Cosmogenic 22Na, 7Be and terrestrial 137Cs, 40K radionuclides in ground level air samples collected weekly in Kraków (Poland) over years 2003–2006

A low background gamma spectrometer with an Etruscan, 2500 years old lead shield and a muon veto detector were applied to study 22Na and 7Be activity concentration in ground level air aerosol samples collected weekly over the years 2003–2006 in Kraków. Each sample was formed with ca 100 000 m3 of passed air, collected with two parallel ASS-500 high volume air samplers. The results for 40K and 137Cs are also presented for reference and comparison. Presented frequency distributions for activity concentration and correlation between the obtained results are discussed. The activity concentration results confirmed seasonal variation of activity to be different for all the investigated radionuclides. Moreover, the seasonal variation in nucleus activity ratio was also noticed for 22Na and 7Be. Cosmogenic radionuclides being mainly of stratospheric origin, are subsequently attached to fine aerosols, via which they are transported to the ground level air. The mean aerosol transport time within the troposphere was estimated as equal to 7.5 days on average, reaching even 50 days in warm seasons. Limitations of the applied model were identified.


Introduction
The cosmogenic radionuclides present in the Earth's atmosphere are produced mainly by secondary cosmic rays in the stratosphere or troposphere, with the relative volume of about 2/3 and 1/3, respectively [1]. The maximum production rate is noticeable at the altitude of approximately 20 km [2,3]. Spallation reactions caused by high-energy particles on oxygen, argon or nitrogen atoms are one of the important processes [4]. For example, 7 Be, one of the most active cosmogenic radionuclides, can origin from several nuclear processes such as: 14 N(p,2a) 7 Be, 16 O(p, 10 B) 7 Be, 14 N(n, 8 Li) 7 Be or 16 O(n, 10 Be) 7 Be [1,5]. The second cosmogenic nuclide studied here, 22 Na, is formed mainly in spallation of argon nuclei by a secondary cosmic ray neutron [4]. 7 Be decays (T 1/2 = 53.3 days) by electron capture to stable 7 Li emitting gamma radiation of 477.6 keV and a monoenergetic neutrino. 22 Na decays (T 1/2 = 2.602 years) by a beta plus process to 22 Ne emitting a photon of 1,274.53 keV, a positron and a neutrino. Both nuclides were also formed due to thermonuclear explosions in the atmosphere [6] conducted in the cold war era. Relatively short half-life times of the nuclides made their current activity levels no longer affected by nuclear tests. The production rate is mostly affected by the solar activity with a reverse proportional relation, i.e. the higher solar activity, the lower production rate. Numerous studies [7][8][9][10][11][12][13][14] have reported the relation between 7 Be activity and the solar cycle. However, the most detailed study was recently published as two papers [5,15]. Such a reverse proportionality to the solar activity results from additional screening of the Earth's atmosphere with the magnetic field generated by charged particles from the solar wind trapped in the magnetosphere (so called Forbush effect). The Sun's activity is basically multi-cyclic, with the main cycle of 11 years. Many parameters characterize the Sun activity; one of such is called the Wolf number and it is a normalized number of black spots on the Sun. Due to the relationship between cosmogenic radionuclides production and the Earth's magnetosphere, a dependence of production rate of such radionuclides on the geomagnetic latitude is observed. A high production rate at high latitudes drops low close to the magnetic equator. Generated nuclides are attached to fine aerosols present at high altitudes and transported along to the ground level air [16]. The effectiveness of such a transport process depends on many factors; in the temperate climate zone one of the most important ones is the temperature gradient in the troposphere, which governs the air convection. This vertical movement of air masses is much more intense in warm seasons than in the cold ones. The resulting seasonal variation of 7 Be is a well known phenomenon [17][18][19]. The typical ground air activity concentration for 7 Be corresponds to few mBq/m 3 , whereas the activity concentration for 22 Na falls below 1 lBq/m 3 . This explains why 7 Be activity concentration in air happens to be the subject to numerous reports, which it is not the case for 22 Na [20]. Contemporary detection of 22 Na faces difficulties. Still, both 7 Be and 22 Na used to be applied in modelling the transport of aerosols from the stratosphere to the ground level air [4,21,22] as an alternative to more typical usage of 7 Be and 210 Pb [23,24]. Recently, the progress on low-background gamma spectrometry achieved mostly due to application of anticoincidence active shields or digital coincidence systems, resulted in the ''renaissance'' of new studies on 22 Na [15,25,26].
