JOURNAL OF ELECTRONIC MATERIALS
	ABSTRACTS Volume 24, Number 8, August 1995 This Month Featuring: Proceedings of the 1994 TMS Fall Meeting in Rosemont, IL, October 4-6. View August 1995 Table of Contents.

SPECIAL ISSUE PAPERS

Twenty-five papers were presented, describing crystalline, nanostructured, and amorphous deposits of metals, alloys, and composites prepared by the electrochemical, chemical (electroless), and vapor deposition techniques. Lattice imperfections, substructure, grain configuration, and crystallogrphic texture were related to internal stresses, surface roughness, nodular defect growth, and stress migration voiding. The impact of structural transformations was analyzed in terms of thermal stability and mechanical, electrical, magnetic, and corrosion behavior.

A selection of eight representative papers is presented in this collection. The editors wish to thank the authors for preparation of the final manuscripts, the reviewers for prompt and judicious criticisms, and JEM for the opportunity to publish this body of work.

Harish D. Merchant
Gould Electronics
Eastlake, OH

Sailesh M. Merchant
AT&T Labs
Orlando, FL

Guest Editors

Thermal Response of Electrodeposited Copper
HARISH D. MERCHANT
Gould Electronics, Eastlake, OH 44095-4001.

KEY WORDS Annealing, copper, electrodeposit, embrittlement, recovery, recrystallization

Thermal response of electrodeposited copper has been characterized by monitoring changes in microstructure, tensile strength, elongation, and microhardness following 30 minute isothermal anneal at temperatures between 23 and 450°C. By judicious control of additives to the electrolyte, considerable enhancements of strength, hardness, and anneal resistance are obtained. When recrystallization is shifted to higher temperatures, annealing proceeds by recovery-like processes; prior to the onset of recrystallization, significant strength loss may occur by recovery. Recrystallization initiates near the substrate side of the deposit and progresses along an uneven front across the deposit thickness. The microhardness and tensile response monitors indicate several unusual anneal-related microstructural processes: (i) precipitation (of microvoids) hardening, (ii) dissolution and reprecipitation, (iii) stepped progress of recovery and recrystallization, and (iv) embrittlement (loss of elongation) at low anneal temperatures and prior to the onset of recrystallization.

Through-Thickness Characterization of Copper Electrodeposit
R.J. DE ANGELIS,¹ D.B. KNORR,² and H.D. MERCHANT²
1--Department of Mechanical Engineering, University of Nebraska, Lincoln, NE 68588. 2--Materials Engineering Department, Rensselaer Polytechnic Institute, Troy, NY 12180. 3--Gould Electronics Inc., Eastlake, OH 44095

KEY WORDS Copper, electrodeposit, embrittlement, dislocation density, particle size, texture, twin spacing

Through-thickness crystallographic texture, defect structure, and tensile embrittlement of 35 µm thick electrodeposit are characterized by successive thinning. An initially random grain structure, inherited from the substrate, evolves into a strong <220> fiber texture. The random to oriented grain transformation begins at the inception of thickening and is complete after about 15 µm deposit thickness, where about 0.9 volume fraction of grains become oriented near <220>. Further thickening of the deposit sharpens the texture, reducing the scatter around the <220> ideal orientation. A duplex coarse/fine particle (coherent domain) structure is obtained. Coarse particles along <220> are less defective and have smaller lattice strains; fine particles along <200>, presumably associated with the random grains, are defect-saturated with finely spaced twins, high dislocation density and enhanced lattice strains. With increasing distance from the shiny surface (of initial film formation), especially following the initial 10 µm deposit thickness, (a) along <220>: particle size and twin spacing increase whereas dislocation density and root mean square (rms) strains decrease, (b) along <200>: particle size increases gradually, dislocation density and rms strains increase sharply and the already fine twin spacing remains unchanged, and (c) the effective particle size ratio D_eff <220>:D_eff<200> exceeds 1.4, suggesting a twinning-induced z-direction particle shape anisotropy. A substantial decrease in tensile elongation is observed at 180°C. The embrittlement increases with the deposit thickness, attributed to the development of low density regions in the morphological boundaries. High elongation and embrittlement directional anisotropies are observed near the shiny surface, perhaps due to preferred nucleation on the substrate asperities.

