Micrometer scale carbon isotopic study of bitumen associated with Athabasca uranium deposits: Constraints on the genetic relationship with petroleum source-rocks and the abiogenic

tters 258 (2007) 378–396www.elsevier.com/locate/epsl

Earth and Planetary Science Le

Micrometer scale carbon isotopic study of bitumen associated withAthabasca uranium deposits: Constraints on the genetic relationshipwith petroleum source-rocks and the abiogenic origin hypothesis

L. Sangély a,b,⁎, M. Chaussidon b, R. Michels a, M. Brouand a, M. Cuney a,V. Huault a, P. Landais a,c

a UMR 7566 G2R-CNRS, BP 239, Bd des Aiguillettes, 54506 Vandoeuvre-lès-Nancy Cedex, Franceb CRPG-CNRS, BP 20, 15 rue Notre Dame des Pauvres, 54501 Vandoeuvre-lès-Nancy Cedex, France

c ANDRA, 1-7 rue Jean Monnet, 92298 Chatenay-Malabry Cedex, France

Received 30 June 2006; received in revised form 7 March 2007; accepted 7 March 2007

Available onli

Editor: R.W. Carlson

ne 16 March 2007

Abstract

In situ analytical techniques – Fourier transform infrared microspectroscopy (μFTIR) and ion microprobe – have been used tounravel the origin of solid bitumen associated with the uranium deposits of Athabasca (Saskatchewan, Canada). Both aliphaticityand carbon isotopic compositions within the samples are heterogeneous but spatially organized in concentric zonations at themicrometer scale. Finally, the δ13C values are positively correlated to the aliphatic contents over an extremely large isotopic rangefrom ∼−49‰ to ∼−31‰. We infer that this positive correlation may be related to the carbon isotopic fractionations associatedwith the synthesis of bitumen through the catalytic hydrogenation of CO2, rather than the result of pre-existing petroleum productprecipitation and/or alteration (such as radiolysis). This explanation is consistent with (i) published results of abiogenic synthesisexperiments, in which the differences in δ13C values between saturated and unsaturated hydrocarbons range from +2 and +19‰,in contrast to the differences systematically observed in conventional bitumen and petroleum ranging from 0‰ to −4‰; (ii) theabsence of a similar positive correlation between aliphatic contents and δ13C values in the other bitumen analyzed in the presentstudy, for which a biogenic origin has been unequivocally established (samples from Oklo, Gabon, and Lodève, France, uraniumdeposits); (iii) the presence of CO2 and H2 in the gas-phase of fluid inclusions in the Athabasca uranium deposits, H2 resulting fromwater radiolysis.

The present results suggest that the δ13C vs. aliphaticity correlation could be used as a criterion to discriminate betweenabiogenic vs. biogenic origin of macromolecular organic matter.© 2007 Published by Elsevier B.V.

Keywords: organic carbon isotopes; secondary ion mass spectrometry; bitumen; uranium deposits; Athabasca; abiogenic synthesis

⁎ Corresponding author. Present address: C.E.A., DIF/DASE/SRCE,BP 12, 91680 Bruyères-le-Châtel, France. Tel.: +33 1 6926 4214; fax:+33 1 6926 7065.

E-mail address: [email protected] (L. Sangély).

0012-821X/$ - see front matter © 2007 Published by Elsevier B.V.doi:10.1016/j.epsl.2007.03.018

1. Introduction

Most Mid-Proterozoic uranium deposits of the Atha-basca Basin (northern Saskatchewan, Canada) contain

mailto:[email protected]

http://dx.doi.org/10.1016/j.epsl.2007.03.018

379L. Sangély et al. / Earth and Planetary Science Letters 258 (2007) 378–396

small amounts of organic matter within an area restricted to∼50 m around the mineralized bodies. Organic matter ishosted in brecciated, hydrothermally altered rocks of theArchean to Paleoproterozoic metamorphosed basementand the Late Paleoproterozoic to Mesoproterozoic sedi-mentary cover [1–4]. This organic matter has been iden-tified as solid bitumen (i.e. allochtonous organic matterfound within rocks in the form of a non-disseminated solidphase). Bitumen is generally considered as deriving from i)the thermal maturation of kerogen in sedimentary rocks orii) the alteration of petroleum (e.g. biodegradation,asphaltenes precipitation, thermal alteration, oxido-reduc-tion reactions) [5]. Since the discovery of the uraniumdeposits of the Athabasca in the late 60's, the origin of theassociated bitumen has remained an unresolved issue des-pite numerous petrographic and geochemical studies [1–4].

Fig. 1. Schematic cross-section of the study area in Athabasca, modified after [unconformity-related uranium deposit (Shea Creek exploration zone, Northlocated: in the uranium-mineralization in the basement (1) and the sedimentarythe sandstone cover (5). Numbers (6) and (7) correspond to graphite occurren(here a meta-greywacke), respectively.

Organic geochemistry studies demonstrate the potential forpetroleum generation of several regional rock units rangingfrom Paleoproterozoic to Paleozoic in age. However, thereis no consensus on which source-rock is actually related tothe bitumen associated with uranium deposits.

Protoliths to the metasedimentary rocks of the base-ment have been identified as Paleoproterozoic organic-rich pelites, psammites, and minor carbonates depo-sited in a continental margin environmental setting. TheMesoproterozoic sedimentary rocks includes fluvial toshallow marine quartz-rich sandstone deposited in anear-shore, shallow shelf environment as well as marinediscontinuous beds of siltstones, mudstones, shales, andstromatolitic carbonates [6]. Occurrences of bitumen arewidely reported in uranium deposits hosted in suchfluvio-deltaic sediments from continental or marginal

18] displaying the occurrences of solid bitumen and graphite within theern Saskatchewan, Canada). Numbers refer to bitumen occurrencescover (2), in the breccia body (3), in the clayey alteration haloes (4), in

ces within the basement: in reverse faults and in metasedimentary rocks

380 L. Sangély et al. / Earth and Planetary Science Letters 258 (2007) 378–396

marine plain environments. However, depositional andformational setting commonly accepted stands in con-trast to some features of the Athabasca.

For instance, genetic link between petroleum source-rocks and bitumen associated with the Paleoproterozoicuranium deposits of Oklo (Gabon) could be establishedon the basis of obvious geometric relationships [7].Bitumen-hosting deltaic sandstones are spatially associ-ated withmarine black shales in structures similar to faulttrap petroleum systems. The observation of mine-ralization paragenesis led some authors to propose thatoils may have acted as a reductant to precipitate uraninitefrom U-bearing hydrothermal solutions permeating intooil reservoirs [8]. As a result, the liquid hydrocarbonsmay have been simultaneously oxidized into solidbitumen. Similar formational process has been proposedfor the bitumen associated with the uranium deposits ofLodève (France). In addition, the genetic link betweenbitumen hosted in the reservoir facies (Cambriankarstified basement, Permian fluvial conglomerates,deltaic sandstones and siltstones interbedded withpyroclastic horizons) and the adjacent Permian lacustrinepetroleum source-rocks has been confirmed by geo-chemical study [9]. Sterane and terpane biomarkerfingerprints in reservoir facies and source-rocks hasbeen shown to be consistent (Landais and Connan,1986).

Uranium deposits of the Athabasca differ from theabove-cited localities by the paragenesis sequence ofmineralization. It is commonly accepted that bitumenpost-date uranium minerals corresponding to the mainmineralization stage, suggesting that oils may haveplayed no role in uranium precipitation [4]. This is

Fig. 2. Photographs illustrating occurrences of bitumen and graphite within thtextures (numbers correspond to occurrences shown in Fig. 1). 1-a: Bitumenbasement (Q = quartz, P = pyrite, D = dravite). 1-b: Back-scattered electronfibrous aspect of bitumen nodules is due to radial internal structures. The condepletion in uranium at the periphery of nodules relative to the core. 2-a: Masnodule separated from its host-rock), both observed within the uranium-ores onodule in section. The lighter rim is enriched in uranium and sulfur relative tnodules in 2-b. 3-a: bitumen within the breccia body. SEM micrograph of coclayey host-rock, 3-c: bitumen nodule in section. The progressive concentrenrichment in sulfur, chlorine and calcium. Pyrite micro-inclusions (white grthe nodules. Bitumen samples in panels 3-a, -b, and -c correspond to the same4-b: bitumen patches, both observed within the alteration haloes in sandstonenclosed in bitumen are quartz (dark grey), pyrite (light grey) and galena (wcavities within the sandstone cover. 5-b: Secondary electron SEM micrographcubes. 5-c: Back scattered electron SEM micrograph of a corresponding bitobserved in bitumen nodules within the basement (1-b). The concentric zoniuranium at the periphery of the nodule. 6-a: massive graphite associated withgneiss. 6-b: Secondary electron SEM micrograph of the corresponding mphotograph revealing the occurrence of graphite distributed along the foliatiowith quartz, muscovite, chlorite and anatase. The identification of graphite inreported first-order Raman spectra exhibiting a single band located around 1

