Characterization of large area photomutipliers under low magnetic fields: design and performances of the magnetic shielding for the Double Chooz neutrino experiment
E. Calvo, M. Cerrada, C. Fernández-Bedoya, I. Gil-Botella, C. Palomares, I. Rodríguez, F. Toral, A. Verdugo
aa r X i v : . [ phy s i c s . i n s - d e t ] M a y Characterization of large area photomutipliersunder low magnetic fields: design andperformances of the magnetic shielding for theDouble Chooz neutrino experiment
E. Calvo, M. Cerrada, C. Fern´andez-Bedoya, I. Gil-Botella,C. Palomares, I. Rodr´ıguez, F. Toral, A. Verdugo
Centro de Investigaciones Energ´eticas, Medioambientales y Tecnol´ogicas(CIEMAT), Av. Complutense 22, 28040 Madrid, Spain
Abstract
This paper describes the characterization studies under low magnetic fields of theHamamatsu R7081 photomultipliers that are being used in the Double Chooz ex-periment. The design and performances of the magnetic shielding that has beendeveloped for these photomultipliers are also reported.
The main goal of the Double Chooz experiment is to measure the mixing angle θ by searching for the disappearance of electron antineutrinos produced inthe CHOOZ nuclear power station in Ardennes (France). Double Chooz will bethe first neutrino experiment with enough sensitivity to improve the currentupper limit [1] in almost one order of magnitude in case no oscillation wasobserved.In order to eliminate the main systematic error source related to the neutrinoflux uncertainty, Double Chooz will use two identical neutrino detectors. Thefirst one (near detector) will be located at 400 m from the two reactors to mea-sure with good precision the flux and the energy spectrum of antineutrinos.The second one (far detector) will search for a deficit of electron antineutrinosand will be installed at 1.05 km from the reactor cores, in the hall built forthe CHOOZ experiment [2], at the end of the nineties. The Double Choozdetectors (Fig. 1) are composed by concentric cylindrical volumes filled withliquids of similar density but different optical properties. The innermost vol-umes conform the target and contain liquid scintillator. Surrounding thesective volumes there is a non-scintillating buffer inside a stainless steel tank,to reduce the events from accidental background. A total of 390 photomul-tiplier tubes (PMT) of 10 in. diameter (Hamamatsu R7081) are distributedon the inside of the stainless tank to collect the scintillation light. A detaileddescription of the Double Chooz detector is given in Ref. [3].Reducing background levels is a key factor to achieve the sensitivity goals inthis type of experiments. The Double Chooz detector, in addition to its un-derground location and the veto systems, is shielded by 15 cm iron bars toprevent the rock radioactivity from reaching the scintillator. The photomul-tiplier tube response may be affected by the Earth’s magnetic field and anadditional non-uniform contribution from the iron shield. The iron bars havebeen demagnetized in order to prevent the magnetic field being higher than 1G in the PMTs region.Although most photomultiplier tubes are affected by the presence of magneticfields, large area PMTs are particularly vulnerable due to the long trajecto-ries that the photoelectrons emitted in the photocathode have to travel untilthey reach the first dynode. PMTs with a large photocathode area have beenextensively used in the last years in high energy physics for neutrino detectionexperiments in order to cover very large detection areas with the minimumnumber of channels. Examples of this are the 20 in. R1449 Hamamatsu PMTsused in the Super-Kamiokande experiment, the 8 in. ETL 9351 in Borexino orthe 10 in. R7081 of Antares [4]. In these experiments the PMTs have to dealwith the uniform Earth’s magnetic field while the Double Chooz PMTs, aspreviously mentioned, will be immersed in a non-uniform and unknown mag-netic field due to the presence of the iron shield. In addition, this field could bedifferent for near and far detectors. The aim of the work reported in this paperis to determine the effect of magnetic fields up to 1 G on the performances ofthe Hamamatsu R7081 PMT and to find a solution for shielding these PMTsagainst the magnetic field. The PMT Hamamatsu R7081 [5] of 10 in. diameter has been chosen due to thesuitable size for Double Chooz detector scale. An appropriate photocathodecoverage of 13% is reached using 390 PMTs. In addition, the PMT providedby Hamamatsu has a very low radioactivity glass and is oil proof.The height of the PMT is 300 mm and the total weight is about 2500 g.2 ig. 1. The Double Chooz detector.
