Journal of Geophysical Research: Atmospheres

Chemical and hygroscopic properties of aerosol organics at Storm Peak Laboratory

Authors


Abstract

[1] A combined field and laboratory study was conducted to improve our understanding of the chemical and hygroscopic properties of organic compounds in aerosols sampled in the background continental atmosphere. PM2.5 (particles with aerodynamic diameters smaller than 2.5 µm) aerosols were collected from 24 June to 28 July 2010 at Storm Peak Laboratory (SPL) in the Park Range of northwestern Colorado. New particle formation (NPF) was frequent at SPL during this campaign, and the samples were not influenced by regional dust storms. Filter samples were analyzed for organic carbon (OC) and elemental carbon (EC), water soluble OC (WSOC), major inorganic ions, and detailed organic speciation. WSOC was isolated from inorganic ions using solid phase absorbents. Hygroscopic growth factors (GFs) and cloud condensation nucleus (CCN) activity of the WSOC were measured in the laboratory. Organic compounds compose the majority (average of 64% with a standard deviation (SD) of 9%) of the mass of measured species and WSOC accounted for an average of 89% (with a SD of 21%) of OC mass. Daily samples were composited according to back trajectories. On average, organic acids, sugars, and sugar alcohols accounted for 12.5 ± 6.2% (average ± SD) of WSOC. Based on the composition of these compounds and that of high molecular weight compounds identified using ultra high resolution mass spectrometry, the organic mass to OC ratio of the WSOC is estimated to be 2.04. The average hygroscopic GFs at RH = 80% (GF80) were 1.10 ± 0.03 for particles derived from isolated WSOC and 1.27 ± 0.03 for particles derived from the total water-soluble material (WSM). CCN activity followed a similar pattern. The critical diameters at a super-saturation of 0.35% were 0.072 ± 0.009 and 0.094 ± 0.006 µm for particles derived from WSM and isolated WSOC, respectively. These GF results compare favorably with estimates from thermodynamic models, which explicitly relate the water activity (RH) to concentration for the total soluble material identified in this study.

1 Introduction

[2] Aerosols affect the Earth's radiation balance directly by scattering sunlight and indirectly through their role as cloud condensation nuclei (CCN) [Twomey, 1974; Twomey et al., 1984; Albrecht, 1989; Charlson et al., 1992; Chuang et al., 1997]. These effects are enhanced for hygroscopic aerosols, which absorb water as a function of relative humidity (RH), grow in size, scatter more light, and serve as cloud condensation nuclei [Charlson et al., 2001]. Absorption of water vapor by aerosols to produce haze and cloud droplets depends on their dry size and chemical composition. Large-scale (climate) models initially treated CCN as though the only soluble material they contained was sulfate [e.g., Chuang et al., 1997; Boucher and Lohmann, 1995]. However, organic material can be a significant fraction of aerosol mass at urban and remote locations [e.g., Chow et al., 1994; Malm et al., 1994; Andrews et al., 2000]. Thermodynamic equilibrium models have been developed to describe the hygroscopic properties of complex mixtures of inorganic salts in solution [Clegg et al., 1998; Zhang et al., 2000; Zaveri et al., 2005].

[3] There is considerable evidence that organic compounds impact aerosol hygroscopic growth and CCN activity. Novakov and Penner [1993] concluded that inorganic constituents could not account for observed cloud droplet number concentrations in the marine environment. Saxena et al. [1995] reported that organic compounds inhibited aerosol hygroscopic growth in urban Los Angeles by 25 to 35% but contributed significantly (25 to 40%) to hygroscopic growth at a remote location in the Grand Canyon. Speer et al. [2003] measured water uptake by PM2.5 (particles with aerodynamic diameters < 2.5 µm) on Teflon® filters from Research Triangle Park, North Carolina in a humidity-controlled chamber. Water associated with inorganic sulfate and nitrate estimated with the Aerosol Inorganics Model [AIM; Clegg et al., 1998; Wexler and Clegg, 2002] accounted for roughly 80% of measured water mass. The remainder was attributed to organic compounds. Dick et al. [2000] measured water uptake as a function of RH using a hygroscopic tandem differential mobility analyzer (HTDMA) and reported significant water uptake by organic compounds during the Southeastern Aerosol and Visibility Study.

[4] Several recent studies examined water-soluble organic carbon (WSOC) hygroscopicity by isolating this material from dissolved inorganic salts. Gysel et al. [2004] performed HTDMA experiments on aerosols nebulized from water extracts of ambient aerosol samples. Hygroscopic growth experiments were done on bulk soluble material and on WSOC isolated with solid phase absorbents. Gysel et al. [2004] found that 20 to 40% of the total particle liquid water was associated with WSOC. Dinar et al. [2006] separated humic-like substances (HULIS) from water extracts of ambient aerosol samples by absorbing them on XAD-8 (Supelco, Inc.) absorbent. They found that the activation diameter (81 nm) of HULIS particles derived from an integrated urban aerosol sample was the same as that of pure ammonium sulfate at a super-saturation (S) = 0.2%. The activation diameter of HULIS in fresh wood smoke was 134 nm at the same super-saturation. Dinar et al. [2007] measured hygroscopic growth factors (GF = diameter of hydrated particles divided by dry particle diameter) with HTDMA for aerosols derived from the HULIS extracts studied by Dinar et al. [2006]. The GF at 80% RH was about 1.3 for urban HULIS. For comparison, the GF for pure ammonium sulfate is about 1.5 at 80% RH [Tang and Munkelwitz, 1994]. In these studies, the solid phase absorbents (e.g., XAD-8) retained high molecular weight WSOC while passing inorganic ions and lower molecular weight WSOC. The latter may also contribute significantly to organic water activity. In general, there is significant variation in reported hygroscopicities of HULIS as reviewed by Kristensen et al. [2012].

