A

eapr(1oalrd©

K

1

cmameeea

0d

Available online at www.sciencedirect.com

Journal of Chromatography A, 1175 (2007) 106–111

Simultaneous determination of gasoline oxygenates and benzene,toluene, ethylbenzene and xylene in water samples using

headspace-programmed temperature vaporization-fastgas chromatography–mass spectrometry

Jose Luis Perez Pavon ∗, Miguel del Nogal Sanchez,Marıa Esther Fernandez Laespada, Bernardo Moreno Cordero

Departamento de Quımica Analıtica, Nutricion y Bromatologıa, Facultad de Ciencias Quımicas, Universidad de Salamanca, 37008 Salamanca, Spain

Received 28 May 2007; received in revised form 27 September 2007; accepted 4 October 2007Available online 22 October 2007

bstract

A sensitive method is presented for the fast analysis of seven fuel oxygenates (methanol, ethanol, tert-butyl alcohol (TBA), methyl tert-butylther (MTBE), ethyl tert-butyl ether (ETBE), tert-amyl methyl ether (TAME) and diisopropyl ether (DIPE)) and benzene, toluene, ethylbenzenend p-xylene (BTEX) in water samples. The applicability of a headspace (HS) autosampler in combination with a GC device equipped with arogrammable temperature vaporizer (PTV) and a MS detector is explored. The proposed method achieves a clear improvement in sensitivity withespect to conventional headspace methods due to the use of the PTV. Two different packed liners with materials of different trapping strengthsglass wool and Tenax-TA) were compared. The benefits of using Tenax-TA instead of glass wool as packed material for the measurement of the1 compounds emerged as better signal-to-noise ratios and hence better detection limits. The proposed method is extremely sensitive. The limitsf detection are of the order of ng/L for six of the compounds studied and of the order of �g/L for the rest, with the exception of the most polarnd volatile compound: methanol. Precision (measured as the relative standard deviation for a level with an S/N ratio close to 3) was equal to or

ower than 15% in all cases. The method was applied to the determination of the analytes in natural matrixes (tap, river and sea water) and theesults obtained can be considered highly satisfactory. The methodology has much lower detection limits than the concentration limits proposed inrinking water by the US Environmental Protection Agency (EPA) and the European Union for compounds under regulation. 2007 Elsevier B.V. All rights reserved.

alysis

vtih

so

eywords: Headspace analysis; Programmed temperature vaporizers; Water an

. Introduction

Fuel oxygenates are generally added to gasoline to increaseombustion efficiency and to reduce air pollution. Com-only used oxygenates in recently developed gasolines include

liphatic alcohols and methylethers. Typical examples areethanol, ethanol, tert-butyl alcohol (TBA), methyl tert-butyl

ther (MTBE), ethyl tert-butyl ether (ETBE), tert-amyl methyl

ther (TAME) and diisopropyl ether (DIPE) [1]. Among thethers, MTBE is the most commonly used octane enhancernd one of the organic compounds with the highest production

∗ Corresponding author. Fax: +34 923 294483.E-mail address: jlpp@usal.es (J.L. Perez Pavon).

iih[ace

021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2007.10.044

; Fuel oxygenates; BTEX

olume worldwide. It was incorporated in gasoline to replaceetraalkyl lead and as result of increasing restrictions on aromat-cs contents. MTBE is a persistent water contaminant due to itsigh water solubility and slow rate of degradation [2,3].

The presence of these chemicals in water is related to fuelpills, leakage from underground storage tanks, and the releasef unburned fuel directly into the atmosphere and surface water.

The BTEX content [4] in a standard gasoline blend is approx-mately 18% (w/w). From a toxicological point of view, benzenes the most hazardous component because it is a confirmeduman carcinogen. The US Environmental Protection Agency

5] includes BTEX in its list of contaminants in drinking waternd limits the maximum concentration levels (MCLs) of theseompounds as follows: benzene (5 �g/L), toluene (1 mg/L),thylbenzene (0.7 mg/L) and xylenes (10 mg/L). The European

romat

Utf

eam[atidmma

tpvgcstHcaucetv

fsfs(m

2

2

seatOzm

2

pc

1a

towtStS

2

2

swf3d9Ttg

2

G(sc

wAhthe liner to the capillary column (0.60 min). The split valve wasthen opened and the liner temperature was held at 250 ◦C for9.00 min. The experimental conditions are shown schematicallyin Fig. 2.

