Characterization of uranium in soil samples from a prospective uranium mining in Serule, Botswana for nuclear forensic application

ABSTRACT This study attempts a nuclear forensic characterization of uranium in soil samples from a prospective Serule mine in Botswana. The analysis involves the determination of forensic signatures found in uranium-bearing materials from the mine. These signatures include 232Th activity concentration, isotopic and activity ratios of 232Th/238U and 235U/238U, impurity concentration, rare earth elements (REE), as well as the mineralogy of the area, all of which were determined using both Inductively Coupled Plasma Mass Spectroscopy (ICP-MS) and X-ray Florescence (XRF) analytical techniques. Isotopic ratios determined reveal that there is a significant difference in the isotopic concentration, activity concentration of 232Th, 238U, and 235U, as well as the 232Th/238U isotope ratio between the uranium ores from each mine. The REE/chondrite analysis indicates clear patterns, suggesting that it is possible to use this feature as a unique identifier for Serule uranium ore source.


Introduction
The growing interest in uranium mining in Africa has created a need for implementation of safety and security measures to prevent and respond to the illegal use of nuclear and other radioactive materials. The use of nuclear forensic techniques to create a national nuclear library is one such measure. Similar to traditional forensics, nuclear forensics rely on the knowledge that there are measurable parameters that are characteristic of a particular material (Hutcheon et al., 2013). These parameters are added into a database to enable comparison with analytical results of materials found out of regulatory control to facilitate the attribution process (Moody et al., 2014). In order to determine the origin of unattributed nuclear/radioactive material, a reference database containing signatures must be available (Keegan et al., 2017).
Although Lawrence Livermore group has created a large database, there are samples from African countries that are not part of the database. Very little research has been conducted pertaining to African countries that have uranium deposits (Mathuthu & Khumalo, 2017). Some countries such as South Africa, Namibia, Zambia, and Tanzania have been actively mining uranium as a primary or secondary product (Dasnois, 2012). South Africa has 71 uranium mines, Namibia has 42, Senegal, Tanzania, and Zambia have 10, Botswana has 9 and Nigeria has 2 (Mathuthu & Khumalo, 2017). These countries must observe the International Atomic Energy Agency (IAEA) nonproliferation Treaty, which aims at safeguarding nuclear and other radioactive material to prevent illicit trafficking. Countries should therefore have National Nuclear Forensic Programs aimed at the characterization of nuclear and other radioactive material found in and out of regulatory control (Dunworth, 2013). This characterization involves the determination of chemical and isotopic concentration as well as physical parameters, which form signatures of the origin. Currently, there is no African country with a developed National Nuclear Forensic Library. Although South Africa has started, it does not possess a fully functional nuclear forensic laboratory (Mathuthu & Khumalo, 2017).
This study focuses on the Southern African region of Botswana, which has prospective uranium mining projects in the regions under investigation. The aim of this study is to analyze and characterize soil samples from Serule. The measured parameters will include the isotopic composition, concentration of the rare earth elements and the impurity spectrum. In addition to these parameters, the mineralogy of the samples will be determined. Thus, information and signatures found in this study can therefore be added to a National Nuclear Forensics Library and used as reference data to aid in source attribution of nuclear material found in and out of regulatory control. projects are foresighted. Botswana lies between latitudes 17̊ and 27̊, longitudes 20̊ and 30̊ occupying 581,730 km 2 land area (Wikipedia contributors, 2019). Uranium deposits are primarily sandstone hosted rollfront/calcrete deposits, found in Gojwane, Serule, and Gorgon regions. The samples under investigation were sandstone collected from Serule region of Botswana shown in Figure 1.

Sample collection
In this study, the samples collected were in the soil matrix. Soil samples are easily collected, prepared, and stored, and they serve as proficient sorbents of actinides, lanthanides, and transition-metal analytes. Nuclear analytes sorb on soil by wet and dry atmospheric deposition processes, accidental release, and transport activities. Thirty (30) slightly disturbed soil samples were randomly collected from five sampling areas S5-S9 across the planned uranium mining area. At the sampling points, a bucket auger was used to drill into the ground and the samples were collected at a depth of 60 cm. Each sample was transferred into a polythene zip lock bag, weighed, and labeled accordingly. For detailed information on the sample collection procedures see Ref (Masok et al., 2018). The samples were then transported to the University of Witwatersrand (WITS) in the Republic of South Africa for subsequent measurements and analysis.

