INTRODUCTION
Organic aerosols (OAs) comprise a significant portion of the atmospheric fine particulate matter and their concentration mainly depends on the geographic position. It has been shown that OA concentrations can amount to 20–50% at continental midlatitudes and up to 90% in tropical forests (Kanakidou et al. Reference Kanakidou, Seinfeld, Pandis, Barnes, Dentener, Facchini, Van Dingenen, Ervens, Nenes, Nielsen, Swietlick, Putaud, Balkanski, Fuzzi, Horth, Moortgat, Winterhalter, Myhre, Tsigaridis, Vignati, Stephanou and Wilson2005). Depending on its chemical composition, OA may have a negative influence on both human health and climate. OAs are classified into primary OA (POA) and secondary OA (SOA). POAs are composed of organic compounds that are directly emitted into the atmosphere as particulate matter from different sources. The primary fraction can also contain a considerable amount of inorganic substances. The SOA fraction is produced in the atmosphere by oxidation of volatile organic compounds (VOCs), reducing their volatility so that they partition from the gas to the aerosol phase (Robinson et al. Reference Robinson, Donahue, Shrivastava, Weitkamp, Sage, Grieshop, Lane, Pierce and Pandis2007; De Gouw and Jimenez Reference De Gouw and Jimenez2009; Hodzic et al. Reference Hodzic, Jimenez, Prévôt, Szidat, Fast and Madronich2010). The total carbon (TC) from the OA is divided into two fractions, i.e. organic carbon (OC) and elemental carbon (EC), of which OC comprises less refractory, non-light-absorbing chemical species and EC consists of highly polymerized and optically active structures (Pöschl Reference Pöschl2005). These two fractions can be separated using thermo-optical devices like the commercial Sunset OC/EC analyzer using specified protocols like Swiss_4S (Zhang et al. Reference Zhang, Perron, Ciobanu, Zotter, Minguillon, Wacker, Prevot, Baltensperger and Szidat2012).
Radiocarbon measurements with accelerator mass spectrometry (AMS) can apportion organic aerosol emissions into fossil and nonfossil sources and quantify such emissions (Szidat Reference Szidat2009). This speciation relies on the principle that 14C is extinct in fossil material (F14C=0), whereas in modern materials 14C is found on the contemporary level (F14C~1). The 14C analysis should be performed separately in OC and EC, since both fractions have different emission origins and affect human health and Earth climate differently (Szidat et al. Reference Szidat, Jenk, Gäggeler, Synal, Fisseha, Baltensperger, Kalberer, Samburova, Reimann, Kasper-Giebl and Hajdas2004a, Reference Szidat, Jenk, Gäggeler, Synal, Hajdas, Bonani and Saurer2004b, Reference Szidat, Jenk, Synal, Kalberer, Wacker, Hajdas, Kasper-Giebl and Baltensperger2006; Pöschl Reference Pöschl2005; Currie and Kessler Reference Currie and Kessler2005; Szidat Reference Szidat2009).
Currently, we perform 14C analysis of ambient aerosol samples by using various analytical combustion instruments coupled to a MICADAS (Mini Carbon DAting System) AMS by means of a gas inlet system (GIS). The analysis involves OC/EC thermal separation, combustion, trapping the CO2 pulse, releasing the gas, and slow injection or infusion into the ion source where the CO2 is converted into C–, O–, as well as other ionic and neutral species (Perron et al. Reference Perron, Szidat, Fahrni, Ruff, Wacker, Prévôt and Baltensperger2010; Ruff et al. Reference Ruff, Fahrni, Gäggeler, Hajdas, Suter, Synal, Szidat and Wacker2010; Wacker et al. Reference Wacker, Fahrni, Hajdas, Molnar, Synal, Szidat and Zhang2013; Agrios et al. Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015; Salazar et al. Reference Salazar, Zhang, Agrios and Szidat2015). This coupling allows for high-throughput routine 14C analysis of aerosols with small carbon mass (>10 µg C). It is possible to analyze the OC and EC fractions separately from one sample with the Sunset OC/EC analyzer coupled to the GIS-AMS system in an online trapping mode (Agrios et al. Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015) because the separation protocol allows for a time gap between the OC and EC peaks of around 20 min. However, subfractions separated with a modified thermal ramp cannot be analyzed by trapping the CO2 because these subfractions appear too close to each other. Therefore, for improved source apportionment an interface is required that injects the CO2 from the separated subfractions in a real-time continuous-flow mode into the AMS. This paper continues the study of a continuous-flow interface developed earlier (Agrios et al. Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015). Here, we characterize and validate the interface using standard materials. In addition, this paper shows the application of this interface to the 14C analysis of thermal subfractions of OC for selected atmospheric aerosol filters as examples of its performance.
