Nickel geochemistry of oil sands fluid petroleum coke deposits, Alberta, Canada

An amphibious track-mounted sonic drill rig was used to collect continuous core samples at six locations in CB and three locations in CW. The cores were collected by advancing casing in sequential 2.0 m intervals to either 6.0 or 8.0 m below surface. Upon retrieval of each interval, the core was extruded from the casing and, immediately, sub-samples were collected every 0.5 m and transferred into polyethylene (PE) bottles. Samples were shipped to the University of Saskatchewan on ice and stored at −20 °C until analysis.

Prior to analysis, the frozen samples were transferred to an anaerobic chamber (<5% v/v H₂, balance N₂) where they were thawed, vacuum-filtered to remove excess pore water, and sealed in gas-tight amber glass vials. The sealed bottles were removed from the anaerobic chamber and transferred to a −20 °C freezer. Once the samples were frozen, the bottles were removed from the freezer, uncapped, and immediately transferred to a freeze drier, which was operated at −40 °C under vacuum until the samples were completely dry (∼24 h).

Polished thin sections were prepared for a subset of dried samples. The dried coke was vacuum embedded in medical-grade epoxy and mounted to 27 mm × 47 mm quartz glass slides using cyanoacrylate adhesive (Nesbitt et al. 2017). The mounted sections were ground to a thickness of 30 μm and polished with 0.5 μm diamond paste prepared with synthetic kerosene to avoid dissolution of water-soluble phases.

Micro-X-ray fluorescence (μXRF) spectroscopy and micro-X-ray absorption near edge structure (μXANES) spectroscopy data were collected at beamline 13-ID-E at the Advanced Photon Source, Argonne National Laboratory. The beamline utilized a dual Si(111) crystal cryo-cooled monochromator to select incident beam energy. μXRF maps of several samples were collected at an incident beam energy of 15 keV with a 2 μm × 2 μm focused beam. Data were obtained in continuous scanning mode using a four-element Si-drift detector (Vortex ME4; Hitachi, Ltd., Chiyoda, Tokyo, Japan) with a 25 ms effective dwell time for each pixel. Nickel K-edge μXANES spectra were then collected at locations within coke particle cross-sections identified from the Ni K μXRF maps. The spectra were collected by scanning the incident energy from −100 to +644 eV relative to the theoretical Ni K-edge (8333 eV). Fluorescence counts for both μXRF and μXANES were dead-time corrected and I₀ normalized.

Processing of the μXRF mapping data was carried out using both beamline-specific software (GSECARS X-ray Microprobe Map Viewer, version 8) and a custom matrix manipulation code was developed in MATLAB v. 2010a (MathWorks, Inc., Natick, Massachusetts, USA). Data reduction and analysis of μXANES spectra was performed using the Athena module of the Demeter software package (version 0.9.21) (Ravel and Newville 2005). Reference spectra for Ni(II) octaethyl porphyrin (NiOEP), Ni(II) tetraphenyl porphyrin (NiTPP), Ni(II) oxide [NiO], Ni(II) hydroxide [Ni(OH)₂], hydrated Ni(II) sulfate [NiSO₄·6H₂O], and Ni(II) carbonate hydroxide [NiCO₃·2Ni(OH)₂] were previously collected (Nesbitt et al. 2017). A reference spectrum for Ni(II) sulfide [NiS] was also obtained. Linear combination fitting (LCF) of μXANES was performed from −30 to +95 eV relative to the theoretical Ni K-edge (8333 eV). The fit sum was unconstrained (i.e., not forced to 1) so that both the fit sum and R-factor could be used as an assessment of the quality of each fit.

Sequential selective chemical extractions were performed to assess the potential for Ni release from coke under relevant environmental conditions. Consequently, the extraction protocol used for this study included chemical analyses of the water-soluble (F1), exchangeable (F2), reducible (F3), and acid-soluble (F4) fractions (Table 1). The water-soluble extractant solution was prepared by equilibrating ultrapure water (18.2 MΩ·cm) with ambient laboratory air for 24 h. The remaining solutions were prepared in the anaerobic chamber by dissolving high-purity reagents (i.e., ≥99.0%) in ultrapure water that was first purged with N₂ gas for 24 h. The final pH of the F2 (pH = 6.7) and F3 (pH = 7.0) were adjusted using phosphoric acid (H₃PO₄) and sulfuric acid (H₂SO₄), respectively. The final pH of the F1 and F4 fractions were measured but not adjusted.

