Identification and Implications of Trimethyl-n-Alkylbenzenes in Marine Oils from the Deep Tarim Basin

State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou 510640, China College of Chemistry, Chemical Engineering and Biotechnology, Donghua University, Shanghai 201620, China Tarim Oilfield Company, PetroChina, Kuerle 841000, China College of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing 100049, China


Introduction
Aryl isoprenoids (AIPs) with various trimethyl substitution patterns, e.g., 2,3,6-, 2,3,4-, and 3,4,5-trimethyl, exist in sediments and oils and are thought to be the diagenetic products of pigments of photosynthetic sulfur bacteria or algae [1][2][3][4][5][6][7][8]. The source of these novel monoaromatic compounds can be determined based on their carbon isotopic signatures [2,[5][6][7][9][10][11], and the depositional environment for the oil and the sediment source can be characterized. For example, AIPs that are characterized by a significant enrichment in 13 C suggest a major contribution from green sulfur bacteria Chlorobiaceae and can be used as an indicator for photic-zone anoxia. These compounds can also be bound into geological macromolecules, such as kerogens, resins, and asphaltenes [5,6,12,13]. The AIR ratio (the proportion of short-chain C 13-17 versus intermediate-chain C [18][19][20][21][22] AIPs) in the sediments has been used to indicate the degree and persistence of photic-zone anoxia [14]. The source of marine oils from the Tarim Basin of China has long been in a debate, due to both multiple source strata and resulted multiple oil charging processes and several secondary oil alteration processes [15]. The detection of AIPs with 2,3,6-trimethyl substitution in the oil and highly abundant 1,2,3,4-tetramethylbenzene (TTMB) in oil asphaltene pyrolysates, along with the carbon isotope values of TTMB, has suggested a euxinic depositional environment of the oil source [12,13,16].
Previous studies have also noted that the relative abundances of AIPs are sensitive to the maturity of sediments and oils. Ratios of C 30 or C 18 AIPs with 2,3,6-trimethyl substitution pattern relative to C 40 diaromatic carotenoids show a good correlation with the C 17 n-alkane/pristane (n-C 17 /Pr) and C 18 n-alkane/phytane (n-C 18 /Ph) ratios that increase progressively with elevated oil maturity, and their abundances decrease rapidly after the source rocks entered the oil window [4,7]. The abundances of AIPs in oils decrease with the maturity [7,17], and the short-chain homologues of AIPs become more pronounced with increasing maturity compared with the long-chain homologues [3,16,17]. Moreover, the diagnostic products released from the flash pyrolysis of these structures bound to the kerogens, namely, 1,2,3,4tetramethylbenzene (TTMB), is less abundant at the relatively higher maturity level, and the dominance of TTMB in C 4 alkylbenzenes will no longer be observed on the pyrolysates of type II kerogens when their Ro values are greater than 0.7% [18]. A similar result was also reported in the aromatic fraction of oils from the Tarim Basin [17].
In this study, a series of alkylbenzenes has been identified from marine oils in the Tarim Basin along with AIPs, through further examination of the m/z 133+134 mass chromatograms of aromatic fractions of the previously studied oils [19]. The regular gas chromatographic (GC) elution pattern and mass spectra of these alkylbenzenes combined with the coinjection of synthesized reference compounds in the aromatic fraction of one oil sample allowed us to assign this series of alkylbenzenes as 2,4,5-trimethyl-n-alkylbenzenes with coelution of 2,3,5trimethyl-substituted isomers. Variations in the ratios of these trimethyl-n-alkylbenzenes to AIPs show a similar tendency to that in the ratios of n-alkanes to isoprenoids and the AIR ratio, which has been attributed to the oil maturity effect. These results are further supported by an oil asphaltene pyrolysis and thus reinforce the critical influence of thermal maturity on the source-indicating parameters of deep oils, which is worthy of a full consideration for the oil source determination of this basin.

