TMS Logo

1997 EMC: Thursday Afternoon Sessions, Part 1

June 25-27, 1997 · 39TH ELECTRONIC MATERIALS CONFERENCE · Fort Collins, Colorado

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


CHAIR: Kang L. Wang, 66-147 Engineering IV, UCLA, Los Angeles, CA 90095-1594
CO-CHAIR: Clivia Sotomayer-Torres, Bergische Univestät, HG Wuppertal, Germany
ROOM: Theater

1:30 pm

Giant Optical Anisotropy in Semiconductor Heterostructures with No-Common Atom: O. Krebs and P. Voisin, Laboratoire de Physique de la Matière Condensée de l'Ecole Normale Supérieure, 24 rue Lhomond, F75005, Paris, France

The no-common atom systems Ml-M2 with generic composition Ml=ClA1, M2=C2A2 (where C stands for the Cation and A for the Anion species) are fundamentally different from those where both materials share a common atom, in general the anion like in GaAs-(AlGa)As. Indeed, in these no-common atom systems, the interfaces involve chemical bonds which differ at the Ml-M2 and M2-Ml interfaces, and differ from the chemical bonds in either material. A first consequence is that the interface dipole contributions to the band offset are not the same, hence the band offsets are not commutative. Band offset asymmetries in the 70-100 meV range have indeed been measured in the InGaAs-InP and InP-AlInAs systems, in close agreement with the theoretical predictions of Priester and co-workers. In addition, at the M1-M2 interface, all the Cl-A1 bonds lie in the {110} plane while all the A1-C2 bonds are in the perpendicular {-110} plane. This local asymmetry is not compensated at the other interface, which allows some optical anisotropy around the growth axis. From polarization-resolved optical transmission measurements in both systems, we have indeed observed a considerable optical anisotropy (linear polarization rate up to 40%!) in the near band-gap spectral region, while the common atom InGaAs-AlInAs heterostructures show perfect isotropy. A remarkable aspect of these effects is that they are clearly allowed by the group-theoretical analysis, but strictly forbidden in the classical envelope-function theory which is almost universally used for calculations of quantum well properties. In the last months, three different theoretical approaches have clarified this intriguing situation: refined sp3s* tight binding calculations by Bertho et al. have established the intrinsic character of these large effects, while Ivchenko et al. have shown that envelope-function boundary conditions should be generalized and produce a zone-center mixing of the heavy and light holes; independently, Krebs et al. proposed an heuristic formulation of a k.p hamiltonian taking local symmetries into consideration and leading to a convenient perturbation treatment of these effects. The three approaches give equally satisfying descriptions of the experimental facts. The optical anisotropy results from a mixing of the heavy and light holes at the mini-zone center due to the breakdown of the roto-inversion symmetry of the zinc-blend structure. The effect is concentrated in the spectral region between the H1-E1 and L1-E1 optical transitions. The asymmetry and resulting optical anisotropy can be further controlled by an electric field applied along the growth axis, giving rise to a new electro-optical property, the << Quantum Confined Pockels Effect >>.

1:50 pm, Student Paper

Carrier Capture Times In Low-Dimensional Semiconductor Lasers Characterized by Dynamic Lasing Emission Measurements: J. Wang, U.A. Griesinger and H. Schweizer, Physikalisches Institut, Universität Stuttgart, Pfaffenwaldring 57, D-70550 Stuttgart, Germany

We report experimental and theoretical studies of carrier capture times in low-dimensional semiconductor distributed feedback lasers with three different typical active region structures: quantum-well (QW), wire and box. The wire and box lasers are pre-pared by deep dry and wet etching method starting from the QW structure. The total effective carrier capture times of the lasers at low temperature are directly determined by means of the dynamic lasing emission measurements. By systematical comparison of QW, wire and box lasers which have quite different packing densities of the active regions but approximately equal carrier diffusion length before capture, we evidence a strong dependence of the total effective carrier capture time on the packing density of the active region, which indicates the significant contribution of the quantum capture time to the total effective carrier capture time. Moreover, the intrisinc local carrier quantum capture time can be deduced from this kind of study. The determined local quantum carrier capture time for the In-GaAs/InGaAsP QW laser is about 3ps at 2K, which is well consistent with a performed detailed quantum mechanics calculation. Furthermore. by comparison of box lasers with the approximate same box size but different box densities at 2K and 77K, we find that the determined quantum capture time of the box lasers is only about 2.4ps. We believe that this is a direct experimental indication of existence of an efficient channel for carrier capture and relaxation in the investigated quantum-box system. The systematic comparison of the QW, wire and dot lasers reveal the dominant limitation of geometry effect on the high speed modulation of quantum wire and dot lasers except when the quantum wire and dot are packed with a high density.

2:10 pm

Selective Intermixing of AlxGa1-xAs and (AlxGa1-x)0.5In0.5P Quantum Well Structures: D. Sun, K. Beernink and D.E. Treat, Xerox Palo Alto Research Center, 3333 Coyote Hill Road, Palo Alto, CA 94304

Intermixing of group III atoms in semiconductor quantum well (QW) heterostructures has attracted considerable interest for the fabrication of optoelectronic devices, since it allows modification of the potential profiles of the QW after the material growth. One technique commonly used for enhancing the interdiffusion of Al and Ga is the use of a SiO2 cap for vacancy-enhanced layer intermixing. We find that the interdiffusion rate of Al-Ga in a GaAs/Al0.4Gao6As QW is much larger compared to a GaxIn1-xP/(AlyGa1-y)0.51n0.5P QW grown together in a stacked QW heterostructure under SiO2. It is possible to enhance the intermixing rate of GaAs/AlGaAs QW selectively using impurities and high Al content potential barrier layers, without intermixing the GaxIn1-xP/(AlyGa1-y)0.5In0.5QW. Materials of dual (840 nm) GaAs/AlGaAs and (670 nm) GaxIn1-xP/(AlyGa1-y)0.5In0.5P QWs stacked in 100 nm separation sandwiched by Al0.5In0.5P cladding layers are grown on GaAs substrates. The samples are patterened with SiO2 stripes for thermal anneal at a temperature range of 840 to 1065°C. Photoluminescence is used to measure the shift of the transition wavelengths from the dual QWs after therrnal anneal. We find that the interdiffusion coeffcient of Al-Ga in the GaAs/AlGaAs QW is two orders of magnitude larger than that of Al-Ga in the GaxIn1-x P/(AlyGa1-y)0.5In0.5P QW under SiO2 at 1000°C. Utilizing the differential intermixing rates of GaAs/AlGaAs and GaxIn1-xP/(AlyGa1-y)0.5In0.5P QWs, it is possible to selectively intermix the GaAs/AlGaAs QW and shift its transition wavelength to the blue side of that of the GaInP QW. By adding AlxGa1-xAs (x>0.8) layers and/or Si impurities to the GaAs QW confinement region surrounding the GaAs QW, the intermixing rate is further increased; the transition wavelength is shifted to 640 nm so that the GaInP QW has the minimum energy states. This technique is promising for fabrication of monolithic side by side multiwavelength lasers.

