GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO. 21, PAGES 4075-4078, NOVEMBER 1, 2001
Spectral dependence of visible light absorption by
carbonaceous particles emitted from coal combustion
Tami C. Bond
Pacific Marine Environmental Laboratory, NOAA, Seattle, Washington
Abstract. Optical characteristics of particles that absorb
visible light are needed to model their effects on atmospheric
radiation. Light absorption by particles emitted from lowtechnology coal combustion has exhibited a strong spectral
dependence. I investigate various explanations for this phenomenon and conclude that a spectrally dependent imaginary refractive index is the most plausible. Following previous work on the structure of amorphous carbon, I propose
that both the magnitude and spectral dependence of light
absorption are controlled by the size of graphitic clusters
within the material, and can be described using the optical band-gap theory. This hypothesis is an alternative to
the current measurement divisions of light-absorbing “black
carbon” and non-absorbing “organic carbon,” and offers an
explanation for preferential absorption at blue wavelengths
that may extend to ultraviolet wavelengths.
1. Introduction
Particles that absorb visible light—often called “black
carbon” or BC—are emitted from the incomplete combustion of carbon-based fuels. These particles affect the radiative balance of the Earth by absorbing solar radiation,
warming the atmosphere, and cooling the surface. Climatic
effects of BC have been assessed using simulations of transport, chemistry, and radiation [Haywood and Ramaswamy,
1998; Myhre et al., 1998; Penner et al., 1998; Jacobson,
2001]. Uncertainties in optical properties affect the accuracy of such models.
Modelers have simulated the spectral dependence of light
absorption by using spectrally dependent imaginary refractive indices [Penner et al., 1998] or by identifying compounds that absorb ultraviolet (UV) light [Jacobson, 1999].
However, light absorption by atmospheric particles is often
measured at a single wavelength [e.g., Malm et al., 1994;
Martins et al., 1998; Anderson et al., 1999]. To extrapolate
the measurement to other wavelengths, the spectral dependence of absorption is sometimes represented as a power-law
relationship using an “Ångstrom exponent” (åap ) similar to
that used for extinction:
αap = Kλ−åap
(1)
Here, αap is the absorption efficiency of the particles (m2
g−1 ), K is a constant, λ is the wavelength of light, and the
subscript ap indicates absorption by particles. For “small”
spherical particles with wavelength-independent refractive
index, åap = 1.0 [van de Hulst, 1957]. When absorption is
stronger at shorter wavelengths, the particles appear yellow,
and åap is greater than 1.0. I do not defend the power-law
relationship here, but use it to parallel previous treatments.
2. Observations
Rosen et al. [1979] measured åap = 1.0 in an urban environment, a value that is often assumed in other situations. However, Patterson and McMahon [1984] reported
stronger spectral dependence for particles from smoldering
combustion. Millikan [1961] found a range of åap (0.6–1.8)
depending linearly on hydrogen content for flame-generated
particles. Bond et al. [1999a] reported values of åap between
1.7 and 2.5 for industrial combustion of lignite. Absorption
by biomass-burning particles at UV wavelengths allows detection by the Total Ozone Mapping Spectrometer (TOMS)
[Herman et al., 1997].
Elsewhere, we have reported details of emission measurements from domestic combustion of bituminous coal [Bond
et al., 2001]. This type of combustion contributes heavily to
the budget of BC in some regions—nearly 50% of the emissions in China [Streets et al., 2001]. Briefly, particles were
collected on Nuclepore filters (0.2 µm pore size) to measure
mass and light absorption by integrating plate (435, 525,
660, 800 nm). Light scattering was measured at 550 nm, and
particle size distributions showed number maxima around
100 nm diameter. Some samples encompassed all burning
phases (heating, flaming, and solid-phase), while other samples isolated specific burning conditions (choked or naturaldraft combustion). Samples from this experiment showed a
strong spectral dependence of absorption (1.0 < åap < 2.9)
and an inverse correlation between åap and absorption efficiency. In the following discussion, I seek to explain the
light-absorbing characteristics of these particles.