J.L. Perez Pavon et al. / J. Ch

nion has also included benzene in its list of 33 priority pollu-ants in waters [6] and has established a limit MCL of 1 �g/Lor benzene in drinking water [7].

Gas chromatography is one of the techniques most widelymployed to quantify mixtures of organic compounds. In thenalysis of fuel oxygenates and BTEX in water, gas chro-atography is generally combined with mass spectrometry (MS)

8–14] or flame ionization detection (FID) [15–19]. Prior tonalysis, extraction and preconcentration of compounds fromhe aqueous samples are generally required. Analytical methodsnclude purge and trap (P&T) [8–10,], headspace solid-phaseynamic extraction (HS-SPDE) [11], headspace solid-phaseicroextraction (HS-SPME) [12,15,17,18], headspace solventicroextraction (HS-ME) [16], headspace (HS) [14,20] or direct

queous injection (DAI) [13,19].Recently, a new alternative to improve sensitivity, main-

aining the simple headspace (HS) instrumentation, has beenroposed. It consists of the use of a programmed temperatureaporizer inlet (PTV) to inject the samples into the chromato-raphic column [21–23]. The conditions are chosen such that theomponents are retained in the liner by cold trapping, while theolvent is eliminated through the split line. This has the advan-age that it is possible to inject large volume coming from theS. The PTV injector is equipped with an efficient heating and

ooling system, in which cooling is carried out by means ofir, liquid nitrogen, carbon dioxide, or by electrical systems. Bysing liners [24] packed with different materials, the range ofomponents that can be trapped in the liner can be significantlyxtended. This setup, not very used up to date, offers benefits inerms of simplicity and automation possibilities together with aery good sensitivity.

In the present work, we propose the determination of sevenuel oxygenates and BTEX in water samples with a highly sen-itive method, using a programmed temperature vaporizer inletollowed by fast capillary gas chromatography coupled to masspectrometry in the selected ion-monitoring mode acquisitionPTV-fast GC/MS(SIM)). Previously, two packed liners withaterials of different trapping strengths were compared.

. Experimental

.1. Chemicals

The solvents used were purchased from the followingources: methanol was from Merck (Darmstadt, Germany);thanol from Scharlau (Barcelona, Spain); tert-butyl alcoholnd diisopropyl ether from Fluka (Buchs, Switzerland); methylert-butyl ether, toluene, ethylbenzene and p-xylene from Acrosrganics (Geel, Belgium); ethyl tert-butyl ether and ben-

ene from Sigma–Aldrich (Steinheim, Germany) and tert-amylethyl ether from Supelco (Bellefonte, PA, USA).

.2. Standard solutions and samples

Solutions of the 11 analytes indicated in Section 2.1 wererepared in ultra-pure water. They were employed to obtain thealibration curves and detection and quantification limits.

ogr. A 1175 (2007) 106–111 107

To perform the measurements, 5 mL of sample was placed in0 mL vials sealed with silicone septum caps. Each sample wasnalyzed in triplicate.

The calibration curves obtained in ultrapure water were usedo predict the concentrations of these compounds in a samplef tap water. The determination was also performed with seaater and river water using the standard additions method. The

ap water sample was taken from the public water system ofalamanca (Spain); that of sea water from Salou (Spain) and

hat of river water from the River Tormes in its passage throughalamanca. Each sample was analyzed in triplicate.

.3. HS-PTV-fast GC–MS measurements

A schematic diagram of the apparatus used is shown in Fig. 1.

