Investigations of SP-AMS Carbon Ion Distributions as a Function of Refractory Black Carbon Particle Type

Abstract

The soot particle aerosol mass spectrometer (SP-AMS) instrument combines continuous wave laser vaporization with electron ionization aerosol mass spectrometry to characterize airborne, refractory black carbon (rBC) particles. The laser selectively vaporizes absorbing rBC-containing particles, allowing the SP-AMS to provide direct chemical information on the refractory and non-refractory chemical components, providing the potential to fingerprint various rBC particle types. In this study, SP-AMS mass spectra were measured for 12 types of rBC particles produced by industrial and combustion processes to explore differences in the carbon cluster (C_n⁺) mass spectra. The C_n⁺ mass spectra were classified into three categories based on their ion distributions, which varied with rBC particle type. The carbon ion distributions were investigated as a function of laser power, electron ionization (on/off), and ion charge (positive or negative). Results indicate that the dominant positive ion-formation mechanism is likely the vaporization of small, neutral carbon clusters followed by electron ionization (C₁⁺ to C₅⁺). Significant ion signal from larger carbon cluster ions (and their fragment ions in the small carbon cluster range), including mid carbon (C₆⁺ to C₂₉⁺) and fullerene (greater than C₃₀⁺) ions, were observed in soot produced under incomplete combustion conditions, including biomass burning, as well as in fullerene-enriched materials. Fullerene ions were also observed at high laser power with electron ionization turned off, formed via an additional ionization mechanism. We expect this SP-AMS technique to find application in the identification of the source and atmospheric history of airborne ambient rBC particles.

Copyright 2015 American Association for Aerosol Research

INTRODUCTION

Black carbon (BC) containing particles are generated as industrial products and as a byproduct of incomplete combustion processes. Industrial applications include reinforcing material in rubber compounds, ink pigments as well as fullerenes and carbon nanotubes for a variety of technical and medical products (Baughman et al. 2002; Scida et al. 2011). Combustion-generated soot particles, which contain black carbon, are an important airborne pollutant (Bond et al. 2007). Black carbon, also termed as refractory black carbon (rBC) or sometimes elemental carbon (EC) (Andreae and Gelencsér 2006; Petzold et al. 2013), is the dominant light-absorbing particulate material that heats the atmosphere (Bond et al. 2013). In addition to directly absorbing sunlight, BC particles can become incorporated into cloud droplets and deposit on snow and ice surfaces, changing cloud properties and surface reflectivity, respectively (Jacobson 2006; Hadley and Kirchstetter 2012). These processes significantly modify the earth's radiation balance (Ramanathan and Carmichael 2008; Bond et al. 2013). Furthermore, atmospheric BC containing particles have been demonstrated to be deleterious to human health (Dockery et al. 1993; Brugge et al. 2007; Lee et al. 2010).

The chemical composition and structure of BC particles, and thus their physical and optical properties (Bond and Bergstrom 2006), depend on their methods of production (Vander Wal and Tomasek 2004). For example, BC-containing soot produced rapidly by spark discharge is highly disordered, and therefore highly reactive (Schmid et al. 2011). At the other extreme, thermally treated and laser-heated BC contains more-ordered graphitic structures (Oberlin 1984; Vander Wal and Jensen 1998), which are correspondingly less reactive (Schmid et al. 2011). Environmentally relevant BC samples produced by diesel engines, aircraft turbines, or biomass burning exhibit a wide range of chemical structures, which may aid in the identification of ambient soot sources (Vander Wal et al. 2010, 2014). This range of chemical and physical properties, which depends upon the efficiency of combustion, affects the BC optical properties and has been described as soot “maturity” (López-Yglesias et al. 2014). Traditional measurements of chemical and physical properties of BC particulate materials include a variety of offline, filter-based techniques, such as Raman and X-ray-based spectroscopies, transmission electron microscopy (TEM) and scanning electron microscopy (SEM), and EC/OC thermal optical techniques (Vander Wal 1997; Bond and Bergstrom 2006; Lack et al. 2014). Laser vaporization and mass spectrometric measurements are other important approaches with the potential for rapid, on-line analysis, including soot particle aerosol mass spectrometry (SP-AMS) (Onasch et al. 2012; Corbin et al. 2014) and single particle laser ablation mass spectrometry (Sinha 1984; Ferge et al. 2006; Maricq 2009).

Mass spectrometric studies of the carbon vapors formed from heated (<2500 K) graphitic carbon filaments were initiated in the 1950s (Chupka and Inghram 1953a,b). Laser vaporization mass spectrometry was employed to study carbon clusters (i.e., pure carbon molecules) in supersonic beams, specifically for clusters with greater than four carbon atoms (Rohlfing et al. 1984; Pflieger et al. 2008). Such studies led to the discovery of buckminsterfullerene (C₆₀) and the classification of fullerenes, in general, as a carbon allotrope (Kroto et al. 1985; Krätschmer et al. 1990). Following the discovery of C₆₀, a wide range of laser vaporization mass spectrometric studies were conducted on BC-containing material. Most of these studies utilized single wavelength, high intensity pulsed lasers for desorption and ionization, which have been shown to directly influence the measured carbon cluster ion distributions; in particular, the distribution of fullerene species (Elsila et al. 2005; Spencer et al. 2008). Previous studies have shown that various carbon materials undergo temperature-driven annealing or graphitizing to different extents (i.e., graphitizing and non-graphitizing carbon) (Harris 2005). High-resolution transmission electron microscopy (HR-TEM) studies have shown partial annealing of BC particles due to heat treatment via pulsed laser light (e.g., laser-induced incandescence techniques), even for extremely rapid (greater than 10⁹ K/s) heating and cooling rates and total times at elevated temperatures of the order of microseconds (Vander Wal and Jensen 1998; Vander Wal and Choi 1999). A subgroup of these studies focused on vaporizing carbon using laser beams with light intensities below the threshold for plasma formation to investigate the thermodynamic parameters of vaporized carbon clusters or the presence of fullerenes (Buseck et al. 1992; Wilson et al. 1993; Pflieger et al. 2008).

