key: cord-0894182-u0p5vf4d authors: Xu, Lei; Yan, Fuwu; Wang, Yu title: A comparative study of the sooting tendencies of various C5-C8 alkanes, alkenes and cycloalkanes in counterflow diffusion flames date: 2020-10-15 journal: nan DOI: 10.1016/j.jaecs.2020.100007 sha: 03cf31d8bc66f0ea6e232d7e15f9c8c4845d75b9 doc_id: 894182 cord_uid: u0p5vf4d In this work, sooting tendencies of various vaporized C5-C8 alkanes, alkenes, cycloalkanes were investigated in a counterflow diffusion flame (CDF) configuration. Mie scattering and laser induced incandescence (LII) were respectively employed to measure the sooting limits and soot volume fraction, both of which were used as quantitative sooting tendency indices of the flames. Research foci were paid to analyze the impacts of fuel molecular structure on sooting tendencies, ranging from the length of the carbon chain, fuel unsaturation, presence and position of the branched chain as well as the cyclic ring structure. It was found that the present CDF-based sooting tendency data did not always agree with existing literature results that were obtained from either coflow or premixed flame experiments, indicating the importance of considering flame conditions when assessing fuel sooting tendencies. Several interesting observations were made; in particular, our results showed that the sooting limits of C5-C8 n-alkane is comparable, regardless of the carbon-chain length, indicating an interesting fuel similarity pertinent to sooting tendency. In addition, we found that cyclopentane, with a five-membered ring, has even stronger sooting propensity than the six-membered cyclohexane; this trend occurs because cyclopentane prefers to decompose to more odd-carbon radicals of cyclopentadienyl and allyl, which are efficient aromatic precursors. Our results are expected to serve as necessary complements to existing sooting index data for a deeper understanding of the correlation between fuel molecular structures and their sooting tendencies. Emission of soot particles originating from incomplete combustion of hydrocarbon fuels are known to negatively affect our living conditions. The adverse effects include, but are not limited to, global warming [1, 2] , air pollution [3] , and human health issues. Perhaps not surprisingly, there is recent evidence that suggests air PM pollution is positively correlated with both the transmission efficiency of SARS-CoV-2 and the case fatality rate of patients who contracted this virus [4, 5] . These negative issues pertinent to soot emissions necessitate the development of efficient soot mitigation techniques, which in turn requires a deep understanding of the soot formation processes. Decades of research has shown that soot formation involves rather complicated physicochemical processes ranging from gas-phase/surface chemistry to particle dynamics [6] [7] [8] [9] . Considerate research efforts have been paid to understand the driving mechanism of soot formation [10] [11] [12] and its dependence on various combustion parameters such as pressure [13] [14] [15] [16] , temperature [15, 17, 18] , strain rate [13, 19, 20] and fuel molecular structures [21] [22] [23] [24] . Due to the diversity of fuel types and the complexity of detailed fuel chemistry, the relationship between fuel structure and soot formation is probably the least understood among the above-mentioned parameters. Nevertheless, it has been well established that fuel structure is a critical factor in affecting soot formation process and thus the fuel's chemical propensity to soot (i.e., sooting tendency). Fundamentally, different fuels would produce different pools of intermediate radical products, which would notably affect the formation of molecular soot precursors as well as subsequent soot growth processes [21] . Obviously, the sooting tendency of a particular fuel has a direct impact on the soot emissions of practical combustion devices utilizing the very fuel [25] . The needs to design clean fuels and to develop low-emission combustion techniques motivate research efforts to establish robust quantitative sooting indices for different fuels. Since measurements in practical devices are apparatus-dependent and fuel-consuming [26] , researchers typically chose to develop sooting indices based on well-controlled laboratory flames. Existing soot indices include the critical equivalence ratio (or C/O ratio) where soot starts to appear in premixed flames [27] or in a micro flow reactor (with maximum temperature below 1500 K) [28, 29] , the smoke point (SP) and threshold soot index (TSI) as determined from the critical flame height at which soot begins to emit from the flame tip in a wick-fed lamp or coflow burner [30] [31] [32] , and the yield sooting index (YSI) based on the measurement of peak soot volume fraction in a methane/air coflow diffusion flames by doping small amounts of target fuel [33, 34] . Several numerical predictive models pertinent to YSI were also developed [25, [35] [36] [37] [38] . Moreover, Barrientos et al. [39] extend the TSI methodology and developed oxygen extended sooting index (OESI) to capture the sooting tendency of oxygenated fuels; Lemarie et al. [40] proposed the fuel equivalent sooting index (FESI) to characterize sooting tendency in turbulent spray flames. These available sooting index data (e.g., YSI and TSI) significantly expand our knowledge and deepen our understanding on fuel sooting propensities. Nevertheless, it is important to note that flame configuration also has a role to play in determining the relative ranking in sooting tendencies among different fuels, which is understandable considering the whole processes of fuel decomposition, formation of molecular soot precursors as well as soot evolution are highly dependent on flame thermal and flow conditions. The widely-used YSI and TSI indices were developed based on either coflow or smoke point flame; these flames behave like a soot formation/oxidation (SFO) flame [41, 42] where soot once formed (in the fuel-rich region) will be transported to the high-temperature oxidizing reaction zone to get oxidized. Therefore, the TSI or YSI for a specific fuel essentially represents the net effects of soot formation and oxidation. However, it may be sometimes useful to compare fuel's sooting propensity in the absence of soot oxidation. For instance, pure fuel endothermic pyrolysis conditions also practically occurs in engines for powering hypersonic aircrafts or rockets [43] , serving as heat sink to cool the engine structures/subsystems [43, 44] . However, the unexpected formation of carbonaceous deposits (cooking) also occurs in the meanwhile, becoming the major limitation of the implementation of