MATERIALS AND METHODS
Sunset OC/EC Analyzer and Temperature Programs
Samples and standard materials are loaded on prebaked quartz filters in a Sunset OC/EC analyzer (Model4L, Sunset Laboratory, USA), which is specially equipped with a nondispersive infrared (NDIR) detector (Zhang et al. Reference Zhang, Perron, Ciobanu, Zotter, Minguillon, Wacker, Prevot, Baltensperger and Szidat2012). The Sunset OC/EC analyzer thermally desorbs and/or combusts aerosol samples by applying a temperature ramp. Afterwards, the evolving products are oxidized catalytically, producing a pulse of CO2 gas, which then passes through the NDIR and a H2O trap containing Sicapent (Merck, Germany) entering the interface, which is connected to the AMS instrument. Here, the CO2 is ionized and detected as a 12C– peak in a Faraday cup on the low-energy side of the AMS. The setup of the system and a proof-of-principle of this method was described by Agrios et al. (Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015).
Three different thermal protocols, termed the TC, OC/EC, and OC subfraction, are used in this work. Pure O2 (99.9995%) is used as carrier gas in order to diminish the charring effect that takes place in anoxic conditions (Zhang et al. Reference Zhang, Perron, Ciobanu, Zotter, Minguillon, Wacker, Prevot, Baltensperger and Szidat2012). In the total carbon (TC) protocol, only one step is used at 760°C for 400 s. Our OC/EC thermal protocol is based on the 4-step Swiss_4S protocol (for details see Zhang et al. Reference Zhang, Perron, Ciobanu, Zotter, Minguillon, Wacker, Prevot, Baltensperger and Szidat2012), but excluding the intermediate helium step. Our OC/EC protocol defines OC as the aerosol fraction that combusts in pure oxygen at 375°C for 400 s in the first step. Then, residual OC is completely removed in pure oxygen at 475°C for 400 s in the second step, where some premature EC evolution occurs as well, and subsequently EC is combusted in pure oxygen at 680°C for 300 s in third step. Two protocols are used to separate the OC subfractions. Their temperature steps are 200, 260, 300, and 375°C for the first protocol and 200 and 375°C for the second protocol with a total duration of 400 s for both protocols. Cooling is not applied between the different steps in order to ensure complete desorption of less volatile components and diminish peak-to-peak carryover. The duration of the thermal steps is long enough to allow baseline separation of the 12C– peaks.
Reference Materials
C7 (oxalic acid), C6 (sucrose), and C5 (wood) are reference materials from the International Atomic Energy Agency (IAEA) (Rozanski et al. Reference Rozanski, Stichler, Gonfiantini, Scott, Beukens, Kromer and van der Plicht1992; Le Clercq et al. Reference Le Clercq, van der Plicht and Gröning1998). The primary standard oxalic acid II (HOxII, SRM 4990C) is provided by the National Institute of Standards and Technology (NIST) and 14C-free material sodium acetate (NaOAc, p.a.) from Merck. They are used for standard normalization and blank subtraction, respectively. Measured F14C values of the standards are given in Table 1. The standards are analyzed with the TC protocol. Grains of the standard materials are loaded on a clean filter and combusted on the sample holder of Sunset OC/EC analyzer. For the calibration of the Sunset OC/EC analyzer, aqueous solutions of sucrose (Sigma Aldrich, D(+)-Sucrose, Fluka Analytical) are employed. Some 5–10 μL from sucrose stock solutions of 1–10 μg/μL are pipetted on 1.5-cm2 prebaked quartz-fiber filters. The wet samples are inserted in the Sunset OC/EC analyzer sample holder and dried during 2 min at 70°C under a helium stream before initiating the total carbon combustion temperature program.
Table 1 14C/12C analysis of IAEA standards (Rozanski et al. Reference Rozanski, Stichler, Gonfiantini, Scott, Beukens, Kromer and van der Plicht1992; Le Clercq et al. Reference Le Clercq, van der Plicht and Gröning1998) using the continuous-flow method with 1σ uncertainties for n=3 repetitions for each material. Measured carbon amounts ranged between 15 and 20 μg C. The P value describes the level of significance from a t test comparing nominal and measured F14C, indicating a statistically significant difference for P values<0.05.
