SEARCH

SEARCH BY CITATION

Keywords:

  • single particle analysis;
  • Supersite;
  • source apportionment;
  • mass spectrometry;
  • aerosol;
  • Houston

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[1] Between 23 August and 18 September 2000, a single-ultrafine-particle mass spectrometer (RSMS-II) was deployed just east of Houston as part of a sampling intensive during the Houston Supersite Experiment. The sampling site was located just north of the major industrial emission sources. RSMS-II, which simultaneously measures the aerodynamic size and composition of individual ultrafine aerosols, is well suited to resolving some of the chemistry associated with secondary particle formation. Roughly 27,000 aerosol mass spectra were acquired during the intensive period. These were classified and labeled based on the spectral peak patterns using the neural networks algorithm, ART-2a. The frequency of occurrence of each particle class was correlated with time and wind direction. Some classes were present continuously, while others appeared intermittently or for very short time durations. The most frequently detected species at the site were potassium and silicon, with lesser amounts of organics and heavier metals.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[2] The development of single-particle mass spectrometers is rapidly progressing to the point that most field experiments aimed at aerosol characterization include at least one of these instruments. In the present issue, the deployment of four types of single-particle instruments in the 1999 Atlanta Supersite Experiment is described [Rhoads et al., 2002; Wenzel et al., 2002; Lee et al., 2002; Middlebrook et al., 2002; J. L. Jimenez and J. T. Jane, Ambient aerosol sampling using an aerosol mass spectrometer in Atlanta and Houston, submitted to Journal of Geophysical Research, 2002]. Results from all four instruments provide complimentary results, as different size and composition biases result from the various measurement techniques [Middlebrook et al., 2001]. An understanding of the strengths and weaknesses of each instrument is essential to determining the extent to which results may be considered quantitative and to designing an optimized field experiment involving multiple single-particle instruments and multiple sampling sites. The 1999 Atlanta Supersite Experiment featured all four instruments operating simultaneously and at the same location, thus revealing some important differences that could be exploited in future field campaigns that involve multiple sampling sites.

[3] About one year later, during the 2000 Houston Supersite intensive experiment, single-particle mass spectrometers were deployed at three sampling locations. A relatively localized high concentration of industrial sources just east of Houston allowed for strategic placement of these sites such that one was generally upwind (Deer Park), one just downwind (Houston Regional Monitoring site 3), and one far downwind (Aldine) of the main emissions. Figure 1 depicts the locations of the three sites in relation to the city, the major freeways, and the Houston ship channel, where much of the industry is concentrated. The single-ultrafine-particle mass spectrometer, RSMS-II, was deployed at Houston Regional Monitoring site 3 (HRM3), the site closest to the sources, where much of the chemistry associated with new particle formation may be observed. RSMS-II has the capability of analyzing particles with aerodynamic diameters as small as 15 nm. Since the chemistry associated with atmospheric processing of the aerosol may be observed at the farther sampling sites, ATOFMS [Wenzel et al., 2002], which analyzes particles larger than a few hundred nanometers in aerodynamic diameter, was deployed at Aldine and Deer Park. Although this configuration does not allow for direct comparison of results between the two instruments as in Atlanta, it may yield more meaningful information concerning the formation and subsequent evolution of aerosols as they grow in an air parcel tracking over the city.

image

Figure 1. Location of sampling sites with respect to Houston. The 610 freeway circles the city, and the Houston ship channel passes just south of HRM3.

