Band 94, Heft 3

The thermally grown aluminum oxide formed on binary β-NiAl alloys has a propensity to spall from oxidation-induced cavities. Microstructural observations of the spalling failures reveal that the oxide fails as a result of a sequential process. Initially, cavities nucleate and grow in the alloy at its interface with the growing oxide. On cooling, the oxide buckles above the cavities and separates from the remaining alloy by propagation along the intact oxide/alloy interface, allowing adjacent buckles to link together until the mechanical condition for macroscopic spalling occurs and the oxide spalls away from the alloy. The buckling and spalling is motivated by the compressive stress in the oxide on cooling as a result of the thermal expansion mismatch with the alloy. Since failure requires fracture along the oxide/alloy interface, it is susceptible to segregation of impurities and, possibly, sub-critical crack growth.

The high-temperature stability of zirconias co-doped with Y and Gd and synthesized as metastable single phases within the equilibrium tetragonal + cubic field is investigated. The motivation arises from strategies to enhance the insulating efficiency of thermal barrier systems by co-doping the conventional ZrO 2 -7.6 % YO 1.5 composition with trivalent rare-earth cations. The issue is the partitioning of the metastable tetragonal (t′) or cubic (c′) phases into the equilibrium t + c phases dictated by the phase diagram, whereupon the tetragonal phase becomes susceptible to the disruptive monoclinic transformation. The experiments are based on compositions synthesized by precursor pyrolysis, all of which yield initially supersaturated single-phase solid solutions. It is found that t′ stabilized by Gd alone is significantly less resistant to partitioning at high temperature than its Y counterpart with the same amount of stabilizer. However, modest substitution of Gd for Y does not degrade the stability, and may improve it in some cases. Increasing the total amount of stabilizer generally enhances phase stability. The roles of thermodynamics and kinetics on the relative stability of the system are discussed.

The lateral extension of alumina scales as they grow at high temperature can induce substantial deformation of the substrate, especially under thermal cycling. The phenomenon is relevant to the understanding of failure mechanisms in thermal barrier coatings, and is explored here using model FeCrAlY substrates, bare as well as coated with electron beam physically vapor deposited (EB PVD) yttria-stabilized zirconia. Model surface imperfections consisting of shallow grooves experience significant distortions that evolve over time. The effect is most pronounced under cyclic oxidation, absent the thermal barrier coating (TBC). Distortions around the grooves are considerably reduced when the TBC is present. A procedure for simulating thermally grown oxide (TGO) displacements has been used to predict the groove distortions. Those absent the TBC are shown to closely match those found experimentally for certain choices of TGO growth stress. It is also shown that the presence of an intact TBC entirely suppresses the distortions. Overlaying the simulated and measured TGO profiles absent the TBC reveals that, for consistency, the TGO growth stress should be lower on convex segments of the TGO than on flat segments, by about a factor 2.

Apart from reactive element additions, commercial FeCrAl alloys contain a number of poorly controlled or unavoidable impurities, which might significantly affect the material’s oxidation behaviour. In the present study it was found that magnesium, if present in a concentration of only 80 ppm, can substantially affect composition and morphology of the alumina surface scales. This was found for scales formed after short (a few minutes) as well as very long (up to 1000 h) oxidation treatments. Magnesium appeared to diffuse outward through the alumina scale and it became enriched in the outer part of the oxide scale, forming a layer of MgAl 2 O 4 spinel. An important effect associated with the Mg enrichment and spinel formation was the development of porosity within the scale which likely affects the scale growth rate and adherence, however, did not fundamentally change the sub-parabolic character of the scale thickening rate, which has previously frequently been reported for pure α-alumina scales.

The electric activity of grain boundaries in donor-doped SrTiO 3 bicrystals is studied using a combination of variable temperature impedance spectroscopy, current–voltage measurements and scanning probe microscopy. The dominant transport mechanism changes from thermionic emission to diffusive at about 220 K. The temperature dependence of interface capacitance is interpreted in terms of field-dependent dielectric constant in the vicinity of the interfaces. Simplified expressions for the temperature dependences of interface potential, charge and capacitance for a weakly nonlinear dielectric are derived to describe the experiment. The complex nature of these materials requires incorporation of such effects as nonlinear polarization and multiple simultaneous transport mechanisms.

