Anal. Chem. 2005, 77, 7288-7293
Carbon-Specific Analysis of Humic-like Substances in Atmospheric Aerosol and Precipitation Samples Andreas Limbeck,* Markus Handler, Bernhard Neuberger, Barbara Klatzer, and Hans Puxbaum
Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9/164, A-1060 Vienna, Austria
A new approach for the carbon-specific determination of humic-like substances (HULIS) in atmospheric aerosols is presented. The method is based on a two-step isolation procedure of HULIS and the determination of HULIS carbon with a dissolved organic carbon analyzer. In the first step, a C18 solid-phase extraction is performed to separate HULIS from inorganic and hydrophilic organic sample constituents in aqueous sample solutions. The second isolation step is conducted on a strong anion exchanger to separate HULIS from remaining carbonaceous compounds. This ion chromatographic separation step including the subsequent on-line detection of HULIS carbon was performed fully automated to avoid the risk of sample contamination and to enhance the reproducibility of the method. With a 5-mL sample volume, a limit of detection of 1.0 mg C/L was obtained; this corresponds to an absolute amount of 5 µg of HULIS carbon. The reproducibility of the method given as the relative standard deviation was 4.3% (n ) 10). The method was applied for the determination of water-soluble HULIS in airborne particulate matter. PM10 concentrations at an urban site in Vienna, Austria, ranged from around 0.1 to 1.8 µg of C/m3 (n ) 49); the fraction of water-soluble HULIS in OC was 12.1 ( 7.2% (n ) 49). The importance of atmospheric aerosols to the global radiation balance, cloud formation, and alleged human health effects has motivated numerous studies on aerosol formation and growth as well as on aerosol composition. Although hundreds of individual organic compounds have been identified in the organic atmospheric aerosol so far,1 together they constitute only a minor fraction of the organic carbon (OC) of urban and rural aerosol.2,3 However, the major contributors to the continental organic aerosol are water-soluble macromolecular compounds with chemical characteristics very similar to naturally occurring humic acids. Evidence for the presence of such humic-like substances (HULIS) in airborne particulate matter was reported as early as 1980 by Simoneit.4,5 More recently, quantitative techniques for the deter* To whom correspondence should be addressed. Tel.: +43 1 58801 15180. Fax: +43 1 58801 15199. E-mail:
[email protected]. (1) Saxena, P.; Hildemann, L. J. Atmos. Chem. 1996, 24, 57-109. (2) Rogge, W.; Mazurek, M.; Hildemann, L.; Cass, G.; Simoneit, B. R. T. Atmos. Environ. 1993, 27A, 1309-1330. (3) Puxbaum, H.; Rendl, J.; Allabashi, R.; Otter, L.; Scholes, M. J. Geophys. Res. 2000, 105, 20697-20706. (4) Simoneit, B. R. T. Phys. Chem. Earth 1980, 12, 343-352.
7288 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
mination of HULIS emerged, indicating a substantial contribution of 20-50% to the water-soluble organic aerosol at urban and rural sites in Europe.6-9 This means that HULIS have an impact on the hygroscopicity and the cloud condensation nuclei formation potential of the atmospheric aerosol and are, therefore, of climatic relevance.10,11 Several techniques were reported for the analysis of this compound class in airborne particulate matter. An early technique used for the isolation of HULIS from the aerosol was identical to that commonly used for the separation of humic acids from soils.12 The aerosol constituents that were soluble in hot aqueous alkali were precipitated from the derived sample solution by acidification. After a washing and drying step, the derived residue was investigated with infrared spectroscopy. A different approach was described by Havers et al.;6 they extracted the humic-like fraction from the aerosol samples with an aqueous sodium hydroxide solution; subsequently, the extract was neutralized with hydrochloric acid, and finally the HULIS were separated from the sample solution using an anion exchange column and investigated with different spectroscopic techniques. More recently, size exclusion chromatography in combination with UV detection,7 capillary electrophoresis,13 fractionation on different separation columns,14,15 and liquid chromatography coupled to mass spectrometry16,17 has (5) Simoneit, B. R. T. Atmos. Environ. 