Freeze-dried coke samples (n = 11) obtained at different depths from locations AB, AM, CX, and DM were sieved into three particle size fractions (f_i): f₁ ≤ 149 μm; 149 μm < f₂ ≤ 350 μm; 350 μm < f₃. Extractions were conducted in triplicate for each of these size fractions. The extraction procedure was initiated by weighing 1.0 g of dried coke into acid-washed 50 mL polypropylene (PP) (SuperClear™ Ultra-High Performance; VWR International, Radnor, Pennyslvania, USA) centrifuge tubes, dispensing 40 mL of ultrapure water (F1) into each tube, and shaking at 80 rpm for 16 h. Following shaking, the tubes were centrifuged (5804R; Eppendorf, Hamburg, Germany) at 4000 rpm (2934g) for 40 min. The supernatant was then collected using a syringe (30 mL HSW NORM-JECT; Henke-Sass Wolf, Tuttlingen, Germany) and passed through a 0.45 μm polyethersulfone (PES) syringe filter (Acrodisc^® 25 mm; Pall Corporation, Port Washington, New York, USA). The filtered supernatant was acidified with nitric acid (HNO₃) to <pH 2 and stored at 4 °C until analysis by inductively coupled plasma—mass spectrometry (ICP-MS) (NexION 300D; PerkinElmer, Inc., Waltham, Massachusetts, USA). This process was repeated for solutions F2 through F4. Between each extraction step, the samples were washed by adding another 20 mL of ultrapure water (18.2 MΩ·cm), shaking at 80 rpm for 20 min, centrifuging at 4000 rotations per minute (rpm) for 20 min, and discarding the supernatant. Nickel release was normalized to the dry coke mass and results were recalculated in mg·kg⁻¹.

The pH point of zero charge (pH_pzc) was determined using a modified version of that described by Pourrezaei et al. (2014). Briefly, a 0.1 mol·L⁻¹ sodium chloride (NaCl) solution was prepared with deionized water. The solution was transferred to an anoxic chamber (≤5 vol.% H_2(g), balance N_2(g)) and bubbled with high-purity nitrogen gas (N_2(g)) to remove dissolved carbon dioxide (CO_2(aq)). After approximately 12 h, 25 mL of this solution was transferred into nine 40 mL amber glass vials and the initial pH was adjusted to between 2 and 10, at 1 pH unit intervals, using 0.1 N hydrochloric acid (HCl) or sodium hydroxide (NaOH) solutions. Next, 0.5 g of homogenized and vacuum-filtered coke was added to each vial, which was sealed in the anoxic chamber using gas-impermeable rubber-lined septa. The sealed bottles were shaken at 60 rpm for 48 h, after which the final pH was measured. The difference between the initial pH and final pH was plotted against initial pH, and the intersection point of the curve with initial pH was taken as the pH_pzc (Pourrezaei et al. 2014).

Multi-level groundwater wells were fabricated using 0.3 cm inner diameter (ID) polyethylene (PE) tubing, which was perforated and screened with 125 μm Nitex mesh over the bottom 10 cm. Eight to 10 of these wells were bundled around a 1.9 cm ID PE standpipe, which was also screened over the bottom 10 cm. The bundles were prepared such that individual well screens were spaced 0.25, 0.5, or 1.0 m apart, with closer spacing at shallower depths to be positioned near the water table. The bundles were installed within 3 m laterally of all continuous coring locations, except BM, using the same sonic drill rig. Briefly, a 7.6 cm diameter steel drill casing fitted with an aluminum knock-out tip was advanced to 6.0 or 8.0 m below ground surface. The well bundle was advanced down the casing, which was retrieved while retaining the well bundle and knock-out tip at the desired depth. Water was added to minimize buoyancy of the bundle during this process. Coke readily collapsed around the well screens as the casing was retrieved.

Water sampling was delayed for at least 30 d after installation. The static water table elevation was measured in the central standpipe at each site prior to purging three well volumes from each well using a peristaltic pump (Geopump Series II; Geotech Environmental Equipment Inc., Denver, Colorado, USA). Clean Pt-cured silicone tubing was connected to the 0.3 cm ID well tubing and fed through the pump to facilitate sampling. The other end of the silicone tubing was attached to a custom-built in-line flow-through cell. Measurements of pH, redox potential (Eh), electrical conductivity (EC), and temperature were performed as water was pumped through the cell. Calibration of the pH electrode (Orion 8156BNUWP ROSS Ultra; Thermo Fisher Scientific, Waltham, Massachusetts, USA) was checked before each measurement and recalibrated using NIST-traceable pH 7, 4, and 10 buffer solutions if the measured and theoretical pH values differed by more than 0.02 pH units. The performance of the redox electrode (Orion 9678BNWP; Thermo Fisher Scientific, Waltham, Massachusetts, USA) was checked between measurements using both ZoBell’s solution (Nordstrom 1977) and Light’s solution (Light 1972). Measured redox potentials were corrected to the standard hydrogen electrode and Eh values are reported.

