1997 MRS Fall Meeting & Exhibit

Chairs

Mohammad Alim,
Rosario Gerhardt, Georgia Inst of Technology
S. Ray Taylor, Univ of Virginia 

Symposium Support 

• Keithley Instruments
• Solartron

1997 Fall Exhibitor

Proceedings published as Volume 500 
of the Materials Research Society 
Symposium Proceedings Series.

* Invited paper

SESSION EE1: ADVANCES IN EXPERIMENTAL TECHNIQUES 
Chair: S. Ray Taylor 
Monday Morning, December 1, 1997 
Republic A (S)

8:30 AM *EE1.1 
UNIVERSALITY OF DIELECTRIC RESPONSE AS AN AID TO DIAGNOSTICS. Andrew K. Jonscher, Royal Holloway, University of London, Surry, UNITED KINGDOM.

The advent of frequency response analysers and the, formulation of the universality of dielectric responses have between them greatly enhanced the usefulness of dielectric measurements as a diagnostic tool. The available frequency range was extended to between eight and twelve decades, making it possible to determine spectra with some precision. In addition, we now have a much more complete understanding than was available previously of the significance of various spectral shapes. For charge carrier systems the universal approach is based on the fractional power law of frequency dependence of the real and imaginary components of the susceptibility ()=()-i()(i)n-1, where the exponent falls in the range 0 9:00 AM *EE1.2 
THE USE OF COMPUTER SIMULATIONS TO INTERPRET AND UNDERSTAND ELECTRICAL MEASUREMENTS. Edward J. Garboczi, NIST, Gaithersburg, MD.

Two of the principal difficulties in interpreting and understanding microstructure based on electrical measurements lies in the shape and topology of the material composing the microstructure. Shapes that are found can usually not be solved analytically, and second phases that are topologically connected also usually rule out analytical treatments. The use of computer simulations, both to generate material shape and topology and numerically solve the electrical equations, is discussed. The length scales considered range from the individual particle level through the microstructural level to the structural (sample shape and size) level.

10:00 AM *EE1.3 
TIME DOMAIN RESPONSE OF ELECTRICAL CERAMICS - MICROSECONDS TO MEGASECONDS. F. A. Modine, Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN.

The electrical properties of ceramics can be measured in either the time domain or in the frequency domain, and, in principle, the same information is obtained. Frequency-domain measurements are most often employed. But time-domain measurements provide somewhat different insights that may be more appropriate, particularly for electrically nonlinear ceramics such as varistors. The interpretation of electrical measurements in the time domain is profoundly influenced by such practical matters as source impedance and waveform characteristics. The measurements are easier to interpret when the source impedance is either high or low relative to the sample being tested. But such conditions may be difficult to achieve in electrically nonlinear materials because the sample impedance can change dramatically. At short times, parasitic reactance is a problem, and at long times stability becomes a concern. Time-domain electrical measurements on grain-boundary-barrier devices such as varistors are valuable because of the many insights they provide into both device performance and the underlying device physics. Barrier heights and widths are not only disclosed but are found to change with time and be influenced by such variables as voltage and temperature. Capacitive and inductive responses are seen and long- and short-term memory effects are both present. The time-domain electrical response of varistors reveals not only the grain-boundary barriers that contribute electrical nonlinearity but also the trapped electrons and holes that are the origin of the barriers. To a large extent the electrical time response of varistors is a reflection of the time response of the electrons and holes that both control over-barrier conduction and contribute directly to the electrical transport. The electron and hole trapping are distinctly non-Debye processes.

10:30 AM *EE1.4 
LOW-FREQUENCY SCANNING CAPACITANCE MICROSCOPY. Stefan Lanyi, Inst of Physics, Slovakian Academy of Sciences, Bratislava, SLOVAKIA; Miloslav Hruskovic, Faculty of Electrical Engineering and Information Technology, Slovak Univ of Technology, Bratislava, SLOVAKIA; Gian Bartolo Picotto, CNR-Istituto di Metrologia, Gustavo Colonnetti, Torino, ITALY.

Scanning Capacitance Microscopy is one of the less known of the many scanned probe microscopies. It is partly due to the lack of suitable equipment on the market in the past,and because the usual setups work at a fixed high frequency and could image only the topology of surfaces, conducting or covered by an insulating film. Usually most of it can be done easier and better with Scanning Tunneling Microscopy or Scanning Force Microscopy. The low-frequency SCM working in the MHz range is able to image the components of the complex capacitance of the probe/sample capacitor, i.e. also dielectric losses. The stray capacitance of the probe could be reduced to hundreds of aF. At present a sensitive I/U converter with a virtual input capacitance of about 8 fF was built to make a combined SCM/STM operation, using the same probe, possible. To extend the working frequency range to lower frequencies an extremely high input impedance is needed. This could be achieved with an electrometric input stage having an input resistance of 10 G and capacitance of about 1 fF. The useful frequency range is from 1 kHz to a few MHz. This makes a combination of microscopy with local impedance spectroscopy possible. The impedance spectroscopy, though in a limited frequency range, is relatively slow and requires an excellent stability and elimination of the creep of the piezoelements by independent position monitoring and feedback.

11:00 AM EE1.5 
RESISTOMETRIC MAPPING USING A SCANNING TUNNELING MICROSCOPE. C.I. Lang, Department of Materials Engineering, University of Cape Town; J. Tapson, Department of Electrical Engineering, University of Cape Town, SOUTH AFRICA.

We present a method whereby spatial variations in the electrical resistance of bulk conductive specimens may be detected on the same scale as the microstructural variations from which they arise. This technique, a new development of scanning tunneling potentiometry, offers significant benefits for microstructural characterization and for investigation of microstructure/resistivity relationships in metals. The technique also detects anisotropies in specimen resistance, offering promise for identification of low-resistivity paths for characterization or optimization purposes. Resistometric mapping is performed using a standard scanning tunneling microscope (STM) with the addition of AC bias and demodulation circuitry. Topographic information is extracted simultaneously with potentiometric information in two in-plane dimensions, providing correlated maps of topographic and resistive variations. The resistometric maps do not however reflect surface topography, but show the location and distribution of regions of enhanced conduction electron scattering which arise from subsurface microstructural features. The use of large, high-frequency currents allows detection of resistive variations even in bulk metallic specimens. The method has been successfully applied to a range of materials, including pure metals, multiphase alloys, composites, and materials exhibiting resistive anisotropy. Detected spatial resistive variations have been associated with phase and grain boundaries, and with grain orientation in anisotropic materials.

11:15 AM EE1.6 
QUANTITATIVE MICROSCOPY OF COMPLEX ELECTRICAL IMPEDANCE WITH SUBMICRON RESOLUTION USING A SCANNING TIP MICROWAVE NEAR-FIELD MICROSCOPE. X.-D. Xiang, Tao Wei, Chen Gao, Fred Duewer, Ichro Takeuchi, Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA.

Quantitative microscopy of complex electrical impedance is of great interest of both academic and industrial communities. We have recently developed instrumentation and model theory for quantitative microwave impedance microscopy with resolution down to 100 nm (/106). The development and applications of a scanning tip microwave near-field microscope (STMNM) will be discussed. The design of our STMNM involves a sharpened metal tip mounted on the center conductor of a high Q coaxial resonator and extended beyond an aperture formed in the endwall of the resonator (a sapphire plate with a thin metal film)[1, 2]. This STMNM has been used to study LiNbO3 single crystals with periodic dielectric constant modulations and ferroelectric domain structures[3]. Both linear and nonlinear dielectric constant images and loss tangent images with sub-micron spatial resolution have been obtained. Edge dislocation defect induced strain field distribution has also been observed in LiNbO3 single crystals, which is not observable by optical microscopy due to the large birefringence of the crystal. A low temperature version of the microscope has been used to study the microwave properties of YBa2Cu3Ox patterned thin films and other superconducting materials. Theoretical model analysis has been performed which allows quantitative microwave impedance microscopy of dielectric and metallic materials. Electrically based microstructure studies of several other materials and potential applications will be discussed.

11:30 AM EE1.7 
MICROMACHINED SFM PROBES FOR HIGH-FREQUENCY ELECTRIC AND MAGNETIC FIELDS. Daniel van der Weide, Vivek Agrawal, Pavel Neuzil, Toralf Bork, University of Delaware, Dept of Electrical and Computer Engineering, Newark, DE.

We present recent results from micromachined localized high-frequency electric (coaxial) and magnetic (loop) field probes integrated with scanning probe microscopes. Our approach enables simultaneous acquisition of both field and topography in the radio-frequency through millimeter-wave regime, enabling more complete characterization of both materials and devices.

11:45 AM EE1.8 
LOCAL PROBE INVESTIGATION OF THE ELECTRICAL PROPERTIES OF SEMICONDUCTING LAYERED COMPOUNDS THIN FILMS. C. Ballif, M. Regula, F. Levy, Institute of Applied Physics, Ecole Polytechnique Federale de Lausanne, Lausanne, SWITZERLAND; Ph. Niedermann, Centre Suisse d'Electronique et de Microtechnique SA, Neuchatel, SWITZERLAND; W. Gutmannsbauer, R. Bucher, Paul Bucher Company, Basel, SWITZERLAND.

An atomic force microscope (AFM) with a conductive tip, is used to investigate locally the electrical properties of WS2 thin films. These local measurements provide information on the conduction within and between the grains and on the surface electronic properties. The studied films consist of large, flat WS2 crystallites (1-5m), with the occurrence of some CoS phases due to the fabrication process [1]. The WS2 films are p-type semiconductors, and exhibit optical properties, photoconductivity and electrical conductivity typical of WS2 single crystals. For the AFM experiments, boron doped diamond has been deposited by chemical vapor deposition onto silicon tips [2]. The diamond coated tips are conductive and suitable for current spectroscopy. Due to the high hardness and chemical inertness of diamond, the tips are resistant to wear and remain conductive with a good spatial resolution even after intense contact AFM imaging. A Topometrix Explorer AFM, with integrated software facilities for electrical measurements, enables simultaneous acquisition of topography and tip-sample current, as well as local current-voltage measurements. With a 100mV voltage applied to the tip (sample contacted by a planar gold electrode), the current images show that no current flows between the WS2 crystallites and the tip. A current in the nA range flows between some crystallites edges and the tip, indicating a high level of impurities or metallic phases at the edges. On the CoS grains, a higher current (20-100nA) flows and linear I-V characteristics are obtained. On the WS2 crystallites, the I-V curves are typical of metal/p-type semiconductor Schottky contacts. Our results indicate that the CoS phases and crystallites edge build non injecting contact to the film, which prevents the polycrystalline thin film to be used in a macroscopic junction. However, the density of metallic phases is not high enough to ensure an in-plane percolative path for the current. Therefore, the electrical conduction and photoconduction measured on macroscopic distances, are mainly related to the WS2 crystallites properties.

SESSION EE2: SEMICONDUCTORS AND MICROELECTRONIC APPLICATIONS 
Chair: Daniel W. van der Weide 
Monday Afternoon, December 1, 1997 
Republic A (S)

1:30 PM EE2.1 
A NOVEL APPROACH TO SEMICONDUCTOR ELECTRICAL PROPERTIES - THE ADVANCED METHOD OF TRANSIENT MICROWAVE PHOTOCONDUCTIVITY (AMTMP). Serguei Grabtchak and Michael Cocivera, Guelph-Waterloo Centre for Graduate Work in Chemistry, University of Guelph, Guelph, Ontario, CANADA.