1982, 9A, 2139-2159. (6) Havers, N.; Burba, P.; Lambert, J.; Klockow, D. J. Atmos. Chem. 1998, 29, 45-54. (7) Zappoli, S.; Andracchio, A.; Fuzzi, S.; Facchini, M.; Gelencser, A.; Kiss, G.; Krivacsy, Z.; Molnar, A.; Meszaros, E.; Hansson, H.; Rosman, K.; Zebu ¨ hr, Y. Atmos. Environ. 1999, 33, 2733-2743. (8) Facchini, C.; Fuzzi, S.; Zappoli, S.; Andracchio, A.; Gelencser, A.; Kiss, G.; Krivacsy, Z.; Meszaros, E.; Hansson, H.; Alsberg, T.; Zebu ¨ hr, Y. J. Geophys. Res. 1999, 104, 26821-26832. (9) Krivacsy, Z.; Gelencser, A.; Kiss, G.; Meszaros, E.; Molnar, A.; Hoffer, A.; Meszaros, T.; Sarvari, Z.; Temesi, D.; Varga, B.; Baltensberger, U.; Nyeki, S.; Weingartner, E. J. Atmos. Chem. 2001, 39, 235-259. (10) Facchini, C.; Mircea, M.; Fuzzi, S.; Charlson, R. Nature 1999, 401, 257259. (11) Charlson, R.; Seinfeld, J.; Nenes, A.; Kulmala, M.; Laaksonen, A.; Facchini, M. Science 2001, 292, 2025-2026. (12) Mukai, H.; Ambe, Y. Atmos. Environ. 1986, 20, 813-819. (13) Krivacsy, Z.; Kiss, G.; Varga, B.; Galambos, I.; Sarvari, Z.; Gelencser, A.; Molnar, A.; Fuzzi, S.; Facchini, M.; Zappoli, S.; Andracchio, A.; Alsberg, T.; Hansson, H.; Persson, L. Atmos. Environ. 2000, 34, 4273-4281. (14) Varga, B.; Kiss, G.; Ganszky, I.; Gelencser, A.; Krivacsy, Z. Talanta 2001, 55, 561-572. (15) Andracchio, A.; Cavicchi, C.; Tonelli, D.; Zappoli, S. Atmos. Environ. 2002, 36, 5097-5107. (16) Kiss, G.; Tombacz, E.; Varga, B.; Alsberg, T.; Persson, L. Atmos. Environ. 2003, 37, 3783-3794. (17) Feng, J.; Mo ¨ller, D. J. Atmos. Chem. 2004, 48, 217-233. 10.1021/ac050953l CCC: $30.25
© 2005 American Chemical Society Published on Web 10/11/2005
been used for the determination of HULIS in aqueous extracts of particulate aerosol samples. Although much work has been dedicated to the determination of HULIS in environmental samples, their analysis poses still problems. Nonselective detection procedures combined with the large sample amounts required for analysis are considered to be the major difficulties of the reported analytical techniques. However, the most important drawback of the currently applied methods for the determination of HULIS in airborne particulate matter is that in all procedures only spectroscopic properties of HULIS were used for quantification. To overcome this problem, response factors with reference samples were determined to correlate the UV signal of separated HULIS with the dissolved OC of the investigated samples.18 This type of quantification assumes that UV-absorbing compounds are representative for the whole water-soluble fraction of the organic aerosol and that the chemical composition of the investigated samples is similar to that of the reference samples. Until now, no method has been reported for the direct determination of the HULIS carbon in environmental samples such as airborne particulate matter, cloudwater, fog, or rain. Here we present a procedure for the selective determination of HULIS in environmental samples, which is based on a C18 solidphase extraction followed by an automated ion chromatographic separation of the enriched species with subsequent carbon-specific detection of the isolated HULIS. The method is applied for the analysis of HULIS in aerosol samples from an urban site in Central Europe (Vienna, Austria). EXPERIMENTAL SECTION Reagents and Standard Solutions. High-purity and organicfree water was obtained from a Millipore ultrapure water system (Milli Q-plus 185) fed with distilled water. Nitric acid, sodium hydroxide, ammonia, and methanol were of p.a. grade purity (Merck, Darmstadt, Germany). The water-soluble fraction of a commercially available humic acid (Fluka, Buchs, Switzerland) has been used as a model substance for method development. This approach is a suitable way to overcome the problems associated with the absence of proper HULIS standard solutions, since HULIS have been reported to have chemical characteristics very similar to naturally occurring humic acids.7 A humic acid stock solution was prepared by dissolving a certain amount of dried (for 24 h at 105 °C) humic acid in organic-free water. To ensure a complete dissolution of the water-soluble fraction, ultrasonic agitation was used. After sedimentation