Download figure to PowerPoint

[4] The present work focuses on measurements made by RSMS-II at HRM3 during the sampling intensive. Over 27,000 mass spectra were obtained over a period of 3 weeks. Unlike in Atlanta, sampling was continuous so that diurnal variations in ultrafine aerosol composition can be resolved from the data. The instrument has been previously described in detail [Mallina et al., 2000; Phares et al., 2002; Rhoads et al., 2001], but several changes were made in the experimental design and data acquisition procedure to facilitate continuous sampling. These changes are detailed in the experimental section. The significant particle classes identified at the site are listed in the measurements section. Many of the observed classes are consistent with expected products of the oil refining process; while others are not as readily identified. Since data are acquired in real time, correlations with time of day and wind direction are possible. These correlations in addition to the observed size distributions among the various classes provide very specific information about particle origins. However, no attempt is made in the present study to extrapolate relative mass loadings of the various particle types from the acquired data. Despite this lack of quantitation, unambiguous differences between results from the Houston Supersite and the Atlanta Supersite exist in terms of the dominant particle classes. These differences reflect the differences in the local sources that characterize the two sites.

2. Experiment

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

2.1. Instrument

[5] RSMS-II sizes particles aerodynamically by only transmitting particles with aerodynamic diameters, Dp, that are log-normally distributed such that [Phares et al., 2002]:

  • equation image

where σg = 1.14. The optimally focused size, Dp*, is determined from

  • equation image

where ρp is the particle density, U is the gas velocity through the inlet (sonic), Cc is the slip correction, and μ is the gas viscosity in the inlet. The size selectivity of aerosol transmission into the mass spectrometer was analyzed numerically by Mallina et al. [2000] and experimentally by Phares et al. [2002]. Both of these references contain detailed descriptions of the instrument.

[6] Since RSMS-II is designed to analyze particles that are too small to be detected by light scattering, an excimer laser is pulsed at a high frequency in order to maximize the probability of hitting a particle that is transmitted through the size-selective inlet. Therefore, much more stress is placed on the laser than in aerosol mass spectrometers that use particle detection to trigger the ablation laser. Since reasonable sampling efficiency requires a minimum laser fluence through the ion source region of the mass spectrometer [Kane and Johnston, 2000; Phares et al., 2002], continuous sampling requires an excimer laser that can maintain a threshold pulse energy for many laser shots. Therefore, the main difference between the present instrument and that deployed in Atlanta is the ablation laser. During the Houston intensive, we employed a GAM model EX10 (Orlando, FL) excimer laser (193 nm), which was capable of firing almost 10 million times at a laser energy above roughly 5 mJ per pulse. Also, a longer focal length lens was used to focus the laser beam to the particle ablation point. The new lens resulted in a wider minimum beam width, thereby increasing the size of the active laser volume where particles are ablated and ionized. The minimum beam width in the ion source region was 1.5 mm; more than twice the size of that produced by the lens used in Atlanta.

[7] Ions produced in the source region are accelerated through a linear flight tube by a dual gradient, and detected by a microchannel plate (Burle Electro-Optics model 1331-3200). A digital oscilloscope card (Gage Applied model CS8500-2M), triggered by the excimer laser pulse, digitized the resulting signal and determined whether a particle had been hit from the maximum ion current in the mass spectrum. Due to the increased size of the active laser beam volume in the source region, a new set of voltages were used in the mass spectrometer. These voltages and the plate spacings are listed in Table 1. The difficulty forming negative ions from laser ablation of ultrafine particles [Carson et al., 1997] made positive ion spectra acquisition the chosen mode of study.

Table 1. Mass Spectrometer Specifications
PlatePosition, cmVoltage, kV
Backing05000
First accelerator1.674.3
Second accelerator2.67Gnd
Front of microchannel117−2.2
Back of microchannel118.5−1.4

[8] The sampling procedure involved collecting mass spectra of particles at a selected aerodynamic size until it was determined that a statistically significant data set had been obtained, as described in the next section. The instrument parameters were subsequently adjusted to select another particle size. When the entire desired range of particle sizes was covered, a new size scan was immediately started. Nine aerodynamic sizes were selected for the present study to optimize comparison with speciation results from bulk impactor samples: 35, 50, 70, 100, 140, 170, 320, 590, and 1140 nm. A single scan through the entire size range lasted roughly 110 minutes.