Microwave measurements of material properties by inserting a sample into a microwave cavity and measuring changes in the cavity parameters have been known since the late 40's. The so-called perturbation theory was developed by Slater and relates changes in cavity parameters directly to and . This type of steady-state measurement is widely used to obtain the values of the complex dielectric constant components. Another well known variation of microwave measurements is known as transient microwave photoconductivity (TMP) and is based on detection of a signal which is proportional to the light induced changes in the sample's conductivity. The origin of the TMP method is of the same perturbative nature as mentioned above, but has been focused mainly on the interpretation of kinetics and development of models to interpret a single measured decay. As a result, the traditional TMP method excludes the harmonic oscillator model, which although providing a qualitative description, has to date, been limited quantitatively to simple processes. In this work we have put all three components (cavity perturbation theory, microwave photoconductivity and harmonic oscillator model) together, have modified them, and have ended up with the advanced method of transient microwave photoconductivity (AMTMP) [1-3]. in this method two kinetic decays are registered: one, proportional to the conductivity changes, and the other, arising from the photodielectric effect. This approach facilitates the interpretation, since we can relate the observed decays to and by considering the light induced changes as a second perturbation. This enables us to distinguish the relative contributions of free and trapped electrons. Thus, the harmonic oscillator model provides not only qualitative, but also quantitative information on the charge carriers. Overall, the AMTMP preserves all the advantages of microwave photoconductivity, but significantly extends its capabilities. This new method was applied to various materials: thin film polycrystalline CdSe, single crystal SI GaAs, single crystal Si, and porous Si.

1:45 PM EE2.2 
MICROSTRUCTURAL AND ELECTRICAL CHARACTERIZATION OF MISFIT DISLOCATIONS AT THE InAs/GaP HETEROINTERFACE. V. Gopal1, T.P.Chin2, A.L. Vasiliev1, J.M. Woodall2 and E.P. Kvam1.School of Materials Engineering, Purdue University, W. Lafayette, IN. School of Electrical and Computer Engineering, Purdue University, W. Lafayette, IN.

InAs is a narrow band gap semiconductor with transport properties that show potential for such applications as IR detectors, low temperature transistors, etc.. However, the lack of suitable substrates for growing epilayers has hampered progress in the development of InAs based devices. Recently GaP has been shown to have promise as a substrate material for InAs and AlInAs epilayers for device applications. In the present study, InAs was grown by Molecular Beam Epitaxy (MBE) on (001) GaP substrates. Though this system has a high lattice mismatch, (11), certain MBE growth conditions result in 80 relaxed InAs layers on GaP with the misfit accomodated predominantly by 90 pure edge dislocations. Varying growth conditions can result in samples consisting almost exclusively of 90 dislocations or those consisting of a mixture of 90 and 60 dislocations. Misfit dislocation microstructures were studied using Transmission Electron Microscopy (TEM). Electrical characterization using lateral conductance and Hall effect measurements were also performed. Preliminary results indicate the possibility of misfit dislocation related conductivity. The possible correlation between misfit dislocation type and electrical properties will be discussed.

2:00 PM EE2.3 
ELECTRICAL MEASUREMENT OF THE BANDGAP OF N+ AND P+ SiGe FORMED BY Ge ION IMPLANTATION. A. Nishiyama and M. Yoshimi, Toshiba Corporation, R&D Center, Advanced Semiconductor Devices Labs, Yokohama, JAPAN; O. Arisumi, Toshiba Corporation, Microelectronics Engineering Lab, Yokohama, JAPAN.

The bandgap of heavily doped n+ and p+ SiGe layers formed by Ge ion implantationÅ@into Si was measured electrically. Recently, new Si device structures using the bandgap engineering have attracted an attention with materials such as SiGe, SiC and SiGeC. We have proposed the SiGe source structure for SOI (Silicon On Insulator) MOSFETs in order to suppress the floating-body effects. We used Ge implantation technique to form the SiGe layers, considering its simplicity and compatibility to the ULSI processes. Therefore, the bandgap of the n+ and p+ SiGe layers formed in the SOI by the Ge implantation was an essential factor for the suppression effect in this structure. In the experiment, LOCOS isolations and gate oxide layers with a thickness of 6nm were formed on SOI wafers with a thickness of 90nm. Poly-Si gate layers with a thickness of 300nm were then deposited and patterned. Ge atoms were implanted at 25keV with dosages of 0.5, 1.0 and 3.0 cm-2 into source/drain regions. The subsequent annealing was performed after As or BF2 ion implantation at 850C for 30 min in order to incorporate Ge into n+ and p+ source/drain regions. The temperature dependence of the base currents of the lateral bipolar transistors has revealed that the bandgap of Ge implanted source region decreased as the Ge dosage increased. In this measurement, source/channel/drain electrodes of the MOSFETs were used as emitter/base/collector. The measured bandgap decreased by 50meV for n+ with 3.0 cm-2 Ge implantation, compared to that of n+-Si. On the other hand, the decrease was 20meV for p+ with the same dosage. It was confirmed that the decrease in the bandgap was the major reason for the suppression of the floating-body effect.

2:15 PM EE2.4 
EVALUATION OF GAP STATES IN HYDROGEN-TERMINATED SILICON SURFACES AND ULTRATHIN SiO2/Si INTERFACES BY USING PHOTOELECTRON YIELD SPECTROSCOPY. Seiichi Miyazaki, Toshinobu Tamura, Tetsuhiro Maruyama, Atsushi Kohno and Masataka Hirose, Dept. of Electrical Engineering, Hiroshima University, Higasi-Hiroshima, JAPAN.

The energy distributions of gap state densities in H-terminated Si(100) and (111 ) surfaces and thermally-grown SiO2/c-Si interfaces have been studied by means of photoelectron yield spectroscopy (PYS). Silicon wafers with boron or phosphorous concentrations of 1.3x1015 to 2.0x1019 cm-3 were precleaned by conventional wet-chemical steps. H-terminated Si(100) and atomically-flat, monohydride-terminated Si(111) surfaces, as confirmed by FT-IR-ATR and UPS measurements, were prepared by 4.5%HF and 40%NH4F treatment, respectively, followed by pure water rinse. The oxidation of c-Si surfaces thus prepared was carried out at 1000C in dry O2. Provided that a cube power low characterizes the energy dependence of photoelectron yield, for the monohydride-terminated surfaces of Si(111) with impurity concentrations below 1017 cm-3, the threshold energy which corresponds to the energy separation between the valence band edge with nearly parabolic density of states and the vacuum level is determined to be 5.12eV, which agrees well with the reported value for cleaved Si(111) surfaces[1]. In addition, PYS spectra below the threshold energy indicate that there exist occupied surface states with a density of 2x1011 cm-2eV-1 below midgap. The origin of the observed surface states remains unclear. For heavily-doped Si surfaces, the valence band tailing is clearly observed with increasing the impurity concentration. The states above midgap are observable for n+ Si surfaces and evaluated to be 5x1010 cm-2eV-1. The interface state distribution in an ultrathin SiO2/Si system, for which the capacitance-voltage characteristics are no longer a well-defined technique because of high tunneling current through the ultrathin oxide layer, has been examined by PYS. It is found that, for an as-grown 2.5nm-thick SiO2/n+ Si, the interface state density near midgap is as high as 1012 cm-2eV-1.

3:00 PM EE2.5 
IN SITU VAN DER PAUW AND HALL ANALYSIS OF THE AMORPHOUS-TO-CRYSTALLINE TRANSFORMATION IN THIN FILM INDIUM TIN OXIDE. Larry Lee, Fidel Amaro, Greg Crawford, Cleva Ow Yang, and David C. Paine, Division of Engineering, Brown University, Providence, RI; Christine Caragianis-Broadbridge, Trinity College, Department of Engineering, Hartford, CT.

Tin-doped indium oxide (ITO) is used as a transparent conductor in FPD applications. Low temperature low energy deposition methods result in the deposition of partially or fully amorphous ITO. The kinetics of the transformation from amorphous to crystalline material have been studied using in situ Hall effect and Van der Pauw measurements and a mechanistic description of the process has been developed. A heated stage with tungsten pressure contact probes was designed and built to allow the transport properties of ITO thin films to be studied as a function of time at temperature. The resulting resistivity-, mobility- and carrier-density- versus time at temperature plots yield a host of microstructural and kinetic insights. For example, the activation energy for the transformation process in air of e-beam deposited ITO was measured to be approximately 1.2 eV. The resistivity of the ITO does not vary monotonically during the transformation. Through the monitoring of the Hall voltage during crystallization the dopant activation (i.e. substitutional Sn) effects can be separated from scattering due to the evolving microstructure. We report on the use of in situ transport measurements which, along with plan view and cross sectional TEM, have been used to develop a model for describing the amorphous to crystalline transformation process as a function of film deposition parameters, annealing temperature and gas ambient.

3:15 PM EE2.6 
VACUUM AND CONTROLLED ATMOSPHERE HALL EFFECT MEASUREMENTS ON WO3 FILMS. W. Dittrich, B. Allen, M.C. Doogue, L.J. LeGore, D.J. Frankel, R.J. Lad, Laboratory for Surface Science and Technology, University of Maine, Orono, ME.

Highly resistive, semiconducting WO3 films are used as sensing elements in conductometric and surface acoustic wave microsensors due to their change in electrical conductivity when exposed to oxidizing or reducing gases. Variations in film deposition techniques and deposition parameters affect the microstructure of these films and therefore have a large influence on the sensor response. An important factor for relating the gas response to microscopic film properties is the knowledge of the carrier concentration and mobility. We have built a Hall effect measurement system, that is an integral part of our ultra-high vacuum thin film synthesis and processing facility, which allows studies of the conductivity, carrier concentration and mobility in vacuum directly after the film deposition, prior to any air exposure, as well as during and after controlled gas exposure up to atmospheric pressure. In vacuum, epitaxial WO3 films with room temperature conductivities in the order of 0.1 -1cm-1 were found to have mobilities in the order of 1 cm(2(2%%(2/Vs. Mobilities of films grown with a polycrystalline microstructure are lower and below our detection limit of 10^-2cm^2/Vs. For epitaxial films, four point van der Pauw conductivity and Hall measurements will be presented as a function of exposure to H_2S.

3:30 PM EE2.7 
EFFECTS OF INTERFACIAL STRESS DISTRIBUTION ON LOCAL ELECTROPHYSICAL PROPERTIES OF METAL-INSULATOR-SEMICONDUCTOR STRUCTURES. Vladimir Popov, Res Inst of Microelectronic Devices, Phys & Techn Res Center, Kiev, UKRAINE.

The influence of gate induced interfacial stress on local electrophysical properties of insulator-semiconductor (IS) systems in metal-insulator semiconductor (MIS) structures has been investigated. The behavior of electrically active defects (EAD) responsible for generation and recombination of minority carriers in silicon MIS structures was studied. These defects are situated at IS interface in silicon and play an important role in IC performance and reliability. New method for investigation of local electrophysical properties of EAD has been developed. This method enables to study internal properties of submicron size EAD using test structures with electrodes which have much higher dimensions than the defects being investigated. The method is based on analysis of pulse-modulated dynamic current-voltage characteristics of MIS structures. The main feature of the method is that special conditions in MIS structures are created owing to which local EAD properties can be strictly distinguished from integral properties of the whole structures. Thus local parameters of MIS structures (threshold voltage, charge in oxide, fast surface state charge and others) in EAD regions and corresponding local degradation phenomena can be investigated under the influence of different technological and external factors. Many of these parameters in EAD significantly differ from their mean values in IS,MIS structures in whole. We have found that evident correlation exists between gate electrode characteristics (material, geometry) and local values of fast surface states and fixed oxide charges in EAD regions of Al-SiO2 - Si, Si poly-SiO2 - Si structures. Electrodes introducing high compressive stresses at Si-SiO2 interface affect on charge density in EAD through the rate of generation of minority carriers remains almost unchanged. The effect of gate-induced stress on behaviour of charges in Si-SiO2 system after ionizing irradiation of MIS structures was also studied. Method for investigation of spatial distribution of radiation induced charges in MIS structures has been created. Method is based on electron beam induced current (EBIC) analysis of MIS gated p-n junctions in scanning electron microscope. The microprofiles of radiation-induced charges in MIS structures after low-energy electron (10-25 keV) irradiation have been measured which correlate with stress distributions caused by gate electrodes at Si-SiO2 interface.

3:45 PM EE2.8 
EFFECT OF Cu ALLOYING PROCESS ON ELECTROMIGRATION OF PLASMA ASSISTED CHEMICAL VAPOR DEPOSITED ALUMINUM. Dong-Chan Kim, Byung-Il Lee and Seung-Ki Joo, Div. of Materials Sci. and Eng., Seoul National University, Shillim-dong, Kwanak-ku, Seoul, KOREA.