The mass spectrometric studies described in this work were conducted using a recently developed instrument, designated as SP-AMS (Onasch et al. 2012), that relies on a continuous wave (CW) laser to vaporize sampled airborne BC particles. The refractory and non-refractory particulate matter (R-PM, including rBC, and NR-PM, respectively) components of absorbing rBC particles are detected using electron ionization mass spectrometry (hereafter, BC detected by the SP-AMS will be referred as refractory black carbon [rBC]). Previous SP-AMS work has shown that the sum of the measured carbon cluster ions is proportional to sampled particulate rBC mass (Onasch et al. 2012) and that differences in the C_n⁺ mass spectra may be related to the chemical structure of rBC (Corbin et al. 2014). Here we focus on the measured C_n⁺ mass spectra for 12 different rBC particle types, including graphitic, amorphous, and fullerenic carbon, and investigate the mechanisms behind the formation of these ions. We investigated the C_n⁺ mass spectra as a function of the SP-AMS laser power, electron ionization (on/off), and the charge of the resulting ions (positive or negative). The real-time, online aerosol particle measurements of SP-AMS are shown to provide potentially useful information for apportioning the mass loading of ambient rBC particles to different sources.

EXPERIMENTAL SECTION

Soot Particle Aerosol Mass Spectrometer

The SP-AMS instrument measures the mass spectra of vaporized material from sampled particles containing rBC. A detailed description of the SP-AMS is presented in Onasch et al. (2012). The SP-AMS is a high-resolution aerosol mass spectrometer (HR-AMS) (DeCarlo et al. 2006; Kimmel et al. 2011) equipped with a CW laser vaporizer based on the Single Particle Soot Photometer (SP2) instrument design (Baumgardner et al. 2004; Schwarz et al. 2006, 2010). The HR-AMS 600°C thermal vaporizer may be used in combination with the CW laser vaporizer to study both NR-PM and light-absorbing R-PM (e.g., rBC). In the experiments presented here, the SP-AMS instrument was operated with only the CW laser vaporizer, making it selectively sensitive to laser light absorbing particles, such as those containing rBC (Cross et al. 2010).

Airborne particles are sampled by the instrument through an aerodynamic lens inlet that focuses particles into a converging beam while the sampled gas is removed via multiple stages of differential pumping. The particle beam intersects the laser vaporizer in the detection region and vaporized R-PM and NR-PM components are ionized by a 70-eV electron beam (electron ionization) and detected by high-resolution time-of-flight mass spectrometry (HR-TOF-MS).

The CW laser vaporizer is an intracavity Nd:YAG laser (1064 nm) pumped using an 808-nm diode laser. The laser cavity, defined by the distance from the Nd:YAG crystal to the mirror and the curvature of the mirror, is designed to generate a Gaussian laser beam with ∼1-mm diameter (defined for a width of 4σ, ±2σ from mean, where σ is the standard deviation) inside the ionization/detection region of the SP-AMS. Willis et al. (2014) estimate that the effective laser beam diameter in the SP-AMS for rBC detection is smaller than the optical diameter (4σ ≤ 0.72 mm). An externally mounted CCD camera is used to track the power, alignment, and the transverse electro-magnetic (TEM) modes of the laser beam by monitoring the small fraction of light transmitted through the mirror. The laser beam power density in the cavity (0.2 MW/cm²) was estimated by measuring the transmission coefficient of the mirror (3 × 10⁻⁴). During standard operation, the laser power is monitored in real-time using the light leaking through the mirror. The laser energy deposited into a single rBC particle is estimated to be ∼1 nJ (or ∼1 erg/pg), assuming a typical rBC particle mass of ∼6 fg (volume equivalent diameter of 200 nm and effective density of 1.4 g/cm³) (Schwarz et al. 2006). Previous studies suggest that this energy density is below the threshold for plasma formation from black carbon materials (Buseck et al. 1992; Gamaly et al. 1999; Pflieger et al. 2008), although the particles’ temperature increase and heating rates may induce partial annealing (Vander Wal and Choi 1999).

The HR-TOF-MS is an orthogonal extraction, reflectron, time-of-flight mass spectrometer from Tofwerk AG (HTOF Platform, Thun, Switzerland), which is standard on all HR-AMS instruments and has been described in detail by DeCarlo et al. (2006). The ion formation chamber is a cross-beam ion source (Pfeiffer Vacuum GmbH) modified to place the electron filament on the ion-extraction axis and with holes drilled on the sides for the laser vaporizer. Ions are accelerated with a 100–200 V potential between the ion source and pulsers and extracted orthogonally into the time-of-flight path every 30–100 μs, depending upon the m/z range of interest. The ion detector comprises two multichannel plates in chevron operated at ∼2 kV with a post-acceleration voltage of ∼3.5 kV. As reported by Gilmore and Seah (2000), the detection efficiencies for high mass ions (>300 m/z) are a strong function of the ion impact energy at the MCP detector. While the current system has not been tested for detecting high mass ions in this regard, all the data in the current study were obtained with SP-AMS instruments operated with similar MS tuning voltages, implying similar detection efficiencies as a function of m/z. This study focuses on qualitative, rather than quantitative, changes to the measured carbon cluster ion mass spectra with the intent of explaining the origins for the observed carbon ion signals in the SP-AMS, and our observations and conclusions should not be significantly affected by this issue. Future work to quantitatively assess SP-AMS carbon cluster ion signals will need to explicitly account for this issue.

The HR-TOF-MS was operated with a mass resolving power, m/Δm, of ∼2000–3000 (V-mode) measured at m/z 200 (m is the mass of a specific ion; Δm is the measure of the peak width, e.g., full width at half maximum [FWHM]) (DeCarlo et al. 2006). This resolution enables the identification, separation, and quantification of carbon ions from typical organic and inorganic ions at a given m/z, such as SO₂⁺ (47.96698), C₄⁺ (48.00000), and CH₄O₂⁺ (48.02113) at 48 m/z. Polycyclic aromatic hydrocarbons (PAH's), which can be co-generated by rBC particle sources, produce ions in the midrange region ∼300 m/z (Dzepina et al. 2007), which may also be resolved from carbon cluster ions.

The ion optics in the SP-AMS instrument were equipped with a bipolar power supply, enabling the sequential study of both positive and negative ions. Standard impact energies of 70 eV typically generate positive ions from neutral molecules; however, secondary electrons may also attach to neutral molecules, generating negative ions. Positive ions may also be generated via other ionization mechanisms, such as thermal ionization, which may occur during particle vaporization for materials with low ionization potentials (e.g., alkali metals).

