Environ. Sci. Technol. 2006, 40, 3327-3335
Fast Determination of the Relative Elemental and Organic Carbon Content of Aerosol Samples by On-Line Single-Particle Aerosol Time-of-Flight Mass Spectrometry
derived from the ATOFMS single-particle aerosol mass spectrometry data. The EC/TC values measured by ATOFMS were compared with the TC/EC values determined by the thermal standard techniques (thermooptical and thermocoulometric method). A good agreement between the EC/TC values obtained by on-line ATOFMS and the offline standard method was found.
T . F E R G E , †,‡ E . K A R G , § A . S C H R O ¨ PPEL,§ K. R. COFFEE,⊥ H. J. TOBIAS,⊥ M. FRANK,⊥ E. E. GARD,⊥ AND R . Z I M M E R M A N N * ,†,‡,# GSF-Forschungszentrum, Institut fu ¨r O ¨ kologische Chemie, Ingolsta¨dter Landstrasse 1, 85764 Neuherberg, Germany, Analytische Chemie, Lehrstuhl fu ¨ r Festko¨rperphysik, Institut fu ¨ r Physik, Universita¨t Augsburg, Universita¨tsstrasse 1, 86159 Augsburg, Germany, GSF Forschungszentrum, Institut fu ¨r Inhalationsbiologie, Ingolsta¨dter Landstrasse 1, 85764 Neuherberg, Germany, Lawrence Livermore National Laboratory, 7000 East Avenue, L211, Livermore, California 94550, and BifAsBayerisches Institut fu ¨ r Umweltforschung und technik, Abteilung Umwelt und Prozesschemie, Am Mittleren Moos 46, 86167 Augsburg, Germany
Introduction
Different particulate matter (PM) samples were investigated by on-line single-particle aerosol time-of-flight mass spectrometry (ATOFMS). The samples consist of soot particulates made by a diffusion flame soot generator (combustion aerosol standard, CAST), industrially produced soot material (printex), soot from a diesel passenger car as well as ambient particulates (urban dust (NIST) and road tunnel dust). Five different CAST soot particle samples were generated with different elemental carbon (EC) and organic carbon (OC) content. The samples were reaerosolized and on-line analyzed by ATOFMS, as well as precipitated on quartz filters for conventional EC/OC analysis. For each sample ca. 1000 ATOFMS single-particle mass spectra were recorded and averaged. A typical averaged soot ATOFMS mass spectrum shows characteristic carbon cluster peak progressions (Cn+) as well as hydrogen-poor carbon cluster peaks (CnH1-3+). These peaks are originated predominately from the elemental carbon (EC) content of the particles. Often additional peaks, which are not due to carbon clusters, are observed, which either are originated from organic compounds (OCsorganic carbon), or from the non-carbonaceous inorganic content of the particles. By classification of the mass spectral peaks as elemental carbon (i.e., the carbon cluster progression peaks) or as peaks originated from organic compounds (i.e., molecular and fragment ions), the relative abundance of elemental (EC) and organic carbon (OC) can be determined. The dimensionless TC/EC values, i.e., the ratio of total carbon content (TC, TC ) OC + EC) to elemental carbon (EC), were * Corresponding author phone: +49 89 3187 4544; fax: +49 89 3187 3371; e-mail:
[email protected]. † GSF Forschungszentrum, Institut fu ¨r O ¨ kologische Chemie. ‡ Universita ¨ t Augsburg. § GSF Forschungszentrum, Institut fu ¨ r Inhalationsbiologie. ⊥ Lawrence Livermore National Laboratory. # BifA. 10.1021/es050799k CCC: $33.50 Published on Web 04/13/2006
2006 American Chemical Society
The overall emissions of carbonaceous particulate matter from anthropogenic sources into the atmosphere are estimated to be 8.0 Tg year-1 for black carbon and approximately 34 Tg year-1 for organic carbon (1). These carbonaceous particles influence the earth climate and are suspected to cause severe adverse health effects when inhaled. Usually, these particles are a result of incomplete combustion processes and consist of a large variety of chemical species. Regarding particle-related health effects, little is known about the specific factors that are responsible for the observed effects. One possibility to address particle-related health effects are epidemiological studies. Epidemiologic studies investigate the correlation or association of time-resolved health data (e.g., hospital admissions, occurrence of asthma or cardiovascular illnesses) of a large population with timeresolved information of the ambient aerosol properties such as mass, size distribution, or chemical parameters of the ambient aerosol. Soot- and particle-associated organic compounds are assumed to be relevant factors for the observed health effects. For example, particle-associated oxygenated organic compounds as well as polycyclic aromatic hydrocarbons are suspected to induce oxidative stress in lung cells, which can be associated with inflammatory processes and thus also with heart and lung diseases. Commonly used sum parameters for description of the carbonaceous material in particulate matter are the elemental carbon content (EC, in µgC/m3), the sum content of organic compounds (OC, in µgC/m3), as well as the total carbon content (TC, in µgC/m3; TC ) OC + EC). Common also is the use of dimensionless ratios, such as the EC/TC value (i.e., the ratio of elemental carbon mass to total carbon mass) or the EC/OC value (i.e., the ratio of elemental carbon mass to organic carbon mass). Due to the intensive application of these sum parameters and dimensionless ratios for description of the carbonaceous fractions of aerosols (e.g., for epidemiologic studies) it is important that new measurement methods for organic and elemental carbon fractions of aerosols are compared with the current standard methodology for the EC, OC, and TC determination. Dahmann et al. (2) proposed parameters such as the dimensionless EC/TC value to evaluate diesel exhaust concentrations in mines and workplace atmospheres. In a similar way such parameters have been introduced and used by the National Institute for Occupational Safety and Health (NIOSH). This led to the established NIOSH 5040 method (3-5) for the determination of the organic and elemental carbon content of aerosol samples. Another example for the application of EC/TC values is the study of Cadle and Groblicki (6). The conventional methods for the determination of EC/ OC or EC/TC values use thermal desorption of semi- and low-volatile hydrocarbons (OC) and combustion of the nonvolatile elemental carbon (EC) during a rising temperature protocol. All thermal desorption techniques use PMloaded quartz fiber filter samples for analysis. Several VOL. 40, NO. 10, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3327
