Table of contents

The formation of dendritic crystals in two dimensions by crystallization from a supersaturated solution is studied by a Green's function method. The dendrites develop sidebranches in qualitative agreement with experiments in three dimensions. The size and the growth rate of the primary dendrite scales with respect to supersaturation and crystalline anisotropy in agreement with the results for needle-crystals. The wavelength of the sidebranches scales with the stability length as conjectured.

Solidification patterns in mixtures are discussed and it is shown that, although the understanding of the factors affecting dendritic growth is improving, more experiments on cellular patterns are needed. The use of transparent eutectic materials is emphasized and the need for more theoretical work on atomic attachment kinetics is stressed.

Computer simulations and experiments on viscous fingering are used to investigate the effects of fluctuations, driving force and anisotropy on the growth of two dimensional unstable interfaces. It is demonstrated that variations of the diffusion-limited aggregation model capture many of the most important features of Laplacian pattern formation. In the viscous fingering experiments carried out in a radial Hele-Shaw cell with nematic or smectic liquid crystals a number of unexpected morphological phase transitions can be observed including crossovers from tip splitting to dendritic growth and from fractal to homogeneous structures. The investigations reviewed here suggest that the role of noise, driving force and anisotropy is crucial in the formation of patterns and it is the complex interplay of these factors which produces the great variety of morphologies found in nature.

An important factor influencing the mechanical properties of polycrystalline materials is the grain shape. A new model of generalised grain shape changes due to plastic deformation is presented, for homogeneous strains occurring when the principal axes are at an arbitrary orientation.

Two-dimensional sections through an idealised polycrystal, taken at an arbitrary orientation, have been produced computationally for the first time, and these correspond closely with observations made on uniaxial test specimens subject to known tensile strains. One application of these idealised sections is to assess the methods of quantitative stereology in a rigorous, quantitative way, which has not previously been possible.

In vitreous silica below 1 THz one finds not only sound waves, tunneling states and structural relaxations, but also additional harmonic vibrations. Inelastic neutron scattering measurements show that the additional harmonic vibrations and the structural relaxations have the same eigenvector, namely relative rotation of connected SiO₄ tetrahedra. The results imply a common origin of additional harmonic vibrations, relaxations and tunneling states.

Optical transitions on dopant ions in glass exhibit a large inhomogeneous broadening due to the random nature of the sites occupied by the dopant ions. The technique of fluorescence line narrowing (FLN) can overcome this inhomogeneous broadening and permit sharp luminescence transitions to be observed. These sharp transitions can carry detailed information about the behaviour of the dopant ions in the glass and about the dynamical properties of the host material. A number of such studies are reviewed.

Neutron spin echo measurements have been performed on the mixed ionic system Ca_0.4K_0.6(NO₃)_1.4 around the glass transition temperature T_g in order to determine the time dependence of the density-density correlation function ϕ_q(t). This quantity plays a central role in recent mode-coupling theories of the glass transition in schematic models. On the liquid side of the transition our results reveal that there are two distinct slow relaxation processes. The faster one is situated in the 10^-13-10^-12s time domain. The second one (a) shows a rapid slowing down on T approaching the transition, which scales with the Stokes-Einstein diffusion constant (b) has a stretched exponential form exp (-(t/τ)^β] with β ≃ 0.6 and (c) tends to become the non-ergodic fraction of the structure factor at the transition. The amplitude ratio of the second and the first relaxation steps varies between 3 and 6 in a smooth oscillatory fashion over the studied wave-number range of q = 0.1-2.1 Å^-1. These findings are in qualitative agreement with theoretical predictions, but only if the theoretically relevant glass transition temperature T₀ is assumed to lie considerably above the phenomenological glass temperature T_g. This implies that the theory breaks down at relaxation times longer than some 10^-5s. In contrast, the q dependence of the characteristic time τ of the slower relaxation step revealed a fully unexpected feature, a strong maximum at q₀/2, where q₀ is the position of the first peak in the structure factor. We propose that this slowing down of the mass density fluctuations reflects the maximum of the charge density structure factor. This new phenomenon is expected to allow the study of the static charge density correlations directly in the q space via the induced dynamic effects.