2:30 pm

Valence-Bank Splittings of Ordered GaInP Measured by Absorption Bleaching: B. Fluege., Y. Zhang, H.M. Cheong, A. Mascarenhas, J.F. Geisz, J.M. Olson and A. Duda, National Renewable Energy Laboratory, Golden, CO 80401

Using time-resolved exciton absoption bleaching, we have measured the optical transition energies from all three valence bands in spontaneously-ordered GaInP. The improved accuracy of the new method allows simultaneous determination of the crystal-field splitting and spin-orbit splitting parameters. Eight Ga0.52In0.48P samples with varying degree of ordering were grown by atmospheric-pressure organometallic vapor phase epitaxy, and then removed from their substrates. Broadband linear absorption and differential absorption (DA) were measured at 5 K. Linear absorption shows results similar to photoluminescence excitation measurements: heavy-hole-like and light-hole-like exciton absorption peaks are only partially resolved from the continua and from each other. In contrast, the DA is a direct measurement of the near-band-edge absorption changes 20 ps after femtosecond optical excitation into the continuum. DA shows sharp bleaching features at each exciton, including the spin-orbit peak, with very little contribution from continuum states. From these spectra, the bandgap reduction, crystal-field splitting and spin-orbit splitting parameters were independently extracted for each sample. The spin-orbit splitting parameter is nearly constant with ordering, at 103 meV. The ratio of bandgap reduction to crystal field splitting parameter is slightly higher than previous works.

2:50 pm, Student Paper

High Efficiency Energy Up-Conversion Induced by Carrier Localization in Ordered (Al0.5Ga0.5)0.5In0.5P/GaAs Heterointerfaces: K. Yamashita, T. Kita and T. Nishino, The Graduate School of Science and Technology and Faculty of Engineering, Kobe University, Rokkodai 1-1, Nada, Kobe 657, Japan

We observed strong up-converted photoluminescence (UPL) enhanced at (Al0.5Ga0.5)0.5In0.5P/GaAs heterointerfaces. It was found that a potential fluctuation in the (Al0.5Ga 0.5)0.5In0.5P layer causes the strong UPL. UPL spectra of the (Al0.5Ga0.5)0.5In0.5P on (a) GaAs(115)A and (b) GaAs(001) were observed at 1.91 eV excitation. A strong ordering of the column-III sublattice appears on GaAs(001). A reduction of the band-gap energy, which comes from band-folding phenomena in the ordering material, is observed. The UPL peaks can be explained by an interface induced Auger process; (1) the energy released by carrier recombinations in the GaAs (narrow energy-gap material) pumps carriers into the (Al0.5Ga0.5)0.5In0.5P (wide energy-gap material). (2) The excited carriers are prevented to feed back to the GaAs layer by the carrier localization in a fluctuated potential of the (Al0.5Ga0.5)0.5In0.5P. Excitation-power dependencies of the UPL intensities for both samples show a superlinear response with an exponent of 1.31. However, the UPL intensity for the (Al0.5Ga0.5)0.5In0.5P/GaAs(ll5)A deviates from the superlinear relationship above about 2.0 W/cm2. The potential fluctuation, which is caused by a statistical distribution of order parameter in the epitaxial alloy, is large for a sample showing the strong ordering. Since the localized states of the (Al0.5Ga0.5)0.5ln0.5P /GaAs(001) efficiently confine carriers, radiative recombinations are enhanced.

3:10 pm, Break

3:30 pm

Microstructure of Double-Variant Ordered GaInP Films: S.P. Ahrenkiel, Y. Zhang, A. Macarenhas, D.J. Friedman and J.M. Olson, National Renewable Energy Laboratory, 1617 Cole Blvd., Golden, CO 80401

Spontaneous atomic ordering in semiconductor alloys often results in material with modified optoelectronic properties with respect to random alloys. We discuss the microstructure and evolution of CuPt-B ordered domains in Gal-xInxP (GaInP) films with (x 0.5) grown on vicinal GaAs substrates by MOCVD and examined by TEM. Double-variant structure is observed at all stages of growth on exact [001] substrates and those miscut towards <111>-A, while small miscuts towards <111>-B or <110> produce a difference in ordering strength of the two variants. Double-variant samples exhibit a fine, interwoven structure at early stages of growth that consists of alternating lamellae of closely associated microdomains lying nearly parallel to the substrate plane with an average vertical period of roughly 40-50-A. Weak lateral variant segregation is detectable at relatively early stages of growth, and large, laterally segregated superdomains form beyond two microns thickness that contain numerous internal structural imperfections attributed to antiphase boundary formation as in single-variant samples. Thick, double-variant samples show increased surface topography associated with the superdomain formation. We also detect narrow contrast bands Iying in the growth plane that correspond to source flux variations during growth. Polarized photoluminescence (PPL) was used to reveal optical anisotropy associated with double-variant ordering. These studies showed enhanced I[]/I[110]PPL ratios from thin, double-variant samples compared to single-variant and thick, double-variant samples. The increased anisotropy is attributed to quantum effects associated with the fine interweave of ordered microdomains at early growth stages that resembles an orientational superlattice.

3:50 pm, Student Paper

Determination of Ordering Induced Birefringence in (Al)GaInP: R. Wirth, C. Geng, F. Scholz and A. Hangleiter, Physikalisches Institut, Universität Stuttgart, Pfaffenwaldring 57, D-70550 Stuttgart, Germany

Ternary and quaternary semiconductors under certain growth-conditions in metalorganic vapor phase epitaxy exhibit a chemical ordering of their group III components in {111}B planes. There has been great interest in the new properties, resulting from the reduced symmetry of this systems, in particular in the band-gap reduction and splitting of the valence bands. We used the properties of the modes in ordered planar waveguides to determine the ordering induced birefringence. To restrict the growth of the ordering planes to one of the variants, GaAs substrates misoriented 6o from (100) towards (1-11) have been used. The eigenmodes of such a waveguide structure are not the normal transverse electric (TE) or transverse magnetic (TM) modes. The properties of the eigenmodes are determined by a competition between the anisotropy of the dielectric tensor and the boundary conditions of the waveguide, resulting in modes tilted with respect to TE/TM. In combination with a transfer-matrix method for anisotropic waveguides it is possible to determine the anisotropy of the dielectric tensor, characterized by the difference of the ordinary and extraordinar index of refraction n = ne-no. We find ordered GaInP, (Al.33Ga)InP, and (Al.66Ga)InP to be positively birefringent far below the band-gap and becoming negatively birefringent at the band-gap. The dispersion of the birefringence emerging from the highly anisotropic interband matrix element is well understood using a 6-band k.p approach. Far below the band-gap we find anisotropies up to n = 0.008, at the band-gap we find a birefringence as strong as n = -0.02. At E = 1.61eV and E = l.91eV the ordered GaInP and (Al.33Ga)InP crystals are isotropic, respectively. A detailed understanding of the birefringence in ordered (Al)GaInP leads to applications, such as lasers with tilted polarization output or tunable TE/TM-polarization converters.

