Treatment of Oil Sands Process-Affected Water (OSPW)

Abstract

The treatment of oil sands process-affected water (OSPW) is important because of the challenges associated with the direct discharged to aquatic environments. This study compared five constructed wetland treatment (CWT) designs by assessing their performance for the treatment of OSPW. Both the aerated and non-aerated treatment wetlands were effective in transformation of classical NAs and NAFCs. The aerated wetland; 2B and the non-aerated wetland; 6B were the most successful systems for the transformation of NAs. Although the preferences based on extent and rates of transformation for naphthenic acid (NA) species were different between the aerated wetland system and the non-aerated wetland systems. The aerated system demonstrated the most efficient transformation of NA species with fewer carbon number and DBE while anaerobic transformation is likely the primary fate of the more complex species (i.e. species with more carbon numbers and DBE). Analysis indicated that the various design configurations played a key role in influencing the preferential transformation NA species in the constructed wetlands. The result further suggest that for efficient treatment of OSPW, the combination of the aerated system and non-aerated system designs and conditions (i.e. 2B and 6B systems) and the marginal cost associated with non-aeration is a good approach.

Keywords: Constructed wetland, Orbitrap-MS, Aeration, Non-aeration, Naphthenic acids, Oil sands process-affected water, Kinetics, Transformation.

Highlights

• Five different constructed wetland treatment designs were compared to gain more insight into preferential treatment of NA species.
• The five different CWT systems effectively transformed NAs.
• Aerated wetland system (2B) showed higher transformation of NAs with fewer carbon number and DBE.
• Non-aerated wetland system (6B) showed higher transformation of NAs with more carbon number and DBE.
• CWT designs contributed to preferential transformation of NA species.

1. Introduction

Oil sand upstream operations using the alkaline hot water extraction technique generate high volumes of oil sand process-affected water (OSPW) traditionally contained in tailing ponds (Siddique et al., 2011). OSPW arouse environmental concerns because of their large accumulations (CAPP, 2014), tailing ponds spatial coverage in the Athabasca oil sand (AOS) area is about 170 km² (Siddique et al., 2011; Pereira et al., 2013) and plausible aquatic toxicity (Anderson et al., 2012; Kavanagh et al., 2012; Marentette et al., 2015; 2017). The presence of classical naphthenic acids (NAs) are the primary contribution to OSPW toxicity (Morandi et al., 2015).  NAs are naturally-occurring alkyl-substituted saturated cyclic and alicyclic carboxylic acids with general description C_nH_2n+zO_x where n is the carbon number, z is hydrogen replacement due to the presence of rings or double bond respectively and x is the number of oxygen atoms (Grewer et al., 2010). The compound classes with x=2 are defined as classical NAs whereas species with X≥3 are classified as oxidized NAs (oxy-NAs) Grewer et al., 2010; Headley et al., 2013b).

In addition, sulfur-containing species (C_nH_2n+zSO_x) and nitrogen-containing species (C_nH_2n+zNO_x) are found in OSPW (Headley et al., 2016). The classical NAs, oxy-NAs, sulfur-containing and nitrogen-containing compounds in OSPW are generally referred to as naphthenic acid fraction compounds (NAFCs) Headley et al., 2013a).

The chemical structure of NAs (carbon numbers and rings plus double bonds) control behaviours like toxicity (Yue et al., 2015a), the rate of degradation (McMartin et al., 2004; Scott et al., 2005; Biryukova et al., 2007; Smith et al., 2008) and their biotransformation (Rowland et al., 2014; West et al., 2014). In addition, environmental persistence of NAs is correlated to the presence of more ring structures and carbon chains at positions apart from β-position corresponding to the carboxylic group of the main aliphatic chain and methyl substitution on the cycloalkane rings prolong their existence in the environment than species with fewer carbon chains and rings (Scott et al., 2005; Smith et al., 2008; Han et al., 2008; Whitby, 2010; Misiti et al., 2013b, 2014). The toxicity of classical NAs is linked to species with more carbon numbers (Frank et al., 2010; Scarlett et al., 2012; Hughes et al., 2017). The toxicity and complex nature of NAFCs imply that comprehensive identification and characterization is necessary (Grewer et al., 2010).

Moreover, the diversity in the structure of NAs and NAFCs complicates knowledge of their environmental fate. These varying characteristic present threats in remediation as reliable measurements at the molecular-levels may be hindered by interferences from compounds such as fatty acids. However, application of high-resolution Orbitrap mass spectrometry (MS) analysis provides more detailed characterization of the individual NAFC species in OSPW at molecular-level and knowledge of their fate and toxicity in the environment (Headley et al., 2013a).

Because the direct disposal of OSPW to aquatic system creates environmental concerns such as toxicity, restoration in their quality prior to disposal is required by regulatory framework (Shell Canada Ltd, 2014; Hughes et al., 2017). Therefore, the treatment of OSPW prior to disposal is important as to reduce their environmental threat. Several treatment options have been available for the improvement of the quality of OSPW for the purpose of meeting the requirement for disposal into aquatic systems such as advanced oxidation processes (AOPs), Biodegradation, membrane filtration, Coagulation/flocculation and Adsorption (Quinlan and Tam, 2015; Dong et al., 2015).

