Effect of gas additives on soot formation in flat laminar premixed ethylene/air flames under low pressure

This study reports on the effect of the additive gases, namely N 2 , CO 2 , and Ar, on the soot formation in laminar premixed C 2 H 4 /air flames at low pressure of 40 kPa. The flames blended with N 2 , CO 2 , or Ar were experimentally investigated, where the emission spectra of CH* and C 2 *, qualitative concentration of PAHs, and soot volume fraction ( f v ) were measured using laser and optical diagnostic methods. The flame temperature was measured using a thermocouple. The results reveal that additive gases significantly influence flame height above the burner and fuel combustion as well as reduce soot formation and flame temperature. With respect to N 2 and CO 2 , Ar proves the most effective in reducing soot volume fraction, achieving a 100-fold reduction compared to the reference flame. Moreover, the additive gases were found to delay the ignition, leading to a 5 mm downstream shift in soot inception. Despite the difference in the properties of the three gas additives, it was found the first incepted soot particles were detected at a common temperature, T inception , of 1515 ± 70 K. Three regimes related to the soot appearance rate were identified, which are the fast-increasing range from the soot inception location, the plateau region, and the decreasing range to the flame front of 20 mm. According to the Lagrangian time-derivative of soot volume fraction ( df v /dt ) as a function of f v , the constant of surface growth ( k SG ) was determined to be 170 s (cid:0) 1 in the flame with additives of CO 2 and Ar, measured at 40 kPa. The identified turning point between the plateau region and the negative soot increase regions can be used as an indicator of the transition from higher to lower soot formation rates in soot modeling.