3. Possible Causes
As stated earlier, assuming small spherical particles with
constant refractive index predicts a value of åap = 1.0. Several departures from assumptions could explain the higher
values of åap observed in the coal emissions: (1) Artifacts
resulting from filter analysis are spectrally dependent. (2)
The particles are not “small” relative to the wavelength of
light. (3) The particles are not spheres, but aggregates of
spherules or coated spheres. (4) The refractive index of the
particles has a spectral dependence resulting from the molecular composition. Below, I examine each of these hypotheses
in turn.
Copyright 2001 by the American Geophysical Union.
3.1. Measurement Artifacts
Paper number 2001GL013652.
0094-8276/01/2001GL013652$05.00
Filter-based measurements of light absorption respond to non-absorbing particles (the “scattering artifact”)
4075
4076
BOND: LIGHT ABSORPTION BY CARBONACEOUS PARTICLES
0.0
Imaginary refractive index, k
1.2 1.0
0.5
0.1
0.0
1.2
1.5
1.2
1.0
1.5
0.5
0.5
0.01
1.2
1.2
1.5
1.5
1.2
1.0
1.2
1.0
0.001
10
100
1000
Diameter (nm)
Figure 1. Values of light absorption exponent, åap , estimated
by Mie calculations with real refractive index 1.55 for lognormal
size distribution with count median diameter as shown and geometric standard deviation of 1.5.
[e.g., Horvath, 1997]. Spectral dependence in either the scattering artifact or the light scattering could affect the apparent spectral dependence of absorption. In a laboratory calibration of the integrating plate, the scattering artifact was
estimated at 550 nm [Bond et al., 1999b]. Unpublished data
from that experiment include measured scattering and absorption by non-absorbing particles at multiple wavelengths.
Based on these data, I found that the scattering artifact at
450 and 700 nm does not differ significantly from the artifact
at 550 nm (9% ± 3% of scattering measured as absorption).
For each coal sample, I estimated light scattering at each
wavelength by applying Mie theory to the measured size distributions. Adjusting the absorption with the wavelengthdependent scattering correction changed åap by only 1–6%.
I conclude that the high values of åap do not result from
filter artifacts.
3.2. Particle Size
The condition that particles must be small relative to
the wavelength of light is given by (2πr|m|/λ 1) [van de
Hulst, 1957], where r is the particle radius and m = n − ik
is the complex index of refraction. This criterion depends
upon the value of |m| as well as the particle size. Figure 1
shows values of åap derived from Mie calculations for wavelengths between 450 and 550 nm, for a range of particle sizes
and spectrally independent imaginary refractive indices (k).
The figure shows that åap = 1 for a wide range of particle
sizes and k, and that the value of åap decreases as particles
become larger. The value of åap exceeds 1.0 for a narrow
range of particle sizes when k < 0.1, and it is greater than
2.0 only when k < 0.0004. The values of k that correspond
to high åap are much lower than any typically assumed for
black carbon. For the measured particle size distributions
and a refractive index of 1.55–0.5i, the theoretical value of
åap varies from 0.2 to 1.2, with smaller values associated
with larger particles.
3.3. Morphology
BC particles often consist of aggregated spherules around
10 nm in diameter, and Mie calculations that assume spherical particles may not be applicable. However, light absorption by aggregates is the sum of the absorption by the
component spherules, as long as the fractal dimension is below 2.0 [e.g., Nelson, 1989]; that is, each spherule acts alone
if there is no shadowing. The limiting cases are the sum
of the component spherules (open particle) and the solid
sphere (closed particle), with realistic particles probably ly-
ing between these cases. Neither case will produce values of
åap higher than those in Figure 1, although the relationship
between size and åap may not be exactly as shown. I therefore argue that the optics of aggregates cannot explain the
high values of åap .