4:10 pm

Polarized Low-Temperature Cleaved-Edge Photoluminescence Study of Spontaneously-Ordered GaInP2: H.M. Cheong, Y. Zhang, A. Mascarenhas, J. Geisz and J.M. Olson, National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401

GaInP2 alloys grown by MOCVD on (001) GaAs substrates exhibit a spontaneous CuPt-type ordering of various degrees along the [] or [] directions, depending on the growth conditions and the substrate misorientation. These structures resemble monolayer superlattices of Ga(l+)In(1-)P2/Ga(1-)In(1+)P2 (01) along the ordering direction. For a [] single-variant material, the symmetry of the crystal changes from the Td symmetry of zinc-blende to trigonal C3v symmetry, with the ordering direction z'=[] as the high symmetry axis. One of the major consequences of this ordering is the polarization of the luminescence: the lowest-energy band-to-band transition is dipole-forbidden for the polarization of the photons parallel to the z' axis. However, in the case of excitonic transitions, since the coulomb interaction relaxes this selection rule, the z'-polarized transition becomes partially allowed, and a finite luminescence intensity is expected. We present results of the first low-temperature photoluminescence (PL) study of 10-µm thick GaInP2 samples in the backscattering geometry on the (110) cleaved edge. The samples were mounted in a liquid helium cryostat, and a reflective microscope objective with a numerical aperture of 0.5 was used to focus the 514.5-nm excitation laser, and also to collect the luminescence. The size of the focus spot was about 5 µm or smaller. The PL spectra for the polarizations parallel and perpendicular to the z' direction were compared at 4.2 K. For an ordered sample, we observed a finite intensity of the excitonic transition peak in the PL spectrum for the polarization direction parallel to the z' direction, as expected. We also observed that the excitonic peak is stronger by a factor of ~16 in the polarization perpendicular to the z' direction than in the z' polarization. A comparison of the experimental results with calculations of the exciton transition intensity based on the kp model, as well as the temperature dependence of the cleaved-edge PL polarization will be presented.

4:30 pm,Student Paper

Characterization of Unicompositional GaInP2 Ordering Heterostructures: C.E. Inglefield, M.C. DeLong and P.C. Taylor, Department of Physics; Y.S. Chun, I.H. Ho and G.B. Stringfellow, Department of Materials Science and Engineering, University of Utah, Salt Lake City, UT 84112; J.H. Kim and T-Y. Seong, Department of Materials Science and Engineering, Kwangju Institute of Science and Technology, Kwangju 506-303, Korea

When GaInP2 is grown by organometallic vapor phase epitaxy(OMVPE), there is a tendency for the cations to be segregated into {111} planes leading to CuPt-type ordering. The degree to which this ordering takes place may be controlled with growth conditions. Ordering affects the material properties, in particular causing a reduction in the bandgap. This ability to control the bandgap without changing the composition of the alloy is potentially very useful in eliminating the problem of lattice matching the layers in epitaxial heterostructures. Two growth parameters which can be used to control the degree of order in a layer are the growth temperature and the V/III ratio. For technical reasons the V/III ratio is the more attractive and is the focus of this work. We will show that for growth based on a change in V/III ratio the unusual optical properties of GaInP2 make it impossible to characterize the resulting heterostructures with photoluminescence (PL) alone. A combination of transmission electron diffraction and microscopy (TED and TEM), PL, and photoluminescence excitation (PLE) is needed to effectively characterize the heterostructures. Using these tools, we have demonstrated the ability to grow high quality, abrupt interfaces in a heterostructure based on a change in V/III ratio. Evidence for carrier transport from the higher bandgap layer into the lower bandgap layer was observed. We will report quantitative measurements of the bandgap differences between two layers of single heterostructures. In addition, we have determined the degree of order in one of the layers from the valence band splitting measured with polarized PLE.

4:50 pm

Electroreflectance Measurements of Intrinsic Electric Fields in Spontaneously-Ordered GaInP2: J.D. Perkins, H.M. Cheong, Y. Zhang, A. Mascarenhas, J.F. Geisz, W.E. McMahon and J.M. Olson, National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401

GaInP2 alloys grown by MOCVD on (001) GaAs substrates exhibit a spontaneous CuPt-type ordering of various degrees along the [] or [] directions, depending on the growth conditions and the substrate misorientation. These structures resemble monolayer superlattices of Ga(0.5+/2)In(0.5-/2)P/Ga(0.5-/2)In(0.5+/2)P (01) along the ordering direction. Recent calculations predict an intrinsic electric field of order 1600 kV/cm for fully ordered (=1 ) GaInP2 and of order 400 kV/cm for the partially ordered (=0.5) grown by MOCVD. Photolurninescence experiments measuring the Stark shift of the exciton deduce an intrinsic field of 320 kV/cm in partially ordered GaInP2. For such strong internal electric fields, clear Franz-Keldysh Oscillations (FKO) would be expected in an electroreflectance measurement. We report electroreflectance measurements of GaInP2 layers embedded in GaAs. A typical device structure is (substrate)/(p-GaAs)/(GaInP2)/(p-GaAs) with the 400 Å top layer acting as a partially transparent conducting front contact. For a 1000 Å thick GaInP2 layer, electroreflectance spectra measured at T = 100 K with no extemal bias do not show FKO. When a positive or negative DC bias voltage of ~1 volt is applied, clear FKO are observed. At the lowest bias voltages for which FKO are seen, the electric field in the GaInP2 layer, as calculated from the FKO, is ~75 KV/cm. Since no FKO are seen with no extemal bias applied, we conclude that, in our samples, any intrinsic electric field in the GaInP2 layer is less than 75 kV/cm.


CHAIR: Parvez Uppal, Sanders, Lockheed Martin Co., MA 6-1551, 65 Spit Brook Road, Nashua, NH 03061
CO-CHAIR: Ralph Dawson, MS 0601, Sandia National Laboratories, Albuquerque, NM 87185-0601
ROOM: Senate

1:30 pm

Low Temperature Amorphous Silicon Films and Solar Cells from Dichlorosilane/Silane Source Gas Mixtures: B. Crone, Princeton University, Department of Electrical Engineering, MST-11, Mail Stop D429, Los Alamos National Laboratory, Los Alamos, NM 87545; S. Wagner, Princeton University, Department of Electrical Engineering, Princeton, NJ 08544