Geological Background
The geological background and details of the investigated 34 oil samples are given by Jia et al. [19]. Briefly, the Tarim Basin in northwest China is one of the largest oil/gas-prone basins in China. Two source units were identified in the basin including the Cambrian-Lower Ordovician and Middle-Upper Ordovician strata. The source and genesis of marine oils have attracted the attention of Chinese organic geochemists over the past two decades [16,17,[20][21][22][23][24][25][26][27][28][29][30] due to the multiple hydrocarbon charge events related with two sets of source rocks. The Cambrian-Lower Ordovician strata were interpreted to be the main source rocks using a detailed oilsource correlation [15]. This finding is supported by recent exploration advances of ultradeep Lower-Middle Cambrian reservoirs in the basin [31][32][33].
Marine oils in this basin were discovered mainly in Ordovician carbonate and Carboniferous clastic reservoirs, and some were found in Silurian, Triassic, and Jurassic clastic strata. The oils show great diversity of physical properties, and heavy, normal, waxy, and light oils were all found in the production zone. The studied oil samples have various types and were collected from reservoirs of different ages in the two main tectonic units (Tabei and Tazhong uplifts). Geochemical and isotopic characteristics of the oils show large variations, and these samples were classified into two distinct groups using biomarkers, carbon, and hydrogen isotopic compositions of n-alkanes [19].

Experimental Techniques
Routine procedures for the molecular and compoundspecific isotopic analysis of oil can be found in Jia et al. [19]. One oil asphaltene sample has been isothermally pyrolyzed for 72 h at five different temperatures (260-380°C) in vacuumed glass tubes, and the variations in the hydrogen isotopic compositions of released n-alkanes have been investigated by Jia et al. [34].
The trimethylhexylbenzenes were synthesized by Friedel-Crafts acylation with 1,3,5-, 1,2,4-, and 1,2,3-trimethylbenzenes as the starting material, followed by reduction with sodium borohydride. Briefly, hexanoic acid was used to react with thionyl dichloride to form hexanoyl chloride, which was then reacted with three isomers of trimethylbenzene catalyzed by AlCl 3 to yield trimethyl-substituted hexanophenones. The trimethyl-substituted hexanophenones were reduced to trimethyl-substituted n-hexylbenzene by sodium borohydride in the presence of AlCl 3 in tetrahydrofuran. The trimethyl-n-hexylbenzene isomers were purified by flash chromatography and characterized by 1 H-nuclear magnetic resonance ( 1 H-NMR, Bruker AM-400 spectrometer) with tetramethylsilane as an internal standard in deuterated chloroform.
Aromatic fractions of a few selected oil samples were chromatographed using Al 2 O 3 (100-200 mesh) to eliminate possible influences from polycyclic aromatic hydrocarbons on the alkylbenzene mass spectrum. The monoaromatic fractions were eluted with n-hexane/dichloromethane mixture (99 : 1, v/v) and analyzed using a Thermo DSQII mass spectrometer to which a Trace ultra GC was interfaced, and a 60 m HP5 capillary column (0.32 mm i.d. and 0.25 μm film thickness) was used for the separation of organic compounds. Analysis was conducted using both selective ion monitoring (SIM) and full scan detection.