Martins et al. [1998] showed a strong spectral dependence
of absorption by BC inclusions with nonabsorbing coatings,
and suggested that this spectral dependence could indicate
the size and BC content of particles. Since smaller BC inclusions are associated with lower bulk absorption efficiencies,
the optics of coated spheres might explain the inverse correlation between αap and åap . However, I found that the
values of åap I obtained from the coated-sphere calculation
program with the parameters used by Martins et al. [1998]
were all below 1.65. Spectral dependence and absorption efficiency were strongly correlated for only a narrow range of
particle sizes (200–300 nm diameter). For larger particles,
the absorption efficiency is fairly constant, and for smaller
particles, åap is fairly constant. For size distributions similar to those of bituminous-coal emissions, I could reproduce
neither the range nor the magnitude ofåap using the coatedsphere program.
3.4. Molecular Structure
I hypothesize that the variation in light-absorbing properties results from the molecular form of the organic components. Jacobson [1999] suggested that aromatic hydrocarbons were responsible for UV absorption. I propose that
a broader theory, currently employed by amorphous-carbon
researchers, describes the absorption characteristics of some
atmospheric particles. While the measurements presented
here are limited to visible wavelengths, the theory may encompass UV absorption by carbonaceous compounds.
Tauc et al. [1966] related the absorption behavior of germanium to the optical gap, the energy required for an electronic transition between the highest ground state and the
lowest excited state. He derived a relationship between the
absorption coefficient α and the optical gap Eg :
√
αE = B(E − Eg )
(2)
where E is the energy of incident radiation and B is a constant. According to this relationship, the value of Eg affects both the absorption efficiency and the spectral dependence. The power-law dependence on wavelength only approximates the spectral dependence over a limited band.
It is generally accepted that light absorption by aromatic compounds shifts toward longer wavelengths as the
7
Actual αap (m2 g-1 , 525 nm)
1
6
5
4
3
2
1
0
0
1
2
3
Predicted
Figure 2. Measured absorption efficiency compared with predicted relative absorption (arbitrary units). Optical gap was estimated based on spectral dependence and used to predict absorption in Equation 2.
BOND: LIGHT ABSORPTION BY CARBONACEOUS PARTICLES
graphitic cluster size grows [Clar, 1964]. Robertson and
O’Reilly [1987] use optical-gap theory to explain that trend
for carbonaceous material. They describe amorphous carbon as islands of sp2 -bonded (graphitic) clusters that control the optical gap and hence the light-absorbing properties, surrounded by bands of sp3 -bonded (diamond-like) carbon. Chhowalla et al. [2000] show that absorption materials
with the same overall content of sp2 -bonded carbon can vary
when cluster sizes differ. The optical gap is inversely proportional to the diameter of a circular cluster or to the square
root of the number of aromatic rings in a cluster [Ferrari
and Robertson, 2000]. Optical-gap theory is also relevant to
flame-generated carbon [Minutolo et al., 1996]; in fact, the
linear relationship between hydrogen content and åap over
the range observed by Millikan [1961] could be explained if
H content varied with cluster diameter, and C content varied
with cluster area.
Compounds emitted from coal combustion may also contain a continuum of sp2 cluster sizes that govern the lightabsorbing properties. One end of the range is pure coal tar,
containing compounds with only a few adjacent aromatic
rings [Watt et al., 1996]. Thermal processing of the tar results in progressing degrees of graphitization, or an increase
in the cluster size [Marchand, 1997]. The other extreme is
complete graphitization, or infinite cluster sizes.
Following other researchers, I predict Eg for the coal
samples using the spectral dependence of absorption. The
absorption efficiency αap is a surrogate for the absorption
coefficient α, since both are proportional to k/λ for small
particles. I normalized both Equation (2) and the measured
absorption to a single wavelength, thereby removing the effect of the constant B, and then determined a value of Eg
using a least-squares fit to the data. This prediction of Eg
results from the spectral dependence, and not from the absolute magnitude of the absorption.
The largest optical gap in the coal-combustion samples is 1.0 eV. This value of Eg corresponds to a cluster
size of about 35 aromatic rings [Robertson, 1991]. Coalcomposition correlations [Genetti et al., 1999] indicate that
the average cluster size in the raw coal is about 8 rings.
For the highest-absorbing samples, the fit estimates Eg ≈ 0,
with a few (unphysical) negative values.
Next, I used the predicted values of Eg for each sample to predict absorption efficiency according to Equation 2.
