Compusi Organistanici Ape

Embed Size (px)

Citation preview

  • 7/30/2019 Compusi Organistanici Ape

    1/9

    Available online at www.sciencedirect.com

    Journal of Chromatography A, 1180 (2008) 122130

    Organotin speciation in French brandies and wines bysolid-phase microextraction and gas

    chromatographyPulsed flame photometric detection

    Julien Heroult , Mate Bueno,Martine Potin-Gautier, Gaetane Lespes

    Laboratoire de Chimie Analytique Bio Inorganique et Environnement, UMR 5254,

    Universite de Pau et des Pays de lAdour, B.P. 1155, 64013 PAU Cedex, France

    Received 18 July 2007; received in revised form 16 October 2007; accepted 27 November 2007

    Available online 4 December 2007

    Abstract

    An analytical method devoted to organotin compounds (OTC) determination in brandy and wine was developed. It is based on solid-phase

    microextraction (SPME) of ethylated organotins. The following operating factors were examined: SPME mode/nature of fibre coating, sample

    volume/dilution, and sampling time. The optimisation work led to dilute the sample in an aqueous buffer (1/11, v/v ratio) in order to satisfactorily

    decrease the matrix effects due to competitive sorption of non-OTC species onto/into fibre coating. The optimised operating conditions consist of

    polydimethylsiloxane (PDMS) coated fibre used in headspace mode for 30 min. In wines, the limits of detection (LOD) and quantification (LOQ)

    ranged from 1 to 40 and 3 to 80 ng(Sn) L1 respectively, according to the species. The analytical validation was made by evaluating the accuracy of

    OTC determination in spiked samples with various concentrations over the whole calibration range, i.e. from LOQ to 1000 ng(Sn) L1. Recovery

    was around 80110% and precision (relative standard deviation, RSD) was between 12% and 25%. Despite the presence of two chromatographic

    peaks corresponding to sulphur compounds during brandy analysis, the selectivity of the method is adequate. The analysis confirmed the analytical

    performances and applicability of the method to wine and brandy samples. The obtained results emphasise the contamination of brandy and wine

    by organotins, the storage in plastic container seeming to be confirmed as the main OTC source. 2007 Elsevier B.V. All rights reserved.

    Keywords: Organotin; Wine; Brandy; Solid-phase microextraction

    1. Introduction

    Organotin compounds (OTC) are widely used in a lot

    of human activities in agriculture and industrial processes.

    Tributyl- (TBT) and triphenyltin (TPhT) are employed as wood

    preservers or pesticides. Mono- and dibutyl- (MBT and DBT),

    mono- and dioctyltins (MOcT and DOcT) are applied as cat-

    alysts and stabilisers in plastic [poly(vinyl chloride) (PVC)]

    production. According to grape culture and wine-making, OTC

    can also be present in brandy and wine. In previous studies,

    these species were detected in various wines, with concentra-

    tions varying between 50 and 80,000 ng(Sn) L1 [16]. The

    extremely high concentrations in some Canadian wines were

    Corresponding author.

    E-mail address: [email protected](J. Heroult).

    attributed to the storage in PVC flasks [7]. Other authors did

    not find differences in butyltin concentrations in wines stored

    in plastic and glass bottles [5]. In most of the analysed sam-

    ples, MBT and DBT were determined in concentrations ranging

    from 100 to 500 ng(Sn)L1. There is no information available

    on simultaneous determination of butyl-, phenyl- and octyltins,

    and no data about phenyltin speciation in wines and brandies.

    Because of the high OTC toxicity at very low concentrations,

    sensitive and selective analytical procedures are required. Ana-

    lytical development devoted to OTC has given rise to a great

    deal of interest, especially in environmental samples. Studies

    about OTC speciation in manufactured foodstuffs such as oils,

    fruit juices or alcoholic drinks remain relatively limited, despite

    quality control of food and drink [2,48]. However, analytical

    request is increasing in more and more food and agrochemi-

    cal industries. Consequently, constant analytical improvements

    are required in order to have methods able to answer to the

    0021-9673/$ see front matter 2007 Elsevier B.V. All rights reserved.

    doi:10.1016/j.chroma.2007.11.084

    mailto:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_5/dx.doi.org/10.1016/j.chroma.2007.11.084http://localhost/var/www/apps/conversion/tmp/scratch_5/dx.doi.org/10.1016/j.chroma.2007.11.084mailto:[email protected]
  • 7/30/2019 Compusi Organistanici Ape

    2/9

    J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130 123

    need and regulation. It is especially urgent since the interna-

    tional institution dealing with food safety and normalisation

    is working to define specific regulation dedicated to OTC in

    manufactured media. The most recent analytical methods pro-

    posed for organotin determination in wines concern butyltins

    [5,8]. They were based on derivatisation andeither liquidliquid

    extraction (LLE) in hexane or headspace solid-phase microex-

    traction (HS-SPME), before gas chromatography (GC) with

    flame photometric detection (FPD) or mass spectrometry (MS)