fuel cooling technologies [45] [46] [47] . In this aspect, by focusing attention on fuel pyrolysis/soot formation process may provide practical guidance for understanding and resolving the coking issue occurring in the pure fuel pyrolytic condition. In addition, it is the authors' opinion that the sooting tendency results obtained in soot inception/formation process (without soot oxidation) also provided attractive complements to YSI data in several aspects. First, since sooting limit represents a critical flame condition where soot starts to appear, it provides a direct measure to predict whether a fuel can form soot under given flame conditions. With a comprehensive knowledge of sooting limits, it is possible to control soot formation at the very beginning and develop soot-free combustion techniques. Second, understanding fuel's sooting propensity in separated soot evolution stages (i.e. soot inception, growth or oxidation) also help to determine which stage is the critical factor in determining the final soot emissions, thus benefiting the design of soot-inhibiting techniques. Third, from scientific viewpoints, understanding the correlation between sooting tendency and fuel types in different environments may also enrich our fundamental understanding on the soot formation mechanism. In such case, it would be beneficial to conduct sooting tendency studies using a counterflow flame configured as the soot formation (SF) type, in which soot particles incepted in the fuel-rich zone continue to grow while being transported away from the oxidizing reaction zone so that oxidation of soot particles are not present. Note the flame and sooting structures of CDFs of both the SF and SFO types have been detailed in a recent review [8] and are thus not elaborated here. Additional merits of using CDF to study sooting tendencies include the flexibility to study strain rate effect [19] , and the ease of data analysis thanks to the quasi-1D nature of CDFs [8] . Therefore, it is our belief that the CDF-based sooting index could serve as necessary complements to the coflow-based data. Although not abundant, several sooting tendency studies have been performed in CDFs, through the measurement of either peak soot volume fraction [48] [49] [50] or critical sooting condition (i.e. sooting limit) [51, 52] . In a previous study [19] , we also investigated the sooting tendencies of various C 1 -C 4 gaseous fuels in CDFs, and proposed an alternative sooting index for rating sooting tendency of different fuels, i.e. the sooting temperature index (STI) defined as the critical flame temperature at sooting limit condition; a sooting sensitivity index (SSI) was also introduced to describe the strain rate effects on sooting tendency. In the present work, we intend to investigate the sooting tendency of C 5 -C 8 alkanes, alkenes and cycloalkanes in CDFs. Petroleum-derived fuels generally contain a wide range of hydrocarbons featured with different chemical functional groups, such as linear and branched alkanes, alkene, cycloalkanes and aromatics [53] [54] [55] . Studying the relationship between fuel structure and its sooting tendency for different hydrocarbons is therefore of practical significance. While McEnally and coworkers have performed a comprehensive YSI measurements for various alkanes, alkene, cycloalkanes, aromatics, alcohols and esters in coflow flames [25, 34, 56, 57] , the relevant sooting tendency information are rather scarce in CDFs, especially for heavier liquid fuels (e.g. with carbon atom larger than 5). As will be detailed later, our CDF-based results do not always agree with the YSI data, suggesting the necessity of performing sooting tendency studies in various complementing thermochemical conditions. The present work is focused on the understanding of the effects of fuel structure on sooting tendency, ranging from the effects of carbon-chain length (comparing C 5 -C 8 n-alkanes), fuel unsaturation (comparing n-hexane and 1-hexene), isomers effects (comparing selected C 6 H 14 isomers of 2-methylpentane, 3-methylpentane, 2,3-dimethylbutane) and the cyclic ring structure (comparing cyclopentane, cyclohexane and methylcyclohexane). Numerical modelling studies were also performed to help understand the kinetic mechanism that lead to the experimental observations. In the following sections, we shall firstly introduce the experimental setup of light scattering and laser induced incandescence (LII) for the measurement of sooting limits and soot volume fraction, respectively. Subsequently, the numerical approach for kinetic modelling was briefly described. Results on the sooting tendency trend of C 5 -C 8 n-alkanes, selected C 6 species and cycloalkanes were then discussed in sequence. Finally, major findings of this work were highlighted. The counterflow burner consists of two vertically placed nozzles; each nozzle has a diameter of 10 mm and the separation distance between them was 8 mm. The fuel and oxidizer stream were introduced from the lower and upper nozzles, respectively, with the same nozzle exit velocity (V 0 ). Nitrogen was provided as the shield gas to minimize the disturbance of surrounding air. The flow rates of gaseous species (including N 2 and O 2 ) were controlled by calibrated thermally-based mass flow controllers (MKS GE50A), while those of C 5 -C 8 liquid fuels were controlled by a high-precision syringe pump (KDS 230); appropriate dynamic range of these controllers were chosen for sufficient resolution and accuracy. The liquid fuel was firstly evaporated in a commercially evaporation system with N 2 as the carrier gas and was then transferred to the bottom burner, which, along with the whole fuel transfer lines, was heated to avoid fuel re-condensation. The initial temperature of fuel and oxidizer stream were controlled at 473 K and 293 K, respectively. A light scattering/extinction system was set up to obtain the sooting limits (i.e., critical conditions for soot formation in terms of fuel and oxygen mole fraction, to be detailed later). A more complete description of the optical layout can be found elsewhere [19] , while a schematic is shown in Fig. 1 . As can be seen, a 514.5 nm Argon-ion laser was used as light source. The laser beam was firstly expanded before being refocused to reduce beam diameter at the flame region, contributing to improve the spatial resolution of the measurements. A photodiode (Thorlabs, DET100A) assembled with an integrating sphere was used to detect the attenuated beam for extinction measurement, and a photomultiplier tube (PMT) placed at 90° to the incident beam path was employed to detect the scattering signal. The convex lens, linear polarizer, pinhole (50 μm) and