1997 EMC: Thursday Morning Sessions, Part 1

The following sessions are among those that will be held during the 39th Electronic Materials Conference (EMC) on Thursday morning, June 26, at Colorado State University, Fort Collins, Colorado. To view the other Thursday morning sessions as well as other programming planned for the meeting, go to the EMC Calendar of Events.

8:20 am

Origin of Photoluminescence Red Shift in InP Quantum Dots: H. Fu and A. Zunger, NREL, 1617 Cole Blvd., Golden, CO 80401

The origin of photoluminescence(PL) in quantum dots (QDs) and its red shift relative to the excitation are both controversial and important. Two main mechanisms were previously considered to explain the red-shifted PL in QDs: (1) trapping of carrier by surface states, and (2) emission from exchange-split forbidden states. However, current theoretical investigations do not take into account the dots surface, and the exchange interactions are not calculated via ab-initio methods. Here, an atomistic nonlocal pseudopotential method is used. It allows incorporation of surface state and provides explicit microscopic wave functions that are used for calculating exchange interactions. We apply the method to InP quantum dots. The results show: (i) For fully passivated QDs, the predicted dependence of band gap on the dot size agrees well with experiment. (ii) No surface states exist in the band gap of fully passivated QDs. (iii) The In dangling bond at the dot surface gives rise to a surface defect state. The calculated energy of this surface defect state and its emission lifetime are in good agreement with the experimental observations in TOPO-passivated InP dots4. This kind of surface defect predicts a large red shift (~100meV), so emission from such surface defects is not the origin of the (~l0 meV) red shift observed in etched InP QD. (iv) The exchange splitting in QD is found to increase significantly relative to the splitting in bulk materials. With a reasonable screening, this exchange splitting can explain the red shift in etched InP QDs, as evolving from excitation into a singlet state, followed by emission from an exchange-split spin-forbidden state. *Supported by BES/OER/DMS under contract No. DE-AC36-83-CH10093.

8:40 am

Lateral Association of Vertically Coupled InGaAs Quantum Dots in a GaAs Matrix: A.F. Tsatsul'Nikov, A.Yu Egorov, A.R. Kovsh, V.M. Ustinov, A.E. Zhukov, N.N. Ledentsov, M.V. Maximov, A.V. Sakharov, A.A. Suvorova, N.A. Bert and P.S. Kop'ev, A.F. Ioffe Physical-Technical Institute, 194021 Polytekhnicheskaya 26, St. Petersburg, Russia; M. Grundmann and D. Bimberg, Institut für Festkörperphysick, Technische Universität Berlin, Hardenbergstr. 36, D-10623, Berlin, Germany

Recently a large attention is attracted to studies of InGaA s-GaAs quantum dots (QDs). Injection lasing via the QD states has been demonstrated. However, to realise good device characteristics one needs to increase the carrier localisation energy in a QD to suppress the carrier evaporation into the GaAs matnx at elevated temperatures. Formation of vertically coupled QDs separated by ultrathin GaAs spacer layers is a promising way to solve this problem. In this wade we investigate the new possibility to increase further the carrier localisation energy in the In_0.5Ga_0.5As QDs embedded into a GaAs matrix by formin the formation of laterally associated QDs. Deposition of several sheets (N) of In,Ga)As QDs, separated by thin (smaller or about the QD height) GaAs spacer layers, results in a transfer of the InGaAs material from the each lower to the each upper QD, and, consequently, in a progressive increase in a lateral size of the QDs with an increase in N. TEM images show that for dense arrays of dots, dhe increase in N finally results (N > 10) in a lateral overlap between neighbouring QDs For N= 1-6 photoluninescence (PL) spectra demonstrate only one narrow PL line, associated with an electron-hole (exciton) recombination via the QD ground state. For N>10 a new line appears on the low energy side of the spectrum. We attribute this line dominating the PL spectrum for N 15 to emission of laterally-associated quantum dots, having a larger volume and thus, smaller size quantization energy. We demonstrate a transport between the laterally-isolated QDs in the lower rows the and laterally-associated QDs in the upper rows via the tunnelling effect caused by vertical coupling of QDs.

9:00 am

Effects of GaAs-Spacer Strain on Vertical Ordering of Stacked InAs/GaAs Quantum Dots: S. Ruvimov, Z. Liliental-Weber, J. Washburn and E.R. Weber, Lawrence Berkeley National Laboratory, Berkeley, CA 94720; A. Sasaki, Osaka Electro-Communication University, Neyagawashi 570, Japan; A. Wakahara, Y. Furkawa, T. Abe and S. Noda, Kyoto University, Kyoto 606-01, Japan

We have proposed to use the islands formed at the InAs/GaAs heteroepitaxy as quantum dots and investigated luminescence properties from the dots. The stacked layers of InAs/GaAs quantum dots are needed as an active layer of semiconductor laser to reduce the threshold current and temperature dependence of operation. The vertical ordering of quantum dots in stacked layer has been observed in the cross sectional view of TEM. However, ordering mechanism has not been thoroughly disclosed. To investigate it, we grow the InAs/GaAs stacked layers with various thickness of GaAs and observe thickness dependence of quantum dots size. Further, we calculate the strain distribution on GaAs layer by valence-force field approach. We grew the InAs/GaAs stacked layers by MBE on the (001)-directed GaAs substrate at 480°C with beam flux ratio V/III = 8. The InAs flux is supplied for 1.8-2.0 ML and the GaAs thickness is varied from 5 to 20 nm, and the InAs/GaAs is repeated ten times. In the case of 1.8 ML of InAs flux, no vertical ordering is observed for 15 nm of GaAs spacer. On the other hand, vertical ordering can be seen in 17 nm of GaAs spacer when the InAs is 2.0 ML These vertical ordering occur on relatively large InAs dots on the first layer and it tends to become larger in size and regular in plain position by increasing stacked number. These results suggest that the vertical ordering is not affected with strain transfer from the InAs dots on the lower layer, but also total strain energy in the multi-stacked structure.