micrograms of PM need to be sampled on a carefully pretreated filter. During the thermal desorption process the evolving carbon species are detected integrally as a function of the temperature. The definition of the separation line between OC and EC depends on the respective method, in particular on the applied temperature protocol and the switching point from the thermal desorption mode (OC analysis with inert gas) and the combustion mode (EC analysis with oxygen added). Often pyrolytic reactions of low-volatile organic material such as oligomeric and polymeric fractions (e.g., humic-like substances, HULIS) in the absence of oxygen evoke complex reactions, including carbonization. The black, refractive fraction originated from low-volatile organic material pretends higher values of elemental carbon (5). The NIOSH 5040 method corrects for this effect by using a laser light transmission measurement during the oxidizing phase (i.e., distinguishing between (colorless) OC and (black) EC components). We used two well-established conventional thermal methods, including the NIOSH 5040 method, in this study as reference (5, 7-9). Aerosol time-of-flight mass spectrometry (ATOFMS) with laser-induced desorption/ionization (LDI) has proven to be a valuable analytical tool for the size-resolved chemical online characterization of aerosol samples (10). With this technique the classification of several different particle types is possible due to their chemical fingerprints, and in principle, elemental carbon as well as organic species (16-18) can be addressed. This raises the question if ATOFMS can be applied for the analysis of the EC/OC ratio. In ATOFMS organic and inorganic carbonaceous material is desorbed and ionized from the particles by an intense laser pulse (laser desorption/ ionization process, LDI) and subsequently analyzed by a bipolar time-of-flight mass spectrometer for simultaneous recording of the mass spectra of anions and cations from a single-particle LDI event. The positive mass spectrum (cations) primarily shows electropositive elements (e.g., metal ions such as the alkali metal cations) or cluster ions thereof, whereas in the negative spectrum, among others, peaks of complex anions, like sulfate, phosphate, and nitrate are present. Due to volatilization, condensation, reaction, and fragmentation of elemental carbon, typical carbon cluster peak series can be observed (as anions and cations), where the ion signals show progressions with a distance of 12 mass units. To evaluate whether single-particle aerosol mass spectrometry is a suitable method for EC/TC analysis, different particulate matter samples, exhibiting a broad range of relative elemental and organic carbonaceous content, were investigated by ATOFMS and conventional methods for EC/OC determination.
Experimental Section Samples. A total of 11 PM samples were analyzed by ATOFMS in this study. This includes five samples from a diffusion flame soot generator (combustion aerosol standardsCAST, Matter Engineering, Wohlen, Switzerland; sample nos. 1-5). CAST is a particle generator for the controlled production of soot particles in a wide size and concentration range. It uses a laminar diffusion flame to produce soot particles. In a diffusion flame, the oxygen is transported from an outer sheath air flow to an inner burner gas flow by diffusion while oxidizing the fuel gas. The flame is “cut open” halfway to its tip, using inert gas for quenching the reaction and stabilization of the formed soot particles. The soot particles are extracted via a sampling probe. Depending on the ratio of burner gas and sheath gas (mixing parameter, MP) the size of the primary particles can be influenced. The generator was running in the fine (MP1, dp ) 180 s200 nm, sample nos. 1 and 2) and ultrafine (MP4, dp ∼90 nm, sample nos. 3 and 4 and MP5, dp ∼60 nm, sample no. 5) particle size 3328
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 10, 2006
TABLE 1. Particulate Matter (PM) Samples Used in the Study sample label
specification
no. 1
diffusion flame soot, CAST, MP 1, high OC, primary particle diameter ∼180-200 nm no. 2 diffusion flame soot, CAST, MP 1, low OC, primary particle diameter ∼180-200 nm no. 3 diffusion flame soot, CAST, MP 4, high OC, primary particle diameter ∼90 nm no. 4 diffusion flame soot, CAST, MP 4, low OC, primary particle diameter ∼90 nm no. 5 diffusion flame soot, CAST, MP 5, no OC adjustment possible, primary particle diameter ∼60 nm no. 6 soot aerosol from pure graphite (GFG 1000) no. 7 printex 90 no. 8 printex G no. 9 diesel soot (sampled directly at passenger car) no. 10 NIST urban dust, SRM1649a no. 11 soot aerosol sampled in road tunnel
