Isotope fractionation leads to unequal distribution of heavier and lighter isotopes between the reactants and products of a chemical reaction since the lighter isotopes often react faster while the heavier isotopes remain in the residual substrate. This is a well-known phenomenon for CO2 fixation during photosynthesis,1 methanogenesis,2 and methane oxidation.3, 4 Rudolph et al.5 also described a kinetic fractionation effect on the stable carbon isotopes of organic compounds during the oxidation by OH radicals in the atmosphere, similar to the study by Iannone et al. for ozone.6 This isotope effect can be used to study the different processes that contribute to the atmospheric concentration of carbonaceous aerosols. However, only little information is available on the stable carbon isotopic composition of carbonaceous aerosols and their gas-phase counterparts in the atmosphere.7, 8 This is largely due to the lack of a coherent method to measure the δ13C signature of organic samples from the atmosphere since their concentrations are much smaller compared to those in other media like water or soil. Recently, some effort has been made to measure the δ13C values of organic compounds. For example, Simoneit9 described the use of gas chromatography (GC)-interfaced isotope analysis of organic compounds. However, this method cannot be used for water-soluble organic carbon (WSOC), e.g. organic acids, due to difficulties in separating polar organic compounds using GC. Often ion chromatography (IC) or liquid chromatography (LC) is used to measure polar organic compounds; these methods include a step where the compounds are dissolved in water. Determining the δ13C value of the water-soluble fraction, however, needs conversion of organic carbon into carbon dioxide gas (CO2) prior to isotopic analysis using isotope ratio mass spectrometry (IRMS). The new online coupling of the IRMS instrument commercially available for LC systems (Isolink, Thermo-Finnigan) may boost the potential of isotope studies for liquid samples. However, its suitability for IC has not yet been established.
The wet oxidation method using sodium persulfate as oxidizing agent is the most commonly used method to convert organic carbon into CO2 for δ13C measurements:10, 11
At temperatures exceeding 100°C, sodium persulfate decomposes to [SO]− (sulfate radical) and readily reacts with organic compounds to form CO2. The δ13C of the resulting CO2 is then measured using IRMS.
In this study we adapted the wet oxidation method for compound-specific stable carbon isotope analysis of organic acids in the atmosphere for both the gas and the aerosol phase, as well as for water-soluble organic carbon (WSOC) in aerosols obtained from filter samples. Using an IC system interfaced with an automated fraction collector, different organic acids were sampled separately and oxidized to CO2 using the oxidation method for subsequent δ13C measurements.
Sampling of organic acids
Sampling of organic acids was made using a wet effluent diffusion denuder/aerosol collector (WEDD/AC) coupled to an IC system. Briefly, the WEDD/AC is a custom-built instrument used for sampling water-soluble ionic species in both the gas and the aerosol phases. It consists of a glass denuder and an aerosol collector. The denuder provides a continuously wetted sampling surface for the gaseous compounds, which diffuse to the walls and are absorbed by the flowing liquid. The aerosols, on the other hand, pass through the denuder and enter the aerosol collector connected to the denuder. There, a continuous flow of steam (100°C) is supplied to transform the aerosol into cloud droplets which are then collected at the end of the aerosol collector. The resulting effluents from both the denuder and the aerosol collector are concentrated on a trace ion concentrator (TAC-LP1). The concentrated ions are then measured using an IC system (Dionex, DX600) equipped with a conductivity detector. A mixture of acetic and formic acid from a permeation source was passed through the denuder to test performance. The collection efficiency of similar systems has been found to be greater than 99%.12 Even though losses on the surface of the sampling line and some uncertainty due to the aging of the permeation source lowered the collection efficiency of the denuder in our system, it was calculated to be at least 85 and 75% for acetic and formic acid, respectively. A more detailed description of the WEDD/AC and IC is given elsewhere.13, 14 The IC effluent is directed to an Isco Foxy 200 fraction collector equipped with two sample trays to collect individual acids separately into glass vials, using pre-programmed internal software that is integrated in the software of the IC system (Chromeleon). Hence, fractions of chromatographically separated organic acids are collected based on their peak detection.
The objective of this study was to further develop the wet oxidation method for atmospheric applications. Therefore, the data presented here are only for standard solutions of organic acids prepared from salts of the corresponding acids. Data on organic acids derived from atmospheric measurements using this method are presented elsewhere.15
Sodium persulfate (oxidation reagent)
A 0.15 mol/L solution of sodium persulfate (Sigma-Aldrich, S-6172) solution was used.16
Organic acid standards
Standard solutions with different concentrations were prepared from their corresponding salts. The organic acids used for this study were acetic, formic, oxalic, and humic. The humic acid was a sodium salt (NaHA, product No. H16752) from Aldrich, which is isolated from crude lignite.17 Diesel fuel was also analyzed. All chemicals used were analytical grade, and ultrapure water (MQ water) (18M Ω · cm) was used throughout the experiment.