Figure 2 shows the comparison with measured absorption
efficiency, with all values relative to the constant value of B.
With the exception of one outlier, 87% of the variability in
absorption can be explained by changes in the optical gap.
The optical-gap theory explains the concomitant variation of absorption efficiency and spectral dependence, as
well as the high values of åap . If it is applicable, the
light-absorbing properties of these samples primarily depend
upon the size of the graphitic clusters. Since most of the material participates in light absorption, the samples cannot
be readily divided into highly absorbing and non-absorbing
components. That division might be useful when purely
black particles (e.g., from diesel engines) are responsible for
the absorption. It is less likely to represent reality in areas
where light-absorbing particles come from low-temperature
coal combustion.
What fraction of coal-burning emissions could consist of
weakly absorbing, yellowish material? The answer depends
on burning conditions, especially the temperature. Tarry
4077
material is often ejected just after fuel addition, when coal
tar devolatilizes from the solid matrix and the combustor is
not hot enough to either oxidize or graphitize the particles. I
isolated this tarry component in one experiment. About 4.8
grams of “yellow” particles with åap = 2.5 escaped during
the first 6 minutes after fuel addition. Burning the remainder of the coal required 1.5 hours, during which 8.5 grams
of “black” particles with åap = 1.1 were emitted. If the
tarry and black samples were collected on the same filter,
as is usually done in this type of source testing, the average
value ofåap would be 1.3. The overall spectral dependence of
absorption is most influenced by the highly absorbing black
particles.
4. Implications
The component of carbonaceous particles that absorbs
visible light is typically called “black carbon.” “Lightabsorbing carbon” may be a better term for this material,
which does not always appear black. Some of the implications of the presence of weakly absorbing, yellowish material
in the atmosphere are listed below.
Measurement at Single Wavelengths. Absorption is often
measured at a single wavelength and extrapolated across the
solar spectrum by assuming a value of åap . The value of åap
obviously affects this extrapolation. If the single measurement is taken at wavelengths greater than 500 nm, total
absorption is underestimated if the value of åap is too low,
and the error is higher for longer wavelengths. For example,
the assumptions åap = 1.0 and åap = 1.5, combined with
absorption measured at 550 nm, result in a difference of
only 2% in absorption integrated across the solar spectrum.
The same calculations differ by 8% when they are based on
absorption at 632 nm.
Chemical Nature. If yellowish carbon contains a higher
fraction of sp3 -bonded carbon, its chemical transformation
in the atmosphere and hygroscopic-growth behavior are
likely to differ from those of “black” carbon. Since “yellow”
and “black” particles result from different combustion conditions, these particles are probably externally mixed when
emitted.
Thermal Measurements. The “thermal-optical” measurement of “elemental carbon” or EC [e.g., Chow et al., 1993] is
widely used to characterize atmospheric and source samples.
This technique corrects for charring of “organic” carbon by
monitoring sample transmittance. Reported EC is carbon
evolved after the transmittance returns to its initial value. It
is acknowledged that EC is an operational definition, equivalent to the mass of refractory carbon with the same absorption properties as the original sample. For correction
at the standard 632-nm wavelength, the reported EC would
absorb less light (integrated across the solar spectrum) than
an original sample with a strong wavelength dependence.
Cody [1984] finds that higher temperatures reduce semiconductor optical gaps, increasing absorption. In the thermaloptical method, the sample returns to the original transmittance at a temperature above ambient. Models of optical
properties [e.g., Lee and Tien, 1981] predict that absorption at 300◦ C is greater than that at 25◦ C. The difference is
only 1% at 632 nm, but 12% at 550 nm. These predictions
should be confirmed to ensure that optical correction does
not systematically bias EC measurements.
Expanded Measurements. Current characterization of
carbonaceous particles identifies individual compounds or
4078
BOND: LIGHT ABSORPTION BY CARBONACEOUS PARTICLES
functional groups, but not the compounds or structures that
control light absorption. Assessing characteristics relevant
to light absorption may require techniques that probe the
cluster size. These methods, including 13 C nuclear magnetic resonance (NMR) or Raman spectroscopy [Akhter et
al., 1985], can require large sample masses and have not yet
been quantitatively applied to atmospheric samples. It is
desirable to have direct measurements of light absorption
at wavelengths ranging from ultraviolet through the visible
spectrum, especially in regions where low-temperature combustion contributes to the burden of particulate matter.