.3.1. Headspace samplingHS sampling was performed with a model 7694 headspace

ampler from Agilent Technologies. This sampler is equippedith a tray for 44 consecutive samples and an oven with positions

or six sample vials. The oven temperature was kept at 90 ◦C for0 min. The sampling system consisted of a stainless steel nee-le, a 316-SS six-port valve with a 3 mL nickel loop (heated to5 ◦C), and two solenoid valves (for pressurization and venting).he headspace sampler was coupled to a PTV injector through a

hermostatted transfer line heated to 100 ◦C (Fig. 1). The carrieras was helium N50 (99.995% pure; Air Liquide).

.3.2. Programmed-temperature vaporizationAll experiments were carried out in a PTV inlet (CIS-4;

erstel, Baltimore, MD, USA). Two different packed linersdeactivated glass wool and Tenax-TA) were compared. Theolvent vent injection mode was used in all cases. Cooling wasarried out by means of CO2.

The injector starting temperature was 5 ◦C. The vent flowas adjusted to 50.0 mL/min and the vent pressure to 5.00 psi.fter 1.70 min, the split valve was closed and the liner was flash-eated at 12 ◦C/s to 250 ◦C. The analytes were transferred from

Fig. 1. Schematic diagram of the apparatus used.

108 J.L. Perez Pavon et al. / J. Chrom

2

l(p4tTtgs

2

5a2p

2oTvbNs

2fi(iaa

e5etii2

2

Gg

3

3

3

utirpitaw

ctt

3

dwmstT3tMapoetr

s

Fig. 2. Sequence of events for solvent vent injection.

.3.3. Gas chromatographyTo perform the gas chromatographic measurements, an Agi-

ent 6890 GC device equipped with a DB-VRX capillary column20 m × 0.18 mm × 1 �m) was used. The column oven tem-erature program involved an initial temperature of 35 ◦C for.50 min; this was increased at a rate of 20 ◦C/min to 70 ◦C,hen increased at 70 ◦C/min to 190 ◦C, and held for 1.0 min.he latter temperature ramp is the maximum one permitted by

he instrumental configuration employed. The total chromato-raphic run time was 8.96 min. The experimental conditions arechematized in Fig. 2.

.3.4. Mass spectrometryThe detector was a quadrupole mass spectrometer (HP

973 N). It was operated in the electron impact mode using70 eV ionization voltage. The ion source temperature was

30 ◦C and the quadrupole was set to 150 ◦C. The analyses wereerformed in the scan and SIM modes.

.3.4.1. Scan detection mode. Initially, this detection mode wasnly used for the identification of the compounds in the samples.he m/z range was 25–120 amu, and the abundance thresholdalue was set to 0. The different compounds were identifiedy comparison of the experimental spectra with those of theIST’98 database (NIST/EPA/NIH Mass Spectral Library, ver-

ion 1.6).

.3.4.2. SIM detection mode. Six SIM groups were used. Therst contained the most abundant ions of methanol and ethanol

29, 31, 32, and 45). The second was formed by the characteristicons of tert-butyl alcohol and methyl tert-butyl ether (41, 43, 59nd 73). The third contained five m/z variables (43, 45, 57, 59nd 87) characteristic of diisopropyl ether and ethyl tert-butyl

tvv(

atogr. A 1175 (2007) 106–111

ther. The next group was formed by six m/z variables (43, 51,5, 73, 77 and 78) characteristic of benzene and tert-amyl methylther. The next group was formed by the characteristic ions ofoluene (65, 91 and 92). The last group contained the typicalons of ethylbenzene and p-xylene (77, 91, 105 and 106). Theons were acquired with a dwell time of 50 ms for groups 1 and, 30 ms for groups 3 and 4 and 20 ms for groups 5 and 6.

.4. Data analysis

Data collection was performed with Enhanced ChemStation,1701CA Ver. C 00.00 software [25] from Agilent Technolo-ies.

. Results and discussion

.1. Study of HS-PTV-fast GC–MS data

.1.1. Solvent injection parametersThe sequence of steps involved when solvent injection was

sed is shown in Fig. 2. Initially, the sample from the headspacehrough the transfer line (100 ◦C) was injected in the cooled PTVnlet (5 ◦C). The split valve was opened to allow the solvent to beemoved, whilst the analytes would remain trapped in the lineracking material. The next step involved the PTV transfer time;.e., the time during which the split valve remained closed andhe PTV injector was heated at 12 ◦C/s up to 250 ◦C. With this,nalyte desorption and transfer to the chromatographic columnere achieved.Finally, the split valve was opened again, a stream of helium

leaning the liner and hence leaving it ready for the next injec-ion. Then chromatographic separation was begun, with theemperature program also shown in Fig. 2.

