For intense proton beam production with compact ion sources: the ALISES ion source family developed at CEA Saclay

The production of intense proton beams in CEA Saclay started in the 90’s with the development of the SILHI source to inject the IPHI accelerator. This ECR ion source is still in operation nowadays. ECR plasma is a well-known heating process and plasma chamber internal dimensions were investigated in order to reduce size and maintenance time. In 2012 R&D investigations on those simple plasma chamber parameters were started with the design of the first ALISES ion source that will give birth to a new source family still in progress today. This paper will summarize the different steps of this development.


Introduction
At CEA/Saclay, in the middle of the 90's the SILHI source developments [1] [2] were initiated and more than 100 mA proton or deuteron beams at energy of 100 keV are now routinely produced in pulsed or in CW (continuous wave) mode.SILHI source is part of the IPHI accelerator [3] and the CEA/Saclay demonstrate its "savoir faire" for other international projects like IFMIF [4] or FAIR [5] where intense beams of light ions were required.Another national project was also based on that source but for lower beam intensity: the Spiral2 deuteron ion source [6].This paper will summarize all the work that has been carried out at CEA Saclay to produced intense light ion beam in the last 13 years and more specifically the constitution of a family of ion sources named ALISES for Advanced Light Ion Source Extraction System.

SILHI ion source
The SILHI ion source is based on the ECR heating process, with magnetic field configuration produced by two independent coils.Magnetic orientation follows the source axis and the magnetic iso-surface of 87.5 mT (called resonant zone) is located inside plasma chamber.Plasma chamber dimensions are 100 mm long and Ø=90 mm, diameter which fit with the diagonal section of the internal dimensions of a WR284 waveguide.The microwave injection is realized through a 3-steps ridged waveguide which concentrated the power onto the source axis.The energy for plasma heating is delivered by a 2.45 GHz industrial magnetron of 2 kW by SAIREM company, driven by an analogic pulse generator.Both ends of plasma chamber are covered with boron nitride plates, a high secondary electron yield material in order to improve plasma electron density.With time boron nitride is sputtered all over plasma chamber internal surfaces which also helps increasing plasma density.Extraction column is composed of 5 electrodes and is water-cooled.The inter-electrode gaps are optimized for the lowest extraction RMS emittance: calculations were run with Axcel-INP® code with 100 mA extracted proton current at 95 keV.

Criteria
The aim of an ion source is to produce a high quality and stable ion beams.Several criteria are implicit, but others are also mandatory for accelerators applications, such as:  Intensity and energy requested by the project  Reproducibility of measurements in time  Reliable exploitation with a Mean Time Between Failure (MTBF) as high as possible  Easy maintenance or a Mean Time To Repair (MTTR) as low as possible As the ion source will be used in the context of accelerators, the LEBT (Low Energy Beam Transportline) conducts particles to inject particles into the RFQ (radio frequency quadrupole), the second stage of acceleration.As RFQ has its own injection acceptance, the beam size and angle must be controlled and adjusted to fit to this acceptance with the help of a pair of solenoids installed in the LEBT.
Intense beams have a specific way to propagate in the LEBT.Beam particles interact with the residual gas and lead to the emission of electrons.Those electrons are trapped by the beam transverse potential and they reduce its value: this is the Space Charge Compensation (SCC) process.Everything that modifies this compensation leads to an emittance growth:  Magnetic field in solenoids of the LEBT and fringe field,  Aberrations of electric field at source exit in the extraction column,  Strong beam focalization while transport in LEBT To preserve as much as possible the initial value of emittance at source exit, solenoid length must be reduced while increasing magnetic flux.LEBT length must be shortened as much as possible to minimize beam interaction with the residual gas.And the beam transport must be smooth to avoid crossover while transport.

ALISES v1
The ALISES source was built to solve the emittance growth problematic by reducing the accelerating column length by 20 cm, a 10% size shrink of usual LEBT (Figure 1).To achieve that, the electrical insulating structure (spare part of the SILHI source) were used and they were positioned before the extraction aperture.That forced us to remove the SILHI type coils due to a lack of space and to design a new magnetic coil [7] that was positioned after the extraction hole at ground potential.This new configuration source and was patented in 2010 at INP [8].
The ALISES v1 ion source was not meant to be a source for accelerators but its second interest was to test new developments around the plasma chamber size.The ion source has also a 3-steps ridged waveguide for microwave injection which is fixed on a piston on high voltage side.This piston can be moved from outside, changing the plasma chamber length from 20 to 100 mm without changing the operating pressure inside the source.We also can insert inside plasma chamber some tube to reduce the plasma chamber internal diameter.