mode. Fine and ultrafine particles were produced with higher (sample nos. 1 and 3) and lower (sample nos. 2 and 4) organic content, respectively, by variation of the oxygen contends of the sheath air. The sheath air composition was 20% oxygen and 80% nitrogen for particles with an increased organic load (“high OC”) and 24% oxygen and 74% nitrogen for particles with lower organic load (“low OC”). Particles (sample nos. 1-5) were taken from the CAST output using a dilution unit (model MD19, Matter Engineering, Switzerland) in order to cool and stabilize the particle population. Particles were subsequently sampled on PTFE filters (no. 11807-50-N, 0.2 µm pore size, Sartorius, Germany) at 9 L/min. The particles were mechanically removed from the filters and stored under argon gas in sealed brown glass bulbs to avoid UV light interaction. Furthermore, three relatively pure elemental carbon samples from different sources were investigated (sample nos. 6-8). Sample no. 6 is soot directly derived from pure graphite by generating sparks between two graphite electrodes in argon gas (GFG 1000, Palas, Karlsruhe, Germany). The vaporized graphite subsequently condenses to very fine particles, which coagulate to form agglomerates similar to natural soot particles. The sampling process from the GFG 1000 generator (sample no. 6) was identical to the one performed at the CAST generator. Sample nos. 7 and 8 are the commercially available carbon black materials (printex90 and printex G respectively, Degussa, Du ¨ sseldorf, Germany). These samples are used as models for pure EC particles. Finally, some real-world ambient and emission PM samples are included, namely, emission particles from a diesel passenger car tail pipe (no. 9), commercially available urban dust standard particles obtained from the U.S. National Institute of Standards (NIST 1649a, no. 10), and particles from a road tunnel (no. 11). The diesel particles (sample no. 9) were taken from the tail pipe of a commercial passenger car (model FIAT Ulysses 2100). Samples were drawn by a 50 cm tube, immerged into the tail pipe at nonloaded motor conditions. The sampling volume flow was smaller than the exhaust gas flow of the motor. Particles were sampled on PTFE filters (no. 11807-50-N, 0.2 µm pore size, Sartorius, Germany) by a vacuum pump. The tunnel dust particles (sample no. 11) were collected as powder from the ventilation system of the Plaputsch tunnel at the Autobahn (highway) at Graz, Austria. Up to the time of analysis all samples were stored in hermetically sealed glass vials. Table 1 summarizes the samples used in this study. Aliquots of the samples were analyzed by single-particle aerosol mass spectrometry
(ATOFMS) as well as by thermocoulometric measurements (2) at the “Institut fu ¨ r Gefahrstoff-Forschung” (IGF) in Bochum, Germany and by NIOSH 5040 measurements (thermooptical measurement) at the Department of Chemical Engineering of Clarkson University, Potsdam, NY. Aerosol Mass Spectrometry (ATOFMS). For analysis, the samples were suspended by inserting a few milligrams of sample into a 150 mL, 0.22 µm filter/sterilizer (Corning Inc., Corning, NY) which then was agitated continuously. The resuspended particles were conducted through copper tubing to an aerosol time-of-flight mass spectrometer (ATOFMS model 3800, TSI Inc. St. Paul, MN) (11). In this instrument, air and entrained particles are drawn through a converging nozzle into the vacuum. Each particle is accelerated to a specific terminal velocity, which is a direct function of the particle’s aerodynamic diameter. Particles continue through three stages of differential pumping and pass two continuous laser beams, scattering light from these beams. The time between the scattering events is a direct measure of the particle velocity and therefore the particle size. This timing information is also used to calculate the time of arrival in the center of the ion source of a bipolar time-of-flight mass spectrometer. A pulsed Nd:YAG laser with a wavelength of 266 nm is fired at this time and subsequently desorbs and ionizes material from the particle (laser desorption/ionizationsLDI). Positive and negative mass spectra are recorded simultaneously for each single particle. The output of the laser (pulse width, 5 ns; pulse energy, 0.2 mJ) was focused to a spot with a diameter of 400 µm resulting in a power density of approximately 3.2 × 107 W/cm2. For each ATOFMS spectrum as shown in the Figures 1-3, 1000 single-particle mass spectra (bipolar) were added. Note that the obtained size information was not used in this study, as the used resuspended material show altered size distributions. Conventional EC/OC Analysis. To compare the mass spectrometric approach to the current thermal standard techniques, the EC and TC values of all samples were measured by independent laboratories. For the EC and TC determination with thermal methods the samples need to be transferred on quartz fiber filters. Therefore, the particles were redispersed from stock by a rotating brush aerosol generator (model RBG-1000, Palas, Germany). Particles were loaded as dust samples into the piston and dispersed into a 100 L mixing box at ambient pressure. Samples were taken from this box on quartz filters by means of a vacuum pump. Aliquots of the loaded quartz fiber filters were used for the thermal EC and TC analysis. The NIOSH method 5040 is based on the volatilization and oxidation of organic compounds and elemental carbon at different temperatures from a filter sample. First, organic carbon is liberated under different temperatures in a pure helium atmosphere. The evolving, volatilized carbon compounds are oxidized to carbon dioxide (MnO2 oxidizer), transformed into methane (nickel catalyst), and subsequently are quantified by a flame ionization detector (FID). In a second phase, 2% oxygen is added to the helium to oxidize and volatilize elemental carbon. A correction for pyrolytically generated elemental carbon by monitoring the filter transmission is performed (4, 12). In contrast the thermocoulometric measurement volatilizes organic carbon in a nitrogen atmosphere and elemental carbon in oxygen. The volatilized material is oxidized (CuO catalyst), and the resulting CO2 is transferred into an electrolyte solution (e.g., BaOH). The rising pH is back-titrated, yielding a quantitative measure of carbonaceous material.
