Aircraft measurements of vertical profiles of aerosol mixing states

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, D11305, doi:10.1029/2009JD013150, 2010

for

Full Article

Aircraft measurements of vertical profiles of aerosol mixing states Kerri A. Pratt1,2 and Kimberly A. Prather1,3 Received 3 September 2009; revised 22 November 2009; accepted 4 January 2010; published 8 June 2010.

[1] To examine the overall impact of aerosols on climate, direct measurements of the size‐resolved mixing states of atmospheric particles are needed as a function of altitude. During the Ice in Clouds Experiment‐Layer Clouds, the recently developed aircraft aerosol time‐of‐flight mass spectrometer directly measured the vertical profiles of size‐resolved single‐particle chemistry in cloud‐free air over Wyoming and northern Colorado. These represent the first aircraft‐based, dual polarity mass spectrometry measurements, allowing a detailed examination of in situ single‐particle mixing state as a function of altitude. Measurement of both positive and negative ions for each individual particle provides the ability to identify the primary particle type/source, such as biomass burning, organic carbon, or soot, and examine the extent of mixing with secondary species, such as ammonium, nitrate, sulfate, and sulfuric acid. For the primary particle cores, biomass burning represented the largest source of submicron particles: ∼33–39% by number from 1 to 7 km. Organic carbon particles were the second most abundant type (maximum of ∼33% by number from 1.2 to 2.0 km) with elemental carbon (soot) particles comprising 14–22% by number from 1 to 7 km. In general, biomass burning, organic carbon, and soot particles were frequently internally mixed with ammonium, nitrate, and sulfate at lower altitudes, switching to sulfate and sulfuric acid mixtures at higher altitudes. Further, the number fraction of externally mixed sulfuric acid particles increased with altitude from 1 to 9%, likely because of cloud processing of SO2. The variance of particle mixing state with altitude significantly changes absorption and hygroscopic properties, and must be taken into account in models calculating aerosol direct and indirect radiative forcings. Citation: Pratt, K. A., and K. A. Prather (2010), Aircraft measurements of vertical profiles of aerosol mixing states, J. Geophys. Res., 115, D11305, doi:10.1029/2009JD013150.

1. Introduction [2] Atmospheric aerosols strongly influence the energy balance of the Earth and the hydrological cycle by scattering and absorbing solar and terrestrial radiation and acting as cloud condensation and ice nuclei [e.g., Poschl, 2005]. The physical, optical, and hygroscopic properties of particles strongly depend on whether different chemical species are mixed within a single particle (internal mixing) or in different particles (external mixing) [Fuzzi et al., 2006]. During transport in the atmosphere, aerosol particles undergo physical and chemical transformations through cloud processing, gas‐particle partitioning, and reactions with atmospheric gases [Poschl, 2005]. Because of the evolving physical and chemical properties of the distribution of aerosols, quantifying aerosol impacts on radiative forcing and clouds represents a challenging task [Solomon et al., 2007]. For example, as 1 Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, California, USA. 2 Now at Department of Chemistry, Purdue University, West Lafayette, Indiana, USA. 3 Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California, USA.

Copyright 2010 by the American Geophysical Union. 0148‐0227/10/2009JD013150

externally mixed soot particles age, they convert to internally mixed particles with a soot core surrounded by sulfate, ammonium, organics, nitrate, and water [Moffet and Prather, 2009]. This results in a positive radiative forcing second only to CO2 [Ramanathan and Carmichael, 2008], as well as increased particle hygroscopicity [Zuberi et al., 2005], showing the importance of particle aging and understanding particle mixing state. The majority of in situ aerosol measurements have been made at ground level; however, the direct and indirect radiative forcings depend on aerosol vertical and horizontal profiles [Solomon et al., 2007]. Thus, in situ, size‐dependent aerosol chemical composition and mixing state (internal/external) measurements are needed as a function of altitude for the further development and evaluation of global climate models [Fuzzi et al., 2006]. [3] Aircraft‐based measurements of individual particles have traditionally been made using off‐line microscopy techniques. For example, using electron microscopy, Pósfai et al. [1999] found an increasing fraction of sulfate particles containing soot with increasing altitude, likely due to aircraft emissions; thus, these more hygroscopic soot particles are more effective cloud condensation nuclei (CCN) at altitudes where cloud formation is more probable. Real‐time aerosol analysis methods reduce sampling artifacts and allow direct measurements of changing aerosol populations with high

