Analysis of BTEX and other substituted benzenes in water using headspace SPME-GC-FID: method validation

Abstract

The analysis of BTEX and other substituted benzenes in water samples using solid phase microextraction (SPME) and quantification by gas chromatography with flame ionization detection (GC-FID) was validated. The best analytical conditions were obtained using PDMS/DVB/CAR fibre using headspace extraction (HS-SPME) at 50 °C for 20 min without stirring. The linear range for each compound by HS-SPME with GC/FID was defined. The detection limits for these compounds obtained with PDMS/DVB/CAR fibre and GC/FID were: benzene (15 ng L⁻¹), toluene (160 ng L⁻¹), monochlorobenzene (54 ng L⁻¹), ethylbenzene (32 ng L⁻¹), m-xylene (56 ng L⁻¹), p-xylene (69 ng L⁻¹), styrene (35 ng L⁻¹), o-xylene (42 ng L⁻¹), m-dichlorobenzene (180 ng L⁻¹), p-dichlorobenzene (230 ng L⁻¹), o-dichlorobenzene (250 ng L⁻¹) and trichlorobenzene (260 ng L⁻¹). This headspace SPME-GC-FID method was compared with a previously validated method of analysis using closed-loop-stripping analysis (CLSA). The headspace SPME-GC-FID method is suitable for monitoring the production and distribution of potable water and was used, in field trials, for the analysis of samples from main intakes of water (surface or underground) and from the water supply system of a large area (Lisbon and neighbouring municipalities).

---

1. Introduction

The production of safe drinking water is an important issue and legislation has established the levels of chemical substances allowed in drinking water, whether occurring naturally, as deliberate additions or as contaminants. As a consequence of European Union legislation (Council Directive 98/83/CE) and their recent implementation for national law, the number of organic compounds in water to be monitored is higher.

Organic compounds in water derive from three major sources: the breakdown of naturally occurring organic materials, domestic and economic activities, and reactions that occur during water treatment and distribution.

Domestic and economical activities may be sources for a large number of synthetic organic chemicals (SOCs) to wastewater discharges, agricultural runoff, urban runoff, and leachate from contaminated soils. Most of the organic contaminants identified in water supplies as having adverse health concerns are part of this group. Some of the SOCs are also VOCs (volatile organic chemicals). This term “volatile organic chemical” refers to the characteristic evaporative abilities (or the vapour pressure) of these compounds, which can also apply when they are dissolved in water. Three broad groups of VOCs have been found in drinking water. One group includes compounds found in petroleum products, especially aromatics like benzene, toluene and xylenes.¹ Major sources of these compounds are leaks in fuel and gasoline tanks and piping due to deficient installation or corrosion especially in old underground storage tanks. Another group is the halogenated VOCs, used as solvents and degreasers. Their former use as septic tank cleaners also accounts for many instances of private well contamination.¹

The third group includes some of the chlorinated organic disinfection by-products, particularly the trihalomethanes. They easily volatilize from water used in the home and can be absorbed by inhalation. The air in bathrooms can represent a major source of exposure because of volatilisation from shower and bathwater. In addition, dermal absorption (across the skin) can occur in bathwater because the VOCs are lipophilic. The combination of dermal and inhalation absorption can represent as much or more exposure to VOCs as ingestion. Their lipophilicity also enables VOCs to cross into the brain and can cause a reversible anesthetic effect, including dizziness, nausea, and cardiac depression.¹

High exposure to most VOCs over many years causes brain damage and moderate exposure can alter kidney and liver function or cause damage.^1,2 The World Health Organization (WHO) defined guidelines values for BTEX and other substituted benzenes in water in order to limit human risks and to retain the aesthetic quality of drinking water: benzene (10 µg L⁻¹), toluene (700 µg L⁻¹), xylenes (500 µg L⁻¹), ethylbenzene (300 µg L⁻¹), styrene (20 µg L⁻¹), monochlorobenzene (300 µg L⁻¹), o-dichlorobenzene (1000 µg L⁻¹), p-dichlorobenzene (300 µg L⁻¹) and trichlorobenzene (20 µg L⁻¹).³ For substances that are considered to be carcinogenic, such as benzene, the guideline value is the concentration in drinking water associated with an excess lifetime cancer risk of 10⁻⁵ (one additional cancer per 100000 of the population ingesting drinking water containing the substance at the guideline value for 70 years).³ For related aromatic compounds, concentrations of the substance at or below the health-based guideline value may affect the appearance, taste or odour of the water.³

In this work, we studied the BTEX and other substituted benzenes in water. A number of aromatic and certain aliphatic hydrocarbons have been detected in ground and surface waters. The origins of these compounds in water supplies are frequently unknown and often controversial, but in the case of surface waters the sources usually are fuel and/or oil spills. In this group of compounds, benzene has been extensively studied.³ The major sources of benzene in water are atmospheric deposition (through rain and snow) and chemical plant effluents (with minor contributions from urban runoff and sewage plants).³ Acute exposure to benzene results in central nervous system depression. Although most studies of benzene toxicity have involved exposure by the inahalation route, limited animal studies suggest that exposure by other routes of administration leads to similar sequelae. Chronic exposure to benzene leads to haemopoietic tissue changes in the form of anaemia and leukopenia.^2,3

Benzene and lower alkylbenzenes are volatile and comparatively unreactive in the environment, but because of the suppression of the evaporation process, they can be present in groundwater at higher levels than those usually found in surface-water.³