Results and Discussion Figures 1 and 2 show averaged combined mass spectra of all soot samples. All mass spectra shown in this paper are due to averages of over 1000 individual single-particle mass
spectra and therefore can be considered be a good statistical representation of the whole sample, respectively. This was supported by the obtained size distributions, which are not skewed for any particle size. In this scope, it is important to note that of course the size distribution measured by the ATOFMS aerosol mass spectrometer does not represent the original size distribution of the aerosol due to the sampling, storage (agglomeration), and resuspension procedures. In our study, the maxima of the particle size distributions of the investigated PM samples is shifted due to the agglomeration effects to larger diameters compared to those of the freshly emitted/formed soot aerosol. However, as the EC/TC values are derived from averaged spectra, representing a large number of individual sampled single particles covering a broad size range, the size of the single measured particles should not influence the results. For potential future realtime on-line analysis of particles from ambient or emission sources it must be considered that for very small particles (e.g., primary soot particles) the achievable ATOFMS sensitivity might be an issue for determination of EC/TC values (i.e., the tiny OC fraction in small particles might fall below the ATOFMS detection limit). Furthermore, the lower limit of active size measurement with the ATOFMS system is in the 250 nm region. In the case of the present study this was not a problem as due to the mentioned storage and particle agglomeration the main fraction of analyzed particles was well above 400 nm. Each of the full bipolar mass spectra covering positive and negative ions from the same particles was produced by first separately calibrating the “half” spectra from positive and negative ions and subsequent linking them (13). Anions have negative mass-to-charge ratio (m/z < 0), and cations consequently have positive mass-to-charge ratios. In the averaged spectra (except sample nos. 10 and 11s the ambient PM samples) the typical peak pattern of carbon clusters with an m/z progression of multiples of 12 is clearly visible. The carbon cluster ions are produced upon the LDI process by fragmentation, volatilization, and recondensation of carbonaceous material. This carbon cluster pattern is characteristic for carbonaceous soot particles in singleparticle aerosol mass spectrometry studies (14, 15). With the use of high laser power densities for LDI (>108 W cm-1) these cluster peaks can only be assigned as peaks resulting from any carbonaceous matter, regardless of the elemental or organic nature of the carbon material. However, in recent studies (16-18) it was shown that by carefully adjusting the laser power (>5 × 107 W cm-1) also the ions of the organic fraction of aerosols are directly accessible in the LDI singleparticle aerosol mass spectrum. Although this is true with respect to molecular ions predominately for polycyclic aromatic hydrocarbons (PAH), most organic molecules show fragments at other m/z values than those of the typical carbon cluster series. Keeping these results from previous studies in mind, the peaks from carbon clusters in the mass spectra of sample nos. 1-5 (diffusion flame soot) shown in Figure 1 can be assigned to be predominately originated from elemental carbon. Taking a closer look at the positive part of the mass spectra of the diffusion flame soot samples, where the content of organic carbon is kept high (sample nos. 1, 3, and 5) due to adjustment of the oxygen supply during the particle production, the progression of peaks from elemental carbon is superimposed with a series of smaller peaks between these clusters (Figures 1 and 3a). On one hand in these spectra in the mass region above m/z ) 120 several mass peaks can be assigned to PAH (m/z ) 128 (naphthalene), m/z ) 150 (acenaphthylene), m/z ) 152 (acenaphthene), m/z ) 166 (fluorene), m/z ) 178 (e.g., phenanthrene), m/z ) 202 (e.g., chrysene)ssee Figures 1-3) which can be ionized without fragmentation under the chosen conditions (16). On the other hand, in addition to the PAH masses, sample VOL. 40, NO. 10, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3329
FIGURE 1. Averaged ATOFMS mass spectra (1000 single-particle spectra, respectively) of the diffusion flame soot samples (nos. 1-5) and of the spark-generated particles (sample no. 6). The observed masses of polycyclic aromatic hydrocarbons (PAH) are indicated with dotted lines (m/z )128, 150, 178, and 202). nos. 1, 3, and 5 do show further mass peaks between the carbon cluster progression peaks. These peaks are originated from fragmented aliphatic organic species (note that on the basis of the particle production procedure, a non-carbonaceous inorganic content of the particles of sample nos. 1-5 can be excluded). In Figure 3 this is shown in more detail exemplary for sample nos. 1 (high OC content) and 2 (low OC content). Considering the negative mass spectra (Figures 1 and 2), solely cluster ions are present in the spectra, regardless of the organic content of the sample. Only sample nos. 3 and 3330
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 10, 2006
5 show a small peak at m/z ) -42, which can be tentatively assigned to CNO-. In summary, no peaks in the negative spectra can be unambiguously specified as organic carbon. This is a known feature of the anion LDI mass spectra. Even particles made of pure organic compounds predominantly show carbon cluster peaks in the negative mass spectrum (17). Accordingly, the negative mass spectra of soot particles are dominated by carbon clusters andswhen presentspeaks from inorganic compounds such as nitrates, sulfates, and phosphates. A detection of organic species in combination with elemental carbon or even an assignment of organic
