Formation energies of neutral and charged oxygen vacancies in MgO, ZnO, Al₂O₃, In₂O₃ and SnO₂ have been calculated by a first principles plane-wave pseudopotential method. Two kinds of polymorphs, i.e., an ordinary phase and a high-pressure or an hypothetical negative pressure phase, have been chosen in order to see the effects of crystal structure. Supercells composed of 54 to 96 atoms were employed, and structural relaxation around the vacancy within second nearest neighbor distances was taken into account. Defect levels were obtained from the difference in total energies of the neutral and charged supercells that contain a vacancy. Ionization energies of the vacancy were calculated as the difference in the bottom of the conduction band and the defect levels. They are found to be proportional to band-gaps with a factor of approximately 0.5, which are prohibitively large for the n-type conduction.

View full abstract

The Si–K and Si–L_2,3 edges of the electron energy-loss near-edge (ELNES) spectra of a model of an extended inactive defect in Si with no dangling bonds were calculated using an ab-initio method which includes the electron-hole interaction. In this method, atom-by-atom excitation is possible. The calculated results are discussed in the context of the subtle structural differences in the local atomic environment. Comparison of the results with measured data shows satisfactory agreement. The method can be applied to other more complicated defective systems such as grain boundaries and interfaces for effective materials characterization.

View full abstract

First-principles relativistic configuration-interaction (CI) calculations of cation L_2,3-edge electron-energy-loss near-edge structures (ELNES) of SrTiO₃, NiO and CaF₂ have been carried out based on the direct approach where the ground state (GS) configuration and the excited state (ES) configuration were calculated simultaneously by direct diagonalization of the many-electron Hamiltonian. The obtained theoretical spectra were compared with our recent results based on the indirect approach where the GS configuration and the ES configuration were calculated separately and their energy separation was adjusted to the value predicted by the Slater’s transition-state calculation. Although both approaches well reproduced the overall features of the experimental spectra, the direct approach tend to slightly overestimate the absolute transition energies. In the case of Ti L_2,3-edge of SrTiO₃, however, the absolute transition energies calculated by the direct approach were comparable to those by the indirect approach and were also in good agreement with experimental values, indicating sufficient inclusion of electron correlations through pure configuration interactions.

View full abstract

First-principles plane-wave pseudopotential calculations have been conducted to investigate Co impurities and native defects associated with oxygen excess in ZnO. The electronic states and formation energies are evaluated using the total energies of supercells. The electronic states indicate that Co impurities are donor-like, while native defects associated with oxygen excess are acceptor-like. Among the native defects, Zn vacancies are likely to be a dominant species in view of the much lower formation energy than those of the others. Calculations for Co impurity-Zn vacancy complexes imply that Co impurities stabilize Zn vacancies and hence oxygen excess in their vicinities. This is suggested to be a role of Co impurities at grain boundaries in ZnO varistor ceramics.

View full abstract

The optimized geometries and formation energies of hydrogen in the perovskite-type oxide, SrZrO₃, have been calculated using the density functional theory under the generalized gradient approximation (GGA). The stable hydrogen sites in the oxide are examined with special interests. It is shown that the hydrogen is tightly bound to an oxygen ion and located slightly outside the ZrO₆ octahedron in SrZrO₃. The O–H bond orientates towards the second-nearest-neighbor oxygen ion, which induces large local displacements of adjacent oxygen ions. For example, the nearest-neighbor oxygen ions tend to approach the hydrogen. The addition of an acceptor dopant, yttrium, modifies such local distortion of oxygen ions around the hydrogen as to increase protonic conductivity. In addition, the calculated hydrogen formation energy shows that the hydrogen behaves as a shallow donor in SrZrO₃. The fundamental principles of doping and atmosphere control in proton conducting oxides are also given in view of energetics.

View full abstract

Positron states for the bulk and vacancy in intermetallic compounds CoAl and CoTi crystallizing in the B2 structure have been calculated using the DV-Xα electronic structure calculation and compared with experimental results. The calculated positron lifetimes in CoAl can explain the compositional dependence of the positron lifetimes in CoAl. The calculated positron lifetime at the Ti vacancy in CoTi is longer than that at the Co vacancy. The charge transfer from Ti to Co leads to the difference of the localization of the positron wave function at the vacancy. At the Co vacancy site surrounded by the positively charged Ti atoms, the positron wave function does not localize as well as at the Ti vacancy site, which decrease the positron lifetime. The calculated results suggest that the positron lifetimes in intermetallic compounds can not be directly correlated to those of the pure component and the charge transfer between the constituent atoms affects the localization of the positron wave function.

View full abstract

First principles calculations by a plane-wave basis pseudopotential method have been made for Mg, Ca, and Sr-doped lanthanum chromites (LaCrO₃). A supercell composed of 40 atoms of a high temperature phase of LaCrO₃ having a cubic-perovskite structure has been used. Solution energies of a neutral solute and a compensated solute by an oxygen vacancy were systematically computed. They were obtained for two kinds of cation sites and four thermodynamical conditions with different chemical potentials of constituent atoms. Mg shows lowest solution energy when it substitutes for the Cr ion. On the other hand, both Ca and Sr prefer to be located at La site. The charge neutral states are preferred by all of them. The results are consistent to experimental results regarding the site preference and the charge state.