Chemical vapor deposition of aluminum has been extensively studied for the metallization of ULSI circuits which is required to fill submicron contact and via holes with a high aspect ratio. In our previous results, plasma assisted CVD process improved the surface morphology and step coverage of aluminum film. We achieved the perfect filling of submicron contact holes with more than 3 aspect ratio and planarized metallization with smooth surface morphology. However, one of the most serious problems in Al CVD studies has been Cu incorporation, because chemical vapor deposited aluminum films have not contained the necessary level of Cu elements essential to improve electromigration reliability. To obtain Copper-alloyed CVD aluminum, we approached a post annealing process for diffusing Cu from sputtered pure Cu or binary compound Cu thin layer deposited on the PACVD Al film. Especially, the compound Cu film has a nearly eutectic composition(Al-33wt%Cu), which has a lowest melting temperature(548.2C) in the Al-Cu equilibrium phase diagram. The efficiency of the compound Cu as a diffusion source was compared with that of pure Cu. Electromigration life time of the PACVD Al doped by the composite Cu diffusion source is several ten times longer than that of the PACVD Al doped by the pure Cu diffusion source. We have characterized the influences of microstructure and distribution of Al-Cu second phases on electromigration performance of the PACVD Al film alloyed by diffusion process.

4:00 PM EE2.9 
EFFECTS OF Sc OR Tb ADDITION ON THE MICROSTRUCTURES AND RESISTIVTIES OF AI THIN FILMS. Shinji Takayama, Hosei Univ., Dept of System and Control Engineering, Tokyo, JAPAN; and Naganori Tsutul, ITES Ltd, Shiga, JAPAN.

High thermal stable microelectronic conductor lines have become high attention for the application of advanced TFT-LCD and VLSI devices. For this purpose, we have currently investigated sputtered Al thin films added rare-earth metals and revealed that some rare-earth elements largely suppressed growth of hillocks at high temperatures associated with a large number of very fine segregated metallic compounds, mostly at grain boundaries. In this report, subsequent works was conducted by the addition of Sc and Tb to Al thin films and investigating the effects of these elements on the change of microstructures and resistivities during annealing. These alloy thin films about 400 nm thick were deposited on a 7059 glass substrate by using a DC magnetron sputtering apparatus. X-ray diffraction measurements showed that the growth of the Al (111 ) plane was largely suppressed by adding only a small amount of the rare-earth-metal-elements to pure Al. These as made alloy films show very fine grain less than 100 nm in size. Their isochronal annealing up to 450C revealed that the resistivities of Al-Sc and Al-Tb alloy films started to decrease largely at 150C and 250C, respectively, while growth of hillocks on the film surfaces takes place at 200 C and 450C for Al-Sc and Al-Tb, respectively. Both X-ray diffraction and TEM observation showed that fine metallic compounds of Al3RE (RE= Sc and Tb) were segregated in Al matrix (mostly at the grain boundaries of Al matrix) for the samples annealed at 350C.

4:15 PM EE2.10 
ELECTRICAL PROPERTIES OF NOVEL ANODIC FILMS FORMED IN NONAQUEOUS ELECTROLYTE SOLUTIONS. Fumikazu Mizutani, Sachie Takeuchi, Tetsuo Nishiwaki, Nariaki Sato and Makoto Ue, Mitsubishi Chemical Corporation, Tsukuba Research Center, Ibaraki, JAPAN.

Recently, valve metals covered with barrier type anodic oxide films are considered attractive candidates for metallizaton in micro device application, for example, thin film transistor liquid crustal display (TFT/LCD). Main advantage of using such anodic oxide films is the suppression of hillock formation in aluminum film, which is promising for its low resistivity. For anodic films, low permittivity and sufficient insulation are desirable. Here, electrical properties of anodic films formed in nonaqueous electrolyte solutions, such as -butyrolactone (GBL), propylene carbonate (PC) and ethylene glycol (EG), were investigated. Before anodization, thin films of valve metals were deposited on glass substrates by sputtering. Then they were anodized in nonaqueous electrolyte solutions. These anodic films were characterized by transmission electron microscope (TEM), auger electron spectroscopy (AES) and other surface analysis method. The films were amorphous and contained components originated from electrolytes. Aluminum counter electrodes were deposited onto the films to make metal-insulator-metal (MIM) elements. Impedance spectroscopic measurements showed that permittivities of the films were smaller than those of the usual films formed in aqueous solutions. Current - voltage characteristics were also measured. The leakage currents of the films formed in nonaqueous electrolyte solutions were lower. The applications of these films in the area of micro devices will be discussed.

4:30 PM EE2.11 
NUCLEATION AND GROWTH AT REACTIVE INTERFACES FOLLOWED BY IMPEDANCE MEASUREMENTS. Frederic Voiron, Michel Ignat, INP Grenoble, LTPCM, St. Martin d'Heres, FRANCE; Tom Marieb, Harry Fujimoto, Components Research, INTEL, Santa Clara, CA.

To evaluate the thickness of different constitutive layers of a Ti/Al multilayer, two different experiments, based on impedance measurements on multilayers, have been performed: … in the first one, the thickness of the layer is deduced from resistivity measurements, applied to a parallel resistance model, … in the second one, the thickness of the layer is deduced from comparing calculated and measured values of surface potentials. These electrical methods have permitted to validate a model describing the nucleation and growth of a TiAl3 phase from Al/Ti interface. The model is based on transport of matter considerations, activated at the interface. It describes the displacement of the plane where stand the reaction. This model represents the TiAl3 formation, and further growth of this phase from the interdiffusion of the constituting elements in the TiAl3.

SESSION EE3: NON-LINEAR CONDUCTION AND DEVICE DEGRADATION 
Chair: Mohammad A. Alim 
Tuesday Morning, December 2, 1997 
Republic A (S)

8:30 AM *EE3.1 
QUANTITATIVE ICTS MEASUREMENT OF INTERFACE STATES AT GRAIN BOUNDARIES IN ZnO VARISTORS. Kazuo Mukae, Akinori Tanaka, Fuji Electric Corp. Research and Development, Ltd., Kanagawa, JAPAN.

Electrical properties of ceramic semiconductors are strongly dependent on the double Schottky barriers(DSB) formed at grain boundaries in the ceramic semiconductors. Since DSB is built by electrons trapped by the interface states at the grain boundary, it is very important to characterize the electronic states to investigate their electrical properties. Isothermal capacitance transient spectroscopy(ICTS) measurement is a intensive method to characterize interface states. Application of ICTS measurement to ZnO:Pr varistor at elevated temperature gave a simple peak corresponding to the interface states at grain boundaries. The energy level of the interface states revealed to be monoenergetic and located around 0.8eV below the conduction band and the cross section area was calculated as around 10-16 cm2. ICTS measurement of PTC thermistor based on ceramic BaTiO3 allowed the first observation of the interface states of the PTC thermistor. This result has indicated the existence of DSB in PTC thermistors. However, most of these ICTS analyses concerned only to emission rate, energy level or cross section area of the interface states and the ICTS intensity has never been analyzed quantitatively. We have established the quantitative treatment of the ICTS intensity in relation to the density of interface states at grain boundaries. From the linear relation between ICTS intensity and reciprocal carrier density(ND), density of interface states was obtained. According to experiments on series of rare-earth doped ZnO varistors, the interface states of Pr, Tb or Nd doped ZnO varistors had higher density of states than La doped varistors. Moreover, the higher density was found to give higher nonlinearity of I-V relation. Application of ICTS measurement to single grain boundary using microelectrodes gave detailed information about the individual grain boundary.

9:00 AM *EE3.2 
Withdrawn.

10:00 AM *EE3.3 
CURRENT LOCALIZATION, NON-UNIFORM HEATING, AND FAILURES OF ZnO VARISTORS. M. Bartkowiak, University of Tennessee, Dept of Physics and Astronomy, Knoxville, TN and Oak Ridge National Laboratory, Solid State Div, Oak Ridge, TN.

Non-uniform heating of ZnO varistors by electrical pulses occurs on three different spatial scales: (1) microscopic (sub-micron), (2) intermediate (sub-millimiter), and (3) macroscopic (of order of millimeters or centimeters). Heating on these scales has different origins and different consequences for device failure in large and small varistors. On the microscopic scale, the heating in thin (e.g., 100 microns thick) varistor slices is observed to be localized in strings of tiny hot spots. The hot spots occur at the grain boundaries in a conducting path where the potential is dropped across Schottky-type barriers and the heat is generated. These observations are interpreted by applying transport theory and using computer simulations. It is shown that the heat transfer on a scale of the grain size is too fast to permit temperature differences that could cause a varistor failure. On an intermediate size scale, the heating in small (e.g., 10 mm diameter and 1 mm thick) varistor disks is most intense along localized electrical paths. The high electrical conductivity of these paths has microstructural origin, i.e., it derives from the statistical fluctuations in properties that inevitably occur in polycrystalline materials (e.g, fluctuations of grain sizes and grain boundary properties). It is found that influence of this statistical disorder changes with varistor size in a manner that can be described with relatively simple mathematics. Current localization on the intermediate size scale appears to be significant only in small varistors. On the macroscopic scale, heating in large (e.g., 42 mm diameter and 42 mm thick) varistor blocks is usually the greatest near the block edges and is approximately radial symmetric in blocks fabricated at a low aspect ratio. In blocks fabricated at a higher aspect ratio, the heating is less symmetric, presumably because uniform properties are more difficult to achieve. Current localization in large blocks can be attributed to inhomogeneities in the electrical properties which originate during ceramic processing (e.g., compaction and sintering). Destructive failures (puncture and cracking) of large varistor blocks are shown to be caused by this non-uniform heating on macroscopic scale.

10:30 AM *EE3.4 
CURRENT FLOW AND STRUCTURAL INHOMOGENEITIES IN NONLINEAR MATERIALS. Felix Greuter, Joachim Glatz-Reichenbach, and Thomas Christen, ABB Corporate Research Ltd., Baden-Daettwil, SWITZERLAND.

The current flow in real 3-dimensional nonlinear materials like varistors or PTC resistors is far more complex than in the simple brickstone model, which is usually used to extract information on the micro-junction properties. Recent 2-dimensional numerical simulation work, e.g., by Vojta and Clarke (J. Appl. Phys. 81, 985 (1997)) and by Bartkowiak et al. (J. Appl. Phys. 79, 8629 (1996)), provides a first basis for a deeper understanding of the electrical path phenomena in these materials. The experimental counterpart, however, is still missing to a great extent. After briefly reviewing the present state of knowledge, we will illustrate the influence of artificial structural inhomogeneities of different lengths scales on the local and the integral electrical behavior of varistors and PTC materials. Admittance properties and thermal images provide complementary information on the true current flow pattern. A clear correlation to the overall distribution of the grain size and the second phases is observed for varistor ceramics, which makes an important link to the material processing routes.

11:00 AM EE3.5 
HIGH POWER SWITCHING IN CONDUCTOR-FILLED POLYMER COMPOSITES. Anil R. Duggal, Lionel M. Levinson, GE Corporate Research and Development, Niskayuna, NY.

It has generally been assumed that the switching properties of conductor-filled polymer composites are based on a positive temperature coefficient of resistance (PTCR) effect where, at a certain switch temperature, the material resistivity increases by orders of magnitude. Here we present studies of the electrical switching behavior at high current densities which demonstrate that, in the high power regime, the observed switching is not based on the PTCR effect. Instead, we show that this type of switching appears to be a general feature in conductor-filled polymer composite materials and a qualitative model for the switching phenomenon is proposed. These results suggest that conductor-filled polymer composite materials can provide a new non-mechanical way of rapidly limiting high power short circuit currents. This should have broad applications in the circuit protection industry.

11:15 AM EE3.6 
TRANSIENT CURRENT GENERATION AS A DYNAMIC PROBE AT MOVING CONDUCTOR/INSULATOR INTERFACES. J. T. Dickinson, L. Scudiero, and S. C. Langford, Washington State University, Dept. of Physics, Pullman, WA.