9:20 am

Carrier Dynamics in Self-Assembled GaInAs/GaAs Single-and Multi-Quantum Dot Layers: K. Kamath, T. Sosnowski, H. Jiang, N. Chervela, T. Norris, J. Singh and P. Bhattacharya, Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI 48109-2122

Single layer self-assembled quantum dot lasers realized by single-step epitaxy are now demonstrating impressive performance (J_th=650A/cm², [partialdiff]g/[partialdiff]n=l.5xl0^-14cm²) at room temperatures and we have recently measured a small-signal modulation bandwidth of 7 GHz Some of the major limitations of these devices are: (i) the differential gain is low, (ii) the optical confinement factor is low; and (iii) lasing takes place from an excited state of the quantum dots, possibly due to gain saturation. Use of mild-quantum dot (MQD) active region will increase the differential gain and confinement factor, but the carrier dynamics of the ground and excited states of such structures are not well known. We have made a systematic study of the growth and time-resolved photoluminescence behavior of single- and multi-layer Ga_0.6In_0.4As/GaAs quantum dots. Growth of the Ga_0.6In_0.4As/GaAs single-layer and multi-layer quantum dots was done by MBE. Growth temperature, growth rate and other parameters were optimized with insitu RHEID and photoluminescence (PL) measurements The nominal thickness of the dots are 1 ML in all the layers and they are vertically coupled. The PL spectra for 8-layer dots are characted by emissions from the ground state (~ 1 µm), excited state (~ 096 µm) and wetting layer (~ Q96 µm) at 18K The PL intensity increases somewhat with increase in the number of dot layers. The PL is observed in multi dot layers up to 300K and there is no saturation of the PL intensity at low (~ 18K) or higher temperatures up to an incident power of ~ lKW/cm² in these samples. More importantly, the PL linewidth (at 18K) decreases steadily with increase of dot layers, from 55 meV for a single dot layer to 36 meV for a 8 dot layer. Fine structures, with a linewidth of 4 meV, were also observed. These values indicate a higher degree of size uniformity of the pyramidal quantum dots. Time-resolved PL measurements were made with a streak camera. For a single layer dot structure the PL decay times of the ground and excited states are 2.5ns and 200 ps, respectively. While the decay time of excited state PL remains fairly constant with increase in number of dot layers, the decay time of the ground state steadily decreases, and is ~750 ps for a 8-layer dot structure. The larger decay times of the ground state emission indicate that the oscillator strength of the transition might be small, which could be a direct manifestation of the dot shape and associated strain tensor. Theoretical calculations indicate that the strain tensor in the dots are far from biaxial and varies along the dot base length. The reduction of the PL decay time with increase in the number of dots possibly due to a gradual change in the dot shape and strain tensor. The shorter decay times of the ground state emission indicate that multi-quantum dot lasers, which lase from the ground state, can be modulated at high frequencies. *Work supported by NSF, ARQ and DARPA.

9:40 am

Contactless Electroreflectance and Surface Photovoltage Spectroscopy Study of a Vertically and Laterally Coupled Quantum Dot-based InAs/GaAs Laser Structure: L. Aigouy, T. Holden and F.H. Pollak, Physics Department, Brooklyn College, Brooklyn, NY 11210; N.N. Ledentsov, V.M. Ustinov and P.S. Kop'ev, A.F. Ioffe Physical-Technical Institute, 195251 St. Petersburg, Russia; D. Bimberg, Technische Universität Berlin, D10623 Berlin, Germany

We have performed a contactless electroreflectance (CER) (300K and 20K) and surface photovoltage spectroscopy (SPS) (103K) investigation of the optical transitions in a InAs/GaAs quantum dot (period N=10)-based laser structure. Signals have been observed from all the relevant regions of the sample including the quantum dots (QDs), InAs wetting layer (WL) and GaAs sections. Two of the energies of the QD transitions are related to the vertical coupling between the dots while a weaker feature at lower energies is in agreement with a lateral coupling between neighboring dots on the upper rows. The two observed WL features correspond to transitions involving heavy- and light-hole excitons in an InAs quantum well, formed by the WL, with an effective thickness of about one monolayer. We have fit the CER spectra and the numerical derivative (with respect to photon energy) of the SPS spectra from the QDs and WL to the first derivative of a Gaussian profile to accurately determine the energies of the observed features. The GaAs portion of the spectra was accounted for on the basis of a third-derivative band-to-band functional form. Three features, at 1.2 eV, 1.26eV and 1.31 eV, have been observed from the QDs. The intense feature at 1.26 eV agrees well with the intense peak in PL and EL spectra and is attributed to the ground state transition in the VECOD. The weaker structure at 1.2 eV corresponds to the low energy tail of the PL emission and is attributed to lateral coupling between neighboring dots on the upper rows. This is due to the higher probability for lateral coupling and association of the QDs as the QD lateral size gradually increases with N. The WL exhibited resonances at 1.41 eV and 1.45 eV, in good agreement with an envelope function calculation of an InAs QW with an average thickness of about 1.4 ML. For the GaAs portion the two CER features at 1.51/1.42 eV and 1.56/1.47 eV at 20K/300K, respectively, correspond to the band gap of undoped and doped (Burstein-Moss sheft) GaAs. The Gaussian nature of the QD and WL features indicates an inhomogeneous broadening this is probably due to fluctuations in the size and thickness of the nanostructures.

10:00 am, Break

10:20 am

Atomic Hydrogen Assisted MBE on Patterned GaAs (311)A Substrates: Formation of Quantum-Wire and Quantum-Dot Arrays: R. Nötzel, Z. Niu, L. Däweritz and K. H. Ploog, Paul-Drude-Institut für Festkörperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany

Strong modification of the surface morphology on a mesoscopic scale occurs in atomic hydrogen assisted MBE of GaAs on GaAs (311)A substrates. With atomic hydrogen the surface structure changes from a smooth morphology exhibiting shallow islands on a µm length scale to a quasi-periodic step array along [-233] with about 40nm lateral periodicity which is comparable to that observed in MOVPE for similar growth conditions. Hence, atomic hydrogen is identified to be responsible for the accumulation of the microscopic surface corrugation due to step bunching. In the following we combine this natural wirelike surface structure with patterned growth to produce arrays of quantum dots: On patterned GaAs (311)A substrates mesa stripes along [01-1] exhibit a fast growing sidewall. Preferential migration of Ga atoms from the mesa top and bottom towards the sidewall forms a smooth convex curved surface profile without facets. For step heights of 10-15 nm quasi-planar quantum wires with excellent structural and electronic properties can thus be fabricated. When atomic hydrogen is introduced during growth on patterned GaAs (311)A substrates the morphology changes again to the distinct periodic step array along [-233] (ie., perpendicular to the mesa). The step structure is continued over the convex curved surface profile without any displacement. Therefore, the natural wirelike surface corrugation can be combined with pattemed growth to form arrays of quantum dots with precise control over the position that is important for many applications of quantum dots in novel electronic and optical devices.