View full abstract

Migration energies of Li ion in Li₃N, Li₂X (X=O, S, Se, and Te) and LiX (X=F, Cl, Br, I) via vacancy mechanism have been calculated by first principles pseudopotential method using plane-wave basis. The energy was obtained as the difference in total energies of supercells by two separate calculations; one with a Li⁺ ion at the normal point and the other with a Li⁺ ion at the saddle point. Positions of atoms within the second nearest neighbor of the jumping ion were fully relaxed. Two kinds of diffusion paths, i.e., direct and indirect jumps, were considered. Results show rough agreement with available experimental data. The migration energies for the indirect jump in both halides and chalcogenides show a tendency to decrease with the increase in the periodic number in the Periodic table. This is consistent with the widely accepted view. However, the migration energies for the direct jump of chalcogenides do not obey the rule. Comparison of two polymorphs of LiF implies that not only the anionic species but also the crystal structure plays an important role in determining migration energy.

View full abstract

X-ray fluorescence holography (XFH) is a promising method for determination of a local environment around a particular element. Since the holographic signal is about 0.3% of the background isotropic fluorescent radiation, it takes a few months to record a set of complete XFH data using a conventional energy dispersive detector. In order to overcome this difficulty, we designed an XFH setup with a combination system of a cylindrical LiF analyzer and an avalanche photo diode (APD) for bulk samples, and with a multi-element SSD for impurity samples. The holography experiments of an Au single crystal and Zn atoms doped in a GaAs wafer performed with these systems at the synchrotron radiation facility SPring-8 enables us to obtain high quality hologram data within a practical measurement time.

View full abstract

The temperature dependences of the nuclear-electric-quadrupole frequency ω_Q of ¹¹⁷In and ¹¹¹Cd doped in ferroelectrics LiTaO₃, Li_1−xIn_x⁄3TaO₃ with x=0.2, and Li_1−xCd_x⁄2TaO₃ with x=0.167, measured by the perturbed-angular-correlation technique, indicate that In behaves like Li; further, in a certain temperature range above the phase-transition temperature, a local system consisting of In (Li) and oxygen ions is very stable, by taking resonance structures, whereas Cd does not behave like Li, and a system consisting of Cd and oxygen ions does not take resonance structures, because the ionic size of Cd is large. It is considered that the resonance structures are due to a disordering of the oxygen ions, and that the order-disorder of oxygen ions is the driving mechanism for ferroelectric instability in the LiNbO₃–LiTaO₃ system.

View full abstract

The feasibility of the determination of local atomic structure was tested by computing Ge X-ray fluorescence hologram patterns of a Ge film 3 atomic layers thick on a Si substrate. An atomic image obtained from a single-energy hologram pattern exhibits many small oscillations and a precise atomic image is difficult to evaluate. By summing plural images reconstructed from holograms obtained at several different energies, the final atomic image is improved due to the disappearance of the oscillations. The full-width at half-maximum of peaks of the atomic image reconstructed from six holograms calculated at 27.0–29.5 keV with 0.5 keV steps is 0.03 nm. This clearly demonstrates that the XFH method has strong potential as an experimental tool for evaluating the local atomic structure.

View full abstract

The hyperfine interaction at ¹⁴⁰Ce nucleus in CeRu₂Ge₂ has been investigated, using the ¹⁴⁰Ce Time-Differential Perturbed Angular Correlation (TDPAC) method. ¹⁴⁰Cs ions, which are fission products from ²³⁵U, have been implanted into a single crystal of CeRu₂Ge₂ at room temperature. ¹⁴⁰Cs decays to ¹⁴⁰Ce 4+ intermediate state through ¹⁴⁰Ba with the life-time of 12.752 days and ¹⁴⁰La with a life-time of 1.6781 days. TDPAC spectrum at 4.2 K clearly shows the existence of the magnetic perturbation, which is absent at room temperature. Larmor frequency observed at 4.2 K is 1.32(1) Grad/s corresponding to the hyperfine magnetic field of 24.2(2) T at ¹⁴⁰Ce nucleus. Magnetization measurements of CeRu₂Ge₂ performed by using the SQUID magnetometer revealed the magnetic moment of 1.848 μ_B which indicates that the hyperfine coupling constant of Ce is 13.1(3) T/μ_B.

View full abstract

The order-disorder transformation and lattice defects in Ni₃Fe have been studied by positron lifetime measurements. Anomalous vacancy-generation during ordering transformation, which was originally found on the ordering process of super-cooled disordered Cu₃Au, has been confirmed on the ordering transformation of Ni₃Fe. Disordered fcc solid solution of Ni₃Fe was brought to room temperature by quenching the specimen from temperatures above the order-disorder transformation point T_C. The ordering process into L1₂ structure was promoted by heating the sample isochronally or isothermally. It has been found that vacancies are generated in both heating processes, i.e., during the ordering process of super-cooled disordered Ni₃Fe. Generated vacancies are not stable up to T_C and annealed out at temperatures below T_C.