Ambient Aerosol Filters
Ambient aerosol filters were selected from different sites: (a) from an urban background station in Zurich, Switzerland, which is located in a park-like courtyard in the city center close to the main railway station (Szidat et al. Reference Szidat, Jenk, Gäggeler, Synal, Fisseha, Baltensperger, Kalberer, Samburova, Reimann, Kasper-Giebl and Hajdas2004a; Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014a); (b) from a suburban station in Sissach, Switzerland, which is influenced by local traffic (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014a); as well as (c) from a suburban station on the Caltech campus in Pasadena, CA, as part of the CalNex 2010 field campaign in the Los Angeles Basin, which aimed at studying the urban plume from Los Angeles and the regional background air (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014b). The samples from Zurich and Sissach were heavily loaded with particulate matter containing typically 30 and 10 μg C/cm2 for OC and EC, respectively. Filters from Los Angeles were loaded with low amounts of 10 μg C/cm2 TC. For the combustion of the Zurich filters, a 0.3-cm2 quartz filter punch per sample is used. In the case of the Los Angeles filters, three 1.5-cm2 quartz filter punches (i.e. the normal size of the Sunset EC/OC analyzer) were piled up in the sample holder due to the low loading of the samples. Therefore, it should be noted that O2 carrier may have accessed the surface of each piled filter inhomogeneously during combustion. We cannot exclude that this situation may have produced artifacts during the measurement.
Cu Reduction Reactor Interface
The direct coupling of the Sunset OC/EC analyzer to the MICADAS requires that the O2 carrier gas excess is scrubbed away prior to the CO2 pulse feeding into the ion source. For this reason, we employ a heated reactor filled with Cu wire bits that has been described before by Agrios et al. (Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015). In brief, the oxygen combusts the samples and the CO2 gas is carried through an oven filled with MnO2, an NDIR detector, and then through a H2O trap. Afterwards, the exhaust stream passes through the heated Cu reactor, forming copper oxide and reducing the O2 flow from 20 mL/min down to ~0 mL/min. Since the O2 is scrubbed away, an additional, minimal He carrier flow is added after the exhaust of the H2O trap via a union Tee and in front of the Cu reactor. This allows the CO2 to be transported into the ion source in a splitless mode via a 90-cm-long fused silica capillary located after the Cu reactor. The minimal He flow is continuously supplied at flow rates that maintain the high vacuum and ionization efficiency of the ion source. The normal pressure range in the ion source of our MICADAS AMS is 4 to 10×10–6 hPa, which limits feasible inner diameters (ID) of the inlet capillary to 0.10–0.13 mm. We use He carrier flow rates of 1–4 mL/min as a compromise, which are high enough to overcome dead volumes and sample losses, but low enough to keep the vacuum of the ion source.
14C Analysis
The 200kV AMS MICADAS installed at the Laboratory for the Analysis of Radiocarbon with AMS (LARA) uses a gas ion source, which accepts CO2 (Wacker et al. Reference Wacker, Fahrni, Hajdas, Molnar, Synal, Szidat and Zhang2013; Szidat et al. Reference Szidat, Salazar, Vogel, Battaglia, Wacker, Synal and Türler2014). The measured 14C/12C ratios are presented as absolute ratios as qualitative information or as F14C values as defined by Reimer et al. (Reference Reimer, Brown and Reimer2004). During gas feeding, the 14C counting and the stable isotopes currents are averaged every 3 s. To calculate the sample F14C, the 14C from the blank and the residual isobaric molecular interference are subtracted from the 14C/12C ratio. Then, the ratio is corrected for fractionation with the 13C/12C ratio and normalized with the 14C/12C ratio of the standard.
RESULTS AND DISCUSSION
The results for the continuous-flow Sunset OC/EC analyzer coupling to the MICADAS AMS consist of three main parts: method optimization, validation, and application to real samples. In the method optimization, we first study how to control the carbon mass flow within the injected pulse of CO2 delivered into the ion source. Secondly, we show how the carbon mass flow affects the ionization efficiency. For the real samples, we investigate the thermal separation of the OC subfractions by measuring the 14C/12C ratio in real time.