10:40 am

Self-assembled Silicon Quantum Wires on Ultrasmooth Sapphire Substrates and Photoluminescence Properties: S. Yanagiya, Y. Moriyasu, S. Kamimura, M. Fujii and M. Ishida, Department of Electrical and Electronic Engineering, Toyohashi University of Technology, Toyohashi 441, Japan; M. Yoshimoto, T. Ohnishi, K. Yoshida, K. Sasaki and H. Koinuma, Materials and Structures Laboratory, Tokyo Institute of Technology, Nagatsuta 4259, Midori-ku, Yokohama 226, Japan

Nanofabrication techniques, allowed to realize quantum effect devices, have been extensively studied in many material systems by various methods. Particularly natural formation techniques, such as self-assembled quantum wires (QWRs) and dots (QDs), have been an extremely active area of current research, because these methods feature easy fabrication process of uniform and high-density QWRs and QDs without lightographic processes. However, as regards the silicon, only a little work has been reported the self-assembled nanostructures. On the other hand, Si nanostructures have been attracted considerable interest for realizing Si-based visible light emitting devices. The results of previous experimental and theoretical studies on Si-related materials, for instance, porous SI and Si nanoparticles embedded in SiO₂, etc. indicated that silicon nanostructues were indispensable system for visible light emission. In this work, we demonstrated fabrication of self-assembled silicon QWRs on ultrasmooth sapphire substrates and observed visible photoluminescence from the wires. Self-assembled Si wires were grown on ultrasmooth sapphire substrates (R-plane) with atomic steps and atomically flat terraces by gas-source molecular beam epitaxy adopting Si₂H₆ as the source gas. Although the wires are self-assembly formed by step-flow growth mode only under proper growth conditions, especially the formation process is sensitive to flatness of the terraces. Si wires with 80 nm-width and 1 nm-height were clearly observed along the atomic steps of the substrate by atomic force microscopy. The wires are perfectly isolated each other, uniform and sufficiently small to be considered quantum wires. Photoluminescence (PL) was measured from 9K to 300K on a standard cryostat system by using 488nm argon laser line. Visible photoluminescence was observed from the Si QWRs at temperatures in the 9K-300K range. The PL spectra indicated broad emission by peak photon energy of 1.65eV with FWHM of about 50nm. The peak photon energy is almost independent of the temperature, while PL intensitiy gradually decreased with increasing temperature. These characteristics are very similar to those from porous Si and other nanostructured Si. Hence, the visible emission from the self-assembled Si QWRs is considered to be concerned with the quantum size effect and Si oxide. These PL results are well interpreted with three-region model which is suggested by Kanemitsu et al. We speculate that there is a thin oxide region at the interface between the Si QWRs, and the visible luminescence is emitted through the radiative recombination center which exists in the disordered localized states in the interface region.

11:00 am

Formation of Arrays of In_0.53Ga_0.47As Coupled Quantum Wire-Dot Structures by Selective Molecular Beam Epitaxy on Patterned InP Substrates: H. Fujikura, Y. Hanada, M. Kihara and H. Hasegawa, Research Center for Interface Quantum Electronics and Graduate School of Electronics and Information Engineering, Hokkaido University, Sapporo 060, Japan

For realization of ultralarge scale integrated circuits based on quantum devices such as single electron devices, it is necessary to establish a suitable technology for formation of high density arrays of coupled quantum structures consisting of high-quality wires and dots. This paper presents a novel selective MBE growth method for formation of arrays of InP-based In_0.53/Ga_0.47As high quality coupled quantum structures consisting of wires and dots embedded in In_0.52Al_0.48As barriers. The basic idea is to apply the selective MBE growth technique to InP substrates having a special mesa-pattern consisting of an array of mesa-pedestals connected with each other by mesa-stripes in order to realize coupled wire-dot structures after growth. In order to find appropriate growth patterns and conditions for growth of such complicated structures, selective MBE growth properties of InGaAs and InAlAs layers, such as growth selectivity and uniformity, development of facets, modulation of growth rates and so on, were investigated in detail first, using various kinds of patterned substrates. It was found that arrays of the isolated InGaAs quantum wires and dots can be successfully realized on -oriented mesa-stripes and square-mesas with <100>-oriented edges, respectively. Then, growth was done using an array of <100> square-mesas connected by -oriented mesa-stripes. SEM observations have shown that pyramidal-shaped dot structures surrounded by four (521) facets appear on the square-mesas and arrow-headed quantum wires appear on the stripes, respectively. Strong emission lines and spots were observed in spatial resolved CL images at the position corresponding to the stripe-mesa and square-mesa regions, respectively, indicating the formation of high quality InGaAs quantum wires and dots. Existence of a potential barrier between the wire and the dot was also indicated from the CL images. The width and height of this potential barrier can be obviously controlled by the growth condition.

11:20 am

Fabrication and Transport Properties of GaAs and InGaAs Quantum Wires Controlled by Novel Schottky In-Plane and Wrap Gates: H. Okada, S. Kasai, H. Fujikura, T. Hashizume and H. Hasegawa, Research Center for Interface Quantum Electronics and Graduate School of Electronics and Information Engineering, Hokkaido University, Sapporo 060, Japan

Gate controlled quantum wires (QWRs) and quantum dots (QDs) are interesting structures for application to future electron devices. However, the previous split-gate geometry widely used to realize GaAs-based quantum structures on high-quality 2DEG wafers realizes only weak confinement potential for electrons, thereby limiting the device operating temperature in mK up to a few K. To overcome this difficulty, we have recently proposed novel Schottky in-plane and wrap gate structures (WPG) controls 2DEG from both sides and top. Using these structures, we have successfully fabricated quantum wire transistors and single electron transistors working at several ten K. However, the basic transport properties of these novel structures have to be clarified for future improvements in structures and performance. The purpose of this paper is to fabricate and characterize GaAs and InGaAs quantum wires controlled by IPGs and WPGs. IPG controlled QWRs were fabricated on AlGaAs/GaAs 2DEG wafers. 2DEG bars with width of 400-100nm were defined by electron beam lithography and wet chemical etching WPG controlled QWRs were formed on InAlA/InGaAs ridge QWR structure which was formed by selective MBE growth on patterned InP substrate. InGaAs wire possesses arrow-headed cross section with geometrical width of 3-100nm. Pt Schottky IPG and WPG electrodes were formed by a novel in-situ electrochemical process where native oxide etching and Pt deposition were done in the same electrolyte by bias polarity change at room temperature. As compared with the conventional metal deposition process, this low-energy process gave substantially increased Schottky barrier heights (GaAs: 1.1eV, InAlAs: 0.9eV). Both IPG QWR and WPG QWR showed good gate control of the wire conductance. Both showed clear Shubnikov-de Haas (SdH) magnetoresistance oscillation, and non-linear Landau plots in low magnetic fields confirmed the one-dimensional electron transport in both points in the Landau plots. Effective wire width of IPG QWR was found to be linearly decreased with the negative gate bias without changing the effective sheet electron density. On the other hand, the width of WPG QWR remained almost constant for small gate biases, changing only effective electron density by applying further negative bias. Short wires showed conductance quantization in the unit of 2e²/h, whereas long wires showed steps with much smaller step heights. The narrowest GaAs IPG wire showed first conductance plateau up to 100K, indicating increased confinement potential. Near pinch-off, both QWRs showed clear Coulomb blockade type conductance oscillations which persisted up to 50K in the case of InGaAs WPG wires. Output current voltage characteristic showed regular diamond-shaped behavior vs. gate voltage, indicating large effective Coulomb gap (~30meV). The result is explained in teens of the dot segment formation near pinch-off.

