Ion flux measurements using a Mach-Langmuir probe in the ITER prototype neutral beam injection ion source

Neutral beam injection systems as foreseen for ITER use radio-frequency (RF) ion sources at low pressure, where negative hydrogen ions are mainly produced via surface conversion of neutral atoms and positive ions at a plasma facing grid (PG). Up to now there is only limited knowledge about how fluxes and directed velocities of the positive ions are affected by external parameters such as power, pressure and the horizontal magnetic filter field which causes plasma drifts and vertical asymmetries in the vicinity of the PG. For this reason a combined Mach-Langmuir-probe diagnostic is used at multiple positions in the expansion and close to the extraction system in the prototype RF ion source (1/8 of the full ITER ion source size) to measure the positive ions directed velocity and flux as well as the plasma parameters simultaneously. With increasing RF power the flux towards the PG is found to increase linearly, its magnitude being controlled by the plasma density. Towards ITER-relevant pressures the ion flux decreases, in contrast to the directed velocity, which increases non-linearly, reaching around 5 km s−1 at a pressure of 0.3 Pa. The magnetic filter field is discovered to strongly bent down the ion flow in front of the PG. As a result, the ions at the lower half of the PG flow almost exclusively parallel to it, wherefore the flux which impinges onto the lower PG half is reduced by around one order of magnitude.


Introduction
In the fusion experiment ITER [1] a powerful neutral beam injection (NBI) system is needed.Here negative hydrogen or deuterium ions (H − or D − ) have to be created, extracted, * Author to whom any correspondence should be addressed.
Original Content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.accelerated and neutralized before being injected into the magnetically confined fusion plasma to heat the latter via collisions [2].The negative ions are generated in an ion source, where a low temperature hydrogen or deuterium plasma is produced via inductive radio-frequency (RF) coupling in multiple cylindrical drivers [3].Each of these drivers has around ten liters volume.To comply with the negative ion current density as demanded by ITER, up to 100 kW of RF power are needed per driver [3].At the same time the pressure in the ion source must not exceed 0.3 Pa, since otherwise too many negative ions are destroyed by collisions with the background gas within the extraction and acceleration grid stages.The drivers are attached to one common expansion region, where plasma species such as positive ions and electrons, but also atoms and neutral molecules expand into.The majority of negative ions originate from the surface of a plasma facing grid (herein abbreviated as PG), which is located across from the drivers and is the first grid of the multi-stage extraction and acceleration grid system.The surface creation mechanism of negative ions at low work function surfaces is complex and a comprehensive review of this topic is presented in [4].In the case of the RF negative ion source a thin layer of cesium covers the PG surface, in this way lowering its work function and thus favoring secondary emission of negative ions.
From this mechanism it becomes evident that the particle fluxes of atoms and positive ions impinging on the PG (along with the absolute value of the work function of a cesiated surface, as investigated in [5]) are important for the negative ion yield.Information about the atomic density, flux and temperature in front ot the PG was recently obtained via a TALIF diagnostic [6].Measurements of the ion Mach number, i.e. the ratio of ion flow velocity compared to the local ion sound speed have been performed in different RF negative ion sources by [7,8].However, the Mach number has to be combined with information about the local ion sound speed and the plasma density.Only then it is possible to obtain the directed ion flux in the expansion and directly in front of the PG, which is not well known until now.Consequently, it is also not clear how different external parameters such as RF power and pressure affect the positive ion flux.Furthermore there is a transverse magnetic filter field B filter , which cools down the electrons in front of the PG to reduce the number of co-extracted electrons as well as the negative ion losses [9].This filter field is known to cause the electrons in the plasma to drift [10,11], in this way influencing the positive ion flux onto the PG, which in turn affects the negative ion yield.Besides that, positive ions are needed at the extraction apertures for compensation of the space charge (created by negative ions), which is important for the extraction of negative ions.Related to the plasma drifts, plasma parameters such as e.g. the plasma density are also affected by the magnetic filter.Therefore an impact on the number of co-extracted electrons and on the produced beam-which also shows vertical asymmetries [12]is to be expected.Knowledge about the positive ion flux and how it depends on external parameters is therefore needed as a necessary requirement to identify measures for optimizing the negative ion yield, its homogeneity as well as the number and symmetry of co-extracted electrons in NBI ion sources.
In this work a combined Mach-Langmuir probe diagnostic is used to obtain the above mentioned parameters and dependencies in the prototype RF negative hydrogen ion source at the BUG test bed [3] for the first time.In section 2 the experimental setup as well as the newly designed diagnostic are introduced.Obtained positive ion fluxes at various positions in the expansion and in front of the PG are shown and discussed in section 3 for the different external parameters power, pressure and magnetic filter field strength.In the conclusion section 4, potential use of the obtained data and insights are highlighted.This mainly concerns input and validation data for fluid and kinetic models [11,[13][14][15], which could then be used for optimizing the positive ion and atomic fluxes towards the PG.In this way it becomes possible to improve the negative ion yield and minimize the asymmetry of the co-extracted electrons.