The aforementioned techniques show great performance in OSPW remediation, however, the chemical and energy demanding aspects of these treatments can have relatively high cost which may limit their large-scale application (Quinlan and Tam, 2015). Presently, advancement in OSPW treatment require cost effective methods to degrade at least the environmental recalcitrant NA species (Wang et al., 2015; Johnson et al., 2011) in accordance with directives for prospective OSPW remediation. Passive techniques such as constructed wetland treatment (CWT) are considered as cheaper and yet technically viable systems for wastewater treatment without demand for anthropogenic energy inputs or active chemical with minimal operational requirements. The efficiency of constructed wetlands for the upgrade of quality of different wastewaters has been demonstrated (Kadlec and Wallace, 2009).

The appropriate CWT design for removal of contaminants is primarily driven by the interplay of essential components of plants, soil and microbes (Rodgers and Castle, 2008; Haakensen et al., 2015; Valipour and Ahn, 2016). In CWT design, flow paths include subsurface flow (SSF; i.e. water level below the soil is maintained) and free-water surface (FWS; i.e. water only flows through the soil surface). SSF is further divided into vertical surface flow (VSF; i.e. the inflow of sample is vertically from the top and through the soil to the outflow) and horizontal surface flow (HSF; i.e. the water is fed at the inflow of the system and is allowed to slowly run horizontally through the soil to the outflow) and these categories belong to sub-surface flow (SSF) CWT (Tee et al., 2012).

The flow paths of the VSF wetland design can be vertical upflow (VUF) if the water is passed from the surface or down flow (VDF) when the water is passed from beneath the system (Valipour and Ahn, 2016). The targeted contaminant for treatment, treatment goals, cost implication, geography and spatial availability are factors taken into considerations for the choice of suitable flow direction (Gessner et al., 2005). The essential design characteristics included in pilot-scale CWT allows flexibility in experimentation (Tao et al., 2006) and results can be extrapolated to advance full-scale designs (Rodgers and Castle, 2008).  The design and operational conditions (i.e. aerobic or anaerobic) of constructed wetland may affect its performance for the removal of contaminants.

Most OSPW-NAs biodegradation investigations were performed under aerobic conditions. For example, NAs were completely degraded below detection limits in aerobic bioreactor (Demeter et al., 2014). NA removal of 34.8% was achieved using membrane bioreactor (MBR) in aerobic environments (Shi et al., 2015). Despite these evidences indicating the removal of NAs under aerobic conditions, presence of electron acceptors (e.g. sulfate) other than oxygen can promote their degradation (Ramos-Padrón et al., 2011). A recent investigation explored the biodegradation of OSPW-NAs and revealed their significant removal under anaerobic conditions (Clothier and Gieg, 2016). Gunawan et al. (2014) studied the degradation of a model naphthenic acid trans-4-methyl-1- cyclohexane carboxylic acid under anaerobic conditions in continuous stirred-tank and biofilm. They showed that NA removal was twice more rapid under anaerobic conditions relative to aerobic conditions. The performance of MBR at hydraulic retention time (HRT) of 48 hours promoted 25% removal of OSPW-NAs which was facilitated by its design under both aerobic and anaerobic conditions (Xue et al., 2016).

The evaluation of the performance and direct comparison of different CWT designs for OSPW remediation have not been investigated to date. This study examined the treatment of OSPW in five different pilot-scale constructed wetland systems. The key objective of this study was to investigate whether the treatment of OSPW would differ in different of CWT designs. Specifically,  the focus were:

(i) to characterize OSPW samples from different CWT systems and assess changes in classical NAs and NAFCs composition relative to the untreated OSPW by high-resolution Orbitrap-MS;

(ii) elucidate and compare the performance efficiency of each CWT design for the treatment of OSPW by applying some metrics to evaluate the extent and rates of transformation with the intent to determine the most efficient design(s) for the transformation of NAs;

(iii) to determine if the different CWT designs can influence preferential transformation of NAs.

The result obtained from this study may be useful for future design and application CWT for OSPW treatment. Given the limited application of constructed wetlands in OSPW treatments, this research represents a significant development in assessing the impact of different constructed wetland designs on the treatment of OSPW.

2. Materials and methods

2.1. Experimental set-up of the pilot-scale systems

The study consisted of five different pilot-scale constructed wetland treatment (CWT) designs, each composed of 4 identical cells Fig.1. The designs differ mainly in terms of flow direction, substrate type, media saturation, plant type and aeration. The nomenclature of the CWT systems representative of the series 1/2, 3/4, 5/6, 7/8 and 9/10 are hereafter denoted as 2B, 4B, 6B, 8B and 10B respectively. The wetlands were constructed and operated at greenhouse facilities of Contango Strategies Ltd., Saskatoon, Canada.