11:40 am

New Route to Reduce Ionized Impurity Scattering in Modulation-Doped GaAs Quantum Wells with High Electron Densities: R. Hey, K.-J. Friedland, H. Kostial and K.H. Ploog, Paul-Drude-Institut für Festkörperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany

The design of GaAs/AlAs layer sequence is optimized in order to open a new route for the reduction of ionized impurity scattering in modulation-doped GaAs single quantum wells (SQW). The barriers of the GaAs SQW are formed by AlAs GaAs short-period superlattices (SPS). Modulation doping on both sides of the SQW is achieved by equally spaced Si single -doping sheets in GaAs layer of the SPS. In these type-II SPS barriers heavy-mass X-electrons smooth the potential fluctuations caused by the randomly distributed ionized Si dopants. By varying the Si density and their separation from GaAs SQW low-temperature mobilities (µ) of 120 m²/Vs at electron densities (n) as high as 1.4x10¹⁶m^-2.are obtained in a 10 nm wide GaAs SQW in the one-subband conductivity mode without any parallel conductance. The population of only the lowest subband even at these high two-dimensional electron gas (2DEG) densities manifests itself in the single-mode resistance oscillations of the Shubnikov-de-Haas measurements. The existence of X electrons were additionally proved by differential capacitance measurement and self-consistent calculations. The heavy-mass X electrons have a high screening capabilities and reduce the ionized impurity scattering considerably. These charges do not contribute to the conductivity because they are localized close to the dopants and hence no parallel conductance is observed. The concept to reduce impurity scattering is valid if dopant segregation is negligible and intemixing does not lift the type II SLS character. The conductivity enhancement is effective as well if screening of ionized impurities is achieved by doping on both sides of the SQW. The high electron conductivities even at 77 K with a (n x µ) product higher than 2xl0¹⁷ (Vs)^-1 makes this new heterostructure very attractive for application in low-noise cryogenic amplifiers operating at very high frequencies.

8:20 am Invited

Surface Photovoltage Spectroscopy of Semiconductor Materials, Structures and Devices: L. Kronik, Department of Physical Electronics, Tel-Aviv University, Ramat-Aviv 69978, Israel

The contactless, nondestructive technique of surface photovoltage spectroscopy (SPS) has been developed into a powerful and versatile tool to study a variety of semiconductor systems including bulk/thin film structures, micro- and nano-structures (e.g., quantum wells, superlattices, quantum dots), and surfaces/interfaces (including heterojunctions) as well as actual device structures (e.g., HBTs, QW lasers, and solar cells). SPS results can yield information about interband (intersubband) transition energies, minority carrier diffusion lengths and lifetimes, surface recombination velocities, and quantitative information about surface and bulk gap states. In addition, by using SPS it is possible to obtain a complete quantitative construction of heterojunction band diagrams, including the band offsets. SPS utilizes a Kelvin probe to measure surface potential variations as a function of incident photon energy (as well as illumination intensity and time). Since its emergence in the early seventies, it was used primarily for the determination of surface state properties. Lately, it has been increasingly applied to more general semiconductor characterization problems. In this talk, some recent applications of SPS which demonstrate the current status and capabilities of the technique are presented. These include: quantitative analysis of surface state properties (demonstrated at II-VI and m-V surfaces), determination of semiconductor band offsets (demonstrated for the InP/In_0.53Ga_0.47As heterojunction), monitoring of quantized energy levels in GaAs/AlGaAs multi-quantum well and superlattice structures, and quality control and characterization of Cu(In,Ga)Se₂-based solar cells.

9:00 am, Student Paper

Characterization of Semiconductor Quantum Well Laser Structures by Surface Photovoltage Spectroscopy: N. Ashkenasy, M. Leibovitch and Y. Shapira, Department of Physical Electronics, Faculty of Engineering, Tel-Aviv University, Ramat-Aviv 69978, Israel; F.H. Pollak, Physics Department, Brooklyn College of CUNY, New York, NY 11210

Surface photovoltage spectroscopy (SPS) has been applied for determining important performance parameters of novel devices, such as quantum well (QW) diode lasers and heterojunction bipolar transistors (HBTs). This technique satisfies the characterization demands of devices containing very thin and/or highly doped layers. These demands include non-destructive, contactless operation at room temperature, wafer-size sample handling, and resolution of quantum level transitions. SPS measurements and their analysis precisely determine photon-induced electron transitions and the accompanied electric field redistribution in the entire device. A GaAs/ AlGaAs-based HBT and an InGaAs/ GaAs/ AlGaAs graded index of refraction separate confinement heterostructure (GRINSCH) laser have been studied. It is shown that monitoring the internal electric fields in the device and their changes due to illumination is a crucial step for device analysis. Such an analysis makes it possible to determine the electron mobility at the HBT base, from which the common-emitter current gain-beta may be extracted. Also QW electric fields and transition energies, may be independently monitored. Since in low dimensional structures, the latter two are dependent through the quantum confined Stark effect, the SPS results may shed light on this effect and its influence on device performance. In addition, it is shown that the field redistribution is dominated by the doping. Thus, the doping levels at different layers of the structure may be extracted. It is shown that a detailed analysis of the spectrum yields growth parameters, such as layer width (especially of well layers), compound mole fraction (even of highly doped layers), and the doping in such highly doped layers. In addition, device parameters, such as the diode built-in- voltage, lasing frequency and the HBT beta may also be extracted.

9:20 am

NFPVS-Near Field Photovoltage Spectroscopy, a New Non-Destructive Method for Semiconductor Characterization: T. Meoded, N. Fried and Y. Rosenwaks, Department of Physical Electronics, Faculty of Engineering, Tel-Aviv University, Ramat-Aviv 69978, Israel

In this work we present a new non-destructive method for characterizing semiconductor optoelectronic properties with nanometer spatial resolution. This technique, which we hereafter call near field photovoltage spectroscopy (NFPVS), combines the measurement of contact potential difference (CPD) with near field, optical excitation. This new technique is likely to have important contributions in at least four main areas: 1) surface state spectroscopy of semiconductors, 2) recombination mechanisms (bulk and surface) in semiconductors, 3) electronic properties of quantum confined semiconductor structures, and 4) characterization and failure analysis of very large scale integrated (VLSI) circuits. The key feature of the technique is that the excited semiconductor sample is in the optical near field region of a tip that measures the electric force between the tip and the sample; in such a case the lateral resolution is determined by the diameter of the aperture at the end of the tip and is not limited by diffraction effects. In addition, the intensity of the light is greatest within a few nm of the tip falling off exponentially with increased depth in the Z direction (perpendicular to the crystal surface). This, in addition to the Beer-Lambert light absorption by the semiconductor makes the NFPVS an ultrasensitive technique for probing surface phenomena. The talk will describe in detail the proposed NFPVS method and preliminary results obtained on bulk semiconductors like CdSe, CdTe, and GaAs.