D11305

1 of 10

D11305

PRATT AND PRATHER: VERTICAL AEROSOL MIXING STATES

temporal resolution [Sullivan and Prather, 2005]. Extensive airborne, real‐time, single‐particle chemistry measurements have been made using particle analysis by laser mass spectrometry (PALMS) [Murphy et al., 1998], showing over 90% of accumulation mode particles away from local sources to generally be internal mixtures of sulfates and carbonaceous material [Murphy et al., 2006]. In contrast to PALMS, the aerosol time‐of‐flight mass spectrometer (ATOFMS) detects both positive and negative ions simultaneously from individual particles [Gard et al., 1997], providing an increased understanding of particle mixing state [Murphy, 2007]. We report results from the first set of flights using the recently developed aircraft (A)‐ATOFMS [Pratt et al., 2009c]. Size‐ resolved single‐particle mixing state measurements of both refractory and nonrefractory species, acquired in real time as a function of altitude, provide insight into the range of mixing states of atmospheric aerosols. Measurement of both positive and negative ions within the dual polarity mass spectrometer for each individual particle provides the ability to identify the primary particle type/source, such as biomass burning, organic carbon, or soot, and examine its mixing with secondary species, such as ammonium, nitrate, sulfate, and sulfuric acid, as discussed herein. This added information is important since the addition of hygroscopic species during aging significantly changes absorption and hygroscopic properties of particles, impacting estimations of aerosol direct and indirect radiative forcing [McFiggans et al., 2006].

2. Experimental Section [4] Measurements of individual ambient aerosol particles by A‐ATOFMS were conducted during the Ice in Clouds Experiment‐Layer Clouds (ICE‐L) from November to December 2007 aboard the National Center for Atmospheric Research/National Science Foundation (NCAR/NSF) C‐130 aircraft based in Broomfield, Colorado. Data are presented from 14 flights (1, 5, 7, 13, 16, 18, 20, 29, and 30 November; 10, 11, 12, 13, and 16 December) from ∼1.2–7.0 km above mean sea level (AMSL) over northern Colorado, western Nebraska, southwestern South Dakota, and Wyoming (approximately 39.8–44.5°N, 102.8–109.7°W). Latitude, longitude, and altitude were measured at 10 Hz by a Garmin GPS and averaged to 1 Hz. Ambient condensation nuclei (CN) concentrations were measured using a condensation particle counter (CPC, TSI Model 3760), detecting particles greater than ∼13 nm in size. [5] The A‐ATOFMS, described in detail elsewhere [Pratt et al., 2009c], measures the vacuum aerodynamic diameter (dva) and dual polarity mass spectra of individual particles with sizes ranging from ∼70–1200 nm in real time. Briefly, following a 210Po neutralizer and pressure‐controlled inlet, particles are focused in an aerodynamic lens system. Particles are optically detected by two continuous wave 532 nm lasers spaced 6.0 cm apart, providing particle velocity and, thus, dva; polystyrene latex spheres of known physical diameter from 95 to 1500 nm are used for single‐particle size calibration. Particles are desorbed and ionized using 266 nm radiation from a Q‐switched Nd:YAG laser operating at ∼0.4–0.6 mJ during ICE‐L. Positive and negative ions resulting from individual particles are detected within the time‐of‐flight mass spectrometer. Decreased fragmentation of carbonaceous ions is achieved within the ATOFMS compared to the

