Radon and thoron radiation exposure of the Angolan population living in adobe houses

This study investigates radon and thoron activity concentrations in adobe houses of Angola. Activity concentrations were recorded by passive detectors in rainy and dry seasons in 40 dwellings located at three areas with different climatic and geological backgrounds (Cabinda, Huambo, Menongue). Regarding seasonal and spatial variations, radon activity concentrations are higher in the rainy than in the dry season and both radon and thoron levels are the highest in Huambo, the central part of the country. The number of adobe houses above certain radon and thoron levels were estimated and the contribution of thoron to the inhalation dose was significant.


Introduction
For a couple of decades, radon has been a subject of research due to its potential risk causing lung cancer. It is considered the second most important cause of lung cancer after smoking [1]. Radon-222 is a radioactive gas in the naturally occurring decay chain of 328 U. Radon is hazardous because of its mobility and due to its 3.8 day half-life what consequently leads to accumulation in closed places [2]. In another hand, radon-220, called otherwise thoron, another isotope of radon from the 232 Th decay chain, has been usually ignored given its short half-life of 55.6 s [3].
Thoron contribution to the inhalation dose has been already proven [4][5][6][7] what further triggered surveys focusing on this isotope [8][9][10][11][12]. Many studies have proven high levels of thoron in houses made of unburnt earthen material (like soil or clay), or adobe, a block made of soil and water dried in ambient conditions in sunlight. In Germany, a study conducted in houses made of clay, the thoron annual inhalation dose was found to be as high as 4 mSv a −1 [3]. In China, a research made in traditional houses proved a thoron contribution up to 56.6% to the inhalation dose [13]. In Hungary, the same contribution went up to 30% in adobe houses [14].
In Africa, research works mostly have been focusing on radon but not thoron. In some countries like Ghana, Nigeria, Kenya and Cameroon [8,11,15,16], an effort is being made to collect reference data based on which national radon action levels can be derived. Nevertheless, a few studies on both isotopes are reported. In Kenya (at the coast region), in houses made of soil, 25% of the surveyed houses showed values above 1000 Bq m −3 of thoron activity concentration and the mean was 652 Bq m −3 [8]. In Cameroon, a study at a high natural radiation area of Bikoue and Ngombas located in the uranium region of Lolodorf reported a contribution of thoron to the inhalation dose by up to 78.5% [17].
Despite the fact that adobe houses are popular in Africa mostly in the sub-Saharan part, research on their natural radiation is scarce. A survey in a modern suburb of Ghana where most houses were built from the 90′s to date shows that 28% of houses are still made of adobe [18]. In Angola, no recorded studies were made on radon and thoron, and especially not in adobe houses where the inhalation doses received by residents are potentially elevated. The main aim of the present study is to evaluate the radon and thoron radiation risk in the adobe house of Angola.

Studied areas
Three geologically and climatically different areas, Cabinda, Huambo and Menongue have been chosen for this study from the North, central and South parts of Angola, respectively (Fig. 1). Basic information of their geology, NORM (Naturally Occurring Radioactive Materials) content in local adobe [19], especially that of the parents of radon and thoron, and their climate are provided bellow separately for the three areas. Information about geological features described in this section was taken from "Notícia Explicativa da Carta Geologica de Angola", a summary of the geology of Angola made by the National Institute of Geology in 1977, reviewed and updated in 1992 [20]. In general about the climate, Angola is a tropical country and only two main seasons are identified, the rainy (November-May) and the dry (June-October) seasons. Weather data detailed are taken for the monitoring period between November 2014 and October 2015 from the "OpenWeather" website history [21], which uses data from local weather stations. The NORM content of adobe  [20]. Geological features on the map are as follows, a) Pleistocene to Cretaceous marine sediments, b) Archean to Proterozoic metamorphic and igneous rocks, c) Belts of the upper Proterozoic (Pan-African of age), d) Sedimentary from Tertiary to Quaternary, e) Mesozoic to Paleozoic sediments building material was previously investigated [19] and below a relevant summary of these data is provided.