of the remaining particles, the solution was filtered through a syringe filter of 0.2µm pore size (Whatman, Anotop 25). Finally, the carbon content of the derived clear solution was determined with a dissolved organic carbon analyzer (described in the next section). The prepared humic acid stock solution was stored in the refrigerator at 4 °C until time of use. Low concentrations of standard solutions were prepared daily by dilution of this stock solution with water. Instrumentation. C-18 Polygosil solid-phase extraction (SPE) cartridges (IST, 221-0020-H) have been used for the enrichment of HULIS and other weakly polar compounds. For the separation of HULIS from the remaining sample constituents, a flow injection
(FI) system consisting of two six-port injection valves (VICI, Cheminert C22), two HPLC pumps (Merck Hitachi L-6000), and an additional peristaltic pump (Ismatec, Germany) was used. Teflon microcolumns (inner diameter l.0 mm, length 15.0 mm) filled with ∼10 µL of a strong anion exchanger (Isolute SAX, IST 500-0020-H, average particle diameter 55 µm, mean pore size 54 Å) and sealed with porous disks were used as separation columns. The connections between the individual parts of the FI system were made with PTFE tubes having an inner diameter of 0.7 mm. To minimize the dead volumes, all tubes were kept as short as possible. For the on-line determination of the separated HULIS, two different types of detectors connected in series were used. A Kontron Uvikon 720LC UV detector was used to monitor the absorbance at a wavelength of 254 nm. Carbon-specific detection was performed using a standard method for the determination of dissolved OC in liquid samples.19 It is based on a commercial instrument (GOTOC 100, Gro¨ger & Obst) that operates with a continuous flow rate of 0.6 mL/min. The liquid samples are combusted in a catalytic combustion furnace operating at 800 °C; the carbon dioxide resulting from the combustion is determined using a nondispersive infrared detector. Using a six-port injection valve fitted with a 300-µL loop for sample introduction and 0.01 M nitric acid as carrier solution, a quantification limit of 0.4 µg of C/mL was achieved, which was calculated as the amount of analyte necessary to yield a peak area equal to the sum of the blank signal plus six times (6σ) the standard deviation of the blank signals (organic-free water). Enrichment and FI Separation procedure. A C-18 SPE procedure was used for the separation of HULIS and other weakly polar compounds from aqueous sample solutions. Since polyacidic organic compounds were usually present in the ionic form under neutral or basic conditions, an acidification of the sample solution is necessary to convert them into the protonated form, leading to a more hydrophobic behavior, which enables the isolation of these compounds from the aqueous solution using SPE.14 Prior to sample application, a pretreatment of the sorbent material was necessary. For conditioning of the used C-18 cartridge, 5 mL of organic-free water, followed by 3 mL of methanol, were aspirated through the SPE column. The preparation of the SPE cartridge was finished by purging with another 5 mL of organic-free water. After this conditioning step, the prepared sample solution was aspirated through the sorbent column with a flow of ∼2 mL/min. The aqueous solution passing the SPE cartridge contained those compounds which were not retained on the column such as inorganic ions and organic compounds, which were still hydrophilic due to the presence of highly polar functional groups (e.g., short-chain monocarboxylic acids such as acetic acid). The adsorbed fraction, including HULIS and other polar organic aerosol constituents such as long-chain monocarboxylic acids, aromatic alcohols, or aldehydes, was eluted from the SPE cartridge with pure methanol. The HULIS content of the eluate was determined using a FI procedure coupled to a detection system consisting of an UV detector and a dissolved OC analyzer for carbon-specific detection. To achieve a separation of the humic-like fraction from the solvent (methanol) and the remaining organic sample constituents, the
(18) Decesari, S.; Facchini, M.; Matta, E.; Lettini, F.; Mircea, M.; Fuzzi, S.; Taggliavini, E.; Putaud, J. Atmos. Environ. 2001, 35, 3691-3699.
