Oil sand pyrolysis: Evolution of volatiles and contributions from mineral, bitumen, maltene, and SARA fractions

Fuel 224 (2018) 726–739

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Oil sand pyrolysis: Evolution of volatiles and contributions from mineral, bitumen, maltene, and SARA fractions

T

Fan Niea, Demin Hea, Jun Guana, Xueqiang Lic, Yu Honga, Linfei Wanga, Huaan Zhengc, ⁎ Qiumin Zhanga,b, a

Institute of Coal Chemical Engineering, School of Chemical Engineering, Dalian University of Technology, No. 2 Linggong Road, Dalian 116024, PR China State Key Laboratory of Fine Chemicals, Dalian University of Technology, No. 2 Linggong Road, Dalian 116024, PR China c Shanxi Coal and Chemical Industry Group Co. Ltd, Block B City Gate Building, 1 Jinye Road, High-Tech Zone, Xi’an City, PR China b

G R A P H I C A L A B S T R A C T

A R T I C LE I N FO

A B S T R A C T

Keywords: Oil sand Pyrolysis SARA Volatile evolution

In order to investigate the pyrolysis behavior through the volatile evolution, pyrolyzer-evolved gas analysis-mass spectrometry and thermogravimetric analysis-mass spectrometry were adopted for in situ analysis and comparison the pyrolytic volatiles derived from two different oil sands as well as their mineral, bitumen, maltene and SARA fractions. For both oil sands, the weight loss and volatile release during the oil-producing process exhibited two stages: a devolatilization stage (< 350 °C) and a thermal-cracking stage (350–600 °C). These two stages afforded different volatile yields and displayed distinct activation energy distributions. Additionally, heart-cut analysis indicated that the volatiles were in disparate compositions at each stage. The volatile evolution of the oil sand bitumen, maltene, and SARA fractions revealed individual contributions on volatile release with increasing pyrolysis temperature. Specifically, volatiles in the devolatilization stage predominantly originated from the saturates in maltene. These resulted in a significant amount of polycyclic biomarkers and contributed toward the release of higher-molecular-weight substances over a temperature range of 200–350 °C. The thermal-cracking stage was the main stage in which most of the gaseous and light pyrolytic products, including amounts of alkanes and olefins, were generated. These compounds mostly originated from cracking of resin and asphaltenes. The aromatic fraction was observed that released volatiles in both stages. Thus, due to the disparity in temperature region for volatile release, the volatile yields during the different stages of the process were mainly determined from the organic constituents of the oil sand. Notably, the volatile compositions predominantly correlated to the original organic structures. Moreover, the minerals exhibited little influence during

⁎

Corresponding author at: Institute of Coal Chemical Engineering, School of Chemical Engineering, Dalian University of Technology, No. 2 Linggong Road, Dalian 116024, PR China. E-mail address: [email protected] (Q. Zhang).

https://doi.org/10.1016/j.fuel.2018.03.069 Received 29 November 2017; Received in revised form 7 March 2018; Accepted 11 March 2018 0016-2361/ © 2018 Elsevier Ltd. All rights reserved.

Fuel 224 (2018) 726–739

F. Nie et al.

the oil-producing stage of the two oil sand samples under the tested heating conditions. However, the presence of interactions between these organic sub-fractions during oil sand pyrolysis is suggested.

1. Introduction

especially aimed at gaseous products. Zhao et al. [9] investigated the behavior of evolved volatiles during pyrolysis of an Alberta oil sand by thermogravimetric mass spectrometry (TG-MS). Through the evolution curves of specific mass fragments, they discovered that the release of organic compounds such as fragments of aliphatic hydrocarbons appeared as two peaks at 130 °C and 480 °C, respectively. On the other hand, inorganic compounds such as H2, H2S, COS, and SO2 were observed from ∼380 °C. Similar results were reported by Hao et al. [10] who studied the evolving tendency of several typical gaseous products with varying temperature by thermogravimetric analysis coupled with Fourier transform infrared spectroscopy (TG-FTIR). However, the gaseous products were only a component of the released volatiles and most gaseous products originated from intense decomposition reactions, especially during the thermal-cracking stage. Thus, simply tracing these gaseous products probably leads to a loss of information during the devolatilization stage. Volatile generation during pyrolysis is rather complicated due to the complex and varied compositions and structures of the organics present in oil sand. However, the pre-separation of the comprehensive organic mixture and the subsequent study of the individual species is beneficial toward understanding the whole mixture. According to the differences in polarity and aromaticity, oil sand bitumen can be separated into saturate, aromatic, resin, and asphaltane fractions (SARA fractions) [11]. To investigate the thermal cracking behavior of each fraction and their contributions toward volatile generation during oil sand pyrolysis, Hao et al. [12] utilized a TG-FTIR technique for the gas-releasing online-analysis of the SARA fractions. They reported that the release behavior of CO2, CO, methane, ethylene, and light aromatics was different due to the structural differences of these species. The activation energy of each SARA fraction, representing the degree of difficulty of each pyrolytic reaction, was also variant [13,14]. However, weight-loss analysis has revealed that the SARA fractions displayed fewer interactions during co-pyrolysis than in their independent pyrolysis [15]. To date, several coupling techniques have been widely used for in situ evolved gas analysis (EGA). Among them, TG-coupled methods are the most popular as they require a short analysis time and offer quantitative information on weight loss. Although ∼10 mg of sample is usually required for TG measurements because of the sensitivity of the balance, a smaller sample mass is preferable to avoid volatile condensation in the transfer system between the TG analyzer and the