A system close to a bulk phase transition presents peculiar response to external boundaries, a single wall may induce the presence of wetting layers, with the appearance of a surface phase transition and critical point. The confinement of the system between the two parallel walls or cylindrical tube, produces a shift of the bulk phase transition, which may be observed in a gravitational field as the capillary rise. The two phenomena may interact: the capillary properties are strongly affected by the wetting regime and the prewetting transition may be frustated by the capillary shift of the bulk transition. Symmetry breaking phase transitions present also the competition between conmensurate and inconmensurate structures which produce a richer variety of phenomena.

We present a short review of recent experimental and computer simulation results on surface roughening and surface initiated melting. These phenomena appear to be completely different although mutual interactions, not yet identified, can exist. The critical examination of various experimental results indicates that locating the start of surface melting at temperatures well below the melting point results from interpretations which should be considered cautiously. To clarify this point more precise experiments are needed. We show that a combination of computer simulations and experiments is the more efficient approach for the investigation of both the mechanisms of surface roughening at the atomic scale and the relationship between surface roughening and surface melting.

Temperature-dependent Rutherford backscattering measurements on a Pb(110) crystal face have provided evidence for an anomalous surface thermal expansion. This structural change can be considered as a precursor of a pronounced disordering ("melting") of the surface, which takes place as the temperature is raised towards the bulk melting point. These phenomena are found to be completely absent on the close-packed (111) surface of Pb. The implications of these results for current models of surface melting are discussed.

A brief review of the electronic properties of semiconductor reconstructed surfaces is given, the case of Si(111)2 × 1 being considered in detail. Optical transitions between surface state bands show a large anisotropy that strongly supports the so called chain model of the reconstruction.

Recent results on the dependence of such transitions on temperature show a considerable electron-lattice interaction at the surface. An Einstein mode with energy around 10 meV, analogous to that recently found with He scattering, seems necessary to explain the observed results. The shape of the spectrum and the number of phonons involved suggest a localized process. Discussion of the results and speculations on a surface polaron will be presented.

The origin of surface stress on metals is discussed. Self-consistent pseudo-potential calculations of the surface stress tensor at the aluminium [111] and [110] surfaces are reported. In both cases the stress is tensile, favouring contraction in the plane of the surface. The results for aluminium are compared with those for jellium and a simple interpretation is developed in terms of the smoothing of the electronic wavefunctions at the crystal surface. Some implications of the presence of a surface stress for surface phonon dispersion and for surface instabilities are discussed.

In the glue model the total cohesion of a metal is determined by a pairwise atom-atom effective interaction plus a many-body force (the "glue") which is introduced to ensure optimal coordination. Using parameters optimized for gold, we have studied the structural behaviour of the low index surfaces Au(100), Au(110) and Au(111). We have used a simulated annealing strategy based on molecular dynamics to search the lowest surface energy configuration. In all cases the optimal structures are found to be reconstructed, and remarkably similar to some experimentally suggested reconstruction models. The main driving mechanism is the formation of close-packed triangular surface layers favoured by the glue term.

At the surface of a ferromagnetic metal, or at its interface with another metal, the reduced coordination number, the symmetry breaking and the possible lattice distortion induced by strains, give rise to local changes of the energy band structure. This leads to important modifications of local magnetic properties, such as atomic magnetic moment and anisotropy, Curie temperature T_c.

The experimental study of these interface effects can be conveniently performed on multilayered ferromagnetic ultrathin films with good crystallinity and sharp interfaces.

After a brief review of the related recent developments for 3d transition metals, we present experimental results in Ni and Co ultrathin films layered by noble metal. For fcc Ni (111) sandwiches, the vanishing of ferromagnetism below about 2 monolayers, is consistent with recent predictions. The T_c variation versus thickness is discussed. The Co films on Au and Cu show an interface magnetic anisotropy determined by Ferromagnetic Magnetic Resonance experiments which favours a perpendicular orientation of the magnetization. The competition with shape anisotropy produces a rotation from the out of plane orientation to the in plane one when increasing the Co thickness.

We discuss the application of a beam of spin-polarized neutrons at grazing incidence to study ferromagnetism in ultrathin films of Fe(001) and Co(001) epitaxially grown in the metastable f.c.c. phase. Orders of magnitude increase in signal sensitivity are obtained by covering the magnetic layer with an epitaxial coating of Cu(001), thereby creating neutron interference enhancement in the thin-film sandwich structure. The ferromagnetism in cobalt films down to a single monolayer thickness can be measured with this technique.