Results and Discussion
4.1. Occurrence and Mass Spectral Characteristics. Typical AIPs (see Figure 1) detected in the Tarim marine oils were confirmed to have a 2,3,6-trimethyl substitution pattern by comparing them with those in oils from the Peace River Basin of Canada [35] (kindly provided by Prof. R.E. Summons). This series of compounds can be recognized easily 2 Geofluids by the unusual low abundance of the C 17 homologue and increased intervals of the retention time between the C 14 and C 15 homologues and between the C 19 and C 20 homologues, which is a result of irregular isoprenoid substitution on the aromatic ring [2]. The total amount of AIPs has been shown to decrease with elevated oil maturity, whereas the relative abundances of short-chain homologues increase with oil maturity or decreasing oil density [16,17]. For the investigated 34 oil samples, abundant AIPs were detected frequently in heavy and normal oils, whereas they were detected at trace levels in waxy or light oils except for the C 13 homologue ( Figure 1). However, some normal oils from the Tazhong uplift also display low relative abundances of AIPs, e.g., oil TZ10C ( Figure 1). Other pseudohomologues of the long-chain alkylbenzenes, abbreviated here as aryl alkanes (AAs, Figure 1), are visible in the m/z 133+134 mass chromatograms. In general, they show high relative abundances in waxy and light oils, which is opposite to the variations in the relative abundances of AIPs. This newly detected series of alkylbenzenes elutes at regular intervals of retention time in the chromatograms, which suggests that an n-alkyl side chain, and not an isoprenoid, connects to the benzene ring ( Figure 1). n-Alkylbenzenes and branched alkylbenzenes [36][37][38] (also termed phenyl alkanes) and 1,2-di-n-alkylbenzenes [39] were also reported to elute at regular retention time intervals. The reported phenyl alkanes in the sediments and oils generally have a single-branched alkyl side chain on the benzene ring, and no additional methyl groups have been found on the benzene ring [36][37][38]. Therefore, except for the 2-phenyl alkanes that show a base peak at m/z 105 in their mass spectra, other isomers of phenyl alkanes are characterized by a base peak at m/z 91 in their mass spectra. 1,2-Di-nalkylbenzenes show a base peak at m/z 105 in their mass spectra, except for ethyl-n-alkylbenzenes that show a base peak at m/z 119 [39]. Two isomers of C 15 trimethylhexylbenzenes detected from a coal-bed wax show mass spectra with a base peak at m/z 133 [40], and C 22 2-ethyl-4and 2-ethyl-6-methyl-n-alkylbenzenes also show a base peak at m/z 133. Then, Guthrie [9] identified 2-ethyl-4/6-methyln-alkylbenzenes from the Ordovician shale samples containing abundant G. prisca. They show mass spectra with a base peak at m/z 133 and also highly abundant fragment ions at m/z 105 and 119. These two series of compounds are characterized by high abundances of C 20 and C 22 homologues, which may be related to the contribution from G. prisca. The newly detected AAs from the Tarim oils can therefore be trimethyl-n-alkylbenzenes, as indicated by the base peak at m/z 133, the very low relative abundances of the fragment ions at m/z 105 and m/z 119 in their mass spectra ( Figure 2(a)), and by the regular intervals of elution time ( Figure 1).
AAs could be distinguished from the AIPs by a much lower relative intensity of the m/z 134 ion in their mass spectra ( Figure 2). An elevated relative intensity of the m/z 134 ion was observed in the mass spectra of AIPs with increasing length of alkyl side chain, which results from easy γ-H transfer and McLafferty rearrangement [2,3]. Consistent with these previous results, the mass spectra of AIPs obtained in this work display an elevated relative intensity of the m/z 134 ion with increasing carbon numbers, and the m/z 134 ion becomes the base peak on the mass spectrum of C 20 AIP (Figure 2(b)). A progressively elevated intensity of the m/z 134 ion with increasing side chain length was also observed in the mass spectra of AAs, which is similar to AIPs. However, the m/z 133 ion is always the base peak for AAs from C 13 to C 20 . The nalkyl side chain linked to the benzene ring can suppress the γ-H transfer and McLafferty rearrangement relative to the isoprenoid side chain. The methyl position on the benzene ring also affects the relative intensity of fragment ions of trimethyl-alkylbenzenes. Methyl substitution on the para-and ortho-position can suppress the hydrogen transfer reaction [2], and accordingly, two isomers of AIPs with methyl on two meta-positions, i.e., 2,3,5-trimethyland 3,4,5-trimethyl AIPs, have the m/z 134 ion as their base peak [2,5]. By contrast, the base peak at m/z 134 was not observed in the mass spectra of AAs (Figure 2(b)). Therefore, the characteristic ion at m/z 133 appeared in the mass spectra of these trimethyl-n-alkylbenzenes in previous studies and this work may result from the n-alkyl side chain and the position of methyl substitution on the benzene ring.

