenges. Current CMOS geometry includes structures where a higher percentage of the atoms/molecules can be considered as "surface" than "bulk". The widely accepted SEMATECH Technology roadmap calls for 34\Ao\ gate oxides by the year 2004. Typical CMOS process control calls for no more than 3 sigma variance, which is 2\Ao\ for a 30\Ao\ oxide. The oxide thickness determines crucial properties such as threshold voltage. Ellipsometry is currently used for in-line oxide thickness measurements, but is unreliable below 40\Ao\. The shallow sampling depth and low energy primary radiation of X-ray photoelectron spectroscopy(XPS) are well suited to characterization of ultra-thin gate oxides. We investigated oxide thicknesses in the range of 20\Ao\ to 100\Ao\. The static effect of X-ray source exposure time on the i)stoichiometry of the film, ii)surface contamination, and iii)calculated thickness was examined. The dynamic precision of multiple separate measurements of samples over time was determined. The potential of thickness mapping across a 200mm wafer will be explored. Finally, we will discuss the viability of XPS as an in-line oxide metrology tool.
EM-TuP-7 Ba\sub 1-x\Sr\sub x\TiO\sub 3\(BST) Thin Films Made by Turbodisc PE-MOCVD Techniques T. Li, R. Stall (Emcore Corporation) Ba\sub 1-x\Sr\sub x\TiO\sub 3\(BST) is a leading candidate as a new dielectric material in capacitors for the next generation of ultra-large-scale integrated dynamic random access memories (DRAMs). However, it is difficult to deposit high quality ultrathin BST films on large size wafers at low temperature. In order to solve these problems, liquid delivery system, TurboDisc and Plasma Enhanced MOCVD techniques are combined for deposition of BST thin films. Two step processes are also used to improve interface mismatch between film and substrate. The first step is a nucleation step to make a very thin nucleation layer of the desired materials or buffer layer at a low temperature; the second step is grain growth step that grain grows from the nucleation layer or buffer layer at a higher temperature. The experimental results show that BST ultrathin films on 4" wafers with uniformity of better than 4% and with good electric properties at deposition temperature 700øC. Typically, the about 1000 \Ao\ thick Ba\sub 0.6\Sr\sub 0.4\TiO\sub 3\ thin films on Pt electrodes have dielectric constant around 600, and a leakage current of less than 1x10\super -7\A/cm\super 2\ at 100 KV/cm\super 2\ and room temperature. In addition, the relationship among compositions, microstructure and electric properties of BST thin films were also investigated.
EM-TuP-8 Composition Control of MOCVD BST S. Bilodeau, R. Carl, P. VanBuskirk (Advanced Technology Materials); D. Kotecki (IBM Microelectronics Division) In the last several years it has become evident that the dielectric properties of BST make it attractive for a variety of integrated capacitor applications. For applications such as DRAMs, conformality is also a requirement, driving the need for a manufacturable MOCVD BST process. The dielectric properties of this material are known to be strongly dependent on the A:B site ratio. For this reason composition control is critical in developing a manufacturable process. We have examined the sensitivity of the MOCVD BST process to process flows, substrate temperature, and growth pressure using central composite experimental design methodology. Composition and thickness were the primary responses and were measured using x-ray fluorescence for over 200 different deposition conditions. Regions of process space have been identified where the composition of thin BST films are controllable to <0.25 At% Ti.
