1353
Electrophoresis 2014, 35, 1353–1361
Monique Teich Dominik van Pinxteren Hartmut Herrmann Chemistry Department, ¨ Leibniz-Institut fur ¨ Tropospharenforschung (TROPOS), Leipzig, Germany
Received September 13, 2013 Revised November 28, 2013 Accepted December 2, 2013
Research Article
Determination of nitrophenolic compounds from atmospheric particles using hollow-fiber liquid-phase microextraction and capillary electrophoresis/mass spectrometry analysis A hollow-fiber liquid-phase microextraction method was developed to enrich nine nitrophenolic compounds from aqueous extracts of atmospheric aerosol particles. Analysis was performed by CE coupled with ESI MS. The BGE composition was optimized to a 20 mM ammonium acetate buffer at pH 9.7 containing 15% methanol v/v. Several extraction parameters (composition of organic liquid membrane, pH of acceptor phase, salting-out effect, extraction time) were investigated for their effect on the analyte recoveries. The donor phase consisted of a 1.8 mL sample solution kept at pH 2 while the acceptor phase was a 15 L 100 mM aqueous ammonia solution. Dihexyl ether served as supported liquid membrane. Low detection limits in the range of nanomole per liter were achieved. Recoveries of aqueous standard solutions were found to be between 11 and 90% with enrichment factors between 10 and 100. Interday and intraday repeatabilities were in an acceptable range for most compounds (6–15% and 7–10%, respectively) but somewhat higher for 4-nitrocatechol (59 and 48%) and 2-nitrophenol (17 and 35%). The developed method was found to be competitive with more established method and was successfully applied to samples of atmospheric particulate matter from field experiments. Keywords: Atmospheric particles / CE-MS / Liquid-phase microextraction / Nitrophenols DOI 10.1002/elps.201300448
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction Nitrogen-containing compounds represent an important fraction of aerosol organic matter (2–7%) [1]. Large amounts
Correspondence: Professor Hartmut Herrmann, Leibniz-Institut ¨ Tropospharenforschung ¨ fur (TROPOS), Permoserstr. 15, 04318 Leipzig, Germany E-mail: [email protected] Fax: +49-341-2717-99-7024
Abbreviations: DHE, dihexyl ether; 2,4DNP, 2,4dinitrophenol; 2,6D4NP, 2,6-dimethyl-4-nitrophenol; 3,4DNP, 3,4-dinitrophenol; EF, enrichment factor; ExCal, external calibration; HF-LPME, hollow-fiber liquid-phase microextraction; 2M4NP, 2-methyl-4-nitrophenol; 4M2NP, 4-methyl-2nitrophenol; 4NC, 4-nitrocatechol; 2NP, 2-nitrophenol; 3NP, 3-nitrophenol; 4NP, 4-nitrophenol; OC, organic carbon; ON, organic nitrogen; R, recovery; SA, standard addition; SLM, supported liquid membrane; TOPO, trioctylphosphine oxide; WSON, water-soluble organic nitrogen
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
are water-soluble and it has been reported that water-soluble organic nitrogen (WSON) can cover 20–80% of the total soluble nitrogen of aerosol particles [2–4]. Comparisons of organic nitrogen (ON) and WSON with organic carbon (OC) and water-soluble organic carbon showed ON/OC ratios of about 0.19 for marine aerosol [5]. Miyazaki et al. [6] found ON/OC ratios of 0.43 for submicrometer particles being influenced by high marine biological activity. At continental conditions, the average WSON/WSOC mass ratio was found to be 5% during measurements in southeastern United States [7]. Attention has been drawn to WSON because it can, in contrast to most of the other organic constituents, absorb ultraviolet and visible light, thus influencing the earth’s radiative balance. This characteristic makes WSON an important contributor to the so-called brown carbon [8]. Furthermore, organonitrogen compounds can act as cloud condensation nuclei without being mixed with inorganic salts [9]. WSON can be of various natural or anthropogenic origins. Biomass burning is considered to be a major contributor to nitrogen-containing compounds [8, 10–12]. Secondary
www.electrophoresis-journal.com
1354
M. Teich et al.
formation due to oxidation reactions in the presence of NOx with volatile organic compounds is also possible, followed by gas-to-particle condensation, which can lead to the formation of secondary organic aerosol [13, 14]. WSON is a complex mixture consisting of hundreds of different compounds. So far several classes of nitrogencontaining compounds were identified, such as aliphatic and aromatic amines, amino acids, nitriles, amides, hydrazines, nitrosamines, heterocyclic compounds, organonitrates, and nitroaromatic compounds [1, 10, 15–18]. Nitrophenols have recently received increasing attention [15]. They are known for their negative effect on human health [19] as well as their potential phytotoxity [20] and, thus, are responsible for forest decline in polluted areas [21]. Quantification of nitrophenols in atmospheric particles requires an appropriate separation technique as well as a preconcentration step prior to separation because of their low concentrations in the range of nanogram per cubic meter. In previous studies, mainly HPLC (e.g. [11, 15, 22]) or GC (e.g. [15,23]) have been used for the determination of nitroaromatic compounds in atmospheric particles, usually combined with solvent evaporation and/or derivatization. In this study, CE is used for separating nitrophenols as it provides high separation efficiency, low sample consumption, short analytical times, and low operation costs [24], making the CE-based technique an attractive alternative to HPLC techniques if competitive LODs can be achieved. The very low injection volume needed for CE is well suited for coupling with modern microextraction techniques, like the hollow-fiber liquid-phase microextraction (HF-LPME), which can provide extraction volumes in the range of a few microliters [25]. In several previous studies, various substituted phenols were studied in different environments using CE analysis, for example, nitrophenols in tap water, snow water, the Yangtze River water [26], and farm water samples [27] as well as methoxy phenols [28] and 4-nitrophenol (4NP) [29] in atmospheric particles. HF-LPME is a relative new membrane extraction technique providing easy handling, high enrichment factors (EFs), low organic solvent consumption, and low costs. In contrast to the more common SPE, no evaporation of organic solvent is needed and potential chemical losses due to evaporation can be avoided [25]. In principle, a water-immiscible liquid is used to form a supported liquid membrane (SLM) inside the pores of a hollow fiber [30]. After impregnating the hollow fiber, either an organic liquid (two-phase HF-LPME) or an aqueous solution (three-phase HF-LPME) is inserted into the fiber (acceptor phase) and afterward placed into a donor solution. The target compounds are extracted from the donor phase through the SLM into the acceptor phase where they are trapped. The recoveries (Rs) and EFs obtained are dependent on the partition coefficient of the analytes between the different phases and the volumes of the donor, liquid membrane, and acceptor phases [25]. Very low LODs can be achieved depending on the method parameters and the used instrument for determination. LODs range from a few nanograms per liter [31] C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2014, 35, 1353–1361
to several micrograms per liter [27], respectively, for nitrophenols extracted by HF-LPME and analyzed by HPLC or CE. Several studies have reported the extraction of various nitrophenolic compounds via HF-LPME from environmental water samples (lake water, river water, sea water, wastewater). Bishop and Mitra [31] determined nitrophenols in air using impinger sampling and subsequent preconcentration with HF-LPME. Those studies differ in their usage of the supported organic liquid membrane, acceptor phase, donorphase composition, extraction time, and LODs. Nitrophenols were also extracted by other kinds of SLM techniques, as reviewed by J¨onsson and Mathiasson [32]. Field and laboratory measurements of nitrophenols in the atmosphere have been reviewed by Harrison et al. [15]. The aim of this study was to optimize CE-MS parameters for good separation of all target compounds, to explore the potential of HF-LPME coupled with CE-MS for a simple and sensitive determination of nitrophenols from atmospheric particles, and to evaluate the performance of the developed method in comparison to available other techniques.
