During the first set of axial measurements with the hairpin probe, an increase in the electron density through the magnetic filter in the ICPS was immedi- ately noticed for hydrogen. The operating pressure of 50 mtorr was the first pressure used during initial testing which is higher than that used some ap- plications including NBI sources, however it will be shown that the may be formed at lower pressures which are applicable to NBI relevant conditions.
The electron density profiles were found to be repeatable over several days of testing and so a series of profiles were taken at different pressures to quantify the pressure dependence of the density peak. Six pressure ranges were then chosen from 10 mtorr to 150 mtorr to investigate if the increase in density through the filter occurred only at 50 mtorr. The forward power for the ex- periments was 200 W and the applied magnetic field used was the standard 250 Gauss filter.
Note that in inductive hydrogen plasmas with less than 1 kW power, the dom-
inant ion present in the plasma is H+3 which increases also with higher gas
pressures. This has been shown theoretically in ICP modelling [105] [106] [93] and also through early experimental verification using a mass spectrometer which was installed on the ICPS.
Figure4.1shows the results for the axial electron density profiles at a variety
of pressures in the ICPS with the red profiles taken with the magnetic filter applied and the black profiles show the results without the filter. The elec- tron density profiles taken without the filter show the common exponential decline in density with distance from the driving antenna (linear in the log- scale shown here). With the application of the filter however the formation of a peak in the electron density profile can be seen. The peak forms initially at a pressure range between 10 mtorr and 25 mtorr and reaches a maximum between 50 mtorr and 75 mtorr. At 150 mtorr the density profile looks to be transitioning away from the trapping regime. The density peak appears lo- calized around the center of the filter and is predominantly bound to the area between physical edges of the permanent magnet blocks. The largest dif- ference between magnetically filtered and unfiltered densities occurs in the center of the filter at x = 5 cm and is more than three times higher for the 50 mtorr and 75 mtorr peaks. Electron densities for hydrogen at this power are seen to range from 2 ×1015 m−3 up to 7×1016 m−3.
To test whether or not the density peak was an artefact from anisotropic elec- tric permittivities arising from an effect of the magnetic filter, density profiles were measured using both parallel and perpendicular orientations of the hair- pin probe with respect to the magnetic field lines. The results were shown
previously in the diagnostics chapter (Figure2.18in Chapter2) and show the
density peaks are not a measurement error from anisotropy in the plasma permittivity.
(a) 10 mtorr. (b) 25 mtorr.
(c) 50 mtorr. (d) 75 mtorr.
(e) 100 mtorr. (f) 150 mtorr.
Figure 4.1: Hydrogen plasma electron densities measured using the hairpin probe.
This same density increase seen in Figure 4.1 was also seen by Cho et al.,
[133] in their negative ion source showing a similar width and increase in-
side the peak. While the magnetic field strength here of 250 G is similar to their peak field of 220 G, the peak seen here is centered on the peak magnetic field position rather than a few centimeters upstream of the peak. This may be due to the presence of a bias electrode near the filter in their work. The pressure of the peak formed in this work begins above 10 mtorr and has a maximum around 50 mtorr to 75 mtorr which is above the pressures used in
NBI devices. However, in the work of Cho et al., [133] the peak forms at a
lower pressure of 3 mtorr (0.4 Pa) and so this feature is reproducible at NBI relevant pressures.
The ability to produce the density peak and tailor its creation at different gas pressures is due to a number of factors including collisional frequency, ion mass, magnetic field strength, and the presence of a tandem cusp field. These
factors will be discussed later with transport modelling in Section4.4and the
criteria for generating this trapping effect in a source will be presented. It will
be also be shown in Chapter 5that optimization of this density peak may be
exploitable for use in fusion relevant plasma sources using different magnetic field combinations.
While electron-driven processes are usually used to explain cross filter trans- port, these results for hydrogen can instead be explained by considering ion- driven plasma transport initiating the cross filter diffusion followed by parti- cle trapping occurring over microsecond time scales for ion diffusion.