Method Optimization: Controlling Carbon Mass Flow Rate
Figure 1 illustrates the behavior of the 12C– current when continuous-flow CO2 pulses are injected into the ion source. Due to the dead volume of the Cu reactor, the 12C– peaks are broader by a factor of 1.5 in comparison to the NDIR peaks. Figures 1a and 1b show a minimum of the 12C– current in the middle of the peak for C mass flows of 1.54 and 2.4 µg C/min, respectively. [The carbon mass flow is defined as the carbon mass injected divided by the time of the full width at half maximum (FWHM) of the 12C– peak.] The drop of the 12C– current happens at relatively low He carrier flow rate (2.5 mL/min) or high carbon masses (20 µg C). This inconvenient ionization suppression causes double peaks from single CO2 injections. It is known that the O– signal is much higher than the C– signal for direct gas ionization as shown by Fahrni et al. (Reference Fahrni, Wacker, Synal and Szidat2013). At high carbon mass flows, the CO2 concentration at the target is higher; therefore, it is possible that the oxygen capture most of the available electrons, suppressing the C– formation (Salazar et al. Reference Salazar, Agrios, Eichler and Szidat2016). The feeding of the CO2 pulse into the ion source with a continuous-flow system cannot be kept as constant and controllable as performed by the gas-tight syringe of the GIS. Controlling the C mass flow depends on knowing the CO2 partial pressure and the pressure at both sides of the ion source inlet capillary, which are unknown for a CO2 pulse injection system.
Figure 1 12C– currents from the analysis of HOxII: (a) 6 μg C transported to the ion source by using a 0.13-mm-ID inlet capillary at 2.5 mL min–1 (dashed line) and 3.5 mL min–1 (solid line) He carrier flow rate; and (b) 6 μg C (dashed black line), 10 μg C (red solid line), and 20 μg C delivered (dotted line) via a 0.10-mm-ID inlet capillary and at a constant He carrier flow rate of 1.5 mLmin–1. Corresponding average carbon mass flow rates are indicated in the legend.
Because the C mass flow affects the signal, it is important to find out how to control it by changing the He carrier flow, the dimensions of the inlet capillary, and combusting different carbon masses. In Figure 2a, the helium carrier gas flow rate is kept constant but the carbon mass is changed in the range of 3–50 μg C, causing the carbon mass flow to increase linearly. Injections of samples higher than 10 μg C are affected by the ionization suppression and such measurements are denoted with a cross symbol in the graph. In Figure 2b, the sample mass is constant at 6 μg C (HOxII) and injections with different He carrier flow rates are performed. The carbon mass flow exhibits a linear relationship with the inverse of the He carrier flow. This reveals that the C mass flow is directly proportional to the injected mass and inversely proportional to the He carrier flow. The importance of these relationships among the experimental parameters allow us to control the carbon mass flow of the analysis.
Figure 2 Carbon mass flow rate of HOxII depending on (a) the C mass injected at 1.5 mL min–1 and (b) the inverse of the He flow using 6 μg C and He flows of 0.5–3.5 mL/min. Measurements with ionization suppression are indicated by crosses. In all measurements, CO2 was transported to the ion source using a 0.10-mm-ID inlet capillary. Same experimental details as in Figure 1.
Method Optimization: Effect of Carbon Mass Flow Rate on the Ionization
Figure 3a demonstrates how the negative ionization is affected by the carbon mass flow rate. For a constant mass of 6 μg C HOxII, we trace the behavior of the system for He carrier flows between 0.5 and 3.5 mL min–1. For higher carrier flow rates, eventually the 12C– signal is lost due to back-pressure exerted by the narrow capillary. The ionization suppression starts at a certain carbon mass flow rate as indicated by the transition between the closed symbols and the asterisks. For our instrumental setup, the suppression thresholds are ~1.8 and ~1.25 μg/min depending on the capillary ID. As indicated before, the suppression may be due to the high concentration of oxygen from CO2 suppressing the C– formation. For the 0.13-mm ID, it seems to be possible to achieve higher efficiencies, but those measurements are affected by ionization suppression. The 0.10-mm-ID capillary enables a wider range of flows with unsuppressed signal. Unfortunately, in our case it is not possible to use higher He carrier flow rates than 1.5 mL min–1 for the smaller capillary because the back-pressure affects the performance of the Sunset OC/EC analyzer. It should be noted, however, that the strong differences between the ionization yields for both capillaries may partially be caused from day-to-day changes of the state of the ion source.
Figure 3 Influence of the carbon mass flow rate on (a) ionization yields using fixed sample masses of 6 µg C and variable He flows and (b) 12C– currents depending on the mass of the sample and geometry of the gas target. Measurements with ionization suppression are indicated by crosses. Same experimental details as in Figure 1.