In 2016, the world’s primary energy consumption was equivalent to 13276.3 million tons of oil. Moreover, due to the projected economic development and increase in productivity, mostly related to Africa and South and East Asia, energy consumption is predicted to increase over the next decades. In fact, according to world’s energy outlook, fossil fuels will remain the dominant sources of energy powering the world economy, accounting for more than three-quarters of the total energy supply by 2035 [1,2]. Consequently, to mitigate the resultant environmental and energy security issues, there is an ever-growing demand for the production of clean diversified fuel supplies and the technological development of and research on the conversion process of fossil resources. Oil sand is an unconventional petroleum resource with abundant reserves that is attracting much attention as an alternative energy supplement. Generally, oil sand is a natural mixture of clays (minerals), bitumen (organics), and a small amount of water. The critical step during its utilization is bitumen recovery, with related methods including pyrolysis, hot water extraction, organic solvent extraction, and some in situ methods. Among these techniques, pyrolysis is considered as a promising method, especially for oil-wet types in which bitumen is less efficiently recovered by hot-water extraction [3]. Moreover, pyrolysis is also an important method for the production of high valueadded light oils from oil sand [4]. Pyrolysis of oil sand is a thermal chemical process during which oil sand particles generate volatiles through desorption of the dissociative and thermal-cracking of chemical bonds. The separation of organics and minerals is finally accomplished by exportation and condensation of these volatiles. Therefore, the temperature greatly influences the degree of devolatilization, yield and quality of the pyrolytic products. Recently, several studies have been reported on the evolved volatiles during oil sand pyrolysis through indirect or direct methods [5]. Studies through thermogravimetric analysis revealed the presence of two relatively distinct weight-loss stages during the process [6,7]: devolatilization of light organics at temperatures < 350 °C and carbonaceous decomposition of heavy organic macromolecules at temperatures ranging between 350 and 600 °C. Simultaneously, the distinct Arrhenius parameters of both stages were also determined according to the firstorder kinetic model [8]. Other studies involved in situ analysis,

Oil Sand

Extraction by n-Hexane

Extraction by toluene

Bitumen

Minerals

Maltene

Residues

Column chromatography

n-Hexane

Toluene

Saturate

Extraction by toluene

Toluene+Alcohol (Volume ratio=1)

Aromatic

Resin

Asphaltene

Fig. 1. Separation flow diagram of each fraction of oil sand.

727

Minerals

Fuel 224 (2018) 726–739

F. Nie et al.

sample (ground to < 160 mesh) was placed in a ceramic crucible and heated from ambient temperature to 650 °C at a constant heating rate. Pure argon was employed as the carrier gas at a constant flow rate of 60 mL/min. The generated volatiles passed into the MS through a quartz capillary (< 1000 mbar). The MS scan range was 1–50 amu at a scan rate of 200 ms/amu.

detector; this prevents the condensates from inducing further memory effects in subsequent analyses [16]. EGA via a temperature programmable pyrolyzer coupled with mass spectrometry (Py-EGA-MS) is another promising technique. In this method, the sensitivity of the MS detector allows for a smaller weight of sample. Additionally, this method provides EGA thermograms with wider mass ranges. Recently, Py-EGA-MS has been widely employed to study the evolved volatiles during the pyrolysis and oxidation of polymers [17,18]. However, to the best of our knowledge, few Py-EGA-MS studies on oil sand and bitumen have been reported to date. The general goals of this study are to reveal the pyrolysis behavior of oil sand from a volatile evolution perspective. Therefore, the evolution behavior of pyrolytic volatiles from two different oil sands were investigated and compared by in situ EGA techniques. For better understanding, the individual behavior and contributions toward volatile release of the oil sand bitumen, maltene, and SARA fractions were also analyzed and compared. Additionally, based on the volatile evolution profiles, heart-cut analysis (HCA) was employed to identify the distribution of pyrolytic products in volatiles over temperature ranges corresponding to each pyrolysis stage.

2.3. Py-EGA-MS analysis Py-EGA-MS analysis was carried out on an EGA/PY-3030D multishot pyrolyzer (Frontier Laboratories Ltd., Japan) coupled with a TRACE ULTRA-ISQ GC/MS (Thermo Scientific, USA) system. During analysis, about 0.8 mg samples were placed in a sample cup and then moved into the pyrolyzer with a sampler tool. The sample was then moved into a free-fall pyrolyzer furnace and heated from 40 °C to 650 °C at a rate of 10 °C/min. The interface temperature was auto-controlled up to a temperature of 300 °C to avoid condensations of the volatiles prior to the sample entering the GC system. Pure helium was employed as the carrier gas at a flow rate of 1.0 mL/min. The volatiles were split at the GC inlet (split ratio 30:1) and subsequently passed through a stainless steel capillary EGA column (Ultra alloy-DTM, 2.5 m × 0.15 mm(I.D.) into the MS. The MS detector comprised an electron ionization (EI) source (70 eV) with the full scan mode ranging from 33 to 550 amu. The temperature of the GC inlet, oven, MS transfer line, and EI source was maintained at 300 °C.

2. Experimental and equipment 2.1. Oil sand samples The two oil sand samples were collected from Indonesia (IO) and the Philippines (PO). Proximate analyses, except for moisture content, were performed according to the Chinese standard GB/T 212-2008. Furthermore, the carbonate contents were determined from the amount of CO2 release (Chinese standard GB/T 218-1996) to elucidate the inorganic matter in the volatile content. Moisture, bitumen, and the SARA fractions were separated by Soxhlet extraction and column chromatography [11]. The separation flow diagram is illustrated in Fig. 1. Ultimate analysis was conducted with an element analyzer (vario EL III, Elementar, Germany). Additionally, the mineral components of each oil sand sample were analyzed by X-ray fluorescence (SRS3400, Bruker, Germany) and X-ray diffraction (Rigaku SmartLab, USA) spectroscopy. The two oil sand samples displayed good utilization values (> 10 wt% bitumen content) but different inorganic and mineral compositions (Table 1 and Fig. 2). Specifically, the PO sample contains more maltene, while less asphaltene is present in its bitumen. On the other hand, the minerals in the IO sample are predominant and abundant in calcium carbonates. For the organic part of the oil sand samples, the number-average molecular weights of the bitumen in the IO and PO samples were determined as 1179 Da and 949 Da, respectively, by gel permeation chromatography. One-dimensional 13C NMR spectrometry (Bruker AVANCE III 500 MHz spectrometer) was subsequently employed to further investigate the structural features of the oil sand bitumens as follows: ∼90 mg bitumen was dissolved in 1 mL deuterated chloroform (CDCl3); the spectrum was collected using the zgpg30 pulse program; a relaxation delay (D1) of 1 and 12,800 scans were employed to sufficiently improve the signal-to-noise ratio (S/N). The 13C NMR spectra were resolved according to the literature [19]. The afforded structural parameters (Table 2) revealed that both samples consist of numerous methylene groups. However, the IO sample comprises more aromatic carbons (aromaticity = 0.38) than the PO sample (aromaticity = 0.19). Thus, the average molecule of the IO bitumen has larger aromatic cores. Moreover, less but relatively longer aliphatic chains were observed in the bitumen of the PO sample.