An important component of friction during rubbing of two surfaces arises in the rapid, transient making and breaking of adhesive bonds between asperities. When conductors are drawn across polymers and inorganic crystalline materials (i.e., insulators), continuous detachment between the two surfaces generates charge separation due to contact electrification. In MEMS devices, such charge transfer can have enormous effects on performance. We have devised sensitive circuits for detecting instantaneous transient currents generated by this process while simultaneously measuring the normal and lateral forces as a conducting stylus is moved across insulating surfaces. The experiments are performed in high vacuum as well as in controlled atmosphere. Using hard conducting tips on softer substrates (polymers such as PMMA, polyethylene, polycarbonate, and polystyrene, and single crystal inorganics such as MgO, Al2O3, and SiO2) we measure these currents vs. obvious parameters (e.g., normal force, stylus velocity) and relate the observed transient currents to the extent of damage to the substrate, the contribution of adhesion to the frictional force, and the physics of contact charging. Present tip radii used are a few µm in dimensions with efforts to extend down to much smaller sizes. Time resolved measurements reflect temporal statistics of make-break interactions - presently, time resolutions of µs have been achieved. The analysis of both the magnitude and fluctuations in the current are informative regarding the complex micromechanics involved.

11:30 AM EE3.7 
CONDUCTANCE TRANSIENTS STUDY OF SLOW TRAPS IN Al/SiNx/Si AND Al/SiNx/InP METAL-INSULATOR-SEMICONDUCTOR STRUCTURES. Salvador Dueñas, Rosa Pelaez, Elena Castan, Juan Barbolla, Universidad de Valladolid, Dept de Electricidad y Electronica, Valladolid, SPAIN; Ignacio Martil, German Gonzalez, Universidad Complutense de Madrid, Dept de Electricidad y Electr onica, Madrid, SPAIN.

In this work conductance transients in the SiNx:H/Si and SiNx:H/InP interfaces will be presented. 550 thick silicon nitride films were directly deposited on both n-type silicon and n-type indium phosphide by low temperature electron-cyclotron-resonance plasma method. Aluminium was finally deposited to fabricate MIS diodes. As in previous works hysteresis phenomena were observed in C-V measurements. Here we present conductance transients produced by the interface states when the sample is carried from deep to weak inversion. We carried out these measurements at several frequencies observing that the shape of these conductance transients varies with the frecuency at which they are obtained. In addition, we made measurements keeping constant the frequency and varying the temperature. We have detected conductance transients at temperatures as low as 200 K. This behavior is explained in terms of a disorder-induced gap-state (DIGS) continuum model. This model was developed to explain the hysteresis phenomena in the C-V curves previously observed. In this model the interface states are distributed both in energy and in space into the insulator, so that emission and capture of free electrons by states located far from the interface can occur by mean of tunneling mechanisms. When capture process takes place, the empty DIGS states capture electrons coming from the conduction band. This process is assisted by tunneling and is time consuming. In this work we prove that the conductance transients observed at low temperature also support the assumption of tunneling assisted capture process. The measurement of these anomalous conductance transients observed in the Si3N4/Si and Si3N4/InP interfaces allows us to obtain the spatial and energetic distribution of their interface states. As it is known, these distributions are exponentially space decaying into the insulator and U-shaped in energy.

11:45 AM EE3.8
COMPUTER SIMULATION OF ZnO VARISTORS FAILURES. M. Bartkowiak, G.D. Mahan, University of Tennessee, Dept of Physics and Astronomy, Knoxville, TN and Oak Ridge National Laboratory, Solid State Div, Oak Ridge, TN; M.G. Comber, Hubbell Power Systems, Centralia, MO; M.A. Alim, The Ohio Brass Co, Wadsworth, OH.

Thermal and mechanical behavior of high power ZnO surge arresters under current pulses of various magnitude and duration is simulated. By solving heat transfer equations for a varistor disk with nonuniform electrical properties, we compute the time dependence of the temperature profile and the distribution of thermal stresses. This simple theoretical model can identify the energy handling limitations of ZnO varistors imposed by three different failure modes: puncture, thermal runaway, and cracking. Each failure mode can be limiting, depending on the disk shape, its electrical uniformity, and the current magnitude. The model conforms to the available failure data, and explains the observation that energy handling improves at high current densities. Cracking and puncture are caused by a localization of the current, which causes local heating leading to nonuniform thermal expansion and thermal stresses. Puncture is most likely in varistor disks with low geometrical aspect ratio and when the current density has intermediate values. Cracking dominates at higher current densities and for disks with high aspect ratio. Puncture and cracking do not occur when the current is small, because the time evolution of the nonuniform heating is slow enough for the temperature distribution to flatten. They are also unlikely at very large currents corresponding to the upturn region of the I-V characteristic, since in this case the current becomes uniformly distributed. For low and very high current densities the most likely failure mode is thermal runaway. The model is also applied to evaluate the influence of the nonuniformity of varistor disks used in surge arresters on their energy handling capability. Puncture is the dominating failure mode for slightly nonuniform disks, but cracking becomes more likely as the degree of nonuniformities increases. It appears to be possible to minimize the chance of a failure of varistor disks at high-current pulses by adjusting their resistivity in the upturn region of the I-V characteristic.

SESSION EE4: COMPOSITES AND MECHANICAL DEGRADATION 
Chair: Edward J. Garboczi 
Tuesday Afternoon, December 2, 1997 
Republic A (S)

1:30 PM EE4.1 
GIANT MAGNETOIMPEDANCE: A RELEVANT APPLICATION OF IMPEDANCE SPECTROSCOPY. K.L. Garcia and R. Valenzuala, Institute for Materials Research, National University of Mexico, Mexico, MEXICO.

Giant magnetoimpedance (GMI) refers to variations in the impedance response of a soft ferromagnetic materials (submitted to a current, i, in the 100 kHz-10 MHz frequency range) when a DC magnetic field, H, is applied. The largest variations in GMI have been observed in wires of low and negative magnetostriction, where i generates a circular magnetic field, h, which interacts with circumferential domain walls, resulting in an additional contribution to the total impedance of the system. The application of H leads to damping of domain walls, and if the DC applied field is high enough, to elimination of domain walls by magnetic saturation of the sample. The impedance response is therefore a strong function of H, decreasing from a high value (for H = 0), down to a value close to that observed in a non-magnetic wire, when it is saturated. This phenomenon is currently used in many applications such as field and current sensors. In this work, we show that GMI can be effectively investigated by means of impedance spectroscopy. We show that it can be represented by a simple equivalent circuit, where circuit elements are clearly associated with physical parameters of the sample such as domain wall permeability, domain wall viscous damping and spin permeability.

1:45 PM EE4.2 
COMPARISON OF TECHNIQUES FOR MICROWAVE CHARACTERIZATION OF PERCOLATING DIELECTRIC-METALLIC MEDIA AND RESOLUTION OF DISCREPANCIES IN MEASURED DATA. Ricky L. Moore, Paul Friederich, Edward Hopkins and Stephen Blalock, Georgia Institute of Technology, Georgia Tech Research Institute, Atlanta, GA.

The measured microwave effective dielectric and magnetic properties of metal-dielectric materials show discrepancies when data from free space, resonant cavity or waveguide measurements are compared. Discrepanices are especially evident for materials where the metallic concentration is near the percolation threshold. This paper will present theory and measured data which resolove these discrepancies by utilizing electrical correlation length as the relevant parameter. Guidelines are presented for proper choice of measurement technique.

2:00 PM EE4.3 
PARTICULATE BASED COMPOSITES EXPLOITING PERCOLATION-RANGE MICROSTRUCTURE. Dilhan M. Kalyon, Shawn Walsh*, Bahadir Karuv, Rahmi Yazici, Herman Suwardie and Jason Garrow, Highly Filled Materials Institute, Stevens Institute of Technology, Hoboken, NJ; *Army Research Laboratory, Aberdeen Proving Ground, MD.

Experimental work has been carried-out to develop particulate based high-gain composites exploiting percolation-range microstructure. A thermoplastic rubber, tri-block copolymer with polystyrene end blocks and poly(ethylene-butylene) mid block, was used as the primary matrix material. Various metal and graphite powders with controlled particle size distributions were used as conductive fillers. Systematic studies were carried-out varying the volume percent and mixing distribution characteristics of the filler particles to optimize the high-gain strain-sensor capabilities of the materials. The mixing distributions of the main ingredients were quantitatively determined by x-ray diffraction and electron probe methods. Four-probe conductivity measurements were employed for evaluating the percolation-range characteristics of the filled formulations and effects of the applied strains on electrical conductivity. The optimum high-gain sensor microstructures developed at percolation-range regions exhibited seven orders of magnitude variation in specific resistivity with up to 30% applied strains. The microstructural distributions involving the spatial distribution of the binder and conductive solid phases and the state of agglomeration of the conductive particles were found to affect the percolation behavior and hence the response of composites produced from these particulate based materials.

2:15 PM EE4.4 
DIELECTRIC DISPERSION MEASUREMENTS ON TWO NEARLY IDEAL CONTINUUM PERCOLATION SYSTEMS. Jungie Wu and D.S. McLachlan, Physics Department, University of the Witwatersrand, Johannesburg, SOUTH AFRICA.

Compressed discs and powders made from Graphite (G) and, hexagonal Boron Nitride (BN) are nearly ideal percolation systems, as the ratio of their conductivities if l0-18 and there is very little scatter of the data points near the critical volume fraction . The powder systems are O.5OG-0.5013N and 0.55G-0.45BN loosely packed powders, undergoing quasi-continuous compression, where the volume fraction of Graphite () goes from about 0.085 to 0.15 and the packing fraction (G+BN) from about 0.17 to 0.30. Results for the dielectric constant and the AC conductivity, as a function of frequency, from 30 hz to 100 Mhz will be presented. Percolation theory predicts that, near the percolation threshold(), ) and ) and x+y=1. This behaviour is observed with x = 0.82, 0.94 & 0.87 and y = 0.147 0.07 & 0.10 for the disc samples, 0.50G-0.5OBN and 0.55G-0.45BN powders respectively. All three sets of results with , with close to , are shown to lie on a single universal curve )/,0) against , where ,0) is the measured value and is not the calculated value but is chosen so the particular set of results ( value) lie on the universal curve. The results for can also be made to lie on a universal curve but here neither ,0) nor ,0) corresponds to the calculated value. A new analytical expression for the scaling function will be examined and compared with these results.

2:30 PM EE4.5 
DIELECTRIC SPECTROSCOPY OF INSULATING MATRIX COMPOSITES. Julie R. Kokan and Rosario A. Gerhardt, School of Materials Science and Engineering, The Georgia Institute of Technology, Atlanta, GA.

The dielectric properties of composites are affected in different ways by a number of parameters. These include the electrical properties of both the filler and matrix, the wetting properties of the matrix on the filler, the size and shape of the filler, and the amount of the filler. A number of models have been proposed to mathematically and physically describe these effects. Most composite systems have been studied via dc resistivity measurements which clearly show the effect of the addition of a second phase to an insulating matrix. However, frequency dependent measurements can provide additional insight into the mechanisms of various composites containing conducting and insulating fillers will be given. Comparison to effective medium modelling and percolation theory will be included.

3:15 PM EE4.6 
COMPUTER SIMULATION OF IMPEDENCE FOR 2-D CONDUCTOR-INSULATOR COMPOSITE. Dae Gon Han, Gyeong Man Choi, Dept of Materials Science and Engineering, Pohang University of Science and Technology, Pohang, SOUTH KOREA.

A computational methodology is presented to simulate a.c. transport properties of two-dimensional composite using the personal computer. Using this method, a.c. impedance spectra of macroscopic mixtures of hard spheres, which have the random or regular arrangement of the components, are studied in order to investigate the effects of geometrical arrangement. This numerical method is sufficiently general to simulate the effects of electrode, or grain boundary on the electrical properties as well as the geometrical arrangement effects. From the simulation results, a.c. impedance spectra of two-component composite may show three arcs when two arcs are expected. The simulations on various shaped clusters show that the third semicircle in impedance spectra plots, which may be analyzed as a new electrical component of the equivalent circuit, originates from the isolated clusters formed in the composite. When the elongated clusters are arranged along the preferred current line, the clearer third semicircle appears. It is emphasized that the effects of geometrical arrangement must be seriously considered in the analysis of the electrical properties of components, especially when the system exhibits percolation phenomena.