Due to the concern with toxicological properties of benzene and related compounds, there is an interest in the development of simple and specific analytical procedures to measure them in water^2,4,5 and air samples.^6–9 In water samples, a pre-treatment is often necessary to isolate the components of interest from sample matrices, to purify and concentrate the analytes. Solid-phase micro extraction is a very convenient technique used for the analysis of volatile and semivolatile compounds in liquid samples. SPME requires no solvents or complicated apparatus. The theory of SPME both in immersion and or in headspace technique (HS) has been described in detail by Zhang and Pawliszyn.^10–12

The HS-SPME technique is usually applied in an equilibrated situation with the analytes distributed between the fibre coating and HS gas present in a sealed vial. In theory, the HS-SPME has several advantages: the fibre does not come in contact with the sample, the background adsorption and matrix effects are reduced, which also enhances the life expectancy of SPME fibre. The extraction by SPME is influenced by several factors: sample matrix, stirring, temperature, the amount of sample, the size of the HS vial, the ratio of the HS to aqueous phase and the position of the coated fibre in the HS, which can all affect the time required for the analyte to equilibrate between the HS vial contents and the SPME fibre coating.^11,13 The effects of these factors and the advantages of HS-SPME vs immersion-SPME have been studied by several authors using many groups of compounds and many sorts of samples.^14–24

CLSA is an alternative method for concentration of BTEX and other substituted benzenes in water samples. This method is a powerful method for trace enrichment of volatile and semi-volatile compounds in water^25–27 and was validated for these compounds.²⁸ In this method there is a continuous recycling of the sample headspace in a closed circuit through a trapping cartridge.

A method using headspace SPME-GC-FID was validated and compared with the CLSA method. The headspace SPME-GC-FID was applied to the analysis of samples of water of main intakes (surface and underground) and supply (pipelines, nets and reservoirs) of municipal water of Lisbon. These compounds are potentially toxic and therefore their presence should be monitored both in environmental water and water for human consumption.

2. Experimental

2.1. Instrumentation

The chromatographic analysis was performed using a Fisons MFC 800 series II gas chromatograph coupled to a flame ionization detector (Milan, Italy). The gas chromatograph was equipped with a split-splitless injector, operating in the splitless mode, a Spectra Physics Analytical Integrator (SP4400) and a 30 m length, 0.53 mm I.D., 3 µm film thickness DB-624 column (6% cyanopropylphenylmethylpolysiloxane, J & W Scientific, Folsom, CA). EPA suggests this column for analysing this type of compounds (EPA Method 503.1 and 502.1).

A fibre holder for manual use was purchased from Supelco (Bellefonte, USA). Microextraction fibres were also from Supelco and coated with five different films: poly(dimethylsiloxane) (PDMS) 7 and 100 µm, poly(acrylate) (PA) 85 µm, poly(dimethylsiloxane/divynylbenzene) (PDMS/DVB) and poly(dimethylsiloxane/divynylbenzene/carboxen) (PDMS/DVB/CAR). All fibres were conditioned in the hot injector of the gas chromatograph according to instructions provided by the supplier: PDMS 7 µm (320 °C/1 h), PDMS 100 µm (250 °C/0.5 h), PA 85 µm (300 °C/2 h), PDMS/DVB 50/30 µm (250 °C/0.5 h) and PDMS/DVB/CAR 50/30 µm (270 °C/2–4 h).

The stirring and heating of water samples was performed using a hot/stirring plate, DataPlate^® Digital, Biomolecular Inc (Cole-Parmer International, Vernon Hills, USA) with stirrer bars 13 mm × 3 mm from Azlon (Bibby Sterrilin, Staffordshire, UK).

The sample temperature during analysis by SPME was monitored using a 5″ thermometer for SPME from Supelco (Bellefonte, USA).

The closed loop stripping was carried out with an apparatus (Brechbühler AG, Switzerland) including a control unit, water bath thermostat, water-bath glass container, glass bottle inlet/outlet assay with 2 rotulex connections, pump, glass retainer for charcoal, precision charcoal filter and heater. This technique was optimised for BTEX and other substituted benzenes²⁸ using 1.5 mg of charcoal (particle size: 0.05–0.1 mm). After 90 min of purging, the charcoal trap was eluted with 20 µL of solvent (carbon disulfide) and 2 µL of eluate injected in a gas chromatograph.

The reagent water producing apparatus was a Millipore, Milli Q Gradient A10 (Molsheim, France).

2.2. Chemicals and standard solutions

The analytes studied [benzene, chlorobenzene, ethylbenzene, (o-, m-, p-)dichlorobenzene, styrene, toluene, 1,2,4-trichrobenzene, (o-, m-, p-)xylene and 1-chlorooctane] were supplied by Fluka, Sigma-Aldrich (Sigma-Aldrich Chemie GmbH, Germany), Carlo Erba (Carlo Erba Reagenti, Milan, Italy) and Supelco (Bellefonte, USA), quality >99,5%, gas chromatography grade. Stock solutions of selected solutes were prepared by weighing and dissolving them in acetonitrile (pesticide quality or equivalent). This stock solution was diluted (1∶1000) in acetonitrile to obtain an intermediate standard solution. Appropriate amounts of the intermediate standard solutions were added to water to give final concentrations in the µg L⁻¹ level for SPME optimization studies. These solutions were stored at 4 °C in the absence of light. For the study of the factors that may influence the extraction, the final concentrations of BTEX and other substituted benzenes in solution were in the range 0.25–0.74 µg L⁻¹. Fourteen solutions containing all standards with concentrations between 15 ng L⁻¹ and 15 µg L⁻¹ were prepared to study linear range: the approximate concentrations were 0.015, 0.025, 0.035, 0.050, 0.10, 0.25, 0.50, 1.0, 2.0, 5.0, 7.5, 10, 12.5 and 15 µg L⁻¹. For the estimation of concentration ratios, a standard solution was prepared containing: benzene (0.98 mg L⁻¹), toluene (2.9 mg L⁻¹), monochlorobenzene (2.3 mg L⁻¹), ethylbenzene (2.1 mg L⁻¹), m-xylene (2.0 mg L⁻¹), styrene (2.6 mg L⁻¹), m-dichlorobenzene (2.2 mg L⁻¹), p-dichlorobenzene (2.8 mg L⁻¹), o-dichlorobenzene (2.7 mg L⁻¹) and trichlorobenzene (2.6 mg L⁻¹).