Table 1 Basic properties of the oil sand samples. Properties

IO

PO

Proximate analysis (wt%, dry basis) Ash Volatiles CO2 (for carbonates) Fixed carbon*

52.78 13.49 25.69 8.04

75.38 12.13 11.63 0.86

Ultimate analysis (wt%, dry-ash-free basis) C H N S O*

75.13 7.09 0.66 6.12 11.03

80.23 9.60 0.47 0.37 9.31

Oil sand components (wt%, as received) Bitumena Malteneb Asphaltenec Moisture Solid dreg

19.33 13.72 5.61 2.38 78.29

12.23 9.75 2.48 0.86 86.91

Maltene components (wt%) Saturate Aromatic Resin

12.29 42.86 44.85

21.56 29.83 48.61

2.2. TG-MS analysis

Mineral components (%) SiO2 CaO Al2O3 Fe2O3 SO3 MgO K2O Na2O SrO P2O5 TiO2 Cl BaO ZrO

11.50 76.30 3.97 2.92 1.95 1.80 0.46 0.35 0.27 0.10 0.24 – 0.14 –

31.37 46.44 7.95 3.67 0.62 3.04 3.21 2.70 0.44 0.16 0.20 0.13 – 0.07

TG-MS analysis was performed with a Mettler Toledo TGA/ SDTA851e thermogravimetry analyzer coupled with a quadrupole mass spectrometer (GSD-301 T3, MS). During analysis, ∼15 mg of the

* By difference. a Toluene soluble. b n-heptane soluble. c n-heptane insoluble.

728

Fuel 224 (2018) 726–739

F. Nie et al.

The mass ratio of the organics was ∼0.25. Finally, the CH2Cl2 in the mixed solution was removed by rotary evaporation and the afforded residues were ground and stored in a dryer. The actual amount of loaded organics was determined as the percentage weight loss after calcining at 800 °C for 3 h. 2.4. Heart-cut analysis While EGA investigates the response of a sample towards thermal stress by analyzing the temperature-resolved overall release of volatiles without chromatographic separation (short deactivated column), HCA provides the compositions of the volatiles at individual temperature ranges based on EGA data [18,20]. During HCA, the volatiles within a set temperature range were trapped by cooled pure nitrogen (∼−190 °C) in the fore-part of the column. In this study, ∼2 mg of the sample was pyrolyzed non-isothermally at a heating rate of 10 °C/min; the pyrolysis thermal regions were divided according to the thermograms obtained by Py-EGA-MS. The GC/MS was equipped with a weak polarity stainless steel capillary column (Ultra alloy + −5, 30.0 m × 0.25 mm × 0.25 μm, Frontier Laboratories Ltd., Japan) for separation. The initial oven temperature was set at 40 °C. After trapping, it was increased to 200 °C at a rate of 4 °C/min and maintained at 200 °C for 5 min. Finally, the oven temperature was programmed to increase from 200 °C to 300 °C at 3 °C/min; this final temperature was held for 15 min. The ion source and transfer line temperatures were kept at 300 °C. Components were detected by MS with a full scan mode ranging from 40 to 400 amu. Other settings were similar to those employed for Py-EGA-MS analysis. Chromatographic peaks were identified using data from the NIST mass spectral data library and semi-quantified via the percentage peak areas. All experiments were performed multiple times to ensure the reproducibility of the experimental data. 3. Results and discussion 3.1. Thermogravimetric analysis and pyrolytic kinetics Fig. 3 presents the TG and DTG curves of both oil sand samples. The total weight losses of the IO and PO samples at 650 °C were determined as 14.2 wt% and 12.8 wt%, respectively. Similar to results reported in literature, the weight loss processes within the temperature range could be divided into three stages [6,21]. The first stage (< 150 °C) comprised moisture evaporation. However, the weight loss in the PO sample at this stage was not obvious for the lower moisture content in the sample. The second stage was the devolatilization stage whereby light organics were desorbed at a temperature ranging from 150 °C to 350 °C.

Fig. 2. X-ray diffraction (XRD) patterns of raw minerals in oil sand.

To facilitate weighing and sampling, the extracted bitumen, maltene, and SARA fractions were loaded onto deactivated silica gel prior to EGA. The extraction and separation processes are illustrated in the flow sheet in Fig. 1. The extracted fractions were next dissolved in CH2Cl2 and silica gel (80–100 mesh), which was calcined at 800 °C for 3 h for the deactivation and desorption of moisture and other organics.

Table 2 Structural parameters of bitumen derived from the oil sand samples. Chemical shift (ppm)

Symbol

Description (mol%)

IO

PO

55.00…10.00 18.60…10.00 23.00…18.60 32.50…23.00 34.60…32.50 42.70…34.60 55.00…42.70 145.00…115.00 129.30…115.00 137.10…129.30

Cal CH3 CH3α CH2 CH CH2α CHα Car CHar Cqp

Aliphatic carbon Terminal methyl groups in aliphatic chain Methyl groups branched in α or β position from an aromatic or naphthenic ring Methylene groups of alkyl chains Methine groups in aliphatic chains Methylene groups branched in α or β position from an aromatic ring Methine groups of naphthenic rings Aromatic carbon Aromatic protonated carbon atoms Aromatic bridgehead quaternary carbon atoms

62.10 9.92 7.70 22.61 3.29 10.18 8.40 37.90 11.53 17.97

81.20 11.46 14.51 41.31 4.54 9.42 0.10 18.80 8.78 6.06

145.00…137.10

Cqs

Aromatic substituted quaternary carbon atoms

8.39

3.81

Symbol

Equation

Description (per 100 carbon)

IO

PO 3.81

Rs

Cqs

Number of alkyl substituents

8.39

n

Cal/ Cqs

Number of carbon atoms per alkyl substituents

7.40

21.29

Ra Canb

(Car −1)/4 CHar + Cqs

Number of aromatic rings Number of non-bridge aromatic ring carbon atoms

9.22 19.92

4.45 12.59

729

Fuel 224 (2018) 726–739

F. Nie et al.

Fig. 3. Thermogravimetric (TG) and differential thermogravimetric (DTG) curves of the raw oil sand samples (10 °C/min).

oil-producing stage, afford most of the organic pyrolytic volatiles and are therefore critical for bitumen recovery. To better clarify the oilproducing stage, the DTG curves were deconvoluted by the Bi-Gaussian function as follows [6]:

Table 3 Fitting parameters of the differential thermogravimetric (DTG) curves (150–550 °C). Samples