The CWTs were set up in a Rubbermaid^® plastic bin with the dimensions of 56.52 cm height by 58.42 cm width by 48.26 cm depth (systems 2B, 4B, 6B and 8B) and 77.47 cm height by 57.15 cm width by 51.77 cm depth (system 10B). The OSPW was distributed by vertical upflow (VUF) in systems 2B and 4B and by horizontal surface flow (HSF) in systems 6B, 8B, and 10B. System 2B was aerated by addition of air stones; 2 stones per cell and systems 4B, 6B, 8B and 10B were operated as non-aerated systems. The depth of the gravel media (porosity 0.26) planted with sedge (Carex aquatilis) was 40 cm in system 2B and 4B. In systems 6B, 8B and 10B, the depth of the sand media was 30 cm (porosity 0.23) planted with sedge (Carex aquatilis), cattail (Typha latifolia), and bulrush (Schoenoplectus acutus) respectively.

The CWTs were operated in a closed system basis by diverting the outflow to the beginning of the series to promote extended hydraulic retention time (HRT) of 4.73 days in 2B and 4B, 5.68 days in 6B and 8B and 9.89 days in 10B at flow rate of 20 ml/min using FMI® QG 400 pumps to deliver continuous supply of OSPW. The OSPW inflow passed via each of the passive constructed wetland configurations, into the media that contains plants, to promote significant plant root effects by evapotranspiration processes (Beebe et al., 2014). Measurements of losses due to evapotranspiration have been previously reported (Ajaero et al., 2018). OSPW was added to system 2B to compensate for evapotranspirative losses. Evapotranspiration raises the concentrations of NAs, and consequently can affect wetland performance. The OSPW was supplied in July 2015 from the Muskeg River Mine external tailings facility on Shell Lease site north of Fort McMurray, currently owned and operated by Canadian Natural Resources Ltd. (CNRL) since June 1, 2017.

The OSPW was transported in 1000 L plastic tote to Contango facility in Saskatoon (SK, Canada), where it was stored at 4°C prior to the wetland treatment study. The characteristics of the untreated OSPW are provided in Table S1. Between October and November 2016, samplings were performed by collecting 500-mL of aqueous samples from each outflow cells 2B, 4B, 6B, 8B and 10B in plastic containers two times per week. The samples were stored in refrigerator at 4^oC prior to extraction and analysis. The temperature at the greenhouse facility was maintained at 22^°C from 7 am to 7 pm and 16^°C from 7 pm to 7 am throughout the experiment.

2.2. Preparation of sample for analysis

A detailed description of the extraction method can be found in (Ajaero et al., 2017). Briefly, weak anion exchange (WAX) solid phase extraction (SPE) cartridge (200 mg, 33 μm polymeric; Phenomenex (Torrance, CA, USA) were conditioned with 5 mL of methanol followed by 5 mL of Milli-QH₂O. Then 10 mL aqueous outflow sample from the treatment wetlands (pH~ 7.0) was loaded onto the cartridges. The NAFCs adsorbed to the cartridges were eluted using 5mL methanol + 5% ammonium hydroxide. The samples were dried under steady stream of nitrogen and reconstituted in 1 mL 50/50 Acetonitrile/Milli- QH₂O + 0.1% ammonium hydroxide before Orbitrap-MS analysis.

2.3. High-resolution Orbitrap-MS analysis

To characterize of NAFCs in this study, high resolution electrospray ionization Orbitrap-MS operated in the negative-ion mode was used. Surveyor MS pump (Thermo Fisher Scientific Inc.) was employed to loop inject 5 µL of OSPW extract into the mass spectrometer using mobile phase of 50:50 acetonitrile/water containing 0.1% ammonium hydroxide set at a flow rate of 200 µL/min. Dual pressure linear ion trap-Orbitrap-MS (LTQ Orbitrap Elite Thermo Fisher Scientific, San Jose, USA) was used for mass spectrometric analysis.  Mass spectra data were collected over the range of 100-600 m/z in full scan mode with a resolution of 240,000. To ensure greater mass accuracy, n-butyl benzenesulfonamide (212.07507 m/z scan-to-scan mass calibration correction) was used as lock mass compound.

The conditions of the ESI source is described elsewhere (Ajaero et al., 2018). Data acquisition, instrument operation and semi-quantitative data analysis were controlled using Xcalibur version 2.2 software (Thermo Fisher Scientific San Jose, CA). Data processing was done with Composer version 1.5.3 (Sierra Analytics, Inc. Modesto, CA) by assignment of chemical molecular formulas according to heteroatom class and double bond equivalent (DBE) of classical NAs and NAFCs. DBE denotes the number of rings plus the number of double bonds in a molecular formula.  Molecular assignments were based on the relative abundances of the classical NAs and NAFC species by accurate mass measurements (i.e. mass accuracies of <2 ppm) of the unknown compounds determined by a high-resolution negative-ion ESI Orbitrap-MS.

The quantification of the unknown NAFCs was estimated from a five-point external standard calibration of NAFCs at known concentrations as reported in past study (Armstrong et al., 2008) within the working range of the calibration curve (10-100 mg/L) using semi-quantitative analysis. NAFC analysis is based on semi-quantification methods because of absence of standard compound for quantification of individual NAFC compounds up to the present time.

The use of external standard calibration curve method for semi-quantification of NAFCs exists in literatures (Headley et al., 2013a; Huang et al., 2018). The measured concentrations by high-resolution Orbitrap-MS were adjusted for evapotranspiration as described earlier (Ajaero et al., 2018).