Cabinda
The geology of Cabinda province is characterized by various formations aging from Precambrian to Holocene, including Pleistocene marine deposits at the coastal area. The climate is humid tropical. The radon and thoron activity concentration measurements were done at the coastal part of the main city, where the humidity is very high in both seasons around 85% (Fig. 2a). Higher temperatures are registered during the rainy season with an average of 27 °C than in the dry season with an average of 23 °C (Fig. 2b). During the monitored year, total precipitation was 244 mm for the rainy and 2 mm for the dry season (Fig. 2c). Previous results of the NORM content of adobe in this area showed values ranging between 15-45 and 27-49 Bq kg −1 for radium ( 226 Ra) and thorium ( 232 Th), respectively ( Fig. 3; [19]).

Huambo
The geology of Huambo province consists of old (Archean to Proterozoic) metamorphic (gneiss, micaschist, metasediments) and igneous (granite, rhyolite to andesites) rocks. At Huambo city, Palaeocene-Eocene laterites were also mapped. The climate is highly influenced by the altitude, i.e. 1700 m above the sea. In the rainy season, the average humidity is 84% and in the dry season, it is 51% (Fig. 2a). The average temperature during the rainy season is 19 °C and during the dry season is 20 °C (Fig. 2b). Precipitation registered during the monitoring period was 522 mm for the rainy and 79 mm for the dry season (Fig. 2c). This area showed the highest NORM contents in adobe building materials with the following ranges: 49-116 and 39-121 Bq kg −1 for 226 Ra and 232 Th, respectively ( Fig. 3; [19]).

Menongue
The geology of Menongue study area is based on Archean gneiss, Proterozoic rhyolitye-andesite and Tertiary-Quaternary Kalahari sediment formations. Menongue has a semi-desert climate influenced by the desert of Namibe. The humidity is much higher during the rainy (78%) than during the dry season (33%, Fig. 2a). The temperature difference between daytime and night is around 10 °C during the rainy season and 20 °C during the dry season. The average temperature in the rainy season is 21 °C and in the dry season is 22 °C (Fig. 2b). Precipitation data registered during the monitored period gives a total of 573 mm and 24 mm for the rainy and the dry seasons, respectively (Fig. 2c). Results of NORM in adobes ( Fig. 3; [19]) showed 15-56 and 17-57 Bq kg −1 for 226 Ra and 232 Th, respectively.

Sampling strategy
A pair of passive radon and thoron detectors were used separately for the rainy and dry seasons in a one year monitoring period in 2014-2015. The rainy season was recorded by a measurement campaign from the beginning of November 2014 to the end of April 2015, whereas the dry season campaign took from the beginning of May to the end of October 2015. The exposure time was 180 ± 5 days. Fifteen houses per area, in total 45 houses were monitored. Only those results were analysed where values were available about both radon and thoron in both rainy and dry seasons, making a total of 40 dwellings. For the placement of the detectors, two aspects were considered. The first one is the possibility of disturbance of the measurement by the residents or visitors of the surveyed dwellings, and the second one is the efficiency of measuring thoron. The distance of the detector from the wall, therefore, was around 9 ± 1 cm and the height around 2 m (when not placed behind a cupboard). All detectors were placed in living rooms as the only place made available by the owners.

Etched track passive detectors
In this study, to monitor radon and thoron activity concentrations, RADUET type etched track passive detectors were used which are commercially available at several locations of Radosys Ltd. (headquarter at Budapest, Hungary). This is a detector composed of two electroconductive cylindrical plastic chambers containing CR-39 plastic nuclear track detector (PNTD). The mentioned chambers are made in a way that one is restricted to detect thoron and the other can detect both radionuclides. Radon and thoron activity concentrations are determined after the PNTDs are etched and alpha tracks are identified and counted with a proper reading system [22]. These detectors are a result of collaborative work between NIRS, Japan and Radosys Ltd., Hungary. Calibration, detector analysis, reading results and radon and thoron activity concentration calculations together with uncertainty estimation were all performed by the manufacturer, Radosys Ltd.

