Forward Beam Monitor for the KATRIN experiment
A. Beglarian, E. Ellinger, N. Hau?mann, K. Helbing, S. Hickford, U. Naumann, H.-W. Ortjohann, M. Steidl, J. Wolf, S. Wüstling
PPrepared for submission to JINST
Forward Beam Monitor for the KATRIN experiment
KATRIN collaboration
A. Beglarian 𝑏 E. Ellinger ,𝑎 N. Haußmann 𝑎 K. Helbing 𝑎 S. Hickford 𝑐,𝑎
U. Naumann 𝑎 H.-W. Ortjohann 𝑑 M. Steidl 𝑒 J. Wolf 𝑐 and S. Wüstling 𝑏 𝑎 Department of Physics, Faculty of Mathematics and Natural Sciences, University of Wuppertal, Gauss-Str.20, D-42119 Wuppertal, Germany 𝑏 Institute for Data Processing and Electronics (IPE), Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany 𝑐 Institute of Experimental Particle Physics (ETP), Karlsruhe Institute of Technology (KIT), Wolfgang-Gaede-Str. 1, 76131 Karlsruhe, Germany 𝑑 Institut für Kernphysik, Westfälische Wilhelms Universität Münster, Wilhelm-Klemm-Straße 9, 48149 Mün-ster, Germany 𝑒 Institute for Nuclear Physics (IKP), Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz1, 76344 Eggenstein-Leopoldshafen, Germany
E-mail: [email protected]
Abstract: The
KArlsruhe TRItium Neutrino (KATRIN) experiment aims to measure the neutrinomass with a sensitivity of 0 . β -electron spectrum of tritium decay. The electrons from tritium β -decay are produced in the Windowless Gaseous Tritium Source (WGTS) and guided magneticallythrough the beamline. In order to accurately extract the neutrino mass the source properties, inparticular the activity, are required to be stable and known to a high precision. The WGTS thereforeundergoes constant extensive monitoring from several measurement systems. The
Forward BeamMonitor (FBM) is one such monitoring system.The FBM system comprises a complex mechanical setup capable of inserting a detector boardinto the KATRIN beamline inside the
Cryogenic Pumping Section with a positioning precision ofbetter than 0 . s − mm − .The detector board contains a hall sensor, a temperature gauge, and two silicon detector chips of p-i-n diode type which can measure the β -electron flux from the source with a precision of 0 . = Corresponding author. a r X i v : . [ phy s i c s . i n s - d e t ] J a n ontents p-i-n diodes 114.4 Data acquisition 12 The KATRIN experiment will improve the sensitivity of neutrino mass measurements to 𝑚 𝜈 = . 𝜎 discovery potential for a mass signal of 𝑚 𝜈 = .
35 eV [1, 2]in the most sensitive direct neutrino mass experiment to date. The neutrino mass will be derivedfrom a precise measurement of the shape of the tritium β -decay spectrum near its endpoint at 𝐸 = (
18 573 . ± . ) eV [3]. The source of β -electrons is a Windowless Gaseous Tritium Source (WGTS) which has an activity of 10 Bq. – 1 –he layout of the KATRIN beamline [4] is shown in figure 1. The
Source and TransportSection (STS) consists of the WGTS, the
Differential Pumping Section (DPS), the
CryogenicPumping Section (CPS), and several source monitoring and calibration systems [5]. Along thebeamline superconducting solenoids generate a magnetic field of several Tesla strength which adia-batically guides the β -electrons towards the spectrometers while excess tritium is pumped out of thesystem. The Spectrometer and Detector Section (SDS) consists of the pre-spectrometer, the main-spectrometer, the monitor-spectrometer, and the
Focal Plane Detector (FPD). All spectrometers areof MAC-E-Filter type which transmit electrons with energies above a chosen retarding energy [6],and reject those with lower energies. The main-spectrometer can perform an energy analysis of the β -electrons with an energy resolution of 0 .
93 eV at 18 . Windowless, Gaseous Tritium Source (WGTS)Rear Section(RS) Differential Pumping Section (DPS)Cryogenic Pumping Section (CPS) Pre-Spectrometer Main Spectrometer Focal-Plane Detector (FPD)Monitor SpectrometerCondensed Kr Source (CKrS)Forward Beam Monitor (FBM)Gaseous Kr Source (GKrS)Ion source(ELIOTT) Photoelectron sourceBeta-Induced X-ray Spectroscopy System (BIXS)
Figure 1 . The KATRIN beamline. The FBM is located at the end of the CPS and represents the final sourcemonitoring system before the β -electrons enter the spectrometer and detector section. The source-related parameters associated with the main systematic uncertainties in the deter-mination of the neutrino mass are activity fluctuations of the WGTS, energy loss corrections (of β -electron scattering in the WGTS), the final state distribution, the source magnetic field, and thesource plasma condition.In order to analyse the tritium β -spectrum and determine the neutrino mass the WGTS needsto be extremely stable, particularly in its isotopic composition and column density. Therefore, theWGTS properties need to be known with high precision, and are continuously monitored for shortand long term fluctuations. There are several monitoring and calibration subsystems associatedwith the WGTS [5].Results from the various subsystems are combined over long time periods during extendedmeasurement time. This paper focuses on one such activity monitoring system, the Forward BeamMonitor (FBM). The FBM is the final monitoring subsystem for β -electrons from the source beforethey enter into the spectrometer and detector section. It has been commissioned prior to theKATRIN krypton measurement campaign in June 2017 [7]. Initial data was then obtained duringthe krypton measurement campaign and during the KATRIN first tritium measurement campaignin May 2018 [8]. The FBM is capable of continuously monitoring variations of the electron flux– 2 –nd changes in the measured shape of the β -decay spectrum during the KATRIN neutrino massmeasurement phases.This paper is organised as follows. In section 2 the WGTS and its operating parameters areintroduced and in section 3 the FBM measurement principle for the monitoring of the relevantWGTS parameters is explained. Section 4 contains a technical description of the FBM. In section 5the FBM commissioning and results from the krypton and first tritium measurement phases arepresented, and section 6 contains the conclusion. The