[5] Particulate OC is composed of numerous compounds with a range of chemical and thermodynamic properties [Seinfeld and Pankow, 2003]. However, only a minor fraction of this material has been specifically identified [Sempéré and Kawamura, 1994; Saxena and Hildemann, 1996; Turpin et al., 2000; Löflund et al., 2002; Mader et al., 2004]. The principal problem with incorporating organic compounds into thermodynamic models is that the chemical and physical properties of most of this material are unknown. The Hygroscopic Properties of Aerosol Organics (HPAO) study at Storm Peak Laboratory (SPL) was conducted to characterize the chemical and hygroscopic properties of aerosol organics in the remote continental atmosphere with laboratory and modeling studies. The objectives of the HPAO study are to: (1) determine the chemical structure and physical properties of atmospheric WSOC; (2) measure the hygroscopic behavior of isolated WSOC under sub- and super-saturated conditions; and (3) estimate the contributions of the identified organic compounds to water activity using the E-AIM model. This paper provides an overview of the study and presents results pertaining to the physical and chemical properties of the aerosol observed during the field project. Papers in preparation provide greater detail on the organic chemistry, hygroscopicity, and modeling components. Additionally, a parallel study of hygroscopic growth and CCN activity of HULIS from filters collected immediately after this campaign is presented in Kristensen et al. [2012].

2 Methods

[6] The approach for the field and laboratory studies is similar to that used by Lowenthal et al. [2009]. A field study was conducted at the Desert Research Institute's (DRI) SPL (3210 m ASL), a remote continental site located on the west summit of Mt. Werner in the Park Range near Steamboat Springs in northwestern Colorado. This site has been used in cloud and aerosol studies for more than 25 years [e.g., Lowenthal et al., 2002; Borys and Wetzel, 1997; Hallar et al., 2011a]. SPL is situated at tree line on a 70 km ridge oriented perpendicular to the prevailing westerly winds. SPL experiences transport from distant sources including urban areas, power plants, and wildfires [Obrist et al., 2008]. Additionally, there are abundant monoterpene emissions from pine trees at SPL [Hardik et al., 2012] during summer months.

[7] The aerosol sampling strategy was designed to collect milligram amounts of organic material for detailed chemical analysis and hygroscopic growth measurements. Aerosol sampling was conducted daily from 24 June to 28 July 2010. Four Tisch Environmental, Inc. high-volume (hivol) samplers collected PM2.5 on precleaned 8” × 10” Teflon-impregnated glass fiber (TIGF) filters (Fiberfilm T60A20, Pall Life Sciences) at a flow rate of approximately 1 m3/min. Volumetric flow controllers for the hivols were calibrated at the factory before the project and calibrated independently in the field against a Roots® (Dresser, Waukesha, WI) meter.

[8] Filters were changed daily at approximately 1000 Mountain Daylight Time (MDT). The filters were stored under refrigeration after sampling. A single-channel medium-volume (medvol) sampler preceded by a Bendix 240 cyclone inlet collected particles on prefired (900ºC for 4 h) 47 mm quartz-fiber (2500 Pallflex QAT-UP) filters in Nuclepore filter packs at a flow rate of approximately 109 L/min. At this flow rate, the 50% cut point of the cyclone was 3.4 µm. These samples were used for analysis of bulk elemental and organic carbon (EC and OC) and water-soluble OC (WSOC). A back-up quartz filter was used to account for potential positive sampling artifacts caused by absorption of volatile organic compounds by the front quartz filter [Turpin et al., 1994]. The medvol was operated concurrently with the hivols and samples were stored under refrigeration before and frozen after sampling.

[9] Size distributions of particles with mobility diameters from 8.8 to 334 nm were measured at 5-min intervals with a TSI, Inc. (Shoreview, MN) Scanning Mobility Particle Sizer (SMPS) (model 3936) coupled with a TSI (model 3022) condensation particle counter [CPC] [Wang and Flagan, 1990]. The sample and sheath flows were 1.0 and 10 L/min, respectively. Corrections for multiple charge and diffusion losses were applied in the TSI Aerosol Instrument Manager® software. Stand-alone TSI model 3025 and 3010 CPC's measured aerosol concentration for particles with diameters larger than 3 and 10 nm, respectively. A TSI Aerosol Particle Sizer (APS) (Model 3321) measured aerodynamic particle sizes from 0.54 to 20 µm [Chen et al., 1985]. Ambient temperature, pressure, wind speed and direction, and RH were measured continuously with Campbell Scientific, Inc., Met One and Vaisala sensors. Ozone (O3) was measured with a Dasibi Ozone Monitor (Dasibi Environmental Corp., Glendale, CA). The meteorological and ozone data were uploaded to the Western Regional Climate Center (http://www.wrcc.dri.edu/) database every 5 min.

[10] After initial WSOC and inorganic ion analysis, the hivol water extracts were composited to provide sufficient WSOC mass for subsequent analyses. Daily extracts were combined based on analysis of NOAA Hybrid Single-Particle Lagrangian integrated trajectories [Draxler and Rolph, 2003; Rolph, 2003] to determine the origin of air masses arriving at SPL. Seventy-two hour back trajectories were calculated for each day of the field campaign starting at 0000, 0800, and 1600 MDT. Back trajectories were calculated in ensemble forms, which calculate 27 trajectories from a selected starting point. Each member of the trajectory ensemble is calculated by offsetting meteorological data by one meteorological grid point (1º) in the horizontal (both latitudinal and longitudinal) and 0.01 sigma units (250 m) in the vertical for the selected starting point. The National Centers for Environmental Prediction's Eta Data Assimilation System (EDAS), covering the United States (http://www.arl.noaa.gov/edas40.php) was used for this analysis. The computational height was selected based on the height which most closely represented the common pressure at the lab, approximately 675 mb. Wind speed and direction at SPL were used in combination with the back trajectories as verification.

[11] Daily filter samples were combined into six composites based on similarities in their source regions revealed by the back trajectories. Three of the composite samples represent daily periods impacted by a coherent pattern and specific source region. Three samples represent a range of conditions. The groupings are described below:

  1. [12] Sample 1 - 6/24 to 6/28: This sample is a fairly coherent group with air arriving from the southwest with an origin in Southern California 72 h prior. As time progressed, the air arrived from the west and ultimately the northwest, and there are increasing numbers of back trajectories originating from Oregon. This sample includes a total of five sample days.

  2. [13] Sample 2 - 6/29 to 7/5: This sample represents a range of conditions and includes six sample days. The sample on 4 July was eliminated due to potential contamination from aerial fireworks, which were launched from high elevation locations above Steamboat Springs, Colorado.

  3. [14] Sample 3 - 7/6 to 7/12: This sample also represents a range of wind directions and includes seven sample days.