    [5,8]. In HS-SPME, a 100m-polydimethylsiloxane (PDMS)

    coated fibre was used. The derivatisation step was achieved by

    hybridization with potassium tetrahydroborate (KBH4), Grig-

    nard propylation or ethylation with sodium tetraethylborate

    (NaBEt4). LLE led to too high limits of detection (LODs), over

    300 ng(Sn) L1. By using HS-SPME, the LOD reported were

    16200, 22100 and 1550ng(Sn)L1, respectively for MBT,

    DBT and TBT. Although HS-SPME was used, strong matrix

    effects were observed, mainly leading to important variation of

    extraction yield. Thus, the LODs could be increased five times

    according to the type of wine, e.g. from 20 to 100 ng(Sn) L

    1

    for DBT [8]. Other work dealing with volatile sulphur com-

    pounds analysis in beer related matrix effects observed in

    alcoholic drinks and induced by the presence of ethanol [9].

    Two hypotheses were given to explain the effect of ethanol on

    organic and sulphur compounds analysis: ethanol could either

    act as co-solvent where analytes remained preferentially or

    be responsible for sorption competition onto/into fibre coat-

    ing.

    Given the different projects concerning OTC regulation in a

    wide range of media, the general aim of the present work is to

    investigate an analytical procedure able to reach organotin spe-

    ciation in manufactured solutions with complex content, quitedifferent from environmental media. The challenge rests on the

    understanding of HS-SPME step, with regard to the presence of

    matrix components sufficiently volatile to be co-extracted. The

    acquired information is especially useful to complete theknowl-

    edge about theSPME fibre coating behaviour. More specifically,

    the study focuses on the optimisation of a method for simulta-

    neous determination of butyl-, octyl- and phenyltins in routine

    conditions. The media chosen were various types of wine and

    brandy. The OTC traceability required a particular attention

    regarding the sensitivity and robustness of the method. The

    reduction of matrix effects due to co-extracted volatile com-

    pounds was investigated.The analyticaldevelopment was based

    on SPME GC-pulsed flame photometric detection (PFPD), pre-viously demonstratedas suitable for OTC speciation in complex

    environmental matrices [1015].

    2. Experimental

    2.1. Instrumentation

    A Varian 1079split/splitless temperature programmable

    injector (Walnut Creek, USA) associated to a Varian 3800

    gas chromatograph coupled with a pulsed flame photomet-

    ric detector was used. The GC system was equipped with

    a 30 m

    0.32mm capillary column (film thickness 0.25m)

    coated with (5% phenyl)methylpolysiloxane, 007-5 (Quadrex,

    New Heaven, CT, USA). The operating conditions were

    optimised previously in our laboratory [15]. The following tem-

    perature program was applied for the separation of OTC: the

    column temperature was held at 80 C for the first minute,

    raised to 125 C at a heating rate of 20 Cmin1, to 190 C

    at 5 Cmin1, to 225 C at 10 Cmin1, then to 270 at

    30 Cmin1 and held at the final temperature for 3 min. The

    splitless injector port was kept at 250 C and the temperature of

    the detector was 350 C. Nitrogen was used as the carrier gas

    (2mLmin1).

    The detector operating conditions have been precisely

    described elsewhere [12]. The signal corresponds to the emis-

    sion of the Sn H molecular bond and was measured via a high

    transmission band filter (> 610 nm; OG590, Schott, France).

    According to the tin emission profile, signalacquisition was car-

    ried out with a gate delay of 3.0 ms and a gate width of 2.0ms

    after each flame ignition.

    The investigation about co-extracted matrix componentswas

    performed by using a Hewlett-Packard 6890 GCMS system(Palo Alto, CA, USA). The GC system was equipped with

    a capillary column HP-5 (30m 0.32mm I.D.) coated with

    (5% phenyl)methylpolysiloxane (film thickness 0.25m) (Agi-

    lent Technology, Massy, France). Ultrapure helium was used as

    carried gas, with a flow rate of 1.7 mL min1. The mass spec-

    trometer was operated with electron impact (EI) at 70 eV as

    ionisation potential, in positive ion mode. The injector temper-

    ature was kept constant at 250 C. The transfer line was held at

    280 C. Mass range from m/z 19500 was recorded in full scan

    mode.

    The manualSPME device was obtained from Supelco (Belle-

    fonte, PA, USA). The fibres used were coated with either100m polydimethylsiloxane (PDMS) or 75m carboxen-

    polydimethylsiloxane (CAR-PDMS) phases. These two types

    of fibres were previously found to give satisfactory results for

    OTC extraction [10,13].

    Mechanical shaking during the SPME procedure was per-

    formed using an elliptical table (KS 2502 basic) from Prolabo

    (Fontenay Sous Bois, France). The maximum adjustable shak-

    ing speed corresponded to 420 rpm. This shaking mode was

    previously found to be the most efficient [11].

    2.2. Standards and reagents

    Monobutyltin trichloride (MBT, 95%), monophenyltintrichloride (MPhT, 98%) and diphenyltin dichloride (DPhT,

    96%) were purchased from Aldrich (Milwaukee, WI, USA).