2 Materials and methods 2.1 Chemicals The chemicals used in this study were obtained from different suppliers as follows: 2-propanol and methanol (LC-MS Chromasolv, ࣙ99.9%), aqueous ammonia solution (NH3 (aq), 5 M), sulphuric acid (puriss. p.a., 95–97%), acetic acid (eluent additive for LC-MS), 1-octanol (⬎95.5%), 3,4-dinitrophenol (3,4DNP, ⬎99%), 2,4-dinitrophenol (2,4DNP, ⬎99%), 4NP (⬎99.5%), 3-nitrophenol (3NP, ⬎99%), 2-nitrophenol (2NP, ⬎99%), 4-nitrocatechol (4NC, ⬎98%) from Fluka (Munich, Germany); acetone (ACS reagent, ࣙ99.5%), undecane (ࣙ99%), dihexyl ether (DHE, ⬎97%), and trioctylphosphine oxide (TOPO, 99%); 2,6-dimethyl-4-nitrophenol (2,6D4NP, 98%), 2-methyl-4-nitrophenol (2M4NP, 98%), 4-methyl-2nitrophenol (4M2NP, 98%) from Sigma-Aldrich (Munich, Germany). EDTA (Titriplex II) was obtained from Merck (Darmstadt, Germany). Five millimolar stock solutions of each nitrophenol were prepared in 100 mM NH3 (aq) and stored at −20°C. Standard mixtures were prepared at micromolar levels in ultrapure water (Milli-Q, Millipore, Schwalbach, Germany) and stored for a maximum time of 4 weeks at 4°C. No degradation of analytes was observed during the storage period.
2.2 Atmospheric particle sampling and extraction Atmospheric particles were sampled at the research station Melpitz [33] in 2012 with a Digitel DHA-80 filter sampler (Riemer, Hausen, Germany). Melpitz is a rural site located about 50 km NE of Leipzig, Germany. The atmospheric particles with an aerodynamic diameter less than 10 m (PM10 ) www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1353–1361
were collected on quartz fiber filters (MK 360, Munktell, Falun, Sweden). Sampling time was 24 h (00:00 to 24:00 CET) at a flow rate of 0.5 m3 /min. To reduce the blank content of carbonaceous material, the filters were heated for 24 h at 105°C before sampling. After sampling the filters were stored at −20°C until extraction. For the analysis of watersoluble nitrophenolic compounds, 12 pieces (78.5 mm2 each) were punched out of one filter and placed into one disposable 5 mL syringe (Omnifix, Braun, Melsungen, Germany). The syringe was filled with 3 mL of MilliQ-water and shook for 2 h at 700 rpm on a laboratory shaker (KS 125, IKA, Staufen, Germany). After shaking, the aqueous particle extract was filtered through a precleaned syringe filter (0.45 m, Acrodisc 13, Pall, Dreieich, Germany). Aliquots of this solution were used for the HF-LPME experiments. Precleaning of the disposable syringes was carried out by ultrasonication in ultrapure water. The syringe filters were cleaned with methanol before use.
2.3 HF-LPME The HF-LPME device used is the same as described before [34]. Briefly, a 2 mL amber glass vial with a screw cap containing a silicone septum (Agilent, Waldbronn, Germany) served as container for the donor phase. A polypropylene hollow fiber (Accurel Q 3/2, 0.6 mm id, 1 mm od, 0.2 m pore size; Membrana, Wuppertal, Germany) was cut into pieces of 11 cm length and cleaned twice by ultrasonication in acetone for 15 min to remove possible contaminants originating from the polypropylene material. Afterward, the fibers were allowed to air dry. Two 1-cm-long pieces of Teflon tubing (1/16 in. od, 1 mm id) punched through the silicon septum served as guiding tubes to hold the hollow fiber into place. The fibers were threaded through the guiding tubes so that 1 cm protruded while shaking. Filling and discharging of the fiber was conducted with analytical syringes (1700 RN series, Hamilton, Bonaduz, Switzerland) and a tapered needle (0.64 mm od, gauge 23S, point style AS, Hamilton Bonaduz, Switzerland). The fiber was flushed with an organic liquid to build up the SLM inside the pores of the hollow fiber. As described by van Pinxteren et al. [34], contaminants could originate from the liquid membrane. Therefore, in a next step, the hollow fiber was filled with 100 mM NH3 (aq) and placed in a vial containing 1.8 mL of 100 mM NH3 (aq) for 15 min to eliminate contaminants. Afterward, the solution was removed by flushing the fiber with air and the precleaned fiber was filled with 15 L of the acceptor phase. NH3 (aq) served as the acceptor phase instead of NaOH because of its high volatility and MS compatibility. The filled hollow fiber was placed in a vial with 1.8 mL of the donor solution acidified to pH 2 with sulphuric acid and shaken for 60 min with a VXR basic Vibrax shaker (IKA) at 2000 rpm. The method optimization and validation were carried out with aqueous standard solutions. After extraction, the acceptor phase was removed and transferred into a polypropylene vial (Agilent) and subsequently analyzed by CE coupled with MS (CE-MS). C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1355
2.4 CE-MS conditions An Agilent 3D CE instrument coupled with an Esquire 3000+ ITMS (Bruker, Bremen, Germany) was used for analysis. ESI was applied as an interface. Precision-cleaved fused-silica capillaries (360 m od, 50 m id, 75 cm length, Polymicro Technologies, Phoenix, AZ, USA) were used to avoid unstable electrospray, which can be caused by poor handmade cuts. At both sides of the capillary, the polyimide coating was removed by an open flame (1–2 cm). The BGE consists of 20 mM ammonium acetate solution containing 15% v/v methanol adjusted to pH 9.7 using 5 M NH3 (aq). Hydrodynamic injection was done by applying a pressure difference of 50 mbar for 10 s to the inlet vial. At the beginning of each series of measurements, the capillary was flushed with 1 M NH3 (aq) (10 min) and BGE (30 min). Prior to each analysis, the capillary was flushed with 5 mM EDTA solution (5 min) and BGE (3 min). The usage of EDTA solution improved the peak shape of 4NC during analysis (see Section 3.1). After each run, a short plug of 1 M NH3 (aq) was injected to avoid carryover of analytes. The sheath liquid consisted of an isopropanol/water (80/20) mixture and the flow rate was set to 0.4 mL/h. The pressure of nebulizing gas was set to 4 psi (276 mbar) and the flow rate of the drying gas was 10 L/min at a temperature of 200°C. ESI was operated in negative mode with a voltage of 4.5 kV and an endplate offset of −500 V. All compounds were detected in their deprotonated form as [M−H]− . Samples were analyzed twice taking the average peak area for data interpretation.
2.5 Figures of merit Method validation was carried out using aqueous standard solutions for HF-LPME extraction. The EFs and Rs were calculated based on real concentrations. During optimization experiments, the peak areas instead of real concentrations were used to estimate the Rs. In general, the EF is given as EF = c(acceptor)/c(donor). c(Acceptor) stands for the concentration of analyte in the acceptor phase after extraction and c(donor) represents the initial concentration in the donor phase. R was calculated as: R=
n(acceptor) EF × V (acceptor) × 100 = × 100, V (donor) n(donor)
(1)
where V(acceptor) is the volume of the acceptor phase, V(donor) is the volume of the donor phase, n(acceptor) is the amount of substance of the respective analyte in the donor phase, and n(donor) is the initial amount of analyte in the donor phase. The intraday repeatability was characterized by 5 parallel extractions on one day and interday repeatability by running 14 extractions in total on four consecutive days. Both are expressed as RSD of the peak area. Repeatability of CE-ITMS is based on runs with standard solution having 120 times higher concentration than in the donor phase due to the volume ratio standard solution-to-donor phase. Instrumental LODs were estimated by measuring a series of diluted standard mixtures. HF-LPME LODs were determined by www.electrophoresis-journal.com
1356
M. Teich et al.
Electrophoresis 2014, 35, 1353–1361
Figure 1. Extracted ion chromatograms of target compounds after CE-MS separation. (A) BGE: 20 mM ammonium acetate (pH 9.1) with 10% v/v methanol; sheath liquid: isopropanol/water (80:20). (B) After optimization, that is, BGE: 20 mM ammonium acetate (pH 9.7) with 15% v/v methanol; sheath liquid: isopropanol/water (80:20).
extracting a series of diluted standard mixtures. The LOD was defined as the concentration where the corresponding peak in the extracted ion electropherogram shows a S/N of 3. Calculation of LOD in terms of atmospheric concentration was done based on the HF-LPME LOD, taking into account the sampled air volume, the volume of water for filter extraction, and the molar mass of the analytes.