consistent with a recent study that points out the lack ofhydrocarbons in the gas-phase of pre- and syn-mine-ralization fluid inclusions in quartz overgrowth [10]. Ithas been proposed that bitumen could rather haveprecipitated in the environment of the uranium depositsthrough radiolytic polymerization of oil on contact ofuranium-bearing minerals. In addition, spatial relation-ship between potential source-rocks in the AthabascaBasin and the bitumen-hosting rocks in the vicinity ofuranium deposits does not display obvious petroleumsystem-like architectural features [1]. Known potentialpetroleum source-rocks in the Athabasca Basin in-cludes i) the Phanerozoic black shales responsible forthe generation of the bitumen regionally distributed inthe Early Cretaceous Athabasca tar sands (i.e. LateDevonian–Early Missippian Exshaw Formation) ii) theMesoproterozoic siltstones and mudstones of theDouglas Formation and dolostones of the CarlswellFormation. The stratigraphic localization of thesepotential petroleum source-rocks in the AthabascaBasin implies that the oil generated from theseformations may have migrated downward through the1500 m thick sandstones of the Athabasca Group toreach the unconformity and the brecciated basement.The bitumen is actually distributed as uranium miner-alization along structures consisting in graphite-richshear faults rooted in the Paleoproterozoic metamor-phosed basement. It should be noted that molecularstudies by gas chromatography-mass spectrometry (GC-MS) appeared of little help in characterizing the originthe bitumen associated with the uranium deposits of theAthabasca. As thermally or radiolytically altered organ-ic matter, Precambrian uraniferous bitumen generally

e different host-rocks (see arrows) and the corresponding microscopicnodules observed in fractures within uranium mineralized bodies in theSEM micrograph of a section in a corresponding bitumen nodule. Thecentric zoning (darker rims underlined by dashed curves) results from asive layered bitumen and 2-b: bitumen nodules (SEM micrograph of af the sandstone cover. 2-c: SEMmicrograph of a corresponding bitumeno the core and corresponds to the light areas observed at the surface ofrresponding bitumen nodules, 3-b: bitumen nodule separated from itsic zoning (lighter rims underlined by dashed curves) results from anains) display a parallel corona distribution mostly limited to the core ofoccurrence as the nodules analyzed in Fig. 6. 4-a: Layered bitumen ande. 4-c: SEM micrograph of a bitumen patch in section. Large mineralshite), micro-inclusions are uraninite. 5-a: bitumen occurring in quartzof quartz cavities showing the bitumen nodules associated with pyriteumen nodule in section. The radial internal structures resemble thoseng (lighter rim underlined by dashed curve) is due to an enrichment insulfide in fracture within the shear zone affecting basement aluminousaterial showing well-crystallized graphite flakes. 7: Reflected lightn of meta-sedimentary rocks of the basement. Graphite occurs togetherbasement rocks is also supported by Raman microspectroscopy. [19]575 cm−1 which corresponds to C–C vibrations in aromatic layers.


contains (i) only trace amounts of molecules extractibleby common organic solvents and (ii) no detectable levelof biomarker compounds are released from this bitumenduring on-line pyrolysis [9].

Carbon isotopic study may provide some informationabout bitumen origin. δ13C value of bitumen depends onboth i) δ13C value of kerogen in the related source-rockand ii) the physical and chemical processes involved in

bitumen formation and alteration. Bulk-rock analysesreported in a previous study revealed a noteworthycontrast in δ13C value between two spatially independentbitumen occurrences [11] (∼−28‰ and ∼−49‰ atCigar Lake and Cluff Lake, respectively). It is still unclearwhether the isotopic variability observed between the twolocalities results from the contribution of several source-rocks with different δ13C values or from different bitumen

Fig. 2 (continued ).


formation/alteration histories. Addressing this issue in theparticular context of uranium deposits requires to evaluatethe influence of radiation-induced alteration (i.e. radiol-ysis) on δ13C values of bitumen. The passage of particlesthrough bitumen may significantly alter the chemicalproperties of irradiated medium through C-chain scissionand crosslinking reactions. Bitumen C-chain scissionreactions resulting in the production of hydrocarbon hasbeen shown to fractionate C-isotopes [12]. The classical

way to evaluate the influence of radiolysis is to investigatethe relationships between δ13C values, uranium contentand/or some relevant chemical properties in the bitumen,e.g. [12,13]. However, difficulties arise in the interpreta-tion of such bulk rock analytical data because radiolysis isa spatially localized process. The radiation dose absorbedby the bitumen is applicable only to the mass of thematerial reached by the particles emitted from uranium-bearing minerals. Most of uranium energy decay is

Fig. 3. Absorption index spectra obtained using specular reflection Fourier transform infrared microspectroscopy on bitumen samples from theAthabasca (A), the Oklo and Lodève (B) uranium deposits. Numbers refer to band assignments, 1: νCHali=3000–2800 cm−1, 2: νC_O=1800–1655 cm−1, νC=Caro=1655–1520 cm−1. Solid lines depict an example of peak fitting according the so-called “valley-to-valley method”.


released in the form ofα particles. The range ofα particlesin organic matter (i.e. the distance of penetration up fromwhich the energy of the particle drops below the point thatthe particles no longer alter the material) has beenestimated to 50 μm. This is supported by microscopystudies commonly reporting increases in reflectance andanisotropy of bitumen b50 μm from uranium minerals[2]. These changes have been interpreted as resulting fromthe development of aromatic structures displaying a shortrange ordering.

The aim of this work was to investigate in situ thevariability in δ13C values and chemical composition ofthe bitumen associated with the uranium deposits of theAthabasca in order to get some clues on their origin. Inthe present study, we developed a precise in situ ap-proach combining Fourier transform infrared micro-spectroscopy (μFTIR) in specular reflection mode [14]and multicollector ion microprobe sensing to investi-gate δ13C variations as a function of chemicalcomposition at the micrometer scale in solid bitumensamples [15]. A set of 9 standards representative of thechemical variability of organic matter was prepared toprecisely calibrate matrix effects during ion microprobeanalysis. External precision of the analyses was ±0.7‰(1σ) on the δ13C values. Because no information isavailable in the literature on δ13C variability intrinsic to“conventional” bitumen at the micrometer scale, wealso studied two other bitumen samples from thespatially independent uranium deposits of Oklo (Paleo-proterozoic, Gabon) and Lodève (Permian, SE France).Both additional samples have been selected because

their origin and their formation processes are wellunderstood [7,9].

2. Geological setting

2.1. Athabasca

The Athabasca uranium ore-bodies are spatiallyassociated with graphite-bearing faults in the vicinityof an unconformity between an Archean to LowerProterozoic gneissic basement and a Mid-Proterozoic,non-metamorphosed sandstone basin [6,16] (Fig. 1).The major uranium mineralization stage has been datedat 1461±47 Ma [17]. Bitumen is found associated withthe uranium mineralization from centimeter to decame-ter scale, in both the sandstone cover and the basement[1–4,11] (Fig. 1). Bitumen occurs in diverse host-rocksincluding brecciated gneiss, sandstone, sandstone-hosted breccias as well as clays within the alterationhaloes of the mineralized bodies (Fig. 2). Bitumen canform more or less massive layers within fractures,patches as well as brittle millimeter size nodules. Whenpresent, bitumen displays total organic carbon contentsin rocks ranging from 0.06 wt.% to 25 wt.% [2].Reflectance and paragenesis data indicate that bitumenassociated with U-deposits probably have formedduring the hydrothermal event which took place bet-ween 1450 and 900 Ma at P–T conditions of 180±20 °Cand 0.6 kbars [10]. Textural relations between uraniumminerals and solid bitumen suggest that bitumen post-dated mineralization [4].


2.2. Oklo

The Oklo uranium deposit is located in theFrancevillian series (Paleoproterozoic) of the France-ville basin in Gabon. It is situated at the top of the basalformation overlaid by the black shales. The basalformation is mainly composed of conglomerates andfine to coarse grained sandstone deposited in a fluvialand deltaic environment. First oil migration event likelyoccurred ca. 2000 Ma ago, when black shales kerogen

Fig. 4. Effects of the aliphaticity on instrumental mass fractionation (αinst) dA: αinst varied by ∼0.005 (i.e. Δinst varied by ∼5‰) over the set of standa(νCHali /Σν), over a range from 0 (for graphite) to 0.7 (for alginite coal).repeated analyses on each standard. The diameter of data points correspondsrepeated measurements on reference materials (i.e. ±0.02). In B: diffeinstrumental mass fractionation and δ13C values determined by conventiomaterials. The linear regression of measured αinst values against aliphaticityhas been used to calculate accurate αinst values from aliphaticity ratios (measurements have been calculated according to the relationship: δ13Ccor

correction for instrumental mass fractionation has been evaluated to ∼±0.

reached maximal burial temperature (∼180 °C asestimated by fluid inclusion study [20]). Uraniumbodies occur in tectonic structures affecting theFrancevillian series, which correspond to the maincharacteristics of petroleum fault traps [7]. Bitumenoccurs either associated with uranium in sandstone-conglomerate reservoirs overlain by impermeable blackshales or in the sandstones and dolomites interbedded inthe black shales. The overlying black shales have beenshown to act as petroleum source-rocks for the bitumen

uring δ13C measurements in organic matter for a set of 9 standards. Inrds and appeared negatively correlated to infrared aliphaticity ratiosError bars on αinst values correspond to the standard deviation overto the maximal absolute standard deviation in νCHali /Σν obtained byrence between ion microprobe δ13C measurements corrected fornal analysis as a function of infrared aliphaticity ratios in standardratios (αinst =−0.0069× (νCHali /Σν) +0.9747, R=95 — see panel A)νCHali /Σν) for each standard. Then, corrected ion microprobe δ13Cr=1000× [(δ

13Cuncorr / 1000+1) /αinst−1]. The uncertainty due to the2‰ (σ /n1 / 2).


trapped in sandstones-conglomerates. Bitumen accountsfor organic carbon contents ranging between 0.5 and1 wt.%, continuously distributed throughout the sand-stone-conglomerates of the Francevillian Series. Ther-mal alteration and interaction with oxidizing uranium-bearing fluids have been shown to be responsible forbitumen precipitation.