.1.2. Comparison of the different packed linersA comparative study of the signals obtained when using two

ifferent packed liners was carried out. The packing materialsere glass wool, an inert support, and Tenax-TA, a porous poly-er designed to trap organics without retaining water. Fig. 3

hows the chromatogram obtained upon injecting a solution ofhe analytes in ultrapure water using a liner packed with Tenax-A. The concentrations of the analytes in the solution were 68.4,.15 and 0.700 mg/L for methanol, ethanol and TBA, respec-ively and 123, 8.75, 14.0, 35.0, 35.0, 14.0, 14.0, 14.0 �g/L for

TBE, DIPE, ETBE, benzene, TAME, toluene, ethylbenzenend p-xylene, respectively. When glass wool was used as theacking, the chromatogram corresponding to the same solutionf compounds had shorter retention times for the analytes thatluted in the first part of the chromatogram. This can be attributedo the volatility of these compounds, which are less stronglyetained in the liner when an inert packing is used.

The signals obtained with the glass wool packing were con-iderably less intense for most of the analytes. The reason is that

hese compounds were to a large extent removed through the splitalve, while they were retained by the Tenax-TA during the sol-ent removal step. The retention times and signal-to-noise ratiosfor the concentrations given in the previous paragraph) corre-

J.L. Perez Pavon et al. / J. Chromat

Fig. 3. Extracted ion chromatograms for m/z 31, 45, 59, 73, 78 and 91 for alaboratory-prepared solution containing the 11 compounds studied in ultrapurewater. The concentrations were: 68.4, 3.15 and 0.700 mg/L for methanol, ethanolaMr

si

fsti

tmtatirwTi

TRc

C

METMDEBTTEp

(

blv

aTtT8tr5eiwt

i

D

wc3tlF(dcdDto

nd TBA, respectively and 123, 8.75, 14.0, 35.0, 35.0, 14.0, 14.0, 14.0 �g/L forTBE, DIPE, ETBE, benzene, TAME, toluene, ethylbenzene and p-xylene,

espectively. A Tenax-TA liner was used.

ponding to the two materials compared in the liner are shownn Table 1, together with the boiling points of the analytes.

After the initial study, two calibration curves were built, oneor each type of liner, preparing standards in ultrapure water witheven uniformly distributed concentration levels. The concentra-ion ranges are shown in Table 2. Each standard was analyzedn triplicate and the linearity of the method was evaluated.

The variables used in the calibrations were the area underhe curve of methanol, ethanol, tert-butyl alcohol (TBA),ethyl tert-butyl ether (MTBE), ethyl tert-butyl ether (ETBE),

ert-amyl methyl ether (TAME) and diisopropyl ether (DIPE))nd benzene, toluene, ethylbenzene and p-xylene (BTEX) inhe extracted ion chromatogram for the quantitation ions shownn Table 2: m/z 31, 31, 59, 73, 45, 59, 78, 73, 91, 91 and 91,

espectively. All the calibrations showed good linear behavior,ith coefficient of determination (R2) values above 0.99.he intercept included zero in all cases, except for p-xylene,

ndicating the presence of this in the ultrapure water. This could

able 1etention times, signal-to-noise ratios, boiling points and log Kow for the 11ompounds studied

ompound tR (min) S/N Boiling pointa (◦C)

1 2 1 2

ethanol 2.17 2.20 3366 7050 65thanol 2.43 3.11 162 2820 78BA 2.98 3.66 354 2252 82TBE 3.65 4.46 185 10098 55IPE 4.29 5.03 16 3660 69TBE 4.76 5.35 23 8099 73enzene 5.73 6.13 174 18169 80AME 5.97 6.30 124 4258 86oluene 7.12 7.21 143 29157 111thylbenzene 7.68 7.71 596 22343 136-Xylene 7.74 7.76 493 12089 139

1) Glass wool liner was used; (2) Tenax-TA liner was used.a Reference [2].

rt

l

Q

wc3ps

a

3a

dwp

ogr. A 1175 (2007) 106–111 109

e attributed to trace level pollution of the air and water of theaboratory, which could not be removed, as seen upon insertingials containing only air.

