JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. D8, 4250, doi:10.1029/2002JD002486, 2003
Size-resolved particle emission indices in the stratospheric plume of an
Athena II rocket
O. Schmid,1,2 J. M. Reeves,1 J. C. Wilson,1 C. Wiedinmyer,1,3 C. A. Brock,4 D. W.
Toohey,5 L. M. Avallone,6 A. M. Gates,6 and M. N. Ross7
Received 26 April 2001; revised 27 June 2002; accepted 4 July 2002; published 25 April 2003.
[1] Simultaneous measurements of carbon dioxide (CO2) mixing ratio and alumina
(Al2O3) particle abundances between 0.004 and 1.2 mm were obtained in the stratospheric
plume wake of an Athena II solid-fueled rocket motor (SRM). A multimode model of the
particle size distribution is used to determine the number (N) and surface area (SA)
emission indices as 8.7 ± 2.0 1015 per kg and 6.9 ± 2.1 1014 mm2 per kg, respectively.
Integration of the size distribution shows that 37% of the total SA and 8% of the total mass
(M) are contained in the submicron size range (diameter < 1mm) that most influences
the stratospheric impact of alumina particles emitted by SRMs. These values are
significantly greater than reported by most previous studies and they imply that the annual
global ozone loss associated with SRM alumina emissions is about half of the ozone
loss from SRM chlorine emissions alone. Comparison of our results to previous studies
raises the possibility that submicron M and SA fraction, and therefore global ozone loss,
INDEX TERMS: 0305 Atmospheric Composition and
are inversely related to SRM thrust.
Structure: Aerosols and particles (0345, 4801); 0340 Atmospheric Composition and Structure: Middle
atmosphere—composition and chemistry; 1610 Global Change: Atmosphere (0315, 0325)
Citation: Schmid, O., J. M. Reeves, J. C. Wilson, C. Wiedinmyer, C. A. Brock, D. W. Toohey, L. M. Avallone, A. M. Gates, and
M. N. Ross, Size-resolved particle emission indices in the stratospheric plume of an Athena II rocket, J. Geophys. Res., 108(D8), 4250,
doi:10.1029/2002JD002486, 2003.
1. Introduction
[2 ] Combustion emissions from solid-fueled rocket
motors (SRMs) include particulate Al2O3 and a mix of
gases such as HCl, Cl2, H2O, NO and CO2 [Prather et al.,
1990]. Although plume wake ozone loss associated with the
reactive gas phase emissions has been observed [Ross et al.,
1997], numerical models suggest that its relative contribution to the global impact of rocket launches is small
[Danilin et al., 2001a]. Global mixing and gas phase
chemistry of HCl emissions from SRMs have been extensively modeled [e.g., Jones et al., 1995; Jackman et al.,
1998; Danilin et al., 2001b] and the various models are in
broad agreement that the change in annually averaged
global total ozone (AAGTO) due to present-day HCl
1
Department of Engineering, University of Denver, Denver, Colorado,
USA.
2
Currently at Max-Planck Institute for Chemistry, Mainz, Germany.
3
Currently at National Center for Atmospheric Research, Boulder,
Colorado, USA.
4
Aeronomy Laboratory, National Oceanic and Atmospheric Administration, Boulder, Colorado, USA.
5
Program in Atmospheric and Oceanic Sciences, University of Colorado, Boulder, Colorado, USA.
6
Laboratory for Atmospheric and Space Physics, University of
Colorado, Boulder, Colorado, USA.
7
Environmental Systems Directorate, The Aerospace Corporation, Los
Angeles, California, USA.
Copyright 2003 by the American Geophysical Union.
0148-0227/03/2002JD002486$09.00
AAC
emissions from SRMs does not exceed the relatively small
value of 0.03%.
[3] In contrast, the global impact of heterogeneous reactions on the surface of alumina particles emitted by SRMs is
much more uncertain, mainly due to limited quantitative
information describing the microphysics, reactivity, and size
distribution of SRM alumina. A variety of reactions could
present a significant ozone loss mechanism including halocarbon decomposition [Robinson et al., 1994] and various
chlorine activation reactions. Molina et al. [1997] measured
the reaction probability g of the chlorine activation reaction
ClONO2 + HCl ! HNO3 + Cl2 on alumina. The relatively
large g = 0.02 value they found emphasized the potential
importance of this reaction. Subsequent modeling studies
using the Molina et al. g value [Jackman et al., 1998;
Danilin et al., 2001b] confirmed that the chlorine activation
reaction could significantly increase the ozone loss associated with SRMs, potentially doubling AAGTO for some
emission scenarios.
[4] These studies also document the significance of submicron alumina particles for global impact studies.
Although SRM particles range from a few nanometers to
about 10 mm in diameter [Ross et al., 1999a; Cofer et al.,
1987], the diameter range between 0.01 and 1 mm is of most
significance for global impact studies, since particles
smaller than 0.01 mm and larger than 1 mm have relatively
short stratospheric lifetimes (weeks to months) due to
efficient removal by coagulation and gravitational settling,
respectively [Danilin et al., 2001b] (hereafter D2001b).