narrow band filter (FWHM =1 nm) in front of the PMT were used to direct the scattered light to the PMT and reject background illumination. A lock-in amplifier (SRS SR830) with the reference signal provided by a mechanical chopper (SRS SR540) was used to condition the scattering signal for efficient noise rejection. Note the scattering setup has been confirmed to be free from interference of spurious light reflected from other surfaces. Sooting limits in this work are defined as the critical X F /X O (defined as the fuel/oxygen mole fraction in the N 2 -diluted fuel/oxidizer stream) where soot starts to appear, determined by comparing the Raleigh scattering signal from the gas molecules and Mie scattering from the soot particles [19, 52] . The procedure was illustrated in Fig. 2 where experimental scattering intensity profiles were shown for n-hexane flame at (a) X F = 0.8 with different X O and (b) X O =1.0 with different X F ; the scattering intensity reported in Fig. 2a and b were all obtained by multiplying the raw PMT signal intensity with the same scaling factor. Figure 2a shows that at the non-sooting condition (X O = 0.28), the Raleigh scattering intensity shows a decreasing trend from the fuel to the oxidizer side; this trend occurs because the fuel molecules have larger scattering cross-section than the oxidizer gases. When soot starts to appear (X O = 0.285), a local peak appears due to the larger Mie scattering from soot (than the gas molecules). With further increase in X O (X O = 0.29), the Mie scattering intensity start to increase rapidly. Thus, the sooting limit is defined as X O,cr = 0.285. Such case corresponds to a SF flame where flame region is located on the oxidizer side of stagnation plane. The formed soot will be transported towards the stagnation plane, away from the high temperature flame region zone with little soot oxidation; Similarly, for a fixed X O of 1.0, the sooting limit of X F,cr =0.135 can be determined from Fig. 2b . This kind of flame can be classified as of SFO type for which the flame region was located on the fuel side of stagnation plane, meaning that soot particles, after being incepted, will be directed towards the flame region and get oxidized. Since the interval of X F or X O is 0.5% (0.005), the experimental uncertainty of sooting limits (in terms of X O,cr and X F,cr ) is expected to be within ±0.25%. All sooting limits were measured at the nozzle velocity of V 0 =45cm/s. It is worthwhile to point out here that by varying the fuel/oxygen mole fraction (X F or X O ) or fuel type, the strain rate would also be changed due to the variation of fuel/oxidizer stream density [58] . Nevertheless, we have confirmed that the variation is rather small and has negligibly small effects on the observed difference in sooting tendency among tested fuels. Besides sooting limits, LII measurements were also performed to compare sooting tendency using the index of peak in-flame soot volume fraction (SVF). This yield-based sooting index may provide complementary information to the above-discussed sooting limits, as the latter is more relevant to the limiting soot onset conditions. Regarding the LII setup, a 10 Hz pulsed Nd-YAG laser with fundamental emission (1064 nm) was used [50, 59, 60] . The laser beam was manipulated by a series of cylindrical lens to form a laser sheet (8 mm high) at the burner center. The LII signals were collected by an intensified CCD camera after going through a band-pass filter with a center wavelength of 400 nm (FWHM = 40nm). The camera gate open time was synchronized to the incident laser and the gate width was set to 80 ns. The LII signals were averaged from 600 laser shots to improve signal to noise ratio, with an estimated measurement uncertainty of less than 5%. LII signals were calibrated by performing additional light extinction measurements with a near-infrared laser beam (λ = 980 nm). The purpose of using a near-infrared laser wavelength is to minimize the interference of light absorption by PAHs, since PAHs could absorb light in the UV-vis spectral range. In deriving the absolute SVF, we used the soot refractive index value of m =1.570.56i [42, 61] . And, we performed Tomographic inversion [62] to obtain the local light extinction coefficient K ext . More details concerning the data processing procedure can be found in our previous studies [42, 63] . In our setup, the calibration factor was 10,000 LII signal counts to an SVF of 0.363 ppm. The systemic uncertainty of absolute SVF was estimated to be around 20%, which primarily originates from the uncertainty in the soot refraction index m [64, 65] . However, since this systematic uncertainty is the same for all the tested fuels, it has no effects on their difference in SVF. On the other hand, the random uncertainty of SVF was estimated to be less than 5% (through over 600 repeated LII measurements). The LII measurements were performed for all target fuels at identical flame condition of X F = 0.35 and X O = 0.3, with V 0 = 20 cm/s. The purpose of using a lower nozzle velocity (V 0 =20 cm/s) in LII measurements than that of scattering experiments (V 0 =45 cm/s) is to achieve moderately sooting flames for all the target fuels. Note since no cross-comparison was made between sooting limits and SVF, as they represent totally different sooting condition, the choice of nozzle velocity may be arbitrary for scattering or LII measurements. Kinetic modelling with detailed fuel pyrolysis and oxidation chemistry were performed to help understand the experimentally observed sooting tendency results. The main objective is to elucidate the fuel structure-dependent pathways of fuel decomposition and soot precursors (e.g., benzene) formation, which are expected to provide essential insights into the relationship between fuel structure and sooting propensity. This approach was rationalized that for aliphatic fuels such as those investigated in this work, benzene formation is thought to be the rate-limiting step to the subsequent polycyclic aromatic hydrocarbon (PAH) and soot formation process [66] [67] [68] . To this end, instead of simulating the experimental flames, all the simulations were performed with the Closed Homogeneous Reactor (CHR) model in CHEMKIN-PRO [69] for simplification. The simulations were set at thermodynamic condition (P=1 atm and T=1500 K) for mixture of 10% fuel and 90% Ar. In particular, the chosen temperature of T=1500 K is more closely to the diffusion flame temperature environment where fuel pyrolysis processes occurs, although the maximum flame temperature would be higher [70] . Such approach of focusing on the fuel pyrolysis and soot precursors formation pathway analyses