Absolute activity measurement of radon gas at IRA-METAS

ARTICLE IN PRESS Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 www.elsevier.com/locate/nima Absolute activity measurement of radon gas at IRA-METAS Philippe Springa,, Youcef Nedjadia, Claude Bailata, Gilles Trisconeb, Franc- ois Bochuda a Institut Universiataire de Radiophysique Appliquée, Grand Pré 1, 1007 Lausanne, Switzerland b École d’ingénieurs de Genève, rue de la Prairie 4, 1202 Geneva, Switzerland Received 12 June 2006; received in revised form 5 July 2006; accepted 25 July 2006 Available online 23 August 2006 Abstract This paper describes the system of the Swiss national metrological institute (IRA-METAS) for the absolute standardisation of radon gas. This method relies on condensing radon under vacuum conditions within a specified cold area using a cryogenerator, and detecting its alpha particles with an ion-implanted silicon detector, through a very accurately defined solid angle. The accuracy of this defined solid angle standardisation technique was corroborated by another primary measurement method involving 4pg NaI(Tl) integral counting and Monte Carlo efficiency calculations. The 222Rn standard submitted by IRA-METAS to the Système International de Référence (SIR) at the Bureau International des Poids et Mesures (BIPM) has also been found to be consistent with an analogous standard submitted by the German national metrological institute (PTB). IRA-METAS is able to deliver radon standards, with activities ranging from a few kBq to 350 kBq, in NIST-Type ampoules, and glass or steel containers usable for calibrating radon-measuring instruments. r 2006 Elsevier B.V. All rights reserved. PACS: 29.40.Wk; 29.30.Ep Keywords: Radon; Solid-angle particle counting; Primary standard; Traceability; 4pg counting 1. Introduction Radon is known to be the major component of natural irradiation of the population. Over 20 occupational epidemiological studies of radon-exposed underground miners unequivocally demonstrated that prolonged exposure to radon increases the risk of lung cancer [1]. Pooled studies of 65 000 miners found a linear relationship between radon exposure and lung cancer deaths [2,3]. This relationship was maintained even among a subgroup of miners that had lower exposures extending into the range for some homeowners [4]. These findings therefore suggest that residential radon also carries a risk of cancer. Improving radon metrology is then necessary for the traceability of secondary measurements of radon in air concentration. In spite of its important contribution to irradiation, only few intercomparisons of either radon standards or radon in air concentration have been carried Corresponding author. E-mail address: [email protected] (P. Spring). 0168-9002/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2006.07.055 out, and furthermore these have revealed a spread in the results of about 5% [5] to 40% [6]. Traceability chains for radon measurement reference systems in the national metrology institutes (NMI) are realised in two ways. In some cases, laboratories measure the gamma emitting daughters of 222Rn gas traceable to 226 Ra Standard Reference Materials (SRMs) [6–9]. Alternatively, measurements performed in some laboratories are traceable to radon primary standards in which radon itself is measured with an absolute method. To our knowledge, only two absolute measurement methods have been implemented. The first one, introduced by Piccolo [10], is based on the detection of alpha particles under an accurately measured solid angle. This system has also been implemented at the PTB [11] and at IRA-METAS. A second absolute method described in Ref. [12] involves a proportional counter of calculable 222Rn efficiency and quantifiable active volume. This paper describes the system developed at IRAMETAS, the procedures to validate it and the first primary standardisation of 222Rn submitted [13,14] to the Système ARTICLE IN PRESS P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 International de Référence (SIR) for activity comparison at the Bureau Internationnal des Poids et Mesures (BIPM). 753 2.2. Measurement chamber 2.1. Radium source 2.2.1. Alpha-detector The measurement chamber consists of a closed cylindrical volume. An ion-implanted silicon detector (ORTEC ULTRA)—with 450 mm2 active area, 100 mm minimum depletion depth and 17 keV guaranteed maximum resolution for alpha particles—is located at the top of this volume. There is a copper diaphragm right underneath this alpha detector. A 75 mm stainless 316L steel foil closes the bottom part of the chamber. In its centre, a laser-welded nickel rod (5.8 mm of diameter) is linked to the cold head of a cryogenic system (EBARA 531-120). The external part of the stainless-steel foil is kept at room temperature. This volume can be evacuated down to 10