Acknowledgments. The application of the optical-gap
theory was inspired by the work of Patrizia Minutolo, CNRNaples. The bituminous-coal measurements were made in conjunction with David Covert at the University of Washington.
I thank Warren Wiscombe for the coated-sphere program, and
Robert Charlson, Patricia Quinn, Timothy Bates, and Jost
Heintzenberg for useful comments. T. C. Bond was supported
during part of this research by a Fannie and John Hertz Foundation Fellowship, and is currently a NOAA Climate and Global
Change Postdoc with UCAR Visiting Scientist Programs. PMEL
Contribution No. 2373.
References
Akhter, M. S., A. R. Chughtai, and D. M. Smith, The structure of
hexane soot I: Spectroscopic studies, Appl. Spectrosc., 39(1),
143–153, 1985.
Anderson, T. L., D. S. Covert, J. D. Wheeler, J. M. Harris, K. D.
Perry, B. E. Trost, D. J. Jaffe, and J. A. Ogren, Single scattering albedo of marine and continental aerosols: observed values
and uncertainty at a coastal station in the Pacific Northwest,
J. Geophys. Res., 104(D21), 26,793–26,807, 1999.
Bond, T. C., M. Bussemer, B. Wehner, S. Keller, R. J. Charlson, and J. Heintzenberg, Light absorption by primary particle
emissions from a lignite burning plant, Environ. Sci. Technol.,
33, 3887–3891, 1999a.
Bond, T. C., T. L. Anderson, and D. Campbell, Calibration and
intercomparison of filter-based measurements of visible light
absorption by aerosols, Aerosol Sci. Technol., 30, 582–600,
1999b.
Bond, T. C., D. S. Covert, J. C. Kramlich, T. V. Larson, and
R. J. Charlson, Primary particle emissions from residential coal
burning: optical properties and size distributions, submitted
to J. Geophys. Res., February 2001.
Chhowalla, M., A. C. Ferrari, J. Robertson, and G. A. J. Amaratunga, Evolution of sp2 bonding with deposition temperature in tetrahedral amorphous carbon studied by Raman spectroscopy, Appl. Phys. Lett., 76(11), 1419–1421, 2000.
Chow, J. C., J. G. Watson, L. C. Pritchett, W. R. Pierson,
C. A. Frazier, and R. G. Purcell, The DRI thermal/optical
reflectance carbon analysis system: description, evaluation
and applications in U.S. air quality studies, Atmos. Environ.,
27A(8), 1185–1201, 1993.
Clar, E., Polyaromatic Hydrocarbons, chapter 7. London, Academic Press, 1964.
Cody, G. D., The optical absorption edge of a-Si:H, in Semiconductors and Semimetals, volume 21, Part B, edited by J. I.
Pankove, p. 11–82, Academic Press, Inc., 1984.
Ferrari, A. C., and J. Robertson, Interpretation of Raman spectra
of disordered and amorphous carbon, Phys. Rev. B, 61(20),
14,095–14,107, 2000.
Genetti, D., T. H. Fletcher, and R. J. Pugmire, Development and
application of a correlation of 13 C NMR chemical structure
analyses of coal based on elemental composition and volatile
matter content, Energy Fuels, 13, 60–68, 1999.
Haywood, J. M., and V. Ramaswamy, Global sensitivity studies
of the direct radiative forcing due to anthropogenic sulfate and
black carbon aerosols, J. Geophys. Res., 103(D6), 6043–6058,
1998.
Herman, J. R., P. K. Bhartia, O. Torres, C. Hsu, C. Seftor, and
E. Celarier, Global distribution of UV-absorbing aerosols from
Nimbus 7/TOMS data, J. Geophys. Res., 102(D14), 16,911–
16,922, 1997.