Experimental setup and diagnostic
The ITER prototype RF ion source (1/8 size of the ITER source) at the BUG test bed [3] is used as experimental setup (cf the cutaway drawing in figure 1).Here hydrogen or deuterium gas is injected into the driver to produce a low filling pressure p fill , which is typically between 0.3 and 0.7 Pa.During the performed campaign only hydrogen is used as feed gas.Furthermore, no cesium is evaporated in order to guarantee a high reproducibility of the measurements and to protect the Mach-Langmuir probe from the chemically active cesium.The plasma grid can be electrically biased with respect to the source walls.In all performed discharges the bias grid voltage is kept below the local plasma potential, which results in an electron repelling sheath [16].Besides that, no high voltage and consequently no negative ion beam extraction is applied.
A low temperature plasma is generated via inductive RF coupling, where RF powers P RF of several tens of kW are usually applied.In the expansion region there are two magnetic fields.First, there are the electron deflection magnets embedded in the extraction grid, which is the second grid after the PG [3].These magnets deflect co-extracted electrons onto the extraction grid and thus remove them from the beam.The deflection field is spatially highly non-uniform and acts close to the PG in |z| < 15 mm distance, where its magnitude is in the range of 5 to 25 mT, thus dominating the one of the magnetic filter field B filter [17].The latter is used to dilute and cool down the electrons in the expansion.The filter field is generated by a DC current flowing from the bottom to the top of the PG.Hence B filter is mostly horizontal, i.e. pointing in the positive x-direction and decreases in z-direction, as indicated for three different values of the filter field current (0, 1.5, 3 kA) by the plot shown in the lower left of figure 1.Typical strengths of B filter in the expansion are in the single-digit mT range, whereas it is lower than 1 mT in the driver.Here its direction is also inverted.At the largest filter field strength of around 6 mT the electron gyro radius is around 1 mm and the Hall parameter, i.e. the ratio of gyro frequency to collision frequency in the order of 100 ≫ 1. Hence the electrons can be considered fully magnetized.For the heavier and colder ions, the gyro radius is around 1 cm and the Hall parameter is around 1. Hence, the ions are not fully magnetized.
During the measurement campaign the external parameters RF power P RF , filling pressure p fill and magnetic filter field strength B filter , are varied systematically within the given ranges, as stated in table 1.
The red circles in figure 1 indicate where the combined Mach-Langmuir probe diagnostic can be attached: to five different horizontal and one vertical flanges at z = 2.5 cm axial distance to the plasma grid as well as to three horizontal  (of which only two are visible) and one vertical flange in the expansion, which are at z = 19 cm.As an example, the probe diagnostic is placed horizontally in the vertically central flange close to the plasma grid.The probe has an axial length of 50 cm to fully cover the plasma grid surface and the expansion height.
As can be seen in the zoomed in picture at the upper left of figure 1, there are two individual diagnostics: at the far left there is a Langmuir probe with a Tungsten tip (diameter ≈ 0.3 mm) and a length of 8 mm.Spaced by a ceramic cylinder of 6 mm length there is the compensation electrode as a vital part of the RF compensation, used to separate the RF oscillations from the measured voltages, as introduced and explained in [18].From a typical current-voltage-characteristic I(V), various plasma parameters such as plasma potential ϕ plasma , electron temperature T e , and density n e are derived using standard methods described in [19].This involves the following steps: (i) The plasma potential ϕ plasma is evaluated by numerically deriving d 2 I/dV 2 and determining where it crosses zero.During the numerical derivation procedure smoothing is applied by means of a Savitzky-Golay filter, which typically extends over 2-3 volts and in this way determines the error of ϕ plasma .(ii) The electron temperature is determined from a fit onto the linear part of the slope of ln(d 2 I/dV 2 ), where ln(•) denotes the natural logarithm function.(iii) The plasma density is obtained by evaluating the current at the plasma potential.Note that when the magnetic field is parallel to the probe axis, this evaluation yields an underestimation of the plasma densities, which needs to be corrected.How this is done is explained in appendix.
Located at a center-to-center distance of 2.5 cm behind the Langmuir probe, there is the Mach probe.It consists of four cylindrical tungsten pins (same diameter and exposed length as the tip of the Langmuir probe), which are equally distributed around the circumference of the ceramic cylinder, which has a diameter of 4 mm.Each pin is embedded into the ceramics, such that only 180 • of its lateral surface is exposed to the plasma.To account for different contact sensitivities of each of the four Mach pins, the probe is rotatable around its central axis via a magnetic manipulator, as shown in light green in figure 1. Contact sensitivity correction was performed in regular intervals throughout the whole campaign.It was found that contact sensitivities of the different pins deviated by around 10%.
The directed ion flux n ions u ions is defined as the product of the plasma density n ions (≈ n e ) and the ions directed velocity u ions .The absolute value of the ions directed velocity u ions can be calculated with the help of the Mach number M ions and the Bohm velocity u Bohm , i.e., u ions = M ions u Bohm . (1) Experimentally, the Mach number can be obtained by comparing the ion saturation currents of two opposed Mach probe contacts up-and down-stream, as explained in [20], i.e.
Herein, K is a calibration factor, which is obtained from a kinetic particle-in-cell model of a spherical probe immersed in a flowing plasma [21].Values of K resulting from the conditions as in the performed campaign are K = 1.34, when no magnetic filter field is present, and K = 1.38 when a magnetic filter field is present (corresponding to B filter ≈ 6 mT, as measured in the expansion of the ion source).Note that from equation ( 2) follows that it is not possible to obtain the fluxes of the individual ion species separately, since the upstream and downstream saturation currents I upstream and I downstream include all positive ion species.The Bohm velocity u Bohm in equation ( 1) is a function of the electron temperature and the ions mass, i.e., Herein, e denotes the elementary charge and T e is measured in units of electron volts.For the calculation of the Bohm speed it is necessary to estimate an effective ion mass, which is done following [18].This results in m ions = 1.8 m H + , according to the individual fractions of 40%, 40% and 20% for H + , H + 2 and H + 3 , respectively.Using superposition, it is possible to obtain the ion flux component in the direction perpendicular to the probe axis.Hence, for the horizontally attached probe, as shown in figure 1, the flux components perpendicular as well as parallel to the PG are obtained.