(19) Bauer, H.; Kasper-Giebl, A.; Zibuschka, F.; Hitzenberger, R.; Kraus, G.; Puxbaum, H. Anal. Chem. 2002, 74, 91-95.
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7289
Table 1. Operating Sequence of the FI On-Line System (Sample Volume ∼0.5 mL) FIAS stepa
time (s)
1 2 3 4
30 175 60 90
pump P1 (mL min-1) 0.60
pump P2 (mL min-1)
pump P3 (mL min-1)
0.60 0.60 0.60 0.60
2.0
valve V1
valve V2
fill/inject load fill/inject fill/inject
fill/inject load fill/inject fill/inject
a (1) Nitric acid is moved into the FI system; thus, eluent loop, microcolumn, and detection system are rinsed with nitric acid, and sample solution is directed into sample loop. (2) NH4OH is pumped into the FI system and moves the sample solution onto the microcolumn; the stream of nitric acid present in the eluent loop is moved into the sample loop and subsequently through the microcolumn; thus after sample loading, the microcolumn is rinsed with nitric acid, simultaneously the eluent loop is filled with NH4OH, and in the meantime the detection system is continuously purged with nitric acid. (3) Nitric acid is pumped into the FI system and moves the content of the eluent loop (NH4OH) through the microcolumn. Thereby retained HULIS are eluted from the microcolumn. (4) The eluate is transferred into the detection system, and concurrently, the FI system including the microcolumn is rinsed with nitric acid.
Figure 1. FI manifold for on-line matrix separation and HULIS detection: P1 and P2, HPLC pumps; P3, peristaltic pump; MC, microcolumn; V1 and V2, six-port valves; SL, sample loop; EL, eluent loop.
solution derived from the SPE procedure was acidified with nitric acid, diluted with organic-free water, and subsequently pushed through a strong anion exchanger (SAX). Thereby HULIS and other compounds capable of carrying a negative charge were extracted from the sample solution, whereas neutral compounds passed the column without retention. After sample loading, a wash step with diluted nitric acid was performed to separate anionic species bearing only one or two charges per molecule (e.g., monoor dicarboxylic acids) from polycharged HULIS. Subsequently, the remaining fractionswhich is assigned to humic-like substancesswas eluted from the SAX column with NH4OH and directly introduced into the detection system. To exclude sorption of carbon dioxide from ambient air, which is a prerequisite for an accurate carbon measurement, and to improve the reproducibility of the procedure, a fully automated FI procedure has been used for HULIS isolation and detection. The setup of the developed FI manifold is depicted in Figure 1. In the fill/inject position, the sample loop was filled with the sample solution, whereas the SAX microcolumn and the detection system was rinsed with 0.01 M nitric acid. The peristaltic pump placed behind the sample loop was used to aspirate the sample solution into the loop; thus, the sample volume used for analysis was not in contact with the pump tubes, which reduces the possibility of sample contamination and analyte losses considerably. In the sample load position, 0.05 M NH4OH moved the nitric acid present in the eluent loop into the sample loop, thereby the content of the sample loop was pushed through the microcolumn. Due to their negatively charged groups, HULIS were retained on the anion exchanger, whereas sample constituents that did not interact with the sorbent material passed through the column. Sample application was followed by a wash step performed to remove the residual matrix from the sorbent material. As wash solution acted the stream of nitric acid, which was moved through the column by NH4OH. Before the NH4OH solution reached the 7290 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
end of the sample loop, the valves were turned back to the fill/ inject position. In the last step, the content of the eluent loop was pumped by nitric acid through the column leading to an elution of HULIS from the microcolumn. The eluted HULIS were directly introduced into the detection system. The detailed FI operating sequence is described in Table 1. RESULTS Enrichment on the SPE Column. Several investigations were performed to optimize the conditions for sample application and elution. In a first series of experiments, the dependence of the humic acid retention from the sample pH was examined. The influence of the sample pH was investigated at a humic acid