Search

In many operations such as breaking a circuit, arc initiation, ion emission from molten metals, the ensuing plasma interacts with electrodes, even in low voltage (3 V) conditions. The physics of metal behaviour in such cases is discussed, and the evidence for formation of a microplasma assessed. An account is given of investigation of the plasma properties temperature, neutral metal gas density, and electron concentration obtained by time-resolves spectroscopy, and the behaviour of the microscopic molten metal bridge and the formation of a microplasma are demonstrated by high-speed cine films (lo4 f/s and 0.4 x lo6 f/s).

Ionization currents in spectroscopically pure helium between plane parallel silver electrodes at separations up to 0.75 cm were measured in the range 6 leq E/p leq 100 v cm-1 mmHg-1 and 3 leq p leq 115 mmHg where E is the applied electric field and p the gas pressure. The values of the primary ionization coefficient α/p deduced from these measurements were lower than previously published experimental values, particularly at the lowest range of E/p. Within the limits of experimental error α/p was found to be a function of E/p only. The present values of α/p were in good agreement with the values calculated by Dunlop in 1949, by Abdelnabi and Massey in 1953 and by Phelps in 1960. This result suggests that the Smit distribution of electron velocities, Maier-Leibnitz's inelastic collision cross section data and Nielsen's electron mobility values are applicable to the data.

The dependence of the secondary emission from the cathode on the energies of the impinging positive ions is of importance in the mechanism of the corona, or asymmetric, discharge through gases. In particular, it is important to find whether positive ions are able to liberate electrons from an electrode on to which they drift when their mean energy of agitation is no greater than that of the atoms of the gas through which they move. This paper describes experiments performed with positive ions of argon moving in that gas under weak electric fields in order to measure the secondary emission produced at a copper cathode. No secondary emission was detected when the electric intensity Z was less than 20 V./cm. and Z/p was less than 40, but the secondary emission attained a value of about 1.1 per cent when the electric field was increased above 60 V./cm. corresponding to values of Z/p greater than 100. This result is compared with those previously obtained with hydrogen and helium ions under similar conditions.

In elucidating the phenomena of breakdown it is necessary first to find the general mechanism by which the various processes of ionization interact to produce the necessary amplification of current, and then to identify the particular ionization processes themselves. It was the exact prediction, both of the form of the growth of ionization and of the breakdown potentials, when the spark parameter pd was less than about 200 (mm of mercury × cm), that first established the validity of the Townsend theory based on the continuous development of ionization by primary and secondary ionization processes; the main characteristics of this theory are given. The problem of breakdown at higher pressures when pd is 760 (mm of mercury × cm) is next discussed, and recent experimental and theoretical work on the growth of ionization currents and breakdown is described from which it is concluded that the general Townsend mechanism can also account for breakdown in this range of the spark parameter. This mechanism is then discussed in relation to other properties of breakdown involving the spatial-temporal development of currents, breakdown under impulse voltages, under non-uniform fields, and, finally, in highly compressed gases at very high voltages.

This paper describes a theoretical investigation of the mechanism of ejection of secondary electrons at the cathode surface in electric discharges in hydrogen. Theoretical estimates of the magnitude of the coefficient δ/α (due to photons) are obtained from the energy balance equation of an electron stream moving in a gas under the influence of an electric field. The magnitudes of the coefficient γ (due to ions) are obtained from the theory of resonance neutralization and Auger neutralization and de-excitation of an ion at a metal surface. Comparison is made between such theoretical estimates and those values of γ and δ/α obtained from an analysis of the measurements of the temporal growth of ionization currents. Satisfactory agreement is obtained when the comparison is made at low values of the ratio of the electric field E to gas pressure p (i.e. in the range E/p similar 50 v cm-1 mmHg-1). At larger values of E/p (similar 200 v cm-1 mmHg-1) the agreement is satisfactory only when the mobility of hydrogen ions is taken to be about 20 cm2 sec-1 v-1 when computing γ and δ/α from the experimental data on temporal growth of ionization. This value of the ion mobility is about three times greater than the previously accepted value obtained by linear extrapolation of experimental data for values of E/p < 100 v cm-1 mmHg-1. The separate contributions to the total γ of the various processes of neutralization and subsequent electron ejection from the cathode are estimated over the full range of E/p from 50 to 250 v cm-1 mmHg-1.