  4. [15] Sample 4 - 7/13 to 7/20: This sample was a coherent group. Air arrived at SPL from the west with origins in California and Nevada 72 h previously. This sample includes eight sample days.

  5. [16] Sample 5 - 7/21 to 7/26: This sample represents a range of conditions. Air arrived at SPL generally from the west, varying from the southwest to northwest quadrants. This sample represents seven sample days.

  6. [17] Sample 6 - 7/27 and 7/28: This sample was a coherent group. Air arrived at SPL in clockwise flow. The wind direction shifted from the south to the southwest during this period. The 72-h source region was predominately from northern New Mexico and northwestern Arizona. This sample represents two sample days.

[18] The organic chemical analysis protocols are described in detail by (Samburova, V., et al., Composition of the water-soluble organic fraction in atmospheric remote aerosols, submitted to Environmental Chemistry, April 2013.) Samburova et al. [2013] and Mazzoleni et al. [2012]. A brief outline is presented here. Prior to the extraction in water, several punches (25 mm diameter) were taken from each TIGF filter and stored in a laboratory freezer. The four daily TIGF filter samples were combined, cut into small pieces (approximately 2.5 cm2), extracted twice with 70 mL high purity water (18 mega ohms) (NanoPure, Barnstead, USA) using an ultrasonic bath for 30 min at room temperature, and then filtered through 0.2 µm PTFE membrane filters (Whatman, Inc.). After removing aliquots for analysis of the daily samples, the extracts were combined into six composites and one blank based on the back trajectory data (described above) and preconcentrated to 55–75 mL final volume with a freeze drying technique. The water-soluble aerosol extracts were placed in precleaned amber bottles (250 mL) and stored at −20°C until analysis. The daily and composite extracts were analyzed by ion chromatography (IC) on Dionex ICS 3000 (Thermo Fisher Scientific, Inc., Waltham, MA) analyzers for sulfate, nitrate, and chloride with a Dionex AS14 column and conductivity detector. Sodium (Na+), potassium (K+), calcium (Ca2+), and magnesium (Mg2+) were analyzed in the composites only by IC on a Dionex CS16 column. Organic acids (formic, glutaric, lactic, maleic, malonic, oxalic, and succinic) were analyzed by IC with Dionex IonPac AG11HC guard and AS11 HC columns. Ammonium (NH4+) were measured by automated colorimetry with an Astoria 301A analyzer. Organic and elemental carbon (EC) on the medvol front and backup quartz filters were measured by thermal-optical reflectance analysis [Chow et al., 1993]. WSOC in extracts from the medvol and hivol samples was determined with a Shimadzu total organic carbon analyzer (Model TOC-VCSH).

[19] WSOC in the composite samples was isolated from inorganic ions using the procedure of Duarte and Duarte [2005]. Two nonionic macro-porous resins, XAD-8 and XAD-4, were applied sequentially. The aqueous extracts were acidified to pH = 2.2 with HCl before being applied onto the XAD-8 column. Effluent from the XAD-8 column was collected and applied to the XAD-4 column. After these two concentration stages, both columns were washed with one column volume of ultra-pure water to remove inorganic species. The organic matter retained on both resins was then back-eluted with a mixture of methanol/water (40% methanol). The XAD-8 column was eluted again with 50% methanol in water to improve recovery of the retained organic matter. The methanol/water eluates containing isolated WSOC were combined and freeze-dried in order to remove methanol and preconcentrate the sample analytes [Lowenthal et al., 2009] (V. Samburova et al., 2013, submitted). Both nontreated and XAD-treated samples were analyzed by gas chromatography/mass spectrometry (MS) for polar organic compounds including sugars, sugar alcohols, sugar anhydrates, lignin derivatives, and organic acids [Mazzoleni et al., 2007] (V. Samburova, 2013, submitted).

[20] WSOC compounds were characterized by ultra-high resolution Fourier transform-ion cyclotron resonance MS (FT-ICR MS) [Marshall and Hendrickson, 2008; Mazzoleni et al., 2012]. Electrospray ionization (ESI) coupled with FT-ICR MS provides detailed molecular characterization of complex organic samples due to its extremely high resolving power (m/Δm > 400,000) and mass accuracy [Marshall et al., 1998; Kujawinski et al. 2002; Kim et al., 2003; Reemtsma et al., 2006; Sleighter and Hatcher, 2007; Reinhardt et al., 2007; Altieri et al., 2008; Wozniak et al., 2008; Mazzoleni et al., 2010]. ESI is a “soft” ionization technique that offers minimal fragmentation of the analytes, thus allowing for mass analysis of intact molecules [Stenson et al., 2003]. The ultrahigh resolving power of the FT-ICR mass spectrometer provides an accurate mass measurement which can be used to determine the chemical formulas of the mass-to-charge ratios (m/z) [Kim et al., 2006]. FT-ICR MS analysis was performed using the Thermo Scientific LTQ FT Ultra mass spectrometer at the Woods Hole Oceanographic Institution. The analysis and results are described in detail by Mazzoleni et al. [2012].

[21] Aliquots of water-extracts containing all water-soluble material (WSM) (total WSM, including organic and inorganic species) and isolated WSOC were subjected to hygroscopic growth measurements to determine the contribution of WSOC to water uptake as a function of RH. The procedure is described in Lowenthal et al. [2009], and detailed results are presented by Taylor et al. [2013]. A TSI model 3076 constant-output atomizer was used to generate particles from the WSM and WSOC extracts. Approximately 30 mL of extract was required to sequentially characterize the upper and lower legs of any hygroscopic growth hysteresis loops between 25 and 90% RH, as described in Santarpia et al. [2004]. The design of the HTDMA is described by Gasparini et al. [2004]. The particles were dried with a Nafion® dryer and introduced into the HTDMA where the first DMA selected particles with a dry diameter of 70 nm. Each measurement consisted of two scans: an “efflorescence scan” and “deliquescence scan.” These are intended to account for hysteresis in hygroscopic growth. The efflorescence scan maps the size response to varying RH of an initially hydrated (i.e., exposed to 90% RH) aerosol. Thus, it can detect efflorescence (crystallization) at low RH. The deliquescence scans map the response of an initially dry (~15% RH) aerosol and can capture deliquescence. The CCN activity of WSM and WSOC particles was measured using an SMPS paired with a Droplet Measurement Technologies, Inc. (Boulder, CO) CCN-100 spectrometer [Roberts and Nenes, 2005]. The RH and supersaturation retrievals from the HTDMA and DMA-CCN systems were periodically calibrated using their response to a well-characterized ammonium sulfate aerosol. The theoretical ammonium sulfate response for each instrument was calculated using parameterizations of the density, surface tension, and activity of an ammonium sulfate solution provided in work of Tang [1996, 1994, 1997]. In the case of CCN activation, an ideal solution was assumed. The sheath flows and column voltage in the DMA columns of both instruments were also routinely calibrated.