View full abstract

The pre-martensitic phenomena in Ni-rich NiTi alloys have been studied by means of positron lifetime spectroscopy and electrical resistivity measurement. We have found anomalous positron lifetime changes above the martensitic transformation temperatures of Ni₅₀Ti₅₀–Ni₅₁Ti₄₉ alloys. Positron lifetime increases anomalously with decreasing temperature even at temperatures 100 degrees higher than M_s of Ni₅₁Ti₄₉ alloys. Almost the same change in positron lifetime has been observed on the subsequent heating run with little hysteresis. We have also found the anomalous behavior of positron lifetime and negative temperature dependence of electrical resistivity even in NiTi alloys that do not exhibit martensitic transformation. The observed pre-martensitic changes of positron lifetime and electrical resistivity directly reflect the intrinsic electronic-structural changes of NiTi alloys that cause so-called pre-martensitic phenomena.

View full abstract

Lattice defects induced by initial hydriding and their effect on residual hydrogen content in LaNi₅ have been studied by means of positron lifetime spectroscopy and hydrogen thermal desorption measurement. Component analyses of positron lifetime spectra show that surprising amount of vacancies together with dislocations are generated by the initial hydriding. Vacancy migration in LaNi₅ after hydrogen desorption at room temperature is observed around 423–673 K, while dislocations in LaNi₅ are much more stable. Hydrogen thermal desorption measurement shows that the release of residual hydrogen occurs mainly in the temperature range from 450 to 650 K, and it ceases at about 800 K . The release temperature of residual hydrogen closely corresponds with the temperature of vacancy migration and annihilation in LaNi₅. Residual hydrogen in LaNi₅ is most likely trapped by vacancies and vacancy clusters, which are induced by hydriding.

View full abstract

The positron lifetimes in the B2 intermetallic compounds Co_100−XAl_X (X=46.9 to 54.1) water-quenched, air- and furnace-cooled from high temperatures have been measured to reveal their defect structures. The positron mean lifetime depends entirely on the composition, irrespective of the cooling rate, and steadily increases from 161 to 180 ps as the Al concentration increases from 46.9 to 54.1 at%. A water-quenched Co_53.1Al_46.9 was isochronally annealed for 900 s successively at 25 K intervals up to 1073 K, but its positron lifetime remained unchanged by the annealing. These results clearly show that vacancies are formed not only on the Co-sites but also on the Al-sites, and the total vacancy concentration is more than 10⁻⁴ in both the Co-rich and the Al-rich CoAl. It was furthermore found that the population ratio of Al-vacancies to Co-vacancies gradually increases with the Co concentration, and the number of Al-vacancies is comparable with that of Co-vacancies in the vicinity of stoichiometric composition.

View full abstract

A ¹⁴⁰La(I^π=3⁻, T_1⁄2=40.3 h)-doped layer has been produced in CaB₆ by means of radioactive isotope (RI) beam technique: ¹⁴⁰Cs(I^π=1⁻, T_1⁄2=63.7 s) was implanted into CaB₆ and the radioactive equilibrium of ¹⁴⁰Ba–¹⁴⁰La was achieved. The concentration of La in CaB₆ was La/Ca∼0.001 and ∼0.005. Obtained TDPAC spectra of the 2083 keV level of ¹⁴⁰Ce (I^π=4⁺, T_1⁄2=3.4 ns, μ=+4.35±0.10μ_N) followed by the β decay of ¹⁴⁰La showed the existence of hyperfine magnetic fields: B_hyp=−15.0±0.5 T and −1.00±0.15 T for La/Ca∼0.001 and B_hyp=−1.51±0.12 T for La/Ca∼0.005.

View full abstract

We carried out molecular dynamics simulation of amorphous silicon nitride containing boron and carbon, in order to investigate the short-range atomic arrangement and diffusion behavior. In amorphous Si–B–N, boron atoms are in a nearly threefold coordinated state with nitrogen atoms, while boron atoms in amorphous Si–B–C–N have bonding with both carbon and nitrogen atoms. Carbon atoms in Si–B–C–N are also bonded to silicon atoms. The self-diffusion constant of nitrogen in Si–B–N becomes much smaller than that in amorphous Si₃N₄. Also, amorphous Si–B–C–N exhibits smaller self-diffusion constants of constituent atoms, even compared to Si–B–N. Addition of boron and carbon is important in decreasing atomic mobility in amorphous Si–B–C–N. This may explain the increased thermal stability of the amorphous state observed experimentally.

View full abstract

We have succeeded in synthesizing ZnS films on substrates such as glass or Si wafer by using CBD (chemical bath deposition) method. These semiconductor films were photocatalytic and showed high productivity of hydrogen gas from the solution containing HS⁻ ions. From observation of SEM/TEM, it was found that the cluster film consisted of 100 nm sized agglomerated particles composed of nano-crystals. The result of EDS analysis for the cross-section of the film prepared by using FIB showed the presence of small amounts of oxygen along with zinc and sulfur. The information of the local quantum structures obtained from EXAFS/XANES analysis suggested that the cluster film has unique atomic-distribution and metastable electronic state different from bulk ZnS and ZnO. Based on the results of the above analysis, the relation between local quantum structures and properties was proposed.

