The thermoluminescent dosimeters, TLD-700H, (3.2 × 3.2 mm² surface and 0.89 mm thickness, Thermo Fisher Scientific, United States) were characterized by a high gamma sensitivity and signal stability [17, 18]. They were calibrated at National Laboratory of Frascati (LNF) of the INFN with a¹³⁷Cs source (dose rate: 1,455.22 µGy/h) in terms of air-kerma in the range (5–350) µGy. For each dose point five TLD were simultaneously irradiated. An individual sensitivity factor was calculated for each dosimeter as the ratio of the batch average value and the individual value of the TL response. The TLD signal was corrected for individual factors, background of the TLD reader and the signal cumulated in the trip for their positioning and return in the ISS laboratory for their measurement. The air-kerma value was converted into ambient dose equivalent, H*(10), using the coefficient, H*(10)/ka, of 1.21 [19]. Sets of 5 TLDs were housed in vials similar to those used to keep Drosophila; the dosimeters were put at about 2 cm from the bottom corresponding to the surface of the feed layer. Measurements were done keeping the TLDs in situ for periods of about 4 weeks and of about 3.5 months (the latter is a proper time period in which no fading effects are expected). TLD-700H show a relative response to ¹³⁷Cs better than 0.98 in the range (0.016–1) MeV.

This is a portable 2″ spectrometer (Mirion Technologies Inc., United States) that is capable of measuring dose-rate with an internal Geiger-Mueller (GM) detector (for high dose/count rate measurements) as well as calculating dose-rate, performing nuclide identification and activity measurements with an external NaI(Tl) gamma scintillation detector probe for gamma up to 3 MeV (dimension of 5.08 cm × 5.08 cm). The InSpector 1000 uses the count rate data from the internal GM tube to determine whether the GM data or the NaI probe data will be used for data display. For each NaI probe, two threshold points are used to implement switching hysteresis. Hysteresys stabilizes the dose reading by preventing erratic swithching from one detector to the other. When a falling count rate passes the “Low threshold”, the InSpector 1000 begins displaying the NaI probe’s count rate data. When a rising count rate passes the High Threshold, the InSpector 1000 begins displaying the GM tube’s count rate data. Considering our experiment conditions with dose rates below the level of μSv/h, the “high threshold” for passing from NaI to GM has not been overcome and the InSpector 1000 has been used to provide measurements of the gamma ambient dose equivalent, H*(10), from NaI data.

This detector consists of a calibratable Scintillator Probe (Automess - Automation und Messtechnik GmbH, GER) which, in conjunction with a GM dose rate meter, can be used to measure photon radiation (gamma and X-rays) in the operational quantity ambient dose equivalent H*(10). A cylindrical 7.62 cm × 7.62 cm organic scintillator serves as the radiation detector. The energy range of the instrument with (without) its protective cover is 38 keV–7 MeV (20 keV–7 MeV).

The detectors described in sections 2.3.2 and 2.3.3 were used as independent detection systems with a standard methodology of measured data evaluation supplied by the manufacturers.

This detector consists of a 7.9-L high pressurized ionizing Chamber (Reuter Stokes, United States) that contains approximately 25 atm of ultrahigh purity argon and provides measurement of the ambient dose equivalent H*(10). It has a nearly flat energy response from 70 keV to 10 MeV (±5% accuracy at background) and an omni directional response.

The response of Automess and Reuter Stokes detectors was calibrated at the beginning of the measurement campaigns at secondary standard certified laboratory with Am-241 (60 keV), Cs-137 (662 keV) and Co-60 (1,250 keV) sources. The calibration of the InSpector 1000 was provided by the manufacturer and the response goodness of this detector has then been compared with respect to that of Automess and Reuter Stokes detectors.