[22] A flow diagram of the filter analysis is provided in Figure 1 to provide details of the analysis performed and the measured outputs. The intent of this diagram is to relate each part of the analysis to the whole project.

Figure 1.

Flow diagram illustrating the analyses performed on filter samples collected during the HPAO study at Storm Peak Laboratory.

3 Results and Discussion

3.1 Particle Concentrations and Size Distributions

[23] The average and standard deviation (SD) of condensation nucleus (CN) concentrations are shown for each of the six composite sample periods in Table 1. Data are presented for the CPC 3010 and CPC 3025, which measure concentrations of particles with diameters larger than 10 and 3 nm, respectively. The difference between the two is the average nanoparticle (3 – 10 nm) concentration. Sample four, from mid-July, had the highest total and nanoparticle concentrations. Sample six, from the last week of July, had the lowest concentrations.

Table 1. Average and the Standard Deviation of Condensation Nucleus (CN) Concentrations (cm−3) During the Six Composite Sample Periods
InstrumentSample 1Sample 2Sample 3Sample 4Sample 5Sample 6
CPC 3010 (> 10 nm)2904 ± 23232543 ± 31082446 ± 33282905 ± 35732164 ± 19991124 ± 660
CPC 3025 (> 3 nm)3920 ± 30633476 ± 42613116 ± 45304226 ± 52533167 ± 27131729 ± 706
3 – 10 nm101693367013211003605

[24] NPF events are common at SPL [Hallar et al., 2011a]. Between 2001 and 2009, NPF occurred on 248 of 474 (52% of the time) days with valid SMPS data at SPL. Figure 2 shows particle size distributions from 8.8 to 334 nm measured with the SMPS. Filters were not considered for 4 July, due to the fireworks event and probable contamination. Thus, the period from 4 July 1000 MST to 5 July 1000 MST is not plotted. The SMPS was not operating between 6 July 2100 and 7 July 0900 MST. The three-dimensional size distributions shown in Figure 2 were examined for NPF using the methodology of Venzac et al. [2008] and Dal Maso et al. [2005]. An NPF event was defined as the appearance of a clear new mode in the ultrafine size range for a significant period of time (hours) followed by the growth of new particles to larger sizes.

Figure 2.

SMPS size distributions during the HPAO field study.

[25] Figure 2 shows that NPF was frequent during the HPAO study, occurring on 67% of the sample days. For composite sample 3, NPF occurred on each of the seven sample days. For sample composite 6, NPF did not occur on either of the two sample days. Figure 3 presents the average SMPS and APS size distributions for each of the six sample periods described above. Figure 3 shows that NPF significantly influenced the concentration of particles smaller than about 30 nm (ultrafine). Composite sample 6 had distinctly lower ultrafine mode concentrations compared with the other sample periods.

Figure 3.

Combined SMPS and APS size distributions for each of the six composite samples.

[26] SPL aerosol loading is occasionally influenced by regional dust storms [Hallar et al., 2011b]. These events are commonly seen in the coarse aerosol mode measured by the APS. The APS coarse mode size distribution (Figure 3) is similar for all six composite samples. Additionally, these distributions closely resemble previously demonstrated background conditions at SPL. Concentrations during regional dust storms are significantly higher [Hallar et al., 2011b], and thus it can be assumed that regional dust storms did not impact the filters collected during this study.

3.2 Aerosol Chemistry

[27] Figure 4 presents a daily time series of the major species concentrations and the percentage of the sum of major species during HPAO. Organic and EC were based on the medvol samples, with backup filter OC subtracted from front filter OC to correct for the absorption artifact on the front quartz filter [Turpin et al., 1994; Subramanian et al., 2004; Chow et al., 2006]. The backup filter thus serves as the dynamic blank for OC and WSOC on the front filter. The positive artifact for OC was 28% with a SD of 10% in [uncorrected] OC on the front filter. There was also a significant positive artifact for WSOC. The concentration of WSOC on the backup quartz filter was 27 ± 6%(SD) of [uncorrected] WSOC on the front filter. Ion concentrations were taken from the hivol samples. The medvol sampler malfunctioned on 7/8/10 and data for this day are not plotted. Concentrations of the sum of the major species and composition were relatively invariant over the field study. If it is assumed that the organic aerosol mass (OM) is 1.8OC [Pitchford et al., 2007], the average of the sum of the major species concentrations (OM + EC + SO42− + NO3- + NH4++other [defined in Figure 4]) was 2.4 ± 0.5(SD) µg/m3. The sum of species was dominated by organic mass (OM) (62 ± 9[SD]%), followed by sulfate (21 ± 7[SD]%), EC (5.4 ± 1.0[SD]%), nitrate (4.8 ± 1.8[SD]%), ammonium (3.9 ± 1.3[SD]%), and other (3.2 ± 1.1[SD]%).

Figure 4.

Time series of chemical composition. OM [1.8OC, from Pitchford et al., 2007] and EC from medium-volume sampler, sulfate (SO42−), nitrate (NO3-), chloride (Cl-), ammonium (NH4+), and other (Ca2+, Mg2+, K+) from the high-volume sampler. Ca2+, Mg2+, and K+ concentrations were measured in the six composite extracts only, and these concentrations are shown for each respective daily period. Na+ concentrations were contaminated by the glassware for individual samples and water extracts.