    Dibutyltin dichloride (DBT, 97%), tributyltin chloride (TBT,

    96%), triphenyltin chloride (TPhT, 95%) and tripropyltin chlo-

    ride (TPrT, 98%) were obtained from Merck (Darmstadt,

    Germany). Monooctyltin trichloride (MOcT, 97%), dioctyltin

    dichloride (DOcT, 97%) and trioctyltin chloride (TOcT, 97%)

    came from Lancaster (Strasbourg, France). Organotin standard

    stock solutions (1000mg(Sn)L1) were prepared in methanol.

    Working standard solutions (10 mg(Sn)L1) were prepared

    weekly from stock standard solutions by dilution in Milli-Q

    water (18.2M

    ) (Millipore, Bedford, MA, USA). Standard

  • 7/30/2019 Compusi Organistanici Ape

    3/9

    124 J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130

    solutions of 100g(Sn) L1 were daily prepared. All the stan-

    dards were stored in the dark at 4 C.

    Ethanoic acid, ammoniac and nitric acid were obtained from

    J.T. Baker (Paris, France). Methanol was purchased from Pro-

    labo. Sodium tetraethylborate (NaBEt4) was purchased from

    Galab products (Geesthacht, Germany). NaBEt4 was dissolved

    in Milli-Q water to provide a 2% (m/v) ethylating solution.Glassware was decontaminated overnight in 10% (v/v) nitric

    acid solution and rinsed with Milli-Q water.

    2.3. Samples

    Various brandies, white and red wines from west, south-west

    and south of France were analysed. They were chosen on the

    basis of different sample matrices, according to their production

    process and storage. Three types of brandy were studied: (1) a

    pure wine distillation product, directly put into glass flask after

    distillation, (2) a brandy let soaked with oak shavings and (3)

    a brandy corresponding to a distillation product stored in oak

    barrel for two years. The brandy alcohol percentage (AP) was40%. Sweet white wines were originating from two different

    wines (AOC type) and glass-bottled. Their AP were 12.5% and

    17%, respectively. White and Red table wines (blended; 11%

    AP), stored in plastic bottles were also analysed.

    For analytical development and optimisation, an OTC-free

    brandy sample was used. It was chosen because of its high alco-

    holic degree and the most important matrix effects observed

    during preliminary tests. It was spiked with the eight OTC of

    interest, i.e. butyl-, phenyl- and octyltins. It is called brandy-test

    sample later on.

    2.4. Analytical procedure

    An aliquot of brandy or wine sample of varying vol-

    ume (540 mL) was directly introduced into a derivatisation

    reactor. Ethylation was carried out using NaBEt4 solution

    (0.5 mL, preliminary tested and validated volume) in a sodium

    ethanoateethanoic acid buffer (0.2 mol L1, pH 4.8, volume

    varying from 20 to 80 mL) [13]. For SPME extraction, both

    direct and headspace (HS) sampling modes were used in a pre-

    liminary study. In direct mode, the fibre was immersed in the

    sample/buffer mixture immediately after NaBEt4 addition, for

    30min [11]. For HS sampling, the SPME fibre was inserted in

    theheadspace10 minafterNaBEt4 addition,stirring timeneeded

    to equilibrate liquid and gaseous phase analyte content as pre-viously studied [13]. HS-SPME sampling was then performed

    undermechanicalstirring.Differentextraction timeswere tested

    and results have been presented later on.

    After SPME sampling, the fibre was introduced into the GC

    injector port where the analytes were thermally desorbed for

    1 min. The same procedure, with exception that no sample was

    added, was applied to determine blanks. All the analyses were

    made in duplicate at room temperature (22 2 C).

    Quantification of OTC was performed by applying the

    standardadditioncalibration method usingTPrT as internalstan-

    dard. OTC standards were added into the reactor just before

    NaBEt4 addition.

    2.5. Chemometrics

    Simultaneous optimisation of operating factors was per-

    formed by using experimental design. This approach allows the

    data to be modelled and transformed into continuous informa-

    tion on the whole experimental field [16]. 3NF design(whereNF

    represents the number of studied factors) was chosen because

    it is easy to use and sufficiently robust. Three different levels

    of adjustment, respectively noted 1, 0 and +1 were used for

    each factor (X), involving N= 3NF experiments. This type of

    design also offers a sequential approach, i.e. the possibility to

    extend the experimental field during the optimisation. This is

    due to the regular distribution of the experiments in the stud-

    ied field. The standard deviation (s) of the measured responses

    (Y) was estimated by duplicating each experiment [17]. The

    effect (A) of a factor or an interaction between two factors on

    a response was calculated by using the least-squares method.

    Correspondingerror(dA) was calculated byusing themean stan-

    dard deviation (s) of the response [17]. The effect was regarded

    as significant if it was higher than the precision. The modellingrelied on a second-order polynomial fitting on the variation of

    the response as a function of the significant factors (Y=AX).