FIGURE 2. Averaged ATOFMS mass spectra of printex materials (sample nos. 7 and 8), as examples of pure EC particles, as well as diesel soot (no. 9), NIST urban dust SRM1649a (no. 10), and dust sampled in a street tunnel (no. 11), which represent ambient soot aerosols. The observed masses of PAH are indicated with dotted lines (m/z )128, 150, 178, and 202). species in the negative mass spectrum is therefore difficult, if at all possible. Therefore, the distinction between EC and OC in not possible on the basis of the negative ion mass spectrum solely. In Figure 3 the assignment of peaks as EC and OC is depicted in more detail. For estimation of the EC content, peaks with masses of m/z ) n12, m/z ) n12+ 1, m/z ) n12 + 2, and m/z ) n12 + 3 (with n ) 1, 2, ...) were used. The series with m/z ) n12 represents the pure carbon cluster progression. The inclusion of the progressions with m/z ) n12 + 1-3 is due to two effects: First, one has to consider the isotopic peak pattern of carbon. However, the relatively high peak areas for the observed m/z ) n12 + 1-3 peaks, which are in general higher than expected, cannot be explained by the isotopic pattern alone (see Figure 3). It is known that the elemental carbon fractions of aerosol particles do not represent chemically pure graphite-carbon but still contain some saturation with hydrogen atoms (i.e., at the border of graphitic moieties). This is also reflected in the
behavior of the carbonaceous material in particulate matter in a thermal analysis. If a stepwise heating is performed, such in the protocol used by Chow et al. (5), more than one “elemental carbon” fractions are observed. In consequence, some influence of the residual hydrogen atoms in the LDI mass spectra of elemental carbon particles and of fractions hereof is to be expected (19). The CnHy peaks with three and more carbon atoms and low relative hydrogen content (y e 3) predominately originate from EC-like polymeric carbonaceous fractions, while CnHy peaks with higher relative hydrogen content are probably due to fragmented organic compounds (hydrocarbons or organic polymeric fractions) or more OC-like polymeric fractions (such as HULIS in SOA). Therefore, in this work the peaks CnH+, CnH2+, and CnH3+ were included for the calculation of the EC content. This procedure is supported also by the fact that particle mass spectra of almost pure carbonic particles still exhibit these peaks (i.e., the m/z ) n12 + 1-3 peaks), while typical organic fragment peaks are absent. VOL. 40, NO. 10, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3331
FIGURE 3. Average positive ATOFMS mass spectra (top) of diffusion flame samples and the respective bar graph representation (bottom) with bars assigned to elemental (full bars) and organic (hatched bars) carbon. The bars represent the integrated MS peak areas (see text). The observed masses of PAH are indicated with dotted lines (m/z )128, 150, and 178). (a) Diffusion flame sample no. 1 (high organic carbon). (b) Diffusion flame sample no. 2 (low organic carbon). The obtained mass spectra of sample nos. 6, 7, and 8 are depicted in Figures 1 and 2. Although these carbon black materials are considered pure EC particles, the spectra of the printex materials show significant peaks from sodium (m/z ) 23) and potassium (m/z ) 39/41). Due to the low ionization potentials (IP) of alkali metals (Na, IP ) 5 eV; K, IP ) 4.3 eV, compared to C, IP ) 11.96 eV and C2, IP ) 12 eV (20)), the LDI process is much more sensitive to these electropositive elements (18). Therefore, the mass spectra were corrected for both sodium and potassium. Although m/z ) 39 can be assigned as both potassium and one of the isotopic peaks of C3+ and/or C3H3+, it can be assumed that it is mainly due to potassium traces present in the samples. It is a well-known fact that in LDI-based aerosol mass spectrometry the mass spectra can be heavily skewed in the presence of even relatively small amounts of potassium in the analyzed particles. The error made by neglecting the area fraction of m/z ) 39, however, is marginal. In real-world samples (no. 9 diesel soot, no. 10 ambient PM, no. 11 tunnel dust; see Figure 2) the influence of inorganic compounds becomes more prominent. The negative spectra show carbon clusters as well as peaks from inorganic compounds as sulfates, phosphates, and nitrates (not explicitly assigned here). For the diesel emission aerosol (no. 9), the positive mass spectra are still dominated by carbon clusters, whereas in the spectra of urban dust (no. 10) and tunnel dust (no. 11) alkali species prevail. Note that in the spectrum of diesel exhaust several PAH peaks are clearly detectable (m/z ) 128, 166, 178, and 202). In the case of sample no. 11 this yields a positive mass spectrum, which is dominated by peaks of sodium and potassium. Elemental carbon and organic carbon are only relatively small peaks in the positive mass spectrum. However, they are present with 3332
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 10, 2006
TABLE 2. Peaks of Typical Inorganic Compounds, Which Were Subtracted from the Spectra Prior to Further Discrimination between Elemental and Organic Carbon m /z
assignment
7 23 24 27 39/41 40 43 54/56 64/66/68 206/207/208/209
Li+ Na+ Mg+ Al+ K+ Ca+ AlO+ Fe+ Zn+ Pb+
high signal intensities in the negative mass spectrum. The small intensity of the carbon cluster ions in the positive mass spectrum is also due to saturation of the MCP detector caused by the very intense sodium and potassium signals. High intense signals at low masses influence the sensitivity for the heavier ions arriving later at the MCP detector (21) (MCP saturation effect). As mentioned above, the determination of the relative content of EC and OC only can be performed with the positive mass spectra (i.e., no anionic OC signals). Therefore, a thorough correction for inorganic peaks in positive MS was performed. Peaks which are due to known inorganic compounds (e.g., metal ions and oxides) are not considered for the EC and OC analysis. Table 2 gives the prominent mass signals from inorganic compounds present in the ATOFM mass spectra of ambient PM and an assignment of these peaks.
