The Role of Continental Crust in the Formation of Uraninite-Based Ore Deposits

This study reports trace element abundances and Pb, Sr, and U isotopic signatures of uraninite from a variety of ore deposits in order to establish baseline forensic information for source attribution of raw, natural U-rich samples. Trace element concentrations, reported here, provide insights into uraninite crystal substitution mechanisms and possible crustal sources of U, including mobility of trace elements between pristine versus altered fractions. Spatially resolved laser ablation (LA) multicollector (MC) inductively coupled plasma mass spectrometry (ICP-MS) analyses were used to determine secondary 207Pb-206Pb isochron ages, and these were validated by corroborative results obtained by solution mode (SM) MC-ICP-MS for the same sample. Secondary Pb-Pb isochron ages obtained, in this study, indicate that uraninite alteration occurs shortly after ore mineralization. Initial 87Sr/86Sr values correlate in general with host craton age, and therefore suggest that uraninite ore formation is closely linked to the nature of the bedrock geology. The δ238U values are explained by invoking multiple physicochemical conditions and parameters such as temperature, nuclear field shift, oxidation, and source rock composition. The δ234U values indicate that the uraninites, investigated here, have undergone recent alteration, but the latter has not perturbed the Pb-Pb secondary isochron ages.


Introduction
Uranium deposits of economic interest are located on most continents and are classified according to their host rock lithology, nearby tectonic structures, and mode of alteration [1]. The physicochemical conditions prevailing during U ore formation are complex and evolve continuously with time as evidenced by the occurrence of several generations of uraninite within one deposit (e.g., [2]), which is the main constituent mineral. Trace and major element incorporation by uraninite (UO2+x) is extremely complex and results in a diverse chemical composition; thus, a more representative formula is (U 4+ 1−x−y−z−uU 6+ xREE 3+ yM 2+ z□ 4− u)O2+x−0.5y−z−2u, where M are divalent metal ions and □ represents a vacancy [3]. Uraninite has been the focus of numerous past investigations to understand its variable chemical nature, and because it is the most important raw material used for the production of fuel destined for nuclear reactors.
Natural uranium has three main isotopes, 234 U, 235 U, and 238 U, of which only 235 U is fissile. The latter feature of uranium has prompted the illicit trafficking of this material for the past several decades, in particular subsequent to the demise of the former Soviet Union [4]. In efforts to combat shift is a mass-independent effect used to predict the dependence of isotopic fractionation on temperature and is particularly important for heavy elements [23,24]. It has been argued previously that the difference in 238 U/ 235 U ratios between low-and high-temperature U ore deposits is related to the temperature dependence of the nuclear field shift [25,26]. As temperature increases, the magnitude of the isotope shift decreases [26]. Oxidation states also affect uranium mobility. In contrast to other oxidation dependent isotopes (e.g., Mo), the heavier 238 U isotope is favored in the lower oxidation state [24], thus, the U 4+ state is linked to higher 238 U/ 235 U values. At low temperature, oxidizing fluids mobilize U 6+ until it encounters a reducing environment resulting in the crystallization or re-crystallization of uraninite as the insoluble U 4+ state. Uranium leached from ore during alteration/recrystallization results in minerals enriched in 234 U [9], and therefore the 238 U/ 234 U ratio is primarily used to evaluate recent fluid alteration events. Given the short half-life of 234 U, the 238 U/ 234 U ratio yields a result consistent with secular equilibrium if the deposit or mineral has not experienced alteration within the past 2.5 million years [9].
This study reports trace element abundances and Pb, Sr, and U isotopic ratios of uraninites (n = 15) from ore deposits within North America and one from the Democratic Republic of the Congo in relation to establishing baseline signatures for nuclear forensic applications. Secondary Pb-Pb isochron ages for several uraninite samples were obtained by both SM-and LA-multicollector (MC) ICP-MS. The reported ages and isotopic ratios are discussed in relation to providing insights into possible source rock compositions and environmental conditions present during the crystallization of uraninite.

Sample Descriptions
Uraninite samples, examined here, are from the "Ewing Collection" housed at the University of Notre Dame. Fourteen uraninite samples are from locations throughout North America and one from Shinkolobwe, Democratic Republic of the Congo (Table 1). Several samples listed in Table 1 have been examined previously for their trace elements (i.e., REEs) and Sr isotope compositions [27,28] with the exception of those from Mitchell, Marshall 2, and Moonlight. None of the samples listed in Table 1 have previously been analyzed for their U or Pb isotope compositions. The uraninite from the counties of Yancey and Mitchell (NC, USA) are both hosted by the Spruce Pine pegmatite, which crosscuts Precambrian age interlayered mica and amphibole gneiss and schist [29]. The uraninite from Ruggles Mine in Grafton County (NH, USA) occurs as dendritic intergrowths within a pegmatite hosted by Devonian-aged Littleton Formation; the latter consists of quartz-mica schist, quartzite, amphibolite, and other high-grade metamorphic rock [30][31][32].