[28] WSOC accounted for 89 ± 21(SD)% of OC in the medvol samples. This level of variation is relatively small and random, i.e., there was no trend with respect to time. This probably reflects consistently aged air masses arriving at this site from the west. Zappoli et al. [1999] found that 77% of OC was water soluble at a background site in Sweden. Kiss et al. [2002] reported that WSOC comprised 71% of OC at K-puszta, a rural site in Hungary, from January to September, 2000. Miyazaki et al. [2009] reported an average WSOC/OC ratio of 0.60 ± 0.17 at a rural site in southern China. Park et al. [2012] reported an average WSOC/OC ratio of 0.56 with a range of 0.42–0.72 in urban Seoul, Korea. WSOC/OC ratios ranging from 0.31 to 0.64 were reported for St. Louis, MO [Sullivan et al., 2004]. Sullivan and Weber [2006] reported average summer and winter ratios of WSOC to OC of 0.47 and 0.42, respectively in Atlanta, GA. Miyazaki et al. [2006] reported WSOC ratios of 0.20 and 0.35 for winter and summer/fall, respectively in Tokyo. It is clear that WSOC ratios are higher in remote than urban areas and that this is likely due to oxidation of organic aerosols and production of secondary organic compounds during transport. The WSOC/OC ratios measured during the HPAO study at SPL are among the highest reported for remote areas. It should be noted that the volatile OC absorption artifact was accounted for in this study for both WSOC and OC. Examination of previously used sampling strategies indicates that this has not always been the case, e.g., Zappoli et al. [1999], Kiss et al. [2002], Miyazaki et al. [2009].

[29] Figure 4 demonstrates that sulfate was the major inorganic constituent of the aerosol during HPAO. The hygroscopic properties of the inorganic fraction of the aerosol depend on sulfate speciation. SPL is located directly west of three coal-fired power plants, at distances of approximately 50, 80, and 250 km. As demonstrated in Hallar et al. [2011a] an increased probability of NPF is associated with westerly winds. Previous measurements of SO2 at SPL show a diurnal cycle varying from a nighttime value of 0.16 ppb to a daytime average value of 0.28 ppb from 31 March to 15 June 2004. Sulfuric acid (H2SO4) is formed by oxidation of SO2 emitted mainly by fossil fuel combustion. Sulfuric acid can be partially neutralized by ammonia (NH3) to form ammonium bisulfate (NH4HSO4) or fully neutralized to form ammonium sulfate ((NH4)2SO4). The degree of sulfate neutralization determines its hydration state and the amount of water it absorbs as a function of RH [Taylor et al., 2011]. Dry ammonium sulfate will not begin to absorb water until the RH is raised to the deliquescence point (~80%). However aqueous ammonium sulfate droplets subjected to a decreasing RH will not necessarily become dry particles at 80% but will continue to become more concentrated (super-saturated) as water is lost until efflorescence occurs at ~35% RH [Martin et al., 2004]. Conversely, pure sulfuric acid grows smoothly with increasing or decreasing RH.

3.3 Organic Characterization

[30] Mazzoleni et al. [2012] reported the results of ultrahigh-resolution FT-ICR MS analysis of the isolated WSOC from composite sample 4, described above. Molecular formulas were identified for approximately 4000 compounds in the mass range of 100 to 800 Da. A majority of these compounds were aliphatic and olefinic in nature. A significant number of compounds were similar to previously determined compounds associated with secondary formation from monoterpene and sesquiterpene precursors. The relative abundance weighted molecular formulas identified by this method were used to determine an average OM to OC (OM/OC) ratio of 1.87 for the isolated WSOC [Mazzoleni et al., 2012]. The OM/OC ratio was derived from the assigned molecular formulas where OM was the measured mass (C and other elements such as O, N, and S), and OC was calculated from the number of C atoms in a molecular formula. The OM/OC ratio is a fundamental property of organic aerosol because OC is commonly measured or estimated from emission inventories and identified individual compounds account for a relatively small or undetermined fraction of OM.

[31] The OM/OC (f) ratio has been used to estimate the contribution of OM to particle mass [Andrews et al., 2000] and aerosol light extinction [Malm et al., 1994]. A range of values of f has been reported in the literature. OM/OC ratios from 13 studies, including the current study, are presented in Table 2 for urban and remote locations. Grosjean and Friedlander [1975] estimated f = 1.4 based on the elemental composition organic compounds in water and organic solvent extracts of aerosols collected in Los Angeles, CA. The values in Table 2 range from 1.4 (Los Angeles) to 2.1 (Hong Kong) for urban sites and from 1.4 (western Pacific) to 2.1 (generic rural) at rural sites. Most are based on direct measurements. However, Turpin and Lim [2001] estimated f based on organic compounds expected to be found at urban and remote locations. Pitchford et al. [2007] adopted a value of 1.8 for remote U.S. National Parks based on a literature “consensus”. Except for the value of 1.4 determined by Russell [2003] for the western Pacific and Caribbean Oceans, all of the remote values are 1.8 or greater. While there are some lower values for the urban sites, values of 1.8 and 1.91 were reported for Pittsburgh, PA by Zhang et al. [2005] and Polidori et al. [2008], respectively. The value of 2.0 reported here for the HPAO study is the highest directly measured value for remote sites in Table 2. It is consistent with the expectation that organics in aged aerosols at remote sites should contain more oxygen.

Table 2. OM/OC Ratio Determined in Various Studies at Urban and Remote Locations
   OM/OC 
  1. a

    PM (Particulate matter); PMx (PM with diameter smaller than x); WSOC.

  2. b

    Fourier Transform Infrared Analysis.

  3. c

    Aerodyne Aerosol Mass Spectrometer.

  4. d

    Fourier Transform Ion Cyclotron Resonance Mass Spectrometry.

StudyPhaseaMethodUrbanRemoteLocation

Grosjean and Friedlander [1975]

PMElemental analysis of water and organic solvent extracts1.4 Los Angeles, CA

Krivácsy et al. [2001]

WSOCElemental analysis 1.9Jungfraujoch, Switzerland

Turpin and Lim [2001]

PMComposition of characteristic compounds1.62.1Various

Kiss et al. [2002]

WSOCWSOC by weighing after separation; OC by elemental analysis 1.93K-puszta, Hungary

Russell [2003]

PMFTIRb 1.4Western Pacific and Caribbean Oceans

El-Zanan et al. [2005]

PM2.5OC by thermal-optical analysis; OM by weight of acetone and dichloromethane extract residues 1.93U.S. National Parks

Zhang et al. [2005]

PM1Elemental analysis by AMSc1.8 Pittsburgh, PA

Chen and Yu [2007]

PM2.5OC by thermal- optical analysis; OM by difference in weight after heating2.1 Hong Kong