When BaTiO 3 powder compacts with 0.2 mol% excess TiO 2 are sintered in air at 1250 °C, below the eutectic temperature, some grains grow abnormally to very large sizes with double twins in the 〈111〉 directions at their centers. The abnormal grains, elongated in the 〈111〉 directions, form mostly straight grain boundaries, lying on their {111} planes with the fine matrix grains. Some grain boundaries between the nearly (111) face of an abnormal grain and the matrix grains are faceted with some facet planes lying on either the (111) plane of the abnormal grain or (210) plane of the matrix grain. These flat grain boundaries, as observed by transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM), must be singular, corresponding to cusps in the polar plots of the grain boundary energy against the boundary normal. A schematic analysis with the capillarity vectors shows that the straight {111} grain boundaries across the triple junctions with the matrix grains are consistent with their singular characteristics. When the specimen is heat-treated further in hydrogen atmosphere, the flat {111} grain boundaries and the faceted grain boundaries between the abnormal grain and the matrix grains become curved when observed under TEM and HRTEM. The singular grain boundaries thus become rough in hydrogen. The results represent a direct observation of the grain boundary roughening transition, which appears to be analogous to the roughening transitions of crystal surfaces.

Sintering of magnesium oxide ceramics to high density requires high temperatures (1500–1800 °C), resulting in coarse-grained microstructures, or additives that promote sintering. An alternative is to use a fine-grained MgO starting powder with an impurity level < 1 % that allows appropriate processing conditions. Here, we report about processing by slip casting, resulting in ceramic green bodies that can be sintered to closed porosity at 1600 °C. X-ray diffraction and scanning electron microscopy combined with energy-dispersive X-ray spectroscopy showed that the sintered bodies are single-phase. The process described here is suitable to sinter large substrates up to 200 mm × 200 mm and 96 % of the theoretical density with closed porosity. The MgO substrates are suitable for partial melt processing of high-quality Bi-2212 thick films.

An amorphous covalent silicon-, carbon- and nitrogen-containing solid is prepared by the thermolysis of a polyvinylsilazane. Crystallization of the metastable material is induced by subsequent heat treatment at 1600 °C. For that purpose, microwave and conventional heating are applied and compared. The resulting materials are investigated by conventional and analytical transmission electron microscopy and X-ray diffraction. In the case of the conventionally heated material, thermal degradation occurs accompanied by a strong reduction of the nitrogen content. In contrast, kinetic stabilization of the Si–C–N solid by microwave heating is observed. Based on thermodynamic considerations, a model for the high-temperature behavior of the material is presented.

Liquid-phase-sintered SiC materials, all doped with 3 vol.% YAG (Y 3 Al 5 O 12 ) and sintered under identical conditions but processed in different furnaces (laboratory vs. industrial furnaces), revealed a variation in electrical resistivity of five orders of magnitude (10 2 vs.10 7 Ωcm). It was expected that, due to different cooling rates, different interface structures had evolved that strongly affected electrical resistivity, i. e., changes in intergranular film chemistry and corresponding thickness. In order to verify this hypothesis, the materials were characterized employing various techniques. High-resolution transmission electron microscopy of SiC interfaces revealed an unexpected result: clean interfaces for all samples. Elemental analysis confirmed yttrium and aluminum segregation at grain boundaries. Electron holography and Fresnel-fringe imaging revealed a change in mean inner potential across SiC interfaces. It is concluded that the segregation of acceptor ions at interfaces lowers the grain-boundary Fermi energy, resulting in the formation of potential barriers (Schottky barriers) along SiC interfaces, which in turn strongly affect electrical resistivity of SiC polycrystals.

It is shown that the defect chemistry of acceptor-doped SrTiO 3 is understood in great detail, and that this permits the accurate description, not only of the equilibrium electronic and ionic conductivities, but also of non-equilibrium situations generated by electrical or chemical driving forces, and even of the quenched state. Recent investigations on internal and external boundaries provide astonishingly precise insights into the pertinent structure–property relations.

The measured optical and transport properties of singlecrystal SrTiO 3 have been determined as a function of oxidation and reduction using spectroscopic ellipsometry, Hall and resistivity measurements. A decrease in the valence to conduction band transitions across the band gap, from O 2p to Ti 3d states, is observed by decreases in both the index of refraction and the extinction coefficient at the region of the band gap energy, in the ultraviolet region. This supports the model that oxygen nonstoichiometry depletes the O 2p densities of states in the top of the upper valence band, in agreement with previous photoemission studies. The degree of reduction decreases the band gap energy, calculated from the optical absorption coefficient spectra. The direct gap energy varies from 3.88 to 3.58 eV, and the indirect band gap energy ranging from 3.77 to 3.0 eV with increasing reduction. The resistivities of the samples decrease with increased reduction, yet free carrier densities are also found to decrease, suggesting an increase in electron mobility with reduction.