Inhalation dose estimation
To evaluate the health risk of the population living in the monitored houses, inhalation dose from radon and thoron were estimated based on Eq. 1 [23].
where D i is the estimated annual inhalation dose from radon or thoron (mSv y −1 ), C av is the annual average radon or thoron activity concentration (Bq m −3 ), F eq is the equilibrium factor which is 0.4 for radon [24] and 0.04 for thoron [25], F c is the dose conversion factor, which is 9 nSv/(Bq h m −3 ) for radon and 40 nSv/(Bq h m −3 ) for thoron [24] and finally, O is the annual indoor occupancy time (7012.8 h y −1 ). Note that without direct determination of thoron decay product activity concentrations and equilibrium factors, the thoron doses estimated with the applied average discrete value (0.04) are loaded with high uncertainty.

Statistical analyses
Various statistical methods have been used to allow variable analyses, data comparison and interpretation by the "Origin" and "Statgraphics" software. Basic statistics have been calculated and represented in box and whisker plots. Simple regression was plotted to check the relationship between different variables. Statistical distributions were analysed by the cumulative frequency histogram and tested for normality and lognormality by the Shapiro-Wilk hypothesis test. Lognormality was tested using the natural logarithm of the data. If the P value is less than 0.05, we can reject the idea that the sample comes from the suggested distribution with 95% confidence. For comparison of medians of sample groups and for checking if the differences are statistically significant, the Mann-Whitney hypothesis test was performed. The null hypothesis is that the median of a chosen group of samples is equal to the median of the other one. The alternative hypothesis is that the medians are different. For P values less than 0.05, the alternative hypothesis is accepted, whereas for P value greater or equal to 0.05, it is rejected at the 95% confidence level.

Basic statistics of seasonal indoor radon and thoron activity concentrations
Basic statistics of radon and thoron activity concentrations are summarized in Table 1 and the box and whisker plots of Fig. 4a and b. Averages with corresponding standard deviations and medians in brackets [average ± standard deviation(median)] of radon during rainy and dry seasons respectively are 38 ± 11 (36)  Results of the P values for radon and thoron, respectively, are as follows, 0.0003 and 0.73 in Cabinda, 0.046 and 0.74 in Huambo and 0.067 and 0.92 in Menongue. Accordingly, for radon, differences are statistically significant between the two seasons, except in Menongue. For thoron activity concentrations, the differences between rainy and dry seasons are statistically not significant in all areas.

Annual average indoor radon and thoron activity concentrations
Results of the annual average indoor radon and thoron activity concentrations are summarized in box and whisker plots (Fig. 5a, b)

Statistical distributions
The sample distribution was ascertained by the frequency histogram and Shapiro-Wilk test (Fig. 7a, b). For both radon and thoron, normal distribution was rejected by the Shapiro-Wilk test that shows a P value equal to 0.00055 for radon and 0.000091 for thoron. On the other hand, the test does not reject the hypothesis of radon and thoron coming from a lognormally distributed population with P values equal to 0.14 for radon and 0.09 for thoron. Both tests are performed with 95% confidence.