    It was validated by evaluating the precision (via the determi-

    nation coefficient, R2), the signification and lack-of-bias (by

    FisherSnedecor statistics, F-tests respectively noted Fsign and

    Fbias) [17]. All statistical tests and uncertainty calculations were

    made in a confidence level of 95%. Mathematical and statisti-

    cal calculationswere performed by using NEMRODW (2000-D

    version from LPRAI, Marseille, France) and EXCEL software.

    3. Results and discussion

    3.1. SPME optimisation

    In this part, brandy aliquots were spiked with the different

    organotins with concentrations adapted to the chromatographic

    responses of the analytes. Extracted amounts (and so extraction

    efficiency) were evaluated by considering the peak areas [11].

    3.1.1. SPME mode and fibre coating

    Direct and headspace sampling modes were first compared

    by using PDMS fibre, as this coating was previously shown to be

    well-adapted to simultaneous extraction of butyl-, phenyl- and

    octyltins from aqueous samples [11,14,15]. HS-SPME should

    allow competitive co-extraction of low-volatile non-organotinspecies into the fibre to be reduced, and has experimentally

    shown the best analytical performances, especially if several

    organotins (including butyl- to octyltins) have to be analysed

    [9,13]. According to Fig. 1, HS-SPME from the brandy-test

    sample actually provides higher peak areas in comparison with

    direct extraction. This mode was thus selected for further devel-

    opment. Because of the particular nature of the sample matrix,

    thenatureof fibre coating was also tested, especially with regard

    to the co-extraction of non-organotin species currently found in

    alcoholic drink. In the literature, both PDMS and CAR-PDMS

    fibres were used for the analyses of volatile organic and sulphur

    compounds in different beverages such as beer or fruit juices

  • 7/30/2019 Compusi Organistanici Ape

    4/9

    J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130 125

    Fig. 1. Typical GCPFPD chromatograms according to SPME mode. Other

    operating conditions: Direct SPME: 30 min sampling, 20mL sample volume;

    Head Space SPME: 15 min sampling, 20mL sample volume. Spiked sample:

    from 100 (MBT) to 2000 (TPhT) g(Sn)L1.

    [9,18,19]. From a previous study concerning aqueous samples,

    PDMS coating allowed higher quantitative simultaneous extrac-

    tion of butyl-, octyl-, MPhT and DPhT, whereas CAR-PDMS

    was showed to be well-adapted to the most volatile methyltins

    [13]. TPhT was also more efficiently extracted by using this

    fibre than with PDMS. A comparison of both fibre coatings in

    HS sampling mode for the brandy-test sample has been carried

    out. Results showed that PDMS coating gave higher OTC peaks

    than the ones obtained with CAR-PDMS fibre (up to 20 times

    higher), which is inagreementwith ourprevious results obtained

    in water samples [13]. Consequently, PDMS fibre was selectedin the analytical procedure.

    3.1.2. Operating factors

    3.1.2.1. Preliminary experiments. In HS-SPME, the mass

    transfer at fibre/headspace and headspace/solution interfaces

    directly determines the amounts of analytes extracted by the

    SPME coating [20,21]. These extracted amounts depend on the

    analyte concentrations and volumes of the three phases. The

    main limiting parameter in HS-SPME is the transport of the

    analytes from the solution to the headspace. It can be enhanced

    by an efficient shaking of thesolution, which insures continuous

    renewal of OTC at the solution surface [22].The matrix effects induced by the presence of ethanol and

    other organic volatile compounds forming aroma were first con-

    sidered. GCPFPD and GCMS analyses of a brandy sample

    and an ethanol/water mixture with the same ethanol content

    (40%) were made, each one being diluted in ethanoic buffer

    and OTC spiked. The results are presented in Fig. 2. An

    important decrease of peak areas can be observed from the

    brandy sample. According to the corresponding GCMS chro-

    matograms,different non-OTCvolatilecomponentsaredetected

    in this sample. Indeed, organic volatile compounds frequently

    present in alcoholic beverages such as fatty acids (octanoic,

    decanoic, dodecanoic acids) and brandy aromas (vitispirane

    Fig. 2. Compared GCPFPD and GCMS chromatograms from (a) an

    ethanol/water mixture (40/60, v/v), (b and c) a diluted brandy sample, with

    respectively (5/55, v/v) and (20/40, v/v) dilution. OTC concentration in solution

    (sample+ buffer): from 167 (MBT) to 1667 (TPhT) ng(Sn)L1 for GC-PFPD

    analysis and from 1.67 (MBT) to 16.7 (TPhT) g(Sn)L1 for GCMS anal-

    ysis. MS Identification: 1. ethanol, 2. octanoic acid, 3. vitispirane, 4. 1,2

    dihydro-1,1,6-trimethylnaphtalene (TDN),5. decanoic acid,6. dodecanoic acid,

    7. silicate compounds.