Windowless Gaseous Tritium Source (WGTS) is the origin of β -electrons whose observedspectrum will ultimately lead to the measurement of the neutrino mass [9]. The general setup of theWGTS is shown in figure 2. It is a column of tritium gas inside a cylinder with a diameter of 90 mmand a length of 10 m. The latter is situated in a homogeneous magnetic field of 3 . 𝑝 in = 10 − mbar, and is pumped out at both ends with a constant outletpressure of 𝑝 out = . 𝑝 in . Figure 2 . Setup of the WGTS. Tritium is injected into the centre of the cylinder and pumped out at bothends. The flux tube is surrounded by superconducting magnets to guide the β -electrons. The longitudinaldensity profile of the tritium molecules along the column is shown above. The column density is defined as tritium molecule density integrated along the central axis of thesource, i.e., the number of tritium molecules per source cross section area. The neutrino massmeasurement depends on the accurate description of inelastic scattering of electrons by the gasmolecules inside the source. There are several key parameters of the WGTS that need to be keptstable with high precision in order to achieve a high sensitivity in the neutrino mass measurement.These include • Beam tube temperature
The molecular tritium gas must be at cryogenic temperatures of <
80 K to minimise corrections– 3 –o the electrons energy due to thermal movement of the decaying mother atoms. The coolingconcept is based on a two-phase liquid neon thermosiphon [10, 11]. • Pressure
The amount of tritium inside the source scales with the inlet pressure. Stabilisation is achievedusing a pressurised control vessel from which tritium flows via a capillary to the beam tube. • Tritium purity
A high isotopic purity of molecular tritium gas ( >
95 %) is required. The tritium purity 𝜖 T isgiven by the ratio of the number of tritium atoms to the total sum of atoms in the WGTS. Inaddition to T other isotopolouges include DT, HT, D , HD, and H . The tritiated hydrogenisotopolouges differ in their mass, recoil energies, and the rotational and vibrational finalstate distributions of their daughter molecules following tritium decay. The gas compositionis measured via LAser RAman spectroscopy (LARA) [12, 13].These key parameters have an effect on the rate and/or energy of the electrons emitted from thesource. There are several control and monitoring systems in the KATRIN experiment with thepurpose of meeting the precision and stability requirements of the key source parameters.The column density, N , can be obtained by combining an in-situ measurement of the tritiumpurity with an activity (decay rate) measurement. The count rate 𝑆 of β -electrons from the sourceas measured by activity detectors scales as 𝑆 = 𝐶 · 𝜖 𝑇 · N (2.1)where 𝐶 is a proportionality constant encompassing experimental properties such as detectorefficiency and acceptance, and the half-life of tritium. Small fluctuations of the source parameterslead to changes in the observed shape of the differential β -electron spectrum. Fluctuations in thecolumn density are expected to be in the 10 − regime. Given the targeted sensitivity for the neutrinomass measurement, column density and tritium purity must not give rise to an uncertainty beyond 𝛿𝑚 𝜈 = . × − eV to the neutrino mass analysis. The β -electrons resulting from the decay of the tritium are adiabatically guided towards the spec-trometer and detector section. The transport section is also used to eliminate the tritium flowtowards the spectrometers which must be free of tritium in order to meet the necessary backgroundrequirements for neutrino mass measurements. The transport section consists of a DifferentialPumping Section (DPS) and a
Cryogenic Pumping Section (CPS).The DPS consists of five beam tube segments within superconducting solenoids with turbo-molecular pumps between each [14]. The CPS traps all remaining traces of tritium by cryo-sorptionon argon frost at 4 K condensed on the gold plated surfaces of the beam tube [15, 16]. Both the DPSand CPS have 20 ° chicanes to block the line of sight for the diffusing tritium gas and to increase theprobability that the tritium molecules get pumped away or hit the walls of the beam tube.At the end of the transport section the tritium flow is suppressed by 14 orders of magnitudecompared to the center of the WGTS. The electron flow is unaffected and all electrons are guidedadiabatically towards the spectrometer and detector section.– 4 – .3 Activity detectors Two activity detectors measure the count rate of β -electrons from the decay of tritium in the WGTS.These detectors1. provide information about fluctuations of the WGTS activity on a timescale of minutes and2. are used (together with the measured tritium purity) to monitor the column density with 0 . β -electrons impact on the rear wall [5]. The second activitydetector is called the Forward Beam Monitor (FBM). It is located in the transport section, mountedbetween the last two superconducting solenoids of the CPS. Here the tritium flow has been sup-pressed by a factor of 14, to approximately 10 − mbar l s − , which minimises background effectsand contamination from tritium. The magnetic field in this position is axially symmetric with amagnitude of 0 .
84 T so the spatial homogeneity of the source profile can be studied. The FBM isthe final measurement component before the spectrometer and detector section.