View full abstract

Cross sections and plan views of LiNb_0.5Ta_0.5O₃ films were investigated mainly by high-resolution transmission electron microscopy. These films were deposited on (0001) sapphire substrates by thermal plasma spray chemical vapor deposition method at various feeding rates of liquid raw materials. It was found that the crystallinity and the preferential orientation of the LNT films depend on the feeding rate. The LNT film formed at the feeding rate of 7 mL/min was epitaxially grown on the substrate, and the orientation relationship between the film and the substrate was (0001)_LNT⁄ ⁄(0001)_sapphire, [11\\bar20]_LNT⁄ ⁄[11\\bar20]_sapphire. The LNT films fabricated at the higher and the lower feeding rate were polycrystalline. These films included twin crystals and other phases such as Li(Nb, Ta)₃O₈. The calculations based on the coincidence of reciprocal lattice points revealed that the epitaxial orientation relationships observed by transmission electron microscopy satisfied the geometrically optimal coherency across the interface.

View full abstract

The atomic structures of Mo–Ni alloys electrodeposited from aqueous solutions containing various molybdenum and nickel concentrations were studied by X-ray diffraction. The films are arranged with respect to molybdenum concentrations in the films. That is, the Group I (Mo<10%), the Group II (20%<Mo<30%) and the Group III (Mo≈40%). In the Group I, the structures of the films are fcc crystals of nickel solid solutions. The films in the Groupes II and III are amorphous. The hydrogen evolution as an electrode in alkaline water electric cell was also compared. At about 20% molybdenum, the nickel film shows the maximum hydrogen evolution.

View full abstract

A new method to synthesize fullerene and sulfur compounds, C₆₀S₁₆ and C₇₀S₁₆ compounds of about 1 mm length, has been developed. The C₆₀S₁₆ crystal is a C-centered monoclinic structure of a=2.0874(1) nm, b=2.1139(1) nm, c=1.05690(6) nm and β=111.9 degree. The C₇₀S₁₆ compound has a primitive monoclinic structure of a=1.5271(2) nm, b=1.49971(7) nm, c=2.18024(9) nm and β=109.791(1) degree. The crystalline structure of these compounds is constructed by fullerene and S₈-rings. A very small amount of charge-transfer between fullerenes and sulfur rings are expected by first-principle calculation.

View full abstract

Growth of a TiO₂ film on a Si wafer by CVD method was studied for a new design of chemically stable device which is useful in the visible region and increases the efficiency of the photocatalytic decomposition process of water. This design is made up of a monolithic structure of a TiO₂ thin film and a Si solar cell. The TiO₂ film was deposited on a Si substrate by the vapor phase reaction of Ti(OPrⁱ)₄ with water at temperatures from room temperature to 473 K. At room temperature, a rough TiO₂ film with a nano-sized cone shape deposit was observed by AFM. The smooth TiO₂ film of ∼10 nm thickness was obtained by the reaction of 80 Pa Ti(OPrⁱ)₄ and 493 Pa water vapors at 473 K for 10.8 ks. The surface and the thickness of the film became rougher and thicker, respectively, with increasing pressure of Ti(OPrⁱ)₄.

View full abstract

Grain boundary structure and current–voltage (I–V) characteristics were investigated for Nb-doped SrTiO₃ bicrystals having small angle tilt boundaries with misorientation angles of 2^° and 4^° against [001]. The bicrystals were fabricated by using hot-joining technique at 1400°C for 10 h under a pressure of 0.4 MPa. High resolution transmission electron microscopy study revealed that the joined boundaries are free from any secondary phases such as amorphous phases even on an atomic scale. The structures of the two boundaries are composed of edge type dislocations whose Burgers vector is [010]. But the density of boundary dislocations differs between the two boundaries. They exist at an interval of 10 nm in the 2^°-boundary and 5.2 nm in the 4^°-boundary. On the other hand, it was found that non-linearity in I–V relation across the boundary increases with an increase in the misorientation angle. This result clearly indicates that the potential barrier height is closely related to the density of boundary dislocations in the case of small angle type boundaries.

View full abstract

The atomic structures and surface morphologies of three types of GaN films were investigated by high voltage atomic resolution microscopy (HVARM). By HVARM, each atomic column of Ga and N could clearly be resolved and the polarity of the film and inversion domains could be directly determined. The GaN film was grown on a sapphire substrate by molecular beam epitaxy (MBE) after nitridation of the sapphire surface. Inversion domains (IDs) crossed the whole film to the surface and made small pyramids on the surface. The small pyramids had Ga-polarity and the rest had N-polarity. A GaN film with In exposure during film growth had an almost Ga-polarity flat surface. In exposure process reduced the density of inversion domains that have a N-polarity. While a GaN film grown on an AlN buffer layer was unipolar, with a Ga-polarity. HVARM observation revealed that the density of the IDs determine the qualities and the polarity of the film.