We present results from experiments to lower the deposition temperature of amorphous silicon by (DCS) / silane source gas mixtures. It has been shown that DCS / silane source gas mixtures have growth rates as high as 50 Å/sec , ~10 times higher than those from pure silane source gases (M. Nakata, S. Wagner, APL, 65, 1994, pp. 1940-42). In this work we have investigated using DCS to achieve lower substrate temperatures, rather than the high growth rates previously found. Films and solar cells were grown in our three chamber plasma enhanced chemical vapor deposition chamber using DC electrical plasma excitation. Films were grown at 300, 250, 200, 150, and 100°C substrate temperatures, with and without DCS in the source gas mixture, and with different amounts of hydrogen dilution. They were analyzed using transmission, photothermal deflection constant photocurrent method, photoconductivity, and dark conductivity. For selected films, solar cells were grown of the structure p+ (10nm) / film i-layer ( 150nm) / silane i-layer (50nm)/n+ (20nm). The cells were characterized by light current voltage measurements, and by quantum efficiency measurements. The dependence of the defect density on deposition temperature fell into two regimes. The films grown from pure silane, and those gown from DCS / silane mixtures with high hydrogen dilution, showed defect densities increases from -2x1015 cm-3 to -2x1017 cm-3 as temperature was lowered from 300 to 100°C. Those films grown with DCS / silane mixtures with no or moderate hydrogen dilution had defect densities of ~2x1016 cm-3 at 300°C, but the defect densities did not increase until the deposition temperature was lowered below 150°C. Below 150°C all films had similar defect densities. The cells all had low efficiencies, except the 300°C pure silane cell. This is largely due to the low photoconductivities of the other films, the 300°C pure silane films photoconductivity was 1x10-4 S/cm, more than 10 times higher than that of any other film. Addition of DCS to silane as a source gas considerably changes the growth mechanism. This leads to growth regimes where film defect density is independent of substrate temperature. Films grown to date with DCS / silane mixtures have high defect densities and low photoconductivities.

1:50 pm, Student Paper

Near-field Scanning Optical Microscopy Studies of Cu(In,Ga)Se2 Solar Cells: A.A. Mcdaniel and J.W.P. Hsu, Physics Department, University of Virginia, Charlottesville, VA 22901; A.M. Gabor, Energy Photovoltaics, Princeton, NJ 08543, Current: Evergreen Solar, Inc., 211 Second Avenue, Waltham, MA 02154

Thin film devices are among the most promising candidates for affordable Solar cells. In order to understand what is necessary to reliably produce good cells, devices must first be illy characterized. Using near-field scouring optical microscopy (NSOM), we Study the spatial variations in photoresponse of two Cu(In.Ga)Se2 (CIGS) solar cells. The cells are imaged on both the surface and the cross-section to examine how the microstructure of the sample affects photoresponse and the quality ofthe p-n,junction. In the cells that we Studied, the -2.7 µm p-type CIGS film was grown on Mo coated Soda-lime glass. A ~500 Å n-type layer of CdS and then a -4500 Å n-type layer of ZnO were deposited on top of the CIGS to form the p-n junction. The efficiencies of these cells studied are <9% and >14%. Topographic images of the low efficiency cell surface reveal two feature sizes: Small (~1 µm) particles which form large (~10 µm) grains with deep crevices in between. Photovoltage images show drastic reduction in photoresponse associated with most crevices. The high efficiency cell shows no large grain stmcture or deep crevices. The photoresponse is more uniform, with small fluctuations which correlate with, but are not identical to, the topography. Surface images of both samples reveal grain to grain variation in response. We also see areas on both samples which show reduction in the photovoltage image with no corresponding topographic changes. Some of these features are as small as ~250 µm, making them extremely difficult to detect using any other optical technique. Cross-section images enable us to locate the p-n junction. We find that the depth of the junction varies, though the average depth roughly corresponds to the nominal n-CdS/p-CIGS interface. We also see nonuniformities in the response along the junction on a length scale of -500 µm. Supported by NSF.

2:10 pm

Nanoparticles as Spray Deposition Precursors to Cu(In,Ga)Se2 Solar Cells: Solution Synthesis of Cu-In-Ga-Se Colloids, Thermal Processing of Colloid-Derived Films, and Device Performance: D.L. Schulz, C.J. Curtis, H. Wiesner, R. Noufi, D.S. Ginley, Center for Photvoltaic and Electronic Materials and Center for Basic Sciences, National Renewable Energy Laboratory, 1617 Cole Blvd., Golden, CO 80401-3393

The use of nanopartide colloids for spray deposition of Cu-In-Ga-Se precursor films and subsequent fabrication of Cu(ln,Ga)Se2 (CIGS) solar cells has been investigated. Nanoparticle colloids might offer advantages over conventional spray deposition precursors owing to reduced particle sizes in the former (e.g., 10 to 1000 Å) versus the latter (e.g., 1 to 5 µm). This approach has the potential to yield films similar in quality to those obtained by vacuum techniques but without the expense and complexity. In the present study, metathesis reaction between Na2Se in methanol and the metal iodides (i.e., CuI, InI3, and Gal3) in pyridine produced amorphous Cu-Se (Eq. 1), In-Ga-Se (Eq. 2), or Cu-In-Ga-Se (Eq. 3) nanoparticle colloids. Purified colloid was sprayed onto heated molybdenum-coated sodalime glass substrates to form Cu-Se, In-Ga-Se, and Cu-In-Ga-Se precursor films. After thermal processing of these films under a Selenium ambient and evaporation of additional metallic reagents, CIGS Solar cells were fabricated. The Cu-In-Ga-Se colloid and precursor films were characterized hy transmission and Scanning electron microscopy, thermal gravimetric analysis, inductively coupled plasma atomic emission spectroscopy, and atomic force microscopy. It was determined that the chemical composition of the porous Cu-In-Ga-Se precursor films was nearly identical to that of the precursor colloid. Standard I-V characterization was performed on the CIGS Solar cells with the best devices to date exhibiting solar conversion efficiencies bom 4 to 5 %. 2CuI+Na2Sepyridine/methanol, -2Nal Cu-Se 0.72InI3 + 0.28Gal3 +1.5Na2Sepyridine/methanol, -3NalIn-Ga-Se 1.10CuI+ 0.68lnI3 + 0.23GaI3 +1.91Na2Sepyridine/methanol, -382NalCu-In-Ga-Se

2:30 pm, Student Paper

Melt Growth and Characterization of Bulk Crystals in the System Cu-In-Ga-Se: C. Beilharz and K.W. Benz, Kristallographishes Institut der Albert-Ludwigs-Universität Freiburg, Hebelstr. 25, 79104 Freiburg, Germany; I. Dirnstorfer and B.K. Meyer, I. Physikalisches Institut de Justus-Liebig-Universität, Henrich-Buff-Ring 16, 35392 Gießen, Germany