In this study, we focus exclusively on the carbon cluster ion signals (C_n⁺ or C_n⁻). The sum of these C_n⁺ (C_n⁻) ions is a measure of sampled particulate rBC mass (Onasch et al. 2012). In addition to the C_n⁺ (C_n⁻) ion signals, the SP-AMS mass spectra contained ion signals due to organic and inorganic compounds associated with rBC materials. For laboratory-generated rBC particles with little or no associated organic components (R_BC < 0.5), there is negligible (<1%) interference at these C_n⁺ ion signals (Corbin et al. 2014). For rBC particles with significant associated organic material (R_BC ≥ 10), the fraction of C_n⁺ ion signal from organic materials can become significant (>10%), depending upon the chemical composition and fragmentation of the associated organic material. However, the C_n⁺ fragmentation patterns from organic and rBC material differ, with C₃⁺ the dominate fragment (∼0.45) of the low carbon ion signal from rBC material and C₁⁺ the dominant fragment (>0.6) of the low carbon ion signal from organic material. Note that one could use the correction to C₁⁺, based on the measured C₃⁺ ion signal from Onasch et al. (2012) to reduce the potential impact of C_n⁺ ion signal interference from associated organic material, although this correction was not used in the current study.

Typical measurements using the SP-AMS are done with the laser vaporizer set at maximum power (∼10⁵ W/cm²), 70 eV electron ionization, and positive ion detection. Here we varied the following instrument settings to investigate the parameters affecting C_n⁺ ion signal distributions: (i) positive and negative ion detection, (ii) ionizing electron filament on (70 eV) and off (i.e., thermal ionization only), and (iii) laser vaporizer power.

Materials Studied

For this study, we selected 12 rBC particle types representing a variety of refractory carbon nanostructures, ranging from amorphous to graphitic to fullerenic (Vander Wal et al. 2010). A complete list and description is given in Table 1. The materials included are as follows: graphitic carbon nanoparticles (Aquadag), carbon black (Regal black), amorphous carbon (carbon nanoparticles and glassy carbon spheres), surface-oxidized carbon (NoritSX-activated charcoal), combustion-generated soots (premixed ethylene flame and methane diffusion flame soots generated at several fuel equivalence ratios, φ), and fullerene-enriched carbon (Alfa Aesar fullerene soot and Nano-C fullerene black). In addition, we aerosolized and sampled particles of 99.9% pure C₆₀ (buckminsterfullerene), specified by the manufacturer (MER Corporation). All of these rBC particle types were measured by the SP-AMS to have rBC mass fractions greater than 0.65 (R_BC < 0.5), with the majority having rBC mass fractions greater than 0.9.

TABLE 1 Refractory black carbon (rBC) materials sampled in this work

#	Particle Type	Carbon type (CAS) and usage	Source of particles (lot #)	Generation technique
1.	Aquadag	Graphite (7782-42-5). Used for coatings for light absorption and electrostatics.	Acheson Colloids Company, Port Huron, MI (Product 5307562, lot 8867)	Atomized
2.	Biomass burning	Open air biomass burning emissions.	FLAME3 study (McMeeking et al. 2009)	Open air flame
3.	Carbon nanoparticles	Amorphous carbon (7440-44-0), made by laser decomposition, average particle size 30 nm.	Sigma-Aldrich Inc. (Produce 633100)	Atomized
4.	Diesel exhaust	On road diesel truck exhaust.	Caldecott Tunnel (Dallmann et al. 2014)	Internal diesel engine
5.	Ethylene pre-mixed flame soot	Fuel equilivance ratios from 2–5.	Premixed flat burner flame source.	Flame (Cross et al. 2010)
6.	C60	Buckminsterfullerene, Buckyballs, Carbon-60 (99685-96-8), sublimed.	Materials and Elecrochemical Research (MER) Corporation	Dry powder aerosolization (Tiwari et al. 2013)
7.	Alfa Aesar fullerene soot	Fullerene-enriched soot generated by Kräetschmer-Huffman arc process (Kräetschmer et al. 1990).	Alfra Aesar (Product 40971, lots F12S011 and L20W054)	Atomized
8.	Nano-C fullerene black	Fullerene-enriched carbon black generated by Nano-C's proprietary II-G low-pressure, combustion-based technology.	Nano-C, Westwood, MA (lot DR-070224)	Atomized
9.	Glassy carbon spheres	Amorphous carbon (7440-44-0).	Yokai Carbon Co. and Alfa Aesar (Product 398004, type 1, lot A06Q23)	Atomized
10.	Methane diffusion flame soot	Net equivalence ratios from 0.6–0.9.	Diffusion flame source	Flame (Stipe et al. 2005)
11.	Norit SX-activated charcoal	Acid washed, steam activated carbon with high S/V ratio and adsorptive capacity.	Norit	Atomized
12.	Regal black	Carbon black or furnace black (1333-86-4). Used as an additive for plastic and rubber, pigment, chemical reagent, batteries, and refractories.	Cabot Corporation, Billerica, MA (Product 400R, lot GP-3901)	Atomized

All the manufactured materials, except for C₆₀, were dispersed in distilled water via sonication, aerosolized using a TSI constant output atomizer (model 3076), and dried in a diffusion drier to less than 20% relative humidity (RH) prior to sampling by the SP-AMS instrument. The materials were sampled as either polydisperse or monodisperse aerosol, with peak and selected mobility diameters in the range of 100 to 400 nm. The C₆₀ material was milled in a nitrogen atmosphere and aerosolized via a dry powder dispersion process (Tiwari et al. 2013). The laboratory-based combustion soot particles, generated using either premixed ethylene (Cross et al. 2010) or methane diffusion flame (Stipe et al. 2005) sources, were sub-sampled from the flame effluent, diluted with particle-free air, and mobility size selected prior to measurement.

In addition to the laboratory-generated rBC particle types, we sampled rBC particles in situ using extractive sampling methods from open flame burning of biomass materials obtained during the FLAME3 study at the Montana Fire Science Laboratory (McMeeking et al. 2009) and from the exhaust of diesel trucks in traffic at the Caldecott Tunnel in Berkeley, California (Dallmann et al. 2014). The diesel truck exhaust was measured to have an average rBC mass fraction of 0.38 ± 0.12 (R_BC ∼ 1.6) (Dallmann et al. 2012). The biomass burning soots exhibited measured average rBC mass fractions that ranged from 0.07 (lodgepole pine and turkey oak; R_BC ∼ 13) to 0.13 (gallberry; R_BC ∼ 7) to 0.47 (manzanita; R_BC ∼ 1.1).

RESULTS AND DISCUSSION

Positive Carbon Ion Mass Spectra

We obtained SP-AMS C_n⁺ mass spectra for all of the rBC materials listed in Table 1. Figure 1 shows the positive ion mass spectra for four laboratory-generated particle types selected for their distinct spectral profiles: (a) Regal black, (b) premixed ethylene flame soot generated with a fuel equivalence ratio (φ) of 2, (c) Nano-C fullerene black, and (d) C₆₀ (buckminsterfullerene). The number of carbon atoms comprising C_n⁺ in each mass spectrum are separated into three basic categories (low carbon: C₁⁺ to C₅⁺, mid carbon: C₆⁺ to C₂₉⁺, and fullerenes: C₃₀⁺/C₆₀²⁺ to C₁₆₆⁺), which may be related to the stable structures of C_n⁺ ions (i.e., linear, rings, and fullerenes) as indicated at the top of Figure 1 (Bowers 2014).

