High current H + 2 beams from a filament-driven multicusp ion source
Daniel Winklehner, Janet M. Conrad, Joseph Smolsky, Loyd Waites
HHigh intensity H +2 beams from a filament-driven multicusp ion source Daniel Winklehner, ∗ Janet Conrad, Joseph Smolsky, and Loyd Waites
Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, USA (Dated: August 28, 2020)For the proposed IsoDAR experiment in neutrino physics, a dedicated H +2 ion source (MIST-1) wasdesigned and built at MIT. The MIST-1 ion source is a filament-driven multicusp ion source, optimizedfor the production of H +2 over protons and H +3 . In this paper, we report the commissioning resultsof MIST-1 and first systematic measurements of beam current and beam composition as functions ofgas load, discharge voltage, and discharge current. The commissioning setup includes a Faraday cupdirectly after the source (for total beam current measurements), a mass separator consisting of a dipolemagnet, slits and another Faraday cup (for beam composition measurements), as well as a set of Allisonemittance scanners (for beam quality measurements). Highlights of the results are total beam currentdensities as high as 40 mA/cm and a H +2 ion species fraction of up to 90 % in DC mode (non-pulsed).The measured emittances are well-reproduced in simulations and are low, as expected for this typeof ion source. Thus, MIST-1 is well suited to inject beam into an RFQ for bunching and subsequentacceleration in a compact cyclotron. I. INTRODUCTION
Neutrino flux measurements have a history of produc-ing unexpected results, starting with the Homestake ex-periment [1]. More recently, LSND and MiniBooNE bothobserved an excess of events in ν µ → ν e appearance ex-periments [2, 3]. These results can be explained by asterile neutrino with a ∆m around 1 eV [4]. The Iso-DAR experiment is designed to be a definitive test of eVscale sterile neutrinos and is described in detail in otherpublications [5–7]. The novelty of IsoDAR is in con-structing an intense anti-neutrino source near a kilotonscale neutrino detector. To reduce backgrounds, neutrinooscillation experiments are ideally done underground. Acartoon of IsoDAR near the KamLAND detector is shownin FIG. 1.The anti-neutrino production is dependent on a 10 mAcontinuous wave (cw) beam of 60 MeV protons. To saveon space and costs, IsoDAR will use a compact cyclotronas a driver instead of a linear accelerator. Commerciallyavailable cyclotrons in this energy range typically haveintensities around 1 mA. Space charge is the primary ef-fect limiting higher currents. In order to reduce spacecharge effects, IsoDAR will accelerate H +2 instead of pro-tons or H − . The presented ion source was developed atMIT to produce high currents of H +2 .Typically, proton sources are designed such that H +2 and H +3 contamination is low. Amongst the multitudeof proton sources that routinely produce tens to hun-dreds of mA of beam, two types lend themselves eas-iest to re-design or de-tuning for improved H +2 cur-rent: The 2.45 GHz, Electron Cyclotron Resonance (ECR)ion source [8], and the multicusp ion source - radio-frequency (RF)-driven [9], or filament-driven [10]. Inseveral measurement periods, the 2.45 GHz ECR versa-tile ion source (VIS) [11, 12] was tested by the MIT ∗ [email protected] FIG. 1. Artists rendition of the IsoDAR experiment paired withthe KamLAND detector at Kamioka. From left to right: Thecyclotron (ion source on top), generating a 60 MeV/amu H +2 beam, the medium energy beam transport line, the neutrinoproduction target [16, 17], and the KamLAND detector[18]. group in collaboration with INFN-LNS and Best Cy-clotron Systems, Inc. While the performance for protonswas excellent, the finding was that the maximum H +2 fraction was limited to 50 % [13, 14]. The high protoncontamination was deemed undesirable for a compactsystem like IsoDAR. On the other hand, Ehlers and Leungat LBNL demonstrated that a filament-driven, multicuspion source is capable of producing extractable total cur-rent densities of 50 mA/cm with up to 80 % of the beambeing H +2 ions [15]. Multicusp Ion Source Technologyat MIT - v1 (MIST-1) is such an ion source. Where theresults of the LBNL source were for sub-mm diameterextraction apertures and pulsed beams, MIST-1 was de-signed to deliver 15 mA total beam current in DC mode,requiring design changes and cooling upgrades.Here we present the preliminary commissioning re-sults of MIST-1. Section II contains a detailed descrip-tion of the source and diagnostic setup. Section III cov-ers simulations of extraction system and test beam line.Measurements and data are in Section IV . a r X i