6-1
AAC
6-2
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
While the alumina mass emission index (EIM) (the total
alumina mass emitted per kg of SRM propellant) is well
known for the various SRMs on different rockets, knowledge of the mass and surface area fractions contained in the
various modes that make up the complete alumina distribution is limited. Published estimates span several orders of
magnitude and predictions from the various global models
reflect this uncertainty (D2001b). A reliable assessment of
SRM alumina ozone loss is not possible until more certain
knowledge of SRM particle size distributions is obtained.
[5] In this paper we use measurements of CO2 mixing
ratio and size resolved particle abundance between 0.004
and 1.2 mm obtained in the stratospheric plume wake of an
Athena II SRM to determine modal particle number (N),
surface area (SA) and mass (M) emission indices. Our data
represent the first SRM alumina size determination based on
instruments with a long history of successful stratospheric
measurements and constrained by a known tracer of SRM
combustion. We compare our results with previously
reported size distributions, estimate the relative importance
of gas phase and heterogeneous ozone loss from SRM
emissions, and discuss the implications that our results have
for assessments of the global ozone loss from SRMs.
Finally, we discuss the limitations of our data and conclusions and propose future measurements that could resolve
the remaining uncertainties.
2. Experimental
[6] The stratospheric plume wake of an Athena II rocket
launched from Vandenberg Air Force Base, California (34
480N, 120 370W) was sampled on September 24, 1999 at
66060 s universal time during the Atmospheric Chemistry
of Combustion Emissions Near the Tropopause (ACCENT)
mission [Ross et al., 1999b]. The NASA WB-57F high
altitude aircraft carried a comprehensive instrument suite to
measure a wide variety of gas phase and aerosol compounds
[Gates et al., 2002; Popp et al., 2002] and encountered the
Athena II plume five times at an altitude of 18.7 km
between 3.6 and 26.3 minutes after launch. Each encounter
lasted between 6 and 20 seconds. Here, we limit our
attention to measurements of CO2 abundances and particle
size distributions between 0.004 and 1.2 mm. A sixth plume
encounter at a significantly lower altitude of about 12 km
was not included in this study of stratospheric plumes due to
its close proximity to the tropopause.
[7] Particle measurements were performed at a 1 Hz
sampling frequency with two instruments developed at the
University of Denver, the nucleation mode aerosol size
spectrometer (NMASS) and the focused cavity aerosol
spectrometer (FCAS). Both instruments have demonstrated
successful operation on board stratospheric aircraft during a
variety of field campaigns [e.g., Fahey et al., 1995; Brock et
al., 2000]. The FCAS is an optical particle counter, which
measures particle concentration as a function of size over a
diameter range from 0.1 to 1.2 mm [Wilson et al., 1992;
Jonsson et al., 1995]. The NMASS consists of five condensation nucleus counters (CNCs) operating in parallel at
fixed internal pressure, temperatures and flow rates. The
supersaturation in each of the five CNCs is controlled at a
different value resulting in CNC-specific activation diameters of 4, 8, 15, 30 and 60 nm, respectively. Using a version
of the smoothed Twomey algorithm [Markowski, 1987] the
combined NMASS and FCAS data are inverted to yield
continuous number size distributions ranging from 0.004 to
1.2 mm [Brock et al., 2000].
[8] Since to good approximation alumina particles emitted by SRMs are spherical [Strand et al., 1981; Cofer et al.,
1987], the second and third moments of the measured
particle size distributions provide particle SA and volume
(V), respectively. Volume is converted into M assuming an
alumina density of rAl2O3 = 2 g per cm3 [Gates et al., 2002].
Numerical corrections for non-isokinetic sampling [Jonsson
et al., 1995], coincidence losses [Saros et al., 1996] and
detection efficiencies [Brock et al., 2000] were applied.
Random errors due to fluctuating operating conditions and
numerical uncertainties in the inversion algorithm were
estimated based on laboratory calibrations and Monte Carlo
simulations. The small sampling losses due to diffusion and
impaction were accounted for according to Willeke and
Baron [1993]. Measurement errors due to uncertainties in
particle density (for EIM) and refractive index (for EISA and
EIM) were estimated based on literature values [e.g., Strand
et al., 1981; Kim et al., 1993; Beiting, 1997]. We estimate
the overall uncertainties (1s; 68% confidence level) for N,
SA, and M as 15, 25 and 50%, respectively. However, since
the temperature of the sample air increases from about 210 K
(ambient) to 300 K in the detectors, an additional bias in SA
and M due to evaporation of volatile material prior to
entering the detection volume cannot be ruled out [Gates
et al., 2002].
[9] Carbon dioxide abundances were measured at a 1 Hz
sampling frequency by infrared absorption using a nondispersive infrared detector (Li-Cor, Model 6251, Lincoln,
Nebraska) adapted for high altitude measurements [Gates et
al., 2002]. In this paper, we are concerned only with the
plume enhancements of CO2 above the background of 365–
370 ppm. The 1s accuracy of these enhancements is limited
to about 0.25 ppm mainly due to fluctuations in the background abundances. For average CO2 enhancements of >1.5
ppm, as encountered here, the relative uncertainty is <17%.