9:40 am

In-Situ Contactless C-V and PL Monitoring of Free and Processed Semiconductor Surfaces in UHV-based Growth/Processing System: T. Yoshida, K. Ikeya, M. Mutoh, B. Adamowicz, T. Saitoh, H. Fujikura, T. Hashizume and H. Hasegawa, Research Center for Interface Quantum Electronics (RCIQE) and Graduate School of Electronics and Information Engineering, Hokkaido University, Sapporo, 060, Japan; T. Sakai, Dainippon Screen Mfg. Co., Ltd., Fushimi-Ku, Kyoto, 612, Japan

One of the promising approaches for fabrication of Gbit-scale silicon ULSIs and future compound semiconductor nano devices is use of an entirely UHV-based growth/processing system. Here, in-situ monitoring and control of growth and processing become vitally important. However, there has been no well established method for characterizing the electronic properties of "free" and "processed" semiconductor surfaces which are directly related to device performance. This paper describes two novel in-situ contactless monitoring techniques of free surfaces of semiconductors after various processings, developed for use in the UHV-based multi-chamber growth /fabrication facility (containing 15 chambers). One is a contactless capacitance-voltage (C-V) method, where the C-V measurement can be performed from the field electrode that is separated from the sample surface by a thin "UHV-gap" (300-400 nm). Thus, a MIS assessment of free surfaces or processed surfaces, such after MBE growth, plasma treatment, dry etching, etc., becomes possible by an "UHV-gap insulator". In this system, the "UHV-gap" is maintained to be constant (300-400 nm) by a piezo-mechanism with capacitance feedback from the three parallelism electrodes. This enables monitoring of surface/interface qualities as well as the profiles of shallow impurities and process-induced deep level defects. The other is an in-situ photoluminescence (PL) technique where excitation power dependences of PL intensity are measured and analyzed on computer to monitor qualities of surfaces /interfaces and near surface quantum structures. In this technique, the magnitude and slope of the PL efficiency can give the density and distribution shape of surface/interface states, respectively. The slope of unity corresponds to discrete states, while the state continuum gives a slope less than unity. Accurate state distributions can be determined quantitatively by fining the measured data to the calculated curves taking account of all possible recombination processes. The novel in-situ techniques were applied to various MBE-grown, thermally cleaned and UHV-processed surfaces of silicon and III-V semiconductors. It was found that there exist high-density surface states at the hydrogen-terminated Si surface and ultrathin-oxide covered Si surfaces formed by low-temperature thermal oxidation (300 - 400°C) in UHV. A well-behaved ultrathin-oxide/Si interface was formed by the low-temperature ECR N₂0-plasma oxynitridation process. It was also found that the ultrathin Si interlayer-based passivation process successfully reduces surface states on GaAs and AlGaAs surfaces and on near-surface quantum wells and wires.

10:00 am, Break

10:20 am

Real-Time Control of Group-V Flux in III-V Molecular-Beam Epitaxy Using a Valved Cracker: D.J. Friedman and A.E. Kibbler, National Renewable Energy Laboratory, 1617 Cole Blvd., Golden, CO 80401

For the molecular-beam epitaxy (MBE) growth of III-V materials, there are several cases in which precise control of the group-V flux can be important. For instance, dopant incorporation of certain amphoteric dopants can be quite sensitive to group-V pressure. Also, operation at low group-V overpressure can be used to maximize surface adatom mobility, or to reduce the load on the chamber Pumping system when growing phosphides; but at low group-V overpressures, even a small unintended dip in the overpressure can lead to complete breakdown of the surface morphology. In this paper, we demonstrate a method for achieving when the group-V source is a valved cracker. The method takes advantage of the following III-V MBE generalities: (1) the group-V flux is a significant fraction (i.e. 50%) of the beam equivalent pressure (BEP) during growth, even for low group-V overpressures; (2) the flux from a valved cracker can be changed very fast by adjusting the valve position - much faster than could be achieved by changing an oven temperature, and (3) the group-III fluxes, once allowed to stabilize, are highly stable over the course of a run. To achieve control of the group-V flux, we position the beam-flux monitor (an ion gauge) so as to intercept part of the beam, but not so far as to shadow the sample from the sources. A computer reads the BEP at frequent intervals (we use 1/sec),compares the reading to an operator-defined setpoint, and uses the difference in a proportional feedback scheme to change the valve setting to keep the measured BEP at the setpoint. Although the BEP is affected by both the group-III and group-V fluxes, the stability of the group-III's mean that it is effectively the group-V flux which is being controlled. The control achieved with this scheme is excellent. Without the control mechanism activated, the during-run stability and run-to-run repeatability we measure from our P and As group-V crackers is in the range of ± 10-20% (or worse), which is large enough to be of importance in the cases described above. In contrast, the control mechanism, once activated, gives short- and long-term BEP stability and reproducibility of better than ± 1%, and provides fast, highly-precise switching from one BEP setting to another. The use to which these capabilities can be put is illustrated by an study of the growth of Ga_0.5In_0.5P at very low group-V overpressure: the group V pressure is decreased in precise steps of 0.20x10^-6 torr while monitoring the surface morphology by measuring the intensity of laser light diffusely scattered from the surface. The surface breaks down as the BEP is stepped from 2.50x10^-6 to 2.30x10^-6, illustrating the sensitivity of the surface morphology to small changes in the BEP. The flux control makes this sort of study very easy to do; without the flux control, such a study would be difficult. Additional applications of the flux control, such as smoothing of the overshoot often observed on the opening of the valve, will be discussed. Other assumptions and necessary conditions for successful flux control will be described, and subtleties in transitioning between layers will be considered.

10:40 am, Student Paper

Endpoint Detection for Plasma Etching of Si_1-xGe_x: T.J. Knight, D.W. Greve, B.H. Krogh, Carnegie Mellon University, Department of Electrical and Computer Engineering, 5000 Forbes Avenue, Pittsburgh, PA 15213