Nodular Defect Growth and Structure in Vapor Deposited Films
R.N. TAIT,¹ T. SMY,² S.K. DEW,³ and M.J. BRETT³
1--Alberta Microelectronic Centre, #318, 11315-87 Avenue, Edmonton, Alberta, Canada T6G 2T9. 2--Department of Electronics, Carleton University, Ottawa, Ontario, Canada KIS 5B6. 3--Department of Electrical Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2G7.

KEY WORDS Microstructure, modeling, nodule defect, thin film, vapor deposition

Nodular defect growth is a problem in evaporated, sputtered, chemical vapor deposited (CVD), and electrodeposited films. In this work, the formation and structure of nodular defects in evaporated MgF₂ and sputtered Ti were studied and modeled. A structurally similar but fundamentally different runaway growth structure was observed in plasma enhanced CVD (PECVD) Si on Ni. Nodule formation was initiated by seeding the substrate with polystyrene spheres for both the evaporated and sputtered deposits. These nodules grew in a characteristic cone shape with a domed top, and with the nodule separated from the bulk of the film by a low density boundary. Simulation of nodule growth using the SIMBAD Monte Carlo deposition model reproduces the nodular shape as well as the low density region which extends from the seed to the film boundary. Sputtering pressures were found to have little influence on the nodule characteristics, although film thickness and surface diffusion can have important effects. Evidence was found to indicate that the defect structure in the CVD films is due to Ni diffusion into the growing Si film. The structure differs from the physical vapor deposition nodule in two respects: extent of the growth and the adhesion of the defect to the film.

Thermal Stability of Electroless NiMeP Amorphous Alloys
N. KRASTEVA,¹ S. ARMYANOV,¹ J. GEORGIEVA,¹ N. AVRAMOVA,² and V. FOTTY¹
1--Institute of Physical Chemistry, Bulgarian Academy of Sciences, Sofia 1113, Bulgaria. 2--College of Chemistry, University of Sofia, Sofia 1226, Bulgaria.

KEY WORDS Amorphous alloys, crystallization parameters, differential scanning calorimetry, electroless ternary alloys, magnetic properties, thermal stability

The thermal stability and the preservation of paramagnetic state of amorphous electroless NiP alloys are important for many industrial applications. The addition of third component (Me = Cu, Sn, or Sb) to NiP alloy should reduce the magnetic moment of precipitated ferromagnetic phase. The influence of Me on the thermal stability of ternary alloys is studied. Using differential scanning calorimetry, analysis of the crystallization kinetics is carried out. The remanence after annealing at different temperatures is presented.

Substructure Formation and Texture in Electrodeposits
OLEG B. GIRIN
State Metallurgical Academy of Ukraine, Prospekt Gagarina 4, Dnepropetrovsk 320635, Ukraine.

KEY WORDS Corrosion, electrodeposit, substructure, texture

The author's work on electrodeposit substructure in relation to the texture is reviewed. Addressed are substructure anisotropy in electrodeposits and the axial and the random components of their texture, periodicity of the substructure with increasing nuclear charge of constituent atoms, regular variation in the substructure with increasing supercooling in the electrodeposition, and occurrence of an amorphous state during electrocrystallization of metals. Interrelation between the substructure and the texture in electrodeposits, and the influence of structure on electrodeposit properties are also discussed.

Electrochemical Processing of Metallic Nanowire Arrays and Nanocomposites
P.C. SEARSON,¹ R.C. CAMMARATA,¹ and C.L. CHIEN²
1--Department of Materials Science and Engineering, The Johns Hopkins University, Baltimore, MD 21218. 2--Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, MD 21218.