The cathode sensitivity range goes from 300 to 600 nm, being the maximumquantum efficiency about 25% at 400 nm. The dynode structure is
Box andLine with 10 stages. The typical gain is about 10 for voltages about 1500 V.Low magnetic fields, about 500 mG, are expected to modify the electron tra-jectory in our PMT. As mentioned before, the magnetic field in Double Choozdetectors is expected to be always below 1 G and not uniform. A dedicatedsetup has been made to quantify the effect of B-fields up to 2 G on the PMTin the three perpendicular directions. A device has been designed and built to provide a fixed and homogeneousmagnetic field in any direction, compensating the Earth’s field. The deviceconsists of three Helmholtz coils, each one placed perpendicularly with respectto each others, as it is shown in Fig. 2, defining the coordinate system. Thecoils are large enough to provide a homogeneous field in a 500 x 500 x 500mm volume where a black box containing the PMT is placed. A current of 1A feeding one Helmholtz coil yields a uniform field of 2 G in that direction.As Fig. 2 shows, the PMT is vertically placed inside a light tight box whichis long enough to accommodate also, on the top, the light source used for thetests. The PMT inside the black box is placed at the center of the Helmholtzcoils system with the dynode chain symmetric to the Y axis.The relationship between current and magnetic (B) field is measured priorto the inclusion of the black box using a three-axis magneto-resistive sensor,HMR2300 [6], which measures in the range from -2 to 2 G with an accuracy3 ig. 2. Helmholtz coils system. The light tight box is placed inside in such a waythat the PMT is in the center of the system. The right plot shows the PMT positionwith respect to the Helmholtz coils: Dy1 is the first dynode and Dy2 is the secondone, behind which the rest of the dynodes are placed. of 1 mG.The light source is a blue LED (DHC-SLR030NB40) controlled by a pulsercircuit designed at CIEMAT. The LED is placed at 70 cm from the PMTand its light is diffused to have a homogeneous illumination over the PMTphotocathode. Studies performed on the LED circuit stability over time givesan RMS variation of 0.2% and a maximum dispersion of 0.9% over 13 hours.A single RG-303 cable connects the PMT to the high voltage power supplyand to the DAQ electronics. Therefore, a circuit is necessary to split the PMTsignal from the high voltage: the so-called splitter , which is placed outside theblack box. The signal cable is sent from the splitter to the ADC (CAEN V265)where the signal is integrated in a 250 ns window and digitalized, in such a waythat 1 ADC count corresponds to 0.033 pC. The pulser circuit and the QDCare VME standard boards and are controlled through a LabView acquisitionprogram running on a PC. A diagram of the setup is shown in Fig. 3.The measurement procedure is the following: First, the Earth’s magnetic fieldis compensated. The response of the PMT is quantified in this environment, asreference, and the LED is calibrated (pulse time vs. emitted light). Afterwards,the B-field is increased up to 1 G in 250 mG steps and from 1 G to 2 G in4 ig. 3. Diagram of the experimental setup
500 mG steps in each of the 3 axis, keeping at zero the B-field in the otherdirections. At every step, the response of the PMT is analyzedThe expected effects of the B-field on the PMTs are:(1) A loss of collection efficiency due to the modification of the electrontrajectory between the photocathode and the first dynode.(2) A decrease of the gain and a degradation of the single-photo-electron (spe)spectrum and the timing properties due to the deviation of electrons inthe amplification chain.Disentangling both contributions would be only possible in spe regime bymeasuring the modified gain. However, a large amount of light would allow tobetter evaluate the global effect of the magnetic field up to higher fields thanfor the spe and would help to compute the effect over the collection efficiency.Therefore, the PMT response is tested at two light levels: spe and about 50photo-electrons (pe).All the following measurements have been performed on two different R7081PMTs showing very similar results. Accordingly, only the results of one ofthem will be reported in the next sections.