4 hPa with a dryrunning-type pumping system Picodry (membrane and turbo-molecular system manufactured by Edwards). In addition to the detector, the spectrometry acquisition chain includes an Ortec 478 high-voltage supply operated at 50 V, an Ortec 142C preamplifier, an Ortec 472 amplifier and an IBM PC with an Ortec 8K MCA Trump card. The spectrum is acquired using Interwinner 5 (Ortec). Radon is generated by a 500 kBq commercial source of Ra (type RN-1025-500 Pylon Electronics Inc.). Radium is in a dry powder form and the emanation rate is assumed to be close to 100% and unaffected by normal variations of pressure and temperature. A special loop allows the rinsing of the source with gaseous helium to remove possible water condensation before growing the radon. The source is flushed then evacuated (to 10
2 hPa) before each operation of the system. Radon growth within the sealed 226Ra source reaches equilibrium after about 65 days. In our routine procedures radon is grown for about 10 days yielding 420 kBq of 222Rn (84% of the nominal activity), which allows measurement with a sufficient statistic for our experimental conditions and leaves enough time for quality controls before sending the standard to the end user. 2.2.2. Radon condensation and geometry The top of the nickel rod, shown in Fig. 2, is used as the condensing surface for the 222Rn. Its base is rigidly linked to the cold head of the cryogenic system. Temperature is regulated with a Lakeshore 331 temperature controller supplemented with a heating resistor circuit. Operating temperature can be adjusted from about 15–400 K. Measurements are made at low temperature, typically 40 K, in order to have a negligible saturation vapour pressure [10]. Because of the strong temperature gradient between the nickel rod and the external part of the stainless-steel foil, the surface where the radon condenses is very well defined. The diameter of the condensed radon source can be measured by autoradiography with a pinhole camera system, as described by Dersch [11]. A device with a pinhole collimator was mounted at the place of the detector. Its dimensions were calculated to give an image 2. IRA-METAS radon measurement system A 226Ra source is connected through vacuum circuitry to the measurement system. Gaseous radon diffuses into the evacuated measuring chamber where it condensates gradually onto a pre-cooled defined surface (cold-finger). The activity is measured by counting the alpha particles emitted by 222Rn. The detection efficiency only depends on a geometrical efficiency factor G ¼ O/4p which is the fraction of the total solid angle O under which alpha particles can be detected. After measurement, radon can be transferred into a liquid nitrogen-cooled container by heating the cold finger. The radon source thus obtained can be used as a calibration standard. Fig. 1 shows a general scheme of the method. 226 alpha detector radium Source collimator solid radon gaseous radon (1) (3) cold finger (2) liquid nitrogen (4) measuring assembly solid radon Fig. 1. General scheme of the measuring method: (1) radon gas is introduced into the system, (2) radon is frozen on the cold finger and measured by alpha spectrometry, (3) radon is heated and then frozen into an ampoule bathing in liquid nitrogen and (4) the ampoule is sealed while radon is still solid. ARTICLE IN PRESS 754 P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 Alpha detector Diaphragm Diaphragm axis a z e Fig. 2. The nickel rod soldered using laser technique to the 75 mm stainless 316L steel foil. b Cold finger Fig. 4. The distances involved in the calculation of the geometric efficiency G are: the diaphragm radius a ¼ 5.997(2) mm, the source radius b ¼ 3.2(2) mm, the off-axis eo0.3 mm, and the source–diaphragm distance z ¼ 100.18(9) mm. Fig. 3. Autoradiography of around 200 kBq of condensed radon. Exposure time is 14 h. The measured source radius is 3.2(2) mm. enlargement factor of 1. Fig. 3 shows the autoradiography obtained using a Kodak MIN RH film. Fig. 5. Top flange incorporating the calibrated micrometre. 2.2.3. Source–detector geometry At the operating pressure, which is below 10