Power and pressure variation
Figure 2 shows the effect of variations of the RF power.For this sweep the probe is horizontally attached to the vertically central flange and extended into the center of the PG at an axial distance of z = 2.5 cm, as indicated in figure 1.The sweep is performed at a fixed filling pressure of 0.3 Pa and a magnetic filter field current of 1.5 kA resulting in a strength of around 3 mT at the measurement position.Plotted in the left picture is the ion flux towards the PG.With increasing P RF , an almost linear increase of the flux is observed, as indicated by the dashed red line.As shown in the centre and right pictures of figure 2, also the plasma density increases linearly with power, whereas the ion velocity towards the PG saturates for P RF > 40 kW at around 5 km s −1 .Note that the electron temperature is around 2 eV, and hardly changes with power.The conclusion being that the increasing ion flux with power is largely controlled by the behavior of the plasma density.Also the Mach number increases monotonically from 0.2 at 10 kW to 0.45 at 70 kW (not shown in figure 2).The error bars result from a Gaussian error propagation scheme, where measurement errors of the plasma density, the Mach number, the electron temperature and the effective ion mass are accounted for.The observed peak in the ion velocity towards the PG at 20 kW seems to be genuine, since it is also seen at 30 kW when no magnetic filter field is present.However, the reason for it is not known to the authors.
Figure 3 shows a pressure variation at the same position of the diagnostic.In this case the power is fixed to 60 kW and the magnetic filter field is again around 3 mT corresponding to 1.5 kA of PG current.It can be observed that the flux decreases towards lower pressures.As in the case of the power variation, the flux qualitatively follows the trend of the plasma density.The ion velocity increases non-linearly towards lower pressures.This increase can be explained by the larger electron temperature and plasma potential at low pressure.The electron temperature is around 2.4 eV at 0.3 Pa, whereas it is around 1 eV at 0.7 Pa.The plasma potential in front of the PG is around 28 V at 0.3 Pa and 17 V at 0.7 Pa.Hence the ions gain considerably more kinetic energy in the plasma sheath.Besides, there are less collisions between ions and neutrals at lower pressures, wherefore less momentum is transferred resulting in a larger directed ion velocity at the PG.The Mach number is roughly constant at M ≈ 0.45 (not shown).

