Introduction

Indoor radon (222Rn) and thoron (220Rn) and their decay products represent the largest source of exposure to ionizing radiation in the population, causing lung cancer (WHO, 2009; UNSCEAR, 2010). 222Rn and 220Rn are noble gases formed in the radioactive decay series of uranium (238U) and thorium (232Th), respectively, and occur naturally in rocks and soils. Research efforts have been primarily focused on the study of 222Rn concentration in dwellings, workplaces, building materials, water, caves, soil and bedrock units (e.g., Neznal et al., 1996; Åkerblom & Lindgren, 1997; Sainz et al., 2007; Appleton, 2013; A. Pereira, et al., 2013; Domingos & Pereira, 2018; Jorge & Pereira, 2020; Sêco et al., 2020; Domingos et al., 2021). The assessment of indoor 220Rn concentrations has been underrated due to its short half-life and technical difficulties in its measurement, among other factors (e.g., Somogyi et al., 1984; Virk & Sharma, 2000; Tokonami, 2010; Vaupotic and Kávási, 2010). An increasing effort to discriminate and evaluate the concentration of both isotopes in dwellings has, however, been observed in the last few years (e.g., Kávási et al., 2007; Prasad et al., 2010; Mayya et al., 2012; Ramola et al., 2016; Zunic et al., 2017).

It is recognized that indoor 222Rn and 220Rn concentration are directly related to exhalation from bedrock, soil and the building materials (e.g., Kumar & Chauhan, 2014; Mann et al., 2015). Several lithological units such as granites, metamorphic rocks and sedimentary rocks present a distinctive radon exhalation potential, leading to an increased risk of exposure to ionizing radiation (e.g., Kemski et al., 2009; Scheib et al., 2013; A. Pereira et al., 2017; Sêco et al., 2020). The activity concentrations of 222Rn and 220Rn, and their progeny in dwellings are also influenced by anthropic factors and meteorological conditions, among others (see references above).

In the southernmost region of the African continent, 222Rn and 220Rn studies have been carried out mostly in Namibia, South Africa and Swaziland (e.g., Mahlobo & Farid, 1992; Nsibande et al., 1994; Farid, 1995; Mahlobo et al., 1995; Lindsay et al., 2008; Kgabi et al., 2009; Oyedele et al., 2010; Njinga et al., 2016; Botha et al., 2017; Munyaradzi et al., 2018). In Angola, studies on this subject are limited to the determination of 222Rn and 220Rn in 45 dwellings along with the concentration of radioelements in building materials (adobe) in Cabinda, Huambo and Menongue (Salupeto-Dembo et al., 2018, 2020). Measurements of 222Rn were also conducted in 75 dwellings located in Lubango (Bahu et al., 2021). The relationship between the geological and anthropic factors, and indoor radon and thoron concentrations is still far away from being understood, not only in Angola but in the African continent. Beyond this, few studies have reported the results of simultaneous measurements of indoor radon and thoron concentrations in the literature in other regions of the world.

The Angolan territory is vast and composed of several geological units which may present highly variable concentrations of radionuclides. The national geological database of Angola has been recently updated, being subjected to several improvements (e.g., Carvalho et al., 2000; Delor et al., 2008; Pedreira & Waele, 2008; Batumike et al., 2009; Ferreira da Silva, 2009; E. Pereira et al., 2011, 2013a, 2013b; Lopes et al., 2016) which are crucial tools to the study of 222Rn and 220Rn distribution in the environment. The increase in knowledge on 222Rn and 220Rn distribution is critical to estimate the exhalation potential of geological materials that are commonly used as building materials in African countries, to model their spatial distribution and for the development of risk maps in the near future.

The main goal of this paper is to evaluate the distribution of 222Rn and 220Rn concentration and to identify the factors that may lead to an increased risk of exposure to these isotopes for the population. A survey of indoor 222Rn and 220Rn measurements in buildings of several counties located in the southwest region of Angola was carried out accounting for different types of bedrock units. To evaluate the role of the geological factors in the distribution of these gases, the concentration of their parent isotopes, namely 226Ra and 224Ra, and the emanation and exhalation power of 222Rn and 220Rn were determined in rock samples collected in outcrops and in a borehole. The anthropic factors were evaluated taken into account the building materials, the type of building usage and the habits of the inhabitants.

Of a broader significance, we demonstrate an effective link between the geological factors and indoor concentrations data and discuss the factors that may lead to higher indoor radon concentrations, posing a potentially significant health hazard due to an increased dose received by the population.

Geological framework

The Lubango, Quilemba, Humpata, and Matala counties are part of the Huila Plateau which is a part of the Angola portion of the Proterozoic Congo Craton (i.e., Angola-Kasai Block; e.g., Marques, 1977; Delor et al., 2008; Pedreira & Waele, 2008; Batumike et al., 2009; Lopes et al., 2016). The region comprises a large variety of lithological units that have been thoroughly studied (e.g., Torquato et al., 1979; Matias, 1980; Carvalho, 1983, 1984; Carvalho & Alves, 1993; Carvalho et al., 2000; Ferreira da Silva, 2009; E. Pereira et al., 2011, 2013a, 2013b; Lopes et al., 2016; and references therein). E. Pereira et al. (2013b) presented a simplified classification for the units studied in the present work that include the: (1) Pre-Eburnean basement (PEB); (2) Eburnean episode units; (3) Post-Eburnean rocks; and (4) Kibarean distension units which are described summarily in Fig. 1.

