1 Commercial hydrocarbon production in New Brunswick extends back to the mid-Nineteenth Century, when the province exported bitumen mined from oil-shale and albertite (solid bitumen) deposits at Albert Mines, located approximately 25 km southeast of Moncton (Martin 2003). Early hydrocarbon exploration focused on known oil and gas seepages in the area south of Moncton (Foley 1989), and one of the first oil wells in North America was drilled in 1859 at Dover, east of the Peticodiac River (St. Peter 2000). Oil-bearing rock in the first production well was encountered at an approximate depth of 27 m (Martin 2003), indicating near surface occurrences of hydrocarbon-bearing units. Until recently, however, the most significant oil and natural gas (ONG) production has come from the Stoney Creek field (St. Peter 2000). It is located west of the Peticodiac River, approximately 20 km south of Moncton (Fig. 1). In that area of the field, well depths range from approximately 800 to 1500 m. The field began producing primarily gas in 1910, with a peak annual production of 2.4 × 10⁷ m³ (860 mmscf). In the mid-1920s, oil production increased, peaking at 4500 m³ (30 000 bbls) per annum. In the late-1940s, the field saw a rapid drop in gas production, while oil production declined more slowly until production stopped in 1991. Oil production recommenced in 2007 and continues today, with a maximum annual production since then of 5000 m³. The Stoney Creek oil field (SCOF) currently includes approximately 160 wells: three are horizontal wells and the rest are vertical wells. Provincial records indicate that most wells have been abandoned, but approximately 34 are either shut in or producing. Another ~50 nonproducing wells are located outside the main SCOF area. In 2000, 55 km to the southwest of the SCOF, gas was discovered at a depth of approximately 2500 m. The field is identified as the McCully gas field (MCGF). Production at the MCGF began in 2003 and had peak gas production of 2.1 × 10¹⁶ m³ in 2008. Provincial records indicate that there have been 39 ONG wells drilled in the field. The wells are vertical or inclined and were hydraulically fractured (Jiang et al. 2016).

2 Hydrocarbon resources from the two ONG fields, and from Albert Mines, are associated with the lacustrine hydrocarbon-bearing shales and sandstones of the Albert Formation. Conventional and fracture-stimulation extraction techniques have been used in the SCOF throughout its history (St. Peter 2000), whereas the MCGF has required fracture stimulation to achieve commercial production rates (Corridor Resources Inc. 2019). With recent advances in horizontal well drilling and hydraulic fracturing (i.e., unconventional ONG extraction), the natural gas and oil held in the Albert Formation shales and tight sandstones may be extracted at economical rates in other areas of the province. However, the surge in unconventional ONG extraction has brought intense scrutiny to the petroleum industry from the public, governments and academics (Vidic et al. 2013).

3 Oil and natural gas production generates concerns regarding the risk to freshwater resources from contamination resulting from wellbore integrity failure, waste-water discharge and accidental spills at the surface (Vengosh et al. 2014). This has prompted a wide range of research, including many regional groundwater studies attempting to determine impacts from pre-existing unconventional ONG production (McMahon et al. 2017) and in areas of potential unconventional development (Harkness et al. 2018; McIntosh et al. 2014; Moritz et al. 2015). Other evaluations have been conducted before, during and after unconventional ONG extraction (Harkness et al. 2017). Studies of this type commonly focus on shallow potable water aquifers in the region of interest and include the collection of data on the molecular composition of the gas, methane (C₁), ethane (C₂) and propane (C₃) content, and the isotopic compositions of carbon (δ¹³C) and hydrogen (δ²H) in C₁. The isotopic composition of hydrocarbon gases can indicate if the gases were formed through microbial action, commonly referred to as biogenic gas, or through abiotic cracking of oil or kerogen, commonly referred to as thermogenic gas (Schoell 1983; Whiticar 1999; Milkov and Etiope 2018). The isotopic compositions of hydrocarbon gases may also be used as tracers to identify natural connections to hydrocarbon reservoirs or pathways for stray gas entering shallow aquifers due to ONG resource developments (Sharma et al. 2014). However, to be useful as a tracer, the isotopic composition of the gas source must be known. Barton (2018) characterized gases from the MCGF, but isotopic characterization of gas from the SCOF has not been carried out. Therefore, the first objective of the work presented here was to determine the isotopic composition of hydrocarbon gases from the SCOF.