    and1,2 dihydro-1,1,6-trimethylnaphtalene [TDN]) areextracted

    in high quantity, simultaneously to OTC. This co-extraction

    induces the OTC sorption onto the SPME fibre to be widely

    decreased [19,23]. These results highlight that ethanol is not

    the only responsible factor for the observed matrix effects. In

    order to decrease these matrix effects, Azehna and Vasconce-

    los proposed a sample pre-treatment based on a preliminary

    extraction of OTC from wine matrix by an aqueous solution

    of tetramethyl ammonium hydroxide (TMAH) [8]. However,

    this pre-treatment had no significant effect and was rather time-

    consuming (50 min). In the present study, other pre-treatments

    were tested in order to remove co-extracted volatile species:

  • 7/30/2019 Compusi Organistanici Ape

    5/9

    126 J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130

    Fig. 3. Influence of solution volume (sample+ buffer) on extraction effi-

    ciency. Volumes (mL): headspace/[sample + buffer]: (a) 60/[20 + 40] and (b)

    120/[40+ 80]. Spiked sample: from 100 (MBT) to 2000 (TPhT)g(Sn)L1.

    heating (30 C) nitrogen degassing (for 3075min), vacuum

    degassing (5 min), ultrasonic bath (10 min), headspace vacuum

    extraction (2 min). Results obtained showed more or less impor-

    tant OTCpeak area decrease,whichis probably theconsequence

    of OTC losses during pre-treatments. Dilution of the sample in

    buffer was also tested. Because a significant improvement ofthe OTC peak areas was obtained, further investigations were

    performed and are presented later on.

    The effect of solution volume (i.e. brandy-

    test sample + buffer) was also tested using a fixed

    headspace/[sample + buffer] ratio (1/1, v/v) in reactors

    with same geometry but different total volumes, respectively

    120 and 240 mL. The positive effect of 120mL reactors on

    the HS-SPME efficiency is obvious for all organotins, as

    illustrated in Fig. 3. The use of smaller reactors leads to not

    less than a two-fold increase of OTC peak areas in comparison

    with 240 mL reactors. Such observation may be explained by

    different stirring efficiencies due to different solution volumesto be agitated as a function of reactor volume. In the present

    conditions (mechanical stirring at 420rpm), a solution volume

    of 60 mL in 120 mL reactor appears as the most convenient.

    3.1.2.2. Experimental design. According to the description of

    the analytical process (see Section 2.4.), literature and pre-

    liminary tests, the following three factors were taken into

    consideration in their respective experimental field (i.e., and

    + levels, respectively):

    (1) headspace/solution [i.e, sample+ buffer] volume ratio

    (HS/S) studied at 1/1 (60 mL/60 mL) and 1/2 (40 mL/

    80 mL);

    (2) sample/aqueous buffer volume ratio (S/B), first considered

    between 1/2 (20 mL/40 mL for 60 mL total volume of solu-

    tion and 27mL/53mL for 80 mL total volume of solution)

    and 2/1 (40 mL/20 mL for 60 mL and 53 mL/27 mL for

    80 mL);

    (3) extraction time: theexperimental field of this factorwas pre-

    liminary studied between 5 and 45 min. As no significant

    response improvement was observed over 30 min, experi-

    mental field was then considered between 5 and 30 min.

    Because of operating constraints and in order to avoid a too

    high number of experiments, the experimentation wasorganised

    in two 32 designs, each one corresponding to the low and high

    levels of factor (1) (i.e. 60 or 80 mL of solution). Consequently,

    each design addressed the two factors (2) and (3).

    The organotin chromatographic peak areas were considered

    individuallyas theresponsesto optimise.Experimental setswere

    performed with constant OTC amount [5 ng(Sn) for MBT to

    100 ng(Sn) for TPhT] in the reactor in order to evaluate the

    actual extracted OTC amount considering the matrix effects.Some selected HS-SPME time profiles of the eight analytes

    are presented in Fig. 4. They were obtained from the mod-

    elling according to the approach described in chemometrics

    part (model validation: R2 > 0.86, Fsign > 115 and Fbias < 5). In

    order to compare them, scale was kept similar between some

    of these figures. Because of the simultaneous ethylation/HS-

    SPME, extracted amounts are dependent on both ethylation and

    extraction yields, but time profiles are mainly dependent on

    extraction efficiency as no limitation is supposed to occur from

    ethylation step (no kinetic limitation and NaBEt4 in excess).

    Individual profiles present significant differences according to

    the degree of substitution and nature of the organic groupbonded to tin atom. Concerning mono- and disubtituted species,

    equilibrium generally appears to be reached after 1520 or

    2030 min, according to volume ratios. For trisubtituted com-

    pounds, extracted amounts show a continuous weak increase in

    the whole sampling time range.

    In the octyltin series, mono- and disubtituted species appear

    quantitatively extracted whatever be the experimental condi-

    tions. This may be due to their lower volatility with regard to the

    other organotins, leading to lowest extracted amounts. For the

    two other organotin series, i.e. butyl- and phenyl-, the extracted

    amount generally decreases from mono- to tri-substituted tin.