1 hPa in all cases, all the alpha particles reaching the detector are assumed to be counted. The intrinsic detection efficiency is 1 for the 5.5 MeV alpha particles emitted by radon since they loose only about 0.7 keV along the 10 cm distance between the source and the detector—their stopping power is 71.3 eV/cm in air at this pressure. The efficiency then depends solely on the solid angle O. As indicated in Fig. 4, the dimensions involved in the calculation of O are the diaphragm radius (a), the source radius (b), the distance between the source and diaphragm (z) and the off-axis of the source and detector distance (e). Several algorithms are available for calculating the solid angle [15,16]. For our configuration, these differ by less than 0.05%. The first critical parameter from the uncertainty point of view is the diaphragm radius a. It was therefore measured by an accredited laboratory (Brown & Sharpe technology, Crissier, Switzerland) traceable to international standards. Its value is 5.997(2) mm. Because the diaphragm stays at room temperature during radon measurements, contraction effects can be disregarded. The source–diaphragm distance z is the next critical parameter. This length was measured with a purpose-built special device (Fig. 5) incorporating a calibrated micrometre installed at the place of the flange containing the alpha detector. A z value of 100.18(9) mm was found. ARTICLE IN PRESS P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 755 3. Primary measurements 3.1. Measurement conditions Fig. 6. Absolute variation (DZ) of the source–diaphragm distance (z) as a function of temperature. Full symbols are measurements from 2005 and open symbols are measurements from 2002 Measurements performed at different temperature and pressure conditions resulted in an increase of 0.3 mm between normal conditions (room temperature and atmospheric pressure) and typical measuring conditions (40 K, 10
3 mbar). Measurements of the source–diaphragm distance z at operating conditions are repeatable within 0.1% (standard deviation). The absolute variation of z between 300 and 40 K has been very stable over the years as shown in Fig. 6. The eccentricity e was well controlled during the set-up of the system. The nickel rod was placed in front of the diaphragm with a centring device. This ensures that e is lower than 0.3 mm. The geometry factor G ¼ O/4p of the system, assuming a uniform distribution of the activity over the cold finger, thus amounts to 8.927(18)  10
4. 2.3. Source–chamber radon transfer Radon is transferred from the 226Ra source to the measurement vacuum chamber. During the growth of 222 Rn, pressure in the radium source raises due to its imperfect sealing allowing constituents of atmospheric air to enter the source volume. During the transfer, these constituents condense with radon on the cold surface and degrade the alpha spectra. In order to remove these residual elements, especially water vapour, which is the most critical, a commercial gas clean filter (Varian Chrompack) was integrated into the vacuum circuit between the radium source and the measuring chamber. Typical measurements are realised with about 250–350 kBq of condensed 222Rn at the cold point at 40 K. Piccolo showed that at 50 K, uncondensed radon represents only a few part per million of the total activity [10]. To ensure that all alpha particle emitted in the measured solid angle are detected, measurements are made at sufficiently low pressures, typically between 10
3 and 10
2 hPa. Our measurement system maintains this vacuum level for about 6 h without pumping, which is more than enough to carry out the measurements. A typical radon measurement sequence consists in background measurements, then a transfer of 222Rn. Six or seven successive 10-min alpha spectra are acquired so as to reach a counting precision of 0.1% or less. Radon is then transferred into an appropriate container, maintained at low temperature in liquid nitrogen, by warming the nickel rod (the cold finger) up to 340 K. Once the transfer is complete, the measurement chamber is sealed again, the cold finger is cooled once more at 40 K, and the activity of possible radon residues is measured after about 12 h. 3.2. Measurements results The activity of radon is measured by detecting the 5.49 and 4.99 MeV alpha particles emitted by 222Rn. The sharp resolution of the semiconductor detector discriminates easily between these alpha emissions and those emitted by the 218 Po and 214Po offspring of radon, at 6.00 and 7.68 MeV, respectively. A typical spectrum is represented on Fig. 7. For each 10-min measurement, the counts under the two radon peaks are integrated. The lower energy limit— denoted ‘—is set to include the 4.99 MeV peak. The exact placement of ‘ is not critical because the count rate in this energy region is about that of the background. By contrast, the position of the upper energy limit—referred to as h—of the radon peak leads to a significant uncertainty over the count rate because of its partial overlap with the 218Po peak. The method used to deal with this problem is as follows. h is set midway between the 5.49 MeV peak of radon and the 6.00 MeV peak of 218Po assuming both are symmetrical. One may compute the following set of integrated count rates Rn between ‘ and 720 channels about h: Z h