2.3. Lodève

The Lodève basin is located on the southern limitof the French Massif Central (France). Major miner-alizations are located in the Autunian series (Permian)which include black shales and repeated sedimentarysequences of silty and dolomite facies in addition tosome cineritic layers. Uranium mineralization is foundmostly in association with bitumen occurring inreservoir facies (such as siltstones, cinerites as well asin fault breccias) which acted as migration pathwaysfor the oils generated from the black shales of thePermian series. Two mineralization events have beendated at 173±6 and 108±5 Ma [21]. The genetic linkbetween the bitumen and the petroleum source-rockshas been demonstrated by biomarkers studies [9]. Oilbiodegradation and interaction with oxidizing uranium-bearing fluids have been shown to be responsible forbitumen precipitation [9].

Fig. 5. Ternary plot showing the in situ relative intensities of stretching mode vRelative intensities νCHali,νC=Caro, and νC_O correspond to aliphatic C–H, arEach symbol corresponds to a different nodule collected: within the ore-bodybody (U=14 ppm, open diamonds), about 1 m from the ore-body (U=4 ppmand Lodève (closed circles) uranium deposits.

3. Methods

3.1. Fourier transform infrared micro-spectroscopy

Fourier Transform microspectroscopy (μFTIR) spec-tra were acquired in specular reflection mode at aspatial resolution of 60 μm, according to a methoddescribed in [14]. Distinct absorption index bands due tostretching (ν), bending (δ), and out-of-plane deforma-tion (γ) mode vibrations were observed (Fig. 3).Assignments of the infrared bands were determinedafter the tables available in [22,23]. Only bands relat-ed to stretching vibration mode were considered be-cause they provided the most reproducible peak fittingresults. The bands corresponding to wave numbers of3000–2800 cm− 1, 1800–1655 cm− 1, and 1655–1520 cm−1 were related to aliphatic C–H (νCHali),carbonyl C_O (νC_O), and aromatic C_C (νC = Caro)bonds, respectively. Spectra in the νCHali range exhibitseveral (up to 4) more or less defined peakscorresponding to asymmetric and symmetric stretchingvibrations in both –CH3 and –CH2-groups [24]. OMICsoftware was used to fit the bands of interest using thecommon method described in [23]. The baseline chosenfor band intensity measurements was the “valley-to-valley” line. The precise determination of integrationrange for the different peaks depends on band profiles.

ibrations calculated for uranium deposits bitumen from μFTIR spectra.omatic C_C, and C_O bonds, respectively. In a: Athabasca bitumen.(U=111 ppm, closed diamonds and open triangles), 25 m from the ore-, open and closed squares). In b, bitumen from the Oklo (open circles)


In the case of well-defined peaks, integration limitscorrespond to the intersections between the peak and thebaseline. In the case of peak overlapping, integrationlimits correspond to peak shoulders. An example ofpeak fitting is depicted in Fig. 3. Peak areas were used tocalculate an aliphaticity ratio) which equals to νCHali /(νCHali +νC=Caro+νC_O). The reproducibility on νCHali /

Fig. 6. Zonations in δ13C values and in aliphaticity ratios observed within twofrom the clayey breccias surrounding themineralization (corresponding to occucorresponds to open squares and open diamonds on Fig. 5 and 7, respectively.aliphaticity ratio, 1b–2b). Black dots in panels 1a–2a and 1b–2b are ion micprogramwas used for data mapping. A grid (i.e. one data point at each mesh intthe coordinates of analysed areas and z the corresponding measured values (i.of all scattered points' z values. Each data point is weighted inversely by its dinorm parameter was set to 4). Then, contour lines were generated by interpolacontour lines displayed in Fig. 6 differ from what would be obtained by directinto account in such mapping procedure since each measured value is attribu

Σν has been determined by repeated measurements(n=4) on several organic standards including resinite,vitrinite, and anthracite. The relative standard deviationsover the 4 measurements range from 2% in aliphatic-rich materials (νCHali /Σν∼0.7) to 20% in aliphatic-poormaterials (νCHali /Σν∼0.1). The uncertainty associatedto νCHali /Σν values given in the present dataset

bitumen nodules from Athabasca deposits. Both bitumen nodules comerrence number 3 on Fig. 1). Bitumen nodule in panels 1a–1b and 2a–2bCurves correspond to equal δ13C (1a–2a) and νCHali /Σν values (infraredroprobe holes and FTIR analysis areas, respectively. Gnuplot plottingersection) were created from the scattered data set (x, y, z), where x, y aree. δ13C or νCHali /Σν). The z values are computed as weighted averagesstance to from the grid point raised to the norm power (in this study theting grid points. The use of grid-data mapping routine explains why theinterpolation between nearby data points. Spot dimensions are not takented to a point (namely the center of analysis areas).

Fig. 7. Plot of δ13C values against infrared aliphaticity (νCHali /Σν)for various bitumen from the Athabasca uranium deposits (all symbolsbut circles, each symbol corresponding to a different nodule) andbitumen from the Oklo and the Lodève uranium deposits (opencircles). Error bars on ion microprobe δ13C values are ±0.7‰. Thediameter of data points corresponds to the maximal absolute standarddeviation in νCHali /Σν obtained by repeated measurements on refer-ence materials (i.e. ±0.02).


corresponds to the maximal absolute standard deviationobtained on standards (i.e. ±0.02).

3.2. Ion microprobe δ13C measurements

The ion microprobe carbon isotope compositionsreported hereafter were determined using the CAMECAIMS 1270 at CRPG-CNRS, Nancy, France. Analyticaltechniques developed for the present study are givenelsewhere [15]. Carbon isotopic compositions are ex-pressed in the conventional per mil delta notation re-lative to the Pee Dee Belemnite (PDB) marine carbonatestandard (δ13C):

d13Csample ¼13C=12C sample13C =12CPDB

− 1

( )� 1000

where 13C/12CPDB=1.1237×10−2.

Samples were mounted in epoxy resin. Polishedsections were sputtered with primary beam of Cs+ ionsof 3–5 nA intensity accelerated to 20 keV impactenergy. The primary ion beam was defocused usingKohler illumination to produce a roughly circular, flat-bottomed crater of ∼20 μm diameter and ∼1–2 μmdepth. A normal-incidence electron flood gun was usedto compensate for sample charging during analysis.Secondary negative carbon ions were accelerated at10 keV without energy filtering, and analyzed at a massresolution of ∼5000 to resolve the interference bet-ween 12CH− and 13C−. The 12C− and 13C− ions werecounted in multi-collection mode on two Faraday cups.A within-run precision ∼0.1‰ at 1σ standard errorof the mean is typically achieved after ∼2 min count-ing with a 12C intensity of ∼5×108 cps. Approxi-mately 20 measurements were performed on eachsample, which resulted in a ∼100–200 μm step grid.The major difficulty for this type of isotopic mea-surement is the calibration of the instrumental massfractionationwhich can be very sensitive tomatrix effects.For carbon, it is defined by:

ainst ¼ ð13C=12C Þmeasured

ð13C=12C ÞtrueIn addition, it can be reported in units of permil, calcu-

lated by the relationship:

Dinst ¼ 1000� lnðainstÞ

Chemical and carbon isotopic compositions of stand-ards (including graphites, anthracite, two vitrinite coals,

two alginite coals, resinite coal, and type II kerogen)span a range from 0.04 to 1.75 in H/C atomic ratiosand from −31.1‰ to −6.8‰ in δ13C values. The repro-ducibility of repeated δ13C measurements on standards(n∼20) (i.e. external precision) is∼ ±0.7‰ (1σ). It wasobserved that αinst varies as a linear function of thealiphaticity in standards investigated, with typical valuesof 0.975 and 0.970 for aliphaticity ratios (νCHali /Σν) of0 (graphite) and 0.7 (alginite coal), respectively (Fig. 4).The aliphaticity ratios of area previously analyzed forδ13C were determined by μFTIR. We checked onstandard materials that aliphaticity ratios measured byμFTIR were not affected by primary ion exposure [25].The sampling depth of μFTIR (in the μm range) is muchlarger than the 10 nm depth chemically altered as theresult of primary ion implantation and of differentialsputtering rates during ion microprobe measurements.However, a slight polishing step was necessary toremove gold coating and to restore smooth and planarsurface suitable for μFTIR used in reflection mode,when non-destructive μFTIR was not performed beforeion microprobe analysis. The precision of the correctionfor matrix effects on αinst is of ±0.2‰ (1σ standard errorof the mean) over the whole range of aliphaticity. Theprecision of ion microprobe δ13C measurements isestimated to be ∼±0.7‰ (1σ) from the quadratic sumof the uncertainties derived from counting statistics(±0.1‰), the point-to-point reproducibility on standards


(±0.6‰) and the correction for matrix effects oninstrumental mass fractionation (±0.2‰).