The increasing interest in Si_1-xGe_x, as an enhancement to existing silicon-based bipolar transistor technology for high speed devices has created additional processing challenges. The variety of configurations which device engineers are pursuing would be greatly enhanced by a reliable mechanism to plasma etch silicon selectively to Si_1-xGe_x, and vice versa. Chemistries for etching Si_1-xGe_x, selectively to Si and vice-versa both exist, however they are only partially selective. The simplest method for controlling etching processes is well-characterized, carefully-timed, selective etch chemistries. In practice, however, decreasing device dimensions, typical process fluctuations, and etches which are not completely selective, make endpoint detection necessary. Optical emission spectroscopy is used as an established, unobtrusive endpoint detection technique for many etching processes. However, even this method is usually restricted to etching between different material systems such as metal on silicon. We present here for the first time a method for endpoint detection in the Si_1-xGe_x system. Our technique uses quadrupole mass spectrometry (QMS) to sense reactive etching byproducts. QMS sensing is able to sense a wide range of species simultaneously based on mass/charge. Our system uses a parallel-plate plasma etch chamber with RF power supplied to the top plate and samples resting on the bottom plate. Sulfur Hexafluoride (SF₆) gas is used which etches both Si and Si_1-xGe_x. The differentially-pumped quadrupole mass spectrometer (MKS #DPS300) is placed in direct line-of-sight of the plasma, allowing for clear detection of neutrals and radicals. The spectrum of signals we observe when etching Si_1-xGe_x includes SF_x source gas fragments and SiF_x and GeF_x reaction byproducts. These are clearly identifiable by comparison with their known mass and isotopic abundance. To demonstrate endpoint detection, a 3300 Å Si_0.8Ge_0.2 -on-silicon sample with an exposed area of 24 cm² was etched (RF power = 0.13 W/cm², pressure = 100 mTorr, SF₆ flow = 25 SCCM, T=40°C) while observing the SiF_x and GeF_x signals measured by the QMS sensor. At the beginning of the etch both signals became apparent. As the Si_1-xGe_x layer removal is completed, the GeF_x signals dropped and the SiF_x signals increased slightly, reflecting the higher silicon content of the substrate. There is a decay in the GeF_x signal with a time constant of about 30 seconds which may either be due to film nonuniformity or a Ge-rich etch byproduct on the wafer surface towards the end of the Si_1-xGe_x layer etch. These results indicate that QMS sensing is feasible for endpoint detection in the Si_1-xGe_x system. This method can be used in conjunction with the partially-selective etch recipes for optimal performance.

11:00 am

High Density Plasma Etching of GaAs: Correlation Between Surface Chemistry and Surface Damage: C.R. Eddy, JR., O.J. Glembocki, R.T. Holm, D. Leonhardt, V.A. Shamamian and J.E. Butler, U.S. Naval Research Laboratory, 4555 Overlook Ave., SW, Washington, DC 20375

Pattern transfer into compound semiconductors remains a critical processing step in the realization of numerous high speed and high temperature, electronic and optoelectronic devices. Due to the ever decreasing critical dimensions and damage tolerances of these and more conventional devices, pattern transfer by high density plasma processes has grown in popularity. Despite the critical importance of such processes to the future of compound semiconductor devices, they largely remain empirical in their development. In this work we make careful examinations into the variations in surface chemistry and correlate the results with damage imparted to the semiconductor. Through the combined observations we establish particular regions of process space that promote highly anisotropic, low damage pattern transfer. Experiments are performed in an electron cyclotron resonance microwave plasma source configured for reactive ion etching and using chlorine chemistry. Surface chemistry is monitored by sampling etch product species (AlCl_x and GaCl_x) through a rf biased, temperature controlled platen. Substrate temperature is monitored by optical bandgap thermometry, permitting strict control of changes in thermal surface chemistry as the plasma/surface environment is altered by varying the ion and neutral fluxes and the ion energy. Both neutral and ionized atomic chlorine fluxes to the substrate are deemed critical for maximum etch product formation rate. These fluxes are obtained at low microwave powers ( 300 W) and total pressures (0.5-1.0 mTorr). Etch anisotropy is largely controlled by ion-driven surface chemistry and is optimum for incident ion energies in the range of 50-200 eV. Surface damage assessments are made by in-situ and ex-situ photoreflectance (PR) spectroscopy. The use of both in-situ and ex-situ PR allows us to determine both the nature of the etched surface and its behavior when exposed to air. This is critical, because of the complex nature of oxides in III-V semiconductors and because their formation may depend on the physical and electronic properties of the surface. Significant surface damage is imparted for ion energies > 75 eV resulting in a shift of the surface Fermi level to the mid-gap. In-situ and ex-situ passivation techniques ate demonstrated to recover the surface Fermi level for ion energies up to 200-250 eV. Thus, low damage, anisotropic processing is achieved for ion energies between 50 and 200 eV.

11:20 am

Synthesis and Photoelectrochemical Investigation of II-IV-V₂ Semiconductors: Y.C. Wen and B.A. Parkinson, Department of Chemistry, Colorado State University, Fort Collins, CO 80523

The II-IV-V₂ chalcopyrite semiconductors are isoelectronic with III-V semiconductors and so applications associated with nonlinear optics and photovoltaic solar cells are possible. The naturally abundant group II (Zn, Cd) and group IV (Si, Ge, Sn) elements which compared to the relative scarcity and expensive of group III elements (In, Ga), make the II-IV-V₂ materials more attractive for a large scale applications. The use of photoelectrochemical techniques for the determination of the solid state properties of semiconductor materials has many advantages. Several electronic properties, such as bandgap, doping density, flatband potential, and current-voltage (I-V) characteristics can be obtained from these techniques. They are relatively simple and non-destructive and provide easy access to the Schottky-like junction allowing many different experiments to probe the interfacial junction chemistry on one sample. In addition, the energetics of the junction can be varied quite easily by appropriate choice of redox species. We have synthesized a variety of II-IV-V₂ single crystals including ZnSiAs₂, CdSiAs₂, CdSiP₂, and ZnSiP₂ by the chemical vapor transport (CVT) technique. We have characterized these crystals by photoelectrochemical methods. The doping levels from Mott-Schottky analysis were compared to those obtained from Hall measurements. The band positions, quantum yields for photocarrier collection and current voltage curves for different crystals were also determined in a variety of electrolytes. Photoluminescence has also been used for the study of various optical transitions in CdSiAs₂. Anisotropic optical behavior was observed for sulfur doped samples. A number of different surface treatments, which have been studied to gain insight into the surface of the semiconductor/electrolyte interface will be discussed

11:40 am

Non-Destructive Assessment of In_0.5(Ga_1-xAl_x)_0.5P Films Grown by Low Pressure MOCVD: Z.C. Feng, D. Collins, P. Zawadzki and R.A. Stall, EMCORE Corporation, 394 Elizabeth Avenue, Somerset, NJ 08873; J.Z. Wan, L. Malikova, F.H. Pollak, Physics Dept., NYS Center for Advanced Technology in Ultrafast Photonic Materials & Applications, Brooklyn College, Brooklyn, NY 11210

In_0.5(Ga_1-xAl_x)_0.5P, lattice-matched to a GaAs substrate, has a direct band-gap transition in the wavelength range between green and red and is very useful in optoelectronic applications such as visible light emitting diodes (LEDs) and laser diodes. During the epitaxial growth by metalorganic chemical vapor deposition (MOCVD), atomic ordering occurs under certain conditions, resulting in the reduction of the alloy bandgap and the photoluminescence (PL) emission peak energy. In order to control and optimize the growth conditions for the production of high quality InGaAlP epilayers, a variety of non-destructive techniques, including PL, Raman scattering (RS), photoreflectance (PR) and atomic force microscopy (AFM), have been applied to evaluate the epitaxial films. Two sets of In_0.5(Ga_1-xAl_x)_0.5P layers were grown on GaAs by low pressure (LP) MOCVD employing a high speed rotating disk reactor under different growth conditions. PL and PR spectra showed the variations of the InGaAlP PL peak and the bandgap energy with the growth pressure. While Raman measurements exhibited appropriate variation with composition for one set of samples doped with Si and Te, they showed no significant changes in the In and Al composition from another set of undoped films grown at different pressures and hydrogen carrier gas flows. Spectral line shape analysis gives information about the sample crystalline quality. AFM was used to study the surface morphology of these quaternary compounds and the dislocations were observed from the doped materials non-destructively. The combination of these four non-destructive techniques offers us a better understanding of the MOCVD-grown In_0.5(Ga_1-xAl_x)_0.5P/GaAs and a useful way to optimize the parameters for the growth of high crystalline quality epitaxial layers.