in 0-D reactor and thus to provide kinetic insights into the observed sooting tendency trend in diffusion flames, has also been widely used by other researchers [71, 72] , as fuel pyrolysis processes play critical roles to the soot formation process of diffusion flames of the SF type. Since no single mechanism describing the decomposition chemistry of all the target C 5 -C 8 fuels exists, the chemical kinetic mechanism JertSurF 2.0 [73] was used for C 5 -C 8 n-alkanes, while the mechanism developed by Zhang et al. [74] and Tian et al. [75] were employed for selected C 6 fuels (hexane isomers/1-hexene) and cycloalkanes, respectively. Note we actually used the same kinetic model to analyze each individual fuel structure effect (carbon-chain length, isomers/fuel unsaturation or cyclic ring structure), that is to say, no comparisons were made between data obtained from different mechanisms. Note the present experiments and simulations were performed for all the fuels under the same temperature boundary. Although the different in fuel types would induce variation of flame temperature which may in turn affect soot formation, we have confirmed that the temperature difference among the tested fuels is rather small which is believed to have negligible small effects on their sooting tendencies. Therefore, we would like to attribute the observed ranking of sooting tendency to the difference in fuel structure. In this section, experimental data and kinetic analyses were first presented for C 5 -C 8 n-alkanes to understand the effects of carbon-chain length on sooting tendency for normal alkane fuels. Subsequently, the sooting tendency of selected C 6 fuels were compared to study the effects of fuel unsaturation, position and numbers of side chains. Finally, the effects of cyclic ring size and the presence of ring-attached side chains were discussed for selected cycloalkanes. Before comparing the sooting tendency trend of the C 5 -C 8 n-alkanes, we would like to elucidate the general features of the sooting limit map, which provides rich information for deep understanding on the sooting characteristics of counterflow flames. Firstly, for SF flame with large X F , the critical X o,cr increases quite slowly as X F decreases from X F =0.8 to 0.4, in such a way that the sooting limit curve has a nearly horizontal boundary. As X F further decreases to 0.2, the critical X o,cr starts to increase sharply. Secondly, in the SF flame regime, the sooting limit curve closely follows the adiabatic flame temperature (iso-AFT) line (T ad =2580 and 2890 K were shown, which were calculated for n-hexane using equilibrium calculation in CHEMKIN-PRO using the mechanism of JetSurF 2.0 [73] ). This suggests that flame temperature plays critical roles in determining the sooting limits of SF flames; The reason for that may be due to the lacking of soot oxidation process in SF flames, and therefore the temperature-dependent fuel pyrolysis and soot precursor formation processes take the dominating effects. On the other hand, for SFO flames, the sooting limit curves differ evidently from the iso-AFT line, as the sooting limits of SFO flames is also influenced by oxidation process, in addition to flame temperature. Thirdly, we noticed that the sooting limit curves bend slightly towards larger X F in the large X O regime. In other words, there exists two corresponding X O,cr for a given X F , indicating an non-monotonic variation of sooting tendency with X O . This interesting phenomena was previously observed elsewhere [19, 41, 52, 76] , and has been successfully explained in our previous work [42] . Briefly speaking, the observed non-monotonic variation of soot concentration with the increase of X O is caused by the competing effects of soot inception, soot surface growth, and soot oxidation process [42] . In particular, increasing X O would lead to the reduction of soot inception rates and the enhancement of soot oxidation, both of which tend to inhibit soot formation. However, increases of X O would also enhance the soot surface growth rate which, on the contrary, enhances soot formation. Generally, the fuel having stronger sooting propensity would have corresponding larger sooting region in the X F -X O plane [19] . However, it was found in Fig. 3 that the sooting limit curve boundary of the C 5 -C 8 n-alkanes is close to each other, especially in the soot formation flame region (large X F ). This indicates that these n-alkanes has comparable sooting tendency, being insensitive to the carbon-chain length. Presumably, this trend occurs because the dominating pathways and rates towards soot formation is similar among the C 5 -C 8 n-alkanes. This kind of fuel similarity pertinent to sooting tendency is rather interesting, as it may help to simplify the description and prediction of sooting tendency for heavier alkanes such as those C 8 n-alkanes. However, this conclusion cannot be projected to smaller alkane, as it has been reported in [19] that the sooting region of smaller C 1 -C 4 n-alkanes in the X F -X O plane was noticeably expanded with the increase of carbon atoms. The different sooting propensity among smaller alkanes (e.g. between methane and propane) may be due to the different pathways to form soot precursors and soot. This point will be further elucidated later. Considering the significant effects of flame temperature on sooting characteristics [17, 77] , it is interesting to further examine the relationship between the present sooting limits (X O,cr , X F,cr ) and flame temperature. To this end, we plotted the X F,cr and X O,cr in relation to the adiabatic flame temperature (T ad ) in Figure 4 . Note the value of T ad at corresponding (X O,cr , X F,cr ) is not necessarily the real temperature where soot just appears, but is used here as a quantitative indicator of ranking fuel's sooting tendency (as will be explained below). It shows in Fig. 4a that, the limiting temperature of SF flames at sooting limits condition remains nearly constant over a wide range of X F,cr ; this implies that there exists a limiting threshold temperature for soot onset (T cr,SF ) for SF flames: the flame will be sooting if it has a higher temperature than T cr,SF . In contrast, the X F,cr of SFO flames remains reasonably insensitive to the temperature. On the other hand, Fig. 4b shows an approximately linear dependence of log X O,cr on 1/T ad ; the flame temperature of SFO flames increases with increasing X O,cr . More importantly, it has been demonstrated [19, 52] that the limiting threshold temperature T cr,SF of SF flames is fuel structure-sensitive and, could be used as an quantitative sooting index. That is, the fuels with lower T cr,SF