v : . [ phy s i c s . acc - ph ] A ug FIG. 2. Cross section view of MIST-1 ion source. Inner partsare labeled: 1. Faraday cup, 2. Extraction System, 3. Perma-nent magnets (Sm Co ), 4. Filament feedthroughs, 4. Watercooling fittings. The backplate also has a gas inlet, throughwhich the hydrogen gas may enter into the source. II. EXPERIMENTAL SETUPA. Ion Source .The MIST-1 ion source is a filament-driven, multi-cusp ion source and is described in detail in previouspublications [6, 7, 19]. A stainless-steel chamber withsamarium-cobalt permanent magnets creates a multicuspfield that confines the plasma around the extraction hole.Hydrogen enters the source through the backplate, andis ionized by a tungsten filament. The filament is a tung-sten alloy, mixed with copper and nickel for corrosionresistance. The ions are then extracted through a holein the center of the front-plate. The source is on a highvoltage platform can be raised up to 20 kV. Ions leavingthe source are focused and accelerated in the extractionsystem.Hydrogen travels via gas line to the hydrogen inlet inthe back of the source. The pressure inside the sourceis controlled by a mass flow controller (MFC), (MKS In-struments, Model: GV50A, with 5 sccm full range) whichmoderates the amount of hydrogen which may enter thesource from the hydrogen bottle. The MFC is controlledvia a RS485/USB with optical interface connected to ourcontrol system PC.Inside the source,the filament is heated via power sup-plies on the high voltage platform ( FIG. 3). The sourcebody, back plate, and front plate are all electrically insu-lated. This allows each piece to be at a different poten-tial, and therefore test the effects of different longitudi-nal fields. In this study, all are at the same potential. Fu-ture studies at which they are at different potentials areforthcoming. The filament emits electrons, which ionizethe gas and forms a plasma. The geometry of the source
FIG. 3. The wiring schematic of the ion source. Red is the highvoltage reference potential, blue is the data cables, and blackare power cables. The power supplies are computer-controlledvia an optical USB extender cable. The source back plate, body,and front plate can all be held at different potentials (they areseparated by insulator rings). During the measurements pre-sented here, they were held at the same potential. causes H to be the dominant ion species over other hy-drogen ion species [6, 15, 19].In this study, we have used a spiral shaped, 0.8128 mmthick filament which was 40 mm from the back plate,see FIG. 2. We have experimented with several differentfilament shapes, thicknesses, and positions. This is anongoing study, and will be discussed in more detail in anupcoming paper.The ions drift out of the source through the extractionhole. Once outside they are transported by a low energyextraction system. The low energy extraction system isa series of copper electrodes that shape the beam whenleaving the ion source. The different electrodes are heldat different potentials, changing the energy of the beamas it goes through the extraction system, and the beamshape and quality. These voltages are controlled by sev-eral high voltage power supplies. The electrodes follow-ing the source plate are the puller (typically kept at a low,negative voltage -2 kV) and the einzel lens. The einzellens is made up of a total of three electrodes, the outerones being grounded, and the larger central piece beingadjustable (typically held at 1 kV).The electrodes are aligned via the compression of sev-eral ceramic balls. This system is then mounted insidethe 6-way cross so that it is aligned to the end flangemounting the ion source. The extraction system is mod-eled using the IBSimu code [20]. This will be discussedin more detail in Section III. B. Low Energy Beam Transport and Diagnostics
A Faraday cup following the extraction system is usedto measure the total current coming from the source.This allows us to know the total current of the beam, butwithout separation of species. In order to differentiatethe beam into species fraction, the beam must continuedown the low energy beam line. To allow for this, whilenot compromising our previous measurement, this Fara-day cup can be retracted. The beam may then continuedown the beamline unblocked.The ion source extraction system is followed by a muchlonger low energy beam line used for beam diagnos-tics (see FIG. 4). Three electromagnets are used forbeam transport: 2 quadrupoles and 1 dipole. The ar-rangement of the magnets, beamline, and diagnostics 6-way are shown in FIG. 4 (inventor full assembly model).Both quadrupoles can be used