Shapes of soot aerosol particles and ... - Wiley Online Library

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, D15206, doi:10.1029/2009JD012868, 2010

Shapes of soot aerosol particles and implications for their effects on climate Kouji Adachi,1,2 Serena H. Chung,3 and Peter R. Buseck1,2 Received 21 July 2009; revised 28 January 2010; accepted 5 February 2010; published 14 August 2010.

[1] Soot aerosol particles (also called light‐absorbing, black, or elemental carbon) are

major contributors to global warming through their absorption of solar radiation. When embedded in organic matter or sulfate, as is common in polluted areas such as over Mexico City (MC) and other megacities, their optical properties are affected by their shapes and positions within their host particles. However, large uncertainties remain regarding those variables and how they affect warming by soot. Using electron tomography with a transmission electron microscope, three‐dimensional (3‐D) images of individual soot particles embedded within host particles collected from MC and its surroundings were obtained. From those 3‐D images, we calculated the optical properties using a discrete dipole approximation. Many soot particles have open, chainlike shapes even after being surrounded by organic matter and are located in off‐center positions within their host materials. Such embedded soot absorbs sunlight less efficiently than if compact and located near the center of its host particle. In the case of our MC samples, their contribution to direct radiative forcing is ∼20% less than if they had a simple core‐shell shape, which is the shape assumed in many climate models. This study shows that the shapes and positions of soot within its host particles have an important effect on particle optical properties and should be recognized as potentially important variables when evaluating global climate change. Citation: Adachi, K., S. H. Chung, and P. R. Buseck (2010), Shapes of soot aerosol particles and implications for their effects on climate, J. Geophys. Res., 115, D15206, doi:10.1029/2009JD012868.

1. Introduction [2] Aerosol particles influence global climate by absorbing and scattering sunlight and modifying cloud properties. Soot particles consist of aggregated carbon spherules that contain curved graphitic layers and typically have diameters of a few tens of nanometers. Soot particles are produced from incomplete burning of fossil fuel, biofuel, and biomass and are the dominant particle type having a warming effect on global climate [Moosmüller et al., 2009]. The Intergovernmental Panel on Climate Change (IPCC) report [IPCC, 2007] indicates that their global, annual mean direct radiative forcing (RF) is approximately 0.34 W m−2, making them the largest contributor to global warming after carbon dioxide and methane. Others believe that the role of soot is even greater and that it could be the second largest contributor to global warming, assuming it is internally mixed with other aerosol particles [Jacobson, 2001; Ramanathan and Carmichael, 2008], as is commonly the case. How1 School of Earth and Space Exploration, Arizona State University, Tempe, Arizona, USA. 2 Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona, USA. 3 Department of Civil and Environmental Engineering, Washington State University, Pullman, Washington, USA.

Copyright 2010 by the American Geophysical Union. 0148‐0227/10/2009JD012868

ever, as the IPCC report indicates, current understanding of global warming from soot is poor. Uncertainties include its shapes, global emissions [Bond et al., 2004], refractive index [Bond and Bergstrom, 2006], and lifetime [van Poppel et al., 2005] in the atmosphere. [3] Soot particles are commonly embedded within other materials, and these can affect optical properties such as absorption, scattering, and asymmetry parameters (hereafter “optical properties” collectively indicates these three properties). Embedding materials and coatings (hereafter “host materials or particles”) act as lenses that focus light on soot and thus amplify absorption [Bond et al., 2006; Fuller et al., 1999]. The change in its optical properties in turn affects its RF. Additionally, calculations using a spherical soot particle and its coating suggest that these optical properties are also influenced by the position of the soot within its host particle, i.e., whether it is in the center or at the surface [Bond et al., 2006; Fuller et al., 1999] (in our study, “position” refers to that of soot within its host particle). However, the actual three‐dimensional (3‐D) shapes of individual soot particles embedded within host materials can be complicated, and the light absorption changes caused by the shapes and position are uncertain. [4] Climate models assume that soot particles are spherical and (1) uncoated (external‐mixing model), (2) concentrically coated (core‐shell model), (3) homogeneously mixed with other materials on a molecular scale (volume‐mixing