FIG. 1. Normalized positive ion mass spectra for (a) Regal black, (b) ethylene soot, (c) Nano-C fullerene black, and (d) C₆₀. Mass spectra represent the measured refractory carbon ion count rate (Hz) normalized to the total carbon ion signal as a function of mass-to-charge ratio (m/z) up to 2000 m/z. The displayed m/z scale corresponds to carbon cluster ions from C₁⁺ (12 m/z) to C₁₆₆⁺ (1992 m/z) as shown on top axis. The C₁⁺–C₅⁺ region has been expanded for clarity. The top of the figure shows the SP-AMS mass spectrometric categories and the related structures for stable carbon clusters.

A common feature of all the mass spectra shown in Figure 1 is the presence of significant signals at low carbon numbers (C₁⁺ to C₅⁺). The higher signals of cations with odd numbers of carbons is consistent with their greater stability (Drowart et al. 1959). In the mass spectra shown in Figure 1, the mass spacing for the dominant C_n⁺ ion signals below C₃₀⁺ is 12 m/z (C₁), whereas the spacing for dominant carbon ion signals above C₃₀⁺ is 24 m/z (C₂). This spacing change may represent a transition from carbon clusters with open 1D and 2D structures (e.g., linear chains, monocyclic and polycyclic rings) to closed or hollow 3D, geodesic structures (Rohlfing et al. 1984; Bloomfield et al. 1985; von Helden et al. 1993; Handschuh et al. 1995; Bowers 2014).

The mass spectrum of Regal black particles (Figure 1a) consists primarily of the carbon cluster ions C₁⁺ to C₅⁺ (∼99% of the total C_n⁺ signals), with a minor contribution from clusters greater than C₅⁺. Similar carbon ion distributions were measured for Aquadag (graphitic carbon), carbon nanoparticles and glassy carbon spheres (amorphous carbon), methane diffusion flame soot (φ_net = 0.7), and NoritSX (activated charcoal).

Mass spectra of ethylene flame soot (φ = 2) contained significant fractions of ions greater than C₅⁺ (Figure 1b), including measureable, although small, ion signals for carbon clusters beyond 2000 m/z (C₁₆₆⁺) (not shown). In particular, ethylene flame soot exhibited significant ion signals in the mid carbon range between 60 m/z (C₅⁺) and 360 m/z (C₃₀⁺ and C₆₀²⁺) (∼0.25 of the total C_n⁺ signals). The rBC materials with significant mid carbon ion fractions (ranging from >0.5 to ∼0.03) in decreasing order included (i) laboratory-based ethylene soot generated in a pre-mixed, flat-burner flame source (φ = 5, 4, 3, 2), (ii) biomass burning soot (gallberry, turkey oak, lodgepole pine, and manzanita), and (iii) Alfa Aesar fullerene soot. All other samples generated negligible (<0.03) fractions of carbon ion signals in the mid carbon category. Mid carbon ion signals were not observed consistently, even for similar rBC particle types. This observation will be discussed later in more detail. These C_n⁺ ion signals appear to be sensitive either to source conditions (e.g., combustion conditions), instrument parameters, or both.

The C_n⁺ mass spectrum of Nano-C fullerene black (Figure 1c) was dominated by a series of ion signals greater than 360 m/z (corresponding to C₃₀⁺ and C₆₀²⁺) (>95% of the total C_n⁺ signals), although it also exhibited signals at low carbon numbers. While the large carbon cluster ion distribution peaks around 1200 m/z, there are large signals at 720 m/z (C₆₀⁺) and 840 m/z (C₇₀⁺), which stand out in the sequence (i.e., as “magic numbers”), indicating the presence of especially stable fullerenes (Fowler 1986). This positive carbon ion series has been identified as being dominated by fullerene ions, especially for carbon numbers greater than 45 (Bowers 2014). In this work, we define total positive fullerene ion signals as being the sum of C_n⁺ signal from C₃₀⁺/C₆₀²⁺ to C₁₆₆⁺.

The carbon ion mass spectrum for C₆₀ (Figure 1d) was dominated by signals at 720 m/z (C₆₀⁺) and 360 m/z (C₆₀²⁺). The carbon-13 isotope abundances at 720 to 726 m/z match those of C₆₀⁺; in addition, the carbon-13 isotope abundances at 360 m/z also match those of C₆₀⁺, but at every half mass (i.e., doubly charged). The formation of doubly charged fullerenes, C₆₀²⁺ and C₇₀²⁺ in particular, is also observed in the spectra of Nano-C fullerene black in Figure 1c. The C₆₀ mass spectrum (Figure 1d) contained ions reflecting the loss of C₂ from both C₆₀⁺ (m/z spacing of 24) and C₆₀²⁺ (m/z spacing of 12) during fragmentation of the parent ions ( O’Brien et al. 1988; Scheier et al. 1994). The C₆₀ mass spectrum also shows ion signals at low carbon numbers, which may represent C_n⁺ ion signals due to potential break up of C₆₀ during preparatory ball milling or carbon vapor (e.g., C₂) lost due to fullerene fragmentation during laser vaporization and electron ionization.

In addition to the laboratory rBC material, we also examined rBC containing particles from two common ambient sources. Figure 2 shows measured SP-AMS C_n⁺ ion mass spectra for (a) on-road diesel truck exhaust and (b) open air biomass burning soot. The diesel truck exhaust was sampled from a single on-road vehicle in the Caldecott Tunnel (Dallmann et al. 2012). The C_n⁺ mass spectrum for the single diesel truck matches the average rBC mass spectrum obtained over 2 h (12:00–14:00 local time) of continuous sampling, during which ∼5% of the vehicles sampled were diesel trucks, suggesting that diesel soot was the dominant source of rBC (Dallmann et al. 2014). The diesel soot C_n⁺ mass spectra were dominated by low carbon ion signals (>0.97).

FIG. 2. Positive carbon ion mass spectra measured in (a) exhaust emissions from a single on-road diesel truck, and (b) smoke from the biomass burning of lodgepole pine. The C₁⁺–C₅⁺ region (1–60 m/z) has been expanded for clarity, and the intensity scales for the C₆⁺ to C₈₃⁺ region (60–1000 m/z) have been expanded to 1:10 (right axes).