4. Samples

4.1. Athabasca

The bitumen samples of Athabasca analyzed in thepresent study were collected from the Shea Creekuranium prospect, located 10 to 20 km south of theCarswell impact structure in the western part of theAthabasca basin (northern Saskatchewan, Canada).They were sampled from drill cores corresponding toan average depth of about 750 m. Rock samples en-closing non-fibrous bitumen nodules were collected inthe ore-body occurring within the sandstone cover(corresponding to number 2 in Fig. 1, and numbers 2-band 2-c in Fig. 2) as well as in clayey breccias sur-rounding the mineralization (corresponding to number 3in Fig. 1, and numbers 3-b and 3-c in Fig. 2). Typicalorganic carbon content of the rock samples is lessthan 0.5 wt.%. The uranium content of the whole rockranges between 100 ppm in barren samples and 10 wt.%in mineralized samples. The uranium content of thebitumen itself, after being mechanically isolated fromthe host-rock, ranges from 5 to 110 ppm. The H/C andO/C atomic ratios of the bitumen range between 0.6 and0.7, and between 0.03 and 0.15, respectively [15].

4.2. Oklo

The Oklo bitumen sample comes from the uraniummine of Oklo, located 50 km northeast of Franceville inthe northwestern edge of the Franceville basin (Gabon).It was collected from underground mining operation ina massive bitumen layer at a depth of about 300 m,near the core of the “natural fission reactor 10”. Thisbitumen is strongly mineralized and encloses quartzrelics remaining from the host sandstone. Its uraniumand organic carbon content are 42 wt.% and 8.7 wt.%,respectively. Bitumen H/C and O/C ratios are 0.46 and0.13, respectively [15].

4.3. Lodève

The Lodève bitumen sample was collected at theMas Lavayre uranium mine, located 5 km southeast ofLodève, in the northwestern edge of the Lodève basin(SE, France). The sample was located in a massive, blackand brittle bitumen layer hosted in siltstone fromunderground mining operation at a depth of about300 m. Its uranium and organic carbon content are of

110 ppm and 86.1 wt.%, respectively. The bitumen H/Cand O/C atomic ratio are 1.23 and 0.02, respectively [15].

5. Results

5.1. Fourier transform infrared micro-spectroscopy

Fourier transform infrared micro-spectroscopy spec-tra reveal that the relative aliphatic content is heteroge-neous at the micrometer scale, within individual samplesof bitumen of Athabasca. At the same scale, it is morehomogeneous in the two other deposits (Figs. 3 and 5). Inthe Athabasca uranium deposits, νCHali decreases at agiven νC=Caro /νC_O and at the micrometer scale fromcenters to edges of each analyzed bitumen nodule. Inaddition, the ranges in aliphatic content vary betweenthe different samples analyzed. No clear relationshipappears between FTIR aliphaticity, uranium content ofbitumen, and distance of samples to the mineralizedbodies (see caption of Fig. 5). On the other hand, νC=Caro

increases relative to νC_O at low νCHali proportion inuranium-rich areas of Athabasca bitumen nodules(Fig. 5). Similar variations were observed in the Oklobitumen whereas the Lodève bitumen exhibits a homo-geneous aliphatic-rich composition.

The proportion of aliphatic hydrocarbons in thebitumen (aliphaticity ratios, hereafter noted νCHali /Σν)has been estimated by integration of μFTIR spectra(aliphaticity ratio values calculated from all the infraredanalyses are given as supplementary material). Within agiven bitumen nodule of Athabasca, νCHali /Σν defineapproximately concentric zonations of a few hundredmicrometer radius with the presence of aliphatic-richcenters and aliphatic-poor rims (Fig. 6).

5.2. Ion microprobe δ13C measurements

Carbon isotopic compositions are also strongly vari-able at the micrometer scale within the Athabasca bitu-men with δ13C values ranging from −51‰ to −23‰(all the ion microprobe δ13C values are given as sup-plementary material). In contrast, δ13C values arerather homogeneous in Oklo and Lodève, δ13C=−18.6±0.7‰ (standard deviation of over 15 measure-ments) and −20.5±1.1‰ (standard deviation of over 17measurements), respectively. The range obtained at themicrometer scale for Athabasca bitumen is in goodagreement with bulk δ13C values reported for bitumenat the scale of the uranium deposits, from −53‰ to−23‰ [1–3,11]. In addition, bitumen nodules from theAthabasca show a well-defined zonation for δ13Cvalues, a zonation similar to that observed for


aliphaticity ratios, with 13C-rich centers and δ13C valueswhich can range from −42‰ to −24‰ within a singlenodule (Fig. 6, panel 1-a). When all the data areconsidered together, δ13C values appear positivelycorrelated with the aliphaticity ratios at the micrometerscale over a range from ∼−49‰ to ∼−31‰ forAthabasca bitumen (Fig. 7). It is of particular note that,in contrast, such a correlation is absent in the Oklo andLodève bitumen.

6. Inference of organic carbon isotopics in petroleumsystems

Two hypotheses will be discussed below on thepossible causes for the micrometer scale variability inaliphaticity and δ13C values in the bitumen nodulesassociated with the uranium deposits of the Athabasca.The observed concentric zonation may be due to i)reaction gradient related to bitumen precipitation/alter-ation processes; ii) multistage bitumen precipitation re-sulting from several oil events.

6.1. Carbon isotopic fractionation during bitumenprecipitation and alteration

Although radiolysis has traditionally been invoked toexplain the alteration of uraniferous bitumen, it cannotbe the cause of the large δ13C variations observed at themicrometer scale in bitumen from the Athabasca ura-nium deposits for two reasons. i) In fact, as thermalalteration, radiolysis reduces the content in heteroatomicand aliphatic relative to aromatic compounds [13], thesechanges being accompanied by an enrichment in 13C[12]. Carbon isotopic fractionation is observed duringartificial maturation experiments, in which the residualcarbon is continuously enriched in 13C as a result of thegeneration of 13C-depleted gases. The carbon isotopicfractionation factors are at maximum between CH4 andthe organic precursors, with for instance experimentallydetermined values of 0.983 and 0.996 for algal and landplant kerogens, respectively [26]. Thus, the increase inδ13C values observed together with a decrease of ali-phatic contents during alteration stands in contrast to thetrend observed in the bitumen samples from Athabasca.ii) In contrast to Athabasca bitumen nodules, the δ13Cvalues observed in the samples from Oklo (Fig. 7),which is likely radiolyzed given the high uranium con-tent of its host-rock (42 wt.%), are rather homogeneous.

However, some radiolysis effects have been notedlocally at the periphery of some of the nodules fromAthabasca, within uranium-rich, discontinuous ∼50 μmwide rims, exhibiting local 13C-enrichements and lower

aliphaticity ratios. For instance, local 13C-enrichmentsdue to radiolysis are visible in Fig. 6 at the peripheryof the bitumen nodule on the left bottom corner in panel2-a. These effects on δ13C values are however verylimited, 4.6‰ at maximum. These radiolytic effectsare only visible in the rim of the present bitumennodules containing uranium inclusions (16 over 104analyses on a total of 5 nodules). Analytical spots loca-ted in U-bearing areas exhibit low νCHali, more or lessνC=Caro-enriched and νC_O-depleted compositions rel-ative to U-free parts of the bitumen nodules. In contrast,these compositions are similar to what can be observed inU-rich Oklo bitumen (Fig. 5). The U-bearing rims of theAthasbasca bitumen (e.g. panel 2-c in Fig. 2) are thusexcluded in the following discussion.

Moreover, our results indicate that the segregationof aliphatic and non-aliphatic hydrocarbons during theprecipitation of bitumen from a petroleum fluid can alsobe excluded as the major cause of the micrometer scalechemical and isotopic variations in the Athabasca bitu-men nodules. The positive correlation between aliphati-city ratios and δ13C values observed here is oppositeto the systematics reported for the soluble fractionof conventional (biogenic) bitumen and petroleum:aliphatics are 13C-depleted compared to aromatics(δ13Caliphatics–δ

13Caromatics=−0.4‰ to −4‰) which arein turn 13C-depleted relative to heteroatomic compounds(δ13Caromatics–δ

13Cheteroatomic compounds=0 to −0.79‰)[5,27–29].

6.2. Multistage bitumen precipitation resulting fromseveral oil migration events?

Since alteration processes have been ruled out, thelinear relationship between δ13C values and aliphaticityratios observed for the bitumen of Athabasca mayindicate the mixing of two bitumen components, onebeing aliphatic-rich and having δ13C values N−31‰,and the other one being aliphatic-poor and having δ13Cvalues b−49‰. However, there is no indication for thepossible occurrence of source-rocks within the Atha-basca Basin likely to have produced petroleum productswith δ13C values as low as b−49‰. As mentioned inthe introduction, potential for petroleum generation hasbeen demonstrated for several rock units in the Atha-basca Basin [1,4]. Bitumen generated from Phanerozoicsource-rocks and distributed in the Early CretaceousAthabasca tar sands exhibit δ13C values of ∼−30‰[30]. Kerogen δ13C values of ∼−30‰ are also reportedfor Mid-Proterozoic potential petroleum source-rocksidentified in the Douglas Formation [1]. A genetic linkwith extremely low δ13C bitumen (∼−49‰) is not


supported by the maximum δ13C difference of ∼5‰commonly reported between kerogen and related oils (orbitumen) [25]. The lack of large C isotopic fractionationduring oil generation, migration and alteration tobitumen is also supported by numerous experiments[25]. Moreover, the hypothesis that extremely low δ13Cbitumen may result from any petroleum source-rock tobe identified across the sedimentary cover of the Atha-basca Basin represents a problematic issue with regardto the general trend observed in organic carbon isotopiccomposition over the geological time. It has beenestimated that sedimentation began in the AthabascaBasin at about 1740–1730 Ma [31], considering meta-morphic ages on titanite, as young as 1750 Ma in thebasement rocks [32]. Kerogens younger than 2000 Magenerally have δ13C values in the −35‰ to −25‰range, data being corrected for the effects of thermalalteration [33].