The arc phase of electrical contacts of the type used in low-power communication engineering is discussed from the point of view of the fundamental processes which occur both in the arc column and at the electrode surfaces. A theory of electrode erosion due to spark discharges is applied to the case of contacts, and the results of recent experiments on discharge initiation and arc transfer are discussed. Attention is drawn to the practical consequences of the short time in which an arc can be established. It is shown that the state of the surface of the electrodes themselves is important in arc quenching; restriking of the spark or arc as the contacts separate caa be effected by electron emission from thin oxide films or from dust on the surfaces. The influence of the various disposable parameters on the wear of the electrodes is examined; conclusions are drawn concerning the desirable physical properties of the electrode material and the conditions of operation, in order that the wear shall be least. Recent experiments confirm that dust can be extremely effective in spark initiation, and it is suggested that the maintenance of the electrode surfaces clean and dust-free would reduce the tendency for the striking of the spark or arc with separating contacts, and so reduce the value of the parallel capacitance required for discharge suppression. The use of preferred gas atmospheres and the treatment of the electrode surfaces can assist in arc quenching; and the presence of even very slight self-inductance (about 10 -8 henry) in the local contact circuit (values less, perhaps, than are generally realized) can lead to arcing under certain conditions.

A quantitative theoretical investigation is made of the spatio-temporal growth of ionization between plane parallel electrodes. A numerical method is described for integrating the ioniza­tion growth equations which enables the growth of the discharge to be traced from its initia­tion, in the uniform field region, as far as the transition to the glow discharge, involving space-charge-distorted fields, on the basis of the Townsend primary and secondary processes. A new method, more accurate than Poisson’s equation in one dimension, is used to calculate the field distortion. Experimental data in hydrogen are analyzed, and it is shown that the theoretically predicted curves of voltage collapse agree with those observed experimentally when the field distortion is taken into consideration. The magnitudes of the various Townsend secondary coefficients required to describe in detail the voltage collapse and the observed formative times of breakdown depend on the cross-sectional area of the discharge. Limiting values of the secondary coefficients are given.

It was shown in part I that there is a need for further experimental data on the drift velocity W of ions in gases such as nitrogen and argon where there are still unresolved problems. In addition, measurements of quantities such as the diffusion coefficient D and mean energy e are required for ions in most gases since published data on these quantities are scanty. In principle, it is possible to determine these quantities by measuring both the drift velocity W and the ratio W/D for selected groups of ions in an apparatus combining a time of drift electrode system, similar to that used by Tyndall et al. (1928) and Tyndall (1938), with a lateral diffusion system similar to that used by Townsend (1925) for the measurement of the ratio W/D for electrons. The advantages of using a time of drift selector for the ions rather than, say, a mass spectrometer are first, that there is no pressure gradient between the selector and the diffusion apparatus, so that no differential pumping is necessary, and secondly, that not only are the ions selected but their mobility is measured at the same time.

The time rate of growth of ionization currents in a gas in a uniform electric field greater than that corresponding to the static sparking potential is investigated theoretically. This theoretical analysis is then applied to the breakdown of a gas at high values of the parameter pd. It is shown that the same primary and secondary ionization processes which lead to a growth of pre-breakdown currents in agreement with experiment, and to the calculation of static sparking potentials in agreement with those measured, also lead to a rapid decrease of the formative time lag with increasing overvoltage. The introduction of some other quite different process to account for the short formative time lag is unnecessary. The present theoretical investigation, together with previous experimental and theoretical studies, therefore, lay the basis of a comprehensive view of the electrical breakdown of gases covering a wide range of parameters. Curves showing the dependence of the formative time lag on overvoltage, calculated by means of the above analysis, are given; these curves may be used to elucidate the various secondary ionization processes operative in the breakdown mechanism.

In part II, the initial design of an apparatus to measure both the drift velocity W and the ratio of drift velocity to diffusion coefficient ( W/D ) for positive ions in gases was described. Measurements of W in argon and nitrogen using this apparatus were discussed in parts II and III, but it was shown that large leakage currents prevented the measurement of the ratio W/D . As a result of investigations of the leakage current, the apparatus was redesigned, and the purpose of the present part is to describe this new apparatus and to discuss the results obtained in hydrogen.