4 In 2014, the New Brunswick Energy Institute commissioned a regional baseline groundwater study across southeastern New Brunswick (Loomer et al. 2016). The first objective of the study was to collect and report on private well-water chemistry in regions that the petroleum industry had identified as potential targets for unconventional ONG exploration and development. In 2017, additional regional well-water sampling in the vicinity of the SCOF was carried out (Loomer and MacQuarrie 2018). The groundwater chemistry data compiled in these two studies included dissolved hydrocarbon gas content and their isotopic compositions. The second objective of the present work was to evaluate the isotopic composition of hydrocarbon gases detected in shallow groundwater, including the rarely reported isotopic composition of C₂, and compare them to the chemical and isotopic compositions of gases from the ONG fields. Similarities in the molecular and isotopic compositions of the hydrocarbon gases at depth and in shallow groundwater can be an indication of direct and rapid transport of the gases toward the surface, possibly facilitated by conduits such as natural open fractures or leaking well bores. However, the isotopic compositions of hydrocarbon gases may be altered as a result of secondary effects such as migration, mixing and microbial action (Whiticar 1994). Using the data from the two regional groundwater studies, data from the MCGF (Barton 2018), and additional hydrocarbon data from the SCOF, this work seeks to identify the presence of such effects and to use additional hydrocarbon data, along with knowledge of the local geology, to aid in the assessment of the gas origins in shallow groundwater. In doing so, it incorporates a more complete, updated and detailed interpretation of the groundwater data from Loomer et al. (2016) and Loomer and MacQuarrie (2018).

5 The study is set within the late Paleozoic Maritimes Basin of Atlantic Canada (Fig. 1), a successor basin that formed subsequent to the Early to Middle Devonian Acadian Orogeny. It developed in five cycles of subsidence and deposition punctuated by basin inversion/uplift and erosion throughout the Carboniferous Period (Wilson and White 2006). It is characterized by thick deposits of predominantly sedimentary rocks that were formed in alluvial, fluvial, lacustrine and marine environments. St. Peter and Johnson (2009) provided a compilation of previous work on the stratigraphic and structural evolution of the Maritimes Basin in New Brunswick. Briefly, the basin is divided into six main stratigraphic units. In order of decreasing age, they are the Horton, Sussex, Windsor, Mabou, Cumberland, and Pictou groups. The Horton Group is composed of alluvial and freshwater lacustrine strata and overlies a pre-Late Devonian crystalline basement complex. The Sussex Group, a sequence of continental clastic rocks deposited during a more arid interval, truncates the Horton Group with an angular unconformity. The Windsor Group formed as a result of a marine incursion and unconformably overlies the Sussex Group. It comprises coarse-grained clastic rocks and red mudstones, together with carbonates, gypsum, anhydrite, halite and local potash beds. The coarse-grained clastic red beds of the Mabou Group overlie the Windsor strata. The Cumberland Group unconformably overlies the Mabou Group and consists primarily of terrestrial red and grey conglomerates and coarse sandstones with lesser mudstones and minor coal seams. The youngest group in the sequence, the Late Pennsylvanian Pictou Group, is made up of fining-upward fluvial cycles of grey and red sandstones and mudstones with coal seams. The Maritimes Basin includes a series of deep, northeast-trending grabens, commonly referred to as subbasins, separated by intervening horsts. The New Brunswick Platform is a horst and the subbasins are infilled with Horton, Sussex, Windsor and Mabou group rocks. The Cumberland and Pictou groups form a cover sequence that unconformably overlies strata in the older subbasins, with the Pictou Group covering most of the New Brunswick Platform.