These observations raise the hypothesis of a geneticlink between extremely low δ13C bitumen and any extra-basinal petroleum-source rock older than 2000Ma. In theAthabasca, carbonaceousmaterial in the Paleoproterozoicmetasedimentary rocks of the basement exhibit δ13Cvalues of −25±5‰ [3]. However, δ13C values in the−40‰ to −50‰ range are commonly reported inArchean to Paleoproterozoic sedimentary rocks. Forinstance, δ13C values as low as−45‰ have been reportedfor kerogen in the 3000–2300 Ma and 2700–2100 Masource-rocks of the Witwatersrand (South Africa) and theFranceville Basins (Gabon), respectively [7,20,34].However, timing of regional metamorphism indicatesthat Paleoproterozoic source-rocks may have lost theirpotential for hydrocarbon generation at time of bitumenemplacement (after ∼1461±47 Ma [17]). The Paleopro-terozoic rocks were actually metamorphosed to theamphibolite–granulite grade during the 1930 Ma Talstonorogeny [6]. As a consequence, the kerogens in theserocks and the possible migrated hydrocarbon productshave been metamorphosed to graphite.

This led some authors to consider an alternativeorigin for the solid bitumen which would have been –directly or indirectly – produced through the conversionof graphite present in basement faults to hydrocarbonsduring hydrothermal alteration events [3,11]. Manyscenarios have been proposed in which bitumen re-presents (i) a solid product resulting either from graphiteradiolysis or in situ hydrogenation or (ii) a precipitateresulting from the radiolytic alteration of liquid orgaseous graphite-derived hydrocarbons once migrated.However, there is no consensus on the reactionresponsible for the strong carbon isotopic fractionationwhich could explain the large difference in δ13C values

observed between the extremely low δ13C bitumen andgraphite (−25±5‰) [3]. Thermodynamic calculationsindicate that liquid hydrocarbons are generated in traceamount only by hydrothermal alteration of graphitefor the P–T conditions prevailing around the Athabascauranium deposits and geologically relevant oxygenfugacity values [35]. There are also no experimentalstudies supporting such hypotheses.

In the following discussion we consider the hypoth-esis of bitumen formation through the catalytic hydro-genation of CO2 in the vicinity of uranium deposits.CO2 bearing fluid was likely produced by graphite dis-solution resulting from the interaction of metasedimen-tary basement rocks with oxidizing basinal fluids. Astrong decrease in graphite content has been reportedwithin clay-rich hydrothermal alteration haloes com-pared to adjacent unaltered rocks of the basement [36].In addition, [19] observed at distances b5 km from theunconformity an intense fragmentation and deformationof graphite flakes as well as the intercalation of clayminerals in-between. Carbon isotopic study of alteredgraphite by [3] supports the hypothesis of graphite de-composition through oxidation to CO2. Shear zone fluidflow pathways likely allowed the migration of CO2-bearing fluid from the basement to the environment ofthe uranium deposits at the level of the unconformity.

7. Can abiogenic synthesis be responsible for themicrometer scale chemical and isotopic variations inthe Athabasca bitumen?

7.1. The context of abiogenic synthesis of organicmatter in the Earth's crust

Even if it has recently been demonstrated that in-organic chemical reactions make a minor contribution tohydrocarbons occurring in the Earth's crust compared tothose produced by biogeochemical cycles [37], thepossibility of an abiogenic origin has been previouslydiscussed for some organic compounds in various geo-logical environments on Earth. This includes for in-stance organic compounds occurring in hydrothermalventings [38–44], in gas-emissions from volcanic areasand geothermal fields [45–47], in gas discharge fromcrystalline basement [37,48,49] and serpentinizedultramafic rocks [50,51], in fluid inclusions, vacuoles,grain boundaries of igneous rocks [52–55], and as filmsdeposited on crack surfaces in crystals from mantlexenoliths [56,57]. The organic compounds in terrestrialsamples for which an abiogenic origin has been dis-cussed are mostly hydrocarbons: methane, C2–C5 andC16–C19 saturated aliphatic hydrocarbons (including n-


alkanes, i-alkanes, and neo-alkanes), ethylene, acetylenee.g. [54,58], benzene, toluene but also include phenoland methyl phenol, benzaldehyde and benzoic acid, O-,N-heterocompounds, Cu-, Ni-organometallic com-pounds, chlorinated compounds [59–61], and possiblyamino acids [38].

In addition, an abiogenic origin has been recentlyconsidered for some occurrences of macromolecularcarbonaceous material (i.e. material insoluble in com-mon organic solvents and containing mostly highmolecular weight polymers and/or macromolecularmolecules, e.g. [62]). These samples, which are alsoregarded by some authors as the earliest traces of life onEarth, includes graphite in apatites from the 3.8 Ga Isuasupracrustal belt [63] as well as macromolecularamorphous organic matter in preserved microbe-likefeatures or disseminated in the 3.5 Ga Apex chert ofWarrawoona Group, Western Australia [64,65]. Severalmechanisms are documented both experimentally and insome natural occurrences for graphite precipitation fromnatural carbon bearing fluids such as those containingCO2, CO and CH4 in geological environments [66]. Inaddition, thermodynamic study indicates that graphiteprecipitation can be achieved in the C–O–H system byseveral means including, for instance, the cooling of aC-bearing fluid and/or changes in ambient oxygen fuga-city. In contrast to graphite, the identification of mecha-nisms for abiogenic macromolecular organic matterfrom natural systems on Earth is still elusive.

However, thermodynamic calculations by Zolotovand Shock [67] indicate that abiogenic synthesis of highmolecular weight hydrocarbons is favoured under con-ditions prevailing during the cooling of CO, CO2 andH2-rich magmatic gases and at temperature ∼below250 °C through overall reactions which may be writtenfor for n-alkanes and naphthalene as:

n CO þ ð2n þ 1ÞH2→CnH2nþ2 þ nH2O ð1Þ

n CO2 þ ð3n þ 1ÞH2→CnH2nþ2 þ 2nH2O ð2Þ

10 CO þ 14 H2→C10H8 þ 10 H2O ð3Þ

10CO2 þ 24H2→C10H8 þ 20H2O ð4Þ

An analogy can be drawn between reactions (1) and(3) and the Fischer–Tropsch (F–T) industrial processwhich corresponds to the synthesis of gaseous, liquidand solid hydrocarbons, aliphatic alcohols, aldehydes,and ketones by the catalytic hydrogenation of CO usingenriched synthesis gas from passage of steam overheated coke. The F–T reactions are heterogeneously

catalyzed by a group VIII metal in its native form or asan oxide (e.g. Fe, Ni, Co). The initial step of the cata-lysis is the adsorption of gaseous CO and H2 reactantsonto the surface of the catalyst. Binding and interactionwith metal atoms promotes (i) the deoxygenation of COto give a surface carbide, (ii) the hydrogenation ofsurface carbide to give surface methylidyne, methylene,methyl, and eventually methane, (iii) the polymeriza-tion of adsorbed C1 monomers to yield n-alkenes andn-alkanes and (iv) the many reactions undergonesubsequently by these primary products, including hy-drogenation, hydrogen migration, skeletal isomeriza-tion, cyclization, dehydrogenation, and oxidation [68].

In addition to magmatic process, a frequently infer-red pathway for hydrocarbon synthesis in geologicalenvironments involves reduction of CO2 during water–rock interaction, with minerals playing a role in gene-rating high dissolved H2 concentrations [69]. Fe-bearingminerals (e.g. magnetite and montmorillonite) as wellas Fe–Ni alloys have been shown to provide catalyticsurfaces during hydrogenation of CO2 [70,71]. It hasbeen proposed that the abiogenic reduction of inorganiccarbon to hydrocarbons could be favoured by theserpentinization reactions occurring in hydrothermallyaltered ultramafic rocks [72]. Shock [73] showed fromtheoretical considerations that the hydrogenation ofaqueous CO2 to organic compounds is favourable inthe oceanic crust under hydrothermal conditions (tem-perature below 500 °C). However, the potential forformation of hydrocarbons other than CH4 during ser-pentinization reactions remains to be demonstrated ex-perimentally [69,72]. More recently, [74] experimentsdemonstrated the potential of alternative geologicalprocesses such as the thermal decomposition of sideriteat 350 °C in presence of water to form abiogenic highmolecular weight hydrocarbons and macromolecularorganic matter [74]. These authors indicate that syn-thesis of organic compounds other than CH4 may pro-ceed in these experiments from reduction of CO2 inwater-saturated vapour phase. All these results supportthe view that an abiogenic origin is worth to be dis-cussed for solid organic matter in some local environ-ments of the Earth's crust, in the absence of biologicalindicators such as biomarkers or microfossils.