Empty polypropylene vials (20 pieces, total mass 130.5 g), cellulose acetate plugs (27 pieces, total mass 40.9 g) and food (total mass 218.09 g) were first measured separately and then all together in vials prepared to host the flies (n. 12 vials, total mass 180.6 g). No special sample treatment was required, as we wanted to assess the background coming from these parts as they are used in the real experiment. The measurements were carried out in the low background counting laboratory “SubTErranean Low Level Assay” STELLA of LNGS [22]. Different detectors were used to assess the samples, all of them being p-type, semi-coaxial HPGe detectors, built in ultra-low background (ULB) configuration, i.e., with especially selected radio-pure materials, surrounded by a large lead/copper shielding, designed to reduce the ambient gamma-ray background by several orders of magnitude. The lead used in these shielding set-ups was selected with respect to its low intrinsic content of ²¹⁰Pb minimizing its contribution through Bremsstrahlung to the intrinsic background of the detector. Therefore, the innermost part of the set-up is also made of high purity copper. Moreover, the set-ups are enclosed in thick, airtight plastic boxes, continuously flushed with highly pure nitrogen gas to avoid any influence from radon gas present in the environment. The nitrogen gas is produced through boil-off from liquid nitrogen. The quality of the liquid nitrogen is 5.0 (i.e. 99.999% purity). Measurements performed on this boil-off nitrogen confirmed that it contains less than 50 μBq/m³ of radon. Data analysis was carried out searching for the relevant peaks in the spectra for the radionuclides of interest (natural radioactivity from ²³⁸U, ²³⁵U, ²³²Th and ⁴⁰K, as well as possible presence of man-made radioactivity, i.e., mainly ¹³⁷Cs, and cosmogenic activity, i.e., mainly ⁷Be). The efficiencies for the full energy peaks were determined through Monte Carlo simulations using well-tested computer codes. This is a commonly used method in case of complex geometries as were vials and plugs.

The spectra, measured at the center of the two laboratories RRE and LRE, placing the 3″ NaI(Tl) scintillator on a 1-m-high desk, are reported as an example in Figure 2A. The total number of counts was always such as to make the statistical error less than 1%. The dose rate values, expressed as absorbed dose rate by the NaI(Tl) crystal, obtained for the various measurement points are shown in Table 1. The dose values outside and inside the incubator were similar. Therefore, the incubator did not significantly perturb the radiation field in the two laboratories.

Regarding the RRE laboratory, the dose rate value measured near the walls was much higher than in the center of the laboratory. This was evidenced by comparing the spectra normalized for the acquisition time measured in the two positions of the RRE laboratory (Figure 2B). The spectrum acquired near the wall roughly showed a double events rate respect to the center of the laboratory and the value on the floor was approximately 50% higher than that on the 1-m-high desk. This is probably due to the presence of gamma emitters in cladding tiles of the walls and floor. Indeed, such ceramic tiles easily contain radioactive elements of the thorium or uranium family. Based on these measurements, the detectors used for the H*(10) rate measurements were placed on the 1-m-high desk at the center of the RRE laboratory to minimize the spatial dependence of dose on the distance from the tiles. These detectors were also placed on the 1-m-high desk at the center of the LRE laboratory. Table 2 shows the H*(10) rate values measured using all detectors available for RRE and LRE laboratories. The RRE laboratory is characterized by the presence of both gamma rays and muons. Each of these detectors measures the total dose rate due to both gamma rays and muons [H*(10)/t]_mis = [H*(10)/t]_γ + [H*(10)/t]_μ. Reuter Stokes accurately measures both [H*(10)/t]_γ and [H*(10)/t]_μ, while Automess and TLD-700H accurately measures [H*(10)/t]_γ but may underestimate [H*(10)/t]_μ [23]. Moreover NaI(Tl) Canberra InSpector 1000 accurately measures [H*(10)/t]_γ but has a negligible muon detection efficiency, i. e., [H*(10)/t]_mis ≈ (H*(10)/t)_γ for this detector. A value of [H*(10)/t]_mis = [H*(10)/t]_γ + [H*(10)/t]_μ = 78.3 ± 4.5 nSv/h was measured using the high pressurized ion chamber, Reuter Stokes, while values of [H*(10)/t]_mis equal to 55.0 ± 4.7 nSv/h and 67.0 ± 1.8 nSv/h were obtained using Automess and TLD-700H respectively. Finally, a value of [H*(10)/t]_mis ≈ (H*(10)/t)_γ = 31.6 ± 2.5 nSv/h was obtained using the NaI(Tl) Canberra InSpector 1000 (corresponding to the ambient dose equivalent rate value of the gamma rays component at RRE), The H*(10) rate values corresponding to the muon component can be obtained for Reuter Stokes, Automess and TLD detectors, by subtracting from their measured values, the value provided by the NaI(Tl) Canberra InSpector 1000 detector. The values obtained were 46.7 ± 5.1 nSv/h (for Reuter Stokes detector), 23.4 ± 5.3 nSv/h (for Automess detector) and 35.4 ± 3.1 nSv/h (for TLD-700H). The underestimation of the muon dose rate of Automess detector and TLD-700H with respect to Reuter Stokes detector, evaluated as percentage variation coefficient, was of 50% for Automess and of 24% for TLD. These results agree with the literature data mentioned by other authors [23, 24].

