The light instrumentation setup for Phase II of the Gerda experiment has to meet several requirements of mechanical and of functional nature.
The quality of the scintillator, in this case liquid argon, plays a fundamental role for the veto performance. Impurities decrease light yield and attenuation length and therewith the probability to detect scintillation light in a given distance of the germa- nium detectors. The triplet lifetime of the scintillation is an indicator for the quality of liquid argon [82]. In 2012, the triplet lifetime of the LAr in the Gerda cryostat has been measured to be (922 ± 31) ns, as described in Sec. 5.1.2. As this seems sufficient for an effective LAr veto, based on the experience with the LArGe setup, the Gerda collaboration decided to not replace the LAr for Phase II.
As a consequence, the light instrumentation has to be inserted as a ready-build entity into a filled cryostat through the lock system, instead of being assembled in the cryostat. This decision allows to replace or work on the light detectors during operation. The instrumentation setup has to accommodate seven germanium detector strings
3.4. REQUIREMENTS FOR LIGHT INSTRUMENTATION IN GERDA
a) b)
Figure 3.7: Schematic drawing of the Phase II lock system of the LAr cryostat. a)
with up to eight detectors each. Due to the clean room height, the maximal height inside the lock which could be realized is 2800 mm and the diameter is 530 mm. One part of the lock system is composed of a movable tube of 930 mm height. This tube has to be connected to the rest of the lock by closing flanges at the top and bottom. If the total height of the lock should be used for the light instrumentation system a part of the light instrumentation setup has to be stored inside this tube and then connected to the rest prior to the closure of the lock. The maximal outer diameter of the setup should be ≤ 500 mm which allows for several millimeters tolerance at each side.
The setup should provide permanent access to the germanium detectors. To ease practical handling it should be possible to work on the germanium detectors with- out dismounting the whole fiber cylinder including the cabling. This requires that germanium detectors with their support can be moved independently from the light in- strumentation setup. In addition, the radioactive calibration sources have to be lowered inside the light instrumentation down to the height of the germanium detectors.
The deployed light detectors should perform stable over the whole measurement time of Phase II of the experiment. In conclusion, the Gerda collaboration decided in the beginning of the design process to explore two different light readouts, namely scintillating fibers read-out by silicon photomultipliers (SiPMs) and cryogenic photo-
CHAPTER 3. THE LIGHT INSTRUMENTATION CONCEPT
multipliers. The fibers seemed promising since it was assumed they can be put much closer to the germanium detectors due to their low internal radioactivity and the high expected self-vetoing. Additionally a fiber shroud would not be optically closed to the volume outside the shroud. The LArGe facility had proven in the past that impressive suppression factors can be reached with cryogenic photomultipliers and they were con- sidered as reliable technique. However, it was clear that the ETL 8” PMTs which were used in the LArGe experiment would be too radioactive to be operated in the Gerda cryostat and even new types of cryogenic PMTs with lower internal radioactivity would have to be placed at a distance ≥ 50 cm. The task of testing cryogenic PMTs with re- spect to their long-term stability has been undertaken by the group of Prof. Lindner at Max-Planck Institut f¨ur Kernphysik (the PMT assessment is described in Ch. 4). The group of Prof. Sch¨onert at TU Munich decided to investigate the light detection using scintillating fibers connected to SiPMs. However, since both light read-outs had not been operated in LAr for such a long time period the idea of using both simultaneously came up soon.
Another important component of a scintillation light veto system are wavelength shifting reflector foils and coatings for fibers and PMTs. As liquid argon emits scintilla- tion light at a wavelength of 128 nm it is not directly detectable by most light detectors. The common strategy is to shift the UV light to visible light using fluorescent chemi- cals, such as Tetraphenyl-butadiene, with good conversion efficiencies. The foil together with the wavelength shifter should be radiopure and mechanically stable at cryogenic temperature during Phase II of the experiment. In addition, the reflector foil should be highly reflective for visible light. As described inSec. 3.3, a VM2000 mirror foil coated with a matrix of TPB in polystyrene was used in the LArGe experiment. Due to a lack in long-term stability it is not sufficient for the use in the Gerda experiment and a new foil is developed for the light instrumentation setup in Gerda (seeSec. 6.1).
In Phase I of the experiment so-called mini-shrouds made off copper were installed around each detector string in order to minimize background from 42K. These mini- shrouds have to be replaced by transparent mini-shrouds in order to allow detection of scintillation light that is created inside the mini-shroud volume1. A short description is given in Sec. 6.1.