The biomass burning soot (Figure 2b) was sampled during the open-air combustion of lodgepole pine during the FLAME 3 laboratory experiments (McMeeking et al. 2009). Three other biomass fuel types (gallberry, turkey oak, and manzanita) were included in this study to show a range of measured C_n⁺ ion distributions (mass spectra not shown). All of these biomass fuel types exhibited SP-AMS C_n⁺ ion distributions with significant ion signals at low carbon numbers (0.68 to 0.92), although lower than diesel soot. It is apparent from Figure 2 that two distinct combustion sources (diesel fuel and lodgepole pine) generated different SP-AMS C_n⁺ ion mass spectra. These differences illustrate the potential of the SP-AMS C_n⁺ measurements to provide identification of ambient particulate rBC sources. In the following sections, we explore some of the complexities in the observed mass spectra that are inherent in the SP-AMS technique.

Multiple Sources of Positive Fullerene Ions

Positive ion mass spectra were obtained with 70-eV electron ionization as well as with the electron source (heated tungsten wire filament) turned off. With the filament off, the only ions observed from the sampled rBC particles were alkali metal ions, such as sodium and potassium, and fullerene ions (no multiply charged ions or smaller carbon fragment ions were observed). For rBC particles with no measureable fullerene ion signal using electron ionization (e.g., Regal black and other graphitic and amorphous rBC particle types), no C_n⁺ ions were observed with the filament off.

Figures 3a and b compare the positive ion mass spectra obtained with the filament on and off for (a) Nano-C fullerene black and (b) Alfa Aesar fullerene soot, respectively. Figure 3c shows the ratio of the positive ion signal with the filament off-to-on for Nano-C fullerene black, Alfa Aesar fullerene soot, and C₆₀ as a function of m/z. Note that the C₆₀ filament off-to-on ratios were only included if the absolute signals had sufficient signal-to-noise; the C₆₀ ratios at 720 m/z were lower than the ratios observed for the fragment ions at C₅₈⁺, C₅₆⁺, C₅₄⁺, and C₅₂⁺. As is evident in Figure 3c, the ratio of fullerene ions during filament off-to-on (i) increased with increasing carbon number, with a few exceptions, and (ii) varied significantly between the fullerene-containing materials.

FIG. 3. (a) Normalized carbon ion mass spectra for Nano-C fullerene black with electron beam on (black) and off (gray/red). (b) Normalized mass spectra for Alfa Aesar fullerene soot with electron beam on (black) and off (gray/red). The laser vaporizer is on in both cases. (c) The ratio of carbon ions with electron beam OFF-to-ON for both materials as a function of m/z. The fractional carbon ion signals for electron beam OFF-to-ON from pure C₆₀ is also included in (c) for the five m/z ion peaks with reasonable signal-to-noise; C₆₀⁺ is the lowest point at the highest m/z in this case.

These observations indicate that in addition to electron ionization, there is, at least, one other mechanism capable of producing positive fullerene ions during CW-laser vaporization. It is likely that the additional ion formation mechanism is via thermionic emission (i.e., thermal energy-driven ionization). Two possible mechanisms for thermionic emission include (1) incoherent, sequential photoexcitation of fullerene molecules in gas phase after vaporization from the laser-heated rBC particles (Ding et al. 1994), and (2) direct release of fullerene ions from laser-heated rBC surfaces during the vaporization process. Mechanism (1) would depend on the laser power density (W/cm²) within the ion formation chamber of the instrument and the transit time of the vaporizing fullerene molecules within the laser beam. Mechanism (2) would depend on fullerenes remaining on the laser-heated rBC particle surfaces until heated to the high temperatures of incandescence/sublimation, at that temperature thermal ionization may become efficient. We are unable to positively identify the ionization mechanism that generates positive fullerene ions with the electron filament turned off. However, it is apparent that positive fullerene ion signal levels (i.e., quantification) and ion distributions in the SP-AMS may be affected by other ionization mechanisms in addition to electron ionization, and therefore may change as a function of the laser vaporizer power.

Laser Vaporizer Power Dependence

Laser vaporizer power is an important instrument parameter which influences the total amount of energy available per rBC particle and its corresponding heating rate, and must be controlled to ensure the laser light intensity is high enough to achieve complete vaporization of the rBC particles passing through the laser (Onasch et al. 2012). Increasing the laser power increases both the total light intensity and the laser beam cross section, which is important, given the potentially incomplete overlap between the laser beam and sampled particle beams (Willis et al. 2014). The results of laser power experiments represent ensemble averages over all of the different trajectories of rBC particles through a given laser beam profile, as governed by the laser beam–particle beam overlap.

Figure 4 shows selected carbon ion signals (C₁⁺ to C₅⁺, C₆₀²⁺, C₅₈⁺, C₆₀⁺, and the total fullerene ion signals) as a function of laser vaporizer power for three rBC particle types: (a) Regal black, (b) Nano-C fullerene black, and (c) C₆₀. All of the carbon ion signals show apparent saturation at the highest laser power attained for each rBC particle type except the fullerene signals in the Nano-C fullerene black sample.

FIG. 4. Laser power studies for (a) Regal black, showing the invariance of C₁⁺–C₅⁺ with laser power, (b) Nano-C fullerene black, showing different laser power dependences for C₃⁺, C₆₀²⁺, C₆₀⁺, and the total fullerene ion signal (C₃₀⁺/C₆₀²⁺ to C₁₆₆⁺), and (c) C₆₀, showing similarities in laser power dependences for the parent C₆₀⁺ and C₆₀²⁺ ions and the fragment C₃⁺ and C₅₈⁺ ions. The laser power axis is a relative measure of the intracavity laser vaporizer power obtained by measuring light leaking through the mirror end of the laser cavity (see text for details).

Regal Black

The carbon ion distribution (C₁⁺–C₅⁺) for Regal black was invariant with laser power, that is, the relative ratios between individual carbon ions (C₁⁺–C₅⁺) were constant with laser power, even as the total ion signal for all ions increases with increasing laser power. The plateau of carbon ion signals with increasing laser power suggests that all Regal black particles passing through the effective region of the laser vaporizer beam were fully vaporized and detected at the higher laser powers. Therefore, the measured C₁⁺–C₅⁺ ion distributions may be a fundamental signature of this carbon type (i.e., carbon black), as there was no evidence for interferences from other ionization mechanisms or fragmentation from larger carbon cluster ions.