Formative times tf of ionization growth in uniform electric fields E were measured for spectroscopically pure helium at pressures p (similar cmHg) between plane parallel silver electrodes in the range 8.61 ≤ E/p ≤ 38.08 v cm-1 (mmHg)-1 over which the observed growth times lay between 10-3 and 10-1 sec for over-voltages of the order of 5% of the static sparking potential Vs. Quantitative analysis of the experimental data showed that the observed slow growth rates could not be accounted for, over the whole range investigated, by the action of secondary processes based on secondary cathode emission due to the incidence of unscattered photons, of positive ions, or of metastable atoms either separately or in any combination of these separate processes.

Cold electron emission was measured from untreated and heat-treated nickel electrodes in dry, oxygen-free nitrogen. The electron currents were measured in the range 103 to 4 × 104 elect/sec for electric field strengths between 9 × 104 v/cm and 3 × 105 v/cm and gas pressures of 500 to 700 torr. The results are discussed in relation to the interaction of photons, metastable molecules and positive ions with the cathode surface. It is concluded that the emission was produced by positive ions liberating electrons from the cathode, subsequent to their neutralization in Auger processes.

Values of the photoelectric secondary coefficient δ'/α due to the incidence of delayed non-resonance photons, having energies of approximately 20 eV on a silver cathode in a low pressure (10 < P0 < 39 mmHg) discharge in helium in the range 9 < E/P0 < 22 v cm-1 mmHg-1 of electric field E to gas pressure P0, have been obtained from measurements of the ionization growth times. Estimates of absolute photoelectric yield have been deduced from the values of δ'/α and are found to lie between 0.2 and 0.3 electrons per quantum.

This paper describes measurements of the formative time lag to breakdown in hydrogen in uniform electric field for a range of E/p from 50 to 250 v cm-1 (mm Hg)-1. The preliminary work showed that the time lag was critically dependent upon the state and nature of the cathode surface. The formative time lags were therefore measured with different cathode surfaces, and the state of each surface was specified by measuring its work function. Previous work in this laboratory has shown that measurements of formative time lags can be analysed in terms of Davidson's current growth equation to yield determinations of the individual secondary ionization coefficients. A similar analysis has been carried out of the results of the present experimental data, and it was found that if the cathode work function was decreased by 0.25 ev, the magnitudes of the secondary coffiecients were increased by a factor of two.

The conditions in which electrical discharges can occur at low-power contacts are surveyed, special attention being paid to the cases of make and break at medium voltage. The mechanism of production of the electrons which can initiate such discharges has been investigated experimentally by measuring the electron emission from the cathode under a succession of voltage pulses. Results showed that the emission was a cold-field process determined by the nature of the surface tarnish layers on the metal, the mechanism for thin (≃10−7 cm) layers being different from that for thicker surface layers, which are usually oxide. The influence of the nature of the electrode surface and of the gas atmosphere was examined, and it was concluded that an insulating surface-tarnish layer was essential for the comparatively high rate of production of the initiatory electrons under the low electric fields (104 ≃ volts/cm) usually set up at small contact gaps. A theory of the mechanism is outlined.

Ordinary un-outgassed metal surfaces exhibit substantial electron emission (similar 105 electrons per sec) under the influence of electric fields of the order of 103-105 v cm-1. The electron emission from a workshop-turned nickel surface under a field of 5 × 104 v cm-1 is about 2 × 104 electrons per sec in air. Such emission rates can be of considerable importance in the operation of switches, contacts, high pressure gas insulation or the initiation of discharges in gases or in vacuo. This paper gives an account of the theory and design of an apparatus for measuring the rate of electron emission from given metal surfaces under given values of applied electric intensities. The method consists of applying a succession of N unit-function pulses to the metal surface acting as the cathode in a small parallel-plate gap, and the statistical time lag ts of sparking measured for each pulse. The rate of electron emission r is, in appropriate conditions, N/Σts; Σts is counted and electronically recorded thus giving r directly. Some typical data found for metals - brass, steel, aluminium, nickel and molybdenum - with various surface conditions are given. The technological application of such data is indicated, and some physical problems in which the apparatus has been used are briefly discussed.

Determinations of ionization coefficients from measurements of the growth of ionization in air under uniform field conditions show that the ratio of the apparent primary ionization coefficient to pressure (α'/p) decreases linearly with increase in pressure in the range 100 ≤ p ≤ 1000 mmHg at E/p approximately equal to 35 v cm-1 mmHg-1 (E is the electric field). This variation is attributed to the influence of the process of three-body electron attachment, and is used to obtain values of the coefficient a2 for this process. The values obtained in this way are shown to be consistent with results obtained by extrapolation of data on three-body attachment obtained by previous workers for oxygen-nitrogen mixtures at values of E/p < 1 v cm-1 mmHg-1.

