Characterization and long-term performance of the Radon Trapping Facility operating at the Modane Underground Laboratory

The investigation of very rare processes in particle or nuclear physics, such as neutrinoless double beta decay (0νββ) [1–3] or direct dark matter detection [4], requires very low radioactive background conditions. The background radiation may have different origins: cosmic radiation, radioactivity of the environment, radioactive impurities in the detector, materials and shielding. One of the ultimate sources of the radioactive background is radon (Rn), a noble gas naturally present in the air and formed, in case of the long-lived isotope (²²²Rn), in the ²³⁸U radioactive decay chain through the radium (²²⁶Ra) parent nucleus. To reduce this source of background radiation, a common solution in some experiments consists of producing and flushing a radon-free carrier gas inside or around the detector. A dedicated facility has been built at the Modane Underground Laboratory (French: Laboratoire Souterrain de Modane—LSM) in order to produce radon-free air primarily flushed around the NEMO-3 experiment looking for the 0νββ process [5] but also used nowadays for other ultra-low background experiments running in the laboratory.

The LSM air is renewed twice per hour with fresh air from outside delivered through the tunnel ventilation system. The radon exhalation rate is extremely variable and correlated to various parameters like day/night effect, weather and conditions of the tunnel ventilation. In 2004, it was observed that this rate had a huge impact on the level of the radon background inside the NEMO-3 detector. In order to reduce the radon activity in the detector volume, the airtight tent with radon-free air delivered from the Radon Trapping Facility (RTF) was built. This method allowed the reduction of radon concentration in the incoming air and thus inside the NEMO-3 detector [6–8].

The air purification system has been initially developed by the Super-Kamiokande collaboration [9] as well as the high sensitive radon detectors able to measure the radon activity down to few mBqm⁻³ [10]. Based on the Super-Kamiokande system and by using of the activated charcoal, the RTF was installed at the LSM in September 2004. Online radon monitoring performed at the exit of the RTF using a high sensitive electrostatic radon detector has demonstrated a reduction of the radon activity in the LSM air by three orders of magnitude, from about 20 Bq m⁻³ down to 10 mBq m⁻³ [11]. After more than nine years of running, the RTF was stopped and disassembled for maintenance operations. It was thus a unique opportunity to analyze the K48 activated charcoal filled in the RTF by several techniques. The 2.6 m high charcoal column has been divided into several layers in order to study both the dynamic adsorption coefficient (K-factor) along the column with a possible ageing effect and the radon spatial trapping profile by measuring a long-living isotope ²¹⁰Pb (half-life = 22.3 years) by the low background gamma spectrometry.

In section 2, we shall recall the main ingredients of the radon adsorption processes that motivated the design of the RTF and the use of several techniques to analyse the activated charcoal. Details on the RTF itself will be given in section 3. Section 4 is describing the protocol adopted to divide the charcoal column in several layers and to prepare the samples using different containers. The description of setups for the K-factor, the gamma spectrometry measurements and the associated results will be given in sections 5 and 6, respectively. The results will be discussed in section 7 before the conclusions.

Radon is a radioactive, chemically inert noble gas. Thus, the only way to trap it is by a physical adsorption called physisorption on a given material through the van der Waals forces. Radon atoms are not definitely trapped but slowed in their path due to adsorption/desorption processes. The goal is to slow down radon atoms in such a way that they will decay before exiting the material. Radon daughters are not chemically inert so they are definitely trapped in the adsorbent material. When flushing a carrier gas mixed with the radon through a given material, the trapping lifetime of the radon is dependent on several parameters: the temperature, the porosity of the solid material, the size of the pores, the partial pressure and the competitive adsorption of the radon and the carrier gas.