Nano-C Fullerene Black

The laser power dependence of the selected ion signals for the fullerene black sample, shown in Figure 4b, is more complex. As shown in Figure 1, Nano-C fullerene black samples exhibited ion signals at low (C₁⁺–C₅⁺) and high (fullerene ions >C₃₀⁺, with some doubly charged fullerene ions, such as C₆₀²⁺) carbon numbers. The low carbon ion signals for Nano-C fullerene black, for clarity shown in Figure 4b using C₃⁺ only, followed a similar trend to the low carbon ion signals of Regal black (Figure 4a), saturating at high laser power with invariant relative ratios (latter not shown in the figure). On the other hand, the total fullerene ion signals for Nano-C fullerene black increased with increasing laser power.

Several distinct trends with laser power can be identified for select fullerene ions in the Nano-C fullerene black case. Specifically, the C₆₀⁺, which is zero at zero laser power, rapidly reaches a relatively constant value at intermediate laser vaporizer powers and subsequently increases significantly with increasing laser power. The C₆₀⁺ power dependence at higher laser power matches the total fullerene ion signal trend. In contrast, C₆₀²⁺ appears to rapidly saturate with increasing laser power, with only a slight decrease at high laser power. These observations suggest that C₆₀⁺ from Nano-C fullerene black are produced by two different mechanisms as a function of laser power, one that dominates at low laser power that also generates C₆₀²⁺ and another at high laser power that does not generate C₆₀²⁺ ions. All these observations appear to be consistent with the conclusions of the previous section of an additional laser power-dependent ionization mechanism for fullerene ions that does not generate multiply charged fullerene ions or fragment ions.

C₆₀

The selected C₆₀ carbon ions shown in Figure 4c did not change significantly with laser power. The parent ion signals, C₆₀⁺ and C₆₀²⁺, and the small carbon cluster ion signals, C₃⁺ and C₅₈⁺ (generated via fragmentation of parent ions during ionization or potentially due to non-C₆₀ carbon material generated by the breakup of C₆₀ during preparatory ball milling), all increased with increasing laser power and appeared to level off at higher laser powers, similar to the behavior of low carbon ion signals for Regal black and Nano-C fullerene black but different from the fullerene ion signals (C₆₀⁺, C₆₀²⁺, and total fullerene ions) of the Nano-C fullerene black. In contrast to Nano-C fullerene black, the fullerene ions from the C₆₀ sample appear to be generated by only one dominant ionization mechanism over the studied range in laser power, which generates both C₆₀⁺ and C₆₀²⁺ in a constant ratio. This observation agrees with the ratios in Figure 3c, which suggests that the C₆₀ sample generates fewer C₆₀⁺ ions (approximately a factor of ∼7 lower) with the filament off than Nano-C fullerene black sample under high laser power conditions. We do not know why the C₆₀²⁺ ions, and associated C₆₀⁺ ions, saturate at higher laser power conditions for the C₆₀ sample, compared with the Nano-C fullerene black. We included the C₆₀ sample in this study as a known quantity, in contrast to the other fullerene-enriched carbon samples whose compositions are not well characterized; however, pure C₆₀ may have a different absorption cross section than other rBC samples, which may affect its laser power dependence in the SP-AMS.

Negative Ion Mass Spectra

In addition to positive ion mass spectrometry, we also obtained mass spectra of negatively charged carbon ions for several of the sampled rBC materials. This mode of operation provides additional information relevant to understanding the SP-AMS technique. Figure 5 shows normalized negative carbon ion mass spectra for three (Regal black, Nano-C fullerene black, and C₆₀) of the four rBC materials shown in Figure 1.

FIG. 5. Normalized negative carbon ion mass spectra for (a) Regal black, (b) Nano-C fullerene black, and (c) C₆₀. The scales for the right axes of (a) and (b) are 1:70; the scale for the left axis of (c) is 1:200. The C₁⁻–C₁₀⁻ region has been expanded for clarity.

Five features are notable in Figure 5. First, at low carbon numbers, the C_n⁻ ion distributions extend to larger cluster sizes (C₂⁻–C₁₀⁻ compared with C₁⁺–C₅⁺) and exhibit an even carbon number preference. This trend is opposite to that observed with the low carbon C_n⁺ distributions and is consistent with previous observations (Bloomfield et al. 1985). Second, virtually no negative mid carbon ions (here defined as between C₁₀⁻ and C₃₀⁻) were observed. Third, negative fullerene ions were only observed for rBC samples which produced positive fullerene ions (e.g., Nano-C fullerene black, and C₆₀ but not for Regal black, Aquadag, carbon nanoparticles, glassy carbon spheres, or methane diffusion flame soot). Fourth, the apparent fragmentation of C₆₀ in negative ion mode (Figure 5c) was significantly reduced compared with fragmentation observed in the positive ion mode (Figure 1d); low carbon negative ion signals were very low (note the scale changes for Figure 5c), and the negative C₂-loss ions (C_(60–2n)⁻) were absent. There were also no multiply charged negative fullerene ions. Fifth, the negative fullerene ion distribution for Nano-C fullerene black (Figure 5b) was significantly different from the positive fullerene ion distribution (Figure 1c); the negative ion distribution exhibited a narrower range of fullerene ions with the exceptionally stable C₆₀⁻ and C₇₀⁻ ions more prominent than their positive ion counterparts (Figure 1c).

Origins of Carbon Cluster Ions in SP-AMS

Low Carbon C₁⁺–C₅⁺

All rBC materials sampled using the SP-AMS exhibit measureable positive ion signals in the low carbon category (C₁⁺–C₅⁺). There are two potential sources for these low carbon ion signals: (i) electron ionization of small carbon clusters in the vapor phase, and (ii) ions generated by vapor phase fragmentation of larger carbon clusters during laser vaporization and/or electron impact. Our observations suggest that both sources may be important, although the latter may be secondary.

Previous work has shown that neutral C₁–C₅ carbon clusters represent the main vapor components from heated graphitic carbon materials (Pflieger et al. 2008) and the most stable geometries of these clusters are linear (von Helden et al. 1993). The positive and negative ion mass spectra for Regal black show that the measured carbon ion fraction in the low carbon category (C₁⁺–C₅⁺ or C₂⁻–C₁₀⁻) is more than 99%. As the distribution of low carbon ions did not change as a function of laser power (Figure 4a), and low carbon ions were absent in the filament-off mass spectra (Figure 3), we conclude that the low carbon ion signals from Regal black particles, a high temperature carbon black, are generated by electron ionization of neutral carbon clusters vaporized directly from the rBC particulate material at ∼4000 K (i.e., incandescence) (Schwarz et al. 2006; Moteki and Kondo 2010). Other rBC materials that also exhibited only low carbon ion signals in the SP-AMS included graphitic nanoparticles (Aquadag), amorphous carbon (carbon nanoparticles and glassy carbon spheres), and “mature” soot (methane diffusion flame soot).