Seasonal variation of radon and thoron activity concentrations
With the aim of evaluating the effect of climatic conditions on indoor radon and thoron activity concentrations in Angolan adobe houses, the one year survey period was divided into two seasons (rainy and dry). As seen in Fig. 4a and proven by Mann-Whitney tests (P value equals 0.0003 in Cabinda, 0.046 in Huambo), medians of the indoor radon during the rainy season show statistically significantly higher values than in the dry season in Cabinda and Huambo. The elevated indoor radon activity concentrations during rainy seasons can be explained by the influence of precipitation. Adobe building material, showing soil properties, has the ability to absorb water [14]. It is also important to emphasize that most indoor radon comes from the ground under the dwellings. The moist content of pore spaces both in the adobe and the soil may increase the emanation coefficient and consequently the exhalation of radon [26,27], thus the level of radon activity concentration in the rainy season. Another factor might be the behaviour of the residents who tend to seal their homes when the weather is worse. Results of the present study are consistent with one carried out in Ghana [28] and in Cameroon [12] where radon levels were also elevated during the rainy season (medians of rainy and dry seasons were respectively found to be 39.3 and 26.9 Bq m −3 in Ghana and 131 and 80 Bq m −3 in Cameron). Both studies explained the difference by the common practice of inhabitants of closing windows during rainy season. Regarding Menongue, the difference is statistically not significant (P value equals 0.067, Fig. 4a). The possible explanation of this variance is that having a semi-desert climate, diurnal temperatures of Menongue are very high (above 35 °C) causing rapid evaporation of the water accumulated in the soil. Consequently, only a small amount of water can accumulate in the pore space influencing radon exhalation. The inhabitants are also less probable to close their windows. Although temperature could play a principle role in seasonal variations of the indoor radon levels, the differences in the average temperatures among seasons are generally very small in Angola as it was the case during the monitored period (4 °C in Cabinda and 1 °C in Huambo and Menongue, Fig. 2b). A comparison of the medians of the annual averages of the indoor thoron, on the other hand, does not show significant differences between the seasons (P value equals 0.73 in Cabinda, 0.74 in Huambo and 0.92 in Menongue; Fig. 4b).

Spatial variation of radon and thoron activity concentrations
Box and whisker plots in Fig. 5a and b show the annual average indoor radon and thoron activity concentrations at the three studied areas. As seen in results the Mann-Whitney (P values of the different comparisons for radon and thoron are, respectively, 0.000009 and 0.021 for Huambo versus Cabinda, 0.000033 and 0.000064 for Huambo versus Menongue, and 0.000013 and 0.025 for Menongue versus Cabinda) test ascertains that activity concentrations in Huambo, the central part of the country are the highest among the three areas. These results can be explained by the differences in geology. In fact, in this area, the geology is composed of metamorphic (gneiss, micaschist, metasediments) and igneous (granite, rhyolite to andesites) rocks which can contain the higher abundance of mother nuclides of radon and thoron in the area. In the previous study of the authors [19] the highest values of 226 Ra and 232 Th in adobe building materials from the same dwellings (116 and 121 Bq kg −1 respectively; Fig. 3), confirm the above stated. Radon and 226 Ra [19] show a correlation coefficient of 0.46 with a P value of 0.0042 (Fig. 8a). This proves that there is a relationship between radon and its source. However, it worth taking into consideration that the correlation would be better if we had radon coming specifically from the building material. As known, most indoor radon comes from the underlying dwelling foundation and some studied dwellings are without coverage on the floor. In these cases, radon coming from the floor may considerably contribute to the indoor levels. This result is consistent with those of Saidou et al. [17] who studied adobe dwellings from Cameroon. Th. The activity concentrations of 226 Ra and 232 Th were measured in adobe building material samples taken from the same dwellings where the indoor radon and thoron measurements took place. These data and more details of the methodology were published in a previous study of the authors [19] Contrary to the case of radon, there is not a statistically significant relationship between thoron and its source, 232 Th [19], which show a correlation coefficient of 0.29 and a P value of 0.08 (Fig. 8b). This is even more surprising considering that at 9 ± 1 cm distance from the adobe walls and at a 2 m height, where the detectors were placed, the only source of thoron is the building material itself (in case of radon the contribution of the soil cannot be excluded). The explanation can be that thoron measurements are more disturbed by other factors like ventilation and humidity than radon, because of its shorter half-life [14,29]. The equilibrium factor of thoron is generally influenced by indoor atmospheric parameters such as ventilation more strongly than that of radon [3]. Because thoron diffuses very slowly, any ventilation can prevent its accumulation near the walls which disturbs the measurements. Thus, it is hard to have the full amount of thoron coming from the building. For this reason, the correlation between thoron and the source can be even less significant than between radon and its source.

