Table of contents

The semiconductor device field is reviewed with the aim of illustrating the main trends. The most striking are the emergence of silicon as the dominant material and the pressure to refine its technology to an unprecedented extent, mainly as a result of its use in integrated circuits. This is supported by research into improved fabrication techniques and efforts to obtain a clearer understanding of device behaviour. In addition there is active interest in novel devices, many of which require more unusual semiconductor materials, and the over-all trend is a rapidly increasing penetration of semiconductor devices into the electronics industry.

X-ray techniques have been used to study the structure both of vacuum-evaporated films of tungsten on glass and of polycrystalline tungsten foil following extended heat treatment in an ultra-high vacuum. A previous prediction that the contact potential difference between the two surfaces should be zero has been checked experimentally and found to be correct. The agreement is probably fortuitous however, in view of the structural differences between the evaporated films and the polycrystalline foil. The former have a completely random orientation, while in most cases the latter show a (100) texture.

A unified model of the oxide-coated cathode is suggested which attempts to combine the essential features of Nergaard's single-crystal model with those of Loosjes and Vink's pore-conduction model. The most important characteristic of the new model is that it regards the cathode as consisting of many small crystals (1 μm) whose surface properties predominate over their bulk properties. A consideration of the effects of localized impurity levels on such small crystals shows that they could produce a depletion layer which would have a depth varying from 10% to 100% of the mean size of the crystals. The band bending caused by these localized levels could also raise the work function of the crystals by amounts ranging from 0·05 to 1 v. This surface depletion layer offers an alternative hypothesis to the usual donor depletion layer, which had been assumed to be caused by electrolysis in the single-crystal model. The effects of these two types of depletion layer probably supplement each other in the decay of pulsed emission currents from the cathode. The surface depletion layer would predominate in a well-activated cathode, but field penetration effects would be important if it were poorly activated or in a high electric field. A large part of the activation process, and probably the whole of the process of the poisoning and recovery of a cathode's emission, can be accounted for simply in terms of changes in the concentrations of surface states.

Part I of this paper gives an analysis for thermally stimulated conductivity curves which takes into account both retrapping and recombination for a single trapping level and recombination centre. Three forms of the final expression are derived: one describing the slope at any point, a second describing the position of the current peak maximum and a third describing the integrated area under the curve up to any point. The application of the analyses to experimental results is illustrated. The use of the more complete analyses appears to lead to less uncertainty in the validity of the results obtained.

Part II describes a modified constant-temperature method of trap parameter determination which offers advantages in simplicity in the curve analysis and is also, to a large extent, self-checking against the assumed model. The technique is particularly useful for the case of fast retrapping and initial full occupancy, when values for trap depth, density, initial occupancy, capture cross section, trapping time and trapping factor can be obtained. It is immediately apparent from the observed decay whether or not fast retrapping at high initial occupancy is taking place.

A non-linear theory of acoustoelectric gain and current in piezoelectric semiconductors is developed. An expression is derived for the dependence of the acoustic gain on the amplitude of the wave. A corrected form of the Weinreich relation is also found together with a formula for the effective trapping factor for electrons in the acoustic wave.

Discrepancies of up to 10% exist in the published values of the elastic constants of alkali halides and these are too large to be accounted for by individual experimental errors. Static, resonance and ultrasonic techniques have been applied to the determination of the elastic constants of LiF, NaF, NaCl and KCl at room temperature. All the specimens of each material were taken from the same parent single crystal which was grown by the Stockbarger method.

Close agreement was obtained with the three methods, indicating that the discrepancies which do exist in the published data may be due to differences in the single crystals used by the various investigators.

The elastic moduli of LiF, NaF, NaCl and KCl single crystals have been measured in the temperature range 20-700°C by an audio frequency resonance method. The moduli were determined from the resonances of flexural and torsional modes of vibration of long thin rods and a knowledge of the room temperature moduli values which had been obtained previously. The elastic compliances s₁₁, -s₁₂ and s₄₄ all increase with increasing temperature, typically by a factor of two in the range considered. Values extrapolated to 0°K are given, and the Debye temperatures are calculated from these.

A high-temperature x-ray study of pure iron has shown that a small, but easily detectable, lattice parameter anomaly, representing a decrease in the mean linear thermal expansion coefficient, occurs over a narrow range of temperature (similar 35 degc) about the Curie point. The total volume change associated with the ferromagnetic-paramagnetic transition in pure iron obtained from the experimental measurements gave values in good agreement with those calculated directly from volume magnetostriction data, although for an alternative method of calculation the agreement was poor.