An additional measurement campaign was carried out using the Reuter Stokes detector exposed in a boat at Civitavecchia harbor, at a distance greater than 200 m from the pier/land and with a depth greater than 12 m to obtain an independent determination of the H*(10) rate value corresponding to the muon component. The result was 47 nSv/h, that is the muon dose rate value at sea level. This value must be corrected for the factor 1.28 due to the altitude of the LNGS RRE and, also, for the attenuation factor of the building where the RRE laboratory is located, equal to 0.8 (according to the UNSCEAR 2000 report Vol. I: Sources) [25]. Therefore, an H*(10) rate value corresponding to the muon component at RRE of 48.1 ± 5.6 nSv/h was obtained, in agreement with the 46.7 ± 5.1 nSv/h value mentioned above. The H*(10) rate value corresponding to the gamma rays component was determined by subtracting this value 48.1 nSv/h from the value 78.3 nSv/h measured using the Reuter Stokes at the RRE laboratory (gamma + muons); it was equal to 30.2 nSv/h, which is compatible with 31.6 nSv/h measured with the NaI(Tl) Canberra InSpector 1000 detector. An average of the values, determined with these two methods, was considered thus obtaining about 47 nSv/h for muons and about 31 nSv/h for gamma rays at RRE laboratory.

The situation is different at LRE laboratory where, since the muon flux value is reduced by a factor of 10⁶, the muon dose rate value was assumed to be negligible respect to the external value and the only low-LET component of the radiation field is given by gamma rays. H*(10) rate values in the range 25–28 nSv/h were measured at LRE using all available detectors. The lowest value of about 25 nSv/h was obtained using the NaI(Tl) InSpector 1000 detector and the highest value of about 28 nSV/h was obtained using TLDs and Reuter Stokes detector. These values agree within the experimental uncertainty, as we expected, since the muon component is not present in this underground site. An average of these values was considered thus obtaining a value of about 27 nSv/h for gamma rays at LRE laboratory. The H*(10) rate mean values for gamma and muon components obtained at RRE and LRE laboratory, are summarized in Table 3. Therefore, a dose rate reduction of the low LET components (photons + muons) of about 3 can be obtained at LRE laboratory compared to RRE.

The dosimetric characterization was also performed inside the devices used to modulate the low LET dose rate component, i.e., in the Marinelli beaker containing tuff, and inside the shielding consisting of a 10 cm thick lead hollow cylinder placed at RRE laboratory and a 5 cm thick iron parallelepiped placed at LRE laboratory. The dosimetric characterization of the lead shielding at RRE laboratory was performed using both the 7.62 cm × 7.62 cm NaI(Tl) scintillator and TLDs. The NaI(Tl) scintillator allows to measure the gamma dose rate reduction factor since this detector has a muon detection efficiency equal to zero. This factor was equal to about 16. In addition, sets of TLDs were placed inside the lead shielding for about 3.5 months to measure the total dose rate reduction factor of gamma and muons components together. This factor was equal to about 2, much lower than that measured with the NaI(Tl) scintillator since the muons are only slightly stopped by the Pb shielding. The dosimetric characterization of the iron shielding at LRE laboratory was performed using both the 7.62 cm × 7.62 cm NaI(Tl) scintillator and TLDs. The gamma dose rate reduction factor for the iron shielding was equal to about 6. Moreover, a dose rate reduction of the low LET components (photons + muons) of about 17 can be obtained inside the iron shield at LRE laboratory compared to the value at RRE laboratory.

