Effects of Meteorological Conditions on Aerosol Composition and

May 2, 2002 - The scaling function used for this study (α = 3432; β = −5.436) compares well with those determined by Allen et al. and falls within...
38 downloads 14 Views 863KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 2345-2353

Effects of Meteorological Conditions on Aerosol Composition and Mixing State in Bakersfield, CA JEFFREY R. WHITEAKER,† DAVID T. SUESS,† AND K I M B E R L Y A . P R A T H E R * ,‡ Department of Chemistry, University of California, Riverside, Riverside, California 92521, and Department of Chemistry and Biochemistry, University of California, San Diego, 9500 Gilman Drive, La Jolla, California 92093-0314

Particle and meteorological instrumentation were used to characterize ambient atmospheric conditions, aerosol size distributions, aerosol mass concentrations, and single particle size and chemical composition in Bakersfield, CA for the period January 9, 1999 through January 28, 1999. The sampling period included four distinct meteorological periods of stagnation, clearing, haze, and rain. Particle number and mass concentrations were the highest during the stagnation episode when a heavy and extensive fog developed. Mass and number concentrations also approached these high levels during the haze period. Single particle size and composition data from an aerosol time-of-flight mass spectrometer (ATOFMS) are used to provide unique continuous information on the diversity in types of particles present, the effects of meteorology on particle size and composition, and the distribution of important chemical species within individual particles. Aerosol composition and mixing state are found to vary with meteorological conditions. Single particle data show that carbonaceous aerosol with secondary ammonium, nitrate, and sulfate dominate the aerosol concentration during a stagnation period with a dramatic composition shift occurring to sodium type particles during the haze period. The aerosol is internally mixed with respect to carbon, nitrate, sulfate, and ammonium during the stagnation period. The mixing state changes significantly over the haze period when much greater diversity in the associations of chemical species within individual particles occurs.

Introduction The San Joaquin Valley (SJV), located in central California (Figure 1), represents an excellent venue for studying ambient aerosol processes. It constitutes a large area of California, supports a population of over 3 million inhabitants, and represents the largest agricultural production area in the state. The potential for air pollution to adversely affect the population and crop production of the SJV necessitates a detailed study of aerosol sources and particle chemistry occurring in the region (1). The study described herein was performed as part of an instrument intercomparison study * Corresponding author phone: (858)822-5312; fax: (858)534-7042; e-mail: [email protected]. † University of California, Riverside. ‡ University of California, San Diego. 10.1021/es011381z CCC: $22.00 Published on Web 05/02/2002

 2002 American Chemical Society

and part of the California Regional PM10/PM2.5 Air Quality Study (CRPAQS). This program was established to gain further insights into particulate matter formation by obtaining continuous monitoring data on particles in the SJV (2, 3). Primary particle emission sources in the SJV include farming operations, road dust, industrial processes, and fuel combustion (4). Geographically, the valley is contained to the west by the Coastal Range (reaching 1530 m above sea level) and the east by the Sierra Nevada (reaching heights in excess of 4300 m above sea level). The valley is further isolated to the south by the Tehachapi Mountains with mountain passes leading to the Los Angeles basin and the Mojave Desert at elevations of approximately 1200 m above sea level. The study described in this paper was performed in Bakersfield, a large urban center found in the southern portion of the SJV. The meteorology of the region has been characterized previously (5). Briefly, inversion conditions are frequent, and the high mountains surrounding the valley lead to trapping of pollutants. This entrapment of air becomes particularly severe during the winter when conditions are characterized by extensive stagnation periods. Stable air masses can last several days with extreme episodes lasting up to 2 weeks. Extensive and persistent fogs can also be present during these events causing a dramatic reduction in visibility. Occasional frontal passages associated with low-pressure systems interrupt these periods and can increase visibility. Under typical meteorological conditions, winds contribute to pollutant transport from the northwest and can have significant effects on sites downwind in the valley (6). The transport of primary particulate matter has been shown to occur over subregional scales (10-30 km), and gas-phase precursors of secondary aerosol have been found to disperse over large distances (100 km) (7). In effect, geography and meteorology play significant roles in the transport and buildup of pollutants in and around the southern portion of the SJV, including Bakersfield. Previous air quality studies concentrated on sources of particulate matter (PM) in the valley and development of a conceptual model to understand the causes of elevated PM levels (8-12). Results show that Bakersfield, an urban commercial area located in the southern portion of the valley, consistently shows higher PM values than other valley sites. The highest levels are found in winter during air stagnation episodes. When extremely stable air masses are present, the PM2.5 (PM from particles < 2.5 µm) fraction dominates the overall PM10 (PM for particles < 10 µm diameter) concentration (9). A large PM2.5 fraction may be due to the accumulation of particles and formation of secondary aerosols which are primarily found on particles with diameters smaller than 2.5 µm (13). Of the particles present in winter, 75-80% of PM2.5 consists of nitrate, sulfate, ammonium, organic carbon, and elemental carbon (9, 10). Application of a Chemical Mass Balance receptor model showed that secondary ammonium nitrate, carbonaceous aerosol, secondary ammonium sulfate, and geologic material dominate PM in the winter months (8, 11, 14). Aerosol optical properties have been used to identify secondary aerosol as the dominant contributor to light extinction during wintertime episodes (15). Other apportionment efforts show that a large source of particulate nitrate may be the reaction of nitrate radical with water vapor at night to form nitric acid with subsequent condensation on preexisting aerosols (16). Investigations into secondary organic aerosol formation reveal that most carbonaceous aerosol present is primary carbon (17). The influence of fog chemistry on aerosol VOL. 36, NO. 11, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2345