Spin-flip electronic transitions have been recently observed in ferromagnetic metals by electron energy loss spectroscopy (EELS) with spin analysis. The strong energy dependence of the exchange excitation process involved allows the identification of these transitions also by ordinary EELS by taking the ratio of spectra obtained with low energy and high energy primary electrons. In rare-earth metals dipole-forbidden transitions between f levels with and without spin-flip are observed as the strongest features in the EELS spectra at low primary electron energies. In iron the spin-flip transitions (Stoner excitations) give rise to an additional contribution at about 2eV in the EELS spectra at low primary energies. The energy of the Stoner excitation is found to be constant up to the Curie temperature.

After an introduction to what is required of a high-performance permanent magnet material, a brief description is given of the Concerted European Action on Magnets, a unique cooperative research programme which involves more than fifty participants (a third of them industrial) from nine countries in the EEC. There follows an account of recent progress in understanding the intrinsic magnetic properties of Nd₂Fe₁₄B, and other isostructural compounds and solid solutions. Finally some directions in the search for new iron-rich phases with a high Curie point and uniaxial anisotropy are indicated.

The hydrogen absorption/desorption (HAD) behaviours of alloys based on Nd₁₅Fe₇₇B₈ (Neomax composition) and the stoichiometric phase Nd₂Fe₁₄B have been examined in an attempt to relate these effects to the microstructures of these alloys. The HAD processes were studied by microbalance, Differential Thermal Analysis (DTA) and mass spectroscopy and these showed that the absorption process in the Nd₁₆Fe₇₆B₈ alloy is multi-stage and occurs readily at room temperature. The Nd₂Fe₁₄B alloy on the other hand could not be activated at room temperature and hydrogen absorption could only be achieved at elevated temperatures. This confirmed the crucial role of the grain boundary phase(s) in the activation process of the Nd₁₆Fe₇₆B₈ alloy. The mass spectroscopy/vacuum degassing studies showed a two stage desorption process whereby hydrogen was first desorbed from the matrix (Nd₂Fe₁₄B) phase and then from the grain boundaries. The mass spectroscopy peaks indicated that almost equal amounts of hydrogen were desorbed from the matrix and the grain boundary phases.

The effects of uniaxial pressure parallel to (001) on the magnetic ordering and on the critical fluctuations of CeSb have been studied by neutron diffraction and diffuse critical neutron scattering. We find that the system exhibits a critical end point at p = 2.0 ± 0.3 k bar and T = 18.1 ± 0.2 K, which is understood in terms of competing bilinear and higher-degree magnetic interactions favoring antiferromagnetic type-I ordering with a second-order transition and a modulated ordering with a first-order transition, respectively. On the basis of a simple mean-field model we are able to explain the salient features of the magnetic phase diagram of CeSb, in particular, the realization of a critical end point.

The concept of phase slippage was introduced in superfluid physics by Anderson in 1966. Direct experimental evidence of this basic phenomenon was looked for unsuccessfully for decades. We describe observations of highly reproducible, discrete dissipation events in the flow of superfluid ⁴He through a microscopic orifice. These events occur at a rate given by the a.c. Josephson equation and they are shown to correspond to the creation of trapped quantised circulation. The onset threshold of the critical velocity at which these events occur is found to be strongly altered by minute traces of ³He impurities and to vary with temperature as 1 - T/T₀ with T₀ = 2.46 K from 1.2 K down to 5 mK. We interpret these events as slips of the quantum-mechanical phase difference across the orifice by 2π in the sense of Anderson.

At temperatures below about T_c/4, the mean free paths of quasiparticles in superfluid ³He-B become extremely long. In consequence the quasiparticles behave entirely ballistically, making possible a number of simple but important experiments:

(i) The damping of a vibrating wire measures the momentum density of the quasiparticles which is dominated by the exp (-Δ/kT) gap Boltzmann factor. The damping may be used as a direct ³He thermometer to measure a number of properties including, for example, the effective thermal conductance across a liquid to solid interface. This conductance also reflects the gap Boltzmann factor.

(ii) A solid object can move through the liquid at low velocities without dissipation, but above a critical velocity produces quasiparticles by pair-breaking of the condensate seen as a sudden onset of damping.