The uncertainties in EIN, EISA and EIM are then obtained
from the quadrature sum of the CO2 uncertainty and the
uncertainties for N, SA, and M yielding 23, 30 and 53%,
respectively.
3. Results
[10] Figure 1 shows plume-averaged differential particle
distributions (dN/dlogD) acquired during the second
through fifth plume encounter at plume ages of 9.7, 15.4,
20.3, and 26.3 minutes, respectively, contrasted with clean
background air (90 second average) sampled at 18.7 km, the
altitude of the plume intercepts. It is noteworthy that while
plume distributions are dominated by alumina, the main
component of stratospheric background aerosol is sulfuric
acid. The error bars in the background distribution represent
the observed (1s) fluctuations about the mean values. The
first plume encounter 3.6 minutes after launch was discarded owing to undercounting in the NMASS due to high
particle concentrations (coincidence loss). Although particle
and CO2 concentrations varied considerably during each
encounter, Figure 1 indicates good reproducibility of plumeaveraged size spectra, which show a dominant peak at about
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
AAC
6-3
Figure 1. Differential size distributions for plume encounters 2 (circles), 3 (triangles), 4 (diamonds),
and 5 (squares) contrasted to the background distribution. Plume and background aerosol is thought to be
predominantly alumina and sulfuric acid, respectively. Error bars indicate the 1s fluctuation about the
mean background concentrations.
0.065 mm adjacent to a secondary peak at about 0.15 mm.
For encounters 2, 3, 4, and 5, integration of the size
distributions yields average total number densities of N =
13,200, 9450, 6000, and 9500 particles per cm3, respectively, similar to the values reported by other studies [Strand
et al., 1981; Ross et al., 1999a].
[11] The differential number distributions shown in Figure 1 were converted into differential EIs (DEIX) with
respect to N, SA and M following Schröder et al. [2000]
R
fX ð DÞ nð DÞdt
dEIX ð DÞ
R
DEIX ð DÞ ;
¼ EICO2
CO2 dt
d log D
ð1Þ
where EICO2 = 382 g/kg denotes the mass emission index of
CO2 for the Athena II (P. F. Zittel, personal communication),
fX(D) equals 1, pD2, and prAl2O3D3/6 for X = N, SA, and M,
respectively, and CO2 and n(D) are the enhancements of
CO2 (mass per volume air) and dN/dlogD (particles per
volume air) at diameter D over the background, respectively.
The integrals are performed over the entire plume encounter.
The measured mixing ratios of CO2 are converted into mass
densities by multiplication with (44/29)r, where r is the
ambient air density and 44/29 is the molar mass ratio of CO2
and air.
[12] Figures 2a, 2b, and 2c depict DEIX with respect to N,
SA, and M, respectively, averaged over all four plume
encounters. The origin of the lines in Figure 2 is explained
below. While all DEIX show peaks at about 0.065 and 0.15
mm, a dominant mode largely beyond the detection limit
(1.2 mm) appears for DEISA and DEIM. Thus the distribution
displays a trimodal structure with two submicron and (at
least) one supermicron mode. Because the only partially
characterized supermicron mode does not contribute substantially to the total number of particles (but dominates the
surface area and mass), EIN can be derived directly by
integrating DEIN. We find EIN = 9.9 1015, 1.1 1016,
5.5 1015, and 9.9 1015 per kg of propellant for
encounters 2, 3, 4 and 5, respectively, with a mean value
of 9.1 1015 per kg. Since the 1s scatter (26%) about the
mean value is about equal to the measurement uncertainty
(23%), EIN is constant within the accuracy of the measurement, i.e., the observed variability in N is consistent with
plume dilution.
[13] The situation for EISA and EIM is more difficult, since
the measured mean values of 3.1 1014 mm2 per kg and
36 g per kg, respectively, represent only a fraction of the
entire distribution (see Figure 2). In order to model the
entire distribution, we represent all DEIX distributions as
multimodal lognormal distributions of the form
" 2 #
ln DX ;k ln D
dEIX
lnð10Þ X EIX ;k
;
exp
¼ pffiffiffiffiffiffi
2
d log D
2p k ln sX ;k
2 ln sX ;k
ð2Þ
where the emission index EIX,k, the geometric mean
diameter DX,k, and the geometric standard deviation sX,k
are the fit parameters of each mode denoted by subscript k.