    The most important dilution of the sample (S/B = 1/2) promotes

    the HS-SPME of mono- and disubstituted species. According tothis observation, the sample/aqueous buffer volume ratio was

    then studied down to 1/11 (i.e. 5/55, v/v ratio). The results

    obtained in this extended experimental field (see Fig. 5) confirm

    that the OTC extraction efficiency is greater for higher dilution,

    especially when sample/buffer volume ratio is below 1/5. Sev-

    eral hypotheses can be made to explain these observations. On

    the one hand, as the solubility of organotins in the liquid phase

    varies according to their substitution degree, the liquid phase

    composition (mainly based on ethanol/water ratio) could lead to

    differences in species partitioning between headspace and solu-

    tion phases and then, differences in extraction yields. However,

    from preliminary experiments, similar extraction yields were

  • 7/30/2019 Compusi Organistanici Ape

    6/9

    J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130 127

    Fig. 4. Organotin HS-SPME timeprofiles plotted accordingto experimentaldesign modelling: (A) butyltins, (B) phenyltins, (C) octyltins. Solution= sample + buffer.

    Headspace/solution volume ratios (HS/S): (1/1): Solution volume= 60 mL; (1/2): Solution volume = 80 mL. Sample/buffer volume ratios (S/B): (1/2): 20mL/40mL

    (for 60mL solution) or 27mL/53 mL (for 80 mL solution); (2/1): 40mL/20 mL (for 60 mL solution) or 53mL/27mL (for 80 mL solution).

    obtained from HS-SPME sampling in water and ethanol/water

    mixture (40/60, v/v). On the other hand, discrimination between

    OTC according to their volatility could occur. This phenomenon

    has been already observed when OTC were extracted by HS-

    SPME from aqueous samples [13,14]. Moreover, in the present

    case, importantsorption competitionsalso existbetween thedif-

    ferentvolatilematrix components, as highlighted in Fig. 2. They

    even could lead to desorption of the most volatile OTC when the

    alcoholic matrix is not sufficiently diluted (as seen for S/B = 2/1,

    on Fig. 4, curves in grey lines). Sorption competition between

    non-OTC and OTC compounds could also explain the signifi-

    cant positive effect of the matrix dilution clearly emphasised in

    Fig. 5. Finally, as shown in Fig. 2b andc, silicate compoundsare

    detected with a 20/40 (v/v) dilution, whereas they are not with a

    1/11 S/B ratio. These compoundsprobably come from reactions

    between fibre bulk (containing silicon) and matrix organic com-

    pounds. It could explain the rapid fibre degradation observed

    during experiments when the sample is not sufficiently diluted.

    Thus, dilution of the sample in aqueous buffer allows lower

    matrix effects and protects fibre against a too early degradation.

    The headspace/solution volume ratio was also found to have

    a significant effect on extraction efficiency, HS/S= 1/1 (corre-

    sponding to 60 mL solution in a 120 mL reactor see Fig. 4)

    giving the highest extraction yield especially for MBT, DBT,

  • 7/30/2019 Compusi Organistanici Ape

    7/9

    128 J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130

    Table 1

    Analytical performances in brandy

    LOD (ng (Sn) L1) LOQ (ng (Sn) L1) Sensitivity (mVng(Sn)1) [*RSD] **R2 Recovery (%) [*RSD]

    MBT 3.4 8 5.6 0.1 [1.8] 0.9933 104 19 [18]

    DBT 1.6 4 1.76 0.01 [0.5] 0.9993 96 6 [6]

    TBT 4.6 11 0.70 0.01 [1.4] 0.9952 98 23 [23]

    MPhT 17 40 3.21 0.07 [2.2] 0.9973 107 10 [9]

    DPhT 30 71 0.065 0.001 [1.5] 0.9954 109 10 [9]MOcT 52 123 0.103 0.001 [1.0] 0.9968 93 10 [11]

    *RSD: relative standard deviation; **R2: determination coefficient.

    MPhT andDPhT. This ratio does notcorrespond to theminimum

    headspace volume, although a minimum HS was expected to

    give an optimal extraction [9,13]. This result is probably related

    to the stirring efficiency as already noticed [11] and/or matrix

    effects induced by higher amounts of non-OTC volatile com-

    ponents as discussed just before. As a consequence, a strong

    interaction between HS/S and S/B ratios occurs, as clearly

    enhanced for octyltins (see Fig. 4). It is in agreement with thepreliminary results presented before in this paper.

    With regard to the influence of the different studied fac-

    tors, the optimised operating conditions were the following: (1)

    30 min extraction time; (2) 1/11sample/buffer volume ratio (i.e.

    5 mLofsample+55 mLof buffer)and (3)1/1 headspace/solution

    volume ratio (i.e. 60 mL [sample + buffer] in a 120mL reactor).

    3.2. Analytical performances

    Thelimitsof detection(LOD) andquantification (LOQ) were

    calculated by using the calibration curve method [24,25]. The

    method accuracy (i.e. trueness and precision defined according

    toISO [26]) wasalsodetermined. Precisionwas evaluated by the

    relative standard deviation (RSD) of the calibration slope (i.e.

    sensitivity) and the determination coefficient. Trueness was esti-

    mated via themean recoveryobtained by determining OTC with

    concentrations covering the calibration curve in spiked samples.

    The results for brandy and wine samples are summarised in

    Tables 1 and 2 respectively.