Introduction
Soot emission from hydrocarbon flames has been under intense investigation for decades.The underlying aims are to better understand the complex interdependency of the different parameters that control the formation, oxidation, and emission of these carbonaceous nanoparticles, especially under practical flame conditions.[1] Soot emitted into the ambient atmosphere has been identified as a substantial threat to human health and results in climate change.[2] On the other hand, the formation and oxidation of soot particulates within the flames' envelope is highly desirable in applications where thermal radiation is the main heat transfer mechanism, e.g., boilers, furnaces, and cement kilns.Similarly, carbon particles, such as 'carbon black', are produced using hydrocarbon flames for use in tires, plastics, batteries, and rubber among others.[3] Hence, it is crucial to better understand the mechanisms of formation and destruction of soot in flames, develop models to predict it, and devise strategies to control its formation and emission.[4] The formation of soot is a highly complex process that involves the transition of gases into solids.This mechanism depends on the type of fuel being used and begins with the combustion of hydrocarbon or fossil fuels, which occurs under high temperatures, pressures, and chemical reactions, as well as electric interactions.[5] The most important step in soot formation is pyrolysis.It leads to the fragmentation of molecules, resulting in the formation of hydrogen ions, acetylene, phenyl radicals, and other soot precursors.The specific composition of these precursors, including the formation of aromatic rings, is dependent on the reaction temperature.[6][7] The PAHs are considered to be an active precursor in soot formation.The combustion of hydrocarbons, including natural gas, oil, and methane gas, releases PAHs with molecular weights ranging from 500 to 2000 amu.These PAHs are also considered to be cancercausing agents, posing a significant health risk to human beings and the environment.Due to coagulation and the presence of the active precursor, new soot particles are formed with surface growth.[8] These soot particles mainly consist of carbon compounds originating from particulate matter, volatile organic compounds, NO, and CO. [5,[9][10].
It is well established that one mechanism to suppress soot formation in flames is the addition of inert gasses to the fuel which can also result in improving the efficiency of internal combustion engines.[10][11][12] For example, additives such as Ar, He, and N 2 , can suppress soot emission and reduce soot volume fraction in hydrocarbon flames due to the dilution effect of changing the segment concentrations, the thermal effect of changing the heat capacity, and the chemical effect of participating in the reactions specifically.[13][14][15][16][17][18][19][20][21] Over the last decades, high-quality measurements have been performed in various flames with the additive gases, e.g., the premixed CH 4 /air flames with CO 2 addition [22], C 2 H 4 diffusion flames diluted with He gas [20], and C 2 H 6 and C 3 H 8 flames with H 2 addition [23].Among these studies, C 2 H 4 has frequently been selected as the fuel in the literature due to its high soot yields and relatively simple chemistry in soot formation [24], particularly in the laminar C 2 H 4 /air premixed flame.Haynes et al. [19] reported an effect on soot formation in a C 2 H 4 / air flat flame as a result of the addition of 3 % N 2 , H 2 O, H 2 , NH 3 , NO, H 2 S, SO 3 , SO 2 , HCl, CH 4 , C 2 H 4, and O 2 .These additives did not influence soot particle coagulation, but soot volume fraction was reduced significantly.Soot suppression was much stronger with the addition of H 2 O and CO 2 .Tang et al. [14] investigated the effect of CO 2 addition on sooting behavior in premixed ethylene flat flames at atmospheric pressure.They found that CO 2 addition gave rise to a reduction in soot nucleation rate and a slower mass growth rate significantly lowering soot yield.De Iuliis et al. [25] reported the role of H 2 addition to fuel in soot formation and growth mechanisms in a rich C 2 H 4 /air-premixed flame.They found the soot conversion efficiency was reduced when increasing the concentration of H 2 additions.Although H 2 addition strongly affected the soot aggregation, no effect was observed on the growth rate of soot diameter.In addition, Renard et al. [26] studied the flame structure C 2 H 4 /air-premixed flat flame with CO 2 , NH 3 , or H 2 O at 50 mbar.With gas additions, the flame front was shifted downstream about 1 to 2 mm and caused flame inhibition.Argon is another additive that is usually used to reduce soot formation.Guo et al. [27] developed a numerical model to study the influence of additives in a C 2 H 4 /air diffusion flame, and they found that argon is more efficient at suppressing soot formation due to the thermal diffusivity.Such studies could help to enable a detailed understanding of the mechanism about the soot inception and growth with additives.
Despite the significant progress on the subject, questions remain about the effects of additive gases on the formation of soot in the laminar premixed flame, particularly under low-pressure conditions.As the flame thickness and soot zone expand under low-pressure conditions, it provides a good opportunity to investigate the soot formation process.Low pressure can help to improve the spatial resolution, which can benefit the investigation of the soot inception conditions and enable the measurements of soot growth rates.[28] For the investigation under low-pressure conditions, there has been some research directed into the soot particle formation in hydrocarbon flames, such as CH 4 and C 3 H 8 .[29] To study the soot formation, most of the published studies have been conducted using laminar flames and utilized laser-induced fluorescence (LIF) and laser-induced Incandescence (LII) in several premixed flames at the low-pressure environment.[30] However, none of them provides a systematic study that specifically focuses on the C 2 H 4 premixed flame under low-pressure conditions, especially with the influence of additive gases on soot inception, PAH formation, and the formation of critical radicals.
Additionally, the soot surface growth is the key factor in soot formation in the combustion environment.It's worth noting that all previously reported data regarding the constant of soot surface growth (K SG ) in C 2 H 4 /air-premixed flames were obtained under atmospheric to high-pressure conditions.[28,31] To the best of the authors' knowledge, it has not been reported of such conditions with additive gases at low pressure.
This paper aims to understand the mechanism of soot formation in a laminar premixed C 2 H 4 /air flame at low pressure with additive gases of N 2 , CO 2 , and Ar using a Mckenna burner.The low-pressure condition was selected as the soot inception zone is expanded providing a better opportunity to improve the spatial resolution.[28] The formation of CH* and C 2 * radicals as well as soot were measured at the height above the burner surface (HAB) up to 22 mm.The qualitative PAH LIF, CH, and C 2 radicals' emissions and the gas temperature are recorded for further understanding of the conditions of soot inception.Finally, three regimes related to the soot appearance rate are identified.The constant of soot surface growth is estimated for the flame with CO 2 and Ar additives.