6 The units most relevant to commercial ONG production in the province are the Dawson Settlement, Frederick Brook and Hiram Brook members of the Albert Formation in the Horton Group (Fig. 1c). The present-day maximum burial depth is approximately 5 km in the Sussex area (Park 2014). Keighley (2008) described the members of the Albert Formation in terms of a tectonically controlled, wedge-shaped lacustrine basin, where the Dawson Settlement and Frederick Brook members represent the sediment-starved, lacustrine-dominated stage. The basin then infilled with the predominantly deltaic and fluvial-floodplain sediments of the Hiram Brook Member. The three members are considered to be gradational with lateral interfingerings of lithologies (sandstones and shales) (St. Peter and Johnson 2009). The primary zone of gas production from the SCOF was the Hiram Brook Member, and the primary oil producing zone involved sandstones of the Dawson Settlement Member (Foley 1989; St. Peter and Johnson 2009). However, oil-rich zones in the Hiram Brook Member at the SCOF are also known (Chowdhury et al. 1991). In the MCGF, the targets for gas have been the Hiram Brook and Frederick Brook members (Park 2014; Barton 2018).

7 The present study is focused on the five areas within the Maritimes Basin that were included in two recent regional groundwater studies (Fig. 1b; Loomer et al. 2016; Loomer and MacQuarrie 2018). Three of these — the Central, Kent and Shediac areas — had been targeted for ONG exploration but have yet to see resource development activities. The other two areas have seen limited (Sussex) to extensive (Stoney Creek) ONG development. The Sussex area includes the MCGF, and the Stoney Creek area extends over a 10-km radius from the centre of the SCOF. The Central and Kent areas are located on the New Brunswick Platform; the majority of the groundwater wells in these two study areas are completed in the Pictou Group. The Shediac area straddles the Cocagne subbasin, the Indian Mountain deformed zone, and the Westmorland uplift (St. Peter and Johnson 2009); however, the surficial bedrock primarily belongs to the Pictou Group. The Sussex and Stoney Creek areas are located within the Moncton subbasin, with faulted and folded strata from all geologic groups of the Maritimes Basin outcropping at surface.

8 The SCOF produces mainly oil; however, natural gas is captured from the wellheads and used to power machinery on site. On 29 October 2018, samples of this gas were collected at the wellhead from seven producing wells (Fig. 2). The sampled wells include three horizontal boreholes drilled in 2006 and 2008 that targeted the Dawson Settlement Member (wells A-89, I-88 and N-78). Four previously shut-in wells (12, 14, 111 and P8), which date from as early as 1910, were re-entered in 2007, and these were also sampled. Well depths range from 800–1580 m, and current oil production is from the Dawson Settlement Member (Acadian Securities 2007; Orlen 2019). The older wells were originally open holes below the Hiram Brook Member (St. Peter and Johnson 2009), and natural gas production was primarily from the Hiram Brook Member, whereas the main oil producing unit was the Dawson Settlement Member (Foley 1989). Casing and perforation details from the horizontal and recompleted wells were not disclosed, but extraction from multiple units is possible (Acadian Securities 2007). Therefore, the production of natural gas cannot be constrained to a specific unit. Gas samples were collected in Isotubes® (IsoTech Laboratories, Champaign IL, USA) using a pressure-controlling sampling manifold.

9 Groundwater data are reported for a subset of 52 wells from the two regional studies (Loomer and MacQuarrie 2018; Loomer et al. 2019). The dataset represents groundwater with dissolved C₁ concentrations >0.1 mg/L and includes wells located within all five study areas. All wells sampled in the Central, Kent, Shediac and Sussex areas are located more than 800 m from an ONG well. In the Stoney Creek area, the horizontal distance between the water wells and the nearest ONG well ranges from 85–4900 m. The water wells were sampled twice, the second including the collection of samples for the determination of δ¹³C and δ²H in C₁. The isotopic composition of C₂ was also determined from Stoney Creek well water samples when C₂ concentrations were >0.1 mg/L. Samples were collected between June and October in 2014, 2015 and 2017. The time between the collection of the first and second samples for each well ranged from 13 to 349 days, but was typically on the order of 40 days.