Sample preparation
Sample preparation is a significant part of the analytical process. For better precision and an accurate representation, samples need to be homogeneous prior to analyses. To guarantee homogeneity, all the samples collected were taken to the WITS University Geosciences Laboratory for crushing and pulverization. The milling media and grinding discs were washed with ethanol, and in addition, extra pure sand was grounded before milling of a new sample to avoid cross-contamination of samples. Once milled, the samples were then transferred into sealed bags, re-tagged, re-weighted, and transferred for storage awaiting measurement and analysis. Measurements and analysis of the samples were conducted at these two laboratories; WITS Analytical Chemistry Laboratory and WITS Earth Laboratory.

Instrumentation
The analysis required in the study was performed using ICP-MS and XRF analytical methods. Several techniques can be used for elemental analysis; of these techniques, ICP-MS offers a wider range of detection limits, multi-element analysis, and isotopic capabilities and time-efficient. Although highly accurate, ICP-MS cannot be directly used on solid samples but requires samples to be in liquid form. Samples in the soil matrix, such as those under investigation were therefore dissolved prior to measurements. In this study, Agilent Technology, 7700 Series ICP-MS was used for the measurement.
For the complete dissolution of the samples, a microwave system (Anton Paar Multiwave, PerkinElmer Germany) was used. 0.5 g of each sample was transferred into Teflon microwave vessels in preparation for acid digestion. The samples were diluted with deionized water, aqua rigia (a 1:3 mixture of HNO 3 and HCl, respectively), and H 2 O 2 , then sealed and placed in a microwave. The system starts and slowly increases to a maximum of 190°C then drops back over a period of 30 minutes. An aqueous was then obtained. Parameters used in the microwave digestion of milled soil samples into the liquid matrix are listed in Table 1.
For ICPMS measurements, the 0.5 g of each of the samples in the batch was taken, duplicated, and labeled. The samples were then diluted with 10 ml of HCl and 50 ml HNO 3 . The digested samples batch together with the calibration standards (5 ppb, 20 ppb, 50 ppb, 100 ppb, 500 ppb, 1000 ppb), blank samples, and quality control samples (20 ppb, 100 ppb) were loaded into the autosampler for measurements. The batch of samples was introduced to the nebulizer containing plasma (argon) at atmospheric pressure and a temperature of 10 000°C. The plasma ionizes the samples and transfers them through a vacuum interface to the hyperbolic quadrupole mass analyzer which measures elements in the sequence. The separated ions are then detected by the detection system, a fast simultaneous dual-mode detector (of nine orders dynamic range). The detection limits of the ICP-MS system in standard mode are shown in Figure 2.
Enhanced data analysis and reporting were performed using MassHunter software. A total of 40 nuclides were observed in the samples under investigation and results given in parts per billion (ppb). The isotopes measured include 232 Th, 238 U, REE, and impurity elements, which are of particular interest in nuclear forensic studies. The concentrations of the isotopes were provided in units of ppb for the different dilution factors.
For XRF, the samples were taken to the School of Geosciences Earth LAB where they were measured using Malvern Panalytical Axios WDXRF spectrometer. Once the experiment was done and measurements were concluded, the results were then provided for further analysis as discussed below.

Determining the Isotopic concentration of 235 U
The concentrations of 235 U were not determined directly from the ICP-MS measurements. However, the isotopic abundance of uranium is constant in nature (99.27% 238 U, 0.711% 235 U) (Masok et al., 2018;Moody et al., 2014). Therefore, the concentration of 235 U can thus be determined from the concentration of 238 U by applying Eqn. 1.
where C U 235 and C U 238 are the concentrations of 235 U and 238 U, respectively.

Determining the Activity concentration of 232 Th, 238 U, and 238 U
The natural radioactivity of an environment is due to its geological and geographical conditions. These conditions are related to the activity concentrations of Th, U, and K of each rock type and have different levels in soil across the different regions of the globe (Bajoga et al., 2015). For the purpose of this study, the focus is primarily on Th and U. The elemental concentration of these three radionuclides ( 232 Th, 235 U, and 238 U) was converted into activity concentration in Bq.kg -1 using Eqn. 2 (Bajoga et al., 2015):  Figure 2. 7700 series ICP-MS detection limits (Garcia, 2020).
where A E is the activity concentration of the nuclide E in the sample, M E , λ E and f AE are the atomic masses (kg.mol -1 ), decay constant (s -1 ), and the measured elemental concentrations (ppm) of E, respectively. N A is Avogadro's number (6:023 � 10 23 atoms.mol -1 ) and C is a constant with the value 10 6 . The activity concentrations of 232 Th, 235 U and 238 U can also be calculated from their measured elemental concentrations (A =f AE ) using the IAEA recommended conversion factors stated below (Spano et al., 2017; United Nations Scientific Committee on the effects of Atomic Radiation. Unsccear, 1998);

Results and discussion
This study was conducted to find the elemental composition of the uranium mined soil samples and determine the measured characteristic parameters that are relevant to nuclear forensics. In this section, the experimental results were analyzed and the characteristic signatures determined.