The spreading of liquid Ag and Ag–Mo alloys on molybdenum substrates has been studied, using a drop-transfer setup. Even though initial spreading velocities as fast as ≈ 1 m/s have been recorded in some experiments, a large variation in the spreading dynamics has been observed, and there is no unique relationship between the contact angle and the spreading velocity. This can be attributed to the formation of ridges at the triple junction, the movement of which controls spreading. The fastest spreading rates are consistent with results reported for low-temperature liquids; these can be described, using a molecular-kinetic model. Spreading kinetics and final contact angles were similar for pure Ag and Ag–Mo liquids.

Three low-energy prototypes of alloyed two-shell nanowires Ag m Au n have been investigated with density functional band structure calculations, using a plane wave basis set and norm-conserving pseudopotentials. All optimised structures are local minima of the heat of formation, H f , with reference to the pure bulk crystalline phases. Especially for low Ag contents, the curved wire surfaces are of comparable energy as the unreconstructed free flat (111) surfaces; thus, wires and flat surfaces can coexist. Alloyed nanowires with low Ag contents are even more stable than the corresponding pure Au or Ag analogues. For the most stable thin wires, negative values of the enthalpy of mixing are obtained which signify full miscibility over the whole compositional range. When including the constitutional entropy, the minimum of the free energy is shifted to higher Ag contents; thus, it is concluded that the Ag : Au ratio can be influenced by the preparation conditions.

The structure and associated translational states of the Σ = 3(112) symmetrical tilt grain boundary in niobium and molybdenum are investigated theoretically by three distinct computational methods. The results are compared with those of previous theoretical studies as well as with transmission electron microscopy observations. It was found that for this grain boundary, when fully relaxed, two possible translation states – “reflection” and “sheared” – are energetically almost degenerate in the two transition metals studied and thus comparison with experiments is ambivalent. This prevents the Σ = 3(112) boundary from being a suitable benchmark for validations of semi-empirical theoretical models of interatomic interactions.

Different alumina polymorphs (alpha, gamma and a transient phase) are formed in a copper matrix by oxidation of (Cu, Al) at 900 °C, at an oxygen activity of a O2 = 10 −13 . Spatially resolved electron energy loss spectroscopy is used to determine bonding and electronic structure at the different interfaces. Compared to adjacent bulk phases, at the interfaces, the near edge fine structures are modified. Cu L 2,3 shows an increased intensity compared to metallic copper and a chemical shift to higher energy. The size of both varies for the different interfaces, smallest chemical shift and highest L 23 intensity are observed at {111} γ-alumina // {111} Cu interfaces, while larger chemical shift and smaller L 2,3 intensities are found for γ-alumina/Cu interfaces in the order of {112̅0} α-alumina //{100} Cu , {0001} α-alumina // {001} Cu , {112̅3} α-alumina //Cu. In the interfacial O K edge, an additional pre-edge intensity is observed. For γ-alumina//Cu, it appears as a clear pre-peak, indicating hybridization with strong contributions of Cu 3d and O 2p states at the interface, while for α-alumina/Cu interfaces modifications in the interfacial O ELNES reveal mixed Cu–O–Al bonding with a strong covalent component. While interfacial Al L 2,3 shows no chemical shift or new fine structural features compared to the bulk precipitate, a decrease in the aluminum coordination is revealed for the α-alumina/Cu interfaces by changes in the intensity distribution in the first peak, while for gamma interfaces no such observation is made. The observations show that interfacial bonding does not only vary with the thermodynamic parameters temperature and oxygen chemical potential, but also with the crystallographic characteristics of the interface.

Cu 2 O can be electrodeposited epitaxially onto Si (001) single-crystal substrates with an orientation relationship that can be described as a 45° rotation around the 〈001〉 direction that both crystals have in common along the growth direction. We show that this orientation relationship corresponds to maximum “reciprocal-space overlap”. In apparent contradiction to the unique orientation relationship, conventional, high-resolution, and high-resolution analytical transmission electron microscopy has revealed that an amorphous interlayer, which has a thickness of about 4 nm and mainly consists of SiO 2 , separates the Cu 2 O layer and the Si substrate. Potential micromechanisms for the evolution of this structure during early stages of growth are discussed.