It is also possible that a fraction of low carbon ion signals will be generated from fragmentation of larger carbon clusters. Previous work has shown that fullerene ions, such as C₆₀⁺, can fragment by expelling small carbon clusters, such as C₂, and reclosing into smaller fullerene ions, such as C₅₈⁺ (Kroto et al. 1991; Foltin et al. 1993); smaller carbon cluster ions (<C₂₀⁺) may fragment expelling C₃ (Geusic et al. 1987). We observed evidence for fullerene ion fragmentation; the positive ion mass spectrum of C₆₀ in Figure 1d shows significant C_n⁺ ion signals in the low carbon category (C₁⁺–C₅⁺; ∼20%) and for several fullerene ions smaller than C₆₀⁺ (e.g., 696 m/z C₅₈⁺), the latter consistent with the loss of C₂. The negative ion mass spectrum of C₆₀ (Figure 5c) exhibited reduced fragmentation, which may be due to the lower energy inherent in the negative ion formation of C₆₀⁻ (Cox et al. 1991; Huang et al. 1995). Finally, recent work has shown that the relative ratios of low carbon ions (e.g., C₁⁺/C₃⁺) appear to be correlated with the presence of larger carbon cluster ions, potentially due to the fragmentation of larger clusters (Corbin et al. 2014). Corbin et al. (2014) noted that rBC particulate materials with higher graphitic content exhibited lower C₁⁺/C₃⁺ ratios than less graphitic rBC material. These observations agree with the C_n⁺ positive ion distributions shown in Figures 1–3. The comparison between diesel exhaust and lodgepole pine combustion shown in Figure 2 illustrates that diesel exhaust has a lower C₁⁺/C₃⁺ ratio and less mid and fullerene carbon ion signals than lodgepole pine combustion, suggesting diesel exhaust is a more graphitic rBC particulate material (Corbin et al. 2014) or a more mature soot (López-Yglesias et al. 2014). These combined observations suggest that fragmentation of larger C_n⁺ clusters, such as C₆₀⁺, is likely to be a function of internal energy in vapor phase (Kroto et al. 1991; Foltin et al. 1993) and may affect the formation and relative ratios of low carbon ion signals.

All ambient and ambient-related rBC particle sources measured to date appear to exhibit SP-AMS-measured C_n⁺ ion distributions that are dominated (>60% of the total C_n⁺ ion signals) by ion signals in low carbon numbers (C₁⁺–C₅⁺), with source-specific differences in the low, mid, and high carbon ion signals being secondary features. We have not observed SP-AMS C_n⁺ ion distributions from ambient measurements that resemble the industry-generated fullerene-rich rBC samples included in this study. Based on our current observations, we recommend using low carbon number ions (C₁⁺–C₅⁺) for ambient rBC mass calibrations and quantitative rBC mass and size measurements for the following reasons: (1) ambient C_n⁺ ion measurements are dominated by low carbon ion signals, and (2) low carbon ion signals appear to be generated solely via electron ionization in the SP-AMS, unlike fullerene ions, and therefore appear to be most directly related to the vaporized components of rBC particulate material.

Mid Carbon C₆⁺–C₂₉⁺

The mid carbon category includes C_n⁺ ions in the mass spectral region from 72 to 348 m/z (C₆⁺ to C₂₉⁺), excluding multiple charged larger carbon cluster ions. Carbon clusters in this category have been reported to be typically the most stable ring structures (von Helden et al. 1993) and likely fragment via C₃ expulsion (Geusic et al. 1987). The rBC materials during this study that exhibited significant mid carbon ion signals were generated by fuel-rich, incomplete combustion processes, suggesting that mid carbon ion signals may originate from incompletely graphitized rBC or not mature soot. The mid carbon ions were always associated with rBC materials that also exhibited fullerene ions, suggesting that mid carbon ions may form, at least in part, via fragmentation of fullerenes.

As previously noted, the mid carbon ion signals were sensitive to either source conditions for rBC particles or instrument parameters. This variability is illustrated by three C_n⁺ mass spectra for nominally the same Alfa Aesar fullerene soot (Product 40791, different lots) sampled by different SP-AMS instruments (not shown). In all spectra, low carbon and fullerene ions are present, but the mid carbon ion fractions varied from significant (0.2) to essentially zero (<0.01). The reasons for the observed variability for a specific rBC material are unclear as studies of Alfa Aesar fullerene soot using a single SP-AMS instrument and varying laser power, filament emission current, electron impact energies, laser beam spread, and mass spectrometer tuning voltages could not reproduce these observed changes in the C_n⁺ mass spectra. Thus, the mid carbon ion distributions appear to be affected by a currently unconstrained instrument or sampling-related variable, such as the extent of sampled rBC particle annealing in the laser beam discussed in the next section.

Fullerenes C₃₀⁺/C₆₀²⁺ to C₁₆₆⁺

As is evident in Figures 1–3, fullerene ion signals vary widely for the rBC materials studied. An important issue here is the extent to which the fullerene ion signals represent the original fullerene content of the rBC material, or whether these signals are generated in the process of laser vaporization. This has implications on the distribution of smaller carbon ions as well, which may potentially form from fragmentation of larger carbon clusters. Several previous laser-assisted studies have reported the mass spectrometric detection of fullerene ions from rBC materials (Rohlfing et al. 1984; Spencer et al. 2008; Maricq 2009), and high laser power densities (>10⁸ W/cm²) have been shown to produce fullerene ions as artifacts of the vaporization process (Buseck 2002). In contrast, low laser power densities (10³–10⁵ W/cm²) have been shown to be less prone to generate species not originally present and yield carbon cluster distributions similar to thermal desorption results, indicating the minimal influence of laser power density on the resulting carbon clusters, including fullerenes (Buseck et al. 1992; Wilson et al. 1993; Pflieger et al. 2008).

In our SP-AMS experiments, the laser power density was low (∼10⁵ W/cm²), minimizing the potential for producing fullerene structures by the laser vaporization process (Greenwood et al. 1991). This is supported by the fact that the SP-AMS-measured fullerene ion signals where expected in the sampled rBC material (i.e., fullerene black/soot and C₆₀) and none where fullerenes were not expected (e.g., graphitic and amorphous carbon) for both positive and negative ion detection. However, laser power dependence was observed for fullerene ion signals (Figure 4b), indicating that the laser power density plays a substantial role in the fullerene ion production. Our studies suggest that increasing the laser power density may increase the measured fullerene ion signal due to an additional ionization mechanism (i.e., thermionic emission). In addition to this observation, HR-TEM studies in the literature highlight the potential for rapid annealing of rBC particles in laser-based systems such as SP-AMS (Ugarte 1994; Vander Wal and Choi 1999). We discuss both issues here.