The hydrochloric acid and sodium thiosulfate (pro-analysis grade) supplied by Merck (Darmstadt, Germany) were used for sample preservation.

2.3. Sampling procedure

For sample storage, method validation and analysis, glass vials taking 20 mm crimp seals from Supelco (10, 25 e 50 mL) were used. Vials were fitted with crimped aluminium caps lined with PTFE-coated butyl rubber septa. Vials, septa and seals were washed with detergent, rinsed with tap water and reagent water, dried at 105 °C for 1 h and stored in zones free from organic vapours. The water samples (40 mL) were collected directly into the vials, sealed and kept at 4 °C until analysis.

When there were delays between sampling and analysis, 4 drops of 6 M HCl were added to the sample (40 mL) for its preservation and for samples containing residual chlorine it was also necessary to add a reducing agent: 3 mg sodium thiosulfate were added to the sample (40 mL) for its preservation.²⁹ After collection, the samples were refrigerated at 4 °C immediately and they were analysed within 14 days.²⁹ The water samples were allowed to reach room temperature before starting the analysis.

2.4. Chromatographic conditions

The chromatographic conditions were previously optimised and validated for the analysis of these compounds.²⁸ The detector and injector temperatures were set at 300 and 200 °C, respectively. The splitless time was 90 s. A desorption time of 1 min at 200 °C was enough for a quantitative desorption of all the analytes studied as the reinsertion of the fibre after the run did not show any carry over. The carrier gas was helium at 1.8 bar and, for FID, air at 0.9 bar and hydrogen at 0.4 bar were used. The following temperature program was used: 40 °C (5 min) and 5 °C min⁻¹ to 210 °C.

On the beginning of each working day, a column blank was followed by a fibre blank and a reagent water blank to detect any possible laboratory contamination.

2.5. Solid-phase microextration procedure

For optimization of analytical conditions in the analysis of BTEX and other substituted benzenes in water by SPME, an aliquot of 40 mL of reagent water was spiked with 10 µL of an intermediate standard solution. After placing a stir bar in each vial, it was sealed with an aluminium seal with a PTFE septum. The vials were placed on a hot/stirring plate at a controlled temperature. In order to monitor temperature, a 5″ thermometer was placed in a vial containing reagent water under the same conditions and in parallel on the same hot/stirring plate. The samples were heated for 30 min to allow the equilibrium between gas and aqueous solution. For extraction, the fibre was pushed out and exposed directly to the headspace above the sample (the stainless steel needle was kept 1 cm below the septum) or by immersion in solution for 20 min. After extraction, the fibre was immediately inserted into to GC injector for desorption. A desorption time of 90 s at 200 °C was enough for a desorption of all the analytes. In each experiment, the analyte concentration was the same and for each factor studied all solutions were analysed in triplicate (N = 3).

The temperature effect in extraction using HS-SPME was studied by analysing three replicates of standard solutions under the same conditions (PDMS/DVB/CAR fibre, 40 mL, 0.5 µg L⁻¹ of each analyte, 10 min, 1500 rpm) but using different temperatures (25, 30, 35, 40, 50 and 60 °C).

The repeatability of the analytical procedure (as relative standard deviation, RSD%) was estimated analysing six replicates of reagent water spiked with an intermediate standard solution of BTEX and other substituted benzenes (5, 10, 15, 20 and 50 µL).

The apparent recovery was determined by analysing six replicates solutions containing two concentration levels of each compound. These solutions were prepared spiking 40 mL of tap water with 10 µL (N = 6) and 20 µL (N = 6) of an intermediate standard solution. The tap water without fortification was analysed in the same conditions.

The amounts of analyte present in the spiked sample and in water without fortification were estimated from the calibration curve. The difference between amount of analyte present in spiked and unspiked sample was divided by the amount of spike to calculate the apparent recovery.

2.6. Water samples

The collection of samples in fieldwork was carried out within the sampling program for the quality control of water distributor company of Lisbon, EPAL (Empresa Portuguesa das Águas Livres). The water sample (40 mL) collected in fieldwork was heated at 50 °C on a hot/stirring plate and then kept at this temperature for, at least, 30 min to allow the equilibrium between gas and aqueous solution. The fibre (PDMS/DVB/CAR) was exposed directly to the sample headspace during 20 min (the stainless steel needle was kept 1 cm below the septum) and without stirring.

3. Results and discussion

The EC Directive 98/83 has only a guideline value of 1 µg L⁻¹ for benzene and there are no guideline values for aromatic compounds related to benzene but the World Health Organization³ has guidelines values for BTEX and other substituted benzenes (presented in the Introduction). The requirements of the EC Directive in terms of detection limit of benzene and other organic compounds point for at least 25% of guideline value. Therefore, the aim of this study was to optimize a method with a low detection limit for each compound and define a linear range for each compound for screening the main intakes of water and the supply system of Lisbon. The concentration of each compound (0.25–0.74 µg L⁻¹) for optimization of SPME analysis was selected taking into account the requirements of WHO guidelines values and the requirements for detection limits for benzene of EC Directive 98/83. The concentrations are very low compared with guideline values and therefore the capacity for saturation of fibre was not studied.