Experimental apparatus
In this study, a 60 mm diameter McKenna burner was utilized and mounted onto an electrically controlled traverse system, which was described in detail in our previously reported work.[28] The setup allowed the measurements to be taken at various heights along the flame centerline.As shown in Fig. 1, the burner was placed inside a 39.8-liter stainless steel chamber equipped with four optically accessible windows, which is labeled as Part 2 in the figure.To maintain a low-pressure environment within the chamber, a vacuum pump (Edwards, EDM12) was employed.The chamber pressure was measured using a Bartron device (MKS, 122AA-0100AB) with a precision of ± 0.2 kPa.Additionally, an electronic pressure regulator and control system (Equilibar, QPV 1) were installed in the exhaust gas pipeline.This setup ensured accurate pressure measurements and provided very good pressure stability within the chamber, with a variation of only 0.66 kPa.
To detect the emission spectra of CH* and C 2 *, the flame emission was imaged on the end of an optical fiber (Thorlabs, BFL200HS02), with a core diameter of 0.2 mm, and the position of its end was controlled with an accuracy of ± 0.5 mm, and using two spherical lenses with focusing lengths of f 1 = 300 mm and f 2 = 250 mm.For a better signal at different heights of the various flames, the fiber end was traversed vertically along the centerline of the flame.The emission spectra of CH* Fig. 1.Schematic diagram of the experiment system, including the stainlesssteel chamber, which featured four optically accessible windows.
S. Algoraini et al. and C 2 * were collected using a spectrometer (Andor, Shamrock 500i) with a grating of 150 lines/mm.
The qualitative concentration of the corresponding PAHs was measured using Laser-induced fluorescence (LIF).The fourth harmonic (266 nm) of a 10 Hz pulsed Nd:YAG laser (Quantel, Q Smart) was used as a laser source for the PAH LIF.The incident laser energy was maintained at 23.3 mJ/cm 2 per pulse to minimize possible interference from laserheated soot particle emission, induced by laser-induced incandesces (LII).The LIF signal was collected at right angles to the laser propagation by a 100 mm focal lens.The lens was positioned so that the fluorescence was coupled to an optical fiber.The LIF signals were collected using a spectrometer (Andor, Shamrock 500i) which was attached by an ICCD camera.The delay time and the gate width of the ICCD were set to 0 ns and 30 ns, respectively.The LII signal intensity was determined by summing the counts from all pixels in the recorded images.
The soot volume fraction, f v , was measured using LII with a fundamental output (1064 nm and ~ 8 ns in duration) from a Nd:YAG laser (Surelite II).The laser power was adjusted using a combination of a halfwave plate and a Glen-laser polarizer.Then, the laser beam was reformed into a horizontal laser sheet of ~ 1 × 5 mm using a cylindrical lens (f = 750 mm).A laser fluence of 1.3 J/cm 2 with 15 % energy fluctuation was used in all measurements.LII signal was collected using an ICCD camera with a short pass filter centered at 900 nm to suppress the 1046 nm radiation to the laser beam.The ICCD was set to operate at gate delay and gate width of 120 ns and 50 ns, respectively, and 200 shots were recorded for each image.
The flame temperature was measured using a 75-μm-diameter Pt/Pt-R thermocouple (Omega) to minimize the conductivity and radiation effects at the junction.The thermocouple temperature has been corrected based on thermal radiation and heat conduction.Catalytic effects of the junction and errors because of conduction along the wires were expected to be negligible.For each measurement, the thermocouple stayed in the flame until the temperature reading became stable.The accuracy of the thermocouple was within ± 0.25 %.The difference between the measured temperature and the corrected temperature is ± 60 K.
The additive gases of N 2 , CO 2 , and Ar were selected to dilute the reference flame of C 2 H 4 /air.The operating flow conditions of different flames are shown in Table 1.The total flows of 7.6 L/min at a fixed equivalence ratio of Φ = 2.1 and fixed dilution rate of 0.6 L/min were examined and the pressure was fixed at 40 kPa.It is worth noting that the flow rate of the reference is ~ 8 % lower than the additive flames.It was assumed the contribution from this difference is minor.It was assumed the contribution from this difference is minor.