Once the Earth’s magnetic field is compensated, the PMT is illuminated with avery small amount of light to obtain the, so-called, spe spectrum. In Fig. 4, thespe spectrum of our PMT is shown. The gain can be obtained very preciselyfrom the fit of the spe spectrum to a parametric function describing the PMTresponse [7]. However, a statistical procedure has been chosen for this studyin order to determine the gain independently of the spectrum shape, which is5 / ndf c – m – – s – – s – a – – s – w – B – ADC counts200 250 300 350 400 450 500 550 E n t r i es / ndf c – m – – s – – s – a – – s – w – B – SPE spectrum
Fig. 4. Spe spectrum from Hamamatsu R7081 PMT in absence of magnetic field.The function which parameterizes the PMT response is superimposed in black. Theparameters describing the noise and signal are also shown [7]. expected to be strongly modified by the magnetic field. This method is basedon the Poissonian distribution of the number of pe and on the fact that thespectrum is divided in two contributions: noise (pedestal Gaussian) and signal(the rest of the distribution). The average number of pe ( µ ) is obtained fromthe number of events contained in the pedestal Gaussian . And, then, the gainis determined as the mean of the distribution, subtracting the pedestal, dividedby the average number of pe previously determined. The gain is measured tobe 50 ADC counts, equivalent to 10 electrons .The LED is also calibrated to obtain the pulse time window correspondingto a determined average number of pe for the PMT under study. Since thedetected light includes the PMT collection efficiency, which could be modifiedby the magnetic field, the light emitted by the LED should be measured atzero B-field. The spe spectrum is obtained for a LED pulse time of 11 ns,which corresponds to 0.2 pe at zero B-field. On the other hand, to obtain 50pe a pulse time of 80 ns is applied. In this regime, the PMT is illuminated with a LED pulse of 80 ns. The PMTresponse for different values of the magnetic field along X,Y and Z directionsis shown in Fig. 5. The largest effect is observed in -X direction (60% of thesignal is lost for 500 mG). As it is shown in Fig. 2, a field in -X direction The probability to have no pe is P (0) = e − µ The voltage applied to the PMTs is tuned to obtain a gain of 10 electrons. DC counts500 1000 1500 2000 2500 3000 3500 4000 E n t r i es B=0B=250 mGB=500 mGB=750 mGB=1000 mGB=1500 mGB=2000 mG
B field +X axis
ADC counts500 1000 1500 2000 2500 3000 3500 4000 E n t r i es B=0B=250 mGB=500 mGB=750 mGB=1000 mGB=1500 mGB=2000 mG
B field -X axis
ADC counts500 1000 1500 2000 2500 3000 3500 4000 E n t r i es B=0B=250 mGB=500 mGB=750 mGB=1000 mGB=1500 mGB=2000 mG
B field +Y axis
ADC counts500 1000 1500 2000 2500 3000 3500 4000 E n t r i es B=0B=250 mGB=500 mGB=750 mGB=1000 mGB=1500 mGB=2000 mG
B field -Y axis
ADC counts500 1000 1500 2000 2500 3000 3500 4000 E n t r i es B=0B=250 mGB=500 mGB=750 mGB=1000 mGB=1500 mGB=2000 mG
B field +Z axis
ADC counts500 1000 1500 2000 2500 3000 3500 4000 E n t r i es B=0B=250 mGB=500 mGB=750 mGB=1000 mGB=1500 mGB=2000 mG
B field -Z axis
Fig. 5. PMT response to a LED pulse time of 80 ns. The two top plots correspondto a B-field along X direction, the middle ones to a B-field along Y axis and thebottom ones to the Z direction. would drive the electron away from the first dynode. If the B-field is appliedalong the Z axis, the degradation of the PMT response is smaller because theLorentz force is null for electrons moving vertically. On the other hand, theeffect is symmetric in Y and Z, as expected.A summary of these measurements is shown in Fig. 6. Low fields of the orderof the Earth’s magnetic field ( ≈
500 mG) along -X axis degrade the PMTresponse to a 40%. In addition, a B-field of 1 G applied in any direction of thePMT transverse plane reduces the PMT signal by more than 80%.7 field (mG)-2000 -1500 -1000 -500 0 500 1000 1500 2000 R e l a t i ve ou t pu t BxByBz
PMT: TA-4369
Fig. 6. Relative response of the PMT with respect to zero B-field for different valuesof the magnetic field. The red curve corresponds to values of B along X direction,the blue one to Y axis and the green one to a B-field along Z axis.