4.2.
Distributions of n-Alkylbenzenes, Mono/Dimethyl-n-Alkylbenzenes, and AAs. Long-chain n-alkylbenzenes, together with methyl-and dimethyl-substituted n-alkylbenzenes, are one of the most common series of aromatic compounds identified in sediments and oils [  3 Geofluids three series of compounds could be detected in both oil samples containing abundant AIPs and those containing AAs ( Figure 3). For methyl-n-alkylbenzenes, the meta-and ortho-methyl isomers were detected in much higher abundance relative to the para-methyl isomer (Figures 3(b) and 3(f)). Among the six isomers of dimethyl-n-alkylbenzenes, the   4 Geofluids relative abundances of 3,5-, 2,5-, 2,4-, and 3,4-dimethyl isomers are much higher than those of 2,3-dimethyl isomer (Figures 3(c) and 3(g)). These results are consistent with those reported by Ellis et al. [43]. Several previous studies have detected trimethyl-n-alkylbenzenes in sediments and oils. For example, n-pentylbenzene elutes before all six trimethyl-ethylbenzene isomers (C 11 ), and the elution order for the six trimethylethylbenzene isomers was also reported for the pyrolysates of kerogen [5]. In this work, the C 13 homologue of the AAs elutes slightly later than the C 13 n-alkylbenzene, and with increasing carbon number, the difference in elution time between the AAs and n-alkylbenzenes decreases rapidly (Figures 3(e) and 3(h)). Individual long-chain 2,3,6-trimethyl-n-alkylbenzenes have been reported in coal, oils, and shales [41,44]. A series of long-chain trimethyl-n-alkylbenzenes were detected from coal-bed wax, but the precise structure of this series was not determined [40]. Up to now, the distributional characteristics of six possible isomers of long-chain trimethyl-n-alkylbenzenes (C 13+ ) are still not well understood, and the structures of the AAs in the Tarim oils thus need to be determined.
Since all ortho-and para-positions are substituted by methyl groups, the substitution pattern with 2,4,6-trimethyl can suppress the hydrogen transfer reaction and result in a more pronounced m/z 133 ion in the mass spectra of AIPs relative to methyl substitutions at the 2,4,5, 2,3,6, and 2,3,4 positions [2]. As expected, the synthesized 2,4,6-trimethyln-hexylbenzene in this work shows a slightly higher relative intensity of the m/z 133 ion compared with the 2,3,4-trimethyl-n-hexylbenzenes (Figures 4(a) and 4(c)). In contrast, AIPs with substitutions of 2,3,5-/3,4,5-trimethyl, with two meta-positions occupied by the methyl, show the most intense hydrogen transfer reaction and a base peak at m/z 134 [2,5]. Consistent with this observation, synthesized 3,4,5-trimethyl-n-hexylbenzene shows a much higher relative intensity of the m/z 134 ion relative to 2,4,6-/2,3,4-trimethyl-n-hexylbenzenes (Figures 4(a) and 4(c)). The mass spectrum of 3,4,5-trimethyl-n-hexylbenzene still shows a base peak at m/z 133 rather than at m/z 134, which is distinct from the AIPs with 3,4,5-trimethyl substitutions [5]. Therefore, the isomer obtained from the reaction of 1,2,4-TMB (Figure 4(b)), which shows a very low relative intensity of the m/z 134 ion, could be assigned as 2,4,5-trimethyl-n-hexylbenzene rather than 2,3,5-trimethyl-n-hexylbenzene. This assignment is consistent with the orientation of the Friedel-Crafts acylation. In 1,2,4-TMB, there is only one position ortho to one of the methyl groups and para to the other. Electrophilic substitution occurs primarily at this position to afford 2,4,5-trimethyl-n-hexylbenzene. The position