FIGURE 1. Map showing locations of urban centers, major roadways, and geographical features of the San Joaquin Valley (SJV), California. The boundary of the San Joaquin Valley Air District as defined by the California Air Resources Board is indicated by the bold line. Aerosol sampling was performed in Bakersfield, an urban center located in the southern SJV for the period January 9, 1999 to January 28, 1999. processing in the SJV has also been studied (18-20). Lastly, investigations into the contributions of suspended road dust from unpaved shoulders to elevated PM10 levels indicate that it may be a significant source of geologic material (21). These and other studies identified significant sources of particulate matter in the SJV. A great deal has been learned from previous studies involving the characterization of PM utilizing conventional filter aerosol sampling techniques for collection and analysis of the ambient aerosol. However, long sampling times and reported chemical losses associated with filter sampling limit the comprehensive analysis of aerosol properties and processes (22, 23). Furthermore, analysis of filter and impactor samples determines bulk aerosol properties which provide no indication of the mixing state of the aerosol or the associations of chemical species within particles. In addition, the high cost associated with collecting and analyzing multiple filter samples precludes establishing a continuous picture of the size-segregated composition of aerosol particles over a long term study. To overcome many of the reported problems associated with off-line sampling and analysis, a number of research groups are developing real-time monitoring techniques for individual particles in the atmosphere (24-26). These techniques allow for the spatial and temporal analysis of chemical species present in ambient atmospheric aerosols (27, 28). Individual particle analysis techniques also offer the possibility of determining the degree of chemical heterogeneity among particles in a given size range, information that is essential for a complete understanding of the atmospheric aerosol (29). By making these measurements in real time, the temporal trends of chemical species can be evaluated on relatively short time scales. The aerosol timeof-flight mass spectrometer (ATOFMS) has proven to be capable of continuous (i.e. 24 h a day) aerosol characterization for long time periods (30). It has also been shown that instruments using aerodynamic sizing in conjunction with single particle mass spectrometry are capable of producing quantitative compositional information through the cor2346

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 11, 2002

rection of raw number counts to atmospheric particle concentrations (31). This paper applies these advances to produce the first quantitative size-resolved aerosol composition information at high temporal resolution for a long-term field study. The goals of this paper are to present an overview of measurements made with a transportable ATOFMS and complementary continuous particulate monitoring instrumentation in Bakersfield, CA. The characterization of the predominant particle types, the influence of meteorology on the presence and abundance of particle types, and the distribution of important chemical species within individual particles including a measure of the degree of internal versus external mixing are presented.

Experimental Methods Particle size, composition, and mass measurements along with meteorological data were collected nearly continuously for 3 weeks from January 9, 1999 to January 28, 1999. Instruments were operated at an urban site in Bakersfield, CA, in the southern part of the SJV. The sampling site was located in an office suite neighboring a California Air Resources Board (CARB) air quality monitoring station. All sample inlets were placed on the roof of the building at a height of approximately 10 m and located approximately 50 m from the nearest roadway. The sampling site is not located near any significant source of atmospheric aerosols. An aerosol time-of-flight mass spectrometer was used to perform single particle size and composition measurements. The transportable ATOFMS has been described in detail elsewhere (32). Briefly, particles are drawn into a vacuum through a convergent nozzle where they are accelerated to velocities dependent on their aerodynamic size. Two subsequent skimmers collimate the particle beam. The particles then enter a light scattering region where they pass through two perpendicular continuous wave diode pumped Nd:YAG lasers (532 nm) positioned orthogonal to the particle beam. Scattered light signals from each particle are collected by

FIGURE 2. Time series nephelometer light scattering coefficient (bscat) data for the sampling period and identification of weather conditions. photomultiplier tubes and sent to a timing/logic circuit. The timing circuit sends a signal to fire a frequency quadrupoled Nd:YAG laser (266 nm) at the time the particle reaches the source of a time-of-flight mass spectrometer. Molecular and ionic components are desorbed and ionized by the laser pulse. The resulting positively and negatively charged ions are analyzed by a dual polarity reflectron time-of-flight mass spectrometer. Particle velocity and the corresponding positive and negative mass spectra are recorded. Aerodynamic diameter (da) can be determined from the particle velocity by generation of an external calibration curve using particles of known size. In this experiment, polystyrene latex spheres (Nanosphere Size Standards, Duke Scientific Corp., Palo Alto, CA) of 0.3-2.4 µm diameter were suspended using a homebuilt Collison atomizer to create monodisperse aerosols used for size calibration. A nephelometer (Model M903, Radiance Research, Seattle, WA) was used to measure the light scattering coefficient (bscat) of the ambient aerosol. The light scattering coefficient is an important particle characteristic as it is inversely proportional to visibility. Determination of bscat is accomplished by drawing particles into a chamber where a light source (flashlamp, 530 nm) illuminates the aerosol. Integrating the intensity of scattered light produces a value for bscat. Pressure and temperature sensors automatically correct for changes in air Rayleigh scattering thus isolating the scattering contribution due to particles. Data points were recorded in five minute averages. Meteorological data including wind speed, wind direction, relative humidity, and ambient temperature were taken directly next to the sample inlets for the monitoring equipment. All meteorological instruments were mounted on a 10 foot tripod on the roof of the field site, and measurements were taken nearly continuously. Single particle mass spectra were saved as peak lists and imported into a database for data analysis. Peak thresholds were set to record only those peaks with heights greater than 10 units and areas greater than 20 units in order to distinguish peaks from the background noise in the mass spectra. Database analysis was carried out using a database constructed in Microsoft Access (33). The ATOFMS particle

