In the frame of the project PURAZOT we have tested purification of nitrogen gas evaporated from the dewar containing 100 l of liquid nitrogen. In this way it was possible to produce about 50 m3 of gaseous N2 with Rn concentration at the level of 12 microBq/m3. Operating with the flux of 0.2 – 0.3 m3/h we can continuously provide the gas for about 1 week without dewar refilling.
     To perform the final test of our system we plan to use the g-ray (High Purity Germanium, HPGe) detectors operated at the LNGS by the Borexino collaboration: GePv [1] and GeMPI [2]. Both show in the background spectra visible lines coming from the Rn daughters. We suspect that this is the result of flushing spectrometers chambers with standard quality nitrogen taken from the gas cylinders (this gas contains Rn at the level of 1 mBq/m3). We plane to measure background of both detectors for one week while they are operated a) under standard conditions (using N2 from gas cylinders), b) under low-Rn atmosphere generated by our system. N2 flux in both cases is assumed to be the same. Comparing obtained spectra in the case a) and b) it is possible to find in a very simple way a Rn influence on the background of the spectrometer and check the performance of the low-Rn gas generator.
     Since nitrogen is produced by the separation from air contains besides Rn also other radioactive elements like Ar or Kr isotopes, e.g. 37Ar, 39Ar, 81Kr or 85Kr. Some of them (39Ar, 85Kr) can be measured via b-decay using low-level proportional counters designed for the GALLEX experiment [3]. A new system allowing such measurements has been already developed in the low-level laboratory at Max-Planck Institute. The absolute sensitivity we can achieve for e.g. 85Kr detection is 100 microBq. In principle using GNO counting system (extremely low counters background) it should be also possible to identify 37Ar and 81Kr due to emission of X-ray after K-capture by 37Ar (2.8 keV) and 81Kr (13.5 keV). However for direct counting of Ar and Kr isotopes their pre-concentration is needed. This can be done in a similar way like for Rn, using trap filled with suitable adsorber. Much easier are measurements of natural argon and krypton (and next recalculations of the results to concentrations of radioactive isotopes) with special tuned rare gas mass spectrometry. We can use VG3600 mass spectrometer (existing at Max-Planck Institute) coupled with the sample preparation and purification section. In the first one the volume of the samples can be determined (or if necessary divided into smaller fractions), in the second one gas is purified before it is introduced into the spectrometer. For purification we use Al and Ti getter pumps. The system can be calibrated with air or Ar/Kr pipettes. Measuring 1 cm3 of nitrogen gas we can detect Ar down to 1 ppb and Kr down to 0.1 ppt. It means we can see e.g. 39Ar down to 1.4 nBq/m3 N2 and e.g. 85Kr down to 100 nBq/m3 N2. If necessary the sample volume can be increased up to 10 – 15 cm3.
     In the frame of the project we propose to investigate (taking the advantage of one of the mentioned above measurement techniques) the content of radioactive isotopes of Ar and Kr in nitrogen used at the Gran Sasso laboratory.  This information can be useful for several low-level experiments installed underground. In the second step we plane to use our low-Rn nitrogen generator and check its ability to reduction of Ar and Kr concentrations in produced gas, comparing to the quality of N2 evaporated directly from the storage tanks or dewars (no purification). Since we expect that adsorption coefficient for Ar and Kr (for charcoal used as a adsorber in our trap) can be smaller than that for Rn (breakthrough can happen for smaller gas volume), such a test requires continuous monitoring of Ar and Kr content in purified gas. It can be done by taking small samples (~ 1 cm3) at different time points of purification and test them later using mass spectrometer. If we find the breakthrough moment, we can also estimate the Ar/Kr adsorption coefficient for the used adsorber. In the case of significant reduction of Ar and Kr in N2 by our system, we can try to investigate their influence on the background of low-level detectors in a similar way like for Rn.
     It would be also interesting to extend abilities of our system with regard to the maximum volume of the low-Rn nitrogen gas, which can be generated. As mentioned before 100 l of liquid nitrogen can assure the gas production for one week. This time is not always sufficient, especially during investigations of special materials expected to be very clean in terms of radiopurity or if higher nitrogen flux than 0.3 m3/h is needed. In principle the design of the trap allows dewar refilling during operations, but much safer (probability of contamination with air) would be usage of bigger LN2 volume since the beginning. A 200 l volume dewars are available at LNGS and our trap is compatible with them allowing continuous gas generation for two weeks. However we have to be sure that for doubled gas volume it will be still possible to produce high purity nitrogen. This needs to be tested and we propose such a test as a third part of the project.
    
References
[1]  C. Arpesella, A low background counting test facility at Laboratori Nazionali del Gran Sasso,
       Appl. Rad. Isot. 47 (1996) 991.
[2]  H. Neder, G. Heusser, M. Laubenstein, Appl. Rad. Isot. 53 (2000) 181.
[3]  R. Wink et al., The Miniaturized Proportional Counter HD-2(Fe)/(Si) for the GALLEX Solar Neutrino Experiment, 
       Nucl. Instr. and Meth. A 329 (1993) 541.

     According to the presented project we need access to the cryogenic facilities and infrastructure for low-level measurements, including Rn counters and gamma spectrometers.
     Liquid/gaseous nitrogen available underground will be used as a object of investigations and as a cooling medium for Rn, Ar and Kr traps working as purification/pre-concentration columns. 
     Infrastructure for low-level measurements is needed to determine the influence of Rn on the background of selected detectors and (indirectly) to determine Ar/Kr concentration in not-purified and purified nitrogen gas.

Indicative time schedule of work
Phase 1 (01.01.2004 – 01.05.2004): Determination of Ar and Kr (and their radioactive isotopes) concentration in 
                                                          liquid/gaseous nitrogen available at the Gran Sasso laboratory. Tests of Ar and Kr 
                                                          removal from N2 by using our low-Rn nitrogen gas generator.
Phase 2 (01.05.2004 – 01.08.2004): Investigations of Rn (Ar/Kr) influence on the background of selected detectors operated
                                                          underground by using our gas generator.
Phase 2 (01.08.2004 – 31.12.2004): Tests of system performance with a dewar containing 200 l of LN2. Measurements of 
                                                          Rn content in purified gas produced with different flow rate.