Pitchford et al. [2007]

PM2.5Literature consensus 1.8U.S. National Parks

Aiken et al. [2008]

PM1Elemental analysis by AMS1.71 Mexico City

Cozic et al. [2008]

PM1Elemental analysis by AMS 1.84Jungfraujoch, Switzerland

Polidori et al. [2008]

PM2.5Organic extract residues analyzed for OC by thermal-optical analysis and for OM by weighing1.91 Pittsburgh, PA
This studyWSOCElemental composition of identified compounds and molecular structures determined by FT ICR-MSd 2.0Storm Peak Laboratory, CO

[32] Samburova et al. (2013 submitted) report the concentrations of 47 individual water-soluble organic species in the six composite WSOC samples. Twenty-nine organic acids, 11 sugars and sugar anhydrides, and 7 sugar alcohols were identified, and their percentages of WSOC were found to be 9.8 ± 3.4(SD)%, 4.9 ± 5.1(SD)%, and 1.1 ± 0.5(SD)%, respectively. Thus on average, organic acids, sugars, and sugar alcohols accounted for 12.5 ± 6.2(SD)% of WSOC. The most abundant compounds with respect to WSOC were glucose (1.3%), sucrose (1.2%), oxalic acid (1.2%), and malonic acid (1.0%). The weighted average OM/OC ratio of the identified individual organic compounds in composite sample 4 was 2.86 (Samburova et al., 2013, submitted). Nearly 80% of the mass of these species was not retained by the XAD separation procedure and is thus not represented by the compounds identified by the FT-ICR MS analysis presented in Mazzoleni et al. [2012].

[33] Samburova et al. (2013, submitted) estimated that these identified individual water-soluble organic compounds (again with an OM/OC ratio of 2.86) accounted for 17% of the WSOM (WSOC plus elements including O, N, and S) in composite sample 4, before XAD separation. If it is assumed that the remainder (83%) was characterized by an OM/OC ratio of 1.87 (see above), then the overall OM/OC ratio of the WSOC in composite sample 4 was 2.04. It should be noted that the chemical and hygroscopic properties of HULIS (isolated with procedures such as the one used by Varga et al. [2001], Kiss et al. [2002], Gysel et al. [2004], Stone et al. [2009], Lin et al. [2010] and in this study) do not fully represent the contributions of the low molecular weight polar compounds. Thus, losses of the low molecular weight polar compounds via XAD separation can account for a significant fraction of the WSOC mass, while significantly altering the OM/OC ratio of the sample.

[34] These losses due to XAD extraction are further highlighted in Table 3, which shows the WSOC/OC ratio, percent of WSOC recovery after XAD treatment, and the percent of WSOC for identified compounds excluding the amount of those compounds in the extract after XAD treatment. These numbers are given on a carbon mass basis, and it should be noted that this estimate may not account for all compounds, as it is possible that compounds could be irreversibly retained on the XAD column.

Table 3. WSOC/OC, Recovery After XAD Treatment, and Recovery Including Identified Compounds Calculated From Concentrations on a Carbon Mass Basis
CompositeWSOC/OCaWSOC Recovery (%)bIdentified %c of WSOCTotal WSOC Recovered and Identified (% of WSOC)d
  1. a

    Ratio of water-soluble organic carbon to total organic carbon.

  2. b

    The percent of initial WSOC recovered after XAD treatment.

  3. c

    Percent of initial WSOC identified as organic acids, sugars, and sugar alcohols excluding those identified in the XAD treated extract.

  4. d

    Sum of columns 3 and 4.

10.825518.373
20.85874.491
30.90819.590
40.849210.9103
50.909312.5106
60.87524.657

3.4 Aerosol Hygroscopicity

[35] Taylor et al. [2013] present detailed results on hygroscopic growth and CCN activity experiments on aerosols derived from WSM and WSOC extracts from the composite samples. An HTDMA was used to measure hygroscopic growth at RH from ~20 to 90% in two separate scans: (1) efflorescence scans to map the upper branch of hysteresis loops, detecting efflorescent phase transitions; and (2) deliquescence scans to map the lower branch of hysteresis loops, detecting deliquescent phase transitions. These differed in the inter-DMA RH profile. Whereas deliquescence scans simply hydrate the initially dry aerosol selected by the first DMA column to the range of controlled RH, efflorescence scans first re-hydrate the aerosol to ~90% RH, forcing them to deliquesce, before reaching a final controlled RH.

[36] Figure 5 presents average GFs at 25, 30, 40, 50, 60, 70, 80, and 90% RH for efflorescence (crystallization) and deliquescence scans of all WSM samples and WSOC composite samples 1, 2, 3, 4, and 6. The error bars in Figure 5 represent the SDs of the individual sample values. GFs for the WSOC sample 5 were anomalously high (1.4) and were assumed to represent contamination, since the flask was cracked during the concentration procedure. This sample is not included in Figure 5. The dry mobility diameter for all HTDMA measurements, D0, was 0.070 µm.

Figure 5.

Average hygroscopic growth factors (DRH/D0) for 70 nm aerosols generated from five of the six isolated WSOC extracts. Scans with increasing RH are labeled D for deliquescence branch and E for efflorescence branch. Error bars indicate the standard deviation between the measurements of samples included in each average.

[37] The average WSOC GF80 was 1.10 ± 0.03(SD). The average GF80 for WSM was 1.27 ± 0.03(SD). Lowenthal et al. [2009] reported a WSOC GF80 of 1.13 ± 0.03(SD) at Great Smoky Mountains National Park in Tennessee using similar methodology. The average GF90 in this study was 1.17 ± 0.04(SD) for the efflorescence scan. Gysel et al. [2004] reported GF90 ranging from 1.08 to 1.17 for HULIS isolated from aerosol samples collected at K-puszta, a rural site in Hungary. Figure 5 shows that while the WSOC absorbed a significant amount of water, it was considerably less hygroscopic than the total WSM.