Results and discussion
To obtain spatially resolved emission spectra, the fiber optic was scanned vertically, and the spectra were recorded at each HAB from 0 to 22 mm.Fig. 2  The qualitative concentration of the corresponding PAHs can be approximately indicated by the signal strengths at the corresponding wavelength bands.Fig. 3 presents a typical example of PAH LIF spectra for the premixed C 2 H 4 /air flame with N 2 at 40 kPa and HAB = 14 mm.The PAH LIF signal in the wavelength range of 320 nm to 360 nm is attributed to 2 to 3-membered PAHs, while the signals within the range of 370 nm to 410 nm are attributed to the presence of 3 to 4-membered PAHs.Thus, the detection wavelength band was used to distinguish the relative size groups of PAHs.However, it is important to consider that the interference from liquid-like condense species and PAHs attached to soot surfaces may also contribute to the recorded LIF signals.[32,33]    The elastic scattering, due to the presence of soot particles and other large species, occurs at 532 nm and 266 nm and will overlap with PAH LIF recorded spectra.To compute the PAH signal level, these two ranges were excluded to avoid the contribution from the laser source.Fig. 4 displays the LII images captured under four flame conditions and four typical HAB from 8 mm to 18 mm at a pressure of 40 kPa.In the LII measurement, the laser beam was reformed into a horizontal laser sheet with a thickness of only 1 mm.The burner was vertically moved to capture LII images at different heights.In the figure, each image exclusively represents the LII intensity from soot particles at the corresponding locations, specifically at 8 mm, 10 mm, 15 mm, and 18 mm, respectively.The LII signal intensity was determined by summing the counts from all pixels in the recorded LII images after subtracting the background.To ensure adequate signal-to-noise ratios, positions within the flames with a high soot volume fraction were chosen for analysis, ensuring the signals were sufficiently high.Consequently, the LII signals selected in this study were identified at HAB values ranging from 8 to 18 mm.Based on the images, additive gases can shift the location of soot inception downstream compared to the reference flame.This phenomenon becomes even more evident when the flame is diluted with Ar.In addition, the asymmetry in LII images may be caused by the stainlesssteel central part of the McKenna burner.A similar observation was reported of this effect by Migliorini et al. [34].As the setup and optical arrangement remained the same during the measurement, the impact of asymmetrical LII images can be neglected.The variation of laser fluence introduces a source of error in the LII measurement.The laser fluence fluctuation of 1.2 to 1.4 J/cm 2 gives rise to a 15 % uncertainty in the evaluated value of f v among 200 images.
To determine df v /dt, the velocity field of the flame was modelled by using the Ansys-Fluent CFD software package, version 18.0.The CFD model was validated by the velocity profile measured using a hot wire in a cold flow.[28] The error between the simulated and measured velocity was less than 15 %.
The local reaction time (dt HAB=i ) can be calculated based on equation (1) below: where x HAB=i+1 -x HAB=i is the distance between two adjacent HAB, which equals 1 mm, and v HAB=i is the velocity at location HAB = i.The time origin was set at the location where the first soot particles were detected.
Fig. 5 contains measured centreline axial profiles of the temperature, the flame emission of CH*, C 2 *, PAH (2-3R), and (3-4R), soot volume fraction, f v , and Lagrangian time-derivative of soot volume fraction, df v / dt, for the reference flame and the flames with additive gases of N 2 , CO 2 , and Ar at 40 kPa.Along with the increasing HAB, three regimes of df v /dt are clearly identified, i.e., (i) the fast increasing range from the soot inception location, (ii) the plateau region, and (iii) the decreasing range to the flame front of 20 mm.The fast growth in region (i) is due to the formation of new soot particles, while the change between region (ii) and (iii) corresponds to surface growth, oxidization, and PAH addition.In comparison to the reference flame, although the additive gases shifted the starting point of region (i) to the higher HAB, the location of the turning point between region (ii) and (iii) appears 3 mm before that of the reference flame.In this region, a significant reduction in f v was observed, namely, 20-fold, 40-fold, and 100-fold lower than the reference flame for the N 2 , CO 2, and Ar addition, respectively.The value of df v /dt at higher f v is important to determine the surface growth rate in premixed ethylene/air flames and it has been discussed in the following section.
Table 2 summarizes the key results for the four flames in Fig. 5. Compared to the reference flame, it is obvious that when diluting the flame with additive gases, the location of the maximum flame temperature was shifted downstream, which resulted in the corresponding shift of the peak concentration of CH*, C 2 *, and PAH.For example, for the reference flame, the maximum intensity of CH* is at HAB = 3 mm, while for the flame with N 2 , CO 2 , and Ar additives, the maximum intensity of CH* is observed at HAB = 8 mm, 8 mm, and 10 mm, respectively.