Copper Indium Selenide (CIS) is one of the most promising materials for thin film solar cells. Solid solutions containing Ga and S besides Cu, In and Se may lead to an even higher efficiency of the energy conversion. Investigations of bulk material can contribute to a better understanding of some properties of the thin film samples. The investigated compositions were chosen from the In2Se2-rich side of the quasibinary section Cu2Se-ln2Se3 in the system Cu-In-Se close to the congruent melting point of CIS, where 10 and 20 atom%, respectively, of the Indium were replaced by Gallium. These values are relevant far Copper Indium Gallium Selenide (GIGS) solar cells. The material for the growth experiments was synthesized from the elements in a fused quartz ampoule. Bulk crystals with cylindrical shape were grown in the synthesis ampoules using the Vertical Gradient Freeze method. All crystals are p-type. They consist of a few grains and are twinned, usually with a certain orientation to the growth direction. The grains have a size of several cubic centimeters. The composition along the growth direction was determined with EDX in a Scanning Electron Microscope. A bole with an initial content of 55 mol% (ln0.9Ga0.1)2Se3 showed a significant variation of the composition whereas the crystals equivalent to 52 mol% In2Se3 were to be found more homogeneous over the whole length. Mechano-chemical polished samples were characterized using Interference Contrast Microscopy and Scanning Electron Microscopy. In the crystals with 55 mol% (ln0.9Ga0.1)2Se3 as well as in that With 52 mol% (ln0.8Ga0.2)2Se3 a micro-sized lamellary network of two phases, similar to that in the system Cu-In-Se, was found. Additional Photoluminescence (PL) studies were carried out using the 488 nm line of a Ar-ion laser at 4 K. Samples prepared from a crystal with 55 mol% (ln0.9Ga0.1)2Se3 show three transition peaks around 0.86, 1.0 and 1.1 eV. All of them shift to higher energies With increasing (Ga+In)/Cu-content. In the case of 52 mol% (ln0.9Ga0.1)2Se3 initial content of the crystals a strong PL peak is found at approximately 0.9 eV. Additional, this material shows a weak peak at 1.1 eV. That indicates the existence of a trace of a second phase in the samples. For GIGS crystals With a composition close to the stoichiometric one density measurements with the buoyancy method gave a value of 5.57 g/cm3.

2:50 pm

Unusually Long Bulk Lifetime and Low Surface Recombination in CdTe Single Crystals: R. Cohen, Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot 76100, Israel; B. Mishory and Y. Rosenwaks, Department of Physical Electronics, Faculty of Engineering, Tel-Aviv University, Ramat-Aviv 69978, Israel

Minority carrier lifetime and surface recombination of the excess photogenerated carriers is of peak importance for the performance of solar cells. Thus it is essential to measure and understand the recombination mechanisms in solar cells materials in order to improve photovoltaic efficiencies. In this work we present a series of time-resolved photoluminescence (TRPL) and surface photovoltage (SPV) measurements which demonstrate that n-doped CdTe single crystals posses extremely long bulk lifetime and very low surface recombination velocities (SRV). The commercial p-CdTe single crystals were doped in a novel procedure that involved annealing of the CdTe crystals in a capsule containing a mixture of CdTe and lnxTey. powders. The TRPL measurements showed that the SRV of etched low doped (n=1.5 x 1016 cm-3 ) samples was below 500 cm/s and the effective bulk recombination time was 160 nanoseconds. By conducting the TRPL measurements under different injection levels and at different temperatures, it was found that the radiative bulk recombination was the dominant mechanism in the low doped crystals. The unusually low SRV values were confirmed also by SPV and surface photovoltage spectroscopy measurements.


CHAIR: Carol Ashby, Sandia National Lab., P.O. Box 5800, Albuquerque, NM 87185-0603
CO-CHAIR: Russell Dupuis, Microelectronics Research Center, University of Texas, Austin, TX 78712-1100
ROOM: Cherokee

1:30 pm, Student Paper

Defect Generation and Diffusion in Wet-Oxidized AlAs: Song S. Shi, Evelyn L. Hu, Center for Quantized Electronic Structures (QUEST), University of California, Santa Barbara, CA 93106

Wet thermal oxidation of Al(Ga)As has found numerous applications in electronic/optoelectronic device fabrication, such as the formation of current apertures in vertical cavity surface emitting lasers (VCSELs) and gate oxides in GaAs-based metal-oxide semiconductor field effect transistor. The quality of this oxide and the oxide-semiconductor interface is thus extremely important in determining the performance of these devices. We had earlier found that the luminescence of quantum well placed in close proximity to the interface between the formed oxide and the adjacent semiconductor serves as a sensitive probe of the quality of that interface. We utilize that technique here to further explore the consequences on the interface quality of the oxidation temperature. The samples used here consist of a 70 Å GaAs quantum well (QW) separated from the AlAs 'oxidation layer' by an Al0.3Ga0.7As barrier with thicknesses varying from 200Å to 400Å. The layer to be oxidized is 500 Å of AlAs, and the whole structure is capped by 500 Å GaAs. In all cases, the oxidation proceeds laterally in from the edges of a lithographically defined stripes (100 µm in width on 150 µm center-to-center spacing). Oxidation times were chosen to just fully oxidize the patterned stripes, and hence varied according the temperature used. For all conditions chosen, the photoluminescence (PL) intensity of the probing quantum well is significantly reduced after the oxidation process. As shown in figure, the PL degradation appears to be systematically worse at the lower oxidation temperatures. Those data also show that for the lower oxidation temperatures, the degree of PL degradation is almost the same whether the quantum well is placed 200Å or 400Å from the oxidation layer. At the highest oxidation temperature shown, there is an improvement in PL intensity when the QW is placed farther away from the oxidation layer. Our earlier work had shown that oxidation-induced degradation in the QW PL could be attributed to the generation of defects which diffused to the QW region, as well as the structural modification of the material immediately surrounding the oxidation layer. We believe that the critical parameter in these experiments is the total oxidation time, which influences the effective diffusion lengths of oxidation-generated defects. The data shown here can help us to further characterize the diffusion of those oxidation-generated defects. A preliminary fit to our data give a value of diffusivity ~4.1x10-16cm-2/s, which agrees well with the published value of diffusion coefficient of As in GaAs.

1:50 pm

The Role of Barrier Layers for Oxidation Depth Control in Al-Containing Heterostructures: Carol I.H. Ashby, Olga Blum, Hong Q. Hou, Sandia National Laboratories, P.O. Box 5800, Albuquerque, NM 87185-0603

One standard approach to controlling the depth of oxidation of different Al-containing layers in a particular heterostructure is the variation of the Al mole fraction, since higher Al-fraction materials generally oxidize at a faster rate. An alternative approach to depth control is based on the interposition of thin GaAs barrier layers between adjacent layers of higher and lower Al mole fraction. We have demonstrated the utility of this latter approach for the formation of integrated optical elements, such as lenses, in AlGaAs/GaAs vertical-cavity devices. We have controlled lateral oxidation depths in mesa structures by placing thin (typically <150 Å) GaAs barrier layers between AlGaA layers of different Al mole fractions. Our initial studies have employed Al0.94Ga0.06As and Al0.98Ga0.02As of variable thickness. The resultant oxidation profiles for the lower-Al-content layer have been modeled as the summation of a constant lateral oxidation component determined by the Al0.94Ga0.06As layer thickness and of an enhanced oxidation component due to species diffusing to the Al0.94Ga0.06As through the barrier layer from a porous feeder channel provided by the more rapid oxidation of Al0.98Ga0.02As to Al2O3. The magnitude of the enhancement is controlled primarily by the GaAs barrier thickness, provided the oxidized depth of the adjacent Al0..98Ga0.02As layer exceeds that of the Al0.94Ga0.06As layer. The dependence on oxidation time, temperature, and the thickness of the GaAs and Al1-xGaxAs layers will be discussed and a possible mechanism for the enhanced oxidation will be presented. The implications for oxidation-profile definition by initial heterostructure design will also be discussed. This work was performed at Sandia National Laboratories and supported by the U.S. Department of energy under Contract No. DE-AC04-94AL85000. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy.