Discharges
While commissioning the source on BETSI (Banc d'Etude et de Tests des Sources d'Ions) test bench in Saclay in 2012, we started to produce the plasma without extraction voltage: the pink plasma rapidly was observed, color that is typical of the hydrogen Balmer's lines.ECR heating was effective with a flux of several sccm (standart centimeter cube per minute) of H2 gas and 500 W of microwave power.When the high voltage and the magnetic field was set simultaneously on, the HV power supply voltage dropped down with a maximum current flow.This behaviour was systematic [9] even without a spark.We suggested that it could be a Penning discharge occurring inside the accelerating column.Several simulations with OPERA code [10] showed that electrons were trapped at several locations (Figure 2).Experimental measurements using Kapton foils wrapped around extraction electrodes were burnt while turning on the ion source.The location of those burn marks on the Kapton foils validates the presence of those discharges and their localization: • far axis: in order to annihilate those discharges, 5 ceramics parts were installed inside empty spaces under vacuum to stop electron motions inside the insulating structure.• around extraction electrodes edges: three tubes made of glass (insulating material) where positioned around the electrodes to fill the gaps where the Penning discharge process occurs.With both protections, ALISES ion source can be turned on, plasma is ignited and positive particles extracted: 20 mA at 23 kV with a 6 mm diameter aperture hole.

Plasma chamber length
If we imagine a "infinite long" plasma chamber, without any calculations, we can assume that all ions produced at the resonance zone where hot electrons are produced by ECR heating, will get neutralized before reaching the extraction hole of the plasma chamber.On the other hand, if the plasma chamber is too short: the number of produced particles at the resonant zone will also tend to zero due to few gas molecules in this tiny volume.For both cases (infinite and zero length plasma chamber) respectively no or few particles are extracted from the ion source.We can assume that there must be an optimum for the plasma chamber length to produce maximal beam intensity out of the ion source.
Plasma chamber length variation was the first measurements to be done, because this length can be adjusted manually from outside with the piston on the high voltage side.For each length, we tuned the ion source to get the maximal extracted intensity for the same extraction voltage.Figure 3 show a quasilinear behaviour of the extracted current versus the plasma chamber length due to increasing plasma chamber volume.Maximum is reached for 95 mm.

Plasma chamber diameter
The cut-off frequency (fc) of an empty cavity is related to the cavity radius and the selected excited mode.For the same mode but with a frequency below fc, that mode cannot exist the cavity.In our case of our circular plasma chamber, the dominant mode is the TE11 mode.For this mode, with a 90 mm diameter the cut off frequency is 1.9522 GHz.
In our first attempt to reduce plasma chamber diameter, we use a 45 mm and then a 30 mm diameter tube fixed inside plasma chamber onto the boron nitride disc.With both reductions, the cut off frequency is well above the 2.45 GHz magnetron frequency.Nevertheless at low extraction potential of 17 kV, plasma was still visible and the current of extracted positive particles are very similar (Figure 4) to the value measured for the nominal 90 mm internal diameter.

ALISES v2
With all the experienced obtained with the ALISES v1 ion source, a new design effort was undertaken to build a second ion source that will be much more compact and will produce high intensity light ion beams.ALISES v1 demonstrate the viability of a shorter accelerating column.For this new source, the microwave injection ridged waveguide and the plasma chamber dimensions were kept similar to the SILHI ion source but the insulating structure was made in a single ceramic structure of 200 mm [11] The source body is in direct contact with the ceramics to avoid Penning discharges.This was also motivated by the fact that Viton gasket number are greatly reduced from 8 to 2 and in the new source those gasket are not involved in the electrode positioning.This reduces alignment procedure difficulties and reduces MTTR which remain one of the key point for an ion source in an accelerator infrastructure.This ion source is completely described in [12].ALISES v2 has been installed on BETSI test bench in March 2015 [13] and produced at once a 18 mA proton beam with a =6 mm extraction hole diameter at 35 kV due high voltage power supply limitation.During optimization, the extracted beam intensity surprisingly increased from 18 to 23 mA with coil current respectively from 60 to 180 A (Figure 5).The resonance zone of 87.5 mT is usually positioned at the end of the 3 steps ridged waveguide in the plasma chamber, this configuration is achieved with 181 A in the ALISES coil (Figure 6 filled diamond).Thus for values of Icoil below 180 A, there is no resonant zone on source axis for ECR heating and for the lowest values of Icoils there is no resonant zone inside the plasma chamber at all (Figure 7

left).
After optimization at 50 kV, the source produced at end of March, a proton beam of 35 mA in CW mode with gas injection of 1.65 sccm, a microwave power of 980 W and 108 A in the ALISES source coil (Figure 6 empty diamond).We tried to understand this production mode by adding a second coil (SILHI type) installed around the ceramic structure.With the 2 coils, we were able to test several different magnetic profiles.Investigations were mainly done in pulsed mode and sometimes the microwave power was switched to CW mode.On Figure 6 several magnetic profiles with combined value of ALISES and SILHI coils are plotted with measured extracted intensities in the legend.The highest intensity extracted was reached with the single SILHI coil at 104 A (Figure 6 full square).In this configuration, the resonant zone is also located outside the plasma chamber (Figure 7 right).The pulsed beam intensity reaches 40.7 mA with 700 W microwave power and 1.95 sccm gas injection and intensity rise up to 48.5 mA with 2.1 sscm gas injection for the same microwave power in CW mode.In this configuration, the beam-dump at the end of BETSI LEBT collected only 26 mA.At 40 kV, a copper insert was installed inside the plasma chamber to reduce its internal diameter.Measurements showed that extracted current increase from 36.7 to 40.5 mA with chamber reduction respectively from 90 to 45 mm, with almost the same coil value.This result confirm the plasma chamber reduction for ALISES v1 and v2 ion sources.
After that ALISES v2 source was used for irradiation purposes: scintillator irradiation for a 4D emittance meter [14], for S3 project for Spiral 2 in GANIL at Caen and also for FAIR Project for debugging Control and Command of an Alisson scanner diagnostic.This first emittance measurement confirmed that proton ratio was lower than expected: intensity lines of H2 + and H3 + were much too intense than expected.Also the presence of heavier pollutants in the emittance figure (Figure 8) partially explains the difference between the 26 mA collected on the LEBT beam dump when 48 mA was recorded on the source high voltage power supply.The ALISES v2 source did not completely reveal all its potential and we still do not understand the heating mechanism, as the ECR resonant zone stands outside the plasma chamber.Additional work is necessary to understand clearly the mechanism of plasma ignition in these conditions.