D15206

1 of 9

D15206

ADACHI ET AL.: SOOT SHAPES AND RADIATIVE FORCING

D15206

Figure 1. 2‐D and 3‐D views of soot embedded within host aerosol particles. (a) TEM image of a particle with embedded soot. The inset in the upper right shows a schematic drawing of the components. Blue dots, OM; gray, soot; red, voids where beam‐sensitive material (presumably ammonium sulfate) was present; and yellow, lacey carbon substrate. (b) 3‐D isosurface image of the particle in Figure 1a. The 3‐D shape was obtained using ET. (c) TEM image of a host particle with embedded soot; the thin fiber is part of the substrate. (d) Cross‐section image of the particle in Figure 1c. The spherules are soot, most of which is located near the surface of the host particle. (e) 3‐D isosurface image of the particle in Figure 1c. Gray and blue indicate soot and OM, respectively. The particle in Figure 1a was collected on March 8 and that in Figure 1c on March 10. model), or (4) mixed with other aerosol particles according to rules such as the Maxwell‐Garnet effective medium approximation (MGEMA), which assumes that isolated soot spherules are suspended in an embedding material [Bond et al., 2006]. Particles matching models 2, 3, and 4 are called internally mixed. The calculated RF values differ by almost a factor of 2 [Bond et al., 2006], depending on the mixing model that is used to compute optical properties. [5] Most ambient soot particles are internally mixed, and the shapes and positions of the soot within their host particles are more complicated than those used in the models. Analyses by using single‐particle mass spectrometry confirm that the external‐mixing model is inadequate because most particles consist of mixtures of different components [e.g., Murphy et al., 2006; Moffet et al., 2008]. Observations using transmission electron microscopy [Adachi and Buseck, 2008; Buseck and Pósfai, 1999; Pósfai et al., 1999; Niemi et al., 2006; Wentzel et al., 2003; Worringen et al., 2008] suggest that the internal‐mixing models do not properly represent the shapes of observed soot particles, which are commonly coated by, embedded in, or aggregated with other materials but retain their graphitic structure and chainlike shapes within the host materials. [6] We used a discrete dipole approximation (DDA) [Draine and Flatau, 1994] to estimate the optical properties of 3‐D particles with arbitrary shape and multiple components, similar to what was done to determine the optical properties of internally mixed particles having modeled

shapes [Martins et al., 1998; Worringen et al., 2008; Lindqvist et al., 2009]. However, the details of the 3‐D shapes of actual particles can only be determined using electron tomography (ET). In this study, ET was used to obtain the 3‐D shapes of soot particles from a series of 2‐D transmission electron microscope (TEM) images obtained systematically along different viewing directions (Figure 1) [van Poppel et al., 2005; Friedrich et al., 2005; Adachi et al., 2007; Friedrich et al., 2009; Weyland and Midgley, 2004; Midgley and Dunin‐Borkowski, 2009]. Combining ET and DDA (ET‐DDA), we estimated the optical properties of individual soot particles embedded within host materials. [7] The contributions from tropical megacities to the global soot budget and the soot warming effect are large [Ramanathan and Carmichael, 2008; Bond et al., 2004], and our samples provide examples of such aerosol particles. The goals of this study are to (1) determine the actual 3‐D shapes of embedded soot particles together with their host materials, (2) calculate the optical properties of such soot particles, and (3) evaluate the effects of shapes and positions of such soot particles on the RF.