2:10 pm, Student Paper

Photoluminescence Properties of III-V Heterostructures Having Native-Oxide Layers: B.P. Tinkham, M.R. Islam, and R.D. Dupuis, Microelectronics Research Center, The University of Texas at Austin, Austin, TX 78712-1100; A.P. Curtis, G.E. Stillman, Center for Compound Semiconductor Microelectronics, University of Illinois at Urbana-Champaign, Urbana, IL 61801-2991; D.T. Mathes, R. Hull, Dept. of Materials Science and Engineering, University of Virginia, Charlottesville, VA 22903-2442

Native-oxide layers formed by steam oxidation of Al-containing epitaxial layers are of interest for a variety of electronic and optoelectronic III-V devices. It is important to determine the effect of the oxidation process upon the properties of the underlying "active" device layers. In this talk, data are presented on the electrical and optical characteristics of heterostructures grown by metalorganic chemical vapor deposition (MOCVD) both before and after oxidation. We have studied GaAs and InGaP doped and undoped "active" layers having various native-oxide "cladding layers" produced by the steam oxidation of AlGaAs and "window" layers. The optical properties of the heterostructures has been studied using 300K and 4K photoluminescence (PL) and 300K time-resolved PL. Impurity concentrations have been analyzed using secondary-ion mass spectrometry and the structural characteristics have been studied using transmission electron microscopy analysis. The luminescence intensity and lifetime from GaAs "active regions" drop dramatically when the adjacent AlGaAs "window layer" is oxidized completely. However, there is an increase in the PL efficiency and decay time of the luminescence with the oxidation of InAlP window layers. Furthermore, we observe that there is a marked change in the impurity-related luminescence in the 4K PL spectra of the GaAs layers after oxidation, indicating a change in the BULK properties of the GaAs layer. For example, the luminescence from a GaAs:Zn "active layer" shows a much smaller Zn-related PL emission after oxidation. Furthermore, the use of a "P-based" oxide compared to the "As-based" oxide leads to dramatically different post-oxidation PL spectra for the GaAs "active region". These results have important implications for devices, including vertical-cavity surface-emitting lasers and LED's. Work supported in part by NSF, ARO, and the State of Texas.

2:30 pm

Control of Wet-Oxidation of AlAs/GaAs Superlattices by Impurity-Induced Layer Disordering: Decai Sun, P.D. Floyd, R.L. Thornton, C. Dunnrowicz, C. Chua, and D.W. Treat, Xerox Palo Alto Research Center, 3333 Coyote Hill Road, Palo Alto, CA 94304

The wet oxidation of Al-based alloys has attracted a lot of attention over the last few years, especially for use in the fabrication of high performance laser diodes. Typically an AlxGa1-xAs (x>0.9) layer inserted into the cladding region of a laser structure is laterally oxidized from an etched mesa to form a circular or stripe aperture for optical and electrical confinement. It has been observed, however, that the final size and shape of the oxide aperture is hard to control since the oxidation rate of AlGaAs is crystal-orientation dependent and crystal defects affect the oxidation rate microscopically. Oxidation of AlAs from an etched circular mesa tends to lead to a diamond shape. The problem is more severe in fabricating edge emitting devices in which a uniform oxide aperture about 3 to 4 µm wide but hundreds of micrometers long needs to be formed. In this paper, we present a method of controlling oxidation of Al-based alloys by utilizing the oxidation rate difference of the AlGaAs alloys of different Al contents, which allows the formation of an oxide aperture of any shape. It is known that the oxidation rate of AlxGa1-xAs has an exponential relation with Al composition x; the larger the x is, the faster the oxidation. By exploiting the difference in oxidation rate between AlGaAs alloys with different Al contents, the formation of small AlGaAs-oxide apertures can be controlled. The lateral variation of Al composition in an AlGaAs heterostructure can be realized by selective layer intermixing. We use the Si-impurity induced layer disordering technique to disorder an AlAs/Al0.15Ga0.85As superlattice heterostructure outside a SiNx circular mask of 50 µm in diameter with Si impurities diffused in from the surface. The disordered region is fully intermixed into Al0.6Ga0.4As. The AlAs layers inside the intermixed region are then oxidized from an etched mesa. The oxidation front is fully contained by the intermixed boundary and well defined. In another structure, an Al0.96Ga0.04As(300Å)/Al0.4Ga0.6As heterostructure with Si modulation doping in the Al0.96Ga0.04As layer is disordered selectively under a 50 µm wide SiNx stripe. Oxidation of Al0.96Ga0.04As laterally from a 70 µm wide stripe mesa was stopped by the disordered AlGaAs alloy interface. The oxidation aperture is smooth along the stripe with uniform width. Other techniques for selective layer disordering and oxidation such as vacancy-enhanced layer disordering will also be discussed.

2:50 pm, Student Paper

Spatially Selective Disordering of InGaAs/GaAs QW Using AlAs Native Oxide and Rapid Thermal Annealing: Chao-Kun Lin, X. Zhang, P.D. Dapkus and D.H. Rich, University of Southern California, Los Angeles, CA 90089

The native oxide of AlAs has received a great deal of interest recently because of its unique potential to realize complex electronic and optoelectronic device structures. VSCEL's employing native oxide current constriction and mode control layers have demonstrated ultralow threshold current laser operation and GaAs MESFET's fabricated on AlxOy layer have also exhibited excellent performance. In this study, the application of AlAs native oxide as a source for QW disordering is presented. The native oxide, used as a source of vacancies in vacancy-mediated compositional disordering, can be placed in an arbitrary location in a multilayer structure and thus offers the possibility to construct three dimensional integrated electronic and optoelectronic structures. In this talk we demonstrate for the first time that buried AlxOy layers can be used as sources for localized disordering. The structures in this study were grown by MOCVD. One such structure is shown in Fig. 1. The graded layers on both sides of AlAs layer are designed to relieve the stress between the native oxide and semiconductor and the 2 µm GaAs cap is used to block surface defects that could cause InGaAs QW disordering in the heat treatments used. To examine the disordering potential of the oxide, 500 µm mesa stripes separated by 50 µm were fabricated by using photolithography and wet chemical etching to expose the AlAs layer. This was followed by wet oxidization at 425°C with 300 sccm N2 gas bubbled through 88°C DI water. After 90 minutes, 70 µm wide AlxOy layers were formed laterally on both edges of the mesas. Subsequent heat treatments were carried out using RTA with flowing N2 gas. The samples were sandwiched by two GaAs substrates to provide proximity protection and heated up to various temperatures at a 1°C/S ramping rate. This slow rate was found to be necessary to avoid cracking of the oxide. After removal of the 2µm GaAs cap layer, the samples were then examined by both photoluminescence and cathodoluminescence to determine the change in the emission wavelength of the InGaAs quantum well. Fig. 2 shows the room temperature PL emission wavelength results for various annealing temperatures. The InGaAs QW's in both the oxide and non-oxide region have been blue shifted by the thermal treatment but by different amounts. The blue shift of the QW in the oxide region is greater than in the unoxidized region. The maximum energy shift difference, 44.8 meV, occurred at 900°C. Room temperature CL results are shown in Fig. 3. The transition of the CL peak across the AlxOy/AlAs boundary is very sharp and no intermediate peak has been observed although the CL intensity is reduced near the oxide interface. Thus, spatially selective InGaAs/GaAs QW disordering facilitated by AlAs native oxide and thermal annealing has been demonstrated. Since the AlAs epitaxial layer can be placed very close to the QW, the QW disordering effect can be carried out easily. The talk will discuss the spatial resolution of this technique towards forming lateral carrier confinement in quantum wells and will also discuss its use in GaAs quantum well structures.