KEY WORDS Electrodeposition, magnetic properties, nanocomposites, nanostructured materials, nanowires

Electrochemical methods for surface modification and materials processing play an important role in many modern technologies. Electrochemical fabrication techniques are becoming increasingly attractive for the deposition of new materials and novel microstructures. We report on property enhancements observed for electrochemically deposited Co and Ni nanowire arrays and Ni/Al₂O₃ nanocomposite films.

Magnetron Sputter Deposition of A-15 and BCC
Crystal Structure Tungsten Thin Films
M.J. O'KEEFE,¹ J.T. GRANT,² and J.S. SOLOMON²
1--Wright Laboratory Solid State Electronics Directorate, Wright-Patterson AFB, OH 45433-7319. 2--Research Institute, University of Dayton, Dayton, OH 45469-0167.

KEY WORDS Body centered cubic (bcc), sputtering, thin films, tungsten

The effect of processing parameters on the crystal structure and electrical resistivity of magnetron sputter deposited tungsten thin films was investigated. Formation of body centered cubic (bcc) W was favored when the concentration of impurity oxygen atoms in the films was <5 at.% while the formation of A-15 W was favored between 6 and 10 at.% oxygen. A transition from A-15 W films to bcc W films occurred as the oxygen was removed from the deposition chamber by presputtering the target for extended periods of time. The binding energies of the W atoms in A-15 and bcc W films are similar, as is the binding energy of the O atoms in the two different crystal structures, indicating that the oxygen is not present as a tungsten oxide compound. The resistivity of A-15 W films is always higher than the resistivity of bcc W films due to the increased oxygen concentration and small grain size of the A-15 films. However, the sputter deposition pressure is found to have a greater effect on resistivity. This is attributed to the formation of cracks in the film.

Enhanced Stress-Migration Reliability for ULSI Interconnect: An Insight into the Perils of Screening Al Depositions Based on Grain Size
V. RYAN,¹ and S.M. MERCHANT²
1--AT&T Bell Laboratories, Allentown, PA 18103. 2--AT&T Bell Laboratories, Orlando, FL 32819.

KEY WORDS Aluminum, grain size, reliability, stress migration (SM), void

The recent trend toward increasing packing density and the demand for improved reliability have dramatically challenged ultra large scale integrated circuit technologists to develop more robust fabrication processes. Finer feature sizes and the addition of layers of interconnect, combined with large mechanical stresses, have greatly exacerbated the insidious problem of stress-induced voiding--open circuits can appear in Al lines immediately after fabrication or after years of shelf storage. Finite element analysis shows that one of the prime candidates for controlling the root cause of the failure mechanism, stress-assisted grain-boundary diffusion, is carefully engineered Al grain size. This paper examines various parameters related to grain size that directly result in mechanically weak populations of interconnect elements. We highlight some critical factors that give rise to unique implementation issues, including the complex statistical nature of the susceptible sites and the evolution of the microstructure during device fabrication. The resulting experimental reliability enhancement can be startling--a new deposition process accounted for a 350% increase in lifetime in sub-micron wide lines.

REGULAR ISSUE PAPERS

Influence of the Liquid Phase Epitaxial Growth Conditions on Composition Profile for Hg_1-xCd_xTe Film
B. LI, J.H. CHU, X.Q. CHEN, J.Y. CAO, and D.Y. TANG
National Laboratory for Infrared Physics, Shanghai Institute of Technical Physics, Shanghai 200083, China.

KEY WORDS Hg1-xCdxTe, liquid phase epitaxy, phase diagram

In this paper, the expressions for the effective segregation coefficient and composition depth profile of Hg_1-xCd_xTe liquid phase epitaxial (LPE) film are presented and compared with the experimental data. The expressions are derived based on the associated solution model and the phase diagram theory of Hg_1-xCd_xTe. The results show that supercooling, cooling rate, liquid composition, and nonequilibrium Hg pressure over the solution will result in the changes of average composition and composition-depth profile of the grown film. Adequate Hg loss, slow cooling rate, and proper prescription of liquid composition can improve the uniformity of longitudinal composition of Hg_1-xCd_xTe LPE film.