counts must be corrected for the instrument transmission efficiency to obtain atmospherically relevant particle concentrations. For a complete description of particle number correction procedures see Allen et al. (31). Busy-time corrections as used by Allen et al. are not applied. Particles are binned into corresponding Micro Orifice Uniform Deposit Impactor (MOUDI) size bins (0.32-0.56; 0.56-1.0; 1.0-1.8 µm). Particle detection efficiency curves are determined by comparing the binned ATOFMS mass distributions to the particulate mass as measured by MOUDI. Slip correction factors and aerodynamic diameter conversion used in the calculations assume a mean free path of 0.066 µm and a particle density of 1.3 g‚cm-3, a value between inorganic and organic substances found in ambient aerosols.

Results and Discussion. Meteorology. The entire sampling interval at Bakersfield can be divided into four meteorological periods based on observations of the general conditions present, including a stagnation episode, a clearing period, light haze, and rain. The first period was characterized by air stagnation in which a dense fog was present in the evening and occasionally during the day. This period lasted from January 9 through January 16. Temperatures ranged from -2 to 18 °C, and relative humidity was in the range 41-100%. The average wind speed was less than 2 m/s with wind direction predominantly from the northwest. Visibility was poor during this period due to the dense fog at night and thick haze during the day, a cycle that has been previously described as the smog-fog-smog cycle (34, 35). Values for bscat are inversely related to the visual range and therefore serve as good indicators of relative visibility (36). The bscat values recorded during the early period (Figure 2) are indicative of the poor visibility present. The second period was characterized as a clearing episode. A weak front passed through the area from the northwest to the southeast on the afternoon and evening of January 17 and lasted through January 20. Average wind speeds reached a maximum on January 18 of approximately 5 m/s. Typical conditions were clear visibility with overcast and cloudy skies. Nephelometer data show the dramatic change in visibility VOL. 36, NO. 11, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2347

FIGURE 3. Positive and negative ion mass spectra of individual particles for the predominant particle types identified in Bakersfield. Single carbonaceous particle mass spectra (a), single elemental carbonaceous particle mass spectra (b), single sodium containing particle mass spectra (c), and single calcium-rich dust particle mass spectra (d). in the values for bscat (Figure 2). Temperatures ranged from 7 to 21 °C with relative humidities in the range 41-92%. This period was significant in that it eliminated the intense stagnation episode. The third period consisted of sun and haze. Overall, conditions were very mild during this period, which spanned from January 21 to January 23. Temperatures ranged from 4 to 17 °C and wind speeds decreased to values below 2 m/s. The wind direction oscillated between the northwest and southeast. The stability of the air mass was evident as there was a marked decrease in visibility for these days as indicated in the nephelometer data (Figure 2). There was also the formation of light fogs in the evenings. A large winter weather system made up the final period. Rain began to fall on the evening of January 23 with intermittent showers continuing until the morning of January 25. Ambient temperatures fell from approximately 7 degrees Celsius at 12:00 pm January 24 to around 0 degrees Celsius at 12:00 p.m. January 25. Wind speeds dropped from 5 m/s to around 1 m/s at 10:00 p.m. January 24 and remained calm until approximately 11:00 a.m. January 25. Snow fell from approximately 12:00 a.m. to 6:00 a.m. January 25 with overcast skies constituting the remainder of the day. Skies remained cloudy until it cleared on January 28 when conditions were mild with clear skies. Temperatures returned to the range of 4-11 °C. The large variety of weather patterns made for excellent sampling conditions during this study. Identification of Particle Types. The mass spectra of 418 442 individual particles were obtained over the entire sampling period. The composition of single particles is assigned based on the ion peaks found in the mass spectra. The presence of specific peaks can be used as markers for classification of distinct particle types. A combination of peaks may be used to assign the particle to a known source, assess the complexity of particle composition, or infer possible chemical transformations that may have occurred in or on the particle. In the Bakersfield data set, general particle types are identified. These types are carbonaceous, calcium-rich 2348