generally have stronger sooting propensity. In addition, it was suggested that [19] T cr,SF can be defined as the temperature at X F,cr = 1.0 for gaseous fuels or at a diluted X F,cr for liquid fuels, when a carrier gas is needed to achieve vaporization. Here, we artificially defined the T cr,SF as the temperature of SF flames at X F,cr = 0.6. In this regard, we may also conclude from Fig. 4a that the large n-alkanes have comparable limiting sooting temperature, especially for C 6 -C 8 n-alkanes (n-pentane has relatively smaller value). This temperature-based sooting index was in agreement with the sooting region in X F -X O plane, both supporting the fuel similarity in sooting tendency for large n-alkanes, regardless of their carbon-chain length. Nevertheless, the present CDF-based sooting limits seem to different from the coflow-based YSI data [34] , which was seen to increase steadily with carbon-chain length for C 5 -C 8 n-alkanes (n-pentane: YSI -8.9, n-hexane: YSI 0, n-heptane: YSI 8.7, n-octane: YSI 18.9). Such disagreement is likely due to the different sooting index used for ranking the sooting tendency. Note the present sooting limits essentially represent a critical condition where soot just appears, which is more relevant to soot inception process. In contrast, the YSI data is a measure of soot volume fraction in which soot growth and oxidation process (besides soot inception) also take important roles. To validate whether the choice of sooting index matters, we also performed additional LII measurements at identical flame condition of X F = 0.35 and X O = 0.3 (SF flames); the peak LII intensity was used as yields-based index for ranking fuel's sooting tendency (called LII-based sooting index), which represents the effects of both soot inception and soot growth process. The LII data shown in Fig. 5 confirm that longer n-alkane has slightly higher soot yields, being consistent with the YSI data. These observations suggest that the LII-based sooting index is more sensitive to carbon-chain length as compared to the sooting limits. However, it is the authors' opinion that the sooting limits may complement the YSI data in representing the fuel's chemical propensity to form soot, by focusing on critical sooting conditions and excluding interference from the soot oxidation process. [73] for mixtures 10% fuel+90% Ar at condition of T=1500 K and P= 1 atm. Fuel consumption rate is approximately 80%. The numbers indicate the percent contribution to the consumption of the source species. What remains unclear is why these large n-alkanes has comparable sooting limits (i.e. sooting propensity), in spite of the difference in carbon-chain length. To this end, mechanistic analyses were performed to focus on the fuel decomposition and benzene formation pathway. The simulation results on fuel pyrolysis process with the Closed Homogeneous Reactor model (see Fig. 6 ) showed that these large n-alkanes in the same chemical family essentially undergo similar decomposition and benzene formation process. Specifically, fuel consumption was initiated by the C-C fission and H-atom abstraction reactions; this process produces a wide range of n-alkyl radicals, which were subsequently decomposed to produce ethylene and other larger alkenes. These alkenes then grow (if the alkane was ethylene) or decompose (for large alkanes) to produce benzene precursors such as propargyl (C 3 H 3 ) and C 4 species, which determined the benzene formation rate via the reaction of C 3 H 3 + C 3 H 3 and C 4 + C 2 . In addition, since all the C-C bond and C-H bond in these large n-alkanes has quantitatively similar bond strength , the C-C fission (or H-abstraction reaction) occurs at a comparable rate [21] . The simulations showed that at fuel consumption rate of 80%, the percentage contribution of C-C fission (or H-abstraction reaction) is at similar level of 32-38% (68-62%). Consequently, the rate to produce intermediate alkenes and thus benzene precursors are also quantitatively similar for these alkanes. Specifically, since ethylene takes the most proportion among various decomposed alkenes (more than 80%, regardless of carbon-chain length), the pathway of ethylene reacting with C 1 and C 2 species to form benzene precursors C 3 H 3 and C 4 become dominant. Note although longer n-alkanes tend to produce more large alkenes beyond ethylene, it effects on the overall sooting limits to contribute negligibly small to. However, it should be pointed out here that the slightly higher amounts of intermediate large alkenes from longer n-alkane may explain why larger n-alkane produce evidently more amounts of soot yields [34] . Hence, we may tentatively conclude that the comparable sooting limits among the C 5 -C 8 n-alkanes is due to their similar fuel decomposition and benzene formation pathways and rates. On the other hand, the notable difference in sooting propensity among smaller n-alkanes (e.g. methane and propane) [19] may be attributed to the different fuel decomposition activity. This is because the C-H bond for methane (438 kJ/mol) is much larger than the C-C bond energy of propane (371 kJ/mol), making it much more difficult to break to produce initial methyl radical (CH 3 ). Moreover, the growth of CH 3 to form benzene precursors also requires relatively slower addition reactions towards important C 2 and C 3 species for methane than that of propane. Now we move on to discuss the effects of fuel unsaturation (the C＝C double) and isomers effects (the position and numbers of side-chains) on sooting tendency, by comparing n-hexane/1-hexene and n-hexane/C 6 H 14 isomers including 2-methylpentane, 3-methylpentane, 2,3-dimethylbutane, respectively. Figure 7 shows the sooting limits (X F,cr , X O,cr ) in X F -X O plane for these selected C 6 fuels. Also included is the relation of critical X F,cr to the calculated adiabatic flame temperature (T ad ) to compare their sooting tendencies through the alternative index of limiting sooting temperature (T cr,SF ). Figure 7 shows that 1-hexene has noticeably larger sooting region and smaller T cr,SF than n-hexane, indicating the stronger sooting propensity of 1-hexene. This result is not surprising as many other sooting tendency studies [19, 34, 39, 56, 78] also reported that the unsaturated fuel (esters, oxygenates) with the C＝C double is more sooting than its saturated counterparts. However, mechanism investigations into the effects of C＝C double is relatively limited. To obtain further information on how the presence of C＝C double in the fuel molecules could lead to higher sooting tendency, the fuel decomposition process of 1-hexene was numerically investigated and compared with that of n-hexane. As