The Switch-On Mechanism of the Current Emission

The Switch-On effect of the current emission is known since long time. This effect is usually attributed to the existence of a Switch-On Voltage, after which the electrodes in the vacuum switch from a non-conductive to a conductive condition. Once reached the Switch-On point, the changes in the electrodes are permanent and the electric current can be measured even at lower voltages. In our experiments we find that when a constant electric field, not too intense (40kV/mm), is applied to smoothed, unconditioned new brand electrodes, the current output is initially negligible (certainly below our current measurement sensitivity $10^{-8}A$). After a long-lasting (in the order of tenths of hours) constant dc voltage has been applied, a sudden increase in current is observed. By then decreasing and/or increasing the voltage around the constant value, we found regular and reversible Fowler-Nordheim type diagrams. These transitions are interpreted as changes in the electrode surface structure. Our research aims to characterize these changes of current conductivity in terms of transition time, current and voltage level for electrodes made of different materials and/or with different surface treatments. Considerations are finally exposed to explain this Switch-On effect as a consequence of the accumulation of electric charge at the metal-insulator cathode interface.


I. INTRODUCTION
High voltage holding in vacuum is one of the most challenging issues in the realization of the Neutral Beam Injectors (NBI) for ITER [1].To produce the required 16 MW Deuterium beam at 1 MeV energy, D-ions are produced and accelerated by a 5 stages accelerator for a total voltage of minus 1M V , and then neutralized before the injection within the ITER plasma.The whole Ion Source is has then to be insulated in vacuum to withstand minus 1M V to ground with a gap length of about 100cm.Figure 1 sketches the Ion source and the accelerator of MITICA, the full-scale prototype of the NBI ITER Injectors under construction at the Neutral Beam Test Facility in Padova [1].So far, the goal to attain 1M V voltage holding with a gap length even larger than 100cm has not yet been reached with pressure lower than 10 −3 P a.Only operating at the pressure of 10 −2 ÷ 10 −3 P a the voltage can be hold; but this is an operating condition not easy to be maintained, being the operation point in the curve V BD vs d very close to the left branch of the Paschen Curve.The reason of the difficulty to reach voltage holding in the M V range is due to the well known Total Voltage Effect; this definition indicates the experimental evidence that the breakdown occurs not only when the electrode electric field Identify applicable funding agency here.If none, delete this.exceeds a given value, but depends also by the voltage applied.In its simplest formulation the total voltage effect is described by the Cranberg [2] breakdown condition EV > const which -for parallel plane electrodes -is expressed by the equation V bd = kd 0.5 d being the gap length.Splitting the gap with the use of intermediate screens connected to intermediate voltages is the brute-force solution to increase the breakdown voltage level.This solution has been tested in the QST labs in Naka and it is going to be implemented in MITICA.Nevertheless, this solution poses complex problems, especially as far as the remote handling procedure for the maintenance is concerned for the ITER NBIs.The effective solution of the issue of voltage holding in the ITER NBI accelerator is then strictly related to a deeper understanding of the breakdown mechanism across long gaps.In the frame of the Breakdown Induced by Rupture of Dielectric (BIRD) model [3], the main voltage breakdown driver is the Explosive Emission from the cathode of electrons; due to the high voltage, they gain enough energy to cause, when hitting the anode surface, desorption of gas in a quantity sufficient to triggers an avalanche process.This breakdown mechanism in somehow merges the "ecton" theory on the Explosive Electron Emission (EEE) [4] with the Clump Theory [2], [5], in which the "ecton" electron cluster takes the place of the micro-particles (clumps).
In the BIRD model the EEE is consequence of localized micro breakdown in the non-conductive Cr 2 O 3 layer of the stainless steel cathode.The micro-breakdown are due to the growth of the internal layer E field until the dielectric strength value is reached, due the mismatch between the Fowler-Nordheim [6] Emission current from the layer to vacuum and the conduction current in the layer.The macroscopic effect of these micro-breakdowns of the layer is the occurrence of current bursts, associated to out-gassing and X ray emission.The characterization of the FE current emission from a nonconductive layer is then a prerequisite for a better understanding of the burst formation and, eventually, for controlling their occurrence.Focus of the paper are the mechanisms underlying the FE emission from insulated layer, analyzing the effect of the applied electric field on the shape of the energy barrier in the two interfaces metal-to-insulator and insulator-to-vacuum.This analysis gives an explanation of the FE current switchon caused by a permanent modification of the energy barriers.Tests have been carried out on samples of stainless steel electrodes with a standard treatment and with electropolishing treatment and on a sample of electrodes covered by a 100µm alumina layer; the results are compared and discussed in terms of the energy barrier modification.