2. Materials and Methods 2.1. Sampling [8] Samples were collected from inside and outside of Mexico City (MC) plumes during the Megacity Initiative:

2 of 9

D15206


D15206

Table 1. Samples Used in This Study Sampling Plume

Date (Mar 2006)

MC plume MC plume Outside MC plume Outside MC plume

08 10 10 19

Location 19.23°N, 19.71°N, 21.00°N, 24.25°N,

Starting Local Time

Collection Time (s)

Number of Particles

15:17:29 14:44:45 10:35:25 17:13:37

277 332 267 246

15 10 14 7

260.97°W 260.86°W 262.27°W 263.57°W

Local and Global Research Observations (MILAGRO) campaign in March 2006 [Adachi and Buseck, 2008; Molina et al., 2008] (Table 1). Particles were collected on TEM grids by using 3‐stage impactor samplers (MPS‐3, California Measurements, Inc.). Details of the compositions, sizes, and mixing states of aerosol particles and the sampling conditions were reported by Adachi and Buseck [2008]. For the current study we used four samples collected inside and outside MC plumes. Because of the rapid ventilation of the MC basin, the MC plume samples probably aged for less than 1 day and those from outside the plume are older [Adachi and Buseck, 2008]. The samples have aerodynamic diameters of 50 to 300 nm and were collected aboard the National Center for Atmospheric Research/NSF C130 aircraft. 2.2. Electron Tomography and Discrete Dipole Approximation [9] We used a 200 kV TEM (Tecnai F20; FEI Corp.) for ET analyses. Soot particles embedded within host materials were selected for analyses using the method as described by Adachi et al. [2007]. Between 7 and 15 particles whose images did not overlap with other particles or with the edges of the TEM grid at high tilt angles were chosen from near the center of each grid. During ET measurements, we took ∼140 images for ∼1 h per particle, avoiding unnecessary exposure of the particles to the electron beam. A total of 46 particles (called “standard particles”) were used for the ET‐DDA calculations as representatives of embedded soot particles. Their sizes are within 1s of the median of 8000 particles that were determined by conventional TEM images obtained from the same samples [Adachi and Buseck, 2008]. [10] The 3‐D tomograms of particles that embed soot were reconstructed using ET as described by Adachi et al. [2007]. The resulting volume uncertainty is 7% ± 6%. The reconstructed host particles were separated into soot, organic matter (OM), and sulfate (Figure 1). Except for sulfate and nitrate, which decomposed rapidly, the samples were stable during analysis, with no differences noted before and after analysis. However, we did not consider volatile and semivolatile OM since the samples were analyzed under vacuum (∼10−5 Pa). [11] Even when embedded, soot can be distinguished from OM in TEM images because their structures are markedly different, i.e., graphitic and amorphous, respectively. There is also a contrast difference of ∼10% in the ET images (the contrast between black and white is 100%). Positions and sizes of soot spherules within aggregates were identified from the tomograms (e.g., Figure 1d) and checked using the TEM images. The soot particle shapes are also used for the “uncoated aggregates” in Figure 2 to show the effect of host materials on optical properties. The voids within the host particles are from materials that decomposed

in the electron beam. The most likely and abundant is ammonium sulfate, although nitrates and other sulfates may also have been present [Moffet and Prather, 2009]. We did not consider their detailed compositions here since the refractive indices of these materials are similar to one another. [ 12 ] We used the DDSCAT 6.1 program [Draine and Flatau, 1994] for the DDA calculation in which the 3‐D shapes of the particles were approximated using an array of polarizable points (dipoles). Errors depend on the number of dipoles used to represent the shapes. We commonly used more than 10,000 dipoles for a particle, and the resulting errors are less than 1% for both scattering and absorption [Draine and Flatau, 1994]. We used a wide range of wavelengths (0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.9, 1.0, 1.25, 1.5, 1.75, 2.0, 2.5, 4, 6, 10, and 14 mm) and 100 incident light angles, assuming the particles are randomly oriented in the atmosphere. 2.3. Climate Model [13] The top of the atmosphere RF of soot particles having different shapes and mixing states was calculated using the Goddard Institute for Space Studies General Circulation Model II‐prime [Hansen et al., 1983; Rind and Lerner, 1996] as the differences in shortwave radiative fluxes with

Figure 2. Absorption cross sections as a function of wavelength for various mixing models relative to those obtained using the ET‐DDA calculation for the standard particles. Mie theory was used for the uncoated sphere, core‐shell, and volume‐mixing particles using a code by Bohren and Huffman [1983]. The plots indicate the median values. The wavelength‐dependent refractive indices from the Optical Properties of Aerosols and Clouds (OPAC) model [Hess et al., 1998] were used for soot and sulfate. A refractive index of 1.33 [Myhre and Nielsen, 2004] was used for OM.