To determine a ratio of elemental and organic carbon the respective peak areas were analyzed after a baseline correction. The substantial tailing of some of the potassium peaks (see Figure 2, sample nos. 6-11) is also taken into account. Therefore, the tailing is extrapolated and subsequently subtracted from the spectrum. The single peak areas were added up for the peaks assigned to elemental and organic carbon, respectively. These sum values were then used to calculate EC/TC ratios as a measure of the relative content of elemental carbon according to TC ) EC + OC (TC ) total carbon). Note that only ratios of peak areas are considered. As only the total intensity of all peaks but not their peak ratios are influenced by the MCP saturation effect, no additional correction for detector saturation is necessary. The real-world sample nos. 10 and 11 show organic ions in the mass range above >200 m/z which likely are due to polymeric and oligomeric carbonaceous fractions (i.e., humic-like substances, HULIS), an indication of the contribution of secondary organic aerosol (SOA). To evaluate the ATOFMS-based method of determining EC/TC ratios, the obtained values were compared to the results of measurements with two independent standard techniques (NIOSH method 5040 (3, 12) thermocoulometric measurements (22, 23); see the Experimental Section). In Figure 4 the values of EC/TC ratios determined by the standard methods as well as the values which were achieved with the ATOFMS single-particle aerosol mass spectrometry method are compared; Table 3 summarizes the exact values. For the thermocoulometric method error bars are given (three replicates); the NIOSH 5040 measurements were performed only once per sample. The error for the mass spectrometric method was estimated by culling the spectra of one sample into equal bins of 250 spectra each and calculation of EC/TC values for each average spectrum obtained for each bin. If the thermocoulometric measurements are compared with the results from the NIOSH method 5040 (Table 3), systematically higher EC/TC values are obtained for all soot samples (sample nos. 1-9) for the thermocoulometric method. The overestimation of elemental carbon by the thermocoulometric method due to the missing correction for the pyrolytic generation of refractive, black carbon is well-known (22) In contrast, for the more complex ambient and tunnel dust samples (nos. 10 and 11) the thermocoulometric measurements yield lower EC/TC ratios compared to the NIOSH method 5040. This can be explained by the fact that no acid treatment for removal of inorganic carbon (carbonates) is performed in the thermocoulometric method in contrast to the NIOSH method 5040. However, this is more evident for the urban dust (sample no. 10) than for the tunnel dust (sample no. 11), where occurrence of higher quantities of carbonates is not very likely. It is clearly visible that the values derived from singleparticle aerosol mass spectrometry, in principle, are comparable to the values obtained from the standard techniques. This is especially true for the correlation of single-particle aerosol mass spectrometry and the averaged thermocoulometric and thermooptical values (Figure 4b). The only sample showing a more obvious deviating ATOFMS single-particle aerosol mass spectrometry EC/TC value is the tunnel dust aerosol (no. 11). Here, the relative EC content is underestimated by about 20%. The positive mass spectrum of this sample is dominated by the mass signals of sodium and potassium (m/z ) 23 and m/z ) 39/41). After the correction for the inorganic compounds, only peaks of minor relative intensity remain in the spectrum. This is, as mentioned above, partly due to MCP detector saturation effects in the positive mass spectrum. Tiny variations in signal strength of the small remaining peaks thus have a strong impact on the resulting EC/TC value and therefore can cause larger errors. This means that with increasing content of inorganic compounds (metals)
FIGURE 4. Direct comparison of the resulting EC/TC values derived from all three methods for all samples. Panel a shows the comparison of the ATOFMS results with the values derived from thermocoulometric and thermooptical measurements; panel b shows the comparison of the ATOFMS results with the averaged value of the standard techniques.
