WorksPhysics and Astronomy
Study of polysterene based scintillator ageing in the DANSS experiment
Abstract
DANSS is a spectrometer of reactor antineutrino based on plastic scintillator. The sensitive volume of the detector is made of 2500 polystyrene based scintillator plates with wavelength shifting (WLS) fiber readout (strips). We present a study of the strips light yield during 6.5 years of DANSS continuous running. Overall ageing at the rate % per year is observed that is considerably smaller than in other similar experiments. We also observe the WLS fiber attenuation length shortening at the rate % per year.
Introduction
The idea of the DANSS detector [1, 2] was to have a scintillation neutrino spectrometer on a movable platform as close as possible to the core of the industrial nuclear reactor. The site just below the reactor body provides several advantages like very high neutrino flux () and moderate overburden ( m w.e.), but at this site only low flammable materials are allowed due to the fire safety reasons. This justified to assemble the sensitive volume from polystyrene based solid state scintillator. Continuous operation of the experiment since October 2016 in a climate controlled room provides a very good data sample to study ageing effects of the detector.
Ageing of the plastic scintillators is already known for many years [3]. There are several mechanisms which are usually considered. An excess mechanical stress at a level above elastic and anelastic deformation results in nonrecoverable damage to the scintillator structure like shear plasticity and crazing [4], creating obstacles on the passage of the light. Humidity of the environment can penetrate into the bulk of the scintillator and produce fogging effect in it [5]. Atmosphere oxygen together with ultraviolet radiation could develop photolysis of the polystyrene resulting in its yellowing [6]. All of these effects affect transparency of the plastic and result in smaller amount of scintillation light collected. Plastic exposition in the DANSS experiment was very mild avoiding mechanical stresses, excessive humidity and ultraviolet radiation.
This work was inspired by the recent study performed by T2K experiment [7]. The T2K paper covers 10 years of the operation and reports light yield decrease in the range of 0.9 ÷ 2.2 % per year for different subsystems of the detector. Measured degradation of polystyrene scintillator strips from the MINOS experiment [8], which are similar to ours, is 2 % per year. Though ageing as large as 7 ÷ 10 % per year was observed in MINERvA [9] at 80°F (27°C).
Experimental conditions
The DANSS experiment [1] is located at the reactor unit of the Kalininskaya NPP about 350 km NW from Moscow. The detector is placed in a room just below the core of the reactor. For the sake of the electronics cooling, the room is air conditioned; thus, the temperature around is maintained all the time. The relative humidity in the room is also very low, below the lower limit of of the humidity detector used as the environment monitor. Despite of the close vicinity of the reactor, the room is clean in terms of radiation background and corresponds to a laboratory room environment.
The detector sensitive volume is built from scintillator plates (strips) which were extruded from polystyrene by the Institute of Scintillating Materials, Kharkiv, Ukraine. The strip design was developed especially for this experiment with very specific material composition and manufacture procedure [10]. The core plastic is doped with PPO (2,5-Diphenyloxazole) and POPOP (1,4-bis(5-phenyloxazol-2-yl) benzene) fluors. The surface is covered by about mm co-extruded diffuse reflection layer. The layer consists of polystyrene with admixture of rutile and of gadolinium oxide . Gadolinium has natural isotope composition and is used to capture neutrons from the inverse beta-decay after their moderation. The final gadolinium density is about , which corresponds to with respect to the whole detector body. The layout of the strip is shown in Figure 1. The strip has 1 m length, 40 mm width and 10 mm thickness. There are three grooves on the top surface of the strip with wavelength shifting (WLS) fibers Y-11(200)M by Kuraray [11]. The fibers are glued into groves with a two component siloxane based optical gel from SUREL SUREL-SL1 (St. Petersburg, Russia) [12]. The central fiber is read out by a SiPM S12825-050C by Hamamatsu. Two side fibers from each of the 50 neighbor strips are collected into 100-fiber bundles and led to a conventional PMT R7600U-300 by Hamamatsu. The opposite ends of the fibers are polished and covered by reflective paint “Silver shine” M415001K00250 by Manoukian Argon (Italy). The strips are put in rows of 25 each. There are 100 rows, which form a cubic meter of the sensitive volume of the detector. Strips in adjacent rows are perpendicular to each other. The strips are supported by copper plates, which serve as an inner layer of the passive shielding. The copper layer is 5 cm thick on average. The aim during the assembly was to have most of the weight residing on the copper frame, though there is some sag of the strips in their middle and some touching between the strips could not be excluded. The outer part of the shielding is assembled from 8 cm of borated polyethylene, 5 cm of lead and another 8 cm of borated polyethylene. The shielding significantly decreases both gamma and neutron backgrounds, making radiation doses in the strip location well below common room conditions. Three centimeter thick scintillator veto counters cover the detector outside the shielding on all sides with the exception of the bottom and serve for tagging muons. The front-end electronics are placed on two sides of the cube in a slit between copper and borated polyethylene. The electronics are cooled by a chilled water cooling system, providing a stable temperature for the central part of the detector. Unfortunately, the temperature at the SiPM location varies from for top and bottom rows to for middle rows. DANSS data acquisition system (DAQ) is based on 64-channel, 12-bit, 125 MSPS waveform digitization modules (WFD) [13], which sample analog signals from photodetectors in 512 ns windows.