can be seen in Fig. 8 , the predominant decomposition route of 1-hexene is the unimolecular dissociation via allylic C-C fission (i.e., the bond between C 3 and C 4 ), which accounts for 86% of total 1-hexene consumption. As a result of the allylic C-C fission, large amounts of n-propyl (nC 3 H 7 ) and allyl radical (aC 3 H 5 ) were produced; similar pyrolysis result of 1-hexene were reported in the literature [79] . In contrast, n-hexane is primarily decomposed via C-C fission and H-abstraction reaction, producing various n-alkyl radicals (such as PC 4 H 9 , nC 3 H 7 and C 2 H 5 ) which were then converted to ethylene. Due to the large amounts of the decomposition products of aC 3 H 5 , it is relatively easy for 1-hexene flame, as compared to n-hexane, to produce the important benzene precursor of C 3 H 3 via the dehydrogenation of aC 3 H 5 (aC 3 H 5 →aC 3 H 4 /PC 3 H 4 →C 3 H 3 ). On the other hand, the decomposition product of nC 3 H 7 are mostly (~ 60%) converted to unsaturated C 2 H 4 and C 3 H 6 , which also in turn contribute to C 3 H 3 formation. Although not shown, the maximum concentration of C 3 H 3 from 1-hexene decomposition is more than twice than that from n-hexane, leading to its much higher benzene production rates (via C 3 H 3 +C 3 H 3 ). This fact may explain why 1-hexene has stronger sooting propensity than n-hexane. (nC 3 H 7 ) ( Fuel decomposition pathways of n-hexane, 1-hexene and 2,3-dimethylbutane. Calculated using the mechanism of Zhang et al. [74] at condition of T=1500 K and P= 1 atm for mixture of 10% fuel+90% Ar. Fuel consumption rate is approximately 80%. The numbers indicate the percent contribution to the consumption of the source species. With respect to the isomer effects, it can be found from Fig. 7 that all the branched C 6 H 14 isomers were more sooting than the straight chain n-hexane. While among these C 6 H 14 isomers, the difference in the sooting limit curve boundary and T cr,SF is not evident; but a general trend of smallest T cr,SF for 2,3-dimethylbutane can be observed, followed by 3-methylpentane and 2-methylpentane. On the other hand, the LII-based sooting index in Fig. 9 shows that 2,3-dimethylbutane with two methyl-branch has significantly larger soot volume fraction than mono-branched one; while the difference in soot yields between 2-methylpentane and 3-methylpentane is much smaller (with 3-methylpentane being slightly larger). The present LII results show that the sooting tendency of C 6 alkane tends to increase with the number of methyl-branch, but are not very sensitive to the specific position of the branch, which agree with the literature-based YSI data [34] . The fuel decomposition pathways of 2,3-dimethylbutane is included in Fig.8 to elucidate the reason why more branched C 6 alkane has stronger sooting propensity than straight-chained counterpart. The kinetic simulation shows that 2,3-dimethylbutane is likely to produce large amounts of iso-propyl (IC 3 H 7 ), which were then decompose primarily to C 3 H 6 ; this situation differs significantly from n-hexane decomposition which mainly produce nC 3 H 7 and then C 2 H 4 . Quantitatively, the ratio in maximum concentration of C 3 H 6 to C 2 H 4 is 0.9 for 2,3-dimethylbutane, larger than the ratio of ~0.44 for mono-branched C 6 H 14 and 0.14 for straight-chained n-hexane. Consequently, the formation rate of benzene precursor C 3 H 3 in 2,3-dimethylbutane is governed by the C 3 H 6 initiated reactions (via dehydrogenation of C 3 H 6 to form C 3 H 3 ), which is faster than the C 2 H 4 -dominated pathway (C 2 H 4 + C 1 species) in n-hexane. In this section, the sooting tendency of selected cycloalkanes, i.e. Cyclopentane (cC 5 H 10 ), Cyclohexane (cC 6 H 12 ) and Methylcyclohexane (cC 7 H 14 ), were compared to investigate the fuel structure effects of cyclic ring size (comparing cC 5 H 10 and cC 6 H 12 ) and the presence of ring-attached side chain (comparing cC 6 H 12 and cC 7 H 14 ) . Figure 10a compares the sooting limits in X F -X O plane for these cycloalkanes; their sooting region does not show significant difference, expect that the sooting limit curve of cC 5 H 10 (cC 7 H 14 ) slightly expands towards large X F (X O ) region. Interestingly, if we plot the critical X F,cr with respect to T ad in Fig. 10b , it is easy to see that cC 5 H 10 has smaller limiting sooting temperature (i.e. T SF,cr at X F =0.6) than cC 7 H 14 or cC 6 H 12 , indicating the sooting tendency trend of cC 5 H 10 ﹥cC 7 H 14 ﹥cC 6 H 12 . This trend is consistent with the LII-based sooting index shown in Fig. 11 , in which the peak LII intensity also shows the order of cC 5 H 10 ﹥cC 7 H 14 ﹥ cC 6 H 12 . It is noticed that the stronger sooting propensity of cC 7 H 14 than cC 6 H 12 obtained in the present CDFs agree with the coflow-based smoke point data [80] and YSI data [34] . However, disagreement appears with respect to the relative sooting tendency trend of cC 5 H 10 and cC 6 H 12 . Specifically, the present sooting limits/LII data was consistent with the smoke point data [80] , which reported the sooting tendency trend of cC 5 H 10 (33.1 mm) > cC 6 H 12 (51.9 mm). Note the fuels with smaller smoke point (i.e. the critical flame height at which soot begins to emit from the flame tip) has stronger sooting propensity. However, the YSI data [34] indicated an opposing sooting tendency trend of cC 5 H 10 (YSI 14.0) < cC 6 H 12 (YSI 19.1). One possible reason for such disagreement may be the potential chemical cross-linking effects between methane and the dopant fuel during YSI measurement [34] , which is different from our works and the smoke point measurements [80] in which pure fuel was used. However, it is difficult to explicitly point out the origins of such disagreement which, of course, deserves further investigations. In this work, we are more interested in the reason why the five-membered cC 5 H 10 , in despite of its smaller ring size, is more sooting than the six-membered cC 6 H 12 . To uncover the underlying mechanism, we performed fuel decomposition pathway analyses for cC 5 H 10 and cC 6 H 12 , using the mechanism developed by Tian et al. [75] with CHR model. As far as the authors aware of, this may be the first study trying to elucidate the mechanism why an increase in ring size conversely reduce the sooting tendency of cycloalkanes. Similar to alkane, benzene formation also play a critical role to the subsequent PAH and soot formation rate for cycloalkanes [21, 81] . The results in Fig. 12 shows that cC 5 H 10 decomposition produces higher amounts of benzene than that of cC 6 H 12 , being consistent with their observed sooting tendency trend. Further ROP-based analyses revealed that benzene formation