The flow-through cell was then disconnected, and water was passed through inline 0.45 μm PES filters (Aquaprep; Pall Corporation, Port Washington, New York, USA). Field determinations of alkalinity, H₂S, and NH₃-N were performed immediately upon sample collection. Alkalinity was determined by titration with normalized H₂SO₄ to the bromocresol green methyl red end-point. Dissolved concentrations of hydrogen sulfide (H₂S) and ammonia as nitrogen (NH₃-N) were quantified on a portable spectrophotometer (DR 2800; HACH Company, Loveland, Colorado, USA) using the methylene blue (HACH Method 8131) and salicylate (HACH Method 10031) methods, respectively. Samples for inorganic anions, stable isotopes of water, and dissolved organic carbon (DOC) quantification were stored in clean PE bottles at 4 °C. Dissolved inorganic anion concentrations were quantified by ion chromatography (IC) (ICS-2100; Dionex Corporation, Sunnyvale, California, USA). Vapor equilibration and cavity ring-down spectroscopy (L2120-i Isotopic Water Analyzer; Picarro, Santa Clara, California, USA) were used to measure δ¹⁸O and δ²H values (Wassenaar et al. 2008). The DOC samples were transferred into amber glass vials and acidified with 5% (v/v) H₃PO₄ to remove inorganic carbon. Quantification of DOC was performed by wet ultraviolet light–persulfate oxidation of DOC and thermal conductivity detection of evolved carbon dioxide (Model 1010 TOC Analyzer; OI Analytical, College Station, Texas, USA).

Trace element and major element samples were passed through 0.1 μm PES syringe filters (Acrodisc; Pall Corporation, Port Washington, New York, USA), acidified to <pH 2 using concentrated trace metal grade nitric acid (OmniTrace; EMD Millipore, Burlington, Massachusetts, USA), and stored at 4 °C in clean PE bottles until analysis. Trace elements were quantified by ICP-MS, and major element concentrations were determined by inductively coupled plasma optical emission spectrometry (ICP-OES) (Varian Vista RL; Agilent Technologies, Santa Clara, California, USA). Thermodynamic speciation modelling of pore water chemistry was performed using PHREEQCi (version 3.1.5) with the MINTEQA2 V4 database (Parkhurst and Appelo 2013). Correlations between pore water parameters were evaluated by determining Pearson product-moment correlation coefficients for centered log-ratio transformed data.

μXRF maps of coke particle cross-sections at the Ni Kα emission line revealed heterogeneous Ni distribution within coke particles (Fig. 2). These cross-sectional maps revealed multiple rings characterized by elevated Ni concentrations, which is consistent with the concentric layer structure previously reported for oil sands fluid petroleum coke (Furimsky 2000; Zubot 2010; Nesbitt et al. 2017). These maps indicated that Ni concentrations were elevated near the outer margin relative to the inner region of individual layers. Vanadium exhibits a similar distribution within individual layers that comprise fluid petroleum coke particles (Nesbitt and Lindsay 2017). These findings suggest that Ni is concentrated at the outer margin of each concentric layer during the fluid coking process.

Nickel K-edge μXANES spectra exhibited a distinct doublet feature on the white line at ∼8352 and ∼8359 eV. All spectra exhibited a rounded shoulder at 8340 eV and a small pre-edge feature at 8334 eV. Spectra from the inner region of individual concentric layers exhibited a higher magnitude at 8359 eV peak. These spectra were generally consistent with previously reported Ni K-edge XANES spectra (Nesbitt et al. 2017) where Ni(II) occurred in porphyrinic square-pyramidal N coordination. Conversely, the Ni K-edge μXANES spectra obtained for the outer margins of individual concentric layers, where elevated Ni concentrations were observed, generally exhibited a higher magnitude peak at 8352 eV. The Ni μXANES line shape after ∼8370 eV was similar among all measurement spots and was generally consistent with Ni K-edge spectra previously reported for bulk fluid coke samples (Nesbitt et al. 2017). Nickel K-edge XANES spectra for asphaltene samples from Alberta oil sands (Lytle 1983) were generally consistent with spectra obtained for Ni porphyrins and the inner region of individual concentric layers (Fig. 2).

