The dating of petroleum fluid residence time in subsurface reservoirs. Part 1: A radiolysis-based geochemical toolbox

Available online at www.sciencedirect.com

ScienceDirect Geochimica et Cosmochimica Acta 261 (2019) 305–326 www.elsevier.com/locate/gca

The dating of petroleum fluid residence time in subsurface reservoirs. Part 1: A radiolysis-based geochemical toolbox Steve Larter a, Renzo C. Silva a,⇑, Norka Marcano a,1, Lloyd R. Snowdon a, J. Eduardo Villarreal-Barajas b,2, Roshanak Sonei a, Lydia C. Paredes Gutie´rrez c, Haiping Huang a, Andrew Stopford a,3, Thomas B.P. Oldenburg a, Jing Zhao a, Priyanthi Weerawardhena a, Michael Nightingale d, Bernhard Mayer d, Jon H. Pedersen e, Rolando di Primio e a

PRG, Department of Geoscience, University of Calgary, T2N 1N4, Canada b Department of Oncology, University of Calgary, T2N 4N1, Canada c Instituto Nacional de Investigaciones Nucleares, C.P. 52750, Mexico d AGg, Department of Geoscience, University of Calgary, T2N 1N4, Canada e Lundin Norway, 1366 Lysaker, Norway

Received 25 April 2019; accepted in revised form 9 July 2019; available online 17 July 2019

Abstract The radiometric dating of geological events was a crucial achievement leading to the establishment of the geological time scale. The dating of the timing of petroleum charge into, and the determination of petroleum residence times in a geological trap would also be significant, as it would remove ubiquitous speculation concerning the history of reservoir charging in any basin setting. A thorough review of prior strategies to estimate the residence time of petroleum fluids in subsurface reservoirs revealed that few if any methods currently used provide useful estimates of the residence age of fluids. This paper is focused on the age dating of petroleum residence time in reservoirs based on the compositional alterations of reservoired fluids caused by natural radiation. This preliminary paper sets out the geochemical landscape and constraints on radiation-based age dating proxies. We report results on the propagation of radiation through reservoir rocks, which indicate that gamma radiolysis is a primary route to crude oil alteration. Gamma ray radiolysis experiments on crude oils at both natural and elevated radiation doses have been completed and the changing crude oil composition observed using LC, GC-MS and NMR. Reservoir fluids are naturally immersed in radioactive subsurface media that cause systematic alteration to crude oil composition with time, but the chemical changes are small in most natural settings. The degree of radiolysis of individual petroleum compounds was found to depend on chemical class, molecular size, initial compound concentration and the nature of the oil matrix, indicating that a proxy system for dating of petroleum charge times and oil residence times in petroleum traps will likely depend on case-specific calibrations. We define the apparent gamma ray radiolysis susceptibility (kGy
1) of different compounds. Large alkanes, such as high molecular weight normal alkanes (>C22) or hopanes, have high radiolysis susceptibility and show the most rapid decrease in component concentration with increasing radiation dose. In contrast, condensed aromatic hydrocarbons and diamondoid hydrocarbons are more resistant to decomposition through radiolysis. While assessing the decrease in ⇑ Corresponding author. 1 2 3

E-mail address: [email protected] (R.C. Silva). Present address: Schlumberger, Dubai. Present address: Radiotherapy Physics, Royal Devon and Exeter Hospital, EX2 5DW, England, United Kingdom. Present address: Department of Biological Sciences, University of Calgary, T2N 1N4, Canada.

https://doi.org/10.1016/j.gca.2019.07.020 0016-7037/Ó 2019 Elsevier Ltd. All rights reserved.

306

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

concentration of a compound through radiolysis is a practical objective, it is more difficult to assess the production of new, unique radiolysis products given the great diversity and low concentrations of individual compounds produced. However, monitoring the production of specific functional groups during radiolysis of crude oils, using NMR spectroscopy, was found to be a feasible proxy analytical target. Analysis of newly generated carbon-carbon double bonds in bulk crude oils may represent an optimal approach for development as a radiolysis proxy. We propose a route to assessing in-reservoir crude oil radiation dose. Ó 2019 Elsevier Ltd. All rights reserved. Keywords: Residence time; Age dating; Experimental radiolysis; Gamma irradiation; Petroleum geochemistry

1. INTRODUCTION The dating of geological events initiated modern geoscience (Holmes, 1911), ending not only the debate over the age of the Earth but also permitting, for the first time, quantitative evaluation of geological processes. Dating petroleum charge times, oil residence times and hydrocarbon fluid charge rates into a trap would also be significant and would reduce the speculation about reservoir petroleum charging times and migration routes common in discussions about petroleum systems (Larter et al., 2012). While much effort has been put into using forward basin models to estimate the timing of petroleum charging, there are no independent constraints on these determinations, even though fluid residence time in a given reservoir is a fundamental factor in constraining any petroleum system evaluation. Commonly, oil maturity and the location of petroleum accumulations are used together to help constrain basin models (e.g., Higley et al., 2009), but the estimates of the times of petroleum charging from such models are usually non-unique. Driven largely by the timing of petroleum generation and initial primary and secondary migration events, most basin modelling studies do not incorporate late stage basin tilting or other events related to late stage petroleum remigration from initial accumulations. While forward basin models can provide estimates of initial oil charge timing, they are not well constrained by any real measurements made on the crude oil and have large errors associated with them, sometimes extending to a significant portion of the age of a reservoir (Larter et al., 2012). Petroleum charge times and rates are key variables in controlling hydrocarbon prospectivity, as they define volumes of trapped petroleum and the dynamics of trap integrity, including leakage and alteration phenomena. The dating of source and reservoir rock deposition by radiometrically calibrated biostratigraphic assessments is now routine and effective (Erwin, 2006), and some progress has been made in remotely dating source rocks from crude oil analysis, using age-related biomarkers (Peters et al., 2005, p490). However, the dating of fluid flow events in petroleum systems is currently based on indirect methods, which is surprising since many petroleum system phenomena related to formation and alteration of accumulations are highly time dependent. Thus the extent of crude oil alteration by biodegradation is controlled by the thermal and oil charge history of a reservoired petroleum column (Larter et al., 2003; Larter et al., 2006) and the

time-dependent mixing of reservoir fluids (Koopmans et al., 2002; Wilhelms and Larter, 2004). The apparent residence time of fluids in a subsurface reservoir will depend on position in the fluid column, the fluxes of fluids into and out of the trap, and diffusive and advective mixing of fluids in the reservoir (Fig. 1). Such fluid mixing and age assessment challenges are well described for ocean water age assessment (Deleersnijder et al., 2001). In principle, examination of the residence age spatial profiles in a fluid column, coupled to numerical models of fluid flux, might allow estimates of the charge, spillage and leakage fluid fluxes into and out of the trap. Such constraints might also permit more realistic assessment of caprock efficiency when limited cap rock material is available (Larter et al., 2012). In addition to petroleum trap studies, such constraints would also be important in the assessment of caprocks associated with potential CO2 storage sites for carbon

q2

q1

Residence time profile (tres)

q3 Fig 1. Schematic illustration of fluid fluxes (oil, gas, water, CO2) into and out of a reservoir. For example, a charge flux of say, oil (q1), contributes via storage and transport in the reservoir to leakage through the caprock (q2) and spillage from the spill point of the trap (q3). The residence age profile (tres), is hypothetical and may reflect different and time variable fluxes at different locations, multiple charge episodes and permeability heterogeneity within the reservoir unit. By age dating the fluid residence time profile in the reservoir and comparing it with reservoir medium radionuclide distribution using a fluid age mixing model, in principle, the charge history and fluxes may be delimited (after Larter et al., 2012).

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

capture and storage programs (CCS). There have also been suggestions to use biodegradation kinetic approaches to assess charge histories (Larter et al., 2003), and recent attempts to assess oil charge histories using radiometric dating methods (Selby and Creaser 2005), but results remain equivocal. Dating fluid residence times in the subsurface remains arguably the biggest missing proxy system in petroleum geology! Past and current advances in this research area are briefly reviewed in the following sub-sections. 1.1. Age dating approaches Several approaches have tried to assess fluid charge and residence histories in subsurface petroleum reservoirs. Basin modeling principally provides source rock maturation times (petroleum generation), not actual local charge times, but it is commonly assumed that most reservoirs are charged during an initial migration stage initiated by petroleum generation in the source rocks. Complex migration routes and histories, resulting from much later stage basin tilting and reservoir spilling can be impacted both by tectonics and even by very late stage ice sheet withdrawal and basin rebound and tilting effects. These sometimes very late stage processes can drastically affect remigration and change accumulation locations dramatically after ice sheet movement (Grasby and Betcher, 2000; Grasby and Chen, 2005). True oil charge and petroleum residence ages in many major petroleum provinces are in truth essentially unknown but can substantially impact charge history and prospect location. Diagenetic approaches to estimating fluid charge history have been extensively tried, based either on: (a) the impact of varying water saturations on the diagenetic evolution of mineral systems that provide either radiometric dating couples (e.g. K/Ar dating of authigenic illite growing in reservoirs, as in Glasmann, 1992); or (b) the assessment of phase behavior of fluids trapped in fluid inclusions in authigenic mineral phases (Karlsen et al., 1993). Both approaches have deficits. K/Ar dating on illites rarely gives very recent geological ages, and the complexities of incomplete feldspar diagenesis, authigenic illite growth and recrystallization in reservoirs tends to produce complex data sets that are hard to interpret (Glasmann, 1992). Fluid inclusions in quartz or calcite can trap water or live petroleum phases, which can provide trapping temperature and trapping time estimates and, through the use of vapor/liquid ratio and homogenization temperature characterization, provide information on the composition of oil as a function of time during the oil charge process (Aplin et al., 2000). By crushing well-cleaned, separated authigenic mineral phases under solvent, some characterization of the petroleum trapped in the inclusions can be made (Karlsen et al., 1993). In terms of dating residence time of fluids in the petroleum column however, the issue is that fluid inclusions commonly only track the earliest phases of petroleum charge, typically being found close to the detrital mineral grain in the authigenic overgrowth, and there is no clear relationship between fluid inclusion occurrence and the actual volume of petroleum in the porosity of the reservoir at any given time. Therefore, fluid inclusions provide

307

evidence of a date and composition for the earliest charge in the petroleum reservoir but, in truth, provide no direct information on the actual residence time distribution of the fluids currently in the reservoir porosity today. The proposed Re-Os method for dating a variety of petroleum system processes and rock units (Selby and Creaser 2005) looks interesting, as osmium isotopic compositions in oils show a considerable range and correlate positively with proposed source rock age. A correlation with a source rock age may constrain the timing of the reservoir filling to be not older than the source but cannot track an oil charge time. Similarly, the oil charge date cannot be older than the age of the reservoir itself. The method was used to suggest a 112 Ma charge event for the Alberta oil sands, but this date is also near the reservoir age (Hein and Cotterill, 2006), suggesting that the Re-Os signal may be related to authigenic clay minerals deposited along with the sand in the reservoir and recovered along with the high molecular weight and polar compounds during extraction and fractionation rather than as organo-metallic or other migrated components within the oil. Tozer et al. (2014) suggested that, based on a variety of geological evidence related to trap geometry and even the timing of igneous intrusions, the actual charge times in the Alberta oil sands are later, but such arguments merely emphasize the need for a reliable fluid residence time assessment method. The isochrons purported by Selby and Creaser (2005) are well defined, though similar relationships might result from fluid exchange and mixing of measured species between oils and water and, in principle, might not represent a true isochron at all. Both rhenium and osmium present in the system appear to be associated with the resin and asphaltene fractions of oils and thus may well be source related. Selby et al. (2007) suggested that 187Re/188Os and 187 Os/188Os ratios in asphaltene at the time of oil generation are like those of the whole oil, proposing that asphaltene fractions can be used to approximate the Re–Os isotopic compositions of the whole oil. For a system capable of dating oil charge time, however, any radiometric signal carried from the source rock would carry information on either oil source age or expulsion time from the source rock, but it is not obvious how it would carry the signal for the time the oil enters the reservoir. Location of the metals (rhenium and osmium) and their exchange between fluids (oil and water) and solids (source rock, carrier bed, and reservoir), in the system are not well defined, and mechanisms for how an internal radiometric clock can track oil charge time or residence time in the current reservoir remain uncertain. Nevertheless, definition of this system as a proxy represents a key step, in the sense that development of methods for the age dating of oil charge is now an area of active research. Another approach to recovering charge history information from a petroleum column, in principle, is to use inversion of a physical model of petroleum charging and mixing, constrained by compositional variations within a petroleum column. Such an approach, for example, could look at the diffusive decay of the compositional wave produced by injecting a more mature petroleum into an earlier, lower maturity reservoir fluid column. The fluid injection rates and timescales could in principle be inverted from analysis