In the ion source of MICADAS, the CO2 is sputtered by Cs+ after it has diffused on the cathode front via a narrow borehole (Fahrni et al. Reference Fahrni, Wacker, Synal and Szidat2013). Therefore, we used gas targets with a 1.9-mm-ID borehole size in order to study how this affects the C– formation in comparison with the original gas targets with 1.0-mm-ID borehole size (Figure 3b). Since the He flow rate is kept constant, the C mass flow rate is the same for the two targets of different borehole size, when same carbon masses are injected. However, the 12C– currents and ionization efficiency are higher for the 1.0-mm-ID. borehole at every point in Figure 3b. For the 1.9-mm-ID borehole, ionization suppression is not visible within the presented CO2 mass range. We believe that these features are due to the lower partial pressure of CO2 when diffusing through the larger volume of the 1.9-mm-ID borehole. Therefore, the ionization not only depends on the carbon mass flow rate, but also in the CO2 concentration at the target surface. Both parameters, the ID of the capillary and the borehole size of the target, should be investigated further in order to optimize measurement conditions for smaller versus larger samples.
Validation
Table 1 presents mean values from three injections per standard material (C5, C6, and C7) taken with the TC protocol using a He flow of 1.5 mL min–1 and inlet capillary of 0.10 mm ID. The individual F14C values are calculated by averaging the isotope ratios over the FWHM of the 12C– peak. Only one grain of standard material was used per measurement and its carbon mass ranges from 15 to 20 μg. In general, 13C/12C and 14C/12C ratios are relatively stable during the ionization suppression that occurs while injecting standard material from 15 to 45 μg C. The resulting peak broadening caused by the high C mass flow rates extends the measurement time and this leads to higher number of 14C counts and, therefore, to a lower uncertainty. Analyses of HOxII samples between 3 and 45 μg C provided 1274 to 12,031 14C counts, obtaining measurement uncertainties from 5% to 1.5%, respectively. All the P values calculated from the t test comparison between the F14C of the respective reference value and our measurement are much larger than the significance level of 0.05. This indicates that there is no statistically significant difference between both values for all standards.
By following the same procedure as Salazar et al. (Reference Salazar, Zhang, Agrios and Szidat2015) for measuring constant and cross-contamination, solid samples spanning different sizes of fossil NaOAc and modern HOxII were combusted in order to estimate the constant and cross-contamination of the Sunset OC/EC analyzer system coupled to the Cu reactor. The system induces a constant contamination mass of 0.6±0.1 μg C with an F14C value of 0.75±0.05, which is comparable to conditions when the Sunset OC/EC analyzer is connected to the GIS using the trap (Agrios et al. Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015). The cross-contamination factor of the system was estimated by combusting the fossil and modern materials alternately and in separate runs with fresh targets. We estimate that 1.4±0.6% of the previous sample remains and cross-contaminates the following measurement. This is substantially higher than for the GIS trapping system, which revealed a cross-contamination factor of 0.5±0.4% (Agrios et al. Reference Agrios, Salazar, Zhang, Uglietti, Battaglia and Luginbühl2015). Cross- and constant contamination tend to increase with the amount of packed granular material interacting with the CO2. In our case, the Cu reduction reactor contains a lot of wire bits, which may explain the slightly higher cross-contamination. Nevertheless, the constant contamination is not affected in the same negative way.
Ambient Aerosol Filter Analysis
Figure 4 presents a proof-of-principle for the separation run of OC and EC for a selected winter sample from the urban background station Zurich. In a pure O2 method, the OC fraction evolved in the first peak at 375°C and the more refractory, pure EC in the last temperature step (650°C). In between, the intermediate fraction at 475°C characterizes the transition from refractory OC with a more nonfossil signal to nonrefractory EC with a rather fossil signal as indicated by the decreasing trend of the 14C/12C ratio. The slope of the linear fitting has been used as a metric to evaluate to what extent the molecular components of distinct thermal fractions exhibit or not significantly different 14C/12C ratios and if the 14C/12C concentration is uniform within the fraction or subfraction. This metric is important for weighting if the thermal fraction needs further purification and better separation tuning. The 14C/12C slopes of the EC and OC fractions are small (<0.011) compared to the second step at 475°C. This second step reflects the transition from the initial step, which is enriched in OC and depleted in EC, to nearly pure EC at the final step. Consequently, we define, as a first approximation, that there is no significant 14C shift if the slope is lower than 0.011. This threshold is protocol-dependent and more studies are required to arrive into a reliable value. The continuous-flow separation of Figure 4 highlights the importance of the intermediate temperature step to purify the EC on the one hand and the potential of the real-time online 14C analysis to uncover different fossil versus nonfossil signals of chemically similar aerosol fractions on the other hand.
