Determining absolute VUV fluxes for assessing the relevance of photon-surface interaction in ion sources

A portable device was developed to quantify VUV fluxes flexibly at ion source setups. It consists of a VUV sensitive photodiode and optical filters for wavelength selection and is calibrated against a VUV spectrometer down to 46 nm for a variety of discharge gases, including Ar, N2, O2 and H2. It was applied to the negative hydrogen ion source at BATMAN Upgrade to quantify the VUV radiation emitted by the driver as well as in front of the extraction surface (plasma grid, PG). The combined VUV fluxes impinging on the PG with photon energies larger than 6.6 eV has a comparable magnitude as the ion flux. It could be shown that the recently confirmed influence of the ion source plasma on the surface work function of the PG can at least partly be ascribed to the VUV radiation from the driver and that photo-emitted electrons from the PG should not play a role in the sheath physics.


VUV emission in ion sources
Vacuum-Ultraviolet radiation (VUV, ≤ 200 nm, ≥ 6 eV) is predominantly present in ion source plasmas due to resonant transitions of the plasma constituents, including high ionic states.This radiation gives insight into basic plasma physics of the system at hand, but furthermore leads to high energetic photon fluxes towards components within or around the plasma.These may affect surface properties and/or lead to photo-emission of electrons depending on the photon energy and the absolute flux magnitude.
VUV spectroscopic systems are the means of choice to investigate this radiation, but are large and heavy devices and need a direct vacuum connection and differential pumping.Hence, portable systems have been developed, where devices based on a photodiode were already used for application to ion sources [1], with or without optical filters for wavelength selection.However, for any VUV system, absolute calibration is not straightforward and requires synchrotron radiation or a complex procedure via a combination of several secondary radiation standards.
In this contribution, a VUV diode system with a combination of edge and interference filters [2] is presented.Compared to the solution presented in [1], the present device is absolutely calibrated against a VUV spectrometer that itself is absolutely calibrated down to 46 nm [3].The diode system is hence capable to quantify VUV fluxes up to 27 eV and was applied to the ion source at the BATMAN Upgrade facility.

Setup of the VUV diode system
The diode system for quantification of VUV emission is depicted in figure 1 (a).It consists of a silicon photo diode that is sensitive between 1 and 1000 nm, and a manually driven filter wheel that can store up to five optical filters.An aperture stop of 4 mm diameter limits the collection solid angle and the entire system can be flanged to any vacuum system by the KF40 flange.The small volume of the system makes a separate pumping unnecessary.
Five interference and five edge filters are available with which the prominent emission systems of a variety of discharge gases can be selected.For details it is referred to [2], where the filter selection is described specifically for Ar, H 2 , N 2 , O 2 and mixtures thereof.Figure 1 (b) shows a spectrum in the VUV spectral range of a hydrogen discharge obtained at a planar ICP setup (15 cm diameter, 10 cm height, 2 MHz, 2 kW max.), where both the VUV spectrometer and the VUV diode system are attached.The transmission curves of several filters are given and the respective transmission ranges are shaded.A bandpass filter at 230 nm is used to detect the hydrogen continuum (H 2 transition a-b), a bandpass at 154 nm selects the Lyman band (H 2 B-X), and the interference filter at 122 nm dominantly transmits the Lyman-alpha line (H n=2-n=1).Subtracting a measurement without any filter by a measurement with a MgF 2 window as an edge filter, the Werner band (H 2 C-X) together with the remaining Lyman series (H n≥3-n=1) below 113 nm is detected.
Accordingly, absolute calibration of the detected emission is performed by relating it to the absolutely calibrated spectrum from the VUV spectrometer integrated within the transmission range of the respective filter.Like this, calibration values have been obtained for the commonly observed emission systems of the above mentioned gases and a thorough benchmarking was conducted [2].An accuracy of better than 25 % compared to the VUV spectrometer is attested and a dynamic range of more than four orders of magnitude is accessible for each filter.Energyresolved absolute VUV flux measurements up to photon energies of 27 eV are thus made possible with a portable and easily attachable device.