3:30 PM EE4.7 
ELECTROMECHANICAL STUDY OF CARBON FIBER COMPOSITES. xiaojun Wang, Xuli Fu and D.D.L. Chung, Composite Materials Research Laboratory, State University of New York at Buffalo, Buffalo, NY.

Electromechanical testing involving simultaneous electrical and mechanical measurements under load was found to be valuable for studying the fiber-matrix interface, the fiber residual compressive stress in the fiber direction and the degree of marcelling (fiber waviness) in carbon fiber composites. The fiber-matrix interface study involves measuring the fiber-matrix contact electrical resistivity during single fiber pull-out testing. It provides information on the bond strength and the interfacial structure and allows even a small change in bond strength due to a change in interfacial structure to be observed. The fiber residual stress study involves measuring the electrical resistance of a single fiber embedded in the matrix while the fiber is subjected to tension through its two exposed ends. The tensile stress needed to decrease the resistance to a minimum value is taken to be the residual stress. Tlle fiber marcelling study involves measuring the electrical resistance of a continuous fiber laminate in the through-thickness direction while the laminate is subjected to tension (within the elastic regime) in the fiber direction. The greater is the fractional increase in the through-thickness resistance per unit strain in the fiber direction, the higher is the degree of marcelling.

3:45 PM EE4.8 
DISLOCATION FORMATION IN TENSILE DEFORMATION OF SINGLE CRYSTAL NIAL STUDIED BY ELECTRICAL RESISTIVITY MEASUREMENT. Y.Q. Sun, P.M. Hazzledine and D.M. Dimiduk, Wright Laboratory Materials Directorate, WL/MLLM, Dayton, OH.

To investigate the brittle-to-ductile transition of the intermetallic compound NiAI in the [001] orientation, tensile tests have been performed simultaneously with measurements of the changes in the specimens' electrical resistance with stress and strain. The electrical resistivity is used to probe dynamically the evolution of dislocations in specimens under load. The resistance change is plotted together with the stress-strain curves. The experiments show that the electrical resistance increases linearly with strain, but at remarkably different rates in the elastic and plastic regions of deformation, with a sharp transition between the two at the yield point. The resistivity in the plastic region is shown to depend sensitively on the stress, but is approximately unchanged with the plastic strain The abrupt charge in the resistivity at the onset of plastic deformation shows that during plastic deformation dislocations form abruptly at the yield point and that the dislocation density is a sensitive function of the applied stress. The experimental results are consistent with the collective nucleation of dislocations in specimens loaded close to failure.

4:00 PM EE4.9 
CHARACTERIZATION OF GEOMATERIALS WITH ELECTRICAL MEASUREMENTS. J. Carlos Santamarina and Katherine Klein, Georgia Institute of Technology, Atlanta, GA.

Geotechnical engineering deals with rocks and soil masses. Soil-water mixtures consist of a highly continuous fluid phase and a large volume fraction of low conductivity soil particles. Electrical measurements at different frequencies can be utilized to characterize geomaterials and to monitor processes taking place within the medium. Broad band spectral measurements capture various polarization mechanisms. As frequency decreases, the effects of electronic, ionic, molecular, interfacial, and double layer polarizations gradually accumulate. Different measurement techniques must be used to obtain electrical measurements over a broad frequency range. Results are presented for four-electrode, two-electrode, and coaxial probe systems. Various geotechnical processes have been monitored in soils using electromagnetic waves, including contaminant diffusion, hardening of soil-cement slurries (cementation), consolidation, and corrosion of buried metals. Electrical measurements have also been used to assess ore-bearing rocks. It is shown that the volume fraction and connectivity of the sulphide phase determine the bulk conductivity of the specimen. These results lead to important field applications to characterize the near surface with ground penetrating radar, time domain reflectometry, resistivity measurements, and tomographic imaging.

4:15 PM EE4.10 
APPLICATION OF BROAD-BAND DIELECTRIC SPECTROSCOPY FOR INVESTIGATIONS OF LIQUID CRYSTAL - POROUS MEDIA MICROCOMPOSITES. G.P. Sinha, and F.M. Aliev, Dept. of Physics and Materials Research Center, University of Puerto Rico, San Juan, PR.

We applied ultra broad-band dielectric spectroscopy in the frequency range from 10-4 Hz to 109 Hz to investigate the effect of size, shape and volume fraction of the pores in the porous matrices on the dielectric properties of liquid crystals (LC) dispersed in these matrices. Measurements in such a broad frequency range make it possible to obtain detailed information on the important aspects of the electrical behavior of heterogeneous materials such as: conductivity, surface polarization, and influence of confinement on dynamics of molecular motion of polar molecules forming LC. We investigated alkylcyanobiphenyls in the isotropic, nematic and smectic phases dispersed in porous glasses (average pore sizes - 100 Å and 1000 Å) which have randomly oriented, interconnected pores, and anopore membranes (pore diameters - 200 Å and 2000 Å) with parallel cylindrical pores. Dispersion of LC resulted in qualitative changes of their dielectric properties. Analysis of broad-band dielectric spectra shows that in organic (LC) - inorganic (porous matrix) heterogeneous composites conductivity plays an important role at f<1 Hz. We observe the appearance of new dielectric modes: a very slow process with characteristic frequency (1 - 10) Hz and a second process in frequency range about (103 - 106) Hz. The slow process arises due to the relaxation of interfacial polarization at pore wall - LC interface. The origin of this could be due to absorption of ions at the interface. Another possibility is the preferential orientation of the permanent dipoles at pore surface. The second new mode is due to the hindered rotation of the molecules near the interface. Additionally we observed two bulk like modes due to the rotation of the molecules around their short and long axii which are modified. Analysis of the observed temperature dependencies of relaxation times is discussed.

4:30 PM EE4.11 
MEASUREMENT OF STRATIFIED DISTRIBUTIONS OF DIELECTRIC PROPERTIES AND DEPENDENT PHYSICAL VARIABLES. Alexander V. Mamishev, Yanqing Du, Bernard C. Lesieutre, Markus Zahn, Massachusetts Institute of Technology, Cambridge, MA.

Recent advances in the -k (frequency-wavenumber) interdigital dielectrometry are described. Using this technology, the information about microstructure of the dielectric material is obtained by applying to the sensor-dielectric interface a spatially varying electric potential swept in frequency from 0.005 Hz to 10,000 Hz. The penetration depth is proportional to the spatial wavelength of the electric potential. Application of multi-wavelength electrode arrangements allows measurement of stratified distributions of complex dielectric permittivity. Calibration techniques relate the distributed dielectric properties of materials to other physical variables, such as density, porosity, cracking, lamination, and diffusion of contaminants into the material. The output of interdigital sensors is strongly influenced by the microgranularity of the material's surface. Although this dependence complicates interpretation of the measurements in some applications, the variation of the output may also be used to characterize the shape of the surface on the microscale. Theoretical simulation is related to measurements performed on several dielectrics used in the electric power industry.

4:45 PM EE4.12 
STUDY OF THE MOLECULAR MOBILITY OF POLYSACCHARIDE SOLID THIN LAYERS BY DIELECTRIC RELAXATION SPECTROSCOPY. K. Liedermann and L'. Lapcik, Jr., Technical University of Brno, Brno, Czech Republic.

The application of polymers as coatings of the affinity sensors necessitates to cover the surface of the measuring device with the membrane having both controllable pore size distribution as well as the specific affinity to various diffusants. Suitable materials, which find increasing application for these purposes seems to be polysaccharides, particularly hyaluronic acid. From the physicochemical point of view it is desirable to know, understand and quantify the molecular mobility of the polymer main chain, its side-chain groups and its different skeletal segments. Molecular mobility properties of polysaccharides may be studied by means of the Dielectric Relaxation Spectroscopy (DRS), as both the main chain and the side chain groups are of a polar character. Four types of polysaccharides were studied: hyaluronic acid (HA), chondroitin sulfate (CHS), hydroxyethylcellulose (HEC) and carboxymethylcellulose (CMC), in the frequency range 10-4 Hz - 30 MHz. Observed dielectric spectra were interpreted as sum of one A.C. conductivity process (probably due to the presence of Na+ ions) and of up to two relaxation processes. The relaxation processes were described by means of the Havriliak-Negami formula and their origin was attributed to the migration (interfacial) polarization due to internal interfaces in the sample. The relaxation parameters were related to the molecular structure of the polymers. The low value of in CHS is related to its strong coupling due to the presence of two polar groups in its monomeric unit, whereas low values of x are interpreted as being due to the strong stearic hindrances caused by long pendants present in HEC.

SESSION EE5: POSTER SESSION: 
MICROELECTRONIC APPLICATIONS 
Chair: Rosario A. Gerhardt 
Tuesday Evening, December 2, 1997 
8:00 P.M. 
Grand Ballroom (S)

EE5.1 
INVESTIGATION OF THE DOPANT DISTRIBUTION IN THIN EPITAXIAL SILICON LAYERS BY MEANS OF SPREADING RESISTANCE PROBE, CAPACITANCE-VOLATGE MEASUREMENTS AND SECONDARY ION MASS SPECTROMETRY. Ilya Karpov, Robert Garza, Greg Moran, Subraminia Krishnakumar, Mitsubishi Silicon America, Salem, OR; Jack Linn, Ron Choma Harris Semiconductor, Palm Bay, FL; Catherine Hartford, Gilbert Gruber, Solid State Measurements, Pittsburgh, PA.

Thin epitaxial silicon layers with tight doping profile tolerances are required for bipolar, CMOS, and BiCMOS applications in microelectronics. In this paper, we examine the dopant distributions in 1 to 10 micron-thick boron- and phosphorus-doped epitaxial silicon layers. These layers were grown by chemical vapor deposition (CVD) on arsenic-, antimony- or boron-doped (100)- and (111)-oriented substrates. We performed doping profile studies by means of local resistivity measurements using a spreading resistance probe (SRP) and by capacitance-voltage (CV) measurements. Chemical profiles of the dopants were also obtained using secondary ion mass spectrometry (SIMS). We compare SRP, CV, and SIMS profiles and discuss the correlations between the three techniques. We also discuss potential limitations for each technique and show examples where the techniques provide complementary information. Our results indicate that SIMS profiling can follow very sharp transitions at the layer/substrate interface, while SRP profiles are very sensitive to concentration variations in high resistivity epitaxial layers. We found that probe tuning, the implementation of a multilayer correction procedure, as well as a forward modeling algorithm significantly enhance the depth resolution of the SRP technique.

EE5.2 
ANODIC OXIDATION OF NITROGEN ADDED AL-BASED ALLOY FOR THIN-FILM TRANSISTORS. Toshiaki Arai and Hideo Iiyori, IBM Yamato Laboratory, Yamato-shi, Kanagawa, JAPAN.

Bottom-gate thin-film transistors (TFTs) have been widely investigated for use in active-matrix liquid crystal displays (AMLCDs). In recent years, much effort has been devoted to developing low-resistivity gate bus lines, to meet the need for large, high-resolution LCDs. Aluminum (Al) and its alloys are remarkably good materials for gate bus lines, because of their low resistivity. However, under thermal processes in AMLCDs' fabrication, aluminum is susceptible to stress migration phenomena such as hillocks and whiskers, which cause defects as a result of short-circuits and current leakage between the gate and the upper electrodes. The anodized film is therefore used as a protective layer against stress migration. Since it also functions as a gate insulator, its electrical properties are very important. The surface flatness of the film is similarly important, because the next layer takes on the shape of the underlying surface, and a smooth surface improves the mobility of TFTs. In our previous paper, gadolinium (Gd), and neodymium (Nd) were employed as respective alloy components to increase the thermal resistance of the aluminum, and the electrical properties, surface roughness, and nanostructure of the anodized films were investigated. Our results showed that the anodic oxidized Al-alloy films had relatively large leakage currents. In this study, nitrogen was added to Al-Nd and Al-Gd alloys, and the surface of these alloys were anodized. The nanostructure of anodized films was studied by combined atomic force microscopy (AFM) and cross-sectional transmission electron microscopy (TEM). The electrical properties and surface roughness of these films were discussed in relation to the adding nitrogen concentration. Our result shows that the films with nitrogen provide lower leakage currents and higher breakdown electric fields than conventional films.