TABLE 3. EC/TC Ratios as Determined with the Standardized Methods as Mentioned in the Text as Well as the Values, Which Were Established with Single-Particle Aerosol Mass Spectrometry coulometric method sample
TC [%]
EC/TC [%]
no. 1 no. 2 no. 3 no. 4 no. 5 no. 6 no. 7 no. 8 no. 9 no. 10 no. 11
86.7 83.7 74.7 74.2 72.8 80.1 77.4 16.4 93.1 12.8 69.2
93.8 ( 0.9 99.3 ( 0.4 96.8 ( 0.5 98.3 ( 0.3 89.4 ( 0.5 89.9 ( 0.2 99.3 ( 0.4 99.7 ( 0.1 95.1 ( 0.1 72.8 ( 1.3 62.6 ( 1.5
thermooptical method TC [%]
single-particle aerosol mass spectrometry
EC/TC [%]
EC/TC [%]
91.0 98.4 80.8 92.7 74.8 83.1 98.0 99.1 79.6 83.5 69.2
90.4 ( 5.3 94.4 ( 0.9 88.6 ( 3.1 93.3 ( 2.4 79.9 ( 2.2 87.0 ( 2.0 94.8 ( 0.3 93.0 ( 1.7 87.3 ( 0.7 76.6 ( 1.0 50.6 ( 1.6
the observation of carbonaceous species becomes more and more difficult and the differentiation into EC and OC shows an increasing uncertainty and usually an overestimation of VOL. 40, NO. 10, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3333
OC. Even though the estimated error for sample nos. 10 and 11 are relatively low (below 4%) the overestimation of OC may be a serious problem for samples of mainly inorganic composition. Considering the detector saturation effect and the relatively low carbon content of the ambient samples (16.4% and 12.8% for sample nos. 10 and 11), it is traceable that carbonaceous material in these samples mainly appears in the negative ATOFMS mass spectrum. As the negative mass spectra cannot be used for a discrimination of EC and OC, with such samples only the minor peaks in the positive spectrum remain for this purpose. However, the problem of highly intense inorganic peaks, leading to detector saturation, can be overcome in the future by the use of a socalled mass gate unit. The mass gate unit allows the deflection of ions of distinct, preselected masses and therefore prevents the saturation of the MCP detector (21). The device consists of a gated wire mesh electrode, placed close to the first spacefocus of the time-of-flight mass spectrometer. During the recording of the TOF mass spectrum, deflection potentials can be applied for short times (fractions of microseconds), when unwanted ions m (e.g., m/z ) 23 and m/z ) 39, respectively) are passing the mass gate by means of an electronic delay generator (e.g., DG535, Stanford Research Inc., CA). Other masses are not affected. The deflection of such very intense peaks, however, might slightly increase the background noise of the spectrum due to the increased number of scattered ions in the TOFMS unit. Single-particle aerosol mass spectrometry represents a method for on-line determination of EC/TC values, which agree rather well with EC/TC values derived by standard methods. This is especially true for soot aerosols with not too high inorganic content. For the determination of EC/TC values from ambient aerosols the implementation of a mass gate device could help to increase the measurement precision. Previous studies (24, 25) illustrate that carbonaceous material primarily is present in particles with diameters less than 1 µm, while inorganic material dominates in particles greater than 1 µm. The extension of the data analysis to several size bins in the range from 200 nm to 3 µm can be used for generation of size-resolved EC/OC values. Furthermore, ATOFMS will be used to investigate the occurrence of oligomeric and polymeric carbonaceous fractions in the ambient aerosol in the future.
Acknowledgments The authors thank Dr. Fricke at the “Institut fu ¨ r GefahrstoffForschung” (IGF) in Bochum, Germany for the thermocoulometric measurements and Professor Phil Hopke at the Department of Chemical Engineering of Clarkson University, Potsdam, NY for his support with the NIOSH 5040 measurements. This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory under Contract W-7405-Eng-48. The Lawrence Livermore National Laboratory contributed financially to the experiments through Laboratory Directed Research and Development Grant 02ERD-002. This work was carried out in cooperation with the GSF-Focus-Network “Aerosols and Health” which coordinates aerosol-related research within the GSF Research Centre. Funding for travel expenses provided by the Bavarian Californian Technology Center (BaCaTeC) is gratefully acknowledged. T.F. thanks the Deutsche Bundesstiftung Umwelt for a Ph.D. scholarship.