8:20 am, Invited

Growth and Electrical Transport Properties of Epitaxial Thin Films of Conductive Oxides Sr_1-xCa_xRuO₃: C.B. Eom, Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC 27708

For many electronic device applications, it is necessary to have epitaxial growth of conductive oxide thin films in a single heterostructure. We have grown epitaxial thin films of isotropic conductive oxides, Sr_1-xCa_xRuO₃ (01) in situ by 90° off-axis sputtering. These metallic oxides are pseudo-cubic perovskites, which could be ideal electrodes for ferroelectric and electro-optic devices. Single crystal epitaxial (110)° SrRuO₃ thin films were obtained on vicinal (001) SrTiO₃ substrates with a large miscut angle ( = 1.9°, 2.1° and 4.1°) and miscut direction close to the [010] axis. The films grown on vicinal substrates displayed a significant improvement in crystalline quality and in-plane epitaxial alignment as compared to the films grown on exact (001) SrTiO₃ substrates. Atomic force microscopy revealed that the growth mechanism changed from two dimensional nucleation to step flow growth as the miscut angle increased. The electrical transport properties of epitaxial thin films of the conductive oxides may be quite different from the corresponding bulk materials because of the existence of strain and cation disorder in the films. We have observed a strain stabilized metal-insulator transition in epitaxial CaRuO₃ thin films deposited on (001) SrTiO₃ and (001) LaAlO₃ substrates. X-ray diffraction studies showed that while semiconducting films with enlarged unit cells were obtained on single crystal (001) SrTiO₃ substrates, metallic films with lattice parameters close to the bulk material grew on (001) LaAlO₃ substrates. It is believed that a strain induced substitution of the small Ru⁴⁺ cations by the larger Ca²⁺ cations occurs, breaking the conduction pathway within the three dimensional network of the RuO₆ octahedra and leading to a metal-insulator transition. This unique phenomenon - which is not observed in bulk material - can be significant in technologically important epitaxial perovskite oxide heterostructures. This work was supported by the ONR Grant No. N00014-95-1-0513, NSF Grant No. DMR 9421947 and the NSF Young Investigator Award.

9:00 am

Copper Delafossites: P-type Conductive and Transparent Oxides: H. Kawazoe, H. Hyodo, M. Kurita, H. Hosono, Materials and Structures Laboratory, Tokyo Institute of Technology Nagatsuta, Midori-ku, Yokohama 226, Japan

A working hypothesis for finding a wide gap oxide with p-type and high electronic conductivity was proposed. Finding or chemical design of the p-type conductivity in wide gap oxides is a very difficult problem, because in most of oxides the valence band edge is exclusively composed of lone pair bands on oxide ions whose characteristic is a strong localization on a single oxygen. Migration of positive holes cannot be expected for the oxides, even if hole doping is successfully done. The essential characteristic in the hypothesis is how we can modulate structure of valence band edge by selecting an appropriate mother phase and doping method. As a major and essential chemical constituent of the candidate materials Cu⁺ or Ag⁺ were selected. Reasons of the selection are in the following: These ions have an electronic configuration of ns², where n stands for the principal quantum number, 4 for Cu and 5 for Ag, respectively. Energy of the occupied ns levels on the cations accidentally agrees with that of 2p⁶ electrons on the oxide ions. The occupied anti-bonding state between the ns levels on the cations and 2p levels on the oxide ions constitute the valence band edge. Significant covalency is expected between the cations and oxide ions, and positive holes introduced can migrate within the crystal. The semi-closed shell electronic configuration inhibits optical absorption bands in visible range. Secondly preferred condition about crystal structure is a tetrahedral coordination of oxide ions. In this structure valence state of oxide ions is roughly represented as sp³. The eight electrons around a single oxygen participate -type chemical bonds and no lone pair remains on the oxide ions. Two dimensional structure was preferred in the hypothesis, because three dimensional crosslinking of the ns² cations tends to a narrow band gap. These requirements led us to the delafossites of copper with chemical compositions of CuMO₂ (M=Al and Ga) as possible mother phases. Sintered disks of the delafossites were prepared by conventional ceramic processes. Polycrystalline thin films of the materials were deposited at 700°C on glass substrates or sapphire single crystals with c-orientation by sputtering or pulsed laser deposition. In the case of sputtering post annealing at higher temperatures was needed for crystallization. The thin films showed dc conductivities of 1 x 10^-1 S cm^-1, and p-type conductivity was confirmed by the measurements of Hall voltage and Seebeck coefficients. No intentional doping was done for the samples. Optical band gaps were estimated to be around 3.1 eV. On the basis of the measurements p-type and high conductivity of the transparent delafossites were experimentally realized.

9:20 am

Carrier Compensation in Ca_1-xY_xTiO₃ and CaTi_1-xNb_xO₃ Single Crystals: K. Ueda, H. Yanagi, H. Hosono, H. Kawazoe, Materials and Structures Laboratory, Tokyo Institute of Technology, 4259, Nagatsuta, Midori-ku, Yokohama 226, Japan

Electrical and optical properties of Ca_1-xY_xTiO₃ and CaTi_1-xNb_xO₃ (x=0,10^-4,10^-3, 10^-2) single crystals were investigated in order to understand electronic states of donors in a typical oxide semiconductor. The single crystals were grown by FZ method in O₂ flow using an infrared radiation furnace. The crystals obtained were annealed in Ar flow to avoid cracking. In both Y doped and Nb doped cases, the crystals of x=0, 10^-4 and 10^-3 were insulating and showed pale yellowish coloration. On the other hand, the crystal of x=10^-2 was conductive and showed blue coloration. After H₂ reduction at 1000°C, all of the doped samples (x=10^-4,10^-3,10^-2) became conductive, and metallic behavior was observed by the resistivity measurement at low temperature. Simultaneously with the remarkable increase in the conductivities, optical absorptions due to the conduction electrons appeared in infrared region. Carrier densities estimated by the Hall measurement at room temperature were 10^l3-10¹⁵ cm^-3 for x=10^-4, 5-6xl0^l8 cm^-3 for x=10^-3 and 2-3xl0^l9 cm^-3 for x=10^-2. Hall mobilities of the doped samples were found to be 2-4 cm²/Vs at room temperature. The conductivity measurement under air at high temperature found that the conductivity of the as-annealed crystals showed an unexpected dependence on x. The conductivity once decreased with an increase in the dopant concentration from x=0 to x=10^-3 and it remarkably increased with a further increase in the concentration to x=10^-2. In addition, p-type conduction in the sample of x=0 was observed above 600°C by the Seebeck measurement under air. These results suggest that the crystal with x=0 already contained some defects acting as acceptors and the introduced donors were compensated by the acceptors up to x=10^-3, and the compensation was removed by annealing in H₂ atmosphere.