Chemical and physical conditions prevailing duringthe formation of uranium deposits of the AthabascaBasin (Saskatchewan, Canada) may offer favourableconditions for abiogenic synthesis of hydrocarbonsthrough catalytic hydrogenation of CO2 for 3 mainreasons. (i) Some studies report [71,75] the productionof lipids (including oxygenated and hydrocarboncompounds with carbon numbers up to 40) during


experimental hydrogenation of CO2 in the temperaturewindow of 150–400 °C with a maximum yield at∼200 °C. These temperature conditions are close tothose derived from fluid inclusions studies to haveprevailed during the formation of the deposits (180 °C,600 bar) [10]; (ii) Hematite, which may constitute acatalyst of hydrogenation reactions is abundant in thevicinity of the ores [6]; (iii) The reactants of thehydrocarbon synthesis through catalytic hydrogenationprocesses such as reactions (2) and (4) are also availablein the environment of uranium deposits. The presence ofH2 and CO2 has been demonstrated in the gas-phase offluid inclusions in both the basement and the alteredsandstone cover, H2 being interpreted as the product ofwater radiolysis [10,76].

7.2. An abiogenic origin for the chemical and δ13Cvariability of the Athabasca bitumen?

It appears that the fractionation related to the catalyticconversion of CO2 to hydrocarbons may explain both(i) the carbon isotopic compositions and (ii) their vari-ation as a function of the aliphaticity ratios in bitumen inthe Athabasca uranium deposits.

The bulk δ13C values reported for the bitumen ofAthabasca ranging from −53‰ to −23‰ [1–3,11] areconsistent with values expected for the catalyticconversion of CO2 to hydrocarbons. Several studies ofthe catalytic hydrogenation of CO to high molecularweight (HMW) hydrocarbons under dry H2 atmosphereshowed variable kinetic carbon isotopic fractionations(αHMW hydrocarbons–CO), ranging from 0.965 to 0.990 [77–80] for a review. In addition, a recent study ofhydrocarbon synthesis during the decomposition of side-rite under hydrothermal conditions indicates a carbonisotope fractionation of∼0.964 between C10–C26 hydro-

Table 1Carbon isotopic fractionation between saturated and unsaturated hydrocarbo

Synthesis experiments Hydrocarbons

Saturated

Spark discharge in CH4a C2H6

Thermocatalysis of oleic acid b n-alkanesOpen-flow FT processes involving CO and H2

d n-C2–C5 alkaOpen-flow FT processes involving CO and H2

e LMWHCs f

a [82].b [83].c Polynuclear aromatic hydrocarbons.d [78].e [79].f Low molecular weight hydrocarbons.

carbons and dissolved CO2 (αC10–C26 hydrocarbons–CO2)

[81]. Similar isotopic fractionations have been obtainedduring experimental hydrogenation of dissolved CO2 toCH4, with αCH4–CO2 ranging from 0.940 to 0.965 [70].Carbon isotopic composition of theCO2-bearing fluids hasnot been measured, but it is highly probable that CO2 hasbeen produced by the decomposition of basement graphiteduring the percolation of the oxidizing basinal fluids [3].Assuming a temperature of 180°C for the percolatingfluids [10], a δ13C value close to −10±5‰ can becalculated at isotopic equilibrium for CO2-bearing fluids,taking δ13C=−25±5‰ for the graphite (as measured in[3]). This calculated value is in good agreement with theδ13C value of −7‰ measured in calcite parageneticallyassociated with bitumen (αcalcite–CO2

=1.004) [10]. As afirst approximation, a carbon isotopic effect analogous tothat related to the abiogenic formation of HMW hydro-carbons from CO (αHMW hydrocarbons–CO) has been consid-ered between reacting CO2 and the synthesized bitumen.The δ13C values which can be predicted for the bulkhydrocarbons produced (from −45±5‰ to −20±5‰)overlap the δ13C range observed for bulk bitumen fromAthabasca (from −53‰ to −23‰).

In contrast to biogenic hydrocarbons, a positivedifference in δ13C between saturated and unsaturatedhydrocarbons is systematically produced in gas synthe-sized through dynamic FT synthesis reactions [78,79].This is also observed for abiogenic volatile hydrocar-bons produced by spark discharge in methane [82] andfor hydrocarbons extracted from synthetic oils pro-duced by thermocatalysis of oleic acid [83]. It shouldbe noted that no isotopic results are available yet foralkenes and alcohols produced together with n-alkanesin synthesis experiment from dissolved CO2 by [81].The differences in δ13C between saturated and unsatu-rated hydrocarbons reported in the literature range

ns during abiogenic synthesis experiments

products analyzed δ13Csaturated–δ13Cunsaturated

Unsaturated (‰)

C2H4 +3.7±0.3C2H2 +4.9±0.3PAHs c +2

nes C3H6 +7LMWHCsf +3–19


from +2 to +19‰ (Table 1). Thus, the mixing in variousproportions of aliphatic and aromatic hydrocarbonsproduced through abiogenic reactions may explain thepositive correlation observed between the aliphaticityratios and carbon isotopic compositions in the bitumenof the Athabasca uranium deposits (Fig. 7). This mayalso explain the observed δ13C difference between ali-phatics and non-aliphatics of at least +18‰ for theAthabasca bitumen, although this value is in the extremerange of available experimental data.

According to this hypothesis, the chemical zonationobserved in Athabasca bitumen nodules may reflect adecreasing H2 content of the gas phase through time.This would have resulted in a decreasing production ofaliphatics vs. non-aliphatic [84] hydrocarbons duringthe growth of the nodules from the center to the edge.This hypothesis is in good agreement with the lack oflarge δ13C variability at the micrometer scale associatedwith a 13C-enrichment of aliphatics relative to non-aliphatics in the bitumen of demonstrated biogenicorigin from Oklo and Lodève uranium deposits [7,9]. Incontrast, a parallel can be drawn between our results andthe observation that isotopic zoning and small isotopicshifts at the millimeter-scale are common features influid-deposited graphite, by contrast with graphiteresulting from the metamorphic conversion of sedimen-tary organic matter [66].

8. Conclusion

In regard to the geological setting as well as theunusual relationship between aliphaticity and δ13Cvalues of the bitumen samples from the Athabascauranium deposits, an abiogenic origin can be consideredas a reasonable hypothesis for the bitumen of theAthabasca. This may also explain the extremely lowδ13C values of some of this bitumen. Obviously, theamount of abiogenic organic matter found in theAthabasca uranium deposits is minor compared to thelayered bitumen commonly observed in uraniumdeposits associated to conventional petroleum deposits(e.g. Oklo and Lodève). However, the present results arein agreement with the growing recognition that,contrary to the general view, 13C-depleted carbonisotopic composition in organic matter cannot beconsidered as an unequivocal evidence for a biogenicorigin [80,81,85–87]. A criterion for abiogenic origin oflow molecular weight alkanes has been recentlyproposed from the positive correlation between δ13Cvalues and the carbon number observed for C1–C4 gasesrecovered from boreholes in the Canadian Shield [37].The bitumen from the Athabasca uranium deposits

shows at the micrometer scale a similar relationshipwith a correlated increase in δ13C and aliphaticity (Fig.7). The present results suggest that this relationship maybe used as a criterion for “abiogenicity” and may beapplied in future studies to organic matter in EarlyArchean or extraterrestrial samples investigated fortraces of life.

Acknowledgements

This research was carried out thanks to a Ph.D. grantfrom the French Research Ministry. We thank M.Champenois, E. Deloule, D. Mangin and C. Rollion-Bard for assistance with ion microprobe analyses, C.France-Lanord and C. Guillemette for help in gas sourcemass spectrometry, A. Izart for providing standards, O.Barrès and P. De Donato for facilities for μFTIRanalyses, L. Richard and M. Roskosz for discussionsand throughout correction of the manuscript. COGEMAResources Inc. (Areva) is also acknowledged forprovision of some samples.

Appendix A. Supplementary data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.epsl.2007.03.018.

References

[1] P. Landais, J.M. Dereppe, A chemical study of the carbonaceousmaterial from the Carswell structure, Geological Association ofCanada special paper — The Carswell Structure UraniumDeposits, Saskatchewan, vol. 29, Geological Survey of Canada,Toronto, 1985, pp. 165–174.

[2] J.S. Leventhal, R.I. Grauch, C.N. Threlkeld, F.E. Lichte, Unusualorganic matter associated with uranium from the Claude deposit,Cluff Lake, Canada, Econ. Geol. 82 (1987) 1169–1176.

[3] T.K. Kyser, M.R. Wilson, G. Ruhrmann, Stable isotope con-straints on the role of graphite in the genesis of unconformity-typeuranium deposits, Can. J. Earth Sci. 26 (1989) 490–498.

[4] N.S.F. Wilson, L.D. Stasiuk, M.G. Fowler, Post-mineralizationorigin of organic matter in Athabasca unconformity uranium de-posits, Saskatchewan, Canada, Summary of investigations 2002,Saskatchewan Geological Survey, vol. 2, 2002, 6p.

[5] J.A. Curiale, Origin of solid bitumens, with emphasis on bio-logical marker result, Org. Geochem. 10 (1986) 559–580.

[6] V. Ruzicka, Unconformity-associated uranium, in: O.R. Eck-strand, W.D. Sinclair, R.I. Thorpe (Eds.), Geology of CanadianMineral Deposit Types, vol. 8, Geological Survey of Canada,Toronto, 1996, pp. 197–210.

[7] F. Gauthier-Lafaye, F. Weber, The Francevillian (Lower Prote-rozoic) uranium ore deposits of Gabon, Econ. Geol. 84 (1989)2267–2285.

[8] B. Nagy, F. Gauthier-Lafaye, P. Holliger, D.W. Davis, D.J.Mosmann, J.S. Leventahl, M.J. Rigali, J. Parnell, Role of organic

http://dx.doi.org/doi:10.1016/j.epsl.2007.03.018

http://dx.doi.org/doi:10.1016/j.epsl.2007.03.018


matter in containment of uranium and fissiongenic isotopes at theOklo natural reactor, Nature 354 (1991) 472–475.