The FBM measures β -electrons from the tritium source as they are guided to the spectrometer anddetector section. Hence, the β -electrons are following the beamline when they are detected by theFBM. Such a detector must not shadow any part of the electron flux tube that will be used for themeasurement of the neutrino mass. Therefore, the FBM configuration is such that the detector islocated in the outer rim of the electron flux during neutrino mass measurements. The active radiusof the flux tube used for measurement is approximately 71 mm and the outer rim in which thedetector is situated is up to 7 mm wide.The p-i-n diode detectors have an energy threshold of approximately 5 keV, dependent onthe background noise and the type of diode used. This lower energy value is determined duringcalibration of each diode. For an accurate rate measurement the lower energy threshold needs to bestable. Figure 3 . Cross section of the FBM setup with the electron flux tube in the KATRIN beamline. Duringnominal monitoring operation the FBM is situated in the outer rim of the flux tube, up to approximately7 mm in thickness. – 5 –t is assumed that the activity measurement in the outer rim of the flux tube is representativeof the activity across the entire beamline cross section. Variations of the column density in theradial direction are expected to be on the 10 − level [17]. The assumption that the outer rim isrepresentative of the entire flux tube is verified during repeated calibration runs when the FBMis moved across the beamline. These two operation modes of the FBM are standard “monitoringmode” and calibration “scanning mode” and are described in the following sections. Monitoring mode is the standard mode of operation for the FBM. It is intended for permanentand continuous monitoring of the source activity and the main observable is the electron countrate. Together with the measurement of the tritium purity, the FBM monitoring mode providescontinuously information on the column density of the source.
Flux tube scans are performed during calibration of the KATRIN experiment. The purpose ofscanning is to1. confirm that the activity in the beamline outer rim is representative of the entire flux tube,2. map any irregularities in the cross section of the flux tube, and3. define the area of the flux tube entering the spectrometer and detector section (i.e. measurepossible shadow effects by STS instrumentation).During the KATRIN experiment calibration runs are performed between neutrino mass measure-ment runs once every ∼
60 days. During commissioning and initial measurement campaigns thescanning mode was used more frequently.
In the following sections a technical description of the FBM is given. A more detailed descriptioncan be found in [18]. Further information on the basic concept and the early development of theFBM can be found in [19] and [20].
The measurement of the electron flux is performed under ultra high vacuum (UHV) conditions in apotentially tritium contaminated environment. The main mechanical requirements for the vacuummanipulator are:1. to situate the FBM detector in the outer rim of the flux tube without shadowing the maindetector and additionally to move it throughout the cross section of the flux tube,2. to be capable of removing all FBM components out of the CPS allowing full metal sealedvacuum valves to separate the FBM volume from the CPS volume, and– 6 –. to provide a safe enclosure for tritium, complying with all radiation safety regulations of thetritium laboratory.An overview of the complete FBM setup is shown in figure 4 and figure 5. electricalcabinet CPSprofilestandroof CPSvalve FBMvalveFBMsix-waycrosslong bellows small bellows tripodstand
Figure 4 . The FBM hardware setup. The beamline is located in the CPS perpendicular to the FBM mainaxis. The CPS and FBM valve separates the FBM from the CPS if the detector is in parking position withinthe FBM six-way cross. With the help of the 2 m long bellow the detector arm can be driven into the fluxtube within the CPS.
The vacuum components of the FBM setup are separated from the CPS by a gate valve.Behind this valve the FBM detector board is completely removed from the KATRIN beamline.Attached to this volume are the turbomolecular pump and pressure gauges. Behind the main FBMvacuum volume are bellows, support structures, stepper motors, rotary encoders, and electricalfeedthroughs. These components provide the movement of the FBM detector board and the readoutof the measured data.The movement of the detector board is realised by combining two linear drive mechanisms.A long stainless steel support tube with an outer diameter of 54 mm can be moved over a distanceof 1 . 𝜇 p materials (such as stainless steel 1 . 𝜇 p < . supporttrolleysbackendinternalCPSmain body CPSflangefront endinserted sliderlongbellows rotary movement rodsupportflanges rodsupportlongtubeTMP Figure 5 . CAD drawing of the FBM as it is inserted into the CPS. Parts of PP2, the CPS, and half of theFBM’s bellows are invisible for a better illustration of the mechanics.
To facilitate an easy slipping onto the second support flange the cylinder has a chamber at itsforward end. Two cut-outs extend the movement limits in 𝑦 -direction and provide space for theelectrical feeding.The axis of the detector holder is made of steel 1 . . . p-i-n diodes. The full lever arm length from the axis to the tip (includingthe cover) is 130 mm and the maximum width of the detector equals the width of the cover whichis 50 mm. The electrical connector is covered from the electron beam by a thin steel plate.The turbomolecular pump is located vertically above the main FBM vacuum volume and iscapable of pumping speeds up to 260 l s − (nitrogen). Two pressure gauges are mounted belowthe FBM vacuum volume which cover the range from 1 . × − mbar to 1 . × − mbar. In– 8 – ain cylinder coverPEEKconnectorrackpinionlever arm axis (ball bearing mounted) tip of FBMholes for p-i-n diodes back plate Figure 6 . The FBM manipulator front end. The detector board is fixed on the end of a lever arm which isrotated by a rack and pinion drive. order to reach the required vacuum level the setup is baked out periodically after being exposed toatmosphere.
