Aviation and the Global Atmosphere

Other reports in this collection

3.2. Aerosol Emission and Formation in Aircraft Plumes

Aircraft jet engines directly emit aerosol particles and condensable gases such as water vapor (H2O), sulfuric acid (H2SO4), and organic compounds, which lead to the formation of new, liquid (volatile) particles in the early plume by gas-to-particle conversion (nucleation) processes. Other gas-phase species and charged molecular clusters (chemi-ions, or CIs) are also generated at emission, including nitric acid (HNO3) and nitrous acid (HNO2). Emission and formation of H2SO4 depend on fuel sulfur content, or sulfur emission index [EI(S)], and the conversion fraction of fuel sulfur to H2SO4. Formation of HNO3 and HNO2 depends on reactions of nitrogen oxides (NOx = NO + NO2) with hydroxyl radicals (OH). Particle formation depends on mixing of exhaust gases with ambient air, plume cooling rate, plume chemistry, and ambient aerosol properties. Soot particles formed during fuel combustion and emitted metallic particles constitute the solid (nonvolatile) particle fraction present in exhaust plumes. Under certain thermodynamic conditions, emitted water vapor condenses and freezes to form water-ice particles, thereby producing a condensation trail (contrail). These line clouds evaporate rapidly if the ambient humidity is low but may change the size and chemical composition of the remaining liquid aerosol particles. If the humidity is above ice saturation, contrails persist and grow through further deposition of ambient water.

An invisible aerosol trail is always left behind cruising aircraft. Aerosol and contrail formation processes in an aging plume determine the number, surface area, and mass of particles that are formed per mass of fuel consumed. Exhaust particle properties change in the presence of a contrail. Exhaust particle morphology and surface properties and aircraft-induced perturbations of background aerosol surface areas (Section 3.3) are of central importance for ozone changes caused by heterogeneous chemical reactions (Chapters 2 and 4). Particle number and freezing probability are key for the formation of ice (cirrus) clouds after passage of an aircraft in a region where otherwise no clouds would form (Section 3.4). Finally, aviation-produced aerosol can directly or indirectly influence the radiation budget of the atmosphere (Section 3.6 and Chapter 6). For recent reviews see Schumann (1996a), Fabian and K�rcher (1997), Friedl (1997), and Brasseur et al. (1998).

The following subsections provide a description of volatile aerosol precursors and the formation of volatile aerosol particles, a characterization of emitted soot and metal particles, a review of contrail and ice formation, and a discussion of the mutual interactions between these particle types. Comments on reducing the impact of aerosols are given in Section 3.7.4.

3.2.1. Volatile Aerosol Precursors Water Vapor

Water vapor is present in aircraft exhaust in known amounts because the emission index is specified by the stoichiometry of near-complete fuel combustion (Chapter 7). Water vapor concentrations of a few percent at the engine exhaust nozzle far exceed the concentrations of other aerosol precursor gases. Ambient water vapor also participates in aerosol processes, with concentrations that vary widely depending on flight altitude and meteorological processes. Because of its abundance and thermodynamic properties, water vapor participates in nearly all aerosol formation and nucleation processes (e.g., Pruppacher and Klett, 1997). Sulfur Species

Aviation fuels (kerosene) contain sulfur in trace amounts. In the current world market, the sulfur content-hence the EI(S)-of aviation fuels is near 0.4 g S/kg fuel or 400 parts per million by mass (ppm; 1 ppm = 0.0001%), with an upper limit specification of 3 g S/kg (Chapter 7). Of importance for the formation of plume aerosol is the partitioning of sulfur at the engine exhaust nozzle into sulfur dioxide (SO2) and fully oxidized sulfur S(VI) compounds, sulfur trioxide and sulfuric acid (S(VI) = SO3 + H2SO4). Most fuel sulfur is expected to be emitted as SO2 based on combustion kinetics and some observations (Miake-Lye et al., 1993, 1998; Arnold et al., 1994; Schumann et al., 1998). However, a fraction of the SO2 can be converted into S(VI) by gas phase chemical reactions with OH, oxygen atoms (O), and H2O inside the engine. The fractional conversion depends on details of combustion conditions, turbine flow properties, blade cooling effects (Chapter 7), and mixing (Chapter 2). Further oxidation can occur in the plume, where the rate-limiting step is thought to be oxidation of SO2 by OH to form SO3 (Stockwell and Calvert, 1983) or liquid-phase reactions of SO2 with H2O2, O3, metals (Jacob and Hofmann, 1983), or HNO3 (Fairbrother et al., 1997). Once SO3 is formed, the gas-phase reaction with emitted H2O to form H2SO4 is fast (< 0.1 s) under plume conditions (Reiner and Arnold, 1993; Kolb et al., 1994; Lovejoy et al., 1996). Gaseous H2SO4 and HSO4-(H2SO4)n (mostly with n = 1,2) ion clusters have been observed in jet exhaust (Frenzel and Arnold, 1994; Arnold et al., 1998a,b).