Impact of the magnetic filter field
Figure 4 shows the impact of the magnetic filter field on the positive ion flux magnitude and direction, measured at several locations in the horizontal center of the source.
Regarding its direction, without magnetic filter field the ion flux (indicated as black arrows) is almost perfectly symmetric with respect to the horizontal plane at height y = 0.When the magnetic filter is switched on, the ion flux is bent downwards at all measured positions, the bending angle increasing with the filter field strength.In the lower half of the PG this results in a flux, which is almost parallel to the PG (cf blue and red arrows there).More quantitatively, figure 5 shows only the vertical component of the ion flux at the PG for the three different filter field strengths.Indicated by dashed lines is the rectangular shape of the projected PG.Without magnetic field, the flux is almost perfectly top-bottom symmetric, i.e. the ions in the top half of the PG flow upwards and the ones in the bottom half flow downwards, the magnitudes increasing with the vertical distance to y = 0.However, when the magnetic filter field is switched on, also the ions at the top half of the PG start to flow downwards.Results from fluid models [11,14] indicate that this behavior is indirectly caused by the Lorentz force in the electron momentum balance, which affects the vertical component of the electrostatic field, such that the ions experience an additional downward force.Note that the Lorentz force term in the ion momentum balance adds only a negligibly small downward force, which is consistent with the estimation in section 2, that the ions are not fully magnetized.Figure 6 shows the other obtained component of the ion flux, i.e. the one which points towards the PG.The upper row shows a cut through the expansion at z = 19 cm and the lower row a cut in close vicinity of the PG, i.e. at z = 2.5 cm.Without magnetic filter field the magnitude of the flux towards the PG decreases slightly from the expansion towards the PG, as is also shown in the left column (note the different color scales for expansion and PG).However, when the filter field is present, there is a large increase of the flux in the expansion plane, which comes out of the driver (compare the three plots of the upper row).In absolute numbers, the flux component towards the PG increases from around 10 21 to 5 • 10 21 m −2 s −1 .This increase of the flux results partly from an increased plasma density, since the magnetic filter field tends to push the plasma back into the driver, and from the change in the electron momentum balance, where the Lorentz force leads to an increase in the axial component of the electric field [11,14].In front of the PG the situation is different, as can be seen from a comparison of the three plots in the lower row.Due to the downwards bended flux, as already shown in figures 4 and 5 the flux component towards the PG is strongly decreased, especially in the lower half of the PG, where the flux is reduced by almost one order of magnitude.Note that there is almost no left-right asymmetry associated with the filter field observable in the ion fluxes, as shown in figures 5 and 6.This is also true for the obtained electron temperature and the plasma potential.This behavior is expected, since the asymmetries are associated with electron plasma drifts, which are absent in the x-direction.
Figure 7 shows a vertical position sweep, performed at a power of 60 kW and a pressure of 0.3 Pa.As can be seen in figure 1, the position of the vertical flange is slightly off-center at x = 2.5 cm.The three dashed vertical lines separate the bottom from the top half of the PG.The perpendicular ion flux component towards the PG is shown in plot (a), whereas the corresponding plasma density and directed velocity are shown in the plots (b) and (c), respectively.Note that without filter field, all three quantities are symmetric in the top and bottom over the PG.
When the magnetic filter is switched on, the flux onto the top half of the PG decreases only slightly as a result of a slightly increased plasma density and a largely decreased velocity.In contrast to that, at the bottom half of the PG the flux decreases significantly.The largest decrease is at the lowest position of the bottom half of the PG (i.e.y = −15 cm), where the flux impinging on the PG is reduced by a factor of around 50 when compared to the situation without filter field.This decrease follows from the decreased plasma density and directed velocity, as is evident from plots (b) and (c).Note that the very low plasma density at the bottom of the lower PG half (where n ions ≈ 10 16 m −3 ) could potentially harm the production of H − , since there could be too few positive ions to compensate for the space charge created by the H − .
The slight but systematic increase of the plasma density in front of the upper PG half when compared to the lower PG half is shown in more detail in the lower row of figure 8.Note that in the expansion, the situation seems to be reversed, as shown in the upper row of figure 8.The increase of the plasma density in front of the upper PG half could contribute to the amount of co-extracted electrons, which is a limiting factor of the source performance [10].Earlier measurements performed with uncompensated pin probes at two locations (see positions marked with red diamonds in the lower row plots in figure 8) also show a slight increase of the ion density [9] and are thus consistent with the values obtained in this work.In contrast to that, results from optical emission spectroscopy (OES) and collisional radiative (CR) modeling [22] indicate a slight decrease of the plasma density in front of the PG for a horizontal line-of-sight at y = 0. Note however, that the error bars in the case of the OES evaluation of the radiation from a recombining plasma are comparatively large due to the multitude of existing excitation channels.In the expansion however, also the plasma density obtained using OES and CR modeling shows an increase with increasing I PG at the line-of-sight at y = 0, as is also evident in this work (cf upper row of figure 8).In terms of absolute values, it is found that all three diagnostics agree within the error bars.