The two stepper motors mentioned in subsection 4.1 (12 . . . ° resolution) are not directly acting on the spindle axes but with one stage transmissions using toothedwheels. Since the FBM is not equipped with motor breaks the 𝑥 -transmission is chosen such thatthe torque at the motor is sufficiently small to withstand the vacuum forces even if it is not poweredanymore.Since it is possible that the stepper motors miss steps without being noticed, absolute rotaryencoders are used to determine the position of the FBM because they retain the full information ofthe position even during a power cut. These optical encoders work with up to 16-bit single turn and14-bit multi turn resolution, i.e. 2 steps per revolution and in total 2 revolutions can be counted.This sums up to an overall resolution of 2 steps. To minimise mechanical play both encoders areconnected directly to their corresponding spindle axes. The main spindle has a slope of 2 . − µ m can be reached. However, due to mechanical tolerancesthe actual precision is significantly lower as will be described in subsection 5.1.To fulfill stringent safety requirements the motion control of the FBM is implemented on a Field-Programmable Gate Array (FPGA) which continues to run during power cuts with the help ofaccumulators. It directly monitors and controls the motor, encoders, and sensors and also includesa fast full safety retraction of the FBM which allows closure of the safety valves to separate theFBM volume from the CPS.The FPGA communicates with two KATRIN internal database systems: the
ZEntrale daten-erfassung Und Steuerung (ZEUS) server and the
Advanced Data Extraction Infrastructure (ADEI)server [21]. All data obtained by the FPGA is automatically transferred and available on bothservers. Safety-critical systems, such as vacuum pumps, valves, pressure gauges, and end switches,are integrated within the KATRIN PCS7 safety system. Three backup batteries ensure operationalreadiness during a power cut. – 9 – igure 7 . Left:
The FBM detector board is made of polyimide and equipped with SMD parts. The two p-i-n diodes are glued to the tip of the board and their signals are amplified by two separate transimpedanceamplifiers. Close to the p-i-n diodes a PT-1000 temperature sensor and a Hall sensor are located.
Right:
Picture of the Hamamatsu S9055 p-i-n diode in TO-18 casing with the lid removed. The silicon diode itselfis mounted on a ceramic carrier which can be taken out of the casing.
The main tasks of the FBM are to monitor the electron flux within the electron beam and to obtainthe beta spectrum of tritium. Detector chips with a thin entrance window (dead layer) are used toallow the detection of electrons with energies below 10 keV. In addition this also allows detection oflow energy ( <
60 keV) photons which is important for calibrating the detector. The FBM features aUHV compatible two channel detector board, including detector chips of silicon type and additionalsensors, as described below.
The detector board (PCB) is made of polyimide to meet the vacuum and material requirements.To enhance thermal conductivity of the board and to dissipate the heat produced by the electricalcomponents, the PCB is a flexible, thin (0 . .
84 T in the centre of the flux tube and is axiallysymmetric. The magnetic field is measured in only one axis and the electron flux should follow thismagnetic field exactly with the exception of upstream blockages. The measurement of the magneticfield is therefore also useful for additional positioning and alignment measurements.Temperature stabilisation is important as the p-i-n diode leakage current rises exponentiallywith detector temperature. Therefore, the energy resolution and stability of the energy threshold aredependent on the detector temperature and effect the spectra obtained. To record the temperature aPT-1000 sensor is placed on the detector board near the p-i-n diodes and the Hall sensor.The board is mounted on a 5 mm-thick aluminum back plate attached to the moving components.It is glued to the back plate with a UHV compatible two-component adhesive to ease the mountingof the electrical parts and for better thermal conductivity. The electronics are covered by a stainlesssteel metal shield to protect them from electrons and ions in the beamline as well as from radiofrequency interference. The detector board has “cut out” corners in order to reduce the area of the– 10 –ux tube that is covered, and features two holes which allow electrons to reach the p-i-n diodes.The electronics and detectors on the FBM detector board are connected via a custom-made PEEKconnector with cabling running through the FBM manipulator to the vacuum feedthroughs. p-i-n diodes
The preamplifiers of the two p-i-n diode detector channels are DC coupled charge sensitive amplifierswhich operate in a continuous reset mode. Each preamplifier consists of a low-noise JFET frontend in common-source configuration and an operational amplifier (op-amp) connected in a non-inverting scheme. The feedback loop stretching across both stages consists of a 𝑅 = Ω resistorin parallel with a 𝐶 = . 𝜏 = 𝐶 · 𝑅 = . 𝐹 𝐴𝐶 = 𝑈 / 𝑄 = / 𝐶 = − translation factor, but also a current readout can be performed bylooking at the DC voltage offset at the output of the preamplifier with 𝐹 𝐷𝐶 = − .The fundamental components of the FBM are the p-i-n diode detector chips. There are twosilicon p-i-n diodes mounted on the detector board which detect the β -electrons from the tritiumsource. These two p-i-n diodes can have different active sensitive areas. The silicon p-i-n diodesare manufactured by Hamamatsu Photonics and can be type S5971, S5972, S5973, or S9055-01which have sensitive areas of different sizes (see table 1). One advantage of these detectors is thattheir casing and properties are all identical, the only difference is their respective sensitive area.This means the electronic design of the detector board can remain the same and the board with the p-i-n diodes that most suits the measurement purposes can be mounted and inserted into the fluxtube. Furthermore the dead layer does not exceed 1 µ m. Diode 𝐴 𝑠 𝐴 𝑠 𝐴 𝑠 [mm ] 𝐶𝐶𝐶 [pF] 𝐼 dark 𝐼 dark 𝐼 dark [pA] 𝑑 dead 𝑑 dead 𝑑 dead [nm] 𝑇 𝑠 𝑇 𝑠 𝑇 𝑠 [s] 𝑇 𝑚 𝑇 𝑚 𝑇 𝑚 [s]data sheet (at 10 V) (at 10 V) data sheet measured S9055-01 0.008 0.5 2.0 <1000 300–500 498 252S9055 0.031 0.8 2.0 129 64.6S5973 0.126 1.6 1.9 31.7 15.9S5972 0.503 3.0 10 8.0 4.0S5971 1.131 3.0 70 3.5 1.8