3 Results and discussion 3.1 Optimization of separation conditions All experiments for optimizing separation conditions during CE-ESI-MS were carried out using a standard mixture with prescribed analyte concentration in the micromolar range. Previous works of our group demonstrated that for acidic compounds a BGE consisting of 20 mM ammonium acetate with additional 10% v/v of methanol at pH 9.1 gives good separation efficiencies [28, 35]. First, runs with aqueous standard mixtures under this condition are shown in Fig. 1A. No baseline separation could be achieved for 3,4DNP and 2,4DNP and no separation at all for 2M4NP and 3M4NP. The targeted 4NC showed a very broad peak and peak tailing. Further optimization of the separation conditions have been carried out to improve the separation efficiency. Thus, the composition of BGE was varied regarding the concentration of ammonium acetate (20–30 mM) and the methanol fraction (0–20%). Moreover, the BGE was tested at different pH values (9.0–10.0). Slight variations of the ammonium acetate concentration lead to no significant improvement in the resolution of the nitrophenols. Therefore, a 20 mM ammonium acetate BGE was kept for further analysis. However, baseline separation of all target analytes could be achieved by increasing the BGE pH and the fraction of methanol. The optimum was found to be a 20 mM ammonium acetate buffer at pH 9.7 with additional 15% v/v of methanol (Fig. 1B). After optimizing separation conditions, the peak-shape problems of 4NC remained. Similar problems with peak shape of 4NC were ob C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
served by Kitanovski et al. [36] during LC-MS analysis. They supposed that peak broadening is caused by interactions of 4NC with metal ions. Thus, EDTA as a complexing agent was applied to complex metal ions. Coinjection of EDTA resulted in a strong improvement of peak shape [36]. Based on this observation, the effect of EDTA solution on the peak shape of the nitrophenols was studied. The addition of small concentrations of EDTA to the BGE (0.05, 0.1, 0.25, 0.5 mM) led to a slightly improved peak shape, while adding EDTA directly to the standard mixture did not have any effect. Furthermore, the peak intensity of the other compounds decreased with increasing EDTA concentration due to enhanced ion suppression. Finally, flushing the capillary with 5 mM EDTA solution prior to analysis gave acceptable results. A comparison between the peak shapes without usage of EDTA and after flushing the capillary is shown in Fig. 2. With EDTA, the intensity of the peak is higher and less peak tailing occurred. The improvement of peak shape with EDTA shows that there are interferences during the CE-MS measurement, which might be caused by metal ions forming complexes with 4NC 3.2 Optimization of extraction parameters The extraction of nitrophenols with three-phase HF-LPME is driven by pH. It is necessary to keep the analytes in their neutral state in the donor phase and provide a pH in the acceptor phase that is high enough to deprotonate the analytes quantitatively. The pKa values of the target compounds ranged from 4.09 to 8.38. Therefore, pH 2 was chosen for the donor phase kept by addition of sulphuric acid. NH3 (aq) was selected as acceptor phase. 3.2.1 Composition of liquid membrane Liquid membrane optimization was carried out with a donor phase consisting of acidified ultrapure water (pH 2) being spiked with nine nitrophenols. As acceptor phase, 50 mM www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1353–1361
1357
3.2.2 Salting-out effect The “salting-out effect” is known to be potentially beneficial in liquid–liquid extractions by modifying the partitioning coefficient of analytes between the different phases. To study the influence of salt addition, extractions of a standard solution were performed at different concentrations of sodium chloride (0, 0.6, 3.1, and 6.1 mol/L) spiked with standard mixture. The acceptor phase and extraction time were set as described in Section 3.2.1. The resulting data are shown in Supporting Information Fig. 3. No significant effect could be observed from adding salt to the donor phase. The extraction efficiencies were in the same range for all nitrophenols. Consequently, no salt was added to the donor phase for further experiments.
3.2.3 Optimization of acceptor phase pH
Figure 2. Extracted ion chromatogram of 4NC (A) without usage of EDTA solution and (B) after flushing the capillary with 5 mM EDTA solution prior to sample injection.
NH3 (aq) solution was used. The extraction shook for 2 h. The data for 4NC could not be evaluated because EDTA had not been applied at this state of optimization (see Section 3.1). In this study, three different organic liquids, which were also used in other studies [31, 37–40], were tested as possible organic liquid membrane: n-undecane, DHE, and 1-octanol. A diagram of the results can be found in the Supporting Information Fig. 1. 1-Octanol showed very poor Rs (lower than 20%) while only 2,4DNP and 3,4DNP had Rs higher than 50%. Nonetheless, the Rs were still lower than those of DHE or n-undecane. The Rs of DHE and undecane gave similar results for 2NP, 3NP, 2M4NP, and 3M4NP. Slightly higher Rs were obtained for 2,6D4NP using n-undecane. 4NP, 2,4DNP, and 3,4DNP showed highest extraction efficiencies with DHE. Therefore, DHE was chosen as the liquid membrane for further experiments. It has been shown that for polar compounds, addition of TOPO in the membrane phase can enhance extraction efficiencies [41]. In a previous study [31], the addition of 5% TOPO to DHE achieved higher Rs for 2,4DNP, 2NP, 3NP, and 4NP during HF-LPME. Thus, in this work, 0, 2, 4, 6, 8, and 10% TOPO were added to the membrane phase, respectively. The results can be seen in Supporting Information Fig. 2. As can be seen, even small amounts of TOPO resulted in a dramatic decrease of Rs. Only the extraction efficiencies of 2NP and 2,4DNP were less strongly decreasing. Hence, DHE was kept as liquid membrane phase without addition of TOPO. C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The extraction efficiency at different pH values in the acceptor phase was tested by varying the concentration of NH3 (aq) between 10 and 100 mM (pH 10.1–11.1). The results are presented in Fig. 3. As can be seen, a higher concentration of NH3 (aq), and thus a higher pH in the acceptor phase, led to higher extraction efficiencies of the nitrophenols. A slight decrease in Rs was observed for 2,4DNP and 3,4DNP but Rs were still found to be close to 80%. A change in the pH value seems to have no strong effect on the extraction efficiency of 4NC. It was proposed that for nearly complete trapping, the pH should be at least 3.3 pH units higher than the pKa of the analyte [42]. This could explain why the extraction efficiency remained nearly the same with increasing NH3 (aq) concentration for the three compounds with lowest pKa value (pKa ⬍ 7 for 4NC, 3,4DNP, 2,4DNP), while Rs of the other nitrophenols increased (pKa ⬎ 7). For further optimization, 100 mM NH3 (aq) was chosen as acceptor phase because Rs were found to be highest at this concentration for most of the targeted nitrophenols.
3.2.4 Effect of extraction time The effect of the extraction time was studied in the range of 10 to 180 min. The results are shown in Fig. 4. It was found that 4NC is the only compound showing a gradual increase in R reaching no equilibrium after 3 h. 2,4DNP, 3,4DNP, and 4NP reached a plateau after 120 min, while 2M4NP, 3M4NP, and 2,6D4NP had a maximum at 90 min. A continuous decrease in R can be observed for 2NP and 3NP after 30 min. For choosing the extraction time, the Rs had to be considered on the one hand, but, on the other hand, also practical issues like short extraction times for high sample throughput were of interest. Hence, 60 min was chosen as a compromise between short extraction times (for practical reasons) and high Rs. www.electrophoresis-journal.com
1358
M. Teich et al.
Electrophoresis 2014, 35, 1353–1361
Figure 3. Influence of acceptor phase pH on R. Donor phase: 1.8 mL aqueous standard solution, pH 2 with H2 SO4 ; SLM: DHE; acceptor phase: 15 L 10–100 mM NH3 (aq); shaking speed: 2000 rpm; extraction time: 2 h.