308

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

of the fluid composition and property information through the oil column if reliable diffusion coefficient estimates were available. Similar approaches, combining numerical models and measured compositional gradients have been used to study charge and biodegradation rates in heavy oilfields (Larter et al., 2003), and the timescales of biodegradation. Such approaches, using reservoir simulators, could in principle track the time-dependent compositional changes recorded in petroleum columns as different oils enter a reservoir. 1.2. Crude oil radiolysis as a chronometer The most robust petroleum fluid residence history assessment methods must involve compositional or concentration variations in petroleum hosted components that do not exchange chemically with reservoir media (minerals, water, and any organic materials), and which are not affected by secondary alteration processes such as biodegradation, thermochemical sulfate reduction or evaporative fractionation. Ideal methods must also be able to distinguish reservoir aging impacts from any compositional changes relating to prior oil history in the source rock or carrier bed system. Ideally, the method should work with gases as well as liquid petroleum components. Radiation impacts from radioactive nuclides in a reservoir rock, affecting trapped fluid composition in the reservoir seems to potentially be a viable route to an age dating proxy. Crude oils are very reactive under high radiation doses, with excited hydrocarbon and heteroatomic molecules being created through non-selective bond breaking and bimolecular recombination reactions including hydrogen loss, condensation processes and reactions between any species present, including water, N2 and CO2. Classical geochemical proxy routes commonly involve one or a few related precursors, converting reversibly or irreversibly to one or a few closely related products. In contrast, specific radiolysis products are often hard to measure, as large numbers of low concentration species are produced, even from radiolysis of simple binary compound mixtures (Larter et al., 2012). Frolov et al. (1998) and Curiale and Frolov (1998), described the production of alkenes in crude oils from natural radiation damage (radiolysis) and suggested that radiolytic olefins may be indicators of the residence time of oils in the present trap. This was a crucial observation but, as reported for petroleum mixtures (Frolov et al., 1998), radiation-induced unsaturation has a complex distribution through essentially all organic fractions, making detection of radiation-induced daughter species by classical geochemical approaches, such as GC-MS, very challenging. Therefore, spectroscopic approaches, such as nuclear magnetic resonance (NMR), are better suited to accurately assess alkene production (Frolov and Smirnov, 1994; Frolov et al., 1996; Frolov et al., 1998). Simply calibrating chemical compositional changes to an incurred radiation dose is not sufficient for a practical chronometric system, as it has been demonstrated that with very radioactive source rocks, such as the Cambrian Alum shale, both kerogen and in-source hosted petroleum are compositionally altered by radiation (Bharati et al., 1995).

While most source rocks are not exposed to Alum shale levels of irradiation, nevertheless, given the higher radionuclide concentrations in shales versus for example, typical reservoir lithologies, and longer residence times for organic matter in source rocks compared to storage times in reservoirs, the radiation dose experienced by organic matter in a source rock (kerogen or petroleum) will likely exceed that experienced by an oil in a reservoir. Any practical reservoir petroleum fluid residence time proxy system must therefore be able to deal with this phenomenon, so innovative approaches are needed to decouple source and reservoir irradiation effects. Another issue is that materials with the high density of rock minerals rapidly attenuate gamma radiation over quite short distances (see Section 3.1). This observation implies that local radionuclide distribution will represent a local radiation source signal for any reservoir fluid-based radiation impacts. This suggests that detection of locally generated radiolysis profiles in oils is a potential route to effective decoupling of prior radiation impacts from the source rock, or from hoteling events in intermediate reservoirs, from the radiation impacts accrued by the final storage of the oil in its current terminal reservoir. This, in turn, requires that compositional profiles through a petroleum column, rather than just single sample analyses, are the route to a chronometer and that mass transport effects and fluid mixing models are key elements that must be included in any interpretation of a fluid residence age proxy system data set. If calibrated for the reservoir radioactive isotope load, measurement of radiation dose profiles in reservoired petroleum, assessed by chemical analysis of crude oils or gases (e.g. CO2), could in principle provide routes to dating reservoir fluid residence time. Radiation production and transport through rocks is complicated and proceeds by a wide range of mechanisms. Nuclear radiation is primarily generated from the alpha emitting decay of uranium (238U) and thorium (232Th) isotopes, and from the beta particle emitting decay of potassium (40K). The primary particles emitted from decaying radionuclides, however, are rapidly stopped by interactions with atoms in the host minerals and are mostly attenuated within the grains of the host minerals themselves, ultimately releasing, through a complex chain of interactions and daughter species, photon radiation in the form of gamma rays. The linear energy transfer properties of 2.5 MeV alpha particles are around 166 keV/mm (Bailey et al., 2015), and thus alpha particles are rapidly stopped by solid and liquid phases in organic matter (Bailey et al., 2015) and rocks. Alpha particles of the uranium and thorium series are, however, sufficiently energetic to travel around 20 mm through minerals (Farley, 2002) and the high mass ions occasionally produced by nuclear fission, primarily of 238U, and which form the familiar fission tracks in apatites and other uranium-bearing minerals, are commonly attenuated in distances of around 16 mm in the host minerals (Gleadow et al., 1986; Gallagher et al., 1998). Thus, in sandstones, primary ions (or most electron-beta emission) will be retained by the primary mineral grain hosting the radionuclide, except for particles emitted from radionuclides that occur within 20 mm of the edge of the mineral grain. For radionuclides within 20 mm of the edge of the mineral grain,

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

a small proportion of emitted alpha particles will exit the grain, either into an adjacent mineral grain or potentially into a pore fluid. Therefore, gamma radiation produced during the radioactive decay series would be the primary irradiator of pore fluids such as water, oil or gas in conventional reservoirs such as sandstones. In siltstones and shales, a much larger fraction of primary radiation would exit the host grain, entering other grains and pore fluids. The interaction of gamma ray photons with atoms in fluids (or other phases) has several possible outcomes depending on the photon energy. At relatively low photon energies, the photon can be absorbed completely through the photoelectric effect, ejecting an electron, while at high photon energies pair production or even a photonuclear reaction is possible (Bailey et al., 2015). With the typical gamma ray photon energies seen in natural radioactivity in sediments (1MeV) and with the common elements found in rock forming minerals, photon scattering is the predominant mechanism of gamma ray attenuation. A gamma ray photon can be scattered by interaction with electrons in the target atoms but keep its energy (Rayleigh scattering), or it may lose part of its energy through the Compton effect, whereby photons interact with the electrons in atoms, losing energy to produce both lower energy photons and an avalanche of emitted secondary electrons (Bailey et al., 2015). Compton scattered gamma ray photons become lower energy gamma rays, X-rays and ultimately infrared photons. Compton scattering is the predominant mechanism of gamma ray attenuation in rocks and the electron avalanche associated with this process is a primary cause of chemical change in irradiated materials. Ultimately, the energy of radionuclide fission ends up as thermal energy in the rock. Established correlations between gamma-ray emission intensity and in situ heat generation in sediments have been proposed (Bu¨cker and Rybach, 1996). As gamma rays can transit distances of several centimeters in rocks, whereas alpha and beta particles will be stopped within the mineral grains, we hypothesize that gamma ray radiolysis of pore fluids is a key process in coarse-grained sedimentary rocks, with Compton scattered electrons being a key mediator of chemical change. The transit distance of gamma ray photons in sediments (tens of centimetres) means that local variations in source gamma ray signal within the sedimentary column, e.g., the local occurrence of more radioactive shales in otherwise less radioactive sandstone or carbonate reservoir sections for example, could produce inherited signals in fluid chemistry in reservoir sections adjacent to the shales, reflecting lithologically controlled, local gamma ray source variation. This is an important inference and hypothesis that, if validated, would mean that the local, in-reservoir, gamma ray signal heterogeneity should then be evident on a time-dependent basis in the pore fluid chemistry. This of course assumes that suitable proxies are identified and sensitive enough analytical methods are developed. Local, inreservoir control of at least part of the radiolytic signal seen in crude oil or other pore fluids, would mean that reservoir residence effects could, theoretically, be distinguished from radiolytic effects on organic matter composition introduced

309

in a source rock, if the complicating effects of diffusive and effective mixing of proxy components can be accounted for. Bailey et al. (2015) described in detail the impact of nuclear radiation in fluid media in the context of typical nuclear radiotherapy medical practices. Such a scenario is an excellent analogy for the irradiation of organic fluids in reservoir pores. The irradiation of biological material (mostly water with some organic materials) gives rise to the production of a flux of energetic secondary electrons. These move away from where they are produced and through interactions with atoms lose their energy to the surrounding medium as they migrate. This energy absorption process gives rise to radicals and other chemical species and it is the chemical interactions involving these active species that are the mediators of most radiolysis damage in organic molecules. Bailey et al. (2015), in their review, noted that chemical changes associated with radiolysis take place on much longer time scales (around 10
5 s), than the initial ionizing events (timescale of around 10
18 s). Bailey et al. (2015) also noted that the main mechanism by which energy is transferred from incident nuclear radiation to organic molecules is via the avalanche of secondary electrons produced. These authors also indicated that the distinction between the direct action of radiation on organic molecules and subsequent indirect action involving interaction of radiation with atoms or molecules, such as water, to make reactive species is very important, with the production of reactive secondary free radicals causing most alteration and ‘‘radiolysis” of organic molecules. As water is a ubiquitous component of most oil reservoir pore components, indirect reaction effects between oxygen species derived from ionized water and petroleum components might also be expected to be important in reservoirs. 1.3. Chemical responses of crude oils to radiation Nuclear radiation has been proposed as a tool for petroleum refining, including combined radiation-thermal cracking, desulfurization and demetallization of crude oils (Zaykina et al., 2002a; Zaykina et al., 2002b; Basfar and Mohamed, 2011). High energy electron beams have also been used for transformation of crude oil to more desirable lower molecular weight products or to reduce oil viscosity (Mirkin et al., 2003). In general, it appears that combinations of radiolytic and more conventional thermal processing strategies seem to be the most effective (Zaykin et al., 2004; Basfar and Mohamed, 2011). Radiation treatment of crude oil tends to increase the proportion of low boiling components (Zaykin and Zaykina, 2004). However, in some cases, increasing the radiation dose rate to very high levels increases the level of alteration and reduces the yield of lower molecular weight species by isomerization and polymerization reactions of crude oil components during radiolysis (Zaykin et al., 2004). Radiolysis treatment has been applied to environmental cleanup problems such as BTEX (benzene, toluene, ethylbenzene, xylene) removal from groundwater (Almeida et al., 2006). The presence of water also impacts the products of radiolysis of petroleum fractions. In a study of the electron

310

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

beam radiolysis of gaseous propane in the presence of water, alcohols and ethers were found to be the main products, though aldehyde and ketone species were detected, as well as alkenes (Ponomarev et al., 2002). The production of oxygenated species has also been observed in the irradiation of natural bitumen (Kovaleva and Yushkin, 2006), with both carbonyl and ether bridge species formed. Coupling reactions, initiating cross-linking and causing an increase in molecular weight, were also a major process observed by Kovaleva and Yushkin (2006). Curiale et al. (1983) suggested a model in which radiation induced effects explain the occurrence of uraniferous solid bitumen in Southwestern Oklahoma. Lewan and Buchardt (1989) also indicated that cross-linking processes in organic matter were evident in the highly uraniferous and naturally irradiated Cambrian Alum shale kerogen. NMR analysis of the Alum shale from Scandinavia also showed increases in the abundance of oxygenated species in kerogens (Bharati et al., 1995). Though the mechanisms for the production of these oxygenated species were not discussed, the ubiquitous presence of water in petroleum systems means that, inevitably, reactive oxygen species derived from water radiolysis will always be present (Head et al., 2014). With hindsight, the high oxygen contents of the Alum shales probably originated from radiolytic reactive oxygen species produced from water that subsequently reacted with the organic matter in the shale. Reactive oxygen species production from water radiolysis in subsurface environments has also been suggested as a potential activation route for alkanes during the early stages of petroleum biodegradation (Head et al., 2014). In general, the energies of gamma rays and other nuclear radiation particles are sufficient to chemically activate all species present in any irradiated mixture. Frolov et al. (1998) examined the origin of olefins in some Russian oils and performed laboratory experiments irradiating crude oils in ampoules with gamma rays at doses up to 12,000 kGy (1 Gy = 1 J of absorbed radiation energy by 1 kg of matter). Irradiated oils were analyzed by GC, NMR and HPLC methods, combined with ozonolysis and microanalysis to detect alkenes. It was observed that natural radiolysis of crude oil produced carbon-carbon double bonds in essentially all alkyl or cycloalkyl structural elements, in diverse petroleum fractions. These authors assumed that natural radiolysis of petroleum was mainly caused by alpha bombardment, but considered that all nuclear radiation types from uranium, thorium and potassium radionuclide system decay (alpha, beta, gamma radiation) produced essentially the same effects. As previously discussed, the penetration distances of alpha and beta radiation are extremely short, and most crude oil is more likely irradiated by primary and Compton scattered gamma rays (and X-rays), and secondary electrons, rather than from the primary particles. Frolov et al. (1998), also indicated that radiolytic dehydrogenation of crude oil molecules results from high energy (103–107 eV) particles or gamma rays entering the crude oil and generating an avalanche of relatively low energy (20–80 eV) electrons, which results in species created through nonselective elimination of hydrogen from the excited molecules. These authors also suggested that the distribution and structures of radiogenic olefins