Fig. 1 presents the chromatogram of a solution containing the target compounds with the corresponding retention time (t_R) and symbols used to represent them.

Fig. 1 Chromatogram of BTEX and other substituted benzenes by GC/FID. Identification of the compounds by their retention time (t_R) and symbols, including the surrogate (Surr)

1-Chlorooctane was used as surrogate in the optimization of analysis of BTEX and other substituted benzenes by CLSA²⁸ and the same compound was included in solutions used for the optimisation of conditions by SPME. 1-Chlorooctane could not be used as surrogate in SPME analysis because its behaviour is different from the analytes. However we retained the term surrogate to refer to 1-chlorooctane as it could be used to detect major changes in the fibre behaviour during the time required to carry out the experiments.

3.1. SPME fibre selection

Some preliminary tests were used to find out the most suitable fibre type considering the signal/noise ratio for each analyte. Non-polar compounds are best extracted with a poly(dimethylsiloxane) fibre (7–100 µm) and polar compounds with the poly(acrylate) fibre (85 µm)¹³ therefore the compounds studied in this work are usually analysed with the PDMS fibres. Fig. 2 shows the areas obtained for BTEX and other substituted benzenes by HS-SPME extraction in three samples, using different fibres. The analyte concentration (0.5 µg L⁻¹), the sample volume (40 mL) and the extraction conditions (40 °C, 1500 rpm, 10 min) were the same in all cases. The peak areas obtained with the PDMS/DVB and PDMS/DVB/CAR fibre coatings were higher than those obtained with the other fibres. The differences between areas obtained with these fibre coatings were larger for analytes with lower retention times, with the exception for the surrogate compound.

Fig. 2 Comparison of response HS-SPME analysis with different fibres

The responses achieved with PDMS/DVB and PDMS/DVB/CAR fibres were very similar, with the exception of benzene and toluene, with a response higher for the PDMS/DVB/CAR fibre; therefore, all subsequent experiments were performed with this fibre coating.

3.2. Effect of the temperature

The effect of temperature on extraction was studied comparing the average areas for each compound at different temperatures. The results are summarized in Fig. 3 and differences in areas are observed for all compounds with the exception of benzene. If analysis of variance (ANOVA) is applied to these results, a P-value smaller than 0.05 (0.012) is obtained at the 95% confidence level confirming that the extraction temperature is significant for this group of compounds.

Fig. 3 Influence of the temperature in extraction by HS-SPME with PDMS/DVB/CAR coated fibre

The most efficient extractions were obtained at 50 and 60 °C. Higher temperatures increase the concentration of the analytes in the headspace but decrease the partition coefficients between the fibre coating and the headspace,¹⁰ and a temperature of 50 °C appears most suitable for HS-SPME extraction of BTEX and other substituted benzenes.

3.3. Effect of the volume

In all experiments the ratio sample volume/headspace volume was kept constant using vials of 10, 25 and 50 mL for samples of 8, 20 and 40 mL respectively. For each volume, three replicates of standard solutions were analysed under the same conditions (fibre, concentration, time and stirring). The peaks in chromatograms obtained after extraction with sample volume of 8 mL were too small for a quantitative method; therefore, the results obtained with the other volumes (20 and 40 mL) were compared. The volume of sample was only significant for some of the target compounds: m-xylene, ethylbenzene, styrene and surrogate: for a sample volume of 40 ml these compounds showed higher responses, therefore, all subsequent experiments were performed with 40 ml of sample.

3.4. Effect of time

In HS-SPME, the exposure time of the fibre in the headspace is important because analysis with very short times may reduce significantly the response. The fibre was exposed directly to the headspace above the sample for 10, 15 and 20 min, under the same conditions.

There are no significant differences between the results obtained for these three extraction times (P-values >0.05 at the 95% confidence level). An exposure time of 20 min was considered suitable and convenient for a chromatographic run of 25 min.

3.5. Effect of stirring

Stirring is effective in shortening equilibration time when SPME sampling is performed by direct immersion of the PDMS fibre in the aqueous phase but when BTEX and other substituted benzenes were sampled in the HS, only a small difference in the equilibration time was found between static or well agitated aqueous samples.^10,30

Fig. 4 presents the results obtained by sampling a solution containing BTEX and other substituted benzenes from headspace phase using a PDMS/DVB/CAR fibre with and without stirring and keeping all the other conditions. There were no differences in results for static and stirred samples, except for the surrogate.

Fig. 4 Comparison of extraction of BTEX and other substituted benzenes between static and stirred samples by HS-SPME with PDMS/DVB/CAR coating fibre

3.6. Comparison between immersion and headspace SPME

Although the headspace technique may be expected as more convenient, analysis by the immersion technique was done for comparison. Fig. 5 presents results obtained in the analysis of the same solution by immersion (1500 rpm and 25 °C) and headspace (static and 50 °C).

The response obtained by HS-SPME was about nine times higher than the response obtained by direct immersion of the fibre. The highest ratio was observed for o-dichlorobenzene (∼28) and the lowest for benzene (∼3). These results confirmed HS-SPME as most suitable for the analysis of volatile organic compounds.^4,30

Fig. 5 Comparison of extraction of BTEX and other substituted benzenes between immersion and headspace SPME with PDMS/DVB/CAR coating fibre

3.7. Linearity

Taking into account the results discussed in the previous section the best conditions of extraction by SPME for BTEX and other substituted benzenes were as follows: PDMS/DVB/CAR fibre, headspace SPME, 50 °C extraction temperature, 40 mL of sample, 20 min extraction time and non-stirring. The calibration was performed by an external standard.