Recent results are presented on the chemical composition of metal/oxide interfaces using the tomographic atom probe and small angle neutron scattering for both Pd/MgO and Ag/MgO interfaces. The amount of segregated oxygen and hydrogen at the interfaces is in agreement with a model of structural vacancies at dense-packed interface planes. Different to Ag–Mg internally oxidized Pd–Mg alloys swell during internal oxidation as well as during vacuum annealing above 400 °C and after the internal oxidation. This behavior is explained by the formation of water bubbles from excess oxygen and residual hydrogen.

The interface chemistry and structure of two metal/ceramic composites was investigated using analytical and high-resolution transmission electron microscopy. Cu and Ni alumina composites were prepared and doped with predetermined amounts of Ca and Si additions to obtain a glass phase during sintering. The Si and Ca additions resulted in the formation of glass pockets at triple junctions and intergranular films at the alumina grain boundaries and the metal/alumina interfaces. A variation in the thickness and chemistry of the film at the metal/alumina interface was observed between the two composites. The formation of intergranular films and their presence at metal/alumina interfaces of occluded particles indicate that the intergranular films are probably stable.

We briefly outline the factors determining the intrinsic widths of features in electron energy loss near edge structure (ELNES) measured by electron energy loss spectroscopy (EELS) in the transmission electron microscope (TEM). We have made estimates of the differing contributions of both the initial and final state lifetime effects in the ELNES ionisation processes and also show how these may be combined with the instrumental energy resolution. We discuss the potential benefits of source monochromation for ELNES measurements via a comparison of these theoretical estimates with experimental spectra from the literature. We show that for certain core level excitations, solid state broadening mechanisms may be the fundamental limiting factor for resolving fine detail in ELNES.

We have analysed diffractograms of elastically filtered phase contrast images of thin evaporated amorphous silicon and carbon foils. For this analysis the two-particle structure factor was taken into account. It was determined from diffractograms of inelastically filtered images using calculated inelastic transfer functions. Due to the high accuracy of this new method, we were able to explain and reproduce quantitatively the intensity profile in the low spatial frequency range of the elastically filtered diffractograms. We could show that the weak-phase object approximation, which is normally used for the evaluation of such diffractograms is not adequate in this region. Instead, one has to use the weak-object approximation, which allows for a small imaginary part of the atomic scattering factor. In spite of the latter being only 3–10 % of the real part for the analysed materials it is observable, because the two-particle structure factor increases steeply for low spatial frequencies (in the direction of decreasing spatial frequencies).

In transmission electron microscopy (TEM), often the preparation of samples is the most critical part. Specimens have to have disk geometries of 3 mm diameter laterally, and they have to be transparent for the electron beam vertically. Therefore, a specimen thickness in the range of some 1 – 10 nm has to be achieved by the preparation process. While shrinking the specimen dimensions, care has to be taken to recover the materials properties in the nm-regime. We report and shortly discuss some TEM specimen preparation techniques mainly used in the Stuttgart TEM specimen preparation laboratory. Furthermore, we demonstrate how more advanced techniques lead to a more reliable preparation of weakly-bonded metal/SrTiO 3 interfaces. In addition, the advantage of low-voltage ion-milling is demonstrated by a case study for bulk SrTiO3. As a result, low-voltage ion polishing as a final step in the TEM specimen preparation by conventional ion-thinning turns out to significantly increase the specimen quality. In turn, the interpretation of high-resolution TEM micrographs and electron energy-loss near-edge structures becomes much more straightforward.

The quantification of interfacial segregation by spatial difference and one-dimensional profiling is presented in general where special attention is given to the random and systematic uncertainties. The method is demonstrated for an example of Al–Al 2 O 3 interfaces in a metal–ceramic composite material investigated by energy-dispersive X-ray spectroscopy and electron energy loss spectroscopy in a dedicated scanning transmission electron microscope. The variation of segregation measured at different interfaces by both methods is within the uncertainties, indicating a constant segregation level and interfacial phase formation. The most important random uncertainty is the counting statistics of the impurity signal whereas the specimen thickness introduces systematic uncertainties (via k factor and effective scan width). The latter could be significantly reduced when the specimen thickness is determined explicitly.

Elemental segregation to two types of boundaries in a low-alloy steel were studied by X-ray mapping using scanning transmission electron microscopy (STEM). To quantify the acquired X-ray maps, the ζ -factor method was applied, and then the compositional maps and the thickness map were obtained. Based on these quantified maps, further information about the analytical sensitivity of solute-element detection and the spatial resolution of segregation analysis were extracted. Furthermore, maps of the number of excess atoms on the boundary were also calculated from the compositional and thickness maps. It was concluded that Cr, Ni and Mo are co-segregated on the prior-austenite grain boundary and only Ni was segregated on the lath boundary.