3 of 9

D15206

D15206


Table 2. Features of Embedded Soot Particlesa Fractal Radius Soot Number Volume‐ Volume‐ Volume Dimension of Gyration Spherule of Spherules Equivalent Equivalent Volume Fraction Volume Soot of Soot of Soot Diameter in Soot Radius of Soot Radius of Host Fraction of Soot Fraction Particle Particles Particles (nm) (nm) Particles Particles (nm) Particles (nm) of OM Particles of Sulfate Positiona Average Standard deviation

2.2 0.2

192 84

44 12

40 32

69 25

206 71

0.87 0.09

0.07 0.08

0.06 0.06

0.54 0.35

a Soot positions within the host particles are calculated from Di/Rve, where Di indicates the distance between mass center of the soot and that of the host particle. Rve indicates the volume equivalent radius of the host particle (N = 46).

and without soot. For the climate model, we used the standard case of van Poppel et al. [2005] except for the optical properties, which were obtained from the current study. [14] The mass concentrations of soot were calculated online in the climate model based on the emission inventory (as black carbon) of Bond et al. [2004] and aerosol processes described by Chung and Seinfeld [2002, 2005]. Using optical properties determined by the ET‐DDA method discussed above, the mass‐normalized scattering and absorption cross sections of soot for all standard particles were determined by using 1.8 g/cm3 for the soot density [Bond and Bergstrom, 2006]. The median values of mass‐ normalized optical properties were used to calculate the RF values. [15] Sources of uncertainties in the estimate of RF from soot, in addition to its optical properties, include emission rates, atmospheric aging, and lifetime, all of which determine its atmospheric loading and hence directly impact its RF. The RF is also affected by the spatial and vertical distributions of soot in the atmosphere, e.g., soot particles over high‐albedo surfaces such as snow and ice or above clouds. Such particles contribute to greater RF than those do over low‐albedo regions. Except for the optical properties, we used the same assumptions when calculating and comparing the values of the RF for the other mixing models. In those cases, the relative values in RF reflect only the differences in optical properties arising from differences in the mixing states, shapes, and positions of soot within its host particles.

3. Results 3.1. Morphological Features of Embedded Soot Particles [16] In areas of high pollution such as MC and its surroundings, much of the soot is embedded within OM together with materials such as ammonium sulfate and nitrate [Moffet and Prather, 2009; Adachi and Buseck, 2008]. We determined the 3‐D shapes, volume fractions, and positions (mass centers) of the soot and host particles (Table 2). OM is the dominant host material, and soot occupies an average of 7% by volume of its host particles. The average fractal dimension of the soot particles is 2.2, suggesting they are not completely compacted (the fractal dimension of a sphere is 3.0). Relatively long distances between the mass centers of the soot and those of the host particles suggest that most soot lies near the surface rather than center of the host particles. These features are similar for samples from both within and outside the MC plumes.

[17] The average volume‐equivalent diameter of our measured particles is 412 nm (Table 2), which is larger than the sampled aerodynamic diameter (50 to 300 nm), a situation that is common for irregularly shaped particles [DeCarlo et al., 2004]. In MC, particles with aerodynamic diameter of ∼300 nm dominated in both mass and number [e.g., Cross et al., 2009; Kleinman et al., 2009] and provided the dominant contribution to light extinction [Shinozuka et al., 2009]. Moffet and Prather [2009] measured sizes, light scattering, and compositions of internally mixed soot particles from a MC ground site during the same campaign. They estimated soot core sizes to be