Quantification of low‐temperature oxidation of light oil and its SAR fractions with TG‐DSC and TG‐FTIR analysis

The oxidation reaction is the key to determining the success of air flooding. In this paper, experimental and theoretical techniques have been developed to identify the low‐temperature oxidation (LTO) mechanisms for light oil during air flooding by comprehensively analyzing thermal stability and oxidation process of the crude oil and its SAR (ie, saturates, aromatics, and resins) fractions. Experimentally, both a thermogravimetric analyzer coupled with differential scanning calorimetry (TG‐DSC) and a thermogravimetric analyzer coupled with Fourier transform infrared spectrometer (TG‐FTIR) are employed to quantify the LTO process of crude oil and each SAR fraction as well as the corresponding oxidation properties. Theoretically, reaction models have been developed to reproduce the experimentally identified reactions. The results show that the oxygen addition reaction and the bond scission reaction occur simultaneously. The former can be initiated when temperature is higher than 50°C, and it is gradually shifted to the latter with the continuous increase in reservoir temperature. The LTO products of light oil include H2O, CO2, carboxylic acids, alcohols, phenols, and ethers. Saturates, aromatics, and resins are all the sources of H2O, CO2, alcohols, and carboxylic acids, whereas ethers are mainly derived from aromatics and resins. At the beginning of an air flooding process, heat is mainly generated from the oxidation of aromatics and resins. Subsequently, oxidizing saturates gradually dominates the air flooding process with an increase in the reservoir temperature.


| INTRODUCTION
Due to the inherent low permeability in a tight oil reservoir, the primary recovery is still very low, though horizontal wells have been drilled and massively fractured. 1,2 Numerous efforts have been directed to develop suitable enhanced oil recovery (EOR) techniques for such a tight formation due to its rapid decline of initial productions. Gas (eg, CO 2 , N 2 , and natural gas) flooding shows positive responses in either low permeability reservoirs or conventional reservoirs that have advanced to the late stages of waterflooding, 3-6 though its extensive applications are still limited due mainly to the gas sources and the associated operational costs. Comparatively, air is readily available and can be compressed and injected to displace oil with relatively low cost. In 1950s, air injection was initiated in a heavy oil reservoir by means of in situ | 377 WANG et Al. combustion (ISC) 7 and then piloted in the Sloss light oil reservoir in 1963. 8 As for the latter, air is used to oxidize the light crude at reservoir temperature to generate flue gas that will displace the formation of oil to the producer under nearmiscible conditions. 9 This research mainly focuses on the air flooding applied in light oil reservoirs. Physically, the lowtemperature oxidation (LTO) between light oil and oxygen dominates the successful application of air flooding in light oil reservoirs. Therefore, it is essential to identify the inherent reactions associated with the LTO processes prior to designing and optimizing an air flooding project in a technically feasible manner.
So far, extensive efforts have been made to quantify the LTO reaction through investigating the changes in gas composition and system pressure before and after the reaction. 10,11 It is found that part components of the crude oil can be oxidized to oxygenated hydrocarbon compounds (eg, aldehydes, ketones, and alcohols) which can be further oxidized to carbon oxides such as CO 2 . 11 However, both the main components involved in the LTO reaction and the differences associated with the oxidation characteristics among different components have not been well understood. To quantify the oxidation characteristics of crude oils, thermal analysis such as thermogravimetric analysis and differential scanning calorimetry (TG-DSC), pressure differential scanning calorimetry (PDSC), and thermogravimetric analysis coupled with Fourier transform infrared spectrometer (TG-FTIR) have been applied to investigate the oxidation properties of crude oil and its SARA fractions, 10,12-32 as summarized in Table 1. Also, this table tabulates the reaction modes and characteristics during the combustion of light and heavy oils, the kinetics and thermochemical parameters of each reaction mode, the differences of oxidation characteristics between light, medium, and heavy oils, the oxidation relationship of crude oil and SARA fractions during the combustion process, and the combustion process of asphalt binder and its SARA fractions, respectively. So far, the oxidation process of light oil, especially the LTO process of light oil and its SARA fractions, has not been fully identified, and the LTO mechanisms and reaction model require a systematic and comprehensive analysis.
In this paper, experimental and theoretical techniques have been developed to identify the LTO mechanisms for light oil during air flooding by comprehensively analyzing the thermal stability and oxidation process of crude oil and its SAR fractions. Experimentally, the TG-DSC tests are conducted to determine the thermal stability by analyzing the changes of weight loss and exothermic behavior. By applying TG-FTIR tests, the real-time oxidation products of crude oil and SAR fractions are then determined to identify the key factors dominating the LTO reaction process. Theoretically, the relationship of oxidation property between crude oil and its fractions is analyzed, and then, a reaction model based on the peroxidation theory and the free radical reaction theory is developed to reproduce the LTO mechanisms.