Figure 4. Influence of extraction time on R. Donor phase: 1.8 mL aqueous standard solution, pH 2 with H2 SO4 ; SLM: DHE; acceptor phase: 15 L 100 mM NH3 (aq); shaking speed: 2000 rpm; extraction time: 30– 180 min.
3.3 Figures of merit The combined extraction and analysis method was validated using aqueous standard solutions and the optimized parameters as stated in the previous sections. The results for intraday and interday repeatabilies, R, EF, and LOD are presented in Table 1. LODs for atmospheric aerosol particles sampled on filters were calculated by assuming a sampled air volume of 25.9 m3 per 1.8 mL donor phase, taking into account the LODs from HF-LPME CE-MS, volume of water for filter extraction, and molar mass of the analytes. Regarding CE-ITMS, the LODs ranged from 0.025 to 0.125 mol/L except for 2NP, which showed an LOD of 7.5 mol/L. The application of HFLPME led to 1 to 2 orders of magnitude lower LODs caused by the observed EFs between 10 and 100 for aqueous standard mixtures. As the EF in HF-LPME is determined by the analyte R and the acceptor-to-donor phase volume ratio, the maximum EF considering the applied volumes would be 120 (15 L acceptor phase to 1.8 mL donor phase). If needed, higher EFs could easily be achieved by increasing the volume ratio, for example, using a smaller amount of acceptor phase. A comparison of instrumental LODs of other methods that determined substituted phenols in atmospheric particles as well as method LODs for studies using HF-LPME to deter-
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
mine nitrophenols in environmental samples are presented in Table 2. In general, regarding the instrumental LODs, comparable results were achieved. Kitanovski et al. [36], who investigated nitrophenolic compounds using HPLC sytems, obtained two orders of magnitude lower LODs, which might be caused by the usage of the more sensitive MS/MS system instead of an ITMS. Previously, Zhu et al. [39] and Bishop and Mitra [31] developed a similar method to determine nitrophenolic compounds (2NP, 3NP, 4NP, 3,4DNP, 2,4DNP) using three-phase HF-LPME in environmental water and gas phase nitrophenols in air samples, respectively. Regarding the HF-LPME LODs, these authors achieved similar results as in this study. In contrast, Bishop and Mitra [31] obtained very low LODs in the range of picomoles per liter. These LODs could be achieved because of a very high volume ratio (250 mL donor phase to 25 L acceptor phase) and consequently high EFs of about 3000. However, such high volume ratios are not always applicable. In this study, the available donor phase volume was limited by the water volume used for the extraction of the particle filter, which limited the EF that could be achieved. Nevertheless, the obtained LODs were comparable to most other existing HF-LPME methods. This comparison shows that the developed method using CE-MS is competitive to other more established methods and by
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1353–1361
1359
Table 1. Figures of merit for CE-ITMS and the final HF-LPME method (15 L 100 mM NH3 (aq) as acceptor phase, DHE as SLM, 1.8 mL donor phase, pH 2 with H2 SO4 , 1 h extraction time)
m/z
166.16 138.11 152.14 152.14 154.11 183.11 183.11 138.11 138.11
Compound
2,6D4NP 3NP 2M4NP 3M4NP 4NC 3,4DNP 2,4DNP 4NP 2NP
MT (min)
13.3 13.7 13.9 14.0 14.2 14.6 14.7 15.0 15.4
c(Donor)a) (mol/L)
0.4 0.8 0.8 0.4 0.8 0.4 0.4 0.8 2.5
Intraday repeatability (RSD PA, n = 5)
Interday R repeatability (RSD PA, n = 14)
EF
CEITMS (%)
HFLPME (%)
CEITMS (%)
HFLPME (%)
HFLPME (%)
HFLPME (%)
CE-ITMS (mol/L)
HF-LPME (mol/L)
Atmosphereb) (ng/m3 )
12 6 12 9 6 6 12 3 22
10 6 14 15 59 14 12 15 17
9 5 6 6 7 7 6 5 28
7 10 9 10 48 6 8 9 35
65 30 75 51 11 69 63 64 21
78 36 90 62 13 83 75 77 25
0.125 0.1 0.05 0.025 0.5 0.05 0.05 0.1 7.5
0.001 0.004 0.004 0.0004 0.04 0.0004 0.0004 0.0008 0.25
0.01 0.04 0.04 0.004 0.5 0.01 0.01 0.01 2
LOD
a) Donor phase concentration used to determine HF-LPME repeatability, R, and EF. b) Calculated for a sampled air volume of 25.9 m3 per 1.8 mL donor phase. MT: migration time; PA: peak area. Table 2. Literature comparison of LODs from (A) determination of substituted phenols in atmospheric aerosol particles and (B) studies that used HF-LPME for determining nitrophenols in environmental matrices
A Analysis method
HPLC-APCI-MS
HPLC-ESI/MS-MS
CE
CE-ESI-MS
GC-MS
CE-ESI-MS
Range of instrumental LOD (mol/L) Reference
0.08–1 [12]
0.5×10−3 –2×10−3 [36]
ca. 1 [29]
0.1–1 [28]
0.009–0.016 [43]
0.025–7.5 This work
B Analysis method
HPLCa)
HPLCa)
HPLCb)
HPLCa)
cLCa)
HPLCa)
CEa)
CE-ESI-MS
Range of method LOD (mol/L) 3×10−6 –7×10−6 0.5×10−3 –2×10−3 ca. 1×10−3 3×10−5 –4×10−5 3×10−3 –7×10−3 ca. 1×10−3 0.1–0.3 4×10−4 –0.25 Reference [31] [44] [40] [37] [39] [38] [27] This work a) Coupled with three-phase HF-LPME. b) Coupled with two-phase HF-LPME. cLC: capillary LC.
inclusion of the HF-LPME enrichment step very low concentrations of nitrophenolic compounds can be detected. The interday and intraday repeatabilities of HF-LPME CE-MS were found to be similar, showing the robustness of the method. For most of the analytes, repeatability lay below 15% and was within the same range as for CE-MS separation itself. An unreasonable high peak area RSD after HF-LPME was found for 4NC (about 50%), likely related to the poor R (11%) of this compound. 2NP showed a larger deviation in repeatability during CE-MS separation compared to the other analytes, but similar repeatability regarding HF-LPME CEMS. This observation could have been caused by the low peak intensity for 2NP in CE-MS measurements. The Rs of the target compounds are in the range of 11 to 75%. It is apparent that 2NP showed a large difference between calculated Rs based on the peak areas (about 80%, see Fig. 4) and Rs based on real concentrations (about C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
20%, see Table 1). The reason was found to be the very shallow slope of the calibration curve of 2NP, leading to rather small changes in PA with rather large changes in concentration.