2.2. Data Acquisition and Analysis

[9] Since single-particle instruments obtain tens of thousands of mass spectra during typical field campaigns, it is necessary to employ automated schemes to sort and classify the spectra allowing meaningful information to be extracted. The neural networks algorithm, ART-2a, was chosen for the present study because of previous successful application to data obtained with RSMS-II [Rhoads et al., 2002]. Parameters optimized for laser ablation mass spectrometry were employed as determined by Phares at al. [2001]. During the intensive sampling period, ART-2a was applied on line to determine whether a statistically significant number of particles had been obtained before moving on to the next aerodynamic size in a scan. Particles were thus classified in real time allowing a preliminary class distribution, which determined the need to continue sampling at the same size or move on to the next size. The number of classes, Ns, that comprised 80% of total particles measured during a single size step was monitored. The criterion for continuing on to a new size was:

  • equation image

where nT is the total number of particles analyzed during the size step. Therefore, if only one significant class was identified, only 30 particles needed to be analyzed. If two significant classes were identified, then 60 particles needed to be analyzed. A 10 minute time limit was placed on each time step in case ambient aerosol concentrations were low, or too many significant classes were identified. After all the data were collected, the entire set of mass spectra was reprocessed with the neural networks algorithm, yielding a new complete class distribution.

2.3. Site

[10] RSMS-II was deployed at the HRM3 site located in Channelview, Texas, east of Houston and less than 1.0 km north of the ship channel. The immediate vicinity of the monitoring site can be characterized as very industrial; laden with chemical plants, oil refineries, and incinerators. The site is sandwiched between a very active railway segment to the south and the PVS chemical plant to the north. It is therefore reasonable to expect that the majority of ultrafine particles measured at HRM3 are newly formed through nucleation of freshly emitted gas phase pollutants or directly emitted in the aerosol phase. The proximity of the dominant sources reduces the probability of significant aerosol processing in the atmosphere, unlike sites upwind or far downwind of these sources.

3. Measurement Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[11] Over 27,000 mass spectra were obtained during the intensive sampling period. The ART-2a algorithm identified over 60 total classes of particles. However, only 15 of these comprised over 99% of the total particles measured. These classes were labeled based on characteristic ion peaks in the mass spectra and are listed in Table 2.

Table 2. Major Particle Classes Identified at HRM3 During Supersite Intensivea
Class LabelNumberPercent of Total
  • a

    A total of 27,189 particles was measured.

Potassium840531
Silicon/silicon oxide825430
Carbon434116
Sea salt20497.5
Iron16626.1
Zinc4481.7
Amine, type 13311.2
Amine, type 23241.2
Lime2781.0
Vanadium/vanadium oxide1730.64
Organic mineral1360.50
Amine, type 3780.29
Lead/potassium750.28
Aluminum610.23
Lead salt420.16

[12] Since RSMS-II obtains particle spectra in real time, it is possible to correlate the detection of a specific class with wind direction and time of day, as well as particle size. This information may enable accurate source determination if a strong directionality is observed in certain particle classes. Figures 2 through 17 display these correlations for each significant particle class. In each figure, the mass spectrum is the average of all normalized spectra within a particle class, where the y axis represents integrated ion current. The plot in the bottom right corner of the figures displays the fraction of total particles within each respective class as a function of time of day. The fraction is reported to account for the fact that the instrument did not operate uniformly at all hours of the day. The plot in the bottom left corner of each figure displays the normalized hit rate of each major particle class as a function of aerodynamic diameter. These plots were generated by normalizing the fraction of particles hit in each aerodynamic size bin for each particle class. Considering the size bin fraction rather than the absolute number of particle hits in each size bin may help to eliminate some of the size bias inherent to the instrument; however, such plots do not represent ambient size distributions due to composition biases. For example, as particle size decreases, amine particles may break up into smaller, less characteristic ion fragments upon laser ablation [Kane and Johnston, 2000; Phares et al., 2001]. Thus the smaller particles may be erroneously grouped with the carbon class (more typical of combustion aerosol); and the amine class may actually be smaller than its appearance in the plot. Nevertheless, the plots present positively identified particle sizes within each class.

image

Figure 2. Summary statistics for all particles sampled. See text for more details. Size vs. sample time: the periods of time when this compound class was observed. Polar: Wind rose showing directional preference for this class. Lower right histogram: fraction of particles in this class by time of day. Lower left histogram: fraction of particles in this class by particle size.