concentration level of 40 mg of C/L, the applied sample volume was 10 mL, and the sample pH varied in the range from 1 to 7. To avoid acid hydrolysis of the dissolved humic acid, the synthetic sample solution was acidified with nitric acid just before application. During sample application, a yellowish band appeared at the top of the sorbent column, whereas the color of the remaining stationary phase was unaltered, indicating that the analyte was effectively retained in the first part of the column. The retained material was eluted from the dried column with 1 mL of methanol; quantification was performed by measurement of the UV absorption. In all experiments, the elution of the retained material was incomplete; a part remained on the column by irreversible adsorption, causing a permanent discoloration of the stationary phase. Best results were obtained for a sample pH of 3, which lead to a recovery of ∼70%. Since the effluent contained less than 5% of the applied humic acid, it was concluded that the remaining fraction was permanently adsorbed on the column. However, an unneeded excess of methanol for sample elution should be avoided, since the solvent must be separated prior to the carbonspecific detection. Therefore, the smallest amount of methanol necessary for quantitative desorption was determined. The elution process was investigated with humic acid standard solutions containing 40 mg of C/L at a sample pH of 3. It was observed that elution with 200 µL of methanol is sufficient to reach maximum recovery. Identical results were achieved for experiments performed with increased eluent volumes.
FI Procedure. To find optimal conditions for the retention of HULIS on the SAX column, a washing step of the stationary phase to overcome matrix effects, and an effective desorption of the analyte from the separation column several investigations were performed. All of the necessary experiments were carried out with a sample volume of 5 mL at a humic acid concentration level of 11.8 mg of C/L. Prior to solid-phase extraction, the sample pH was adjusted to pH 3 by the addition of 1 mL of 0.005 mol/L nitric acid. The humic acid retained on the C-18 column was eluted with 200 µL of methanol. The derived extract was diluted with organicfree water to a final volume of 2 mL. From this solution, 500 µL was subjected to the FI procedure. Elution of the humic acid from the SAX microcolumn was performed with 0.2 mol/L NH4OH. Possible changes of individual parameters were stated at the respective investigations. In a first series of experiments, the influence of the sample pH on the efficiency of the humic acid retention on the SAX microcolumn was determined. The pH of the methanol solutions derived from the SPE procedure was adjusted to acidic or basic values by addition of nitric acid and sodium hydroxide solutions, respectively. Afterward, the sample solution was diluted with organic-free water to a final volume of 2 mL. Best results were obtained for strong acidic solutions (pH 1-3); for reduced acidities, a decrease in the recovery was observed, reaching a minimum for neutral solutions (pH 6-7). Basic sample solutions (pH 8-10) yielded a slightly improved humic acid retention, but compared to strong acidic conditions, the results were reduced by a factor of 2. Subsequently the influence of the sample loading rate onto the microcolumn was determined in a range from 0.3 to 0.6 mL/min, indicating that the effectiveness of the procedure is not influenced by this parameter. The investigation of higher sample loading rates, which would shorten the whole determination, was not possible due to the limited uptake rate of the carbonspecific detector. Prior to analyte elution, a washing step was performed to remove the residual matrix and remaining traces of methanol from the column, which could interfere the final determination of HULIS in real samples with the carbon-specific detection unit. The effect of column rinsing was investigated at a humic acid concentration level of 11.8 mg of C/L; the nitric acid content of the applied wash solution varied in the range from 0.001 to 0.1 mol/L. Losses due to analyte leaching from the microcolumn did not occur, since for up to the maximum monitored nitric acid concentration and a final volume of 500 µL of wash solution, no decrease of the humic acid signal was observed. For elution of the retained humic acid from the SAX