EE5.3 
THE INFLUENCE OF IONIC ACTIVITY ON THE ELECTRICAL PROPERTIES OF PECVD (TEOS) SILICON DIOXIDE. Alvaro Romanelli Cardoso, Maria L.Pereira da Silva, LSI/PEE/EPUSP, University of Sao Paulo, Sao Paulo, BRAZIL; Jorge J. Santiago-Aviles, Dept.of Electrical Engineering, University of Pennsylvania, Philadelphia, PA.

A PECVD system was modified to receive an equipotential screen between plasma and the silicon wafer. This way, by modulating the screen potential, it was possible either to modify the ion speed or to remove the ion from the bombardment process. The films produced using this technique were analyzed by the use of I-V and C-V characteristics, supplemented by the use of Raman spectroscopy. The results comfirm the importance Oxygen ions play in the TEOS oxidation chemistry. It is of importance to consider not only the reactions occuring at the surface but the ones occuring in the plasma as well. In the solid phase, TEOS oxidation causes the removal of contaminant carbon compounds from the thin film. The electrical parameters measured showed a slight tendency to improve as a function of an increasing bombardment by positive ions (increasing negative screen voltage). Electrical analysis shows the wide influence specific ion bombardment has on the dielectric response of the resulting films. For example it was found that ion bombardment decreases the permitivity at the same time increasing the silicon dioxide film hysteresis. A qualitative model was developed that suggest an explanation of the observed results.

EE5.4 
ANALYSIS OF TEOS SILICON DIOXIDE: THE IDENTIFICATION OF CARBONATIOUS CONTAMINANTS. Maria L. Pereira da Silva, Alvaro Romanelli Cardoso, LSI/PEE/EPUSP, University of Sao Paulo, Sao Paulo, BRAZIL, Jorge J. Santiago-Aviles, Dept of Electrical Engineering, University of Pennsylvania, Philadelphia, PA.

This work presents the analysis on a silicon dioxide film deposited using organometallic precursors (TEOS). The analysis was done in an attempt to correlate the physico-chemical characteristics with the electrical ones. The main measuring techniques utilized were C-V and I-V characteristics supplemented with SEM to evaluate film homogeneity, SIMS and Micoprobe analysis for total carbon content determination, Fourier Transformed-Mass Spectroscopy (using laser desorption) for surface characterization, and Gas Chromatography coupled to Mass Spectroscopy for elucidation of electrochemical reaction taking place during the I-V characterization. Raman Spectroscopy was found to be a versatile tool in the elucidation and evaluation of the carbon compounds present in the film and their spatial distribution. This technique presents high spatial resolution and a detection level superior to that of IR spectroscopy. The result of the electrical characterization points to poor quality film, suggesting an incomplete decomposition of the organometallic precursor as evidenced by the multiple carbonatious compounds found through the film. The expererimental work indicates the process modifications needed for the desired film quality.

EE5.5
POISSON STATISTICAL MODELING OF WEAROUT IN A 53 ÅRPECVD ULTRATHIN SIO2 FILM. C. Silvestre*, J. R. Hauser, Center for Advanced Electronic Materials Processing, North Carolina State University, Raleigh, NC.

We have examined the breakdown characteristics of a 53 remote plasma enhanced chemical vapor deposited thin SiO2 film. Under 0.1 A/cm2 constant current stressing, the 53 SiO2 film exhibits a much smaller voltage shift V (0.70 Volt) at breakdown due to wearout than SiO2 films with thickness between 96 and 145 (> 3.0 Volt). This is due to the 53 oxide film having a high tunneling current component. Since tunneling current does not dissipate energy to create the neutral trap states which lead to breakdown, the 53 film has a much larger charge-to-breakdown than the thicker oxides. As with the thicker oxides, the 53 breakdown data may be fitted with a Poisson distribution. This fit implies that the breakdown occurs in a localized volume containing one trap. For simplicity we assume that the trap volume is roughly symmetric and spans the oxide thickness. We deduce from first principles that (t)bd= constant, where bd is the width of breakdown area and (t) is the charged trap density. Fitting the 53 oxide breakdown data with a (t) calculated from the measured V results in an unphysically small bd (0.13 ). Imposing a physically reasonable bd of 53 requires (t) to be reduced by a factor 1000. The V across the oxide must then be attributed in its entirety to the accumulation of interface trapped charge with =8.8 1012 cm-2 at breakdown. Due to the nature of the oxide, CV measurements do not reliably measure this interface trapped charge as the constant current stressing proceeds. Assuming the value of was proportional to electron fluence for each oxide thickness examined, we calculated the voltage across the oxide due to , corrected (t), and refitted the breakdown data to determine new values for bd and bd, where bd+1 is number of traps in the oxide breakdown volume at breakdown. The results show unchanged values of bd and increased values of bd.

SESSION EE6: POSTER SESSION: 
CORROSION OF METALS AND POLYMER/METAL INTERFACE 
Chair: S. Ray Taylor 
Tuesday Evening, December 2, 1997 
8:00 P.M. 
Grand Ballroom (S)

EE6.1 
UNDERSTANDING COATING AND SUBSTRATE HETEROGENEITIES USING LOCAL ELECTROCHEMICAL IMPEDANCE METHODS. S.R. Taylor, A.M. Mierisch, R.B. Leggat, Center for Electrochemical Science and Engineering, University of Virginia, Charlottesville, VA.

Present limits in the understanding of corrosion performance of coatings on aluminum are linked to the observation that: (1) coatings fail at local sites, (2) failure events in coated metals can be metastable, and (3) present chemical and electrochemical analytical methods which provide surface averaged information either do not detect these events, or cannot provide sufficent detail about the defect site. Thus, a fundamental understanding of the events which lead to the breakdown of a coated metal substrate requires a description of the local chemical and electrochemical events which take place within the coating and at the polymer/metal interface. This research has used local electrochemical impedance mapping (LEIM) and spectroscopy (LEIS) to examine initial breakdown sites on immersed coated aluminum samples, and monitored these sites as a function of time. Additional characterization of the underfilm solution chemistry was performed using capillary electrophoresis (CE). Neat resins of polyurethane (ca. 10 microns thick) were spin cast onto samples of aluminum alloy 2024-T3 having various surface pretreatments. Coated samples were exposed to chloride containing solutions and monitored via LEIM(S). LEIM has determined that failure of contiguous films originates at at least two different types of sites. These sites, termed red and black due to their respective colors, differ in their local impedance spectra and underfilm chemistry. The local spectra of red sites is less insulating than black sites. In addition, local chemical analysis of the solution adjacent to these sites using CE indicates increased Cu++ within the red sites, and Mg+ and Zn+ within the black, suggesting a role of at least two different phases within the alloy. The initiation of red sites can be located via LEIM by the aperance of and impedance increase. This increase may be due to the nucleation of either associated water at the coating/metal interface, or the initial development of a blocking corrosion product.

EE6.2 
IMPEDANCE STUDY ON PHYSIOCHEMICAL PROCESSES AT ELECTRODE/EPOXIDE INTERFACE: GOLD VERSUS TIN/LEAD PLATED ELECTRODES. Jane M. Terry and Laura J. Turbini, Georgia Institute of Technology, School of Materials Science and Engineering, Atlanta, GA.

Low frequency, in situ impedance spectroscopy (105 to 10-4 Hz) was used to investigate the effects of elevated temperature and humidity on the stability of a polymer-electrode interface. Experiments focused on an unfilled expoxide resin, used as a polymer matrix in a conductive adhesive, sandwiched between gold or tin/lead plated electrodes. A temperature of 85C and a relative humidity of 85% were used to accelerate degradation mechanisms at the interface. A comparison study of two impedance samples containing tin-lead and gold-plated electrodes provided information on events occurring at the polymer-electrode interface. A suspected oxidation of tin-lead surfaces was observed in the low frequency region. The passivation was confirmed by the presence of a low frequency Warburg impedance observed in the tin-lead sample and absent in the gold sample. The lower frequency impedance response for the formation of condensed water on the electrode surface was observed in the sample containing the gold-plated electrodes. It was suspected that condensed water layers also formed in the samples containing the tin-lead plating, but only after sufficient passivation occurred which rendered the surfaces unreactive. The presence of passivation and condensed water on the electrode surface indicated regions of adhesion loss. It was proposed that regions of poor adhesion, along the electrode surface, existed after resin cure, and that water and oxygen diffused to these areas and began the passivation and/or delamination process.

EE6.3 
ELECTROCHEMICAL CORROSION OF MOLYBDENUM IN WASTE GLASS MELTS. S.K. Sundaram, Pacific Northwest National Laboratory, Richland, WA.

Electrochemical corrosion of molybdenum in seven different waste glass melts was investigated. A three-electrode cell configuration was used. Platinum was used as the counter electrode and reference electrode. Potentiodynamic scan and impedance data were used in determining electrochemical corrosion rates. Representative samples were characterized using scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS), and x-ray fluorescence (XRF) spectroscopy. The corrosion rates are correlated to the temperature of testing and chemistry of the glass melts. Passivation of molybdenum was observed in some cases. The passivation layer was found to be not stable, leading to further corrosion. Cathodic polarization reduced the transition metal ions in the melt at the electrode. The electrochemical corrosion and protection mechanisms are presented.

EE6.4
CHARACTERIZATION OF OXIDE FILMS ON ZR 2.5NB CANDU PRESSURE TUBES. M. A. Maguire, AECL Chalk River Laboratories, Reactor Material Research Branch, Chalk River, Ontario, CANADA.

Electrochemical Impedance Spectroscopy, EIS, has been used to characterize oxide films formed on Zr-2.5Nb CANDU pressure tubes (PTs). The oxide is modeled with an analog circuit comprising generalized Debye elements connected in parallel. Results from controlled experiments on anodic and thermally formed films are presented to support interpretations of model parameters in terms of physical properties of the oxide. Results are then shown of in-service oxides formed on PTs which have been removed from CANDU power reactors.

SESSION EE7: DIELECTRICS AND FERROELECTRICS 
Chair: Rosario A. Gerhardt 
Wednesday Morning, December 3, 1997 
Republic A (S)

8:30 AM EE7.1 
THE DIELECTRIC LOSS OF SINGLE CRYSTAL AND POLYCRYSTALLINE TIO2. S.J. Penn,N. McN. Alford, A. Templeton and X. Wang, South Bank University, London, UNITED KINGDOM.

TiO2 is an excellent model dielectric material. It has a high dielectric constant of around 100 and can display a very low dielectric loss. Its major drawback is that its temperature coefficient of the resonant frequency, Tcf, is very high at around 500 ppm/K. In this paper the dielectric loss of single crystal rutile in two different crystal orientations and is compared with the dielectric loss of polycrystalline TiO2 prepared in our laboratories. The dielectric loss is examined as a function of temperature from 300 K to 10 K and the behaviour is compared with models of loss based on phonon dispersion. It is found that by careful attention to processing, losses in polycrystalline TiO2 are comparable to the losses in single crystals. The tan delta of polycrystalline TiO2 at 300 K and at 3 GHz is ^-5and is5 10^-6at 15K. For comparison the tan delta of single crystal 001 rutile was found to be5.25 10^-5at 300 K and2 10^-6 at 15 K The oxygen stoichiometry in titanium oxide ceramics is found to be particularly important with respect to the dielectric loss and Raman microscopy has been performed as a tool to probe oxygen variation in a range of samples.

8:45 AM EE7.2 
VERY LOW LOSS CERAMIC DIELECTRIC RESONATOR MATERIALS. N.McN. Alford, S.J. Penn, A. Templeton, X. Wang, South Bank University, London, UNITED KINGDOM; P. Filhol, Tekelec Components, FRANCE; N. Klein, Forschungszentrum Juelich, GERMANY; J.C. Gallop, National Physical Laboratory, UNITED KINGDOM.