    The butyltin LOD are 310 times lower than those reported

    in the literature of wine [5,8]. The detection limits for MPhT,

    DPhT andMOcTaresatisfactory, eveninbrandywiththehighest

    alcohol content. TPhT andDOcT canalso be detected.However,

    due to their lower volatility, their LOD remain very high, over

    1000 ng(Sn)L1.

    Accordingto thedatapresentedin Tables1and2 andthesatis-

    factory recovery (trueness and precision), the optimised methodappears suitable for wine and brandy analysis.

    3.3. Analysis of wine and brandy samples

    Analyses were performed in routine conditions. Some sam-

    ples of brandy originally stored in oak barrel during maturing

    were analysed. They appeared slightly contaminated by MBT

    only. Concentrations ranged between 10 and 50 ng(Sn)L1.

    The analytical precision was around 37%. Some non-OTC

    species were also detected by PFPD, as shown in Fig. 6. These

    compounds were observed in brandy only. They involved chro-

    matographic peaks withthe typical shapeof sulphur compounds.

    Bravo et al. studied sulphur interferences and identified them

    from their retention times [14]. According to this study and

    the retention times measured from Fig. 6, the first peak could

    be possibly attributed to diethyl trisulphur (*IS1: Et2S3) and

    the second one to diethyl tetrasulphur (*IS2: Et2S4). They

    could correspond to species formed during ethylation, as pre-

    viously demonstrated [14]. Unfortunately, these species could

    Table 2

    Analytical performances in white (in white) and red (in grey) wines

    LOD (ng (Sn) L1) LOQ (ng (Sn) L1) Sensitivity (mVng(Sn)1) [*RSD] **R2 Recovery (%) [*RSD]

    MBT 2.2 5.2 44.0 0.3 [0.7] 0.9949 111 18 [16]

    2.5 5.7 44 1 [2.3] 0.9920 123 22 [18]

    DBT 1.2 2.9 61.2 0.7 [1.1] 0.9956 96 12 [12]

    1.7 3.9 92 2 [2.2] 0.9922 120 17 [14]

    TBT 1.6 3.9 18.7 0.3 [1.6] 0.9943 105 16 [15]

    2.1 5.3 34 1 [2.9] 0.9911 94 10 [11]

    MPhT 2.2 5.2 5.77 0.09 [1.6] 0.9949 114 29 [25]

    9.1 20 32 1 [3.1] 0.9884 81 10 [12]

    DPhT 18 43 2.67 0.03 [1.1] 0.9944 79 15 [19]

    37 82 4.9 0.2 [4.1] 0.9915 90 11 [12]

    MOcT 30 71 6.5 0.1 [1.5] 0.9920 104 23 [22]

    35 69 13 1 [7.7] 0.9723 108 17 [16]

    *RSD: relative standard deviation; **R2

    : determination coefficient.

  • 7/30/2019 Compusi Organistanici Ape

    8/9

    J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130 129

    Fig. 5. Organotin extraction profiles according to sample/buffer volume ratio

    Extraction duration, 30 min; total volume, 60mL.

    Fig. 6. Chromatogram of a brandy sample exhibiting non-OTC species (*IS1

    and *IS2).

    not be identified from the GCMS chromatograms due to the

    lack of MS sensitivity. Finally, because of the efficient chro-

    matographic separation of sulphur species and organotins, the

    OTC quantification was not disturbed by these compounds and

    these first applications confirm the suitability of the analytical

    method.

    TheTable3 presents theresults obtainedon selectedbrandies,

    white and red wines. MBT was found in all samples, whereas

    DPhT was never detected. In most of the samples, the OTC

    concentrations are lower than 150ng(Sn) L1, some of them

    being close to LOQ. The precision of the determination remains

    satisfactory for all the different analysed matrices.

    Considering the brandy samples, the most contaminated

    appears to be the one stored with oak shavings. The three

    butyltins were found, showing that wood treatment may con-

    tribute to the contamination. However, TBT and DBT do not

    Table 3

    Organotin speciation in various brandies and wines

    Sample Characteristics (alcohol content) Concentrations (ng(Sn)L1)

    MBT DBT TBT MPhT DPhT MOcT

    Brandy 1 distillation product stored in glass (40%) 27 2 nd nd

  • 7/30/2019 Compusi Organistanici Ape

    9/9

    130 J. Heroult et al. / J. Chromatogr. A 1180 (2008) 122130

    seem very persistent, being not detected in brandy stored for

    several years in oak barrel. Other sources of OTC contamina-

    tion exist since MBT was also found in the distillation product

    and MOcT was present in brandy samples (see Table 3). The

    occurrence of this last compound is usually attributed to plas-

    tic material. The brandy storage in plastic tank could explain

    the presence of this compound. The AOC sweet white wines

    appear slightly contaminated. The highest OTC concentrations

    were found in table wines stored in plastic bottles that led to

    significant MBT and DBT content. These results are in agree-

    ment with those reported in literature [5,8]. They confirm the

    impact of storage in plastic container. An important MPhT con-

    centration was found in the white wine sample. MPhT presence

    in such concentration remains difficult to explain. It could corre-

    spond to an accidental contamination or could come from TPhT

    degradation, TPhT being usually used as agricultural biocide in

    continental media.