In Double Chooz the study of the effect of the B-field on the spe spectrumsignal is crucial because the expected number of pe per PMT for the neutrinosignal is about 1.5. In our setup, the PMT is illuminated with a light pulseof 11 ns. Fig. 7 illustrates the effect of the magnetic field. The spe spectrumfor zero B-field (black histogram) is plotted together with the same spectrumfor a magnetic field of +500 mG along Y axis (red histogram). First, it isobserved that the peak from the signal of one pe is moved to a lower valueand the width of such a peak is larger due to the deviation of the electronsfrom their trajectory along the dynode chain. Second, the region betweenpedestal and signal (valley) is populated by badly amplified pe. A way toquantify the degradation of the spe signal is through the measurement of thepeak-to-valley (P/V) ratio. This magnitude takes into account the size of thepeak (good amplified photo-electrons) and the population of the valley (verybad amplified pe). The measured P/V ratio for different B-fields is shownin Fig. 8. A PMT with a P/V ratio below 2.5 does not fulfill the experimentrequirements. This happens for B-fields about 500 mG applied perpendicularlyto the PMT.In addition, the spe spectrum allows to analyze separately the effect of theB-field over the PMT gain and over the collection efficiency. The effectivegain and the average number of pe ( µ ) are recalculated for B-fields up to 1 G.The relative variation of the collection efficiency is equivalent to the variationof µ if the light source is constant. The variation of µ is obtained from thestrong illumination regime because the LED in this last case is less sensitive8 DC counts200 250 300 350 400 450050100150200250300350400450500
B=0=+500 mG Y B PMT: TA-4369
Fig. 7. Spe spectra of the PMT illuminated by a LED pulse of 11 ns. The blackhistogram corresponds to the response of the PMT for B=0 and the red one for aB-field of 500 mG along Y direction.
B field (mG)-2000 -1500 -1000 -500 0 500 1000 1500 2000 P eak -t o - va ll ey r a t i o BxByBzTA 4369
Fig. 8. Peak-to-valley ratio as a function of the applied magnetic field. The red curvecorresponds to values of B along X direction, the blue one to Y axis and the greenone to a B-field along Z axis. to temperature and voltage supply variations than in the spe range.The relative variation of gain and collection efficiency as a function of theapplied B-field are shown in Fig. 9. The effect of the magnetic field is largeron the collection efficiency than on the gain, as expected for these large pho-tocathode area PMTs. The asymmetric behavior of the PMT gain for B-fieldsapplied along X direction is very important due to the asymmetry of the dyn-ode chain in this axis. The gain even increases for positive fields because thephoto-electrons are more effectively focused from first to second dynode.9 field (mG)-1000 -500 0 500 1000 G / G D - BxByBz
Gain variation
B field (mG)-1000 -500 0 500 1000 E ff / E ff D - BxByBz
Collection Efficiency variation
Fig. 9. Variation of the gain (left) and collection efficiency (right) as a function ofthe applied B-field.