[14] In the submicron range, the curve fits (solid lines in
Figure 2) show two dominant modes at DN,2 = 0.065 mm
and DN,3 = 0.15 mm sandwiched between two less signifi-
AAC
6-4
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
Figure 2. Differential emission indices DEIX with respect to particle number (a), surface area (b), and
mass (c). The points represent the average of encounters 2 through 5, while the curves are lognormal fits
to the data using equation 2 and the parameters in Table 1.
cant modes at DN,1 = 0.037 mm and DN,4 = 0.35 mm. For the
following discussion, we assume a single mode in the
supermicron range (mode 5), which is justified by measurements of other investigators [e.g., Cofer et al., 1987; Ross et
al., 1999a]. While modes 2 and 3 are readily identified in all
three panels of Figure 2, mode 1 (0.037 mm) appears only as
small shoulder of mode 2 in Figure 2a; mode 4 is best
identified in Figure 2b bridging the gap between the modes
at 0.15 mm and >1 mm. Due to the relatively small contribution of modes 2 and 4 to EIX (<11%; see Table 1) these two
modes can be neglected for most practical purposes, i.e., the
size distribution can be considered as trimodal (modes 2, 3,
and 5). Since the fifth mode is not characterized well enough
to sufficiently constrain all three fit parameters of the mode,
we require the integral of DEIM to be equal to the known
alumina emission index EIM = 302 g per kg (= EIAl2O3),
reducing the independent fit parameters of mode 5 to DM,5
and sM,5. Now DM,5 (=1.91 mm) and sM,5 (=1.39) are
determined by least squares fit of DEIM and used to calculate
DN,5, DSA,5, sN,5 and sSA,5 from the known relationships
between the moments of lognormal distributions [Hinds,
1999]. The remaining fit parameters EIN,5 and EISA,5 are now
easily determined by fitting DEIN and DEISA, respectively.
The quality of the model fit is illustrated by the solid lines in
AAC
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
6-5
Table 1. Fit Parameters Equation (2) for the Differential Emission Indices DEIX and the Relative Contribution of Each Mode to the
Emission Indices EIN, EISA and EIM
DEIN
DEISA
DEIM
Fraction of total
Mode
EIN 1015/kg
Dg mm
sg
EISA 1014mm2/kg
Dg mm
sg
EIM g/kg
Dg mm
sg
EINa, %
EISA, %
EIM, %
1
2
3
4
5
0.85
6.04
1.27
0.08
0.06
0.037
0.065
0.148
0.35
1.38
1.22
1.31
1.27
1.64
1.39
0.057
0.971
0.928
0.46
4.52
0.042
0.076
0.162
0.588
1.71
1.27
1.28
1.27
1.62
1.39
–
2.6
5.1
10.0
284
–
0.080
0.165
0.700
1.91
–
1.28
1.27
1.56
1.39
10
69
15
1
1
1
14
13
7
65
0
1
2
3
94
a
The missing 4% in total EIN is due to particles with D < 0.02 mm, which are not represented by the modes listed here. This fraction of the size spectrum
has negligible impact on EISA and EIM. See body for more details.
Figure 2; a complete list of the fit parameters is compiled in
Table 1. It is important to note that for DEIN a purely
lognormal fit fails for D < 0.02 mm requiring an additional
term (7.4 1014 exp(44.1D[mm]), not listed in Table 1)
which represents 4% of EIN. Integration of the DEIX fits
yields EIN = 8.7 1015 per kg, EISA = 6.9 1014 mm2 per kg
and, by definition, EIM = 302 g per kg. The relative
contribution of each mode to EIX is presented in Table 1.
We find that the fractional contribution of submicron
particles to N, SA, and M (Nfrac, SAfrac, and Mfrac) is
99%, 37%, and 8%, respectively. Comparing these values
to the sum of the fractional contributions of modes 1
through 4 (see Table 1) shows that mode 5 which we refer
to as supermicron mode contributes 0% to Nfrac and 2% to
both SAfrac and Mfrac.
[15] In addition to the experimental errors listed above,
there is uncertainty in the fit parameters, mainly for the
largest mode, due to the detection limit of 1.2 mm. In order
to estimate the magnitude of this uncertainty, we assume as
a worst case scenario DM,5 = 4.35 mm (instead of 1.91 mm),
the largest value of the mass weighted mean diameter for
supermicron SRM aerosol found in the literature [Cofer et
al., 1991]. This reduces EISA to a lower limit of 4.3 1014
mm2 per kg and enhances SAfrac to an upper limit of 60%.
On the other hand, if non-volatile components other than
alumina (e.g., metals, soot) contribute to the measured
aerosol mass, normalization of the mass spectrum to EIAl2O3
underestimates the supermicron M (and SA) and therefore
overpredicts SAfrac. Although there is evidence that both
volatile and non-volatile components other than Al2O3 may
contribute to SRM aerosol, their relative contribution is
unknown but probably small [Cofer et al., 1987; Gates et
al., 2002]. For the following discussion we adopt 37% as
our best estimate of SAfrac.