The linear range of FID coupled with the HS-SPME procedure for a PDMS/DVB/CAR was studied by analysis of fourteen solutions containing all standards at different concentration levels (between 15 ng L⁻¹ and 15 µg L⁻¹). Under these chromatographic conditions, m-xylene and p-xylene co-elute (t_R = 13.26 min) as do styrene and o-xylene (t_R = 14.31 min). Therefore, we prepared two groups of fourteen solutions for the linearity study: one group containing all standards including m-xylene and styrene and the second group containing all standards including p-xylene and o-xylene. The study of linearity included the statistical linearity test determining the test value PG³¹ required for the F-test. If PG ≤ F, the non-linear calibration function does not lead to a significantly better adjustment: the calibration function is linear. If PG > F, the working range should be reduced as far as possible to receive a linear calibration function; otherwise the information values of analysed samples must be evaluated using the non-linear calibration function.

For most compounds in this study, the three higher concentration levels are not well represented by a linear calibration function and these observations may be explained, as due to deviation of equilibrium or limited fibre capacity and this is confirmed by PG values (these PG values were larger than tabulated F values of Fisher/Snedecor).

The peak areas obtained for these high concentration levels were smaller than what would be expected for a linear plot. To avoid the reduction of the concentration range, a polynomial equation (second or third degree) was fitted to the calibration data. If the highest concentration is not considered, the trichlorobenzene and the p-dichlorobenzene show linear calibration functions, with PG values of 0.024 and 1, respectively [F(1,10)95% = 5].

The analysis of non-linear calibration curves of target compounds showed a good fit by an equation of second degree with squared correlation coefficients (R²) between 0.9981 and 0.9993 with the exceptions of benzene and toluene. Removing the last point, the toluene also presents a good fit with a polynomial function of second degree.

The study of linearity is required in procedures for chromatographic method validation. The fitting of non-linear calibration curve may extend the working range although this is unusual in routine analysis where dilution of samples is preferred in order to keep concentrations within the working range established according to the linearity criteria. The linear ranges, the PG values and F values of Fisher/Snedecor for each compound are given in Table 1.

Table 1 Regression data for BTEX and other substituted benzenes by optimised HS-SPME-GC/FID using a PDMS/DVB/CAR fibre^a

Compound	Linearity range/µg L^−1	R ^2	N	SE	m	SE(m)	b	SE(b)	PG	F
a R ^2 = squared correlation coefficients, m = slope, SE(b) = standard error of the intercept, N = number of data points, SE(m) = standard error of the slope, PG = test value, SE = standard error, b = intercept, F = value of Fisher/Snedecor (tabled value)
B	0.049–0.37	0.9937	4	1317	96307	5430	2841	1252	0.83	F(1,1)95% = 161
T	0.37–3.7	0.9953	6	8383	92769	2864	19607	5887	0.03	F(1,3)95% = 10
CB	0.58–2.8	0.9969	11	5142	89335	1674	2000	2149	0.1	F(1,8)95% = 5.3
EB	0.027–2.7	0.9967	11	9431	173297	3299	3861	3940	0.2	F(1,8)95% = 5.3
m-X	0.075–3.8	0.9947	9	16433	181980	4206	13522	6399	5.7	F(1,6)95% = 6.0
p-X	0.073–1.9	0.9952	8	10906	225995	6422	10416	6013	10^−4	F(1,5)95% = 6.6
St	0.13–6.4	0.9953	9	18357	117310	2957	15893	8695	1.1	F(1,6)95% = 6.6
o-X	0.059–2.3	0.9945	9	13569	219347	6209	8186	6612	5.3	F(1,6)95% = 6.0
m-dCB	0.28–5.5	0.9971	9	8141	79559	1629	8114	4338	0.04	F(1,6)95% = 6.0
p-dCB	0.35–6.9	0.9971	9	10570	82883	1686	−3034	5633	0.4	F(1,6)95% = 6.0
o-dCB	0.34–6.7	0.9963	9	10592	75934	1740	6070	5645	3.6	F(1,6)95% = 6.0
tCB	0.33–6.6	0.9959	10	9832	67666	1536	379	4657	0.085	F(1,7)95% = 5.6

---

The squared correlation coefficients (R²) of BTEX and other substituted benzenes were between 0.9937 and 0.9971. The linear range of BTEX and other substituted benzenes was one order of magnitude except for ethylbenzene and (m-, p-, o-)xylene (two orders of magnitude).

3.8. Detection and quantification limits

Taking into account the linearity ranges, two groups of solutions were prepared, one group containing BTEX and other substituted benzenes including m-xylene and styrene and the other group containing BTEX and other substituted benzenes including o-xylene and p-xylene. All compounds in these solutions have concentrations at lower values of the linear range for each target compound. Ten replicate solutions were analysed by headspace SPME-GC-FID and the standard deviations (SD) were determined based on the areas for each compound. The values of minimum detection level (MDL) and minimum quantification level (MQL) were calculated using the formula (3 × SD) and (10 × SD), respectively^31,32 and are presented in Table 2 which includes the MDL and MQL values for closed-loop-stripping analysis.