Download figure to PowerPoint

image

Figure 3. Summary statistics for potassium compound class. See text for more details. Large graph: average mass spectrum for this class. Size vs. sample time: the periods of time when this compound class was observed. Polar: Wind rose showing directional preference for this class. Lower right histogram: fraction of particles in this class by time of day. Lower left histogram: fraction of particles in this class by particle size.

Download figure to PowerPoint

image

Figure 4. Summary statistics for the silicon/silicon oxide compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 5. Summary statistics for the carbon compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 6. Summary statistics for the sea salt compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 7. Summary statistics for the iron compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 8. Summary statistics for the zinc compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 9. Summary statistics for the amine type 1 compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 10. Summary statistics for the amine type 2 compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 11. Summary statistics for the amine type 3 compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 12. Summary statistics for the lime compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 13. Summary statistics for the vanadium/vanadium oxide compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 14. Summary statistics for the organic mineral compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 15. Summary statistics for the lead/potassium compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 16. Summary statistics for the aluminum compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

image

Figure 17. Summary statistics for the lead salt compound class. See legend of Figure 3 and the text for a detailed description of each panel.

Download figure to PowerPoint

[13] The polar plot displays the frequency of occurrence of each particle class as a function of wind direction. The data are plotted as a wind rose with the radial distance determined by the fraction of total particles from a particular wind directions. The plot just above the wind rose in each figure displays the size and time, expressed in decimal day of the year, corresponding to the detection of all particles within each class.

[14] Figures 713, 16, and 17 contain an extra graph depicting time series corresponding to passing plumes of those particular particle types. The detection frequency (particle hits per hour) is plotted as a function of hour of the day. In each case, a plume is identified as a sharp increase in detection frequency of a certain particle class. The most commonly observed particle types were present consistently throughout the sampling period, and therefore exhibited no sudden increases in local concentration.

[15] Figure 2 summarizes all of the data taken during the intensive. For convenience, the information is displayed on the same template as described above. Two notable features are evident from Figure 2. First, there is a sizeable gap in the data between decimal days 246 and 250, corresponding to 2 and 6 September. The instrument was not sampling during this time period. Second the wind was blowing predominantly from the north and south during the intensive. The presented wind roses account for this bias since the detection frequencies are normalized with the total number of counts for each wind direction.

3.1. Potassium

[16] The most commonly detected particle type during the intensive was characterized by a single large peak in the mass spectra corresponding to potassium, K+ (see Figure 3). The lack of carbon cluster ions in the spectra suggests that most of these particles may not have resulted from biomass burning. Potassium has a very high ionization efficiency, and is very easily detected, even when present in small concentrations.

3.2. Silicon/Silicon Oxide

[17] Perhaps the most surprising observation from the class-resolved size distributions is the abundance of ultrafine Silicon/Silicon Oxide particles, as small as 35 nm in aerodynamic diameter (see Figure 4). The extremely small size of these particles suggests formation from an incineration process or perhaps through thermal decomposition of Silane [Onischuk et al, 2000]. At this point, the origin of these particles is not known. Even more surprising is that they were present steadily throughout the intensive sampling period. Neglecting composition biases, the silicon particles appear to dominate the particles under 50 nm in aerodynamic diameter. The apparent dominance of silicon particles over a few hundred nanometers may be the result of increased sample flow rate and concomitant background that occurs during sampling of these larger sizes. With the increased gas flow, the smallest particles may be able to transmit through the inlet during sampling of the largest particle sizes. It is also possible that the relatively high frequency of particle hits above 400 nm indicates gas phase contamination. Therefore the presented results above 400 nm are, at best, inconclusive without validation from bulk techniques. In both cases, however, it is clear from the data at the larger sizes that a large amount of silicon and silicon oxide was present in the air at HRM3, either in the particulate phase or in the gas phase.