The size of the pores must be at least of the order of the size of the radon atoms. However, the bigger the size of the pores, the shorter the trapping lifetime will be. The temperature has also a huge impact on the trapping lifetime. The lower the temperature, the longer the trapping lifetime. Therefore, it is usual to work at temperatures well below the room temperature. The radon adsorption in a given material is also dependent on the concentration of the pores or the high degree of microporosity. For active charcoal adsorbent, it is common to have a surface area in excess of 1000 m² per gram. The size and the nature of the molecules of the carrier gas is also very important because it can be in competition with the radon adsorption by the pores.

The radon adsorption can be estimated from the ratio of radon concentration in an adsorbent to the radon concentration in the carrier gas going to the adsorbent. This ratio, called the K-factor expressed in m³/kg, is shown in equation (1):

$$\begin{eqnarray}&&K=\displaystyle \frac{{C}_{a}}{{C}_{g}},\end{eqnarray} \tag{ 1 }$$

where C_a is the radon concentration in the adsorbent (in Bq/kg) and C_g is the radon concentration in the gas (in Bq m⁻³). The K-factor is evaluated when the equilibrium between adsorption and desorption in the sample is reached.

In chromatography, an important parameter is the mean retention time τ_R. It represents the time needed by an atom to cross a trapping column when the gas is flushed at a certain flow [12]. The mean retention time in the adsorbent material is then the result of a competition between the trapping time of the atom in a pore and the transportation time between two trapping processes. It is known that this parameter is related to the gas flow Φ (in m³/h), to the mass of an adsorbent M (in kg) and to the K-factor (in m³/kg) by the following equation (2):

$$\begin{eqnarray}&&{\tau }_{R}=\displaystyle \frac{K\,M}{{\rm{\Phi }}}.\end{eqnarray} \tag{ 2 }$$

In a chromatographic column with a length L, one can also define the mean velocity v of a radon atom in the adsorbent material by v = L/τ_R. Radon atoms are radioactive with a mean lifetime τ. In this case, if the mean retention time τ_R of a chromatographic column is much shorter than the radon lifetime, then most of the radon atoms will exit the material. In contrasrt, if τ_R is of the order of magnitude or greater than the radon lifetime, most of the radon atoms will decay in the material. Therefore, the capture of the radon thus depends both on the mean velocity and on the size of the column.

The time for a radon atom to reach a given distance x in the sorbent material during a time interval t is then defined by t = x/v. Considering the radioactive decay law, one can translate the time relation in a spatial relation in the following equation (3):

$$\begin{eqnarray}&&N(x)={N}_{0}\,{e}^{-\tfrac{x}{X}},\end{eqnarray} \tag{ 3 }$$

where N(x) and N₀ are the numbers of the radon atoms present at the depths x and x = 0, respectively, and $X=\tfrac{{\rm{\Phi }}\,\tau L}{K\,M}$ is the mean free path of the radon atom before decaying.

In this paper, we consider only the ²²²Rn isotope regarding its long lifetime (τ = 5.5 days). After decaying, the daughter nucleus ²¹⁸Po is no more a noble gas and is stopped where the radon has decayed. Thus, one expects a spatial exponential profile of the radon daughter nuclei in the material sorbent along the x direction of the gas flow. To study a posteriori the radon trapping spatial profile, it is suitable to measure its daughter nuclei, especially the ²¹⁰Pb nucleus with a long half-life. This nucleus is a waiting-point that accumulates all the radon atoms that have decayed at a given location in the sorbent material. So, the ²¹⁰Pb activity will be directly correlated to the radon accumulation and is supposed to follow the same kind of equation as equation (3):

$$\begin{eqnarray}&&{A}_{{Pb}}(x)={A}_{{Pb}}(0)\,{e}^{-\tfrac{x}{X}},\end{eqnarray} \tag{ 4 }$$

where A_Pb(x) and A_Pb(0) are the ²¹⁰Pb activities at the depths x and x = 0, respectively. The ²¹⁰Pb activity is an indirect access to the radon trapping spatial profile and is measured by the gamma spectrometry using the 46 keV gamma line.