20 R
20 ¼ SðEÞ dE; R
19 ¼ R20 ¼ Z Z ‘ h
19 SðEÞ dE; . . . ; ‘ hþ20 SðEÞ dE, ‘ where S(E) refers to the energy spectrum. For any given measurement, we take the arithmetic mean of the ARTICLE IN PRESS 756 P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 Fig. 7. Typical spectrum acquired in 10 min. Alpha peaks of daughters of 222Rn are easily discriminated from those of radon. The counts at low energy are attributed to high-energy beta particles of 214Bi. Table 1 Typical uncertainty budget for a primary measurement using the solid defined angle method Type A Upper limit of integration for 222 Rn peak (%) Geometry factor G (%) Background (%) 222 Rn half-life (%) Counting statistics (%) Fig. 8. Seven successive 10-min measurements performed on April 5, 2006. Error bars includes uncertainty (k ¼ 1) for counting statistics and for radon peak position (see Section 3.2). The solid line shows the average value while the dotted one marks one standard deviation. {R
20, y, R20} count rate distribution to be the actual count rate of radon. The results of a typical measurement, performed on April 5, 2006 at 10:00 UT, are shown on Fig. 8. The count rates are corrected for background, and decay during measurement. In the present example, activity was found to be 352.60(99) kBq. 3.3. Uncertainty budget Table 1 presents the uncertainty budget for the radon measurement mentioned above. The main contribution to the uncertainty comes from the geometrical efficiency factor G discussed in Section 2.2.3. Calculation of the uncertainty of this factor gives a relative standard uncertainty of 0.2%. The next main contribution is the statistical uncertainty, which depends on the activity and the measurement Total (%) Combined uncertainty (quadratic sum of type A and B) (%) Type B 0.138 0.201 0.01 0.002 0.09 0.09 0.264 0.28 duration. For a typical 1-h measurement of about 250 kBq of 222Rn, the relative standard uncertainty is about 0.1%. The propagation of the uncertainty on the upper limit of the integration window is taken to be the spread of the Rn count rate set assuming a rectangular distribution. Collected data over 20 spectra ground this assumption. The propagated uncertainty was found to be in this case 0.14%. This, of course, presumes that no uncertainty is entailed by the counting method used by the multi-channel analyser, in our case an Ortec 8 K MCA Trump card, a hypothesis we are currently probing and shall report about in a subsequent publication. Finally, the typical global standard uncertainty amounts to 0.28%. 3.4. Comparison with the 4pg NaI(Tl) integral counting method Comparing the results of this defined solid angle standardisation technique with those of another primary ARTICLE IN PRESS P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 measurement method is useful to probe its accuracy. At IRA-METAS, radon standards are therefore also measured using 4pg NaI(Tl) integral counting with Monte Carlo efficiency calculations. Once a radon condensate has been measured, it can be transferred through a vacuum circuit into a NIST-type ampoule whose bottom bathes in liquid nitrogen (see Fig. 1) and then flame-sealed. After secular equilibrium between radon and its daughters this ampoule is measured in a well-type 500  500 NaI(Tl) monocrystal. The radonloaded NIST ampoule is held at the bottom of the 43.5 cm3 volume well with a closely fitting Plexiglas tube, and integral countings above a 22.6 keV energy cut-off are repeatedly made using an adequate electronic chain. The detection efficiencies for radon and its descendents are computed numerically using Monte Carlo simulations with the GEANT code [17]. At equilibrium, from the point of view of emissions able to reach the sensitive volume of the NaI(Tl) detector, only 222Rn, 214Pb and 214Bi matter. The rest of the daughters of radon do not contribute significantly because either their branching ratios or the intensities of their radiative emissions are negligibly small. If, after the onset of secular equilibrium between radon and its daughters, N counts are registered during a measurement initiated at tini. and lasting Dt, the activity of radon at tini can be shown to be ARn ðtini Þ ¼ N 1 lRn Dt , Dt
Rn þ aPb
Pb þ aBi
Bi 1