between  Figure 5: m/z 133+134 mass chromatograms of the aromatic fractions separated from asphaltene pyrolysates at different temperatures. Filled circles indicate AIPs, and unfilled triangles indicate newly identified AAs. The mass chromatogram of the pyrolysates obtained at 260°C, which is almost the same as those at 280°C and 300°C, was not shown here. values. The C 13-16 AA/AIP ratio represents the sum of C 13-16 AA peak area divided by that of C [13][14][15][16] AIPs; C 17-22 AAs were not included due to their identification with less confidence in the relatively low-temperature pyrolysates. The n-C 17 /Pr and n-C 18 /Ph ratios were extracted from our previous study [34]. 6 Geofluids the two methyl groups (ortho to both) is hindered sterically which excludes the formation of 2,3,6-trimethyl-n-hexylbenzene. The formation of 2,3,5-trimethyl-n-hexylbenzene could also be excluded because the position ortho to one of the methyl groups and meta to the two others in 1,2,4-TMB is less reactive for the Friedel-Crafts acylation. Except for the 2,4,6-trimethyl-n-hexylbenzene, three synthesized compounds were coinjected with the aromatic fraction of oil sample LN14O, and the m/z 133+134 mass chromatograms of the aromatic fraction before and after coinjection are shown in Figures 4(d) and 4(e). Coinjection results show that the C 15 AA is possibly the 2,4,5-trimethyln-hexylbenzene, and 3,4,5-and 2,3,4-trimethyl-n-hexylbenzenes are also present but in lower abundances relative to 2,4,5-trimethyl-n-hexylbenzene. Nevertheless, 2,4,5-trimethyl-n-hexylbenzene likely coeluted with 2,3,5-trimethyln-hexylbenzene, as suggested by the previous study [5]. The C 15 AA detected in oil samples shows a much higher relative intensity of m/z 134 ion in the mass spectrum (Figure 2(a)) than the synthesized 2,4,5-trimethyl-n-hexylbenzene (Figure 4(b)). Therefore, the C 15 AA in the oil could be the 2,4,5-trimethyl-n-hexylbenzene that coeluted with 2,3,5-tri-methyl-n-hexylbenzene, which made this series of isomer stand out relative to other isomers of trimethyl-nhexylbenzenes.

Pyrolysis of Oil Asphaltenes.
Variations in the relative abundances of alkanes and δD values of n-alkanes with increasing pyrolysis temperatures have been reported for the asphaltene pyrolysates [19]; in this study the trimethylalkylbenzene distributions were extracted from the pyrolysates (see Figure 5). The distribution of AIPs in the pyrolysates obtained at 280°C and 300°C is similar to that in heavy or normal oils shown in Figure 1, as indicating by relatively high abundance of C 18-22 AIPs ( Figure 5). The relative abundances of C [15][16][17][18][19][20][21][22] AIPs decrease at 320°C, and even the relative abundance of C 14 AIPs decreases significantly besides that of C [15][16][17][18][19][20][21][22] AIPs at 350°C. When the pyrolysis temperature was increased to 380°C, the AIPs could only be detected at trace level except the C 13 homologue, but the AAs predominate the trimethyl-alkylbenzenes in the pyrolysates. This feature is similar to those observed on the light or waxy oils in Figure 1. Nevertheless, other isomers of trimethyl-n-alkylbenzenes (especially C 13-18 homologues), like the 3,4,5-and  Figure 4, show notably higher relative abundance in the asphaltene pyrolysates than in oil samples. This discrepancy might result from the differences either in molecular composition between oil asphaltenes and aromatic fraction or in thermal maturation between laboratory and geological conditions.