Figure 3-1: Aerosol and contrail formation processes in an aircraft plume and wake as a function of plume age and temperature.

The chemical lifetime of exhaust OH in the early jet regime is determined by reactions with emitted NOx and by OH self-reactions, the latter leading to the formation of hydrogen peroxide (H2O2) (K�rcher et al., 1996a; Hanisco et al., 1997). Measurements indicate OH exit concentrations below 1 ppmv (Tremmel et al., 1998). For an OH concentration of 0.5 to 1 ppmv at the engine's nozzle exit plane and without SO3 emissions, the OH-induced pathway alone yields about 0.3 to 1% S-to-H2SO4 conversion in the plume (Miake-Lye et al., 1993; Danilin et al., 1994; K�rcher et al., 1996a). Model calculations indicate overall S(VI) conversion fractions in the range of 2 to 10% for various supersonic and subsonic jet engines (Brown et al., 1996a; Lukachko et al., 1998; Chapter 7), consistent with some earlier SO3 measurements behind gas turbines (e.g., CIAP, 1975; Hunter, 1982). Chemi-ions

A large number of chemi-ions (CIs) are expected to be present in aircraft exhaust because ion production via high-temperature chemical reactions is known to occur in the combustion of carbon-containing (not necessarily sulfur-containing) fuels (e.g., Burtscher, 1992). In the jet regime, some recent models indicate that CIs effectively promote formation and growth of electrically charged droplets containing H2SO4 and H2O (Yu and Turco, 1997). In addition, CIs may contribute to the activation of exhaust soot. Positive ions include H3O+ and organic molecules like CHO+, C3H3+, and larger molecules (Calcote, 1983), whereas the free electrons rapidly attach to other molecules to form negative ions with sulfate and nitrate cores. Measurements of positive CIs in exhaust plumes are not available, and only very few in situ measurements of negative CIs are available to date.

Arnold et al. (1998a) measured a total negative CI concentration of 3 x 107cm-3 (about 3 x 1015/kg fuel) at plume ages of around 10 ms in the exhaust of a jet engine on the ground, consistent with approximately 109 cm-3 at a plume age of 1 ms (Yu et al., 1998).

That concentration represents a lower bound from diffusion losses of these particles within sampling devices prior to detection and the limited detection range of the employed mass spectrometer. The fact that these measurements yielded only a fraction of CIs in the plume has been partially confirmed by in-flight measurements (Arnold et al., 1998b) showing smaller total CI count rates for high-sulfur fuel compared with low-sulfur fuel. Therefore, current CI data are consistent with a CI emission index of about 2-4 x 1017/kg, corresponding to a concentration of about 2 x 109 cm-3 at the engine exit. This value has been estimated numerically based on coupled ion-ion recombination kinetics and plume mixing (Yu and Turco, 1997; K�rcher et al., 1998b; Yu et al.,1998). Although not directly comparable, CIs have been observed in hydrocarbon flames at concentrations of about 108 to 1011 cm-3 (Keil et al., 1984), supporting the estimated concentration of negative CIs. Nitrogen Species