for vertical focusing of thebeam, however only the first quadrupole is being usedfor the initial studies. When a mass spectrum is beingtaken, the quadrupole is increased in a constant ratiowith the dipole current. The dipole is used for horizon-tal focusing and ion species separation. Following thesecond quadrupole magnet is the analysis 6-way cross,which contains a second Faraday cup and two Allisonscanners with perpendicular axes.Each Faraday cup is equipped with a negative elec-trode in front to suppress electrons from influencing themeasurement. For each test, we ensure that the voltageon the Faraday cup suppressor is sufficient by measuringthe Faraday cup current as a function of the suppressorvoltage. Once the voltage is sufficient, and the electronsare suppressed, the measurement will no longer changewith increased suppressor voltage. It is at this point thatwe run our diagnostic tests.Measurements are taken while varying: • HV platform potential, • Discharge voltage, • Filament heating current, • H gas flow into source, • Filament shape and position.The plasma density is determined in large part by thefilament discharge current. This can vary with voltagedue to the heating of the filament. This can be compen-sated for by varying the filament heating voltage in a PIDloop, helping to make the plasma stable. For each plasmasetting, the total current is measured in the first Faradaycup. To take an ion mass spectrum, the dipole currentis varied from 0 80 A while current measurements aretaken in the second Faraday cup. This process is con-trolled and mass spectra are recorded via an automatedLabVIEW program run on the PC. For selected settings,measurements are also taken with the Allison scanners.A model for the Allison scanners in Figure 5. Parametersfor the dipole are in TAB. II.
TABLE I. MIST-1 ion source parameters.
Parameter Value (nominal)
Plasma chamber length 6.5 cmPlasma chamber diameter 15 cmPermanent magnet material Sm Co Permanent magnet strength 1.05 T on surfaceFront plate magnets 12 bars (star shape)Radial magnets 12 barsBack plate magnets 4 rows of magnets, 6 bars totalFront plate cooling embedded steel tubeBack plate cooling embedded copper pipeChamber cooling water jacketWater flow (both) (1.5 l/min)Filament feedthrough cooling water cooledFilament material W mixed with Cu and NiFilament diameter ≈ . mmDischarge voltage max. 180 VDischarge current max. 24 AFilament heating voltage max. 8 VFilament heating current max. 100 A The Allison scanners have a water cooled plate with aslit on its front face. This slit allows beam to pass intothe scanner at a specific transverse position. Inside thescanner there are two parallel plates, which are held atvarying potentials, followed by a second slit. The volt-age between the plates can be tuned so that only par-ticles of a specific momentum pass through the secondslit, which are then measured by a Faraday cup. There-fore, the scanner can map a series of momenta, based onthe voltages, to a single transverse position. To do a fullscan, this process is then repeated at multiple differentpositions by moving the Allison scanner along a slice ofthe beam. This mapping can be used for to determinethe phase space and emittance of the beam.In order to find the species ratios, the ion source andextraction system are kept at a fixed set of parameters.This is maintained until a stable current is observed inthe first faraday cup. The first faraday cup is then re-tracted, allowing the beam to enter the low energy beam-line. The dipole magnet in the beamline is varied from0 to .4 Tesla. The first quadrupole is adjusted after eachdipole measurement in order to focus the beam and max-imize current in the second Faraday cup at the end of thebeamline. The current in this parameter space is thenmeasured. These are then plotted as a function of dipolefield (see sec IV). Based on the ratio of mass to charge of
TABLE II. Parameters for the Bruker dipole magnet donated bythe University of Huddersfield.
Parameter Value
Maximum field 0.7 TMaximum current 125 AMaximum voltage 47 VBending radius 300 mmPole gap 75 mm
FIG. 4. Model of the low energy beam line diagnostic system. Starting with multiple species coming from the ion source,the species are then focused in the beamline by quadrupole magnets, and separated by the dipole magnet. Their currents andemittances can then be measured by the Faraday cup and emittance scanners at the end. the different ions in the source, it is possible to identifythe different peaks in the spectra as different ions. Fromthese measurements, we can identify the makeup of thebeam coming from our source.