3.4 Application to real samples The developed and validated method was applied to three PM10 filters from Melpitz, Germany, having different OC loadings; hence, three different matrices existed. To additionally investigate the accuracy of the method, quantification was carried out twice for each filter using external calibration (ExCal) and standard addition (SA). The ExCal curve was obtained from aqueous standard solutions at different donor concentration level after extraction with HF-LPME. Thus, the calibration curve already contains the Rs of each compound www.electrophoresis-journal.com
1360
M. Teich et al.
Electrophoresis 2014, 35, 1353–1361
Table 3. Mass concentrations obtained from three filter samples compared with concentrations determined in other studies
Analyte
Mass concentration (ng/m3 ) This worka) Melpitz, Germany
2NP 3NP 4NP 2M4NP 3M4NP 4NC 2,6D4NP 2,4DNP 3,4DNP
Kitanovski et al. [36] a) Zhang et al. [22] b) Ljubljana, Slovenia Mainz, Germany
January 24, 2012 August 30, 2012
October 10, 2012 Winter 2010/2011
SA
ExCal
SA
ExCal
SA
ExCal
Electrophoresis 2014, 35, 1353–1361
Monique Teich Dominik van Pinxteren Hartmut Herrmann Chemistry Department, ¨ Leibniz-Institut fur ¨ Tropospharenforschung (TROPOS), Leipzig, Germany
Received September 13, 2013 Revised November 28, 2013 Accepted December 2, 2013
Research Article
Determination of nitrophenolic compounds from atmospheric particles using hollow-fiber liquid-phase microextraction and capillary electrophoresis/mass spectrometry analysis A hollow-fiber liquid-phase microextraction method was developed to enrich nine nitrophenolic compounds from aqueous extracts of atmospheric aerosol particles. Analysis was performed by CE coupled with ESI MS. The BGE composition was optimized to a 20 mM ammonium acetate buffer at pH 9.7 containing 15% methanol v/v. Several extraction parameters (composition of organic liquid membrane, pH of acceptor phase, salting-out effect, extraction time) were investigated for their effect on the analyte recoveries. The donor phase consisted of a 1.8 mL sample solution kept at pH 2 while the acceptor phase was a 15 L 100 mM aqueous ammonia solution. Dihexyl ether served as supported liquid membrane. Low detection limits in the range of nanomole per liter were achieved. Recoveries of aqueous standard solutions were found to be between 11 and 90% with enrichment factors between 10 and 100. Interday and intraday repeatabilities were in an acceptable range for most compounds (6–15% and 7–10%, respectively) but somewhat higher for 4-nitrocatechol (59 and 48%) and 2-nitrophenol (17 and 35%). The developed method was found to be competitive with more established method and was successfully applied to samples of atmospheric particulate matter from field experiments. Keywords: Atmospheric particles / CE-MS / Liquid-phase microextraction / Nitrophenols DOI 10.1002/elps.201300448
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction Nitrogen-containing compounds represent an important fraction of aerosol organic matter (2–7%) [1]. Large amounts
Correspondence: Professor Hartmut Herrmann, Leibniz-Institut ¨ Tropospharenforschung ¨ fur (TROPOS), Permoserstr. 15, 04318 Leipzig, Germany E-mail: [email protected] Fax: +49-341-2717-99-7024
Abbreviations: DHE, dihexyl ether; 2,4DNP, 2,4dinitrophenol; 2,6D4NP, 2,6-dimethyl-4-nitrophenol; 3,4DNP, 3,4-dinitrophenol; EF, enrichment factor; ExCal, external calibration; HF-LPME, hollow-fiber liquid-phase microextraction; 2M4NP, 2-methyl-4-nitrophenol; 4M2NP, 4-methyl-2nitrophenol; 4NC, 4-nitrocatechol; 2NP, 2-nitrophenol; 3NP, 3-nitrophenol; 4NP, 4-nitrophenol; OC, organic carbon; ON, organic nitrogen; R, recovery; SA, standard addition; SLM, supported liquid membrane; TOPO, trioctylphosphine oxide; WSON, water-soluble organic nitrogen
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
are water-soluble and it has been reported that water-soluble organic nitrogen (WSON) can cover 20–80% of the total soluble nitrogen of aerosol particles [2–4]. Comparisons of organic nitrogen (ON) and WSON with organic carbon (OC) and water-soluble organic carbon showed ON/OC ratios of about 0.19 for marine aerosol [5]. Miyazaki et al. [6] found ON/OC ratios of 0.43 for submicrometer particles being influenced by high marine biological activity. At continental conditions, the average WSON/WSOC mass ratio was found to be 5% during measurements in southeastern United States [7]. Attention has been drawn to WSON because it can, in contrast to most of the other organic constituents, absorb ultraviolet and visible light, thus influencing the earth’s radiative balance. This characteristic makes WSON an important contributor to the so-called brown carbon [8]. Furthermore, organonitrogen compounds can act as cloud condensation nuclei without being mixed with inorganic salts [9]. WSON can be of various natural or anthropogenic origins. Biomass burning is considered to be a major contributor to nitrogen-containing compounds [8, 10–12]. Secondary
www.electrophoresis-journal.com
1354
M. Teich et al.
formation due to oxidation reactions in the presence of NOx with volatile organic compounds is also possible, followed by gas-to-particle condensation, which can lead to the formation of secondary organic aerosol [13, 14]. WSON is a complex mixture consisting of hundreds of different compounds. So far several classes of nitrogencontaining compounds were identified, such as aliphatic and aromatic amines, amino acids, nitriles, amides, hydrazines, nitrosamines, heterocyclic compounds, organonitrates, and nitroaromatic compounds [1, 10, 15–18]. Nitrophenols have recently received increasing attention [15]. They are known for their negative effect on human health [19] as well as their potential phytotoxity [20] and, thus, are responsible for forest decline in polluted areas [21]. Quantification of nitrophenols in atmospheric particles requires an appropriate separation technique as well as a preconcentration step prior to separation because of their low concentrations in the range of nanogram per cubic meter. In previous studies, mainly HPLC (e.g. [11, 15, 22]) or GC (e.g. [15,23]) have been used for the determination of nitroaromatic compounds in atmospheric particles, usually combined with solvent evaporation and/or derivatization. In this study, CE is used for separating nitrophenols as it provides high separation efficiency, low sample consumption, short analytical times, and low operation costs [24], making the CE-based technique an attractive alternative to HPLC techniques if competitive LODs can be achieved. The very low injection volume needed for CE is well suited for coupling with modern microextraction techniques, like the hollow-fiber liquid-phase microextraction (HF-LPME), which can provide extraction volumes in the range of a few microliters [25]. In several previous studies, various substituted phenols were studied in different environments using CE analysis, for example, nitrophenols in tap water, snow water, the Yangtze River water [26], and farm water samples [27] as well as methoxy phenols [28] and 4-nitrophenol (4NP) [29] in atmospheric particles. HF-LPME is a relative new membrane extraction technique providing easy handling, high enrichment factors (EFs), low organic solvent consumption, and low costs. In contrast to the more common SPE, no evaporation of organic solvent is needed and potential chemical losses due to evaporation can be avoided [25]. In principle, a water-immiscible liquid is used to form a supported liquid membrane (SLM) inside the pores of a hollow fiber [30]. After impregnating the hollow fiber, either an organic liquid (two-phase HF-LPME) or an aqueous solution (three-phase HF-LPME) is inserted into the fiber (acceptor phase) and afterward placed into a donor solution. The target compounds are extracted from the donor phase through the SLM into the acceptor phase where they are trapped. The recoveries (Rs) and EFs obtained are dependent on the partition coefficient of the analytes between the different phases and the volumes of the donor, liquid membrane, and acceptor phases [25]. Very low LODs can be achieved depending on the method parameters and the used instrument for determination. LODs range from a few nanograms per liter [31] C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2014, 35, 1353–1361
to several micrograms per liter [27], respectively, for nitrophenols extracted by HF-LPME and analyzed by HPLC or CE. Several studies have reported the extraction of various nitrophenolic compounds via HF-LPME from environmental water samples (lake water, river water, sea water, wastewater). Bishop and Mitra [31] determined nitrophenols in air using impinger sampling and subsequent preconcentration with HF-LPME. Those studies differ in their usage of the supported organic liquid membrane, acceptor phase, donorphase composition, extraction time, and LODs. Nitrophenols were also extracted by other kinds of SLM techniques, as reviewed by J¨onsson and Mathiasson [32]. Field and laboratory measurements of nitrophenols in the atmosphere have been reviewed by Harrison et al. [15]. The aim of this study was to optimize CE-MS parameters for good separation of all target compounds, to explore the potential of HF-LPME coupled with CE-MS for a simple and sensitive determination of nitrophenols from atmospheric particles, and to evaluate the performance of the developed method in comparison to available other techniques.