Figure 1. Scintillation strip layout.
The SiPM bias voltages were set once at the very beginning of the detector operation and fixed for the rest of the running period. The setting was carried out based on a single pixel signal with the aim of about 12 ADC units per pixel. The voltages were fixed once, and no voltage adjustment for temperature changes or any other reason was ever carried out. The DAQ operated with a common trigger, fired when either an instant digital sum of all PMTs was above a 0.5 MeV threshold or a veto system was triggered. In addition to the common trigger, each SiPM channel was triggered individually once per 50,000 pulses in it. These individual triggers were used to collect the SiPM thermal noise data, out of which single pixel charge and crosstalk were extracted. Such calibration was performed for each SiPM approximately once per half hour [14]. For the neutrino analysis stream, the energy scale of all detectors is calibrated by atmospheric muons approximately every two days. The present analysis does not use the energy scale from the main analysis stream and takes SiPM data only in the form of number of photo carriers (ph.c.) per hit. This number is corrected for SiPM saturation caused by the finite number of pixels. SiPM S12825-050C has 667 pixels in its nearly square matrix. If more than one photon triggers a single pixel, a signal equivalent of only one pixel is produced. Only of SiPM pixels are effectively illuminated by the WLS fiber in DANSS strips [14]. This number was used for the saturation correction. This paper is based on the statistics collected during the period from October 10, 2016 till March 15, 2023.
Analysis
Though the detector has notable overburden, about 40 atmospheric muons pass each second through the detector [15]. Energy deposit from close to vertical muons is used as a standard candle. We start with selection of triggers with vertical muons. The event should have number of hits in the above 90. This effectively selects events with muons passing through the most of the detector. Track coordinates in the lowest and the highest 8 planes are calculated by averaging hits with a weight proportional to their energy deposit. A muon track is defined as a line connecting the points with these averaged coordinates. In order to select vertical muons only events with distances below 20 cm in the horizontal plane between these points are selected. So the tangent of the track zenith angle is below 0.2 and the difference between lengths of track in a strip is less than 2%. An example of a selected event is shown in Figure 2. A hit energy is shown by a color and the line corresponding to the track is shown also. Dark blue rectangles show low energy hits, produced by SiPM noise.