View full abstract

In order to study interface boundary structures, a three dimensional Near-Coincidence Site lattice (Near-CSL) model based on Bollmann’s O-lattice method has been proposed. Previous studies have shown that intragranular α-phase precipitates in a Ti–22V–4Al alloy have a morphology of lath-like, with two well defined facet planes. A detailed TEM observation of the laths has been made in this study and the results correlated to the present analysis. Accommodation mechanism of misfit dislocations and the Burgers vector of the defects observed at the broad face and side facet plane of the intragranular α-phase precipitates have been determined. The structural ledges at the side facet, as predicted by the present analysis, is consistent with HRTEM observation. The Burgers vector of misfit dislocations at the broad face is in good agreement with that previously reported for a Zr–2.6Nb alloy, that has almost the same lattice parameter ratio as the Ti–22V–4Al alloy used in the present study. Although the Burgers vector of the defects on the broad face can be determined by HRTEM observation, the boundary plane is not well defined. However, it can be demonstrated that the interface boundary structure of the broad face can be rationalized from the results of both the Near-CSL analysis and the HRTEM observations.

View full abstract

Internal friction originated from sliding and diffusional flow along grain boundaries was monitored (as a function of both temperature and damping frequency) in model SiC polycrystals with an intergranular SiO₂ glassy film. Emphasis was placed on the change of the internal friction characteristics upon additional Ca cation. It is expected that Ca cation segregates to the intergranular glass phase and modifies its network structure. The presence of a relaxation peak of internal friction due to grain-boundary sliding enabled quantitative evaluation of the activation energy for viscous flow of the intergranular glass and the related viscosity magnitude. A peak-analysis procedure, according to the peak-shift method (i.e., monitoring peak shift upon damping frequency change), is proposed, which quantitatively revealed the activation energy for viscous flow in various impurity-doped intergranular glasses. The presence of chemical gradients at grain boundaries, namely the presence of families of boundaries within the SiC polycrystal with different chemical characteristics, has also been analyzed by taking into account the dependence of peak morphology on damping frequency.

View full abstract

Internal friction of alumina polycrystals with engineered grain boundaries was measured in the low frequency range of the torsional forced-vibration. The maximum shear stress amplitude was 50 MPa. Owing to the very high temperature reached during these experiments (1900 K), internal friction spectra could be detected which reveal new important features of the intrinsic rheological behavior of alumina grain boundaries. The monotonically rising internal friction background and the torsional creep behavior, both of a diffusive origin, were systematically characterized upon modifying the grain-boundary structure of alumina with the addition of selected cation dopants (e.g., Titanium and Lutetium). The combined characterizations of internal friction spectrum and torsional creep rate enabled us to precisely assess the role of different cation dopants on “locking” or promoting the diffusive grain-boundary flow, which governs polycrystal deformation.

View full abstract

Orientation controlled alumina bicrystals were fabricated by a hot joining technique at 1773 K in air to obtain [0001] symmetric tilt boundaries including coincidence grain boundaries. The grain boundary energies were measured by the thermal grooving technique, and they were found to strongly depend on the grain boundary character. Atomic structures of those grain boundaries were observed by high-resolution electron microscopy (HREM). It was found that the atomic structures did not always correlate with the grain boundary energies. This finding indicates that the grain boundary energy originates not only from the atomic bonding on the grain boundary but also from the strain of the grain interior in the vicinity of the boundary. The grain boundary sliding was also investigated by the high-temperature creep test. As the results, the grain boundaries with the same energy showed different sliding behavior. The occurrence of grain boundary sliding is considered to depend on the atomic bonding of a grain boundary.

View full abstract

High-temperature plastic deformation characteristics such as flow stress and elongation to failure in polycrystalline Al₂O₃ with an average grain size of about 1 \\micron is remarkably changed by the doping of 0.1 mol% YO_1.5, SiO₂, TiO₂ or ZrO₂ at 1400°C under an initial strain rate of 1.2×10⁻⁴ s⁻¹. The difference in the flow stress and the tensile ductility is considered to be originated from change in the grain boundary chemistry in Al₂O₃ due to segregation of the dopant cation in the vicinity of the grain boundary. A change in the chemical bonding state in the cations-doped Al₂O₃ is examined by first-principle molecular orbital calculations using DV-Xα method based on [Al₅O₂₁]²⁷⁻ cluster model. Chemical shift observed in electron energy-loss spectra (EELS) at the grain boundary due to the dopant segregation can be roughly reproduced by the molecular orbital calculations. A correlation is found between the flow stress and product of net charges of aluminum and oxygen ions. Moreover, the tensile elongation to failure seems to correlate with bond overlap population between Al and O. The change in the chemical bonding strength at the grain boundary must dominantly affect to the grain boundary diffusion and bond strength at the grain boundary, and thus seems to be an important factor to determine the plastic flow behavior in polycrystalline Al₂O₃.

View full abstract

The effect of Cu local interconnect on intrinsic gate delay is analyzed. Beyond the 70-nm node, the intrinsic gate delay is limited by the local interconnect by taking account of electromigration in fine routings. Developments of materials and processes with improved electromigration for local interconnect are required.