4. Discussion
[16] We wish to compare our alumina size distributions
with the results of previous SRM studies. The most frequently quoted SRM particle size distributions are provided
by Ross et al. [1999a] (hereafter Ross), Beiting [1997] and
Brady and Martin [1995] (hereafter BM) who reported
SAfrac values of 3%, 33% and 83%, respectively. Although
the value determined here, 37 ± 11%, initially appears to
support Beiting, we note the following caveats. Beiting and
BM derived size distributions by combining measurements
of different limited portions of the size distribution,
obtained in the plumes of different rockets, encountered
under different atmospheric conditions [Strand et al., 1981;
Zolensky et al., 1989; Cofer et al., 1991]. Hence skepticism
must be acknowledged regarding the degree to which the
Beiting and BM distributions are representative of any
particular SRM under stratospheric conditions. In contrast,
the distribution of Ross represents an average of measurements obtained in the plumes of rockets with similar SRMs
(Space Shuttle and Titan IVA) and under similar stratospheric conditions and so could reasonably be expected to
represent the distribution of high thrust SRMs. The large
difference between Ross and our results could be explained
by discrepancies in instrument performance (e.g., Ross did
not account for sub-isokinetic particle enhancement) or by
supposing actual differences in the size distributions
between large and small SRMs. Support for the latter view
is provided by Hermsen [1981] who showed that the volume
weighted mean diameter D43 (defined as ratio of the fourth
and third moment of the size distribution) of alumina emissions increases with SRM thrust across a wide range of
SRMs. Consistent with this relationship, D43 reported by
Ross for the large SRMs (3.2 mm) exceeds D43 reported here
for the smaller Athena II (2.6 mm) so that SAfrac reported by
Ross for the large SRMs (3%) is significantly less than SAfrac
reported here for the smaller Athena II (37%). While we
cannot rule out instrument sampling effects for Ross, taken
together these data suggest that large SRMs have smaller
SAfrac (and Mfrac) than small SRMs.
[17] Insofar as the models of global ozone depletion
[Jackman et al., 1998; D2001b] agree that the predicted
ozone loss from the chlorine activation reaction is sensitive
to the assumed SAfrac, it is instructive to consider how our
results might influence the model predictions. Since SRM
emissions produce only small perturbations in background
aerosol surface area density (SAD) [Jackman et al., 1998],
we can estimate global ozone loss for a given size distribution by linearly scaling existing model predictions. The
validity of this assumption is confirmed when comparing
case studies B and D of D2001b, representing the modeled
global ozone loss for two different size distributions under
otherwise identical conditions.
[18] Ozone loss from alumina emissions will increase
with EISA, although not necessarily in a simple linear way,
since the steady state global surface area density SADss for a
given EISA will depend on the interaction between size
dependent stratospheric lifetimes of particles and the details
of the particle size distribution. D2001b calculated SADss as
zonally averaged annual alumina SAD over the entire
stratosphere using a 13 bin alumina size distribution. They
present several calculations with different assumptions
regarding removal mechanisms for alumina particles. For
AAC
6-6
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
comparison purposes, we focus on their most realistic case
(Case C), which combines tropospheric washout, coagulation with background sulfate aerosol and gravitational
sedimentation as removal mechanisms. Since for a given
size bin ozone loss is proportional to SADss and the SADss
is proportional to the annual alumina mass emission rate
(MER) in this bin, we can determine the global ozone loss
for any size distribution by scaling SADss according to
MER. Hence, the estimated change in AAGTO is given by
P
AAGTO ¼ AAGTODan
SADDan; j MERj =MERDan;j
j
SADDan
ð3Þ
where SADDan and AAGTODan are the total SADss and
AAGTO calculated by D2001b, respectively, based on
their assumed size distribution (see Table 2 in D2001b), and
the summation is taken over all particle size bins j. The
summation represents the total SAD for a given size
distribution where MERj and MERDan,j are the annual
alumina mass emission rates in each size bin for the size
distribution of interest and D2001b’s distribution, respectively. Table 2 in D2001b gives SADDan,j and MERDan,j. In
order to focus on the relative influence a particular size
distribution has on the predicted AAGTO, we maintain the
total MER (=MERj) assumed by D2001b of 1120 tons per
year (chosen to represent nine Space Shuttle and four Titan
IV launches but arbitrary in that the ozone loss linearly
scales with any given total MER). Hence, MERj are
calculated by integrating the size distribution of interest
over the range of each bin and normalizing the total MER to
1120 tons per year.
[19] Table 2 presents the total SADss and AAGTO for
the various size distributions discussed in this paper using
equation (3). Depending on the assumed distribution,
AAGTO ranges from 0.0028% (Beiting) to 0.0128%
(BM). As an aside, we note that while D2001b claim that
their assumed distribution is that of Beiting, in fact it is not,
since they replaced particle diameter by radius in Beiting’s
expressions without modifying the corresponding modal
coefficients. While this point is irrelevant for the scaling
procedure described above it explains the somewhat lower
AAGTO of 0.0021% reported by D2001b compared to
the Beiting value of 0.0028%. The Ross size distribution
for the large SRMs yields a AAGTO of 0.0031%, close
to Beiting’s value. Our measured size distribution for the
Athena II produces AAGTO = 0.0104%, closer to BM.
The AAGTO reported by D2001b for the SRM chlorine
emissions alone corresponding to the assumed total MER of
1120 tons of alumina equals 0.0123%. For the following
discussion, we utilize D2001b’s global ozone model to
compare gas phase and heterogeneous ozone loss due to
SRM emissions. While we are aware that other modeling
studies found different values for both gas phase and
heterogeneous ozone loss [e.g., Jackman et al., 1998],
D2001b is the only study providing enough detailed sizeresolved information on heterogeneous ozone loss to allow
scaling their modeling results to different alumina size
distributions. Hence utilizing D2001b’s model does not
imply an endorsement of D2001b over other studies; this
choice is purely based on practical considerations.