The Great Bear uraninite is hosted by late Aphebian-aged units comprised of pumice-dominated pyroclastic flows with subordinate ash and plutons within the Great Bear Lake region of Northwest Territories (Canada). This region is known for the occurrence of several U-Ag-Bi-Cu-Co-Ni-As minerals that are found within quartz and carbonate gangue [33]. Several remobilization events have been recorded at the Echo Bay location that are associated with a diabase intrusion [34,35]. The Shinkolobwe Mine (Democratic Republic of the Congo) is part of the Shaban area of the Katanga system, known for ore deposits of U, Cu, Co, and Ni. Fractures that occur within dominantly siliceous dolomite, and dolomitic and carbonaceous shales partially affected by Mg-metasomatism are host to the uraninite mineralization [34,35].
Four uraninite samples (Marshall 1, 2, 3, and 4) are investigated from the areas of Marshall Pass and Sargents, Colorado (USA). Hydrothermal activity resulted in the formation of colloform and finegrained uraninite within fault-controlled veins and breccia zones of Pennsylvanian-aged limestone proximal to intersections of Proterozoic and Paleozoic sequences [36,37]. Billiken Lode and Jefferson uraninite are both from Jefferson County (CO, USA) from deposits that contain complexly folded and faulted Proterozoic metasediments, which host uraninite and accessory minerals ankerite, quartz, calcite, and potassium feldspar [35,36].
The uraninite within the collapse breccia structure from Orphan Lode (Grand Canyon National Park, AZ, USA) is found disseminated within the Pennsylvanian and Permian host rock matrix of limestone, sandstone, and shale [38,39]. Moonlight Mine ore (Navajo County, AZ, USA) occurs as both grains and cement and is found within an Upper Triassic channel deposit of sandstone and conglomerate. Skyline Mine uraninite (Monument Valley, UT, USA) is located within tabular sandstones of Upper Triassic age of the Chinle Formation [40].

Uranium Deposits
Uraninite occurrences on a global scale are categorized by their respective deposit type and subtype using the classification scheme outlined by the International Atomic Energy Agency (IAEA) [2]. Deposit types that were classed using an older identification system have been renamed, for example, "vein type" is no longer a utilized term. A total of thirteen types are investigated, here, and are listed below with respect to their temperature of formation (from high to low): intrusive, graniterelated endogranitic, granite-related perigranitic, polymetallic iron oxide breccia complex, volcanicrelated, metasomatite, metamorphite-monometallic, metamorphite-polymetallic, Proterozoic unconformity basement-hosted, Proterozoic unconformity contact, palaeo-quartz-pebble conglomerate, collapse breccia pipe, and sandstone (Table 1). Several U ore deposit types are not listed in Table 1, however, these are described below since they are cited in the discussion section for comparative purposes.
Intrusive deposits usually form as a result of either partial melting or fractional crystallization [41]. Intrusive deposits included in this study are from pegmatites and are considered high-grade ore [2]. Granite-related endogranitic deposits are located within veins or disseminations within the granite. Granite-related perigranitic deposits originate in veins surrounding the granitic plutons [2]. Low grade ore is produced in polymetallic iron oxide breccia complex deposits; these are broadly linked to iron oxide-copper-gold deposits. Volcanic-related deposits occur within or near volcanic calderas filled with volcanic sediments and consist of medium grade ores. Metasomatite deposits, which are associated with low to medium grade ore, are related to Na-or K-metasomatism.
Metamorphite deposits involve fluids characterized by higher temperatures, 200 to 400 °C [24], and these are variable in grade, size, and tonnage. Metamorphite deposits that are structure-bound (examined here) occur as monometallic veins associated with traces of other metallic minerals, or as polymetallic veins found with Co, Cu, Fe, Mo, Ni, Pb, Zn, Ag, and As metallics [2]. Uraninite from metamorphite deposits, examined here, that formed within hydrothermal environments are associated with lower temperatures of formation than typical metamorphites. Uranium deposits that occur immediately above, below, or span an unconformable contact that separates Archean-Paleoproterozoic crystalline basement rock from Proterozoic red beds are defined as Proterozoic unconformity deposits. Unconformity contact deposits are situated directly above the unconformity at the base of the overlying sediment, whereas basement-hosted deposits are found below the unconformity in metasedimentary rocks [2].
Palaeo quartz-pebble conglomerate deposits consist of uraninite and brannerite hosted in pyriterich quartz-pebble conglomerates [2]. Collapse breccia pipe deposits produce high-grade ore within cylindrical, vertical filled pipes in sedimentary basins; currently the only known examples are within the Grand Canyon region of the USA sandstone deposits, which form at ambient temperatures when ground water removes and transports the mobile U 6+ through the sandstone until a reductant is encountered; consequently, U 4+ minerals form, such as uraninite [24]. Several episodes of mobilization and crystallization can occur in a single sandstone deposit resulting in several generations of uraninite.