IO PO

Temperature (°C)

Reliability parameters

Peak 1

Peak 2

χ2

R2

442.6 455.0

329.5 321.2

3.717 × 10−8 4.628 × 10−8

0.9873 0.9966

( (

2

) )

⎧ Hexp − (x − x c ) , x < x c 2w12 ⎪ f (x ) = ⎨ (x − x c )2 ⎪ Hexp − 2w22 , x ⩾ x c ⎩

In this stage, an evident peak (∼320 °C) was observed in the DTG curves of the PO sample, while only a sub-peak was found in the IO region. The third stage (350–550 °C) comprised the thermal-cracking of heavy organics where chemical bonds break down under severer thermal stress. For both oil sand samples, the third stage contributed toward the greatest weight loss and weight loss rates, accordingly presenting higher and shaper peaks on the DTG curves. In general, the second and third stages, collectively known as the

(1)

where H is the peak height, x c represents the position of the peak center, and w is the modulus at full width half maximum. According to the R2 and χ2 values in Table 3, the simulated DTG curves exhibit a moderate goodness-of-fit with the raw curves. For each stage, the temperature corresponding to the peak center of both oil sands approached 325 °C for the devolatilization stage and 443 °C for the thermal-cracking stage. Moreover, the initial temperature of the thermal-cracking stage was ∼350 °C, while thermal cracking prevailed 730

Fuel 224 (2018) 726–739

F. Nie et al.

Associated with the peaks on the DTG curve, the shift points on the activation energy distribution curves approximately represent the boundary between the devolatilization and thermal-cracking stages. Therefore, the two processes also exhibit disparate activation energy distributions according to the distributed activation energy model.

after the temperature reached 400 °C (Fig. 4). To further distinguish the two processes during the oil-producing stage, the distributed activation energy model (DAEM) [22] was employed to determine the variation in apparent activation energy with the degree of pyrolysis. Related figures from these calculations are listed in the Supporting information. Fig. 5 presents the activation energy at each conversion ratio (w/w0) at a temperature range of 150–600 °C. During the oil-producing stage, the activation energy of the PO sample was higher than that of the IO sample. This indicates that the pyrolytic reactions occur less easily in the former sample. Moreover, the plots of activation energy versus increasing w/w0 ratios reveal that each curve exhibits an marked shift point (Fig. 5). For the IO sample, the activation energy begins to increase monotonically at w/w0 = 0.25. This corresponds to a temperature of 360 °C at a heating rate of 10 °C/ min. Due to higher conversions during the lower temperature period, the shift point for the PO sample was observed at w/w0 = 0.5, corresponding to a temperature of ∼380 °C at a heating rate of 10 °C/min.

3.2. Evolved gas analysis by TG-MS Oil sand pyrolysis affords volatile species of varying molecular weights and structures. The test scan range (1–50 amu) was selected to target the gaseous products. Fig. 6 presents the evolution curves of typical gaseous products including H2 (2 amu), CH4 (16 amu), CO (28 amu), and CO2 (44 amu) at a heating rate of 10 °C/min. H2 is a key gaseous product during the pyrolysis of hydrocarbondominated substances. It is mainly afforded from the dehydrogenation of hydroaromatic structures and condensation of aromatic cores [23,24]. Fig. 6 illustrates that the release of H2 begins when the

Fig. 4. Fitting curves from differential thermogravimetric (DTG) analysis at 150–550 °C.

731

Fuel 224 (2018) 726–739

F. Nie et al.

However, a relatively higher intensity is observed for the CO curve of the IO sample. On the other hand, CO2 afforded three peaks in the PO sample, while only two distinct peaks were observed in the IO curve. The peak at ∼180 °C was ascribed to desorption by the absorbed CO2 in the oil sand matrix. CO2 is primarily derived from the decarboxylic reaction (> 350 °C) and the decomposition of carbonates in minerals (> 600 °C) [25,26]. Moreover, cleavage reactions of oxygen heterocycles generate CO and CO2 when intermediate products combine with oxygen atoms through free radical mechanisms [27]. TG-MS analysis reveals that although the evolution curves for the two oil sand samples differ in shape and intensity, the temperature at which most of the abovementioned gaseous products are released during the thermal-cracking stage is ≤350 °C. It is generally supposed that pyrolysis follows a free radical mechanism [28]. Thus, the yield of gaseous products of a sample mirrors the degree of thermal decomposition of its organic components, indicating that the organic cores of oil sand are damaged predominantly during the thermal-cracking stage. Meanwhile, temperatures ≥350 °C favor the conversion of macromolecules to light fractions [29], thereby facilitating the recovery of oil sand bitumen. Regrettably, no further detail on the devolatilization stage could be elucidated from the evolution of gaseous products.

Fig. 5. Activation energy distribution versus conversion ratio (150–600 °C).

temperature reaches ∼300 °C and turns a peak at ∼450 °C for both oil sand samples. However, the peak of the IO sample occupies wider ranges of temperature, while that of the PO is relatively sharper. CH4 production is attributed to the cleavage of aliphatic compounds and aliphatic side chains [10,12]. CH4 release commenced earlier in the PO than in the IO sample (∼300 °C versus 400 °C, respectively). Moreover, the PO sample afforded a peak at a lower temperature than the IO sample (∼460 °C versus 500 °C, respectively). During oil sand pyrolysis, elemental oxygen is mainly released in the form of CO and CO2. For both samples, the CO curves are similar to those afforded by H2.

3.3. Py-EGA-MS analysis 3.3.1. Raw oil sand samples Py-EGA-MS allows for the recording of thermograms within a wider mass range. Such thermograms display the release of volatiles by application of a well-defined temperature program using a deactivated column [16]. Py-EGA-MS data (Fig. 7) revealed the volatile evolutions (33–550 amu) of the raw oil sand samples at 10 °C/min. These are

Fig. 6. Evolution curves of H2, CH4, CO, and CO2 by thermal gravimetric-mass spectrometry (TG-MS) at 10 °C/min.