This is the reason why we proposed to sample the charcoal column layer by layer and to perform the K-factor (see the section 5) and the low background gamma spectrometry measurements (see the section 6).

Based on the proposal of the Super-Kamiokande collaboration, the RTF was designed by the Czech team of the NEMO-3 collaboration and built by the Czech company ATEKO a.s. Before the installation at the LSM it was tested at the IEAP CTU in Prague. The principle of the RTF operation is based on the radon trapping from cooled compressed air by the physical adsorption method applied on an activated charcoal. The air is cooled down to the −55 °C before entering in the charcoal column. The advantage of this commonly used method is a huge effective surface area of the porous charcoal. The RTF technical specifications are as follows: a compressor (7 bar), a filtration with an oil separator (0.03 μm) and a dust separator (0.1 μm), an air dryer with a dew point at −70 °C for 8.5 bar, a cooling unit and an adsorption column with an internal diameter of 0.6 m and with a height of 2.6 m loaded with 425 kg of charcoal (activated charcoal K48 made of the coconut shells from the Silcarbon Aktivkohle¹¹ ). A photo and a schematic drawing of the RTF are shown in the figure 1. The radon-free air is flushed through the all LSM experiments, i.e. NEMO-3 (dismantled in 2012) and EDELWEISS-III [13] experiments and the HPGe platform with a typical flow of 120 m³/h. The input and output air have a temperature of approximately 25 °C and 20 °C, respectively.

Zoom In Zoom Out Reset image size

The charcoal sampling (see the figure 2) was realized from the top inlet of the column with the help of a vacuum cleaner. The first 5 cm of the column were free of sorbent. Then the first charcoal layer with a non-homogeneous thickness between 5 cm and 13 cm located on the column sides was sucked out in order to start the sampling at a well-defined depth along the radius. This non-homogeneity was caused by removing the column plug. Thus, the 13 cm depth has been defined as the 'zero' mean layer depth (MLD) in the following analysis. Further, the sampling of layers with the thickness from 5 cm to 10 cm was performed on the whole depth of the column. Each layer is then defined by its own MLD. The uncertainty of the MLD has been evaluated to be 1.5 cm.

Zoom In Zoom Out Reset image size

Several different containers have been used for charcoal sampling. Due to self-absorption, large volume containers have been used in order to measure the high energy gamma lines from radionuclides ⁴⁰K and ¹³⁷Cs with good efficiency and also to have a more representative mass sample. Smaller containers have been also used in order to measure the low energy gammas from ²¹⁰Pb and ²²⁶Ra (see the section 6). Typical information about the type of containers and the mass of each sample are presented in table 1.

4.1. Marinelli beakers

Two types of Marinelli beakers with a volume of 4 l and 0.6 l were used. Before filling the Marinelli beaker with the required amount of the charcoal, it was placed in a plastic bag and mixed in order to homogenize it (see figure 3(a)). Every Marinelli beaker was hermetically sealed with silicon glue and additionally with an adhesive tape (see the figure 3(b)). The number of collected samples, their mass and the depth range are given in table 1. These samples have been measured using a high efficiency coaxial-type HPGe detector located at the National Radiation Protection Institute (NRPI) in Prague (Czech Republic), suitable for the detection of the high energy gamma lines coming from ⁴⁰K and ¹³⁷Cs.

Zoom In Zoom Out Reset image size

4.2. Petri boxes and plastic tubes

4.3. Containers for the K-factor measurements

The K-factor measurements as a function of the depth were performed at the Center for Particle Physics in Marseille (CPPM). The used experimental setup is shown in figure 4. The carrier gas (nitrogen) was mixed with a radon source consisting of a metal plate coated with a thin ²²⁶Ra layer maintained at a fixed temperature. The gas mixture with well-defined radon concentration of C_a = (900 ± 24) Bq m⁻³ is introduced in a copper cartridge filled with 2 g of the active charcoal located in a freezer at −30 °C (or −50 °C). All measurements have been carried out with a flow of (10.0 ± 0.2) l/h, at the pressure of 1 bar.