This paper describes accurate experimental measurements of the spatial growth of ionization in steady state conditions in hydrogen at high pressures (100 to 450 mm Hg) using an apparatus and procedure previously described (Crompton, Dutton and Haydon 1956). The values of the secondary ionization coefficient ω/α obtained from these measurements by means of the method of analysis developed in that paper show that, at a constant value of E/p (E the field, p the gas pressure), the value of ω/α is dependent both on the nature of the cathode and on the pressure of the gas. The variation in the value of ω/α with the material of the cathode shows that the predominant secondary ionization processes in hydrogen at high pressures are cathode dependent processes, thus confirming the previous work. A theoretical investigation of the possible cathode-dependent processes in hydrogen shows that the variation in the value of ω/α with gas pressure may be satisfactorily explained on the assumption that the photo-electric process is predominant and that excited molecules undergo collisions of the second kind with neutral gas molecules, so that the molecular collision frequency affects the rate of production of photons. Theoretical computations of the value of ω/α based on the assumption that the photo-electric process at the cathode is the predominant secondary ionization process, and using previously measured values of the atomic constants involved, give good agreement with experiment. Measurements of sparking potentials to within 1% show, in agreement with theory, that such measurements are not sufficiently precise to detect the departures from Paschen's law consequent on the observed dependence of the value of ω/α on gas pressure.

The breakdown mechanism was investigated by examining the dependence of the shape of the cathode emission (?/?, X/p) curves on the deposition of electro-positive or electro-negative atoms on cathodes of nickel, aluminium, silver, copper and molybdenum. X is the electric intensity, and p the gas pressure at about 15? C, and ? is the generalized secondary ionization coefficient which includes processes of cathode emission due to impact of positive ions, of excited atoms, and of photons generated in the ionization current; ? is the primary ionization coefficient due to the electrons. Deposition of positive atoms on the cathode produced sharp photoelectric peaks in the curves in the region X/p 150 v./cm.mm. Hg, but ?/? was not greatly affected at higher values of X/p. These results are analysed in terms of the varying effective work function of the surface due to the deposition of atoms. It is concluded that the high photoelectric emission from cathodes of low effective work function was due to low-energy photons generated in the pre-breakdown currents, but that for clean metals the secondary emission was due to impact of positive ions; photo-emission due to any high-energy photons generated in the ionization currents was negligible in comparison.

High-frequency breakdown between wires and coaxial cylinders in air and in hydrogen was investigated over the range of frequencies f from 3.5 Mc/s. to 70 Mc/s. at pressures below 20 mm. Hg. It was found that geometrically similar systems broke down at the same potential provided that both parameters ap and f/p were invariant, a being a linear dimension. Under conditions such that the total amplitude of oscillation of the electron cloud was smaller than the electrode interspace, the electric field at the wire was the controlling factor, and breakdown was found to be independent of the nature of the surface of the cylinder. Consequently, secondary ionization processes did not appear to play an important part as in the case of static discharges. The experiments were in agreement with the view that the high-frequency discharge in air and in hydrogen is determined by primary ionization in the gas and loss of electrons to the walls by diffusion only, provided the electron mean free path is small compared with the linear dimension of the discharge tube, and the collision frequency is very much greater than the oscillation frequency. The significance of the similarity relation is briefly discussed

Pre-breakdown currents were measured from polished nickel electrodes after various outgassing treatments at pressures of 10−6 to 10−9 torr. Breakdown was initiated by electron currents emitted from cathode sites characterized by work functions greater than 4 ev, electric field magnification factors of about 200 and emitting areas of 1 × 10−12 to 1 × 10−8 cm2. A breakdown mechanism is postulated which consists of the electron bombardment of the anode, the growth and ‘blow-out ’ of emitting sites and the transport of material from the anode to cathode.

Spark time lags in short gaps with tungsten electrodes were measured for different degrees of oxidation and activation, and the distributions of lags examined and compared. The results, which showed a reduction of mean lag and a narrowing of distribution following oxidation and activation, are discussed in relation to the mechanism of electron emission from oxidized cathodes for impulse and static conditions, and also for the Geiger counter.