    4. Conclusion

    In the field of manufactured solutions, wine and brandy are

    at once complex matrices and very good supports to evalu-

    ate SPME fibre coating behaviour. However, analytical studies

    devoted to this kind of matrices remain scarce in the literature.

    Onemajor difficulty comes from aromasandflavours formedby

    numerous volatile compounds. According to the OTC analytical

    process, these species are co-extracted into the SPME fibre con-

    currently with ethylated OTC.This phenomenon leads to drastic

    decrease of analyte chromatographic signals, poor repeatability

    and insufficient sensitivity.

    In the present study, a rapid method for simultaneous deter-

    mination of butyltins, MPhT, DPhT and MOcT without anypretreament was optimised and validated. The matrix effects

    were simplyovercomeby diluting thesample. With regardto the

    analytical performances, this method appears robust and precise

    with RSD generally around 1020% over a range of calibration

    from LOD up to 1000ng(Sn)L1. The results obtained from

    brandy and wine analysis show the applicability and suitability

    of the developed method in routine conditions. The presence

    of butyl-, phenyl- and octyltins in various concentrations in

    these samples emphasises the interest of analytical develop-

    ments adapted to speciation of all these OTC. TPhT and DOcT

    can not be quantified at satisfactory level and further investi-

    gation is needed to overcome this difficulty. Finally, the results

    obtained open new perspectives in the SPME-based analysis of

    manufacturedmedia, which is especially importantaccording to

    the increasing need of adapted analytical procedures and future

    regulations.

    References

    [1] D.S. Forsyth, D. Weber, C. Cleroux, Food Addit. Contam. 9 (1992) 161.

    [2] D. Forsyth, D. Weber, K. Daglish, Talanta 40 (1993) 299.

    [3] D.S. Forsyth, W.F. Sun, K. Dalglish, Food Addit. Contam. 11 (1994)

    343.

    [4] G.B. Jiang, F.Z. Xu, F.J. Zhang, Fresenius J. Anal. Chem. 886 (1999) 256.

    [5] J.Y. Liu, G.B. Jiang, J. Agric. Food Chem. 50 (2002) 6683.

    [6] M. Azenha, M.T. Vasconselos, J. Agric. Food Chem. 50 (2002) 2713.

    [7] D. Forsyth, D. Weber, L. Barlow, Appl. Organometal. Chem. 6 (1992) 579.

    [8] M. Azhena, M. Vasconcelos, Anal. Chim. Acta 458 (2002) 231.

    [9] P.G. Hill, R.M. Smith, J. Chromatogr. A 872 (2000) 203.

    [10] G. Lespes, V. Desauziers, C. Montigny, M. Potin-Gautier, J. Chromatogr.

    A 826 (1998) 67.

    [11] S. Aguerre, Ch. Bancon-Montigny, G. Lespes, M. Potin-Gautier, Analyst125 (2000) 263.

    [12] Ch. Bancon-Montigny, G. Lespes, M. Potin-Gautier, J. Chromatogr. A 896

    (2000) 149.

    [13] M. Le Gac, G. Lespes, M. Potin-Gautier, J. Chromatogr. A 999 (2003) 123.

    [14] M. Bravo, G. Lespes, I. De Gregori, H. Pinochet, M. Potin-Gautier, J.

    Chromatogr. A 1046 (2004) 217.

    [15] T. Zuliani, G. Lespes, R. Milacic, J. Scancar, M. Potin-Gautier, J. Chro-

    matogr. A 1132 (2006) 234.

    [16] J. Goupy, in: Plans dExperiences pour Surface de Reponse, Dunod, Paris,

    1999.

    [17] G. Sado, M.C. Sado, in: Les Plans dExperiences, Afnor Technique, Paris,

    1991.

    [18] M. Riu-Aumatell, M. Castellari, E. Lopez-Tamames, S. Galassi, S. Bux-

    aderas, Food Chem. 87 (2004) 627.

    [19] M. Riu-Aumatell, J. Bosch-Fuste, E. Lopez-Tamames, S. Buxaderas, FoodChem. 95 (2006) 237.

    [20] Z. Zhang, J. Pawliszyn, Anal. Chem. 65 (1993) 1843.

    [21] J. Ai, Anal. Chem. 69 (1997) 3260.

    [22] A.G. Oomen, P. Mayer, J. Tolls, Anal. Chem. 72 (2000) 2802.

    [23] A.C. Silva Ferreira, P. Guedes de Pinho, Anal. Chim. Acta 513 (2004)

    169.

    [24] AFNOR, XP T90-210, 1999, Protocol for the evaluation of an alternative

    quantitativephysico-chemical analysismethod against a reference method.

    [25] L.A. Currie, Anal. Chim. Acta 391 (1999) 127.

    [26] ISO 5725, Accuracy (trueness and precision) of results and measurement

    method, 1994.