This paper presents the results of measurements of the changes in strain, as determined from x-ray diffraction line broadening, stored strain energy, particle shape, size and size distribution in samples of yttrium iron garnet powder brought about by a variety of milling treatments. For similar starting material the strain induced by the various milling procedures was found to vary by two orders of magnitude. An example is given of the build-up of strain and the accompanying changes in particle characteristics with milling time.

The field emission from negatively charged spherical solid particles has been studied analytically, and some numerical results depicting the variation in the number of emitted electrons with the charge, radius and work function of the particle are presented.

The ionization distribution in depth has been determined experimentally for two elements, lead and titanium, at various incident electron energies, by the sandwich tracer technique previously used by Castaing and Descamps.

The experimental curves of the ionization distribution in depth at 29 kv for lead and titanium are compared with curves calculated from the distribution in depth and energy of the incident electrons computed by Bishop. Agreement between the experimental and theoretical results is good.

The surface ionization straightphi₀, which is mainly due to backscattering of the electrons, has been determined at various accelerating voltages for titanium, nickel and lead. straightphi₀ is a function of the atomic number of the anticathode and of the accelerating voltage. Good agreement is obtained between the experimental values and values calculated by Duncumb and Melford. The x-ray absorption correction curves f(×) for titanium and lead have been computed from the experimental curves of the distribution in depth of the characteristic emission. These are compared with the curves predicted by the `Philibert correction'.

Experimental values of the trapping probability of low-energy inert gas ions in polycrystalline tungsten are presented. This trapping probability is shown to be critically dependent upon the surface topography and the defect state of the tungsten, and it is concluded that experimental data determined without proper regard for these variables cannot be used for informative comparison with theoretically determined values of trapping probability. A theoretical model accounting for the possible relaxation of the tungsten lattice during ion penetration is also presented.

Sound absorption and velocity measurements were made in cyclopropane, ethylene and sulphur hexafluoride at 400 and 700 kHz in the frequency/pressure range 0·2-30 MHz atm⁻¹ and in the temperature range -70 to 150°C. Relaxation times for the transfer of vibrational energy were deduced and compared with previous measurements. These results are discussed in terms of the theories of molecular energy transfer. Evidence for double relaxation processes is found in the case of ethylene and sulphur hexafluoride, but the measurements in cyclopropane fit a single relaxation process within the estimated experimental error.

A hot-wire type of thermal diffusion column is used to determine the thermal conductivity of helium in the temperature range 350-1350°K. The description of the apparatus, the theory of the method, and the various corrections to be applied to the apparent conductivity values are discussed briefly. The values for the thermal conductivity λ of helium agree well with those given by the relation λ×10⁵=17·201+6·764×10⁻²T-1·25×10⁻⁶T². (λ is in cal cm⁻¹ s⁻¹ degk⁻¹ and T is in °K). The present results are compared with the experimental data of other workers as well as with the predictions of the Chapman-Enskog theory and the modified Buckingham exp-six potential. The precision of our data is about ±2%, and our results agree with the theory to within an average absolute deviation of 1·9%.

An analytical expression for the intensity distribution of a reflection echelon for a Doppler-broadened Lorentzian profile (Voigt profile) has been obtained. The results of calculations have been presented in a graphic form which can be used to determine both the parameters characteristic of a line shape.

The theoretical treatment of a positive column using fluid equations to describe the ion motion and including ion inertia is extended to the case where quasi-neutrality is not assumed. Calculations have been performed in the approximation of electron temperature large compared with ion temperature. The range of discharges from collisionless to collision-dominated and from quasi-neutral to freely-diffusing can then be treated. Charged-particle, ion-velocity and potential distributions have been obtained.

The relevance of the calculations is examined in the low-current (low number density) case when electron temperature becomes dependent on current in an active discharge. Firm conclusions cannot be drawn because of the lack of experimental data at low enough pressures.

A study has been undertaken of the change in optical extinction direction in deformation bands produced by tensile and shear tests on oriented polyethylene terephthalate sheets. The change in extinction direction has been related theoretically to the band direction, the shear strain and the normal strain. Good agreement was obtained between experimental results and the prediction of this theory. It can be concluded that under the conditions of initial orientation and deformation within the bands, the optical anisotropy is directly related to the strain history.

The behaviour of a gas phase and combined gas and surface phase flow in a model microporous medium is examined. The model system consists of a bundle of cylindrical tubes each of which is constructed from a series of sections which have radii chosen from a distribution function. Structure factors are derived and calculated for this medium using mainly chi-square and Gaussian radius distribution functions, and their variation with the properties of the flow system is discussed. It is concluded that the particular aspect of pore structure examined in this paper is of considerable importance in the interpretation of surface flow results.