Two potential mechanisms could account for the observed differences in Ni coordination between the inner region and outer margin of each concentric layer. The first potential mechanism is distortion of the organo-metallic asphaltenic Ni site during coking. Finite difference modelling of the near-edge structure (FDMNES) (Bunău and Joly 2009) for Ni(II) porphyrins indicated that slight inflation and deflation of the overall molecular structure strongly influenced the doublet feature (Nesbitt et al. 2017). When the local porphyrin structure (i.e., Ni-N, N–C, and C–C bonds) surrounding the Ni metal centre were equally inflated by between 3% and 10%, the characteristic porphyrin doublet white line feature was replaced with a single sharp asymmetrical white line at ∼8352 eV. The second potential mechanism is distortion or degradation of metalloporphyrins and accumulation of inorganic Ni phases during fluid coking. For example, Ni(II) oxide exhibits a strong whiteline feature at ∼8350 eV but no secondary peak at ∼8359 eV (Fig. 2). This phase also exhibits a pre-edge peak at 8333 eV without a shoulder at 8340 eV. Although there may be distortions and variability in Ni coordination proximal to the organo-metallic Ni centres in asphaltenic micelles, the spectral similarity among all measurement spots at higher energy implies an underlying constant long-range Ni coordination.

Combinatorial fitting was performed for each Ni K-edge μXANES spectrum (n = 18) to assess all unique combinations of up to four of the six reference spectra (i.e., NiOEP, NiTPP, NiO, NiS, Ni(OH)₂, NiSO₄·6H₂O, NiCO₃·2Ni(OH)₂). The best fits for all spectra, which exhibited the smallest R-factors and residuals, consistently included NiTPP, NiOEP, NiO, and NiS. Fitting Ni μXANES spectra with porphyrin, NiO, and NiS standards generally yielded quantitatively acceptable (R-factor <0.009), albeit occasionally qualitatively poor, fits (Table 2). Spectra obtained for Ni(OH)₂, NiSO₄·6H₂O, and NiCO₃·2Ni(OH)₂ reference materials exhibited similar features to those of NiO. Therefore, including more than one of these reference spectra in the LCF produced negative component weightings and therefore did not improve overall fits. Nickel(II) sulfide can be formed during hydroconversion processes via hydrodemetallation reactions of porphyrinic Ni prior to coking (Gray 2015). Fine Ni/Mo-sulfide catalyst particles may also be transported to the coking process through attrition of larger catalyst pellets in upstream hydroconversion reactors (Gray 2015). The shoulder at ∼8340 eV in the NiS reference spectrum contributes to the shoulder observed at the same energy in the sample spectra, which improved fits for almost every sample. The LCF results indicated that the NiO and NiS reference spectra explained up to a maximum of 43% of the sample spectra, with the remaining proportion comprised of varying proportions of NiTPP and NiOEP reference spectra. Moreover, Ni μXANES spectra obtained from spots near outer margins of individual concentric layers exhibited the highest NiO and NiS weightings, which suggests elevated Ni concentrations also correspond to the highest relative proportion of inorganic or non-porphyrinic Ni phases.

To further assess contributions to the lineshape of these Ni K-edge μXANES spectra, a second LCF scenario was also considered. Linear combination fits were generated using a Ni K-edge XANES spectrum for AOSR asphaltene (Lytle 1983) and NiS, plus one of the following: NiO, Ni(OH)₂, NiSO₄·6H₂O, or NiCO₃·2Ni(OH)₂ (Fig. 2). The Ni K-edge spectrum for this asphaltene sample is likely more representative than the synthetic porphyrin reference materials of pre-coking Ni speciation. Consequently, fit quality—determined based upon R-factor and residuals—was improved considerably with this second scenario. Nickel coordination in this asphaltene sample was, however, less well constrained than in the NiTPP and NiOEP reference materials. In this second LCF scenario, different inorganic Ni reference spectra, all of which are geochemically feasible within coke deposits, were included in the best fits of various sample spectra (Table 3). Due to the similar features of Ni K-edge spectra for the various inorganic reference materials, the LCF results do not necessarily confirm the presence or absence of a given inorganic Ni-bearing compound. Nickel(II)-sulfide was expected in the samples and, therefore, was included in fits for all sample spectra. The NiS reference spectrum differs substantially from that of the other inorganic reference materials and is unlikely to produce non-unique fits.