The study presented in this paper shows that the Hamamatsu R7081 PMTresponse is reduced between 30% and 95% for transverse B-fields from 500 mGto 1 G. This so large effect will reduce dramatically the performance of ourdetectors. A degradation of the PMT response will affect the measurement ofthe energy due to the loss of collected pe. A worse energy resolution wouldaffect the separation signal to noise and the sensitivity of the neutrino energyspectrum to θ . On the other hand, since the B-field is expected to be differentin both detectors, an accurate PMT calibration should be required in orderto avoid additional uncertainties between them.We have observed that the gain can be recovered increasing the applied volt-age but not the collection efficiency. Therefore, a magnetic shielding becomesmandatory to keep the good performance of these PMTs. A possible way to shield the PMTs against the magnetic field is by using com-pensating coils for the whole detector. However, this solution presents severalproblems: It does not cancel completely a non-uniform magnetic field and usesan extra space not available in the tight pit of the Chooz lab. Therefore, theDouble Chooz collaboration has chosen a passive individual shield for everyPMT. 10 .1 Design
CIEMAT group has designed a passive magnetic shield considering the differ-ent constraints and requirements. The loss of PMT signal due to the presenceof the magnetic field is required to be smaller than 10%. Considering the re-sults reported in the previous sections, that is equivalent to accept a transverseB-field smaller than 150 mG. The performance of the shield is measured bymeans of the transverse shielding factor, defined as the ratio between the ex-ternal and internal transverse B-fields. If the maximum external field is 2 G(a conservative value for the design), the required shielding factor should be13.The simplest design is a cylinder without endcaps fitting the PMT dimensions:300 mm diameter and about 300 mm high. The material and its thickness werechosen according to its permeability. The mu-metal is a very known highpermeability material that satisfies our requirements. A very detailed study [8]has shown that the magnetic properties of the material as specified by themanufacturer are not reliable at these so very low fields. In the gauss range,the magnetic behavior of the mu-metal depends strongly on the size of themagnetic domains, which depends, in turn, on the mu-metal sheet thicknessand thermal treatment, and should be determined experimentally for eachsample. Several materials of different thickness and heights have been testedand some of them satisfied our requirements. Finally, the mu-metal from MecaMagnetic Company [10] of 0.5 mm thickness was chosen due to its very goodperformance and other practical aspects as the price and delivery time. Thelength of the shield is reduced to 275 mm because of mechanical constraints.The last step in the shielding design is to define the position of the PMT insidethe cylinder. The transverse shielding factor is not homogeneous along the mu-metal cylin-der axis. The shielding factor (SF) is maximum at the center and decreasessteeply as moving away from the center because of the openings. Fig. 10 showsan example of this behavior. In the openings, the SF does not depend on thematerial but only on the geometry of the shield. The photocathode to the firstdynode region is the most affected by the magnetic field and has to be shieldedas much as possible. Therefore, the PMT should be placed below the shieldedge, where the shielding factor is smaller than our requirements (SF ≈ Mu-metal is a registered trademark of Magnetic Shield Corp. [9]. ig. 10. Shielding factor as a function of the position on the axis for a mu-metalcylinder from Meca Magnetic of 305 mm high, 280 mm diameter and 0.5 mm thickin presence of a 2 G external field [8]. a 300 mm diameter cylinder). In addition, the PMT acceptance would be re-duced due to the cylinder shadow over the photocathode area. The final shieldposition should be a compromise between