Fig. 1
figure 1

Simplified geological map of the study area displaying the location of indoor radon measurements, bedrock samples collected and the executed borehole. Adapted and reinterpreted from the third sheet of the Geological Map of Angola at the scale of 1: 1.000.000 (LNICT, 1980) and E. Pereira et al. (2013a, 2013b). The lithostratigraphical framework was retrieved from the published literature listed under references

Full size image

Material and methods

The detectors for the measurement of indoor 222Rn and 220Rn concentration and bedrock samples were prepared and analyzed at Laboratory of Natural Radioactivity of the Department of Earth Sciences of University of Coimbra (LRN-UC), which complies with the accreditation criteria for testing laboratories established in ISO/IEC 17,025:2017 for the measurement of indoor radon using passive detectors according to ISO 11665-4 and the activity concentration of 226Ra, 232Th and 40K in construction natural and man-made materials, including soils and bedrock samples according to ISO 18589-3.

Indoor radon and thoron

About 98 passive detectors of the RADUET type (from Radosys Ltd.) were used for the measurement of indoor 222Rn and 220Rn concentration in dwellings and workplaces through integration methods. RADUET detectors were developed and calibrated by the National Institute of Radiological Sciences (Zhuo et al. 2002; Tokonami et al., 2005; Sorimachi et al., 2012). They comprise two CR-39 discriminative polyallyl diglycol carbonate sensors (10 × 10 mm) with two selective chambers (denoted main and sensitive chambers) that record 222Rn and 220Rn tracks (see Tokonami et al., 2005). The performance of these detectors has been evaluated elsewhere (see Sorimachi et al., 2012), and they have been shown to provide similar results to other types of discriminative 222Rn and 220Rn passive and active detectors (see Zhuo et al. 2002; Szeiler et al., 2012; Omori et al., 2020).

Several factors were considered in the distribution of the 98 RADUET detectors for sampling of indoor 222Rn and 220Rn including: (1) the type of building usage (dwellings or workplaces); (2) the type of building materials (e.g., clay, concrete, metal, stone, filled or hollow bricks; see Fig. 2); and (3) the underlying bedrock materials (see Fig. 1).

Fig. 2
figure 2

Examples of building materials and styles: A Clay (Adobe); B Concrete block; C Metal (Metallic plates); D filled clay brick; E hollow clay brick; F Stone; G Adobe built housing; H Dwelling built with concrete blocks; I Construction with metal plates; and J Dwelling built with stones

Full size image

About 81 detectors were placed at a height between 0.8 and 2.0 m from the ground in ground floor rooms with high occupancy factors. The detectors were placed in secure locations to prevent damages and away from heat and ventilation sources. The ventilation conditions were variable between the sampled buildings since the detectors were placed in dwellings and workplaces built with different construction materials and styles. Contrary to the recommendations stated in ISO 11665-4, the detectors were placed closely to the walls to attain a higher efficiency in thoron detection. The remaining detectors (n = 17) were placed strategically in 6 distinct buildings to evaluate the influence of the proximity of the detectors to the floor and walls on indoor 222Rn and 220Rn results and thereby, the effectiveness of thoron detection.

The detectors were retrieved on average, after three months of exposure; however, some detectors were exposed for periods of up to six months. The duration of sampling was adjusted between sampling sites to suit the assumed radioactivity as recommended in ISO 11665-4.

Following exposure, each detector was sealed and sent to LRN-UC. The detectors were then chemically revealed using the Radobath equipment (Radosys Ltd.), at a temperature and duration defined by the manufacturer. The tracks were counted with an automatic image analysis system, the Radometer 2000 Dosimetric Microscopy (Radosys Ltd.). Indoor 222Rn and 220Rn activity concentration were calculated according to ISO 11665-4. Background measurements from 10 non-exposed detectors were detracted from the total tracks count of the exposed detectors. Calibration factors were provided by the manufacturer (Fc in tracks/cm2 per Bq.h/m3). Data below the lower limit of detection (LLD) of 10 Bq/m3 for indoor 222Rn and 15 Bq/m3 for indoor 220Rn were replaced by 0.65 LLD after Palarea-Albaladejo and Martín-Fernandez (2013).

The study of the influence of the proximity of the detectors to the floor and walls on indoor 222Rn and 220Rn revealed a fivefold increase in the median indoor 220Rn between detectors placed near the floor and/or walls (of 242 Bq/m3), compared to detectors placed in a mid-ground level (49 Bq/m3) (see Fig. 3). However, according to the results of the Mann–Whitney–Wilcoxon test, indoor 222Rn (W = 33.0, p value = 1.000) and indoor 220Rn (W = 47.5, p value = 0.159) were not significantly different according to the detector placement strategy, which is probably due to a low sample size.