The formation of positive fullerene ions with the filament off (i.e., ion formation via thermionic emission) indicates that there is more than one mechanism for generating fullerene positive ions. An additional ionization mechanism complicates the quantification and interpretation of fullerene content distribution within sampled particles. The additional ionization may be the reason for the large increase observed in total fullerene ion signals with increasing laser power. Figure 4b shows that the mechanism responsible for increase in C₆₀⁺ (and total fullerene ions) in Nano-C fullerene black at high laser power does not generate C₆₀²⁺ ions, consistent with the lack of C₆₀²⁺ ion formation with the filament off in Figure 3. As C₆₀²⁺ appears to be generated via electron ionization only, the C₆₀²⁺ ion signal may be useful in separating the measured C₆₀⁺ ion signal into fractions generated by the dominant ionization mechanisms: electron ionization and thermionic emission. The C₆₀²⁺ ion signal (and presumably the corresponding C₆₀⁺ ion signal) appears to be laser power-independent (Figure 4b), suggesting that C₆₀ structures are not produced by the laser and that the carbon cluster ion signals generated by electron ionization may represent the pre-existing C₆₀ content. Therefore, SP-AMS fullerene ion signals may reflect pre-existing fullerene content in the sampled rBC particulate material, although the amount and distribution are not yet quantitatively determined.

Another laser power density-dependent mechanism that may be playing a substantial role in governing the C_n⁺ mass spectra obtained with the SP-AMS may be the annealing of the sampled rBC particles at high temperatures in the laser vaporizer. While the SP-AMS laser is considered to have a relatively low laser power density, rBC particles transit time through laser of the order of microseconds and heat to incandescence (∼4000 K), and generate heating rates of the order of 1 e7 K/s. Heating rBC higher than 2000 K allows disordered rBC materials to anneal, meaning that the disordered carbon structures of the initial rBC undergo solid-state rearrangement to a more thermodynamically stable configuration. Previous HR-TEM studies have shown partial annealing occurring in rBC particles due to pulsed laser light (i.e., laser-induced incandescence techniques) at power densities ∼10⁷ W/cm², where the heating (cooling) rates were in excess of 1e11 K/s (1e9 K/s) and the total time at elevated temperature conditions is of the order of microseconds (Vander Wal and Jensen 1998; Vander Wal and Choi 1999). Therefore, some level of annealing may be expected to occur within rBC particles in the SP-AMS.

The chemical processes that govern the annealing of carbon materials at high temperatures (3000–4000 K) are not well understood, although some form crystalline graphite (i.e., graphitizing carbon) and others do not (i.e., non-graphitizing carbon) (Harris 2004, 2005). HR-TEM and Raman results indicate that some non-graphitizing carbons generate fullerene-like structures during heat treatment (Ugarte 1994; Burian and Dore 2000; Harris 2005). Thus, it is possible that the laser power dependence of the fullerene ion signals from the Nano-C fullerene black sample (Figure 4b) may be due, in part, to heat-related annealing processes prior to vaporization. In this case, strong fullerene ion signals in the SP-AMS may result from rBC materials with pre-existing carbon structures that facilitate fullerene formation during heating. This potential mechanism is consistent with both our observations of fullerene ions only being generated from a subset of rBC materials, not including graphitic (e.g., mature soot) or amorphous carbon, and our observations of an additional ionization mechanism for generating positive fullerene ions. Furthermore, variations in the initial chemical composition of the rBC material and particle heating rates in the SP-AMS may influence the extent of annealing prior to complete vaporization and may explain the variations observed in the fullerene and mid carbon ions distributions.

CONCLUSIONS

The SP-AMS utilizes a new technique designed to detect and identify, in real-time, airborne rBC containing particles using a CW laser vaporizer. The instrument aims to measure the size, mass, and chemical composition of rBC particles, including both refractory and non-refractory materials. The present study focused on the origin and nature of refractory carbon cluster ions produced in the SP-AMS. The C_n⁺ mass spectra were measured for 12 types of rBC particles produced by industrial and combustion processes, including diesel truck particulate exhaust, and biomass combustion soot. The measured carbon cluster ion distributions varied as a function of rBC particle type and may serve as a fingerprint for the nature of the particles.

The C_n⁺ ion distributions were classified into three categories (low, mid, and fullerene carbons), based on the carbon ion cluster size and the measured mass spectra. The low carbon number (C₁⁺–C₅⁺) ion signals are consistent with a direct measure of small, neutral carbon clusters in the vapor phase, with potential contributions from ions generated by the fragmentation of larger carbon clusters. The mid carbon ions may be related to the fragmentation of fullerene-like structures but have not been consistently reproducible and warrant further study. The fullerene ions were likely produced by the electron ionization of pre-existing fullerenes or fullerene-like structures vaporized from the sampled rBC materials and modified by an additional, thermionic ionization mechanism under high laser power densities. The SP-AMS carbon ion distributions are likely governed, in part, by the annealing of the rBC particulate material at high temperatures in the laser vaporizer prior to complete vaporization, a process that is dependent upon the initial chemical composition of the rBC material and the rate of particle heating.

To date, the most observed ambient and ambient-related rBC particle sources exhibit SP-AMS measured carbon ion distributions that are dominated by ion signals in low carbon numbers (C₁⁺–C₅⁺) (Fortner et al. 2012; Massoli et al. 2012; Corbin et al. 2014; Dallmann et al. 2014), with the source-specific differences presented here being a secondary perturbation. Therefore, we recommend using low carbon number ions (C₁⁺–C₅⁺) for ambient rBC mass calibrations and quantitative rBC mass and size measurements. We expect our observations to provide a basis for the interpretation of future SP-AMS studies aiming to identify the source and atmospheric history of airborne ambient rBC particles.

ACKNOWLEDGMENTS

The authors acknowledge Jason Olfert and Rouzbeh Ghazi for help with the inverted diffusion flame source, and helpful discussions with Joakim Pagels and Axel Eriksson from Lund University.

Funding

This research was supported by DOE ASR grants DE-FG02-05ER63995, DE-SC0006980, and DE-SC0011935, DOE SBIR DE-FG02-07ER84890, NASA SBIR NNX-10CA32C, NSF Atmospheric Science Program grants 1244918 and 1244999, NSF Center for the Environmental Implications of Nanotechnology grant EF-0830093, an EPA STAR Graduate Fellowship to Andrea J. Tiwari, and EPA STAR grants R833747 and 83503301. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not reflect the views of the funding agencies.