Due to the low ionization fraction in the ICPS the dominant collisional trans- port processes in the plasma are the elastic collisions between charged par-
ticles and neutrals [92] [134]. At pressure ranges between 1 mtorr and 100
mtorr, the light-weight H+3 ion dynamics of hydrogen plasmas have several
different properties which can contribute to this trapping behaviour. Mod-
elling of the filter dynamics in Section 4.4 of this trapping effect will show
that the filter particle transport is dominated by the ion transport properties
and ∇B drift fields but a qualitative explanation of the process will be ex-
plained briefly here.
At the lower measurement pressures here of around 10 mtorr the H+3 ions ex-
hibit weakly collisional magnetized kinetics characterized by long mean free
At intermediate pressures between 25 mtorr and 75 mtorr the combination of moderate mean free paths and Larmor radii on the order of the source tube width allows for strong cross field transport from upstream near the antenna into the central region of the magnetic filter where they become magnetized and trapped.
At higher pressures of 100 mtorr to 150 mtorr, the ions are highly collisional and freely diffuse throughout the source across field lines at long timescales
of 100 µs to 500 µs. Table 4.1 summarizes the ion properties of H+3 through
the filter region. At the trapping pressure of 50 mtorr the Table4.1shows that
the momentum exchange collisions per cyclotron orbit (νm ·ω−ci1) is approxi-
mately 1.91 meaning that collisions occur at each half cycle which allows for the maximum cross field displacement per collision. This idea is qualitatively
supported by Hayashi et al., [135] who used different ion mass plasmas to
show an indirect relationship between ion mass and filter transport. Their work reasoned that the ion Larmor radius is related to to filter transport, but they provided no density measurements or transport theory.
Table 4.1: Leading collision parameters for dominant hydrogen ions (H+3) in the
ICPS. 0.05 eV ion energy at (10 Gauss / 250 Gauss) respectively in the brackets
Parameter 1 mtorr 10 mtorr 50 mtorr 100 mtorr
Mean free pathλm f p(cm) 41 4.1 .83 0.04 Cyclotron frequencyωci(kHz) (5 / 126.9) (5 / 126.9) (5 / 126.9) (5 / 126.9)
Collision frequencyνm(kHz) 4.85 48.5 242.8 485 Larmor radiusrL(cm) (6.3 / 0.25) (6.3 / 0.25) (6.3 / 0.25) (6.3 / 0.25)
Collisions per cyclotron orbitνm·ωci−1 (0.97/0.38) (1.7/0.38) (48.56/1.91) (97/3.82)
These initial experimental results for hydrogen on the ICPS show that the transverse filter can cause particle traps to occur which increase the local plasma density by several factors. This trap has been previously shown to
form at NBI relevant pressures [133] and so this chapter will provide a suite
of measurements and modelling of the source to elucidate the mechanisms which lead to its formation. This work could help inform a future set of ex-
periments which could specifically target this phenomena and optimize the process on an up-scaled negative ion device at lower pressure.
It should be noted that when the magnetic field is in place, a ’dark’ region occurs in the vicinity of where this peak in plasma density occurs (See Fig- ure 2.2(b) from Chapter 2 for a photograph of the dark region). Previously
shown ion saturation and electron current data (Figs. 2.13) support these den-
sity measurements which indicate that this region is certainly not devoid of plasma. Further supporting evidence of the density peak using EEPF mea-
surements will be shown in Section4.2. The lack of light emission under the
magnets implies low levels of electronic de-excitation are taking place within the filter region. Recombination events are rare within the plasma bulk due
to the low cross section for radiative recombination [136] and the majority of
recombination occurs due to charged particle flux to the walls with an emis- sion in the invisible ultraviolet region. Despite this, successful measurements of the optical emission through the dark region were made and will be shown in Section4.3.