D11305

PALMS instrument, allowing for differentiation between elemental and organic carbon within the A‐ATOFMS [Murphy et al., 2007; Spencer and Prather, 2006]. [6] While the field campaign focused on cloud studies, A‐ ATOFMS size‐resolved dual polarity mass spectra were collected in clear (cloud‐free) air for 15,071 individual particles; this does not include data from within two smoke plumes. Single‐particle mass spectra were imported into YAADA (www.yaada.org), a software toolkit for MATLAB (The MathWorks, Inc.). Single‐particle mass spectra were tagged with the altitude, latitude, longitude, and CN concentration present at the time that each particle was sampled. The fraction of particles producing a mass spectrum did not vary significantly (standard deviation of 0.3%) with altitude, suggesting that the A‐ATOFMS did not miss more pure particles, as observed in certain studies [Spencer et al., 2008; Wenzel et al., 2003]. An adaptive resonance theory‐based clustering method (ART‐2a) [Bhave et al., 2001; Rebotier and Prather, 2007; Song et al., 1999] was used to classify single‐particle mass spectra with a vigilance factor of 0.80, learning rate of 0.05, and 20 iterations. ART‐2a classifies particles into separate clusters based on the presence and intensity of ion peaks in individual single‐particle mass spectra. Clear air particles from all flights were analyzed together, resulting in 233 clusters with 88% of the chemically analyzed particles classified and described herein. On the basis of characteristic chemical species or a possible source, resulting ART‐2a clusters were manually classified into 13 general particle classes: biomass burning, organic carbon (OC), aromatic, amine, elemental carbon/organic carbon (ECOC), elemental carbon (EC), biological, salt (Na‐K‐Cl), mineral dust, metal, sulfate‐nitrate (particles with no positive ions), and sulfuric acid. The characteristic ion markers defining each particle class are noted in Table 1 with more detailed descriptions provided below. Chemical species have different affinities for the formation of positive and negative ions [Murphy, 2007], resulting in positive ion mass spectra that can be linked to the primary source signature and negative ions, which, in general, provide information regarding internal mixing with secondary species [Pratt and Prather, 2009]. Thus, these particle type classification labels do not necessarily reflect all of the species present within a particular particle type. Peak identifications correspond to the most probable ions for a given m/z ratio based on previous lab and field studies. The relative contributions of each particle type with respect to altitude are shown in Figure 1; for a more detailed examination, Figure 2 illustrates the relative fractions of these particle types with respect to particle diameter, as well as altitude. Standard errors of particle type number percentages were calculated assuming Poisson statistics and reflect the number of clear air particles sampled over all noted flights for each altitude bin. [7] To investigate single‐particle mixing state, the presence of ammonium (m/z 18(NH+4 )), nitrate (m/z −62(NO−3 )), sulfate (m/z −97(HSO−4 )), and sulfuric acid (m/z −195(H2SO4HSO−4 )) [Froyd et al., 2009; Miller et al., 2005] was examined with respect to altitude for the major particle classes [Moffet et al., 2008; Sullivan et al., 2007]. First, individual particles containing each of these m/z markers were identified using a peak area threshold of 100 (arbitrary units). The average peak area of each specific m/z is proportional to the amount of each particular species on a particle [Bhave et al., 2002; Pratt et al.,

2 of 10

D11305


D11305

Table 1. Aircraft‐Aerosol Time‐of‐Flight Mass Spectrometer Particle Type Classificationsa A‐ATOFMS Particle Class Biomass burning Organic carbon (OC) Aromatic Amine Elemental carbon/organic carbon (ECOC) Elemental carbon (EC) Biological Salt (Na‐K‐Cl) Mineral dust Metal Sulfate‐Nitrate Sulfuric Acid

Characteristic Mass Spectral Ion Markers m/z 39(K ); less intense m/z 12(C+), 27(C2H+3 ), 36(C+3 ), 37(C3H+), −26(CN−) [Hudson et al., 2004; Silva et al., 1999] m/z 27(C2H+3 /CHN+), 37(C3H+) [Silva and Prather, 2000; Spencer and Prather, 2006] m/z 77(C6H+5 ), 91(C7H+7 ); less intense m/z 27(C2H+3 /CHN+), 37(C3H+) [Silva and Prather, 2000] m/z 58(C2H5NHCH+2 ); less intense m/z 27(C2H+3 /CHN+), 37(C3H+) [Angelino et al., 2001] m/z 12(C+), 24(C+2 ), 36(C+3 ), 37(C3H+) [Moffet and Prather, 2009; Spencer and Prather, 2006] Carbon cluster ions: C+n , C−n [Moffet and Prather, 2009; Spencer and Prather, 2006] Inorganic ions (m/z 23(Na+), 39(K+), and/or 40(Ca+)), m/z 27(C2H+3 /CHN+), 37(C3H+), −26(CN−), −79(PO−3 ) [Fergenson et al., 2004; Pratt et al., 2009a; Russell, 2009] m/z 23(Na+), 39(K+), −35, −37(Cl−) Inorganic ions (m/z 27(Al+), 39(K+), and/or 40(Ca+)), m/z −79(PO−3 ), often m/z −60(SiO−2 ), −76(SiO−3 ) [Silva et al., 2000] Intense m/z 7(Li+), 56(Fe+), and/or 138(Ba+), for example No positive ions; m/z −62(NO−3 ), −97(HSO−4 ) No positive ions; m/z −195(H2SO4HSO−4 ) [Miller et al., 2005] +

a Detailed descriptions and mixing with ammonium, nitrate, sulfate, and sulfuric acid are noted in the text. A‐ATOFMS, aircraft‐aerosol time‐of‐flight mass spectrometer.

2009b]; therefore, average peak areas for the above noted secondary species were calculated for those particles containing the species within each particle type, or matrix, to examine the relative amount of that species on a particular class of particles. Uncertainties in average peak areas are reported as standard errors.