Conclusion and outlook
A newly designed combined Mach-Langmuir probe diagnostic was successfully applied in the ITER prototype RF ion source.With this probe it is possible for the first time to simultaneously obtain the directed ion velocity and density and in this way the directed flux of positive ions.A large number of spatial measurement positions yields densely populate 2D flux maps in the expansion and in front of the extraction grid system.The impact of several external parameters, such as RF power, pressure and the strength of the magnetic filter field are investigated.It is found that the positive ion flux, which impinges on the PG, scales linearly with power, its magnitude mainly driven by the plasma density, whereas the directed ion velocity towards the PG saturates for P RF > 40 kW at around 5 km s −1 .Towards lower pressures the positive ion flux as well as the plasma density decrease, in contrast to a non-linearly increasing ion velocity.Regarding the magnetic filter field, it is found that an almost perfectly top bottom and left right symmetric positive ion flux (when no magnetic filter field present) is strongly bent downwards when the filter field is switched on.This leads to a top bottom asymmetry, where the flux onto the lower half of the PG is largely decreased, whereas the one onto the upper half is only slightly reduced.Also the plasma density is found to become vertically asymmetric when the magnetic filter is present.Since the impacts of the above mentioned three major external parameters on the fluxes and plasma parameters are quantified, a further experimental campaign should target to investigate the influence of the used isotope (i.e. by using deuterium instead of hydrogen) as well as the role of the electrostatic plasma grid bias.
The trends and absolute values of the fluxes and plasma parameters obtained in this work provide valuable input and validation data for fluid and kinetic models that attempt a rigorous description of the physics in the full ion source as well as in the region close to the extraction.Hence this data enables more detailed and reliable control of the atomic and ionic fluxes onto the plasma grid in order to optimize the surface produced negative ions as well as the number and symmetry of the co-extracted electrons.

Appendix. Influence of the magnetic filter field on the plasma density evaluation
As briefly noted in section 2, when the magnetic filter field is present the electrons are fully magnetized.Hence it is to be expected, that the evaluation of the electron branch, where the plasma density is deduced from the current at the plasma potential, is influenced by the magnetic field.To investigate this influence, the probe is inserted horizontally (i.e.parallel to the magnetic filter field) and vertically (i.e.perpendicular to B filter ). Figure A1 shows the evaluated electron densities at the crossing point (x, y, z) = (2.5, 0, 2.5) cm at an RF power of 60 kW and a pressure of 0.3 Pa for three different values of the magnetic filter field corresponding to a PG current of 0, 1.5 and 3 kA.Because of the reduced electron mobility perpendicular to the magnetic field, it is found that considerably less electrons reach the horizontal probe when the filter field is present (compare black circles versus blue squares).
To correct this influence in the electron densities obtained with the horizontal probe, the ion branch is considered.As also noted in section 2, the ions are not fully magnetized and the ion gyro radius is around 1 cm, thus considerably larger than the probe radius, which is 0.15 mm.Therefore it is valid to assume that the evaluation of the ion branch will not be subject to magnetic field effects.As is shown in figure A1, this is the case, since the plasma densities evaluated from the ion branch do not differ, regardless of how the probe is oriented with respect to the magnetic filter field (see red diamonds and gray crosses).Furthermore it is found, that between the plasma densities evaluated from the ion branch using OML theory and the ones obtained from the electron branch in the vertical probe, there is a factor of two, which does not depend on the magnetic field strength.Note that also other authors found a systematic overestimation of the plasma density, when OML is used for evaluation [23], the factor depending on the probe specifics.The conclusion from this is that the vertical probe evaluation is valid also when the magnetic filter field is present, whereas the one from the horizontal probe must be corrected according to the strength of the magnetic field at the probe tip location, i.e. to make the black circles coincide the blue squares.The density correction scheme as explained above is evaluated not only in front of the plasma grid center but at all other available crossing points in front of the PG, which are five in total and are located at x = 0, z = 2.5 cm and y ∈ {−13, −6, 0, +6, +13} cm.Since the variation of the magnetic filter field strength in y direction is negligible, the average value of the density correction factor of these five crossing points is used.This results in 0.97 (standard deviation of 0.07) when no magnetic field is present.This shows that the results from the horizontal and vertical probe are within the error bars.For I PG = 1.5 kA (corresponding to a filter field strength of roughly 3 mT at the measurement position (cf figure 1)) the obtained correction value is 1.28 (standard deviation of 0.16) and for I PG = 3 kA (corresponding to roughly 6 mT at the measurement position) the correction value is 1.93 (standard deviation of 0.79).The given density correction factors are multiplied with all plasma densities evaluated from the electron branch of the horizontal probes, when a magnetic filter is used.
Note that it was checked that the plasma potential and electron temperature, as evaluated by the second derivative of the electron current, are almost not affected by the orientation of the probe with respect to the magnetic field, wherefore no correction is applied to the plasma potential and electron temperature values obtained with the horizontal probe.