The primary nitrogen emission from aircraft is in the form of NOx (Chapter 2). In reactions of NOx with OH in the plume, gaseous HNO2 and HNO3 are formed. Despite larger reaction rates, less HNO3 is formed than HNO2 because the ratio of NO2 to NO at the engine exit is small (< 0.2) (e.g., Schulte et al., 1997) (see Chapter 7). HNO3 can also form in the plume even in the absence of NO2 emissions (K�rcher et al., 1996a). In situ measurements in young plumes revealed both HNO2 and HNO3 concentrations above background levels (Arnold et al., 1992, 1994; Tremmel et al., 1998). HNO3 can be more abundant in plumes than H2SO4, especially for low EI(S) values. These acids (especially HNO3) are important because they can be taken up by water-soluble exhaust particles and form stable condensed phases such as nitric acid trihydrate (NAT = HNO3�3H2O) and liquid ternary (H2O/H2SO4/HNO3) solutions under cold and humid plume conditions (Arnold et al., 1992; K�rcher, 1996). Hydrocarbons

Aircraft engines emit non-methane hydrocarbons (NMHCs) as a result of incomplete fuel combustion. These species include alkenes (mostly ethene), aldehydes (mostly formaldehyde), alkines (mostly ethine), and a few aromates. A few (8 to 10) species were found to account for up to 80% of NMHC emissions (Spicer et al., 1992, 1994). High levels of carbonyl compound emissions (on the order of 0.2 ppmv) also have been observed in a combustor (Wahl et al., 1997). Some in-flight data indicate that NMHCs with up to 8 carbon atoms have EIs in the range 0.05 to 0.2 g C/kg fuel and represent approximately 70% of total NMHC emissions (Slemr et al., 1998). However, the current database on NMHC emissions and on partitioning between individual compounds is small and perhaps not representative for all engine types. Some emitted NMHCs might act as aerosol-forming agents in nascent plumes and may be adsorbed or dissolved in plume particles, thereby possibly contributing to the total amount of volatile aerosol found in plumes (K�rcher et al., 1998b). In addition, the presence of trace NMHCs amounts may facilitate nucleation (e.g., Katz et al., 1977) and alter the hygroscopic behavior and growth rates of particles (Saxena et al., 1995; Cruz and Pandis, 1997). Engines also may emit volatile particles containing engine oils or other lubricants, but this effect has not been quantified.

Aircraft also occasionally introduce hydrocarbons by jettisoning fuel at low altitudes in the troposphere. Most of the fuel evaporates while it falls to the ground (Quackenbush et al., 1994), which leads to a small increase of hydrocarbons in this region. Because of the small amounts of fuel released in this way, no essential impacts on atmospheric aerosols are expected.

3.2.2. Volatile Particles Basic Processes

Volatile particles form in the exhaust plume of an aircraft as a result of nucleation processes associated with the emission of aerosol precursors (Hofmann and Rosen, 1978). Typical aerosol parameters in a young plume are included in Table 3-1 for reference. The newly formed particles grow by condensation (uptake of gaseous species) and coagulation (particles collide and attach) in the expanding plume. Coagulation processes involving charged particles originating from CI emissions are more effective because charge forces enhance collision rates. These processes are schematically presented in Figure 3-1. Key processes at young plume ages are determined mainly from the results of simulation models because of the lack of suitable plume measurements.