III. SIMULATIONS
Ions extracted from MIST-1 are simulated using twopackages: IBSimu and Warp [20, 21]. IBSimu is used ini-tially because of it’s capability to accurately model ionstraversing the plasma sheath. However, IBSimu wouldbe too computationally expensive for a simulation ofthe entire test beamline. Warp is able accurately model
FIG. 5. Model 6-way cross housing the y axis Allison scan-ner and Faraday cup. This is immediately following the secondquadrupole magnet on the left. The Allison scanner can be re-tracted so that beam current can be measured by the Faradaycup in order to quickly measure mass spectra. FIG. 6. IBSimu simulation of the low energy extraction system.This simulated with 80% H (yellow) and 20 % protons (red)to correspond to the ion source developed at LBNL [15]. Theelectrodes are in blue, and the equipotential lines are in green. spacecharge effects and beam transport, with less com-puting power than IBSimu. However, the plasma sheathmodeling in Warp is not as well-established. This led tothe decision to use both packages in series. The IBSimusimulations run from the ion source to midway throughthe first 6-way cross. The Warp package is used to sim-ulate ions from the first 6-way cross to the end of thebeamline. A. Ion Source Extraction (IBSimu)
The extraction system was designed and simulated us-ing IBSimu, a particle in cell code which uses iterativeprocesses to calculate the particle trajectories throughelectromagnetic fields. IBSimu has successfully beenused to design and simulate several extraction systemsand can be considered well-benchmarked against exper-iment [22–25]. IBSimu uses electrode geometries whichcan be imported from CAD files. The electrodes are setto static potentials. These are then used to calculate thefields in the system. The particle trajectories are simu-lated by tracing the particle through this field, then tak-ing into account space charge into the electric field. Thisprocess is repeated until the simulation converges. Theparticle trajectories together show a beam profile.Multiple species can be simulated simultaneously, ac-counting for the space charge of each. The format ofthis code creates an accurate plasma model, which is im-portant for understanding the beam behavior in areas ofhigh plasma density. In the case of our ion source, thehigh current density near the extraction hole makes anaccurate plasma model a necessity.The parameters of the ion source listed in Section IIare input into the IBSimu code. The geometry of theplasma electrode is imported to model the plasma aper-ture. Each of the electrodes is assigned the voltage corre-sponding to the experiment, and the plasma density usedin IBSimu can be used as a proxy for many of the filamentsettings. The plasma density can be approximated by us-ing the size of the plasma aperture and the total currentextracted the in experiment.IBSimu is used to simulate the extraction from the ionsource, however, over long distances the code becomescomputationally expensive. (The ion source extractionis only 0.1 m, where the full LEBT is several meters inlength.) Due to this, after the extraction the particle dis-tribution is transferred to WARP, a less computationallyexpensive code. After the extraction, the plasma modelis no longer crucially important, and so the use of WARPrather than IBSimu is acceptable. This is done by us-ing IBSimu to print a text file with phase parameters ofevery particle in the beam at a certain point along thebeamline, then importing that data into WARP. WARP isthen used to calculate the behavior of the beam in theremainder of the low energy beam line.
B. Low Energy Beam Transport (Warp)
Warp is a particle-in-cell Python package developed atLawrence Livermore National Laboratory and LawrenceBerkeley National Laboratory and has been in develop-ment since the 1980s. The particle distributions fromIBSimu simulations are loaded at the initial distributionsfor the Warp simulations. Particles are propagated usingthe wxy-slice package to solve for the fields at each step.Space-charge compensation is treated as a free param-eter and simulated by modifying each species’ current: I comp = I i · (1 − f e ) with f e the space-charge compensa-tion factor from literature [26]. A more accurate space-charge compensation model will be implemented in fu-ture simulations [27]. Simulations are done for emit-tance measurements and mass spectrometer measure-ments.Included in the simulations are models of the mag-netic fields and vacuum components. In order to accu- FIG. 7. Simulation of test beamline using Warp. The top plotshows 2-rms envelopes for H +2 and the plot below it showsthe maximum envelopes. Dotted lines show where the 6-waycrosses and vacuum tubes are. The vertical dashed lines indi-cate where the magnets are along the beamline. The left twoplots on the bottom row show the cross-section and xx’ phasespace of the initial beam. The middle bottom plot is the fi-nal beam cross-section. The two right plots on the bottom rowshow the xx’ and yy’ phase spaces at the end of the beamline. rately model the quadrupole and dipole magnets, CADmodels of the yokes and coils were imported to COM-SOL to simulate 3d fields [28]. These fields are