Fig. 3
figure 3

Box and whiskers plots for comparison of a Indoor 222Rn and b Indoor 220Rn results from detectors placed closely to the floor and/or wall versus detectors placed in a mid-ground level

Full size image

Bedrock samples

Fieldwork was carried out in the Lubango, Quilemba, Humpata and Matala counties. About 58 bedrock samples were collected for the determination of 226Ra and 224Ra activity concentration, 222Rn and 220Rn exhalation data and emanation coefficients, including 6 samples from a borehole. The samples were collected methodically according to the different types of rocks recognized in the study area, being considered representative of the geological unit sampled (Fig. 1). The selection of the sampling sites was supported by a preliminary assessment of the distribution of radionuclides (K, U and Th) in fresh and flat bedrock surfaces using a handheld gamma-ray (GR) spectrometer (RS-230, from Radiation Solutions) provided by LRN-UC following the procedure described in Sêco et al. (2021). Samples were collected taking into account the homogeneity of the GR spectrometer readings in order to improve the representativeness of the samples (Sêco et al., 2021). Areas covered by soil or vegetation were avoided, as well as uneven surfaces (e.g., Sêco et al., 2021).

The samples collected were dried in an oven at 70ºC, milled with a tungsten carbide mill and stored in radon proof Marinelli beakers. After at least 27 days, the samples were analyzed with a benchtop GR spectrometer from Ortec with a NaI (Tl) detector (3″ × 3″) mounted within a lead shield. The 1767.5 and 2614.5 keV isotopic lines were used to measure 214Bi and 208Tl, respectively. The measurements lasted 10 h and the spectra were analyzed with ScintiVision-32 (version 2, Ortec). Given that samples were stored until equilibrium between 222 and 220Rn and their parent isotopes were attained, 214Bi and 208Tl activity are representative of 226Ra and 224Ra activity, respectively. The methodology is discussed in detail by Domingos and Pereira (2018). The estimated LLD for the sample set was 4.1 Bq/kg for 226Ra and 1.7 Bq/kg for 224Ra.

222Rn and 220Rn exhalation and emanation coefficients were determined with the accumulation method according to the procedure described in Domingos et al. (2021) using an AlphaGuard DF2000 monitor from Saphymo GmbH. The flow rate was set to 2 L/min, and the continuous Rn/Tn 2L measuring mode was used to discriminate between 222 and 220Rn. According to Burkin and Villert (2017), the LLD for 222Rn and 220Rn when using the Rn/Tn 2L measuring mode when calculating the mean value of a 60 min period is 6 Bq/m3 for 222Rn and 12 Bq/m3 for 220Rn. The mean activity concentration of 222Rn and 220Rn was estimated by calculating the mean over a period of at least 1220 min (see also Domingos et al., 2021). The background 222Rn concentration was monitored prior to each measurement and subtracted from the equilibrium concentration. The concentration of 222Rn was corrected to correspond to the equilibrium concentration. The 222Rn and 220Rn exhaled per unit mass (EX) were determined with the following equation (Eq. 1):

$$EX = \frac{C \times V}{W}$$
(1)

where V is the free volume of the stainless-steel container, W is the weight of the sample (in kg), C is the equilibrium concentration of 222Rn or 220Rn (in Bq/m3). The 222Rn and 220Rn emanation coefficients were estimated by dividing the 222Rn and 220Rn exhaled per unit mass by the activity concentration of 226Ra or 224Ra (in Bq/kg), respectively. Data below the LLD for 226Ra, 224Ra, 222Rn and 220Rn were replaced by 0.65 LLD (after Palarea-Albaladejo and Martín-Fernandez, 2013) prior to the computation of 222Rn and 220Rn exhalation and emanation coefficients.

To perform geochemical analyses by X-ray fluorescence (XRF), the samples were ultimately dried at 105 °C, grinded with an agate mill and sieved to a grain size lower than 125 μm. Fused beads were prepared by subjecting the samples to a maximum temperature of 1065 ºC in a Claisse LeNeo fusion instrument using a 1:10 sample to flux ratio composed by lithium metaborate and lithium tetraborate (in equal proportion) for measurement of major elements (see Domingos & Pereira, 2018). Pressed pellets were also prepared for the measurement of trace elements, namely U and Th concentrations. Analyses were performed at Instituto Pedro Nunes with an AXIOSmAX (PANalytical) spectrometer, calibrated with certified geological standards. Loss on ignition (LOI) was estimated through the weight loss after subjecting the samples to a temperature of 1065 °C in an oven.

Annual inhalation dose

The total annual inhalation dose (AID) (in mSv/y) due to inhalation of 222Rn and 220Rn and their progeny was estimated with the following equation (UNSCEAR, 2000):

$$\begin{aligned} {\text{AID}} = & \left\{ {\left( {0.17 + 9 \times F_{{{\text{Rn}}}} } \right) \times C_{{{\text{Rn}}}} + } \right. \\ & \left. {\left( {0.11 + 40 \times F_{{{\text{Tn}}}} } \right) \times C_{{{\text{Tn}}}} } \right\} \times {\text{OF}} \times T \times 10^{{ - 6}} \\ \end{aligned}$$
(2)

where FRn and FTn are the equilibrium factors for 222Rn and its progeny and 220Rn and its progeny, respectively, which were assumed to be 0.4 for 222Rn and 0.02 for 220Rn (UNSCEAR, 2010); CRn and CTn are the measured 222Rn and 220Rn activity concentrations (in Bq/m3); OF is the occupancy factor and T is the time of exposure. The values 0.17 and 9 correspond to dose conversion factors for 222Rn and its progeny concentrations (in nSv), respectively, while the values 0.11 and 40 are the dose conversion factors for 220Rn and its progeny concentrations (in nSv), which were retrieved from UNSCEAR (2000). The factor 10−6 allows the conversion of nSv into mSv.