Previous models (Miake-Lye et al., 1994; K�rcher et al., 1995; Zhao and Turco, 1995; Brown et al., 1996b; Taleb et al., 1997; Gleitsmann and Zellner, 1998a,b) show that particles form when the condensing species, primarily H2SO4/H2O, reach concentrations critical for binary homogeneous nucleation in the expanding and cooling exhaust gas. Because concentrations fall below nucleation thresholds as the plume further dilutes, the amount of H2SO4 in the early stages of the plume (< 1 s) controls the formation of new volatile particles. More recent models emphasize the role of CI emissions in volatile particle formation and growth (Yu and Turco, 1997, 1998a,b; K�rcher, 1998a; K�rcher et al., 1998a). Figure 3-2 shows the size distributions of exhaust particles at a plume age of 1 s, as inferred from models and a few measurements. In the radius range below 10 nm, volatile particles containing H2SO4 and H2O dominate the overall distribution. Model analyses of near-field particle measurements strongly suggest that the volatile particle size distribution exhibits a bimodal structure (Yu and Turco, 1997), with smaller particles formed by the aggregation of homogeneously nucleated clusters of hydrated H2SO4 molecules (neutral mode) and larger particles formed by rapid scavenging of charged molecular clusters by CIs (ion-mode). Only particles from the ion mode are expected to grow beyond the smallest detectable sizes (radius ~2 to 3 nm) of particle counters. Soot and ice contrail particles are significantly larger than the volatiles. An approximate stratospheric size distribution is shown for comparison. In contrast to soot and ice particles (see below), volatile particle spectra are mainly derived from numerical simulation models because of the lack of size-resolved, in situ particle measurements in the nanometer size range. However, the use (in field measurements) of multiple particle counters with different lower size-detection limits allows derivation of the mean sizes of observable particles as a function of fuel sulfur content and plume age (K�rcher et al., 1998b; Yu et al., 1998), thereby providing strong criteria to test the validity of model results.

Table 3-1: Summary of number mean radius, number density, and surface area density for sulfate and soot particles in aircraft plumes and in the background atmosphere, and for ice particles in contrails and cirrus. Flight levels of subsonic (supersonic) aircraft are in the 10-12 km (16-20 km) range. Also included are estimates of zonal mean perturbations to sulfate and soot properties caused by the 1992 aircraft fleet.
Density Radius
Number Density
Surface Area Density
(mm2 cm-3)
Plume (1 s)a 0.002 (1-2) x 107 500-1000
Background (10-12 km)b 0.01-0.1 50-1000 1-6 (10-40)
Background (20 km, non-volcanic)c 0.07 5-10 0.5-1
Background (20 km, volcanic)d 0.2-0.5 10-100 10-40
Plume (1 s)e 0.01-0.03 5 x 104- 5 x 104 50-5000
Background (10-20 km)f 0.05-0.1 0. 01-0.1 3 x 10-5- 3 x 10-2
Young Contrail (0.1-0.5 s)g 0.3-1 104-105 104-105
Persistent Contrail (10 min to 1 h)h 1-15 10-500 103-104
Young Cirrusi 5-10 1 102-104
1992 Aircraft Perturbationj      
Sulfate Aerosol (10-12 km, 50-60�N) 0.01 90-900 0.1-1.1
Soot (10-12 km, 50-60�N) 0.02 3-30 0.02-0.2

a) Detectable only by ultrafine particle counters (particles smaller than 2-3 nm radius are not detected). Calculations by Yu and Turco (1997) for average FSC consistent with observed data.
b) Properties highly variable; size distributions often bimodal. Ranges include small (> 10 nm) particles. Large particle mode (~100 nm) often similar to stratospheric aerosol particles (Hofmann, 1993; Yue et al., 1994; Schr�der and Str�m, 1997; Solomon et al., 1997; Hofmann et al., 1998). High range of values inferred from satellite extinction data and represents mixtures of aerosols and subvisible clouds.
c), d) Yue et al., 1994; Kent et al., 1995; Borrmann et al., 1997; Thomason et al., 1997b.
e) Hagen et al., 1992; Rickey, 1995; Petzold et al., 1999.
f) Only largest soot particles with longest atmospheric lifetimes are measured by wire impactors (Sheridan et al., 1994; Blake and Kato, 1995; Pueschel et al., 1997). Uncertainties in total surface area introduced by fractal geometry of particles.
g) K�rcher et al., 1996b, 1998a; Petzold et al., 1997.
h) Values representative of contrail core for low ice-supersaturation (Heymsfield et al., 1998a; Schr�der et al., 1998b) (see also Sections 3.4.4 and 3.6.3). Far larger particles are observed for large ice-supersaturation (Knollenberg, 1972; Gayet et al., 1996).
i) Str�m et al., 1997; Schr�der et al., 1998b. Larger values are observed in warm cirrus clouds (Heymsfield, 1993; see also Sections 3.4.4 and 3.6.3).
j) Results of fuel tracer simulations discussed in Section 3.3.4. Values shown represent upper bounds to zonal mean perturbations caused by emissions of the 1992 aircraft fleet. Results are representative of flight levels at northern mid-latitudes and are calculated using the range of values of computed tracer concentrations from all models and assuming a fuel sulfur content of 0.4 g/kg fuel, a 5% conversion of sulfur to sulfate aerosol, an EI(soot) of 0.04 g/kg fuel, and a mean particle size of 10(20) nm for sulfate (soot) particles.