Many atmospheric processes affect radionuclides concentration in air. A number of studies reported the relationship between the precipitation yield and the level of the observed activity of cosmogenic radionuclides. Rain effectively cleans the air from aerosols, hence from cosmogenic radionuclides. In general, the observed level of cosmogenic radionuclides, based mainly on 7 Be studies, depends also on the exchange of air between the stratosphere and troposphere, particularly including springtime injections of the stratospheric air into the troposphere [1,14,[27][28][29][30][31]. Apart from this depends on the vertical thermal movements velocity in the troposphere governed by temperature distribution [32], the precipitation rates [7,9,15,[33][34][35][36][37], horizontal atmospheric movements of air masses from lower to higher latitudes [1,22,32] and the atmospheric diffusion [38]. All the processes contribute to seasonal changes in 7 Be concentrations in ground level air [9,39], which sometimes ''mask'' the solar cycle dependence [12,40]. In Europe springtime increase of 7 Be concentration in air is typical. In Japan, however, it seems to be accompanied by a similar increase in autumn [41,42] or even winter [43], that have never been observed in Europe. The paper [46] provided an explanation to the peculiarity of Japanese 7 Be observation [41][42][43][44][45] as resulting from the influence of very cold air blown from Siberia in winters, which clashes with much warmer air masses coming from the Pacific Ocean.
The model proposed in [4] used both 7 Be and 22 Na to determine the mean time for the retention of aerosols in the troposphere. One of the goals of our study was to adopt this model to the European data to check its applicability to other areas. We aimed at checking weekly the regularity in 7 Be and 22 Na behaviour reported in our previous study for half-a-year samples [19], and we also wished to demonstrate our capability to determine 22 Na activities on the levels falling below 1 lBq/m 3 , as we apply the spectrometer with a cosmic muons veto detector and 2,500 years old lead in shield. Apart from cosmogenic radionuclides, the activity concentration of two terrestrial nuclides, namely 137 Cs (anthropogenic) and 40 K were measured to find similarities and differences with the cosmogenic nuclides. The study was conducted as a PhD project whose selected main findings are presented here.

Samples
Samples of ground level air aerosols were collected over the years 2003-2006 at the premises of our Institute in Kraków (Southern Poland, 50.088N, 19.928E, 240 m a.s.l.) with two high volume aerosol sampling stations produced by CLOR (Polish Central Laboratory for Radiation Protection), Warsaw, type ASS-500. The nominal airflow rate for both ASS-500 is 500 m 3 /h. [47]. They use Petryanov FPP-1.5-1.5 filters. The ASS-500 is a typical aerosol sampler used presently at 12 locations in Poland for routine air radioactivity monitoring. The monitoring network works on weekly basis, collecting aerosols from 50 000 to 80 000 m 3 of air on a single ASS-500 filter. Our two air samplers are not identical. The filter dimensions of the older model are 0.57 m 9 0.45 m, while the other square filter is 0.45 m by 0.45 m [19]. The exposed filters (with the supporting cloth removed) are pressed to 4-5 mm high pellets with a 51 mm diameter. In routine monitoring such filter pellets undergo a whole day measurement with a typical low background germanium gamma spectrometer equipped in a 10 cm lead with copper and cadmium linings, providing a minimum detectable activity for 22 Na of about 0.7 lBq/m 3 . This enables us to detect 22 Na in some single filters only during warm seasons, i.e. spring or summer. Here we applied a double filter pellet 51 mm in diameter but about 10 mm high. Each examined sample contained aerosols from about 100,000-150,000 m 3 of air.

Measurement system
An ultra low-background homemade gamma spectrometric system ( Fig. 1) was arranged for the measurements [48,49]. It comprised about a 25 % efficiency HPGe detector situated in a stainless steel and a copper U-type aluminium free cryostat, with a carbon composite endcup top window and a shield made of (from inside to outside): 1 cm electrolytic copper, 5 cm Etruscan, 2,500 years old lead, 2 mm cadmium, a standard 10 cm lead and the 10 cm outer steel containers filled with paraffin. An active shield made of a multiwire proportional chamber, served as muon veto detector. 0.44 pulses/second are recorded in the spectrum range from 40 keV to 3 MeV [49]. It was calibrated with a widely available mix multi-gamma source produced by Polatom, Ś wierk (SZN-3) in a form of a single pressed filter 51 mm in diameter, and about 5 mm high. A trick was performed to circumvent production of a dedicated calibration source. The calibration curve was calculated from the data obtained by adding two calibration source spectra, corresponding to two measurements-one in a standard geometry, and another one performed within the same time but with a plain pressed filter used as a separator. Obviously, while calibrating the spectra were added, time retained, and activity was doubled.