732

Fuel 224 (2018) 726–739

F. Nie et al.

process, while the stages display more marked in the EGA curves. However, the corresponding temperature of the DTG peaks was lower than that of the evolution curves, possibly due to the quick devolatilization process of the smaller sample used in the Py-EGA-MS tests [16]. Similar to DTG peaks, the peak sizes during the devolatilization and thermal-cracking stages of the EGA curves also varied. For both IO and PO samples, the peak ratios between devolatilization and thermalcracking were around 36/64 and 54/46, respectively. This discrepancy between the two stages indicates the different yields of the volatile species at each stage, which is mostly determined by the proportions of the relatively thermally unstable and stable factions in oil sand. At lower temperatures, the thermally stable species lack sufficient volatility to escape the bitumen and remain behind to be further cracked until they are sufficiently volatile to escape [30]. In addition, the results can further explain the volatile release of sub-fractions in oil sand. This is discussed in the Section 3.3.2. Based on their shapes and peaks, the EGA curves were factitiously divided into three zones which have been marked in Fig. 7: Zone A (< 350 °C), Zone B (350–550 °C), and Zone C (> 550 °C). The mass spectra (Fig. 8) reveal that the volatiles from the IO and PO samples exhibit similar ion fragment distribution that are concentrated at values < 300 amu. At values > 300 amu, the intensity is lowered by three orders of magnitude. Within the tested temperature region, the highest intensity originated from CO2 at m/z 44. Fragments with m/z values of

Fig. 7. Volatile evolution curves of the raw oil sands from pyrolyzer-evolved gas analysismass spectrometry (Py-EGA-MS; 10 °C/min, 33–550 amu).

represented by the average mass of the total ion intensity curves as a function of temperature. Due to higher scanning values over those recorded for H2O (18 amu), no peaks related to dewatering were observed at ∼100 °C. Compared to the DTG curves in Fig. 3, two stages (two peaks) appear during similar temperature regions in the oil-producing

Fig. 8. Mass spectra corresponding to the zones of the evolved mass spectrometry (EGA) curves of the oil sand samples.

733

Fuel 224 (2018) 726–739

F. Nie et al.

41, 55, and 69, typically [Cn H2n − 1]+ fragments, were also significant; these fragments predominantly originate from structures with chain hydrocarbons. An m/z value of 81 was afforded by cycloolefins (e.g. alkyl substituted cyclohexene), while m/z values of 95 and 109 were probably due to aromatic fragments [31]. Possibly due its heavier bitumen molecules, the IO sample displayed a relatively stronger intensity than the PO sample at higher m/z values (insets in Fig. 8). For both oil sand samples, the distribution and intensity of the ion fragments of each zone were distinct from the total values. This confirmed the presence of different volatile components at each stage. The peak with the highest intensity at m/z = 44 dominated Zone C. This was attributed to the thermal decomposition of the carbonates present in the minerals. Owing to the higher sensitivity of the MS detector in GC/MS system, no significant weight-loss peaks were detected in the

corresponding region in TGA. Zones A and B, temperature regions corresponding to the devolatilization and thermal-cracking stages, comprised fragments as varied as the total fragments. Data once again confirmed that the release of oil sand organics mainly takes place during the devolatilization and thermal-cracking stages. However, in the latter, the thermal decomposition is more severe, presenting ion fragments of higher intensities. Although the EI source can scarcely present molecular ion peaks or identify compounds simply through individual ion fragments, it is thought that the ion fragments are generated by compounds of equal or higher molecular weights. To present the variation of compounds in volatiles within different molecular weight ranges versus increasing temperature, evolution curves with scan ranges of 33–300 amu and 300–550 amu were selected by a mass filter tool. As illustrated in Fig. 9,

Fig. 9. Volatile evolution curves of raw oil sands from pyrolyzer-evolved gas analysis-mass spectrometry (Py-EGA-MS) with different mass scan ranges (10 °C/min).

734

Fuel 224 (2018) 726–739

F. Nie et al.

the two peaks afforded from the 300–550-amu scan is also different. The PO peak in the thermal-cracking stage is significantly weaker than the one in the devolatilization stage. This distinction between the two oil sand samples originates from their different organic constituents and is also explained by the volatile evolution of bitumen, maltene, and SARA fractions in the subsequent section. When the temperature reaches 550 °C, the curve intensity of the 300–550-amu scan becomes weaker, indicating the end of the intense thermal-cracking reactions of the macromolecules in oil sand.

the curves in 33–300 amu of both oil sand samples are similar to the full range (33–550 amu) curves. On the other hand, curves afforded from a scan range of 300–550 amu displayed lower intensities and different shapes. Specifically, for both oil sand samples, the intensity of the curves from the 300–550-amu range increased smoothly before 150 °C and subsequently reached a relatively stronger peak during the devolatilization stage. However, this peak occurs at a higher temperature than that afforded from the full scan. This observation indicates that substances with higher molecule weights also exist in the volatile species from the devolatilization stage and are mostly concentrated at temperatures ranging from 200 to 350 °C. The other peak in the curve of the 300–550-amu range appears during the thermal-cracking stage. In this case, the peak temperature is quite close to that observed in the full scan curve. For both IO and PO samples, the size relationship between

3.3.2. Organic components of oil sand The organic components of oil sand comprise a complex mixture of hydrocarbons and heteroatoms that differ in aromaticity, polarity, and other properties [32]. Fractionation according to solubility and polarity

Fig. 10. Volatile evolution curves of oil sand and its bitumen, maltene, and asphaltene fractions from pyrolyzer-evolved gas analysis-mass spectrometry (Py-EGA-MS; 10 °C/min).

735

Fuel 224 (2018) 726–739

F. Nie et al.

temperature. Generally, however, pyrolysis of bitumen also exhibits a two-stage process for volatile release in the same temperature region as that observed for the raw oil sand. This suggests that under the tested heating conditions, the minerals have little influence on the pyrolysis of the two oil sand samples. For maltene and asphaltene, they are respectively the n-hexane soluble and insoluble fractions of bitumen. The dissimilarity between their curves (Fig. 10) is marked at the devolatilization stage, where little volatile contribution by the asphaltene fraction is observed at temperatures < 350 °C for both oil sand samples. The asphaltene fraction, as the heavy part of bitumen, exhibits higher aromaticity because it comprises larger aromatic cores [35]. Accordingly, higher temperatures are needed for the scission of alkyl and other substituted groups present in the peripheral sites of the asphaltene structure [13]. Thus,

has been widely employed in the separation and characterization of bitumen [33,34]. To investigate the behavior and contributions of the oil sand organic components toward volatile release, the volatile evolution curves of bitumen, maltene, and SARA fractions were obtained by Py-EGA-MS. Fig. 10 displays the volatile evolution curves of bitumen, maltene, asphaltene, and raw oil sand samples. Compared to the curves of raw oil sand, no significant increase in intensity was found in the bitumen curves at temperatures > 550 °C; whereby minerals are removed. However, possibly due to losses during extraction and vacuum rotary evaporation of toluene, the volatile evolution curves of bitumen at the low temperature region display weaker intensities than those of the raw oil sand. This phenomenon is more evident for the PO sample, which comprises more substances that is able to devolatilize at the low

Fig. 11. Volatile evolution curves of maltene and its saturate, aromatic, and resin fractions from pyrolyzer-evolved gas analysis-mass spectrometry (Py-EGA-MS; 10 °C/min).