Once formed, the new volatile particles interact with nonvolatile and contrail ice particles through the processes of coagulation, freezing, condensation, and evaporation (Figure 3-1). Calculations show that the new liquid particles grow and shrink as a function of relative humidity, whereas H2SO4 molecules that enter the droplets stay in the liquid phase because of their very low saturation vapor pressure (Mirabel and Katz, 1974). They also suggest that volatile particles may take up HNO3 and H2O in the near field (K�rcher, 1996) to form particles with compositions similar to those found in cold regions of the stratosphere. These particles may persist in cold (< 200 K), HNO3-rich stratospheric air but will be short-lived (< 1 min) otherwise. As the plume continues to dilute with ambient air, abundant newly formed volatile particles remain at nanometer sizes and therefore add substantially to the overall aerosol surface area and abundance (Danilin et al., 1997). Their efficiency for heterogeneous chemistry and cloud formation, however, is size- and composition-dependent (K�rcher, 1997). They may be too small to act as efficient cloud- or ice-forming nuclei in the background atmosphere unless the air mass containing the aerosol is lifted or cooled or the relative humidity increases. Although studies exist on heterogeneous plume processing along selected trajectories (Danilin et al., 1994), systematic investigations of heterogeneous chemistry coupled to plume aerosol dynamics remain to be performed (Chapter 2).

Figure 3-2: Size distribution of various aerosol types present in young jet aircraft exhaust plumes (adapted from K�rcher, 1998a).

The evolution of volatile particles is significantly altered if a contrail forms. In contrails, volatile particles have to grow to sizes greater than about 100 nm via uptake of ambient H2O before most of them freeze (Section As ice particles grow in size by deposition of H2O, they may also scavenge other volatile and soot particles (Anderson et al., 1998a,b; Schr�der et al., 1998a). Thus, contrails are expected to contain fewer small particles than non-contrail plumes because of enhanced scavenging losses. After evaporation of contrail ice crystals, the residual volatile and soot cores remain as particles in the atmosphere (Figure 3-1). This contrail processing is expected to modify the particle size distribution and composition and may lead to efficient cloud condensation nuclei production (Yu and Turco, 1998b). Observations and Modeling of Volatile Particles and Sulfur Conversion

Volatile particle abundances observed in situ in the plumes (mostly young, < 100 s) of subsonic and supersonic aircraft are summarized in Figure 3-3a. The data have been compiled from various field studies (Fahey et al., 1995a,b; Schumann et al., 1997; Anderson et al., 1998a,b; Schr�der et al., 1998a). The results show EIs for ultrafine volatile aerosol particles (nominal radii > 2 to 3 nm) in the range of 1015 to 1016/kg fuel for low to average fuel sulfur content values and exceeding 1017/kg fuel for high-sulfur fuel. Besides the obvious dependence on fuel sulfur content, the spread in EI values may be explained by differences in the emission characteristics of the engines, variations in the lower size detection limits of the particle counters, and differences in plume ages at the time of the observations. The increase in ultrafine particle abundance with increasing fuel sulfur content for the Advanced Technology Testing Aircraft System (ATTAS), T-38, and B757 aircraft strongly suggests an important role for fuel sulfur in the growth of volatiles from molecular clusters to detectable particles.