A major use for ceramic dielectric resonators is in cellular communications. The massive increase in demand has caused severe congestion in the bands allocated to mobile communications (0.9 and 1.8 GHz). Far better pre-selection filters are required to handle the increasing volume and one low cost and realistic option is to improve the properties of dielectric resonators by attention to processing. Certain dielectric oxide single crystals display very low loss at microwave frequencies. On cooling the loss is generally observed to drop and Q's of sapphire at 10 GHz exceed 106 at low temperatures of around 10 K. However, single crystals are expensive and the purpose of this research is to explore inexpensive, sintered polycrystalline alternatives. By very careful attention to porosity, purity, processing and microstructure Q values approaching those of single crystals have been achieved and in alumina and titania. We have achieved the highest Q values reported in these materials at microwave frequencies. The loss of polycrystalline ceramics of Al2O3, TiO2, Ba(Mg1/3Ta2/3)O3 (BMT) and Zr0.875Sn0.25Ti0.875O4 (ZTS) has been studied. Al2O3 has been studied as a model material for dielectric loss. Theory predicts that the loss in single crystal sapphire should follow a T5 dependence. However at low temperatures the loss is dominated by extrinsic and the T5 dependence does not hold. In polycrystalline alumina the intrinsic loss is immediately masked by these extrinsic losses, even at room temperature, and a simple T dependence is observed. Results on polycrystalline alumina show that Q's of > at 10 GHz and at room temperature are possible and Q's well in excess of 105 at 10 GHz and 77 K can be achieved. In higher permittivity materials with dielectric constant of around 100, Q of at 3 GHz are possible by careful attention to processing.

9:00 AM EE7.3 
THE AC ELECTRICAL CHARACTERIZATION OF STRONTIUM TITANATE CERAMICS. A. Kohli, C.C. Wang, S.A. Akbar, P. Gouma, and M. Mills, Center for Industrial Sensors and Measurements (CISM), The Ohio State University, Columbus, OH; M.A. Alim, The Ohio Brass Company, Hubbell Inc., Wadsworth, OH.

SrTiO3-based ceramics (STO) are used as internal boundary layer capacitors, substrates for growth of thick-film superconductors, and oxygen sensors. The microstructures and electrical properties of STO are strongly dependent on the non-stoichiometry, doping level, and processing conditions. In this study, immittance measurement is used to investigate the effect of excess titania on the microstructures and electrical properties of STO. The small-signal frequency-dependent (10 Hz f 32 MHz) immittance data acquired at elevated temperatures (500 to 1000 C) as a function of oxygen partial pressure (1 to 104 Pa) are utilized to resolve underlying competing phenomena. The influence of sintering temperature and dopants on the transport behavior, trapping and relaxation processes is also discussed. Microstructure and electrical properties correlation thus established is further verified with the results obtained from electron microscopy.

9:15 AM EE7.4 
COMPLEX DIELECTRIC SPECTROSCOPY CHARACTERIZATION OF A Li0.982Ta1.004O3 FERROELECTRIC SINGLE CRYSTAL. Ming Dong and Rosario A. Gerhardt, School of Materials Science and Engineering, The Georgia Institute of Technology, Atlanta, GA.

The dielectric properties of a c- oriented ferroelectric Li0.982Ta1.004O3 single crystal have been investigated. This single crystal has a Curie temperature of 600C. The frequence and the temperature dependence of the dielectric properties have been measured from 500 to 650C at frequencies ranging from 10 to 106 Hz. A low frequence despersion was previously reported 1 to be due to the contribution of Li+ ionic carriers. Considering this fact, both blocking and non-blocking electrodes were used fro separating the dielectric relazation from the ionic conduction contribution. The effect of a superimposed dc bias voltage was also evaluated in order to check fro any electrode polarization effects. Based on the electrical measurement data and complex nonlinear least squares fitting, an equivalent circuit is proposed to represent the dielectric properties of the single crystal.

10:00 AM EE7.5 
MAGNETIC TRANSITION STUDIED BY ELECTRICALLY BASED METHODS IN Mn-Zn FERRITE. P. Gutierrez, A. Pelaiz*, A. Huanosta andR. Valenzuela, National University of Mexico, Mexico, D.F., MEXICO; * On leave from University of La Havana, Cuba.

Complex impedance measurements were carried out on polycrystalline samples of Mn-Zn ferrites in the 5 Hz-13 MHz frequency range and 25-200C temperature range. Cole-Cole plots exhibited semicircles which can be attributed to grain boundaries and bulk impedance responses. The relaxation frequency associated with grain boundaries showed a clear maximum at 132C. This temperature is in very good agreement with the order (ferrimagnetic)-disorder (paramagnetic) transition for this ferrite as measured by an independent magnetic technique. These results show that order-disorder transitions can be effectively investigated by means of electrical methods.

10:15 AM EE7.6 
DISLOCATION DOMAINS IN FERROELECTRIC LIQUID CRYSTALS. A. Beresnev and W. Haase, Institute of Physical Chemistry, Darmstadt University of Technology, Darmstadt, GERMANY.

Ferroelectric liquid crystals (FLC's) with high spontaneous polarization can possess periodically distributed disclocation walls, separating regions with oppositely inclined smectic layers (dislocation domains). Precise X-ray investigations of the structures of these domains (period of domains 10-15 m at room temperature) were accomplished in collaboration with A. Iida et al. using a microfocus X-ray beam of synchrotron radiation. Here the presence of regions with oppositely inclined smectic layers was confirmed. But the interchange of smectic layer inclinations while scanning the domain texture was not found, though the resolution of the microfocus was sufficient to record the domain structure. Thus a modified model is proposed, where the inclination of smectic layers in one domain changes the sign in the middle of the cell if one goes from one electrode to another. We present also the observations of dislocation walls, visible in the free surface of the FLC film, obtained by disassambling the FLC cell. On both electrodes the FLC film saves the dislocation walls, going from electrode to free surfaces. Also a mechanical-optical effect was found, connected with the proposed domain structure. If the upper electrode of a FLC, is shifted on very small distance in the direction perpendicular to the dislocation walls the initial transparent cell is converting to opaque state. The opposite shift returns the cell again to transparent state. This transformation is totaly reversible and can be explained by a shift of the upper domain texture on a distance less than half of the domain period, and with the fact, that the dislocation walls of the upper half of the FLC layer do not coincide with the dislocation walls of the bottom half of the FLC layer.

10:30 AM EE7.7 
CORRELATION BETWEEN ELECTRICAL PROPERTIES AND COMPOSITION / MICROSTRUCTURE OF OF SI-C-N CERAMICS. Christoph Haluschka, Christine Engel, Ralf Riedel, Technische Hochschule Darmstadt, Fachbereich Materialwissenschaft, Fachgebiet Disperse Feststoffe, Darmstadt, GERMANY; Hans-Joachim Kleebe, Universitaet Bayreuth, Institut fuer Materialforschung, Bayreuth, GERMANY; Rainer Franke, Universitaet Bonn, Physikalisches Institut, Bonn, GERMANY.

In this paper we report on the measurement of the electrical properties of multielement ceramics in the ternary Si-C-N system and the correlation with compositional and microstructural characteristics. The ceramics were derived by thermally induced ceramization of polysilazane. The chemical composition of these materials as well as their structural properties can be controlled by varying the conditions of either the polymer to ceramic transformation and of the final heat-treatment. At high temperatures, ternary silicon carbonitride separates into the binary phases SiC and Si3N4 at high temperatures. The impedance of these materials was measured using auto-balancing bridges in the frequency range between 20 Hz and 30 MHz. The measurements continuously covered the temperature range from 100 K to 1100 K. D.c. and a.c. conductivity as well as the permittivity were determined by this technique. Especially the d.c. conductivity varies by several orders of magnitude depending on the synthesis conditions. This was correlated to the chemical composition, the hybridization state and the microstructure characterized by chemical analysis, XANES, Raman spectroscopy, HRTEM and XRD. The electrical measurements provide also information about the conduction mechanism, which is influenced by the structure, in particular, by the short range order of these materials.

10:45 AM EE7.8 
ELECTRICAL PROPERTIES OF INTEGRATED Ta2O5 METAL-INSULATOR-METAL CAPACITORS. B. C. Martin, Motorola Inc., Materials Research and Strategic Technologies / Semiconductor Products Sector, Mesa AZ; C. Basceri, S. Streiffer, and A. Kingon, North Carolina State University, Department of Materials Science and Engineering, Raleigh, NC.

The electrical properties of reactively sputtered, thin-film Ta2O5 as a high-permittivity dielectric have been found to be promising for integrated metal-insulator-metal (MIM) capacitors. The typical stack capacitor used in this study comprised a 2 K TiN top electrode, 400 Ta2O5, and a 2 K TaN bottom electrode. All three film layers in the MIM capacitor structure were reactively sputter deposited in a MRC Eclipse system and analyzed to determine the microstructure. Electrical data from our 6-inch, 0.5 m BiCMOS flow yielded a mean specific capacitance of 5 fF/m2, and a leakage current density 1 10-7 A/cm2 at 3.5 V for a 400 Ta2O5 film. The dielectric properties along with the current-time behavior as a function of the dielectric thickness, field, and temperature will be presented.

11:00 AM EE7.9 
DIELECTRIC BREAKDOWN IN SILICON NITRIDE CAPACITORS USED IN MONOLITHIC MICROWAVE INTEGRATED CIRCUITS. S. Dayal, G.C. Dubey, Solid State Physics Laboratory, Delhi, INDIA.

Various types of rupture patterns are observed during dielectric breakdown of insulating films. These have been classified as propagating single hole and streak breakdowns. These are sensitive to the structure of films and also dependent on the dielectric strength. On the basis of dielectric strength three types of break down have been reported i.e. primary, secondary and tertiary. The present paper reports the rupture patterns observed in Si3 N4 insulating films. Microwave Monolithic Integrated Circuits (MMIC) utilize these capacitors for many applications. The rupture pattern observed in these films are of two types only. These have been correlated with the processing parameters of the film and with dielectric strength of the film.

11:15 AM EE7.10 
RELATIONS OF MICROSTRUCTURE AND ELECTRICAL PROPERTIES FOR BaRuO3 THIN FILM. C.M. Chang, C.H. Tai and Y.F. Chou, Industrial Technology Research Institute, Material Research Laboratory, Taiwan, CHINA.

Oxide materials possess enormous potentials for electronic device applications. Electronic device performance requires excellent chemical and structural compatibility between thin film layers of different materials. Metallic alkaline ruthenates, BaRuO3, has been expected to be a potential candidate for electrode materials.(1) We find that the electrical transport behavior of BaRuO3 thin film can be correlated to its microstructure. BaRuO3 thin films of different thermal treatment histories show distinct electrical properties. In this study, BaRuO3 thin films were prepared on (100)Si substrates using pulsed laser deposition technique following by in situ or ex-situ annealing processes. Resistivity of thin films were examined by the four-probe method. X-ray diffractometer was used to determine phase structure and lattice constants. Surface morphology and microstructure were observed by a scanning electron microscope (SEM) and an atomic force microscope(AFM). Cross-section transmission electron microscope(TEM) is suggested to observe interfacial microstructure if necessary.

11:30 AM EE7.11 
MEASUREMENTS OF NON UNIVERSAL PERCOLATION EXPONENTS IN SYSTEMS WITH DIFFERENT CONDUCTING COMPONENTS, WHICH HAVE SIMILAR 3D DISTRIBUTIONS. C. Chiteme and D.S. McLachlan, Physics Department, University of the Witwatersrand, Johannesburg, SOUTH AFRICA.

The conductivity and dielectric constant of percolation systems obey the following equations, =( for (; = and = for (). Measurements on computer simulations, model systems and many continuum percolation systems, when fitted to these equations, give the universal or close to the universal values of s and t (0.87 and 2.0). However, many continuum systems give rise to s and t values which differ from the universal values. For t, values as high as 4.0-6.0 have been observed. The experiments reported here are designed to clarify what physical parameters determine s and t in continuum systems. The systems studied contain a number of different granular conducting powders (scale 3-10 m), whose macroscopic distribution (scale 300 pm) is the same, but where the various conducting powders distribute themselves differently on the granular scale. This leads to different links, blobs and nodes configurations (or fractal like patterns), with different inter granular conductance. To date the conducting powders used include carbon black, graphite, graphite boron-nitride mixtures, nickel (magnetic) and granular NbC (superconducting). A wide range of s and t values have been observed in these systems and the correlation between these exponents and the nature and distribution of the conducting powders will be reported on.