[9] P. Landais, Organic geochemistry of sedimentary uranium oredeposits, Ore Geol. Rev. 11 (1996) 33–51.

[10] D. Derome, Evolution et origines des saumures dans les bassinsprotérozoïques au voisinage de la discordance socle/couverture.L'exemple de l'environnement des gisements d'uranium associésaux bassins Kombolgie (Australie) et Athabasca (Canada), Ph.D.thesis, Université Henri Poincaré, France, 2002.

[11] P. Landais, J. Dubessy, J.M. Dereppe, R.P. Philp, Characteriza-tion of graphite alteration and bitumen genesis in the Cigar Lakedeposit (Saskatchewan, Canada), Can. J. Earth Sci. 30 (1993)743–753.

[12] J.S. Leventhal, C.N. Threlkeld, Carbon-13/Carbon-12 isotopefractionation of organic matter associated with uranium oresinduced by alpha irradiation, Science 202 (1978) 430–432.

[13] P.A. Eakin, Isotopic and petrographic studies of uraniferoushydrocarbons from around the Irish Sea Basin, J. Geol. Soc.(Lond.) 146 (1989) 663–673.

[14] O. Ruau, B. Pradier, P. Landais, J.L. Gardette, Influence of theconditions of deposition on the chemistry and the reflectancevariations of the Brent coals, Org. Geochem. 25 (1996) 325–339.

[15] L. Sangély, M. Chaussidon, R. Michels, V. Huault, Microanalysisof carbon isotopic composition in organic matter by secondaryion mass spectrometry, Chem. Geol. 223 (2005) 179–195.

[16] J. Hoeve, T.I.I. Sibbald, On the genesis of Rabbit Lake and theother unconformity-type uranium deposits in Northern Saskatch-ewan, Canada, Econ. Geol. 73 (1978) 1450–1473.

[17] M. Fayek, T.M. Harrison, R.C. Ewing, M. Grove, C.D. Coath,O and Pb isotopic analyses of uranium minerals by ionmicroprobe and U–Pb ages from the Cigar Lake deposit,Chem. Geol. 185 (2002) 205–225.

[18] G. Lorilleux, M. Jebrak, M. Cuney, D. Baudemont, Polyphasehydrothermal breccias associated with unconformity-relateduranium mineralization (Canada): from fractal analysis tostructural significance, J. Struct. Geol. 24 (2002) 323–338.

[19] A. Wang, P. Dhamelincourt, J. Dubessy, D. Guérard, P. Landais,M. Lelaurain, Characterization of graphite alteration in an ura-nium deposit by micro-Raman spectroscopy, X-ray diffraction,transmission electron microscopy and scanning electron micros-copy, Carbon 27 (1989) 209–218.

[20] D.J. Mossman, F. Gauthier-Lafaye, S.E. Jackson, Carbonaceoussubstances associated with the Paleoproterozoic natural nuclearfission reactors of Oklo, Gabon: paragenesis, thermal maturationand carbon isotopic and trace element compositions, PrecambrianRes. 106 (2001) 135–148.

[21] J.R. Lancelot, B. de Saint André, H. de la Boisse, SystématiqueU–Pb et évolution du gisement d'uranium de Lodève, Miner.Depos. 19 (1984) 44–53.

[22] P.G. Rouxhet, P.L. Robin, G. Nicaise, Characterization of kerogensand of their evolution by infrared spectroscopy, in: B.Durand (Ed.),Kerogen, Editions Technip, Paris, 1980, pp. 163–190.

[23] P. Landais, A. Rochdi, In situ examination of coal macerals oxidationby micro-FT-i.r. spectroscopy, Fuel 72 (1993) 1393–1401.

[24] S.H. Wang, P.R. Griffiths, Resolution enhancement of diffusereflectance IR spectra of coals by Fourier self deconvolution I:C–H stretching and bending modes, Fuel 64 (1985) 229–236.

[25] L. Sangély, Development of in situ isotopic analysis in organiccarbon by ion microprobe: contribution to the study of the originof bitumen associated with uranium deposits of the Athabasca(Canada) and the Witwatersrand (South Africa), Ph.D. thesis,Institut National Polytechnique de Lorraine, France, 2004.

[26] U. Berner, E. Faber, G. Scheeder, D. Panten, Primary crackingof algal and landplant kerogens: kinetic models of isotope varia-tions in methane, ethane and propane, Chem. Geol. 126 (1995)233–245.

[27] W.J. Stahl, Source rock–crude oil correlation by isotopic type-curves, Geochim. Cosmochim. Acta 42 (1978) 1573–1577.

[28] Z. Sofer, Stable isotope compositions of crude oils: application tosource depositional environments and petroleum alteration, Am.Assoc. Pet. Geol. Bull. 68 (1984) 31–49.

[29] V.E. Andrusevich, M.H. Engel, J.E. Zumberge, L.A. Brothers,Secular, episodic changes in stable carbon isotope composition ofcrude oils, Chem. Geol. 152 (1998) 59–72.

[30] K.N. Jha, J. Gray, O.P. Strausz, The isotopic composition ofcarbon in the Alberta oil sand, Geochim. Cosmochim. Acta 43(1979) 1571–1573.

[31] R.H. Rainbird, R.A. Stern, N. Rayner, C.W. Jefferson, Age,provenance, and regional correlation of the Athabasca Group,Saskatchewan and Alberta, constrained by igneous and detritalzircon geochronology, in: C.W. Jefferson, G. Delaney (Eds.),EXTECH IV: Geology and Uranium EXploration TECHnologyof the Proterozoic Athabasca Basin, Saskatchewan and Alberta,Geological Survey of Canada, 2005.

[32] S.E. Orrell, M.E. Bickford, J.F. Lewry, Crustal evolution and ageof thermometric reworking in the Western hinterland of theTrans-Hudsonian orogen, Northern Saskatchewan, PrecambrianRes. 95 (1999) 187–223.

[33] D.J. DesMarais, Isotopic evolution of the biogeochemical carboncycle during the Proterozoic Eon, Org. Geochem. 27 (1997)185–193.

[34] Y. Watanabe, H. Naraoka, D.J. Wronkiewicz, K.C. Condie, H.Ohmoto, Carbon, nitrogen, and sulfur geochemistry of Archeanand Proterozoic shales from the Kaapvaal Craton, South Africa,Geochim. Cosmochim. Acta 61 (1997) 3441–3459.

[35] L. Richard, L. Sangély, Thermodynamic and carbon isotopicconstraints on the origin of unusual bitumens in the uraniumdeposits of Athabasca (Saskatchewan, Canada), 14th AnnualGoldschmidt Conference, Copenhagen, Denmark, 2004.

[36] P. Bruneton, Geology of the Cigar Lake uranium ore-deposit(Saskatchewan, Canada), in: C.F. Gilboy, L.W. Viagrass (Eds.),Economic Minerals of Canada, vol. 8, Saskatchewan GeologicalSpecial Publication, Toronto, 1987, pp. 99–119.

[37] B. Sherwood-Lollar, T.D. Westgate, J.A. Ward, G.F. Slater, G.Lacrampe-Couloume, Abiogenic formation of alkanes in theEarth's crust as a minor source for global hydrocarbon reservoirs,Nature 416 (2002) 522–524.

[38] D.E. Ingmanson, M.J. Dowler, Unique amino acid compositionof Red Sea brine, Nature 286 (1980) 51–52.

[39] J.A. Welhan, H. Craig, Methane, hydrogen and helium inhydrothermal fluids at 21 degrees N on the East Pacific Rise,in: P.A. Rona, K. Bostroem, L. Laubier, K.L. Smith Jr. (Eds.),Hydrothermal Processes at Seafloor Spreading Centers, PlenumPress, New York, USA, 1983, pp. 391–409.

[40] J.A. Welhan, J.E. Lupton, Light hydrocarbon gases in GuaymasBasin hydrothermal fluids: thermogenic versus abiogenic origin,Am. Assoc. Petrol. Geol. Bull. 71 (1987) 215–223.

[41] R. Botz, D. Stüben, G. Winckler, R. Bayer, M. Schmitt, E. Faber,Hydrothermal gases offshore Milos Island, Greece, Chem. Geol.130 (1996) 161–173.

[42] N.G. Holm, J.L. Charlou, Initial indications of abiotic formationof hydrocarbons in the Rainbow ultramafic hydrothermalsystem, Mid-Atlantic Ridge, Earth Planet. Sci. Lett. 191 (2001)1–8.


[43] C. Riedel, M. Schmidt, R. Botz, F. Theilen, The Grimsey hydro-thermal field offshore North Iceland: crustal structure, faulting andrelated gas venting, Earth Planet. Sci. Lett. 193 (2001) 409–421.

[44] J.L. Charlou, J.P. Donval, Y. Fouquet, B.P. Jean, N. Holm,Geochemistry of high H2 and CH4 vent fluids issuing fromultramafic rocks at the Rainbow hydrothermal field (36 degrees14N, MAR), Chem. Geol. 191 (2002) 345–359.

[45] T.M. Gerlach, Chemical characteristics of the volcanic gases fromNyiragongo lava lake and the generation of CH4-rich fluid inclusionsin alkaline rocks, J. Volcanol. Geotherm. Res. 8 (1980) 177–189.

