Toward Elimination of Soot Emissions from Jet Fuel Combustion

Soot from jet fuel combustion in aircraft engines contributes to global warming through the formation of contrail cirrus clouds that make up to 56% of the total radiative forcing from aviation. Here, the elimination of such emissions is explored through N2 injection (containing 0–25 vol % O2) at the exhaust of enclosed spray combustion of jet fuel that nicely emulates aircraft soot emissions. It is shown that injecting N2 containing 5 vol % of O2 enhances the formation of polyaromatic hydrocarbons (PAHs) that adsorb on the surface of soot. This increases soot number density and volume fraction by 25 and 80%, respectively. However, further increasing the O2 concentration to 20 or 25 vol % enhances oxidation and nearly eliminates soot emissions from jet fuel spray combustion, reducing the soot number density and volume fraction by 87.3 or 95.4 and 98.3 or 99.6%, respectively. So, a judicious injection of air just after the aircraft engine exhaust can drastically reduce soot emissions and halve the radiative forcing due to aviation, as shown by soot mobility, X-ray diffraction, Raman spectroscopy, nitrogen adsorption, microscopy, and thermogravimetric analysis (for the organic to total carbon ratio) measurements.


INTRODUCTION
About a million tons of carbonaceous (soot) nanoparticles are released every year by aviation through incomplete combustion of jet fuel. 1 These emissions have a major impact on the health of airport workers and communities living near airports due to their cytotoxicity. 2 In addition, soot nanoparticles typically form clusters (agglomerates) that strongly absorb light, reducing visibility and increasing the radiative forcing, RF, and thus the Earth's temperature. 3 Most importantly, aircraft soot emissions act as ice nuclei and form contrail cirrus clouds. 4 The RF from such contrails makes up about 56% of the total RF induced by aviation. 5 Thus, eliminating soot emissions from aircraft engines is essential to limit their impact on public health and substantially reduce their climate forcing. 6 To this end, bio-based (e.g., hydrotreated esters and fatty acids, HEFA 7 ) or synthetic fuels derived by the Fischer− Tropsch (FT) process 8 have been explored to reduce the soot emissions from the combustion of petroleum-based jet fuels in aircraft engines. For example, a 50:50 blend of jet A and HEFA fuels decreased the total number concentration, N t , of soot nanoparticles 7 by 50−70%. Similarly, the combustion of a 60:40 blend of jet A1 and FT-derived fuels lowered by 34− 50% the soot N t . 8 Blending jet fuel with such alternative fuels decreases the mean mobility, 7 d m , and primary particle diameters, 9 d p , of soot by about 15 and 30%, respectively. Raman and microscopy analyses indicate that the combustion of biofuels results in more amorphous soot than jet fuels, while FT-derived fuels yield more graphitic soot. 9 Despite the rather large (50−70%) reduction of aircraft soot emissions, using blends of jet with bio-based or synthetic fuels reduces only up to 20% the RF from contrail cirrus clouds. 6 In this regard, climate modeling revealed that a 90% decrease of soot N t can reduce this RF 6 up to 50%. This can be attained through gas (or air) injection downstream of the aircraft combustors. 10 For example, the design of quite a few of the current aircraft combustors is based on the rich quench lean (RQL) concept 11 where swirling and cross-flow jets are used in the primary zone to produce high concentrations of soot. 12 This zone is followed by a lean dilution zone, where additional air is injected to oxidize that soot. 12 Similarly, O 2 was introduced downstream of model laboratory RQL combustors burning ethylene 13 to oxidize soot and reduce its volume fraction, 14 f v , and N t up to 99%. However, soot produced by ethylene combustion contains a higher organic and amorphous carbon content than aircraft soot from jet fuel combustion. 