2 Materials and methods 2.1 Chemicals The chemicals used in this study were obtained from different suppliers as follows: 2-propanol and methanol (LC-MS Chromasolv, ࣙ99.9%), aqueous ammonia solution (NH3 (aq), 5 M), sulphuric acid (puriss. p.a., 95–97%), acetic acid (eluent additive for LC-MS), 1-octanol (⬎95.5%), 3,4-dinitrophenol (3,4DNP, ⬎99%), 2,4-dinitrophenol (2,4DNP, ⬎99%), 4NP (⬎99.5%), 3-nitrophenol (3NP, ⬎99%), 2-nitrophenol (2NP, ⬎99%), 4-nitrocatechol (4NC, ⬎98%) from Fluka (Munich, Germany); acetone (ACS reagent, ࣙ99.5%), undecane (ࣙ99%), dihexyl ether (DHE, ⬎97%), and trioctylphosphine oxide (TOPO, 99%); 2,6-dimethyl-4-nitrophenol (2,6D4NP, 98%), 2-methyl-4-nitrophenol (2M4NP, 98%), 4-methyl-2nitrophenol (4M2NP, 98%) from Sigma-Aldrich (Munich, Germany). EDTA (Titriplex II) was obtained from Merck (Darmstadt, Germany). Five millimolar stock solutions of each nitrophenol were prepared in 100 mM NH3 (aq) and stored at −20°C. Standard mixtures were prepared at micromolar levels in ultrapure water (Milli-Q, Millipore, Schwalbach, Germany) and stored for a maximum time of 4 weeks at 4°C. No degradation of analytes was observed during the storage period.
2.2 Atmospheric particle sampling and extraction Atmospheric particles were sampled at the research station Melpitz [33] in 2012 with a Digitel DHA-80 filter sampler (Riemer, Hausen, Germany). Melpitz is a rural site located about 50 km NE of Leipzig, Germany. The atmospheric particles with an aerodynamic diameter less than 10 m (PM10 ) www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1353–1361
were collected on quartz fiber filters (MK 360, Munktell, Falun, Sweden). Sampling time was 24 h (00:00 to 24:00 CET) at a flow rate of 0.5 m3 /min. To reduce the blank content of carbonaceous material, the filters were heated for 24 h at 105°C before sampling. After sampling the filters were stored at −20°C until extraction. For the analysis of watersoluble nitrophenolic compounds, 12 pieces (78.5 mm2 each) were punched out of one filter and placed into one disposable 5 mL syringe (Omnifix, Braun, Melsungen, Germany). The syringe was filled with 3 mL of MilliQ-water and shook for 2 h at 700 rpm on a laboratory shaker (KS 125, IKA, Staufen, Germany). After shaking, the aqueous particle extract was filtered through a precleaned syringe filter (0.45 m, Acrodisc 13, Pall, Dreieich, Germany). Aliquots of this solution were used for the HF-LPME experiments. Precleaning of the disposable syringes was carried out by ultrasonication in ultrapure water. The syringe filters were cleaned with methanol before use.
2.3 HF-LPME The HF-LPME device used is the same as described before [34]. Briefly, a 2 mL amber glass vial with a screw cap containing a silicone septum (Agilent, Waldbronn, Germany) served as container for the donor phase. A polypropylene hollow fiber (Accurel Q 3/2, 0.6 mm id, 1 mm od, 0.2 m pore size; Membrana, Wuppertal, Germany) was cut into pieces of 11 cm length and cleaned twice by ultrasonication in acetone for 15 min to remove possible contaminants originating from the polypropylene material. Afterward, the fibers were allowed to air dry. Two 1-cm-long pieces of Teflon tubing (1/16 in. od, 1 mm id) punched through the silicon septum served as guiding tubes to hold the hollow fiber into place. The fibers were threaded through the guiding tubes so that 1 cm protruded while shaking. Filling and discharging of the fiber was conducted with analytical syringes (1700 RN series, Hamilton, Bonaduz, Switzerland) and a tapered needle (0.64 mm od, gauge 23S, point style AS, Hamilton Bonaduz, Switzerland). The fiber was flushed with an organic liquid to build up the SLM inside the pores of the hollow fiber. As described by van Pinxteren et al. [34], contaminants could originate from the liquid membrane. Therefore, in a next step, the hollow fiber was filled with 100 mM NH3 (aq) and placed in a vial containing 1.8 mL of 100 mM NH3 (aq) for 15 min to eliminate contaminants. Afterward, the solution was removed by flushing the fiber with air and the precleaned fiber was filled with 15 L of the acceptor phase. NH3 (aq) served as the acceptor phase instead of NaOH because of its high volatility and MS compatibility. The filled hollow fiber was placed in a vial with 1.8 mL of the donor solution acidified to pH 2 with sulphuric acid and shaken for 60 min with a VXR basic Vibrax shaker (IKA) at 2000 rpm. The method optimization and validation were carried out with aqueous standard solutions. After extraction, the acceptor phase was removed and transferred into a polypropylene vial (Agilent) and subsequently analyzed by CE coupled with MS (CE-MS). C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1355
2.4 CE-MS conditions An Agilent 3D CE instrument coupled with an Esquire 3000+ ITMS (Bruker, Bremen, Germany) was used for analysis. ESI was applied as an interface. Precision-cleaved fused-silica capillaries (360 m od, 50 m id, 75 cm length, Polymicro Technologies, Phoenix, AZ, USA) were used to avoid unstable electrospray, which can be caused by poor handmade cuts. At both sides of the capillary, the polyimide coating was removed by an open flame (1–2 cm). The BGE consists of 20 mM ammonium acetate solution containing 15% v/v methanol adjusted to pH 9.7 using 5 M NH3 (aq). Hydrodynamic injection was done by applying a pressure difference of 50 mbar for 10 s to the inlet vial. At the beginning of each series of measurements, the capillary was flushed with 1 M NH3 (aq) (10 min) and BGE (30 min). Prior to each analysis, the capillary was flushed with 5 mM EDTA solution (5 min) and BGE (3 min). The usage of EDTA solution improved the peak shape of 4NC during analysis (see Section 3.1). After each run, a short plug of 1 M NH3 (aq) was injected to avoid carryover of analytes. The sheath liquid consisted of an isopropanol/water (80/20) mixture and the flow rate was set to 0.4 mL/h. The pressure of nebulizing gas was set to 4 psi (276 mbar) and the flow rate of the drying gas was 10 L/min at a temperature of 200°C. ESI was operated in negative mode with a voltage of 4.5 kV and an endplate offset of −500 V. All compounds were detected in their deprotonated form as [M−H]− . Samples were analyzed twice taking the average peak area for data interpretation.
2.5 Figures of merit Method validation was carried out using aqueous standard solutions for HF-LPME extraction. The EFs and Rs were calculated based on real concentrations. During optimization experiments, the peak areas instead of real concentrations were used to estimate the Rs. In general, the EF is given as EF = c(acceptor)/c(donor). c(Acceptor) stands for the concentration of analyte in the acceptor phase after extraction and c(donor) represents the initial concentration in the donor phase. R was calculated as: R=
n(acceptor) EF × V (acceptor) × 100 = × 100, V (donor) n(donor)
(1)
where V(acceptor) is the volume of the acceptor phase, V(donor) is the volume of the donor phase, n(acceptor) is the amount of substance of the respective analyte in the donor phase, and n(donor) is the initial amount of analyte in the donor phase. The intraday repeatability was characterized by 5 parallel extractions on one day and interday repeatability by running 14 extractions in total on four consecutive days. Both are expressed as RSD of the peak area. Repeatability of CE-ITMS is based on runs with standard solution having 120 times higher concentration than in the donor phase due to the volume ratio standard solution-to-donor phase. Instrumental LODs were estimated by measuring a series of diluted standard mixtures. HF-LPME LODs were determined by www.electrophoresis-journal.com
1356
M. Teich et al.
Electrophoresis 2014, 35, 1353–1361
Figure 1. Extracted ion chromatograms of target compounds after CE-MS separation. (A) BGE: 20 mM ammonium acetate (pH 9.1) with 10% v/v methanol; sheath liquid: isopropanol/water (80:20). (B) After optimization, that is, BGE: 20 mM ammonium acetate (pH 9.7) with 15% v/v methanol; sheath liquid: isopropanol/water (80:20).
extracting a series of diluted standard mixtures. The LOD was defined as the concentration where the corresponding peak in the extracted ion electropherogram shows a S/N of 3. Calculation of LOD in terms of atmospheric concentration was done based on the HF-LPME LOD, taking into account the sampled air volume, the volume of water for filter extraction, and the molar mass of the analytes.