Table 2 Minimum detection level (MDL) and minimum quantification level (MQL) for BTEX and other substituted benzenes by optimised HS-SPME-GC/FID, using a PDMS/DVB/CAR fibre and by closed-loop-stripping analysis

Compound	HS-SPME		CLSA
	MDL/µg L^−1	MQL/µg L^−1	MDL/µg L^−1	MQL/µg L^−1
Benzene	0.015	0.050	0.031	0.105
Toluene	0.16	0.54	0.026	0.087
Chlorobenzene	0.054	0.18	0.025	0.083
Ethylbenzene	0.032	0.11	0.012	0.041
m-Xylene	0.056	0.19	0.018	0.061
p-Xylene	0.069	0.23	0.016	0.055
Styrene	0.035	0.12	0.017	0.057
o-Xylene	0.042	0.14	0.019	0.062
m-Dichlorobenzene	0.18	0.59	0.027	0.089
p-Dichlorobenzene	0.23	0.77	0.020	0.067
o-Dichlorobenzene	0.25	0.83	0.017	0.057
Trichlorobenzene	0.26	0.87	0.024	0.081

---

The MDL for BTEX and other substituted benzenes were in the range 0.015–0.26 µg L⁻¹ and the MQL were in the range 0.050–0.87 µg L⁻¹ by HS-SPME followed by GC/FID, using a PDMS/DVB/CAR fibre.

There are significant differences between the quantification limits obtained for these two methods (P value = 0.0021 at the 95% confidence level): they are lower for CLSA than for HS-SPME with the exception of benzene. The differences between quantification limits of the two methods were higher for the compounds with higher retention times. The compound with lowest quantification limit by HS-SPME is the compound with lower retention time, benzene. The quantification limits for the different compounds by CLSA are very similar (<0.1 µg L⁻¹).

3.9. Reproducibility

The reproducibility of the headspace SPME-GC-FID procedure was studied by analysing six replicate samples of reagent water spiked at five different concentration levels of each compound. For the purpose of this method it may be considered as acceptable a RSD of 10% or less and Table 3 shows that practically all values are below 10%. There are no significant statistical differences (P-value = 0.21) between relative standard deviations obtained for the different spiked levels; therefore, the precision of this method looks acceptable.

Table 3 Reproducibility of the optimised HS-SPME procedure using the PDMS/DVB/CAR fibre

	5 µL/40 mL (N = 6)			10 µL/40 mL (N = 6)			15 µL/40 mL (N = 6)			20 µL/40 mL (N = 6)			50 µL/40 mL (N = 6)
	µg L^−1	Area	RSD (%)	µg L^−1	Area	RSD (%)	µg L^−1	Area	RSD (%)	µg L^−1	Area	RSD (%)	µg L^−1	Area	RSD (%)
B	0.12	11120	11.0	0.25	23199	5.4	0.37	25545	4.2	0.49	31734	6.1	1.23	39704	7.1
T	0.37	34535	4.8	0.74	95312	2.2	1.10	96866	3.9	1.47	125686	3.7	3.68	297345	4.5
CB	0.27	39160	2.9	0.54	105259	3.1	0.80	107319	5.4	1.07	143278	3.0	2.68	441013	3.2
EB	0.25	37443	2.9	0.50	103251	4.3	0.75	103654	4.8	1	138809	3.4	2.50	448569	3.1
m-X	0.32	30229	4.0	0.64	97357	3.4	0.96	85390	2.7	1.28	118359	6.7	3.20	455411	3.7
St	0.28	16300	7.6	0.55	49567	5.6	0.83	41165	4.0	1.1	58093	6.9	2.75	241539	3.7
m-dCB	0.35	16024	9.0	0.69	57407	2.7	1.04	45304	3.6	1.38	64432	4.1	3.45	293275	4.6
p-dCB	0.34	15298	5.0	0.67	51915	4.6	1.01	42257	1.9	1.34	57304	3.6	3.35	251981	3.6
o-dCB	0.33	62773	2.7	0.66	111137	5.0	0.98	166376	7.4	1.31	225235	6.4	3.28	618481	6.4
tCB	0.33	11481	7.5	0.66	40398	4.4	0.99	31270	6.2	1.32	95385	7.7	3.3	219683	5.7

---

3.10. Repeated analyses on the same sample

The successive extractions on the same sample (40 mL, 0.5 µg L⁻¹ of each target compound), in the same conditions, were done in order to check the possibility of repeated analysis using the same sample, in routine analysis (see Fig. 6).

Fig. 6 Successive extractions on same sample by HS-SPME

Before two successive analyses on the same sample there was a period of 15 min of stabilization before exposure to the fibre (20 min).

In the second extraction, the response (area count) for the majority of compounds was 25–30% lower; therefore, in these conditions it is not possible to repeat the analysis on the same sample.

These results may be explained as due to depletion of the analyte occurring after each extraction. This explanation is confirmed by the observed concentration ratios discussed in next section.

3.11. Concentration ratios

The initial choice of experimental conditions is made in order to get relatively high concentrations of the analytes in the fibre. In order to compare the efficiency of the fibre for the extraction of different compounds using the proposed experimental conditions, estimated concentration ratios were calculated. The polymeric material of this fibre has a complex composition but it was assumed as homogeneous with a phase volume of 0.5 µL.

The area of a peak obtained using SPME for the spiked solution was divided by the area of the corresponding peak of the chromatogram of the standard solution (0.5 µL, direct injection) used for spiking: this quotient was then divided by the dilution factor (spiking volume/40 mL).

The values of estimated concentration ratios should be considered just as operational values because this chemical system is complex: there are three phases (sample, head space and fibre) and the fibre material contains three different polymers (PDMS, DVB and CAR). These values of calculated concentration ratios are presented in Fig. 7.

Fig. 7 Concentration ratios for different concentrations of spiked samples by HS-SPME with PDMS/DVB/CAR fibre

The concentration ratios decrease with the increase of the analytes concentration but there is no direct relationship between concentrations ratios and retention times: benzene and trichlorobenzene have low concentration ratios.