15 In particular, Raman spectroscopy showed that the oxidative reactivity of soot increases with its amorphous carbon content. 16 Recently, the impact of air injection downstream of jet fuel combustion was elucidated in a laboratory RQL combustor. 17 The rather small air flow rates used there resulted in low O 2 concentrations downstream of the combustor 17 that reduced soot f v up to 40%. The limited reduction of soot f v in current RQL combustors can be attributed to the inhomogeneity of temperature and gas profiles that result in regions with high concentrations of soot that survive oxidation and exit the combustor. 18 Large reductions (>90%) of soot N t and f v have been attained by dilution and combustion of ethylene 19 or jet fuel 20 soot in a lean premixed flame. 21 However, this exhaust treatment system does not resemble the dilution zones in common RQL combustors. 18 Similarly, "soot-free" combustion of jet fuel was attained recently in a laboratory-scale lean azimuthal flame (LEAF) combustor of jet A1 fuel by enhancing soot oxidation while injecting hydrogen. 22 Here, enclosed spray combustion (ESC) of jet fuel ( Figure  S1) that produces surrogate aircraft soot emissions 23 is used to explore their elimination. During ESC of jet fuel, soot nanoparticles grow by surface reactions 24 and agglomeration, 25 attaining similar morphology, size distribution, and organic carbon content with those of aviation emissions. 23 Most importantly, the Raman spectrum of soot from ESC of jet fuel is in excellent agreement with that measured from aircraft soot 26 ( Figure S2). This indicates that the oxidative reactivity of such surrogate aircraft soot is similar to that of aviation emissions. 16 So, the elimination of such soot is investigated here by injecting N 2 containing 0−25 vol % of O 2 downstream of ESC of jet fuel. The impact of such O 2 addition on the soot mobility, primary particle size distributions, f v , N t , composition, and nanostructure is elucidated below for the first time to the best of our knowledge. That way, the transformation of soot during oxidation is quantified, providing a basis for optimization of the RQL concept that is already used by some aircraft engine manufacturers. 11

MATERIALS AND METHODS
Soot nanoparticles were generated by ESC. Briefly, soot was produced by jet A fuel (POSF 10325 27 ) spray combustion using an external-mixing, twin fluid nozzle 28 enclosed in two, 30 cm long quartz tubes (each with a 42 mm inner diameter) in series 29 ( Figure S1). So, 4 mL/min of fuel was dispersed into a fine spray with 1 L/min of O 2 . The resulting spray was ignited and sustained by a supporting premixed methane/ oxygen flame (CH 4 = 1.25 L/min, O 2 = 2.25 L/min). Sheath air was fed through 12 evenly spaced holes surrounding the spray flame at 17.2 L/min. A torus ring 30 with 12 jet outlets between the two tubes (height above burner, HAB = 30 cm) was used to introduce 20 L/min of N 2 with or without O 2 in an upward swirled pattern to quench the flame as well as to dilute and oxidize the exhaust soot emissions. The O 2 concentration, [O 2 ], was varied from 0 to 25 vol %. The steel torus ring was made using two pieces of pipe welded to a tube ( Figure S3) having a 0.38 cm inner diameter and 12 outlets, each having a 0.06 cm diameter 30 with an upward azimuth angle of 10°.
The temperature profile, T, was measured using a 1 mm (nominal) bead diameter and an R-type thermocouple (Intertechno-Firag AG) and corrected for radiative heat losses. 31 The T measurements and energy balance used here have been described and validated for ESC of jet A1 fuel. 31 The centerline flame T profiles during ESC of jet A (Figure 1a: circles) and A1 (squares) fuel are quite similar. Figure 1b shows that the centerline T by ESC of jet A1 fuel at HAB = 35 (circles) and 63 cm (triangles) increases with increasing oxygen content in the injected nitrogen jets from the torus ring at HAB = 30 cm, as expected.