In Situ Pb Isotope Ratios by LA-MC-ICP-MS
Small portions (~1 cm 2 ) of uraninite were cut and placed fresh surface down into a 1 inch round mount that was then filled with epoxy and cured before being polished. X-ray fluorescence (XRF) elemental maps were generated using an EDAX Orbis Micro EDXRF with the following conditions: 40 kV voltage, 300 µA, and 100 µs dwell time. XRF maps were used to identify viable areas with sufficient Pb abundances for laser ablation (LA) MC-ICP-MS measurements. In situ Pb isotopes were obtained using a Nu Plasma II MC-ICP-MS instrument located within Midwest Isotope and Trace Element Research Analytical Center (MITERAC) at the University of Notre Dame. Analyses were conducted using 25 µm spot sizes at 8 Hz and corresponding energy density of ~10 to 11 J/cm 2 . Groupings of 5 ablations of samples were bracketed by the amazonite feldspar in-house standard to monitor and correct for instrumental drift and mass bias (procedure after [42,43]).

Bulk Sample Trace Element Abundances by ICP-MS
For each sample of uraninite, aliquots were separated by hand picking, then powdered and digested for SM-ICP-MS and SM-MC-ICP-MS analyses. When possible, the samples were separated into "pristine' and "altered" fractions based on color, morphology, and luster. For example, lustrous black portions were considered pristine sections of samples, whereas yellow and orange fractions were deemed altered uraninite or containing secondary-U minerals. Approximately 50 mg of powdered sample was placed into 15 mL, precleaned Savillex © Teflon beakers for digestion using ~4 mL of high purity, concentrated HNO3 produced with the use of a Savillex © DST-1000, sub-boiling, acid purification system. Aliquots from the digested solutions were used for both trace element concentrations and isotopic measurements. Solution mode analyses were conducted on a Nu Instruments Attom high resolution (HR) ICP-MS operating at medium mass resolution (M/ΔM ≈ 2500). A standard-spike addition method was employed to correct for matrix effects and instrumental drift (after [44]). Trace element abundance determinations using the analytical method, adopted here, were validated by Balboni et al. [28] based on repeated analyses of CUP2 (uranium oxide concentrate) certified reference material, both with and without chemical separation of the U-rich matrix via ion exchange chromatography.

Bulk Sample Pb, Sr, and U Isotope Ratios by SM-MC-ICP-MS
Pb was isolated by processing digested samples through ion exchange chromatography using AG1-X8 (200 to 400 mesh) resin following the procedure by Manhès et al. [45]. The analytical protocol for determining the Pb isotope compositions followed that of Simonetti et al. [46]. The purified Pb aliquot was spiked with a NIST SRM 997 thallium standard solution (2.5 ppb) prior to aspiration into the MC-ICP-MS instrument. Pb and Tl isotopes and 202 Hg were measured using seven Faraday cups on the Nu Plasma II MC-ICP-MS instrument. The 205 Tl/ 203 Tl was measured for monitoring the instrumental mass bias (exponential law, 205 Tl/ 203 Tl = 2.3887), and 202 Hg was recorded for the 204 Hg interference correction on 204 Pb. Prior to sample introduction, a baseline measurement of the gas and acid blank ("on-peak-zero") was conducted for 30 s. Data acquisition involved 2 blocks of 25 scans (each scan was 10 s). A 25 ppb solution of the NIST SRM 981 Pb standard (spiked with 6 ppb NIST SRM 997 Tl standard) was also analyzed periodically throughout the analytical session. Repeated measurements (n = 4) of the NIST SRM 981 + Tl standard solution yielded the following average values and associated (2σ) standard deviations: 206 Pb/ 204 Pb = 16.935 ± 0.003, 207 Pb/ 204 Pb = 15.488 ± 0.002, and 208 Pb/ 204 Pb = 36.686 ± 0.008.
