This doctoral thesis is concerned with rapid solidification of AB5 materials suitable for electrochemical hydrogen storage. The primary objective of the work has been to characterise the microstructure and crystal structure of the produced AB5 materials as a function of the process parameters, e.g. the cooling rate during rapid solidification, the determination of which has been paid special attention to. The thesis is divided in to 6 parts, of which Part I is a literature review, starting with a short presentation of energy storage alternatives. Then a general review of metal hydrides and their utilisation as energy carriers is presented. This part also includes more detailed descriptions of the crystal structure, the chemical composition and the hydrogen storage properties of AB5 materials. Furthermore, a description of the chill-block melt spinning process and the gas atomisation process is given. In Part II of the thesis a digital photocalorimetric technique has been developed and applied for obtaining in situ temperature measurements during chill-block melt spinning of a Mm(NiCoMnAl)5 hydride forming alloy (Mm = Mischmetal of rare earths). Compared with conventional colour transmission temperature measurements, this technique offers a special advantage in terms of a high temperature resolutional and positional accuracy, which under the prevailing experimental conditions were found to be +/- 29 K and +/- 0.1 mm, respectively. Moreover, it is shown that the cooling rate in solid state is approximately 2.5 times higher than that observed during solidification, indicating that the solid ribbon stayed in intimate contact with the wheel surface down to very low metal temperatures before the bond was broken. During this contact period the cooling regime shifted from near ideal in the melt puddle to near Newtonian towards the end, when the heat transfer from the solid ribbon to the wheel became the rate controlling step. In Part III of the thesis the changes of the crystal structure and the grain structure of La0.60Ce0.29Pr0.04Nd0.07Ni3.37Co0.79Mn0.25Al0.74 with increasing cooling rate during chill-block melt spinning are described. Totally, the material was rapidly solidified at 9 different cooling rates. The grain structure, crystallographic texture and the lattice parameters were studied by means of electron microscopy and powder X-ray diffraction. Additionally, the density of the rapidly solidified materials was measured by a gas pycnometer. All these properties were found to change with increasing cooling rate. The grain size decreased continuously with increasing cooling rate and was in the range of 1-5 μm. The strength of the crystallographic texture first increased and then decreased with increasing cooling rate. Transmission electron microscopy studies revealed that the grains contained a large amount of crystallographic twins and that the solidification morphology changed from cellular to plane front at a cooling rate during solidification of approximately 6·104 Ks-1. The unit cell volume and the density followed the same pattern with increasing cooling rate and decreased within each solidification morphology, but at the cooling rate from which the morphology changed, both these parameters suddenly increased. The identical variations in the unit cell volume and the density is explained by formation of excess lattice vacancies during rapid solidification. In Part IV of the thesis rapid solidification of the materials La0.60Ce0.27Pr0.04Nd0.09Ni4.76Sn0.24 and LaNi4.76Sn0.24 at 7 different cooling rates are described. The materials were analysed by means of electron microscopy and powder X-ray diffraction. The grain structures of both alloys were found to be in the nanometer range, and the grain sizes were almost invariant with increasing cooling rate. Furthermore, the lattice parameters of these materials were almost unaffected by increasing cooling rate. However, elemental line scans showed that the tin containing materials were not chemically homogeneous after chill-block melt spinning. The tin and nickel level fluctuated in an opposite manner, and the origin of these fluctuations is possibly due to inhomogeneities in the master alloys produced prior to rapid solidification. Part V of the thesis deals with the effect of heat treatment of the rapidly solidified materials presented in Part III and IV. The first material was heat treated at 400°C and the latter two at 1000°C and 900°C respectively. Electron microscopy investigations showed that the grain structure of the first material remained unchanged during the heat treatment while the latter two were subject to sincere grain growth. The inhomogeneities were removed during the heat treatment, and X-ray powder diffraction showed that the lattice constants were changing towards equilibrium values during the heat treatment. Furthermore, the density variations in the rapidly solidified material in Part III were removed by the heat treatment. This change and the change of the lattice parameters were probably due to annihilation of excess lattice vacancies during the heat treatment. Finally, in Part VI of the thesis the measured variations in the lattice parameters with increasing cooling rate are compared with the electrochemical hydrogen storage properties of the materials, which has been studied in a parallel work. It is shown that the hydrogen storage capacity and the absorption pressure of the material in Part III are controlled by the unit cell volume and hence the cooling rate during solidification.