3:10 pm, Break

3:30 pm

Growth of Ga2O3(Gd2O3) Using MBE Technique - Key to the Demonstration of Enhancement Mode GaAs MOSFET: M. Hong, J.P. Mannaerts, F. Ren, J. Kwo, M.A. Marcus, W-Y. Hwang, S.N.G. Chu, V.J. Fratello, and A.Y. Cho, Bell Laboratories, Lucent Technologies, Murray Hill, NJ 07974

We have recently demonstrated enhancement-mode GaAs MOSFETs (metal-oxide-semiconductor field-effect-transistors) with inversion by combining a novel oxide deposition and conventional ion implantation processes. The oxide is a mixture of Ga2O3 and Gd2O3, which forms an excellent interface with GaAs having an interfacial state density as low as mid 1010 cm-2eV-1. It has been generally recognized that GaAs MOSFETs may offer potential advantages over conventional Si-based MOSFETs in terms of high speed, low power consumption and circuit simplicity. However, during the past three decades, many attempts toward this end did not lead to devices showing commercially acceptable properties, mostly due to poor quality of the gate oxide. These approaches include (a) thermal, anodic, and plasma surface oxidation of GaAs, and (b) deposition of different dielectric materials, e.g., Si3N4, SiOx, Al2O3, and Ga2Ox in combination with dry, liquid, and photochemical surface treatment. In this talk, we discuss the growth of this new oxide using a multi-chamber molecular beam epitaxy (MBE) system. Before the gate oxide deposition, chemically clean and atomically ordered GaAs surfaces of the device wafers were achieved using thermal desorption (580-600°C) in a solid source MBE chamber under an As overpressure. In-situ reflection high energy electron diffraction (RHEED) was used to monitor the desorption of GaAs native oxides and to ensure the attainment of a reconstructed GaAs surface. The wafers were then transferred under a UHV of low 10-10 Torr to another MBE chamber for oxide deposition. The second MBE chamber is As-free, for reasons that excess As on GaAs surface may contribute to the interface states. In addition to the studies of Ga2O3(Gd2O3), other oxides such as SiO2, Al2O3, and MgO were also investigated. Chemical compositions of all these oxide-GaAs interfaces examined by x-ray photoelectron spectroscopy (XPS) revealed no chemical reactions between the deposited oxides and GaAs. In particular, As2O3 and As2O5 are not detectable. However, the interface recombination velocity of SiO2-, Al2O3-, and MGO-GaAs is high, about 107 cm/s, comparable to that of a bare GaAs surface. Only the Ga2O3(Gd2O3)-GaAs interface has a low recombination velocity of 4-5x103 cm/s, which is consistent with the low interface state density of mid 1010 cm-2 eV-1 as inferred from the capacitance-voltage measurements. The concentration of Gd2O3 varies from 0 to 40 at.%, depending on the deposition conditions. The roughness of the Ga2O3(Gd2O3)-GaAs interface as studied by low angle x-ray scattering and cross section transmission electron microscopy will be reported.

3:50 pm, Student Paper

Wet Thermal Oxidation of Low Temperature Al.98Ga.02As: H. Reese, Y. Chiu, A. Shakouri, F. Hu, and J. Bowers, University of California-Santa Barbara- ECE Dept., Santa Barbara, CA 93106

In this report, we have studied wet thermal oxidation at 400 ~ 450°C of low temperature (LT) grown Al.98Ga.02As. This study demonstrates that the oxidation process is dependent on the microstructure of the LT Al.98Ga.02As. The LT Al.98Ga.02As oxidized at about twice the rate of standard grown Al.98Ga.02As. Two Al.98Ga.02As wafers (fig. 1 and 2) are grown by MBE at 320°C, and a reference sample (fig. 3) is grown at 570°C. Sample LT1 and the reference sample are cleaved into five samples each and annealed after growth by RTA (Rapid Thermal Annealer) at 500°C, 590°C, 650°C, 700°C, and 800°C. Sample LT2 is annealed in situ during growth at 590°C. Mesas are then etched using H3PO4/II2O (3:1:50) to expose the side walls of the Al.98Ga.02As. The samples are oxidized in a water vapor ambient, created by an N2 carrier gas bubbling through 80°C water. The oxidation is carried out for various oxidation times in furnace temperatures of 400°C, 420°C, and 450°C, and an N2 flow rate of 550 mL/min. The as-grown samples show very different oxidation characteristics. Sample LT1 has not shown lateral oxidation, and a linear oxidation front is not observed. Instead, a dominant color is observed for each oxidation time. These observed colors of sample LT1 may be due to vertical oxidation, in which the water vapor diffuses through the LT GaAs cap layer to oxidize the LT Al.98Ga.02As. Figure 4 shows the LT Al.98Ga.02As after 10 minutes of oxidation. The color remains constant, which suggests that the 2500 Å of LT Al.98Ga.02As has completely oxidized. The samples annealed at different temperature also have different times of complete oxidation. These samples annealed at different temperatures have different microstructures, which contribute to the different times for complete oxidation. When LT Al.98Ga.02As is annealed, As precipitates from inside the material, and the size of these precipitates increases as the annealing temperature increases. These As precipitates may serve as nucleation centers for the excess As generated during the oxidation process. When the density of the As precipitates decreases, the oxidation rate also decreases.

4:10 pm, Student Paper

Effect of Oxidation of AlxGa1-xAs on Adjacent Semiconductor Layers: HALL (Electrical) and TEM (Structural) Characterization: P.A. Parikh, P.M. Chavarkar, L. Zhu, J. Ibbetson, J.S. Speck, and U.K. Mishra, ECE Department, University of California, Santa Barbara, CA 93106