were not strongly influenced by the type and energy of bombardment, or the presence of water, salts or mineral phases, nor by crude oil sulfur contents or asphaltene concentrations. Olefinic hydrocarbons were found to be generated mainly from nonselective dehydrogenation of saturated hydrocarbons in the oil and increases in olefin concentration were detected at radiation doses as low as 75 kGy, with olefin production linearly correlating with total radiation dose received in laboratory experiments (Frolov et al., 1998). 1.4. Natural radiation dose estimates in geological settings Radionuclide concentrations in rocks can be used to assess their radioactivity, which is usually reported as Becquerel per kilogram (Bq/kg) (Okeyode and Akanni, 2009). Becquerel is the SI unit used for radiation emission whereas Gray is the SI unit for absorbed radiation dose. As radionuclides are mostly in the mineral grains and fluids are in the pore systems, correction factors are then needed to allow for the heterogeneous geometry and internal absorption effects of porous media to calculate effective radiation doses in the pore fluids (nGy/h per Bq/kg). However, such information is generally not available for subsurface environments. The UNSCEAR 2000 Report on the radiation impacts of rocks uses as conversion factors for absorbed dose rate in air, per unit activity, per unit of soil mass, of 0.623 for 232Th series, 0.461 for 238U series and 0.0414 for 40 K series to estimate the radiation dose rates at 1 m above ground level. The assumptions for such calculations are that the radionuclides are evenly distributed in the soil or sediments. If the same conversion factors are used, a hypothetical sandstone with 232Th = 7 ppm, 40K = 2.19 ppm, and 238U = 8 ppm, would generate a conservative radiation dose rate of approximately 540 Gy/Ma to fluids present in its pores. From a perspective more germane to subsurface geological systems, wellbore gamma ray log responses can potentially be used to estimate total in situ radiation dose rates. The radiogenic heat production within a sedimentary rock directly relates to the total emitted radiation from that sediment, as eventually most of the radiation converts into thermal energy. Ultimately, all radioactive decay products and heavy-particle radiation will be stopped or scattered within the rocks, and the produced gamma rays will undergo multiple Compton scattering producing X-rays and ultimately infrared photons, which will be absorbed. A simple correlation between gamma ray log responses (in API units) and radiogenic heat production for various sedimentary rock sections was published by Bu¨cker and Rybach (1996), as:
Heat generation lW=m3 ¼ 0:0158ðGR
0:8Þ where GR is the gamma ray log response in API units. This approach represents a very practical route to estimating local variations in total radiation dose rates in rocks. However, its major limitation is that different combinations of radionuclides can generate the same response in gamma ray log instruments. The closer the 40K/238U and 232 Th/238U ratios are to typical values for the continental

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

crust, the lower is the calibration error (Bu¨cker and Rybach, 1996). Only a fraction of the emitted high-energy radiation is effectively absorbed by the pore fluids in geological settings, depending on porosity and the densities of mineral and fluid phases. Lewan and Buchardt (1989) estimated 10% of the energy from uranium decay potentially impacting the pore fluids in the Alum shale. In fact, a description of calibration factors for effective pore fluid radiation dose rates in rocks is required. Herein, the heat generation equation of Bu¨cker and Rybach (1996) is used to estimate the general range of reservoir radiation dose rates seen in natural settings. Using this approach, typical reservoir sandstones have radiation dose rates of up to 10 kGy/Ma, whereas dose rates within intra-reservoir or cap rock shales can be much higher. Typical subsurface accumulated radiation doses in reservoirs are probably lower than 200 kGy. The low dose irradiation experiments described here have total doses ranging up to 250 kGy and are considered realistic broad analogues for the radiation dosing of pore fluids in subsurface petroleum reservoirs. 1.5. Radiation dose proxy modeling Geochemical proxies are molecular parameters based on the concentrations, isotopic compositions and/or distributions of molecular components that correlate theoretically or empirically with an environmental variable in the system of study. In the current approach, we observe proxies, based on components that can be accurately analyzed in a crude oil sample, that can be correlated with the residence time of a petroleum sample in a given reservoir setting. Hypothetically, natural nuclear radiation derived from a reservoir medium can change the composition of crude oils in a systematic way that could allow the determination of the radiation dose received by the crude oil while hosted in that reservoir. Geochemical studies in a wide variety of areas ranging from environmental to fossil fuel reseach have shown almost uniformly that mass transport phenomena are commonly a factor in proxy development and utilization. In particular, physical processes, especially ubiquitous mixing processes, greatly limit the traditional application of molecular proxies to assessing complex system properties. Thus, petroleum migration and mixing greatly complicate assessment of petroleum system properties such as maturity from analysis of reservoired oils (Wilhelms and Larter, 2004). In the context of dating petroleum fluid residence times, mass transport phenomena are also crucially important. Hence, as in the assessment of age in bodies of water in the oceans, an advection/diffusion mixing model is an essential component (Deleersnijder et al., 2001). Consequently, proxy based organic geochemical systems for assessment of petroleum residence time in a given reservoir require a physical model to interpret data. Such a model must couple the linked effects of: (a) local generation of radiolysis proxies and compound alteration effects; (b) advection of new oil charge being added to a reservoir, resulting in mass transport or bulk flow mixing; and (c) local diffusion of products and reactants within fluid

311

columns resulting in mixing of mobile proxy-precursors and products. 1.6. Objectives and overall approach For the study reported in this paper, a variety of whole crude oil radiolysis experiments using gamma rays emitted from either 60Co or 137Cs, both of which have primary gamma ray energies around 1 MeV, were carried out. These experiments, collectively, gave us insights into key radiolysis mechanisms and identified several fluid residence age proxy systems (more properly, radiation dose proxies). Knowing the accumulated radiation dose of a fluid and the dose rates provided by its reservoir permits the calculation of a ‘‘fluid residence age” estimate. In seeking such molecular proxies for accumulated radiation dose, two general types of proxy systems have been identified: (I) destruction proxies based on the systematic removal of readily identifiable and quantifiable species in crude oil that relate in some way to the incurred radiation dose; and (II) proxies for the production of new compounds or species, such as those with carbon-carbon double bonds, bi-molecular addition products or low molecular weight fragments of larger structures produced during radiolysis, where the concentration of produced species is related in some way to the incurred radiation dose. It is clear that hybrid proxies combining the aspects of type I and type II proxies might also be possible. Note: In our dating parlance, the term ‘‘date”, refers to the timing of an event, and the term ‘‘age” refers to an interpreted event. We use the term ‘‘fluid residence age” to broadly refer to the length of geological time a fluid charge has been in a reservoir location. This paper focuses predominantly on hydrocarbonrelated potential proxy systems and their mechanisms and possible utility in fluid residence age dating. A possible approach to total dose assessment, using GC-MS assessed compound loss strategies as a function of radiation dose, i.e. change in concentration of selected molecular species as a function of incurred dose is discussed. How the radiolysis sensitivity of different petroleum species depends on chemical classes is also described. Additional ongoing investigations include studies of the relative magnitude and mechanism of processes involved in source rock irradiation of organic matter, versus in-reservoir irradiation of crude oils. Additionally, an advection-diffusion-radiolysis simulation model computer program is being developed. Case histories in major petroleum basins are being used to calibrate and assess if a practical age dating proxy system is viable and feasible. These aspects of our research program will be covered in subsequent papers of this series. 2. METHODS 2.1. Experimental overview In one set of experiments, both non-biodegraded and biodegraded marine oils, contained in sealed ampoules, were irradiated at high gamma radiation dose levels at the

312

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

Instituto Nacional de Investigaciones Nucleares (ININ), Mexico, with total doses of up to 10,000 kGy. In the second set of experiments at the Department of Oncology, University of Calgary, Canada, samples were irradiated at total dose levels close to those experienced by crude oils over geological time in typical sandstone reservoirs, with total doses in the range of 0–120 kGy. Some low radiation dose experiments with ternary, model mixtures of standard compounds were also carried out. The resulting irradiated samples were processed by solid phase extraction separation of hydrocarbon fractions, which were subsequently analyzed by quantitative GC-MS, while selected oil samples were also analyzed by NMR. Low molecular weight gaseous species, produced during irradiation, were analyzed using gas chromatography by opening the irradiated ampoules in a specially designed gas sampling system, followed by the analysis of the produced gases by gas chromatography. The 3D geometry of in-reservoir radiolysis physics was examined using gamma ray attenuation studies. Irradiation experiments using water as a co-reactant, as well as further investigations on the crude oil polar fractions, will be reported in subsequent papers of this series. 2.2. Radiation attenuation in reservoir media To assess the penetration range in reservoir rocks of gamma and X-ray photons derived by Compton scattering and absorption of the primary gamma radiation signal, attenuation experiments using reservoir media samples were carried out. A 10 cm diameter cylinder of oilcontaining bituminous sandstone from the Cretaceous McMurray Formation was encased in a hard fiberglass shell for stability. A 60Co radiation therapy unit source was used to irradiate the core sample stack axially though the center (1.25 MeV, 30 Gy). Radiation sensitive GAFChromicTM EBT2 film had been placed at varying distances (0–18 cm) along the core to assess radiation intensity at different locations through the rock cylinder and thus assess the attenuation coefficient of the oil sand reservoir rock. The same experiment was repeated with an X-ray beam at 165 keV, with four EBT2 films placed within 7.5 cm from the radiation source. A calibration curve was built to convert the green pixel channel counts of the digitized films into accumulated radiation doses. 2.3. High dose irradiation experiments (0–10,000 kGy) A JS-6500 industrial gamma irradiator, category IV, fitted with a water filled storage pool (Nordion, Ottawa, Ontario, Canada) and a multi pencil-rack 60Co source was used for the high dose experiments, which were conducted at the Instituto Nacional de Investigaciones Nucleares (ININ). The irradiator has a 1 MCi capacity, and the dose rate at the time of the irradiation experiments was 13.05 kGy/h. The irradiation times to cover a 50– 10,000 kGy dose range ranged from 3.8 to 766 h. 60Co decays to 60Ni, generating mainly two gamma ray photons at 1.33 and 1.17 MeV. One high-viscosity biodegraded marine oil from the Western Canada Sedimentary Basin (Alberta heavy oil)

and one Norwegian North Sea, non-biodegraded marine black oil (North Sea oil) were selected for the irradiation experiments. Oil samples of approximately 1 g were placed in 5 mL glass ampoules, which were then sealed under argon. In total, 65 samples from the two oils were submitted to high dose irradiation experiments. 2.4. Low dose irradiation experiments (0–120 kGy) A Gammacell 1000 irradiator (Atomic Energy of Canada Ltd., Chalk River, Ontario, Canada), with a 137 Cs source was used for the low dose irradiation experiments which were conducted at the University of Calgary. 137 Cs decays into metastable 137Ba, which then emits a 0.66 MeV gamma ray to reach stability. The sample holder was constructed with styrofoam to accommodate up to 48 glass ampoules (1 mL) or 2 mL vials. To account for radiation dose rate inhomogeneities within the sample holder (e.g., Masterson and Febo, 1992), the experiments were conducted with the turntable rotation off, and the radiation dose rate for each sample spot was calibrated individually. Sample spots were filled with a water-based gel to mimic oil density, and GAFchromictm EBT3 radiochromic dosimetry film (0.5  2 cm) was placed inside the gel-filled spots. Independent 2 minute irradiation tests were performed in triplicate, and the films subsequently digitized using an Epson 10000XL professional scanner. Densitometry using the red channel was used to evaluate the radiation dose absorbed by each sample spot (0.15 kGy/h). Three North Sea oil samples were prepared for low dose (18, 69 and 111 kGy) irradiation experiments by placing approximately 1 g of material in 1 mL glass ampoules, which were then sealed under argon and placed in a custom sample holder. These samples were used to extend the dose range of North Sea oil samples analyzed by NMR (see Section 2.6). Three mixtures of standard compounds were also prepared for low-dose irradiation experiments (Table 1). Each mixture component was dissolved in toluene at a final concentration of 1 mg/mL. Approximately 1 mL of the standard solution was placed in 2 mL glass GC-MS auto sampler vials and sealed with plastic caps. As it was not Table 1 Chemical composition of different standard mixtures used in low dose irradiation experiments (0–100 kGy). All compounds in concentrations of 1 mg/mL dissolved in toluene. Mixture

Compounds

Molecular formula

Abbreviation

Mix #1

Pristane n-dodecylbenzene n-docosane

C19H40 C18H30 C22H46

Pr DB C22

Mix #2

Squalane Cholestane 1-triacontanol

C30H62 C27H48 C30H62O

Sq Ch Tr

Mix #3

Dibenzothiophene Dibenzofuran-4carboxylic acid Benzo[a]carbazole

C12H8S C13H8O3

DBT Acid

C16H11N

BC

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

possible to remove all air from the headspace some minor amounts of oxygen undoubtedly remained present. Mix #1 was irradiated at 50 kGy, whereas mix #2 and #3 were irradiated at 100 kGy. Both original and irradiated mixtures were analyzed by GC-MS in full scan mode, where compounds were identified and quantified based on their mass spectra. 2.5. GC-MS analysis Original and irradiated oils were fractionated before gas chromatography-mass spectrometry (GC-MS) analysis based on a modified procedure published by Bastow et al. (2007). In brief, 50 mg of oil is de-asphalted by solid phase extraction (SPE) with FlorisilÒ magnesium silicate cartridges and then loaded on a silica gel (0.6 g, 70–230 mesh) column that was pre-washed with pentane. The saturated hydrocarbon compound fraction was collected by elution of 2 mL of pentane, whereas the aromatic hydrocarbon compound fraction was collected by eluting 2 mL of dichloromethane, followed by 2 mL of isopropyl alcohol. A set of internal standards was added to the oil before the de-asphalting step to allow quantitative analysis by GC-MS. This internal standard set contained squalane, adamantane-d16, cholestane-d4, phenyldodecane-d30, naphthalene-d8, phenanthrene-d10 and 1,1-binaphthyl for the quantification of 200+ individual compounds in each sample. An Agilent GC-MS Instrument (7890B GC and 5977 MS) with a HP-5MS capillary column (30 m  0.25 mm  0.25 mm) was used for the analysis of both saturated and aromatic compound fractions. The injector was set to 230 °C in splitless mode, and the temperature program for the GC column included 5 min at 40 °C, followed by a 4 °C/min ramp up to 325 °C, then an isothermal hold at 325 °C for 15 min. Helium was used as carrier gas at a flow rate of 1.0 mL/min. The mass spectrometer was set to perform scan (50–500 amu) and selected ion monitoring modes, simultaneously. A complete list of quantified compounds and monitored ions is shown in Table 2.