Table 1 . Parameters of the FBM p-i-n diodes. Capacitance 𝐶 and dark current 𝐼 dark are taken from thedata sheets. The thickness of the dead layer 𝑑 dead was determined with an electron gun and Monte Carlosimulations. The two right columns show the time to build a sufficiently detailed spectrum in monitoringmode ( 𝑇 𝑚 ) and scanning mode ( 𝑇 𝑠 ). The casing of these diodes is metal and includes a large glass window. Since the windows ofthese TO-18 casings would prevent the detection of any electrons the diodes are removed from thehousing and directly mounted (using two-component adhesive) onto the FBM detector board. TheHamamatsu S5971 p-i-n diode detector chip is shown in figure 7.The choice of the p-i-n diode size is based on the expected rate from the tritium source withineach measurement phase (larger diodes are used for commissioning measurements where the amount– 11 –f tritium is lower).The statistical error of the measurement is dominated by the number of electrons that arecounted by the detector and is given by Δ 𝑁𝑁 = √ 𝑁 = √︁ 𝐴𝜙𝜖𝑡 (4.1)where 𝐴 is the sensitive area of the p-i-n diode, 𝜙 is the electron flux density, 𝜖 is the detectorefficiency, and 𝑡 is the measurement time. The detector efficiency includes losses due to backreflected electrons and pile-up effects. To reach the required precision of Δ 𝑁 / 𝑁 = . 𝑡 = . 𝐴𝜙𝜖 (4.2)Assuming an energy threshold of 7 keV approximately of the tritium spectrum is measured. Usingthis reduction factor, an electron flux density of 10 s − mm − and a detector efficiency of 𝜖 =
65 %the measurement time needed to reach the required 0 . p-i-n diodes iscalculated and listed in table 1.The one unknown property of these p-i-n diodes is their individual dead layer. During man-ufacturing the thickness of the dead layer is not measured and therefore not available a priori, butlimited to 1000 nm. The thickness of the dead layer is indicated by the minimum energy that canbe detected. The measurement of the dead layer is done by analysing the shape of the peak frommonoenergetic electrons originating from an electron gun (see section 5.2). Figure 9 illustratessuch an analysis. Measurements of the dead layer are performed for each p-i-n diode before theyare mounted on the FBM detector board. It is assumed that the dead layer remains constant overtime, even after bakeout cycles of the vacuum setup. This is because the dead layer is silicon oxidewhich is not affected by heat and requires approximately 10 electrons (on the order of severalyears in the FBM location) to suffer from radiation damage. For the two p-i-n diode detector channels an Amptek PX5 and an Amptek DP5 are used for thedata readout. These are digital pulse processors with build-in amplifiers used to amplify the signalby up to a factor of 100. These Amptek devices are connected to a Mac computer running the
Object-orientated Real-time Control and Acquisition (ORCA) software [22]. An ORCA readoutmodule was specifically designed for the FBM Amptek devices. The raw ORCA data is convertedinto ROOT files for analysis. The preamplifier outputs of the two p-i-n diode detector channels canalso be connected to separate low-pass filters to measure the DC offset occurring from the eventrate on the respective p-i-n diode chip.The pulse processing parameters of each detector channel can be optimised to obtain either thecount rate or a spectrum of the β -electrons from the source. The peaking time is set to • Fast channel: 1 . µ s to measure the count rate (larger p-i-n diode with higher count rate) • Slow channel: 3 . µ s to measure the spectrum (smaller p-i-n diode with lower count rate)– 12 – po s i t i on [ mm ] − − − −
20 0 20 40 60 80 y po s i t i on [ m m ] − − −
200 204060 M agen t i cfi e l d [ T ] . . . . . . . M agen t i cfi e l d [ T ] . . . . . . . FitData x po s i t i on [ mm ] − − − −
20 0 20 40 60 80 y po s i t i on [ m m ] − − −
200 204060 M agne t i cfi e l d [ T ] − . − . − . . . . . . M agne t i cfi e l d [ T ] − . − . − . . . . . . Residuals −
50 0 50 x position [mm] − − − − y po s i t i on [ mm ] . . . . . m ag . F i e l d [ T ] Figure 8 . Left:
Data and fit result of the calibrated and temperature corrected 𝑧 -component of the magneticfield in the CPS. Right:
The residuals of the simulated magnetic field which shows a good agreement withthe data.
During scanning the required measurement time at each point is reduced due to the increasedelectron flux towards the centre of the beam tube. The analysis of the FBM data is based on theestablished analysis systems of the KATRIN experiment. Therefore, all data, slow control, and runfiles are available on the ADEI server and KATRIN databases.
This section presents selected results [18] of the measurements performed with the FBM duringits commissioning phases as well as during the first KATRIN measurement campaigns. Theseresults serve as an evaluation tool for the positioning accuracy of the vacuum manipulator andthe performance of the detector. In some cases the data is compared to the results of numericalsimulations of the detector response.
Positioning reproducibility is the ability of the FBM to find a position relative to a former position.This is different to the absolute positioning accuracy which includes external reference pointswith respect to the KATRIN coordinate system. The reproducibility is validated by using a lasersetup as well as a portable
Coordinate Measuring Machine (CMM). It was determined to be betterthan 0 . 𝜎 alignment 𝜎 alignment 𝜎 alignment [mm] ˜ 𝜎 max ˜ 𝜎 max ˜ 𝜎 max [mm] 𝜎 full 𝜎 full 𝜎 full [mm] Offset [mm] 𝑥 .
28 0 .
042 0 . − . ± . 𝑦 . .
07 0 .