For Sr separation, the ion exchange columns contained 1.7 mL of 200 to 400 mesh AG50W-X8 resin following a modified procedure by Crock et al. [47]. The resin bed volume was cleaned with high purity 6N HCl and 18 MΩ cm −2 H2O and, then, conditioned with 5 mL of high purity 2.5N HCl. The sample aliquot was, then, loaded onto the resin in 0.25 mL of 2.5N HCl, washed with 9.75 mL of 2.5N HCl, and eluted with 4 mL of 2.5N HCl. Subsequent to ion exchange separation, the Sr-bearing aliquots were dried down and later taken up in 2% HNO3 (~2 mL) and aspirated into the ICP torch using a desolvating nebulizing system (DSN-100, Nu Instruments Inc., Wrexham, United Kingdom). Strontium isotope measurements were conducted using a Nu Plasma II MC-ICP-MS instrument following the protocol outlined in Balboni et al. [27]. Strontium isotope data were acquired in static, multicollection mode using 5 Faraday collectors for a total of 400 s, consisting of 40 scans of 10 s integrations. Accuracy and reproducibility of the analytical protocol were verified by the repeated analysis of a 100 ppb solution of the NIST SRM 987 strontium isotope standard during the course of this study, which yielded an average value of 0.71025 ± 0.00004 (n = 4). The ɛSr values, reported here, are calculated using 87 Sr/ 86 Sr initial ratios for the samples and the following equation: where BABI (basaltic achondrite best initial) 87 Sr/ 86 Sr = 0.69908 [48]. Uranium was purified from digested uraninite samples using UTEVA resin as outlined in Martinelli et al. [49]. Two Faraday collectors were used to measure the 238 U and 235 U ion signals, whereas the 234 U ion signal was recorded on a discrete dynode secondary electron multiplier. Ion signals were collected for 40 scans of 10 s integrations each (400 s total). Analyses were conducted using a standard-sample bracketing technique. Instrumental mass bias corrections employed the exponential law and the certified 238 U/ 235 U, 238 U/ 234 U, and 235 U/ 234 U ratios for the CRM 112A standard (New Brunswick Laboratory, Argonne, IL, USA). The internal in-run precision (2σ level) was orders of magnitude lower than the calculated external reproducibility based on the repeated measurements of the CRM 112A standard, and thus the latter uncertainties were reported, here. Delta values for the U isotope measurements were determined using the following equation:  [50]. Additionally, method validation was established by Spano et al. [51] by repeated measurement of uranium standards IRMM-184 (natural U) and IRMM-185 (enriched 235 U ~1.97%) using the analytical protocol, adopted here, and these yielded an external reproducibility (2σ level) of between 0.73‰ and 0.99‰, 13.6‰ and 3.4‰, and 9.3‰ and 5.6‰ for the 238 U/ 235 U, 234 U/ 238 U, and 235 U/ 234 U ratios, respectively. Table 2 lists the trace element abundances determined for the uraninite samples, investigated here. Figure 1 illustrates various trace element ratios and initial 87 Sr/ 86 Sr of the uraninite samples, investigated here, as a function of their deposit type. These ratios indicate the relative preferential incorporation of trace elements into uraninite, and are compared to canonical values for lower, middle, and upper crust [52], and crustal sediments [53,54]. Figure 1a indicates that uraninite samples from metamorphite deposits are characterized by the highest initial 87 Sr/ 86 Sr values followed by (in order of decreasing ratios) those from metamorphite-hydrothermal, intrusive, collapsed breccia, and sandstone. Compared to the La/Yb values for continental crust [52], those for most of the uraninite samples, investigated here, are lower ( Figure 1a). For the same uraninite sample, analyses of altered fractions yield higher La/Yb values as compared with their pristine counterparts (open vs. filled symbols in Figure 1a and Table 2). Figure 1b indicates that Zr abundances are higher relative to both Nb and Hf contents for the uraninite samples, investigated here; relative to continental crust, uraninite samples, studied here, have comparable Zr/Nb ratios but higher Zr/Hf values (Figure 1b). Uraninite samples that are associated with abundant zircons (Ruggles and Mitchell intrusive pegmatite deposits) exhibit higher Hf contents ( Table 2) and, consequently, lower Zr/Hf values. Figure 1c demonstrates that Zr/Hf ratios are at least an order of magnitude higher than their corresponding Rb/Cs values. Ruggles and Mitchell uraninite samples contain similar Zr/Hf ratios relative to continental crust and sediments ( Table 2). The data in Figure 1d (and Table 2) indicates that altered fractions of the uraninite samples contain higher Nb and Ta abundances relative to their corresponding pristine aliquots.

Secondary Pb-Pb Isochron Ages
Pb isotope ratios obtained by SM-and LA-MC-ICP-MS are listed in Tables 3 and 4, respectively. Selected secondary Pb-Pb isochrons based on both SM-and LA-MC-ICP-MS-generated data were produced using Isoplot (v. 4.0; [55]) and are shown in Figure 2 (see Figure S1 for all other Pb-Pb isochrons for remaining samples). Reported ages for the pegmatite host to the uraninites from Mitchell and Yancey counties are Paleozoic in age and range between 252 and 542 Ma [29]. A U-Pb age of 304 Ma (no associated uncertainty) is reported for uraninite from the Ruggles Mine [31]. In this study, uraninite from Mitchell and Ruggles yield Pb-Pb secondary isochron ages of 370 ± 120 Ma and 324.6 ± 8.1 Ma, respectively. In comparison, LA-MC-ICP-MS analyses of uraninite yield ages of 370 ± 110 Ma for Mitchell, 370 ± 1600 Ma for Yancey 1, and 327 ± 110 Ma for Ruggles. Given their proximal geographic location to one another in the Appalachian Mountain belt, a combined secondary Pb-Pb isochron including all three sites results in uraninite ages of 324.5 ± 1.2 Ma and 323 ± 21 Ma for SM-and LA-MC-ICP-MS methods, respectively ( Figure 2).