Characterization of uranium mined samples
ICP-MS was used to determine the elemental and isotopic concentrations of sample under investigation, while the XRF was applied on the samples to determine the major compounds in the samples. Table 2 shows the concentrations in parts per million (ppm) of 232 Th, 238 U, and 235 U as measured by ICP-MS. The results from Table 2, illustrated in Figure 3, were used to calculate the activity concentrations in Bq.kg -1 of 232 Th, 235 U, and 238 U using Eqn 2 and the results are shown in Table 3 illustrated in Figure 4. Table 4 shows the activity ratios 232 Th/ 238 U and 235 U/ 238 U. These parameters, along with the rare earth elements (REE) profile and impurity spectrum are of interest in nuclear forensic investigations.

Uranium and Thorium composition in the samples
The results for the uranium and thorium isotopic concentrations in ppm for the individual sampling area are summarized in Table 2.
• 232 Th According to literature, the average crustal elemental concentration of 232 Th is in the range 8-12 ppm (Erdi-Krausz et al., 2003). S9 has 232 Th concentration of 6.797 ± 0.544 ppm. This range falls below the average crustal concentration. S5, S6, S7, and S8 range from 8.016 ± 0.641 ppm to 12.578 ± 1.006 ppm and fall within the average range. The mean isotopic concentration in this area was found to be 9.68 ± 2.58 ppm, within the world average.  , 2003). S5, S7, S8, S9 had concentrations ranging from 2.542 ± 0.203 ppm to 3.010 ± 0.241 ppm which fall within the world average range. The remaining sample, S6 had concentration of 5.304 ± 0.424 ppm which is greater than the world average. Figure 2(b,c), shows slight differences in the isotopic concentrations of 238 U and 235 U. This showed that the isotopic concentrations obtained were characteristic of that area and did not vary significantly at different points within the area. Overall, the samples can be characterized by 238 U concentration that is within the world average range with a mean isotopic concentration of 3.40 ± 1.13 ppm, and mean isotopic concentration of 235 U of 0.02 ± 0.01 ppm.

Activity concentration of uranium samples
The activity concentration serves as a valuable parameter in nuclear forensics. Table 3 shows the specific activity concentrations of 232 Th, 238 U, and 235 U for the samples in Bq.kg -1 .
• 232 Th The average activity concentrations of 232 Th in the continental crust and in soil are 44 Bq.kg -1 and 37 Bq. kg -1 (Maxwell et al., 2013). S5, S8, and S9, had activity concentrations of 35.95 ± 1.80 Bq.kg -1 , 32.71 ± 1.64 Bq. kg -1 , and 27.73 ± 1.39 Bq.kg -1 , lower than the continental averages. S6 and S7 had activity concentrations of 49.81 ± 2.49 Bq.kg -1 and 51.32 ± 2.57 Bq.kg -1 with significantly higher values than the continental averages. The mean activity concentration of the area was therefore found to be 39.50 ± 10.53 Bq.kg -1 . The activity concentration graph in Figure 3(a) displayed slight variations, the activity concentrations obtained were therefore characteristic of the area.  (Maxwell et al., 2013). The samples had concentrations ranging from 37.32 ± 1.87 to 65.77 ± 2.29 Bq.kg -1 higher than the continental crust with the exception of S5 and S8 which were found to have activity concentration of 32.87 ± 1.64 Bq.kg -1 and 31.52 ± 1.58 Bq.kg -1 . The mean activity concentration in the area was found to be 42.50 ± 13.98 Bq.kg -1 . The low activity concentrations of 235 U compared to that of 238 U were due to the natural abundance of the isotopes in the earth's crust, i.e. 235 U (0.75%) and 238 U (99.8%). The activity concentration of 235 U had a range of 1.47 ± 0.74 Bq.kg -1 to    (Maxwell et al., 2013) 1.200 0.047 3.08 ± 0.15 Bq.kg -1 and mean value of 1.97 ± 0.65 Bq. kg -1 . The activity concentration shown in Figure 3(b, c) of S5, S6, S7, S8, and S9 had small variations. This showed that the activity concentrations obtained were characteristic of that region and can be added to a nuclear forensic library.

Activity ratio of Uranium samples
The activity ratio of 232 Th/ 238 U shown in column 2 of Table 4 where S9 was distinctly lower than the continental average of 1.2 (Maxwell et al., 2013). S5, S7, and S8 had values that were slightly lower than the continental average. Also, there were significant differences in the values obtained for most of the samples. The 232 Th/ 238 U activity ratio for all the samples ranged from 0.757 to 1.179 with a mean of 0.96 ± 0.20. The activity ratio for 235 U/ 238 U was equal to the continental average of 0.047 for all the samples, this was evident because the natural abundance of 235 U was used to determine its isotopic concentration from the measured isotopic concentration of 238 U.