II. CONDUCTING AND INSULATING LAYERS
During the voltage conditioning of electrodes in vacuum there is the experimental evidence of DC emission punctuated by sudden current peaks, usually referred as micro-discharges.Different processes have been proposed to explain this intermittent behavior: micro-particles [2], [5] or cathode-tip thermal instabilities.Another process, considered in literature [4], [7] but not so far deeply delved, is the breakdown of the poorly-conductive oxides layer that cover most of the cathode metals (Al, Cu, Mg, Stainless Steel).Recently this process has been deeper investigated [3], [8], especially for Stainless Steel cathode, covered by a Cr 2 O 3 layer.In presence of a Metal-to- Insulator interface (MI interface), once an electron emission has started at the layer surface, there is an electrons accumulation at the MI interface that has two consequences: the increase of the field within the dielectric layer and a gradual lowering of potential energy profile along the layer.In the first case, the internal field reaches the dielectric strength value of the material and a micro-discharge occurs inside the layer.In the second case, the lowering of the layer potential energy makes possible current conduction through the dielectric.In this case, the FE switches to a higher DC current [7].Should be noted that both these effects constitute, in our understanding, the fingerprint of the presence of nonmetal interface between the metallic electrode and vacuum.In the following we report experiments that use three types of axialsimmetric stainless AISI 304 electrodes obtained by: 1) standard finishing: the surface was smoothed with abrasive sheets covered with Al 2 O 3 , grain size from 30 to 5 micrometers.The resulting surface roughness is R a < 0.4µm.The electrodes were cleaned in an ultrasonic bath with alkaline detergent for 1.5h at 50°C, rinsed with demineralized water for 1h at 50°C, dried with N2 spray.
2) electropolished electrodes: the surface has been smoothed by electropolishing in a galvanic solution .The electrodes have been cleaned in ultrasonic bath with alkaline detergent for 1.5h at 50°C, rinsing with demineralized water for 1h at 50°C , drying by N2 spray.
3) alumina coating :a substrate of Al 2 O 3 has been deposited by plasma spray up to a thickness of 0.2mm.

III. EXPERIMENTS
The tests have been performed in a classical (emi)sphere (cathode) and plane (anode) configuration of stainless steel AISI 304 electrodes, with three different treatments.Figure 2 shows the electrode with their dimensions.

A. Standard machined AISI 304 electrodes
We operated the first pair of standard machined AISI 304 electrodes at a gap distance d = 1.0mm.Referring to Fig. 2, for the first half hour the voltage was set at 30kV , then after another half hour break, the voltage was set at 35kV and left for seven hours.After the application of 35kV current started to be measured and some micro-discharges occurred.However, the zoomed area of Figure 3 shows that the application of Voltage and the start of the DC component of the current occurs with a dealy of about 60s.The start of the DC current, to few µAmps is referred as the First-Switch-On point (FSO).Although the voltage did not change, another sudden current jump to several tens of µAmps, called the Last-Switch-On (LSO), occurred about six and a half hours later.In the following days of operation, the current level did not change, though the electrodes have been left in free air for several days.The tests have been repeated with new electrodes with the same size and the same treatment process.The first day the same voltage profile of the previous experiment, has been applied, but no change in the current appeared.The second day the voltage has been increased to 38kV and after more than four hours a very small direct current (FSO) showed itself.The second current jump (LSO), took place, at a voltage of 48kV , about ninety hours after the start of the experiment, a much longer time than what happened in the first experiment.These two experiments show very similar behavior, albeit with Fig. 4. AISI 304 -electrode couple N.2 -Normally treated very different timings.One possible explanation is that the switch-on timing depends upon difference of the status of the surface in the higher electric field region (the lower the surface roughness, the longer the time).To test this hypothesis, the next experiment has been done using electro-polished electrodes, a treatment that guarantees the lowest roughness .

B. Electropolished AISI 304 electrodes
Electropolishing [9] technique uses an electrochemical process that removes more efficiently than mechanical treatment the surface tips Moreover, this treatment improves stainless steel passivation, with a surface enrichment of chromium.
Also in this case we tested two pairs of electrodes.In the first test, the AISI 304 electrodes have been subjected to voltages from 35kV up to 60kV , with increasing step of 5kV , for tens of hours each.No switch-on occurrence was found.Then the test started on a second pair of electropolished electrodes.In this case the same voltages from 35kV to 60kV were used, with an overall time of less than 3 hours.We held 60kV for about 170 hours and then increased it to 65kV for another 100 hours.Apart some current bursts, no switch-on occurrence has been detected.Later, instead of increasing the voltage, and then the risk of voltage breakdown (that damages the status of the surfaces) it was preferred to halve the gap distance (0.5mm).We reached the FSO after about 360 hours, but this DC current of few µAmps disappeared after ten hours, as can be seen in the zoomed area of Figure 5. Next, after 425 hours, at a voltage of 39kV (electric field of 78kV /mm), the LSO occurred.The current jumped to more than 60µA.As expected, the time required to reach the switch-on points, is much longer than standard machined electrodes.

C. Alumina covered cathode
In the next experiment we used an alumina coated cathode.The voltage was raised from 35kV up to 54kV where a sudden current jump occurred.As can be seen from Figure 6, in the phase preceding the current jump, there are many more microdischarges than in the AISI 304 experiment.Furthermore, we see that after the switch-on point, the evolution of the direct current is very jagged.It therefore seems that even after the activation of the DC current the micro-discharge activity continued.Clearly a more thorough investigation is needed to clarify this behavior.In any case, we can undoubtedly attest that even in this experiment the switch-on point is clearly achieved.