[46] N.E. Podkletnov, E.K. Markhinin, New data on abiogenic syn-thesis of prebiological compounds in volcanic processes, Orig.Life Evol. Biosph. 11 (1981) 303–315.

[47] B. Capaccioni, M. Martini, F. Mangani, L. Giannini, G. Nappi, F.Prati, Light hydrocarbons in gas-emissions from volcanic areasand geothermal fields, Geochem. J. 27 (1993) 7.

[48] B. Sherwood-Lollar, S.K. Frape, S.M. Weise, P. Fritz, S.A.Macko, J.A. Welhan, Abiogenic methanogenesis in crystallinerocks, Geochim. Cosmochim. Acta 57 (1993) 5087–5097.

[49] B. Sherwood Lollar, G. Lacrampe-Couloume, G.F. Slater, J. Ward,D.P. Moser, T.M. Gihring, L.-H. Lin, T.C. Onstott, Unravellingabiogenic and biogenic sources of methane in the Earth's deepsubsurface, Chem. Geol. 226 (2006) 328–339.

[50] T.A. Abrajano, N.C. Sturchio, J.K. Bohlke, G.L. Lyon, R.J. Poreda,C.M. Stevens, Methane-hydrogen gas seeps, Zambales Ophiolite,Philippines: deepor shalloworigin?Chem.Geol. 71 (1988) 211–222.

[51] T.A. Abrajano, N.C. Sturchio, B.M. Kennedy, G.L. Lyon, K.Muehlenbachs, J.K. Bohlke, Geochemistry of reduced gas re-lated to serpentinization of the Zambales ophiolite, Philippines,Appl. Geochem. 5 (1990) 625–630.

[52] I.A. Petersilie, H. Sörensen, Hydrocarbon gases and bituminoussubstances in rocks from the Ilimaussaq alkaline intrusion, SouthGreenland, Lithos 3 (1970) 59–76.

[53] D.S. Kelley, Methane-rich fluids in the oceanic crust, J. Geophys.Res. 101 (B2) (1996) 2943–2962.

[54] S. Salvi, A.E. Williams-Jones, Fischer-tropsch synthesis of hy-drocarbons during sub-solidus alteration of the Strange Lakeperalkaline granite, Quebec/Labrador, Canada, Geochim. Cos-mochim. Acta 61 (1997) 83–99.

[55] D. Kelley, G. Früh-Green, Abiogenic methane in deep-seatedmid-ocean ridge environments: insights from stable isotopeanalyses, J. Geophys. Res. 104 (B5) (1999) 10,439–10,460.

[56] E.A. Mathez, Carbonaceous matter in mantle xenoliths: compo-sition and relevance to the isotopes, Geochim. Cosmochim. Acta51 (1987) 2339–2347.

[57] R. Sugisaki, K. Mimura, Mantle hydrocarbons: abiotic or biotic?Geochim. Cosmochim. Acta 58 (1994) 2527–2542.

[58] N.G. Holm, J.L. Charlou, Initial indications of abiotic formationof hydrocarbons in the Rainbow ultramafic hydrothermal system,Mid-Atlantic Ridge, Earth Planet. Sci. Lett. 191 (2001) 1–8.

[59] T.N. Tingle,M.F. Hochella Jr., C.H. Becker, R.Malhotra, Organiccompounds on crack surfaces in olivine from San Carlos, Arizonaand Hualalai Volcano, Hawaii, Geochim. Cosmochim. Acta 54(1990) 477–485.

[60] T.N. Tingle, E.A. Mathez, M.F. Hochella Jr., Carbonaceous mat-ter in peridotites and basalts studied by XPS, SALI, and LEED,Geochim. Cosmochim. Acta 55 (1991) 1345–1352.

[61] E.A. Mathez, D.M. Mogk, Characterization of carbon com-pounds on a pyroxene surface from a gabbro xenolith in basalt bytime-of-flight secondary ion mass spectrometry, Am. Mineral. 83(1998) 918–924.

[62] M.A. Sephton, I. Gilmour, Pyrolysis-gas chromatography-isotoperatio mass spectrometry of macromolecular material in meteor-ites, Planet. Space Sci. 49 (2001) 465–471.

[63] M.A. Van Zuilen, A. Lepland, G. Arrhenius, Reassessing theevidence for the earliest traces of life, Nature 418 (2002)627–630.

[64] M.D. Brasier, N.V. Grassineau, O.R. Green, A.P. Jephcoat, A.K.Kleppe, M.J. Van Kramendonk, J.F. Lindsay, A. Steele, Question-ing the evidence of Earth's oldest fossils, Nature 416 (2002)76–81.

[65] J.D. Pasteris, B. Wopenka, Necessary but not sufficient: Ramanidentification of disordered carbon as a signature of ancient life,Astrobiology 3 (2003) 727–738.

[66] F.J. Luque, J.D. Pasteris, B. Wopenka, M. Rodas, J.F.Barrenenechea, Natural fluid-deposited graphite: mineralogicalcharacteristics and mechanisms of formation, Am. J. Sci. 298(1998) 471–498.

[67] M.Y. Zolotov, E.L. Shock, A thermodynamic assessment of thepotential synthesis of condensed hydrocarbons during coolingand dilution of volcanic gases, J. Geophys. Res. 105 (B1) (2000)539–559.

[68] P.M. Maitlis, Fischer–Tropsch, organometallics, and other friends,J. Organomet. Chem. 689 (2004) 4366–4374.

[69] T.M. McCollom, J.S. Seewald, A reassessment of the potentialreduction of dissolved CO2 to hydrocarbons, Geochim. Cosmo-chim. Acta 65 (2001) 3769–3778.

[70] J. Horita, M.E. Berndt, Abiogenic methane formation andisotopic fractionation under hydrothermal conditions, Science285 (1999) 1055–1057.

[71] T.M. McCollom, G. Ritter, B.R.T. Simoneit, Lipd synthesisunder hydrothermal conditions by Fischer–Tropsch-type reac-tions, Orig. Life Evol. Biosph. 29 (1999) 153–166.

[72] M.E. Berndt, D.E. Allen, W.E.J. Seyfried, Reduction of CO2

during serpentinization of olivine at 300 °C and 500 bar, Geology24 (1996) 351–354.

[73] E.L. Shock, Geochemical constraints on the origin of organiccompounds in hydrothermal systems, Orig. Life Evol. Biosph. 20(1990) 331–367.

[74] T.M. McCollom, Formation of meteorite hydrocarbons fromthermal decomposition of siderite (FeCO3), Geochim. Cosmo-chim. Acta 67 (2003) 311–317.

[75] A.I. Rushdi, B.R.T. Simoneit, Lipid formation by aqueousFischer–Tropsch-type synthesis over a temperature range of 100to 400 °C, Orig. Life Evol. Biosph. 31 (2001) 103–118.

[76] J. Dubessy, M. Pagel, J.M. Beny, H. Christensen, B. Hickel, C.Kosztolany, B. Poty, Radiolysis evidenced by H2–O2 and H2

bearing fluid inclusions in three uranium deposits, Geochim.Cosmochim. Acta 52 (1988) 1155–1167.

[77] M.S. Lancet, E. Anders, Carbon isotope fractionation in theFischer–Tropsch synthesis and in meteorites, Science 170 (1970)980–982.

[78] K. Wedeking, R. Ship, S. Chang, E. Rightor, T.J. Pinnavaia,Carbon isotopic compositions of individual hydrocarbons, carbondioxide, and oils from the Fischer–Tropsch reaction, 16th Lunarand Planetary Science Conference, Houston, USA, 1985.

[79] G.U. Yuen, J.A. Pecore, J.F. Kerridge, T.J. Pinnavaia, E.G. Rightor,Carbon isotopic fractionation in Fischer–Tropsch type reactions,21st Lunar and Planetary Science Conference, Houston, USA,1990.

[80] J. Horita, Some perspectives on isotope biosignatures for earlylife, Chem. Geol. 218 (2005) 171–186.


[81] T.M. McCollom, J.S. Seewald, Carbon isotope composition oforganic compounds produced by abiotic synthesis under hydro-thermal conditions, Earth Planet. Sci. Lett. 243 (2006) 74–84.

[82] S. Chang, D. Des Marais, R. Mack, S.L. Miller, G.E. Strathearn,Prebiotic organic syntheses and the origin of life, in: W.J. Schopf(Ed.), Earth's Earliest Biosphere — Its Origin and Evolution,Princeton Univ. Press, Princeton, 1983, pp. 53–92.

[83] E.M. Galimov, The Biological Fractionation of Isotopes,Academic Press Inc, Orlandor, 1985.

[84] G.P. van der Laan, Kinetics, selectivity and scale up of theFischer–Tropsch synthesis, Ph. D. thesis, Groningen University,Netherlands, 1999.

[85] P. Szatmari, Petroleum formation by Fischer–Tropsch syn-thesis in plate tectonics, Am. Assoc. Pet. Geol. Bull. 73 (1989)989–998.

[86] E.M. Galimov, Isotopic criteria for identification of organiccarbon on Earth and meteorites, Space Sci. Rev. 106 (2003)249–262.

[87] M. Brasier, O. Green, J. Lindsay, A. Steele, Earth's oldest(3.5 Ga) fossils and the “early eden hypothesis”: questioning theevidence, Orig. Life Evol. Biosph. 34 (2004) 257–269.

Micrometer scale carbon isotopic study of bitumen associated with Athabasca uranium deposits: Constraints on the genetic relationship with petroleum source-rocks and the abiogenic

Documents