EM-TuP-9 Luminescence Spectra of Nanocrystalline Ba\sub 1-x\Pb\sub x\TiO\sub 3\ R. Katiyar, J. Meng (University of Puerto Rico) Barium titanate (BaTiO\sub 3\), one of the most important ferroelectric semiconductor materials, has been extensively studied by a variety of techniques. However, the luminescence of BaTiO\sub 3\ has not yet been understood fully. In our present experiments, we have, for the first time, measured the luminescence spectra of nanocrystalline Ba\sub 1- x\Pb\sub x\TiO\sub 3\ (PBT) prepared by a sol-gel method. The spectra were obtained using a triple grating monochromator attached with a microscope (ISA Mode T64000) with an excitation line of 514.5 nm of an Argon ion laser, indicating an apparent intrinsic emission centered at 2.07 eV at room temperature. The intensity of the visible luminescence in PBT decreased and its peak position shifted to a higher energy side upon increasing the Pb concentration. The Raman spectra and powder x-ray diffraction for PBT with grain size of 40 nm revealed that the dependence on the Pb concentration of the visible luminescence band in PBT is closely connected with the PBT structure transformation from the cubic phase of BaTiO\sub 3\ to the tetragonal phase of PbTiO\sub 3\. Acknowledgement: This work was supported by NASA-NCCW-0088, DE- FG02-94ER75764, and NSF-OSR-9452893 Grants.
EM-TuP-10 Nitridation of Al\sub 2\O\sub 3\ Surfaces by Control Enhanced Compact ECR Plasma P. OKeeffe, H. Mutoh, S. Den, Y. Hayashi (Irie Koken Co., Ltd, Japan); S. Komuro, T. Morikawa (Toyo University, Japan) Despite the extensive use of electron cyclotron resonance(ECR) plasma in group III-nitride growth procedures very little quantitative information is available with regard to the exact plasma-material interaction processes involved. At this conference an advanced MBE compatible compact ECR source with enhanced control features is described. Nitrogen plasma characteristics as a function of discharge magnet position are evaluated using Langmuir probe diagnostics, ion energy analysis and emission spectroscopy. Selective control of the ion energy distribution independent of the plasma density is demonstrated for the first time in a compact ECR plasma source. The ion energy can be tuned from 10eV to 30eV. Nitrogen plasma induced surface reaction processes on Al\sub 2\O\sub 3\ substrates were investigated as a means of converting the Al\sub 2\O\sub 3\ surface to that of AlN, thereby providing a AlN buffer layer for nitride growth. The surfaces were analyzed by RHEED, XPS and SEM. Prior to ECR plasma nitridation in-situ surface cleaning of the substrates was performed by hydrogen plasma for 30min. Plasma characteristics combined with XPS results suggest that it is the atomic species and not the ionic species which are the dominant reactive species responsible for nitridation of the surface. However SEM photographs reveal that the morphology of the Al\sub 2\O\sub 3\ surface can be directly related to the plasma ion energy. It is demonstrated that control of the average ion energy independent of plasma flux provides enhanced surface morphology while maintaining a high rate of nitridation.
EM-TuP-11 Full Characterization of AlxGa\sub 1\\sub -\\sub x\N/In\sub x\Ga\sub 1\\sub -\\sub x\N/GaN LED Devices by SIMS Y. Gao, S. Mitha, J. Erickson, C. Huang (Charles Evans & Associates) SIMS (Secondary Ion Mass Spectrometry) is applied to the characterization of AlxGa1-xN /InxGa1-xN/GaN LEDs. We used a combination of MCs+ SIMS and normal SIMS for our analysis. MCs+ SIMS is a specialized technique recently developed for the quantification of matrix composition and some trace elements. We describe depth profiling for the concentration of p- and n-dopants (Mg, Zn, Si doping and diffusion profiles) and impurities (C, O, H and metals contamination ), determining In and Al compositions in the active layer (InxGa1-xN/GaN quantum well) and AlxGa1-xN confinement layers, as well as the sharpness of the interface and the thickness of each layer. The wealth of information so obtained has proven very useful for solving problems encountered in both research and production of GaN LEDs.