13 4 . ± . Table 2 . The overall positioning accuracy 𝜎 full results from the combination of the uncertainties of thealignment 𝜎 alignment and the positioning reproducibility ˜ 𝜎 max . There is a misalignment between the FBM andthe flux tube expressed by a constant offset which was determined from flux tube scans. – 13 –o calibrate the movement system, as well as to find the center of the flux tube, the magneticfield in the CPS can be used (see left panel in figure 8). The shape of the magnetic flux can bedescribed by a two-dimensional Gaussian. The required calibration values, namely the encodervalue for the horizontal lever arm and the offset of the magnetic flux center to the FBM system(listed in the last row in table 2), are given by the free parameters in a fit of data taken during a fluxtube scan.To demonstrate the excellent positioning accuracy of the manipulator a thin (0 .
14 mm diameter)electron beam was scanned with the FBM by moving the detector (type S5971 with 1 . . p-i-n diode, it is rather the diode being scanned by the beam than vice versa. The plotin figure 10 shows the measured intensities as a function of detector position. The large circularcontours represent the entrance window of the diode (small, 1 . . 𝑥 FBM = − . 𝑦 FBM = . For calibration KATRIN is equipped with an electron gun which is situated in the rear section andcan provide a mono-energetic electron beam with energies up to 20 keV. In the left panel of figure9 the measured detector response to 18 . p-i-n diode. The best matchwas obtained with a dead layer thickness of 340 nm (see figure 9). The simulations overestimatethe data in the low energy tail which is caused by an incomplete model which does not includethe magnetic field configuration in the CPS. The small bump at approximately 5 keV is the resultof reflected electrons which are guided back to the detector within the peaking time for the DAQdue to magnetic mirroring in the CPS. It was possible to determine the dead layers of the FBM p-i-n diodes which range from 300 nm to 500 nm causing an energy dependent shift of the measuredpeak of 0 . Before the actual tritium measurement an alternative front end, equipped with a Faraday cup, wasinstalled to the FBM in order to check ion blocking, measure the radial ion distribution in thebeamline, and check the simulated source gas models by measuring secondary electrons [24]. Themeasurements with the p-i-n diode detector started with the “first tritium measurement campaign“[25] which took place from the 5 th to the 20 th of May 2018 with a gas mixture of 0 .
10 15 20
Kinetic energy [keV] R a t e [ / s ] electron gun at 18.2 keVsim. with 340 nm dead layer Energy [keV] . . . . . . E f fic i en cy []
340 nm dead layer / ◦ pitch ε d ε r ε th (5 keV) ε i (5 keV) Figure 9 . Left:
Measured and simulated electron gun peaks obtained during the first tritium campaign. Thesimulation includes a detector energy resolution with FWHM = .
35 keV. The best match was obtained witha dead layer thickness of 340 nm.
Right:
The simulated efficiencies for electrons not to get reflected fromthe detector ( 𝜖 𝑟 ), not to get stopped in the dead layer ( 𝜖 𝑑 ), and for exceeding the energy threshold of 5 keV( 𝜖 th (5 keV)). The intrinsic efficiency of the detector is then given by 𝜀 𝑖 (5 keV) = 𝜀 𝑟 · 𝜀 𝑑 · 𝜀 th (5 keV). y p o s i t i o n [ mm ] rate cut at 100 cps C o un t r a t e [ / s ] Figure 10 . Scan of stationary electron gun beam. The FBM detector is moved through the beam in a gridwith 0 . . p-i-n diode. The detector chip profiles are positioned such that they comprise the highest rate. With a fraction of only 0 . − mm − was expected at the FBM measuring plane. Therefore, the largest p-i-n diodes havebeen chosen (1 . ) to optimise counting statistics. The peaking time of the DAQ for bothchannels was 6 . µ s, resulting in a pile-up rate of about 3 % which can be neglected for stabilityanalyses (see section 5.3.4).Acceptance tests were performed prior to the campaign to extract calibration parameters, energyresolutions, and noise thresholds of the detectors. These measurements were performed with an Am source in the vented system with the FBM in parking position. The source was placed at aclose distance between the two p-i-n diodes. The desired diode could then be irradiated using themovement mechanics and be adjusted to find the maximum count rate. Figure 12 shows one of the
Am spectra extracted from these measurements. The calibration parameters are obtained by a– 15 – igure 11 . β -electron rate trend summary of the first tritium campaign. The full available data from thestability measurements at the monitoring position for both channels is plotted. The count rates for channel1 are approximately 0 . . .
02 % per hour while for the single regions this value is smaller than 0 .
01 % per hour.