[38] Figure 6 shows the results of CCN measurements on aerosols derived from the WSM and WSOC extracts. The CCN activity measurements utilized an SMPS to select initial dry particle sizes. Total downstream particle concentrations (CN) were measured with a TSI 3762 CPC. As the size selected by the SMPS increases, the ratio of CCN to CN concentrations sharply transitions from zero to unity at a size corresponding to the critical activation diameter associated with the super-saturation (SS) in the CCN counter. The D50 is defined as the diameter at which 50% of the particles activated at a specific SS. CCN measurements were conducted at three nominal SS: 0.2, 0.4, 0.6, and 0.8%. Figure 6 plots the D50 against actual SS (%) (derived from calibrations) for WSM and WSOC particles. As seen in Figure 5, WSM particles were significantly more CCN active than WSOC particles. The single hygroscopicity parameter kappa (κ) was estimated for each SS and D50 according to Petters and Kreidenweis [2007]. Excluding sample 5, κ ranged from 0.112 for sample 1 to 0.173 for sample 6.

Figure 6.

Average critical diameter (D50) as a function of super-saturation (SS %) for WSM and WSOC particles. Error bars indicate the standard deviation between the measurements of samples included in each average.

[39] These results must be evaluated in terms of potential contamination of the isolated WSOC extract by inorganic ions, which would increase the apparent GFs and lower the D50’s. Table 4 presents sulfate “cleaning” efficiencies (one minus the ratio of sulfate concentration after XAD treatment to the initial sulfate concentration, expressed as a percentage), the ratio of the sum of the measured inorganic anions to WSOC in the XAD-treated WSOC extracts, GF80, GF90, and D50 at SS's of 0.35, 0.52, and 0.68% for the six composite samples. Samples 5 and 6 had the lowest cleaning efficiencies (95.1 and 93.4%, respectively, and the highest anionic content (3.0 and 0.94%, respectively). Note that the WSOC concentration is expressed as OC and not OM. The percentages in Table 4 would be roughly half as large if WSOC was expressed as OM. It is clear that samples 5 and 6 have anomalously high GF80 and GF90 and κ as well as smaller D50’s at all super-saturations than the other samples. Sample 2 has somewhat higher GFs and κ and lower D50’s than samples 1, 3, and 4, although its sulfate removal efficiency is the same and its relative ionic content is lower than those of sample 4. While it is not clear that sample 2 is contaminated, if only samples 1, 3, and 4 are considered, then the average GF80 and GF90 would be 1.3 and 2.6% lower, respectively, than the averages for samples 1, 2, 3, 4, and 6 in Table 4. Note also that samples 1, 3, and 4 represent 57% of the study period. In general, this table illustrates that as the sulfate cleaning efficiency approaches 100%, the GF approaches a minimum and the D50’s approach a maximum.

Table 4. Relationship Between Residual Inorganic Ions in Isolated WSOC Extracts and WSOC Hygroscopicity
 SO42−       
 CleaningΣ Ions/      
 EfficiencyaWSOCb  D50 (µm)D50 (µm)D50 (µm)Kappa (κ)
Composite(%)(%)GF80cGF90cSS = 0.35%SS = 0.52%SS = 0.68% 
  1. a

    One minus the ratio of sulfate concentration after XAD treatment to the initial sulfate concentration, expressed as a percentage

  2. b

    Cl- + NO3- + SO42− + NH4+ in XAD-treated water extract. WSOC represents OC, not OM.

  3. c

    The uncertainty in the RH determinations associated with the measurements is approximately 2%

199.40.641.081.140.0980.0760.0630.112 ± 0.005
298.30.261.111.200.0900.0700.0570.145 ± 0.007
399.00.601.081.150.0980.0750.0610.116 ± 0.006
498.30.331.091.140.0970.0760.0620.114 ± 0.005
595.13.01.401.600.0600.0470.0390.47 ± 0.02
693.40.941.151.240.0840.0660.0540.173 ± 0.009

[40] The average value of κ for samples 1, 3, and 4 is 0.114 ± 0.002. Kristensen et al. [2012] presented κ derived from CCN measurements on particles from HULIS extracts for samples collected at SPL in the month following the HPAO study as well as at urban sites in Copenhagen, Denmark and Melpitz, Germany. The respective κ at these sites were 0.10, 0.08, and 0.09. Kristensen et al. [2012] attempted to theoretically correct κ for the effect of inorganic ions present in the HULIS extracts. For SPL, the corrected κ was 0.08. For comparison, Kristensen et al. [2012] presented values of κ derived from CCN measurements of HULIS from previous studies at K-puszta during winter (κ = 0.08, Fors et al. [2010]) and Budapest, Hungary (κ = 0.11 in spring and 0.14 in summer [Ziese et al., 2008]).

3.5 Modeling Results

[41] GFs can also be calculated based upon the chemical composition of the aerosol and a knowledge of the chemical thermodynamics of the various components and their mixtures. Table 5 lists GFs for composite sample 4 at 70, 80, and 90% RH, both for the total WSM and for the XAD extract (WSOC). The amounts of water associated with the inorganic ions, polar organic compounds, and WSOC extract were calculated separately as a function of RH, and the volumes summed. This is equivalent to using the Zdanovskii-Stokes-Robinson approach to estimate aerosol water content, with the three analyzed fractions as components, and allows the relative contributions of the fractions to total water uptake to be assessed. The water associated with the ions was calculated using the model of Clegg et al. [1998] (E-AIM, at http://www.aim.env.uea.ac.uk/aim/aim.php) at the reference RH of 10%, and that of Pitzer [1991] at higher RH. The relationship between RH and water content of the polar organic fraction was calculated using UNIFAC [Fredenslund et al., 1975]. This structure-based model was also used for the molecular formulas identified by FT-ICR MS in the WSOC extract. This is the largest analyzed fraction of the aerosol by mass, and the elemental composition (numbers of C, H, N, O and S atoms), relative abundances, and “double bond equivalents” are known for each molecular formula identified by FT-ICR MS [Supplementary Information to Mazzoleni et al., 2012]. This information is not sufficient to uniquely assign the structures and chemical group compositions required by UNIFAC (i.e., the numbers of –OH, –COOH, –CH2– units, etc., that make up the molecule). These were therefore assigned using a method in which it was assumed, initially, that each molecule had a linear carbon chain “backbone”. UNIFAC structural groups, each of which could be weighted to express a preference for its selection, were assigned to match both the elemental composition and the double bond equivalents of the molecule. In cases where this was not possible, an aromatic molecule was assumed and the same procedure carried out. In only 10 cases out of 3737 molecules was no assignment possible. We note that the structures cannot be uniquely assigned for each molecule, nor even the identities and numbers of each functional group present. Overall, it appears that the element of chemical composition that the GF is most sensitive to is the numbers of highly polar groups present—chiefly –OH and –COOH. Increases in the numbers of these groups tend to increase the water uptake at any given RH.