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 11, 2002

dust, and sodium containing. The chemical species found in these types agree with those previously described in source apportionment studies in the SJV (9, 10). Representative single particle mass spectra are shown in Figure 3 (a-d). Figure 3a shows a typical mass spectrum from a carbonaceous particle identified by the presence of peaks in the positive ion mass spectrum corresponding to C+ (m/z 12) or C3+ (m/z 36) or C2- (m/z 24) or C3- (m/z 36) in the negative ion spectrum. These peaks are chosen because they represent the predominant mass fragments from carbonaceous compounds in the particles. This particle mass spectrum also contains markers for chemical species related to the addition of secondary species. Significant ion intensities for ammonium and nitrate species are shown in the positive ion mass spectrum by the presence of NH4+ (m/z 18) and NO+ (m/z 30) and the negative ion mass spectrum by NO2(m/z 46) and NO3- (m/z 62) ions. NH4+ and NO+ ions have been shown to be appropriate markers for ammonium nitrate (37). The presence of ammonium nitrate in particles may be from the reaction of ammonia and nitric acid (38). Sulfate is also present in this organic particle as indicated by the HSO4- (m/z 97) ion in the negative ion spectrum. Sulfate may be incorporated by heterogeneous oxidation of gaseous SO2 as well as a result of combustion processes. The majority of organic particles sampled during the stagnation period contain markers for nitrate and sulfate. These associations will be discussed in more detail later. As shown here, the negative ion spectrum allows for the detection and identification of chemical species in the particle which do not yield a significant positive ion response. Negative ion spectra were acquired for 90% of the particles detected in Bakersfield. In contrast to the carbonaceous particle spectrum in Figure 3a, the mass spectra of a single particle containing an elemental carbon signature are shown in Figure 3b. Carbon chain fragments extending from C+ (m/z 12) to C12+ (m/z 144) in the positive ion spectrum and C- (m/z 12) to C13(m/z 156) are identified in the spectra shown. Elemental carbon particles are emitted from combustion sources. For

FIGURE 4. Compositionally resolved size distributions for January 14, 1999 and January 24, 1999 taken in Bakersfield, CA. Unscaled number counts for each particle type are shown as a function of the aerodynamic diameter size as determined by the ATOFMS. this paper, these particles are included in the carbonaceous particle classification. The mass spectra for a typical sodium type particle are shown in Figure 3c. These particles are identified by the presence of the sodium ion, Na+ (m/z 23). The mass spectra for the single particle shown also contains a large signal from K+ (m/z 39, 41), Na2O+ (m/z 62), and Na2Cl+ (m/z 81). In addition to these markers, the mass spectra contain species indicating heterogeneous chemistry has occurred on this particle. These reaction peaks include Na2NO3+ (m/z 108), Na3SO4+ (m/z 165), and Na(NO3)2- (m/z 147). The presence of these species indicates this particle may have undergone reactions with gas-phase nitric oxide species (39) and/or sulfuric acid. The mass spectra shown are typical for sodium containing particles sampled in the SJV, indicating extensive reactions occur on particles of this type during transport to Bakersfield. Sodium type particles likely result from the transport of marine aerosols or suspension of sodium-rich dust, such as those from salt playas. The mass spectra for a typical calcium-rich dust particle found in the SJV are shown in Figure 3d. This particle type was identified by peaks corresponding to Ca+ (m/z 40), CaO+ (m/z 56), Ca2O+ (m/z 96), or Al+ (m/z 27). This particle also contains a large signal corresponding to Na+. The positive ion spectrum indicates a dust particle containing soil minerals

and metals. The negative ion spectrum for the particle shown, however, contains a multitude of carbon mass fragments that are indicative of organic components in the particle. Like the carbonaceous and sodium type particles, peaks corresponding to nitrates, sulfates, and ammonium are used to identify secondary species in the particle. The single particle spectra provide an indication of the large degree of heterogeneity of particle composition over the course of this study. Compositionally Resolved Size Distribution. Unscaled particle counts for the general particle types are presented in Figure 4 as compositionally resolved size distributions for all classified particles sampled on January 14 and 24. The numbers of particles as a function of aerodynamic diameter are plotted for the three most abundant general particle types identified on January 14 and 24 in Bakersfield. Carbonaceous plus secondary inorganics, dust plus secondary inorganics, and the sodium type plus secondary inorganics refer to those particles containing marker peaks for the general particle type plus secondary aerosol species ammonium, nitrate, or sulfate. To keep the distributions on a scale where the differences are apparent, the counts for the carbonaceous particle plus secondary inorganics counts on January 14 and the sodium type particle plus secondary inorganics on January 24 were scaled down by a factor of 2. These VOL. 36, NO. 11, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2349