VUV flux quantification at the BUG ion source
The system was applied at the ion source for negative hydrogen ions at the BATMAN Upgrade (BUG) test stand [4], see figure 2 (a).A H 2 plasma is generated in the driver and expands towards a low-work function converter surface (plasma grid, PG) at which negative hydrogen ions are produced by surface conversion, and subsequently extracted.In order to enhance the survival probability of the H − ions in the sheath plasma, a transverse magnetic filter field is present in the expansion region of the ion source to effectively decrease electron temperature and density close to the PG.The filter field is produced by an electrical current I PG of several The conversion efficiency of H − steeply depends on the work function of the converter surface that is covered with Cs to lower the work function down to values below 2 eV [5].The caesiated surface is affected by plasma-surface interaction due to impinging fluxes of hydrogen radicals, ions as well as energetic photons.Heiler [6] has shown that each species, in particular also VUV photons, contributes to a variation of the surface work function upon plasma impact.Furthermore, VUV photons can induce the emission of photoelectrons, especially for low-work function surfaces [7], and hence affect the plasma in the sheath region.
To quantify the photon flux onto the converter surface at the BUG ion source, the versatile VUV diode system was used.It was attached at two different positions: one having a line-of-sight (LOS) through the driver plasma and another observing the plasma close to the PG. Figure 2 (a) illustrates the approximate detection cones for the diode system at the respective LOS.
Figure 2 (b) shows the time trace of an exemplary measurement raw signal.The diode system was installed at the driver LOS and the filter for the Lyman-alpha line was installed.The emission can be followed nicely with little to no noise and a high temporal resolution.The integration time of the data acquisition system (PREMA DMM 5017) was set to 100 ms.After plasma ignition and stabilization of pressure and RF power a stable emission plateau is reached over which the signal is averaged for the succeeding data evaluation.

VUV emissivities
In order to obtain absolute values of emissivity, the calibration values need to be transferred from the planar ICP setup where calibration was performed to the setup under investigation.Here, the volumes within the plasma, from which radiation is collected, need to be taken into account.These so-called effective volumes are typically calculated by solid angle integration over the entire collection cone.For simple geometries like the planar ICP or the driver at BUG, this can be done analytically.For complex geometries, as for instance at the LOS close to the PG at BUG, ray tracing codes may be applied.In the present case, the code described in [8] was used.While for the driver the ratio of the effective volumes is close to unity, the LOS at the PG collects three times more light than during calibration.The resulting emissivities are shown in figure 3. VUV emission of more than 11 eV is detected and is most intense for both LOS, where the driver plasma emits about two orders of magnitude higher VUV intensities for any spectral range.This difference can be roughly reproduced by collisional-radiative modeling using YACORA [10] and hence arises mainly from the local plasma parameters: T e = 9.8 eV, n e = 5×10 17 m −3 in the driver, and 2.3 eV, 6.6×10 16 m −3 at the PG, determined by Langmuir probe and OES.