3.3. Carbon

[18] The particle sizes of the carbon containing particle class as determined by the instrument are quite small, mainly between 50 and 140 nm in aerodynamic diameter (Figure 5). This suggests that the class is comprised of combustion aerosol and secondary organic particles. Without more characteristic ions to aid in identification of specific organics (see the amine classes) it is difficult to distinguish between combustion aerosol and secondary organics. The measurement site is located in a heavily industrialized area near the Houston ship channel. There are numerous industrial and power production facilities in the area that may emit carbonaceous particles into the atmosphere. In addition, there is a major freeway (I-10) about 0.5 km to the north of the site. It is clear from the wind rose that the sources are predominantly located in three petals of the wind rose including one to the north suggesting the freeway, but further back trajectory analysis will be needed to more specifically identify these sources.

3.4. Sea Salt

[19] The sodium dominated class portrayed in Figure 6 is most probably sea salt. This particle class appeared consistently throughout the entire study, and most prominently when the wind was blowing from the Gulf of Mexico, southeast of the site. There appears to be an abundance of sea salt in the 70–200 nm aerodynamic size range, and even as small as 35 nm. The dominant presence of sea salt in ultrafine marine boundary layer aerosol has been identified by Murphy et al. [1998].

3.5. Iron

[20] The iron dominated particles in the class portrayed in Figure 7 appear exclusively from the south-southwest, indicating that the source of these particles was likely close to the site. The detection frequency does not seem to tail off at the larger particle sizes, suggesting that the instrument is detecting the tail end of a coarser mode. Also notable in Figure 7 is that a major plume of ultrafine iron particles was detected between 3 and 5 am on 29 August.

3.6. Zinc

[21] Although the more frequently detected particle classes represent the typical composition of ultrafine aerosol at HRM3, the minor particle classes may provide important information concerning transient plumes that passed through the site or pollutants that are too directional to be detected frequently. For example, a class of Zinc-containing particles, comprising 1.7% of total measured particles, appeared exclusively when the wind was blowing from the west (see Figure 8). This particle class appears fairly consistently throughout the study, but only in the early morning hours, exhibiting clear maxima at 2 am and 7 am, corresponding to zinc plumes that passed the site on 1 and 15 September, respectively.

3.7. Amines

[22] Three amine classes were detected during the intensive. All of these exhibit peaks at 30, 44, 58,and 72 Da, corresponding to the CnH2n+2N+ ion fragments indicative of aliphatic amines; however, one (labeled type 1) is dominated by the peak at 30 Da, another (type 2) is dominated by the peak at 58 Da, and the third (type 3), by the peak at 72 Da. Figures 9, 10, and 11 reveal a notable feature that would not otherwise be realized through sole consideration of the mass spectra. Type 1 and type 3 correlate with wind from the south-southwest and the east-southeast, respectively. These two amine types may have originated from different sources, or they may have the same origin but experienced different chemical reactions depending on the trajectory followed by the air parcel.

[23] Type 1 exhibits the largest size distributions, peaking at 320 nm, and also exhibits the smallest ion fragments. This indicates that type 1 is a different amine species from 2 and 3, since the smaller of two particles of the same composition would break into smaller ion fragments upon laser ablation.

[24] As shown in Figures 910, all three amine types were detected as passing plumes. One strong plume of type 2 occurred at 2 pm on 24 August. This coincided with a plume of type 3, indicating that types 2 and 3 may have the same origin. However, types 2 and 3 generally do not correlate well with wind direction. Strong plumes of type 1 were detected at 8 am on 24 August and 10 September, and then again in the early morning hours of 17 September.