Table 5. Measured and Calculated Growth Factors of Material From Sample 4a
RH (%)Growth Factor (meas.)Growth Factor (calc.)bGrowth Factor (calc.)cVolume Fraction (ions)Volume Fraction (polar organics)Volume Fraction (extract)
  1. a

    The calculated growth factors assume fully liquid aerosol particles, and are referenced to unity at 10% RH.

  2. b

    Based upon a water activity/concentration relationship for aqueous solutions of the organic extract calculated using UNIFAC. The calculated volume fractions of the ions, polar organics, and extracted organics correspond to this growth factor.

  3. c

    Based upon a water activity/concentration relationship for aqueous solutions of the organic extract calculated assuming Raoult's law.

Total Water Soluble Material (WSM)
701.131.121.120.360.1260.51
801.211.171.180.420.1280.46
901.441.311.340.530.1320.348
WSOC in XAD extract
701.061.031.040.0120.0500.94
801.091.041.060.0150.0570.93
901.151.061.140.0270.0800.89

[42] Densities, and hence solution volumes and GFs, were estimated using equation (12) of Semmler et al. [2006]. Densities of electrolyte components in water were taken from Clegg and Wexler [2011] and densities of the (liquid) organic solutes were estimated using the method of Girolami [1994].

[43] The range of GFs calculated using this approach, and listed in Table 5 for composite sample 4, are in broad agreement with the measurements for the total WSM. The calculated GFs are similar to those measured with the HTDMA. At RH of about 90% and above the uptake of water by the inorganic ions has the greatest influence in the GF. The difference between the model and measurements for the WSOC in the XAD extract is less than 0.1 in the GF at all three RH. However, there is some uncertainty associated with the GF of the total WSM sample, because of a charge imbalance between the inorganic cations and anions measured in the sample. (This was corrected, for these calculations, by increasing the amounts of inorganic anions assumed to be present to balance the total charge on the cations.) The assumption of a Raoult's law relationship between concentration and RH, for the compounds in the WSOC extract identified by FT-ICR MS, yields higher GFs than the UNIFAC-based calculation (compare columns three and four in Table 5). However, the difference is quite small which suggests that the WSOC contribution to the GF is relatively insensitive to the assumed physico-chemical properties of this fraction.

4 Summary

[44] Organic compounds compose a significant fraction of the aerosol in remote locations. However, the chemical composition and hygroscopic properties of organic compounds are not well understood. To address these issues, a multi-institutional collaborative study was undertaken. A field experiment was conducted from 24 June through 28 July 2012 at SPL, a high-altitude remote continental site in the Park Range of northwestern Colorado. PM2.5 samples (particles with diameters smaller than 2.5 µm) were collected on filters daily. Organic (OC), elemental (EC), and WSOC were measured on quartz-fiber filters. The positive sampling artifact for OC caused by absorption of volatile compounds by the quartz filter was corrected for by subtracting OC on a backup quartz filter. The positive artifact was 28 ± 10(SD)% and 27 ± 6(SD)% of front filter OC and WSOC, respectively. hivol PM2.5 samples were collected on TIGF filters and analyzed for WSOC, sulfate, nitrate, ammonium, and chloride ions. Organic material accounted for 64 ± 9(SD)% of the mass of measured species, followed by sulfate (21 ± 7[SD]%), EC (5.4 ± 1.0[SD]%), nitrate (4.8 ± 1.8[SD]%), and ammonium (3.9 ± 1.3[SD]%). WSOC comprised 89 ± 21[SD]% of OC.

[45] While the chemical composition of the daily samples was relatively stable over time, the hivol TIGF filters were combined sequentially into six composites based on air-mass back trajectories to provide sufficient material for detailed organic analysis and hygroscopicity measurements. High-molecular weight organic compounds were separated from inorganic ions using XAD-4 and XAD-8 solid phase absorbents. The removal efficiency for sulfate, nitrate, and ammonium ions ranged from 93.4 to 99.4% and averaged 97 ± 2%. Forty-seven compounds representing organic acids, sugars, and sugar alcohols were identified which accounted for 12.6% of the WSOC, on average. Based on the chemical structure of these compounds and the elemental compositions of over 4000 compounds characterized by ultrahigh-resolution FT-ICR MS, an OM/OC ratio of 2.04 was estimated for the isolated WSOC.

[46] Hygroscopic growth and CCN activity were determined experimentally for the six composite water extracts. The average hygroscopic GFs at RH = 80% (GF80) were 1.10 ± 0.03(SD) for particles generated from isolated WSOC and 1.27 ± 0.03(SD) for particles generated from the total WSM. CCN activity followed a similar pattern. The critical diameters at a super-saturation of 0.35% were 0.072 ± 0.009(SD) and 0.094 ± 0.006(SD) µm for particles derived from WSM and isolated WSOC, respectively. Considering potential inorganic contamination of the isolated WSOC extracts, GF80 and GF90 would decrease by 1.3 and 2.6%, respectively, and critical diameters (D50) would increase by only 0.003 to 0.004 µm at super-saturations of 0.35, 0.52, and 0.68%. The average value of the hygroscopicity parameter κ, 0.114 ± 0.002, derived from CCN measurements on isolated WSOC compares well with values derived from previous studies. During this study, water-soluble organic compounds contributed significantly to aerosol hygroscopic properties. These GF results compared favorably with estimates from a combination of the E-AIM and UNIFAC models, which explicitly treats the water activity for the total soluble material identified in this study.

Acknowledgments

[47] The National Science Foundation Division of Atmospheric Sciences collaborative grant AGS-0931431, AGS-0931910, AGS-0931505, and AGS- 0931390 supported this work. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The authors thank Ty Atkins for technical assistance with the field campaign. The Steamboat Ski Resort provided logistical support and in-kind donations. The DRI's SPL is an equal opportunity service provider and employer and is a permittee of the Medicine-Bow Routt National Forests. The authors thank Dr. Eric Williams for access to SO2 data from a previous field campaign at SPL. The authors also thank Dr. Allison Steiner for assistance in interpreting the SO2 data.

Ancillary