Each sample, i.e. a double filter exposed for a week, was measured for 7 days. The measurement was performed at least 1 month after collecting the aerosol in order to reduce unsupported activity of short lived radon ( 222 Rn, 220 Rn) progenies, leaving only in both U and Th series supported component for short lived isotopes. A sample spectrum compared to background is presented in Fig. 2. Please note that 40 K peak is not visible in the presented background, hence 22 Na peak is practically undisturbed by Compton edge from 40 K peak.
The minimum detectable activity (using the standard Currie formula, [51] ) for 22 Na was reduced from about 0.7 lBq/m 3 in a routine monitoring regime (i.e. a single filter, 1 day measurement, a standard low background spectrometer) by almost an order of magnitude, and fell to about 0.1 lBq/m 3 for our set-up (a double filter, an ultra low level spectrometer, 7 days measurements).

Quality assurance
The presented measurements were collected over a period of 4 years. A standard source, i.e. 137 Cs and 60 Co, was measured to check stability of the detector efficiency calibration over the time, randomly at first, later regularly once a week, between each of the two measurements. The calibration was also positively double-checked by comparing 40 K, 7 Be and 137 Cs results from the double filter set with the results for the same radionuclides from a routine measurement of a single filter. While conducting the measurements reported here, our laboratory participated in intercalibration runs in the field of gamma spectrometry organized by the IAEA, IRMM and Polish National Atomic Agency, PAA; the results were approved. Soon after finishing the measurements, i.e. since 2008, the laboratory was granted Polish accreditation for gamma spectrometric measurements (ISO 17025). Applied model for aerosol time retention The model for mean aerosol retention time proposed in paper [4] is described with the Eq. 1: where R a -7 Be to 22 Na activity concentration ratio observed in ground air; P-production rate in the stratosphere, index denotes the radionuclides, 7 Be or 22 Na, assumed P Be /P Na was equal to 1.1 9 10 3 ; k-decay constant, indexes as above; T t -mean time of aerosol retention in the troposphere (unknown-estimated from the model), T s -mean time of aerosol retention in the stratosphere (20 days assumed following the original paper [4]) The model concept can be explained as follows. In the stratosphere aerosols are marked with two isotopes of different half-life time, which start to act as a clock. Given their mean activity ratio on the ground level (R a ), with an assumed retention time in the stratosphere and the production rate, the retention time in the troposphere can be obtained by fitting. Recently, this model was applied for initial studies by another Polish team [24].

Activity concentration
The results for activity concentration of 7 Be and 22 Na, as well as 137 Cs and 40 Table 1. A similar scheme has been found to repeat for the following years. Surprisingly apart from such a periodic structure, a tendency of slight decrease in time is also observable, which was rather unexpected, since solar activity kept decreasing over these years: the Wolf number (or Annual Sunspot number) decreased from about 60 in 2003 to below 20 in 2006 [5].
The activity concentration of 40 K in air filters also revealed some seasonal structure. The arithmetic mean was equal to 19.7 lBq/m 3 , with the minimum at 3.0 lBq/m 3 and maximum at 40.8 lBq/m 3 . The lowest values were observed in summer. The main source of 40 K is re-suspension of soil particles [50]. This might explain why at summer time, with high vegetation, 40 K levels are low. High winter levels might come from the local emissionburning wood, biomass and coal for heating [50,52].
Activity concentration of 137 Cs in ground level air (Fig. 3) also displays a kind of cyclic structure with minima in late spring-early summer when high vegetation reduced resuspension. Maxima are observed in winters. It suggests that local resuspension cannot be the dominant effect, as Polish winters are characterised with snow covering lands and frozen soil, hence a direct aeolian erosion resulting in high resuspension is very unlikely. However, other local emissions-wood and biomass related burning seems to be responsible for such winter enhancement of 137 Cs [52] concentration in air. The highest activity concentration of this isotope was recorded in 2003, mean activity for winter was equal to 3.90 lBq/m 3 , ranging from 1.20 to 9.20 lBq/m 3 , which decreased later in spring to a mean of 2.52 lBq/m 3 with respective ranges from 0.80 to 5.60 lBq/m 3 . Summer mean was equal to 0.98 lBq/m 3 and it ranged from 0.40 to 1.90 lBq/m 3 , while autumn mean was higher again and ranged between 0.4 and 2.5 lBq/m 3 .
The activity concentrations for 22 Na have maxima in warm seasons, reaching 0.75 lBq/m 3 whereas minima close to 0.1 lBq/m 3 fall within cold seasons. The recorded values correspond with the results reported for a similar period in Switzerland [15] and Finland [26]. The activity concentration determined for other radionuclides were found also comparable to the literature data.
