Table of contents

A method for two-dimensional deconvolution is applied in the correction of a computer-generated transmission electron microscope image of a planar object subjected to lens aberrations (spherical aberration and defocusing), in an attempt to assess the viability of resolving molecular structure in the transmission electron microscope. The object considered is a phthalocyanine molecule examined in bright-field microscopy at incident electron energies of 100 and 200 keV, and for different objective lens defocus values. The results are presented in the form of intensity profiles along the y axis for the aberrated image in the absence of experimental error, and an image affected by (i) random error and (ii) sinusoidal error of varying frequency. The effect on the deconvoluted result of using an incorrect defocus value in the deconvolution procedure is examined. The results derived for the object structure differ in detail from the known molecular shape; these differences arise from the necessary assumption of a linear relationship between the image intensity and the object structure. Further, the effects of the lens aberrations extend out to 5 nm from the centre of the molecule, and the present results were restricted to a radius of 1·5 nm from this centre.

Experiments are described which demonstrate the feasibility of using secondary-electron-emission storage techniques to give a bistable CRT with a liquid-crystal display medium.

This technique enables many of the problems inherent in other electron-beam-addressed liquid-crystal displays to be overcome, and gives a high-resolution display, with selective erasure capability, that can be viewed in high-ambient-lighting conditions.

Large collision broadening at high pressures results in overlap of the rotational transitions of the CO₂ laser. The effective linewidth of the lasing transitions then becomes so large that picosecond laser pulses would be possible.

An analysis has been conducted on the influence of the reference electrode area on probe characteristics taken in the electron retarding range of potentials, when the reference electrode, and perhaps also the probe, is in the `blocking effect' régime (where probe dimensions are greater than or equal to the electron and ion mean free paths). A new and more severe criterion for sufficiency of reference electrode area is derived, and it is shown that insufficiency of reference area can lead not only to rounding of the `knee' of the probe characteristic, but can cause an extra, spurious, near-linear section to appear in the (log current against potential) characteristic near space potential, leading to the possibility of grossly inaccurate electron temperature measurements. A number of published results are cited to show that this phenomenon is in fact a frequent source of error in probe measurements.

Results are presented which describe the properties of a single-mode low-frequency (ω₀) instability in a current-carrying plasma at low neutral pressures. Comparison of the results with a linear theory suggests its identification as an ion-sound instability caused by the electron stream in the plasma.

At higher neutral pressures, other wave modes are self-excited in the plasma and subharmonics of the fundamental are observed at ω₀/4, ω₀/2, and 3ω₀/4, as well as modes at 5ω₀/4, 3ω₀/2 etc. Theoretical calculations based on a nonlinear equation of the Van der Pol type are developed. The predictions are compared with the experimental results, and good qualitative agreement is achieved.

A comparison is made of the calculated conductivities, including the effects of atomic clusters, of saturated sodium, potassium and caesium vapours. It is shown that the dependence of the conductivity on pressure at a fixed temperature has a form which could enable the presence of clusters to be deduced without requiring an accurate knowledge of the electron scattering cross section.

The cathodic erosion of Cu, Cr and Cd has been measured at DC currents below 2000 A, and was found to be, to a first approximation, dependent on the total charge passed through the arc only. The observed erosion rates were 76, 22 and 400 μg C⁻¹ for Cu, Cr and Cd respectively.

Measurements on different kinds of copper indicate that the crystal grain size is of influence on the erosion rate.

Of the total energy dissipated in the arc, about 25% is transferred to the cathode and 75% is divided between the losses to the surroundings and input to the anode. An attempt is made to correlate the erosion rate to one or more of the material properties of the metal.

Measurements are reported of the total radiation loss from high-current pulsed arcs which burn freely and stably between copper electrodes in atmospheric air. The diagnostic system developed for these measurements is described, and typical results are presented. It is shown that the radiation loss, which has important temporal and spatial variations, is a major component in the arc power balance. The radiation data, together with a knowledge of the electrical input power to the arc column, have enabled the percentage of the input power lost as radiation to be determined.

A mathematical model of the gas atomization of a liquid is developed. This model, in which the liquid is allowed to have infinite depth, takes into account both the compressibility of the gas and the viscosity of the liquid, enabling a prediction to be made of the expected drop size in the spray formed when a gas flows over a liquid. It gives good results for the air atomization of water and an order-of-magnitude agreement for that of molten metals. A universal curve is constructed for molten metals such that the predicted size of the most frequently occurring drop size can be found by evaluating a single formula. The effect of allowing the liquid to have finite depth is investigated in two cases of interest.

Expressions have been obtained for the structure factors κ_s and κ_g (previously defined by the authors) which relate to the steady state flow of dilute gases with and without additional surface flow in a series capillary model and in a bundle of parallel capillaries. These two models represent two opposite limits to the behaviour of the more realistic network models. Radius distribution functions were chosen which would lead to analytical expressions and, at the same time, include the properties of skew, symmetrical, unimodal and bimodal distributions.

For the parallel arrangement κ_s=1 and κ_g>1 for any distribution function but κ_g does not exceed unity by more than 12% unless the distribution is skew bimodal with a rather small proportion of larger than average radii. The serial model behaviour is dominated by constrictions and κ_g is always less than unity. On the other hand, κ_s greater, equals, less 1 and typically has a maximum which depends on the width of the distribution. The influence of distribution shape on κ_s is complicated but is again most strongly marked for the bimodal distributions.

A theory is suggested for predicting the threshold voltage for dynamic scattering in nematic liquid crystals subject to a constant electric field. Because of experimental results showing similar instabilities in isotropic liquids, an isotropic model is used. The electrical behaviour of the liquid is assumed to have a charge carrier dissociation-recombination process in equilibrium. Under these assumptions, typical ion density curves are obtained and a non-zero charge density is found near the electrodes. The resulting electrical forces result in an instability. Calculations are performed to predict the threshold voltage for the instability. Order-of-magnitude calculations give results consistent with experimental data in liquid crystals.