importedto Warp with a scale factor to simulate different fieldstrengths. Vacuum components are accounted for by us-ing Warp’s installconductors function. Conducting cylin-ders are used for the vacuum tubes and 6-way crosses; arectangular box is used for the dipole chamber. Particlescoming into contact with conductors are removed fromthe simulation and the currents are adjusted after each FIG. 8. Simulation of dipole scan using Warp. The horizon-tal axis shows the dipole scale used in Warp, relative to thescale used for H +2 . The scale difference between simulations ischosen to match differences in the dipole settings of an actualscan. The current for each species is calculated at the end of thebeamline for each simulation. The total current of all species inthe simulation is plotted here. The measured current is scaledby a factor of 0.65 so that the H +2 peaks are the same height,for easier visual comparison. simulation step.For phase space simulations, a single dipole setting isused based on the species of interest. At each step arethe maximum and 2-rms, horizontal and vertical, beamenvelopes as shown in FIG. 7. The phase space of everyparticle remaining at the end of the simulation is alsosaved. These particle distributions are used to comparemeasured currents and emittances with simulations. Forsimulations of dipole scans, only the final step informa-tion of each particle species is needed.The simulations from IBSimu result in a circular beam,so the xx’ and yy’ phase spaces are identical. The asym-metries in the cross-section of the final beam arise pri-marily from the dipole field. These features were notpresent when ideal dipole fields from Warp were used.This is also what leads to the distinct features in the finalphase space plots. All ion species are simulated, but onlyselected species are plotted.The beamline was designed using simulations of pro-tons and H +2 . Once assembled, measurements of thebeam’s actual composition were taken as described inSection IV. New simulations are then done based on themeasurements, starting with IBSimu simulations. Themulti-species beam is than simulated in Warp (FIG. 8).By varying the dipole scale, the current measured in thesecond Faraday cup during a actual dipole sweep can besimulated. This allows us to check the accuracy of ourestimated beam species compositions. The simulationresults here include H + , H +2 , H +3 , N + , O + , H O + , N +2 ,and O +2 . Good agreement has been reached with the hy-drogen species. The higher mass ions are due to vacuumleaks and out-gassing of contaminants. Disagreement be-tween simulation and measurement of the higher massions is currently being investigated. Simulations usingmeasurements from the Faraday cup with the 1 cm slitwill be included in the next revision of this paper. IV. MEASUREMENTS
Here we present a first set of measurements, usingthe diagnostics described in Section II.
It should be notedthat these are very preliminary results and will be updatedin the near future!
One particular issue that we arecurrently investigating, is the presence of higher masscontaminants around M/Q = 18 (mass number/chargestate). These are discussed in the following subsection.Performance tests and variations of source parametersare then shown, followed by a brief discussion. As theseresults are currently highly preliminary, we forego theusual error analysis and present data as-is. The plasmaaperture is exchangeable and during the presented mea-surements, we used two different aperture size: 3 mmand 4 mm diameter, hence currents are usually reportedas densities in mA/cm . FIG. 9. Mass spectrum with an H mass flow of 1.5 sccm. Themost prominent species are indicated in the figure. Fluorineand Silicone most likely stem from the out-gassing of rubberO-rings. Water (and the accompanying OH + ) can be attributedto insufficient baking of the source and beamline. A. Source Contamination
The first systematic tests during commissioning of theMIST-1 ion source showed a number of contaminantsthat persisted through all measurements. As an exam-ple, FIG. 9 shows a typical mass spectrum with promi-nent species indicated.We have not found significant contamination aboveM/Q = 40 (a small Ar peak can occasionally be seen)and restrict our dipole scans to M/Q = 0 to M/Q = 45typically. The mass-to-charge ratios of the peaks that arefound in the recorded spectra hint at water contamina-tion ( O + , OH + , H O + ), a small air leak ( O + , O +2 , N + , N +2 , Ar + ), and a third source of contamination, likelyrelated to either the silicone rubber O-rings or the gasdelivery system ( F + , Si + ). Measures are currently takento remove these sources of contamination.In the remainder of the paper, we will report speciesfractions in the following way (e.g.): H +2 = 40 % (70 %HSp), to denote the percentage of total extracted beamand percentage of hydrogen species only (parentheses). B. Performance Tests
The first results we are reporting here are of the ionsource peak performance so far: The highest extractedcurrent density, the highest H +2 fraction, and the highesttotal extracted H +2 current density (a balance between H +2 fraction and total extracted current density). Highest total current.