The dosimetric characterization of the Marinelli beaker, was performed using TLDs at both RRE and LRE laboratories. The Marinelli beaker, containing about 2,800 g of tuff, was placed inside the lead shielding at RRE laboratory, to eliminate the ambient gamma component, and on the 1-m-high desk located at the center of LRE laboratory. Sets of 5 TLDs were housed in five vials similar to those used for Drosophila. Figure 3 shows a sketch of the vials inside the Marinelli beaker (open circles in Figure 3). The dosimeters were put at about 2 cm from the bottom corresponding to the surface of the feed layer. To evaluate the dose rate distribution of the radiation field inside the Marinelli beaker, TLDs were placed at the same height from the base in five vials, four of them around and one in the center. Moreover, the vial placed at the centre of the Marinelli beaker was selected to house the TLDs at two different heights (5,6 in Figure 3) (2 cm and 6 cm respectively). Measurements were performed keeping TLDs in situ for periods of about 4 weeks. The results of these measurements are reported in Table 4. A small variation of the dose rate was evidenced at the center of the cylindrical hole of the Marinelli beaker for different heights. A mean value equal to about 120 nSv/h and 118 nSv/h was obtained at RRE and LRE laboratories respectively. Moreover, a higher dose rate value was found at the perimeter of the Marinelli internal hole (for the minimum height). The mean value of the H*(10) rate into the cylindrical hole of the Marinelli beaker at the minimum height was about 129 nSv/h and 127 nSv/h at RRE and LRE laboratories respectively. The uniformity was such that the variations in the irradiated area are less than 16% (acceptable for most radiobiological experiments).

The dose rate value due only to the tuff was 89 nSv/h at RRE, calculated by subtracting the value obtained in the lead shielding without the Marinelli, 31 nSv/h, from the mean value at the center, 120 nSv/h. The estimate of dose rate due to the tuff alone at the LRE laboratory can also be derived by subtracting the value obtained using TLDs on the 1-m-high desk located at the center of the laboratory, 28 nSv/h, from the mean value at the center of the cylindrical hole of the Marinelli beaker, 118 nSv/h. In this case, a dose rate of 90 nSv/h was obtained, confirming the value found at RRE laboratory.

The average values of radon activity concentration in the air for LRE and RRE laboratories measured from March 2019 to July 2021 were 22 Bq/m³ and 15 Bq/m³ with standard deviations of 11 Bq/m³ and 3 Bq/m³ respectively. For the RRE laboratory, seasonal variations of radon concentration levels, measured during 2020 are shown in Figure 4. There were no seasonal variations of radon concentration observed in the LRE laboratory.

The neutron flux was measured placing the DIAMON spectrometer on the 1-m-high desk located at the center of the RRE laboratory for about 40 h. A measured value of 0.018 neutrons/cm²s^-1 was obtained, with an average experimental data uncertainty of about 7%. The energy spectrum of neutrons at LRE can be seen in Figure 5.