microcolumn, several solvents were investigated. With nitric acid solutions ranging from 0.1 to 2.0 mol/L, only a partial elution was achieved. Similar results were observed for basic solutions with varying amounts of sodium hydroxide. NH4OH was found to elute the adsorbed humic acid more effectively from the sorbent material, but still a part of the applied sample remained on the stationary phase, leading to a discoloration of the SAX microcolumn. Since concentrated NH4OH solutions are known to absorb carbon dioxide from ambient air, which could interfere the subsequent analysis of the eluted HULIS with the carbon-specific analyzer, it was necessary to decrease the NH4OH concentration to enable an accurate detection of HULIS in real samples. From investigations conducted with NH4OH concentrations ranging from 0.01 to 0.2 mol/L, we
determined that 0.05 mol/L NH4OH was sufficient to achieve the sharp elution peaks required for a sensitive detection. In addition, it was observed that the uptake of CO2 is distinctly reduced at this NH4OH concentration, which leads to improved limits of detection and quantification, respectively. Another benefit of the reduced eluent concentration is the increase in the lifetime of the SAX microcolumn from ∼50 to more than 100 measurements. Analytical Performance. For statistical characterization of the developed method including the SPE and the FI procedure, multiple determinations (n ) 4) of blank solutions and humic acid standards with varying concentrations (3-23.7 mg of C/L) were performed. An excellent linearity (r2 ) 0.999) between signal response and analyte concentration was found over the whole investigated concentration range for the UV as well as for the carbon-specific detection, indicating that the performance of the procedure is not influenced by the irreversible adsorption of analyte on the SAX microcolumn. The reproducibility given as the relative standard deviation was determined at a humic acid concentration level of 11.8 mg of C/L (n ) 10). With UV detection, a relative standard deviation of not more than 1.8% was achieved, a value that is comparable with the results reported for other FI procedures with on-line UV detection of analyte traces.21,22 A value of 4.3% was obtained for the carbon-specific detection. This minor decrease in the reproducibility is mainly caused by the solid-phase extraction. Investigations with blank solutions showed that this step of the procedure increases the carbon content of the sample, whereas the UV signal was unchanged, indicating that the adsorption properties of the blank solutions were not influenced by the SPE procedure. The limit of detection and the limit of quantification, respectively, were calculated as the amount of analyte necessary to yield a peak area equal to the sum of the blank signal plus three times (3σ) and accordingly six times (6σ) the standard deviation of the blank signals. Using a sample volume of 5 mL for analysis, a limit of detection of 1.0 mg of HULIS C/L was obtained for the carbon-specific analyzer. The quantification limit was determined with 1.6 mg of HULIS C/L, which corresponds to an absolute amount of ∼8 µg of humic-like material dissolved in a volume of 5 mL. Measurement of the UV signals revealed a detection limit of 0.08 mg of HULIS C/L and a quantification limit of 0.10 mg of HULIS C/L, respectively, for the UV detection. The influence of different sample volumes on the analytical performance of the proposed method was investigated with sample volumes in the range of 2-10 mL. The determined linear correlation (r2 ) 0.989) between the obtained detection signals (UV absorption and carbon-specific detection) and the used sample volume indicates that analyte losses can be excluded up to a sample volume of 10 mL. Since the determined statistical functions showed no dependence on the used sample volume, a further decrease of the detection limits is possible by increasing the sample volume used for the SPE procedure. General problems that may occur in on-line column separation and preconcentration systems are memory effects and analyte losses caused by the adsorption of compounds to pump tubes, (20) Kiss, G.; Varga-Puchony, Z.; Hlavay, J. J. Chromatogr., A 1996, 725, 261272. (21) Ruzicka, J.; Hansen, E. Flow Injection Analysis, 2nd ed.; Wiley: New York, 1988. (22) Trojanowicz, M.; Worsfold, P.; Clinch, J. Trends Anal. Chem. 1988, 7, 301305.