Zoom In Zoom Out Reset image size

In order to establish when the adsorption/desorption equilibrium in the adsorbent is reached, the gas from the trap is continuously monitored with a commercial RAD7 radon detector¹² . When the breakthrough curve in the RAD7 reaches a constant value, the trap is isolated and disassembled from the setup. Then, the radon activity of the charcoal C_a is measured with the HPGe detector using the main gamma lines of ²¹⁴Pb and ²¹⁴Bi. The K-factor is then deduced from the ratio of the radon activity in the gas C_g and in the adsorbent C_a from the equation (1).

Two sets of measurements were performed. In the first one, the radon adsorption of the RTF charcoal without regeneration as a function of the MLD was measured. In the second set of measurements, the charcoal samples were heated during 12 h up to 200 °C in the vacuum (50 mbar). This latter process called regeneration was done in order to remove water traces or any volatile vapours or gases accumulated from the charcoal sample aiming to possibly retrieve the initial K-factor value.

Figure 5 shows the K-factor values measured at −30 °C as a function of the MLD for the two configurations. As one can see, except for the first layer, all the K-factors measured with or without regeneration remain constant within error bars. The adsorption capacity of the first layer, which corresponds to the first 2 cm, is however around 3 times lower than for the rest of the column. It is interesting to mention that these results are in a good agreement with those obtained with the new samples of the K48 charcoal from the same company and measured with the same setup (K = 77 ± 7 m³/kg) [16]. For the first layer, the K-factor increases by a factor of two after the regeneration. However, it still remains lower than the K-factor values from other layers. The lower adsorption capacity of the first layer could be understood as an aging effect due to the presence of some volatile organic compounds (VOC) from the laboratory and tunnel air. In order to verify this assumption, a VOC analysis using a gas chromatography-mass spectrometry (GC-MS) analytical method at the Institute of Rock Structure and Mechanics of the Czech Academy of Sciences (IRSM) was carried out. In this study, a statistically significant content increase of the organic compounds in upper layers was determined. The K-factor measurements proved that most of the adsorption capacity of the RTF (about 95%) remains constant after nine years of running, which means after filtration of 12 million m³ of dried air.

Zoom In Zoom Out Reset image size

Additional measurements have been performed in order to measure the K-factor around the operating temperature of the RTF, i.e. −50 °C. In particular, measurements for layers 6 and 9 using pure nitrogen have been performed. The results are presented in table 2. One can observe that these measured values are consistent within their uncertainties and greater compared to the values at −30 °C as expected from the Arrhenius law. These results are in a good agreement with those using the fresh K48 charcoal (see the table 2) obtained by Noel et al [16]. This is additional proof that there is no ageing effect in the RTF except the first layer.

The RTF was equipped with three temperature sensors able to measure the temperature gradient along the column. Temperatures of the charcoal at the entrance (top of the column) and exit (bottom of the column) were measured to be −54 °C and −52 °C, respectively, with a temperature gradient of only 2 °C. From the measurements with the different temperatures, it was possible to extrapolate the K-factor of (280 ± 18) m³/kg at (−53 ± 1) °C. Considering a total charcoal mass of (425 ± 5) kg and a gas flow of (120 ± 10) m³/h, a mean retention time τ_R of (41.3 ± 6.6) days of the RTF has been derived. This value is 7.5 times greater than the mean lifetime τ of the radon. Thus, a suppression factor of ${1830}_{-1220}^{+4970}$ has been estimated. Large error bars are coming from the impact of the exponential function on the relative uncertainty of the relevant parameters and will not allow us to take into account this suppression factor in the discussion (see the section 7).