EM-TuP-12 Study of the Controlled Surface Oxidation of Zinc Cadmium Telluride by Atomic Force Microscopy and X-ray Photoelectron Spectroscopy A. Burger, M. George, H. Chen, K. Chen, W. Collins (Fisk University) We report results on X-ray photoelectron spectroscopy (XPS) and atomic force microscopy measurements on cadmium zinc telluride crystals (CdZnTe) and describe a model of the chemical and morphological modifications of the surface occurring after H2O2 etching. In this study, CdZnTe wafers were etched with the standard Br:methanol solution and then followed by an etch in 15% H2O2 at room temperature for 5, 10, 15 and 75 minutes. Although the standard Br:methanol etchant is an efficient way of removing the surface damaged layer and producing smooth surfaces it also leaves a non-stoichiometric, tellurium rich layer. The present XPS data establishes the possibility of a controlled surface oxidation that results in a significant reduction of surface leakage currents of room temperature X-ray CdZnTe detectors. Atomic force microscopy (AFM) was employed to determine the surface roughness factors which were correlated to surface leakage currents and detector performance.
EM-TuP-13 The Interaction of Base-stabilized Tris(trimethylsilyl)aluminum and Ammonia on Clean and Oxidized Al(111): Growth of Aluminum Nitride D. Robinson, J. Rogers, Jr. (University of Washington); R. Paine (University of New Mexico) Fueled by the desire to deposit aluminum-containing thin films with low impurity levels and at low temperature, researchers continue to develop new types of organometallic precursors. One such group of novel precursors includes alkyl-silyl aluminum compounds. Tris(trimethylsilyl)aluminum (TTMSA) stabilized with diethyl ether has been synthesized and characterized. The adsorption of TTMSA on clean and oxidized Al(111), its subsequent thermal decomposition, and reaction with ammonia to form aluminum nitride has been studied. Temperature-programmed desorption (TPD), Auger electron spectroscopy (AES), and reaction scattering were used in this investigation. The expected leaving groups (diethyl ether and trimethylsilane) are weakly adsorbed on clean and oxidized A1(111). Diethyl ether desorbs molecularly from Al(111) at ~180K, and trimethylsilane desorbs molecularly at ~115K. The nonreactivity of both of these molecules with clean and oxidized Al(111) suggests that TTMSA's adsorption and decomposition proceeds with low impurity levels. *This work was supported by the National Science Foundation under Grant No. CTS-9303974
EM-TuP-14 Formation of Compensating Defects in Si-doped GaAs C. Domke, P. Ebert, M. Heinrich, K. Urban (Institut fur Festkorperforschung, Julich, Germany) A primary problem hampering applications of compound semiconductors in optoelectronic devices is a lack of doped crystals with reproducible and high carrier concentrations. The limited carrier concentrations are due to a compensation of the dopant atoms. We report that atomically resolved scanning tunneling microscopy should be a generally applicable method of identifying quantitatively the microscopic condensation mechanisms of dopant atoms in compound semiconductors. We demonstrate the methodology on Si-doped GaAs. We identify all defects occurring in STM images of (110) cleavage planes of bulk GaAs crystals and measure the surface densities of Si\sub Ga\ donors, Si\sub As\ acceptors, Si\sub n\ clusters, Ga vacancies (V\sub Ga\) and Si vacancy complexes. After the classification of all observed defects we separate surface defects formed thermally after cleavage (by Langmuir desorption), in order to determine the bulk defects exposed on the surface by cleave of the crystals. This allows us to deduce the critical Si concentrations of the consecutive compensation of Si donors by Si\sub As\ acceptors, Planar Si clusters and (Si\sub Ga\ -Ga-vacancy) complexes. We observe the concentration of charged donors and acceptors does not increase significantly even if the Si concentration increase by more than an order of magnitude. In contrast the concentration of Si incorporated in the uncharged Si pairs and clusters increases strongly. This contradicts the current theoretical expectations. We develop a microscopic model based on the screened Coloumb interaction between charged acceptors and donors, the amphoteric nature of Si, and the Fermi-level-effect. It explains the observed defects, the critical Si concentrations of each identified defect formation mechanisms and predicts the solubility limit of Si in GaAs.