Aerosol samples were collected on quartz filters, and different fractions of the carbonaceous aerosols were separated and analyzed as follows. Two filter punches of 14 mm diameter were taken from the whole filter. For the total carbon (TC) measurement, the first filter punch was packed into a tin cup and combusted with an elemental analyzer (EA-1108, Carlo Erba).18 For the determination of water-insoluble carbon (WINSC), the second filter was soaked in MQ water in an ultrasonic bath for 20 min. The filter was then taken out of the water, dried by putting in the hood overnight, and packed into a tin cup for combustion (details of the filter sampling are described elsewhere15). The difference between TC and WINSC was taken as water-soluble organic carbon (WSOC). WSOC was also measured from the filter extract using the wet oxidation method as described below. The results from the oxidation method were compared with those from the combustion method.
Wet oxidation process
The wet oxidation process consists of two steps, a pretreatment and an oxidation step. For the pretreatment, 3 mL of the prepared sodium persulfate solution were placed in a reaction vial (borosilicate glass vial from Schütt Labortechnik, article no. 3.560 103) and left for 45 min boiling on a hot plate. This step was performed to remove the organic compounds that could be present in the reagent water. The vials were then removed from the hot plate, the CO2 was pumped out with a vacuum pump, and the vials were subsequently filled with helium using a three-way valve. For the oxidation step, 2 mL of the standard or sample solution were injected into the pretreated persulfate solution using a syringe. Then the vials were placed again on the hot plate for 25 min to convert organic carbon in the solutions into CO2. Total analysis time (pretreatment + CO2 removal + sample oxidation) was on average 1.5 h per batch (15 samples in each batch). Blanks were prepared in the same manner as the samples except that no standard solution was injected into the pretreated sodium persulfate solution. The amount of carbon, given as CO2, was determined from the peak area measurement in the mass spectrometer.
δ13C values of the filter samples were determined using a DELTA-S Finnigan MAT isotope ratio mass spectrometer coupled to the elemental analyzer via a Conflo II interface (instrument precision 0.1–0.2‰ depending on the sample size). δ13C values of the salts of the organic acids were also measured by weighing 0.6 to 0.8 mg of the salts into small tin capsules and combusting them to CO2 with the elemental analyzer. The δ13C values of CO2 derived from the wet oxidation were measured using a Delta Plus XL mass spectrometer (Finnigan, Germany) coupled to the Gasbench II peripheral (Finnigan, Germany) and equipped with autosampler.
The stable isotope distribution is expressed in the conventional δ (delta) notation, the relative difference between the isotope ratios of sample and a standard, and is expressed as:
where R is the ratio of the abundances of the heavier to the lighter isotope. The composition of the sample is expressed with respect to the reference material Vienna PeeDee Belemnite (VPDB), which by definition has a δ13C value of 0‰.
RESULTS AND DISCUSSION
Blanks were analyzed to determine any contribution of carbon due to contamination. While the blank amount before evacuation was about 1 µg, the average blank concentration measured after evacuation was 200 ± 50 ng carbon (corresponding to an area of 0.42 Vs) and had a δ13C value of −21.2 ± 1.7‰ (n = 15). The δ13C of the blank indicates that the carbon of the blank was not due to a leak, which otherwise would have δ13C values similar to that of atmospheric CO2, i.e. −8‰. The δ13C value of impurities present in the MQ water (without persulfate) was also measured to identify the source of the carbon in the blank. The amount of carbon measured from the MQ water was often below 50 ng carbon. From this, it was concluded that the blank CO2 mainly resulted from the oxidation of organic contaminants that may be present in the persulfate solution. Hence, the pretreatment time was optimized to produce lowest blanks by retaining the reactivity of the persulfate. The blank amount did not vary strongly as shown by the standard deviation (SD) of only about 50 ng. The detection limit, defined as 3 times the SD of the blank amount, is thus 150 ng.
Different types of septa were also tested to find the most suitable septum with the property to remain gas-tight for at least 12 h and to tolerate a temperature of 100°C during the pretreatment and sample oxidation. Moreover, the septum should also be soft enough for needle penetration in order not to damage the gas-sampling needle. The septum that met all the requirements was the Teflon-faced silicone septum (Supelco, part no. 27166). Since Teflon is chemically inert, it was placed facing inwards to be in contact with the persulfate. This minimized any possible reaction or contamination between the persulfate and the septum.