Analysis of all charcoal samples was done by the use of the different types of the HPGe detectors. Activities of radionuclides in the low energy region (46 keV for ²¹⁰Pb) were measured from Petri boxes and plastic tubes samples using the Broad Energy Germanium (BEGe) and well-type HPGe detectors, located at the LSM and CENBG laboratories, respectively. For activities of radionuclides with the higher gamma energies, we used the Marinelli beakers measured with a coaxial HPGe detector with higher relative efficiency, located at the NRPI institute. All these detectors were calibrated using several standard etalons: a radionuclide dissolved in a silicone gel with the density of 0.98 g cm⁻³ placed in a plastic Marinelli beaker from the Czech Metrology Institute and several IAEA sediments (IAEA-314, IAEA-385 and IAEA-447) measured in different geometries for Petri boxes and plastic tubes. The activities were determined by taking into account the effective density of each charcoal sample. A systematic uncertainty of about 7% on the photopeak efficiency has been estimated for all HPGe detectors.

Several radionuclides have been detected in the charcoal. An example is illustrated in figure 6. It is possible to observe gamma lines coming from ²¹⁰Pb (46 keV), ¹³⁷Cs (662 keV) and ²¹⁴Pb (352 keV), the daughter nucleus of ²²⁶Ra.

Zoom In Zoom Out Reset image size

In the following subsections each of the detected isotopes will be discussed.

6.1. Activity of ²¹⁰Pb

The results of the mass activities given in Bq/kg are shown in table 3 with 1σ statistical uncertainties. Here, the two sets of data correspond to the two types of small containers used. Further, the activities have been corrected to correspond to the same date, i.e 1st of January 2014. The Petri box measurements exhibit nonzero ²¹⁰Pb activities up to the last layer located at the MLD of 254.5 cm. This is due to the fact that they were performed with an ultra-low background BEGe planar detector located at the LSM and exhibiting a very low intrinsic contamination in ²¹⁰Pb [14].

Table 3.  An overview of measured ²¹⁰Pb activities from different sampling with respect to the MLD. Activities have been normalized to the date, i.e. 1st of January 2014.

	Petri box/Planar Detector		Plastic tubes/Well-type-detector
MLD [cm]	A [Bq/kg]	σ [Bq/kg]	A [Bq/kg]	σ [Bq/kg]
2.0	—	—	485	10
12.0	550	29	567	8
22.0	371	20	395	7
32.0	302	18	303	6
42.0	247	14	239	5
54.5	—	—	171	8
59.5	—	—	122	8
64.5	99.3	6.2	100.5	6.7
74.5	—	—	63.1	2.5
84.5	46.2	2.3	42.0	3.8
94.5	—	—	28.7	2.0
104.5	25.9	1.7	24.6	2.6
114.5	—	—	22.1	2.1
124.5	11.6	0.6	9.6	1.3
134.5	—	—	8.5	1.2
144.5	—	—	8.4	1.2
154.5	7.55	0.51	6.4	2.1
164.5	—	—	3.6	0.3
174.5	—	—	3.8	0.8
184.5	3.68	0.34	—	—
194.5	—	—	2.1	0.6
214.5	1.55	0.33	—	—
224.5	2.28	0.34	—	—
234.5	1.30	0.24	—	—
254.5	0.91	0.18	—	—

From these measurements, we observe a clear decrease of the activity as a function of the depth. One noticeable exception concerns the plastic tube measurements for which the activity of the first layer at 2 cm is below the value at 12 cm. This can be explained by the fact that the input air was flushed inside the column through the plug (see figure 2(b)) on its lateral sides. Thus, in the first layer at the top of the column, the charcoal was more exposed to the radon at the edge compared with the center, where the sample has been taken. After 10 cm, this is more homogeneous due to the diffusion of the radon along the radius. This lower value may also be explained by the fact that the K-factor is much lower in the first layer (see the section 5). So, the radon may have been less trapped in the first layer and then less ²¹⁰Pb is observed. Anyway, the fit of the activity curve with the MLD has been applied to the values starting from the MLD of 12 cm.