ALISES v3
ALISES v3 source was designed in 2018 and built while the BETSI test bench was upgraded to 100 kV of extraction voltage.This source has a plasma chamber length and internal diameter of respectively 100 and 45 mm, the insulating structure is made in ceramic of 175 mm external diameter.The source body inserted inside this isolator is a brased copper assembly, which includes the 90° elbow waveguide, the 3-steps ridged waveguide, the water cooled plasma chamber and the gas injection port.Intermediate electrode connector is located on the high voltage flange.The magnetic configuration is composed of a single coil supported directly by the insulator as on ALISES v2 ion source.The source is completely described in this conference [15].The most interesting aspect about this source is that there is no need for a high voltage platform to hold the ion source: it is plugged directly onto the LEBT.
In 2019 first beam is extracted at 50 kV on a specifically developed test bench called TROPICS, with only intensity and beam noise measurement analysis.ALISES v3 source shows good behaviour, with increasing extracted current versus microwave power and versus extraction voltage.
After BETSI test bench upgrade to 100 kV, the ion source was installed and electrode formation procedure started with a slow biased ramp.Long run campaign started in 2022 to analyse beam stability with time.It was necessary to find the right settings to run the source over several days uninterrupted.Many long run tests were made at 40 kV producing 21 mA of protons, lasting up to 85 hours.The longest run achieved at 37 kV exceed over 140 hours (Figure 9).After these experiments, an emittance measurement campaign started with the Allison scanner developed for ESS project [16].The source parameters were set up as follows: the gas, duty cycle and microwave power stayed constant, respectively at 1.3 sccm, 10 ms at 10 Hz and 600 W. Emittance values are plotted versus the intermediate electrode bias value for four levels of extraction voltage.The expected value of ε for a standard ion source dedicated for accelerators is below 0.2 π mm mrad.On Figure 10, this limit is reached for extraction voltage above 45 kV.Great care was taken for high voltage above 50 kV: better results are obtained for higher extraction voltage.Unfortunately, toward 95 kV bias voltage, an electric arc initiated a "carbonized" path inside the ceramic.After manually grinding this path, the source was limited to 50 kV because this path greatly compromises the dielectric properties of the ceramic.A second design of ceramic was made, the length between electrodes along ceramic surface under vacuum were increased.With this new design, higher extraction voltage were tested in early 2023: for 800 watt, in pulsed mode 41 mA at 70 kV were measured.In CW mode, 50mA were extracted at 76kV.This work is still underway to reach SILHI's performances requirements at 100 kV.

Conclusions
Problems with intense beam operations, with shorter LEBT length size gave rise to the ALISES v1 source and R&D program in CEA Saclay.The ALISES v1 source was clearly a first attempt.The second version of ALISES demonstrated the feasibility of reducing plasma chamber internal dimensions while testing a new insulating structure configuration.ALISES v2 simplified assembly allows easier mounting, better alignment for shorter maintenance procedures.Besides that, the heating process without a resonant zone inside plasma chamber is still not understood.The ALISES v3 source shows excellent behaviour and continuous-operation qualification tests are underway.These tests for its qualification will continue to demonstrate the production of high intense proton beams at 100 kV.

Figure 1
Figure 1 Comparison between SILHI and ALISES ion sources.

Figure 2
Figure 2 Simulations of Penning discharges inside the ALISES v1 source.

Figure 3
Figure 3 Extracted intensity versus plasma chamber length.

Figure 4
Figure 4 Extracted intensity and cut-off frequency versus plasma chamber internal diameter.

Figure 5
Figure 5 Extracted beam Intensity versus current intensity in ALISES coil.

Figure 6
Figure 6 Bz field simulated with Opera code for several magnetic configurations.

Figure 7
Figure 7 Both magnetic setups, the resonant zone of 87.5 mT stands clearly outside the plasma chamber.

Figure 8
Figure 8 Emittance measurement at the end of the LEBT.

Figure 9
Figure 9 Intensity on beam dump during long run test at 37 kV.

Figure 10
Figure 10 Emittance measurement versus puller electrode bias voltage at 600 watt microwave power and for different extraction voltage.