the SF achieved and the acceptanceof photons.The response of the shielded PMT is measured in the experimental setupshown in section 2.2 for different positions of the PMT inside the shield. Thisstudy has been done with a 0.25 mm thick cylinder of Magnetic Shield Corp..The length of this cylinder is 380 mm and is moved vertically, defining itsposition by the length of the shield protruding above the PMT top. The inter-nal walls of the shield were covered with a black sheet to avoid the reflectedlight that could distort our measurements. A layout of the PMT inside themu-metal cylinder is shown in Fig. 11. The measurements have been done for5 positions: -8.5 cm (shield edge below the PMT top, Fig. 11-right) , 0 (at thePMT top level, Fig. 11-center), 2.5 cm, 5 cm and 7.5 cm (shield edge abovethe PMT top, Fig. 11-left). The PMT performance has been tested for a largeamount of light, applying the magnetic field separately along each axis whilekeeping null the B-field along the other coordinates.The results are presented in Fig. 12. The shielding is maximum in the trans-verse coordinates (X and Y), as expected, and satisfies our requirements evenwith the shield edge at the PMT top level (h=0). On the contrary, the ver-tical field (Z), for which smaller shielding is required, is very badly shieldeddue to the openings and depends much on the shield position. Moreover, forthese vertical fields, a degradation of the PMT response with respect to thenon-shielded PMT is observed when the cylinder is below the photocathode(h=-8.5 cm). The reason is that the lines of magnetic field are disturbed by12 ig. 11. Layout of the PMT inside the mu-metal cylinder for three different relativepositions. H (cm) Photon yield (pe) Relative reduction (%)10 203.6 155 205.2 140 209.8 12Bare PMT 239.3 0Table 1Photon yield measured by 390 PMTs for a 1 MeV electron generated at the DoubleChooz detector center for different positions of the PMT inside the shield. H is thelength of the shield above the PMT. The relative reduction on photon yield due tothe shield shadow is also shown. the presence of the mu-metal and a transverse field is created in the PMT. Tosummarize, the cylinder edge should be above the PMT top more than 2.5 cmto guarantee a degradation of the PMT response below 10% up to 1 G.On the other hand, the cylinder has been included in the software simulationof the Double Chooz detector in order to determine the new acceptance fordifferent heights above the PMT top. In Table 1, the detected photon yieldfor an electron of 1 MeV generated at the detector center is shown for threepositions of the PMT with respect to the cylinder. The result for a bare PMTis also shown as reference. The light detected by the 390 PMTs is reduceda 12% when the shield edge is at the PMT top and does not decrease muchwhen the cylinder protrudes from the PMT 5 cm or even 10 cm.A mu-metal cylinder from Meca Magnetic has also been tested in three differ-ent positions: 0, 2.5 cm and 5 cm, showing the same behavior but a slightlybetter performance (Fig. 13). This result confirms our previous measurementsin [8].Taking into account that the acceptance of the PMT does not vary so muchfrom 0 to 5 cm shield edge position, and considering also mechanical con-straints for the PMT system support, the conservative solution of placing theshield 5.5 cm above the PMT top has been chosen.13 field (mG)-2000 -1500 -1000 -500 0 500 1000 1500 2000 R e l a t i ve ou t pu t no shieldedh = -8.5 cmh = 0h = 2.5 cmh = 5 cmh = 7.5 cm Magnetic Field X-axis
B field (mG)-2000 -1500 -1000 -500 0 500 1000 1500 2000 R e l a t i ve ou t pu t no shieldedh = -8.5 cmh = 0h = 2.5 cmh = 5 cmh = 7.5 cm Magnetic Field Y-axis
B field (mG)-2000 -1500 -1000 -500 0 500 1000 1500 2000 R e l a t i ve ou t pu t no shieldedh = -8.5 cmh = 0h = 2.5 cmh = 5 cmh = 7.5 cm Magnetic Field Z-axis
Fig. 12. Relative output of the PMT with respect to zero B-field for different mag-netic fields and positions of the mu-metal shield with respect to the PMT top (h).The black histogram corresponds to a non-shielded PMT.