10 All water wells had standard 15-cm-diameter steel casing through overburden and were completed as open holes in bedrock, or with slotted well screens when completed in unconsolidated sediments or erodible shale. Well depths ranged from 6 to 88 m, with a median depth of 37 m. The wells supplied water to private homes, businesses or farm buildings and had pressure tanks to maintain relatively constant pressure. All samples were collected upstream of any water treatment systems present. Sampling and analysis followed the procedures described in Loomer et al. (2016, 2018) and Loomer and MacQuarrie (2018). Samples were collected upon stabilization of the field parameters (pH, specific conductance, dissolved oxygen and E_h). Samples for dissolved hydrocarbon gases were collected using the submersed vial method. Analyses for the molecular composition of the hydrocarbon gases were conducted at the Research and Productivity Council in Fredericton, New Brunswick, using a Bruker Gas Chromatograph coupled to a Flame Ionization Detector (method AQS90). The results from the headspace analyses were converted to aqueous concentrations using Henry’s Law. The instrument detection limit was 0.0001 mg/L and the detection limit based on field blanks was 0.0005 mg/L for C₁ and 0.0001 mg/L for C₂ and C₃. Results from field blank and replicate samples are provided in Tables A1 and A2 in Appendix A. The presence/absence of ethene (C₂₌), propene (C₃₌), butane, pentane and hexane (C₄-C₆ hydrocarbons) were also evaluated in groundwater samples from the Stoney Creek area. Standards were injected into the instrument to obtain the retention time for each compound and the GC software CompassCDS™ (Scion Instruments, Edmonton, Alberta) was used to identify peaks in the two sample chromatographs (Appendix B). Quantification of C₂₌, C₃₌, and C₄-C₆ was not available.

11 Molofsky et al. (2016) found that there was a significant difference in dissolved C₁ concentrations between open direct-fill, semi-closed (similar to this study) and closed (IsoFlask®) techniques when C₁ concentrations were >20 mg/L. To evaluate possible losses from the sampling setup used in this work, 42 wells in the Stoney Creek area were sampled using both the submersed-vial and IsoFlask® methods (Loomer and MacQuarrie 2018). Methane concentrations determined using the IsoFlask® method were up to 15 mg/L higher than those determined from the submersed-vial method when C₁ concentrations were >15 mg/L. Based on the data from Loomer and MacQuarrie (2018), the following relationship was used to correct concentrations from the submersed vial technique (vial, mg/L) to equivalent IsoFlask® concentrations (Corrected, mg/L) for samples that were not collected with IsoFlasks:

12 Samples for dissolved gas isotopic analysis were collected in the same manner as for dissolved gas concentrations, either in IsoFlasks® or in submersed 700-mL glass Wheaton bottles with screw caps and butyl rubber septa, preserved with 6N HCl. Isotopic and compositional analyses for Stoney Creek produced gas and the isotopic analyses of dissolved gases in groundwater were conducted at the Ján Veizer Stable Isotope Laboratory at the University of Ottawa. The chemical composition of the gas samples (O₂, CO₂, N₂ and hydrocarbons C₁-C₅) was measured using an SRI 8610C dual column gas chromatograph fitted with flame ionization and thermal conductivity detectors. Methane and ethane were extracted from the groundwater samples using the headspace equilibrium technique. Stable carbon and hydrogen isotope measurements of the hydrocarbon gases, C₁ to C₃ were conducted if concentrations were above 0.08% in the gas phase. The isotopic compositions of C₄ and C₅ were reported when their peaks in the GC could be resolved. Analyses were performed using a Thermo-Fisher Delta V Isotope Ratio Mass Spectrometer interfaced via a Conflow 4 to an Agilent 8790A gas chromatograph equipped with a combustion furnace (1000°C) for carbon oxidation to CO₂, and a pyrolysis furnace (1420°C) for hydrogen conversion to H₂. The National Institute of Standards certified reference standards NGS1 and NGS2 were used for the ¹³C and ²H in C₁ calibration for groundwater samples from the Central, Kent, Shediac and Sussex study areas. For the analyses of the groundwater samples from the Stoney Creek area and the Stoney Creek production gas samples, the USGS HCG-1, -2 and -3 standards were used. The USGS standards offer standardized values for ¹³C and ²H in C₁–C₃. No certified standards were available for C₄ and C₅. Therefore, peak retention times were confirmed with a C₁–C₆ gas composition standard (Air Liquide, Ottawa), and the isotopic calibration for C₃ was applied to C₄ and C₅. Carbon dioxide concentrations in the gas samples were too low for reliable isotopic measurements.