The shift of the flame temperature and peak concentration of the species is also consistent with the shift of the soot inception location, which indicates the formation of soot is highly correlated with flame temperature and PAHs.From Table 2, it is clear the soot formation starts at the location when the flame temperature and PAH concentration reach the maximum.When exposed to high temperatures in the reaction zone, PAHs can undergo further reactions and lead to aggregation and condensation into soot particles.In addition, the formation of C 2 * typically occurs after the formation of CH* due to the sequence of chemical reactions involved.However, in flames with additives, the peak concentration of C 2 * and CH* are present at the same location simultaneously, which could be caused by the limitation of the HAB resolution, set at 1 mm.After the inception of soot, the soot volume fraction rapidly increased to its peak value.Despite the difference in the thermal conductivity, thermal capacity, and viscosity of the three additives, it was found that the first soot particles were detected at a common temperature, T inception , of 1515 ± 70 K.This is almost the same as the value recently reported for premixed C 2 H 4 /air flames at 27 kPa.[28].
Moreover, additive gases can effectively lower the soot volume fraction.For example, in the flame diluted by Ar, the maximum soot  In addition, according to Bockhorn et al. [35], the relationship between df v /dt and f v may be described by a first-order rate law as the equation below: where f v ∞ is the soot volume fraction at a vast distance away from the burner surface and k SG is a rate constant describing the time of active soot growth.The value of k SG can only be evaluated at the region where f v ∞ is reachable.
To illustrate the differences in the soot surface growth for the different flames, Fig. 6(b) presents the profiles after the normalization of df v /dt and f v .In each flame, a common scaling factor is applied to both df v /dt and fv for normalization.The peak of the profile in each flame is adjusted to the same location and the maximum values of df v /dt are normalized to the value of 1.After the normalization, the profile changing rate between different flames becomes easy to compare while the slope of each profile remains unchanged.The black dashed line is drawn to indicate the value of k SG for the flame with additives of CO 2 and Ar.According to the figure, k SG of the flame with CO 2 and Ar additives shows a quite similar value of 170 s − 1 , which is consistent with the typical value of 100 to 200 s − 1 published in the literature under various flame conditions, including pressure and mixture composition.[29,36] For example, Mätzing and Wagner [37] reported a value k SG = 140 in C 2 H 4 /air flame with a C/O ratio of 0.63 under a high-pressure condition of 5 bar, while Tsurikov et al. estimated k SG of 106 and 102 for the C 2 H 4 /air flame with an equivalent ratio of 2.3 and 2.5, respectively, at the pressure of 3 bar.It further reveals that the value of k SG is independent of pressure or fuel type and mixture composition.
It is worth noting that the translation of the burner was limited to the HAB = 22 mm, the highest soot profile of the reference flame was measured at its maximum f v of 0.04 ppm and did not reach the region of f v ∞ , at the pressure of 40 kPa.A similar phenomenon was also observed in the flame with N 2 additives.Only at the higher HAB with higher f v , the profile changes due to the proximity of f v ∞ .[36] Therefore, the straight line with a negative slope cannot fit accurately in the reference flame and the flame that is diluted with N 2 in this study.This limitation is common in the findings published in the literature, as they often struggle to reach the rich-soot region required to satisfy Equation (2).Therefore, adding additive gases of CO 2 and Ar into the flame could be a feasible way to significantly reduce soot volume fraction, thereby reaching the region of f v ∞ .On the other hand, low-pressure conditions provide a good opportunity to investigate the soot formation process, as flame thickness and soot zone expand.As k SG is independent of mixture composition, it is possible to predict the k SG for the C 2 H 4 /air reference flame under low pressure by determining the k SG for the flame with additive gases.To our best knowledge, it is also the first time that the value of k SG is reported in the C 2 H 4 /air flame under low-pressure conditions with additive gases.On the other hand, when increasing f v , the region with positive slopes is observed firstly due to the formation of new soot particles, while the region with negative slopes is presented after df v /dt reaches the maximum.Within this negative-slope region, the soot appearance rate becomes lower due to the combination of factors, including surface growth, soot oxidation, and PAH addition.[38] When introducing additive gases, the changes of df v /dt in both the positive-slope region and negative-slope region become more evident compared to the reference flame.This could be attributed to the ability of additive gases to lower the flame temperature and dilute the reactants.A further understanding of this process can be achieved by soot modelling.The location of the slope turning point could also be used as an indicator of the transition from higher to lower soot formation rates in such models.For example, the transition point of the reference flame is at f v = 0.025 ppm and df v /dt = 1 × 10 − 6 s − 1 , while that of the flame with Ar additive is at f v = 2 × 10 − 4 ppm and 1.25 × 10 − 8 s − 1 .