9:40 am Late News

10:00 am, Break

10:20 am, Invited

Defects and Transport in Epitaxial, Rare-Earth-Doped SrTiO₃ and BaTiO₃ Thin Films: S. Gilbert, Texas Instruments, M/S 14, P.O. Box 655936, Dallas, TX 75265

Alkaline earth titanate thin films are promising candidates for device applications, including dynamic random access memories and nonvolatile memories. The defect structure of these perovskite oxides plays a central role in determining their electrical and optical properties. To optimize these properties, a fundamental understanding of the thin film point defect chemistry, deep level trapping states, and associated transport phenomena is required. Each issue plays an important role in a variety of long-term degradation mechanisms (i.e. fatigue, imprint, and resistance degradation) observed in these materials. As a model system for investigating the point defect structure and electronic transport properties of perovskite oxide thin films, epitaxial n-type BaTiO₃:La and SrTiO₃:Eu layers were deposited. The thin film electrical properties were tailored by doping in-situ with rare earth donors. Rare earth cations represent a widely-studied group of donors in the alkaline earth titanates. The films were deposited on (100) LaAlO₃ and 0.5 wt.% Nb-doped (100) SrTiO₃ by metalorganic chemical vapor deposition. The deep level electronic structure and carrier transport properties of the doped films will be described. Deep level analysis of epitaxial SrTiO₃:Eu was carried out using photocapacitance spectroscopy of Au/SrTiO₃:Eu/SrTiO₃:Nb Schottky barrier diodes. Photocapacitance spectra were measured at 90 K by focusing monochromatic light onto the semi-transparent Au Schottky contact while measuring the dependence of the steady state junction capacitance on the incident photon energy between 0.5 and 3.4 eV. Carrier transport in epitaxial BaTiO₃:La thin films on LaAlO₃ were examined using temperature dependent (77 to 300 K) dc, four-probe resistivity and slow ac thermoelectric power measurements. Based upon the results of these investigations, models are presented that describe both the point defect structure and carrier transport properties of rare earth-doped SrTiO₃ and BaTiO₃ thin films.

11:00 am, Student Paper

Hydrogen Complexes in Epitaxial BaTiO₃ Thin Films: G.-C. Yi, B.A. Block and B.W. Wessels, Department of Materials Science and Engineering and Materials Research Center, Northwestern University, Evanston, IL 60208

There has been considerable interest in high dielectric constant materials for advanced semiconductor device applications. Barium titanate (BaTiO₃) is a promising dielectric thin film with dielectric constants of the order of 300. The thin films, however, suffer from low resistivity presumably due to the presence of impurities. The nature and identity of these impurities in the films are not well understood. In the present study, hydrogen complexes in epitaxial BaTiO₃ thin films were investigated using Fourier transform infrared spectroscopy. Hydrogen is a well known impurity in bulk ferroelectrics, forming OH complexes. Both undoped and Er-doped layers were grown at 650-800°C using low pressure metal-organic chemical vapor deposition. From the infrared spectra of the films grown at 750-800°C, an absorption peak was observed at 3486 cm^-1. The absorption peak is attributed to one of the vibrational modes of O-H in BaTiO₃. Based upon the anharmonic diatomic oscillator model, the hydrogen in BaTiO₃ favors an interstitial site between the O-O edges of the oxygen octahedron with a 45° angle from c-axis. The OH concentration in BaTiO₃ thin films is estimated to be as high as 3x10¹⁹ cm^-3. Films grown at a low temperature between 650 and 700°C did not show the absorption peak. The presence of C-H complexes in the MOCVD-grown layers was also investigated. Er-donor doped layers showed additional absorption peaks at 2905 and 2964 cm^-l which are ascribed to the vibrational modes of C-H complexes in the Er-doped layers. The C-H complex concentration increased for highly Er-doped layers with an Er concentration higher than 10²⁰ cm^-3. The role of hydrogen on the dielectric properties of BaTiO₃ will be discussed.

11:20 am

Microwave Properties of Ferroelectric Thin Films: S.W. Kirchoefer, J.M. Pond, J.S. Horwitz, A.C. Carter and D.B. Chrisey, Code 6851, Naval Research Laboratory, Washington, DC 20375-5347

The development of new technologies for the deposition of thin dielectric films has resulted in renewed interest in ferroelectric materials for microwave device and circuit applications. The properties of bulk ferroelectric materials have been known for decades. By utilizing the technology of pulsed laser deposition (PLD), it is possible to deposit thin films of these materials with electronic and physical properties that are more compatible with present microwave technology than bulk materials. Employing a target containing the stoichiometric ratios of SrTiO₃ and BaTiO₃ desired, thin films (~0.5 mm thick) are deposited on MgO, LaAlO₃ and SrTiO₃ substrates. Post-deposition annealing has previously been shown to greatly improve the microwave properties of these films. Interdigital capacitors, consisting of 1-micron-thick silver electrodes, are deposited via standard photolithography and metal-liftoff patterning. The interdigital electrodes are configured to facilitate microwave probing of the device using a 200-m-pitch Picoprobe connected to an HP 8510 vector network analyzer. Microwave reflection measurements are employed to characterize the ferroelectric film properties in the frequency range from 50 MHz to 20 GHz under a variety of dc electric field biases. We have measured bias-dependent capacitance tuning greater than 4:1 in some films. We have also observed films with losses at 10 GHz that are comparable to those of state-of-the-art semiconductor varactor devices. The features of the observed data that originate from the electronic properties of the thin ferroelectric films have been isolated and quantified. Several techniques have been employed to analyze the measured data which allow the dielectric losses to be separated from the metal-electrode losses. In all of the high quality films examined in detail, it has been found that the dielectric loss tangent can be described as the sum of a frequency independent term and a term which is proportional to frequency. The frequency dependent term is found to be maximum in the absence of a dc bias field. Studies are in progress to understand the correlation of these microwave losses with film composition, deposition conditions, post-annealing profiles, impurities, grain size, and other microscopic film properties.