The indoor occupancy factor was estimated from the time of exposure that was calculated as the difference between the number of hours in a year (365 × 24 = 8760 h) and the time spent away from home, considered as the normal working period. According to the Angola Labor Law, the normal work period should not exceed 8 daily hours. Workers have the right to weekly days of rest, and the vacation period is 22 working days in each year. Angola has also 15 public holidays (source: https://en.wikipedia.org/wiki/Public_holidays_in_Angola, last accessed on December 7, 2021). Thus, working days were estimated to be 224, giving an estimated time spent away from home solely related to working of 1792 h, and an occupancy factor for workplaces around 0.20. Hence, the estimated time of exposure for indoor radon in dwellings was 6968 h, giving an occupancy factor of 0.79, which is similar to the indoor occupancy factor considered in UNSCEAR (2010) of 0.8.

Data analysis

Firstly, descriptive statistics, such as the mean, standard deviation, median, coefficient of variation, minimum and maximum, were estimated to summarize the data and to evaluate the dispersion. The distribution of the data was studied using the Shapiro–Wilk W test for normality and quantile plots. The Spearman Rank correlation coefficient was used to assess correlations. The influence of building materials, building usage and bedrock on 222Rn and 220Rn data was assessed through the Kruskal–Wallis test using those variables as categorical variables. Statistical significance was established at a level of 0.05. All statistical analyses were performed with R (R Core Team, 2021), using the packages EnvStats (Millard, 2013), Hmisc (Harrell, 2020), psych (Revelle, 2020) and summarytools (Contois, 2020).

Results and discussion

Indoor radon and thoron data

Descriptive statistics of 222Rn and 220Rn indoor measurements, the ratio between 222 and 220Rn indoor concentrations and the total indoor radon concentrations (TIRC), which corresponds to the sum of 222Rn and 220Rn concentrations, are presented in Table 1.

Table 1 Descriptive statistics of indoor 222Rn and 220Rn data
Full size table

The World Health Organization’s (WHO) proposes a RL of 100 Bq/m3 for indoor radon which may be extendable to 300 Bq/m3 if the country-specific conditions inhibit attaining the RL mentioned above (WHO, 2009). Indoor 222Rn data range from 7 to 227 Bq/m3, being lower than 300 Bq/m3 in all buildings studied. Indoor 220Rn data range from 10 to 461 Bq/m3, being higher than 300 Bq/m3 in 3 out of 81 measurements (approximately 4% of the data). The TIRC is higher than 300 Bq/m3 in 5% of the data. The mean and median indoor 222Rn and 220Rn are lower than 100 Bq/m3 separately; however, the combined mean concentration of these isotopes is equal to 100 Bq/m3 (see Table 1). Bahu et al. (2021) report indoor radon concentrations in Lubango similar to the TIRC measured in present work, ranging from 20 to 497 Bq/m3 with a mean of 136 Bq/m3 determined with standard CR-39 detectors, which do not allow for discrimination between 222 and 220Rn.

Indoor 220Rn data are more variable than 222Rn indoor data as shown by the coefficient of variation (Table 1). The ratio between indoor 222Rn and 220Rn varies between 1:21 to 12:1, averaging 3:1. Hence, indoor 222Rn concentrations can be considered generally higher than indoor 220Rn concentrations. In fact, indoor 222Rn concentrations are higher than indoor 220Rn concentrations in 63 buildings (77% of the buildings studied). The proportion of 220Rn in TIRC varies from 8 to 95%, averaging 35%.

222Rn and 220Rn indoor data present a positive asymmetry. 222Rn distribution is platykurtic, whereas 220Rn distribution is leptokurtic. The results of the Shapiro–Wilk W test indicate a significant deviation from normality (Table 1). The normal quantile plots for log-transformed data are presented as supplementary material. 222Rn distribution is approximately log-normal, while 220Rn data deviate from the log-normal distribution due to a considerable number of samples below the LLD (see Table 1). Thus, nonparametric statistical methods were used to investigate the relationship between the studied variables. 222Rn and 220Rn indoor data are not correlated according to the results of the Spearman rank correlation coefficient (R = 0.10, n = 81, p value = 0.377), even if data below the LLD for 222Rn and 220Rn are excluded (R = − 0.05, n = 45, p-value = 0.745).

Data were grouped by geological unit, building materials, and building usage in Fig. 4. The results of the Kruskal-Wallis test for indoor 222Rn and 220Rn data using geological units, building materials and usage as categorical variables are presented in Table 2.