The frequency distributions over all seasons for 7 Be activity concentration in ground level air in Kraków is bimodal, with median or mean values close to 4,000 lBq/m 3 for spring and summer, whereas for autumn and winter to about 2,300 lBq/m 3 . Frequency distributions for 22 Na in ground level air also show at least bimodal distribution. Despite a bit unclear shape, due to relatively large uncertainties, the means are close to medians and both are more than twice higher for warm seasons (spring, summer) than for the cold ones (autumn, winter). On the contrary to both cosmogenic radionuclides the frequency distribution of activity concentration 137 Cs is asymmetric in shape, with the mean from 20 to 50 % higher than the median. Another contrasting features-with regards to cosmogenic radionuclides-are summer minima and winter maxima for activity concentration. Frequency distribution for 40 K activity concentration in ground level air seems rather polymodal.

Correlations
High correlation can be expected for activity concentration of radionuclides of similar origin: cosmogenic ( 7 Be, 22 Na) and terrestrial ( 137 Cs, 40 K). Table 2 presents the correlation between activity concentration for all four radionuclides and four seasons. The correlation between 7 Be and 22 Na (see also Fig. 4) was the highest for spring (squared Pearson factor r 2 = 0.49, significance level p \ 0.0001) and only slightly lower was for summer (r 2 = 0.43, p \ 0.0001), whereas it was much lower for autumn (r 2 = 0.30, p = 0.0022), and the lowest value was found for winter (r 2 = 0.12, p = 0.0215). The correlation plot (Fig. 4) for winter might suggest at least two correlation lines to be present: one with a slope of about 6 9 10 3 (similar to that of warm seasons), and the second one, much stepper with 2 9 10 4 . The majority of 137 Cs observed in ground level air are expected to originate from resuspension and burning of biomass (mostly in winter). However, one cannot exclude that some 137 Cs can appear in ground level air introduced from the stratosphere. The proportion between them might be governed by such factors as the volume and kind of precipitation, temperature, or the tropopause thickness [53,54]. Thus one can assume two components for 137 Cs, one dominant terrestrial, well correlated with 40 K, and a stratospheric one, correlated with cosmogenic nuclides ( 7 Be or 22 Na). Correlation between 137 Cs and 7 Be is rather weak, the highest in spring (r 2 = 0.246, p = 0.0005) and only slightly lower in summer (r 2 = 0.215, p = 0.0004). It is much lower in winter (r 2 = 0.117, p = 0.022), and no correlation at all was found during autumn (r 2 = 0.018, p = 0.516). Similarly, no correlation between 22 Na and 137 Cs was observed. It suggests that even some 137 Cs still comes from the stratosphere, it is much less intense than the one of terrestrial origin. The clearest correlation between 137 Cs and 40 K was noticed for summer (r 2 = 0.374, p \ 0.0001) and autumn (r 2 = 0.346, p = 0.001). It was much lower for spring (r 2 = 0.192, p = 0.0004), and the lowest for winter (r 2 = 0.0613, p = 0.0933). The winter samples are the most puzzling for 137 Cs, as they reveal lack of correlation between 40 K and 137 Cs, accompanied with the highest values for 137 Cs. It cannot be the cosmogenic component, as for cosmogenic radionuclides the minimum is observed in cold seasons, so the emission from burned biomass, namely wood, seems to be the most likely explanation [52]. We know from other studies [55] that in Kraków winters display the lowest activity concentration for plutonium in air, with relatively highest 238 Pu/ 239?240 Pu ratio, which might suggest sea-spray from the North Sea (or other spent nuclear fuel) to be an important source of radionuclides in winter. However, the massive sea spray should increase 40 K, what is not observed. 40 K is present in the air not only due to soil resuspension and ashed biomass (regular heating for houses), but also due to industrial air pollution rich in minerals. Since this fraction is relatively high and it increases only slightly in cold seasons due to releases from large heating centrums, it stabilises the scale of potassium seasonal variations making it poorly correlated with 137 Cs. Similarly to 137 Cs, no correlation between 40 K and both cosmogenic radionuclides ( 22 Na, 7 Be) was noticed for the ground level air in Kraków in all the examined seasons, except for a weak correlation observed in spring. However, this correlation between 40 K and cosmogenic radionuclides noticed for spring (for 22 Na: p = 0.0070 and for 7 Be: p = 0.0005) seems to be apparent-the enhanced 40 K is likely to originate from pollens or resuspension of soil due to agricultural works in the fields, whereas cosmogenic radionuclides are of stratospheric origin. 7 Be/ 22 Na activity ratio for weekly samples and the resulting aerosol retention time The observed activity ratio for 7 Be/ 22 Na in ground level air measured on weekly basis shows a cyclic, seasonal structure with warm periods (spring or summer) maxima and minima in winter or autumn (Fig. 5). It confirms our previous findings [19] where half-year sets of samples were studied and a seasonal variation of 22 Na to 7 Be was suggested. Such a variation of cosmogenic isotopes activity ratio was observed and discussed in recent papers from Finland (26) and Switzerland (15). The Finnish paper explains such features by a relationship with global meteorological parameters such as North Atlantic Oscillation, whereas the latter one provides explanation based on the introduced mixing cycle starting with spring injection of cosmogenic nuclides from the stratosphere, summer intense downward mixing within upper troposphere, lowering of mixing height in autumn, and finally almost no mixing in winter.