Analysis to determine accuracy and precision was performed during method development with replicates (n=3) of commercial standard mixture of naphthenic acid from Sigma-Aldrich (Oakville, ON, Canada). Replicate analysis (n=3) of laboratory blank (Milli-Q water) was carried out similar to the NA standard mixture.

2.4. Data analysis

To determine transformation rates of classical NAs species in the treatment systems were obtained by least square linear regression of best fit plot of natural log of the ratio of concentration: initial concentration (C/Co) versus time. The rate of transformation of NA species was described by first-order kinetics. The slope of the regression line was related to the rate constant (k) and was used to for the comparison of rate of transformation of individual carbon number and DBE.

Linear regression plots were obtained using Originpro software (Ver b9.5.195, Originlab Corporation).  All statistical analyses were performed with Originpro software (Ver b9.5.195, Originlab Corporation). Atwo-way analysis of variance (ANOVA) and Turkey pairwise comparison was used to determine significant differences (p < 0.05) in the rates of transformation of NA species in different CWT designs.

Fig.1.Schematic diagram of the pilot-scale constructed wetland treatments. The abbreviations 2B, 4B, 6B, 8B and 10B are outflows from the CWT systems representative of the series 1/2, 3/4, 5/6, 7/8 and 9/10.

3. Results and Discussions

3.1. Performance evaluation of the different wetland treatment systems

3.1.1. Abundance of NAFC classes

The high-resolution Orbitrap-MS data for classical NA and oxy-NAFCs; O₂, O₃, O₄, O₅ and O₆ species showing how the relative abundance vary in different treatment systems on the final treatment day compared to untreated OSPW is depicted in Fig.2a. The value of relative abundances of individual NAFCs during treatments relative to the untreated OSPW sample is presented in Table S2. The compound class were transformed to different extents in the different wetland treatment systems.  For O₂-NAs, the highest transformation from 56.11% abundance in the untreated sample to 26.38%, was found in 2B, followed by system 6B (29.0%) and systems 10B (40.05%), 4B (39.04%) and 8B (35.65%) respectively. This suggests that different wetland configurations may have different capacity for O₂-NAs transformation. However, transformation of O₂-NAs in all the five treatment wetland systems was substantial.

With respect to O_x-NAFCs (3 ≤ x ≤ 6), their total abundance increased in all treatment wetland system compared to untreated OSPW (Table S2). O_x-NAFCs (X≥3) are hydroxylated NAs which provides good indication of degradation of O₂-NAs (Wang et al., 2013). A plot of the relative abundance of O₂-NAs versus the corresponding sum of relative abundance of O_x-NAFCs at each time during treatment in the different wetlands produced strong negative correlation (R² ≥ 0.97) Fig. S1 and may imply that some of the parent O₂-NAs are oxidized to O_x-NAFCs during treatment in the wetlands. The 2B and 6B systems demonstrated comparable efficiency for the transformation of total O_x-NAFCs and were higher compared to other systems. As displayed in Fig.2b, increase in relative abundance of O₂S, O₄S and O₅S is noticeable in all the treatment wetland systems relative to the untreated OSPW, except in series 10B that slight decrease in relative abundance was observed for the O₅S.

In all the wetland treatments, the relative abundance of O₃S decreased. The molecular breakdown of O_xS species (x≥3) may account for the increased in the relative abundance of O₂S. Furthermore, the oxidation of O₃S to higher O_xS (x≥4) is likely the contributing factor for the observed increase in levels of the O₄S and O₅S. With respect to O₂S, the highest increase in abundance was found in 4B compared to other systems while O₄S and O₅S increased mostly in series 2B compared to other system. In terms of the O₃S, most decrease in relative abundance was observed in series 6B than the rest of the treatment wetlands.

Very low total abundance of nitrogen-containing species (≤0.38%, 0.56%, 0.4%, 0.4%, 0.5 and 1.3% in untreated sample, 2B, 4B, 6B, 8B and 10B respectively) were observed in all the treatment wetland systems and therefore not graphically considered. The data revealed that the relative abundance of NAs among the treatment wetland systems varied. However, the transformation of classical NAs and NAFCs species is based on selectivity and is not properly represented by the change in total relative abundance of the compound classes and is best described by monitoring the fate of individual components during treatment. Therefore, indices that provides better metrics for assessing the extent of transformations individual species of carbon numbers and DBE is important.

Fig. 2. Relative abundance of (a) classical NA and oxidized NAFC and b) sulfur-containing NAFC species detected by high-resolution negative ion ESI Orbitrap-MS based on single run for the untreated OSPW on day 0 (untreated OSPW) and in different wetlands on final day of treatment.

3.1.2. Distribution of classical NAs and NAFC species in CWT systems

Remediation approaches for the upgrade of OSPW quality such as CWT alters the molecular distribution of individual NA species. By monitoring the changes in distributions of classical NAs and NAFCs in the untreated OSPW and during treatment in different wetland systems enable evaluation of the effects of wetland designs on the selective transformation of individual species.  Fig. 3 shows the changes in concentration profile of O₂-NA species on the final day of treatment in the different wetland systems relative to the untreated OSPW sample.