EM-TuP-15 Growth of Hexagonal Wurtzite GaN by Rapid Radiation Heating, Metal-organic Chemical Vapor Deposition H. Shi, B. Shen, K. Yang, Y. Zheng (Nanjing University, P.R. China) Recently, extensive investigations have been done on gallium nitride (GaN) due to its potential applications in electronic and optoelectronic devices, such as UV-blue light emitting diodes, metal-semiconductor field effect transistors, and laser diodes. However, the intrinsic obstacles, e.g. lattice mismatch, different thermal expansion coefficients between substrates and the epilayers, high dislocation densities in the films, etc. have greatly hindered the development of this field. Therefore, other techniques should also be explored in order to increase the epitaxial film quality. In this paper, we report the growth of single crystal GaN films on (0001) Sapphire and (111) Si wafers by Radiation Heating, Metal-organic Chemical Vapor Deposition. Ammonia and Trimethylgallium (TMG) are the main reaction sources, with H2 or N2 as the carrier gas. GaN buffer layer with a thickness of about 300 Angstrom was initially prepared at 820K. After that, crystalline GaN film was grown at about 1120K. Typical growth rates of GaN films on Sapphire and Si substrates are 2 and 6 microns/h, respectively. Raman spectra (with E2 peak at around 569 1/cm), XRD spectra, and SEM, etc., have revealed that the epitaxial films are of hexagonal wurtzite structure and of high quality. It is supposed that the radiation of light in the visible region plays an important role in the growth of GaN films, which is helpful to the increment of N-atoms in the reaction chamber, producing an enormous effect on Ga-N bonding.
EM-TuP-16 Effects of Nitrogen and Hydrogen on the Electrical and Optical Properties of rf-sputtered ZnTe:Cu Thin Films T. Gessert, A. Mason, P. Sheldon, T. Coutts (National Renewable Energy Laboratory) It is observed that typical contaminant gases in an Ar sputtering environment can significantly effect the electrical and optical properties of the rf-sputtered ZnTe:Cu thin films. For ZnTe films containing ~6 at.% Cu, and ambients containing 0.2 vol.% nitrogen (in Ar), the resistivity remains unchanged (at ~1 Ohm-cm) but the carrier type is observed to switch from p- to n-type. For ambients containing higher concentrations of nitrogen (i.e., 0.7 vol.% and 1.3 vol.%), the resistivity increases significantly (to 1 kOhm-cm and 15 kOhm-cm, respectively). Although these observation help explain the sensitivity of the target's surface to contamination by air, it is a curious result in view of the success with nitrogen doping in MBE-grown ZnTe. The optical band gap for ZnTe:Cu films is observed to decrease from ~2.25 eV to ~2.05 eV when the ZnTe:Cu film is sputtered in 1.3 vol.% nitrogen. A similar increase in resistivity is observed for films containing ~0.5 at.% Cu when hydrogen is admitted into the Ar sputtering ambient. However, in this case the optical band gap increases from 2.25 eV to 2.75 eV when 1.2 vol.% hydrogen is added to the sputtering ambient. Residual effects of hydrogen contamination are observed after only 1 hour of pumping hydrogen. XPS measurements of films deposited in hydrogen indicate phase-segregated regions in which the Zn is either in a Zn-Te bond or a Zn-O bond. Similar analysis for films deposited in pure Argon indicate a homogeneous phase in which the Zn is bonded as ZnO. The above results suggest that the effects of these gases may be usefully exploited to produce ZnTe films that are better suited to particular device applications.