Fig. 4
figure 4

Box and whiskers plots according to the geological units (a, b), construction materials (c, d) and building usage (e and f) for indoor 222Rn (left) and indoor 220Rn (right) concentrations. Geological units: EGM—Eburnean granitoids and migmatites; CG—Chela group; LF—Leba Formation; PEMR—Post-Eburnean Mafic Rocks; RGP—A-type red granites and porphyritic rhyolites. Building materials: 1—Clay; 2—Concrete; 3—Metal; 4—Stone; 5—Brick (5A—filled brick; 5B—hollow brick)

Full size image
Table 2 Results of the Kruskal–Wallis test for indoor 222Rn and 220Rn data using geological units, building materials and usage as categorical variables (81 samples)
Full size table

Indoor 222Rn and 220Rn data are not significantly different between geological units (Table 2). Both are higher in A-type red granites and porphyritic rhyolites (RGP), showing median indoor 222Rn and 220Rn levels higher than WHO’s RL of 100 Bq/m3, and lower in Post-Eburnean Mafic Rocks (PEMR, Fig. 4). The RGP extend through intrusive bands with a NE-SW alignment, resulting from (probable) anorogenic (A-type) magmatism (Torquato et al., 1979; Carvalho et al., 1987; Drüppel et al., 2007). According to Drüppel et al. (2007), A-type granites from the felsic rock suite that intrudes the anorthosite rocks of the Kunene Intrusive Complex typically contain abundant incompatible elements (Large Ion Lithophile and High Field Strength elements), including U and Th, which may justify the higher activity concentrations of 222Rn and 220Rn reported in this unit. The PEMR include dykes and sills of dolerites, gabbros and norites (LNICT, 1980). Low concentrations of U and Th are often observed in mafic rocks (e.g., Cinelli et al., 2019; Domingos et al., 2020), which may explain the lower indoor 222Rn and 220Rn activity concentrations observed in these units. The Chela Group (CG) presents higher median indoor 222Rn and 220Rn levels compared to the Eburnean granitoids and migmatites (EGM). The Leba Formation (LF) presents high median indoor 222Rn levels, but comparatively low 220Rn levels.

222Rn data are not significantly different according to the building materials (Table 2), in contrast with indoor 220Rn concentrations which are significantly different according to the building materials (Table 2). Clay and concrete-based building materials are associated with higher median 220Rn levels, while lower 220Rn levels are observed associated with metal and brick-based building materials (Fig. 4).

The influence of building usage was also assessed. Indoor 222Rn are not significantly different between dwellings and workplaces, although higher median values are observed in dwellings (Table 2, Fig. 4). Indoor 220Rn data are significantly different between dwellings and workplaces, with indoor 220Rn levels being also higher in dwellings compared to workplaces (Table 2, Fig. 4).

Exhalation and emanation data

Descriptive statistics for 226Ra and 224Ra activity concentration, 222Rn and 220Rn exhaled per unit mass and their respective emanation coefficients are presented in Table 3. Samples retrieved from the borehole executed within a soil profile of an Eburnean granitoid were excluded from this analysis.

Table 3 Descriptive statistics of 226Ra and 224Ra activity concentration, 222Rn and 220Rn exhalation rate and emanation coefficients
Full size table

224Ra activity concentration and 220Rn exhaled per unit mass are higher on average than 226Ra activity concentration and 222Rn exhaled per unit mass, respectively, despite a similar range (Table 3). The median 222Rn emanation coefficient is two times higher than the median 220Rn emanation coefficient.

The distribution of 226Ra, 224Ra, 222Rn and 220Rn exhaled per unit mass and emanation coefficients are positively skewed. The distribution of 226Ra is leptokurtic. The distribution of 222Rn emanation is approximately mesokurtic, whereas the remaining variables present a platykurtic distribution. The results of the Shapiro–Wilk W test indicate a significant deviation from normality for all variables apart from the 220Rn emanation coefficient (Table 3). The normal quantile plots for log-transformed data and the 220Rn emanation coefficient are presented as supplementary material. The linearity observed in those plots indicates that the distribution of 226Ra and 224Ra activity concentration, 222Rn and 220Rn exhaled per unit mass and 222Rn emanation coefficient is approximately log normal. Hence, nonparametric statistical methods were used to assess correlation between the variables (Table 4).

Table 4 Spearman rank correlation matrix of radiological data
Full size table

Statistically significant positive correlations are observed between all variables apart from 226Ra activity concentration and 222Rn emanation coefficient (Table 4). The correlation between 222 and 220Rn exhaled per unit mass is stronger than the correlation between 222 and 220Rn emanation coefficient, which is, in turn, stronger than the correlation between 226 and 224Ra activity concentration. 226Ra activity concentration and 222Rn exhaled per unit mass, 224Ra activity concentration and 220Rn exhaled per unit mass also present strong correlations. The positive correlations observed indicate a good predictive power of 226Ra and 222Rn data for estimating 224Ra and 220Rn in the study area.

Influence of surface weathering processes on exhalation and emanation data

The coefficient of variation observed in Table 3 indicates a lower variability of 224Ra activity concentration compared to 226Ra activity concentration, as well as a lower variability of 220Rn exhaled per unit mass and emanation coefficients compared to 222Rn values. These results contrast with observations made by Domingos et al. (2021) for granites sampled in the Central Iberian Zone (Portugal), where 224Ra and 220Rn were generally more variable than 226Ra and 222Rn, respectively. Given that this could indicate that the distribution of U and Th isotopes and their progeny in the study area might be constrained by differences in the relative mobility of these isotopes, further studies were conducted to determine the cause of the variability of the radiological variables studied.