Phase and Magnetic Properties of Iron Oxide Clusters in 20CaO·20SiO₂·7Fe₂O₃·6FeO Glasses
MIYUKI HAYASHI, MASAHIRO SUSA, TOSHIO MARUYAMA, and KAZUHIRO NAGATA
Department of Metallurgical Engineering, Faculty of Engineering, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152, Japan.

KEY WORDS -Fe2O3, anistropy energy constant, cluster, Fe3O4, ferrimagnetism, pinning, silicate glass, superparamagnetism

The annealing of 20CaO·20SiO₂·7Fe₂O₃·6FeO glasses at 973Kin vacuo produced clusters of iron oxide, the shape of which was nearly spherical and the diameter distributed in the narrow range 25-115Å. The phase of clusters was identified to be Fe³⁺(Fe³⁺_1.30 Fe²⁺_0.55 V_0.15 )·O₄ in the inverse spinal structure based upon the Mössbauer spectra and x-ray diffraction profiles. The clusters exhibited superparamagnetism and their effective anisotropy energy constant was inversely proportional to the cluster diameter. The magnetization of the glasses measured by a vibrating sample magnetometer was 7.2 x 10^-6Wbmkg^-1 at 10 kOe at room temperature and smaller than the value calculated assuming that the whole clusters have superparamagnetism. These results suggest the pinning of spins near the cluster surface.

A Novel Processing Technique to Fabricate Planar InGaAsP/InP Electroabsorption Waveguide Modulators
Q.Z. LIU, X.S. JIANG, L.S. YU, Z.F. GUAN, P.K.L. YU, and S.S. LAU
Department of Electrical and Computer Engineering, University of California, San Diego, La Jolla, CA 92093-0407.

KEY WORDS InGaAsP/InP, processing technique, waveguide modulators

A novel processing technique has been developed to fabricate planar electroabsorption waveguide modulators in compound semiconductor heterostructures. The lateral confinement of light is achieved by introducing controllable, reproducible, and stable stresses into semiconductor heterostructures using WNi surface stressor stripes, which also serve as electrodes for the waveguide modulators. Self-aligned helium implantation is employed to achieve electrical isolation using the stressors as the templates for the ion masks. An increase as large as 33000 times has been obtained in the dc resistance between the neighboring waveguide modulators 25 µm apart. Propagation loss of 1.7 dB/cm is observed in the photoelastic waveguides at a wavelength of 1.53 µm following the He implantation. A post-implant thermal annealing at 310°C for 40 min increases the dc resistance between the neighboring devices to the maximum value, and at the same time reduces the optical loss to its value before ion implantation (less than 1 dB/cm). Using a combination of the photoelastic effect and helium implantation, planar InGaAsP/InP Franz-Keldysh-effect waveguide modulators 430 µm long with a 10 dB extinction ratio at 3 V for the TM mode have been fabricated. Planar electroabsorption quantum-confined Stark effect waveguide modulators have also been demonstrated. This planar device processing technique may prove valuable in future photonic integrated circuit technology.

Control of Implant-Damage-Enhanced Boron Diffusion in Epitaxially Grown n-Si/p-Si_1-xGe_x/n-Si Heterojunction Bipolar Transistors
T. GHANI,¹ J.L. HOYT,¹ A.M. MC CARTHY,² and J.F. GIBBONS¹
1--Solid State Electronics Laboratory, Stanford University, Stanford, CA 94305. 2--Lawrence Livermore National Laboratory, Livermore, CA 94551.