Rapid dissolution of alumina upon addition to the cryolitic melt is crucial for the modern Hall-Heroult process for aluminium production. The formation of slow - dissolving alumina agglomerates may be detrimental, and irregular dissolution kinetics may cause the loss of process control. So-called anode effects may subsequently ignite, which are a major source of green-house gases from the primary aluminium industry. A literature review and the study of the theory of sintering provides the background for discussing the present work. The most probable mass transport mechanism in the transition alumina-fluoride-moisture system studied here is surface diffusion. Surface diffusion is a non-densifying mass transport mechanism that will result in coarsening (alumina grain growth) but only weak interparticle bonding since no macroscopic shrinkage is involved. Rapid mass transport is known to result when there is a simultaneous phase transformation, and this is the case when transition alumina transforms to α-alumina, catalyzed by the presence of fluorides. The main experimental techniques used in the present work were powder X-ray diffraction (XRD) and Scanning Electron Microscopy (SEM). Supporting techniques used have been speciffc area determination by the BET method and simple thermo-gravimetric techniques. An optical furnace was designed and built in order to study the dissolution of tablet alumina agglomerates. A preliminary agglomeration study of preformed cylindrical alumina samples served to map some of the most important mechanisms involved when alumina powder interacts with alumina-saturated cryolitic melt. The conditions at the alumina-melt interface were studied, but it is concluded that the experimental method could not provide the necessary parameter control in order to study the agglomeration mechanism in further detail. The tablet agglomerate study is the major experimental contribution of the present work. The experimental method provided good control of the sample chemistry and well defined temperature and time variables. It is concluded that liquid cryolitic melt (NaAlF4) provides an effective mass transport route for the transformation assisted growth of α-alumina platelets. The platelets that initially form will provide the limited mechanical strength necessary for agglomerate formation and their persistence in a cryolitic melt. Alumina agglomeration may therefore take place with only partial, initial phase transformation. It is concluded that differences in the agglomeration behavior of various qualities of alumina may be the rate determining property for alumina dissolution kinetics in cryolitic melts. Differences in the agglomeration behavior may be due to a number of physical properties of alumina. It is argued here that the fundamental, but difficult to measure, alumina nano-structure may be most important. The alumina nano-structure is correlated to secondary alumina properties such as the α-alumina content, specific surface area (BET) and moisture content (MOI, LOI). In this study an X-ray diffraction line profile analysis using the Warren-Averbach method shows that there is a significant difference in the nano-structure of the two smelter grade alumina qualities under study. This may explain the different agglomeration behavior that is observed. An optical study of tablet agglomerate dissolution in cryolitic melt proved to be largely unsuccessful due to severe corrosion of the quartz crucibles used. However, a proposed mechanism for the tendency for disintegration of alumina agglomerates, thus dissolving as \snow-flakes" is supported. The temperature response time in the tablet alumina samples was studied in order to determine the experimental limit of the shortest time period possible in the experiments. The exothermal γ -> α transformation is observed for secondary alumina samples containing adsorbed fluorides. An interesting effect of the carbon content in secondary alumina is also shown. The moisture content of smelter grade alumina is a function of the alumina quality, in particular the technology used for the calcination of the aluminium trihydrate precursor. In the current study the moisture content is shown to be a dynamic function of the ambient temperature and relative humidity. The moisture content is an important variable for the study of alumina agglomeration, and for the fluoride emission from the Hall-Heroult process. The kinetics of moisture desorption and absorption for various alumina qualities is studied. The desorption kinetics is concluded to be signifcantly different, while it is also shown that practical absorption kinetics is a function of the sample size and available surface area.