Table 2. Total Steady State SAD and Change in Annually
Averaged Global Total Ozone (AAGTO) Due to Alumina
Emissions for Various SRM Particle Distributionsa
Size Distribution
SADss, 106 mm2/cm3
AAGTO, %
Danilin et al. [2001b]
Ross et al. [1999a] (large SRM)
Beiting [1997]
Brady and Martin [1995]
Present study (small SRM)
Realistic scenariob
41.5
61.3
55.4
262
206
97.5
0.0021
0.0031
0.0028
0.0128
0.0104
0.0049
a
For comparison, AAGTO due to chlorine emission alone is about
0.0123%.
b
This scenario assumes 25% (75%) of the alumina emissions is due to
small (large) SRMs with a characteristic particle size distribution similar to
the one reported by the present study ([Ross et al., 1999a]). See body for
more details.
[20] If our Athena II size distribution is characteristic of
small SRMs and the Ross distribution is characteristic of
large SRMs, the ozone depletion per kg of propellant due to
alumina emissions alone would be about three times greater
for small SRMs than for large SRMs (see Table 2). Considering that the AAGTO due to chlorine emissions alone is
0.0123%, the AAGTO from SRM chlorine and alumina
emissions combined is 0.0227% and 0.0154% for a small
and large SRMs, respectively. Hence the total ozone loss per
kg of propellant is about 50% larger for small SRMs than for
large ones. In light of the possibility that the alumina size
distribution is related to SRM thrust, care must be taken
interpreting Table 2. Impact assessment efforts should weight
the relative contributions of small (e.g., Athena II) and large
SRMs (e.g., Space Shuttle) to the total emission in order to
improve the accuracy of the assessment. Furthermore, the
details of the vertical profile of the alumina mass deposition
into the atmosphere affects SADDan,j and hence AAGTO.
Therefore, Table 2 may serve as illustration of the impact of
commonly used alumina size distributions on global ozone,
but not as representation of actual atmospheric ozone losses.
[21] As a guideline to more realistic AAGTO values we
compare D2001b’s launch scenario of 9 Space Shuttle and 4
Titan IV launches per year to the currently more typical
launch rates of 6 Space Shuttle, 2 Titan IV and about 25
smaller rocket launches per year (Delta II, Atlas IIAS, Taurus,
and various missiles). The annual stratospheric alumina
emissions from this more realistic scenario is approximately
990 tons, 12% less than assumed by D2001b. The year-toyear variability in emissions associated with changing launch
schedules is about 20%. Launches from nations other than the
United States account for approximately an additional 200
tons. Thus, D2001b’s MER of 1120 tons of alumina per year
concurs with current launch rates. For the current launch
scenario, approximately 25% of the alumina emissions are
from the smaller rockets and so could have submicron mass
fractions and, consequently, ozone depletion characteristics
similar to the Athena II reported here. As previously mentioned, Athena II type (i.e., small SRM) alumina emissions
produce about three times as much ozone loss per kg fuel as
Space Shuttle type (i.e., large SRM) emissions (see Table 2).
Thus while the smaller rockets typically contribute only
about 25% of the total alumina emissions, they account for
about 50% of the ozone loss from all alumina emissions
assuming all small SRMs have particle size distributions
similar to Athena II. Accounting for the smaller rockets also
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
increases the alumina related ozone loss from all rockets by
about 60% (from 0.0031% to 0.0049%). Including the
ozone loss due to gas phase chlorine reactions (0.0123%)
total ozone loss increases by about 10% from 0.0154%
(100% large SRM) to 0.0172% (75% large SRM, 25%
small SRM), 30% of which can be attributed to alumina
emissions. Clearly, if our suggestion is correct that the
submicron mass fraction of SRM alumina is strongly related
to SRM thrust, then smaller rockets play a significant role in
determining the complete impact of rocket emissions and,
therefore, should be accounted for in global models.
[22] Finally, we point out that ablation during reentry of
spent satellites may constitute another significant source of
alumina in the stratosphere [Zolensky et al., 1989]. However, this source is purely quantified and its atmospheric
implications are unknown.
5. Conclusions
[23] We performed in situ CO2 and particle measurements
in the stratospheric plume wake of an Athena II rocket
propelled by a single small SRM. Based on lognormal curve
fits and the assumption that EIM = 302 g per kg (= EIAl2O3) we
determined DEIN and DEISA up to 10 mm yielding total
emission indices of EIN = 8.7 ± 2.0 1015 per kg and
EISA = 6.9 ± 2.1 1014 mm2 per kg. The data suggest that
small SRMs (similar to Athena II) favor the production of
submicron particles so that the alumina-induced ozone loss
per kg of propellant for small SRMs could be about three
times the loss for large SRMs. Application of our size
distribution to a recent model of the stratospheric impact of
SRM emissions shows that at least for small SRMs (Athena II
type) ozone loss from chlorine activation reactions on alumina emissions is comparable to the ozone loss from gas
phase chlorine emissions alone. For a realistic launch scenario, alumina emissions contribute about 30% to the estimated annually averaged global total ozone loss of
0.0172%. Given the significant differences among the
reported SRM particle distributions and the possibility that
submicron particle production (and therefore ozone loss) is
related to SRM thrust, further in situ sampling of stratospheric SRM plumes for different rockets using the same
instruments under similar conditions is suggested to confirm
our results and better understand the role of alumina emissions in the stratosphere.