(iii) A supercritically-driven wire resonator thus acts as a quasiparticle source. The ballistic propagation of the quasiparticles can be detected on a distant (∼ mm) wire used as a quasiparticle-wind detector. The mechanical force on the detector wire leads to a small movement of the wire which can be measured with a SQUID.

Our observation of this quasiparticle wind yields direct information about the mean free paths and velocities of the excitations and also on the mechanism of the pair-breaking process. These quasiparticle beam methods open up a wide range of possible scattering experiments.

The NMR linewidth of HD impurities in solid para hydrogen has been measured to be less than the rigid lattice value. Motion due to quantum diffusion, similar to the exchange of atoms in dense hcp ³He, of the HD molecules is believed to be the cause of the narrowing of the linewidth. This motion is severely restricted for ortho-H₂ because of their quadrupole-quadrupole interactions with ortho neighbors. Pulsed NMR experiments designed to measure the motional effects on NMR spin echos are discussed. Pressure and temperature effects on the HD motion are also examined.

Nuclear magnetism in elemental metals is discussed with the emphasis on spontaneous nuclear ordering. Materials feasible for experiments are classified according to the abundancies of magnetic and non-magnetic isotopes which crucially affect their physical properties. The Hamiltonian of these systems consists of the dipolar interaction and the conduction electron mediated indirect interaction. The theory and measurement of indirect spin-spin coupling are described. Copper is still the only metal in which spontaneous nuclear ordering has been observed; the experiments are briefly reviewed. The structure of the observed three antiferromagnetic phases is discussed. A spin-wave calculation predicts a novel antiferromagnetic structure for the ground state. Phase diagrams for silver and gold in a magnetic field are expected to be drastically different from that of copper.

Tumour-seeking agents, such as hematoporphyrin derivative (HPD), in combination with laser radiation provide new possibilities for cancer tumour detection and treatment. The fluorescence emission from tissue irradiated with UV light is used for tumour localization. The development of techniques for maximum discrimination between tumours and normal tissue is of particular interest. Natural tissue fluorescence as well as fluorescence due to injected HPD will be discussed. Results from studies on rats and mice are reported. The use of fluorescence for identifying atherosclerotic plaque in human vessels is also illustrated. The development of clinical instrumentation for point measurements and imaging is discussed.

Nuclear Magnetic Resonance Imaging is a non-invasive method of imaging in vivo of the concentration and relaxation times of the mobile protons in bodies. It yields morphological and sometimes physiological information on patients. After recalling the principle of NMR we describe the methodology of obtention of spatially localized signals and present examples of images of clinical interest.

An overview is presented of the basic principles, the physics constrains and the practical limitations of Positron Emission Tomography. The most recent developments in PET instrumentation and the actual trends are discussed. Finally, some clinical applications are described.

Because coronary disease represents the principal health problem in the Western, industrialized world, and because of the risks and costs associated with conventional methods of visualizing the coronary arteries, an effort has been underway at the Stanford Synchrotron Radiation Laboratory to develop a less invasive coronary imaging procedure based on iodine K-edge dichromography. A pair of line images, recorded within a few milliseconds of each other, is taken with two monochromatic X-ray beams whose energy closely brackets the K-edge of iodine, 33.17 keV. The logarithmic subtraction of the images produced by these beams results in an image which greatly enhances signals arising from attenuation by iodine and almost totally suppresses signals arising from attenuation by soft tissue and bone. The high sensitivity to iodine allows the visualization of arterial structures after an intravenous injection of contrast agent and its subsequent 20-30 fold dilution.

The experiments began in 1979, with initial studies done on phantoms and excised pig hearts. The first images of anesthetized dogs were taken in 1982. The results of experiments on dogs will be reviewed, showing the stepwise evolution of the imaging system, leading to the use of the system on human subjects in 1986. The images recorded on human subjects will be described and the remaining problems discussed.

A review is given of recent experimental and theoretical work on electrical instabilities in semiconductors. Particular attention is paid to the conditions under which stochastic self-oscillations arise of the electric field strength, concentration of the charge carriers and other physical quantities. Possible connection between the stochastic self-oscillations and "1/f" noise is indicated.

The method of recurrence relations allows one to calculate time dependent correlation functions from first principles. It originates from the Kubo scalar product which realizes abstract Hilbert space of dynamical variables. For this realized space, we show that there exists a unique orthogonalization process by recurrence relation.