The O₂-NAs in untreated OSPW consists of individual carbon numbers from 9-20 and DBE from 1-8 (Z=0 to −14 or 0 to 7 ring structure). The most abundant O₂-NA species before and during the treatments were the C12-16 and DBE= 3-4. Overall, significant compositional changes were observed during treatment in all the systems compared to the untreated sample. Any removal of NAs in aqueous system is as a result of the change or loss in the chemical structure of the parent NAs (Demeter et al., 2015). A visual examination of the O₂-NA profile shows that the most representative species with C12-16 and DBE=3 and 4 were remarkably transformed. However, the greatest transformation of the most representative species was observed in aerated wetland 2B, followed by the non-aerated wetland 6B systems.

It was found that O₂-NA species with C12-14 and DBE=2 were almost eliminated in all the systems during treatment. Moreover, it is worth noting the increased concentration of C10 NAs and DBE=1 compared to untreated OSPW sample in all non-aerated treatment wetlands. The presence of C10 and DBE=1 specie in in the non-aerated treatment system maybe associated with the following:

(i) conversion of higher carbon number and DBE species which may account for the increase in concentration of species with C10 NAs and DBE=1 in the non-aerated treatment wetlands. Parent NAs with more carbon number and cyclicity can be converted by microbes to species with fewer carbon number and rings (Xue et al., 2016; Zhang et al., 2016),

(ii) the species with C10 and DBE=1 might be fatty acids from biological sources (microbes and plant materials) formed at higher rate than can be converted by the non-aerated systems compared to the aerated system. It has been previously reported that O₂-NAs with carbon number in the range from 9-24 and DBE = 1 (Z = 0) are mainly fatty acid (Ross et al., 2012; Sun et al., 2017).

It could be speculated that the apparent absence of C10 NAs and DBE=1 in the aerated system is a reflection of CWT design and operational condition substantial contribution in the preferential transformation of the less carbon number and DBE compounds.

In addition, more detailed analysis of the changes in the distribution of classical NAs and NAFC species during treatment was identified using transformation efficiency (%). The percent transformation of sum of the concentrations of each O₂-NA species in the different wetland configurations based on the most abundant DBE (2-8) and carbon numbers (11-19) on the final day of treatment is illustrated in Fig S2 and Table S3.

The aerated wetland; 2B and the non-aerated wetland 6B were found to have higher transformation efficiency for O₂-NA species relative to other systems. However, the fewer carbon number NAs (C≤14) and DBE (≤3) were more effectively transformed ≥80% in 2B whereas efficiency was higher for species with more carbon number (C≥15) in 6B ≥80%.

The O₃-NAFCs and O₄-NAFCs, the most representative O_x-NAFCs were used for the evaluation of transformation of individual oxy-NAFC species in the five wetland systems (Figs. S3 and S4). C13 and C14 NAs and DBE=4 are the most dominant species in O_x-NAFC, followed by C14 and C15 NAs and DBE=5 Figs. S3a and S4a. However, the concentration of O₃-NAFC C13 NAs and DBE=4 and C14-15 NAs and DBE=5 during treatment in the aerated system and non-aerated systems decreased.

No trends specific to carbon numbers and DBE could be observed for the efficiency of transformation for O₃-NAFC species. Some O_x-NAFC species such as DBE=6 (2B), DBE=8 (4B) and DBE= 2 (6B) for O₃-NAFC and DBE=8 (2B), DBE=7 (4B) and C10-13 NAs (4B) for O₄-NAFC increased in proportion with treatment in the systems resulting in the negative transformation efficiency noted in Figs. S5. Some oxy-NAFCs are presumably intermediates produced from the biodegradation of parent O₂-NAs (Grewer et al., 2010; Sun et al., 2014; Klamerth et al., 2015), hence, the negative transformation efficiency noted for some O_x-NAFC during treatment may be due to the conversion of O₂-NAs during treatment which resulted in the increased concentration of the affected species.

The transformation of O_x-NAFCs (X = 3-4) were lower than those of O₂-NAs. The greater resistance of O_x-NAFCs has been shown in the biodegradation of OSPW-NAs (Han et al., 2009).

Another study showed inability to degrade of O₃ to be degraded by biological materials (Headley et al., 2009). On the other hand, Headley et al. (2009) noted the capability of the biological materials to remove of O₄-NAFCs amongst some persistent species. This is in very good agreement with the observation in this study for O₄-NAFCs. The changes in the concentration profile and efficiency of most O_x-NAFCs species observed in this study indicate that the all five wetland systems have potentials to transformation the O_x species.

Individual classical NAs and oxy-NAFCs analyzed during treatment in the wetlands were compared to the untreated OSPW sample. CWT caused changes in the structure of O₂-NAs and O_x-NAFCs contributing to transformation to a large extent for O₂-NAs. The more resistance of O_x-NAFC species to transformation may have little implications with respect to toxicity. Toxicity analysis of NAFC reported O₂-NAs as the primary toxicants in OSPW while the toxicity of O₃-NAFCs and O₄-NAFCs may be of little consequence (Yue et al., 2015b).