A value of ∼30000 for the concentration ratio may be used to illustrate the depletion of analyte in a sample (40 mL) containing 0.50 µg L⁻¹ of analyte. Assuming a volume of 0.5 µL of polymeric material of fibre, the concentration of the analyte in the aqueous solution will be reduced to a calculated value of 0.36 µg L⁻¹ after the first analysis. Each subsequent extraction would reduce the concentration by 28% if the concentration ratio was kept constant.

3.12. Apparent recovery

The apparent recoveries (Rec) for these compounds when analysed by HS-SPME were studied by analysing two groups of six samples of tap water (one group with m-xylene and styrene, and the other group with o-xylene and p-xylene) spiked with two concentration levels of each compound. The apparent recoveries increase with the concentration and the (m-,o-,p-) xylenes and styrene are the compounds with lowest recoveries.

The apparent recoveries of BTEX and other substituted benzenes for higher concentrations were between 68 and 106% (Table 4) with the exception of styrene.

Table 4 Apparent recovery of the optimised HS-SPME procedure using the PDMS/DVB/CAR fibre

	10 µL/40 mL			20 µL/40 mL
	µg L^−1	Rec (%)	RSD (%)	µg L^−1	Rec (%)	RSD (%)
B	0.25	102	14.8	0.49	101	1.4
T	0.74	83	8.2	1.47	94	3.4
CB	0.58	89	12.3	1.15	97	1.6
EB	0.54	70	7.1	1.07	86	1.4
m-X	0.50	56	2.8	1	68	4.5
p-X	0.59	59	36	1.17	71	18
St	0.64	5	6.6	1.28	11	17.6
o-X	0.49	67	31	0.97	78	25
m-dCB	0.55	93	17.2	1.1	98	2.0
p-dCB	0.69	90	14.4	1.38	99	1.4
o-dCB	0.67	95	14.1	1.34	106	11.5
tCB	0.66	94	25.1	1.32	102	2.1

---

Most compounds had a recovery higher than 85% for both fortification levels; however, these were better for the higher concentration (fortification of 20 µL). The precision increases with the concentration levels (when the concentration is 5–10 times higher than the minimum quantification level).

The apparent recovery of styrene was about 10%, therefore, the method is not suitable to quantify this compound. Depending on the fortification level, the apparent recoveries of xylenes vary between 60 and 70%. Since the retention times for styrene and o-xylene are the same (these compounds were not resolved on the chromatographic column), when a sample has a peak with this retention time, it will probably be m-xylene, although confirmation should be obtained by a complementary analysis: GC with mass spectrometry detector.

The m-xylene and p-xylene have similar apparent recoveries and the same retention time. If a peak occurs in the chromatogram at this retention time, a complementary analysis, with a different column separating these isomers, would be necessary to check which of these compounds is(are) present in the sample.

4. Water samples

EPAL is the largest distributor of water in Portugal supplying about 25% of the national population (Lisbon and neighbouring municipalities). A total of 225 water samples were collected between 5th June 2000 and 28th January 2001 at 21 different sampling points. For each sampling point there were at least five samples collected in different days. For the samples with smaller periodicity, the analyses were done in duplicate. Table 5 summarizes the results.

Table 5 Water analysis: samples with BTEX and other substituted benzenes

Compound	SM01A, sample/µg L^−1			SL09A, sample/µg L^−1
	00/06/08	00/07/06	00/07/13	00/07/13
Toluene	<0.16	<0.16	<0.16	0.51
Ethylbenzene	<0.03	0.63	0.23	<0.03
m-, p-Xylene	<0.06	2.49	1.23	0.20
Styrene/o-Xylene	<0.04	1.40	0.71	0.20
o-Dichlorobenzene	<0.25	<0.20	0.76	0.59
Trichlorobenzene	1.01	<0.26	<0.26	<0.26

---

These 225 samples are representative of raw and distributed waters and were collected as part of the sampling program for the quality control of EPAL.

Only four samples had measurable concentration of the analytes and three of these were collected at the same sampling site (Table 5). The values presented in Table 5 are much lower than guideline values from the World Health Organization.

5. Conclusions

The headspace SPME-GC-FID was validated for the analysis of BTEX and other substituted benzenes in water samples.

The different parameters affecting the HS-SPME of BTEX and other substituted benzenes have been studied: SPME fibres with different coatings were tested and PDMS/DVB/CAR was found as the most suitable fibre for this analysis. The best conditions of extraction by SPME for BTEX and other substituted benzenes with this type of fibre were obtained with 40 mL samples, extracted (headspace) at 50 °C during 20 min without stirring.

The optimized method gave good reproducibilities, ranging from 2.9 to 11%, 2.2 to 5.6%, 1.9 to 7.4%, 3.0 to 7.7% and 3.2 to 7.1% as determined at 0.3, 0.6, 0.9, 1.2 and 3 µg L⁻¹, respectively.

Linearity was studied over a wide range of concentrations, and the minimum detection levels (between 15 and 260 ng L⁻¹) were found suitable for analysis, considering the regulatory levels (Council Directive 98/83/CE and guidelines values of the World Health Organization³). These limits were compared with those obtained by the CLSA technique. The SPME fibre has a smaller capacity and therefore the amount of analyte extracted is smaller than with CLSA, resulting in higher detection levels and quantification limits. However, analysis by CLSA are laborious and slower (90 min per sample), a larger volume of sample (1000 mL) is necessary and carbon disulfide has to be used as solvent. Carbon dissulfide is toxic and has high affinity for laboratory environmental contaminants (it is difficult to obtain good blanks).