Steam explosions are possible during granulation of Si and FeSi75. These explosions are a great hazard, and must be avoided. Norwegian ferroalloy producers have initiated a research program to learn more about such violent melt-water interactions, in a joint effort with NTNU and SINTEF. The focus has primarily been on important parameters that can be controlled industrially, such as water temperature and metal composition. This thesis-work has focused on the effect of small additions of Al and Ca in Si-metal and FeSi75. However, within the same project, experiments on the effect of water temperature have also been carried out. The work has primarily been of experimental character. Two experimental apparatuses have been used. The first apparatus allows us to rapidly melt a sample of metal in an inert atmosphere to a desired temperature, expose the surface of the melt to an oxidizing agent (i.e. water) and then rapidly cool the sample to room temperature. The oxide that forms at the surface is examined with a microprobe. Thus, information regarding the composition and substance of the oxide layer is available. The second apparatus is suitable for releasing single drops of melt into a water tank, where they can be triggered and explode. A variety of techniques have been used in order to monitor the experiment: regular video, high-speed film, high-speed video, open-shutter imaging and pressure transducer measurements. Both Si and FeSi75 must be triggered in order to explode. Trigger pressures range from 0.3 MPa (FeSi75) to 2 MPa (Si-metal). We have established at which depths the molten drops can be triggered. Molten drops of FeSi75 can be triggered at depths twice of those of molten drops of Si. The latter can be triggered even if they are partially solidified. The explosion itself is strong enough to trigger neighbor drops as far away as 400 mm. Thus, we cannot rule out the possibility of large-scale steam explosions during granulation of molten Si or FeSi75, which is in accordance with industrial practice. By the use of high-speed imaging techniques and pressure measurements, we have been able to describe qualitatively what happens when a molten drop of Si/FeSi75 fragments rapidly in water. As the melt fragments, the rapid heat transfer generates vapor as bubbles, which expand and collapse in a cyclic manner. Large pressure pulses are generated upon collapse of the steam bubble, that is, when water jets impact in the center of the collapsing bubble. The first step in the oxidation of liquid silicon is the formation of gaseous SiO. The fate of this gas now depends on the flow conditions at the surface of the melt. In the case of a molten drop descending in water, most of the gas is flushed away from the surface. Thus, there are only minor traces of oxygencontaining material (i.e. silica) at the surface of the solidified drop. The addition of small amounts of Al and/or Ca dramatically changes the behavior of the molten drop. A strange effect is the two-fold increase in the fall velocity for molten drops of silicon. A similar effect was detected for molten drops of FeSi75. Alloying elements such as Al and Ca greatly reduce the risk for a steam explosion of molten Si. The significance of these elements is related to the oxidation reactions at the surface of the molten drop of metal. As silicon reacts with water vapor and oxidizes, hydrogen gas is formed. If Al and Ca are present in the melt, these elements will speed up the hydrogen generation considerably. This gas is strongly influencing on the probability for a steam explosion to occur. H2 stabilizes the vapor film around the drop, that is, much stronger trigger pressures are needed to collapse the film. Even if the trigger pressure is strong enough to collapse the vapor film, violent interactions are almost completely absent. A fragmentation of the melt is observed, but the heat transfer is apparently not rapid enough to generate steam bubbles, i.e. the generation of steam is below the critical limit.

Magnetic properties of nanoparticle systems and spin glasses have been investigated theoretically, and experimentally by squid magnetometry. Two model three-dimensional spin glasses have been studied: a long-range Ag(11 at% Mn) Heisenberg spin glass and a short-range Fe0.5Mn0.5TiO3 Ising spin glass. Experimental protocols revealing ageing, memory and rejuvenation phenomena are used. Quantitative analyses of the glassy dynamics within the droplet model give evidences of significantly different exponents describing the nonequilibrium dynamics of the two samples. In particular, non-accumulative ageing related to temperature-chaos is much stronger in Ag(11 at% Mn) than in Fe0.5Mn0.5TiO3. The physical properties of magnetic nanoparticles have been investigated with focus on the influence of dipolar interparticle interaction. For weakly coupled nanoparticles, thermodynamic perturbation theory is employed to derive analytical expressions for the linear equilibrium susceptibility, the zero-field specific heat and averages of the local dipolar fields. By introducing the averages of the dipolar fields in an expression for the relaxation rate of a single particle, a non trivial dependence of the superparamagnetic blocking on the damping coefficient is evidenced. This damping dependence is interpreted in terms of the nonaxially symmetric potential created by the transverse component of the dipolar field. Strongly interacting nanoparticle systems are investigated experimentally in terms of spin-glass behaviour. Disorder and frustration arise in samples consisting of frozen ferrofluids from the randomness in particle position and anisotropy axes orientation. A strongly interacting system is shown to exhibit critical dynamics characteristic of a spin-glass phase transition. Ageing, memory and rejuvenation phenomena similar to those of conventional spin glasses are observed, albeit with weak temperature-chaos effects.