With 5 mA of total beam cur-rent measured in faraday cup 1, the highest current den-sity recorded was ≈ mA/cm (4 mm diameter aper-ture). This was with a high discharge voltage of 150 Vand a H flow of 1 sccm. Accordingly (see discussion be-low) the species balance was shifted towards H +3 , witha H +2 fraction of ≈ % (26 % HSp) and H +3 fraction of ≈ % (60 % HSp). FIG. 10. Mass spectrum for MFC = 0.125 sccm and 80 V dis-charge. The H +2 fraction is 43 % (91 % HSp). Highest H +2 contribution. With a low H gas flowof 0.125 sccm, U discharge = 80 V, and I discharge = 4 A,the highest fraction of H +2 was recorded as 43 % (91 %HSp), the total extracted current was only 0.38 mA(5.4 mA/cm ) , however. The corresponding mass spec-trum is shown in FIG. 10. Highest current with high H +2 contribution. Thehighest recorded total current while H +2 was the dom-inant species was 1.1 mA. The H flow rate was 0.25sccm, I discharge = 5 . A, U discharge = 150
V. The H +2 frac-tion was 43 % (69 % HSp). C. Systematic Parameter Variations
In these preliminary tests, several parameter variationswere performed and mass spectra were recorded for eachset of parameters.
FIG. 11. Variation of hydrogen flow from the MFC. All contami-nant species are summed up and amount to ≈ % of the totalcurrent (held constant at 0.5 mA throughout the measurement.Second order polynomial trendlines are added to guide the eye. FIG. 12. Variation of discharge voltage. Hydrogen flow.
In this study, the total extracted beamcurrent was held stable at 0.5 mA. The MFC was changedfrom 0.5 to 2.5 sccm. The results are plotted in FIG. 11.Notably, the H +2 contribution rises towards lower pres-sures. This has consistantly been observed (see also pre-vious subsection). The contribution from the contami-nant species is nearly constant at ≈ %. The H +2 frac-tion is 21 % (39 % HSp) at 0.5 sccm. A later test withMFC = 0.125 sccm and discharge voltage reduced to80 V yielded a much higher H +2 fraction of 43 % (91 %HSp), albeit at the expense of a lower total beam current(0.38 mA), as discussed in subsection IV B and shown inFIG. 10. Discharge voltage.
Here we varied the discharge volt-age from 70 V to 140 V, while keeping the H flow rateconstant at 0.5 sccm. The total extracted current in fara-day cup 1 was 0.9 mA. As can be seen in FIG. 12, thecontribution of the contaminants is reduced, while thethree hydrogen species slightly increase. No clear trendfor the relative changes of hydrogen species with respectto each other can be seen. We attribute the sudden de-crease in proton current at 120 V and of H +2 current at130 V to changes in the plasma. Similar unstable regimeshave been observed in previous ion sources of the sametype [10]. Discharge current.
Here we varied the discharge cur-rent from 3 A to 3.5 A, while keeping the H flow rateconstant at 0.25 sccm. While increasing the dischargecurrent, the total extracted current increased linearly.This is not surprising, as a higher discharge current usu-ally indicates higher plasma density. As the dischargecurrent is increased, the processes in the plasma becomemore favorable for proton production than H +2 . As thisis a limited data set and only covers a very small portionof a large parameter space, this measurement should betaken with a grain of salt. FIG. 13. Variation of discharge current. Linear trendlines wereadded to guide the eye. -70 -60 -50 -40 -30 -20 -10 0 x (mm) -20-1001020304050 x ' ( m r a d ) H2+ Horizontal Cleaned -100 -80 -60 -40 -20 0 y (mm) -20-1001020 y ' ( m r a d ) H2+ Vertical Cleaned Up
FIG. 14. Preliminary emittance scans. Good qualitative andquantitative agreement was found with WARP simulations.Compare to FIG. 7, lower right phase space plots.