IV. THEORETICAL APPROACH
The previous experiments evidently share some important characteristics, such as, for example, the presence of microdischarges, the sudden current jump to a several tens of µAmps DC level (LSO) and the great resilience demonstrated once this state is reached.Other characteristics, such as the presence of the FSO and the relative DC current levels, clearly visible in some experiments, appear not to be so in others.Furthermore, timing can be very different from one experiment to the other.To get a deeper understanding of the phenomena involved, we have developed a very simple 1D (naive) model, shown in Figure 7.This represents an insulating layer attached to a piece of metal.We assumed, for simplicity, that the potential energy profile U (x) is obtained, by double integration, from a sinusoidal charge density profile ρ(x) at the material boundaries.In Figure 7, a small transfer of negative charge from the insulator to the metal has also been applied to balance the two Fermi levels.The metal work function W and the electron Fig. 7.The neutral Insulator-Metal model with metal work-function W and insulator electron affinity χ.The potential energy profile U (x) is obtained from the sinusoidal charge density ρ(x).Some negative charge was displaced from the insulator to the metal.affinity of the insulator χ are indicated and the band gap of the insulator is highlighted by the dotted area.When a power supply is connected to a pair of metal electrodes covered by a dielectric layer, the main effect, on the cathode side, is to increase the electron density of the metal near the insulator-metal interface, as indicated by the green arrow of Figure 8.This produces two main results: a downward shift in the potential energy profile of the insulator and an increase in the average electric field within the dielectric layer.In principle it is possible for the applied voltage to lower the energy profile such that the Fermi level of the metal is above the bandgap.However this requires a very high vacuum electric field, as shown in the example of Figure 8, where an external field of about 1 M V mm has been simulated.Furthermore, the electric field inside the dielectric also reaches very high values, most likely higher than the dielectric strength of the material.In recent works [3], [8] it has been assumed that in high voltage and high vacuum experiments a DC electron current can be generated, due to the electrons detached from a dielectric cathode covering layer.In this case electrons reach the cathodic metal edge next to the dielectric layer via the anode and the voltage generator and there they build up.This results in a consistent lowering of the potential energy profile, even with much lower applied electric fields.See figure 9.However, the electric field within the dielectric layer is bound to increase to a value greater than the dielectric strength, so a relaxation mechanism of the electric field inside the dielectric layer must also be invoked for consistency.In Latham book [7] it has been suggested that inside the insulator there might be many electron traps and electrons can move (jump) and get trapped.This give rise to a field relaxation mechanism.
In Figure 9 we have moved some insulator negative charge towards the outside of the layer, in order to maintain a lower internal electric field.The model presented is obviously very basic and in no way we claim it is exhaustive, nevertheless it can provide a key to understanding many experimental facts.It depicts a scenario in which the application of an electric field in high vacuum, possibly enhanced by local geometric asperities, gives rise to a current, due to electron detachment from the dielectric surface.These electrons accumulate on the outer metal edge at the cathode side, increasing the electric field and lowering the insulator energy profile.When the metal Fermi level exceeds the insulator band-gap a resonant electron tunnelling occurs (FSO).A further lowering of the energy profile causes the Fermi level of the metal to exceed the whole energy profile of the insulator, with a consequent new sudden increase in the emitted DC current level (LSO).Furthermore, each time the insulator field exceeds the dielectric strength a micro-discharge occurs.The increase in the insulating field can possibly be attenuated by an electron trapping mechanism within the dielectric material.As a last consideration we observe that we have considered the band theory of the solid state, which is valid only if the layer is very thick.In the case of the AISI 304 electrodes, we know that they are covered by a Cr 2 O 3 layer only a few nanometers thick.Anyway, the basic idea of the model still remains valid, in the last case it is sufficient to replace the bands with the discrete states of the system.This likely introduces the possibility of experimentally obtaining various DC current discrete states.

V. CONCLUSION
In this paper we have reported four experiments carried out to better understand the evolution of the current between high voltage high vacuum electrodes.In the first three we used AISI 304 electrodes, standard machined or electropolished.In the forth the cathode was covered with an alumina layer.In each experiment it was found a sudden jump of the DC current to several tenth of microamps.These points are reached at different voltage, but they usually do not appear at the start of the new voltage step.A basic model was presented to elucidate a possible mechanism behind this feature.Ultimately it leads back the phenomenon to the accumulation at the metal-insulator interface of electrons detached from the free surface of the insulator due to the electric field.

Fig. 1 .
Fig. 1.Vertical Section of the MITICA vacuum vessel and beam source (left) and horizontal section of the ion source and electrostatic accelerator (right).

Fig. 8 .
Fig. 8. Highly powered cathode.The metal Fermi level is above the insulator band-gap.Resonant tunnelling of metal electrons occurs.

Fig. 9 .
Fig. 9. Powered cathode.The negative charge Q B detached from the outer edge of the insulator and accumulated at the Insulator-Metal interface, lowering the energy profile.A field relaxation mechanism allows the value of the internal field to be kept low.