The mobility of U is generally greater than Th in surface environments because U is soluble in the hexavalent state (U6+ or UO22+; e.g., Boyle, 1982; Salminen et al., 2005; Regenspurt et al., 2010; Boekhout et al., 2015). Warm and humid paleoclimate conditions are associated with higher U mobility, due to leaching of U (e.g., Ruffell & Worden, 2000; Ruffell, 2016). These climate conditions are present in the Huila Plateau where four types of climates are recorded according to the Köppen–Geiger classification (e.g., Peell et al., 2007), namely: Cwb (Oceanic subtropical highland; West region), Cwa (subtropical humid; East region), Aw (tropical savanna; Northwest region) and BSh (warm semi-arid; South region). Thus, the higher variability of 226Ra and 222Rn may reflect the higher mobility of their parent isotope, suggesting a relevant control of surface weathering processes in the distribution of 226Ra and 222Rn in the study area. High 224Ra/226Ra ratios may also be connected to the climate conditions prevailing in the SW region of Angola (e.g., Burgess et al., 2004; Huntley, 2019).

To investigate the influence of surface weathering processes in the radiological properties of the bedrock, a borehole was executed in a soil profile with a thickness of 33 m developed over an Eburnean granitoid. About 6 samples were retrieved at specified depths in the borehole. The main results of the chemical and radiological analyses performed are shown in Fig. 5.

Fig. 5
figure 5

Variation of chemical and radiological properties with depth (in m) through a borehole executed on a soil profile developed over an Eburnean granitoid. Results are shown for Loss on Ignition (LOI), total iron content (Fe2O3 in %), Magnesium (MgO, %), Titanium (TiO2, %), 224Ra and 226Ra activity concentration, Th and U concentration (in ppm) measured by XRF, 222Rn and 220Rn exhaled per unit mass (EX in Bq/kg), and 222Rn and 220Rn emanation coefficients (EM in %)

Full size image

Loss on ignition (LOI) is a common proxy for chemical weathering due to its relationship with the volatile content (e.g., Gupta & Rao, 2001; Domingos & Pereira, 2018). LOI increases toward the surface, signposting a higher volatile content, which indicates a higher degree of chemical alteration toward the surface (Fig. 5).

224Ra and 226Ra activity concentration, Th and U concentration, 222Rn and 220Rn exhaled per unit mass and emanation coefficient increase until a 15 m depth and then, decrease in depth toward the slightly weathered bedrock sample retrieved at the end of the borehole. Similar trends were observed for the total iron (Fe2O3), magnesium (MgO) and titanium contents (TiO2). These results indicate a concomitant accumulation of 224Ra and 226Ra with Fe2O3 and TiO2 in ferralic horizons as a result of long and intensive weathering processes (e.g., Preetz et al., 2009). These horizons are typically associated with residual concentrations of kaolinite and highly resistant primary minerals (Preetz et al., 2009), which may justify the observed decrease in 222Rn emanation coefficient, that is typically lower in highly resistant U-bearing minerals like zircon (e.g., Sakoda et al., 2011).

Along the soil profile, 226Ra activity concentration ranges from 11.8 to 38.9 Bq/kg (with a coefficient of variation of 40%), showing a threefold increase at a depth of 15 m from the bedrock (base) level. 224Ra is less variable (with a CV of 29%); nonetheless, a twofold increase is also observed at a 15 m depth. 222Rn and 220Rn exhaled per unit mass vary similarly (with a CV of 44 and 45%, respectively), being highest at a 9 m depth, where they present a threefold increase compared to the bedrock base level. The 222Rn emanation coefficient is more variable than 220Rn (CV of 26% compared to 18%), being lower in two scenarios: (1) closer to the surface (likely due to leaching of the parent isotopes which also show lower concentrations closer to surface), and (2) closer to the bedrock (associated with a lower degree of alteration indicated by LOI). The mobility of 226Ra and 224Ra with depth due to surface weathering processes and subsequent variation of the exhaled and emanated 222Rn and 220Rn corroborate the hypothesis of a strong influence of surface weathering processes in the distribution of these radioisotopes in the study area. Given the higher variability in depth of 226Ra activity concentration and 222Rn emanation coefficient compared to 224Ra activity concentration and 220Rn emanation coefficient, surface weathering processes may also justify the higher variability of the isotopes of this decay chain in the study area.

Influence of geological factors on exhalation and emanation data

Data were grouped by geological unit in Table 5 and Fig. 6. The Chela Group samples were divided according to the Formation sampled. The samples collected in outcrops of the Gandarengos granite were also separated from the remaining EGM samples. The results of the Kruskal–Wallis test are presented in Table 6. Apart from 226Ra activity concentration and the 222Rn emanation coefficient, all variables were statistically significantly different between geological units.