D. Emittance Measurements
At this point in time, we only performed a prelimi-nary test of the emittance scanners, which can be seenin FIG. 14. Good qualitative and quantitative agreementwas found when compared to the IBSimu/WARP simula-tion of the setup. The measured emittances were 0.56 π -mm-mrad and 0.45 π -mm-mrad for the horizontal andvertical scan, respectively. E. Discussion
The preliminary measurements indicate that H +2 be-comes the dominant species at low H mass flow (0.25sccm and below) and low discharge voltage (80 V andbelow). This is in agreement with earlier findings byEhlers and Leung [10, 15]. As H +2 has a short meanfree path before it either combines with an H to H +3 ordissociates, the H gas flow must be in balance with dis-charge voltage, and filament position. In the next mea-surement period, the filament position will be varied tofind the maximum extracted current with high H +2 frac-tion. The quality (amittance) of beam extracted from theion source is very good, as is typical for filament-driven, multicusp ion sources. Currently, we only have emit-tance scans after the beam has gone through the non-ideal LEBT, but as the agreement with the simulations isgood, we can extrapolate back to the ion source, whichyields 1-rms emittances in the range of 0.1-0.2 π -mm-mrad at extraction. The contamination of the beam withheavier species is a topic of ongoing study. We are cur-rently in a longer shutdown to remove possible sourcesof this contamination. V. CONCLUSION
We have presented a new filament-driven multicuspion source, designed to produce high currents of H +2 inDC mode for extended periods of time. For the novelRFQ Direct Injection method that is being proposed forthe IsoDAR experiment, the nominal goal is 15 mA of H +2 delivered by the ion source, with 10 mA the minimumsuccess condition. Here we reported maximum currentsof 5 mA from a 4 mm aperture. If this is scaled up to an8 mm aperture and taking into account the species com-position of the ion source, 4.4 mA of H +2 can currentlybe delivered from this source. This is about a factor 2short of the success condition. However, a large por-tion of the total beam current are contaminants and weare confident that these can be eliminated through care-ful cleaning, baking of the ion source and LEBT, and re-placement of certain elements prone to out-gassing. Thiswill further increase the available H +2 beam current. Fur-thermore, improved cooling will allow higher dischargecurrents, also leading to higher beam intensity. Con-sidering only the hydrogen species, we have seen H +2 fractions above 90%. Furthermore, we developed anaccurate simulation model of the ion source and beamline, that was compared with mass spectra and emittancemeasurements in 6-way cross ACKNOWLEDGMENTS
This work was supported by NSF grants PHY-1505858and PHY-1626069, as well as funding from the BoseFoundation. The authors are very thankful for the sup-port of the MIT Central Machine shop, the MIT PlasmaScience and Fusion Center (PSFC), and the University ofHuddersfield for support with machining, lab space andutilities, and equipment, respectively. [1] J. N. Bahcall and R. Davis, Science , 264 (1976), pub-lisher: American Association for the Advancement of Sci-ence Section: Articles.[2] A. Aguilar-Arevalo et al. (LSND), Phys. Rev.