Uraninite from the Great Bear area is the oldest documented sample investigated in this study. The Great Bear region is identified by several uranium deposits with variable ages; mineralization at Echo Bay has been dated between 1500 ± 10 and 1424 ± 29 Ma by U-Pb dating [33]. Here, we report secondary Pb-Pb isochron ages of 1509 ± 19 and 1444 ± 61 Ma for SM and LA-MC-ICP-MS-based data, respectively ( Figure 2). Using SM-MC-ICP-MS, the secondary Pb-Pb isochron age of −68 ± 7.2 Ma for the uraninite from Jefferson does not agree with the reported U-Pb age of 69.3 ± 1.1 Ma for the Jefferson County Schwartzwalder deposit [57]. Similarly, the LA-MC-ICP-MS yielded a secondary Pb-Pb isochron age of 97 ± 20 Ma, which, given the associated uncertainty, overlaps with the age provided by Ludwig et al. [57]. The Billiken uraninite, also located in Jefferson County, yields a secondary Pb-Pb isochron age of 537 ± 18 Ma  [38]. LA-MC-ICP-MS analyses of the Shinkolobwe uraninite has a secondary Pb-Pb isochron that yields an age of 617 ± 1600 Ma, which overlaps (given the large associated uncertainty) with published ages of 670 ± 20 Ma and 620 ± 10 Ma [58], while Decree et al. [34] reported an age of 652.3 ± 7.3 Ma for Sinkolobwe uraninite.

Sr Isotope Data
Sr isotope results, obtained here, are listed in Table 5. Sr concentrations for uraninite samples, investigated here, range from 60 to 1089 ppm ( Table 2). All initial 87 Sr/ 86 Sr ratios are well above the present-day Bulk Earth value of 0.7045 indicative of a significant crustal signature ( Table 5). The Rb-Sr isotope results yield erroneous Rb-Sr isochron ages ( Figure S2). Initial 87 Sr/ 86 Sr and corresponding ɛSr values, as defined above, were calculated based on the secondary Pb-Pb isochron ages, reported here. Literature values were used in the event that a secondary Pb-Pb isochron age was not determined. The ɛSr values define a broad range of values between +14.4 and +114.1 (Table 5 and [59]). In Figure 3, North American craton Nd model ages based on the Bennett and DePaolo's [21] study are compared to the ɛSr values calculated for uraninite investigated in this study (n = 12) and for uraninite (n = 15) analyzed previously [59]. Overall, the ɛSr values increase from the margins to the central regions of the North American craton with the highest values in Canada, with the exception of uraninite from Rabbit Lake, Saskatchewan, Canada at +17.9 (Figure 3). Rabbit Lake uraninite is classified as a Proterozoic unconformity ore deposit, and thus could have experienced higher degrees of alteration during open system processes. This general trend of increasing ɛSr values relative to geographic position in North America is also evident within the southwestern region of the United States (Figure 3b). In the central section of the region shown in Figure 3b    Uraninite samples linked to a pegmatitic origin can be separated into two groups (Figure 3a and Table 5). Uranium deposits located within the eastern coastal region of the USA, where craton ages are <1.4 Ga, are linked to ɛSr values that range between +24.1 and +37.7; whereas uraninite with ɛSr values of +72.3 (>2.7 Ga) and +114.1 (1.7 to 2.0 Ga) are from deposits located within the older cratons of Canada (Figure 3a).
Uraninite samples linked to a pegmatitic origin can be separated into two groups (Figure 3a and Table 5). Uranium deposits located within the eastern coastal region of the USA, where craton ages are <1.4 Ga, are linked to ɛSr values that range between +24.1 and +37.7; whereas uraninite with ɛSr values of +72.3 (>2.7 Ga) and +114.1 (1.7-2.0 Ga) are from deposits located within the older cratons of Canada (Figure 3a).
Bataille and Bowen [60] modeled the 87 Sr/ 86 Sr variation of bioavailable Sr (i.e., isoscape maps) across the USA, which are based on reported literature values for various types of geologic samples (rock lithologies, fluvial, and vegetation). Figure 4 superimposes the sample locations and corresponding ɛSr values for samples investigated in this study within the USA isoscapes map from Bataille and Bowen [60].

U Isotope Data
U isotope ratios for uraninite, investigated here, are listed in Table 5. For this study, the average external reproducibility (2σ level) associated with the 238 U/ 235 U, 234 U/ 238 U, and 235 U/ 238 U ratios are 0.3‰, 8.18‰, and 0.79‰, respectively. Figure 5 displays the δ 238 U values for pristine solutions of uraninite, obtained here, as compared with those from several previous analogous studies of uraninite (Table  S1).