An outline is given of the nature of the fundamental physics processes occurring at an electrical contact and the problems to which they give rise, particularly in relation to light duty electrical contacts. The discussion includes contacts in which currents and potentials are of the order of amperes and volts, and also the so-called electrostatic contacts, in which one or other or both of these quantities may be extremely small. Recent work on microscopic molten metal bridges, micro-arcs (both of which are important in metal transfer) and the problems of `electrostatic' contacts which mainly depend on surface properties, are described. Outstanding problems are discussed and the method by which they are being attacked are indicated.

Values of the secondary ionization coefficient γ due to the Auger process of electron ejection caused by the incidence of atomic and molecular positive ions of helium on a silver cathode in low current discharges in the range 9 < E/p0 < 41 v cm-1 torr-1 of electric field E to gas pressure p0 have been deduced from measurements of ionization growth times. For He+ ions γ is found to be approximately 0.2 electrons per ion, but for He2+ ions the electron ejection efficiency is about 0.05 electrons per ion. The rate of conversion of atomic to molecular helium ions is deduced from the measurements to be approximately 160 p02 sec-1. Because of the p0 and p02 dependence of secondary processes at pressures in excess of about 20 torr the similarity principle for discharges is expected to fail in helium.

An assessment of the method used to measure values of the Townsend primary coefficient of ionization at low pressures is given. This shows that the secondary processes of ionization must, in general, be taken into consideration in the analysis of the data, especially at pressures of the order of a few millimetres of mercury, in order to obtain accurate values of α/p0.

The growth of pre-breakdown ionization currents I in uniform fields in nitrogen has been measured over a range of the spark parameters E/p from 41 to 45 V/cm/mm Hg and pd up to 810 mm Hg cm with sparking potentials up to 38 kV, where p is the gas pressure and d the electrode separation. The graphs of log I against d were initially linear, but at separations near the sparking distance the graphs curved upwards, indicating the existence of both primary α and secondary ( ω/α ) ionization mechanisms. The sparking potentials and distance here measured were found to be exactly in accordance with those predicted by the growth of current equation I = I 0 exp ( αd )/[1 – ( ω/α ) {exp( αd ) –1 }]. There were strong indications that the secondary ionization was, in fact, a cathode process. Possible effects of space charge were also considered, and the spark mechanism discussed.

The electrical breakdown of gases between coaxial cylinders at low pressures was studied in air (a gas mixture), hydrogen (a diatomic gas) and in helium (a monatomic gas). The significant secondary ionization processes occurring were correlated with those which are known to occur in uniform fields. The similarity theorem was used to help in the identification of the fundamental collisional processes. The experimental results were consistent with the view that breakdown involves the same primary and secondary ionization processes as in uniform fields: the relative importance of the dominant secondary electron emission processes depends on the nature of the electrodes and the geometry of the system. The effects of traces of impurities in helium were also studied, and the results were accounted for by the action of metastable atoms.

A full understanding of electrical discharges and plasma requires knowledge of the motion of electrons and ions in gases in electric fields. For electrons, a large amount of published information (see, for example, Townsend 1915, 1925; Thomson & Thomson 1928; 1933; Healey & Reed 1941; Allis 1956; Bates 1962; Hasted 1964; McDaniel 1964) exists concerning collisional properties such as the mean free path, mean energy, diffusion coefficient and mobility. For ions, on the other hand, basic data on the various collisional properties other than mobilities is much more scanty. Positive ions, however, play an important role in the mechanism of electrical discharges not only in the initial breakdown of the gas but also in the development of the current and in the maintenance of the discharge (Llewellyn Jones 1953, 1957). The various collisional processes involving ions such as the liberation of secondary electrons from the cathode, space charge formation, ambipolar diffusion, recombination and charge exchange have been discussed by Mott & Massey (1949) and by Massey & Burhop (1952). Although positive ion mobilities have been studied by many investigators (see Bates 1962) there are considerably fewer results available about the other collisional properties, especially in weak electric fields. This is due to the formidable difficulties encountered in both the theoretical and experimental studies especially in the case of slow ions. Theoretical calculations relating to the collisional properties of positive ions require a knowledge of the forces of interaction exerted between ion and molecule. When the period of interaction is comparatively long, as in the case of slow positive ions colliding with gas molecules, these forces of interaction are often unknown. This is particularly so for positive ions travelling through their parent gas where the phenomenon of charge exchange occurs. Hence, in these circumstances, accurate experimental data concerning such slow collisions are of particular value.