KEY WORDS Boron diffusion in Si, heterojunction bipolar transistor (HBT), implant damage enhanced diffusion, pulsed laser annealing, SiGe

Boron out-diffusion in epitaxially grown n-Si/p⁺-Si_1-xGe_x/n-Si heterojunction bipolar transistors is significantly enhanced during 850°C, 10 s rapid thermal annealing following arsenic emitter contact implantation. In this paper, we introduce three techniques which dramatically reduce boron out-diffusion during implant activation. Limiting the post-implant processing to 600°C for 2 min results in minimal diffusion giving acceptable device performance. A second technique involves pulsed laser annealing of the As implant, which removes residual defects and eliminates enhanced diffusion during subsequent thermal processing. Finally, we show that high bulk concentrations of oxygen in the Si_1-xGe_x (~10²⁰ cm^-3) dramatically reduce the implant-damage-enhanced boron diffusion. In addition to the depth profiles, electrical measurements performed on heterojunction bipolar transistors, incorporating these fabrication techniques, show ideal collector current characteristics and confirm the absence of deleterious boron out-diffusion effects.

Incorporation of Sn on GaAs (111)A Substrates by Molecular Beam Epitaxy
S.J. HU,¹ M.R. FAHY,² K. SATO,³ and B.A. JOYCE²
1--School of Physics, Universiti Sains Malaysia, 11800 Penang, Malaysia. 2--IRC Semiconductor Materials, The Blackett Laboratory, Imperial College, London SW7 2BZ, United Kingdom. 3--Japan Energy Corporation, 10-1, Toranomon-2-Chome, Minato-ku, Tokyo 105, Japan.

KEY WORDS GaAs (111)A, molecular beam epitaxy (MBE), n-type dopant, Sn

Behavior of Sn as donor species in the MBE growth of GaAs on (111)A substrates has been investigated by varying the growth temperature from 460 to 620°C, As₄:Ga flux ratio from 4 to 25, and Sn concentration from 10¹⁶ to 10²⁰ atoms cm^-3. Secondary ion mass microscopy measurements show that Sn does not surface segregate on (111)A substrates under this growth condition, in contrast to that on (001) substrates. Sn is uniformly incorporated throughout the bulk of the grown layer for all samples, apart from the most highly doped ones. To increase the Sn carrier concentration on the (111)A substrates, the measured carrier concentration shows that doping should be carried out at a low growth temperature and/or high As₄:Ga flux ratio.

Defects in CdTe Single Crystals Grown by Very Fast Vapor Growth Technique
H. WIEDEMEIER¹ and G.H. WU²
1--Department of Chemistry, Rensselaer Polytechnic Institute, Troy, NY 12180. 2--Department of Chemistry, Rensselaer Polytechnic Institute, Troy, NY 12180. Present address: Institute of Physics, Chinese Academy of Sciences, Beijing, P.R. China.

KEY WORDS CdTe, defects, growth stability, single crystal, twins, very fast vapor growth

Single crystals of CdTe were obtained by the very fast vapor growth technique. The four major defects, namely, multi-grains, lamellar, lateral, and micro-twins, which are fatal to the single crystallinity, were eliminated or limited by increasing the growth stability. This investigation indicates that the latent heat under high growth rate conditions should not be neglected. A model is developed to explain the effects of latent heat on the growth stability at the interface. A relationship between crystal morphologies and growth conditions was established. It strongly suggests that the above defects are growth stability related. The origin of twinning, dominated by growth stability, is discussed in this paper.

Deep Level Transient Spectroscopy Assessment of Silicon Contamination in AlGaAs Layers Grown by Metalorganic Vapor Phase Epitaxy
E. CALLEJA,¹ F. SANCHEZ,¹ E. MUÑOZ,¹ P. GIBART,² A. POWELL,³ and J.S. ROBERTS³
1--Departamento de Ingeniería Electrónica, E.T.S.I. Telecomunicación, Universidad Politécnica, Ciudad Universitaria, 28040 Madrid, Spain. 2--Centre de la Recherche sur l'Heteroepitaxie et ses Applications, Parc Sophia Antipolis, Av. Bernard Gregory, 6560 Valbonne, France. 3--SERC III-V Central Facility, Electrical Engineering Department, University of Sheffield, Mappin St., Sheffield S1 SJD, United Kingdom.