View full abstract

This paper describes a 0.13-\\micron CMOS made by using highly reliable copper and SiLK^T.M. (DOW CHEMICAL) interconnection technologies. We propose a hybrid interlayer structure with SiLK^T.M. at the trench level and SiO₂ at the via level to improve electrical properties, mechanical strength, and reliability. Using these technologies, we made a fully functional 1.5-Mbit SRAM macro and investigated the reliability of its copper wiring in terms of electromigration.

View full abstract

A CVD-WN film is deposited as a barrier metal for copper interconnection using thermal CVD (Chemical Vapor Deposition) of WF₆/NH₃/SiH₄ gases. Deposition of CVD-WN film with a resistivity of less than 300 \\microΩ·cm at a temperature of 400°C or less has been realized for the first time in the world. The deposited WN film is proved to be excellent in barrier properties and able to prevent Cu diffusion even with the film thickness of 6 nm as well. It is also proved that the CVD film is superior to a sputtered barrier metal in coverage and ECD (Electro-chemical deposition) filling properties. XPS analysis showed that adhesion of the CVD film to low-k materials is deteriorated by the existence of F at the interface between the WN film and low-k. Chemical surface pretreatment for low-k materials proves to restrain the pile-up of F at the interface and improve adhesion.

View full abstract

The role of additives in copper electroplating baths in the damascene process has been investigated. We proposed a bottom-up filling model and confirmed it by comparing the experimental and simulation results. Janus Green B and Basic Blue 3 which absorb on the copper surface and suppress copper deposition were examined for additive use to improve filling capability. Damascene copper grew uniformly in the bath that contained Basic Blue 3. But it grew preferentially from the bottom of the trench for Janus Green B. Addition of Janus Green B produced a continuous concentration gradient in the sub-micron trench when the additive’s diffusion rate and consumption rate on the copper surface were well balanced. We estimated filling profiles from numerical simulation using parameters that were determined by an electrochemical method. These profiles agreed well with the experimental results.

View full abstract

Copper sputtering method for fabrication of high performance logic LSI was studied. Extension of target to substrate distance is effective to improve step coverage of sputtered film combined with reduced operation pressure. Step coverage of low pressure long throw sputtering method also strongly depends upon the feature size of trenches and holes which are formed on silicon wafer. Sub-micron holes and trenches are successfully filled with copper by using this sputtering process followed by re-flow annealing process. Hydrogen annealing process prior to the sputtering deposition on via openings is also investigated to realize good conductivity through the via. This process results in the reduction of copper oxide at the surface of copper film. Using these newly developed processes, 0.2 \\micron node BiCMOS LSI with 4 level copper interconnects was successfully fabricated and high performance of the copper interconnect system was clearly demonstrated.

View full abstract

We have carried out a detailed study of the relevant process techniques by using high-pressure annealing in order to embed Cu interconnections flawlessly into via holes or trenches, and have thereby identified a process technique increase the reliability of electrical circuits. In the present study, Cu interconnections were subjected to heat treatment in a high-pressure argon gas atmosphere (150 MPa) with the prospect that such a process might lead to optimized conditions for creating the embedded Cu interconnection by preventing the occurrence of minute voids, improving the distribution of poly-crystalline orientation and improving the adhesive strength between Cu and TaN (barrier layers), thereby increasing the reliability of Cu interconnections and consequently improve the yield. This paper describes the effects of the high-pressure annealing process for dual damascene Cu interconnections in some detail.

View full abstract

It is essential to decrease RC delay in order to increase the IC device speed. Cu is now adopted to decrease the resistivity of interconnects. Several low k materials have been studied to decrease the dielectric constant. Ag is one of the candidates after Cu as a low resisitivity interconnect metal although it has not yet been adopted in the actual products. It is also studied to decrease the effective dielectric constant values by considering the device configuration such as the cap layer only on the metal. Co–W–P is reported as one of the cap material candidates for Cu. Ni–B is proposed herein as the cap material for Cu and Ag. The study results confirmed that an Ni–B layer can be selectively deposited on Ag metal by the electroless plating method, using DMAB (dimethylamine borane) as the reducing agent; that a deposition rate of 150 nm/min and a B content of 3.2 at% can be obtained under conditions of pH 10 and 353 K; that an Ni–B layer with a 3.2 at%B content provides a barrier effect that prevents Ag and Cu diffusion through thermal processes (in this case, Ni–B remained crystalline before and after a thermal process); and that an Ni–B layer with a 13.5 at%B content does not provide a barrier effect that would prevent Cu diffusion through thermal processes (in this case, the Ni–B structure changed from amorphous to crystalline after the thermal process). Damascene interconnects with Ag metal and a Ni–B cap layer were also formed as part of the trial effort.

View full abstract

The resistivities of thin films and fine lines of copper (Cu) and aluminum (Al) were measured by a resistance ratio method (referred to as “RR method” hereafter), which measures the ratio of room-temperature resistance to liquid-nitrogen-temperature resistance. This method can predict resistivity exactly without the need for precise and detailed line or film dimension measurements. Thinner films and finer lines have higher resistivities in the case of both Cu and Al, with Cu showing larger resistivity increases (films and lines) than Al . From the present data, we estimated the electron mean free path of Cu to be 55 nm, which is close to most of the previously reported values, and that of Al to be 22 nm. Line resistivities depend not only on the line width but also on line thickness. We propose a simple equation for expressing line resistivity in terms of line thickness and width.