| Materials
In this study, a light oil produced from a tight reservoir in the Changqing oilfield in China is collected and used to conduct the experiments. Physical properties of the oil and its SARA fractions are tabulated in Table 2. It is worthwhile noting that the oxidation of asphaltenes is not included in this research due to the fact that the state of asphaltenes in a crude oil is completely different from that of the asphaltenes separated from the crude oil. 33,34 The air composed of 21.0 mol% oxygen and 79.0 mol% nitrogen is supplied by the Qingdao Tianyuan Gas Company. The neutral alumina, reagent-grade n-pentane, HPLC-grade toluene, HPLC-grade methanol, and HPLC-grade tetrahydrofuran used in the SARA fraction separation are all provided by the Sinopharm Chemical Reagent Co., Ltd.

| Experimental setup
In this study, a vacuum oven (YZF-6032; Shanghai Yaoshi Instrument Equipment Factory), an analytical balance (AB105; Shanghai Precision Instrument Company), and an ultrasonic disperser (Scientz-2400F; SCIENTZ) are used to separate the light crudes into SARA fractions. The operating temperatures of the vacuum oven are from the room temperature to 250°C with its temperature accuracy of 0.1°C and the ultimate vacuum <133 Pa. The maximum scale of the balance is 105 g with its weighting accuracy of 0.01 mg, while the frequency of ultrasonic disperser is 19.5-20.5 kHz.
The experimental setup used in thermal analysis consists of a TG-DSC test system (STA6000; PerkinElmer) and a FTIR spectrometer (Fourier Infrared Spectrometer; PerkinElmer). As for the STA6000, its balance sensitivity is 0.1 μg with calorimetric accuracy ±2%, and temperature accuracy <±0.5%. As for the FTIR spectrometer, the scanning wavenumber precision is better than 0.008 cm −1 and the absorbance precision is better than 0.05%.

| Separation of SARA fractions
The crude oil was separated into SARA fractions according to a modified analytical procedure used by Freitag et al. 35 Asphaltenes were recovered from the oil by ultrasonic dispersion in 40 volumes of n-pentane, overnight flocculation, and filtration through 0.8-μm filter paper, while the remaining solvent was removed by evaporation in a vacuum oven. The fractions of saturates, aromatics, and resins were subsequently separated from the maltenes (ie, the residual oil after removal of asphaltenes) by a modified liquid chromatography procedure on an alumina packing. The saturates were eluted from the alumina column using n-pentane, the aromatics were separated using toluene, and the resins were eluted out with a mixture of 12.5 vol% methanol and 87.5 vol% tetrahydrofuran. Any remaining solvents were removed from the isolated fractions by evacuation in a vacuum oven. During the fraction purification and cooling process, nitrogen was used to prevent the oxidation of SARA fractions.

| TG-DSC tests
Thermal stability of crude oil and each fraction was examined by using the TG-DSC test. In this work, 12 mg oil or its fraction sample was placed in the alumina crucible prior to the TG-DSC analysis. The input gas was composed of 21.0 mol% oxygen and 79.0 mol% nitrogen at a flow rate of 30 mL/min. The samples were then heated from 40°C to 600°C at a heating rate of 4°C/min to examine the changes of weight loss and heat flow.