The O₂-NA species implicated in toxicity are C≥17 NAs (Hughes et al., 2017), C15-18 and DBE=4 and C14-17 and DBE=3 (Yue et al., 2015b). It is noteworthy that these species concentrations decreased during treatment in all the systems. In addition, among the NAFC species, evaluation of changes in the abundance of carbon number and DBE for O₂ species affords the most convincing evidence of the extent of biodegradation of OSPW-NAs (Huang et al., 2015). Therefore, transformation of O₂-NAs species at molecular level is of high interest and reminder of this paper will be focused on these species.

Fig. 3. Concentration profile of O₂-NA species using Orbitrap-MS: a) untreated OSPW; b) aerated wetland 2B and non-aerated wetlands: c) 4B; d) 6B; e) 8B and f)10B on final day of treatment. Carbon number is denoted as n.

3.2. Impact of CWT design on transformation of O₂-NA species

The changes in composition which examines the extent of transformation of individual O₂-NA species can be investigated by the application of different indices. The application of different evaluation indices is necessary in the assessment of the progress of biodegradation of contaminants and treatment performances (Tembhekar et al., 2015). Application of different indices for assessment of NA species transformation and the kinetics may provide more useful information on the efficiency of the remediation approach and preferential transformation of NA species in each treatment system.

To better understand the extent of transformation of O₂-NA species based on individual carbon number and DBE in the different wetland systems, detailed focus on the ratio of NA_∑no/NA_∑ni was used for individual carbon number; likewise ratio of NA_∑DBEo/NA_∑DBEi was used for individual DBE. The calculations were based on the concentrations of the O₂-NA species, where NA_∑no and NA_∑ni are the total concentration of O₂-NAs with the same carbon number in the inflow on day zero (untreated OSPW) and outflow samples on the final day of treatment respectively, NA_∑DBEo and NA_∑DBEi, are the total concentration of NAs with the same DBE in the inflow and outflow samples respectively.

To further reveal how selective transformation of O₂-NAs at molecular level is influenced by carbon number and DBE, the ratio of NA_no/NA_ni for each DBE and NA_DBEo/NA_DBEi for each carbon number respectively generated can be seen in Figs.5 and 6. NA_no and NA_DBEo ratio indicates the concentration of NAs with certain carbon numbers and DBE values respectively for samples on the final day of treatment and NA_ni and NA_DBEi indicates the concentration of NAs with certain carbon numbers and DBE values respectively for untreated OSPW on day zero.  Lower ratio (<1.0) is an indication of effective degradation of NA species in the treatment system (Huang et al., 2017; Xue et al., 2017).

Figs. 4a demonstrate the selectivity of different systems based on NA_∑no/NA_∑ni and NA_no/NA_ni ratios over the range of carbon numbers 11-19. Trends in carbon number and DBE transformation were similar in all the non-aerated treatment systems (4B-10B) and may indicate that the non-aerated systems share similar microbial communities but the extent of transformation differed markedly. Transformation trend of O₂-NA species was different in the aerated system (2B).

Overall, the aerated treatments system 2B and non-aerated system 6B has the highest transformation based on carbon numbers compared to other treatment systems (4B, 8B, and 10B). The species with 12≤C≤14 were better transformed in the aerated system; 2B compared to other species. This result is consistent with previously study under aerobic conditions that reported more favorable degradation of NA species with fewer carbon numbers (Clemente et al., 2004; Han et al., 2009).

As shown in Fig. 5, the 2B system showed higher preference for the transformation of species with fewer carbon numbers within DBE≤4 than other systems. For non-aerated system 6B species with 15≤C≤19 within DBE≥7 were more effectively transformed relative to other species (Fig. 4a). Reports also exist on the more favorable degradation of longer chain carbon compounds in crude oil under anaerobic conditions (Bekins et al., 2005; Hostettler et al., 2007). The non-aerated (6B) preferentially transformed of species with more carbon number (C≥17) within DBE≥7 compared to other systems Fig. 5. Contradictory to past studies that reported great recalcitrance to or insignificant influence of carbon number on the biodegradation of classical NAs (Clemente et al., 2004; Han et al., 2008; Wang et al., 2013).

Additionally, the hydrophobic character of NAs increases with more carbon numbers (Zubot et al., 2012). The increased hydrophobicity of NAs may result in aggregation in the wetland system. Aggregation can raise the levels of NAs thus making them more readily available for microbial degradation (Misiti et al., 2013b). However, it is suggested that specialized bacterial cells such as those that possess hydrophobic cell surface have potentials to directly assimilate aggregated hydrophobic pollutants (Peltola, 2008). Hence, organic pollutants that have high hydrophobicity (or higher lipophilicity) have easy sorption on the outside or internal surface within the bacterial cells that can consequently enhanced biodegradation under certain circumstances (Parson and Govers, 1990).

It was also found that systems 4B, 8B, and 10B favorably transformed O₂-NAs with long carbon chain although this trend was not appreciable in the system 4B which may be due to simultaneous transformation or production emanating from microbial conversion of more carbon number NAs. On close examination of each DBE (Figs.4a and 5), it was found that systems 2B and 6B have almost similar extent of transformation for C15-16 NAs within DBE=5-6.