The results presented here allow to adopt a model for aerosol retention developed by Tokuyama and Igarashi [4]. In the original paper the model was applied as follows: the ratio of activity concentration in ground level air 7 Be/ 22 Na (equal to R a in Eq. 1) was used to estimate the retention time for aerosols in the troposphere, T t (Eq. 1). Since this ratio is season dependable, it should be calculated for each season separately. Frequency distribution for the 7 Be/ 22 Na activity ratio does not take a Gaussian shape, it is rather bimodal or polymodal (see Fig. 6). The data for summer (with high 22 Na values, so with relatively smaller uncertainties) display the narrowest distribution. Likely, some of the width observed for the frequency distributions comes from the relatively high uncertainties in 22 Na activity concentration. The mean value for all our results (n = 173) is equal to R a = (15.9 ± 5.1) 9 10 3 , whereas seasonal mean values are shown in Table 3 together with similar results for Japan in 1998. In both countries they were lower for spring and summer (close to 1 9 10 4 ) and higher (close to 2 9 10 4) for autumn or winter, however the seasonal difference is bigger for Poland which likely results from the lack of autumn maximum of cosmogenic radionuclides reported for Japan (see ''Introduction''). Taking into consideration the mean or median value for R a and assuming retention time in the stratosphere T s equal to 20 days [4] the resulting mean retention time T t is equal to about 7.5 days for the entire data set, whereas for warm seasons (when R a is close to 1 9 10 4 ), with the same T s the value of T t assumed, it grows to about 50 days (Fig. 7). The cold season with R a close to 2 9 10 4 seems to be well described with this model. T s massively longer then 20 days assumed in the original model [4], namely ranging from 102 to 205 days, was obtained recently within the same model in another study [24]. In the presented paper, the T t is assumed as known, based on 210 Po/ 210 Pb system, and the model was used to determine T s . With longer T s applied, much shorter T t is obtained, far below 10 days as it can be seen from Fig. 7. It does not seem to be very realistic, revealing rather weakness of the applied model.

Conclusions
Studies on air activity of cosmogenic 7 Be measured together with 22 Na and other radionuclides seem to provide a solidly reliable system for studying the atmospheric processes. The applied detection system and sample collection method allowed to measure in weekly collected samples not only 7 Be, 137 Cs and 40 K but also much less active,  Poland (15.9 ± 5.1) 9 10 3 11.2 9 10 3 11.1 9 10 3 21.3 9 10 3 18.9 9 10 3 Japan [4] (10.0 ± 1.2) 9 10 3 (8.4 ± 1.5) 9 10 3 (8.0 ± 1.2) 9 10 3 (16 ± 4.2) 9 10 3 hence much more difficult to measure, 22 Na. The determined activities are comparable with those reported by other investigators and show seasonal variation. The increase of cosmogenic radionuclides concentration in air in spring and summer seems to be caused by an injection of stratospheric air followed by the intense mixing of the troposphere. The activity ratio for 7 Be and 22 Na allowed us to estimate the retention time for tropospheric aerosols, which appeared to be season dependent: it is longer for warmer seasons (spring, summer), and shorter for colder (autumn, winter) ones. During warmer seasons the temperature gradient between the tropopause and ground level air is much higher, which results in much more intense-in comparison to colder seasons-convectional vertical movements of air masses. This apparently enhances time of aerosol retention in air during warm seasons by nearly an order of magnitude, from less than 10 days as average for all seasons to about 50 days in warm seasons.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.  Fig. 7 Aactivity ratio 7 Be/ 22 Na (R a ) in ground level air as a function of retention time for aerosols in the stratosphere (T s ). The curves present results of the model given by Eq. 1 [4]. The retention time in the troposphere T t can be read out from the picture provided R a is known and T s assumed