3 Results and discussion 3.1 Optimization of separation conditions All experiments for optimizing separation conditions during CE-ESI-MS were carried out using a standard mixture with prescribed analyte concentration in the micromolar range. Previous works of our group demonstrated that for acidic compounds a BGE consisting of 20 mM ammonium acetate with additional 10% v/v of methanol at pH 9.1 gives good separation efficiencies [28, 35]. First, runs with aqueous standard mixtures under this condition are shown in Fig. 1A. No baseline separation could be achieved for 3,4DNP and 2,4DNP and no separation at all for 2M4NP and 3M4NP. The targeted 4NC showed a very broad peak and peak tailing. Further optimization of the separation conditions have been carried out to improve the separation efficiency. Thus, the composition of BGE was varied regarding the concentration of ammonium acetate (20–30 mM) and the methanol fraction (0–20%). Moreover, the BGE was tested at different pH values (9.0–10.0). Slight variations of the ammonium acetate concentration lead to no significant improvement in the resolution of the nitrophenols. Therefore, a 20 mM ammonium acetate BGE was kept for further analysis. However, baseline separation of all target analytes could be achieved by increasing the BGE pH and the fraction of methanol. The optimum was found to be a 20 mM ammonium acetate buffer at pH 9.7 with additional 15% v/v of methanol (Fig. 1B). After optimizing separation conditions, the peak-shape problems of 4NC remained. Similar problems with peak shape of 4NC were ob C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
served by Kitanovski et al. [36] during LC-MS analysis. They supposed that peak broadening is caused by interactions of 4NC with metal ions. Thus, EDTA as a complexing agent was applied to complex metal ions. Coinjection of EDTA resulted in a strong improvement of peak shape [36]. Based on this observation, the effect of EDTA solution on the peak shape of the nitrophenols was studied. The addition of small concentrations of EDTA to the BGE (0.05, 0.1, 0.25, 0.5 mM) led to a slightly improved peak shape, while adding EDTA directly to the standard mixture did not have any effect. Furthermore, the peak intensity of the other compounds decreased with increasing EDTA concentration due to enhanced ion suppression. Finally, flushing the capillary with 5 mM EDTA solution prior to analysis gave acceptable results. A comparison between the peak shapes without usage of EDTA and after flushing the capillary is shown in Fig. 2. With EDTA, the intensity of the peak is higher and less peak tailing occurred. The improvement of peak shape with EDTA shows that there are interferences during the CE-MS measurement, which might be caused by metal ions forming complexes with 4NC 3.2 Optimization of extraction parameters The extraction of nitrophenols with three-phase HF-LPME is driven by pH. It is necessary to keep the analytes in their neutral state in the donor phase and provide a pH in the acceptor phase that is high enough to deprotonate the analytes quantitatively. The pKa values of the target compounds ranged from 4.09 to 8.38. Therefore, pH 2 was chosen for the donor phase kept by addition of sulphuric acid. NH3 (aq) was selected as acceptor phase. 3.2.1 Composition of liquid membrane Liquid membrane optimization was carried out with a donor phase consisting of acidified ultrapure water (pH 2) being spiked with nine nitrophenols. As acceptor phase, 50 mM www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1353–1361
1357
3.2.2 Salting-out effect The “salting-out effect” is known to be potentially beneficial in liquid–liquid extractions by modifying the partitioning coefficient of analytes between the different phases. To study the influence of salt addition, extractions of a standard solution were performed at different concentrations of sodium chloride (0, 0.6, 3.1, and 6.1 mol/L) spiked with standard mixture. The acceptor phase and extraction time were set as described in Section 3.2.1. The resulting data are shown in Supporting Information Fig. 3. No significant effect could be observed from adding salt to the donor phase. The extraction efficiencies were in the same range for all nitrophenols. Consequently, no salt was added to the donor phase for further experiments.
3.2.3 Optimization of acceptor phase pH
Figure 2. Extracted ion chromatogram of 4NC (A) without usage of EDTA solution and (B) after flushing the capillary with 5 mM EDTA solution prior to sample injection.
NH3 (aq) solution was used. The extraction shook for 2 h. The data for 4NC could not be evaluated because EDTA had not been applied at this state of optimization (see Section 3.1). In this study, three different organic liquids, which were also used in other studies [31, 37–40], were tested as possible organic liquid membrane: n-undecane, DHE, and 1-octanol. A diagram of the results can be found in the Supporting Information Fig. 1. 1-Octanol showed very poor Rs (lower than 20%) while only 2,4DNP and 3,4DNP had Rs higher than 50%. Nonetheless, the Rs were still lower than those of DHE or n-undecane. The Rs of DHE and undecane gave similar results for 2NP, 3NP, 2M4NP, and 3M4NP. Slightly higher Rs were obtained for 2,6D4NP using n-undecane. 4NP, 2,4DNP, and 3,4DNP showed highest extraction efficiencies with DHE. Therefore, DHE was chosen as the liquid membrane for further experiments. It has been shown that for polar compounds, addition of TOPO in the membrane phase can enhance extraction efficiencies [41]. In a previous study [31], the addition of 5% TOPO to DHE achieved higher Rs for 2,4DNP, 2NP, 3NP, and 4NP during HF-LPME. Thus, in this work, 0, 2, 4, 6, 8, and 10% TOPO were added to the membrane phase, respectively. The results can be seen in Supporting Information Fig. 2. As can be seen, even small amounts of TOPO resulted in a dramatic decrease of Rs. Only the extraction efficiencies of 2NP and 2,4DNP were less strongly decreasing. Hence, DHE was kept as liquid membrane phase without addition of TOPO. C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The extraction efficiency at different pH values in the acceptor phase was tested by varying the concentration of NH3 (aq) between 10 and 100 mM (pH 10.1–11.1). The results are presented in Fig. 3. As can be seen, a higher concentration of NH3 (aq), and thus a higher pH in the acceptor phase, led to higher extraction efficiencies of the nitrophenols. A slight decrease in Rs was observed for 2,4DNP and 3,4DNP but Rs were still found to be close to 80%. A change in the pH value seems to have no strong effect on the extraction efficiency of 4NC. It was proposed that for nearly complete trapping, the pH should be at least 3.3 pH units higher than the pKa of the analyte [42]. This could explain why the extraction efficiency remained nearly the same with increasing NH3 (aq) concentration for the three compounds with lowest pKa value (pKa ⬍ 7 for 4NC, 3,4DNP, 2,4DNP), while Rs of the other nitrophenols increased (pKa ⬎ 7). For further optimization, 100 mM NH3 (aq) was chosen as acceptor phase because Rs were found to be highest at this concentration for most of the targeted nitrophenols.
3.2.4 Effect of extraction time The effect of the extraction time was studied in the range of 10 to 180 min. The results are shown in Fig. 4. It was found that 4NC is the only compound showing a gradual increase in R reaching no equilibrium after 3 h. 2,4DNP, 3,4DNP, and 4NP reached a plateau after 120 min, while 2M4NP, 3M4NP, and 2,6D4NP had a maximum at 90 min. A continuous decrease in R can be observed for 2NP and 3NP after 30 min. For choosing the extraction time, the Rs had to be considered on the one hand, but, on the other hand, also practical issues like short extraction times for high sample throughput were of interest. Hence, 60 min was chosen as a compromise between short extraction times (for practical reasons) and high Rs. www.electrophoresis-journal.com
1358
M. Teich et al.
Electrophoresis 2014, 35, 1353–1361
Figure 3. Influence of acceptor phase pH on R. Donor phase: 1.8 mL aqueous standard solution, pH 2 with H2 SO4 ; SLM: DHE; acceptor phase: 15 L 10–100 mM NH3 (aq); shaking speed: 2000 rpm; extraction time: 2 h.
Figure 4. Influence of extraction time on R. Donor phase: 1.8 mL aqueous standard solution, pH 2 with H2 SO4 ; SLM: DHE; acceptor phase: 15 L 100 mM NH3 (aq); shaking speed: 2000 rpm; extraction time: 30– 180 min.
3.3 Figures of merit The combined extraction and analysis method was validated using aqueous standard solutions and the optimized parameters as stated in the previous sections. The results for intraday and interday repeatabilies, R, EF, and LOD are presented in Table 1. LODs for atmospheric aerosol particles sampled on filters were calculated by assuming a sampled air volume of 25.9 m3 per 1.8 mL donor phase, taking into account the LODs from HF-LPME CE-MS, volume of water for filter extraction, and molar mass of the analytes. Regarding CE-ITMS, the LODs ranged from 0.025 to 0.125 mol/L except for 2NP, which showed an LOD of 7.5 mol/L. The application of HFLPME led to 1 to 2 orders of magnitude lower LODs caused by the observed EFs between 10 and 100 for aqueous standard mixtures. As the EF in HF-LPME is determined by the analyte R and the acceptor-to-donor phase volume ratio, the maximum EF considering the applied volumes would be 120 (15 L acceptor phase to 1.8 mL donor phase). If needed, higher EFs could easily be achieved by increasing the volume ratio, for example, using a smaller amount of acceptor phase. A comparison of instrumental LODs of other methods that determined substituted phenols in atmospheric particles as well as method LODs for studies using HF-LPME to deter-
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
mine nitrophenols in environmental samples are presented in Table 2. In general, regarding the instrumental LODs, comparable results were achieved. Kitanovski et al. [36], who investigated nitrophenolic compounds using HPLC sytems, obtained two orders of magnitude lower LODs, which might be caused by the usage of the more sensitive MS/MS system instead of an ITMS. Previously, Zhu et al. [39] and Bishop and Mitra [31] developed a similar method to determine nitrophenolic compounds (2NP, 3NP, 4NP, 3,4DNP, 2,4DNP) using three-phase HF-LPME in environmental water and gas phase nitrophenols in air samples, respectively. Regarding the HF-LPME LODs, these authors achieved similar results as in this study. In contrast, Bishop and Mitra [31] obtained very low LODs in the range of picomoles per liter. These LODs could be achieved because of a very high volume ratio (250 mL donor phase to 25 L acceptor phase) and consequently high EFs of about 3000. However, such high volume ratios are not always applicable. In this study, the available donor phase volume was limited by the water volume used for the extraction of the particle filter, which limited the EF that could be achieved. Nevertheless, the obtained LODs were comparable to most other existing HF-LPME methods. This comparison shows that the developed method using CE-MS is competitive to other more established methods and by
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1353–1361
1359
Table 1. Figures of merit for CE-ITMS and the final HF-LPME method (15 L 100 mM NH3 (aq) as acceptor phase, DHE as SLM, 1.8 mL donor phase, pH 2 with H2 SO4 , 1 h extraction time)
m/z
166.16 138.11 152.14 152.14 154.11 183.11 183.11 138.11 138.11
Compound
2,6D4NP 3NP 2M4NP 3M4NP 4NC 3,4DNP 2,4DNP 4NP 2NP
MT (min)
13.3 13.7 13.9 14.0 14.2 14.6 14.7 15.0 15.4
c(Donor)a) (mol/L)
0.4 0.8 0.8 0.4 0.8 0.4 0.4 0.8 2.5
Intraday repeatability (RSD PA, n = 5)
Interday R repeatability (RSD PA, n = 14)
EF
CEITMS (%)
HFLPME (%)
CEITMS (%)
HFLPME (%)
HFLPME (%)
HFLPME (%)
CE-ITMS (mol/L)
HF-LPME (mol/L)
Atmosphereb) (ng/m3 )
12 6 12 9 6 6 12 3 22
10 6 14 15 59 14 12 15 17
9 5 6 6 7 7 6 5 28
7 10 9 10 48 6 8 9 35
65 30 75 51 11 69 63 64 21
78 36 90 62 13 83 75 77 25
0.125 0.1 0.05 0.025 0.5 0.05 0.05 0.1 7.5
0.001 0.004 0.004 0.0004 0.04 0.0004 0.0004 0.0008 0.25
0.01 0.04 0.04 0.004 0.5 0.01 0.01 0.01 2
LOD
a) Donor phase concentration used to determine HF-LPME repeatability, R, and EF. b) Calculated for a sampled air volume of 25.9 m3 per 1.8 mL donor phase. MT: migration time; PA: peak area. Table 2. Literature comparison of LODs from (A) determination of substituted phenols in atmospheric aerosol particles and (B) studies that used HF-LPME for determining nitrophenols in environmental matrices
A Analysis method
HPLC-APCI-MS
HPLC-ESI/MS-MS
CE
CE-ESI-MS
GC-MS
CE-ESI-MS
Range of instrumental LOD (mol/L) Reference
0.08–1 [12]
0.5×10−3 –2×10−3 [36]
ca. 1 [29]
0.1–1 [28]
0.009–0.016 [43]
0.025–7.5 This work
B Analysis method
HPLCa)
HPLCa)
HPLCb)
HPLCa)
cLCa)
HPLCa)
CEa)
CE-ESI-MS
Range of method LOD (mol/L) 3×10−6 –7×10−6 0.5×10−3 –2×10−3 ca. 1×10−3 3×10−5 –4×10−5 3×10−3 –7×10−3 ca. 1×10−3 0.1–0.3 4×10−4 –0.25 Reference [31] [44] [40] [37] [39] [38] [27] This work a) Coupled with three-phase HF-LPME. b) Coupled with two-phase HF-LPME. cLC: capillary LC.
inclusion of the HF-LPME enrichment step very low concentrations of nitrophenolic compounds can be detected. The interday and intraday repeatabilities of HF-LPME CE-MS were found to be similar, showing the robustness of the method. For most of the analytes, repeatability lay below 15% and was within the same range as for CE-MS separation itself. An unreasonable high peak area RSD after HF-LPME was found for 4NC (about 50%), likely related to the poor R (11%) of this compound. 2NP showed a larger deviation in repeatability during CE-MS separation compared to the other analytes, but similar repeatability regarding HF-LPME CEMS. This observation could have been caused by the low peak intensity for 2NP in CE-MS measurements. The Rs of the target compounds are in the range of 11 to 75%. It is apparent that 2NP showed a large difference between calculated Rs based on the peak areas (about 80%, see Fig. 4) and Rs based on real concentrations (about C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
20%, see Table 1). The reason was found to be the very shallow slope of the calibration curve of 2NP, leading to rather small changes in PA with rather large changes in concentration.
3.4 Application to real samples The developed and validated method was applied to three PM10 filters from Melpitz, Germany, having different OC loadings; hence, three different matrices existed. To additionally investigate the accuracy of the method, quantification was carried out twice for each filter using external calibration (ExCal) and standard addition (SA). The ExCal curve was obtained from aqueous standard solutions at different donor concentration level after extraction with HF-LPME. Thus, the calibration curve already contains the Rs of each compound www.electrophoresis-journal.com
1360
M. Teich et al.
Electrophoresis 2014, 35, 1353–1361
Table 3. Mass concentrations obtained from three filter samples compared with concentrations determined in other studies
Analyte
Mass concentration (ng/m3 ) This worka) Melpitz, Germany
2NP 3NP 4NP 2M4NP 3M4NP 4NC 2,6D4NP 2,4DNP 3,4DNP
Kitanovski et al. [36] a) Zhang et al. [22] b) Ljubljana, Slovenia Mainz, Germany
January 24, 2012 August 30, 2012
October 10, 2012 Winter 2010/2011
SA
ExCal
SA
ExCal
SA
ExCal