VUV fluxes onto the plasma grid
Conversion of emissivities to fluxes impinging on the PG needs to account for the geometry of the emission volume.The measured emissivity is assumed homogeneous and isotropic within the plasma volume, i.e. multiplying the emissivity by the emission volume V and dividing it by the volume surrounding surface area A, gives the flux leaving the emission volume.For the LOS close to the PG, the PG surface is one of the boundaries of the emission volume and thus, using the A/V ratio directly yields the impinging flux.For the driver, the mentioned procedure only yields the flux leaving the driver, which afterwards needs to be scaled towards the PG due to beam widening.This is again done by solid angle calculation and approximately yields a factor of 10 reduction of the flux.
Figure 4 (a) shows the obtained fluxes impinging on the PG surface.The VUV bars consist of the contributions from the driver (dashed part) as well as from the plasma close to the PG (empty part).The two orders of magnitude higher emissivity in the driver together with the factor 10 reduction due to beam widening still yields roughly a factor of 10 larger contribution to the total flux than coming from the plasma close to the PG.This means, the driver plasma dominates the VUV influx onto the PG and using the photon energy of 10.2 eV for the Lyman-alpha line exemplarily, the power density of this single line is calculated to about 75 W/m 2 impinging onto the PG.Furthermore, a comparison to the impinging fluxes of hydrogen ions and atoms is shown, calculated from the plasma parameters obtained by Langmuir probe measurements and OES close to the PG [9].The atomic flux is thus dominating and can influence the surface chemistry due to its property of being a radical.Energy influx towards the PG is, however, only given by ion and VUV fluxes and here, summing up all the photon fluxes (dashed line), comparable magnitudes are achieved.
Figure 4 (b) shows measurements performed without the magnetic filter field (I PG =0) used to cool down the plasma in the expansion region.While most of the statements given above still hold, the increased electron temperature close to the PG (5.2 eV instead of 2.3 eV) now yields VUV fluxes that are comparable to the ones coming from the driver.

Conclusions and Implications
Using a detection system comprising a VUV sensitive photodiode and optical filters, high-energy photon fluxes can be quantified with a coarse wavelength resolution.The system is calibrated against a VUV spectrometer for energetic fluxes up to 27 eV for a variety of discharge gases.It was applied to the ion source for negative hydrogen ions at the BATMAN Upgrade facility.
Total VUV fluxes of the order of some 10 20 m −2 s −1 were measured, impinging onto the caesiated plasma grid with energies up to above 11 eV.Heiler [6] has shown, that the work function of caesiated surfaces is affected by VUV fluxes as low as 10 18 m −2 s −1 within hundreds of seconds.Assuming the actual influence on the work function is driven by the time-integrated flux, i.e. the fluence, the PG surface work function at the BUG ion source should be affected by VUV radiation from the driver within seconds.This means, that VUV radiation is at least partly responsible for the known influence of the ion source plasma on the PG work function (see [11] for instance for recent measurements showing beneficial as well as adverse impact).On the other hand, Wünderlich et al. [12] have estimated that even VUV fluxes higher than 10 22 m −2 s −1 do not produce sufficient photo-electrons to affect the plasma sheath in front of the extraction system.The current measurements hence confirm, that those can be neglected for the sheath physics.
In conclusion, the developed VUV diode system was applied to deduce that VUV fluxes from the driver will certainly affect the surface work function of caesiated ion sources for NNBI, but will probably not alter the sheath physics by photo-induced emission of electrons.For ion sources for accelerators, the situation might be different: Apart from a varied spectral composition due to different feed gases and possibly highly-charged ions, the typically much higher power density coupled into the plasma volume and the smaller dimensions may lead to expectedly higher VUV fluxes exposing the ion source walls.

Figure 1 .
Figure 1.(a) Schematic of the portable diode system for quantification of VUV emission.(b) Measurement and calibration principle of the device, exemplarily shown at a H 2 discharge: Radiation transmitted through the filter interval is correlated to specific emission features.For calibration, the measured absolute spectrum from the VUV spectrometer is integrated over the respective transmission ranges and related to the measured value from the VUV diode.

Figure 2 .
Figure 2. (a) The ion source for negative hydrogen ions at the BATMAN Upgrade test stand with illustrated detection cones for the VUV diode system at the two lines-of-sight.(b) Signal from the diode system with the filter for the Lyman-alpha line installed.The stable plateau for data evaluation is highlighted.

Figure 3 .
Figure 3. VUV emissivities measured at BUG for the two LOS through the driver and close to the PG.Reproduced from [9].

Figure 4 .
Figure 4. (a) Particle and photon fluxes towards the PG at the BUG ion source.For the VUV fluxes the respective contribution from the driver plasma is highlighted dashed, the sum of all photon fluxes is given as dashed line, and for ionic and atomic fluxes the underlying plasma parameters are given on top of the bars.Reproduced from [9].(b) Same as in (a), but without the magnetic filter field in the expansion region of the ion source.