distributions are not corrected for the ATOFMS instrumental transmission efficiency. The relative differences in size distributions for the unique particle types are more evident when uncorrected because in corrected size distributions the fine mode (1 µm). The carbonaceous particles with secondary inorganics show a bimodal distribution with significant fine and coarse mode fractions. This distribution is noteworthy because typical ambient distributions show carbonaceous particles predominantly in the fine mode (1 µm) is more typical of those usually detected by ATOFMS, being composed primarily of inorganic species with little influence from carbonaceous particles. During the entire study, calcium-rich dust and sodium type particles are found predominately above 1 µm. These particles are likely generated by mechanical processes and therefore are expected to occur mostly in the coarse mode (42). A large change in the relative number counts of the three particle types can be seen on January 14 and January 24, indicating a major shift in ambient aerosol composition. This shift is representative of the overall change in composition that occurred between the two halves of the study. The size distributions also show distinct differences in the physical properties of the particles present during these time periods. During the stagnation period, the mass distribution had a peak around 0.5 µm. In contrast, the mass distribution during the haze event peaked at particle diameters greater than 1 µm. The difference in size distributions for particulate matter on January 14 and January 21 in Bakersfield is also evident in work presented by Chung et al. (43). They report that approximately 20% of the total mass resides in the 1.0-1.8 µm size bin on January 14, whereas 50% of the total mass exists in the 1.0-1.8 µm size bin on January 21. This contrast in the composition and shift in the size distribution could be indicative of the presence of an aerosol resulting from different formation and/or transformation processes. The single particle size and composition acquired over the course of this study is examined in an effort to understand the observed shift. Compositionally Resolved Number Concentration. The measured particle counts from the ATOFMS can be scaled to produce quantitative number concentrations upon comparison of ATOFMS data with data from conventional aerosol sizing and mass analysis instruments (31). In this study, aerosol mass distributions measured by MOUDI impactors beside the ATOFMS were used for generation of the scaling factors. The scaling function used for this study (R ) 3432; β ) -5.436) compares well with those determined by Allen et al. and falls within the error of the functions presented in that study. Scaling factors generated from the Bakersfield data set allow for generation of atmospherically representative number concentrations for the different particle types. The corrected number concentration and percentages of the number concentration in 12-h time bins for the particle types are shown in Figure 5. These plots are divided into those particles with diameters greater than 1 µm (1.0-2.5 µm) and less than 1 µm (0.32-1 µm) to show the distinct composition differences that occur for these two size modes. The particles in Figure 5 are divided into the major general particle types identified in Bakersfield. Carbonaceous, calcium-rich dust, and sodium type particles can be separated 2350

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 11, 2002

further into those particles containing marker peaks for inorganic secondary species including ammonium, nitrate, and sulfate. For purposes of clarity, these particles are divided into two subclasses. The first grouping contains those particles with mass spectra containing marker ion peaks for all three secondary species of interest: ammonium, nitrate, and sulfate (i.e. “Carbon/NH4/NO3/SO4” represents carbonaceous particles with ammonium, nitrate, and sulfate markers) (Figure 5). The second subclass is composed of those particles with mass spectra that contain any single marker or combination of ammonium, nitrate, or sulfate markers (i.e. “Carbon + sec. inorg.” represents carbonaceous particles with any combination of ammonium, nitrate, or sulfate markers) (Figure 5). This classification scheme was used in order to limit the complexity of the figure while still demonstrating the variability in particle composition. Particles with marker peaks for a combination of major particle types (i.e. carbonaceous and dust; or carbonaceous and sodium) are also distinguished. These particles are further subdivided based on the presence of marker peaks for the secondary inorganic species ammonium, nitrate, or sulfate. A final classification is made for those particles that do not fit the general particle type criteria but do contain markers for ammonium, nitrate, or sulfate. These particles are identified as secondary inorganic particles (“Sec. Inorganics”). Any particles that do not fit any of these types are identified as unclassified. For the entire study an average of 6% of particles are unclassified. It should be noted that this classification scheme is based on the presence or absence of marker peaks to establish a qualitative description of the particle composition. Since it does not use the intensities of the ion peaks, it is not intended to infer anything regarding the absolute concentrations of individual species within single particles. For the case in which there is some combination of secondary inorganic markers, the absence of a particular marker ion may be due to a lack of sensitivity or the presence of a peak below the peak detection threshold. This method of plotting the data does provide a quantitative representation of the number concentrations of the different particle types in the atmosphere. This represents the first report of data at this level of quantitative and temporal resolution from a single particle analysis instrument. Scaling the ATOFMS data into quantitative number concentrations allows for examination of the changes in particle concentrations and composition over time. The corrected number concentrations (Figure 5) display the dependence of aerosol concentration and composition on meteorology in the region. The percentage of the different particle types over the duration of the study is useful in examining the relative contributions from particle types with smaller numbers of counts that are not apparent in the scaled number concentrations. Data from the stagnation period show an elevated concentration of particles in both size ranges (greater and less than 1 µm diameter). The composition of the aerosol during this period is dominated by carbonaceous particles with secondary aerosol species. The abundance of ammonium, nitrate, and sulfate is indicative of the buildup of secondary aerosol. There is a significant change in particle composition beginning January 17 when the number and percentage of carbonaceous particles decrease. Of particular importance is the decrease in the relative percentage of carbonaceous particles in the greater than 1 µm diameter size bin. This corresponds with the transition from the stagnation (fog) event to the clear period. The particle concentration during the haze period approaches that of the stagnation episode. However, the particle composition during the haze event of January 21-23 does not show a large number of the carbonaceous type particles. Rather, those particles classified as the sodium type constitute