during cC 5 H 10 and cC 6 H 12 decomposition is dominated by the route of C 3 H 3 + C 3 H 3 . Since cC 5 H 10 decomposition produces higher amounts of C 3 H 3 than cC 6 H 12 (see Fig. 12 ), the higher benzene formation tendency of cC 5 H 10 can be easily understood. [75] at T=1500 K and P=1 atm for mixture of 10% fuel + 90% Ar. Fundamentally, the difference in C 3 H 3 production between cC 5 H 10 and cC 6 H 12 decomposition is related to the fuel-structure dependent decomposition pathways. As shown in Fig. 13 , the fuel consumption processes of cC 5 H 10 and cC 6 H 12 do share a similar route via the ring-opening isomerization to alkenes which then primarily translate to aC 3 H 5 and n-alkyl. However, difference appears in the other decomposition route from the ring structure of cC 5 H 9 and cC 6 H 11 . Specifically, the subsequent decomposition of cC 5 H 9 prefers to produce odd-carbon radicals of cyclopentadienyl (C 5 H 5 ) and aC 3 H 5 , which then convert primarily to C 3 H 3 via the reaction of C 5 H 5 = C 3 H 3 + H 2 and the dehydrogenation of aC 3 H 5 , respectively. In contrast, cC 6 H 11 tends to produce C 4 and C 2 species; this indeed makes cC 6 H 12 have higher benzene formation rate via the reaction of C 4 + C 2 than cC 5 H 10 , however, it only plays a secondary role to total benzene formation. Consequently, cC 5 H 10 has faster C 3 H 3 and benzene formation rate than cC 6 H 12 , leading to the stronger sooting propensity of cC 5 Fig. 13 Fuel decomposition pathways for cyclopentane and cyclohexane calculated using the mechanism of Tian et al. [75] at T=1500 K and P=1 atm for mixture of 10% fuel+90% Ar. Fuel consumption rate is approximately 80%. The numbers indicate the percent contribution to the consumption of the source species. The above discussions concerning the benzene formation chemistry provide reasonable explanations for the observed sooting tendency between cC 5 H 10 and cC 6 H 12 . However, Further investigations by comparing their PAH growth/soot evolution process is also required. For instance, it has been suggested that the PAH formation pathway via the self-combination of C 5 H 5 (to form naphthalene) may also play a role to soot formation of cC 5 H 10 flames [82, 83] . Unraveling the possible roles of the C 5 H 5 related PAH pathways would help to provide a more comprehensive understanding for the observed sooting tendency among the cycloalkanes. This work experimentally investigated the sooting tendency of liquid C 5 -C 8 alkanes, alkenes and cycloalkanes in counterflow diffusion flames, using both the sooting limits and soot volume fraction as quantitative sooting indices. The main objective is to clarify the fuel structure effects (including carbon-chain length, isomers, fuel unsaturation and cyclic ring structure) on sooting tendency, with the help of numerical analyses on the fuel decomposition/soot precursor formation pathways. Major conclusions and findings are summarized in the following: (1) The large C 5 -C 8 n-alkanes has comparable sooting limits among each other, in contrast to the smaller C 1 -C 4 n-alkane in which notable difference in their sooting limits were previously observed. This result indicates a rather interesting fuel similarity with respect to sooting tendency among these large n-alkanes. This trend occurs because the sooting limits is mostly related to the specific fuel decomposition and benzene formation pathways, which is essentially similar for the C 5 -C 8 n-alkane regardless of their carbon-chain length. This may help to simplify the prediction on sooting tendency of large n-alkanes. (2) The presence of fuel unsaturation in 1-hexene leads to its stronger sooting tendency than n-hexane, as 1-hexene tend to produce more amounts of allyl radical and therefore has faster benzene precursor (C 3 H 3 ) formation rate. On the other hand, all the branched C 6 H 14 (2-methylpentane, 3-methylpentane, 2,3-dimethylbutane) has larger sooting propensity than n-hexane; while the sooting tendency of C 6 H 14 isomers increased with the number of methyl-branch, but is not very sensitive to the position of methyl-branch. Numerical pathway analyses revealed that more branched C 6 H 14 decomposes larger amounts of C 3 H 6 than the straight one (which primarily produces C 2 H 4 ), thus leading to its higher sooting tendency. (3) The present CDF-based sooting limits and LII data shows that the five-membered cyclopentane has even stronger sooting propensity than the six-membered cyclohexane. Numerical pathway analyses showed that, compared to the decomposition of cyclohexane which primarily produces C 4 species, cyclopentane tends to decompose to more odd-carbon radicals (such as allyl and cyclopentadienyl radical) and thus has faster benzene formation rate and stronger sooting tendency. Note our result is inconsistent with the coflow-based YSI data in which a higher sooting tendency of cyclohexane than cyclopentane was reported. This result suggests the necessity of evaluating fuel's sooting tendency in different flame environments. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Bounding the role of black carbon in the climate system: A scientific assessment Air Quality and Climate Connections High secondary aerosol contribution to particulate pollution during haze events in China Association of particulate matter pollution and case fatality rate of COVID-19 in 49 Chinese cities Assessing the relationship between surface levels of and PM10 particulate matter impact on COVID-19 in Combustion-formed nanoparticles Formation of nascent soot and other condensed-phase materials in flames Soot formation in laminar counterflow flames Soot formation in high pressure laminar diffusion flames Reactive sites on the surface of polycyclic aromatic hydrocarbon clusters: A numerical study Surface reactivity of polycyclic aromatic hydrocarbon clusters Resonance-stabilized hydrocarbon-radical chain reactions may explain soot inception and growth Sooting limits of non-premixed counterflow ethylene/oxygen/inert flames using LII: Effects of flow strain rate and pressure (up to 30 atm) Soot formation in counterflow non-premixed ethylene flames at elevated pressures Pressure and temperature dependence of soot in highly controlled counterflow ethylene diffusion flames PAHs and soot formation in laminar partially premixed co-flow flames fuelled by PRFs at elevated pressures Effect of temperature on soot inception in highly controlled counterflow ethylene diffusion flames Temperature and oxygen effects on oxidation-induced fragmentation of soot particles Effect of strain rate on sooting limits in counterflow diffusion flames of gaseous hydrocarbon fuels: Sooting temperature index and