elRn Dt where ei (i ¼ Rn, Pb, Bi) are the Monte Carlo computed total detection efficiencies for 222Rn, 214Pb and 214Bi, respectively, while aPb (aBi) is the ratio of the activity of 214 Pb (214Bi) over that of 222Rn. The fraction on the right, where lRn stands for the radioactive constant of 222Rn, accounts for the decay of radon during measurement. As a typical example, the radon condensate discussed in Section 3.2 was transferred into a NIST ampoule and measured over a 3-week period in the 4pNaI(Tl) detector. The measurements [18] displayed a robust internal coherence and yielded an activity of 353.62(40) kBq, to be contrasted to the defined solid angle value of 352.60(99) kBq at the same date of reference. The 4pg NaI(Tl) measurement is therefore about 0.3% higher but compatible with the defined solid angle primary measurement within the uncertainties, thus lending confirmation to it. This conclusion is supported by three other similar standardisation comparisons we made in recent years. Note that the uncertainty of the 4pg NaI(Tl) result above does not take into account the propagation of the uncertainties of the dimensions of the geometries or those of the decay schemes of 222Rn, 214Pb and 214Bi on the detection efficiencies. It only includes the statistical uncertainty associated with the counting, the Monte Carlo statistical uncertainty, and the propagation of the uncertainties of the half-lives of radon and its daughters on aPb and aBi at secular equilibrium. The latter were estimated by varying stochastically 10 000 times the periods 757 of radon and all its daughters within their distributions— assumed to be Gaussian—in the Bateman equations, and then taking the standard deviations of the distributions of aPb and aBi values as the relevant uncertainties. 4. Delivery of radon gas standards 4.1. Transfer to a container Condensed radon measured with the absolute method has to be transferred into a sealed container in order to be used as an activity standard. Radon is transferred by heating the cold finger onto which radon solidified while, at the opposite end of the vacuum circuit, bathing the container for delivering radon in liquid nitrogen (see Fig. 1). The transfer can be monitored by measuring intermittently the alpha spectra generated by the radon left in the chamber. Fig. 9 displays the evolution of the 5.49 MeV peak of 222 Rn during heating and transfer. During heating, the sublimation of radon can be observed on the alpha spectra since the radon peak shifts towards higher energy and narrows as self-absorption in the evanescent solid radon disk decreases. Counting under the 222Rn peak increases due to the spread of radon over the whole volume of the measurement chamber. Radon progressively condensates into the cold container. After several minutes, the radon peak begins to decrease and the pressure in the measurement chamber falls down to a value close to the pressure at the beginning of the transfer. Finally, the container is sealed and measurements are performed to check whether the transfer was total and quantify possible radon residues as described in Section 3.1. 4.2. Available geometries IRA-METAS is able to produce radon gas standards in three different geometries. The NIST type 5 ml gas ampoule is a flame sealed container. It is the SIR reference geometry for the measurement of gaseous samples. Two different two-valves-containers allowing the flushing of the gas content are also available. The first one is a classical spherical glass bulb of about 30 cm3. The second is a more robust cylindrical stainless-steel container with valves of better quality. Fig. 10 shows a picture of the different geometries. 4.3. Secondary measurements Sealed NIST-type ampoules were used both for the SIR contributions and calibrating IRA-METAS secondary standard ionisation chamber of the IG11 type. This reliable instrument [19] has been under quality control surveillance since 1983, through regular measurements of 137Cs and 226 Ra reference sources, and consistently demonstrated a strong stability. For this geometry, each prepared radon ARTICLE IN PRESS 758 P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 Fig. 9. Evolution of the alpha spectra (radon peak) during radon transfer. The increase in counting is due to an increase in the solid angle caused by the spread of the radon molecules throughout the chamber. Fig. 10. IRA-METAS deliverable geometries for radon standards. From left to right: (a) NIST-type sealed ampoule, (b) glass ampoule with flushing manifolds and (c) stainless-steel container. standard is measured after a 24-h period of latency necessary for the gamma-emitting daughters of radon to reach equilibrium—in the ionisation chamber in order to check whether the condensed radon got transferred entirely into the ampoule, by an alternative to solid angle counting. In the case of the primary measurement reported in Section 3.2, the radon was subsequently transferred into a NISTtype ampoule and then measured in the ionisation chamber; a calibration factor (Ae) was inferred. For geometrical reasons, both types of valve-ampoules cannot be measured in the IG11 ionisation