During physics data taking with germanium detectors it is favorable to operate prob- lematic germanium detectors in anti-coincidence mode. Detectors that are completely switched of count as dead volume as the energy depositions inside these germanium diodes are not detected. This undetected energy cannot be deposited in LAr and thus the veto efficiency of the light readout is artificially decreased. In contrast, the re- duction of the energy threshold for the germanium detector-detector anti-coincidence improves the veto efficiency of a light instrumentation in Gerda in the same manner. In Ch. 5 the expectations from Monte Carlo simulations concerning a light in- strumentation for the Gerda experiment are presented and in Sec. 7.3 a comparison between the LAr veto commissioning test results using radioactive calibration sources and the corresponding Monte Carlo results is made. The final design which takes into account both the mechanical and functional requirements and the results of the Monte Carlo simulation based optimization campaign is described in Sec. 6.1.
1
An alternative option would be to still use copper mini-shrouds but equip them with a light read-out inside the mini-shroud volume. In [103] a discussion of this option is given.
CHAPTER
4
PHOTOMULTIPLIER TUBE ASSESSMENT
For the LAr veto of the Gerda experiment eighteen 3” photomultiplier tubes of type R11065-10/20 from Hamamatsu Photonics K.K. [77] are used. These photomultiplier tubes are specified to have low radioactivity (∼ mBq level) and to work at cryogenic temperature.
This chapter explains briefly the functional principle of photomultiplier tubes (PMTs) along with important parameters. Afterwards, the R11065 photomultiplier type, to- gether with the test stands at MPIK, and the in-house developed voltage divider pro- duction optimized for low radioactivity and pulse shape are described. Sec. 4.4focuses on the long-term tests of the PMTs and the determination of important PMT charac- teristics. In particular, the issue of light emission of several PMTs during the operation in a cryogenic liquid and the countermeasures undertaken by the manufacturer are explained in detail.
4.1
Photomultipliers as light detectors
The functional principle of a photomultiplier tube is as follows: an incoming photon hits the photocathode of a PMT and is absorbed. Subsequently, an electron – a so- called photoelectron – is emitted with a certain probability via the external photoelectric effect if the energy of the incident photon is high enough. The interior of the PMT is kept under vacuum to avoid photoelectrons to collide with atoms or molecules in gas. The produced photoelectron is accelerated towards the first dynode which is at a slightly more positive voltage than the photocathode. To improve the collection of electrons a focusing electrode is placed in between. When the first dynode is hit, several electrons are knocked out, resulting in a multiplication of electrons. The second dynode is on an even more positive high voltage and the process of multiplication is repeated, leading to a growing cascade of emitted electrons. Eventually, all emitted photons are accelerated towards an anode. The electrons on the anode create a measurable electric current which is directly proportional to the number of electrons emitted from the photocathode.
Since amplification factors of approximately 106 can be reached, it is possible to detect single photons with PMTs and to reach a good separation of signal and noise. Voltage divider bases are necessary to provide each dynode with the appropriate voltage
CHAPTER 4. PHOTOMULTIPLIER TUBE ASSESSMENT
and to ensure linear response of the PMT.
The following PMT parameters are important for the successful operation in the low background experiment Gerda.
• Total detection efficiency: is given as the product of quantum efficiency and collection efficiency of the PMT. The total detection efficiency should be as high as possible to be able to detect single photons.
The quantum efficiency (Q.E.) is a measure for the probability that an incident photon creates a photoelectron. Nowadays, by means of a better understanding of the creation process of bi-alkali photocathodes, quantum efficiencies of ≈ 35% at 420 nm are reached [78]. Collection efficiency is a measure for the probability that an electron emitted from the photocathode is accelerated towards the first dynode and detected. It depends on the geometry inside the PMT, especially on the structure of the focusing electrodes and first dynode. In case of the R11065 PMTs it is 95% [43].
• Gain: is the factor by which a single photoelectron emitted from the photocath- ode is amplified by the manifold dynode structure. It is defined as
g = Qspe
e (4.1)
with Qspe the charge from a single photoelectron and e the elementary charge. A
high gain is preferable to discriminate between charge created by photoelectrons and charge created by noise signals. It allows to reduce the number of additional amplification stages by external electronics which itself induce noise.
• Dark count rate: arises mainly from thermal emission from the photocathode. Thermal emission of electrons is randomly distributed in time. The signals created by thermal electrons exhibit the same structure and characteristics as a real signal. Henceforth, they cannot be distinguished from a real signal. However, when operating the PMTs at cryogenic temperature this contribution is negligable. The remaining dark count rate may be due to internal radioctivity. It is important for any experiment to understand the rate of fake signals caused by such dark counts at the specific operation conditions.
• Afterpulses: are caused by photoelectrons that ionize rest gas molecules before reaching the first dynode. The ion drifts back to the photocathode and in turn generates a photoelectron. Afterpulses are registered at the anode with a time delay of typically 200 ns to a few µs, thus in the same time window as the slow component of the LAr scintillation light. Consequently, they contribute to the random coincidence rate.
• Linearity of PMT response: defines in which energy range the relation be- tween the number of incident photons and the detected photoelectrons is linear. It strongly depends on the design of the voltage divider base.
• Radioactivity: A low intrinsic radioactivity of the PMT of a few mBq is neces- sary to minimize the induced background level by the PMTs.