Table 2 GC-MS monitored compounds in oil samples. Compounds

Monitored ions (m/z)

C0-5 Naphthalenes

128, 184, 184, 178, 231

C0-2 Dibenzothiophenes C0-2 Phenanthrenes C20-21 and C26-28 Triaromatic steroid hydrocarbons C21 and C27-29 Monoaromatic steroid hydrocarbons C9-35 n-Alkanes Pristane and Phytane C0-4 Adamantanes C19-30 Tricyclic Terpanes + C24 Tetracyclic Terpane C27-35 Hopanes + Gammacerane C27-29 Steranes

142, 156, 170, 198 198, 212 192, 206

253 85 183 136, 135, 149, 163, 177 191 191 217

313

2.6. 1H NMR analysis The NMR method used in this study was based on previous work by Frolov and Smirnov (1994). The advantage of NMR spectroscopy over an analytical method such as GC-MS is that functional groups such as carbon-carbon double bond environments can be detected and quantitated, irrespective of their location in a complex mixture of many chemical species. Thus, whole oil samples can be analyzed intact. A Bruker NMR spectrometer operating at 600 MHz was used for 1H NMR analysis of original and irradiated North Sea oil. Experimental conditions include: 4100 scans, relaxation delay = 2 s, acquisition time = 3.89 s, filter width = 125 kHz, sweep width = 8400 Hz, flip angle = 30°, sample rotation = 20 Hz. Whole oils were diluted to approximately 200 mg/mL in CDCl3 and were spiked with tetramethylsilane for chemical shift calibration. The eight North Sea oil irradiated samples analyzed by 1H NMR ranged in dose from 0–7600 kGy. In addition, seven North Sea oil samples doped with 1-pentadecene (5–1000 ppmV) were also prepared and analyzed. Spectra baselines were corrected with points selected in regions free of peaks and also from the regions of interest. Olefinic hydrogen (Hole, d 6.3–4.2 ppm) area is measured in relative terms as Hole/(Haro+sat+ole), where aromatic (Haro) and saturated (Hsat) hydrogen chemical shift ranges are d 8.3–6.6 and d 3.5–0.1 ppm, respectively. Then, the Hole/ (Haro+sat+ole) ratio measured for each sample is converted to olefinic hydrogen abundance based on the calibration curve calculated from experiments with 1-pentadecene. Thus, the olefinic hydrogen abundances reflect the proportion of olefinic hydrogen atoms in samples. Because olefinic hydrogen abundance values are typically very low, they are reported in parts per million (ppm, not to be confused with NMR chemical shift). 2.7. Volatile hydrocarbon analysis Gas chromatography analyses with a flame ionization detector (GC-FID) were carried out using a Scion 450/456 GC instrument equipped with an Agilent CP-Sil 5CB (10 m  0.15 mm) column. A 3 mL aliquot of the gas sample was manually injected into the side inlet (100 °C). Certified gas standards (Praxair Distributors Inc.) were used to calibrate the GC immediately prior to the analysis. Hydrocarbon gases C1–6 were quantified, with a lower detection limit of 1 ppm and analytical precision and accuracy better than ±2.5% of the reported concentrations. A subset of the high dose irradiation experiment samples were submitted for gas analysis. In total, 23 Alberta heavy oil samples irradiated from 50–10,000 kGy, and 11 North Sea oil samples ranging from 0–200 and 4400–10,000 kGy were analyzed for the hydrocarbon composition of the produced gases. Gases from irradiated North Sea oil samples with doses ranging from 200–4400 kGy were not analyzed by GC-FID as the ampoules had already been opened for other analytical procedures. A simple experimental apparatus was used to collect gases trapped in the irradiated ampoules. The upper portion of an ampoule was interference fitted into the end of

314

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

a length of thin-walled rubber tubing (Gooch Tubing Series 200, 1.500 width  300 length; 38  76 mm), whereas the other end of the tubing was sealed with a clamp above the conical tip of the ampoule. A needle and valveequipped syringe were used to evacuate and collapse the tubing and the valve closed for 1–2 min to ensure there was no air leaking into the apparatus. The tube was then filled with 2–5 mL of helium, the ampoule cracked open by flexing the tubing, and the gas mixture was collected in a syringe after 1 minute of equilibration before being analyzed using a GC-FID apparatus.

3.2. Radiolysis of standard compound mixtures

3. RESULTS 3.1. Gamma and X-ray attenuation in reservoir media Based on the calibration curves built by irradiating the GAFChromicTM EBT2 films at known doses and measuring their green pixel density, the radiation doses accumulated by the films inserted in a sandstone reservoir cylinder sample were plotted as a function of position along the core sample. The linear attenuation coefficient for the attenuated radiation can be defined by Equation (1), where m is the linear attenuation coefficient of the material in cm
1, and x the material thickness in cm. Io and I, are the incident initial radiation dose and the dose measured at any location (distance), respectively, within the rock material. l ¼ x
1 lnðIo =IÞ

ð1Þ

60

For the Co sourced irradiation experiments, the obtained data were fit to the form ae
lx (Eq. (1), Fig. 2). The experimental value for the linear attenuation coefficient was found to be m = 0.114 cm
1, which is higher than the linear attenuation coefficient of water (0.06323 cm
1) but lower than that of a more dense material (e.g. ordinary concrete, density = 2.4 g/cm3, m = 0.139 cm
1) (Hubbell and Seltzer, 2004), and compatible with the estimated sandstone bulk density of 2.15 g/cm3 (porosity / = 0.3). The half value layer is the thickness of absorbing material needed to reduce the incident radiation intensity by half

1.25 MeV gamma ray attenuation in sandstone

Dose (Gy)

40 30

y=

39.137e-0.114x R² = 0.99

20 10 0 0

2

4

6

(i.e. I0/I = 2, Eq. (1)). For the bituminous sandstone used in our experiments, the calculated half value layer is 6.08 cm. Similar experiments using 165 keV X-rays on the same sandstone revealed that their transmission is reduced to approximately 10% at 7.5 cm, being completely attenuated within reservoir cores within distances of 10 cm. These results indicate that gamma radiation derived from radiogenic decay of radionuclides in reservoirs will impact crude oil composition within a few tens of centimeters of more highly radioactive reservoir sections, such as nearby radioactive shale horizons.

8 10 12 14 16 18 20 Thickness (cm)

Fig. 2. Radiation dose measured by GAFChromicTM EBT2 films placed at varying distances (0–18 cm) into the sandstone cylinder and exposed to 30 Gy at maximum dose at one end of the core. A 60 Co radiation source (1.25 MeV) was used in these experiments, and data fitted to the form ae
lx, where m is the linear attenuation coefficient of the material in cm
1, and x the depth in cm.

Earlier work indicated that gamma ray irradiated binary mixtures of hydrocarbons resulted in a complex and diverse range of products including bimolecular addition reaction products, even if the starting mixtures were extremely simple (Larter et al., 2012). For example, GC-MS analysis of the saturated hydrocarbon fraction of a gamma ray irradiated mixture (1265 kGy) of just cyclohexane and androstane produced a complex mixture of three groups of compounds representing: (a) a series of cyclohexane derived dimers; (b) unreacted androstane plus related ring opened steranes (secosteranes); and (c) a series of steroidcyclohexane reaction products. A large number of products meant that the concentrations of individual new products were very low (e.g. <5 ppm). Further, the recognition of specific radiolysis products at measurable concentrations, using traditional quantitative GC-MS methods focused on individual high molecular weight compound assessment has proven difficult (c.f. Larter et al., 2012). Low molecular weight species are also produced, but in most cases these species coelute with abundant petroleum components that are ubiquitous in crude oils. To further investigate the feasibility of finding characteristic, non-olefinic, radiolysis products as fluid radiation dose proxies, we studied the irradiation of simple mixtures of hydrocarbons and some nonhydrocarbons. From this, an assessment was made of the

Table 3 GC-MS area ratios of standard compounds before (0 kGy) and after irradiation experiments (50 or 100 kGy). Radiation dose (kGy)

GC-MS area ratios*

Mix #1

DB/Pr

C22/Pr

0 50

1.12 1.14

0.91 0.90

Mix #2

Ch/Sq

Tr/Sq

0 100

0.98 1.02

0.60 0.09

Mix #3

Acid/DBT

BC/DBT

0 100

2.41 0.32

0.17 0.90

* DB = n-dodecylbenzene; Pr = pristane; C22 = n-docosane; Ch = cholestane; Sq = squalane; Tr = 1-triacontanol; Acid = dibenzofuran-4-carboxylic acid; DBT = dibenzothiophene; BC = benzo[a]carbazole.

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

possibility of using characteristic non-olefinic radiolysis products as a viable radiation dose proxy. Using the low dose 137Cs irradiation apparatus, three different standard compound mixtures were exposed to radiation doses of approximately 50 and 100 kGy (see Table 1). Within different positions in the sample holder, the measured dose rates ranged from 1.10 to 5.88 Gy/min. Ratios of indicative standard compound peak areas before and after radiolysis, as measured by GC-MS total ion current, are reported in Table 3. Whereas the mixture containing only hydrocarbons (mix #1: pristane, dodecylbenzene, n-C22 alkane) showed negligible differences between the ratios of standard compound area ratios, before and after radiolysis, mix #2 (squalane, triacontanol, cholestane) and mix #3 (dibenzofuran-4-carboxylic acid, dibenzothiophene, benzo[a]carbazole) indicated that, by far, the more labile and easily destroyed compounds during radiolysis

315

were the oxygenated species triacontanol and the dibenzofuran carboxylic acid. With toluene as the base solvent and with air present in the vials, some toluene oxidation products were seen in all irradiated samples (Fig. 3), reflecting the small amounts of oxygen in the headspace of the GC-MS vials. The key observation however, confirming earlier studies, is the presence of abundant bimolecular addition reaction products such as alkylated biphenyls and diphenyl alkanes (highlighted in Fig. 3) formed by reaction between activated toluene and other species. The chromatogram peak area ratios between pristane (Pr), n-dodecylbenzene (DB) and n-docosane (C22) are similar in both original and irradiated mix #1 samples, reflecting broadly similar rates of radiolysis for all hydrocarbons. Traces of squalane were also identified in the mix #1 chromatograms, indicating some cross-contamination with the

Fig. 3. GC-MS total ion chromatograms (TIC) of mix #1: pristane, dodecylbenzene, n-C22 alkane; mix #2: squalane, triacontanol, cholestane; mix #3: dibenzofuran-4-carboxylic acid, dibenzothiophene, benzo[a]carbazol; before (black) and after (gray) irradiation experiments at 50 kGy (mix #1) and 100 kGy (mix #2 and #3). Highlighted areas show toluene (used as solvent) bi-molecular addition (27–37 min) and oxidation products (<25 min), as well as C8-11 hydrocarbon structures at retention times lower than 23 min. Ratios of standard compounds before and after irradiation can be found in Table 3.