The method of recurrence relations exploits geometric properties of the structure of the realized Hilbert space such as the dimensionality and shape. These geometric properties depend on models as well as static constraints, e.g., temperature, wave vector, size. This method has been applied to several physical models including the homogeneous dense electron gas, the spin-1/2 XY and transverse Ising models, the spin van der Waals model and others. For these models, we have obtained the relaxation function, response function and memory function.

Current filamentation and chaotic temporal resistance behavior generated in homogeneously p-doped germanium during low-temperature avalanche breakdown have been investigated. Spontaneous current oscillations showing typical chaotic scenarios were found under tiny variation of distinct control parameters (d.c. bias voltage, transverse magnetic field, temperature). By means of a scanning electron microscope equipped with a liquidhelium stage, two-dimensional imaging of current filaments has been performed. In particular, the filament nucleation and the subsequent growth processes could be resolved. Furthermore, the origin of the spontaneous oscillations was found to be closely linked to the formation of the current filament pattern.

The switching dynamics of thermally induced optical bistability in ZnSe nonlinear interference filters is investigated experimentally and theoretically. The device, maintained on its low transmission level by a cw argon laser beam, is switched on its high transmission level by an external light pulse. Either a very fast switching or an important slowing down are obtained when the holding cw intensity or the switching pulse energy are well chosen. This variation of the switching times is analyzed theoretically and observed experimentally.

How do the properties of a solid gradually evolve as atoms are brought together to form increasingly larger units? In order to answer this question one must study aggregates of atoms too large to be called molecules but still too small to have even the structure of a crystal. Recent advances in experimental methods have made possible some first explorations in this long neglected but facinating field of study.

Without abstract, see full text

The alkali-halide cyanide systems, e.g., (KBr)_1-x(KCN)_x show a cubic room temperature phase with dynamic orientational disorder of the aspherical CN ions. At low temperatures the CN-rich compounds order ferroelastically whereas for lower CN concentrations an orientational glass is formed. The experimental results and the present state of understanding will be reviewed with emphasis on the anomalous structural and elastic behaviour at the crossover from one type of transition to the other.

Although plasticity of metals is complicated due to the string-like nature of its elementary units, the dislocations, it is currently tried to understand various stability aspects of plasticity. Of special interest are the dislocation distributions after cyclic loading ("fatigue"). Well developed periodic patterns are observed in the spatial distribution of the dislocations. Computer simulations of the behavior of straight edge dislocations will be reported, which show that the former phenomenon may be a straight-forward consequence of the interaction between dislocations. More recent ideas to relate the problem to general concepts of non-linear dynamics are also discussed.

It is shown that a coupled study of the structures due to interband transitions and of the free carrier response in heavily doped semiconductors gives a combined picture of crystal order, free carrier concentration, effective mass and relaxation time.

As, P and B ion implanted and laser annealed Si crystals are investigated. Reflectance measurements on samples with free carrier concentrations in the range 10¹⁹-10²¹ cm^-3 are reported and discussed.

The extension of the conventional effective-mass theory for shallow donors in semiconductors is examined. The shallow-deep instability problem is studied. A better incorporation of the intervalley scattering factor is discussed. The binding energies and pressure coefficients of chalcogen donors in Si are calculated using a model impurity potential. Results are compared with experimental values and earlier theoretical estimates.

EL2, the main defect in GaAs, has gathered a lot of attention during the last years because of its interesting physical features as well as its importance for properties of GaAs devices. In spite of much effort put into the identification of EL2, the model of its nature is not fully established as yet. The results of different experiments performed in laboratories all over the world, leading to the determination of EL2 nature are presented and different EL2 models are discussed.

The absorption spectra, as well as the UPS in disordered solids reflect many physical features of the interband optical transitions absent in crystals: the well-known k-smearing, electron-hole kinematical correlation and random matrix elements. For the usual linear stationary spectroscopy, the Kubo formula applies. We compare its use within the CPA to a variety of models, and to realistic computations, both in orbital form and in real space.

An ultrafast strong light pulse transcends the Kubo picture of interband transitions, as it excites a nonlinear, far non-equilibrium transient. New quantal effects emerge, in particular the coherence between the light and the disorder. The powerful tool of the non-equilibrium Green functions is invoked in conjunction with a generalized CPA. A two band model excited by a rectangular light pulse is then soluble and can be analyzed numerically.