For the set of 11 compounds, repeatability (n = 10) was evalu-ted at a level corresponding to an S/N ratio of approximately 3.he repeatability was satisfactory (Table 3) with an RSD equal

o or less than 15% both when the glass wool liner and when theenax-TA liner was used. In many cases, the RSD was lower than%. Likewise, repeatability was studied for higher concentra-ions: 24.4, 1.13 and 0.25 mg/L for methanol, ethanol and TBA,espectively, and 43.8, 3.13, 5.00, 12.5, 12.5, 5.00, 5.00 and.00 �g/L for MTBE, DIPE, ETBE, benzene, TAME, toluene,thylbenzene and p-xylene, respectively. The RSD (Table 3)mproved since it was less than or equal to 12% when the glassool liner was used and less than or equal to 6% in the case of

he Tenax-TA liner.The detection limits (DLs) were estimated using the follow-

ng equation:

L = 3.3σ

S

here σ is the standard deviation of peak response for 10 repli-ates (n = 10) corresponding to an S/N ratio of approximately, S is the slope of the calibration curve and 3.3 is the student-factor (n-1, 0.99). The detection limits obtained were alwaysower when Tenax-TA was used as the filler of the liner (Table 3).or the most polar compounds, with shorter retention timesmethanol, ethanol, tert-butyl alcohol) the improvement in theetection limit ranged between 3- and 11-fold. For the remainingompounds, a more significant increase was observed. The limitecreased 80-, 333-, 500-, 100-, 40-, 200- and 35-fold for MTBE,IPE, ETBE, benzene, TAME, toluene, and ethylbenzene. In

he case of p-xylene, a compound that showed an ordinate at therigin in its calibration line the limit decreased six-fold.

The higher peak response of measurements with Tenax-TAesulted in higher signal-to-noise ratios and hence lower detec-ion limits.

The quantitation limits (QLs) were estimated using the fol-owing equation:

L = 10σ

S

here σ is the standard deviation of peak response for 10 repli-ates (n = 10) corresponding to an S/N ratio of approximately, and S is the slope of the calibration curve. The results forrecision, and detection and quantitation limits in both cases areummarized in Table 3.

In light of the results obtained, we decided to use Tenax-TAs material for the PTV inlet liner.

.2. Determination of the 11 compounds in differentqueous matrices

To check the predictive capacity of the calibration curves,ifferent aqueous matrices were employed: tap, sea and riverater. Initially, we studied the possible presence of these com-ounds in the samples from the chromatograms corresponding

110 J.L. Perez Pavon et al. / J. Chromatogr. A 1175 (2007) 106–111

Table 2Concentrations and m/z ratios selected for the 11 compounds studied

Compound Concentration range m/z

Quantitation ion Qualifier ion Qualifier ion

Methanol 0–83.1 (mg/L) 31 29 32Ethanol 0–3.15 (mg/L) 31 45 29TBA 0–0.700 (mg/L) 59 41 43MTBE 0–96.3 (�g/L) 73 41 43DIPE 0–10.6 (�g/L) 45 43 87ETBE 0–17.0 (�g/L) 59 87 57Benzene 0–35.0 (�g/L) 78 77 51TAME 0–42.5 (�g/L) 73 43 55Toluene 0–17.0 (�g/L) 91 92 65Ethylbenzene 0–17.0 (�g/L) 91 106 77p-Xylene 0–17.0 (�g/L) 91 106 105

Table 3Relative standard deviation and detection and quantitation limits (�g/L) for the methods compared

HS-PTV-GC–MS (1) HS-PTV-GC–MS (2)