| TG-FTIR tests
The TG-FTIR tests were conducted to quantify the oxidation process of the crude oil and its SAR fractions. 12 mg oil or its fraction sample was placed in the alumina crucible of the thermogravimetric analyzer, while air with a flow rate of 30 mL/min was used to ensure an oxidation environment. The sample was first heated to 180°C at a heating rate of 50°C/min and then kept at 180°C for 120 min. By using air as a carrier gas, the oxidized volatiles were directly introduced to the IR gas cell of the FTIR spectrometer for online analysis. The gas cell temperature was kept at 180°C, the resolution was set at 1 cm −1 , and the scanning range was 500-4000 cm −1 . The transfer line between the thermal analyzer and the infrared spectrometer was maintained at 180°C to avoid any condensation of the released gaseous products.

| Thermal stability
The TG, DTG, and DSC curves of crude oil and its fractions are plotted in Figure 1. The thermal stability can be characterized by the weight loss and heat flow changes at different temperatures, while the temperature ranges of each phase are summarized in Table 3.
During the heating process, the TG curves show a continuous decline (see Figure 1A), while the DTG and DSC curves are fluctuated (see Figure 1B, C). The peaks of the DTG and DSC curves correspond to the rapid weight loss and rapid exothermic stages, respectively. The heating process of crude oil and SAR fractions consists of three main consecutive stages, that is, LTO phase, fuel deposition (FD) phase, and high-temperature oxidation (HTO) phase. This finding is consistent with those documented elsewhere. 28,36,37 The LTO phase mainly occurs distillation and LTO reactions. 10,28,38 Due to distillation, the weight loss of saturates reaches 90.9% at the end of the LTO phase. LTO reactions, typically occurring at temperatures below 350°C, 25,39-41 consist of oxygen addition reaction and bond scission reaction, and the rate of heat generation and weight loss of the latter is much higher than the former. [42][43][44] During the heating process, the oxidation peak first appears in saturates at 339°C, indicating that the reaction mode of saturates is the most prone to change from oxygen addition reaction to bond scission reaction with an increase in temperature. The FD reaction involves the oxidation of pyrolysis products 45 where coke and light hydrocarbons with low molecular weight are generated, accompanied by heat absorption. 46 The FD reaction temperatures of aromatics and resins are higher than that of saturates, and their weight losses and exothermic peaks in FD stage are obviously larger. Combustion is the main reaction at the HTO phase which consumes the coke generated from the FD stage and produces large heat and carbon oxides. 47,48 At the HTO phase, the starting and peak temperatures of aromatics and resins are higher than those of saturates, and their combustion reaction intensity is much greater than saturates. Resins have the highest heat production at the HTO phase.
The weight loss pattern of crude oil is similar to that of saturates, that is, both weight losses at the LTO phase are above 80%, and the weight loss patterns are similar at both FD and HTO phases. However, the weight loss of crude oil is greater than that of saturates at T < 200°C due to the inherent deviation associated with the separation of SARA fractions. 35 Such hydrocarbon loss occurs the most frequently with saturates because the initial boiling points of the aromatics and resins are much higher. In addition, the heat production of crude oil at the LTO and FD phases is in a good agreement with that of saturates in the DSC tests, though the heat release Crude oil and its fractions have a measurable heat production at a temperature higher than 50°C, that is, the LTO reactions can occur at T > 50°C. The increase of sample weight at the initial of the TG tests (see Figure 1A) also confirms the existence of the oxygen addition reaction. The peak weight of saturates, aromatics, and resins reaches 100.18, 100.28, and 100.32 wt% at a temperature below 200°C. In addition, the heat release intensity sequence is found to be resins > aromatics > saturates at T < 200°C. Therefore, the oxidation activity of SAR fractions is resins > aromatics > saturates when temperature is lower than 200°C. This finding is consistent with those documented elsewhere. 49,50 The differences in oxidation activity of SAR fractions are due to the different molecular structures: Resins are a heavy fraction of crude oil, which contains a large number of aromatic rings, alicyclic rings, and kinds of short-and longchain branches, resulting in a strong polarity and can react with oxygen more easily than light fractions. 10 In addition, resins contain a large number of heteroatoms, such as S and N atoms, and the bond energy of the C-S and C-N bonds is significantly lower than the C-C bond, so the resins are more susceptible to the attack of oxygen. 51 When the temperature is above 200°C, the heat production rate and weight loss rate are both increased dramatically, especially for saturates. This is because the bond scission reactions with higher heat generation efficiency gradually become dominant with an increase in temperature. 43 Therefore, at the initial stage of LTO, aromatics and resins are the important sources for oxygen consumption and heat generation. After the reservoir temperature is increased due to the heat production, the oxidation of saturates is enhanced and gradually dominates the LTO. The content of saturates in a light crude oil is dominant and ensures the stable oxygen consumption during air flooding. 19 As for a reservoir with good insulation, the reservoir temperature will increase continuously due to the heat generated from LTO reactions. Then, the spontaneous combustion can occur with the reaction mode spontaneously changed to the FD and HTO reactions, during which aromatics and resins dominate the stability of the combustion front.