Notations
Symbols
AAGTO
AAGTO
BM
CNC
D
D43
annually averaged global total ozone
change in AAGTO
Brady and Martin [1995]
condensation nucleus counter
particle diameter
volume weighted mean diameter of
particle size distribution
D2001b Danilin et al. [2001b]
DX geometric mean diameter of DEIX
DEIN, DEISA, DEIM differential emission index with
respect to particle number, surface
area, and mass (dEIX/dlogD)
AAC
6-7
DEIX differential emission index with respect
to particle number, surface area or mass
EIAl2O3 alumina mass emission index
EICO2 CO2 emission index
EIN, EISA, EIM particle number, surface area, and mass
emission index
EIX emission index with respect to particle
number, surface area or mass
FCAS focused cavity aerosol spectrometer
M particle mass per volume air
Mfrac fractional contribution of submicron
particles to total M
MER total annual alumina mass emission rate
into the stratosphere due to rocket
launches
n differential particle concentration
(dN/dogD)
N number of particles per volume air
Nfrac fractional contribution of submicron
particles to total N
NMASS nucleation mode aerosol size spectrometer
SA particle surface area per volume air
SAD annually averaged global surface area
density of background aerosol in the
stratosphere
SADss steady state SAD
SRM solid-fueled rocket motor
V particle volume density
Greek Letters
CO2 CO2 concentration above background
n differential particle concentration above
background
g reaction probability
r ambient air density
rAl2O3 density of alumina particles
s 68% confidence level
sX,k geometric standard deviation of mode k
in DEIX
Subscripts
Dan value taken from D2001b
j indicates size bin in particle distribution
k indicates mode of particle size distribution
(here k = 1,2,...,5)
X represents N, SA, or M
[24] Acknowledgments. We appreciate the efforts of the WB-57F air
and engineering crews and range support personnel at Vandenberg Air
Force Base. This work was supported by NASA, the Air Force Launch
Programs Office, and Thiokol Propulsion.
References
Beiting, E. J., Solid rocket motor exhaust model for alumina particles in the
stratosphere, J. Spacecr. Rockets, 34, 303 – 310, 1997.
Brady, B. B., and L. R. Martin, Modeling solid rocket motor booster exhaust plumes in the stratosphere with SURFACE CHEMKIN, Tech. Rep.
TR-95 (5231)-9, Aerospace Corp., El Segundo, Calif., 1995.
Brock, C., F. Schröder, B. Kärcher, A. Petzold, R. Busen, and M. Fiebig,
Ultrafine particle size distributions measured in aircraft exhaust plumes,
J. Geophys. Res., 105, 26,555 – 26,567, 2000.
Cofer, W. R., G. G. Lala, and J. P. Wightman, Analysis of mid-tropospheric
space shuttle exhausted aluminum particles, Atmos. Environ., 21, 1187 –
1196, 1987.
AAC
6-8
SCHMID ET AL.: PARTICLE EMISSION INDICES OF A SRM
Cofer, W. R., G. C. Purgold, E. L. Winstead, and R. A. Edahl, Space shuttle
exhausted aluminum oxide: A measured particle size distribution,
J. Geophys. Res., 96, 17,371 – 17,376, 1991.
Danilin, M. Y., M. K. W. Ko, and D. B. Weisenstein, Global implications of
ozone loss in a Space Shuttle wake, J. Geophys. Res., 106, 3591 – 3601,
2001a.
Danilin, M. Y., R.-L. Shia, M. K. W. Ko, D. K. Weisenstein, N. D. Sze, J. J.
Lamb, T. W. Smith, P. D. Lohn, and M. J. Prather, Global stratospheric
effects of the alumina emissions by solid-fueled rocket motors, J. Geophys. Res., 106, 12,727 – 12,738, 2001b.
Fahey, D. W., et al., Emission measurements of the Concorde supersonic
aircraft in the lower stratosphere, Science, XX, 70 – 74, 1995.
Gates, A. M., et al., In situ measurements of carbon dioxide, 0.37 – 4.0 mm
particles, and water vapor in the stratospheric plumes of small rockets,
J. Geophys Res., 107(D22), doi:10.1029/2002JD002121, 4649, 2002.
Hermsen, R. W., Aluminum oxide particle size for solid rocket motor
performance prediction, J. Spacecraft, 18, 483 – 490, 1981.
Hinds, C. H., Aerosol Technology, Wiley-Interscience, New York, 1999.
Jackman, C. H., D. B. Considine, and E. L. Fleming, A global modeling
study of solid rocket aluminum oxide emission effects on stratospheric
ozone, Geophys. Res. Lett., 25, 907 – 910, 1998.