736

Fuel 224 (2018) 726–739

F. Nie et al.

fractions during pyrolysis with reported data [12,38]. For oil sand pyrolysis, the volatiles during the devolatilization stage predominantly originate from the saturate fraction. The significant peak of volatile release during the devolatilization stage of the PO sample is attributed to its higher saturate content. The volatile species from the thermalcracking stage of maltene mainly belong to the resin fraction, which is more thermally stable than both the saturate and aromatic species. In conclusion, the yield of volatiles during different stages is greatly determined by the organic constituents of the oil sand.

during oil sand pyrolysis, the release of volatiles during the devolatilization stage primarily originates from the maltene fraction, while the volatiles during the thermal-cracking stage are derived from both the maltene and asphaltene fractions. For the IO samples, a higher asphaltene content in bitumen decreases the temperature susceptibility and improves the resistance of the samples to thermal decomposition [36]. Moreover, owing to the higher maltene content in PO, relatively more volatiles (a larger peak) were generated than from the IO sample during the devolatilization stage. Compared to the two oil sand samples, the peak shapes in the evolution curves of maltene and asphaltene at each stage also differ. This is predominantly due to their different original compositions. Asphaltenes also consist of a bimodal composition that represents fractions of entities that are easily converted and of refractory thermally stable species [37]. Therefore, the differences observed in the asphaltene curves depend on the ratio between the thermally stable and unstable components of their origins. While for maltene fraction, the volatile evolution of its sub-fractions was analyzed to determine the causes of the differences between the two oil sand samples. As is depicted in Fig. 11, the volatile evolution curves of each maltene subfraction exhibit disparate shapes. Generally, the relationship of their released volatile intensity (average amounts) during the devolatilization and thermal-cracking stages is as follows:

3.3.3. Interactions between each fraction during pyrolysis Fractionation of the oil sand organics facilitates the analysis and understanding of the individual volatile release processes. However, potential interactions between these fractions may take place during pyrolysis. This section compares volatile evolution curves of bitumen and maltene between the simulate and the raw ones. The simulate curves are calculated as Eq. (2): n

g (T ) =

∑

wi fi (T )

i= 1

(2)

where g(T) and fi (T) are the respective simulate curve and its subfraction curve as a function of temperature and wi is the corresponding percentage mass. Eqs. (3)–(6) represent calculations for the simulate bitumen and maltene, respectively, while the comparison of the simulate and raw curves are presented in Fig. 12.

Devolatilization stage:saturate > aromatic>resin; Thermal-cracking stage:resin > aromatic>saturate. This behavior can be verified by comparing the weight loss of these

Fig. 12. Comparison between raw and simulate volatile evolution curves of bitumen and maltene (10 °C/min).

737

Fuel 224 (2018) 726–739

F. Nie et al.

Simulate IO bitumen: g (T ) = 0.7098fmaltene (T ) + 0.2902fasphaltene (T )

3.4. Volatile compositions of each stage (3)

As shown in Section 3.3, Py-EGA-MS provides the basic information and differences on volatile release via temperature. According to the volatile evolution curves, HCA identified the composition and yield of the volatiles at each stage. GC/MS analysis of the in-situ trapped volatiles by cooled nitrogen (∼−190 °C) during the devolatilization (40–350 °C) and thermal-cracking (350–600 °C) stages at 10 °C/min, provided the relative content of each species of volatile compounds at the two stages (Table 4; original GC/MS results in the Supporting information). GC/MS results indicated that the volatiles during oil sand pyrolysis comprise a complex mixture including aliphatics, aromatics, heteroatom-containing compounds, polycyclic biomarkers, and inorganic gases. The volatile composition at each stage differs greatly (Table 4). Small amounts of alkanes and olefins were present during the devolatilization stage, while in the thermal-cracking stage, they account for > 45% of the detected compounds. According to the information on organic structures (Table 2), it is illustrated that the main organic structures are damaged at low degrees during the devolatilization stage. Alkanes and olefins in thermal-cracking stage are predominantly afforded from the cracking of the abundant methylene groups in oil sand bitumen. Moreover, due to the longer aliphatic chains in the PO bitumen, the carbon number of the detected alkanes is ≤C40, while that in the IO volatiles only reaches C35. Higher proportions of unsaturated aliphatics, including olefins, cycloolefins, and alkadienes, were also found in the volatiles generated during the thermal-cracking stage. Among the aromatics, benzenes are the main products. However, more benzenes with long alkyl substituents are found in the volatiles afforded during the devolatilization stage, while the length of the alkyl substituents of the benzene species in the thermal-cracking stage were < C4. Additionally, the volatiles from the devolatilization stage of the PO sample comprised a significant amount of hydronaphthalene and hyrdroindene (∼7%). Another great difference between the two stages is in the polycyclic biomarker content, which includes hydro-PAHs, terpenes, sterane, and

Simulate PO bitumen: g (T ) = 0.7972fmaltene (T ) + 0.2028fasphaltene (T ) (4)

Simulate IO maltene: g (T ) = 0.1229fsaturate (T ) + 0.4286faromatic (T ) + 0.4485fresin (T )

(5)

Simulate PO maltene: g (T ) = 0.2156fsaturate (T ) + 0.2983faromatic (T ) + 0.4861fresin (T )

(6)

As illustrated in Fig. 12, none of the simulate bitumen and maltene curves are in a perfect line with the raw curve. We therefore postulated that interactions between the sub-fractions, which follow a group of complex free radical reactions, exist during pyrolysis [27]. The coincidence between the simulate and the raw curve may depend on the degree of thermal decomposition and the priority of the amount of subfraction. Firstly, compared to the devolatilization stage, bond cleavage and the generation of free radicals in the thermal-cracking stage occur under a severer thermal pressure. According to the volatile evolution curves, the maltene, asphaltene, aromatic, and resin fractions exhibit a strong degree of volatile release during the thermal-cracking stage. For all these sub-fractions, the proximate simultaneity on the decomposition temperature leads to interactions between radicals from different origins. As is observed in the PO bitumen, the simulate curve well matches the raw curve during the devolatilization stage. However, the peak intensity decreases during the thermal-cracking stage. Secondly, the simulate volatile evolution curves of bitumen display a relatively better match. In the tested oil sand samples, the bitumen is dominated by the maltene fraction, accounting for > 70 wt%. Thus, the volatile evolution curves of bitumen are greatly influenced by that of maltene. However, due to more and no amount dominating sub-fractions in maltene, the matched-degree between the simulate and the raw curve decreases, simultaneously revealing the possibility of interactions between these sub-fractions during pyrolysis.