EM-TuP-17 Wet Oxidation of Al\sub X\Ga\sub 1-x\As for Optoelectronic and Metal-Insulator-Semiconductor (MIS) Applications: Temporal Evolution of Composition and Microstucture C. Ashby, J. Sullivan, N. Missert, P. Newcomer, H. Hou, A. Baca (Sandia National Laboratories) Wet thermal oxides have been employed in high-efficiency edge-emitting and vertical-cavity surface-emitting semiconductor lasers. Morphology of the oxide and electronic properties at the oxide/semiconductor interface affect device operation. Thermal oxidation of an Al\sub x\Ga\sub 1-x\As surface layer has also shown promise for metal-insulator-semiconductor (MIS) applications. Time- and temperature-dependent studies of the wet thermal oxidation of Al\sub x\Ga\sub 1-x\As (1\>=\ x \>=\ 0.90) on GaAs have been performed to examine the oxidation process and the microstructures and electrical properties of the resulting oxide layers. Two important oxidation regimes have been identified. The first regime involves the oxidation of Al and Ga in the Al\sub x\Ga\sub 1-x\As alloy to form an amorphous oxide layer, as determined by thin-film X-ray and electron diffraction and by Raman spectroscopy. The second regime involves the oxidation and elimination of residual elemental As (both crystalline and amorphous) and a-As\sub 2\O\sub 3\, which have been observed to remain in the oxide layer after the formation of the amorphous Al-oxide appears complete. Further oxidation time or annealing can lead to crystallization and delamination of the oxide film. Residual As plays a vital role in determining the electronic properties of the oxide film. DC transport and capacitance measurements performed on MIS diode structures reveal that residual As can result in up to a two order of magnitude increase in leakage current and up to a 30% increase in the dielectric constant. Residual As also strongly affects the interface-state density of the oxidized Al\sub x\Ga\sub 1-x\As/GaAs interface, as manifested by the onset of strong Fermi-level pinning and high leakage currents. This work was performed at Sandia National Laboratories and supported by the U.S. Department of Energy under Contract No. DE-AC04-94AL85000.
EM-TuP-18 Nickel Doping of Boron Carbide and Corresponding Fermi Level Shifts D. McIlroy (University of Idaho); S. Hwang, N. Remmes, P. Dowben (University of Nebraska) Using the technique of plasma enhanced chemical vapor deposition (PECVD), we have successfully grown Ni doped boron carbide. The source compound closo-1,2-dicarbadodecaborane (orthocarborane) has been successfully used to grow rhombohedral B\sub 5\C p-type materials. With appropriate partial pressures of nickelocene, nickel can be incorporated into the boron carbide to form mildly and strongly doped n-type Ni-boron carbide films. Simple diodes constructed with these materials exhibited tunnel diode character. An understanding of the transformation of rhombohedral B\sub 5\C from p-type to n-type can be understood by examining the electronic structure of metal doped molecular films of orthocarborane. With Na doping, new densities of occupied and unoccupied states are introduced into the highest occupied-lowest unoccupied molecular orbital gap. At a high level of Na doping, a large density of unoccupied states develop at the Fermi level, establishing the n-type character of the Na-doped molecular films. Since the icosahedral molecular structure of orthocarborane is analogous to structures observed for boron rich materials, these results give a qualitative explanation of the transformation of B\sub 5\C from p-type to n-type with nickel doping.
EM-TuP-19 Si & Ge Nanocrystal Formation in an Inductively Coupled Plasma Reactor C. Gorla, S. Liang (Rutgers University); G. Tompa (Structured Materials Industries, Inc.); W. Mayo, Y. Lu (Rutgers University) We have studied Si and Ge nanopartical formation in Silane-Ar and Germane-Ar plasma discharges respectively. These discharges were generated in a helical resonator type Inductively Coupled Plasma (ICP) reactor. Generation of group IV nanocrystals, which emit light in the visible range is the main goal of our research. We have observed Si particles varying in sizes from 5 nm to 15 nm under different conditions. The particles were mostly spherical and made up of a crystalline core with a 1-2 nm thick amorphous shell. The size distribution was narrow for particles formed at a pressure of 200 mtorr, plasma power of 400 W and Silane flow rate of 20 sccm (+ 980 sccm Ar). Conditions for producing particles smaller than 5 nm and having a narrow size distribution are being investigated. Results on Ge nanocrystals formed in this plasma reactor will also be presented. These types of ICP reactors are being developed as high density plasma sources for deposition and etching. Contamination from particles generated in the plasma are a continuing problem in the semiconductor industry. Another goal of our study is to determine conditions which prevent particle formation in the helical resonator reactor. Under very low flow rates and under high plasma powers, the Si particle density decreased considerably and a film like deposition occurred on TEM grids placed in the reactor for collecting the particles.