RSDa RSDb LD LQ RSDa RSDb LD LQ

Methanol 3.4 5.3 1199 3630 5.6 3.2 387 1171Ethanol 5.6 2.7 53 160 14 3.6 5 17TBA 3.4 4.7 2 6 4.1 4.1 0.5 1MTBE 15 6.6 4 11 7.7 6.1 0.05 0.2DIPE 5.4 6.8 2 6 6.3 4.4 0.006 0.02ETBE 9.6 0.2 4 12 8.9 4.6 0.008 0.02Benzene 5.6 5.5 2 5 5.9 4.7 0.02 0.06TAME 11 2.9 4 11 15 3.5 0.1 0.3Toluene 7.3 12 2 5 5.7 3.6 0.01 0.03Ethylbenzene 7.7 12 0.7 2 2.1 2.5 0.02 0.06p-Xylene 7.0 6.9 0.6 2 3.9 3.3 0.1 0.4

(1) Glass wool liner was used; (2) Tenax-TA liner was used.

ol, et5 d p-xy

trTiwdcwtapalowiopasts

saseen that the calibration curves in ultrapure water are suitablefor predicting the content of the compounds in tap water.

In the case of analyte quantification in sea and river water,the standard additions method was used to overcome matrix

Table 4Added and predicted concentrations and confidence intervals (95% probability)in tap water when the ultrapure water model was used.

Compound Added concentration Predicted concentration

Methanol 39.1 (mg/L) 36 ± 3 (mg/L)Ethanol 1.80 (mg/L) 1.5 ± 0.1 (mg/L)TBA 0.400 (mg/L) 0.32 ± 0.02 (mg/L)MTBE 70.0 (�g/L) 53 ± 5 (�g/L)DIPE 5.00 (�g/L) 4.1 ± 0.6 (�g/L)ETBE 8.00 (�g/L) 6 ± 1 (�g/L)Benzene 20.0 (�g/L) 19 ± 2 (�g/L)

a RSD at a level corresponding to the detection limit.b RSD at the following concentrations: 24.4, 1.13 and 0.25 mg/L for methan.00 �g/L for MTBE, DIPE, ETBE, benzene, TAME, toluene, ethylbenzene an

o them and from the mass spectra of the compounds for whichetention times equal to those of the analytes were obtained.he tap water did not contain any of the compounds stud-

ed; in sea water, only p-xylene was found, while in the riverater sample, methanol, MTBE, benzene and p-xylene wereetected. A study was conducted to explore the possibility ofarrying out analyte quantification using an external standard,ith the calibration curves generated in ultrapure water, in order

o predict the concentration of the compounds in the differentqueous samples (tap, river, sea). To accomplish this, the sam-les were spiked at the laboratory with the 11 compounds studiedt the same concentrations as those used to obtain the calibrationines of ultrapure water, employing regression lines. The signalsbtained for the tap, river and sea water matrices, respectively,ere plotted on the OY axis of these lines, while the values found

n ultrapure water were plotted on the OX axis. Thus, each pointf the lines represented the values of the peak areas of a com-ound in two different matrices (tap-ultrapure, river-ultrapure

nd sea-ultrapure). Whereas the analytes studied in river andea water had slopes departing from unity in comparison withhe ultrapure water, in most cases the tap water sample affordedlopes very close to unity.

TTEp

hanol and TBA, respectively, and 43.8, 3.13, 5.00, 12.5, 12.5, 5.00, 5.00 andlene, respectively.

In light of the above results, quantification of the tap wateramples was performed with an external standard. The addednd predicted concentrations are shown in Table 4. It may be

AME 20.0 (�g/L) 14 ± 2 (�g/L)oluene 8.00 (�g/L) 7.6 ± 0.8 (�g/L)thylbenzene 8.00 (�g/L) 7.7 ± 0.9 (�g/L)-Xylene 8.00 (�g/L) 7.4 ± 0.9 (�g/L)

J.L. Perez Pavon et al. / J. Chromat

Table 5Concentration range for the standard additions and predicted concentration andconfidence intervals (95% probability) for the compounds in the sea and riverwater