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7291
Table 2. Separation of Potentially Interfering Compounds C18 SPEa compound humic acid acetic acid citric acid lactic acid oxalic acid succinic acid adipic acid phthalic acid phenol 1,3-dihydroxynaphthalene 4-hydroxybenzoic-acid 2,5-dihydroxybenzoic acid levoglucosan
effluent
extract
FI-SAXa effluent
wash step
O
O
O BDL BDL CT BDL BDL BDL BDL BDL CT
O
O
CT
O
O
CT
O O O O
extractb
O O O O O O O
O
O O O O O O
O
O
BDL
a
O, indicates the fraction in which the major part of the investigated compound was found. b BDL, indicates that the observed signal was below the detection limit. CT, indicates that traces of the investigated substance were detected in the isolated extract.
valves, and other parts of the FI manifold.23 Since no increase of the blank signal occurred after multiple measurements of a humic acid standard solution containing 11.8 mg of C/L followed by measurements of a blank solution, it was concluded that the performance of the proposed procedure was not affected by such effects. Another serious problem of preconcentration and separation techniques are potentially interfering compounds, which could influence the performance of the whole procedure. To demonstrate the selectivity of the proposed method, synthetic solutions containing only one potential interference at a time (with a concentration of 40 mg of C/L) were analyzed. The compounds selected for this interference study (see Table 2) contained one, two, or more of the functional groups typically present in HULIS,6,7,24 some of these components are major contributors to the organic carbon content of atmospheric aerosols.1-3,25 For most of the investigated compounds, the signal obtained in the isolated HULIS fraction was not distinguishable from that of blank solutions (Table 2), indicating that these components were effectively separated prior to sample detection. Although the signals observed for lactic acid, 4-hydroxybenzoic acid, 2,5dihydroxybencoic acid, and 1,3-dihydroxynaphthalene were slightly increased compared to that of blank solutions, their contribution to the HULIS signal in real samples is expected to be negligible, since the concentrations used in the experiments were distinctly higher than reported for environmental samples. This assumption has been confirmed with another set of experiments conducted at a reduced concentration level of 4 mg of C/L, which yielded for all investigated compounds signals below the detection limits of the used detection systems. Another outcome of this interference study is that losses of highly polar HULIS components due to incomplete retention on the C18 column could be excluded, (23) Welz, B.; Yin, X.; Sperling, M. Anal. Chim. Acta 1992, 261, 477-487. (24) Decesari, S.; Facchini, M.; Fuzzi, S.; Tagliavini, E. J. Geophys. Res. 2000, 105, 1481-1489. (25) Fine, P. M.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 2001, 35, 2665-2675.
7292 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
since under the used conditions even strong polar compounds such as lactic acid were partially retained on this sorbent material (Table 2). Owing to the retention principle of the SAX microcolumn, the possibility of HULIS losses on this stationary phase is rather unlikely; even less charged compounds such as oxalic acid were completely retained. Therefore, it was necessary to introduce the wash step with nitric acid to elute the compounds bearing only one or two charges per molecule prior to the carbon-specific detection of the multiple charged HULIS. The accuracy of the proposed method was checked with the standard addition method by determining a defined amount of humic material in a real snow sample. Snow collection was performed at the Institute of Chemical Technologies and Analytics building, a site particularly influenced by traffic in downtown Vienna. Two samples with a sample intake of ∼10 mL of melted snow were analyzed using the procedure described above. Quantification of the derived absorption signals was based on a calibration function for the complete procedure that has been determined with humic acid standard solutions. The standard additions method for calibration was applied by adding diluted humic acid standard solution to a second set of melted snow samples. From the difference between the intensities of the analyte signal measured in the spiked and the previously measured unspiked sample solutions, which is equal to the added amount of humic acid, the concentration of humic-like material in the original snow sample was calculated. Since the results obtained with the standard additions method (1.03 ( 0.04 mg of HULIS C/L) showed very good agreement with the findings derived from the external calibration (1.06 ( 0.05 mg of HULIS C/L), it was concluded that the presence of potentially interfering matrix does not influence the accuracy of the method. Determination of HULIS in Aerosol Samples. PM10 samples were collected from June 1999 to May 2000 in a park next to the central hospital at an inner district site of Vienna, Austria. Aerosol collection was performed on a daily basis with a Digitel DA80H Hi Volume sampler. Precombusted quartz fiber filters (Pallflex, Tissue Quartz 2500 QAT-UP, 150-mm i.d.) were used as sampling substrates. The air volumes collected within 24 h were in the range of 670-720 m3 STP (273 K, 1013 h Pa). To exclude any contamination from the sampling procedure itself, field blanks were collected by purging sample air for 30 s through the filter. After sampling, the filters were stored in Petri dishes with a Parafilm cap in the refrigerator at 4 °C. A more detailed description of the sampling site and the used sampling procedure has been presented recently.26 To investigate the seasonal pattern of the water-soluble HULIS concentration in the PM10 aerosol, weekly averaged samples were constructed by combining single spots (i.d. 17 mm) of samples collected for 24-h periods. Using this procedure, two pooled samples per week (one for analysis and one for replicate analysis) were prepared. The sample volumes of the combined weekly samples varied in a range from 69 to 74 N m3 STP. To obtain the water-soluble fraction of the HULIS, the pooled filter punches were extracted two times with organic-free water using ultrasonic agitation. After addition of nitric acid to the combined extracts (for pH adjustment), the derived sample solution was diluted with (26) Hauck, H.; Berner, A.; Frischer, T.; Gomiscek, B.; Kundi, M.; Neuberger, M., Puxbaum, H.; Preining, O.; AUPHEP-Team, Atmos. Environ. 2004, 38, 3905-3915.