Similarly, for NA_∑DBEo/NA_∑DBEi ratio, transformation was highest in the aerated treatments systems 2B and non-aerated system 6B relative to the other treatment systems Fig. 4b.  However, species with fewer DBE≤4 were more effectively transformed in 2B than in other systems. This observation agrees with earlier findings that biodegradation of NAs with larger DBE sizes (corresponding to higher cyclicity and double bonds) are less obtainable in commercial and OSPW NAs (Han et al., 2008).

The primary reason for better degradation of NAs with fewer DBE is because they are less hydrophobic and more soluble (Jones et al., 20111; Zubot et al., 2012), making them more readily available which result in more preferential transformation. While greater recalcitrance of NAs with more DBE is due to the fact that presence of rings forestalls the tertiary carbon to appear at α or β positions and thus prohibits β-oxidation which is the primary mechanism preferred for microbial degradation (Han et al., 2008).

By contrast, species with DBE≥7 were more preferentially transformed in 6B compared to the other wetland systems. Within species with C≤14, fewer DBE≤4 resulted in greater efficiency of transformation in 2B while within C≥17 species, enhanced transformation of DBE≥6 was observed in 6B Fig. 6. This observation is interesting compared to previous studies which showed that the presence of more rings and double bonds impede the degradation of NAs (Han et al., 2008). The ratios indicate that extensive degradation was achieved in all the treatment systems. Thus, the ratios suggest a pattern of preferential transformation of O₂-NA species in the aerated versus non-aerated treatment systems.

It can be concluded that the aerated system 2B is more suited for the transformation species with fewer carbon numbers and DBE whereas non-aeration is the preferred condition species with more carbon numbers and DBE. Generally, it is impressive that the transformation of the more recalcitrant NAs species with more carbon numbers and DBE was observed in all treatment systems. Moreover, the ratios of the transformation of O₂-NA species were more indicative of and describe the pattern of transformation in the different treatment wetland designs. Thus, they are useful indices for the assessment of trend and extent of NA species transformation.

Fig.4.Relative transformation of O₂-NAs in different wetland treatment designs based on (a) carbon number (NA_∑no/NA_∑ni) and (b) DBE (NA_∑DBEo/NA_∑DBEi).

Fig. 5.The relative transformation of O₂-NAs with different carbon numbers (NA_ni/NA_no) at each DBE in different wetland designs on day zero and final day of treatment.

Fig. 6. The relative transformation of O₂-NAs with different DBE (NA_DBEi/NA_DBEo) at each carbon number in different wetland designs on day zero and final day of treatment.

3.3. Kinetics of O₂-NA transformation

The estimation of removal rates constant (k) and half-lives (t_1/2) of contaminants are key parameters for detailed understanding of remediation timespan (Liu et al., 2016) and for the description of their environmental fate. The appropriateness of the kinetic model that best describe the rate of transformation of NA species in our treatment wetlands based on the highest value of the coefficient of determination (R²) was reported previously (Ajaero et al., 2018). The first-order kinetic model demonstrated excellent fit to the data for O₂-NA species in all the wetland systems. A strong linear relationship is acknowledged when the absolute value of the coefficient of determination is more than 0.9 (Bewick et al., 2004).

The values of R² for the five treatment wetlands studied are listed in Tables S4. This finding is anticipated given the fact that first-order kinetics has been used to best describe the biodegradation of NA species in CWT in earlier studies (Toor et al., 2013; McQueen et al., 2016; Ajaero et al., 2018). The determination of the concentration of O₂-NA species for a given period of time for individual components during treatment in the wetlands was done according to the procedure reported elsewhere (Toor et al., 2013; Ajaero et al., 2018). The rates of transformation were determined for the most representative NA species with carbon number 11-19 and DBE 2-8. The values of the associated kinetic parameters such as rate constant (k) and half-life (t_1/2) of classical NA species are given in Tables S5 and S6.  Significantly different slopes (p < 0.05) within each system are indicated in Table S7. The transformation of individual carbon number and DBE over time is illustrated in Figs. 7 and 8. Generally, the rate of transformation of all species increased with time in the five treatment systems.

The kinetic data showed that the rate of transformation was proportional to the size of the carbon numbers in the non-aerated treatment wetlands; 6B, 8B and 10B (Fig.7). This is supported by our past study on the fate and behavior of NAs in non-aerated CWT that showed more rapid transformation of NAs with more carbon number in non-aerated treatment wetland (Ajaero et al., 2018). O₂-NAs with C≥15 had the highest rate of transformation among the species. Moreover, 6B had the most rapid transformation rates for species with more carbon numbers (C≥15) than other systems. However, in system 4B, the rate of transformation was slowest and comparable for all range of carbon numbers.

In comparison the aerated treatment wetland (2B) had the most rapid transformation for species with fewer carbon numbers (C≤14) among the species compared to other treatment systems. Transformation rates were also monitored for individual DBE (Fig.8). The influence of DBE on the transformation rates of classical NAs showed different trends in the different wetland systems. The rate of transformation was indirectly proportional to the size of the DBE, in which the rate was fastest for species with fewer DBE than those with more DBE in the aerated treatment wetland (2B). This finding is consistent with the pattern observed for the degradation of classical NAs where the rate of biodegradation of species with fewer rings and double bonds was faster than species with a greater number of carbons under aerobic conditions (Smith et al., 2008). It has been previously suggested that the molecular structure controls the extent and rate of biodegradation of NAs in aerobic environment (Misiti et al., 2013a).