Table 5 Descriptive statistics by geological unit
Full size table
Fig. 6
figure 6

Box and whiskers plots of a 226Ra activity concentration; b 224Ra activity concentration; c 222Rn exhaled per unit mass; d 220Rn exhaled per unit mass; e 222Rn emanation coefficient; f 220Rn emanation coefficient by geological unit. Geological units: PEB—Pre-Eburnean Basement; EGM—Eburnean granitoids and migmatites; EGMG—Eburnean granitoids and migmatites: Gandarengos granite; CGT—Chela Group, Tundavala Formation; CGH—Chela Group, Humpata Formation; CGB—Chela Group, Bruco Formation; LF—Leba Formation; PEMR—Post-Eburnean Mafic Rocks; RGP—A-type red granites and porphyritic rhyolites

Full size image
Table 6 Results of the Kruskal–Wallis test for the radiological variables studied using geological units as categorical variable (N = 52 samples)
Full size table

RGP presented the highest median 226Ra and 224Ra activity concentrations, 222Rn and 220Rn exhaled per unit mass, being followed by the Gandarengos granite (EGMG). The EGMG and RGP 226Ra activity concentrations are similar to values reported by Pereira et al. (2017) in granites and orthogneisses associated with magmatic episodes that occurred before the Variscan orogeny close to relevant structural alignments (e.g., Ferreira et al., 1987), in an anorogenic context. The highest median 222Rn emanation coefficients were, however, observed in the Pre-Eburnean Basement (PEB), followed by RGP, while the highest median 220Rn emanation coefficients were observed in RGP, followed by EGMG.

The lowest median 226Ra and 224Ra activity concentration (4.5 Bq/kg and 8.1 Bq/kg, respectively), 222Rn and 220Rn exhaled per unit mass (0.3 Bq/kg and 0.2 Bq/kg, respectively) were observed in LF. The results for the LF are within the range of values reported by Sêco et al. (2020) in dolomitic limestones outcropping in the Lusitanian Basin (Portugal). The 222Rn emanation coefficient with a median of 7% is also within the values reported by Sêco et al. (2020) for dolomitic limestones.

222Rn and 220Rn exhaled per unit mass were both higher in igneous and/or metamorphic basement (PEB, EGM, EGMG and RGP), and lower in sedimentary (CGT, CGH, CGB and LF) and mafic rocks (PEMR). The median 222Rn emanation coefficient was lower than 20% in all geological units studied apart from PEB. The median 220Rn emanation coefficient was lower than 10% in all geological units studied, averaging 6%. Higher median values for the 222Rn and 220Rn emanation coefficients were also observed in igneous and/or metamorphic units (PEB, EGM, EGMG and RGP), compared to sedimentary (CGT, CGH, CGB and LF) and mafic rocks (PEMR).

220Rn exhaled per unit mass exceeded 222Rn exhaled per unit mass in all samples collected in PEB, CGT, CGB and RGP (see Table 5). This is associated with 224Ra/226Ra ratios exceeding 3, and to a higher 220Rn emanation coefficient in CGB. Over half of the samples from EGM, EGMG and CGH present 220Rn exhalation values higher than 222Rn. This indicates 220Rn exhalation rate should not be disregarded in dose assessment studies, particularly if those materials are used as building materials.

Comparison of indoor and bedrock data

226Ra and 222Rn exhalation and emanation data of a building’s underlying materials such as bedrock or soil have been widely used as predictors for indoor 222Rn concentration (e.g., Singh et al., 2005; Gusain et al., 2009; Appleton et al., 2011; Bossew et al., 2013; Forkapic et al., 2017). The scatterplots of the median indoor 222Rn and 220Rn activity concentration by geological units against the median 226Ra and 224Ra activity concentration, 222Rn and 220Rn exhaled per unit mass and emanation coefficients are presented in Fig. 7. A linear relationship between indoor data and exhalation and emanation data appears to exist due to the particularly high median values observed for the RGP unit (Fig. 7). This suggests 226Ra and 224Ra activity concentration, and 222Rn and 220Rn exhalation and emanation could be useful for predicting indoor 222Rn and 220Rn activity concentration. However, further investigations are needed due to the low sample size (n = 5 geological units). Although the main source of indoor 220Rn are the building materials, correlations between indoor 220Rn and the soil content of 232Th have been acknowledged in houses constructed with local soil materials (e.g., Shang et al., 2005). The relationship between indoor 220Rn and bedrock 224Ra activity concentration and 220Rn exhaled per unit mass observed in Fig. 7 may be due to the fact that local raw materials are often used as building materials in the study area.

Fig. 7
figure 7

Scatterplots of the median indoor 222Rn activity concentration by geological units against the median a 226Ra activity concentration; b 222Rn exhaled per unit mass and c 222Rn emanation coefficient and scatterplots of the median indoor 220Rn activity concentration by geological units against the median: d 224Ra activity concentration; e 220Rn exhaled per unit mass; f 220Rn emanation coefficient. The error bars correspond to the median absolute deviation. Legend: EGM—Eburnean granitoids and migmatites; CG—Chela Group; LF—Leba Formation; PEMR—Post-Eburnean Mafic Rocks; RGP—A-type red granites and porphyritic rhyolites

Full size image

Assessment of the annual inhalation dose

The results of the estimated AID are presented in Table 7. The normal quantile plots for log-transformed data of the AID are presented as supplementary material. The linearity observed in the normal quantile plots for log-transformed AID data indicate that the distribution of AID for both dwellings and workplaces is approximately log-normal.

Table 7 Total annual inhalation dose (AID in mSv/y) due to inhalation of 222Rn and 220Rn and their progeny
Full size table

Table 7 shows the estimated AID for dwellings and workplaces, considering the occupancy factors presented in Sect. 3 of this paper. The AID was also estimated according to building material and the geological units using the occupancy factor estimated for dwellings to assess the influence of these parameters in the AID. The AID ranges from 0.51 to 5.28 mSv/y in dwellings (Table 7), with a median of 1.50 mSv/y, which is lower than the estimated average radiation exposure due to natural sources of 2.4 mSv/y (UNSCEAR, 2010). The estimated mean value of the AID is similar to the values reported by Ramola et al. (2016) who conclude that the health risk due to exposure of radon, thoron and their progeny for that range of values is not significant.