Shield length above PMT (cm)-10 -8 -6 -4 -2 0 2 4 6 8 R e l a t i ve ou t pu t Magnetic Shield Co.Meca Magnetic
Bare PMTRequirement
Magnetic Field Z-axis = 1G
Fig. 13. Relative output of the PMT with respect to zero B-field for B Z = − Finally, the performance of the PMT has been tested with the shielding se-lected to be used in Double Chooz: Meca Magnetic of 275 mm high, 300 mmdiameter and 0.5 mm thick. The distance between the top of the PMT andthe shield edge was 5.5 cm. Table 2 presents a summary of the performance14 X (mG) -2000 -1000 -500 +500 +1000 +2000 RMSSignal(B =0)/Signal(B=0) (%) 90.7 95.4 97.5 102.2 104 107.3 ± / Gain (%) -8 -4 -2.5 +1.6 +3.7 +6.3 ± coll ) / E coll (%) -1.3 -0.6 -0.04 +0.6 +0.2 +0.9 ± Y (mG) -2000 -1000 -500 +500 +1000 +2000 RMSSignal(B =0)/Signal(B=0) (%) 99.2 100 100 100 99.8 99.2 ± / Gain (%) +0.1 +0.9 -0.4 +0.5 +0.3 -0.3 ± coll ) / E coll -0.96 -0.88 +0.6 -0.5 -0.5 -0.5 ± Z (mG) -2000 -1000 -500 +500 +1000 +2000 RMSSignal(B =0)/Signal(B=0) (%) 72.7 96 99.2 99.6 96.7 75.2 ± / Gain (%) -12.7 -2.3 -0.9 +0.5 -1.2 -11.8 ± coll ) / E coll -16.7 -1.7 +0.18 -0.9 -2.2 -14.8 ± X , B Y and B Z . The response of the PMT is described by its signal for a LED pulsetime of 80 ns, its gain and the relative variation of its collection efficiency. of the shielded PMT under different B-fields. The possible degradation of thePMT response is measured through the signal of a large amount of light. Thegain and collection efficiency were measured as mentioned in section 2.5. Theerrors (RMS) quoted in the same table have been determined propagating thestatistical error of the corresponding variable. The effect of a magnetic fieldin Y is negligible up to 2 G. For magnetic fields along X axis, while the col-lection efficiency is fully recovered, the gain is still slightly affected. The mostprobable reason is that the lower edge of this shield is near the dynode chainand disturbs the lines of the magnetic field. As a result, an additional fieldalong X is created distorting the amplification chain. Finally, the effect of theB-field in Z is small enough, about 4% at 1 G, to be controlled by calibration.To summarize, the loss of PMT signal for magnetic fields up to 1 G is alwaysless than 5% in any direction. The measurements presented in this paper show that the PMT HamamatsuR7081 is strongly affected by magnetic fields smaller than 1 G. The PMTsignal is reduced more than 60% for B-fields close to the Earth’s magneticfield along X direction. A mu-metal cylinder wrapped around the PMT shieldsvery effectively transverse magnetic fields, while the shielding in the vertical15irection depends on the cylinder length protruding the PMT photocathode.It was demonstrated that, using a 275 mm high, 300 mm diameter and 0.5 mmthick cylinder with its edge placed 5.5 cm above the photocathode, the PMTresponse is reduced less than 5% for B-fields up to 1 G in any direction. Theinstallation of these shields in all PMTs of the two Double Chooz detectorswill, therefore, ensure that no systematic effects due to magnetic field willdegrade the expected sensitivity reach of the experiment in the measurementof θ . Acknowledgements
The authors would like to thank the members of the Double Chooz collab-oration for their comments and suggestions on this work and, specially, F.Suekane for providing us with the Hamamatsu PMTs used in this study.
References [1] Th. Schwetz, New J. Phys. 10:113011, 2008.[2] M. Apollonio et al., Eur. Phys. J. C27, 331-374, 2003.[3] F. Ardellier et al., Double Chooz Collaboration,
Double Chooz: A search for theneutrino mixing angle θ Preprint arXiv:hep-ex/0606025, 2006.[4] Super-Kamiokande: Y. Fukuda et al., Nucl. Instrum. Meth. A501, 418 (2003).Borexino: G. Alimonti et al., Astropart. Phys.16:205-234,2002. Antares: J.A.Aguilar et al., Nucl. Instrum. Meth. A555, 132-141, 2005.[5] jp.hamamatsu.com/products/sensor - Parameterization of the response of the photomultiplierHamamatsu R7081