Literature Cited (1) Bond, T. C.; Streets, D. G.; Yarber, K. F.; Nelson, S. M.; Woo, J.-H.; Klimont, Z. A technology-based global inventory of black and organic carbon emissions from combustion. J. Geophys. Res., D: Atmos. 2004, 109, D14203. 3334
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 10, 2006
(2) Dahmann, D.; Fricke, H.-H.; Bauer, H.-D. Diesel engine emissions in workplace-atmospheres in Germany. Occup. Hyg. 1996, 3, 255-262. (3) Birch, M. E.; Cary, R. A. Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust. Aerosol Sci. Technol. 1996, 25, 221-241. (4) NIOSH. In NIOSH Manual of Analytical Methods, 4th ed.; Cassinelli, M. E., Ed.; National Institute for Occupational Safety and Health, DHHS (NIOSH): Cincinnati, OH, 1996; Vol. 1st Supplement, Publication No. 96-135. (5) Chow, J. C.; Watson, J. G.; Crow, D.; Lowenthal, D. H.; Merrifield, T. Comparison of IMPROVE and NIOSH carbon measurements. Aerosol Sci. Technol. 2001, 34, 23-34. (6) Cadle, S. H.; Groblicki, P. J. In Particulate Carbon: Atmospheric Life Cycle; Klimisch, R. L., Ed.; Plenum Press: New York, 1982; pp 89-109. (7) Countess, R. Interlaboratory analyses of carbonaceous aerosol samples. Aerosol Sci. Technol. 1990, 12, 112-121. (8) Guillemin, M.; Cachier, H.; Chini, C.; Dabill, D.; Dahmann, D.; Diebold, F.; Fischer, A.; Fricke, H.-H.; Groves, J. A.; Hebisch, R.; Houpillart, M.; Israel, G.; Mattenklott, M.; Moldenhauer, W.; Sandino, J. P.; Schlums, C.; Sutter, E.; Tucek, E. International round robin tests on the measurement of carbon in diesel exhasut particulates. Int. Arch. Occup. Environ. Health 1997, 70, 161-172. (9) Hitzenberger, R.; Jennings, S. G.; Larson, S. M.; Dillner, A.; Cachier, H.; Galambos, Z.; Rouc, A.; Spain, T. G. Intercomparison of measurement methods for black carbon aerosols. Atmos. Environ. 1999, 33, 2823-2833. (10) Suess, D. T.; Prather, K. A. Mass spectrometry of aerosols. Chem. Rev. 1999, 99, 3007-3035. (11) Gard, E.; Mayer, J. E.; Morrical, B. D.; Dienes, T.; Fergenson, D. P.; Prather, K. A. Real-time analysis of individual atmospheric aerosol particles: Design and performance of a portable ATOFMS. Anal. Chem. 1997, 69, 4083-4091. (12) Chow, J. C.; Watson, J. G.; Pritchett, L. C.; Pierson, W. R.; Frazier, C. A.; Purcell, R. G. The dri thermal/optical reflectance carbon analysis system: description, evaluation and applications in U.S. air quality studies. Atmos. Environ. 1995, 27A, 1185. (13) Steele, P. T.; Tobias, H. J.; Fergenson, D. P.; Pitesky, M. E.; Horn, J. M.; Czerwieniec, G. A.; Russel, S. A.; Lebrilla, C. B.; Gard, E. E.; Frank, M. Laser power dependence of mass spectral signatures from individual bacterial spores in bioaerosol mass spectrometry. Anal. Chem. 2003, 75, 5480-5487. (14) Noble, C. A.; Prather, K. A. Real-time measurement of correlated size and composition profiles of individual atmospheric aerosol particles. Environ. Sci. Technol. 1996, 30, 2667-2680. (15) Held, A.; Hinz, K.-P.; Trimborn, A.; Spengler, B.; Klemm, O. Chemical classes of atmospheric aerosol particles at a rural site in central Europe during winter. J. Aerosol Sci. 2002, 33, 581594. (16) Zimmermann, R.; Ferge, T.; Ga¨lli, M.; Karlsson, R. Application of single particle laser desorption/ionization time-of-flight mass spectrometry for detection of polycyclic aromatic hydrocarbons from soot particles originated from an industrial combustion process. Rapid Commun. Mass Spectrom. 2003, 17, 851-859. (17) Silva, P. J.; Prather, K. A. Interpretation of mass spectra from organic compounds in aerosol time-of-flight mass spectrometry. Anal. Chem. 2000, 72, 3553-3562. (18) Gross, D. S.; Ga¨lli, M. E.; Silva, P. J.; Prather, K. A. Relative sensitivity factors for alkali metal and ammonium cations in single-particle aerosol time-of-flight mass spectra. Anal. Chem. 2000, 72, 416-422. (19) Reilly, P. T. A.; Gieray, R. A.; Whitten, W. B.; Ramsey, J. M. Direct observation of the evolution of the soot carbonization process in an acetylene diffusion flame via real-time aerosol mass spectrometry. Combust. Flame 2000, 122, 90-104. (20) Weltner, W.; van Zee, R. J. Carbon molecules, ions and clusters. Chem. Rev. 1989, 89, 1713-1747. (21) Mu ¨hlberger, F.; Hafner, K.; Kaesdorf, S.; Ferge, T.; Zimmermann, R. Comprehensive on-line characterization of complex gas mixtures by quasi-simultaneous resonance-enhanced multiphoton ionization, vacuum-UV single-photon ionization, and electron impact ionization in a time-of-flight mass spectrometer: Setup and instrument characterization. Anal. Chem. 2004, 76, 6753-6764. (22) Birch, M. E. Analysis of carbonaceous aerosols: interlaboratory comparison. Analyst 1998, 123, 851-857. (23) 1/120.44, Z. In Method No. 44: Diesel Engine Emissions; Carl Heymanns Verlag: Ko¨ln, Germany, 1995.
(24) Liu, D. Y.; Wenzel, R. J.; Prather, K. A. Aerosol time-of-flight mass spectrometry during the Atlanta supersite experiment: 1. Measurements. J. Geophys. Res., D: Atmos. 2003, 108 (D7), 8426.
aerosol mass spectrometers. Atmos. Environ. 2004, 38, 57455758.
(25) Alfarra, M. R.; Coe, H.; Allan, J. D.; Bower, K. N.; Boudries, H.; Canagaratna, M. R.; Jimenez, J. L.; Jayne, J. T.; Garforth, A. A.; Li, S.-M.; Worsnop, D. R. Characterization of urban and rural organic particulate in the lower Fraser valley using two aerodyne
Received for review April 26, 2005. Revised manuscript received January 16, 2006. Accepted January 28, 2006. ES050799K
VOL. 40, NO. 10, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3335