To further characterize the radioactive background of our experiments, the intrinsic contribution of radioactivity in experimental set up (caps, vials and culture medium, see Figure 6A) was carried out by gamma ray spectrometry with high pure germanium (HPGe) detectors and Inductively Coupled Plasma Mass Spectrometry (ICP-MS) (Element II, Thermo Scientific, United States). The analysis was carried out by these two different techniques taking advantage of the LNGS expertise for the measurement of ultra-low-level radioactivity applied to the study of rare fundamental physics phenomena, such as double-beta decay, nuclear decays and dark matter, [26, 27]. It is worth remembering that these techniques are very different in terms of detectable nuclides (precursors vs. daughters), sensitivity and time of measurement. Furthermore, ICP-MS is considered a destructive technique that requires elaborate sample processing prior the analysis. On the other hand, HPGe spectroscopy allows for very easy sample handling although it requires long time of measurement and a large amount of sample. Usually, the choice of one over the other technique depends on the physical properties of the sample, the amount of available sample and time [26]. Concerning the natural decay chains of uranium and thorium, HPGe spectrometry and ICP-MS address different nuclides. They also have different sensitivity and times of measurement, ICP-MS being usually much faster and more sensitive to uranium and thorium themselves, whereas gamma-ray spectrometry is more sensitive to the gamma-active shorter-lived progenies within their decay chains, but with longer measurements times. A combination of both can usually give a complete picture on the status of the decay chain and identify possible ruptures of the secular equilibrium within it. It is important to point out that ICP-MS analysis directly quantifies the concentration of ³⁹K and the concentration/specific activity of ⁴⁰K is derived based on its natural isotope abundance (0.012%).

Each experimental set consists of a polypropylene vial weighing about 6.6 g, a cellulose acetate cap weighing about 1.5 g and Nutrifly medium weighing about 7 g. As expected, the major contribution in terms of specific activity comes from ⁴⁰K. The results from the two analyses are in good agreement with each other, especially for the ⁴⁰K activity (Table 5). The analysis on fly was carried out only by ICP-MS. Figure 6B reports the values from the ICP-MS analysis, with the contribution of ⁴⁰K calculated for the single element of the experimental set up taking into account the weight of the elements (i.e. 1.5 g plug; 6.6 g vial; 7 g food and 0.9 × 10⁻³ g fly).

The neutron flux measured with the DIAMON spectrometer at RRE was used as input for the Monte Carlo simulation, to evaluate the neutron dose rate at RRE. In the simulation the neutrons were uniformly generated from the surface of a 10 cm sphere around the experimental setup. The angular distribution of neutrons was isotropic, while the energy distribution reproduced the spectrum of Figure 5 from thermal energies to about 20 MeV (see Figure 7). Two simulations were performed varying the sphere radius from 10 cm to 20 cm and considering 10⁷ and 10⁸ simulated events, which resulted in excellent agreement. The 2D map of neutron dose rates calculated in the 1 cm³ water scoring volume are shown in Figure 8, for the simulation with the sphere radius of 10 cm and 20 cm. Each pixel gives the dose rate value calculated for the corresponding voxel. To extract from the maps an average value of the physical dose rate due to neutrons, we did not consider the external layer of pixels to minimize possible border effects. By averaging over the central pixels from the simulation with 10⁸ neutrons, a dose rate of 1.4 nGy/h was obtained with a standard deviation of 0.2 nGy/h.

Simulations were performed to evaluate the dose rate in water inside the vial for Drosophila due to the radioactive contamination of the different components (plug, vial, medium) starting from their measured activities. We focused on the ⁴⁰K contained in the medium that gives the prevailing contribution to the dose rate (as from Table 5). In the simulation, the vial geometry has been implemented as in Figure 9A with ⁴⁰K decays uniformly distributed in the medium. The dose has been estimated in a cube of water with volume equal to 1 cm³ placed at different heights with respect to the medium. The spectrum of gamma energy deposits in the water cube is shown in Figure 9B. The spatial distributions of energy deposit evaluated for distances of the cube with respect to the medium of 2 mm and 30 mm respectively are shown in Figures 9C, D.

In the situation where the water cube is placed at the center of the vial (30 mm distance from the medium), a gamma dose rate in the range 0.025–0.25 nGy/h was calculated. Since the medium has been located at the bottom of the vial, there is a variability of dose rates (lower dose rates with higher distances from the medium). The estimated intrinsic gamma dose rate can be considered low and negligible compared to the environmental contribution.