232
Th/ 238 U activity ratio could be identified as a unique signature that is characteristic of the area. The relationship between uranium and thorium from Serule in Botswana can be considered in terms of the quotient of thorium and uranium. Figure 5 shows the correlation between thorium and uranium with a linear fitting relation type and a correlation coefficient of 0.87. The theoretical expected ration of 232 Th/ 238 U is approximately 3.00 for normal continental crust (Tzortzis & Tsertos, 2004).

Rare earth elements
Samples from uranium mines were analyzed to determine the REE concentrations shown in Table 5. According to literature (Spano et al., 2017), the REE concentrations vary with origin, therefore form part of the fingerprint. The REE data in this study were normalized to examine the statistical variations and thus determine the CN-REE profile of the mine. Figure 6 illustrates these patterns.
The REE pattern shown in Figure 6 indicates a match between all the samples. The shapes of the CN-REE plot were all similar and formed peaks at the same elements (La, Ce, Nd). The REE profile obtained was therefore characteristic of the region that can be added to a nuclear forensic library. Table 6 and Figure 7 present the concentration of the impurity which varies according to the sampling areas. However, the graphs in Figure 6 followed the same trend with the impurity elements displaying peaks at Zr, Ba, P, respectively. Zirconium had the highest abundance in S5, S6, and S7. For S8 and S9, Barium had the highest abundance. Therefore, the elemental concentration of impurities was not characteristic of this region but varies at different sampling points. The spectrum was found to be characteristic of the area and can thus form part of a fingerprint.

Mineral identification and quantification
The mean chemical composition of the samples was investigated and XRF measurements for quarts, aluminum oxide, and hematite were found. Table 7 shows the major minerals identified in the samples.
Quartz (SiO 2 ) was identified as the most dominant mineral ranging from 70.77% for S8 to 82.89% for S5. The abundance of Aluminum oxide (Al 2 O 3 ) ranged from 7.2% for S5 to 14.91% for S6. Hematite (Fe 2 O 3 ) ranged from 1.80% for S7 to 3.07% for S7 and S8. The mean chemical composition was as follows: 74.54% of SiO 2, 11.27% of Al 2 O 3, and 3.96% of Fe 2 O 3 . Other minerals such as MnO, MgO, CaO, Na 2 O, K 2 O, TiO 2 , P 2 O 5 , Cr 2 O 3 were identified and were found to have abundances less than 1%. These results comply with the IAEA guideline (Keegan et al., 2017).  Table 8 and Figure 8 are the results of the activity concentration of 232 Th, 235 U, and 238 U; the rear earth elements (REE); major minerals and the impurity obtained from the soil samples using ICP-MS and XRF measurement techniques. The presence of impurities in soil samples collected in Serule, Botswana, is higher followed by the major mineral and thorium being the highest amongst NORM.    Figure 9 shows a correlation of 0.6576 between REE and impurities, similar observations were reported by other researchers (Keegan et al., 2017) and the results can be used for the forensic identity of NORM associated with the REE and minerals from Serule, Botswana.

Comparison between the current study and the world 238 U and 232 Th activity concentrations Bq.kg −1
The activity concentrations of 238 U for the current study are lower than most of the countries but, higher than the world results. While the activity concentration of 232 Th for the current study is higher than Kubwa, Abuja, and Northern central Nigeria and lower than the rest of the countries as shown in Table 9.

Conclusion
The principal aim of this study was to characterize and identify signatures in uranium soil samples that could be added to a nuclear forensic database. To achieve this, an investigation on the prospective uranium mine 30 soil samples from five sampling points was collected and analyzed. The activity concentration from the current study was compared with the results from other parts of the world as presented in  (Keegan et al., 2017) and are     shown in Table 8. The unique results of the activity concentration of 232 Th, 238 U, and 235 U; Activity ratio 232 Th/ 238 U; CN-REE profile and Impurity pattern (Keegan et al., 2017) were obtained. In terms of the mineralogy of uranium samples from Serule, it was established that even though there were slight variations in the concentrations of the minerals, Quartz was the dominant mineral in all the sites of the region. The data will be stored as sample information if the country approves this storage (Keegan et al., 2017) as sample required. If the data are sensitive and storage is prohibited, the results will be added externally only after obtaining results from comparison or it will automatically be deleted (Keegan et al., 2017). Therefore, approval from Botswana will be requested before the data can be added into a Nuclear Forensic Library and used as reference data.