Energy [keV] − − R a t e [ / s ] NpL α NpL η,β
NpL γ Am γ (26 . Am γ (59 . photoelectrons Am γ s Noise
Am dataGlobal fit
Figure 12 . Data and fit of the
Am spectrum. The spectrum is recorded with the FBM detector with anenergy resolution of 𝜎 FWHM ≈ Am lines are identified and labeled. The full fitfunction consists of a combination of Gaussians and error functions for each (strong)
Am line in the peak.The global fit comprises 33 free fit parameters. Only the Gaussian parts representing the strongest lines areused for the calibration of the FBM. global fit to the whole spectrum. – 16 – .3.2 Spectrum
The spectrum shown in figure 13 is the first tritium spectrum recorded with the FBM. Between 6 keVto 20 keV the spectrum agrees with the expectation, however below 6 keV the slope is unexpectedlyincreasing. This is probably due to background counts from noise and edge effects from the diodes.This may also explain why the spectra of the two channels do not match perfectly for lower energies.Other likely sources for this mismatch, which is also the reason for about 2 % lower rate in channel1 than in channel 2 during the whole campaign, are • uncertainties in the energy calibrations which cause the deviations among the channels forlower energies, • small differences in the active area, or • small differences in the dead layer thickness of the two p-i-n diodes. Energy [keV] R a t e [ / s / . k e V ] Channel 1 / cut at 4.6 keV / rate=1967 cpsChannel 2 / cut at 4.6 keV/ rate=2000 cps
Figure 13 . Tritium β -spectrum measured with both channels of the FBM detector during the first tritiumcampaign. The rate deviation of approximately 2 % between the two channels is probably caused by theuncertainties in the calibrations or differences in the active surface or dead layer thickness of the p-i-n diodes. Several scans of the β -electron flux cross section were performed recording the tritium count rate,the magnetic field, and the temperature. During a scan, the temperature usually drops by about1 ° C. This occurs when the detector is moved further into the cold CPS where the detector directlyfaces the 4 K cold beam tube of the CPS in which the argon frost layer is prepared.Figure 14 shows the results of scans over the cross section of the flux tube for both detectorchannels. The electron flux shows the expected Gaussian shape where the rate drops from the centerto the outer rim by approximately 10 % as predicted by simulations [26]. It can be seen that theevent rate for identical positions changes during the scans which affects the extracted mean of thefits. Nevertheless, the means are compatible to the results from the alignment measurements insection 5.1 which use the magnetic field data. This is expected as the electron flux scales with themagnetic flux. – 17 – − −
20 0 20 40 60 x/y Position [mm] R a t e [ s − ] X : f t mean= ( − . ± . mm / χ red =0.52X : f t mean= ( − . ± . mm / χ red =0.75Y : f t mean= (5 . ± . mm / χ red =2.13Y : f t mean= (6 . ± . mm / χ red =0.81 monitoring position −
50 0 50 x position [mm] − y po s i t i on [ mm ] −
50 0 50 x position [mm] − y po s i t i on [ mm ] R a t e [ s − ] Figure 14 . Radial dependence of the count rate derived from a cross scan during the first tritium campaignwith channel 1 of the FBM.
Top:
1D Gaussians are fit to the data for each horizontal (X , ) and vertical(Y , ) scan. The Gaussian means are compatible with the results from magnetic field measurements. Onecan clearly see that for identical positions slightly different rates are measured, for example the rate increasedduring the 𝑥 -scans such that the mean of the X fit is lower than for X . The Gaussian widths are approximately 𝜎 =
165 mm.
Bottom:
2D scatter plot of the same data. The scans for 𝑦 are not perfectly on a vertical linedue to the chosen scan pattern which explains the larger uncertainties in the fits. During the two weeks of the first tritium campaign the FBM was mainly monitoring the flux inthe CPS at position 𝑥 FBM =
65 mm (outer rim of the flux tube, see figure 14). From time totime background measurements were taken slightly out of the beam at 𝑥 FBM =
80 mm. The fullrate trend graphs are shown in figure 11 for both detector channels including linear fits to thedata. The entire monitoring time is separated into six time regions. There is a long term drift ofapproximately 0 .
02 % / h determined from all regions, while for single regions the drift is generallysmaller, especially for the longer regions 2, 3, 5, and 6, hence the reason for the larger long term driftmust mainly originate from incidences which occur between the regions. Several investigationshave been performed to find the source of this long-term drift, and there are hints that the detectorresponse changes over time due to an increase in the noise level and degrading effects of the detectorchip. Hence this drift is probably caused by the FBM and not by a change of the incoming electronflux. The latter assumption is supported by the results of the other monitoring systems which do– 18 –ot observe such a drift. However, this long-term drift is sufficiently small as the FBM is designedto monitor relative source fluctuations over short time intervals, such as seconds, minutes and atmaximum a few hours. Within these time ranges the drift is within the required sensitivity of 0 . 𝜖 T measurement performed by the LARA system. In this campaignthe tritium amount was limited to about 1 % in deuterium, and consequentially statistical fluctua-tions in the determination of the concentration of the tritiated hydrogen isotopolouges were muchstronger than it is the case for standard operation ( 𝜖 T > . ≈ . The KATRIN experiment aims for a precise measurement of the electron antineutrino mass witha sensitivity of 0 . Forward Beam Monitor (FBM). The FBM has the advantage of being capable ofcontinuously monitoring variations of the electron flux and changes in the observed shape of the β -decay spectrum with high accuracy on short time scales.A UHV compatible vacuum manipulator was commissioned. It is able to place a detector boarddirectly into the beta-electron flux originating from the tritium source. Although the mountingposition of the apparatus demands a movement mechanism with a working stroke of 1 . . p-i-n diodes. The FBM detector reaches an energy resolution of about 𝜎 FWHM = O (10 cps) and thus to measure relative changes in the electron flux with 0 . p-i-n diodes has a large impact on the detector responsewhen measuring electrons. It was found that the dead layer thickness of the p-i-n diodes used forthe FBM range from 300 nm to 500 nm.After commissioning, the FBM was employed for several KATRIN measurement campaigns.The capabilities of the FBM detector were confirmed as well as the positioning accuracy of themanipulator. A small long term (days to weeks) drift of the rate was observed which correlates to– 19 – drift of the noise level of the electronics. On short time scales (hours) the FBM is stable to theper-mille level. With this the FBM is a monitoring device which reaches all its design goals.With its good performance the FBM data already played a key role in reducing the systematicuncertainties of the tritium concentration 𝜖 T fluctuations during the first tritium campaign. Thiswas achieved by combining it with the LARA data which featured a relative statistical uncertaintyof only a few percent on time scales of minutes due to the low amount of source gas molecules.This will be important for an upcoming keV-scale sterile neutrino analysis of the first tritium data. Acknowledgments
We acknowledge the support of the Ministry of Education and Research BMBF (05A14PX3,05A17PX3) and the Helmholtz Association.
References [1] A. Osipowicz et al.
KATRIN: A Next generation tritium beta decay experiment with sub-eV sensitivityfor the electron neutrino mass. Letter of intent . 9 2001.[2] J. Angrik et al.