One of the commonly used thermal maturity indicator (EasyRo) has been calculated according to the reported kinetic parameters [45] and the present pyrolysis procedure, and the variations in selected molecular ratio with thermal maturation level are shown in Figure 6. The relative abundances of short-chain AIPs, denoted by AIR ratio, increase progressively at EasyRo of approximately 0.8% (close to the oil peak), suggesting that long-chain AIPs began to crack secondarily at this point. By contrast, the relative abundances of trimethyl-n-alkylbenzenes (compared with that of aryl isoprenoids, denoted by the C 13-16 AA/AIP ratio) increase slightly at EasyRo of 1.1% (shortly after the oil peak) and then increase rapidly when the asphaltenes have evolved into the condensate/wet gas stage. This result is consistent with a dominance of AAs over AIPs in Figure 5 (380°C). Similar variation trends could also be observed on the relative abundances of n-alkanes compared with acyclic isoprenoids, commonly denoted by the n-C 17 /Pr and n-C 18 /Ph ratios (C 17 nalkane/pristane and C 18 n-alkane/phytane, respectively). This result suggests that the aryl isoprenoids may share a similar thermal evolution trend to the acyclic isoprenoids in the saturated fraction of oil, and this thermal evolution is mainly associated with the stability of isoprenoid side chain.

Variations in the AIPs and AAs in Oils.
The n-alkylbenzenes and their methyl-substituted series are generally thought to be originated from the cyclization and aromatization of the precursors possessing linear carbon skeletons as well as functionalities, such as the straight-chain fatty acids and alcohols [39,40,43]. During the sedimentation and diagenesis, these fatty acids and alcohols could be first reduced to α-carbonium ions that generate n-alkylbenzenes during the subsequent cyclization and aromatization [40]. If the H + migration or isomerization of carbonium ions has happened, the cyclization and aromatization products will be methyl-nalkylbenzenes or dimethyl-n-alkylbenzenes, respectively [40]. The methyl-n-alkylbenzenes in low-maturity oils are usually dominated by the isomer with the ortho-position substituted by methyl, i.e., 2-methyl-n-alkylbenzenes [39,43], which is consistent with its lower formation energy relative to two other isomers [43]. The oil pyrolysis experiments and natural maturation series of oil have further shown that the relative proportions of isomers of (di) methyl-n-alkylbenzenes depend on the thermal stabilities in the isomerization reactions, as a result of the elevated thermal stress and the catalysis by clay [39,43]. By contrast, the trimethyl-n-alkylbenzenes were detected only in a few cases [40], and the structures of two C 15 isomers were not determined in their study. Different nalkylbenzene classes in the waxy oil show similar distributions (Figures 3(e)-3(h)), e.g., the same carbon number range and no carbon number preference. These results may suggest that the relatively abundant AAs are the products of the cyclization, aromatization, and isomerization reactions of the precur-sors having linear carbon skeletons. However, one should bear in mind that the AAs are the coelution products of 2,4,5-and 2,3,5-trimethyl-n-alkylbenzenes, making them show higher relative abundances than other isomers (Figures 4 and 5).
In the above section, we have shown that the detected AAs are relatively abundant in light oils, waxy oils, and some normal oils, which is generally in contrast with the variation in relative abundances of AIPs in various types of oils ( Figure 1). To further explore the maturity effect on the relative abundance of AAs in the oils, the ratios of AAs to AIPs were calculated for the 34 previously investigated oil samples [19]. Biomarkers and isotopic compositions (C, H, and S) of two heavy oils, TZ62S and TD2⋲, have been shown to be different from those of most marine oils from the Tarim Basin, and these two oil samples were classified into a unique genetic group [19]. Up to now, their origin and genesis are still not well understood, possibly due to a hydrothermal effect [15]. Therefore, they were not included in the following discussion. Considering the structure similarity, the n-C 17 /Pr, n-C 18 /Ph, and AIR ratios were used as maturity indicator for oils. It has been demonstrated that the n-C 17 /Pr ratio exhibited excellent linear correlation with the n-C 18 /Ph ratio of the majority of marine oils from the Tarim Basin, suggesting a maturity trend of the variations in these ratios [17,19]. In addition, the ratio of short-chain (C 12-17 ) to long-chain (C 18-23 ) AIPs, which is similar to the AIR ratio, has been shown to follow a positive correlation to the increasing maturity [17].