FIGURE 5. Corrected number concentration and percentage of total aerosol concentration for particle types separated into particles greater and less than 1 µm in 12-h time bins. Particles are divided into carbonaceous type, sodium type, and calcium-rich dust type. Types are subdivided into particles containing marker peaks for ammonium, nitrate, and sulfate (.../NH4/NO3/SO4) and particles containing any single marker peak or any combination of marker peaks for ammonium, nitrate, or sulfate (... + sec. inorg.). a large percentage of the total concentration. The change in composition from predominantly carbonaceous aerosol to a large sodium type is evident predominantly in those particles with diameters greater than 1 µm, but is also noticeable in particles less than 1 µm. Not only does the composition change dramatically, but the number concentration of these particles greater than 1 µm diameter increases as well. The increase in coarse particle concentration was also observed in the size distribution for the haze period. Examining the addition of marker peaks for secondary aerosol species shows that this change in composition is not to a pure sodium type particle but rather to transformed particles, containing markers for the secondary aerosol species ammonium, nitrate, and sulfate. Dust particles show a dependence on meteorological conditions as the percentage increases during storms and the passage of weather systems. The increase in percentage of dust particles on January 18 and 19 occurs during the clearing event where a small weather system passed through the valley. Increased winds during this period may suspend more dust in the atmosphere. It is evident that the meteorology of the SJV plays a large role in the elevation of particle concentration. With the addition of real-time compositional information, it becomes clear that different conditions may produce PM levels similar in magnitude but dramatically different in composition. Aerosol Mixing State. The complexity of the distribution of chemical species in the individual particles is evident in the corrected percent and number plots. The association of

species within individual particles provides a direct measure of the mixing state of the aerosol in Bakersfield. To examine this in detail, the color-coded dot plot, shown in Figure 6, is constructed for particles sampled on two separate days. January 14 and January 21 are chosen because they highlight the effect of changing meteorological conditions on particle composition that occurred over the course of the study. This plot presents the percentage of the total corrected number concentrations for the particle types and major chemical substances. Each size bin contains 100 dots with each dot representing 1% of the particles present in that size bin. A dot that is colored by only one color represents particles that contain marker peaks for one of the classification types or secondary species examined in this study: blue for sodium particles, gray for carbonaceous particles, orange for dust particles, red for nitrate containing particles, green for ammonium containing particles, and yellow for sulfate containing particles. Dots with stripes of different colors represent particles that meet the general search criteria for a number of different particle types or contain secondary chemical components. Each combination of particle components must be present at levels greater than 0.5% to be represented by a dot. Combinations of species that are not observed in greater than 0.5% of the total particles are represented by light blue dots and grouped together and labeled as “many types”. These classifications are based on the presence or absence of marker peaks in the mass spectra of individual particles and are not intended to be a measure VOL. 36, NO. 11, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2351

FIGURE 6. Size-segregated chemical compositions and associations for January 14 and January 21, 1999. Each dot represents 1% of the total particle number concentration within the indicated size range. of the mass fraction of the chemical substances within particles. Instead, Figure 6 is intended to provide an accurate representation of the percentage of the scaled number of particles containing important chemical components in the atmosphere. Using single particle data there are two ways to approach an analysis of the aerosol mixing state. The first is to examine the heterogeneity of individual particle composition as a whole. The general single particle types (i.e. dust, carbonaceous, sodium-type) identified in the Bakersfield data set are externally mixed as they arise from different sources and have unique compositions. These general particle types remain externally mixed over the course of the study. Another approach in analyzing the aerosol mixing state using single particle data involves examining the associations of specific chemical species within particles. A determination of the extent of mixing of the aerosol for two size modes, greater and less than 1 µm, is made here with respect to the major chemical components nitrate, ammonium, sulfate, and carbon. Figure 6 shows that there is a distinct difference in the associations between chemical species within particles for January 14 and January 21. In general, the number of 2352

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 11, 2002

unique combinations of particle types is much greater for January 21, especially for particles greater than 1 µm, compared to January 14 where most particles in all size bins look very similar. Carbon, nitrate, sulfate, and ammonium are internally mixed above and below 1 µm on January 14 during the stagnation period. In contrast, only carbon is internally mixed in particles below 1 µm, while nitrate and sulfate are internally mixed in particles greater than 1 µm on January 21 during the haze period. An interesting feature in Figure 6 is that despite an external mixture of carbon in particles greater than 1 µm during the haze period, nearly half of the sodium type particles contain some carbonaceous component. Also, the distribution of sulfate and nitrate during the haze period shows a strong association with the sodium type particles. This reflects the importance of secondary aerosol formation and addition of carbon, nitrate, and sulfate to sodium type and dust particles. Implications. These findings are very important for interpretation and modeling of the effects of fog and stagnation events on the ambient aerosol and for understanding the role of meteorology in SJV air pollution (12). Previous work has determined that the activation of particles into droplets, rapid growth during fog, and subsequent dissipation of fog can greatly alter the size distribution of the ambient aerosol (44-46, 19). The measurements presented here, showing an abundance of carbonaceous type particles in sizes greater than 1 µm, are direct evidence of the alteration of the size/composition distribution of particulate components during fog and stagnation events in the San Joaquin Valley. Elevated organic and elemental carbon concentrations in PM2.5 and in size-segregated analysis of particles greater than 1.0 µm have been measured at a number of sites in the SJV (9, 43). However, it was not explicitly known if the carbonaceous fraction was composed of pure organic and elemental carbon particles or if carbon was mixed with other particulate species or major particle types such as salt or dust. The results from this study clearly show an internal mixture of carbon during stagnation episodes and the association of an abundance of carbon with inorganic aerosol during other meteorological periods. The largest contributors to wintertime PM in the SJV, secondary ammonium, nitrate, and sulfate are found internally mixed during the stagnation episode. The aerosol during a haze event shows a dramatically different composition with an abundance of sodium associated with nitrate. In general, this period shows a greater diversity in the associations of chemical species within individual particles. Particles greater than 1 µm during this event are internally mixed with respect to nitrate and sulfate indicating extensive transformation of the aerosol through the addition of secondary components. These changes in particle chemical properties and their correlation with meteorological events must be accounted for in aerosol models aimed at understanding air pollution in the SJV. These results also demonstrate size-resolved quantitative particle composition information with high temporal resolution for an extended period of time. The ability to analyze particle chemical and physical properties at such a level will undoubtedly provide further insights into particle formation and evolution in the atmosphere.