Jones, A. E., S. Bekki, and J. A. Pyle, On the atmospheric impact of launching the Ariane 5 rocket, J. Geophys. Res., 100, 16,651 – 16,660, 1995.
Jonsson, H. H., et al., Performance of a focused cavity aerosol spectrometer
for measurements in the stratosphere of particle size in the 0.06 – 2.0 mm
diameter range, J. Oceanic Atmos. Tech., 12, 115 – 129, 1995.
Kim, H.-O., D. Laredo, and D. W. Netzer, Measurement of submicrometer
Al2O3 particle in plumes, Appl. Opt., 32, 6834 – 6841, 1993.
Markowski, G. R., Improving Twomey’s algorithm for inversion of aerosol
measurement data, Aerosol Sci. Technol., 7, 127 – 141, 1987.
Molina, M. J., L. T. Molina, R. Zhang, R. F. Meads, and D. D. Spencer, The
reaction of ClONO2 with HCl on aluminum oxide, Geophys. Res. Lett.,
24, 1619 – 1622, 1997.
Popp, P., et al., The emission and chemistry of reactive nitrogen species in
the plume of an Athena II solid-fueled rocket motor, Geophys. Res. Lett.,
29(18), doi:10.1029/2002GL015197, 1887, 2002.
Prather, M. J., M. M. Garcia, A. R. Douglass, C. H. Jackman, M. K. W. Ko,
and N. D. Sze, The space shuttle’s impact on the stratosphere, J. Geophys. Res., 95, 18,583 – 18,590, 1990.
Robinson, G. N., A. Freedman, C. E. Kolb, and D. R. Worsnop, Decomposition of halomethanes on a-alumina at stratospheric temperatures,
Geophys. Res. Lett., 21, 377 – 380, 1994.
Ross, M. N., et al., Observation of Stratospheric Ozone Depletion in Rocket
Plumes, Nature, 390, 62 – 65, 1997.
Ross, M. N., P. D. Whitefield, D. E. Hagen, and A. R. Hopkins, In situ
measurement of the aerosol size distribution in stratospheric solid rocket
motor exhaust plumes, Geophys. Res. Lett., 26, 819 – 822, 1999a.
Ross, M. N., et al., Study blazing new trails into the effects of aviation and
rocket exhaust in the atmosphere, Eos Trans. AGU, 80, 1999b.
Saros, M. T., P. H. McMurry, J. J. Marti, and R. J. Weber, Ultrafine aerosol
measurement using a condensation nucleus counter with pulse height
analysis, Aerosol Sci. Technol., 25, 200 – 213, 1996.
Schröder, F., C. A. Brock, R. Baumann, A. Petzold, R. Busen, P. Schulte,
and M. Fiebig, In situ studies on volatile jet exhaust particle emissions:
Impact of fuel sulfur content and environmental conditions on nuclei
mode aerosols, J. Geophys. Res., 105, 19,941 – 19,954, 2000.
Strand, L. D., J. M. Bowyer, G. Varsi, E. G. Laue, and R. Gauldin, Characterization of particulates in the exhaust plume of large solid-propellant
rockets, J. Spacecraft, 18, 297 – 305, 1981.
Willeke, K., and P. A. Baron, Aerosol Measurement, Van Nostrand Reinhold, New York, 1993.
Wilson, J. C., M. R. Stolzenburg, W. E. Clark, M. Loewenstein, G. V. Ferry,
K. R. Chan, and K. K. Kelly, Stratospheric sulfate aerosol in and near the
northern hemisphere polar vortex: The morphology of the sulfate layer,
multimodal size distributions, and the effect of denitrification, J. Geophys. Res., 97, 7997 – 8013, 1992.
Zolensky, M. E., D. S. McKay, and L. A. Kaczor, A tenfold increase in the
abundance of large solid particle in the stratosphere as measured over the
period 1976 – 1984, J. Geophys. Res., 94, 1047 – 1056, 1989.
L. M. Avallone and A. M. Gates, Laboratory for Atmospheric and Space
Physics, University of Colorado, 1234 Innovation Drive, Boulder, CO
80303, USA.
C. A. Brock, Aeronomy Laboratory, National Oceanic and Atmospheric
Administration, 325 Broadway, Boulder, CO 80305, USA.
J. M. Reeves and J. C. Wilson, Department of Engineering, University of
Denver, 2390 South York Street, Denver, CO 80208, USA.
M. N. Ross, Environmental Systems Directorate, The Aerospace
Corporation, Los Angeles, CA, USA.
D. W. Toohey, Program in Atmospheric and Oceanic Sciences, University
of Colorado, UCB 311, Boulder, CO 80309, USA.
O. Schmid, Biogeochemistry Department, Max-Planck Institute for
Chemistry, P.O. Box 3060, D-55020 Mainz, Germany. (oschmid@mpchmainz.mpg.de)
C. Wiedinmyer, National Center for Atmospheric Research, P.O. Box
3000, Boulder, CO 80307, USA.