Magnetic and electrical properties of selected compounds containing manganese (Mn) are investigated by SQUID magnetometry and transport measurements. (Ga,Mn)As is a magnetic semiconductor obtained from GaAs by substituting Ga3+ for Mn2+. Mn acts in the alloy as a magnetic impurity, as well as a hole dopant. A carrier mediated ferromagnetic interaction is observed in (Ga,Mn)As single layers, as well as in (Ga,Mn)As/GaAs superlattices. The magnetic and electrical properties of these structures are controlled by the amount of holes, and thus by the amount of compensating defects such as AsGa antisites. Magnetic inhomogeneity appears for thin layers as well as for layers containing large concentration of Manganese. In non magnetic metallic elements containing a small amount of manganese impurities, a magnetic interaction develops, oscillating in sign with the distance between Mn atoms. Due to random distribution of manganese in a Ag(Mn) alloy, competing ferromagnetic and antiferromagnetic interaction appears, yielding magnetic frustration and the appearance of a spin glass phase at low temperature. These disordered systems show aging, chaos and memory phenomena, which are investigated in the three dimensional Ag(Mn) and Fe0.5Mn0.5TiO3 spin glasses using time dependent magnetization measurements. Perovskite manganites of type (R3+1-xA2+x)MnO3 show colossal magnetoresistive e_ects (CMR). For an optimum doping x, a ferromagnetic order is established, and large changes of their electrical resistance with an applied magnetic field are observed; a magnetoresistance which can be tailored by adding oriented grain boundaries in thin films of these materials. The Manganese appears in the system as Mn3+ and Mn4+, and both ferromagnetic and antiferromagnetic interaction is mediated by the charge carriers along the Mn-O-Mn bonds of the perovskite structure. Depending on the cations forming the manganite, and their relative amount, glassy dynamics may appear, yielding aging and memory features similar to those observed in spin glasses.

This thesis treats charged particle transport, mostly in solid materials but also, to some extent, in aqueous media. Three major types of materials have been investigated; dielectric materials, ion conductors, and drug-delivery systems. The frequency-dependent dielectric permittivity of sputtered amorphous thin film tantalum oxide (Ta2O5) has been determined by using impedance spectroscopy. A new interpolation formula has been derived, that interpolates between the two power-law regions at low and high frequencies usually observed in the dielectric spectrum. This formula is based on a regular-singular-point (RSP) analysis of the conduction process, and the power-laws in the dielectric spectrum are interpreted in terms of RSPs of the underlying rate equation for the corresponding polarization-current response function. Lithium transport properties of Ta2O5 have been analyzed by using the galvanostatic intermittent titration technique and by isothermal transient ionic current measurements. Chemical and component diffusion coefficients for intercalated lithium have been extracted. Moreover, the ion conduction process has been analyzed theoretically, and expressions for transient ionic currents derived, both for single ion-conducting layers and for three-layered structures of ion conductors. Electrical measurement techniques have also been applied to pharmaceutical systems. The alternating ionic current technique has been developed as a tool for determining the release of electrically charged drug substances in aqueous media. Tablets made of agglomerated micronized cellulose have been investigated, and sodium chloride has been used as a model drug. An attempt has been made to describe the combined drug dissolution and drug release processes in mathematical terms.

Clearly, there is a need for simple, strong actuators capable of large strokes in miniaturized systems such as valves and optical shutters. The basis for this work is the microstructure technology with processing techniques adopted from the integrated circuit industry. In many cases alternative techniques have been developed to obtain features not achievable with conventional silicon technology. Techniques to fabricate thermally activated phase transition actuators capable of large strokes, as well as strong, piezoceramic actuators, have been investigated Multilayered piezoceramic actuators have been fabricated and used in a miniature linear motor. A technique to build freestanding, three-dimensional structures drop by drop using a micromachined ink jet head and a slurry of piezoceramic particles has been developed. Ion track technology was used to create narrow pores in polyimide. To make bimorph-like structures capable of large strokes, these pores were impregnated with paraffin- a material with a large volume expansion associated with its solid-to-liquid phase transition. Paraffin was used in a silicon thermal switch intended for a passive thermal control system, and in a device to be used as a valve in a gas regulation system. Finally, paraffin actuators for integration in thermoplastic microfluidic systems have been developed. During the course of this work not only the importance of identifying the best materials for a given application has been addressed and acknowledged, but also that of finding a processing route on occasion far from the conventional one, and perhaps most important, that of anticipating the often surprising effects following from miniaturization.