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

mix #2 experiment during sample preparation (Fig. 3). Full scan GC-MS analysis of reaction products was attempted but the unambiguous identification of individual structures has made this challenging due to their low abundance and commensurate low signal/noise ratio in the GC-MS traces. In both mix #1 and #2 experiments, at lower retention times (<20 min) in the GC-MS chromatograms (Fig. 3), hydrocarbon peaks were identified as aliphatic and alicyclic moieties with 8 to 11 carbons, as well as isoprenoidal hydrocarbon fragments. The same species were not detected in mix #3, as all mix #3 standard compounds, as well as the solvent (toluene), are aromatic compounds. In irradiated mix #3, a degradation/oxidation product of benzo[a]carbazole with molecular formula C16H13NO2, was identified, with a peak area of 5.6% of unreacted benzo[a]carbazole (Fig. 3). At higher retention times, a series of small peaks showing m/z 217 fragments in their mass spectra were detected, suggesting the presence of altered benzocarbazoles likely produced via the addition of toluene and/or oxidation. The only radiolysis product detected from DBT irradiation was dibenzothiophene sulfoxide in low quantities (1.4% area of the unreacted DBT). Dibenzofuran was also detected in irradiated mix #3, accounting for 0.5% of the area of unreacted dibenzofuran carboxylic acid. The overall conclusion from these model compound studies is that, not surprisingly, all compounds are reactive under gamma ray radiolysis, with much interaction between starting materials and the production of a diverse and complex range of products with no single abundant component being produced. The presence of additional bimolecular reaction products, albeit in small quantities, appears to be a uniform characteristic of organic compound radiolysis. The most reactive compounds on radiolysis appear to be the same oxygenated compounds that are also less stable and more reactive under GC-MS electron impact conditions. Thus, electron impact mass spectral behavior of individual compounds can be used as a broad general proxy for some aspects of radiolysis sensitivity. With large numbers of reaction products, it is difficult to envisage the identification and quantitation, using compound specific targeted methods such as GC-MS, of specific individual compounds which could be used as good radiation dose proxies. Only the low molecular weight alkanes may have utility in this regard as ubiquitous radiolysis products (see Section 3.3.1), as they are common degradation products of higher alkanes. However, radiolytic light alkane production in natural settings must be quantified against a typically high background concentration of these species in many petroleum system settings. 3.3. Radiolysis proxies Below we report the change in concentration with increased radiation dose of different compound classes analyzed in oils. The significance of the linear regression slope of compound concentration changes at the 95% confidence level (p-value <0.05) were examined to determine whether there are significant relationships between the concentrations of specific molecules and radiation dose, i.e. if experimental errors are small enough to permit distinguishing the

detected concentration changes (with increasing radiation dose) from noise. In the non-biodegraded North Sea oil and the Alberta heavy oil, 8 and 33 monitored compounds (respectively, out of 198), showed no significant concentration-dose relationships whereas all other compounds did. 3.3.1. Normal and isoprenoid alkanes The North Sea oil and Alberta heavy oil have very different n-alkane distributions and concentrations (Fig. 4). Because the Alberta heavy oil had undergone in-reservoir biodegradation processes, it is depleted in n-alkanes, reaching a maximum n-alkane concentration of only 150 ppm at n-C15. Attempts to analyze the gaseous hydrocarbons from the Alberta heavy oil samples failed to detect C1-6 species. Pristane (Pr) and phytane (Ph), which are slightly more resistant to biodegradation than the n-alkanes, are present at higher concentrations in the Alberta heavy oil, at 813 and 674 ppm, respectively. In contrast, the North Sea oil sample contains n-alkanes at much higher concentrations, e.g. n-C17 at 2390 ppm (Fig. 4). The concentration changes of representative n-alkanes with increasing radiation dose are shown in Fig. 5. In the North Sea oil, all compounds analyzed decrease in concentration with increasing radiation dose. For the Alberta heavy oil, both n-C12 and n-C21 show an increase in concentration with increasing radiation dose, while n-C24 and other larger n-alkanes show decreasing concentrations with increasing doses. A negative slope of compound concentration with increasing radiation dose suggests the radiolytic degradation of a given compound dominates. In contrast, an increased compound concentration with increasing dose in irradiated oils indicates the production rate of such species from fragmentation of higher molecular weight species is greater than its simultaneous destruction via radiolysis. Concentration increases of compounds such as low molecular weight n-alkanes in the biodegraded oil (Alberta heavy oil) result from fragmentation of larger species and this is more clearly evident in the Alberta heavy oil sample, given a low background concentration of these components in the initial oil.

Concentration (ppm)

316

2500 2000 1500 1000 500 0

NSO

AHO

Fig. 4. Concentration profiles of n-alkanes (plus pristane and phytane, Pr and Ph) in North Sea oil (NSO) and Alberta heavy oil (AHO) samples as measured by peak areas with GC-MS (m/z 85 and 183 mass chromatograms). Hydrocarbons smaller than n-C10 were neglected due to the evaporative losses that might have occurred during sample preparation. Error bars indicate the 95% confidence interval calculated from 9 replicates.

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326 NSO

AHO 140

2500 n-C21

n-C31

2000 1500

C21

1000

C10

500 C31

Concentration (ppm)

n-C10 Concentration (ppm)

317

120 100

C24

80

C21

60

C12

40 20

0

n-C12

n-C21

n-C24

0 0

2500

5000

7500

0

10000

2500

Radiation dose (kGy)

5000

7500

10000

Radiation dose (kGy)

Fig. 5. Concentration of representative n-alkanes in oils with varying radiation dose. Dotted lines are the linear fit for each compound data series. For the Alberta heavy Oil (AHO), n-C12 and n-C24 were selected based on the very low initial concentration of n-C10-11 and n-C25-31 in the oil (Fig. 4). Note the significant difference in the concentration scales for the two samples. NSO = North sea oil.

AHO 1.26

14

0.64

1.24

12

1.16

2

0.56

1.14

0

2000

1.9

30000

1.8 1.7 1.6 0

1.5 2500 5000 7500 10000 Radiation dose (kGy)

∑n-Alkanes (ppm)

2 35000

20000

Pr/n-C19

4

0.58

2.1

25000

6

1.18

2.2

40000

8

1.2

1500 1000 500 0 0

4 3.5 3 2.5 2 1.5 1 0.5 0 2500 5000 7500 10000 Radiation dose (kGy)

n-C13/n-C26

0.6

10

1.22 Pr/Ph

0.62

45000 ∑n-Alkanes (ppm)

0.66

Pr/n-C19

0.96 0.95 0.94 0.93 0.92 0.91 0.9 0.89 0.88

n-C13/n-C26

Pr/Ph

NSO

Fig. 6. Variation of molecular composition from GC-MS in irradiated North Sea oil (NSO) and Alberta heavy oil (AHO) samples: pristane/phytane and pristane/n-C19 ratios; summed concentrations of C10-35 n-alkanes and n-C13/n-C26 ratios for both oils. Dotted lines represent the linear fit for each compound data series.

Fig. 6 shows how a few molecular parameters vary with radiation dose, which may help to understand radiolysis impacts on normal and isoprenoid alkanes in crude oils. The summed C10-35 n-alkane concentrations in the North Sea oil decreases with increasing radiation dose. Because the opposite trend occurs with the biodegraded Alberta heavy oil, which had low initial n-alkane concentrations (Fig. 4), the results indicate the relevance of the chemical matrix to the radiolysis product composition and the importance of the absolute background concentration of precursor and product species. In the case of normal alkanes, both the major products and the main precursors are normal alkanes, with n-alkyl substituents on other species likely also contributing to low molecular weight alkane production during radiolysis. Fig. 6 also shows that in the North Sea oil, the pristane/phytane ratio increases with

radiation dose, and because the concentrations of both species decrease with radiation dose (e.g. phytane abundance varies from 1338 to 1120 ppm, at 0 and 10,000 kGy, respectively), we have an indication that phytane is more susceptible to radiolysis than pristane. The Pr/Ph ratios in the original samples are close to one, thus changes in Pr/Ph ratios are likely caused by the preferential degradation of pristane. Moreover, pristane is shown to be more susceptible to radiolysis than n-C19, suggesting that the molecular structure also plays a key role in radiolysis mechanisms and impacts. An underlying general mechanism of light n-alkane production during radiolysis can be inferred from Fig. 7, where the compound concentration/radiation dose slopes (ppm/kGy) for individual normal alkanes, obtained by fitting a linear equation to the data, are plotted against

318

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

NSO n-alkanes

AHO n-alkanes

(A) 2500

n-C17

1500

n-C20 n-C23 n-C24

n-C16

n-C21 n-C25

n-C22

1000

n-C15 n-C19

n-C18

2000

n-C26

n-C14 n-C10

n-C9 n-C28

-0.03

n-C11

n-C29 n-C30

-0.04

n-C13 n-C12

n-C27

500 0 -0.05

Initial concentration (ppm)

Initial concentration (ppm)

3000

n-C31 n-C34 n-C33 n-C32

-0.02

-0.01

n-C35

0

180 n-C15 160 (B) n-C17 140 n-C16 n-C24 120 n-C18 n-C19 100 n-C25 n-C14 80 n-C21 n-C20 n-C26 n-C22 60 n-C13 n-C26 40 n-C12 n-C27 n-C9 n-C28 n-C11 20 n-C10 n-C34 n-C29 n-C30 n-C35 n-C31 n-C33 0 n-C32 -0.002 -0.001 0 0.001 0.002 0.003 0.004

Concentration slope (ppm/kGy)

Concentration slope (ppm/kGy)

Fig. 7. Cross plot of initial compound concentration versus the degradation slope (ppm/kGy) of n-alkanes undergoing gamma ray radiolysis, as measured by a linear fit of the concentration versus radiation dose profile for irradiation of (A) North Sea oil (NSO) and (B) Alberta heavy oil (AHO), irradiated at up to 10,000 kGy. In both plots, triangles represent compounds for which concentration slopes are not significantly different than zero (95% confidence level).

0.7 0.6 Gas dryness (C1/C1-5)

the initial concentration of the species in each oil. Compounds that show concentration/radiation dose slopes that are not significantly different from zero (95% confidence level) are shown as triangles. In North Sea oil samples, the n-C26-35 alkane concentration/dose slopes are negative, indicating net compound destruction, but increase linearly with the initial compound concentration. This suggests that compound concentration plays a key role in the ‘‘degradation kinetics,” most likely simply reflecting ‘‘target availability.” More abundant components are more rapidly degraded. It is noteworthy that the n-C10-17 alkanes have less negative concentration/dose slopes, with an almost linear relationship between the slopes and original compound concentration. The lightest normal alkanes have the least negative slope/concentration values, suggesting that these compounds show net production as well as being simultaneously destroyed during radiolysis. However, the measured slope/concentration values of most of these compounds cannot be distinguished from zero. In the Alberta heavy oil samples, where the initial normal alkane concentrations are very low, the production of the light n-alkanes n-C10-22 is evident by the positive concentration/dose slope relationships (indicative of light normal alkane production). This strongly suggests that fragmentation of normal alkanes and n-alkyl structures in other species during gamma ray radiolysis yields shorter chain normal alkanes and this is a primary and dominant pathway. Other processes such as bi-molecular recombination processes are a secondary reaction pathway. It was observed that irradiated oil containing ampoules (initially at one bar) commonly had positive pressure when opened. Analysis of the gases indicated that hydrocarbon gases (C1-5) were also produced during radiolysis, even at low net doses (<100 kGy). The distribution of the nalkanes in Fig. 7 suggests that large n-alkanes are being degraded and that some of the degradation products are low molecular weight n-alkanes. It appears likely from Fig. 7 that intermediate carbon number normal alkanes are both being produced and degraded in significant amounts. In the few initial experiments analyzed by GC-FID, the volatile hydrocarbons present in the

0.5 0.4 0.3 0.2 NSO 0.1 AHO 0 0

2500

5000

7500

10000

Radiation dose (kGy)

Fig. 8. Low molecular weight hydrocarbon analysis reported as dryness of gases trapped in irradiated North Sea oil (NSO) and Alberta heavy oil (AHO) ampoules.

irradiated samples were not quantitated in absolute terms, because there was no conservative internal standard and no control on the total pressure and thus quantity of gas in the ampoule headspace. Instead, we determined C1 to C5 instrumental responses and investigated the evolution of hydrocarbon gas dryness as a function of radiation dose, measured as the ratio of methane (C1) over the sum of all hydrocarbons with 5 carbons or less (C1-5). Fig. 8 shows the evolution of gas dryness with increasing gamma ray radiation dose for both North Sea oil and Alberta heavy oil at gamma ray doses ranging from 0–10,000 kGy. Gases from North Sea oil samples irradiated from 200–4400 kGy were not analyzed for gas dryness by GC-FID because the irradiated ampoules had been previously opened for other analyses. Fig. 8 reveals low gas dryness values for the lowest radiation doses (<50 kGy), an increase in gas dryness at low incident doses between 50 kGy and 400 kGy, followed by a plateau of maximum dryness at more elevated radiation doses. At 400 kGy, Alberta heavy oil sample gases had already achieved dryness values that remained stable at 0.6 over the radiation dose range up to 10,000 kGy. Similar analysis could not be made with the North Sea oil due to the lack of data at