Figure 2. An example of selected muon event. The color scale corresponds to strip energy deposit. A black line shows muon track position reconstructed by its ends.
At the next stage hits are selected. Several criteria are used:
distance from muon track is less than 1.5 strip width;
there is no hit in the neighbor strips of the same layer. This is done to avoid partial track segments shared between adjacent strips;
there are hits in the strips below and above in the same projection.
In order to have more narrow distributions the value of ph.c. is corrected for longitudinal profile along the strip. We use average profiles extracted from the data as described in [14]. Only strips which were alive throughout the whole period and all the time connected to the same DAQ channel were selected. This requirement made a significant cut over available channels leaving only 1713 out of 2500 for further analysis, but it gives a valuable simplification of the analysis.
The whole data was splitted into 100-files chunks, which correspond to approximately 2 days of data taking. A distribution of number of photo carriers per hit for all hits in some data chunk is shown in Figure 3. Median value is used in the analysis as a measure of the light yield (LY).

Figure 3. An example of number of photo carriers per hit distribution.
Results
Despite of efforts to stabilize the temperature of the detector it varies at a level of C. A SiPM breakdown voltage has a strong temperature dependence [16]. As a consequence other properties like gain, crosstalk, afterpulses probability and photon detection efficiency (PDE) also have strong temperature dependence. The detector is equipped with 36 temperature sensors, placed uniformly on the two sides of the copper frame close to SiPM. Readout of the temperatures is done by an independent slow monitor system every 5 minutes and is written separately from the main DAQ data files. Graphs of the temperatures are presented in Figure 4 (top). At the bottom part of this figure single pixel signal dependence of the nearby SiPM is given. A clear anticorrelation is seen.

Figure 4. Temperature dependence as a function of calendar time (the top panel). Different colors show 3 sensors placed in the center (blue), right top corner (red) and left bottom corner (green) of the "Y" readout plane. The calendar time dependence of the single pixel signal (the bottom panel). The same colors are assigned to SiPMs close to the corresponding temperature sensors. To avoid overlap the red graph is shifted up by 5 units and the green one by 10.
The data used in this analysis is taken in the form of ph.c., so variation in the single pixel signal and crosstalk due to temperature is already taken into account, but changes in PDE are still not compensated. This compensation is done in a simple linear approximation based on the single pixel signal measured for each SiPM in each data file. In this approximation the corrected value of the signal is given by an equation:
where is a number of ph.c., upper indexes correspond to event number (), file number () and strip number (). is a single pixel signal and subscript "avg" denotes its average over all files. is a correction coefficient common for all SiPM channels. The value of corresponds to the minimum of deviation from a linear fit to LY over time dependence. The LY dependence over time is shown in Figure 5. The top panel shows LY without the temperature correction. The corrected result is shown at the bottom. It is seen that some uncompensated effects are still present. Nevertheless the LY decrease with time is clearly seen.

Figure 5. Light yield dependence over calendar time without the temperature correction (top) and with the correction according to the equation (1)(bottom). Statistical errors only.
Statistical only errors are shown in Figure 5. In order to estimate systematical errors the detector was split into 50 sections of 50 neighbor strips each (5 horizontally strips to 10 rows of the same direction). Annual LY decrease was calculated in these sections. The distribution of the results is shown in Figure 6. RMS of this distribution gives a good estimate of the systematic error, so the ageing of the whole detector is 0.55 0.05(syst.) %/year.

Figure 6. Light yield decay distribution for sections of the detector.
An observed decrease of the LY could have several reasons in addition to the degradation of the scintillation material. They include ageing of WLS fibers, optical gel used to glue fibers and light receivers1. We can not separate these factors with an exception of WLS fiber light attenuation length decrease, which can be seen as a larger ageing effect for longer distance along the strip. The LY decrease dependence over the length along the strip is shown in Figure 7. A hint of ageing effect increase with the distance to a SiPM is seen. At moderate distances from SiPMs the light attenuation is described by an exponent with the attenuation length m. The observed increase in the ageing effect with the distance could be translated into the annual decrease of the attenuation length with a formula:

Figure 7. Light yield decrease dependence over the length along the strip. The line presents the linear fit in the range 20 ÷ 100 cm.
where is the relative annual decrease of the LY and is the length from muon track to SiPM. The measured WLS attenuation length decrease, calculated from a linear fit in the range 20 ÷ 100 cm shown in Figure 7, is 0.26 ± 0.04(stat.) %/year.
Conclusions
Conclusions
A continuous measurement of the DANSS detector scintillator performance was carried out for 6.5 years. An annual LY decrease (syst.) %/year is clearly seen though it is smaller than in other experiments. T2K measurement is in the range of %/year for different detectors [7], MINOS, which was using a very similar scintillating material, observed a decrease of about 2 %/year [8]. The small decrease of LY at DANSS could be explained by the rather stable and mild environment of the DANSS detector. A contrary example is the fast %/year ageing of the MINERvA [9] scintillation detectors, which are operated at a higher temperature (). Anyway ageing of DANSS scintillation strips is small and does not require special care in the data analysis.
Acknowledgments
DANSS collaboration is grateful to the directorates of ITEP and JINR for constant support of this work. The collaboration appreciates the permanent assistance of the KNPP administration and Radiation Safety Department staff.
The detector DANSS was created with the support of the State Atomic Energy Corporation Rosatom (ROSATOM) via state contracts H.4x.44.90.13.1119 and H.4x.44.9B.16.1006. Long time operation of the experiment and initial data analysis were supported by the Russian Science Foundation grant 17-12-01145 (2017–2021). The presented analysis as well as current detector operation are supported by the Russian Science Foundation grant 23-12-00085 (https://rscf.ru/project/23-12-00085/) (2023–2025).
References
- [1]DANSS Collaboration. (I. Alekseev et al.), JINST 11 (2016) P11011
- [2]DANSS Collaboration. (I. Alekseev et al.), Phys. Lett. B 787 (2018) 56
- [3]C. F. Barnaby and J. C. Barton, J. Sci. Instruments 39 (1962) 176
- [4]R. P. Kambour, J. Polym. Sci. Macromol. Rev. 7 (1973) 1
- [5]E. D. Sword, In Proceedings of the IEEE International Symposium on Technologies for Homeland Security (HST) 2017 (2017) 1
- [6]E. Yousif and R. Haddad, Springerplus 2 (2013) 398
- [7]T2K Collaboration. (K. Abe et al.), JINST 17 (2022) P10028
- [8]MINOS Collaboration, (D. G. Michael et al.), Nucl. Instrum. Meth. Phys. Res. A 596 (2008) 190
- [9]MINERvA Collaboration (L. Aliaga et al.), Nucl. Instrum. Meth. Phys. Res. A 743 (2014) 130
- [10]P. N. Zhmurin et al., Development of Manufacturing Technology, Production and Delivering of Plastic Strip Neutron Detectors. Report on Research and Development Work. Institute of Scintillating Materials Ukraine National Academy of Science: Kharkiv, Ukraine, 2008. (In Russian)
- [11]Wavelength Shifting Fibers from Kuraray Co., Ltd., Tokyo, Japan http://kuraraypsf.jp/psf/ws.html
- [12]SUREL-SL1, http://www.surel.ru/silicone/76/ (In Russian)
- [13]I. G. Alekseev et al., Instruments Exp. Tech. 61 (2018) 349
- [14]I. G. Alekseev et al., Phys. Part. Nucl. Lett. 15 (2018) 272
- [15]DANSS Collaboration. (I. Alekseev et al.), Eur.Phys.J.C 82 (2022) 515
- [16]Stefan Gundacker and Arjan Heerin, Phys. Med. Biol. 65 (2020) 17TR01
- [17]Y. Sun and J. Maricic, JINST 11 (2015) C01078
- [18]B. Dolgoshein et al., Nucl. Instrum. and Meth. Phys. Res. A 563 (2006) 368