E. Aprile et al (XENON Collaboration), Lowering the radioactivity of the XENON1T photosensors, arXiv:1503.07698, Eur. Phys. J. C75 (2015) 11, 546.
The XENON1T experiment employs 242 photomultiplier tubes (PMTs) in the time projection chamber, arranged into two circular arrays. Because the overall background goal of the detector is incredibly low, with less than 1 expected event in a tonne of liquid xenon and one full year of data, the PMTs must be made out of ultra-pure materials. These materials were selected for their content in traces of 238-U, 232-Th, 40-K, 60-Co, 137-Cs and other long-lived radionuclides.
The XENON collaboration joined efforts with Hamamatsu to produce a photosensor that meets the strict requirements of our experiment. The sensor is a 3-inch diameter tube that operates stably at -100 C and at a pressure of 2 atmospheres. It has a high quantum efficiency, with a mean around 35%, for the xenon scintillation light at 178 nm and 90% photon collection efficiency.
The sensor, shown schematically in the left picture, features a VUV-transparent quartz window, with a low-temperature bi-alkali photocathode deposited on it. A 12-dynode electron multiplication system ensures a signal amplification of ~3 millions, which is a crucial feature to detect the tiny signals induced by the rare collisions of dark matter particles with xenon nuclei.
Before the tubes were ready to be manufactured, the construction materials were inspected with gamma-ray spectroscopy and glow-discharge mass spectroscopy (GDMS). For the former, we employed the world’s most sensitive high-purity germanium detectors, GeMPI and Gator, operated deep underground at the Gran Sasso Laboratory. GDMS can detect trace impurities in solid samples and the results were compatible with those from germanium screening. We measured many samples to select the final materials for the PMT production. As an example, specific 226-Ra activities around or below 0.3 mBq/PMT were seen in most of the inspected materials. Such an activity corresponds to 3 x 10-4 226-Ra decays per second and tube, or about 26 decays per day.
The relative contribution of the selected materials to the trace contaminations in U, Th, K, Co and Cs of the final product, seen in the left picture, also tells us how to improve further sensor versions for the XENONnT upgrade. Most of the nuclides in the 238-U and 232-Th chains, especially dangerous for their emission of alpha particles, that can the produce fast neutrons in (alpha,n) reactions, are located in the ceramic stem of the tube. In consequence, finding a new material to replace the ceramic might drastically improve the background expectations.
Once the final production started, and the tubes were delivered in several batches to our collaboration, they were measured in the Gator detector. Its inner chamber can accommodate 15 PMTs at a time, as seen in the left picture. Each batch was screened for about 15 days, and theobserved activities were mostly consistent from batch to batch. For all measured PMTs, we obtain contaminations in uranium and thorium below 1 mBq/PMT. While 60-Co was at the level of 0.8 mBq/PMT, 40-K dominates the gamma activity with about 13 mBq/PMT. The information from screening was considered in the final arrangement of the PMTs in the XENON1T arrays. PMTs with somewhat higher activities are placed in the outer rings, where they are more distant from the central, fiducial xenon region of the detector.
The average activities per PMT of all trace isotopes served as input contaminations to a full Monte Carlo simulation of the expected backgrounds in XENON1T. The results show that the PMTs will provide about 1% and 6% of the total electronic and recoil background of the experiment, respectively. We can therefore safely conclude that the overall radioactivity of the sensors is sufficiently low, and they will certainly not limit the dark matter sensitivity of the XENON1T experiment.