KATRIN design report 2004 . 2005.[3] M. Aker, K. Altenmüller, M. Arenz, M. Babutzka, J. Barrett, S. Bauer et al.
Improved upper limit onthe neutrino mass from a direct kinematic method by KATRIN. Phys. Rev. Lett. , 123:221802, Nov2019.[4] M. Arenz, M. Babutzka, M. Bahr, J. Barrett, S. Bauer, M. Beck et al.
Commissioning of the vacuumsystem of the KATRIN main spectrometer. Journal of Instrumentation , 11(04):P04011–P04011, apr2016.[5] M. Babutzka et al.
Monitoring of the operating parameters of the KATRIN Windowless GaseousTritium Source . New J. Phys. , 14:103046, 2012.[6] A. Picard et al.
A solenoid retarding spectrometer with high resolution and transmission for kevelectrons. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions withMaterials and Atoms , 63(3):345 – 358, 1992.[7] M. Arenz, W.-J. Baek, M. Beck, A. Beglarian, J. Behrens, T. Bergmann et al.
First transmission ofelectrons and ions through the KATRIN beamline.
Journal of Instrumentation ,13(04):P04020–P04020, apr 2018.[8] Max Aker, K. Altenmüller, M. Arenz, W.-J. Baek, J. Barrett, A. Beglarian et al.
First operation of theKATRIN experiment with tritium. The European Physical Journal C , 80(3):264, 2020.[9] Lutz Bornschein, Beate Bornschein, Michael Sturm, Marco Röllig, and Florian Priester for theKATRIN collaboration.
Status of the neutrino mass experiment KATRIN. Fusion Science andTechnology , 67(2):274–277, 2015.[10] S. Grohmann.
Stability analyses of the beam tube cooling system in the KATRIN source cryostat.Cryogenics , 49(8):413 – 420, 2009.[11] S. Grohmann, T. Bode, M. Hötzel, H. Schön, M. Süßer, and T. Wahl.
The thermal behaviour of thetritium source in KATRIN. Cryogenics , 55-56:5 – 11, 2013. – 20 –
12] M. Sturm, M. Schlösser, R. J. Lewis, B. Bornschein, G. Drexlin, and H. H. Telle.
Monitoring of allhydrogen isotopologues at tritium laboratory Karlsruhe using Raman spectroscopy. Laser Physics ,20(2):493–507, 2010.[13] Magnus Schlösser et al.
Accurate calibration of the laser Raman system for the Karlsruhe TritiumNeutrino Experiment. Journal of Molecular Structure , 1044:61 – 66, 2013. MOLECULARSPECTROSCOPY AND MOLECULAR STRUCTURE 2012.[14] A. Marsteller et al.
Neutral tritium gas reduction in the KATRIN differential pumping sections.Vacuum , 184:109979, 2021.[15] W. Gil, J. Bonn, B. Bornschein, R. Gehring, O. Kazachenko, J. Kleinfeller, and S. Putselyk for theKATRIN collaboration.
The Cryogenic Pumping Section of the KATRIN experiment. IEEETransactions on Applied Superconductivity , 20(3):316–319, 2010.[16] Carsten Röttele.
Results of the first Cool-down of the KATRIN Cryogenic Pumping Section. Journalof Physics: Conference Series , 888:012228, sep 2017.[17] Markus Hötzel.
Simulation and analysis of source-related effects for KATRIN . Ph.D. thesis, KIT,Karlsruhe, 2012.[18] Enrico Ellinger.
Development and investigation of the Forward Beam Monitor for the KATRINexperiment . Ph.D. thesis, Bergische Universität Wuppertal (BUW), 2019.urn:nbn:de:hbz:468-20191127-123110-5.[19] Udo Schmitt.
Entwicklung eines Monitordetektors für das KATRIN-Experiment . PhD thesis, KIT,Karlsruhe, EKP, 2008.[20] Martin Babutzka.
Untersuchung eines verfahrbaren Monitordetektors zur Überwachung der Aktivitätdes 𝛽 -Zerfalls in der kryogenen Pumpstrecke des KATRIN-Experiments . Diploma thesis, KarlsruherInstitut für Technologie (KIT), 2010.[21] Marco Kleesiek. A Data-Analysis and Sensitivity-Optimization Framework for the KATRINExperiment . Ph.D. thesis, Karlsruhe U., EKP, 2014.[22] M.A. Howe, G.A. Cox, P.J. Harvey, F. McGirt, K. Rielage, J.F. Wilkerson, and J.M. Wouters.
Sudbury Neutrino Observatory neutral current detector acquisition software overview. IEEETransactions on Nuclear Science , 51(3):878–883, 2004.[23] Agostinelli, S. et al.
Geant4 - a simulation toolkit. Nuclear Instruments and Methods in PhysicsResearch Section A: Accelerators, Spectrometers, Detectors and Associated Equipment ,506(3):250–303, 2003.[24] M. Klein.
Tritium ions in KATRIN: blocking, removal and detection . PhD thesis, Karlsruher Institutfür Technologie (KIT), 2018. 10.5445/IR/1000093526.[25] Max Aker et al.
First operation of the KATRIN experiment with tritium . The European PhysicalJournal C , 80(3):264, 2020.[26] Norman Haußmann.
Simulation and measurement of the Forward Beam Monitor detector signal forthe KATRIN experiment . Ph.D. thesis, Bergische Universität Wuppertal (BUW), 2019.urn:nbn:de:hbz:468-20191009-085446-9.. Ph.D. thesis, Bergische Universität Wuppertal (BUW), 2019.urn:nbn:de:hbz:468-20191009-085446-9.