The size, shape and stability of microscopic molten metal bridges of iron, platinum, gold, nickel, tungsten and molybdenum have been studied in the current range 10-100 A and the rupture voltages measured in each case. High-speed photographs were also obtained, and these gave information concerning the development and rupture of the molten metal bridges and the post-rupture micro-plasma.

Theoretical estimates of the areas of electrode hot-spots are made from a consideration of the lateral spread of the electrons and positive ions due to diffusion and their space charge. Areas thus found show that very high current densities of the order 106 A/cm2 are possible in typical short spark gaps at atmospheric pressure. The energy balance at such hot-spots for an equilibrium temperature equal to the boiling point of the electrode material is then considered and, neglecting chemical attack, an equation is set up relating the volume of electrode evaporated per spark (i.e. the spark erosion) to the physical properties of the electrode material.

A quantitative analysis of ionization growth times measured in pure helium between plane parallel silver electrodes (d = 0.817 cm) for 5.07 < po < 38.51 mmHg, in the range 8.61 < Es/po < 38.08 v cm-1 mmHg-1, for overvoltages up to 6% has been made. This shows that the dominant secondary ionization process at the higher pressures is delayed photoelectric emission due to the incidence at the cathode of collision-induced non-resonance radiation produced in the volume destruction of atoms in the 2 1S metastable state in two-body collisions in the gas. The cross section computed for this process is about 3 × 10-20 cm2/atom. At the lower pressures used secondary electron emission due to the incidence of helium positive ions plays an increasingly important role, contributing as much as 75% of the secondary emission at p = 5.07 mmHg.

The experimental conditions which have to be satisfied in order to measure ionization coefficients accurately at high values of the parameter pd (pressure x gap distance) are assessed. In an apparatus designed to fulfil these conditions, precision measurements were carried out for uniform electric fields E in air at values of the parameter pd up to 2300 mm Hg cm, which corresponds to a sparking potential of about 80 kv, thus extending the range so far investigated by Llewellyn Jones and Parker in 1950 and 1952. The results show that prebreakdown ionization growth in air is due to the action of primary (α) and secondary (ω) ionization processes modified by the process of attachment a, and that significant space charge effects were absent. At a value of pd of 2300 mm Hg cm the sparking potential calculated (81 ± 1.5 kv) from the criterion (expressed in terms of the ionization and attachment coefficients) which gives the static sparking potential was in agreement, within the experimental error, with that (79 ± 0.7 kv) observed. It was also shown that the state of the cathode surface had a marked influence on the secondary ionization even at these higher values of pd. For 35 less-than-or-eq, slant E/p less-than-or-eq, slant 40 v cm-1 (mm Hg)-1 and pressures between 400 and 1000 mm Hg, the apparent secondary ionization coefficient ω/(α-a) was found to be a function of E/p alone, whereas the apparent primary ionization coefficient (α-a)/p showed a small, just detectable decrease with increasing pressure at a given value of E/p.

From experimental investigations on ionization currents and high-voltage sparking in uniform electric fields it has generally been assumed that the growth of pre-breakdown currents followed a simple exponential law involving no secondary ionization, and that therefore to account for critical breakdown it was necessary to assume the sudden introduction of a new ionization mechanism not related to the growth of initial pre-breakdown currents. These data are critically analyzed, and it is shown that they do not in fact preclude the existence of important secondary ionization, as was assumed. The experimental conditions which must be satisfied in order to reveal and measure any such secondary ionization are discussed, and a high-voltage ionization chamber designed and built to satisfy these conditions is described. Measurements of pre-breakdown ionization currents between parallel plates in air under conditions which include those corresponding to a 1 cm gap in the atmosphere ( E / p = 40; pd = 800; V 8 = 30 kV) showed that, while the growth of the currents was simply exponential over the greater part of the sparking distance, in the important region near breakdown itself the currents did, in fact, increase at the faster rate represented by I = I 0 exp ( αd )/[1 — ( ω / α ) {exp ( αd )—1}]. This indicates the existence of a secondary mechanism throughout the complete breakdown process. Furthermore, the measurement of the ionization coefficients α and ( ω / α ) in the early stages of the growth of a small (≃ 10 -14 A) photo-electric current can predict, within the experimental error, the static potential at which the gas breaks down, thus indicating that the mechanism of the spark itself was the same as that which led to the growth of the photo-current even in its early stages.