View full abstract

In order to understand the void formation mechanism in electroplated Cu interconnects used in Si-semiconductor devices, microstructure of Cu/CuO/Cu layered films which were prepared on the Si₃N₄/Si substrates by the sputter-deposition technique was observed by transmission electron microscopy (TEM) and scanning ion microscopy (SIM). A high density of macro and micro voids were observed in the samples annealed in atmosphere containing hydrogen, whereas no voids were observed in the samples annealed in Ar atmosphere. TEM observation suggested that a small amount of oxygen contained in the Cu films (even a native oxide layer) formed water vapor at elevated temperatures, causing formation of the micro-voids when the samples were annealed in hydrogen atmosphere. The present result suggested that the void formation in the electroplated Cu films was induced by existence of impurities such as oxygen in the Cu films, and that the void growth was strongly enhanced by annealing in hydrogen atmosphere.

View full abstract

A detailed investigation has been carried out into the evolution of grain size and grain orientation in electroplated Cu interconnection lines. The specimens were annealed to produce a range of different grain size distributions. Very accurate grain size distributions were obtained from extensive TEM observations of a large number of specimens, followed by computer tracing of grain boundaries. Although annealing causes the average grain size to increase, very small grains are found to persist in the distribution, even when the average grain size is large. These small grains have been investigated in detail to determine the reason for their thermal stability. In addition, the occurrence and redistribution of small micro-voids has been investigated for wafers with different grain sizes. These voids are believed to be associated with seams which are formed in the Cu lines during plating.

View full abstract

Stress voiding was investigated in damascene Cu lines embedded in Ta/TaN/SiO₂/Si. Microstructure was observed before and after heat treatment at 723 K using a focused ion beam (FIB) technique. The distribution of thermal stress was calculated using a three-dimensional finite element method (FEM). FIB observation revealed that slit-like voids were formed at trench shoulders both before and after heat treatment. FEM calculation indicated that a large shear stress concentration occurred at the voided sites. The coincidence between the FIB observation and the FEM calculation suggests that the slit-like voids were formed by shear-mode delamination of Cu from the Ta/TaN barrier layer.

View full abstract

The wear behavior of an A2218 alloy and an A2218/Al₃Fe composite fabricated by the plasma synthesis method was investigated under dry sliding conditions. The wear tests were carried out at a sliding speed of 0.2 m/s and at the load range of 3–600 N. The variation of wear rate with applied load is composed of two regions for both materials and the two regions belong to mild wear. On the basis of observation and analysis on the worn surface, subsurface and the wear debris, the wear mechanism acting on each region was identified. At low loads, oxidative wear is the main wear mechanism and at high loads, the wear is controlled mainly by delamination and adhesive wear mechanism and oxidative wear is an additional wear mechanism. The A2218/Al₃Fe composite exhibits an improved wear resistance at lower loads. On the contrary, at higher loads, due to the presence of the mechanical mixed layer on the worn surface of the A2218 monolithic alloy, both the monolithic alloy and composite exhibit similar wear rate.

View full abstract

The steady-state creep behavior of metal matrix composites was analyzed via consideration of two accommodation processes, diffusion and plastic, which are inevitable for the materials to continue creep deformation. The creep experiments were performed using a model material, in situ TiB fiber-reinforced pure α-Ti matrix composite, which has a good interfacial bonding, a moderate diffusional-accommodation rate and no fine oxide dispersions. A sigmoidal curve of strain rate and stress relation in a double logarithmic plot was observed, indicating the presence of three deformation regions: plastic-accommodation-control region, diffusional-accommodation-control region and complete diffusional-accommodation region, at high, middle and low stresses, respectively. The activation energies in the three regions were close to those of volume, interface, and volume diffusion of α-Ti, respectively.

View full abstract

The appearance and vanishing of the load transfer effect in the steady-state condition of composites’ creep were analyzed via consideration of two accommodation processes, diffusion and plastic, which are inevitable for composites to continue creep deformation. Creep experiments were performed using model materials, Ti–5, 15 and 20 vol%TiB in situ composites, which have moderate diffusional-accommodation-controlled creeps, good interfacial bonding and no fine dispersions. With higher strain rates than the diffusional-accommodation-controlled creep, the stress exponent of all composites was 4.5, similar to that of α–Ti, 4.3, and the strain rates decreased with increasing volume fraction of the reinforcements. Variations of the strain rate agreed with the prediction by the self-consistent potential method, and thus the load transfer effect was confirmed in this region. With strain rates near the diffusional-accommodation-controlled creep, the stress exponents of Ti–15 and 20 vol%TiB decreased to 2. With a strain rate lower than the diffusional-accommodation-controlled creep, the strain rates of the composites did not decrease with increasing volume fraction, and vanishing of the load transfer effect was observed.