Challenges in the realisation of a miniature robot are both to handle the complexity of such a system, and to cope with effects of the actual reduction in physical size of all the parts. In particular, the mechanisms for locomotion have to be analysed. The main achievements presented in the thesis are the evaluation and the development of fabrication techniques for miniature multilayer piezoceramic actuators, the evaluation of different motion mechanisms for miniature robots, and the development of building techniques for piezo-based miniature robots. New piezoelectric drive units for miniature robots were designed and fabricated. To realize these monolithic devices, the fabrication technique for multilayer piezoceramic structures was further developed and evaluated with respect to the potential for miniaturisation. Introducing milling in the green state as a technique for shaping piezoceramic actuators gave a geometrical freedom without impairing the possibility of miniaturisation. A rapid prototype process was also developed. In this process, green machining in a milling machine was not only used to shape the multilayer structure, but also to pattern the internal electrodes. The first prototype was a multilayer telescopic actuator, which proved to have a displacement amplification of about 5 compared to a multilayer stack. The drive units were used to evaluate different motion mechanisms. Experiments showed that for a mass corresponding to a typical miniature robot, i.e. 1-10 g, it is possible to use both dynamic and quasistatic motion mechanisms. Artefacts due to vibrations were identified as the main reason for non-ideal behaviour when the movable mass is small. Design criteria for robots with small masses are presented. A tethered cm3 miniature robot for micromanipulation was successfully built. Application specific integrated circuits and two drive units were integrated with a particular building technique. Three-axial positioning and manipulating operations were demonstrated, allowing for a 5-axial movement of a tool.

This thesis presents the results from the development process of a ceramic dental filling material based on calcium aluminate cement. The main focus of the work concerns dimensional changes during setting, hardening and curing and the understanding of the factors controlling the dimensional stability of the system. A range of compositions for research purposes and the composition of Doxadent™ – a dental product launched on the Swedish market in October 2000 – were evaluated. Furthermore hardness characteristics, flexural strength, porosity and microstructure studies are presented. The studies of dimensional changes led to a thorough investigation of the measuring devices used and their relevance. A split pin expander technique, very simple in function, has been evaluated and improved. The technique is considered to be adequate for detecting dimensional stability in restrained samples, thus mimicking the case for real fillings in most tooth cavities. The dimensional changes in the calcium-aluminate based cement system are mainly controlled by the grain size, the exact composition and the compaction degree. The expansion of the calcium-aluminate cement system was in the early work decreased from several percent down to only tenths of a percent. Results show that Doxadent™ has less than 0.2% in linear expansion after 200 days of storage in water. However, long-term tests have been unable to verify whether expansion stops with time. Long-term in-vitro studies of dimensional changes also affect the test equipment used, which is why the long-term behavior of the dimensional stability has to be clinically evaluated. The material integrates excellently with the tooth structure, has hardness and thermal properties similar to those of enamel and dentine, and is also biocompatible during hardening. A patented process for the preparation of wet compacted specimens was also developed.

Die casting is a very cost-efficient method of forming thin-walled and complex near net-shaped products with close geometric tolerances and good surface finish. A permanent die tool is used to make large quantities of identical products. The performance and tool life are limited by several mechanisms, e.g. thermal fatigue cracking, erosion, and corrosion. To develop new and more resistant tool materials for die casting detailed knowledge of the actual casting conditions and the tool failure mechanisms are essential. This thesis contributes to an increased knowledge of tool failure in die casting by investigating and simulating actual casting conditions and tool failure mechanisms. A method to record the temperature fluctuations in a cavity insert during actual brass die casting was developed, and details of the temperature conditions were obtained. Also, a test method based on cyclic induction heating and internal cooling of hollow cylindrical test rods was developed, where the surface strain during thermal cycling could be measured. This method reproduced the characteristic type of surface cracking observed on die casting tools, and proved to give information of the strains and stresses behind the fatigue failure. In actual die casting, the dominant tool failure mechanism is thermal fatigue cracking. The formation of the cracks is associated to accumulation of the local plastic strain that occurs during each casting cycle. Initial crack growth is facilitated by oxidation of the crack surfaces, and proceeded growth is facilitated by this oxidation in combination with crack filling of cast material, and by softening of the tool material. In addition, local enrichment of Pb at the crack front from the cast alloy melt was also observed to promote the crack growth in die casting of brass. In an investigation of thermal fatigue of two hot work tool steels, quenched and tempered to different conditions, it was found that low-cycle fatigue occurs, although the estimated tensile stress never exceed the initial yield strength of the steel. The reason is a gradual softening of the steel during the thermal cycling, and the presence of stress raising defects. The resistance against thermal cracking improves with initial tool steel hardness, because any initial ranking in hardness among the steels is unaffected by the thermal cycling. Another investigation on a selection of surface engineered tool steels, including common diffusion treatments, PVD coatings and combinations of these, showed that surface engineering generally reduce the resistance against thermal cracking as compared to untreated references, since the engineering processes influence negatively on the mechanical properties of the hot work tool steels. Finally, corrosion tests of CrN PVD-coated tool steels by exposing them to molten aluminium revealed the mechanisms of initiation and progress of liquid metal corrosion of this material combination, and that the corrosion resistance improves with the CrN coating thickness.