A high efficiency, high gain, linear power amplifier technology is in great demand. To this end we are developing a GaAs On Insulator (GOI) technology with Al2O3 formed by the steam oxidation of AlxGa1-xAs as the buffer layer insulator. The object of this work is to study the effect of oxidation of an AlxGa1-xAs layer on the properties of the adjacent semiconductor layers. In this study, we report the first (to the best of our knowledge) HALL measurements on a doped layer (InGaAs channel) over an oxidized layer (Al2O3 buffer formed by wet oxidation of Al0.98GaAs) and we present TEM micrographs illustrating the effect of overoxidation of the Al0.98GaAs layer on adjacent semiconductor layers. The as grown device structure is an Al0.25GaAs/In0.2GaAs p-HEMT grown on 500Å Al0.98GaAs followed by 100 Å Al0.25GaAs spacer layer grown on a conventional GaAs buffer on a S.I GaAs substrate. SiO2 mask is used to define the Hall mesa (the Hall mask is designed so as to enable lateral oxidation of the entire active Hall area), simultaneously exposing the AlGaAs layer laterally. The processing for TEM samples is similar except that the mesa structure is a set of lines 50 µm wide the 50 µm spacing. Next, the steam oxidation is done at 420°C to 450°C with N2 bubbling through water at 85°C. Finally lithography for the contact is done, the oxide cap removed and standard AuGe/Ni/Au alloyed contacts defined. The Hall measurements were done for the unoxidized sample, and samples oxidized at 420°C and 450°C. For the TEM studies, a temperature of 450°C was used with three different oxidation times of 15, 30 and 45 minutes. The mobility and the carrier concentration values were:n=5.1x1012/cm2, µ=3250cm2/V-s for the unoxidized sample, n=2.7x1012/cm2, µ=2630cm2/V-s for the 420°C sample and n=2.8x1011/cm2, µ=360 cm2/V-s for the 450°C sample. For the TEM samples, the one oxidized for 15 minutes shows negligible effect on the InGaAs layer and most of the damage is confined to the 100 Å spacer layer. For the 30 minute and 45 minute oxidation times, clear degradation is observed in the active InGaAs layer. This means that the oxidation times, clear degradation is observed in the active InGaAs layer. This means that the oxidation causes degradation of adjacent regions, which is evident by the inferior mobility and sheet charge density in the oxidized samples, and observed from the TEM pictures. Overoxidation results in more damage to the adjacent regions and oxidizing at lower temperature results in less degradation of channel properties. However reduced oxidation rate at lower temperature is a limiting factor so an optimum temperature has to be determined. This study will feed into the ongoing development of the GOI MESFET and p-HEMT technology.

4:30 pm, Student Paper

CV and DLTS Characterization of MIS Capacitors on In0.53Ga0.47As with Oxidized AlAs0.56Sb0.44 as the Insulating Dielectric: J.G. Champlain, P.M. Chavarkar, P.A. Parikh, and U.K. Mishra, Department of Electrical and Computer Engineering, University of California, Santa Barbara, CA 93106

In0.53Ga0.47As with its high electron mobility and carrier saturation velocity offers the capability to fabricate low power, high frequency field-effect transistors. Oxides as insulators in electronics offer the possibility for reduced leakage currents through the buffer or gate of a field-effect transistor (FET) increasing the efficiency of the device and reducing the power consumption. Incorporating oxide insulators into InP based electronics will enable the fabrication of low leakage, low power high speed MISFETs. Oxidation of AlAs0.56Sb0.44 (which is lattice matched to InP) can be used to obtain an insulating dielectric in the AlInAs/GaInAs/InP material system. Metal-insulator-semiconductor (MIS) capacitors were fabricated on InGaAs with oxidized AlAsSb as the insulating dielectric. The epi-structure of the capacitor is as follows: semi-insulating InP substrate, 1500 Å p+ (Be: 5x1018 cm-3) In0.53Ga0.47As bottom ohmic contact layer, 1000 Å p- (Be: 3x1017 cm-3) In0.53Ga0.47As layer, 500 Å AlAs0.56Sb0.44 layer for oxidation, 500 Å n+ (Si:5x1018 cm-3) In0.53Ga0.47As top contact layer. The epistructure was grown by MBE. The AlAsSb layer was grown using antimony in its tetramer form. The beam equivalent pressures of As2 and Sb4 were 1.0x10-6 Torr and 2.0x10-6 Torr, respectively. Mesas were defined using BCl3:Cl2:SiCl4 (10:10:10 sccm) reactive ion etching (RIE) to expose the AlAsSb layer. Lateral oxidation of the AlAsSb layer was performed at 350°C in an open tube furnace by bubbling N2 through water heated to 90°C for 1 hour. Ti/Au metallization was used for ohmic contacts to the bottom p+ contact layer and the top n+ contact layer. High frequency capacitance-voltage (C-V) and deep level transient spectroscopy (DLTS) measurements were performed on MIS capacitors with a 3848.5 mm2 area. From the results of C-V measurements the capacitance of the oxide was found to be 4.318 pF which gives a relative permittivity of 6.3 for the oxide. Preliminary results from the DLTS measurements give a trap density of 1x1012 /cm2 with the interface/bulk trap energy level 0.153 eV from the valence band. The leakage current is 2 nA at a bias of 1 Volt as obtained from I-V measurements on MIS capacitors. We have demonstrated for the first time CV and DLTS characteristics of MIS capacitors on InGaAs with oxidized AlAsSb as the insulating dielectric. The use of oxidized AlAsSb as gate insulator will enable the fabrication of InGaAs based MISFETs for ultra low-power, high frequency electronics.

4:50 pm

Optical and Electronic Properties of Oxidized AIN Thin Films Grown at Different Temperatures: Enam Ahmed Chowdhury, G. Qui, M. Dashiell, S. Woods, J.O. Olowolafe, D. Van Der Weide, and J. Kolodzey, Electrical Engineering Department, University of Delaware, 140 Evans Hall, Newark, DE 19716

We report on the properties of a new insulator, AlO:N, for compound semiconductors produced by thermally oxidizing AlN thin films. The process steps were similar to the ones for SiO2, creating the possibility of a new technology for nitride-based metal-insulator-semiconductor field effect devices and integrated circuits. In this report, we describe how the properties change versus oxidation temperature. Thin films of AlN were grown on p-type Silicon or quartz substrates by reactive magnetron sputtering, using an Al metal target with an RF power of 200-400 W, and a mixture of N2 and Ar gases with partial pressures of 3-10 mTorr. AlN film thicknesses ranged between 0.1-0.6 µm, and x-ray diffraction revealed a polycrystalline structure. The AlN films were oxidized under O2 flow in a horizontal quartz tube furnace from 800°-1100°C, with a duration of 1-3 hours. Depending on temperature and time, the AlN films were partially, or fully oxidized. The film structure and composition was analyzed with RBS and XRD, showing complete oxidation of AlN above 900°C, with conversion into Al2O3 with small amounts of residual nitrogen. Considering the extreme importance of surfaces in field-effect devices, we used atomic force microscopy (AFM) in air and observed a reduction in the surface roughness in the oxidized films compared to the as-grown AlN. From capacitance-voltage (C-V) measurements of MOS structures (with Al metal contacts), we obtained dialectic constants up to 12.4 for the AlO:N. Effective oxide charge densities lower than 1011 cm-2 were observed. The C-V curves were comparable to thermally grown SiO2 on Si. Optical properties of the oxidized films were measured using reflectance and transmittance at visible and infrared wavelengths. In the near IR range, we observed ~30% reflectance of films oxidized at 1100°C. We will report on the variation of properties versus oxidation temperature. This work was supported by grants from DARPA and the ARO.

The information on this page is maintained by TMS Customer Service Center (

Search TMS Meetings Page About TMS TMS OnLine