chamber. Nevertheless, secondary measurements, with a precision typically better than 1%, can be performed with a well-type NaI(Tl) counter in order to ascertain the entirety of the radon transfer from the measurement chamber into the ampoule and rule out radon leakage after sealing. The container is placed on top of a NaI(Tl) scintillator in a reproducible position, and integral countings above a 121.78 keV threshold are performed in well-defined electronic conditions. The NaI(Tl) detector is periodically checked with a 152Eu source and has shown a stable efficiency over the past years. The detection efficiency is calculated for each radon standard produced. The linearity of the response for this system was tested by measuring a stainless-steel radon standard over a 23-day period covering the range of typical activities requested by customers. The observed standard deviation of the calibration factor was found to be 0.2% at most. 5. Conclusions Radon activity was measured at IRA-METAS with an absolute standardisation technique based on the condensation of radon and the detection of its alpha particles ARTICLE IN PRESS P. Spring et al. / Nuclear Instruments and Methods in Physics Research A 568 (2006) 752–759 through an accurately specified solid angle in vacuum conditions. This method yields a typical standard uncertainty of about 0.3%. The accuracy of this way of making absolute activity measurement of radon was corroborated by comparing the results of the defined solid angle standardisation technique with those of another primary method, 4pg NaI(Tl) integral counting with Monte Carlo efficiency calculations. The latter technique yields measurements that about 0.3% higher but compatible with the former. IRA-METAS was the first NMI to submit a 222Rn sample of known activity to the SIR at the BIPM. Since then, the Physikalisch Technische Bundesanstalt (PTB) also submitted a radon sample to the SIR. The degree of equivalence computed by the BIPM indicates that our activity determination and that of the PTB are consistent with each other [20]. IRA-METAS can now produce radon standards, with activities ranging from several kBq to 350 kBq, in ampoules, and glass or steel containers usable for calibrating radon-measuring instruments. Acknowledgements We are grateful to Pr. Jean-Franc- ois Valley for supporting this project and for his critical reading of the manuscript. Manuel Santos has our special thanks for designing the measurement chamber and carrying out all the mechanical work with great effectiveness. Jean-Jacques Gostely’s useful suggestions for improving the manuscript are warmly acknowledged. This work is dedicated to the memory of Marc Décombaz. 759 References [1] R. Wiliam Field, J. Duport Philippe, Med. Phys. 30 (2003) 485. [2] National Research Council, Health Effects of Exposure to Radon, BEIR VI, Committee on Health Risks of Exposure to Radon (BEIR VI), Board on Radiation Effects Research, Commission on Life Sciences; National Academy, Washington, DC, 1998. [3] J.H. Lubin, et al., J. Natl. Cancer Inst. 87 (1995) 817. [4] J.H. Lubin, et al., Radiat. Res. 147 (1997) 126. [5] J.M.R. Hutchinson, J. Cessna, R. Collé, P.A. Hodge, Appl. Radiat. Isot. 43 (1992) 175. [6] J.C.J. Dean, M. Burke, Nucl. Instr. and Meth. A 339 (1994) 264. [7] R. Collé, J.M.R. Hutchinson, M.P. Unterweger, J. Res. Natl. Inst. Stand. Technol. 95 (1990) 155. [8] P. De Felice, Xh. Myteberi, Nucl. Instr. and Meth. A 369 (1996) 445. [9] T.H. Gan, S.B. Salomon, J.R. Peggie, J. Res. Natl. Inst. Stand. Technol. 95 (1990) 171. [10] J.L. Picolo, Nucl. Instr. and Meth. A 369 (1996) 452. [11] R. Dersch, Appl. Radiat. Isot. 60 (2004) 387. [12] I. Busch, H. Greupner, U. Keyser, Nucl. Instr. and Meth. A 481 (2002) 330. [13] G Ratel, C Michotte and F O Bochud, BIPM comparison BIPM.RI(II)-K1.Rn-222 of activity measurements of the radionuclide 222Rn, Metrologia 41 (Technical Supplement 06002) (2004). [14] G. Triscone, M. Santos, J.-J. Gostely, J.-F. Valley, F. Bochud, metINFO 10 (2003) 4. [15] M.L. Curtis, J.W. Heyd, R.G. Olt, J.F. Eichelberger, Nucleus 13 (1955) 38. [16] S. Pommé, L. Johansson, G. Sibbens, B. Denecke, Nucl. Instr. and Meth. A 505 (2003) 286. [17] M. Décombaz, J.P. Laedermann, Nucl. Instr. and Meth. A 369 (1996) 375. [18] Y. Nedjadi, P. Spring, C. Bailat, F. Bochud, M. Decombaz, G. Triscone, J.-J. Gostely, J.-P. Laedermann, Primary activity measurements with 4pg NaI(Tl) counting and Monte Carlo calculated efficiencies, Appl. Radiat. Isot. A, submitted for publication. [19] J.-J. Gostely, J.-P. Laedermann, Appl. Radiat. Isot. 52 (2000) 447. [20] G. Ratel, C. Michotte, K. Kossert, H. Janben, Update of BIPM.RI(II)-K1.Rn-222 comparison of activity measurements for the radionuclide 222Rn to include the PTB, BIPM 2006.