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

were quantified using GC-MS for both irradiated Alberta heavy and North Sea oils. Fig. 9 shows the change in concentration of C30 hopane as a function of radiation dose for both Alberta heavy oil and North Sea oil samples. Coincidently, the original concentrations of C30 hopane in both oils are very similar at 400 ppm. However, as radiation damage progresses, it seems that the degradation rate of hopane in the North Sea oil is higher than in the Alberta heavy oil, given its higher degradation slope (Fig. 9), suggesting that the oil matrix (e.g. density, chemical composition) also plays a key role in radiolysis-induced reactions of individual compounds. The linear fitted concentration slope (see hopane example in Fig. 9) for each monitored compound type (except n-alkanes, pristane and phytane) is plotted against the original compound concentrations in the oils in Fig. 10 (North Sea oil) and in Fig. 11 (Alberta heavy oil). The data show that the apparently linear relationship between the concentration/dose slope and the initial compound concentration is compound class dependent. For compounds at similar initial concentration, the susceptibility to removal through radiolysis decreases in the sequence hopanes, steranes, triaromatic and monoaromatic steranes > tricyclic terpanes > naphthalenes, phenanthrenes > dibenzothiophenes > diamondoid alkanes. The greater stability of aromatic species (compared to aliphatic compounds) to radiation has been reported before (Williams, 1963). Interestingly, the diamondoids (adamantanes) detected in the North Sea oil sample appear to be the compounds most resistant to radiolysis. Further

low dose, but it seems that the irradiated North Sea oil samples also achieved stable gas dryness values between 0.5 and 0.6 at elevated total doses, slightly lower than those of the Alberta heavy oil samples. Methane and other light normal alkanes are being produced during crude oil radiolysis and that with increasing radiation dose, methane becomes the dominant component. 3.3.2. Other molecular markers Hydrocarbons and aromatic sulfur compounds (198 in total), typically used as proxies in petroleum geochemistry,

450 C30 hopane (ppm)

NSO 400

AHO

y = -0.013x + 397.6 R² = 0.95

350 300 250

y = -0.018x + 398.6 R² = 0.96

200 0

2500

5000

7500

319

10000

Dose (kGy)

Fig. 9. Concentrations of C30 hopane in North Sea oil (NSO, blue) and Alberta heavy oil (AHO, red) with varying radiation dose. Dotted lines represent the linear fit for each compound data series. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3000 2500 2000

Tricyclic terpanes Hopanes Steranes Diamondoids

-0.015 -0.01 -0.005 Concentration slope (ppm/kGy)

0

Phenanthrenes Naphthalenes Dibenzothiophenes MAS TAS

1500 1000 500

(B) 0 -0.08

-0.06 -0.04 -0.02 Concentration slope (ppm/kGy)

0

Inirtial concentration (ppm)

450 400 350 300 250 200 150 100 50 (A) 0 -0.02

160 140 120 100 80 60 40 20 (C) 0 -0.005 -0.004

Initial concentration (ppm)

Initial concentration (ppm)

Initial concentration (ppm)

North Sea oil

350 300 250 200 150 100 50 (D) 0 -0.01

Tricyclic terpanes Hopanes Steranes Diamondoids

-0.003 -0.002 -0.001 Concentration slope (ppm/kGy)

0

0.001

Phenanthrenes Naphthalenes Dibenzothiophenes MAS TAS

-0.008 -0.006 -0.004 -0.002 Concentration slope (ppm/kGy)

0

Fig. 10. Initial compound concentration versus the concentration slope (ppm/kGy), of monitored (A) saturated hydrocarbons, and (B) aromatic compounds, in the North Sea Oil, during the same set of gamma ray radiation experiments. The concentration slope was measured by a linear fit of the compound concentration/radiation dose slope (see example in Fig. 9) for each compound analyzed. n-Alkane data are shown separately in Fig. 7. Boxed areas in plots A and B are expanded in plots C and D, respectively. TAS = triaromatic steroid hydrocarbons, MAS = monoaromatic steroid hydrocarbons. Table 2 describes the actual species analyzed in each labelled compound group. In all plots, triangles represent compounds for which concentration slopes are not significantly different from zero (95% confidence level).

320

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

Initial concentration (ppm)

Initial concentration (ppm)

Alberta heavy oil 500 400

Tricyclic terpanes Hopanes Steranes

300

Diamondoids

200 100 (A) 0 -0.015

500 400

-0.01 -0.005 Concentration slope (ppm/kGy)

0

Phenanthrenes Naphthalenes Dibenzothiophenes

300

MAS TAS

200 100 (B) 0 -0.006

-0.004 -0.002 Concentration slope (ppm/kGy)

0

Fig. 11. Initial compound concentration versus the concentration slope (ppm/kGy) of (A) saturated hydrocarbons, and (B) aromatic compounds in the Alberta heavy oil sample set, during gamma ray irradiation. The concentration slope was measured by a linear fit of the concentration/dose slope (see example in Fig. 9). n-Alkane data are shown separately in Fig. 7. TAS = triaromatic steroid hydrocarbons, MAS = monoaromatic steroid hydrocarbons. Table 2 describes the actual species analyzed in each labelled compound group. Triangles represent compounds for which concentration slopes are not significantly different from zero (95% confidence level).

investigation is needed to verify whether the potential formation of diamondoid cages from the radiolysis of other oil constituents plays a role in our experiments. In Alberta heavy oil, all the analyzed diamondoids showed concentration/dose slopes not significantly different from zero (triangles in Fig. 11A). However, the significant experimental errors associated with their analysis in the Alberta heavy oil samples (likely due to low initial abundance and/or volatility) do not support any firm conclusions regarding their radiolysis resistance. For the compounds investigated here, a correlation between the initial concentration of a given compound and its concentration variation with radiolysis was again observed (trend lines in Figs. 10 and 11). By eliminating the effect of the initial compound concentration, the impact of other molecular properties on resistance to radiolysis may be investigated. Thus, the apparent gamma ray radiolysis susceptibility (kGy
1) of the different compound classes was defined. This parameter represents the component concentration/dose slope relationships, where the compound concentrations are normalized to the initial component concentration in the non-irradiated oil, for each monitored chemical class. The results are summarized in the boxplots shown in Fig. 12.

Positive values of radiolysis susceptibility indicate increases of compound concentration with increasing radiation dose, and only a few compounds showed apparent increases in concentration with radiolysis. In the Alberta heavy oil (Fig. 11), diamondoids, dibenzothiophene, and 1,7- and 2,3-dimethyldibenzothiophene increased in concentration with increasing radiation dose, although none of them showed a concentration/dose slope relationship significantly different from zero (95% confidence level). Therefore, the only compounds that show a statistically significant increase in concentration in the Alberta heavy oil samples are the n-C9–20 alkanes (except n-C18) (Fig. 7). In the North Sea oil samples (Fig. 10), the sum of the 2,8-, 2,7-, and 3,7-dimethyldibenzothiophenes GC-MS peaks, showed a positive but not statistically significant concentration/dose slope, whereas the increase in C30b tricyclic terpane concentration with radiation dose was statistically significant. We note that we do not have full mass spectra for this compound, so the identity is only tentative. A large range of radiolysis susceptibility was detected within the n-alkanes (Fig. 12). Not unsurprisingly, the nalkane class is the one with the largest range of molecular weight analyzed, with the species analyzed from C10 to C35. Fig. 13A shows how the carbon number and radiolysis susceptibility are correlated for all the monitored compounds in both oils, whereas Fig. 13B highlights the same information obtained for the n-alkanes only. The monitored compounds in the Alberta heavy oil samples show, on average, a higher resistance to radiolysis than in the North Sea oil samples, suggesting that the radiolysis

Fig. 12. Distribution of radiolysis susceptibility of monitored compounds, as grouped by compound class, for Alberta heavy oil (red) and North Sea oil (blue) irradiated oil sample sets. The gamma ray radiolysis susceptibility (kGy
1) of the different compounds is derived by normalizing the gradient (slope), for each compound, in the linear trend of compound concentration versus radiation dose, to the compound original concentrations in the parent oil for each monitored chemical compound or class. TAS = triaromatic steroid hydrocarbons, MAS = monoaromatic steroid hydrocarbons, TT = tricyclic terpanes. Table 2 describes the actual species analyzed in each labelled compound group. On the box plot, maximum and minimum values for each compound class are represented by whiskers, while the top and bottom of the boxes represent first and third quartiles of the data. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

Radiolysis susceptibility (kGy-1)

All monitored compounds 1.5E-4

NSO

1.0E-4

AHO

5.0E-5 0.0E+0 -5.0E-5 -1.0E-4

(A) 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Carbon number

Radiolysis susceptibility (kGy-1)

n-Alkanes 1.5E-4

NSO

1.0E-4

AHO

5.0E-5 0.0E+0 -5.0E-5 -1.0E-4

(B) 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Carbon number

Fig. 13. Radiolysis susceptibility as a function of the compound carbon number in Alberta heavy oil (AHO) and North Sea oil (NSO) for (A) all monitored compounds and (B) n-alkanes. In the case of the Alberta heavy oil, the low initial concentrations of normal alkanes make the effects of radiolytic light alkane generation more evident at low carbon numbers than in the case of the NSO oil with its higher initial normal alkane concentrations. Triangles represent compounds for which concentration slopes are not significantly different from zero (95% confidence level).

mechanisms are matrix dependent. Therefore, oil-specific proxy calibrations for the impact of radiation on a crude oil are likely necessary. Moreover, Fig. 13 indicates that the radiolysis susceptibility is also dependent on the molec-

North Sea oil @ 3600 kGy

321

ular size of the component (as represented broadly by the carbon number). This effect can also be observed within compound classes, although it is more pronounced in the aliphatic hydrocarbon classes, such as the tricyclic terpanes (C19-29), than the alkylated naphthalenes (C10-15) and other aromatic compound classes. Larger molecules are more readily degraded during irradiation than smaller analogues, as they represent larger targets with more atoms and hence more bonds available to be cleaved. 3.3.3. Formation of unsaturated species during gamma ray radiolysis The next group of results describes the production of new compounds or species during radiolysis, such as those containing a carbon-carbon double bond, and how the concentration of produced species relates to the incurred radiation dose. Initial investigations by Frolov et al. (1998 and references therein) showed that a complex mixture of principally non-terminal alkenes were generated during radiolysis, chiefly from the corresponding alkanes, and that the concentration of alkenes produced was dependent on the radiation dose received by the crude oil. Frolov et al. (1998) also showed, however, that unsaturated sites were also produced during radiolysis, in essentially all alkyl and cycloalkyl environments, in any hydrocarbon or nonhydrocarbon species in the crude oils with appropriate alkyl carbon substitution. This, and the further observation that biodegraded oils with low initial normal alkane contents produced lower concentrations of alkene hydrocarbon products (Frolov et al., 1998) suggested that alternate approaches to monitoring unsaturated species production in crude oils were needed. As unsaturated species are distributed throughout the traditionally chromatographically separated fractions of crude oil, a more holistic spectroscopic approach to assessing the total unsaturation inventory in irradiated crude oils was investigated. Preliminary work on the application of Fourier transform infrared (FTIR) methods to the quantitation of olefinic structures suggested that FTIR did not have enough specificity, nor the sensitivity to accurately quantify carbon-carbon double bond structures at the concentrations present in naturally

background

Region A Aromatics (Haro) Region B Olefins (Hole)

Region C Saturated (Hsat)

Fig. 14. 1H NMR spectrum of North Sea oil after a 3600 kGy radiation dose. Highlighted areas correspond to aromatic, olefinic (region B, expanded) and saturated hydrogen resonance regions. Each region was integrated after baseline correction and the relative olefinic hydrogen content was determined as the area ratio Hole/(Haro+sat+ole), where Hole represents the area of region B, and the denominator corresponds to the sum of areas A, B, and C. The area ratio was used to estimate the equivalent olefinic hydrogen content (1198 ppm, in this case) from a calibration curve constructed with 1-pentadecene doped oil.