For the non-aerated wetlands, systems 4B and 8B showed fastest rate of transformation for DBE≤4 relative to other DBE species within the two systems while the rate for the species with DBE≥5 were almost similar in 4b and 8B. Conversely, no obvious regularity was observed for the rates of transformation of DBE in the system 6B and therefore, the DBE showed less influence on the rate of transformation as much as it is affected by carbon number in this system. However, the fastest rate of transformation was observed for DBE=5 and 8. No distinctive effect of DBE on the rates of transformation of NA species was observed in the treatment wetland system10B. The half-life of in-situ biodegradation of OSPW-NAs in tailing ponds is estimated to take over a decade (i.e.12.8-13.6 years (Han et al. 2009).

Similarly, a degradation rate of NAs by microbial consortium with half-life of 203 days was reported (Mahdavi et al., 2015) and half-life between 44-240 days in biological system (Han et al., 2008). Interestingly; the CWT systems facilitated the transformation of NA species with the longest half-life of 34 days (DBE=8) observed in 4B system at the same time the shortest half-life of 7 days (DBE=2) observed in 2B system. Therefore, the accomplishment of the systems most especially the 2B and 6B in this study underscores the potential for CWT approach to expedite OSPW-NA remediation.

Based on kinetic data for this study, the rate of transformation of carbon numbers and DBE species suggest a pattern of preferential removal of NAs in the CWT systems.  The possible reason for the differences noted in the preferential transformation of NAs among the CWT systems studied may be attributed to the microbial communities present in the different designs. It is established that the preference for the degradation of OSPW-NAs is related to the microbial communities present. The factors influencing the degradation of OSPW-NA s are not limited to structural effects with respect to carbon number and DBE, also the preference of the different microorganisms for different NA species is significant (Del Rio et al., 2006). Though the characterization of microbial communities present in each CWT system is not within the scope of this study, the existences of different microbial populations under aerobic and anaerobic conditions for treatment of OSPW was previously shown (VanMensel et al., 2017).

Furthermore, the dynamics of the microbial communities in the pilot-scale CWTs might be due to influence from the constructed wetland designs. It has been shown that pilot-scale constructed wetland design is a major factor affecting the type of microbial communities (Button et al., 2015; Lv et al., 2017). For example, significantly different microbial communities were found in vertical flow and horizontal flow constructed wetlands used for the treatment of domestic wastewater (Button et al., 2015). Different designs of constructed wetland such as vertical and horizontal flow systems with different levels of oxygen and HRT are expected to support different microbial communities (Lv et al., 2017).

In summary, aerated wetland system; 2B and non-aerated wetland system; 6B demonstrated optimum treatment performance compared to the other systems. For example, highest transformation rate of NAs with fewer carbon number and DBE were recorded in wetland system 2B whereas highest rates of species with more carbon number and DBE were recorded for 6B. DBE has the most the most significant effects on NA transformation in system 2B while carbon number has the most significant influences on the transformation of NAs in system 6B than in other systems. This provided insight on the impact of constructed wetland designs on the preferential transformation of NA species. Moreover, the non-aerated system (in the absence of aeration) involves low energy requirement.

The results highlight the potential for improved efficiency of OSPW treatment through combination of systems operated under conditions of 2B and 6B systems and the advantage of cheaper cost of non-aeration in case of large-scale application to reduce costs.

It is known that the structure of NAs is based on carbon number and DBE influences its toxicity (Rowland et al., 2011). It is expected that the transformation of NAs in the CWT systems will correlate to reduced toxicity. The reduction in toxicity of NAs by the CWT systems will be communicated elsewhere.

Fig. 7. Transformation performance of the five CWT designs for individual carbon numbers: a) aerated wetland 2B and non-aerated wetlands: b) 4B, c) 6B, d) 8B and 10B.

Fig. 8. Transformation performance of the five CWT designs for individual DBE: a) aerated wetland 2B and non-aerated wetlands: b) 4B, c) 6B, d) 8B and 10B.

4. Conclusions

This study investigated the treatment of OSPW-NAs by five different CWT designs and the extent and rates of transformation were monitored. Both aerated and non-aerated treatment wetlands were effective in transformation of classical NAs and NAFCs. Preferential transformation of NA species was found in different CWT design. The results indicated that the aerated treatment wetland (2B) and the non-aerated treatment wetland (6B) were most successful in transformation on NA species.

Aerated wetland system was most favorable for transformation of fewer carbon number and DBE species whereas more carbon number and DBE species were most preferentially transformed in non-aerated system. Result further indicated that the various designs were likely responsible for the preferential transformation of NA species in the constructed wetland systems.

Based on the general results of the treatment performance and the low cost of non-aeration, if operated under synergistic conditions and designs of the aerated system and the non-aerated system (i.e. 2B and 6B) can be considered technically feasible for more effective treatment of OSPW.

Conflict of interest

None.

Acknowledgements

This work was supported by funding from the Program of Energy Research and Development (PERD). We gratefully acknowledge the University of Regina for graduate student funding. Also, we sincerely appreciate the contribution of Rachel Martz in the routine sampling.