Figure 7 shows the measured values of the activity as a function of the MLD, with an exponential fit for the combined data of Petri boxes and plastic tubes. The activity was fitted by an exponential function plus a constant background as follows:

$$\begin{eqnarray}&&A(x)={A}_{0}\,{e}^{-\tfrac{x}{X}}+{A}_{{res}},\end{eqnarray} \tag{ 5 }$$

where A is the ²¹⁰Pb mass activity (in Bq/kg) at the depth x, A₀ is the initial activity at x = 0, X is the mean free path of the radon atoms (see section 2 and equation (3)) and A_res is the intrinsic ²¹⁰Pb activity of the charcoal. Table 4 summarizes the coefficients A₀, X and A_res derived from the two data sets and their combination from the global fit shown in figure 7. We conclude that the coefficients are consistent from the two sets of measurements within their uncertainties.

Zoom In Zoom Out Reset image size

Table 4.  Summary of the parameters derived from the fit of the two separated datasets and the combined datasets.

Parameter	Unit	Petri box sampling	Plastic tubes sampling	Combined
Activity at depth 0 A0	Bq/kg	824 ± 40	948 ± 43	908 ± 27
Mean free path X	cm	30.1 ± 0.6	28.1 ± 0.7	28.8 ± 0.4
Intrinsic activity Ares	Bq/kg	1.07 ± 0.14	1.07 ± 0.39	1.15 ± 0.13

The initial ²¹⁰Pb activity of about 900 Bq/kg observed at the top of the charcoal column seems to be consistent with a simple estimation. Assuming a constant airflow of 120 m³/h, an input activity of (20 ± 5) Bq m⁻³ and the decay of the ²¹⁰Pb during the operation, one can derive a mass activity of 1000 Bq/kg in the first 20 cm layer which is fully consistent with the measured activity taking into account uncertainties and the fact that the ²¹⁰Pb activity has also decreased during nine years of running.

The radon mean free path X is equal to (28.9 ± 0.4) cm and its mean retention time τ_R in the RTF is (47.6 ± 1.2) days. Considering the total height of the column of (255 ± 5) cm, the reduction factor of the RTF from the gamma spectrometry measurements is ${6790}_{-1720}^{+2370}$. The comparison of the radon reduction factor obtained via different methods will be discussed in section 7. Obviously the height of the column has been well designed in order to decrease the radon activity in the input air by at least three orders of magnitude.

Finally, the intrinsic ²¹⁰Pb activity observed at the level of 1.1 Bq/kg is probably coming from a natural intrinsic activity of the charcoal production itself, as it will be discussed in more detail in the next subsection about the ²²⁶Ra activity.

6.2. Activities of ²²⁶Ra and its daughters

Figure 8 shows the results of the mass activity of ²²⁶Ra measured with the K48 charcoal samples placed in plastic tubes. The ²²⁶Ra activity has been deduced using the gamma rays from the ²¹⁴Pb and ²¹⁴Bi radionuclides (see figure 6) assuming a secular equilibrium with ²²⁶Ra itself. The activities ranged from 0.3 Bq/kg to 1.2 Bq/kg, with a mean value of (0.42 ± 0.03) Bq/kg. This mean value is fully consistent with an activity of 0.46 Bq/kg measured at the end of 2003, which means before the beginning of the RTF operation [17].

Zoom In Zoom Out Reset image size

The ²²⁶Ra activity is obviously coming from the natural intrinsic activity of the coconut shells. It is a long-living radionuclide in the ²³⁸U chain and also a direct progenitor of ²²²Rn and ²¹⁰Pb. Assuming a secular equilibrium of ²¹⁰Pb with ²²⁶Ra, we may expect a ²¹⁰Pb intrinsic activity of about 0.42 Bq/kg in the charcoal. This value is below the measured intrinsic activity of 1.1 Bq/kg obtained from the fit shown in figure 7. This excess of ²¹⁰Pb of about 0.7 Bq/kg may be explained by the exposition of the coconut shells to the radon deposition in the air during its lifetime. Anyway, this is not an issue in our study because it contributes as a constant background almost negligible along the height of the charcoal column that does not affect the measurement of the radon mean free path presented in the previous subsection.