Soot was extracted from the centerline of the flame at HAB = 63 cm using a straight tube sampler. 32 The sampled aerosol was rapidly diluted and quenched by mixing with N 2, followed by compressed air from a rotating disk diluter. The total dilution factor was set to 33.24 at all conditions investigated here. The distribution of the soot mobility diameter, d m , and its total number concentration, N t , were obtained by averaging five 65 s scans of a scanning mobility particle sizer. 32 The soot f v was estimated based on the measured d m and d p distributions, accounting for the soot agglomerate structure: 33 (1) where N i is the number concentration of soot agglomerates having d m,i and mean d p . The index k varies from 1 to 100, i.e., the largest number of d m bins measured by the scanning mobility particle sizer. The exponents for d p and d m,i in eq 1 were validated with aerosol particle mass analyzer data in premixed, 34 diffusion 25 and spray 23 flames. Equation 1 has been derived by capitalizing on a power law for the soot effective density that was obtained by discrete element modeling of soot agglomeration and surface growth. 25 This equation has been used to measure accurately the soot f v in laminar premixed, 35 diffusion flames, and diesel engines, 36 accounting for the realistic morphology of soot. 33 Soot was also collected on a glass fiber filter for off-line analysis. Then, Raman spectra of such soot nanoparticles were obtained using a 515 nm laser having 50 mW power (Renshaw inVia). The laser was focused with a ×20 magnification optical microscope, which gives a 2 μm spot size, while a 10% laser power was focused on the sample for 120 s and three acquisitions. 37 The intensities of the disorder (D ∼ 1350 cm −1 ) and graphitic (G ∼ 1580 cm −1 ) bands 37 were obtained after straight line subtraction of the baseline. 38 The X-ray diffraction (XRD) patterns of soot at diffraction angles, 2θ = 10−70°, were also obtained by an AXS D8 diffractometer (Bruker) at a scan rate of 0.0197°/s. Here, the average interlayer distance, d, of soot was obtained by analyzing the 002 XRD peak using Bragg's law 39 = · d n 2 sin 002 (2) where n = 1 is the order of diffraction, λ = 0.154 nm is the wavelength of the diffractometer, and θ 002 is the center angle of the 002 peak. Similarly, the average crystallite length, L c , of spray flame soot was obtained by 39 where K = 0.89 is the peak shape factor 40 and β 002 is the full width of the half maximum of the 002 peak. The crystallites d and L c were determined here using eqs 2 and 3 with the θ 002 and β 002 derived from the measured XRD patterns that were validated using the patterns of commercial carbon blacks. 41 The organic to total carbon (OC/TC) mass ratio of soot was obtained by thermogravimetric analysis (TGA). 42 The samples were first placed in N 2 to volatilize OC and then in air to oxidize the elemental carbon (EC). The sample heating began at 30°C in N 2 and was ramped up to 900°C at 20°C/ min. The temperature was held at 900°C for 10 min before dropping back to 30°C at 20°C/min. The same temperature profile was then repeated in air. From the TGA mass loss, the OC/TC was estimated as the ratio of mass lost under N 2 divided by the total mass lost in both stages.
Soot nanoparticles were analyzed by N 2 adsorption on a Tristar II Plus surface area and a porosity system (Micromeritics) at 77.3 K after degassing in vacuum (VacPrep 061, Micromeritics) at 200°C overnight. The specific surface area, SSA, was derived from N 2 adsorbed at five relative pressures ranging from 0.05 to 0.25 using the Brunauer−Emmett−Teller method. 43 Soot nanoparticles were also imaged using transmission electron microscopy (TEM, FEI Tecnai F30 FEG). The nanoparticles were dispersed in ethanol and placed in an ultrasonic bath for 15 min to break up large agglomerates. 23 A drop of ethanol solution was then placed on lacey carbon TEM grids with a 200 mesh copper support (LC200-Cu-150, Electron Microscopy Sciences) and allowed to dry. The primary particle diameter, d p , was measured by manually placing ellipses over the primary particles in ImageJ 44 and calculating the area-equivalent diameter. About 150−200 primary particles were counted for each [O 2 ] condition to obtain statistically significant size distributions. 23

Reducing Surrogate Aviation Soot Emissions by O 2 -Containing Jets.