322

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

Olefinic hydrogen (ppm)

North Sea oil (0-7600 kGy)

2000

1500 200 150

1000

100

Slope 0.4 ppm/kGy

50

500

0 -100

100

300

500

0 0

2500

5000

7500

Radiation dose (kGy)

Fig. 15. Olefinic hydrogen abundance as a function of the total gamma ray radiation dose absorbed by the North Sea oil (NSO), from 0–7600 kGy. Inset plot highlights results from separate low dose (0–400 kGy) experiments and shows a rate of olefin formation via radiation damage of around 0.4 ppm/kGy within this interval, which is comparable to dosage experienced by oils in natural environments in crude oil reservoirs.

irradiated crude oils. Therefore, we further explored high resolution 1H NMR as an alternative analytical approach. 1 H NMR responses related to olefinic protons are detected within the d6.3–4.2 ppm chemical shift range, and the relative area of olefinic proton resonances to total proton resonances, Hole/(Haro+sat+ole), can be easily determined with proper baseline correction (Edwards, 2011). Experiments with a crude oil sample doped with 1-pentadecene showed a linear relationship between the relative area of the olefin resonance measured in 1H NMR experiments and the equivalent concentration of olefinic hydrogens (in ppm). Such a relationship was then used to convert measured Hole/(Haro+sat+ole) in oil samples into a concentration unit that better represents the amount of carbon–carbon double bonds present in a given sample. For example, Fig. 14 shows a 600 MHz 1H NMR spectrum for an North Sea oil sample irradiated at 3600 kGy in which olefinic hydrogen content was estimated at 1198 ppm. NMR analyses of irradiated North Sea oil samples at high dose (0–7600 kGy) showed an apparent non-linear increase of the olefinic hydrogen content with applied radiation dose (Fig. 15). The inset plot, which highlights the low radiation dose experiments done on the North Sea oil, shows that there is also a linear relationship between olefinic hydrogen relative abundance and total received radiation dose (a slope of approximately 0.4 ppm/kGy) at dose levels comparable to those in natural petroleum reservoir environments. This is a crucial observation and suggests that this analytical approach is a potential route to a practical radiation dose assessment proxy for reservoired crude oils. 4. DISCUSSION 4.1. Age dating proxy modeling considerations A plausible approach that could be applied to dating reservoir fluid residence times in petroleum reservoirs is

that natural nuclear radiation derived from a reservoir medium can change the composition of a crude oil in a systematic, radiation-dose dependent way. This would allow an observer to determine the radiation dose received by the crude oil while hosted in that reservoir and knowing the accumulated radiation dose, a knowledge of the nuclear radiation dose rates provided by that reservoir would permit, in principle, the estimation of a fluid residence age. The natural radiation dose applied to petroleum fluids in subsurface pore spaces for crude oil traps will be typically <200 kGy. Experiments aimed at calibrating the total radiation dose absorbed by the entrapped fluids and any matrix-dependent compositional effects are needed. The results presented in this paper suggest that most if not all these factors are experimentally feasible. For proxy-based organic geochemical systems designed to assess petroleum residence time in a given reservoir, proxies based on the concentrations, isotopic compositions, and/or distributions of molecular components must be suitable to determine the in-reservoir radiation dose accumulated by a petroleum sample. This in turn, can be correlated with the residence time of a petroleum sample in the given reservoir. However, simply calibrating chemical compositional change in a crude oil in a reservoir to a radiation dose impact is insufficient for a chronometric system, as the radiation dose experienced by pre- or post-generation organic matter in a source rock might exceed that experienced by oil in a reservoir. During petroleum generation, abundant hydrogen transfer occurs, and it is plausible that many olefinic moieties in any petroleum in a source rock are saturated during petroleum generation, but this remains to be quantitated. The different impacts of radiation in source rocks and in reservoirs are being investigated. To be used as a practical tool, any viable approach must be able to uniquely identify the impacts of crude oil radiolysis in a terminal reservoir, irrespective of any prior alteration of kerogen or generated petroleum in the source rock, the compositional legacy of which is transmitted to the reservoir in crude oil during migration. One possible way of isolating the in-reservoir crude oil radiolysis signal from any prior radiation signal in a migrating crude oil is to use our understanding of the systematics of radiation transport within reservoir rocks. In assessing any proxy system, some basic understanding of the systematics of nuclear radiation generation and propagation within reservoir rocks is needed. The key observations for development of a viable radiation dose proxy are that whereas alpha and beta particle radiation has a relatively short transit distance - up to 20 mm - from the source radionuclide in rocks (Farley, 2002), gamma ray and Xray attenuation studies in reservoir sandstones (Fig. 2) indicate that gamma radiation effects can be expected in crude oils several tens of centimeters away from the radiation source. This is encouraging, as it suggests that radioactive shales within or capping a reservoir may provide a variable internal and local radiation dose signal, enabling local in-reservoir irradiation effects to be discriminated by monitoring the local variation of radiolysis proxies at various distances away from in-reservoir shales. A locally varying radionuclide distribution correlating with a corresponding

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

signal in an oil column profile within a reservoir will be a definitive source signal for any reservoir fluid-based radiation impact. Thus, the detection of locally generated radiolysis effect profiles is a potential route to decoupling source versus reservoir radiation impacts, even if source irradiation effects are transmitted to the reservoir. Thus, inreservoir mass transport effects and fluid mixing models (c.f. Wilhelms and Larter, 2004; Larter et al., 2003) are key elements to include in any such data analysis. Physical models of in-reservoir radiolysis of crude oil must, therefore, couple the linked effects of local generation of radiolysis proxies and radiolytic compound alteration effects near the most radioactive reservoir sections. With oil charge being added to a reservoir over time, the impacts of diffusion of radiolysis products and reactants within fluid columns will result in diffusive mixing of proxy profiles in oil columns. 4.2. Crude oil radiolysis products and potential age dating proxies A multitude of radiolysis products are produced from irradiation of even simple mixtures of hydrocarbons and related compounds, which indicates that radiolysis products of a naturally complex organic mixture such as crude oil would be of great complexity. As previously discussed, the radiolysis of oil components is ultimately caused by an avalanche of 20–80 eV electrons generated when gamma rays enter the material phase. While phase densities and conditions are quite different between bulk phase crude oil radiolysis and a mass spectrometer ion source, the electron impact nature of radiolysis processes suggests that product distribution from radiolysis of a given compound can be informed by examination of its mass spectrum, commonly obtained after electron impact ionization at 70 eV. For instance, radiolysis of triacontanol and dibenzofuran carboxylic acid, both easily ionized in a mass spectrometer ion source and containing labile nucleophilic functional groups, is shown to be quite rapid (Table 3). Metastable molecular fragments produced after irradiation can be stabilized by reacting with other chemical species in the chemical matrix. Such findings suggest that the polar constituents of petroleum fluids should be systematically evaluated as potential reservoir residence age dating proxies, as alterations in oil composition after radiolysis at low radiation doses may be more easily detected in the more sensitive polar compound fractions. The balance between the different radiolysis mechanistic pathways is determined by the chemical matrix and other compositional features (e.g. chemical classes) of the targeted molecules. The current study has not investigated the co-radiolysis of organic compounds and water, but in the presence of water oxidation reactions may also take place (Ponomarev et al., 2002; Head et al., 2014). Studies on the impact of water on petroleum radiolysis will be investigated in future work. GC-MS analyses of irradiated oils indicate that aromatic hydrocarbons are more resistant to alteration during gamma ray radiolysis compared to saturated hydrocarbons and this is consistent with literature findings (Williams, 1963). Our results also indicated, that the diamondoid

323

hydrocarbon class is, not unexpectedly, one with enhanced resistance to radiolysis degradation. Aromatic structures more easily accommodate excited electronic states, thus less fragmentation is observed when compared to saturated hydrocarbons. Diamondoid hydrocarbon structures are known to be more resistant to thermal stress (Dahl et al., 1999) and biodegradation (Grice et al., 2000) than other hydrocarbons and now it is apparent that they may also exhibit an enhanced resistance to radiolysis as well. Given the complexity of radiolysis products, monitoring destruction of readily identifiable and quantifiable species in crude oil that relate in some way to the incurred radiation dose could yield a pragmatic and suitable radiolysis proxy. Most of the studied compounds with carbon number greater than 10 show radiation dose-dependent reductions in concentration with increasing gamma irradiation dose (Fig. 13). In general, large alkanes such as high molecular weight normal alkanes, hopanes or steranes show the largest changes in concentration with increasing radiation dose. For experiments with radiation doses of up to 10,000 kGy, the change in concentration of the components such as hopane is a few parts per million, compared to an initial concentration in the oil of several hundred parts per million. This is a change in concentration of about 1%. Standard organic geochemical protocols based on GC-MS procedures are not capable of reliably detecting such small changes without substantial development, though compound quantitation with the accuracy and precision of routine stable isotopic measurements would be usable. While such analytical requirements might seem daunting in organic geochemistry, in other fields analytical contrasts of up to 10 orders of magnitude are being developed, such as the detection of light from exoplanets against a bright stellar light source (Oakley and Cash, 2009). Application of more sensitive quantitative tools in organic geochemistry is required. Production of new compound classes or species during radiolysis, such as olefinic species, bi-molecular addition products and characteristic low molecular weight fragments of larger structures produced during radiolysis are an alternate proxy to one based on compound destruction. The radiation dose-dependent production of alkenes described by Frolov et al. (1998) is an excellent example of one such proxy, recognizing that olefinic species can appear in any alkyl elements in a wide range of precursor components. As indicated in Figs. 15, 1H NMR is capable of detecting carbon-carbon double bond production from both the high irradiation dose experiments and at gamma radiation doses comparable with those likely experienced by petroleum in natural petroleum reservoirs. An analytical technique capable of measuring unsaturation across a variety of petroleum fractions with ultra-low limits of detection would offer an extremely valuable proxy. 1H NMR is known to lack sensitivity, especially compared to other spectroscopic techniques, although it would appear that with modern instruments it has enough sensitivity to monitor total unsaturated species in a crude oil, without prior separation. Noteworthy, as described by Curiale and Frolov (1998), olefin occurrence in crude oils can also have sources other than in situ radiolysis, including olefin generation from

324

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

organic matter pyrolysis due to thermal impacts from igneous intrusions occurring close to the reservoired oil, olefins expelled and migrated from source rocks, and syndepositional olefins captured from immature organic matter during migration. Contamination of reservoir material by olefin containing drilling muds is a problem. Thus, the effective use of olefins as radiation dose proxies involves decoupling the abundance of radiolytic olefins from those derived from other sources. Major challenges to the application of reservoir fluid residence age dating proxies, beyond issues related to fluid mixing and mass transport, are the clear matrix dependent effects on species production and/or destruction during crude oil radiolysis. The dose-dependent reduction in component concentration varies with crude oil matrix, compound class, initial component concentration in the oil, and the compound carbon number and chemical type (size, aromaticity and functional group content) (Figs. 9–13). This would suggest that for any proxy chosen for development, a calibration exercise for individual oils, and perhaps even reservoir media, may be necessary. Fortunately, the analytical methods used in this study seem to be sensitive enough to detect proxies (e.g. alkene production) at natural reservoir fluid history dose levels. Thus, incremental irradiation of a natural sample with a known additive radiation dose permits the determination of dose-dependent proxy behaviour in a given oil system. Such a procedure is an effective means of dealing with oil type- and matrixdependent effects in the crude oil radiolysis system. This would also permit, in principle, the naturally determined proxy to be calibrated for assessments of actual radiation dose. Fortunately, realistic natural radiation doses for fluids in petroleum reservoirs can be attained in experimental timescales. Gaseous hydrocarbon products were also detected after crude oil radiolysis at low doses, suggesting they can potentially be used as radiolysis proxies. The increase in gas dryness at low but increasing radiation doses suggests that the radiolysis product signature slowly dominates the gas composition of the irradiated sample until a plateau is reached. The Alberta heavy oil produced gas dryness values started at 0.40 and stabilized at a dryness value of 0.60 after exposure to approximately 400 kGy, whereas for the North Sea oil samples, gas dryness values range from 0.10–0.55 and plateaued between 0.50 and 0.55. Differences in the dryness values at high radiation doses might be attributed to the cumulative net fragmentation pattern of alkylated structures exhibited by the different gross structures of these two oils, the North Sea oil having a much higher initial nalkane and other linear alkane content, thus producing more radiolysis derived wet gas. Radiolysis products are matrix-dependent and further investigation is ongoing to determine how such differences impact gaseous or any other proposed reservoir fluid residence age dating proxies. Experiments in this study were carried out with ‘dead’ oils (solution gas phase lost during equilibration at atmospheric pressure). However, in a natural system, radiolytically generated gases would be mixed and diluted by the thermogenic gas signatures of most crude oils.

5. CONCLUSIONS A review of prior strategies to estimate the residence time of petroleum fluids in subsurface reservoirs revealed that few, if any methods currently used, provide useful estimates of the reservoir residence age of fluids. An approach based on the quantitative analysis of chemical changes in crude oil composition as a function of accumulated radiation dose, calibrated against assessments of radiation dose rate from oil hosting sedimentary rocks, indicates that various organic geochemical proxies may potentially prove useful. Proxy systems based on either radiation dosedependent destruction of crude oil components, or radiation dose-dependent production of neoformed compounds have been proposed. Our conclusions are:  Reservoir fluids are naturally immersed in radioactive subsurface media that causes systematic alteration to crude oil composition with time, but the chemical changes are small in most natural settings. New methods are needed to quantify these changes.  The degree of radiolysis of individual petroleum compounds (apparent gamma ray radiolysis susceptibility – kGy
1) was found to depend on chemical class, molecular size and the nature of the oil matrix, indicating that a proxy system will likely depend on case-specific calibrations.  Large alkanes, such as high molecular weight normal alkanes or hopanes, have high negative values of radiolysis susceptibility and show the most rapid decrease in component concentration with increasing radiation dose. In contrast, condensed aromatic hydrocarbons such as alkylated naphthalenes and phenanthrenes, and also diamondoid hydrocarbons (alkylated adamantanes) are the most resistant to destruction through radiolysis and show less negative values of radiolysis susceptibility.  Light hydrocarbon species that have been formed through crude oil radiolysis may also have the potential to be reservoir residence age dating proxies. Importantly, such effects can be detected at radiation doses similar to those from natural geological settings.  Monitoring the production of specific functional groups using NMR spectroscopy during radiolysis is a feasible analytical target and analysis of newly generated species containing carbon-carbon double bonds in bulk crude oils may represent an optimal approach for development as a radiolysis proxy.  Daunting constraints on the proposed approach include resistance of proxy system components to secondary inreservoir alteration processes, and requirements of no exchange of precursors or products with the reservoir medium itself. The former requirement can be sensibly examined using standard organic geochemical assessments and protocols once the proxy candidates are finally identified.  To separate source rock and reservoir radiation impacts on oils, one approach made viable by the observation of the short range of gamma radiation

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

in rocks, is to use in-reservoir radioactivity heterogeneities and proxy modelling tools to examine proxy variations near more highly radioactive reservoir sections, which would be uniquely derived from in-reservoir radiolysis. Local in-reservoir signals of radiolysis might thus be identified using an oil charge, advection-diffusion model linked to measured variations in local radiation dose assessed for example, using downhole gamma ray logging tools or other radioactive element assessment methods. Despite the complexity of crude oil radiolysis, monitoring destruction of readily identifiable and quantifiable species in crude oil, or assessing the production of new chemical moieties (e.g. carbon-carbon double bond) both of which relate systematically to the incurred radiation dose, are potential routes to a reservoir fluid residence age assessment system.