and a summer minimum for HULIS concentrations was found. For the cold season of the year (October 1999 to March 2000), a mean HULIS concentration of 0.63 µg of C/m3 (n ) 26) was calculated, whereas for the warm period (June to September 1999 and April to May 2000), a mean concentration of 0.31 µg of C/m3 (n ) 23) was determined. The averaged contribution of watersoluble HULIS to the organic fraction in PM10 was 13.7% (n ) 26) in winter and 10.3% (n ) 23) in summer, which confirms that HULIS are a major contributor to the organic carbon in atmospheric aerosols during all seasons.
Figure 2. Correlation of water-soluble HULIS concentrations derived from UV detection and carbon-specific detection, Results for January and February are not included due to problems with the UV detection system.
Figure 3. Weekly averaged PM10 concentration profiles for watersoluble HULIS carbon at the sampling site next to the central hospital in Vienna, contribution of the water-soluble HULIS carbon to the organic carbon.
organic-free water to a final volume of 10 mL and analyzed using the developed procedure. The content of water-soluble HULIS C determined in the different aerosol samples varied from 0.11 to 1.78 µg/m3 with a mean concentration of 484 ng/m3 (n ) 49), which agrees with previous findings for urban sites.7,18 Comparing the results observed for UV detection with corresponding data from the carbon-specific analyzer, it was observed that the UV signal underestimates the content of HULIS carbon in the watersoluble fraction by a factor of ∼2 (Figure 2). This result might be explained with differences in the UV-absorbing properties between the investigated PM10 samples and the humic acid standard material used for calibration. Interpreting the seasonal variation of water-soluble HULIS in PM10 (Figure 3), a winter maximum
CONCLUSIONS In this work, a method based on an off-line C18 SPE column preseparation and a microcolumn FI system coupled with a detection system for the carbon-specific determination of HULIS in atmospheric samples is presented. The method is applicable for precipitation samples as well as for aerosol samples. The obtained detection limits are among the lowest known values in the literature for humic-like substances with respect to the small sample volumes used for analysis. The efficient sample utilization enables even for samples with restricted sample amounts the direct measurement of the HULIS carbon. The main advantage of the developed method is the considerable chemical selectivity that was achieved by the combination of two separation steps with different mechanism. Thus, the accuracy of the proposed method is not affected by the presence of potentially interfering organic matrix, which makes the presented approach suitable for the determination of HULIS in environmental samples such as snow, rain, or fog and similar matrixes such as airborne particulate matter. The method has been applied to the analysis of watersoluble HULIS in aerosol samples collected in a park next to the central hospital in Vienna, Austria. The content of water-soluble HULIS in PM10 varied from 0.11 to 1.78 µg of C/m3 (n ) 49) in weekly averaged samples and showed a clear seasonal variation with enhanced concentrations in the cold period of the year and reduced levels in the warm time of year. ACKNOWLEDGMENT The authors thank the AUPHEP team for assistance with sample collection and sample pretreatment. Financial support by the regional government of Vienna, Project-AQUELLA, is gratefully acknowledged. Received for review May 31, 2005. Accepted August 1, 2005. AC050953L
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7293