Acknowledgments This work was supported by the California Air Resources Board (Contract # 95-307) and ENSR (Contract # 99064). The authors thank Mike Kleeman of UC Davis for MOUDI support. The authors also thank Phil Powers and the staff of the California Air Resources Board Air Quality Monitoring Laboratory in Bakersfield, CA, Dabrina Dutcher, and Charles E. McDade for their assistance and Don Liu for insight into single particle data analysis and extend thanks to anonymous reviewers for helpful comments.

Literature Cited (1) Ranzieri, A. J.; Thuillier, R. H. In Planning and Managing Regional Air Quality: Modeling and Measurement Studies; Solomon, P. A., Silver, T. A., Eds.; Lewis Publishers: Boca Raton, FL, 1994. (2) Watson, J. G.; DuBois, D. W.; DeMandel, R.; Kaduwela, A.; Magliano, K.; McDade, C.; Mueller, P. K.; Ranzieri, A.; Roth, P. M.; Tanrikulu, S. Aerometric Monitoring Program Plan for the California Regional PM10/PM2.5 Air Quality Study; Draft; Prepared for the California Regional PM10/PM2.5 Air Quality Study Technical Committee, c/o California Air Resources Board, Sacramento, CA,1998. (3) Dutcher, D.; Chung, A.; Kleeman, M.; Miller, A. E.; Perry, K. D.; Cahill, T.; Chang, D. Instrument Intercomparison Study Bakersfield, CA 1998-1999; Final Report; Prepared for California Air Resources Board, Monitoring and Laboratory Division, Sacramento, CA, 1999. (4) California Air Resources Board 1996. Estimated Annual Average Emissions - Kern County. California Air Resources Board, Technical Support Division, Sacramento, CA. (5) Smith, T. B.; Lehrman, D. E.; Reible, D. D.; Shair, F. H. The Origin and fate of airborne pollutants within the San Joaquin Valley; Prepared for California Air Resources Board, Sacramento, CA, 1981. (6) Niccum, E. M.; Lehrman, D. E.; Knuth, W. R. J. Appl. Meteorol. 1995, 34 (8), 1834-1847. (7) Blanchard, C. L.; Carr, E. L.; Collins, J. F.; Smith, T. B.; Lehrman, D. E.; Michaels, H. M. Atmos. Environ. 1999, 33, 4775-4786. (8) Chow, J. C.; Watson, J. G.; Lu, Z.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L. W. Atmos. Environ. 1992, 26A (18), 3335-3354. (9) Chow, J. C.; Watson, J. G.; Lu, Z.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L. W. Aerosol Sci. Technol. 1993, 18, 105-128. (10) Chow, J. C.; Watson, J. G.; Lu, Z.; Lowenthal, D. H.; Frazier, C. A.; Solomon, P. A.; Thuillier, R. H.; Magliano, K. Atmos. Environ. 1996, 30 (12), 2079-2112. (11) Magliano, K. L.; Hughes, V. M.; Chinkin, L. R.; Coe, D. L.; Haste, T. L.; Kumar, N.; Lurmann, F. W. Atmos. Environ. 1999, 33, 4757-4773. (12) Pun, B. K.; Seigneur, C. Atmos. Environ. 1999, 33, 4865-4875. (13) Gray, H. A.; Cass, G. R. Environ. Sci. Technol. 1986, 20(6), 580589. (14) Schauer, J. J.; Cass, G. R. Atmos. Environ. 2000, 33, 1821-1832. (15) Richards, L. W.; Alcorn, S. H.; McDade, C.; Couture, T.; Lowenthal, D.; Chow, J. C.; Watson, J. G. Atmos. Environ. 1999, 33, 4787-4795. (16) Smith, N.; Plane, J. M. C.; Nien, C.-F.; Solomon, P. Atmos. Environ. 1995, 29(21), 2887-2897. (17) Strader, R.; Lurmann, F.; Pandis, S. N. Atmos. Environ. 1999, 33, 4849-4863. (18) Collett, J. L., Jr.; Hoag, K. J.; Rao, X.; Pandis, S. N. Atmos. Environ. 1999, 33 (1999), 4833-4847. (19) Hoag, K. J.; Collett, J. L., Jr.; Pandis, S. N. Atmos. Environ. 1999, 33, 4817-4832. (20) Lillis, D.; Cruz, C. N.; Collett, J. L., Jr.; Richards, L. W.; Pandis, S. N. Atmos. Environ. 1999, 33, 4797-4816. (21) Moosmu ¨ ller, H.; Gillies, J. A.; Rogers, C. F.; DuBois, D. W.; Chow, J. C.; Watson, J. G.; Rangston, R. J. Air Waste Manage. Assoc. 1998, 48, 398-407. (22) Baron, P. A.; Heitbrink, W. A. In Aerosol Measurement: Principles, Techniques, and Applications; Willeke, K., Baron, P. A., Eds.; Van Nostrand Reinhold: New York, 1993. (23) Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Egami, R. T.; Solomon, P. A.; Thuillier, R. H.; Magliano, K.; Ranzieri, A. Atmos. Environ. 1998, 32 (16), 2835-2844.