Nevertheless, the intrinsic ²²⁶Ra activity of the charcoal may be an issue for residual self-emanation of the radon. Indeed, a radon self-emanation may occur from the charcoal itself due to the ²²⁶Ra content. The radon self-emanation of the charcoal have been recently pointed out but it has been measured at higher temperatures, i.e. from 20 °C to 140 °C [12]. In the case of the RTF, running at −55 °C, we believe that the radon self-emanation from the K48 charcoal is highly suppressed. In addition, due to the high trapping efficiency of the RTF, the radon self-emanation may only come from the last layers of the column, which dramatically reduce its effect. A rough estimation gives a hypothetical contribution below 1 mBq m⁻³ in radon for a ²²⁶Ra activity of 0.42 Bq/kg in the K48 activated charcoal. Thus, it seems not to be an issue for our purpose but might be a problem for very challenging low radon activity well below 1 mBq m⁻³.

For the following radionuclides measured with higher gamma energies, we will only present the data from Marinelli beakers, i.e. with a higher mass of the charcoal and with a more homogeneous samples.

6.3. Activities of ⁴⁰K and ¹³⁷Cs

In this section, we will discuss and compare the main parameters of the RTF obtained from three complementary methods: the K-factor measurements, the gamma spectrometry measurements of ²¹⁰Pb and the online radon measurements performed at the exit of the RTF during its operation.

The results of the measurements of the K-factor for the K48 activated charcoal have demonstrated a stability of its adsorption capacity during the nine years of the RTF operation at the LSM (see the section 5). This is proof that such device can be used during many years without any significant ageing, except the very first layer (few cm) of the activated charcoal, probably exposed to some VOC from the laboratory and tunnel air. The K-factor has been estimated to be (280 ± 18) m³/kg at (−53 ± 1) °C using pure nitrogen. With this method, it has been possible to derive both the mean retention time τ_R and the mean free path X of the radon atoms which are microscopic parameters of the radon behavior in the K48 activated charcoal. The results for this study are summarized in the first line of table 5. It has also been possible to derive the radon suppression factor between the entrance and the exit of the RTF using this method. But as mentioned in section 5, the impact of the exponential function used to derive the suppression factor leads to high relative uncertainties. Therefore, the suppression factor derived from the K-factor measurements will not be taken into account in the following discussion. Nevertheless, it is interesting to note that, given the errors on the airflow, the temperature in the column and the K-factor as well as the exponential behaviour of the reduction factor, the design of the RTF must be done in a very conservative way.

Table 5.  Summary of the adsorption parameters and the radon suppression factors using three independent techniques. These results are taking into account a total mass of (425 ± 5) kg, an airflow of (120 ± 10) m³/h, a charcoal column height of (255 ± 5) cm and a radon mean lifetime of 5.5 days.

Technique	K [m3/kg]	X [m]	τR [d]	Suppression factor
K-factor meas. with N2 at (−53 ± 1) °C	280 ± 18	0.339 ± 0.061	41.3 ± 6.6	${1830}_{-1220}^{+4970}$
210Pb γ spectrometry meas.	—	0.289 ± 0.004	47.6 ± 1.2	${6790}_{-1720}^{+2370}$
Radon online meas.	—	—	—	2000 ± 950

