Fluorine Speciation Analysis Using Reverse Phase Liquid

Jun 11, 2012 - Trace Element Speciation Laboratory (TESLA), Chemistry, University of Aberdeen, Aberdeen, AB24 3UE, Scotland, U.K.,. ‡...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/ac

Fluorine Speciation Analysis Using Reverse Phase Liquid Chromatography Coupled Off-Line to Continuum Source Molecular Absorption Spectrometry (CS-MAS): Identification and Quantification of Novel Fluorinated Organic Compounds in Environmental and Biological Samples Zhiwei Qin,‡ David McNee,† Heike Gleisner,§ Andrea Raab,† Kwaku Kyeremeh,⊥ Marcel Jaspars,‡ Eva Krupp,†,∥ Hai Deng,‡ and Jörg Feldmann*,†,‡ †

Trace Element Speciation Laboratory (TESLA), Chemistry, University of Aberdeen, Aberdeen, AB24 3UE, Scotland, U.K., Marine Biodiscovery Centre (MBC), Chemistry, University of Aberdeen, Aberdeen, AB24 3UE, Scotland, U.K. § Analytik Jena AG, Konrad Zuse Strasse 1, 07445 Jena Germany ∥ Aberdeen Centre for Environmental Sustainability (ACES), University of Aberdeen, Aberdeen, AB24 3UE, Scotland, U.K. ⊥ Biochemistry, FGO Torto Building, University of Ghana, Legon, Ghana ‡

S Supporting Information *

ABSTRACT: Driven by increasing demand for the monitoring of industrial perfluorinated compounds (PFCs), the identification of novel fluorine containing compounds (FOCs) and the tracking of organofluorine drugs and their degradation products, there is a clear need for sensitive, fluorine-specific detection of unknown FOCs. Here we report the first ever direct fluorine-specific (speciation) method; capable of individually detecting untargeted FOCs in environmental and biological samples through the application of continuum source molecular absorption spectrometry (CS-MAS) using a commercial CS-AAS. Two model FOCs (2,4,6, trifluorobenzoic acid (TFBA) and 5-fluoroindol-5-carboxylic acid (FICA)) were used, achieving fluorinespecific detection across a range of 0.1 to 300 ng/mL fluorine, corresponding to a limit of detection of 4 pg F and 5.26 nM for both compounds. Both TFBA and FICA showed a similar response to CS-MAS detection, potentially enabling the quantification of fluorine content in novel FOCs without having molecular standards available. This paper also reports the use of reverse-phase high performance liquid chromatography (RP-HPLC) coupled off-line with CS-MAS for the identification of single organofluorines in a mixture of FOCs via fraction collection. The linear range of both FOCs was determined to be from 1 to 500 ng/mL. The limits of detection of those species were just above 1 ng/mL (100 pg) and can therefore compete with targeted analytical methods such as ESI-MS. Finally, as a proof of principle the analysis of a fluoride-containing groundwater sample from Ghana demonstrated that this method can be used in the detection of novel FOCs, with identification achieved through parallel ESI-MS. Coupled HPLC−CS-MAS/ESI-MS is the first analytical methodology capable of selectively detecting and identifying novel FOCs, making possible the quantification of all fluorine containing compounds in one sample. This is the necessary analytical requirement to perform f luoronomics. he carbon fluorine bond is among the strongest bonds in chemistry; hence fluorinated organic compounds (FOC) have extraordinary properties, making them attractive products in the cosmetics, surface coatings, and medical industries.1−4 Perfluorinated compounds (PFCs) are especially well-known

T

© 2012 American Chemical Society

Received: May 4, 2012 Accepted: June 11, 2012 Published: June 11, 2012 6213

dx.doi.org/10.1021/ac301201y | Anal. Chem. 2012, 84, 6213−6219

Analytical Chemistry

Article

None of these methods may be used to screen for the detection of individual novel FOCs in an extract, since all methods require a pretreatment step to convert FOCs into fluoride. Here we report the proof of concept that coupling RPHPLC with parallel CS-MAS and ESI-MS may be used as a direct method for the sensitive fluorine-specific detection of novel and target FOCs in an environmental sample. The principle is based on online pyrolysis of the sample, rapidly converting all FOCs to fluoride, which reacts with added gallium in a secondary molecule formation step to form volatile GaF, a compound with a specific absorption band. The aim of this article is to demonstrate fluorine speciation methodology using the two model FOCs TFBA and FICA in the presence of fluoride. Additionally, this method has been used to analyze an environmental water sample from a fluoride affected area and figures of merit are reported.

for their unusual solubility, being simultaneously hydro- and lipophobic. However, PFCs also display exceptional environmental stability and persistence, presenting the risk of bioaccumulation. As such they have recently been recognized as emerging global pollutants with perfluorooctane sulfonate (PFOS) registered on the Stockholm Convention list of persistent organic pollutants (POPs).5,6 This recognition has triggered a boost in interest in the monitoring of PFCs in water, wildlife and food.7,8 Because of the small concentrations involved, sophisticated analytical methods have been developed over the years to determine PFC content of environmental samples; however, these methods have been hamstrung by the lack of fluorine specific elemental detection. The current method of choice uses isotopically labeled target PFCs and reverse-phase liquid chromatography coupled to an electrospray tandem mass spectrometer.9 MS/MS is necessary as, lacking elemental detection, compounds must be identified based on their fragmentation products, as well as their molecular weight. Although fluorine is the 13th most abundant element in the earth’s crust, today only a dozen naturally occurring FOCs are known.10 The central reason for this is the lack of a sensitive fluorine-specific methodology capable of screening biological extracts and environmental samples without prior molecular information. So far only 19F-NMR is fluorine-specific, though this method lacks sensitivity, while mass spectrometry, sensitive to all ionizable compounds, is not element-specific.11 Even intensive high-resolution data analysis of mass deficiencies and isotope distribution is insufficient to identify trace quantities of FOCs among hundreds of other organic molecules in a biological or environmental sample. This is due to the insufficient mass deficiency effect of fluorine and the fact that fluorine is monoisotopic and does not display a distinct isotopic pattern. Hence no MS data mining algorithm can detect potential fluorine-containing compounds. This is of particular importance in pharmacological studies where the biotransformations of organofluorine based drugs such as the antimalarial drug mefloquine (Lariam) need to be displayed.12 Mass balance approaches comparing the total fluorine content of the identified PFCs and the total fluorine content of an organic extract have revealed that only a fraction of the FOCs in the sample are detected.8,13,14 Furthermore, this approach is time-consuming, requiring the conversion of all FOCs into fluoride then determining fluoride content by ion chromatography with conductivity,15 or forming a complex with aluminum (AlF+) and detecting via inductively coupled plasma mass spectrometry (ICP-MS),16 or using gas chromatography after the formation of volatile fluorosilanes,17 or through high resolution high temperature molecular absorption using continuum source high resolution atomic absorption spectrometer (abbreviated throughout as CSMAS)18 All methods presently available convert FOCs to fluoride or other derivatives before chromatography is applied, hence all species information is lost. This latter method (CSMAS) utilizes online pyrolysis and subsequent formation of metal monofluorides such as AlF, InF or GaF at high temperature, as these monofluorides absorb light between 200 and 900 nm.19,20 Hence, total fluorine has been measured in solids,21 water, toothpaste,22 tea infusion23 or even in cancer cells24 indirectly by hot temperature molecular absorption spectrometry (MAS) using a commercially available CS-AAS instrument..



EXPERIMENTAL SECTION Initially, the CS-MAS was optimized for fluorine detection by identifying the optimal conditions under which the GaF molecular absorption band was of maximal intensity for each compound. This was tested by measuring each compound’s response in a 50:50 methanol/water mixture. As methanol gradients are often used in the mobile phase during RP-HPLC the influence of the methanol solvent was measured. An RP-HPLC method was developed and optimized to achieve separation of fluoride and each of the two compounds within 10 min, with a peak width of approximately 1 min. Figure 1 shows the instrumental setup of HPLC−CSMAS/ESI-MS which has been used for identification and quantification of the FOCs in the sample.

Figure 1. Analytical workflow and instrumental setup for the screening of novel fluorinated organic compounds (FOCs) in environmental and biological samples in order to determine the f luoronome.

The limits of detection as well as linear range for the FOCs were determined and compared to those achieved by state-ofthe-art analytical methods for targeted FOC determination. Finally, a sample of groundwater−water previously established as containing a large quantity of fluorine was spiked with known concentrations of model FOCs to test recovery and highlight this method’s use in screening for novel FOCs in environmental samples. Materials and Methods. Chemicals. All chemicals used were of analytical grade. Ultrapure water was used throughout (Milli-Q water 18 MΩ cm, M-Millipore, U.K.). Chemicals used 6214

dx.doi.org/10.1021/ac301201y | Anal. Chem. 2012, 84, 6213−6219

Analytical Chemistry

Article

fluoride content of the groundwater obtained from deep boreholes has been reported previously.25 The sample was spiked with known concentrations of TFBA and FICA (250 μg/L) which amounts to approximately 1/50th of total fluoride concentration.

for HPLC separation included potassium fluoride, 2, 4, 6trifluoro-benzoic acid (TFBA) and 5-fluoroindole-2 carboxylic acid (FICA). Chemicals used for AAS modifiers, reported by Gleisner et al.,17 included gallium nitride (99.9%, Sigma-Aldrich U.K. Ltd.), sodium acetate (May & Baker, U.K. Ltd.), zirconium (CPI International, USA Ltd.), palladium nitrate (99.999%, Sigma-Aldrich U.K. Ltd.) and magnesium nitrate (Fluka Analytical, U.K. Ltd.). Standard Solutions and Calibration. Initial solutions contained 10 mg of F/L TFBA, FICA, and potassium fluoride. For calibration these were diluted to 0.1, 0.5, 1.0, 10, 20, 50, 100, 250, and 500 μg F/L. UV calibration was carried out with FICA only. CS-MAS parameters for measuring fluorine as GaF were taken from Gleissner et al.18 Gallium nitrate concentration was optimized to be 10 g/L (Supporting Information Figure S1). Instrumentation. HPLC Separation Conditions. HPLC was performed with an Agilent 1200 system fitted with a C8 (4.6 × 150 nm, Agilent) reversed phase analytic column. The mobile phase was an isocratic 50:50 mixture of HPLC-grade MeOH and ultrapure water, injection volume was 100 μL, flow rate was 1.5 mL/m, all compounds were eluted within 10 min, UV absorption was measured at 254 nm. Fraction collection took place once every 10 s across the 10 min runtime; 60 samples altogether. Fractions were then placed in the autosampler of the CS-MAS and subsequently measured for fluorine content. Continuum Source High Resolution Molecular Absorption Spectrometry (CS-MAS). A commercially available contrAA 700 HR-CS-AAS (Analytik Jena, Jena, Germany) was used. The atomizer was equipped with a transversely heated graphite tube. Heating was via ohmic resistance, powered by a low-voltage high-current supply. Continuum radiation source was provided by a special high-pressure 300-W xenon short-arc lamp capable of emitting radiation over a range of 190 to 900 nm. A double monochromator, consisting of a prism premonochromator and an echelle grating monochromator, was used to identify the analytical line of organofluorine compounds at high resolution, by providing a spectral bandwidth per pixel of about 1.5 pm at 200 nm and a linear charge coupled device (CCD) array detector with 588 pixels. Measurement was conducted as previously described by Gleisner.18 Zirconium coating was applied out three times on a new PIN-platform (Analytic Jena, Part NO. 407-A81.025), with a 50 μL injection of 1 g L−1 Zr solution. Pretreatment by chemical modifiers was carried out to stabilize organofluorine compounds at a pyrolysis temperature sufficient to remove the majority of the matrix components (Supporting Information Table S1). The graphite furnace’s temperature program is detailed in Supporting InformationTable S2. Since the absorption of GaF at high temperature was observed, rather than an atomic line, it has often been described as continuum source molecular absorption spectrometry (CSMAS).18,22 HPLC-ESI-MS. A secondary separation was carried out under identical conditions to those described above with the eluent flow directed to an ESI-MS (Agilent MSD XCT). This was operated in positive mode, with MS/MS selection set to automatic. Scanned mass range was 100−2000 amu, with an ion spray voltage of 4500 V. Parameters were optimized for the detection of TFBA and FICA ions at m/z ratios 177 and 180, respectively. Groundwater Sample. A groundwater sample was collected, as part of an ongoing project in the Upper East Region of Ghana (10°54′23.88″N 0°48′17.66″W), a region in which high



RESULTS AND DISCUSSION Pyrolysis and molecule formation temperatures of 550 and 1150 °C respectively were found to give the highest sensitivity

Figure 2. Calibration of individual fluorinated organic compounds (2,4,6,-trifluorobenzoic acid (TFBA) and 5-fluoroindole-2-carboxylic acid (FICA)) expressed as fluorine concentration.

Figure 3. Solvent effect of potential mobile phases on the sensitivity of fluorine of a fluorinated organic compound TFBA using CS-MAS as GaF (218.248 nm). Concentration of fluorine was around 150 ng F/ mL as TFBA, expressed as relative sensitivity with regards to the GaF absorbance in 100% water in percent. Error bars represent SD (n = 3).

for 100 ng F/L as TFBA without increased blank signal (Supporting Information Figure S2). Similar optimized temperatures were found for FICA. The response curves with regards to molarity of FICA and TFBA are shown in the inlet of Figure 2. The difference in the response is explained by the 3:1 molar ratio of fluorine in the two model compounds. When absorbance is related to fluorine concentration, the calibration curves show near identical gradients up to a concentration of at least 500 ng F/mL (Figure 2). This means that none of the FOCs have evaporated from the graphite tube and the fluorine−carbon bond has been quantitatively or at least to the same degree cleaved during online pyrolysis for both TFBA and FICA; hence the response is not compound but fluorine6215

dx.doi.org/10.1021/ac301201y | Anal. Chem. 2012, 84, 6213−6219

Analytical Chemistry

Article

Figure 4. Chromatogram of TFBA and FICA using RP-HPLC coupled to CS-MAS for specific fluorine detection at 211.248 nm for molecular absorption of GaF and online UV absorption at 254 nm for aromatics in a presence of fluoride.

Figure 5. Quantification of individual FOCs in a mixture of FOCs. RP-HPLC-CS-AAS chromatograms of a mixture of TFBA and FICA (a) and the calibration curves of TFBA (b) and FICA (c) in a mixture using the fluoride specific signal of GaF from CS-MAS when separated using RP-HPLC.

followed by its detection using chromatography.8,13,14 However, can CS-MAS also be used as a fluorine-specific detector for chromatography for unchanged FOCs which are often separated by RP-HPLC methods? As RP-HPLC requires a nonpolar solvent, response of the fluorine signal must be tested in the presence of varying concentrations methanol and ethanol. TFBA was prepared in solutions of 20−100% methanol and 20−100% ethanol and the absorbance compared to the response in 100% water. The relative response is shown in Figure 3. It appears that ethanol changes fluorine detection only slightly, but 80% methanol reduces the GaF absorbance signal to approximately 75%. This loss of signal, while significant, is far less than that observed for measurement of elements with high ionization potential such as As, Br, and P by ICP-MS.26−28 This supports the possibility of developing mobile phase solvent gradient programmes for optimizing the separation of different FOCs. Here however an isocratic mobile phase was judged sufficient to separate the two model compounds. A mixture of TFBA and FICA (both 10 mg F/ L) was separated in the presence of fluoride. This can be seen

specific. If this result can be extrapolated from the response of the two model compounds to all other fluorinated organic compounds, this method could offer the opportunity to measure total fluorine content independent of its molecular forms. The correlation between the results from elementspecific and species-specific calibration is nearly R2 = 1 as shown in Supporting Information Figure S3. This means that one compound can be the calibrant for another fluorine containing compound, making a mass balance of all FOCs in a sample possible. Standard deviation of the blank signal and the lowest concentrations is in the region of 2−3%. By calculating three times the standard deviation of the blank signal a detection limit of subppb fluorine (