Declaration of Competing Interest None ACKNOWLEDGEMENTS The PRG research team provided valuable experimental support for this research program. The Department of Oncology at the University of Calgary, and Gabriel Ortiz from the Instituto Nacional de Investigaciones Nucleares – Mexico, are thanked for their experimental support. Petrobras and Lundin Norway are acknowledged for sponsoring the project Rip van Winkle Phase I, and Lundin Norway for sponsoring the Rip van Winkle Phase II project (http://ucalgary.ca/prg/research/rip-van-winkle). Michael Briscoe is acknowledged for conducting the sandstone irradiation experiments. Veronique Fau from the Applied Geochemistry Group – University of Calgary is thanked for assisting with gas analyses. Dan Stoddart is thanked for many pertinent technical discussions and support. Contributions: Renzo Silva and Norka Marcano contributed equally to experimental program design, execution and performed most of the data analysis. Lloyd Snowdon, Jing Zhao, Roshanak Sonei, and Andrew Stopford contributed to design, construction and execution of the crude oil irradiation experiments; Eduardo Villarreal-Barajas and Lydia Paredes Gutie´rrez designed the nuclear physics components of the experiments. Haiping Huang, Thomas Oldenburg and Priyanthi Weerawardhena assisted with geochemical analysis and discussion. Michael Nightingale and Bernhard Mayer supported the gas analyses and assisted with technical discussions. Rolando di Primio and Jon H. Pedersen assisted with geochemical interpretations and discussion. Steve Larter designed and supervised the project. The manuscript benefited greatly from insightful contributions and suggestions made by 3 anonymous reviewers and by the associate editor for the paper.

325

APPENDIX A. SUPPLEMENTARY MATERIAL Supplementary data to this article can be found online at https://doi.org/10.1016/j.gca.2019.07.020. REFERENCES Almeida K. C. S., Oikawa H., Oliveira J. and Duarte C. L. (2006) Degradation of petroleum hydrocarbons in seawater by ionizing radiation. J. Radioanal. Nucl. Chem. 270, 93–97. Aplin A. C., Larter S. R., Bigge M. A., Macleod G., Swarbrick R. E. and Grunberger D. (2000) Confocal microscopy of fluid inclusions reveals fluid-pressure histories of sediments and an unexpected origin of gas condensate. Geol. 28, 1047–1050. Bailey, D., van Aswegen, A., Todd-Pokropek, A. and Humm, J. (eds.) (2015) Nuclear Medicine Physics: A Handbook For Teachers and Students, 1st ed. International Atomic Energy Agency, Vienna, p. 736. Basfar A. A. and Mohamed K. A. (2011) Radiation-induced desulfurization of Arabian crude oil and straight-run diesel. Radiat. Phys. Chem. 80, 1289–1290. Bastow T. P., van Aarssen B. G. K. and Lang D. (2007) Rapid small-scale separation of saturate, aromatic and polar components in petroleum. Org Geochem. 38(8), 1235–1250. https://doi. org/10.1016/j.orggeochem.2007.03.004. Bharati S., Patience R. L., Larter S. R., Standen G. and Poplett I. J. F. (1995) Elucidation of the Alum Shale kerogen structure using a multi-disciplinary approach. Org Geochem. 23, 1043– 1058. Bu¨cker C. and Rybach L. (1996) A simple method to determine heat production from gamma-ray logs. Mar. Pet. Geol. 13, 373– 375. Curiale J. A., Bloch S., Rafalska-Bloch J. and Harrison W. E. (1983) Petroleum-related origin for uraniferous organic-rich nodules of southwestern oklahoma. Am. Assoc. Pet. Geol. Bull. 67, 588–608. Curiale J. A. and Frolov E. B. (1998) Occurence and origin of olefins in crude oils. A critical review. Org. Geochem. 29, 397– 408. Dahl J. E., Moldowan J. M., Peters K. E., Claypool G. E., Rooney M. A., Michael G. E., Mello M. R. and Kohnen M. L. (1999) Diamondoid hydrocarbons as indicators of natural oil cracking. Nature 399, 54–57. Deleersnijder E., Campin J.-M. and Delhez E. J. M. (2001) The concept of age in marine modelling. J. Mar. Syst. 28, 229–267. Edwards J. C. (2011) Chapter 16. Applications of NMR Spectroscopy in the Petroleum Industry. ASTM International, West Conshohoken, p. 657. Erwin D. H. (2006) Dates and rates: Temporal resolution in the deep time stratigraphic record. Annu. Rev. Earth Planet. Sci. 34, 569–590. Farley K. A. (2002) (U-Th)/He dating: Techniques, calibrations, and applications. Rev. Mineral. Geochem. 47, 819–844. Frolov E. B. and Smirnov M. B. (1994) Unsaturated hydrocarbons in crude oils. Org. Geochem. 21, 189–208. Frolov E. B., Melikhov V. A. and Smirnov M. B. (1996) Radiolytic nature of n-alkene/n-alkane distributions in Russian Precambrian and Palaeozoic oils. Org. Geochem. 24, 1061–1064. Frolov E. B., Melikhov V. A. and Vanyukova N. A. (1998) Olefins of radiogenic origin in crude oils. Org. Geochem. 29, 409–420. Gallagher K., Brown R. and Johnson C. (1998) Fission track analysis and its applications to geological problems. Annu. Rev. Earth Planet. Sci. 26, 519–572.

326

S. Larter et al. / Geochimica et Cosmochimica Acta 261 (2019) 305–326

Glasmann J. R. (1992) The fate of feldspar in Brent Group reservoirs, North Sea: a regional synthesis of diagenesis in shallow, intermediate, and deep burial environments. Geol. Soc. London, Spec. Publ. 61, 329–350. Gleadow A. J. W., Duddy I. R., Green P. F. and Lovering J. F. (1986) Mineralogy and confined fission track lengths in apatite: a diagnostic tool for thermal history analysis. Contrib. Mineral. Petrol. 94, 405–415. Grasby S. E. and Betcher R. (2000) Pleistocene recharge and flow reversal in the Williston Basin, central North America. J. Geochem. Explor. 69–70, 403–407. Grasby S. E. and Chen Z. (2005) Subglacial recharge into the Western Canada Sedimentary Basin—Impact of Pleistocene glaciation on basin hydrodynamics. Geol. Soc. Am. Bull. 117, 500. Grice K., Alexander R. and Kagi R. I. (2000) Diamondoid hydrocarbon ratios as indicators of biodegradation in Australian crude oils. Org Geochem. 31, 67–73. Head I. M., Gray N. D. and Larter S. R. (2014) Life in the slow lane; biogeochemistry of biodegraded petroleum containing reservoirs and implications for energy recovery and carbon management. Front. Microbiol. 5, 1–23. Hein F. J. and Cotterill D. K. (2006) The athabasca oil sands — a regional geological perspective, Fort McMurray Area, Alberta, Canada. Nat. Resour. Res. 15, 85–102. Higley D. K., Lewan M. D., Roberts L. N. R. and Henry M. (2009) Timing and petroleum sources for the Lower Cretaceous Mannville Group oil sands of northern Alberta based on 4-D modeling. Am. Assoc. Pet. Geol. Bull. 93, 203–230. Holmes A. (1911) The association of lead with uranium in rockminerals, and its application to the measurement of geological time. Proc. R Soc. A Math. Phys. Eng. Sci. 85, 248–256. J.H. Hubbell S.M. Seltzer 2004 Tables of x-ray mass attenuation coefficients and mass energy-absorption coefficients (version 1.4) Available at: http://physics.nist.gov/xaamdi (accessed August 2, 2016) Karlsen D. A., Nedkvitne T., Larter S. R. and Bjørlykke K. (1993) Hydrocarbon composition of authigenic inclusions: application to elucidation of petroleum reservoir filling history. Geochim. Cosmochim. Acta 57, 3641–3659. Koopmans M. P., Larter S. R., Zhang C., Mei B., Wu T. and Chen Y. (2002) Biodegradation and mixing of crude oils in Eocene Es3 reservoirs of the Liaohe basin, northeastern China. Am. Assoc. Pet. Geol. Bull. 86, 1833–1843. Kovaleva O. V. and Yushkin N. P. (2006) Influence of radiation on the molecular structure of natural solid bitumens. Dokl. Earth Sci. 410, 1145–1147. Larter S., Huang H., Adams J., Bennett B., Jokanola O., Oldenburg T., Jones M., Head I., Riediger C. and Fowler M. (2006) The controls on the composition of biodegraded oils in the deep subsurface: Part II – Geological controls on subsurface biodegradation fluxes and constraints on reservoir-fluid property prediction. Am. Assoc. Pet. Geol. Bull. 90, 921–938. Larter S., Oldenburg T., Marcano N., Snowdon L., Adams J. and Chanthramonti K. (2012) New routes to solutions of the WCSB oil charge conundrum: c-ray Photons and Fourier Transform Mass Spectrometry. In GeoConvention 2012: Vision Calgary, pp. 1–6.

Larter S., Wilhelms A., Head I., Koopmans M., Aplin A., di Primio R., Zwach C., Erdmann M. and Telnaes N. (2003) The controls on the composition of biodegraded oils in the deep subsurface—part 1: biodegradation rates in petroleum reservoirs. Org Geochem. 34, 601–613. Lewan M. D. and Buchardt B. (1989) Irradiation of organic matter by uranium decay in the Alum Shale, Sweden. Geochim. Cosmochim. Acta 53, 1307–1322. Masterson M. E. and Febo R. (1992) Pretransfusion blood irradiation: clinical rationale and dosimetric considerations. Med. Phys. 19, 649–657. Mirkin G., Zaykina R. F. and Zaykin Y. A. (2003) Radiation methods for upgrading and refining of feedstock for oil chemistry. Radiat. Phys. Chem. 67, 311–314. Oakley P. H. H. and Cash W. (2009) Construction of an Earth model: analysis of exoplanet light curves and mapping the next Earth with the new worlds observer. Astrophys. J. 700, 1428– 1439. Okeyode I. C. and Akanni A. O. (2009) Determination of some physical parameters of Olumo Rock, Abeokuta Ogun-state, Nigeria. Indian J. Sci. Technol. 2(7). Peters K. E., Walters C. C. and Moldowan J. M. (2005) The Biomarker Guide, second ed. Cambridge University Press, Cambridge, p. 1155. Ponomarev A. V., Makarov I. E., Saifullin N. R., Syrtlanov A. S. and Pikaev A. K. (2002) Electron-beam radiolysis of gaseous propane in the presence of water under circulation conditions. Radiat. Phys. Chem. 65, 71–78. Selby D. and Creaser R. A. (2005) Direct radiometric dating of hydrocarbon deposits using rhenium-osmium isotopes. Science 308(5726), 1293–1295. Selby D., Creaser R. A. and Fowler M. G. (2007) Re-Os elemental and isotopic systematics in crude oils. Geochim. Cosmochim. Acta 71, 378–386. Tozer R. S. J., Choi A. P., Pietras J. T. and Tanasichuk D. J. (2014) Athabasca oil sands: Megatrap restoration and charge timing. Am. Assoc. Pet. Geol. Bull. 98, 429–447. UNSCEAR (2000) Report to the General Assembly: Sources and Effects of Ionizing Radiation. United Nations Publication, New York, p. 463. Wilhelms A. and Larter S. (2004) Shaken but not always stirred. Impact of petroleum charge mixing on reservoir geochemistry. The Geological Society of London, London, pp. 27–35. Williams F. (1963) Radiation chemistry of hydrocarbons. Q. Rev. Chem. Soc. 17, 101–132. Zaykin Y. A. and Zaykina R. F. (2004) Bitumen radiation processing. Radiat. Phys. Chem. 71, 469–472. Zaykin Y. A., Zaykina R. F. and Silverman J. (2004) Radiationthermal conversion of paraffinic oil. Radiat. Phys. Chem. 69, 229–238. Zaykina R. F., Zaykin Y. A., Mamonova T. B. and Nadirov N. K. (2002a) Radiation methods for demercaptanization and desulfurization of oil products. Radiat. Phys. Chem. 63, 621–624. Zaykina R. F., Zaykin Y. A., Mirkin G. and Nadirov N. K. (2002b) Prospects for irradiation processing in the petroleum industry. Radiat. Phys. Chem. 63, 617–620. Associate editor: Ann Pearson