Table 4 Relative content of each species of volatile compounds at different temperature regions (peak area%). Species of volatile compounds

IO

CO2 SO2 CS2

PO

40–350 °C

350–600 °C

40–350 °C

350–600 °C

43.71 8.68 0.32

38.47 0.28 –*

6.11 0.44 –

23.00 0.26 –

Aliphatics

Alkanes Olefins Cycloalkanes Cycloolefins Alkadienes

8.77 5.71 1.58 1.49 0.10

21.09 24.40 2.39 3.93 0.78

10.30 3.04 4.15 0.88 0.26

36.62 29.58 2.08 1.25 0.15

Aromatics

Benzenes Naphthalenes Indenes Hydronaphthalenes Hydroindenes PAHs

2.87 – 0.41 0.86 0.23 –

2.15 – 0.29 0.16 0.13 –

3.29 0.73 0.06 4.46 2.66 0.19

1.47 – – 0.28 0.24 –

O-containing compounds

Phenols Alcohols Ketones Esters

– 0.28 0.44 –

– 0.47 0.30 –

0.10 – 0.10 0.73

– 0.22 – 0.29

S-containing compounds

Thiophenes Benzo[b]thiophenes

2.89 0.09

1.76 0.88

– –

– –

0.38 17.91 3.30

– 0.66 1.85

0.41 53.07 9.00

– 2.47 2.09

N-containing compounds Polycyclic biomarkers Others * “–” Not detected.

738

Fuel 224 (2018) 726–739

F. Nie et al.

hopance. These were predominantly observed as organic compounds in the volatiles of the devolatilization stage. Terpenes and sterane are the most common biomarkers in sediments that mainly originate from prokaryotes (bacteria) and true nuclear organisms (phytoplankton and higher animals), respectively [39,40]. When combined with the PyEGA-MS results, these polycyclic biomarkers are thought to be dissociated, especially in maltene. Thus, they can easily be released from oil sand at lower temperatures, thereby contributing toward the stronger intensity in the 300–550-amu range in Fig. 8. Additionally, due to the higher amount of polycyclic biomarkers, the PO sample exhibits volatile evolution curves of stronger intensities during the devolatilization stage. Compounds with oxygen, sulfur, and nitrogen hetero atoms were also observed. Owing to the high sulfur content in the IO sample, significant amounts of SO2, CS2, thiophenes, and benzo[b] thiophenes were found in the volatiles.