13 Differences in δ¹³C values for field duplicate gas samples ranged from 0.0–0.7‰, showing no apparent trends with carbon number or isomer. Therefore, the reported δ¹³C uncertainty is based on the maximum difference measured, ± 0.7‰, for all hydrocarbon gases. Differences in δ²H for duplicate samples ranged from 0.3–2.0‰, with uncertainty increasing from C₁ to C₃. Hold times for the dissolved gas samples varied from 1–9 weeks due to capacity limitations of the analytical laboratory. Evaluation of the effects of prolonged hold times showed no correlation between δ²H-C₁ values and sample hold times; however, the δ¹³C₁ values became progressively depleted over time (Loomer et al. 2018). Therefore, the maximum estimated uncertainty in δ¹³C₁ based on the regression analysis presented in Loomer et al. (2018) and the hold times for samples in this study is -4.6‰. The uncertainty in δ²H-C₁, based on duplicate analyses, is ± 4.4‰.

25 In this work we have compiled molecular and isotopic composition data from two regional groundwater studies and compared them to the molecular and isotopic composition data from ONG fields in New Brunswick with the objective of identifying relationships between hydrocarbons found at depth and those observed in shallow groundwater. New isotopic data from the SCOF indicate that a wet thermogenic gas was co-produced with oil in the ONG field. The gas was likely generated under conditions near the onset of the oil-gas transition zone and the δ¹³C and δ²H isotopic profiles are consistent with KIE models. The gas currently produced from the SCOF has a lower thermal maturity than that of the MCGF.

26 The effects of oxidation on the isotopic composition of carbon and hydrogen in aqueous C₁ are apparent when C₁ concentrations are low (≤1 mg/L). Data from these low C₁ samples can lead to conflicting interpretations when considering multiple source-discrimination approaches and we found that including the C₁ concentrations can help decipher the apparent contradictions. All groundwater samples from the Pictou Group have low C₁ concentrations (≤4 mg/L), and the presence of thermogenic gas in this geologic unit is considered unlikely. Elevated C₁ concentrations (>5 mg/L) were observed in groundwater samples from bedrock in the Horton, Sussex, Mabou and Cumberland groups. Multiple samples from wells completed in the Horton Group also show conflicting biogenic/ thermogenic signatures despite gases from the Horton Group having different and distinct thermogenic signatures at depth. The possible influence of biogenic activity is also evident in a well completed in the Mabou Group. Conflicting molecular, C₁-C₆, and isotopic, δ¹³C₁ and δ¹³C₂ data from these wells suggest that biological activity has overprinted an original thermogenic source in near-surface Horton Group bedrock. Therefore, the isotopic composition of dissolved hydrocarbon gases in shallow groundwater may be subject to considerable alteration not only at low C₁ concentrations, but also at elevated concentrations. This has major implications for the understanding of baseline methane occurrences should ONG development proceed in the Maritimes Basin. Future baseline study design and interpretation should be evaluated within this context. Multiple lines of evidence are required for reliable interpretations concerning hydrocarbon origins.

27 Thermogenic isotopic signatures occur in samples with elevated C₁ concentrations from several locations where bedrock of Sussex, Mabou or Cumberland group directly overlies the Horton Group. This suggests C₁ may be migrating through natural pathways from the Horton Group without significant alteration of the isotopic composition. Fracture-controlled transport is the most likely explanation.