SESSION EE8: IONIC AND MIKED CONDUCTING MATERIALS 
Chair: Harry L. Tuller 
Wednesday Afternoon, December 3, 1997 
Republic A (S)

1:30 PM *EE8.1 
NON-DEBYE AND CPA BEHAVIORS OF IONIC MATERIALS. J.C. Wang, Energy Division, Oak Ridge National Laboratory, Oak Ridge, TN.

Non-Debye and constant-phase-angle behaviors associated with the bulk and interfacial processes involving ionic materials have been widely observed with, for example, impedance measurements. In both of the behaviors, the spectrum (impedance, admittance, or dielectric function plot) may have a constant phase angle (CPA) over a wide frequency range. Attempts have been made to connect the CPA behavior to the micro- or defect structures of materials in form of fractal, pore, and ion-hopping models. In this paper, the yielding of a CPA spectrum from these models is reviewed. To aid the discussion, the construction of infinite-frequency range CPA elements using three distinct types of resistor-capacitor networks will be presented and utilized. The observed wide frequency ranges of the CPA behavior suggest that the fractal and pore models, which require a wide range of special geometrical features down to submicron levels, may not be realistic. The ion hopping model is more general and is similar to the Arrhenius model for chemical reactions in a thermally-activated form. Because of thermal fluctuations, the activation energies have an exponential distribution which can yield a CPA spectrum over a wide frequency range. Shortcomings of the ion-hopping model will be discussed.

2:00 PM EE8.2 
INVESTIGATION OF THE COMPLEX CONDUCTIVITY OF NANOCRYSTALLINE Y2O3-STABILIZED ZIRCONIA. Pia Mondal and Horst Hahn Department of Materials Science, Thin Film Division, Darmstadt University of Technology, Darmstadt, GERMANY.

Because of its nearly pure ionic conductivity polycrystalline Y2O3-stabilized zirconia is a well characterized material for oxygen sensors and solid oxid fuel cells for many years. Nanocrystalline structures exhibit extremely fine grain sizes and consequently large fractions of grain boundaries. Hence, the electrical properties like total conductivity and activation energies may be affected by the larger fraction of grain boundaries in the nanocrystalline Y2O3-stabilized zirconia. On account of the possibility of observing the lattice relaxation and the grain boundary relaxation in different frequency and temperature ranges the impedance spectroscopy is an excellent method for the investigation of this question. Nanocrystalline ZrO2- and Y2O3-powders with grain sizes from 5 - 20 nm were prepared by the Inert Gas Condensation Method. The powders were mixed in different fractions, pressed into green bodies and sintered to ceramic samples with 25 - 49 nm average grain size. The phase, grain size and specific surface area of the powders and the ceramic samples was characterized by X-ray diffraction, nitrogen adsorption and High Resolution Electron Microscopy. Densities from 82 to 96 % of the single crystal density were determined using the Archimedes principle. The complex conductivity was measured for frequencies ranging from 20 Hz to 3 MHz at temperatures between 80 K and 1100 K. The lattice relaxation, grain boundary relaxation and the contact process could be resolved and identified. The dc-conductivity for the lattice and the grain boundaries were deduced from the data. The activation energies for ac- and dc-conductivity were found to be equal and were determined for the lattice and the grain boundaries as eV and eV, respectively.

2:15 PM EE8.3 
THERMALLY STIMULATED CURRENT MEASUREMENTS ON CERAMIC SENSOR MATERIALS. B. Lakshminarayanan, C.C. Wang, and S.A. Akbar, Center for Industrial Sensors and Measurements (CISM), The Ohio State University, Columbus, OH; M.A. Alim, The Ohio Brass Company, Hubbell Inc., Wadsworth, OH.

Thermally stimulated current (TSC) measurements are used to study the role of defect states at the grain-boundary regions on the electrical transport properties of the yttria-based thermistors and gas sensor materials. These defect states attributed to a trapping effect are characterized by the trap-filling mechanism. The TSC measurements are conducted on these ceramic materials using an in-house built automated experimental setup, which is capable of measuring low currents (< 10-15 A) in a wide temperature range (from room temperature up to 1000C). A systematic TSC measurement is performed on polycrystalline samples of different compositions and heat-treatment conditions. Additional TSC spectra of the yttria-based thermistors are obtained to study the role of defect states at the grain-boundary regions in the aging behavior. The resulting electrical characteristics are modeled considering the microstructural feature.

2:30 PM EE8.4 
ELECTRICAL PROPERTY AND MICROSTRUCTURAL CORRELATION STUDY OF Y2O3 DOPED CeO2 THIN FILMS. Chunyan Tian, Siu-Wai Chan, Dept. of Chemical Engineering, Materials Science, and Mining Engineering School of Engineering and Applied Science, Columbia University, New York, NY.

CeO2 based electrolytes have been shown to have an exciting future when used in oxygen sensors and intermediate temperature (500-700C) solid oxide fuel cells (SOFC) because of their higher ionic conductivity over YSZ. Furthermore, using thin layers of solid electrolytes can minimize the ohmic loss in the electrolytes so as to lower the operating temperature and maintain the required power output. Therefore, it is highly desirable to produce thin layers of these materials for SOFC applications. In this work, the thin films of Y2O3 doped CeO2 have been prepared on a variety of substrates using an e-beam deposition technique. Rutherford backscattering spectroscopy, x-ray diffraction and transmission electron microscope were used to characterize the composition and microstructure of the films. Both epitaxial and textured films were observed on different substrates. AC impedance spectroscopy has been used to study the electrical properties of these films. Only grain boundary arc was observed in the film complex impedance plots. The resistive grain boundaries were found to dominate the conductivities of the films comparing to the bulk of the same dopant concentration. The electrical properties of the films change with the grain boundary orientations.

3:15 PM EE8.5 
EXTENDED D.C. ELECTRICAL TRANSPORT OF THE MIXED CONDUCTOR CU3CS2. Peter K. LeMaire, Jeffery Benoit, Rodney Speel, Central Connecticut State University, Department of Physics and Earth Sciences, New Britain, CT.

D.C. electrical transport measurents have been done over the temperature range 200 K to 450 K on the mixed conductor Cu3CS2. Above 220 K, the voltage versus time curves follow the Yokota model for mixed conductors. Below 220K the voltages were practically constant with time, suggesting very little ionic transport below this temperature, and an electronic conductivity of the order of 10-5S cm-1 at 200 K. At ambient temperatures, the ionic and electronic conductivities were of the order of 10-3S cm-1, and the chemical diffusion coefficient was found to be of the order of 10-6 cm2s-1, in agreement with earlier work on Cu3CS2. Above 220 K, the ionic conductivity versus temperature plots were of the Arrhenius form with an activation energy of about 0.4 eV. The average jump time and residence time of the mobile ions were estimated to be of the order of 10-12s and 10-6 respectively, confirming hopping as the mode of ionic transport. The electronic conductivity versus temperature plot supported thermal activation as the mode of electronic transport. The results suggest the CuxCS2 group of mixed conductors to be very stable, and the Yokota model, with minor modification, to be very reliable for the analysis of these and other similar mixed conductors.

3:30 PM EE8.6 
CONDUCTIVITY AND STABILITY OF Gd2((Mo1/3Mn2/3)xTi(1-x))2O7. J. J. Sprague, O. Porat, H. L. Tuller, Dept. of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA.

A composite solid state electrochemical device, with (Gd1-xCax)2Ti2O7 serving as the electrolyte and Gd2(Ti1-xMox)2O7 (GT-Mo) as the anode has recently been proposed. The latter exhibits high levels of mixed conduction under reducing atmospheres, but decomposes at high PO2. We have recently succeeded in extending the stability limits of the GT-Mo to higher PO2 with the addition of Mn. In this study, we report on the conductivity and stability of Gd2((Mo1/3Mn2/3)xTi(1-x))2O7 (GMMT) as a function of PO2, T, and composition utilizing impedance spectroscopy, electrode blocking cells, defect modeling, and x-ray diffraction. The addition of Mn extends the stability region of the material to PO2 = 1 atm and retains a level of conductivity higher than that acheivable by Mo alone. The role of this material as a cathode in a composite pyrochlore cell is discussed.

3:45 PM EE8.7 
COMPOSITION-DEPENDENT ELECTRICAL CONDUCTIVITY OF IONIC-ELECTRONIC COMPOSITE. Young Min Park, Gyeong Man Choi, Pohang University of Science and technology, Dept of Materials Science and Engineering, Pohang, SOUTH KOREA.

Ionic-electronic two-phase composite material may have unique electrical properties due to their blocking interface which is not present in the more general insulator-conductor composite. In this study, Oxygen ion conducting zirconia and hole-conducting NiO were selected as the component of the ionic-electronic composite considering their similar magnitude of electrical conductivity and limited solid solubility. The electrical conductivity of yttria(8 mol%) stabilized zirconia(YSZ)-nickel oxide composites was measured by impedance and 4-probe d.c. conductivity between 200 and 1000C. The high temperature electrical conductivity of the composite decreased with NiO addition within the solubility limit of NiO in YSZ. Further addition of more-conductive NiO did not increase the conductivity until the NiO percolates the composite. The percolation was also shown in the activation energy of the conductivity. From the galvanic cell measurement and the oxygen partial-pressure dependence of electrical conductivity, the percolation and thus mixed conduction was observed for x between 0.4 and 0.8(x in xNiO-(1-x)YSZ). Thus the electrical conductivity of YSZ-NiO composite was explained by the blocking effect of NiO on ionic conduction in YSZ-rich composition and the percolation by NiO in intermediate composition. The effect of ionic-electronic interface is also significant below 600C.

4:00 PM EE8.8 
CONTACT EFFECTS IN IMPEDANCE SPECTRA OF ELECTROCERAMICS. J.-H. Hwang, J.D. Shane, S.J. Ford, T.O. Mason, Northwestern University, Dept of Materials Science and Engineering, Evanston, IL; G. Hsieh, Pacific Northwest Laboratory, Materials and Chemical Sciences, Dept, Richland, WA; E.J. Garboczi, National Institute of Standards and Technology, Building Materials Division, Gaithersburg, MD.

Point and imperfect planar electrodes play an important role in the electroding and subsequent electrical characterization of electroceramics. For example, spreading resistance effects can be manifested in the ``bulk'' portion of an impedance spectrum, resulting in misleading interpretations of data. On the other hand, point and imperfect planar electrodes can be used to probe near-interface heterogeneities in electroceramics, e.g., composition gradients, grain boundaries, etc. Contact probe impedance spectroscopy (CPIS) is a potentially powerful technique to study such electrical/dielectric heterogeneities. Additional contact effects are observed in composite systems with a conductive second phase, e g., fibers. Experimental results and simulations are presented for several systems, including coarse-grained zirconia, nanophase ceria, cement paste, and cement-based composites.

4:15 PM EE8.9 
CHARACTERIZING THE DISPERSION OF CONSTITUENTS IN CONCRETE BY ELECTRICAL RESISTIVITY MEASUREMENT. Pu-Woei Chen, Xuli Fu and D.D.L. Chung, Composite Materials Research Laboratory, State University of New York at Buffalo, Buffalo, NY.

The method of characterising the degree of dispersion of constituents in cement paste or mortar by DC electrical resistivity measurement was demonstrated for the case of the constituent being much more conducting than cement (namely short carbon fibers) and for the case of the constituent being much less conducting than cement (namely latex particles). The degree of fiber dispersion is described by the ratio of the measured conductivity to the calculated value. The degree of latex particle dispersion is described by a factor (<1) in a model which considers latex and cement phases partly in series and partly in parallel.

4:30 PM *EE8.10 
Marc Doyle, University of California-Berkeley, Berkely, CA

Abstract Not Available