The δ 238 U values for uraninite from a deposit type yield a range of values that are either all negative or all positive (not mixed) with the exception of those from metamorphite polymetallic, sandstone, and Proterozoic unconformity contact deposits ( Figure 5). Granite-related deposits were separated into two subtypes; deposits of an endogranitic nature yield negative δ 238 U values, whereas perigranitic uraninite are characterized by positive values (Figure 5). Secular equilibrium is affected by fluid-mineral interaction either by post depositional alteration, or chemical weathering induced leaching [24]. Negative δ 234 U values indicate the loss of U, whereas positive δ 234 U values are associated with U gain. The majority (69%) of δ 234 U values, determined here, are between −10‰ and +10‰. Proterozoic unconformity basement-related uraninite is the only type with δ 234 U values greater than +10 (Table 5 and S1). Figure 6 illustrates an overall positive trend between 207 Pb/ 206 Pb and 238 U/ 235 U ratios for uraninite examined in this study. Great Bear uraninite is characterized by the lowest 238 U/ 235 U at 137.56. A positive array is observed when 207 Pb/ 206 Pb ratios are plotted against initial 87 Sr/ 86 Sr values ( Figure  6b). Figure 6c displays Ba/Sr ratios (log scale) versus initial 87 Sr/ 86 Sr values and the result is also an overall positive trend. Figure 5. Natural 238 U/ 235 U ratios of uraninite plotted as δ 238 U (defined in text) based on their deposit type. Uraninites from this study (solid squares) and several other previous studies (X symbols), which include: [9,20,25,50,51,59,61,62]. Solid line in upper right corner represents external reproducibility.

Discussion
A large concentration of uranium deposits are located within the western USA [63]. This presumably results from the prolonged tectonic activity in the area, such as the Laramide orogeny, which occurred between 80 and 35 million years ago [64], although it is most likely not the only factor. Such regional tectonic events could have had a significant impact on the geochemistry and isotopic signatures associated with uraninite mineralization. Figure 1a-d indicates that trace element abundances and their incorporation within uraninite is controlled primarily by the crystal structure, and these display some fractionation relative to canonical values for continental crust [52] and sediments [53,54]. Detailed examination of trace elemental ratios for pristine and corresponding altered sections for the same uraninite sample indicate higher La/Yb ratios for the latter (Figure 1a and Table 2). If the La/Yb ratio serves as a proxy for monitoring the degree of light vs. heavy rare earth elements (LREE/HREE) enrichment, then the results shown in Figure 1a suggest the larger LREEs are preferentially incorporated into uraninite alteration products. The latter feature has also been documented in a recent study by Balboni et al. [29]. Figure 1b,d indicates greater variation for Nb abundances relative to the other trace elements depicted in Figure 1 as the former span >4 orders of magnitude. In general, Figure 1d displays the removal of the more mobile elements Ta and Nb from pristine uraninite into altered sections, which is facilitated by the similar ionic radii of Y and Yb with that of uranium. Moreover, the mobility of Ta and Nb is dependent on the chemistry of the fluid/melt, and in turn the degree of interaction between the host rock and corresponding source rock [6]. The higher and more variable Zr/Hf ratios as compared with Rb/Cs values (Figure 1c) is not a crystallographic-controlled feature but is rather a source-dominated result. For example, sample 625 is from a sandstone-type deposit (Moonlight Mine) and thus its high Zr/Hf ratio could reflect the presence of detrital zircon in the precursor host rock. The remaining uraninite samples with high Zr/Hf values could simply reflect regional host rock source compositions or the result of U ore formation processes, such as metamorphism (e.g., Great Bear). The high Rb/Cs ratio for the altered Billiken sample could be attributed to contamination by Kfeldspar from the host rock ( Figure 1c). Figure 1a illustrates that initial 87 Sr/ 86 Sr values are age dependent, i.e., older metamorphite uraninites is characterized by higher initial Sr values, whereas much younger pegmatitic uraninite plot at lower ratios; intermediate initial Sr isotope values are recorded by uraninite aged between ~44 and 440 Ma. This interpretation is consistent with that postulated by [65]. Figure 6a plots 238 U/ 235 U vs. 207 Pb/ 206 Pb ratios for the uraninite samples, investigated here, and the latter value reflects a combined contribution from both common and radiogenic Pb sources; increased contributions from both components result in higher 207 Pb/ 206 Pb ratios since older deposits contain more radiogenic Pb. However, there does not seem to be a definitive temporal variation in the 238 U/ 235 U ratios (Figure 6a). The Great Bear uraninite is the oldest sample in this study and has the lowest 238 U/ 235 U ratio, which is attributed to vastly different paleo environmental conditions (e.g., oxidation state [19]) at the time of mineralization (Figure 6a). Figure 6b displays an overall positive trend between initial 87 Sr/ 86 Sr and 207 Pb/ 206 Pb, which could be attributed to radiogenic in-growth over time; i.e., samples with higher initial ratios for both isotope systems are older. Moreover, the overall positive correlation between Ba/Sr and initial 87 Sr/ 86 Sr ratios could be attributed to the increasing involvement of K-feldspar, which preferentially incorporates both Ba and Rb within the host crust ( Figure 6c). For example, K-feldspar is an abundant mineral in high pressure and temperature metamorphic terrains, such as the one hosting the Great Bear uraninite deposit. The results, presented here, indicate that despite the complex chemical nature of uraninite and its capacity to incorporate a variety of trace elements, it is nonetheless still possible to identify specific deposit locations that occur within proximity of each other (e.g., <300 km apart in southwestern USA) due to their contrasting isotope signatures (Figures 1, 3, and 6).