EM-TuP-20 Observation of Photoluminescence from Germanium Doped Aluminum Nitride Films E. Forsythe (Structured Materials Industries, Inc.); J. Sprague (Naval Research Laboratory); S. Metha (Lehigh University); D. Morton (Army Research Laboratories); B. Khan (N.A. Philips Laboratories); G. Tompa (Structured Materials Industries, Inc.) The observation of light emission from nanocrystal based materials, such as porous silicon and Si nanocrystalline based films has provided much interest in novel light emitting devices for display and electro-optic applications. We report here the first observation of photoluminescence from heavily Ge doped and annealed AlN films. A dual ion beam assisted deposition process prepared the Ge doped AlN films. The films were post annealed at temperatures ranging from 800C to 1000C in a nitrogen atmosphere. The photoluminescence results are reported for two Ge concentrations, each annealed at different temperatures. The results demonstrate an emission peak at 780nm for annealing temperatures at 1000C. By comparison, the photoluminescence emission was much weaker for annealing temperatures below 1000C. We speculate that below 1000C germanium nitride phases are formed in the film, which disassociate upon annealing temperature at 1000C, in accordance with germanium nitride phase diagrams. In addition, the photoluminescence has an emission peak below 500nm for all annealing conditions attributed to the matrix. The photoluminescence excitation spectroscopy results are reported for the two spectral regions. The films were characterized by TEM and SIMS. Light emission from Ge based precipates in an AlN matrix demonstrates the possibility for novel electroluminescing devices from these nanoscale semiconductor based materials.
EM-TuP-21 Use of Surface Photoabsorption to Study and Monitor the GaAs(001) Surface R. Eryigit, S. Vemuri, J. Eng, Jr., H. Fang, B. Bent, I. Herman (Columbia University) The use of surface photoabsorption (SPA) as a probe of the surface of GaAs(001) is described, both as a real-time monitor during etching and for analysis of the optical response of the surface for different terminations. In experimental studies, SPA is used to monitor the GaAs(001) surface during chemical dry etching by a modulated molecular HCl beam. The reflectance of p-polarized 488-nm light incident near the pseudoBrewster angle is seen to track the relative Ga- and As- richness of the surface during the on and off cycles of the HCl beam, and is seen to correlate with mass spectrometric monitoring of desorbing GaCl and As\sub2\. The more general use of SPA to study the optical properties of GaAs is also described. Specifically, experimentally-available SPA data are used to determine the dielectric tensors of the surface regions of GaAs(001), with Ga- and As-rich terminations, within the three-layer model (TLM) scheme. The validity of the widely-used first-order approximation to the TLM is also investigated. The dielectric tensors obtained from SPA monitoring are compared to those obtained previously using reflectance-difference spectroscopy (RDS) to determine the possible surface reconstructions during atomic layer epitaxy of GaAs. These SPA-derived dielectric functions and the TLM are used to optimize experimental conditions for monitoring GaAs surfaces with SPA; this analysis is also extended to the use of SPA for monitoring more general materials systems. The work was supported by NSF grant DMR-94-11504 and the Joint Services Electronics Program Contact No. DAAH04-94-G-0057.