Figure 4 Continuous-flow 14C/12C measurement overlaid with the 12C– current and the 13C/12C ratio for an aerosol filter from Zurich from January 2008 treated with the OC/EC protocol consisting of three temperature ramps (375, 475, and 650°C) under pure oxygen. The elevated baseline after the second peak is due to an artifact of the measurement and has not been reproduced in other runs.
Figure 5 shows two characteristic examples of real-time 14C measurements of OC subfractions with modified temperature ramps. Taking in account the threshold slope of 0.011, Figure 5a shows that with the exception of the subfraction at 200°C, the 14C/12C ratio does not vary significantly throughout the thermal treatment for this winter sample from Zurich with substantial wood-burning emissions (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014b). The more fossil signal at the onset of the temperature ramp indicates that the wood-burning OC, which carries a purely nonfossil signal, is slightly more refractive than the fossil OC from the urban traffic. The situation is different for a spring sample from the Los Angeles Basin (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014a) shown in Figure 5b. Here, the OC peak is split thermally only into two subfractions, as the application of more thermal steps is not feasible when the filter loading is low. The volatile subfraction evolving in the first peak at 200°C is characterized by more fossil levels with a trend towards more nonfossil values for an increasing refractivity of the OC chemical species, which even continues into the second peak at 375°C. This is corroborated by the observation for the Los Angeles Basin that fresh SOA, which is more volatile, mainly originates from fossil emissions from the urban center, whereas aged (i.e. less volatile) SOA rather originates from a nonfossil rural background outside of the city (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014a). Table 2 compiles the examples from Figure 5 and further real-time 14C analyses of OC subfractions with similar thermal behavior. Each row contains the values for one filter measured once. The isotopic data are the averages of the values laying inside the FWHM of the 12C– peak due to the good stability of the 13C/12C ratio in this range. From the linear fitting evaluation, only the measured thermal subfractions denoted by asterisks in the table demonstrate a significant difference in the 14C/12C concentration within the respective subfraction. Table 2 shows that the shift from fossil to nonfossil emissions in the first peak for the winter sample from Zurich was neither visible for Zurich during summer, nor for the rural station Sissach during winter. This may result from the absence of wood burning during summer for the former and smaller traffic contribution for the latter case (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014b). For the Los Angeles Basin, the additional cases from Table 2 behave differently than the sample from 20/05/2010, which is displayed in Figure 5b, probably because this sample is more influenced by the local urban plume than the other cases (Zotter et al. Reference Zotter, Ciobanu, Zhang, El-Haddad, Macchia, Daellenbach, Salazar, Huang, Wacker, Hueglin, Piazzalunga, Fermo, Schwikowski, Baltensperger, Szidat and Prévôt2014a).
Figure 5 Continuous-flow 14C/12C analysis of OC subfractions (a) evolving at 200, 260, 300, and 375°C for an aerosol filter from Zurich for 28/01/2009; and (b) of OC subfractions evolving at 200 and 375°C for an aerosol filter from the Los Angeles Basin for 20/05/2010. Please note that both examples are also included in Table 2.
Table 2 14C real-time analysis of a selection of ambient aerosol filters from several campaigns. The F14C values are averaged over distinct CO2 subfractions evolving at 200, 260, 300, and 375°C or 200 and 375°C. Each filter was measured only once. The uncertainty is related to the counting statistics. Statistically significant 14C shifts within such subfractions are indicated with an asterisk. For the filters from the Los Angeles Basin, three 1.5-cm2 quartz filter punches were piled up in the sample holder during combustion. ND means not detected.
CONCLUSIONS AND OUTLOOK
A continuous-flow interface for 14C AMS is described that allows a commercial Sunset OC/EC analyzer to be coupled to a MICADAS gas ion source. Dimensions of the transfer capillary and target boreholes as well as the He carrier gas flow are optimized such that carbon masses are adjusted in order to prevent 12C– signal suppression, and the response of the system in the analysis of very small samples is satisfactory. This allows for the first time to measure efficiently ambient atmospheric aerosol combustion products by 14C AMS in real-time mode, revealing the potential of real-time online 14C analysis of chemically similar subfractions that are not possible to be disentangled by conventional thermo-optical methods. Future studies will focus on the evaluation of the peak-to-peak memory by combusting known mixtures of fossil and modern materials in a modified temperature ramp as well as on the usage of targets with intermediate ID borehole sizes as to optimize further the CO2 proportion to carrier gas.