Accurate and rapid assessment of forensic signatures for intercepted illicit nuclear material requires reliable, proven, and state-of-the-art analytical techniques. Traditional isotopic measurements are conducted using acid digestion followed by ion exchange chromatography, which can take weeks to complete [17]. Recent advances in MC-ICP-MS instrumentation combined with laser ablation technology have provided the ability to obtain accurate chemical (e.g., [51]) and isotopic results within days (e.g., [66]). The secondary Pb-Pb isochron ages, obtained here, by both SM-and LA-MC-ICP-MS are in general within good agreement of one another, and with previously documented ages for these same U ore deposits (Figures 2 and S1). Hence, these results support the use of the LA-MC-ICP-MS method for common Pb age dating of uraninite as a viable nuclear forensic tool. Isotopic measurements conducted using SM-MC-ICP-MS are associated with much lower uncertainties as compared with those obtained by LA-MC-ICP-MS. In general, laser ablation analyses are associated with a higher uncertainty because, unlike solution mode analyses, the elements of interest are not separated from the U-rich matrix and the laser ablation ion signals are transient in nature; i.e., are not stable and decrease as a function of time of analysis. For Shinkolobwe, the uraninite matrix appeared to be "softer" as compared with the remaining uraninite samples, which influenced and reduced the efficiency of the ablation process. Marshall Pass samples were heavily altered throughout, which could have affected the ablation process via the presence of micro fractures within the uraninite resulting in higher uncertainties (i.e., less stable ion signals). Nevertheless, the LA-MC-ICP-MS analyses provide valuable information on the relative ages of uraninites, being investigated, i.e., whether it is billions of years old, or formed within the last 100 million years. In a nuclear forensic analysis, this type of information will prove useful in constraining the possible U deposits of origin.
Lewis et al. [67] showed that uraninite could be texturally and compositionally heterogeneous at the micron scale. Application of the LA-MC-ICP-MS method provides spatially resolved ages for both the mineralization and the associated alteration of uraninite, which can both be easily masked by bulk sample analysis by SM-MC-ICP-MS. Analyses conducted using LA-MC-ICP-MS have the capability to specifically target regions of interest on a sample at the 10 to 100 s of micron scale and hence avoid potential contamination from mineral inclusions and host rock materials. Uraninites from Jefferson, Billiken, and Shinkolobwe are the only three samples that did not give reliable ages. Ages determined for the uraninite from Billiken Lode of 537 ± 37 and 733 ± 1200 Ma do not match the previously documented age of 69.3 ± 1.1 Ma; this discrepancy can be explained by possible contamination from the host rock (Proterozoic age). This is supported by further examination of the LA-MC-ICP-MS analyses, since two separate secondary Pb-Pb isochrons associated with lower uncertainties yield a younger (476 ± 110 Ma) and older (1255 ± 65 Ma) age, the latter result is closer in age to that of the host rock for the Billiken uraninite sample ( Figure S1).
Isotopic measurements of both digested pristine and altered uraninite fragments were used to generate two-point secondary Pb-Pb isochrons for SM-MC-ICP-MS analyses. The agreement between ages, obtained here, with those previously documented for uraninite indicates that the alteration event occurred soon after the time of mineralization. Evidence for more recent alteration events is also corroborated by the negative δ 234 U values (Table 5), which can be attributed to uranium leaching [9]. Comparison between ages obtained by LA-MC-ICP-MS for the pristine (>80 wt% UO2) and altered (<80 wt% UO2) regions in Marshall 4 uraninite suggest the alteration occurred relatively soon after its crystallization ( Figure 2). Marshall 3 yields the youngest age via SM-MC-ICP-MS; however, the LA-MC-ICP-MS results for this same sample has pristine uraninite yielding an age of 145 ± 240 Ma and altered areas give −122 ± 55 Ma. Thus, the latter result obtained by SM-MC-ICP-MS for Marshall 3 reflects a very recent alteration event, which is supported by the δ 234 Uvalue of −64.69. On the basis of the ages obtained in this study of 1509 ± 19 and 1444 ± 61 Ma by SM-and LA-MC-ICP-MS, respectively, the uraninite from Great Bear is further confirmed to originate from the Echo Bay mine (1500 ± 10 to 1424 ± 29 Ma) within the Great Bear region.