For screening of the BTEX and other substituted benzenes in distribution water, GC-FID is suitable for routine analyses but the use of GC-MS would be advisable to confirm identification of compounds and get further quantitative information if such information is needed: mass spectrometry is very sensitive and provides both qualitative and quantitative information.

The optimised HS-SPME procedure was applied in field trials monitoring samples from main intakes of water (surface or underground) and from the water supply system of a large area (city of Lisbon and surrounding area).

In routine analysis for control of public water supplies (raw ant treated water), just a few samples are likely to contain benzene and/or related compounds and the proposed method was easily used for screen testing.

If more detailed analysis are necessary for the few samples expected found with contaminants, the use of GC-MS would be necessary to confirm the identifications and search for possible causes of contamination.

This method can be adapted for an automated chromatographic system to make the process easier.

Acknowledgements

This work was partially supported by EPAL (Empresa Portuguesa das Águas Livres), Lisbon, Portugal.

References

1. AWWA (American Water Works Association), Water Quality & Treatment, A Handbook of Community Water Supplies, Mc Graw-Hill Handbooks, 5th edn., 1999, ch. 2.
2. Y.-P. Liu and W.-H. Ho, J. Chin. Chem. Soc., 2000, 47, 415–420.
3. W.H.O. (World Health Organization), “Guidelines for Drinking-water Quality”, Vol.II (“Health Criteria and Other Supporting Information World Health Organization”, 2nd edn., Genéve, 1993, ch. 3 and 4.
4. J. C. Florez Menéndez, M. L. Fernández Sánchez, J. E. Sánchez Uría, E. Fernández Martinez and A. Sanz-Medel, Anal. Chim. Acta, 2000, 415, 9–20.
5. Z. Polkowska, D. Gorlo, A. Wasik, M. Grynkiewicz and J. Namiesnik, Chem. Anal., 2000, 45(4), 537–550.
6. V. Parreira Fabricio, R. de Carvalho Ciomara and Z. de. L. Cardeal, J. Chromatogr. Sci., 2002, 3(40), 122–126.
7. G. Bertoni, C. Ciuchini, A. Pasini and R. Tappa, J. Environ. Monit., 2002, 4, 903–909.
8. L. Tuduri, V. Desauziers and J. L. Fanlo, J. Chromatogr. Sci., 2001, 12(39), 521–529.
9. Supelco, Application, Note 141, Sigma-Aldrich Co, 1998.
10. J. Pawliszyn, Solid Phase Microextraction, Theory and Practice, Wiley-VCH, New York, 1997, ch. 3, 4 and 6.
11. Z. Zhang and J. Pawliszyn, Anal. Chem., 1993, 65(14), 1843–1852.
12. Z. Zhang, J. Y. Min and J. Pawliszyn, Anal. Chem., 1994, 66(17), 844A–852A.
13. Supelco, Bulletin 923, Sigma-Aldrich Co, 1998.
14. W. P. David and J. Pawlizyn, J. Chromatogr., 1992, 625, 247–255.
15. E. Psillakis, A. Ntelekos, D. Mantzavinos, E. Nikolopoulos and N. Kalogerakis, J. Environ. Monit., 2003, 5(1), 135–140.
16. C. Silva Fernando, R. de Carvalho Ciomara and Z. de. L. Cardeal, J. Chromatogr. Sci., 2000, 7(38), 315–318.
17. I. Valor, M. Perez, C. Cortada, D. Apraiz and J. C. Molto, Food Sci. Technol., 2000, 102, 721–752.
18. J. Dugay, C. Miège and M. C. Hennion, J. Chromatogr. A, 1998, 795, 27–42.
19. M. F. Alpendurada, J. Chromatogr. A, 2000, 889, 3–14.
20. M. Llompart, K. Li and M. Fingas, J. Chromatogr. A, 1998, 824, 53–61.
21. J. Bartelt Robert, Anal. Chem., 1997, 69, 364–372.
22. S. P. Thomas, R. Sri Ranjan, G. R. B. Webster and L. P. Sarna, Environ. Sci. Technol., 1996, 30, 1521.
23. T. Górecki and J. Pawliszyn, Anal. Chem., 1995, 67, 3265–3274.
24. T. Nilsson, R. Ferrari and S. Facchetti, Anal. Chim. Acta, 1997, 356, 113–123.
25. C. Almeida, J. Nascimento, M. A. Cavaco, M. J. Benoliel and L. Vilas Boas, Rev. Ind. Água, 1996, 21, 14–23.
26. K. Grob, J. Chromatogr., 1973, 84, 255–273.
27. K. Grob and F. Zürcher, J. Chromatogr., 1976, 177, 285–294.
28. C. Almeida and L. Vilas Boas, et al, Optimização das Condições de Análise do Benzeno, Alquilbenzenos e Clorobenzenos em Amostras de Água, 8° Encontro Nacional de Saneamento Básico, Barcelos, Portugal, 1998.
29. APHA-AWWA-WEF, Standard Methods for the Examination of Water and Wastewater, American Public Health Association, Washington, DC, 19th edn., 1995, Parte 6000.
30. S. Fustinoni, R. Giampiccolo, S. Pulvirenti, M. Buratti and A. Colombi, J. Chromatogr. B, 1999, 723, 105–115.
31. ISO (International Organization for Standardization), ISO Standards Compendium, Environmental Water Quality, 1st edn., 1994, vol. I (General), pp. 279–287.
32. The United States Pharmacopoeia, The National Formulary, 2002, USP 25/NF 20, pp. 2256–2259.

---