Bahu et al. (2021) report AID values ranging from 2 to 7 mSv/y, with a global mean of 3.4 mSv/y for a similar time of exposure (7000 h). Differences in the mean AID between the results reported in the present work and Bahu et al. (2021) may be due to discrimination between 222 and 220Rn carried out in the present work, given that 220Rn may contribute up to 95% of the TIRC measured (see Sect. 4.1), corroborating the importance of discriminating between radon and thoron in dose assessment studies. In workplaces, given the lower occupancy factor, the AID ranges from 0.06 to 1.85 mSv/y, with a median of 0.26 mSv/y (Table 7).

Higher median AID values were observed in concrete-based constructions, followed by clay and brick constructions (Table 7). Lower AID median values were observed in metal and stone-based constructions. With respect to the geological units, higher median AID values were observed in the RGP unit, which was shown to present the highest median 226Ra and 224Ra activity concentrations, 222Rn and 220Rn exhaled per unit mass (see Fig. 6 and Table 5), beyond the highest median indoor 222Rn and 220Rn activity concentrations (Fig. 4). The lowest median AID was observed in the PEMR unit that also presented low values of 226Ra and 224Ra activity concentrations and 222Rn and 220Rn exhaled per unit mass compared to other geological units (Fig. 6 and Table 5). These results indicate that the characterization of the bedrock units is a useful proxy for dose assessment.

The empirical cumulative distribution function (ECDF) for dwellings and workplaces is presented in Fig. 8. The ECDF plot of the AID shows that the 1 mSv/y effective dose threshold established in the Council Directive 2013/59/EURATOM for the purpose of radiation protection in workplaces is exceeded in 12% of workplaces (5 out of 40). The workplaces where this threshold is exceeded are built over EGM (n = 3) and the RGP (n = 2) units, and the building materials used were filled (n = 2) and hollow bricks (n = 3).

Fig. 8
figure 8

Plots of the empirical cumulative distribution function (ECDF) of the annual inhalation dose (in mSv/y) for a dwellings and b workplaces

Full size image

Conclusions

This study presents 226Ra and 224Ra activity concentration, 222Rn and 220Rn exhalation in representative bedrock samples, an assessment of indoor 222Rn and indoor 220Rn activity concentrations measured in buildings and an estimation of the total AID due to inhalation of 222Rn and 220Rn and their progeny in both dwellings and workplaces located in the SW region of Angola. The major conclusions can be summarized as follows:

  • Indoor radon levels are generally below 300 Bq/m3;

  • Indoor 222Rn and indoor 220Rn are not correlated indicating both must be estimated independently for a proper estimation of the contribution of 220Rn to the AID;

  • The contribution of 220Rn to the TIRC averages 35% but may be dominating, reaching 95%;

  • Indoor 222Rn and indoor 220Rn were not significantly different according to geological unit; however, indoor 220Rn was significantly different depending on the building materials used as well as building usage;

  • The AID has a median of 1.50 mSv/y in dwellings which is lower than the estimated average radiation exposure due to natural sources of 2.4 mSv/y (UNSCEAR, 2010), indicating that the health risks due to exposure of radon, thoron and their progeny in the study area are not significant according to Ramola et al. (2016);

  • The highest indoor 222Rn and 220Rn activity concentrations were observed in buildings located above A-type red granites and porphyritic rhyolites;

  • Clay and concrete-based building materials were associated with higher median 220Rn levels and higher AID estimates, compared to metal and stone-based materials;

  • Higher indoor 222Rn and 220Rn concentrations were observed in dwellings compared to workplaces.

    The analysis of bedrock samples revealed that:

  • 224Ra activity concentration and 220Rn exhaled per unit mass were higher but less variable than 226Ra activity concentration and 222Rn exhaled per unit mass, respectively;

  • A significant variability of radiological properties with depth associated with surface weathering processes was observed;

  • The highest 226Ra and 224Ra activity concentration, and 222Rn and 220Rn exhaled per unit mass were observed in A-type red granites and porphyritic rhyolites, similarly to indoor 222Rn and 220Rn indoor radon concentrations;

  • Statistically significant correlations were observed between the radiological parameters studied.

    The results of this study have several implications for the future development of risk maps in African countries, namely:

  • Both indoor 222Rn and indoor 220Rn must be assessed independently to accurately determine the dose received by the population;

  • The underlying bedrock materials, building usage and building materials should be considered in the design of sampling plans and the development of risk maps;

  • Bedrock radiological data should be useful for estimating the risk of exposure to indoor 222Rn and indoor 220Rn.

Studies are still needed to assess the occupancy factors for a proper assessment of the AID based on the predominant lifestyle habits in the Angolan territory, that are significantly different from the lifestyle habits in developed countries. The estimation of the AID pertaining to workplaces is, however, expected to be accurate. If the 1 mSv/y effective dose threshold established in the Council Directive 2013/59/EURATOM is considered, then 12% of the workplaces studied where this threshold was exceeded would require mitigation measures.