3. Results and Discussion [8] Average CN number concentrations decreased with increasing altitude from ∼5000/cm3 at 1–2 km MSL to ∼200/cm3 at 6–7 km with the boundary layer top at approximately 3.1 km (not shown). Consistent with measurements near an urban center (Denver, CO), the highest CN concentrations (∼60,000/cm3) were observed at low altitudes following takeoff at 1729 m from Broomfield, CO. Average CN concentrations from 5 to 7 km (∼227/cm3) are similar to previous CN concentration measurements for November– December at 5–10 km over Wyoming (∼300/cm3) [Hofmann, 1993]. Thus, the single‐particle chemistry discussed below is generally representative of a remote continental location. [9] Contributions from wildfires, prescribed burns, and residential wood burning resulted in the biomass burning class representing the largest source of submicron particles: ∼33–39% by number from 1 to 7 km (Figure 1). This is consistent with previous flight‐based PALMS measurements which found that ∼33% of particles sampled in the North American background troposphere were characterized by a biomass burning signature [Hudson et al., 2004]. Previous studies have observed forest fire plumes to loft to high altitudes [Kahn et al., 2008], influencing particle chemistry in the free troposphere and stratosphere [Jost et al., 2004; Petzold et al., 2007]. Further, long‐term surface measurements have shown that biomass burning contributes ∼30% on average annually to the total fine particle mass in the western United States [Park et al., 2007]. The mass spectra of the biomass burning particles (Figure 3a) were characterized by an intense potassium ion peak with less intense carbonaceous

marker ions (e.g., m/z 12(C+), 27(C2H+3 ), 36(C+3 ), 37(C3H+), −26(CN−)) [Hudson et al., 2004; Silva et al., 1999]. With increasing altitude from 1 to 7 km, the number fraction of biomass burning particles mixed with ammonium decreased from 65 ± 3% to 47 ± 4% (Figure 3b). Similarly, the number fraction of biomass burning particles containing nitrate decreased significantly from 87 ± 1% from 1 to 3 km to 35 ± 4% from 4 to 7 km. In addition, the average peak areas of the ammonium and nitrate marker ions for these individual particles decreased 45 ± 7% and 61 ± 9%, respectively, from 1 to 7 km (Figure 3c). In contrast, the number fraction of biomass burning particles containing sulfate stayed relatively constant with altitude while the average sulfate peak area increased by 67 ± 13% from 1 to 7 km. The increasing accumulation of sulfate on biomass burning particles with increasing altitude is consistent with aging [Li et al., 2003] and cloud processing [Verma et al., 2007]. Last, the number fraction of biomass burning particles containing sulfuric acid was relatively

Figure 1. Relative number fractions of particle classes with respect to altitude, binned by 1 km.

3 of 10

D11305


D11305

Figure 2. Relative number fractions of particle classes with respect to vacuum aerodynamic diameter (dva, nm) and altitude (km above mean sea level (AMSL)). constant (∼32–50%) at all altitudes; however, the average sulfuric acid peak area increased by a factor of 4.0 ± 0.7 from 1 to 6 km, consistent with increasing sulfuric acid content, decreasing ammonium concentrations, and increased aqueous phase processing with altitude. [10] The relative fraction of OC particles was greatest (∼33%) near the surface (1.2–2.0 km) with decreased abundance (∼16–21%) above 4 km, suggesting decreased transport to the free troposphere compared to biomass burning particles. As shown in Figure 4a, the OC particle class mass spectral signature was dominated by OC marker ions, including m/z 27(C2H+3 /CHN+), 37(C3H+), 43(CH3CO+/CHNO+), 50(C4H+2 /C3N+), 59(C3H9N+), and 142(C11H+10/C8H16NO+) [Angelino et al., 2001; Erupe et al., 2008; Silva and Prather, 2000]. Oxidized OC is shown at all altitudes through the presence of the intense m/z 43 peak, likely related to gas phase oxidation leading to secondary organic aerosol formation within the boundary layer [Kanakidou et al., 2005], as well as cloud processing at higher altitudes [Ervens et al., 2004]. Sulfate was present in ∼87–97% of OC particles by number at all altitudes (Figure 4b). The number fraction of OC particles mixed with ammonium decreased slightly (20 ± 5%) from 1 to 7 km, but the average ammonium ion peak area associated with these particles decreased 70 ± 5% (Figure 4c). Similarly, the average amount of nitrate on the OC particles decreased with increasing altitude, and while 82 ± 2% of the OC parti-