D64 , 112007(2001), arXiv:hep-ex/0104049 [hep-ex].[3] A. A. Aguilar-Arevalo et al. (MiniBooNE), Phys. Rev. Lett. , 161801 (2013), arXiv:1303.2588 [hep-ex].[4] A. Diaz, C. A. Argelles, G. H. Collin, J. M. Conrad, andM. H. Shaevitz, arXiv:1906.00045 [hep-ex, physics:hep-ph] (2019), arXiv: 1906.00045.[5] A. Bungau, A. Adelmann, J. R. Alonso, W. Barletta, R. Bar-low, L. Bartoszek, L. Calabretta, A. Calanna, D. Campo,J. M. Conrad, Z. Djurcic, Y. Kamyshkov, M. H. Shae-vitz, I. Shimizu, T. Smidt, J. Spitz, M. Wascko, L. A.Winslow, and J. J. Yang, Physical Review Letters ,141802 (2012), publisher: American Physical Society.[6] M. Abs, A. Adelmann, J. R. Alonso, S. Axani, W. A. Bar-letta, R. Barlow, L. Bartoszek, A. Bungau, L. Calabretta,A. Calanna, D. Campo, G. Castro, L. Celona, G. H. Collin,J. M. Conrad, S. Gammino, R. Johnson, G. Karagiorgi,S. Kayser, W. Kleeven, A. Kolano, F. Labrecque, W. A.Loinaz, J. Minervini, M. H. Moulai, H. Okuno, H. Owen,V. Papavassiliou, M. H. Shaevitz, I. Shimizu, T. M. Shokair,K. F. Sorensen, J. Spitz, M. Toups, M. Vagins, K. Van Bib-ber, M. O. Wascko, D. Winklehner, L. A. Winslow, andJ. J. Yang, arXiv:1511.05130 [hep-ex, physics:physics](2015), arXiv: 1511.05130.[7] D. Winklehner, J. Bahng, L. Calabretta, A. Calanna,A. Chakrabarti, J. Conrad, G. D’Agostino, S. Dechoudhury,V. Naik, L. Waites, and P. Weigel, Nuclear Instrumentsand Methods in Physics Research Section A: Accelera-tors, Spectrometers, Detectors and Associated EquipmentAdvances in Instrumentation and Experimental Methods(Special Issue in Honour of Kai Siegbahn), , 231(2018).[8] R. Miracoli, L. Celona, G. Castro, D. Mascali, S. Gammino,D. Lanaia, R. Di Giugno, T. Serafino, and G. Ciavola, Rev.Sci. Intr. , 02A305 (2012).[9] R. F. Welton, M. P. Stockli, S. N. Murray, and R. Keller,Review of Scientific Instruments , 1793 (2004), pub-lisher: American Institute of Physics.[10] K. W. Ehlers and K. N. Leung, Review of Scientific Instru-ments , 1353 (1979).[11] G. Castro, D. Mascali, L. Celona, S. Gammino, C. Caliri,F. Di Bartolo, D. Lanaia, M. Mazzaglia, R. Miracoli,L. Neri, et al. , Review of Scientific Instruments , 096109(2014).[12] G. Castro, G. Torrisi, L. Celona, D. Mascali, L. Neri,G. Sorbello, O. Leonardi, G. Patti, G. Castorina, andS. Gammino, Review of Scientific Instruments , 083303(2016), publisher: American Institute of Physics.[13] J. R. Alonso, L. Calabretta, D. Campo, L. Celona, J. Con-rad, R. G. Martinez, R. Johnson, F. Labrecque, M. H.Toups, D. Winklehner, et al. , Review of Scientific Instru-ments , 02A742 (2014).[14] J. Alonso, S. Axani, L. Calabretta, D. Campo, L. Celona,J. M. Conrad, A. Day, G. Castro, F. Labrecque, and D. Win-klehner, Journal of Instrumentation (10), T10003.[15] K. Ehlers and K. Leung, Review of Scientific Instruments , 677 (1983). [16] A. Bungau, J. Alonso, L. Bartoszek, J. Conrad, M. Shae-vitz, and J. Spitz, Journal of Instrumentation (03),P03001, publisher: IOP Publishing.[17] A. Bungau, J. Alonso, L. Bartoszek, J. M. Conrad, E. Dun-ton, and M. H. Shaevitz, Journal of Instrumentation (07), T07002, publisher: IOP Publishing.[18] K. Inoue, International Journal of Modern Physics A (2004).[19] S. Axani, D. Winklehner, J. Alonso, and J. M. Conrad, Re-view of Scientific Instruments , 02B704 (2016).[20] T. Kalvas, O. Tarvainen, T. Ropponen, O. Steczkiewicz,J. ¨Arje, and H. Clark, Review of Scientific Instruments ,02B703 (2010), publisher: American Institute of Physics.[21] A. Friedman, R. H. Cohen, D. P. Grote, S. M. Lund, W. M.Sharp, J.-L. Vay, I. Haber, and R. A. Kishek, IEEE Trans-actions on Plasma Science , 1321 (2014), conferenceName: IEEE Transactions on Plasma Science.[22] S. Nishioka, S. Abe, S. Mattei, J. Lallement, T. Kalvas,A. Hatayama, and J. Lettry, in AIP Conference Proceedings ,Vol. 2011-1 (AIP Publishing LLC, 2018) p. 080017.[23] T. Kalvas, O. Tarvainen, J. Komppula, H. Koivisto, J. Tu-unanen, D. Potkins, T. Stewart, and M. Dehnel, in
AIPConference Proceedings , Vol. 1655-1 (AIP Publishing LLC,2015) p. 030015.[24] V. Toivanen, T. Kalvas, H. Koivisto, J. Komppula, andO. Tarvainen, Journal of Instrumentation (05), P05003.[25] O. Midttun, Y. Levinsen, R. Miyamoto, and C. Plostinar, in (JACOW, Geneva, Switzer-land, 2017) pp. 4445–4447.[26] M. Reiser, Theory and design of charged particle beams , 2nded. (Wiley-VCH, Weinheim, 2008).[27] D. Winklehner and D. Leitner, Journal of Instrumentation (10), T10006.[28] Comsol: Multiphysics software for optimizing designs(2020),