[7] Ma Y, Li S. Study of the characteristics and kinetics of oil sand pyrolysis. Energy Fuels 2010;24:1844–7. [8] Liu P, Zhu M, Zhang Z, Zhang D. Pyrolysis of an Indonesian oil sand in a thermogravimetric analyser and a fixed-bed reactor. J Anal Appl Pyrol 2016;117:191–8. [9] Zhao HY, Cao Y, Sit PS, et al. Thermal characteristics of bitumen pyrolysis. J Therm Anal Calorim 2012;107(2):541–7. [10] Hao J, Feng W, Qiao Y, Tian Y, Zhang J, Che Y. Thermal cracking behaviors and products distribution of oil sand bitumen by TG-FTIR and Py-GC/TOF-MS. Energy Convers Manage 2017;151:227–39. [11] Hepler LG, Hsi C, editors. AOSTRA technical handbook on oil sands, bitumens and heavy oils. Edmonton, Alberta, Canada: Alberta Oil Sands Technology and Research Authority; 1989. [12] Hao J, Che Y, Tian Y, Li D, Zhang J, Qiao Y. Thermal cracking characteristics and kinetics of oil sand bitumen and its SARA fractions by TG–FTIR. Energy Fuels 2017;31:1295–309. [13] Varfolomeev MA, Galukhin A, Nurgaliev DK, Kok MV. Thermal decomposition of Tatarstan Ashal’cha heavy crude oil and its SARA fractions. Fuel 2016;186:122–7. [14] Shin S, Im SI, Kwon EH, Na JG, Nho NS, Lee KB. Kinetic study on the nonisothermal pyrolysis of oil sand bitumen and its maltene and asphaltene fractions. J Anal Appl Pyrol 2017;124:658–65. [15] Liu D, Song Q, Tang J, Zheng R, Yao Q. Interaction between saturates, aromatics and resins during pyrolysis and oxidation of heavy oil. J Petrol Sci Eng 2017;154:543–50. [16] Shiono A, Hosaka A, Watanabe C, Teramae N, Nemoto N, Ohtani H. Thermoanalytical characterization of polymers: a comparative study between thermogravimetry and evolved gas analysis using a temperature-programmable pyrolyzer. Polym Test 2015;42:54–61. [17] Materazzi S, Gentili A, Curini R. Applications of evolved gas analysis part 2: EGA by mass spectrometry. Talanta 2006;69:781–94. [18] Kim YM, Han TU, Watanabe C, Teramae N, Park YK, Kim S, et al. Analytical pyrolysis of waste paper laminated phenolic-printed circuit board (PLP-PCB). J Anal Appl Pyrol 2015;115:87–95. [19] Planche JP, Michon L, Martin D, Hanquet B. Estimation of average structural parameters of bitumens by 13C nuclear magnetic resonance spectroscopy. Fuel 1997;76:9–15. [20] Heigenmoser A, Liebner F, Windeisen E, Richter K. Investigation of thermally treated beech (Fagus sylvatica) and spruce (Picea abies) by means of multifunctional analytical pyrolysis-GC/MS. J Anal Appl Pyrol 2013;100:117–26. [21] Meng M, Hu H, Zhang Q, Li X, Bo W. Pyrolysis behaviors of tumuji oil sand by thermogravimetry (TG) and in a fixed bed reactor. Energy Fuels 2007;21:2245–9. [22] And KM, Maki T. A simple method for estimating f(E) and k0(E) in the distributed activation energy model. Energy Fuels 1998;12:864–9. [23] Arenillas A, Rubiera F, Pis JJ, Cuesta MJ, Iglesias MJ, Jiménez A, et al. Thermal behaviour during the pyrolysis of low rank perhydrous coals. J Anal Appl Pyrol 2003;68:371–85. [24] Zou C, Ma C, Zhao J, Shi R, Li X. Characterization and non-isothermal kinetics of Shenmu bituminous coal devolatilization by TG-MS. J Anal Appl Pyrol 2017;127:309–20. [25] Ritchie RGS, Roche RS, Steedman W. Non-isothermal programmed pyrolysis studies of oil sand bitumens and bitumen fractions: 1. Athabasca asphaltene. Fuel 1985;64:391–9. [26] Crawford RW. Pyrolysis of Sunnyside (Utah) tar sand: characterization of volatile compound evolution. Fuel Sci Technol Int 1988;7:823–49. [27] Macphee JA, Charland JP, Giroux L. Application of TG–FTIR to the determination of organic oxygen and its speciation in the argonne premium coal samples. Fuel Process Technol 2006;87:335–41. [28] Poutsma ML. Fundamental reactions of free radicals relevant to pyrolysis reactions. J Anal Appl Pyrol 2000;54:5–35. [29] Shi J, Ma Y, Li S, Wu J, Zhu Y, Teng J. Characteristics of Estonian oil shale kerogen and its pyrolysates with thermal bitumen as a pyrolytic intermediate. Energy Fuels 2017;31:4808–16. [30] Barbour RV, Dorrence SM, Vollmer TL, Harris JD. Pyrolysis of Utah tar sands: products and kinetics. Am Chem Soc Div Fuel Chem 1976;21:278–83. [31] Middleditch BS, editor. Practical mass spectrometry. Boston, MA: Springer; 1979. [32] Yoon S, Bhatt SD, Lee W, Lee HY, Jeong SY, Baeg JO, et al. Separation and characterization of bitumen from athabasca oil sand. Korean J Chem Eng 2009;26:64–71. [33] Xia TX, Greaves M. Downhole upgrading athabasca tar sand bitumen using thai – SARA analysis. In SPE international thermal operations and heavy oil symposium, 12–14 March, Porlamar, Margarita Island, Venezuela; 2001. http://dx.doi.org/10. 2118/69693-MS. [34] Fuhr B, Scott K, Dettman H, Salmon S. Fractionation of bitumen by distillation, sara analysis and gel permeation chromatography. Prepr – Am Chem Soc Div Pet Chem 2005;50:264–5. [35] Trejo F, Rana MS, Ancheyta J. Thermogravimetric determination of coke from asphaltenes, resins and sediments and coking kinetics of heavy crude asphaltenes. Catal Today 2010;150:272–8. [36] Firoozifar SH, Foroutan S, Foroutan S. The effect of asphaltene on thermal properties of bitumen. Chem Eng Res Des 2011;89:2044–8. [37] Hauser A, Bahzad D, Stanislaus A, Behbahani M. Thermogravimetric analysis studies on the thermal stability of asphaltenes: pyrolysis behavior of heavy oil asphaltenes. Energy Fuels 2008;22:449–54. [38] Zhang C, Xu T, Shi H, Wang L. Physicochemical and pyrolysis properties of sara fractions separated from asphalt binder. J Therm Anal Calorim 2015;122:1–9. [39] Patience RL. Fossil fuel biomarkers, applications and spectra. Earth-Sci Rev 1986;23. 230–230. [40] Yang Chun, Wang Zhendi, Yang Zeyu, Hollebone Bruce, Brown Carl E, Landriault Mike. Chemical fingerprints of alberta oil sands and related petroleum products. Environ Forensics 2011;12:173–88.

4. Conclusions (1) Both the weight losses and release of volatiles reveal two stages for the oil-producing process during oil sand pyrolysis: devolatilization (< 350 °C) and thermal-cracking (350–600 °C) stages. These two stages display different activation energy distributions according to the distributed activation energy model. (2) The raw oil sand, bitumen, maltene, and SARA fractions display different volatile release characteristics with increasing temperature. According to the volatile evolution curves of raw oil sand and its bitumen, the effect of the mineral during pyrolysis of the two oil sand samples under the tested heating condition is small. However, the existence of interactions between these sub-fractions during oil sand pyrolysis has been suggested. (3) The yield of volatiles during the different stages is greatly determined by the organic constituents of the oil sand. Moreover, the volatile compositions at the different stages is markedly disparate, which is related to the original organic structure of oil sand. The volatiles in the devolatilization stage predominantly originate from the saturates in maltene. These species comprise a large number of polycyclic biomarkers and contribute toward the release of highermolecular-weight substances over a temperature range of 200–350 °C. The thermal-cracking stage is the main stage in which most of the gaseous and light pyrolytic products (mostly alkanes and olefins) are generated. These are primarily afforded from the cracking of resin in maltene and asphaltenes. The raw aromatic fraction in maltene releases volatiles in both stages. Acknowledgement Support from the National Key R&D (2016YFB0600400) is greatly acknowledged.

Program

of

China

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2018.03.069. References [1] BP statistical review of world energy 2017; June 2017. < http://www.bp.com/ content/dam/bp/en/corporate/pdf/energy-economics/statistical-review-2017/bpstatistical-review-of-world-energy-2017-full-report.pdf > . [2] BP energy outlook; 2017 edition. < http://www.bp.com/content/dam/bp/pdf/ energy-economics/energy-outlook-2017/bp-energy-outlook-2017.pdf > . [3] Misra M, Miller JD. Comparison of water-based physical separation processes for U.S. tar sands. Fuel Process Technol 1991;27:3–20. [4] Krutof A, Hawboldt K. Blends of pyrolysis oil, petroleum, and other bio-based fuels: a review. Renew Sustain Energy Rev 2016;59:406–19. [5] Kapadia PR, Kallos MS, Gates ID. A review of pyrolysis, aquathermolysis, and oxidation of Athabasca bitumen. Fuel Process Technol 2015;131:270–89. [6] Wang Q, Jia C, Jiang Q, et al. Pyrolysis model of oil sand using thermogravimetric analysis. J Therm Anal Calorim 2014;116:499–509.

739