We assign the term of frustration to the general case in which the interactions which compete on short distances in the stability of some material lead to local configurations of molecules which are incompatible on large scales. The standard example in liquid crystals are the blue phases, whose ground state is an assembly of frustrated domains with specific double twist order and characteristic size, separated by defect lines (disclinations). Blue phases are modifications of cholesteric phases with small molecules; the same considerations extend to cholesteric liquid crystal polymers where frustration is also documented. In particular, we discuss observations made in solutions of biological liquid crystal polymers, and in biological objects, like some proteins and the chromosome of dinoflagellate; this last example provides a geometry of double twist different to that one theorized in blue phases. In this model, the finite size of the chromosome is directly related to the characteristic size of the unfrustrated domain of double twist. Finally, we extend the consideration of a space of constant positive curvature already used for blue phases to the case of crystalline arrangements of chiral molecules. The frustration is entirely relieved in such a space. We argue that the same model of curved space could describe short range correlations in chiral or non-chiral amorphous polymers.

The low-temperature behavior of a tunneling atom interacting with a degenerate electron gas is studied by means of the renormalization group. Both electronic screening and electron-assisted tunneling are considered. As a result of the spin degeneracy an anomalous low temperature phase is found, corresponding to the nontrivial fixed point of the multichannel Kondo model.

Some aspects of the motion of charge density waves under a static electric field observed in a number of quasi one-dimensional conductors with an incommensurate superlattice are reviewed. Experiments aimed at the measurement of the CDW phase velocity are described. Recent experiments in the blue bronze Rb_0.3MoO₃ have shown that the local field oscillation frequency measured by NMR is equal to the periodic voltage oscillation frequency and these both agree with that expected from the measured sliding CDW current.

Thermodynamic properties of the random-field Ising model are reviewed and compared with published experimental results on diluted antiferromagnets in a uniform field. The Landau-Ginzburg free energy is calculated in the self-consistent approximation for 3-dimensional systems treating the random field induced fluctuations and the thermodynamic fluctuations of the order parameter separately. At low temperatures and in the presence of random fields two states exist, the stable, long-range ordered state, and the metastable disordered microdomain state. In a heating process, the ordered state changes to the high-temperature phase at the well defined transition temperature. The transition is first order for arbitrarily weak random field. Critical properties are briefly examined and according to the Ginzburg criterion, the classical critical behaviour is predicted for large random fields. For weak fields, nonclassical critical behaviour is expected.

Some "metallurgical" aspects of ³He and ⁴He are briefly described, namely the structure and growth of liquid solid interfaces (roughening transition), and the plasticity of the solid phase. Because orders of magnitude are very unusual, He provides an ideal system in which to test standard concepts of solid state behaviour.

A satisfactory understanding of the electronic properties of layered transition metal halides has been recently achieved on the basis of ultraviolet and photoemission investigations and detailed calculations of their band structure. A general picture of the electronic properties of these magnetic ionic insulators is presented, by discussing the examples of Cr halides and Fe, Co and Ni dihalides, studied by reflectance, electron loss and X-ray photoemission spectroscopy. The optical anisotropy is also considered by presenting the case of FeBr₂ crystals. Finally, recent angle-resolved photoemission investigations of the band structure of NiI₂ crystals open future prospects for the comprehension of Mott-Hubbard insulators, of which Ni-halides can be considered as prototypes.

Zero-phonon lines in impurity optical spectra and their inhomogeneous broadening are discussed. Photoburning of persistent spectral holes is not only an effective method to eliminate the inhomogeneous broadening and perform high resolution spectroscopy, but it serves also as a powerful tool to control by means of illumination the absorption coefficient and refractive index of the matter. It opens new horizons for frequency-selective optical memories, data storage and data processing. Hole burning by picosecond pulses and the time-and space-domain holography based on it are discussed and the results of the experiments described.

We report on the first measurements of two-photon absorption spectroscopy performed using the synchrotron radiation as a tunable vacuum ultraviolet radiation source and the light from a Nd-Yag laser. This new technique opens the possibility of performing non linear spectroscopy in the far ultraviolet region. As an example of application, we present and discuss the two-photon absorption spectra near the fundamental thresholds of KI, KCl and NaCl.