(24) Johnston, M. V.; Wexler, A. S. Anal. Chem. 1995, 67, 721A726A. (25) Suess, D. T.; Prather, K. A. Chem. Rev. 1999, 99, 3007-3035. (26) Noble, C. A.; Prather, K. A. Mass Spectrom. Rev. 2000, 19, 248274. (27) Liu, D.-Y.; Rutherford, D.; Kinsey, M.; Prather, K. A. Anal. Chem. 1997, 69 (10), 1808-1814. (28) Murphy, D. M.; Thomson, D. S.; Middlebrook, A. M. Geophys. Res. Lett. 1997, 24(24), 3197-3200. (29) Heintzenberg, J. Tellus 1989, 41B, 149-160. (30) Hughes, L. S.; Allen, J. O.; Kleeman, M. J.; Johnson, R. J.; Cass, G. R.; Gross, D. S.; Gard, E. E.; Ga¨lli, M. E.; Morrical, B. D.; Fergenson, D. P.; Dienes, T.; Noble, C. A.; Liu, D.-Y.; Silva, P. J.; Prather, K. A. Environ. Sci. Technol. 1999, 33(20), 3506-3515. (31) Allen, J. O.; Fergenson, D. P.; Gard, E. E.; Hughes, L. S.; Morrical, B. D.; Kleeman, M. J.; Gross, D. S.; Ga¨lli, M. E.; Prather, K. A.; Cass, G. R. Environ. Sci. Technol. 2000, 34(1), 211-217. (32) Gard, E.; Mayer, J. E.; Morrical, B. D.; Dienes, T.; Fergenson, D. P.; Prather, K. A. Anal. Chem. 1997, 69(20), 4083-4091. (33) Liu, D.-Y. Ph.D. Dissertation, Department of Chemistry, University of California, Riverside, CA, 2000. (34) Munger, J. W.; Jacob, D. J.; Waldman, J. M.; Hoffman, M. R. J. Geophys. Res. 1983, 88(C9), 5109-5121. (35) Pandis, S. N.; Seinfeld, J. H.; Pilinis, C. J. Geophys. Res. 1990, 95(D11), 18489-18500. (36) Hinds, W. C. Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles; John Wiley & Sons: New York, 1982. (37) Liu, D.-Y.; Prather, K. A.; Hering, S. V. Aerosol Sci. Technol. 2000, 332000, 71-86. (38) Wexler, A. S.; Seinfeld, J. H. Atmos. Environ. 1990, 24A(5), 12311246. (39) Gard, E. E.; Kleeman, M. J.; Gross, D. S.; Hughes, L. S.; Allen, J. O.; Morrical, B. D.; Fergenson, D. P.; Dienes, T.; Ga¨lli, M. E.; Johnson, R. J.; Cass, G. R.; Prather, K. A. Science 1998, 279, 11841187. (40) Noble, C. A.; Prather, K. A. Environ. Sci. Technol. 1996, 30(9), 2667-2680. (41) Hughes, L. S.; Allen, J. O.; Bhave, P.; Kleeman, M. J.; Cass, G. R.; Liu, D.-Y.; Fergenson, D. P.; Morrical, B. D.; Prather, K. A. Environ. Sci. Technol. 2000, 34(15), 3058-3068. (42) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics; John Wiley & Sons: New York, 1998. (43) Chung, A.; Herner, J. D.; Kleeman, M. J. Environ. Sci. Technol. 2001, 35 (11), 2184-2190. (44) Fuzzi, S.; Facchini, M. C.; Orsi, G.; Lind, J. A.; Wobrock, W.; Kessel, M.; Maser, R.; Jaeschke, W.; Enderle, K. H.; Arens, B. G.; Berner, A.; Solly, I.; Kruisz, C.; Reischl, G.; Pahl, S.; Kaminski, U.; Winkler, P.; Ogren, J. A.; Noone, K. J.; Hallberg, A.; FierlingerOberlinninger, H.; Puxbaum, H.; Marzorati, A.; Hansson, H.-C.; Wiedensohler, A.; Svenningsson, I. B.; Martinsson, B. G.; Schell, D.; Georgii, H. W. Tellus 1992, 44B, 448-468. (45) Noone, K. J.; Ogren, J. A.; Hallberg, A.; Heintzenberg, J.; Stro¨m, J.; Hansson, H.-C.; Svenningsoon, B.; Wiedensohler, A.; Fuzzi, S.; Facchini, M. C.; Arends, B. G.; Berner, A. Tellus 1992, 44B, 489-504. (46) Ulevicˇius, V.; Trakumas, S.; Girgzˇdys, A. Atmos. Environ. 1994, 28(5), 795-800.

Received for review October 23, 2001. Revised manuscript received March 14, 2002. Accepted March 20, 2002. ES011381Z

VOL. 36, NO. 11, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2353