The erroneous and negative Rb-Sr isochron ages indicate that the uraninite samples were affected by open system behavior most likely involving hydrothermal fluids. Rubidium has a much larger radius (1.48 Å) compared to both Sr (1.13 Å) and U (1.0 Å). Thus, this large difference in ionic radius can render Rb incompatible within the uraninite structure. In contrast, Sr can substitute for Ca (an important impurity) in eight-fold coordinated sites within UO2+x making Sr more compatible. The different geochemical properties between Rb and Sr could cause a fractionation between the parent and daughter isotope during fluid interaction, which does not affect the Pb isotopes. Alteration events must have occurred during the last 2.5 Ma within the southwestern USA, as indicated by the range of negative δ 234 U values from −2 to −65 (and up to +47), which could also have disturbed the Rb-Sr isotope systematics of uraninite. The elevated 87 Sr/ 86 Sr values, reported here, suggest that the source of Sr, similar to that for the U, is predominantly of crustal origin. This interpretation is corroborated by the fact that older Nd-model craton ages [21] are associated with higher initial 87 Sr/ 86 Sr (and ɛSr values) for uraninites, investigated here ( Figure 3). The influence of the host rock on uraninite mineralization is further emphasized by the correlation between higher predicted 87 Sr/ 86 Sr values based on isoscape maps for the continental USA [60] and higher Sr values, reported here ( Figure 4).
Several recent studies have investigated the main mechanism behind U isotope fractionation. Figure 5 illustrates the U isotope fractionation associated with several types of U deposits. Higher δ 238 U values are preferred within the lower oxidation state as per the nuclear field shift [9]. Although U isotope fractionation is dominated by reduction of U, the oxidation state is overprinted by increasing temperature [24]; as the temperature increases, the degree of isotopic fractionation decreases [9]. However, pegmatitic-type U ore deposits within the Appalachian mountain belt (Ruggles and Yancey) display lower δ 238 U values than previously documented [9], as low as −1.76 ( Figure 5). These negative δ 238 U values indicate that although temperature plays an important role during uraninite formation, it is not the sole mechanism responsible for U fractionation. Similar to the results from Uvarova et al. [9], U deposits associated with lower ore grades tend to have more negative δ 238 U values than those associated with higher ore grades ( Figure 5). Metamorphite and sandstone uranium deposits share similar mechanisms for ore formation with continual remobilization and crystallization of uraninite. The latter could be responsible for the highly variable δ 238 U values [62], shown in Figure 5. The range of δ 238 U values can also be related to the degree of U leaching from the host rock [24], which is linked to fluid interaction and the preferential removal of weakly bound 234 U. However, careful examination of all the data, reported here, indicates that there is no correlation between δ 238 U and δ 234 U values ( Figure S3).

Conclusions
Ages obtained for uraninite based on secondary Pb-Pb isochrons are in good agreement with those reported in the literature for the U ore deposits investigated here. Moreover, age results obtained by SM-MC-ICP-MS corroborate those determined by LA-MC-ICP-MS. The latter method provides spatially resolved age information for mineralization and alteration of uraninite in a shorter period of time as compared with conventional dating methods (e.g., ID-TIMS). LA-MC-ICP-MS measurements are associated with higher uncertainties as compared with those obtained by SM-MC-ICP-MS; however, useful age information can still be obtained from the former, i.e., clearly distinguish between uraninites that are characterized by vastly distinct ages. The trace element concentrations for uraninite indicate that the crystal structure dictates the incorporation of impurities followed by element availability. The large ionic radius of Rb (in particular relative to U and Pb) renders it more mobile during secondary alteration (open system) processes, consequently, eliminating the effectiveness of the Rb-Sr geochronometer for age determination of uraninite. Initial 87 Sr/ 86 Sr ratios for uraninite samples, investigated here, are in general positively correlated with the age of the host craton, i.e., higher initial 87 Sr/ 86 Sr values are found within older cratons. The U isotope signatures determined for uraninite, examined here, further support the variety of processes responsible for isotope fractionation, and these include temperature, the nuclear field effect, oxidation, and source rock composition.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure S1: Secondary Pb-Pb isochrons determined using SM-and LA-MC-ICP-MS for uraninites from this study, Figure S2: Rb-Sr isochrons obtained by SM-MC-ICP-MS. All ellipses are at the 2σ level for uncertainty, Figure S3: δ 238 U and δ 234 U values, Table S1: Complied list of uraninites from several previous studies.