In the first LCF scenario, NiO was consistently favoured over other inorganic standards (i.e., Ni(OH)₂, NiSO₄·6H₂O, or NiCO₃·2Ni(OH)₂), because small changes in the overall spectra could be accounted for by varying the relative proportion of NiTPP and NiOEP. Conversely, in the second LCF scenario, different inorganic standards were favoured over NiO because the asphaltene alone could not account for small variations in the overall spectra. The fit sum was consistently closer to 1.0 for fits generated using the second scenario, further indicating that the Ni K-edge spectrum for AOSR asphaltene is a better fit than any linear combination of spectra for the synthetic NiTPP and NiOEP reference materials.

Linear combination fitting approximated the relative magnitudes of the doublet peaks at ∼8352 and ∼8359 eV reasonably well. Conversely, attempts to simulate the dampening and broadening of the shoulder at 8340 eV, although improved by the inclusion of NiS, were generally unsuccessful (Fig. 2). This inconsistency could result from the presence of an unknown Ni-bearing phase or, more likely, from variability at the nano-scale in the distortion of Ni coordination sites within coke. This interpretation is consistent with results of FDMNES analysis described by Nesbitt et al. (2017).

Results of the sequential extraction protocol revealed that Ni release was dependent on both extractant and particle size (Fig. 3). Water-soluble Ni (F1) averaged 0.25 ± 0.22 mg·kg⁻¹ across all locations and grain-size fractions. The highest water-soluble Ni concentrations, which exceeded 0.5 mg·kg⁻¹, were observed for the smallest grain-size fraction (f₁) and the shallowest (i.e., 0 and 0.5 m below surface) samples. Exchangeable Ni (F2) exhibited a median value of 2.6 mg·kg⁻¹ among all samples and averaged 3.1 ± 0.9 and 2.1 ± 1.1 mg·kg⁻¹ for the smallest and largest particle-size fractions, respectively. Background Ni concentrations in the weak reductant solution (F3) approached values measured in the supernatant following this extraction step. Consequently, Ni release could only be determined for two samples. Finally, acid-soluble Ni (F4) for all locations and grain-size fractions averaged 0.47 ± 0.19 mg·kg⁻¹ with a median value of 0.43 mg·kg⁻¹. Although not statistically significant, there was an apparent inverse relationship between Ni release and grain size for the F1 and F2 fractions. Overall, the summed Ni released during the selective extraction protocol averaged 3.6 ± 1.4 mg·kg⁻¹, which corresponds to 1.6% ± 0.8% (w/w) of total Ni concentrations reported by Nesbitt et al. (2017).

Extraction results indicated that only a small proportion of total Ni present in these coke deposits can be leached under environmentally relevant conditions. This result is consistent with the interpretation that porphyrin complexes are not the principal source of dissolved Ni within oil sands fluid petroleum coke deposits. Although organic Ni phases generally comprise the major component of total Ni (Nesbitt et al. 2017), inorganic phases are a likely source of dissolved Ni within these deposits. Limited Ni release during the water-soluble sequential extraction step suggested that NiSO₄·6H₂O and other soluble inorganic Ni(II) phases did not represent a major component of leachable Ni within the coke samples. In contrast, approximately 80% of total Ni release occurred during the reducible sequential extraction step. The extractant solution used during this step was buffered to pH 6.7, which is near the measured pH_pzc of 6.4 and a value of 6.5 ± 0.3 previously reported for oil sands fluid petroleum coke (Pourrezaei et al. 2014). Consequently, exchange with Na and a pH-dependent change in net surface charge may have contributed to Ni release in the exchangeable fraction. Nickel release was limited in the reducible fraction, but some Ni release was detected with the acid-soluble fraction. Iron and Mn release with these fractions were also limited, suggesting that Fe(III) and Mn(V) (hydr)oxide contents were generally low. Therefore, Ni release in the acid-soluble fraction is more likely associated with carbonate, hydroxide, or sulfide phases and potentially with desorption of remaining Ni at low pH. Overall, these results indicate that Ni release may be enhanced at elevated ionic strength and acidic pH.