Transmission electron microscopy and the method of histogram analysis of electron spectroscopic imaging were employed for studying the effect of heat treatment on the microstructure of Al/Ti metallization multilayers. The multilayers were deposited on a suite of piezoelectric substrate materials (quartz, LiTaO 3 , and LiNbO 3 ) by electron beam evaporation. In the as-deposited state, we did not detect a reaction between the individual Al and Ti layers. However, after annealing in air at 450 °C for 5 h, we found the Ti layers completely dissolved. In the course of this reaction, the microstructure of the Al layers transforms to globular grains consisting of Al and the intermetallic phase Al 3 Ti. Our results demonstrate the feasibility of forming intermetallic metallizations (Al 3 Ti) directly from deposited layers of elemental metals (Al, Ti).

The practical classification of grain boundaries as coincidence site lattice (CSL) boundaries requires the use of an acceptance criterion, of which the classical Brandon criterion is the most commonly used. In this work, we consider the role of the acceptance criterion on the grain boundary character distribution, triple junction distribution, and the boundary clustering behavior in nine experimental microstructures. The Brandon criterion is directly compared with the Palumbo – Aust criterion, which is more restrictive in the allowable deviations from exact CSL misorientation. Although the latter criterion causes changes in the CSL fraction, the triple junction distributions, and the cluster characteristics of the experimental microstructures, these changes are all mutually commensurate. The essential features of boundary connectivity are therefore not fundamentally affected by the choice of a CSL acceptance criterion.

Pure annealed Cu was subjected to shocks produced by a high power laser, producing shock pressures of about 5 GPa. The shock waves traveled through the 1 mm thick specimens. The shocks were found to modify the microstructure of the metal primary through the introduction of a large concentration of dislocations. The mechanical properties of the shock processed Cu were measured by tensile testing and compared to the unshocked material. The shock processing causes an increase in the yield stress, a decrease in the initial work hardening rate, and a decrease in the strain to failure. The deformation microstructures were characterized and found to be consistent with the observed mechanical behavior.

Layers of ε-Fe 3 N and γ′-Fe 4 N on ferrite were produced by nitriding iron single crystals or rolled sheets of iron in flowing ammonia at 520 °C. The nitride layers were characterised using X-ray diffraction, light microscopy as well as scanning and transmission electron microscopy. The compound layer consists of ε-Fe 3 N at the surface and of γ′-Fe 4 N facing the ferrite. After 4 h of nitriding, pores develop in the near surface region of ε-Fe 3 N showing more or less open porosity. Growth of the entire compound layer as well as of the massive and the porous ε-Fe 3 N sublayer is diffusion-controlled and follows a parabolic growth rate. The γ′-Fe 4 N layer is formed as a transition phase within a narrow interval of nitrogen activity, and it shows little growth in thickness. The transformation of γ′-Fe 4 N to ε-Fe 3 N is topotactic, where the orientation of the closed-packed iron layers of the crystal structures is preserved. Determination of lattice plane spacings was possible by X-ray diffraction, and this was correlated to the nitrogen content of ε-Fe 3 N. While the porous layer exhibits an enhanced nitrogen content corresponding to the chemical composition Fe 3 N 11 , the massive e Fe 3 N layer corresponds to Fe 3 N 1.0 . The pore formation in ε-Fe 3 N 11 is concluded to be the result of excess nitrogen atoms on non-structural sites, which have a high mobility. Therefore, recombination of excess nitrogen to molecular N 2 at lattice defects is preferred in ε-Fe3N with high nitrogen content compared to stoichiometric ε-Fe 3 N 1.0 with nitrogen on only structural sites.

The formation and thermal stability of φ phase in as-cast, annealed and very slowly cooled samples have been studied by means of X-ray diffraction, metallographic investigations, differential thermal analysis and magnetothermal analysis. The results of the combination of these data indicate that the φ phase forms from the β phase by a solid-state polymorphous reaction at 650 °C, it is thermodynamically stable below the formation temperature and can be in equilibrium with i, η, β, ε, and ω phases. On the other hand, an ordered β structure (β′) is found in as-cast and moderately cooled alloys, this metastability is characterized by a suppression of φ phase formation during cooling.