It has been also demonstrated (see section 6) that the ²¹⁰Pb activity as a function of the MLD measured by the gamma spectrometry was following an exponential curve as expected from the radon trapping profile. From the gamma spectrometry measurements, the radon mean free path X and thus the mean retention time τ_R have been deduced (see the second line of the table 5). They are fully consistent at 1σ C.L. with the ones derived from the K-factor study but with smaller uncertainties. This is why it has been possible in this case to derive a more reliable suppression factor of ${6790}_{-1720}^{+2370}$. The third method is the online radon measurement performed at the exit of the RTF with the electrostatic radon detectors. A suppression factor of 2000 ± 950 has been measured during the RTF operation, considering the measured radon activity of the air at the input (20 ± 5) Bq m⁻³ and at the output (10 ± 4) mBq m⁻³ of the RTF [11]. One can observed that the suppression factor derived from the last two methods, i.e. 6790 and 2000, is consistent at 2σ C.L. taking into account the uncertainties.

The RTF installed at the LSM has proven its capacity to purify the LSM air from radon by more than three orders of magnitude down to 10 mBq m⁻³ during the nine years. In order to reach even better air purification, e.g. down to 1 mBq m⁻³, one has to increase the height of the column or to choose another sorbent with a higher K-factor. In that case, the radon self-emanation of the sorbent due to its ²²⁶Ra content may be an issue that has to be taken into account, especially for the last layer of the column. In our work, the K48 activated charcoal has the ²²⁶Ra activity of 0.42 Bq/kg, and we do not know the radon emanation rate for the RTF running temperature. As far as we know, such radon emanation measurement of the sorbent at a low temperature has not been performed yet. An ideal scenario for an efficient and cheap device would be to use mainly a sorbent with a high K-factor, even with radium impurities, and to use at the end of the chromatographic column few cm of a very radiopure sorbent in order to adsorb the radon emanated from the former layers of the sorbent.

In order to provide a radon-free air to the various ultra-low background experiments hosted at the LSM, the RTF using the K48 activated charcoal was designed and built in 2004. This device was well adapted to fulfill the requirements in radon suppression in the air by at least three orders of magnitude, i.e. down to 10 mBq m⁻³. After nine years of operation, the RTF has been stopped and dismantled for maintenance operations. This was a unique opportunity to analyze such a device after almost a decade of operation. An original study has thus been conducted in order to understand how the radon was trapped in the column as a function of the depth. In this way, the charcoal column has been sampled in several layers in order to measure both the K-factor and the charcoal radioactivity, especially the activity of ²¹⁰Pb isotope which provides a picture of the radon trapping localization in the RTF.

Based on the K-factor measurements, it has been shown that despite a decade of running, the K48 activated charcoal has still the same adsorption capacity at the RTF running temperature than before along the column height of 2.6 m, except the first few cm. This demonstrates the high reliability of the RTF where more than 95% of the charcoal has still the same adsorption capacity even after having filtered about 12 million m³ of dried air.

Based on the low background gamma spectrometry measurements, it has been demonstrated that the activity of ²¹⁰Pb as a function of the depth has an exponential behavior, as expected from the radon trapping profile. We have deduced a value of the mean free path X for radon atoms of (0.289 ± 0.004) m as well as the radon mean retention time τ_R of (47.6 ± 1.2) days. These results are fully consistent at 1σ C.L. with the ones derived from the K-factor study. The radon suppression factor of the RTF of ${6790}_{-1720}^{+2370}$ has been also derived from the ²¹⁰Pb profile. Taking into account the uncertainties, this latter value is in agreement at 2σ C.L. with the suppression factor deduced from the online radon measurements performed during the RTF operation.

This work may be very useful in the future in order to design a new RTF at the LSM able to provide radon-purified air at the level of 1 mBq m⁻³.

We acknowledge the LSM staff for their technical assistance during the charcoal sampling. We would also like to thank M Nakahata and the SK collaboration for transferring the know-how of the RTF to the LSM. This work was supported by the Ministry of Education, Youth and Sports of the Czech Republic under the Contract Number LM2015072.