Pore water samples from the CB and CW deposits exhibited total dissolved Ni concentrations ranging from 2 to 120 μg·L⁻¹ with a median value of 7.8 μg·L⁻¹ (Fig. 4). The highest concentrations occurred within the upper 1.0 m of the saturated zone, and dissolved Ni generally decreased with increasing depth below the water table. This trend was generally opposite to that observed for pH, which increased with depth below the water table (Fig. 4). The median pore water pH was 7.7 and values ranged from 6.3 to 8.4. A significant (p < 0.05) negative correlation (R = −0.79) between pH and Ni (Fig. 5) suggested that pH-dependent precipitation–dissolution or sorption reactions influenced Ni distribution and mobility within coke deposits. Zajic et al. (1977) observed a similar negative relationship between dissolved Ni concentrations and pH in coke pore water. Puttaswamy and Liber (2011) reported Ni concentrations in oil sands fluid petroleum coke leachate of 145 ± 31 μg·L⁻¹ at pH 5.5 compared with 0.2 ± 0.1 μg·L⁻¹ at pH 9.5. Thermodynamic speciation modelling results indicated that pore water was consistently undersaturated with respect to NiO_(s), NiCO_3(s), and Ni(OH)_2(s), suggesting that these phases were not principal controls on dissolved Ni concentrations.

Pore water alkalinity (median: 230 mg·L⁻¹ as CaCO₃) exhibited a downward increasing trend, with concentrations exceeding 400 mg·L⁻¹ (as CaCO₃) observed in the lowermost wells for most locations (Fig. 4). Dissolved organic carbon concentrations ranged from 0 to 15 mg·L⁻¹ (median 2.2 mg·L⁻¹) and generally increased with depth. Dissolved SO₄ concentrations (median 340 mg·L⁻¹) concentrations exhibited more variability, both among locations and with depth at individual locations. Previous research suggests Ni leaching may be enhanced in the presence of dissolved SO₄ (Puttaswamy and Liber 2012). However, total dissolved nickel concentrations did not correlate with SO₄, alkalinity, or DOC, suggesting these constituents were not important controls on Ni mobility within coke deposits. Elevated H₂S concentrations were detected at several locations with concentrations ranging from <5 to 30 μg·L⁻¹. Consequently, pore water was supersaturated with respect to NiS_(s), mackinawite [FeS_(s)], and pyrite [FeS_2(s)]. Nickel incorporation into mackinawite and pyrite formed under sulfate-reducing conditions can occur (Luther et al. 1980; Huerta-Diaz et al. 1998; Rickard 2012). Precipitation of these sulfide phases, therefore, has potential to limit dissolved Ni concentrations under sulfate reducing conditions within the coke deposits.

Speciation modelling results suggested that Ni²⁺ comprised 30%–90% of dissolved Ni (median 62%), whereas NiHCO₃⁺, NiCO₃⁰, and NiSO₄⁰ accounted for the remainder. In general, the proportion of Ni²⁺ predicted by the model declined with depth corresponding to decreased total Ni concentrations and increased proportions of both NiHCO₃⁺ and NiCO₃⁰. Depth-dependent trends in NiSO₄⁰ abundance were not apparent due to the greater spatial variability in SO₄ concentrations relative to alkalinity. Complexation with Cl⁻ and OH⁻ was not predicted to strongly influence aqueous Ni speciation. The presence of non-ionic complexes (i.e., NiCO₃⁰, NiSO₄⁰) may have partially limited Ni sorption.

The pore water data indicate that pH is a dominant control on dissolved Ni concentrations within fluid coke deposits. Pore water pH ranged from slightly below (pH 6.2) to above (pH 8.4) measured and published (Pourrezaei et al. 2014) pH_pzc values for oil sands fluid petroleum coke. This observation suggests that sorption–desorption reactions involving Ni²⁺ and positively charged aqueous Ni complexes (e.g., NiHCO₃⁺) likely influence dissolved Ni concentrations within the fluid coke deposits. Consequently, the observed negative relationship between pH and dissolved Ni concentrations may result from pH-dependent variation in net surface charge and, therefore, Ni²⁺ and NiHCO₃⁺ sorption. The precipitation of secondary sulfide phases may have limited dissolved Ni concentrations at greater depths below the water table. However, Ni K-edge μXANES spectra could not discriminate between NiS formed by catalytic conversion of porphyrinic Ni and potential secondary NiS precipitated in situ. Nevertheless, the pore water chemistry results indicate that Ni mobility should be limited in mildly alkaline and sulfidic environments, which are commonly observed in tailings deposits in the AOSR (Stasik et al. 2014; Dompierre et al. 2016; Reid and Warren 2016).