Extensive recirculation results in radially rather uniform conditions away from the burner, as has been shown for temperature, T, by computational fluid dynamics (CFD) analysis (e.g., Figure 1a in ref 45). To further confirm this for the soot aerosol, its average mobility diameter, d m , f v , and N t were measured at the centerline (r/R = 0) and inbetween the tube wall and centerline (r/R = 0.5; Table S1) at HAB = 25 cm (i.e., well below the location of the torus ring with the 12 N 2 -jets containing O 2 ). The soot N t , f v , and d m at the centerline are similar (within the measurement variation) to those obtained in-between the tube wall and centerline there. This indicates that the soot aerosol has been largely homogenized across the tube radius when it reaches the torus ring (HAB = 30 cm). Further downstream, the soot size distribution becomes even more uniform across the tube due to its intense mixing with the O 2 -containing N 2 jets, as shown in Figure S4 for two radial locations at HAB = 35 and 63 cm, as well as by the corresponding N t and mean d m (Table S2). This indicates that the soot aerosol is well mixed across the enclosing tube, corroborating CFD simulations at similar gasaerosol mixing configurations. 30 Figure 2 shows the soot mobility (a) and primary particle  The latter is derived by accounting for the realistic agglomerate structure of soot that is essential to close its mass balance. 33 Increasing [O 2 ] from 0 to 5 vol % enhances soot f v by 80% (Figure 3) due to the PAH formation and adsorption on the soot surface, 46 consistent with the soot f v increase after injection of small amounts of air downstream of synthetic fuel combustion. 17 Soot N t also increases by 25%. This could be attributed to the inception of nascent soot enabled by the low concentrations 47 and N t , reducing them by 99.6 and 95.4%, respectively. The reduction of soot N t obtained here is on par with the 99.9% N t reduction measured after air dilution and combustion in a lean premixed flame. 20 This indicates that rather uniform soot concentration profiles are attained here ( Figure S4 and Table S2), similar to those in premixed flames. 20 Furthermore, the soot f v reduction measured here is 50% larger than that attained in a laboratory-scale RQL combustor. 17 This could be attributed to potentially more homogeneous mixing of soot with oxidizing gas by employing the current jet configuration. The soot f v = 5 × 10 −11 obtained here at [O 2 ] = 25 vol % is on par with the f v = 3 × 10 −11 to 6 × 10 −11 measured in a so-called "soot-free" LEAF combustor. 22 In fact, the corresponding N t = 3.1 × 10 6 #/cm 3 is 3 orders of magnitude lower than the N t = 3.5−7.5 × 10 9 #/cm 3 measured in LEAF. 22 The largest 95.4% N t reduction of jet fuel emissions attained here using O 2 -containing N 2 jets is about 25−60% greater than that obtained by blending jets with HEFA 7 or FTderived 8 fuels.
3.2. Soot Nanostructure and Composition. Even though the large reduction of soot N t and f v using O 2containing dilution jets is promising, the nanostructure and composition of the remaining soot emissions have to be characterized to assess their impact on public health and climate. In this regard, Figure 4a Figure 4f shows the specific surface area, SSA (circles and a solid line), and organic to total carbon (OC/TC) mass ratio (diamonds and a broken line) of soot produced here at the conditions shown in Figure 3. The SSA of soot correlates with its cytotoxicity 50 and thus it is essential to quantify its impact on public health. At [O 2 ] = 0 vol %, soot nanoparticles have SSA = 292.3 m 2 /g and OC/TC = 5.1%, consistent with those measured from ESC of jet A1 fuel at similar equivalence ratios. 23 Varying [O 2 ] from 0 to 5 vol % decreases the SSA of soot to 130.6 m 2 /g and increases its OC/TC to 8.7%, as small amounts of O 2 enhance PAH formation 46 and thus the soot OC/TC. As [O 2 ] further increases up to 25 vol %, soot nanoparticles are oxidized and their diameter decreases (as discussed in Figure 2), increasing their SSA up to 445.5 m 2 /g. This 50% enhancement of soot SSA attained here is on par with the 30% increase obtained by blending jet with alternative   51 and thus its direct radiative forcing 3 by up to 17%. Atmospheric transformations of particle composition and morphology (e.g., during water processing 52 ) should be accounted for to most accurately quantify the impact of aircraft soot emissions on public health and climate. The impact of O 2 -containing N 2 dilution jets on soot nanostructure is quantified by X-ray diffraction (XRD) and Raman spectroscopy. Figure 4g shows the XRD patterns along with the mean interlayer distance, d, and