3.8. Lime

[25] The various calcium oxide ions in Figure 12 indicate ultrafine particles composed of lime (CaO). The size distribution of the lime particles appears from the data to be very monodisperse around 70 nm. Lime is commonly used in stack gas scrubbers to reduce SO2 emissions from power plants through conversion to calcium sulfite. It is also used for glass manufacturing, and in extracting silicates from ores to produce pure metals. The highly directional nature (east-southeast) of this class suggests a single source of the lime. Almost 40% of the lime particles were detected in three short events at 8 am on 29 August, 11 am on 8 September, and between 7 and 9 pm on 9 September.

3.9. Vanadium/Vanadium Oxide

[26] Vanadium is present in crude oil, and may be emitted during the refining process. The vanadium dominated class displayed in Figure 13 exhibits a relatively narrow size distribution between roughly 100 and 200 nm and appears to be emitted intermittently. Major vanadium plumes were detected on 24 August at 2 pm, 9 September at 7 pm, and 17 September at 8 pm. In all cases, the wind was generally blowing from the south-southwest.

3.10. Organic Mineral

[27] The class depicted in Figure 14 contains calcium, sodium, and small hydrocarbon fragments. There is also a significant peak at 64 Da, most likely a larger organic fragment ion. The size plot in Figure 14 reveals a relatively monodisperse size distribution close to 100 nm. The class appeared consistently throughout the study and generally from the northwest.

3.11. Lead/Potassium

[28] The lead dominated particle class depicted in Figure 15 appeared almost exclusively (78%) on 31 August and 1 September. The other 22% was distributed fairly evenly throughout the rest of the intensive. The presence of a small potassium (K+) ion peak in the mass spectra may indicate only a very small amount of potassium actually present within the particles, due to the high ionization efficiency of potassium compared to lead. Considering that these particles are almost purely lead, the actual size of the particles in this class is close to 10 nm, compared to the aerodynamic diameter of 140 nm apparent from Figure 15, as aerodynamic diameter is proportional to density in the free molecular limit. These particles probably originated from a smelter.

3.12. Aluminum

[29] Figure 16 depicts a class of ultrafine aluminum particles that appear to be most abundant under 50 nm. As with the previously discussed lead class, the actual size of these particles is close to 10 nm. Thus these particles also formed through nucleation from the gas phase. Unlike the lead class, however, the aluminum particles originated mainly from the south-southwest, indicating a different source. Almost 50% of the particles appeared during two plumes: one at 2 pm on 24 August, and the other at 2 am on 10 September.

3.13. Lead Salt

[30] Another minor class (Figure 17) contained a large lead (Pb+) peak and a significant peak centered at 242 amu, perhaps PbCl+. This particular particle class appeared almost exclusively between 1 and 3 am on 31 August, and then only very sporadically afterwards. Since the particle class never recurred in high concentrations, it might be considered a transient emission, rather than a steady one. A material that is present in ultrafine aerosol for a short time may not easily be detected using bulk techniques; yet may still produce significant health and environmental effects.

4. Comparison With 1999 Atlanta Supersite

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[31] The obtained classifications are very different from those obtained by RSMS-II in the 1999 Atlanta Supersite Campaign, where more than 74% of measured particles were carbonaceous. This indicates the dominant influence of the local sources, which were mainly combustion sources close to the Atlanta site. The sources close to HRM3 are more dominated by refineries, incinerators, and chemical plants, as evidenced by vanadium, amines, ultrafine metals, silicon, and lime. Figure 18 displays pie charts depicting the distribution of classes for each aerodynamic size and wind direction. This view is analogous to that presented for the Atlanta data Rhoads et al. [2002, Figure 3]. The present data reveal a more evenly distributed aerosol class distribution, especially from the main sources just south of the site.

image

Figure 18. Size- and wind-resolved particle class distributions, as detected by the instrument.