Only a few observations have been analyzed using detailed microphysical simulation models (Brown et al., 1996a; Danilin et al., 1997; K�rcher and Fahey, 1997; Yu and Turco, 1997, 1998a; Andronache and Chameides, 1998; K�rcher et al., 1998a,b). Simulations show better agreement between calculated and observed particle concentrations when ion effects are taken into account. More important, the description of plume microphysics using binary homogeneous nucleation failed to explain a field measurement (Yu et al., 1998). In two cases, condensation nucleus observations in the exhaust of the ATTAS and the Concorde (Figure 3-3a) have been explained in detail with a model that includes CI emissions on the order of 1017/kg fuel. The observable (ion mode) particles have mean radii of about 2 to 4 nm in the young plume, for EI(S) ranging from average to high values. For decreasing levels of available H2SO4, the ion mode particles decrease in size. The number of detected particles falls below 1017/kg fuel when the mean radius of their size distribution becomes smaller than the detection limit of the particle counters.

Figure 3-3: (a) Emission indices of detectable volatile particles in number per kg fuel measured in situ in plumes of various subsonic aircraft and the supersonic Concorde. (b) Same as (a), but for the emission indices of soot particles.

The extent of conversion of fuel sulfur to S(VI) necessary to explain the observed mass of volatile aerosol in young plumes seems to be variable. Direct measurements of H2SO4 have provided a lower bound of ~0.4% (for high-sulfur fuel, 2.7 g/kg fuel) and an upper bound of ~2.5% (for low-sulfur fuel, 0.02 g/kg fuel) for the conversion fraction in one case (Curtius et al., 1998), consistent with calculated SO3 emission levels (Brown et al., 1996a,c). For the low fuel-sulfur case, it has been demonstrated that the observed volatile particles cannot be mainly composed of H2SO4 (K�rcher et al., 1998b). In other cases, conversion fractions have been indirectly inferred from mass balance arguments involving observed or inferred particle size and number distributions, available sulfur as measured in fuel samples, and assumptions of aerosol composition. The dependence of the conversion fraction on EI(S) differs in the few studies performed to date (Fahey et al., 1995a; Schumann et al., 1996; Hagen et al., 1998; Miake-Lye et al., 1998; Pueschel et al., 1998). The values deduced from these measurements range from the minimum value of 0.4% to more than 20%. Some of the indirect analyses may be affected by uncertainties (possibly about 20%) regarding the sulfur content in the fuels. Other experimental uncertainties are associated with these determinations, and the range of aircraft engines and operating parameters adds to the observed variability.

Generalization of these results cannot be done with confidence because of limited empirical knowledge of the S(VI) conversion fraction and emission levels of CIs representative of the exhaust as it enters the atmosphere. Part of the observed volatile aerosol may be composed of HNO2 (Zhang et al., 1996; K�rcher, 1997) or NMHCs (K�rcher et al., 1998b), or it may result from unrecognized sulfur oxidation reactions (Danilin et al., 1997; Miake-Lye et al., 1998). Uptake of gaseous SO2 and subsequent heterogeneous oxidation to sulfate in the new volatile particles is likely small (K�rcher, 1997), but the possibility of other condensational or aqueous-phase growth mechanisms has not yet been fully explored. On the other hand, CI emissions of about 1017/kg fuel are consistent with observations (see Section, although they need to be confirmed by further measurements at the engine exit. The best estimate for the S-to-H2SO4 conversion fraction in young plumes is 5%, with an estimated variability, or uncertainty range, of 1 to 20%. Combustion models predict a range of sulfur conversion up to about 10% and a potential for slightly higher values because of turbine blade cooling effects (Sections and 7.6). Low conversion fractions of 2% are sufficient to explain observed volatile particle concentrations in the young plume behind the ATTAS when the effects of CIs are taken into account in simulation models of plume chemistry and microphysics (K�rcher et al., 1998a,b; Yu and Turco, 1998a,b; Yu et al., 1998). In contrast, the Concorde observations can be explained by assuming that about 20% of the fuel sulfur is converted to SO3 before leaving the engine exit in such simulations (Yu and Turco, 1997).

In situ measurements detailing particle volatility and size distributions such as those included in Figure 3-3 have involved relatively young plumes. Further observations in aging plumes (> 1 h) as they dilute with the background atmosphere are currently lacking. Without detailed observations of the microphysical evolution and chemical composition of volatile exhaust particles from the engine exhaust plume to the global scale, important uncertainties remain in assessing the potential global impact of exhaust products on chemistry and cloudiness.

Other reports in this collection