crystallite length, L c , of soot produced at various [O 2 ]. The pattern of unoxidized ESC soot ([O 2 ] = 0 vol %) exhibits a rather broad 002 peak (broken line) at a diffraction angle, 2θ, of about 24°that yields d = 3.7 Å and L c = 1.4 nm, in agreement with the XRD pattern of unoxidized carbon black 41 and aircraft soot. 26 Figure 4h shows the Raman spectra along with the mean ratio of the disorder (D) over the graphitic (G) band of soot produced at various [O 2 ]. Increasing [O 2 ] from 0 to 5 vol % hardly alters the nanostructure and the Raman spectrum of soot. However, further increasing [O 2 ] from 5 to 20 and 25 vol % reduces D/G to 0.85 and 0.84. This D/G reduction indicates that the average PAH size of soot decrease 53 due to oxidation and small PAH adsorption, consistent with Raman spectroscopy measurements of oxidized carbon black, 54 and soot from premixed 55 and diffusion 56 flames. The small PAH sizes attained after soot oxidation with [O 2 ] = 20 vol % enhance the oxidative reactivity of soot 16 and thus its reactions with ozone in the atmosphere. 57 Most importantly, the amorphous soot emitted after such oxidation has smaller ice nucleation activity than graphitic soot 58 produced in the absence of downstream O 2 here. This can further limit the formation of contrail cirrus clouds and thus their radiative forcing! 3.3. Discussion and Outlook. In conclusion, it is shown that injecting air downstream of jet fuel combustion can drastically reduce its soot emissions. By capitalizing on the quantitative understanding of soot oxidation 48 and, in particular, surface growth and agglomeration dynamics in the ESC reactor and torus ring, 31 it was shown that upward injection of 12 swirling O 2 -containing N 2 jets facilitates close contact of the soot aerosol with oxidizing gas to enable drastic reduction of soot emissions (Figure 3). In particular, the injection of N 2 containing 20−25 vol % of O 2 enhances the oxidation of soot nanoparticles and decreases their N t and f v by 87.3−95.4 and 98.3−99.6%, respectively. Oxidation at these conditions increases the amorphous and organic carbon content of the emitted soot, reducing its light absorption, 51 direct radiative forcing, 3 and ice nucleation activity. 58 The number concentration of ice nuclei formed in the contrails of aircraft engines decreases almost linearly as the soot number concentration decreases from 10 16 down to about 8 × 10 13 #/kg of fuel. 4 Recent measurements have shown that aircraft engines combusting jet A1 fuel release 5 × 10 15 #/kg of fuel (see Figure 4 in ref 8). Injection of O 2 downstream of jet A or A1 fuel combustion reduces the soot N t up to about an order of magnitude (Figure 3). In this N t range, the concentration of ice nuclei seems to decrease linearly with the soot concentration. 4 This suggests that the injection of air downstream of aircraft engines may reduce the radiative forcing from their emissions 6 by at least 50%.
To relate the present results to the emissions of actual jet engines, besides matching fuel and oxidant composition (jet fuel A or A1 and air or [O 2 ] = 20%), one has to match the socalled high temperature particle residence time between ESC and jet engines, as has been shown in the combustion synthesis of nanoparticles (i.e., Figure 7 in ref 59). The scale-up of the present spray combustion reactor has been explored experimentally and numerically 59 up to 2 orders of magnitude 60 where it was shown that the characteristics of flame-made nanoparticles can be preserved by maintaining similar high-temperature particle residence times across scales. In this regard, the present set of data is essential to derive and validate CFD 59 and moving sectional models for soot oxidation from jet fuel combustion. 48 Such models can be used to obtain robust oxidation rates for aircraft soot emissions and facilitate The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c01048.
Schematic of the experimental setup for the generation and elimination of soot from ESC of jet fuel; Raman spectra of soot from ESC of jet fuel and aircraft engines operating at two thrusts; 25 top and side views of the torus ring for injection of O 2 -containing N 2 along with a reference ruler; mobility size distributions of soot from ESC of jet A1 fuel and controlled oxidation at the flame centerline (r/R = 0, dotted lines) and in-between the enclosing tube wall and centerline (1 cm away from each, r/R = 0.5; solid lines) at HAB = 35 and 63 cm along with the corresponding total particle concentration, N t ; and soot N t , f v , and d m at the centerline and in-between there and tube wall 5 cm below and above the torus ring as well as at HAB = 63 cm (PDF)