cles were mixed with nitrate at 1–2 km, few (9 ± 7% by number) contained nitrate at 6–7 km. In contrast, the number fraction of OC particles containing sulfuric acid increased from 40 ± 2% at 2–3 km to 70 ± 4% at 6–7 km with the relative amount of sulfuric acid increasing from 1 to 6 km (corresponding average peak area increase of 6.0 ± 0.9). [11] Two minor subclasses of OC particles were observed: aromatic‐ and amine‐containing. The aromatic‐containing OC particles had a minor contribution (∼2–4% by number) below 4 km with contributions of ≤1% from 4 to 7 km; generally, aromatics are associated with “brown” light‐ absorbing carbon [Andreae and Gelencser, 2006]. The mass spectral signatures of these particles (not shown) were characterized by OC marker ions, aromatic fragment ions (m/z 51(C4H+3 ), 63(C5H+3 ), 77(C6H+5 ), 91(C7H+7 ), 115(C9H+7 ), 165(C13H+9 ), 189(C15H+9 ), 219(C17H+15)), polycyclic aromatic hydrocarbon (PAH) molecular ions (m/z 178 (phenanthrene/ anthracene), 202 (pyrene/fluoranthene)), and organic nitrogen [Silva and Prather, 2000]. Overall, the aromatic particles were mixed with ammonium, nitrate, and sulfate; however, sulfuric acid was not observed on these particles. Given the presence of a coniferous wood burning PAH tracer (m/z 234 and its fragments) [Bente et al., 2006] during this early winter (November–December) study, the leading source of these aromatic particles, primarily found within the boundary layer, is hypothesized to be residential wood burning. The second

4 of 10

D11305


Figure 3. (a) Average mass spectra of biomass burning particles at 1–2 and 6–7 km AMSL. (b) Relative number fractions of biomass burning particles containing ammonium (m/z 18(NH +4 )), nitrate (m/z −62(NO−3 )), sulfate (m/z −97(HSO−4 )), and sulfuric acid (m/z −195(H2SO4HSO−4 )) with respect to altitude from 1 to 7 km, binned by 1 km. (c) Average peak areas of ammonium, nitrate, sulfate, and sulfuric acid for biomass burning particles at different altitudes with respect to the 1–2 km AMSL average.

Figure 4. (a) Average mass spectra of organic carbon (OC) particles at 1–2 and 6–7 km AMSL. (b) Relative number fractions of OC particles containing ammonium (m/z 18(NH+4 )), nitrate (m/z −62(NO−3 )), sulfate (m/z −97(HSO−4 )), and sulfuric acid (m/z −195(H2SO4HSO−4 )) with respect to altitude from 1 to 7 km, binned by 1 km. (c) Average peak areas of ammonium, nitrate, sulfate, and sulfuric acid for OC particles at different altitudes with respect to the 1–2 km AMSL average. 5 of 10

D11305

D11305


D11305

Figure 5. (a) Average mass spectra of elemental carbon (EC) particles at 1–2 and 6–7 km AMSL. (b) Relative number fractions of EC particles containing ammonium (m/z 18(NH+4 )), nitrate (m/z −62(NO−3 )), sulfate (m/z −97(HSO−4 )), and sulfuric acid (m/z −195(H2SO4HSO−4 )) with respect to altitude from 1 to 7 km, binned by 1 km. (c) Average peak areas of ammonium, nitrate, sulfate, and sulfuric acid for EC particles at different altitudes with respect to the 1–2 km AMSL average. OC particle subtype were particles with dominant amine signatures, which contributed ≤1% by number at 1–3 km and 6–7 km. However, from 3 to 6 km, the contribution of amine particles increased to ∼3–11% by number with these particles occupying the smallest (i.e., 150–300 nm) size range measured (Figure 2), potentially suggesting nucleation processes could be an important formation mechanism [Barsanti et al., 2009]. As such, the mass spectra of the amine‐ containing particles varied as a function of altitude. From 1 to 3 km, the mass spectral signature (not shown) was typified by OC, ammonium, nitrate, sulfate, and amine markers: m/z 58(C2H5NHCH+2 ), 59(trimethylamine), and 142(C8H16NO+) [Angelino et al., 2001; Erupe et al., 2008]. In contrast, from 3 to 7 km, the amines are not mixed with ammonium or sulfate but instead they were mixed with other OC species, chloride, nitrate, phosphate, and an amine marker at m/z 118((C2H5)3NOH+) [Angelino et al., 2001]. Previous single‐ particle mass spectrometry measurements have shown ∼30–80% of the aerosol mass in the free troposphere to be carbonaceous [Murphy et al., 2006]. While a small soot core, comprising