Conclusions
The influences of gas additives, namely N 2 , CO 2, and Ar, onto the C 2 H 4 /air premixed laminar flames, at the reduced pressure of 40 kPa, were investigated.It was found the first soot particles were detected at a

Table 2
Summary of key results for all flames in the current study based on data presented in Fig. 5.
shows a typical example of the spatially resolved emission spectra of the premixed C 2 H 4 /air flame with Ar at 40 kPa at HAB = 9 mm.In hydrocarbon flame emission, the strong CH* (0-0) emission is observed at 431.5 nm.However, it is partly overlapped by C 2 * emission at 436 nm.The C 2 *(1-0) and (0-0) emissions are detected at 436 nm, 472 nm, and 517 nm.To determine the qualitative concentration, the signals of CH* and C 2 * are determined by the integration of the intensity at the range of 418 to 431 nm and 500 to 525 nm, respectively.The spatially resolved emission spectra enable the identification of the spatial location of the maximum intensity of CH* (I CH_max ) and C 2 * (I C2_max ).

Fig. 2 .
Fig. 2. Spatially resolved emission spectra of the premixed C 2 H 4 /air flame with Ar at 40 kPa and HAB = 9 mm.

Fig. 3 .
Fig. 3. Typical example of the PAH LIF spectra for the premixed C 2 H 4 /air flame with N 2 at 40 kPa and HAB = 14 mm.

Fig. 4 .
Fig. 4. LII imaging of reference flame and the flame with N 2 , CO 2 , and Ar additives at 40 kPa pressure.

Fig. 6 .
Fig. 6.(a) df v /dt as a function of soot volume fraction at 40 kPa for the reference flame and flames with N 2 , CO 2 , and Ar additives.(b) df v /dt as a function of f v after normalization.In each flame, the values of df v /dt and f v are normalized to the peak location using a single scaling factor for the slope comparison.The black dashed line is drawn to indicate the k SG for the flame with additives of CO 2 and Ar.

Table 1
Flow conditions for the laminar premixed C 2 H 4 /air flames.