| Oxidation relationship between crude oil and fractions
The additivity rule is used to quantify the connection of oxidation behavior between the crude oil and its fractions. The weight loss of each fraction is summed according to its content in crude oil to obtain the cumulated weight loss, and the same method is used to obtain the cumulated heat flow. The same calculation method is applied in reference. 17 The relationship between the cumulated and measured weight loss and heat flow is illustrated in Figure 2. Obviously, there exists a similar pattern between the cumulated and measured weight loss, though differences remain when temperature is  Figure 2A). This is mainly due to the loss of the light hydrocarbons during the SARA separation. As for the heat flow (see Figure 2B), the cumulated values are close to the measured ones within the test temperature range except those at 130-240°C (loss of light hydrocarbons) and 480-560°C (excluding asphaltene). Asphaltene is the heaviest fraction in crude oil and important source of combustion fuel at the HTO phase. 16,44 In addition, the interaction between SARA fractions during the reaction is also an important reason for the differences between the cumulated and measured values. 22 In general, the cumulated weight loss and heat flow of crude oil based on the oxidation properties of SAR fractions are qualitatively similar with their measured values. As such, it is an effective method to quantify the oxidation characteristics of crude oil by analyzing that of its individual fraction. This finding is consistent with those documented elsewhere. 19

| LTO reaction process
The TG-FTIR tests are conducted to quantify the LTO reaction process by describing the dynamic evolution of gaseous products during each fraction oxidation. The TG curves, 3D FTIR, and FTIR spectra of the released volatiles during the oxidation of crude and fractions are analyzed and discussed, respectively.

| Weight loss
The weight loss of crude oil and each fraction at 180°C is depicted in Figure 3. As can be seen, the weight losses of crude oil and saturates are large at the beginning and reach 47.96 wt% and 30.11 wt% at t = 10 min, respectively. Then, their weight losses slow down, and the residual weights are 38.49 and 40.21 wt% at the end of the test (ie, t = 120 min). The weight loss at the initial stage is mainly attributed to the distillation of hydrocarbons with low boiling points. The weight losses of aromatics and resins are significantly lower due to their high boiling points. The weight losses of aromatics and resins are only 19.85 and 6.92 wt% at the end of the test. The evaporation of water, carbon oxides, and light oxygenated hydrocarbons produced during the LTO are the contributor for the continued weight loss, especially at t > 20 min when the evaporation of low-molecular-weight hydrocarbons is terminated. This finding is also supported by the spectra changes of the released volatiles (see Figure 4 and Figure 5).
In addition, the sample weight is increased significantly in the first three minutes of the tests (see Figure 3), indicating the existence of oxygen addition reactions at the LTO stage.

| 3D FTIR
The 3D FTIR of crude oil and its fractions can provide the overall information of dynamic evolution of gaseous products in the LTO process (see Figure 4). The absorption peaks in the spectrum represent different functional groups. The 3D FTIR of fractions are obviously different, indicating obvious differences between the LTO processes of each fraction (see Figure 4A-D).

| Gaseous products
FTIR spectra of the emitted volatiles of crude oil and its fractions at different times are selected to analyze the LTO process (see Figure 5A-D).