KEY WORDS AlGaAs, metalorganic vapor phase epitaxy (MOVPE), Si contamination

A systematic silicon contamination has been detected by deep level transient spectroscopy in undoped and n-type doped (Te, Se, Sn) AlGaAs layers, grown in two different metalorganic vapor phase epitaxy reactors. DX center generation by substitutional donors, with very specific capture and emission thermal barriers (fingerprints), is the key to unambiguously identifying their presence, with detection limits well below the standard secondary ion mass spectroscopy capability. We comment on the potential sources of Si contamination (most common in this epitaxial technique), and on the relevance of such contamination to interpreting correctly experimental data related to the microscopic structure of DX centers.

Preparation of (Pb_0.88La_0.12)TiO₃ Thin Films for Dynamic Random Access Memory by Low Pressure-Metalorganic Chemical Vapor Deposition
SEAUNG-SUK LEE and HO-GI KIM
Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, 373-1 Kusong-dong, Yusong-gu, Taejon, 305-701 Korea.

KEY WORDS Dynamic random access memory (DRAM), PLT thin film, MOCVD, ferroelectric

La-modified lead titanate (PLT) thin films were prepared by hot-wall type low pressure-metalorganic chemical vapor deposition method. Pb(dpm)₂, La(dpm)₃, and titanium tetraisopropoxide were used as source materials. The films were deposited at 500°C under the low pressure of 1000 mTorr and then annealed at 650°C for 10 min in oxygen ambient. Sputter-deposited platinum electrodes and 180 nm thick PLT thin films were employed to form MIM capacitors with the best combination of high charge storage density (26.7 µC/cm² at 3V) and low leakage current density (1.5 x 10^-7 A/cm² at 3V). The measured dielectric constant and dielectric loss were 1000~1200 and 0.06~0.07 at zero bias and 100 kHz, respectively.

Passivation of Hg_1-xCd_xTe by Photochemical Native Oxidation: Quantitative Analysis of Oxide Composition
G.H. WINTON, L. FARAONE, and N. WARRINGTON
Department of Electrical and Electronic Engineering, The University of Western Australia, Nedlands, W.A. 6009, Australia.

KEY WORDS Hg1-xCdxTe, photochemical oxidation, Rutherford backscattering spectroscopy (RBS), x-ray photoelectron spectroscopy

The composition of photochemically grown native oxides on Hg_1-xCd_xTe (x = 0.3) has been analyzed and depth profiled using x-ray photoelectron spectroscopy and Rutherford backscattering spectroscopy. The oxide films were grown in either N₂O or O₂ ambients, and differences in the oxidation process were examined by varying the time and temperature of oxide growth. Under all growth conditions, oxides grown in an O₂ ambient exhibited a higher Hg concentration in the bulk oxide region when compared to N₂O grown oxides. The Hg/Te ratio of all the oxides was found to be less than the starting Hg_1-xCd_xTe substrates and, in some cases, this may be leading to an accumulation of Hg in the oxide/Hg_1-xCd_xTe interface region. For growths at higher temperatures (~75°C), the excess Hg was seen to move from the oxide/Hg_1-xCd_xTe interface region to the oxide surface. In O₂ ambients, the Hg accumulated at the surface of the oxide whereas for growths in N₂O, it was lost to the ambient. Previous results on photochemical oxidation of Hg_1-xCd_xTe show an inverse relationship between oxide growth rate and temperature.¹ Evidence obtained in this study from oxide compositions, depth profiles and annealing at higher temperatures, suggest that this relationship between oxide growth rate and temperature is primarily due to temperature induced differences in the oxidizing ambient, and not the result of a change in the film growth mechanism due to changing diffusion characteristics with temperature.