Sea water

Compound Standard additions Added Predicted

Methanol 0–31.3 (mg/L) 9.77 (mg/L) 9 ± 1 (mg/L)Ethanol 0–1.44 (mg/L) 0.450 (mg/L) 0.41 ± 0.04 (mg/L)TBA 0–0.320 (mg/L) 0.100 (mg/L) 0.11 ± 0.01 (mg/L)MTBE 0–56.0 (�g/L) 17.5 (�g/L) 18.8 ± 0.6 (�g/L)DIPE 0–4.00 (�g/L) 1.25 (�g/L) 1.4 ± 0.2 (�g/L)ETBE 0–6.40 (�g/L) 2.00 (�g/L) 2.3 ± 0.3 (�g/L)Benzene 0–12.0 (�g/L) 5.00 (�g/L) 5.1 ± 0.3 (�g/L)TAME 0–16.0 (�g/L) 5.00 (�g/L) 5.5 ± 0.7 (�g/L)Toluene 0–4.80 (�g/L) 2.00 (�g/L) 1.95 ± 0.07 (�g/L)Ethylbenzene 0–4.80 (�g/L) 2.00 (�g/L) 2.2 ± 0.3 (�g/L)p-Xylene 0–6.80 (�g/L) – 2.6 ± 0.4 (�g/L)

River water

Compound Standard additions Added Predicted

Methanol 0–68.4 (mg/L) – 5 ± 1 (mg/L)Ethanol 0–2.03 (mg/L) 0.450 (mg/L) 0.47 ± 0.09 (mg/L)TBA 0–0.450 (mg/L) 0.100 (mg/L) 0.09 ± 0.01 (mg/L)MTBE 0–43.8 (�g/L) – 4 ± 1 (�g/L)DIPE 0–7.50 (�g/L) 1.25 (�g/L) 1.3 ± 0.2 (�g/L)ETBE 0–9.00 (�g/L) 2.00 (�g/L) 2.4 ± 0.3 (�g/L)Benzene 0–12.5 (�g/L) – 2.4 ± 0.4 (�g/L)TAME 0–30.0 (�g/L) 5.00 (�g/L) 5.8 ± 0.8 (�g/L)TEp

eimfi(ia(

mpwtTtimm

4

iatuid

walem

rcsmc

A

(cS

R

[

[

[[

[[[

[

[[

[

[

[

oluene 0–12.0 (�g/L) 2.00 (�g/L) 2.3 ± 0.3 (�g/L)thylbenzene 0–12.0 (�g/L) 2.00 (�g/L) 2.2 ± 0.3 (�g/L)-Xylene 0–14.0 (�g/L) – 2.3 ± 0.5 (�g/L)

ffects. The concentration range for the set of standard additionsn both aqueous matrices is shown in Table 5. All samples were

easured in triplicate. The concentration obtained and its con-dence interval (95% probability) for p-xylene in sea water was2.6 ± 0.4) �g/L (Table 5). The concentrations and confidencentervals (95% of probability) for methanol, MTBE, benzenend p-xylene in river water were (5 ± 1) mg/L, (4 ± 1) �g/L,2.4 ± 0.4) �g/L and (2.3 ± 0.5) �g/L, respectively (Table 5).

With a view to checking the possibilities of the proposedethodology for the analytes not detected initially in the sam-

les, they were spiked with these compounds. Thus, the seaater was spiked with all the analytes except p-xylene, and

he river water was spiked with ethanol, TBA, DIPE, ETBE,AME, toluene and ethylbenzene. The added concentrations andhose predicted with the standard additions method are shownn Table 5. These results reveal the applicability of the proposed

ethodology for the quantification of these compounds in wateratrixes.