View full abstract

We study the relation between microstructures of electrorheological (ER) fluids and their viscosity change by performing Brownian dynamics simulations of a model ER system both in a static state and in a simple steady shear. From large-scale three-dimensional simulations, it is found that (1) under no shear flow there are two principal phases in microstructural changes: first aggregation of particles into chains oriented along the field direction, and the subsequent slow coalescence of chains into columns, and (2) under a simple steady shear there are three stages in viscosity changes with increasing the field: Newtonian at a weak field, non-Newtonian at a moderate field, and Bingham plastic with yield stress at a high field.

View full abstract

The damage rate dependence of the yield stress change in a neutron-irradiated Fe–Cu model alloy was analyzed by a model calculation. The model was based on the rate theory, and focused on the description of the nucleation and growth of point defect clusters and copper clusters. The binding between copper atoms and vacancies and the effect of cascade damage which directly creates small point defect clusters were incorporated in this model. The instability of small point defect clusters caused by thermal dissociation was also included. From the result of the calculation, the yield stress changes were estimated using the Orowan model and the Russel-Brown model. As a result of this calculation, it was clarified that copper clusters are the main factor of yield stress change in almost all irradiation stages below 0.1 dpa. The contribution of copper clusters to yield stress change increased with decreasing damage rate. The nature of the damage rate dependence is not affected by the copper-vacancy binding but by the sink strength, which changes dynamically throughout irradiation.

View full abstract

The structural features of metallic glasses depend on the cooling rate of the melt. The cooling rates for casting processes which are typically employed for preparation of bulk metallic glasses are suggested from microstructures of an eutectic Al–33 mass%Cu model alloy. The interlamellar spacing λ of eutectic Al–CuAl₂ rod-shaped specimens of 50 mm in length and 2 to 5 mm in diameter has been determined by optical microscopy and scanning electron microscopy. From the measured interlamellar spacings ranging from λ=0.18 to 0.5 \\micron, the solidification front velocities and the cooling rates at different positions in the as-cast samples are derived. The decisive effect of the diameter of cast rods is confirmed. For a centrifugal casting technique, the maximum cooling rate decreases from 730 to 95 K/s when the diameter increases from 2 to 5 mm. Moreover, it is revealed that the local cooling rates decrease significantly from the bottom towards the top of the rods. From the estimated cooling rates at a fixed rod diameter the centrifugal casting technique is assessed as superior to other methods applied, namely copper-mould casting and suction casting. The estimated cooling rates are compared with literature data for glass-forming alloys.

View full abstract

Martensitic transformation temperatures and the two-way shape memory effect (TWME) in ductile Cu–Al–Mn alloys deformed at room temperature were investigated using differential scanning calorimetry (DSC) and the bending test. The M_s and A_f temperatures were found to increase slightly with increasing degree of pre-deformation, while the M_f and A_s temperatures decreased. The TWME is strongly influenced by the M_s temperature and the grain size of the specimens, i.e., the highest degree of TWME is always obtained in the specimens with M_s≈−100°C and the TWME increases with increasing ratio of grain size to specimen width, where the maximum value of TWME=3.2% was obtained in the present study. In situ observation of the surface relief at several temperatures on the deformed specimens was also performed and it was found that some restricted martensite variants grow and shrink with temperature change without the typical self-accommodation microstructure in the specimens with large TWME.

View full abstract

The effect of the Al concentration and layer structure on the electrical and microstructural properties of TiAl Ohmic contacts for p-type 4H–SiC were investigated. The Al concentration was found to effect strongly on these contact properties, and the specific contact resistance of 1×10⁻⁵ Ω-cm² was obtained for the TiAl contacts with the Al concentration higher than 77 at% after annealing at 1000°C. However, agglomeration of Al was observed after annealing, which caused the rough surface morphology. On the other hand, the TiAl contacts with the Al concentration lower than 75 at% showed non-Ohmic behavior and had smooth surface morphology after annealing at 1000°C. It was found from X-ray diffraction analysis that the interface structures were strongly influenced by the Al concentrations of the TiAl contacts. For the TiAl contact with high Al concentration, formation of Al₃Ti, Ti₃SiC₂, and Al₄C₃ was observed, and for the TiAl contact with low Al concentration, formation of Al₃Ti, Ti₃SiC₂, and Ti₅Si₃ was observed. It was concluded that the electrical property of the TiAl contact was not affected by the number of TiAl layers and was very sensitive to the Al concentration in the TiAl contacts.

View full abstract

The new chemical layered structure, referred to as a chemical stripe structure, was found as a metastable state in the (bcc→hcp+C15) metastable phase separation in the Ti–Cr alloys. Note that a metastable separation is here defined as a separation characterized by a positive value of the second derivative of Gibbs free energy with respect to chemical composition. The experimental data showed that the layered structure had periodicity of 4×d₀₀₂, about 0.72 nm, and chemical ordering of Ti:Cr=3:1, and that a lot of antiphase boundaries with phase shifts of π⁄2 and π were involved in the structure. The coherent length of the layered structure was estimated to be about 15 nm along the modulated direction. In addition to these features, it was found that the layered structure was formed from the local bct state as another metastable state, not directly from the zone structure consisting of one Cr layer. On the basis of these data, the physical origin of the formation of the chemical layered structure is also discussed.

View full abstract