Number of dwellings with activity concentrations above recommended values
Both radon and thoron activity concentrations show lognormal distributions (Fig. 7). Therefore, a lognormal probability plot for both isotopes can be drawn with the 95% confidence intervals (Fig. 9). The plot for annual radon average therefore indicates that there is a 95% probability, that 10 to 32% of adobe houses in Angola, assuming the representativity of our sampling, are above the recommended limit for indoor radon concentration of 100 Bq m −3 (Fig. 9a; [30]. However, none of the houses exceed the stronger recommendation of 300 Bq m −3 annual average radon activity concentration. For thoron, the lognormal probability plot (Fig. 9b) shows that there is a 95% confidence, that 59 to 81% of adobe houses in Angola are over 100 Bq m −3 and 6 to 28% are over 300 Bq m −3 . Despite the fact that many studies are reporting high levels of indoor thoron [4-7, 17, 31] (Table 2), as well as its high potential contribution to the inhalation dose, no limit concentration was yet recommended for thoron. Therefore, the 100 and 300 Bq m −3 values serve as reference values and not as any kind of limit or recommended values. For the appropriate understanding of these data, one must also note that the thoron activity concentrations were measured at a fixed distance, 9 ± 1 cm away from adobe walls and that it is expected to show high inhomogeneity indoors.

Inhalation radiation risk in Angolan adobe houses
According to the results in the present study (Fig. 6), the annual radon inhalation dose is particularly significantly higher in Huambo (2.83 ± 0.8 mSv y −1 ) than the worldwide average (1.15 mSv y −1 [32]). The other two studied areas are below the reference value. Moreover, when thoron is included in the estimation, total inhalation dose is about two times higher (Fig. 6). As already mentioned by Tokonami et al. 2001 [4], thoron may have an important contribution to the inhalation dose. In the present study, the average thoron contribution is calculated to be 50%, which shows the potentially high significance of this Other studies also showed significant thoron contribution to the inhalation dose. In traditional (made of soil) Chinese houses, thoron contributed up to 57% to the total dose [13]. Hungarian adobe houses showed an average thoron contribution of 30% [14]. A study on radon and thoron in houses made of soil bricks carried out in Cameroon showed a thoron contribution of up to 79% [17]. From the referred results, it is clear that radon is not enough to access the total annual inhalation dose in adobe or similar houses, and that the total radon and thoron radiation risk in Angolan adobe houses is not negligible and worth further considerations.

Conclusion
Seasonal and spatial variations of the radon and thoron activity concentrations in Angolan adobe houses at three, geologically and geographically different study areas were detected in this study. Radon activity concentrations are higher during the rainy season than in the dry season, whereas no statistically significant differences were detected for thoron. Annual average radon and thoron activity concentrations at around 9 cm from adobe walls are the highest in Huambo (central part of the country). Seasonal variations are essentially most probably due to precipitation variability, and spatial variations are mostly due to differences in geology and the mother nuclide content of adobe. Radon and thoron activity concentrations are sampled from a lognormally distributed population based on hypothesis testing. Regarding radon activity concentrations, 10 to 32% of adobe houses in Angola are above the recommended limit of 100 Bq m −3 and when it comes to thoron activity concentrations, 59 to 81% of adobe houses in Angola are expected to be over 100 Bq m −3 , and 6 to 28% of them are probably over 300 Bq m −3 (as reference values). The inhalation dose in Huambo (4.5 ± 1.5 mSv y −1 ) is estimated to be above the worldwide average (1.15 mSv y −1 ). The contribution of thoron to the inhalation dose is shown to be 50% which refers to its significance even when considering the uncertainties of the calculation. The total inhalation dose received by residents in Angolan adobe houses is not expected to be negligible and it worth further investigations. The results of this study are valuable information on radon and thoron in Africa in general and in Angola in particular.