. Conclusions

The proposed methodology has been successfully appliedn different types of water samples. The advantage of thispproach over previously described methods [20] is that in

he present case a PTV inlet was used. Important benefits ofsing Tenax-TA as packed material in the PTV instead of annert support, such as better signal-to-noise ratios, and betteretection limits, were found. Using solvent vent injection, it

[

[[

ogr. A 1175 (2007) 106–111 111

as possible to carry out quantitative analysis for methanolt mg/L level; ethanol, TBA, TAME and p-xylene at �g/Level, and MTBE, DIPE, ETBE and benzene, toluene andthylbenzene at ng/L level. It should be emphasized that theethod showed good precision and accuracy.The proposed method, with headspace sampling, does not

equire prior sample treatment, which reduces the errors asso-iated with this step of the analysis and at the same time highensitivity is obtained owing to the injection mode used. Theethod also has the advantage of simplicity over other precon-

entration modes, such as purge and trap.

cknowledgments

The authors acknowledge the financial support of the DGIProject CTQ2004-01379/BQU) and the Consejerıa de Edu-acion y Cultura of the Junta de Castilla y Leon (ProjectA057A05) for this research.

eferences

[1] F. Ancillotti, V. Fattore, Fuel Process. Technol. 57 (1998) 163.[2] M. Rosell, S. Lacorte, D. Barcelo, Trends Anal. Chem. 25 (2006) 1016.[3] T.C. Schmidt, Trends Anal. Chem. 22 (2003) 776.[4] A. Serrano, M. Gallego, J. Chromatogr. A 1045 (2004) 181.[5] http://www.epa.gov/safewater/contaminants. National primary drinking

water regulations, EPA 816-F-03-016, Washington DC, United States, June2003.

[6] Decision 2455/2001/EC of the European Parliament and of the Council of20 November 2001, Off. J. Eur. Commun. 331, 15 December 2001.

[7] Directive 98/83/EC of the Council of 3 November 1998, Off. J. Eur. Com-mun. 330, 5 December 1998.

[8] A. Tanabe, Y. Tsuchida, T. Ibaraki, K. Kawata, A. Yasuhara, T. Shibamoto,J. Chromatogr. A 1066 (2005) 159.

[9] M. Rosell, S. Lacorte, A. Ginebreda, D. Barcelo, J. Chromatogr. A 995(2003) 171.

10] A.K. Vickers, C. George, Analysis of low concentration oxygenates inenvironmental water samples using purge and trap concentration and gaschromatography/mass spectrometry, Agilent technologies 2003, Folsom,CA, USA.

11] M.A. Jochmann, M.P. Kmiecik, T.C. Schmidt, J. Chromatogr. A 1115(2006) 208.

12] S. Nakamura, S. Daishima, Anal. Chim. Acta 548 (2005) 79.13] L. Zwank, T.C. Schmidt, S.B. Haderlein, M. Berg, Environ. Sci. Technol.

36 (2002) 2054.14] Z. Lin, J.T. Wilson, D.D. Fine, Environ. Sci. Technol. 37 (2003) 4994.15] J. Ji, C. Deng, W. Shen, X. Zhang, Talanta 69 (2006) 894.16] N. Bahramifar, Y. Yamini, S. Shariati-Feizabadi, M. Shamsipur, J. Chro-

matogr. A 1042 (2004) 211.17] I. Arambarri, M. Lasa, R. Garcıa, E. Millan, J. Chromatogr. A 1033 (2004)

193.18] J. Dron, R. Garcıa, E. Millan, J. Chromatogr. A 963 (2002) 259.19] R. Kubinec, J. Adamuscin, H. Jurdakova, M. Foltin, I. Ostrovsky, A. Kraus,

L. Sojak, J. Chromatogr. A 1084 (2005) 90.20] J.L. Perez Pavon, M. del Nogal Sanchez, C. Garcıa Pinto, M.E. Fernandez

Laespada, B. Moreno Cordero, Anal. Chem. 78 (2006) 4901.21] J. Efer, S. Muller, W. Engewald, T. Knobloch, K. Levsen, Chromatographia

37 (1993) 361.22] M.V. Russo, Chromatographia 39 (1994) 645.

23] J.L. Perez Pavon, M. del Nogal Sanchez, M.E. Fernandez Laespada, C.

Garcıa Pinto, B. Moreno Cordero, J. Chromatogr. A 1141 (2007) 123.24] B. Kolahgar, E. Pfannkock, Tecnical Note 36, Gerstel, 2002 Mulheim/Ruhr.25] Enhanced ChemStation, G1701CA, Version C00.00, Agilent Technolo-

gies1999, CA, United States.