sooting sensitivity index The Effect of the Strain Rate on PAH/Soot Formation in Laminar Counterflow Diffusion Flames Studies of aromatic hydrocarbon formation mechanisms in flames: Progress towards closing the fuel gap Investigation of the effect of molecular structure on sooting tendency in laminar diffusion flames at elevated pressure Influences of the molecular fuel structure on combustion reactions towards soot precursors in selected alkane and alkene flames Effects of fuel structure on structural characteristics of soot aggregates Measuring and predicting sooting tendencies of oxygenates, alkanes, alkenes, cycloalkanes, and aromatics on a unified scale A Quantitative Model for the Prediction of Sooting Tendency from Molecular Structure Measurements of sooting limits in laminar premixed burner-stabilized stagnation ethylene, propane, and ethylene/toluene flames Effects of n-butanol addition on sooting tendency and formation of C1 -C2 primary intermediates of n-heptane/air mixture in a micro flow reactor with a controlled temperature profile Study on sooting behavior of premixed C1-C4n-alkanes/air flames using a micro flow reactor with a controlled temperature profile An improved methodology for determining threshold sooting indices from smoke point lamps Evaluation of sooting tendency of different oxygenated and paraffinic fuels blended with diesel fuel Vision based algorithm for automated determination of smoke point of diesel blends Improved sooting tendency measurements for aromatic hydrocarbons and their implications for naphthalene formation pathways Sooting tendencies of oxygenated hydrocarbons in laboratory-scale flames Predicting sooting tendencies of oxygenated hydrocarbon fuels with machine learning algorithms Predicting the soot emission tendency of real fuels -A relative assessment based on an empirical formula Numerical simulations of yield-based sooting tendencies of aromatic fuels using ReaxFF molecular dynamics Numerical investigation of the pressure-dependence of yield sooting indices for n-alkane and aromatic species Group additivity in soot formation for the example of C-5 oxygenated hydrocarbon fuels Analysis of the sooting propensity of C-4 and C-5 oxygenates: Comparison of sooting indexes issued from laser-based experiments and group additivity approaches Soot zone structure and sooting limit in diffusion flames: Comparison of counterflow and co-flow flames Experimental and soot modeling studies of ethylene counterflow diffusion flames: Non-monotonic influence of the oxidizer composition on soot formation Cracking and deposition behavior of supercritical hydrocarbon aviation fuels Coking and gas products behavior of supercritical n-decane over NiO nanoparticle/nanosheets modified HZSM-5 Experimental and modeling study on pyrolysis of n-decane initiated by nitromethane Modelling and numerical simulation of n-heptane pyrolysis coking characteristics in a millimetre-sized tube reactor Effect of Aviation Fuel Type on Pyrolytic Reactivity and Deposition Propensity under Supercritical Conditions Sooting tendency of n-heptane/anisole blends in laminar counter-flow flames with oxygen enrichment The blending effect on the sooting tendencies of alternative/conventional jet fuel blends in non-premixed flames A comparative study on the sooting tendencies of various 1-alkene fuels in counterflow diffusion flames Effect of dimethyl ether (DME) addition on sooting limits in counterflow diffusion flames of ethylene at elevated pressures Sooting limit in counterflow diffusion flames of ethylene/propane fuels and implication to threshold soot index A jet fuel surrogate formulated by real fuel properties Recent progress in gasoline surrogate fuels Experimental comparison of soot formation in turbulent flames of Diesel and surrogate Diesel fuels Sooting tendencies of unsaturated esters in nonpremixed flames Sooting tendencies of nonvolatile aromatic hydrocarbons An experimental and theoretical investigation of the dilution, pressure and flow-field effects on the extinction condition of methane-air-nitrogen diffusion flames Chemical effects of hydrogen addition on soot formation in counterflow diffusion flames: Dependence on fuel type and oxidizer composition On the opposing effects of methanol and ethanol addition on PAH and soot formation in ethylene counterflow diffusion flames The elusive history of m∼= 1.57 -0.56i for the refractive index of soot One-dimensional tomography: a comparison of Abel, onion-peeling, and filtered backprojection methods An experimental study on the spectral dependence of light extinction in sooting ethylene counterflow diffusion flames Soot concentration and temperature measurements in co-annular, nonpremixed CH4/air laminar flames at pressures up to 4 MPa Experimental and numerical study of soot formation in counterflow diffusion flames of gasoline surrogate components Understanding benzene formation pathways in pyrolysis of two C6H10 isomers: Cyclohexene and 1,5-hexadiene Fuel-structure dependence of benzene formation processes in premixed flames fueled by C6H12 isomers Fuel dependence of benzene pathways CHEMKIN-PRO 15131, Reaction design Effect of mixing methane, ethane, propane, and propene on the synergistic effect of PAH and soot formation in ethylene-base counterflow diffusion flames Elucidating the chemical pathways responsible for the sooting tendency of 1 and 2-phenylethanol Experimental and theoretical insight into the soot tendencies of the methylcyclohexene isomers A high-temperature chemical kinetic model of n-alkane (up to n-dodecane), cyclohexane, and methyl-, ethyl-, n-propyl and n-butyl-cyclohexane oxidation at high temperatures, JetSurF version 2.0 An experimental and kinetic modeling study of the oxidation of hexane isomers: Developing consistent reaction rate rules for alkanes Shock Tube and Kinetic Modeling Study of Cyclopentane and Methylcyclopentane Growth of soot particles in counterflow diffusion flames of ethylene Temperature dependence of the fuel mixing effect on soot precursor formation in ethylene-based diffusion flames Structural effects of biodiesel on soot formation in a laminar coflow diffusion flame Experimental and kinetic modeling study of 1-hexene combustion at various pressures Some effects of molecular structure of single hydrocarbons on sooting tendency Multiple benzene-formation paths in a fuel-rich cyclohexane flame Aromatic and Polycyclic Aromatic Hydrocarbon Formation in a Laminar Premixed n-Butane Flame An Experimental Study in Non-Premixed Flames of Hydrocarbon Growth Processes that Involve Five-Membered Carbon Rings This work was supported by the National Natural Science Foundation of China (51606136, 51976142).