Horvath, H., Experimental calibration for aerosol light absorption
measurements using the integrating plate method-summary of
the data, Aerosol Sci., 28(7), 1149–1161, 1997.
Jacobson, M. Z., Isolating nitrated and aromatic aerosols and nitrated aromatic gases as sources of ultraviolet light absorption,
J. Geophys. Res., 104(D3), 3527–3542, 1999.
Jacobson, M. Z., Global direct radiative forcing due to multicomponent anthropogenic and natural aerosols, J. Geophys. Res.,
106(D2), 1551–1568, 2001.
Lee, S. C., and C. L. Tien, Optical constants of soot in hydrocarbon flames, Eighteenth Symposium (International) on Combustion, 1159–1166, 1981.
Malm, W. C., J. F. Sisler, D. Huffman, R. A. Eldred, and T. A.
Cahill, Spatial and seasonal trends in particle concentration
and optical extinction in the United States, J. Geophys. Res.,
99(D1), 1347–1370, 1994.
Marchand, A., Various kinds of solid carbon: structure and optical properties, in Polycyclic Aromatic Hydro-carbons and Astrophysics, edited by A. Léger, 31–54, D. Reidel Publishing
Co., 1997.
Martins, J. V., P. Artaxo, C. Liousse, J. S. Reid, P. V. Hobbs, and
Y. Kaufman, Effects of black carbon content, particle size, and
mixing on light absorption by aerosol from biomass burning in
Brazil, J. Geophys. Res., 103(D24), 32,041–32,050, 1998.
Millikan, R. C., Optical properties of soot, J. Opt. Soc. Am., 51,
698–699, 1961.
Minutolo, P., G. Gambi, and A. D’Alessio, The optical band gap
model in the interpretation of the UV-visible absorption spectra of rich premixed flames, Twenty-Sixth Symposium (International) on Combustion, The Combustion Institute, 1996.
Myhre, G., F. Stordal, K. Restad, and I. S. A. Isaksen, Estimation
of the direct radiative forcing due to sulfate and soot aerosols,
Tellus, 50B(5), 463–477, 1998.
Nelson, J., Fractality of sooty smoke: implications for the severity
of nuclear winter, Nature, 339, 611–613, 1989.
Patterson, E. M., and C. K. McMahon, Absorption characteristics
of forest fire particulate matter, Atmos. Env., 18, 2541–2551,
1984.
Penner, J. E., C. C. Chuang, and K. Grant, Climate forcing by
carbonaceous and sulfate aerosols, Clim. Dyn., 14, 839–851,
1998.
Robertson, J., Hard amorphous (diamond-like) carbons, Prog.
Solid St. Chem., 21, 199–333, 1991.
Robertson, J., and E. P. O’Reilly, Electronic and atomic structure
of amorphous carbon, Phys. Rev. B, 35(6), 2946–2957, 1987.
Rosen, H., A. D. A. Hansen, L. Gundel, and T. Novakov, Identification of the graphitic carbon component of source and ambient particulates by Raman spectroscopy and an optical attenuation technique, in Carbonaceous particles in the atmosphere,
Lawrence Berkeley Laboratory RP-2034, 229–232, 1979.
Streets, D. G., S. Gupta, S. T. Waldhoff, M. Q. Wang, T. C.
Bond, and Y. Bo, Black carbon emissions in China, Atmos.
Env., 35, 4281–4296, 2001.
Tauc, J., R. Grigorovici, and A. Vancu, Optical properties and
electronic structure of amorphous germanium, Phys. Stat. Sol.,
15, 627–637, 1966.
van de Hulst, H. C., Light Scattering by Small Particles, chapter
6, New York, Dover Publications, Inc., 1957.
Watt, M., T. H. Fletcher, S. Bai, M. S. Solum, and R. J. Pugmire, Chemical structure of coal tar during devolatilization,
Twenty-Sixth Symposium (International) on Combustion, The
Combustion Institute, 1996.
T. C. Bond, Pacific Marine Environmental Laboratory,
NOAA, 7600 Sand Point Way NE, Seattle, WA 98115-6349
([email protected])
(Received June 14, 2001; revised August 1, 2001;
accepted August 3, 2001.)