Generally, the C 13-16 AA/AIP ratio increases linearly with elevated n-C 17 /Pr, n-C 18 /Ph, and AIR ratios, thus defining a clear maturity trend (Figures 7(a)-7(c)). The light and waxy oils overall have much higher ratios than heavy and normal oils, which is also consistent with the maturity trend of oil density. A few normal oils from the Tazhong uplift have relatively high ratios, which may be indicative of relative high maturity or a mixing of oil from different maturity levels [19]. Notably, a few light oils (condensates) from the Tazhong uplift (such as TZ452O, TZ261O, and TZ85O) possess higher C 13-16 AA/AIP ratios than most of study samples, making them deviate upwardly from the maturity trend (Figures 7(a)-7(c)). The exact reason for this discrepancy is not clear yet based on the available data. The mixing of oils from different sources might be one explanation. For example, the light oil TZ452O has been interpreted to be the mixing origin as indicated by its biomarker parameters [19]. However, the source of the light oil TZ85O is still obscure, because of its similarity of both carbon isotopic and biomarker characteristics between those of the majority of heavy and normal oils. Other secondary factors like the thermal sulphate reduction and biodegradation may be possible but need a further investigation.
A few waxy oils (such as TZ721O, TZ83O, and LN634O) have lower AIR ratios than most study samples, causing their deviating downwardly from the maturity trend (Figure 7(d)). The migration fractionation [46], such as gas washing of precharged oil and later fluid phase changes, is possibly responsible for the deviation. Waxy oils formed by this way are relatively depleted in short-chain hydrocarbons, leading to lower AIR ratios of waxy oils than condensates. By contrast, 8 Geofluids the C 13-16 AAs/AIPs, n-C 17 /Pr ratio, and n-C 18 /Ph ratios, representing relative abundances of compound classes with different thermal stabilities but similar elution time (evaporation temperature), could thus be much less affected by the fractionation process.

Conclusions
A series of trimethyl-substituted alkylbenzenes has been detected in various types of marine oils from the Tarim Basin. Based on their GC elution patterns, mass spectra, and coinjection with synthesized compounds, this series of compounds has been assigned as 2,4,5-trimethyl-n-alkylbenzenes that coeluted with 2,3,5-trimethyl-n-alkylbenzenes.
Other isomers of trimethyl-n-alkylbenzene were also detected. The relative abundances of this series of trimethyl-n-alkylbenzenes are usually much higher in the light and waxy oils than in the normal and heavy oils, which is opposite to the variations in relative abundances of AIPs in the oils. Detailed examinations reveal that the ratios of AAs to AIPs of studied oils generally show good correlations with molecular maturity indicators. This maturity trend could also be observed on the pyrolysates of oil asphaltenes with increasing pyrolysis temperature. Therefore, much higher abundance of AAs in these deep oils than those of AIPs can be primarily attributed to higher thermal stability of n-alkyl side chain than that of isoprenoids one attached to benzene ring. In addition, the coelution of the two isomers also makes them stand out from the trimethyl-n-alkylbenzenes. The AA and AIP deviations of some condensates and waxy oil molecular ratios are likely influenced by other factors like oil mixing and migration fractionation, which may have influenced the relative abundances of AAs and AIPs. Therefore, the effects of thermal stress and other secondary alteration processes on molecular source indicators should be given a full consideration for tracing the source of deep and ultradeep oils.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper.