Journal of Chromatographic Science, Vol. 44, November/December 2006 Determination of Volatile Organic Compounds by Headspace Trap F. Barani1, N. Dell’Amico1,*, L. Griffone1, M. Santoro2, and C. Tarabella1 1Agenzia Regionale Protezione Ambiente Ligure, ARPAL, Dipartimento della Spezia, Via Fontevivo 21, 19100 La Spezia, Italy and 2Perkin Elmer Italia, Via Tiepolo 24, 20052 Monza, Italy Abstract A new analytical method for the determination of halogenated and aromatic volatile organic compounds in groundwater, mineral water, and drinking water at concentrations ranging between 1–10000 ng/L is developed. A new type of headspace sampler that combines static headspace sampling with a trap is used, yielding very low detection limits and good repeatability without carryover effects. An unexpected transformation of 1,1,2,2-tetrachloroethane into trichloroethene is observed and explained. Introduction Over the last 10 years, analytical techniques for the determination of halogenated and aromatic volatiles have improved, leading to better limits of detection (LOD) and reproducibilities. The three main historical techniques are static headspace (1–7), dynamic headspace (purge and trap) (1,3,4,8,9), and solid-phase microextraction (SPME) (10). Repeatability, detection limit, and elimination of water before the chromatographic run are the three principal variables. Static headspace has an excellent repeatability [coefficients of variation (CV) < 5%], with no interference by water, but sometimes the achieved LOD is not enough to comply with some environmental regulation. For example, it cannot be used to measure concentrations of benzene below 0.5–1 µg/L. SPME has a good LOD (below 0.5 µg/L), but a worse repeatability (CV > 10 %) when using an autosampler (10). The purge and trap technique has a better LOD than static headspace, but the elimination of water may be difficult, and the system is prone to carry-over effects. This makes it is necessary to clean the system using reagent water between every sample. The device must be washed, purged with detergent solution, rinsed with distilled water, and dried for analyses of samples containing large amounts of water-soluble materials, suspended solids, high boiling point compounds, or high levels of volatile compounds. Moreover the trap and other *Author to whom correspondence should be addressed: email [email protected]. parts of the system must be baked and purged (9). A new type of headspace sampler, the PerkinElmer Turbomatrix HS40 Trap, which combines headspace sampling with a trap (PerkinElmer, Wellesley, MA), was used in this study, as it had a good LOD (like purge and trap systems), a very good repeatability (similar to static headspace), and allowed the elimination of water before the chromatographic run. Carry-over effects were not observed in our range of concentration (1–10000 ng/L) so it was not necessary to extensively bake and purge the trap after every sample was analyzed. The Turbomatrix HS 40 Trap instrument is an autosampler for up to 40 vials that can be used to determine volatile organic compounds present in several matrices. It works with the pulsed-pressure approach, combining a slight modification of the balanced pressure principle with the use of an on-board cold and packed trap to extend its detection limits. The analysis is performed through the following steps: Equilibration The vial is warmed at a fixed temperature, defined by the sample characteristics, for a set constant time in order to reach equilibrium conditions. Pressurization The needle pierces the septum, and carrier gas (at a pre-set pressure) is allowed to enter the vial to set the internal pressure to a particular value. Simultaneously, a valve isolates the gas chromatography (GC) column, avoiding any column pressure change. This column isolation flow is manually set by the operator 10 mL/min higher than the column flow. Trap load During this step, the headspace of the vial is sent to the cold packed trap, allowing a flow of the headspace through the trap. This trap load step can be repeated up to four times for each vial (Figure 1). Trap dry-purge The cold packed trap is purged with carrier gas to eliminate water. Even if the adsorbent material is mostly hydrophobic and Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission. 625 Journal of Chromatographic Science, Vol. 44, November/December 2006 the vial equilibration temperature is kept low, a certain amount of water will remain in the trap. This would damage the capillary column and worsen the detection limits by increasing the baseline. Therefore, water vapor must be eliminated. Trap desorb and trap hold During these steps, the trap temperature is increased to the desired high value at a rate of 400°C/min to release the trapped Column isolation Valve Needle Trap Detector Column analytes. It is then kept at that value for a specified time to clean it, avoiding any possible carry-over. As soon as the trap is heated, the isolation of the column from the carrier gas flow is stopped and the GC run begins. The trap is desorbed in the backflush mode, and the analyst decides whether or not to activate the split line present at the end of the trap. During these studies several instrumental parameters were optimized to reach LODs and performances required by the Italian law. Moreover, the instability of some halogenated volatiles, especially 1,1,2,2-tetrachloroethane, were observed and explained. Experimental The instrument used was the PerkinElmer Turbomatrix Headspace 40 Trap connected by an inert heated transfer line to a PerkinElmer Clarus 500 GC with a duel channel flame ionizaVial tion detector (FID) and electron capture detector (ECD) or to a Oven PerkinElmer Clarus 500 GC–MS. HS trap GC The analyses performed in the first part of this study were done with a PerkinElmer Elite Volatiles column 60 m × 0.32-mm i.d. Figure 1. Turbomatrix HSTrap during trap load: column isolation flow is on. (1.8-µm film thickness) that was connected to the headspace trap by an inert column (1.5 m × 0.32 mm) and to the FID and ECD by an inert universal Y splitter. Helium was Table I. Composition of Standard CUS-6068, Calibration Range and tR of the carrier gas. Halogenated and Aromatic VOCs Analyzed GC conditions were: Ar–CH4 flow, 30 mL/min (as make-up gas for the ECD); air flow, 450 Low High mL/min (for the FID); H2, 45 mL/min (for the concentration concentration FID); injector temperature, 150°C; ECD temperacalibration calibration ture, 370°C; FID temperature, 300°C; oven, 35°C Conc. Bias range range tR for 12.00 min, then programmed at 5.0°C/min to Analyte µg/L µg/L (µg/L) (µg/L) min Detector 60°C, after 1 min, programmed at 17.0°C/min to 220°C for 0.0 min, and finally at 30°C/min to Hexachlorobutadiene 501 0.30.003–0.050 0.03–0.5029.45 ECD 240°C. o-Xylene 1504 8 0.075–1.500 0.75–15.0024.56 FID Optimal headspace parameters used were: vial m-Xylene 1503 8 0.075–1.500 0.75–15.0024.06 FID p-Xylene 1504 8 0.075–1.500 0.75–15.0024.06 FID temperature, 70°C; needle temperature, 100°C; Styrene 1506 8 0.075–1.500 0.75–15.0024.43 FID transfer line temperature, 130°C; trap material, Ethylbenzene 1504 8 0.075–1.500 0.75–15.0023.76 FID air toxics; trap load temperature, 40°C; trap desToluene 1503 8 0.075–1.500 0.75–15.0021.10 FID orption temperature, 320°C; thermostatation Benzene 1505 8 0.075–1.500 0.75–15.0014.57 FID time, 20 min; cycles number, 3; pressurization Bromoform 300.2 1.50.015–0.300 0.15–3.0024.11 ECD time, 1.5 min; decay time, 1.5 min; desorption Bromodichloromethane 50.2 0.30.003–0.050 0.03–0.5017.40 ECD time, 0.3 min; trap hold, 0.6 min; dry purge time, Dibromochloromethane 200.6 1 0.010–0.200 0.10–2.0021.67 ECD 12 min; cycle time, 54 min; column pressure, 33 Tetrachloroethene 50.1 0.30.003–0.050 0.03–0.5022.43 ECD psi; vial pressure, 33 psi; desorption pressure, 50 1,1,2,2-Tetracloroethane 1003 5 0.065–1.300 0.65–13.0024.68 ECD psi; purge, on; shaker, on; outlet split, on or off 1,2-Dibromoethane 10 0.10.0005–0.010 0.005–0.1022.09 ECD according to the actual concentration range (low 1,1,1-Trichloroethane 300.8 1.50.015–0.300 0.15–3.0013.43 ECD 1,2,3-Trichloropropane 1001 5 0.050–1.000 0.50–10.0024.76 ECD or high concentration, as specified below, in Table Trichloroethene 100.4 0.50.005–0.100 0.05–1.0017.28 ECD I). 1,1,2-Trichloroethane 1505 8 0.075–1.500 0.75–15.0020.86 ECD The following reagents were used: ultrapure 1,2-Dichloropropane 502 2.50.025–0.500 0.25–5.0017.07 ECD water produced by Millipore (Billerica, MA); Elix Trichlomethane 300.2 1.50.015–0.300 0.15–3.0010.59 ECD 3-MilliQ was boiled for at least 90 min, cooled, 1,2-Dichloroethane 300.6 1.50.015–0.300 0.15–3.0013.14 ECD and preserved under a nitrogen atmosphere; 1,1-Dichloroethane 1504 8 0.075–1.500 0.75–15.008.17 ECD Suprapure hydrochloric acid (30%) was supplied cis-1,2-Dichloroethene 1504 8 0.075–1.500 0.75–15.009.77 ECD by Merck (Darmstadt, Germany). trans-1,2-Dichloroethene 1504 8 0.075–1.500 0.75–15.007.52 ECD Custom Standard CUS-6068 was from 1,1-Dichloroethene 300.1 1.50.015–0.300 0.15–3.005.73 ECD (Ultrascientific, Kingstown, RI). For its composi. 626 Journal of Chromatographic Science, Vol. 44, November/December 2006 Hexachlorobutadiene o-Xylene m-Xylene p-Xylene Styrene Ethylbenzene Toluene Benzene Bromoform Bromodichloromethane Dibromochloromethane Tetrachloroethene 1,1,2,2-Tetracloroethane 1,2-Dibromoethane 1,1,1-Trichloroethane 1,2,3-Trichloropropane Trichloroethene 1,1,2-Trichloroethane 1,2-Dichloropropane Trichlomethane 1,2-Dichloroethane 1,1-Dichloroethane cis-1,2-Dichloroethene trans-1,2-Dichloroethene 1,1-Dichloroethene 225 91 91 91 104 91 91 78 173 83 129 166 83 107 97 75 95 97 63 83 62 63 61 61 61 Qualifier 1 Qualifier 2 (m/z) (m/z) 227 106 106 106 103 106 92 77 171 85 127 164 85 109 99 110 130 83 62 85 64 65 96 96 96 260 105 105 105 78 65 – 51 252 129 131 131 168 – 119 77 132 99 65 87 – – 98 98 98 r2 0.9968 0.9979 0.9993 0.9989 0.9972 0.9962 0.9994 0.9932 0.9975 0.9984 0.9991 0.9986 0.9995 0.9934 0.9992 0.9989 0.9987 0.9979 0.9996 0.9994 0.9989 0.9990 0.9987 0.9992 0.9978 mV mV Time (min) B Time (min) Figure 2. Chromatograms of blank samples by ECD (A) and FID (B). Area Tetrachloroethene Area Pressure (psi) Figure 3. Dependence of peak area on desorb pressure. A mV Analyte Target Ion (m/z) A Time (min) B mV Table II. Halogenated and Aromatics VOCs Analyzed with Selected Ion Recording GC–MS, in the Concentration Range from 0.100 to 2.000 µg/L. Target Ion and Qualifiers Used to Quantify and Correlation Coefficient Obtained Are Shown following conditions: the chromatographic column was a PerkinElmer Elite Volatiles (60 m × 0.25-mm i.d., 1.4 µm) connected to the headspace trap by an inert column (1.5 m × 0.32 mm). The carrier gas was He. The temperatures were: injector, 150°C; oven, 40°C for 12.00 min, then programmed at 5.0°C/min Time (min) C mV tion, see Table I. Standard solution A was freshly prepared by diluting 10 µL of custom standard in 1 mL of ultrapure water. Standard solution B was freshly prepared by diluting 100 µL of standard solution A in 1 mL of ultrapure water. Sodium sulfate anhydrous, volatiles free, was obtained by heating commercial prepared sodium sulfate (Baker, Deventer, Holland) at 400°C for 24 h. Two calibration curves were prepared by using standard solutions A and B, dissolved, respectively, in 0.5, 1.0, 2.0, 5.0, 7.0, and 10.0 µL of solution A (low concentration curve) and of solution B (high concentration curve) in 10 mL of ultrapure water. Then 30 µL of HCl and 100 mg of sodium sulfate under a nitrogen atmosphere was added. All samples were processed by adding 30 µL of acid (sufficient to eliminate 700 mg/L of carbonate) and 100 mg of sodium sulfate to 10 mL of sample. The HCl addition must be superior if the carbonate concentration was greater than 700 mg/L to adjust pH to less than four. This precaution was indispensable to avoid any problem with 1,1,2,2-tetrachloroethane, as better specified during the following discussion. The “high curve” was obtained by opening the outlet split, and the “low curve” was obtained by closing it. The analysis performed in the second part of this study was done working with a PerkinElmer Clarus 500 GC–MS, using the Time (min) Figure 4. Chromatograms of halogenated volatiles at different desorb temperatures: 320°C (A), 250°C (B), and an overlay, where no significant differences are visible (C). 627 Journal of Chromatographic Science, Vol. 44, November/December 2006 The first problem that occurred during the optimization process of the analytical technique was obtaining blank water, which does not contain contaminants invalidating the analysis. The normal ultrapure water produced by Millipore Elix 3–Milli Q may sometimes still contain organohalides, such as trichloromethane, trichloroethene, bromoform, etc., which are present in common Italian drinking water. Therefore, the water was boiled for 90 min to eliminate the majority of these organohalides. This boiled water was preserved under a nitrogen atmosphere in order to prevent any further contamination of organohalides, such as chloroform or dichloromethane, which can usually be found in standard laboratories. Before analysis, the vials were heated for 2 h at 180°C and then stored in a dessicator. Because of the lack of a clean room, it was not possible to analyze dichloromethane. A blank chromatogram is shown in Figure 2. Several tests were done to find the best possible pressure settings to reduce the amount of water vapor that otherwise may interfere with the normal functionality of the detectors. Different pressure levels were also tested to reduce the dry-purge time. However, the lower pressures can shorten the time, and the LOD Pressurization vial effect Time (min) B A Time (min) Area Figure 6. These chromatograms refer analysis to the analysis of two vials of the same sample run consecutively. Only the relative height of the trichloroethene (peak 1) and 1,1,2,2-tetrachloroethane (peak 2) peaks have changed. Pressure (psi) B Area Cycle number effect 1 2 Cycles number Figure 5. Dependence of peak area of tetrachloroethene from vial pressurization (A) and dependence of peak area of tetrachloroethene from cycle number (B). 628 A mV Discussion values were worse. Figure 3 shows how high pressure increases the signal. The best desorption temperature was 320°C. However, this was not a critical variable (Figure 4). Even though 250°C was enough to desorb all the analytes without any difficulty, it was still recommended to use higher temperatures in order to clean the trap from other possible interferences. To increase sensitivity without excessively increasing analysis time, the trap load step can be repeated. Figure 5B shows the importance of this parameter and the three cycles that gave good results. The vial pressure was not a critical parameter. Figure 5A mV to 60°C, after 1 min programmed at 20.0°C/min to 220°C. The MS was operated in the single ion mode, acquiring for each analyte the ions specified in Table II. For each ion, a 0.05 s dwell time was used with a 0.001 s delay between them. Headspace parameters used were the same as previously reported. Trichloroethene 1,1,2,2-Tetrachloroethane Time (min) Figure 7. Two overlaid chromatograms obtained using classical static headspace at different pH levels [pH 9 (chromatogram 1), pH 3 (chromatogram 2)]. Note how the “1,1,2,2-tetrachloroethane” peak disappears at alkaline pH and at the same time the trichloroethene peak increases. Journal of Chromatographic Science, Vol. 44, November/December 2006 mV does not show any signal increase above 30 psi. Finally, a serious problem with 1,1,2,2-tetrachloroethane was observed. The determination of 1,1,2,2-tetrachloroethane was often impossible because it partially or totally disappeared during the chromatographic run. Meanwhile, an increase in trichloroethene’s signal is observed. All official methods (1,2,4,8) describe procedures that can be applied to 1,1,2,2-tetrachloroethane, without reporting data concerning the validation for this analyte. Only Environmental Protection Agency (EPA) 5030C (8) describes a generic interference during purge and trap analysis, if the system is contaminated. Time (min) This phenomenon was first observed (as shown in Figure 6) Figure 8. Chromatogram obtained with acidification of the sample with 30 µL when working with HSTrap, and the use of a lower desorption HCl. temperature did not solve the matter. The same happened with classical headspace analysis (without trap), as shown in Figure 7 and with “syringe-type” headspace. 1 Though it is well known, 1,1,2,2-tetrachloroethane is not a Trichloroethene 1,1,2,2-Tetrachloroethane stable compound, and it decomposes to trichloroethene at high temperatures (11). 1,1,2,2-Tetrachloroethane, in a water solution, was stable at room temperature for one month, at acid/neutral pH, but it decomposed to trichloroethene by E2 elimination 2 at basic pH. The study demonstrated that the previously mentioned pheTime (min) nomenon may appear during thermostating at 70°C using neuFigure 9.Two overlaid chromatograms of the same sample obtained using a tral water. Therefore, it was necessary to acidify all solutions to headspace trap. Na2CO3 is added to the sample (chromatogram 1). An pH < 4 because under these conditions the relative height of the unspiked sample (chromatogram 2). two peaks remained constant (as shown in Figure 8). The data clearly demonstrated that the observed reaction took place during thermostatation of the vials in Table III. Concentrations are the Mean of 10 Replicates. Data Obtained the water phase and was not because of active Splitting the Sample at the End of the Column to an ECD and an FID sites present in the trap, as stated in EPA method 5030C (8). It was also noted that 1,1,2,2-tetraHigh concentration Low concentration chloroethane completely disappeared and deConcentration Concentration LOD Recovery graded into trichloroethene by adding Na2CO3 to Analyte µg/L CV% r2 µg/L (%) µg/L CV% r2 the solution [as suggested by the ISS 00/14 method (1)] in order to eliminate CO2 from Hexachlorobutadiene 0.23 7.1 0.9962 0.030 20.2 0.9916 0.0005 121 o-Xilene 7.36 2.3 0.9953 0.639 4.5 0.9982 0.0496 85 sparkling water (Figure 9). 14.56 2.4 0.9955 1.263 5.1 0.9981 0.0425 84 The conditions of acidification eliminate any m+p-Xilene 7.39 2.1 0.9955 0.617 7.6 0.9937 0.0276 82 problem with a carbonate concentration below Stirene Ethylbenzene 7.33 2.2 0.9960 0.662 3.2 0.9986 0.0056 88 700 mg/L. A variation of the pH did not cause Toluene 7.35 1.7 0.9967 0.654 3.1 0.9991 0.0362 87 problems of repeatability or anomalous increasing Benzene 7.20 2.5 0.9948 0.708 1.8 0.9976 0.0326 94 of other signals. It was also possible to determine Bromoform 1.58 1.8 0.9982 0.140 3.4 0.9985 0.0106 93 repeatability, linearity, LOD, roughness, and accu- Bromodichloromethane 0.25 2.7 0.9909 0.021 4.8 0.9974 0.0010 84 racy using a natural mineral water with a concen- Dibromochloromethane 1.06 1.4 0.9983 0.088 2.3 0.9991 0.0008 88 Tetrachloroethene 0.25 3.1 0.9908 0.031 2.7 0.9959 0.0007 122 tration of approximately 180 mg/L or carbonate. 1,1,2,2-Tetracloroethane 7.05 1.2 0.9962 0.722 1.3 0.9994 0.0026 111 Table III lists LODs (International Union of Pure 1,2-Dibromoethane 0.05 1.5 0.9987 0.004 1.7 0.9991 0.0006 81 and Applied Chemistry definition), repeatability, 1,1,1-Trichloroethane 1.50 2.1 0.9915 0.155 4.4 0.9959 0.0002 103 correlation coefficients of the calibration curves, 1,2,3-Trichloropropane 6.20 4.9 0.9985 0.536 2.6 0.9916 0.0083 107 and recovery using samples of the same mineral Trichloroethene 0.51 2.0 0.9972 0.056 2.9 0.9983 0.0025 112 1,1,2-Trichloroethane 7.95 1.4 0.9987 0.693 4.2 0.9995 0.0060 92 with 7.0 µL of solutions A and B, respectively. 2.37 8.4 0.9987 0.259 5.0 0.9958 0.0180 103 Linearity was calculated with the analysis of the 1,2-Dichloropropane 1.73 2.0 0.9989 0.174 5.7 0.9919 0.0152 116 residues, and no particular deviation was Trichloromethane 1,2-Dichloroethane 1.54 3.1 0.9936 0.123 4.7 0.9914 0.0010 82 observed. 1,1-Dicloroethane 8.67 2.5 0.9918 0.775 3.0 0.9975 0.0213 103 Robustness was estimated at the confidence cis-1,2-Dicloroethene 7.52 2.9 0.9959 0.880 4.4 0.9951 0.0839 117 level of 99%; thus, it was very interesting to trans-1,2-Dicloroethene 7.59 2.4 0.9978 0.729 2.1 0.9912 0.0909 97 observe that changing the following parameters: 1,1-Dicloroethene 1.39 2.6 0.9974 0.183 5.2 0.9969 0.0049 121 operator, desorption pressure, dry purge time, 629 Journal of Chromatographic Science, Vol. 44, November/December 2006 Acknowledgments 1 The authors wish to thank the “Agenzia Regionale Protezione Ambiente Ligure”, especially Head of U.O. Laboratori Dr. C. Grillo, for his support. 2 Time (min) Figure 10. Chromatograms obtained with different traps: 500 injections (chromatogram 1) and 30 injections (chromatogram 2). trap, vial pressure, desorption temperature of the trap, and desorption time did not influence the method. Even by changing the trap (one with about 500 injection and one with 30 injection), no significant change was noted, as shown in Figure 10. Conclusion A new method was developed for the determination of halogenated and aromatic volatiles. The method was very precise and accurate. All parameters were established according to the Italian laws on drinking, mineral water, and groundwater, except for the LOD of 1,2,3-trichloropropane, which had a limit of 1 ng/L, but only 8 ng/L was reached. One nanogram per liter could not be reached because of elevated noise at the retention time of 1,2,3-trichloropropane, which was caused by high column bleeding. The effect of pH on the analysis, which is usually underestimated in the standard methods [EPA 5030B (7) or UNI EN ISO 10301 (2)], was observed and described. It was demonstrated that the transformation of 1,1,2,2-tetrachloroethane into trichloroethene must be inhibited using an acidic pH by adding hydrochloric acid to every sample. Other recent studies have noted that a high pH (> 11 to 12) also influenced other halogenated volatiles. The determination of very volatiles halogenated substances (e.g., vinyl chloride and chloromethane) involving different instrumental parameters is in progress. 630 References 1. M. Ottaviani and L. Bonadonna. Metodi Analitici per le Acque Destinate al Consumo umano, 2nd Vol. Istituto Superiore di Sanità, Rome, Italy, 2000, pp. 15–34. 2. UNI EN ISO 10301:1999. Determinazione degli idrocarburi alogenati molto volatili. Eds. UNI, Milan, Italy, 1999. 3. 5140. Solventi Organici Aromatici, Metodi Analitici per le Acque, Manuali e linee guida, 29/2003, Eds. Agenzia Protezione Ambiente e servizi Tecnici, Rome, Ital, 2003, pp. 791–98. 4. 5150. Solventi Clorurati, Metodi Analitici per le Acque, APAT, Manuali e linee guida, 29/2003, Eds Agenzia Protezione Ambiente e servizi Tecnici, Rome, Ital, 2003, pp. 799–809. 5. ISO 11423-1:1997, Water quality–Determination of benzene and some derivatives, Part 1: Head-space gas chromatographic method. Eds. ISO, 1997. 6. B. Kolbe and L.S. Ettre. Static Headspace–Gas Chromatography: Theory and Practice. Wiley-VCH, New York, NY, 1997. 7. US Environmental Protection Agency. Method 5021A, Volatile Organics Compounds in Various Sample Matrices Using Equilibrium Headspace, 1st rev. Environmental Protection Agency, Washington, D.C., June 2003. 8. US Environmental Protection Agency. Method 5030C, Purge and trap for aqueous sample, 3rd rev. Environmental Protection Agency, Washington, D.C., May 2003. 9. L.S. Clesceri, A.E. Greenberg, and A.D. Eaton. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association, Water Environment Federation, 1998, pp. 6-21–6-26. 10. Metodo Unichim 1210, Determinazione di idrocarburi volatili (VOC) e idrocarburi volatili alogenati (VOX) – Metodo per microestrazione in fase solida (SPME) e gascromatografia capillare. Eds. UNICHIM, Milan, Italy, 1997. 11. SIDS Information, CICADS 3, 1998. Available at: www.inchem.org. (January 8, 2006). Manuscript received February, 3, 2006. 007875_01