Figure 1 .
Figure 1.Sketch of the experimental setup of the ion source at the BUG test bed with the combined Mach-Langmuir probe system and the axial profile of the magnetic filter field created by a vertical current through the plasma grid.

Figure 2 .
Figure 2. RF power variation at fixed p fill = 0.3 Pa and magnetic filter field generated by I PG = 1.5 kA.Probe is positioned at the PG center, as indicated in figure 1. Plotted along in red as straight dashed lines are the corresponding affine linear fits.

Figure 3 .
Figure 3. Filling pressure variation at fixed P RF = 60 kW and magnetic filter field generated by I PG = 1.5 kA.Probe is positioned at the PG center, as indicated in figure 1.

Figure 4 .
Figure 4. In-plane components of the positive ion flux for a vertical cut through the center of the BUG ion source at x = 0.The dashed border indicates the position of the plasma grid.Values are for fixed P RF = 60 kW and p fill = 0.3 Pa and for the three different magnetic filter field strengths according to a PG current of 0, 1.5 and 3 kA, respectively.Arrow lengths are proportional to positive ion fluxes, the horizontal arrow at the bottom right indicating the scale.

Figure 5 .
Figure 5. Ion flux parallel to the PG at an axial distance z = 2.5 cm from the PG.The dashed line indicates the projection of the plasma grid.Shown are the fluxes for three different values of the filter field 0, 1.5 and 3 kA.RF power and filling pressure are fixed at 60 kW and 0.3 Pa, respectively.Measurement positions are indicated by crosses.The probe is inserted from the right.

Figure 6 .
Figure 6.Magnitude of the ion flux component perpendicular to the PG for different values of the magnetic filter field (left: no filter field; center: I PG = 1.5 kA; right: I PG = 3 kA).Upper row: cut through expansion at z = 19 cm.Lower row: cut in front of PG at z = 2.5 cm.RF power and filling pressure are fixed at 60 kW and 0.3 Pa, respectively.

Figure 7 .
Figure 7. Vertical position sweep close to the PG surface at x = z = 2.5 cm.The used power and pressure are 60 kW and 0.3 Pa, respectively.Plot (a) shows the ion flux towards PG, (b) the plasma density and (c) the directed velocity component, which points towards the PG.The three dashed vertical lines indicate the positions of the lower and upper PG half.

Figure 8 .
Figure 8. Magnetic filter field variation of the plasma density in the expansion at z = 19 cm (upper row) and in front of the PG at z = 2.5 cm (lower row).RF power is 60 kW and pressure is 0.3 Pa.The red diamonds indicate the position of the uncompensated pin probes used in former measurement campaigns.

Figure A1 .
Figure A1.Evaluated plasma densities from horizontally and vertically injected probe at the crossing point (x, y, z) = (2.5, 0, 2.5) cm for different values of the magnetic filter field strength.Used RF power is 60 kW and filling pressure is 0.3 Pa.
This work has been carried out within the framework of the EUROfusion Consortium, funded by the European Union via the Euratom Research and Training Programme (Grant Agreement No. 101052200 -EUROfusion).Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Commission.Neither the European Union nor the European Commission can be held responsible for them.