Crude oil
As shown in Figure 5A, in the first 10 min of the test, the released volatiles are mainly hydrocarbons (corresponding to wavenumber 2800-3000 cm −1 ) with low boiling points, but the LTO reaction has already occurred at this time. The wavenumber 3400-4000 cm −1 in the spectra at t = 2.  51 This is also supported by the follow-up tests of the crude oil acidity and the oxygen element content. The acidity is increased from 0.04 mg KOH/g of the original oil to 0.12 mg KOH/g after its oxidation at 120°C for 6 hours, and the oxygen element content is increased from 0.32 wt% to 0.48 wt%. In addition, the weak absorption peak at 1000-1300 cm −1 indicates the presence of a small amount of ether in the evolved gas. Asymmetrical stretching and bending vibrations of CO 2 occur at 2360 and 670 cm −1 , respectively, 43 and both absorption peaks are found in the FTIR spectra. CO 2 is originated from decarboxylation of carboxylic acids generated by oxygen addition reaction. 10,51 The concentration of CO 2 is increased as the test progresses. Accordingly, the bond scission reactions are intensified. After t = 30 min, CO 2 becomes the main component of the gaseous products.
The LTO products of crude oil include H 2 O, CO 2 , and oxygenated hydrocarbons (eg, carboxylic acids, alcohols, phenols, and ethers). The oxygenated hydrocarbons are the main products at the early stages; however, CO 2 is increased with time and becomes the main gaseous product at the later stages of the test. Therefore, both oxygen addition reaction and bond scission reaction occur, but the former is dominant at the beginning of the reaction and the latter is gradually enhanced and dominated as time proceeds.

Saturates
Similar to crude oil, the main released volatiles of saturates in the first 10 min are hydrocarbons with low boiling points (see Figure 5B). The oxygenated hydrocarbons produced in the LTO of saturates include carboxylic acids and alcohols, but phenols cannot be detected. CO 2 production is significant later of the test. During the LTO of saturates, the oxygen addition reaction and the bond scission reaction also occur simultaneously, and the products consist of H 2 O, CO 2 , alcohols, and carboxylic acids. Saturates are an important source of carboxylic acid and CO 2 in the LTO of crude oil.

Aromatics
Different from saturates, the main gaseous products of aromatics at the initial stage are H 2 O and oxygenated hydrocarbons rather than light hydrocarbons (see Figure 5C). Another notable difference is the obvious adsorption peaks at 970-1080 and 1120-1270 cm −1 , indicating the presence of aromatic ether in LTO products of aromatics. The products of aromatics during LTO include H 2 O, CO 2 , phenols, alcohols, carboxylic acids, and ethers. The oxygen addition reaction is prominent at the beginning of LTO, while the bond scission reaction is dominant at later stages. The aromatic ethers produced are an important source of ethers in the LTO of crude oil.

Resins
The gaseous LTO products of resins include H 2 O, CO 2 , phenols, alcohols, carboxylic acids, and ethers, among which H 2 O, CO 2 , and ethers are the main components (see Figure  5D). Different from saturates and aromatics, both oxygen addition reaction and bond scission reaction are obvious in the entire process of the resins test, and the LTO intensity of resins is higher than that of saturates and aromatics. The ethers produced in the LTO of resins are another major source of ethers in the LTO of crude oil. In summary, the oxidation characteristics of SAR fractions are the intrinsic determinants of the crude oil oxidation. Saturates, aromatics, and resins are all the sources of the produced H 2 O, CO 2 , alcohols, and carboxylic acids, whereas ethers are mainly derived from aromatics and resins. CO 2 gradually becomes the main gaseous product with the continuous increase of bond session reactions.