Terrestrial gamma dose rate values in the external laboratories are similar for LNGS and WIPP, while SNOLAB and Modane report a lower and higher value respectively. The elevated gamma dose rate value at Modane laboratory is caused by naturally high radiation levels due to the soil composition [29]. The dose rate values of muon component in the external environment, measured at LNGS and simulated with Monte Carlo Code at SNOLAB and Modane laboratories, are very similar. To our knowledge, WIPP group did not measure or simulate the dose rate corresponding to the muon component. However, for several radiobiological experiments, WIPP group did not consider the external environment as a reference but they used a source consisting of KCl salt (containing ⁴⁰K, a beta-particle, and gamma-ray emitter) within their underground facility to simulate the external reference environment with a dose rate of about 70 nSv/h, a level matching typical average surface background radiation levels [7].

Of course, a single source of radiation does not represent the multitude of natural sources; in particular, the difference is given by the muons that are in above-ground environment but not in underground environment. However, since muons and gamma rays are both low-LET radiation, some biological systems may not be affected by this difference, at least for some endpoints [12].

Muon dose rate values are negligible for all underground laboratories. Higher values of gamma dose rate were obtained at LNGS and Modane laboratories, while a value of only 8 nSv/h was obtained at SNOLAB and WIPP. Overall, the dose rate values of the low LET component are reduced, compared to those obtained in the external laboratories, by a factor of approximately 3 and 8 for LNGS and SNOLAB respectively, 10 for Modane and WIPP. A ‘lead castle’ constructed with 10-cm-thick lead bricks was used in underground environment of SNOLAB, obtaining a strong reduction of gamma dose rate [32]. Shields were also used in the other underground laboratories. At LNGS a 5-cm-thick iron shield was used to reduce the gamma dose rate; a stronger reduction, like that evaluated by SNOLAB, can be obtained by moving the lead shield from LNGS RRE laboratory to LRE laboratory. A gamma dose rate value < 1nSv/h was obtained at Modane underground laboratory using a lead shield consisting of an interior layer of 5 cm of copper, surrounded by 10 cm of lead [30]. Finally, a 15 cm-thick vault made from pre-World War II, low-activity steel was used at WIPP to reduce gamma dose rate up to 0.16 nGy/h [5]. However, a different behavior of the biological system, maintained in parallel in low radiation background conditions with respect to those maintained in normal environmental background conditions, was observed even without using shields [12, 15].

The lead shield at RRE laboratory allowed for a strong gamma dose rate reduction as measured with the NaI(Tl) scintillator detector. Therefore, within the lead shield at the RRE laboratory the biological system is basically exposed only to the muon component. The muon dose rate value obtained inside the lead shield by TLD measurements was 30.8 ± 4.2 nSv/h. We would like to point out that the lead shield at RRE allows for a complementary situation to the one at the LRE where the biological system is exposed only to the photonic component. Studies on the single components of low LET can be made to understand how each of them can affect the response of the biological system. To exploit this possibility radiobiological experiments are planned considering the fruit fly D. melanogaster as biological systems maintained in parallel inside and outside the lead shield at the RRE laboratory and also at the LRE laboratory. Moreover, the iron shield at LRE allows to further reduce the gamma dose rate. Overall, thanks to the use of iron shield the total dose rate (gammas + muons) at LRE is 17 times lower than that at RRE, making the LRE laboratory a suitable location to carry out radiobiological studies with a strong reduced dose rate.

The gamma dose rate can be increased at LRE using the especially designed Marinelli beaker filled with tuff, a natural gamma-emitting building material. The energy spectrum of the tuff was already reported in a previous paper [14]. When this is filled with about 2,800 g of tuff, an H*(10) rate value increase of about 90 nSv/h can be obtained. This value is higher than the low LET environmental dose rate value at both RRE and LRE laboratories. However, it is possible to reduce the intensity of the “Marinelli beaker” source by decreasing the amount of tuff within it. Starting from the measured value of about 27 nGy/h, the approach of modulating the amount of tuff inside the Marinelli beaker allows for a gradual increase in the gamma dose rate at the LRE laboratory in order to understand wheather there is a threshold value of the H*(10) rate that induces a response switch of the biological system, causing it to change from one state to another.