Download figure to PowerPoint

[32] Despite the dominance of combustion particles detected in Atlanta, a greater number of significant particle classes were identified in Atlanta than at HRM3 in Houston. RSMS-II identified 29 particle classes that comprised more than 0.1% of total particles in Atlanta. In Houston, only the 15 particle classes presented in this paper comprised more than 0.1% of total particles measured. This may be due to the proximity of HRM3 to the main aerosol sources. The Atlanta data, therefore, may contain particles that have experienced more processing by the urban atmosphere, resulting in an increase in aerosol complexity.

5. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[33] The deployment of the single-ultrafine-particle mass spectrometer, RSMS-II, during the Houston Supersite intensive yielded information about the general class distribution of ultrafine particles at HRM3, as well as transient plumes of ultrafine particles that were present at the site for short durations. Detection of transient plumes was possible because of the instrument's fine temporal resolution and its ability to run continuously for long periods of time. The most prominently detected ultrafine particle types contained potassium, silicon, and, to a lesser extent, carbon.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[34] This work was supported by the U.S. Environmental Protection Agency via Houston Supersite funding.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information
  • Carson, P. G., M. V. Johnston, and A. S. Wexler, Laser desorption ionization of ultrafine aerosol particles, Rapp. Commun. Mass Spectrom., 11, 993996, 1997.
    Direct Link:
  • Kane, D. B., and M. V. Johnston, Size and composition biases on the detection of individual ultrafine particles by aerosol mass spectrometry, Environ. Sci. Technol., 34, 48874893, 2000.
  • Lee, S.-H., D. M. Murphy, D. S. Thomson, and A. M. Middlebrook, Nitrate and oxidized organic ions in single particle mass spectra during the 1999 Atlanta Supersites Project, J. Geophys. Res., 108(DX), doi:10.1029/2001JD001455, in press, 2002.
  • Mallina, R. V., A. S. Wexler, K. P. Rhoads, and M. V. Johnston, High speed particle beam generation: A dynamic focusing mechanism for selecting ultrafine particles, Aerosol Sci. Technol., 33, 87104, 2000.
  • Middlebroook, A. M., et al., A comparison of particles mass spectrometers during the 1999 Atlanta Supersite Experiment, J. Geophys. Res., 108(DX), doi:10.1029/2001JD000660, in press, 2002.
  • Murphy, D. M., J. R. Anderson, P. K. Quinn, L. M. McInnes, F. J. Brechtel, S. M. Kreidenweiss, A. M. Middlebrook, M. Posfai, D. D. Thomson, and P. R. Buseck, Influence of sea-salt on aerosol radiative properties in the southern ocean marine boundary layer, Nature, 392, 6265, 1998.
  • Onischuk, A. A., A. I. Levykin, P. Strunin, K. K. Sabelfeld, and V. N. Panfilov, Aggregate formation under homogeneous silane thermal decomposition, J. Aerosol Sci., 31, 12631281, 2000.
  • Phares, D. J., K. P. Rhoads, A. S. Wexler, D. B. Kane, and M. V. Johnston, Application of ART-2a to laser ablation mass spectrometry of particle standards, Anal. Chem., 73, 23382344, 2001.
  • Phares, D. J., K. P. Rhoads, and A. S. Wexler, Performance of a single-ultrafine-particle mass spectrometer, Aerosol Sci. Technol., 36, 583592, 2002.
  • Rhoads, K. P., D. J. Phares, A. S. Wexler, and M. V. Johnston, Size-resolved ultrafine particle composition analysis, 1, Atlanta, J. Geophys. Res., 108(DX), doi:10.1029/2001JD001211, in press, 2002.
  • Wenzel, R. J., D.-Y. Liu, E. S. Edgerton, and K. A. Prather, Aerosol time-of-flight mass spectrometry during the Atlanta Supersite, 2, Scaling procedures, J. Geophys. Res., 108(DX), doi:10.1029/2001JD001563, in press, 2002.

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Experiment
  5. 3. Measurement Results
  6. 4. Comparison With 1999 Atlanta Supersite
  7. 5. Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.