Table 1 shows the carbon isotopic composition of each organic acid measured as salt using combustion and their respective solutions using the oxidation method. The observed difference between the δ13C values of the salt measured with the combustion method and the solution measured with the oxidation method was less than 0.5‰, except for humic acid where the difference was 1.2‰. The latter could result from the heterogeneous properties of the humic acid used in this experiment (S. Sjögren, Paul Scherrer Institute (PSI), 2005, personal communication). Although slightly higher than the combustion method, the standard deviation of the oxidation method was found to be small (on average 0.4‰). The amount of carbon required for the combustion, however, was much higher (100 µg) compared to the oxidation method (150 ng). The method was also tested for reproducibility of the carbon content by preparing different concentrations of the organic acids. The correlation between the carbon content measured from the oxidation and the amount of carbon placed into the oxidation vial in the form of organic acid solution is shown in Fig. 1. These results demonstrate that on average the amount of carbon derived from oxidation was within 5% of the expected value for acetic and oxalic acids, while the accuracy and reproducibility for humic acid was worse (13%) due to the heterogeneity of the humic acid standard, as mentioned above.
Table 1. δ13C values against VPDB of the solid salts of the acids measured using the combustion method and δ13C values of the acid solutions derived by the oxidation method
δ13C from combustion (‰)
δ13C from oxidation (‰)
Average (SD (n = 3))
Average (SD (n = 5))
The effect of the blank on the δ13C signature of organic acids was further assessed by determining the δ13C deviation of the standards from the solid salts at small sample sizes. For standards containing less than 1 µg carbon, a deviation in δ13C values towards the corresponding blank values was observed. Therefore, a theoretical weighted average δ13C was calculated using the measured concentrations and the δ13C values of the blanks as follows:
where δ13CT = weighted average δ13C, δ13Cb = δ13C value of the blank, Ab = amount of carbon in the blank, δ13Cs = δ13C of the standard measured using combustion, As = amount of carbon in the standard: δ13CT was calculated for different concentrations of organic acids by varying the value of As. As demonstrated in Fig. 2, there is very good agreement between the measured δ13C and the calculated δ13CT, which implies that the blank correction for samples of less than 1 µg carbon is suitable.
A series of experiments was performed to determine whether the fraction recovery of the organic acids using IC has any effect on the δ13C value of the samples. This was done by passing the standard solutions of the organic acids (10 µg) through the IC system and then measuring the δ13C value of the recovered organic acids using the oxidation method. The results of this experiment are summarized in Fig. 3. In general, we found a good agreement between isotope values of the solid and their corresponding solutions and those of the IC-recovered samples. However, a slight deviation in the δ13C values of the solid acetic acid from the prepared acetic acid solution and from IC-recovered samples was observed (Fig. 3). This indicates contamination in the organic acid solutions, which may come from the atmospheric CO2 dissolved during solution preparation. This effect is pronounced for acetic acid due to the large difference between the δ13C of the atmospheric CO2 (∼−8‰) and δ13C of acetic acid. This effect can be minimized by using an environment that is free of atmospheric CO2 during solution preparation and fraction collection, by flushing helium or nitrogen over the solution. Similarly, the blank in the persulfate solution could also contribute to the shift of acetic acid due to the δ13C difference of the blank and acetic acid.
WSOC from filters
The WSOC results obtained from the oxidation method were compared with those from the combustion method. The amount of WSOC from the filters derived from combustion was calculated from the difference between TC and WINSC. Less than 10% difference was observed between the two methods (Fig. 4). The isotope value of WSOC is derived in an analogous way as Eqn. (1) by assuming TC to be the weighted average of WSOC and WINSC. The averaged δ13C value of the WSOC calculated for all the samples was −25.5 ± 0.6‰ for the combustion and −24.0 ± 0.7‰ for the oxidation process. The 1.5‰ difference between the two methods may come from components that were oxidized by the combustion but not by the wet oxidation process.
The wet oxidation method using sodium persulfate, presented here, shows a high reproducibility and precision for δ13C measurement of organic acids and water-soluble organic carbon (WSOC). Its applicability has been demonstrated to compound-specific analyses of organic acids and WSOC from the atmosphere. The proposed oxidation method has an accuracy of 0.5‰ with a precision of 0.4‰ for δ13C, and can be used to determine the δ13C values of WSOC with a minimum carbon content of 150 ± 50 ng. However, for samples with a carbon content of less than 1 µg, a blank correction has to be made using the weighted mean average. With further experimental procedures, such as cryogenic purification, we believe that this method can be used to determine the concentrations and δ13C signatures of other fractions of carbonaceous aerosol such as water-insoluble organic carbon and black carbon.
This work was supported by the Swiss National Science Foundation.