An apparatus for the measurement of steady-state pre-breakdown ionization currents in helium at high pressures is described, and the sampling procedure necessary to obtain with it reproducible results discussed. The primary and secondary ionization coefficients determined from these measurements are given for 3 3 < E/p0 < 5.0 v cm-1 mmHg-1 and for 150 < p0 < 560 mmHg. The values of ? are discussed in relation to previous theoretically computed values and the experimental values obtained at lower gas pressures but with similar values of E/p0. The values of the generalized secondary ionization coefficient are found to be relatively high (similar 0.1) and to increase as E/p0 decreases.

The use of radioactive isotopes in the measurement of the transfer of metal from one electrode of an electrical contact to the other is described and details of the experimental procedure are given. The relation between matter transfer and circuit inductance at very low values of the inductance has been determined for platinum and palladium contacts operating at potential differences of the order of one volt. It is shown that for these metals there is no range of inductance down to 10-8 H over which the transfer is independent of inductance. The amount of transfer in relation to the volume of the molten metal bridge between the electrodes is considered, and the significance of the results in the light of theories of the phenomenon of transfer is discussed.

Oxide layers on the cathode produced marked irregularities in the Paschen curves for air and hydrogen, especially in the neighbourhood of the minimum Vm. Vm in hydrogen was measured at different stages during the removal of oxide layers from the cathode, and also during the deposition of one metal on another as base. Removal of the oxide layer produced smooth Paschen curves, while a thin film of one metal on a different metal as base gave values of Vm which were not characteristic of either metal in bulk. The deposition of thin metallic films had considerable effect on cathode emission owing to the alteration of the work function

This second slower group was resolved only in the third order maxima and was present in such small numbers that its mobility could not be accurately determined. Nevertheless, when the two groups of ion are resolved in this way the correct value is obtained for the mobility of the majority group. The lower values of mobility, sometimes found when the first and second order maxima were used, were due to the perturbation of the peaks of the majority group by the unresolved minority group. This minority group had a lower drift velocity than the other group and was seen to become more predominant as the conditions were altered so that the number of collisions between the faster ions and the gas molecules was increased.

Experimental investigation of the cold electron emission from metals covered with thin ( ~ 10 -7 cm) tarnish films in gases shows that substantial emission ( ~ 10 5 electrons/s) can be produced by comparatively low applied electric fields ( ~ 10 4 V/cm). The results are consistent with the view that the electrons emitted are extracted from the metal substrate by the high electric field set up across the thin surface film when covered with a layer of positive ions, the presence of positive ions on the thin film being essential to the emission mechanism which follows a field law; this high emission does not occur when the thin surface tarnish film is absent. A theory based on this view is given and quantitative agreement with the experimental data is obtained. The applications of these results to the problems of the initiation of electrical discharges in general, including the electrical breakdown mechanism at high gas pressures as well as the initiation of micro-arcs at electrical contacts, are indicated. Comparison of the field emission from a mechanically polished surface with that from an electrolytically polished surface shows that the surface smoothness obtainable by the two methods of polishing can be of the same order.

This paper discusses briefly the various kinds of discharges which can occur in evacuated apparatus, i.e. , for ranges of pressure approximately equal or smaller than 10 −4 mm. Hg when the mean free path is of the order of dimensions of the apparatus. Under these conditions the processes of ionisation in the gas are not always the predominating processes of either initiation or maintenance of such discharges, which are then greatly dependent upon the nature of the electrode surfaces. The mechanisms of the various discharges which can be produced in this way are briefly discussed, and an indication given of the importance of the possible roles played by magnetic fields and by the nature of the electrode surfaces.

The validity of the similarity theorem for high-frequency discharges in helium is examined experimentally in non-uniform fields over a range of frequencies f from 5 to 70 Mc/s and gas pressure p from 1 to 30 mm Hg. The theorem is satisfied for similar discharge systems in pure helium but not when traces of impurity are present. The theorem is shown to be sensitive not only to changes in the parameters pa and f/p, where a is a linear dimension of the discharge tube, but also to changes in the purity of the gas when the impurity can introduce collisional processes which are not dependent on X/p. The failure of the theorem for impure helium for similar discharge systems is accounted for by the occurrence of collisions of the second kind between metastable atoms of helium and molecules of impurity.