EM-TuP-22 Real-time Control of Polysilicon Etching in a High Density Helicon Source using In Situ UV-visible Spectroscopic Ellipsometry S. Vallon, O. Joubert, L. Vallier, F. Ferrieu (France Telecom-CNET); B. Dr\aa e\villon (LPICM, Ecole Polytechnique, France) A spectroscopic UV-visible ellipsometer from ISA Jobin-Yvon has been installed on a plasma etching chamber. This chamber is also connected with a surface analysis chamber equipped with X-ray photoelectron spectroscopy (XPS). The etching of poly-Si and poly-Si\sub 0.5\Ge\sub 0.5\ gates masked with resist or oxide patterns can be monitored in real time using eight wavelengths recorded simultaneously. A signature of the process is obtained and in particular, the transitions between the layers of the gate stack are detected during the etching process whatever the structure of the mask is. The signatures recorded can be used to design multi-step etching recipes where plasma operating conditions are adapted to the etching of each individual layer of the stack. The ellipsometer can also be used in spectroscopic mode before and after the etching process. Signatures recorded on blanket areas of silicon exposed to the plasma can be used to calculate the thickness perturbated by the plasma. Models developed to fit the signatures obtained by spectroscopic ellipsometry show that the plasma perturbation can be well described by a transparent layer (corresponding to the adsorption of active species) on top of an amorphised layer. The thickness of the transparent layer can also be determined independently by XPS analysis. In addition, by comparing the signatures obtained before and after the etching process, the gate oxide thickness consumption will be determined.
EM-TuP-23 Laser-induced Si(111) Surface Reaction with Cl\sub 2\ and its Characterization using Femtosecond Second-harmonic Generation S. Haraichi, F. Sasaki (Electrotechnical Laboratory, Japan) Laser-induced surface reaction for the Si(111)/Cl\sub 2\ system has been observed using femtosecond second-harmonic generation (SHG). In the dry etching process, the relaxation time of electric excitation and vibration motion of atoms which are probably the origin of surface reaction is lower than the nanosecond order. However an accurate time-resolved analysis of below the nanosecond order has not been undertaken as yet. Surface SHG has been applied to various observations of surface phenomena because of its high surface sensitivity, in-situ observation, and ultrafast response. For the experiment of pump-probe measurements, Ti-sapphire laser at 1300nm and 800nm with a 100fs pulse width were used as the fundamental lights. With the wavelength at 1300nm, a contribution of absorption is small and the surface specific information can be obtained. On the other hand, the 800nm light excites free carriers in Si and induces surface reaction effectively. SHG intensities with exposure to Cl\sub 2\ decay rapidly under the 1300nm probe wavelength, and increase rapidly under the 800nm probe wavelength. When some molecules chemisorb on a Si surface, SHG intensity usually decays because of the decrease of the Si surface dangling bonds. In our experiment the 800nm probe beam excites free carriers in Si substrate, and the photoexcited carriers attribute to the enhancement of the SHG from the Cl\sub 2\-exposed Si surface. With both the Cl\sub 2\ flux and the pump beam irradiation, SHG intensities depend on the time delay between the pump and the probe beams. Only when the pump beam coincides with the probe beam, the photoinduced Cl desorption and/or the Cl etching of Si occur. Surface reaction is also affected by the combination of pump-probe polarization. An interference between the pump and the probe beams excites a carrier grating in Si and induces surface reaction.
EM-TuP-24 Accurate Determination of Singularities in Optical Spectra by Reciprocal-Space Analysis S. Yoo, D. Aspnes (North Carolina State University) We discuss the determination by reciprocal-space analysis of the energies of singularities, e.g., critical point thresholds in reflectance or ellipsometric spectra and exciton and impurity energies in photoluminescence or photoluminescence excitation spectra. Here, singularities are defined by phase coherency of the Fourier coefficients rather than the shape of real-space lineshapes that may be poorly understood. Using a combination of model calculations, first-order approximations, and direct analysis of data, we demonstrate in addition to the usual filtering advantages: (1) singularities can be determined accurately without a detailed knowledge of real-space lineshapes; (2) PL and PLE energies can be determined essentially independently of baseline effects; (3) sensitivity to systematic errors that result from inadvertent overlayers is reduced; and (4) an optimal monochromator resolution (slit width) can be defined and data corrected for finite-monochromator resolution. We also (5) derive an approximate analytic expression for describing the effect of overlayers on singularities; and (6) develop an approach for isolating spectral regions for detailed analysis. These results optimize analytic capabilities for demanding applications, such as the determination of compositions from real-time analysis of spectral data.

Time Period TuP Sessions | Topic EM Sessions | Time Periods | Topics | AVS1996 Schedule