| LTO reaction models
Under reservoir conditions, the oxidation of crude oil and its fractions occur according to the free radical chain reaction mechanism. 22,52 The LTO mechanisms of crude oil and SAR fractions are reproduced by the peroxidation theory of Bach-Engler and the free radical chain reaction theory of Semyonov, 53,54 and thus, the corresponding LTO reaction model is established. The LTO process mainly involves chain initiation reaction, chain propagation reaction, chain degenerate-branching reaction, and chain termination reaction, 53 among which the chain initiation reaction that produces hydrocarbon radicals is not only the most difficult step in the whole oxidation process, but also the key factor causing the reactivity differences between SAR fractions. The radicals at the beginning of the reaction are produced by the action of external energy (eg, heat), 54 and the reaction formula is shown as follows: The chain propagation reactions are mainly the addition reaction and the hydrogen abstraction reaction between hydrocarbon radicals and oxygen molecules or hydrocarbon molecules, and their products are hydrocarbon peroxides and new free radicals listed as follows: In the degenerate-branching chain reaction, the organic peroxides decompose to free radicals such as RO ⋅ , then, the free radicals produced can initiate new chain reactions and produce oxygenated hydrocarbons such as alcohols, aldehydes, and carboxylic acids.  The chain termination reactions are mainly the combination of free radicals, which generate oxygenated hydrocarbons (eg, ethers and alcohols) or high-molecular-weight components, that is, In the aforementioned reaction formulae, -R s and -R l can represent an alkyl group, an alkyl side chain containing an aromatic ring or a cycloalkane ring, or an alkyl side chain containing a fused aromatic ring.
With an increase in molecular weight, the C-C bond energy in hydrocarbons is decreased. Then, the C-C thermal stability sequence of SAR fractions is found to be saturates > aromatics>resins. This means that resins are more prone to initiate the chain initiation reaction [ie, Reaction (1)] under the same conditions. Saturates produce the least free radicals due to the high bond energy. Therefore, the LTO activity of the SAR fractions is in this order: resins > aromatics > saturates. The temperature rise can not only promote the chain initiation reaction and increase the initial amount of free radicals, 53,54 but also accelerate the chain propagation reactions and promote the formation of hydrocarbon peroxides. Therefore, the oxidation reaction intensity increases with the increase of temperature in the DSC tests.
The production of oxygenated hydrocarbons during a LTO process can be explained by the chain degeneratebranching reactions and chain termination reactions. As found in the FTIR tests, the oxygenated hydrocarbons include alcohols, aldehydes, ethers, and carboxylic acids, and these products are related to Reactions (5), (7), (9), (11), and (12). H 2 O is a by-product of these reactions. The decarboxylation reaction (ie, Reaction (10)) is the main source of CO 2 in the LTO. As the amount of carboxylic acids formed by chain degenerate-branching reaction increases, the intensity of decarboxylation increases, and the CO 2 production increases gradually. Since resins contain a large number of aromatic rings and alicyclic rings, resulting in a strong polarity and can react with oxygen more easily, 10 the production of oxygenated hydrocarbons and CO 2 in LTO of resins is obvious and the reaction intensity of resins is higher than that of saturates and aromatics.
The content of heavy components is increased after the LTO of crude oil, 10,55 which can be explained by Reaction (13). The combination of two high-molecular-weight free radicals produces hydrocarbon with higher molecular weight, which is an important mechanism for the increase of heavy components in crude oil after the LTO reaction.

| CONCLUSIONS
1. Oxygen addition reaction and bond scission reaction exist simultaneously in the LTO of light oil. The former (1) R s -R l → R s ⋅ + R l ⋅ (2) R s -CH 2 ⋅ + O 2 → R s -CH 2 OO⋅ (3) R s -CH 2 OO⋅ + RH → R s -CH 2 OOH + R⋅ (4) R s -CH 2 OOH → R s -CH 2 O⋅+⋅OH (5) R s -CH 2 O⋅ + RH → R s -CH 2 OH + R⋅ (11) R s -CH 2 ⋅ + ⋅OH → R s -CH 2 OH (12) R s -CH 2 O⋅ + R⋅ → R s -CH 2 -O-R (13) R s ⋅ + R s �⋅ → R s -R � s can occur when the temperature is higher than 50°C, and it shifts to the latter as temperature rises. Saturates have a lower transition temperature than aromatics and resins. 2. The LTO products of crude oil include H 2 O, CO 2 , carboxylic acids, alcohols, phenols, and ethers. Saturates, aromatics, and resins are all the sources of the produced H 2 O, CO 2 , alcohols, and carboxylic acids, whereas ethers are mainly derived from aromatics and resins. 3. The LTO products of aromatics and resins are important ways of oxygen consumption and heat generation at the initial stage of air flooding. Subsequently, the oxidation of saturates gradually dominates the air flooding process with an increase in the reservoir temperature.