Speciation of organotins in environmental samples by SPME-GC:

comparison of four specific detectors: FPD, PFPD, MIP-AES and

ICP-MS

Sandrine Aguerre,*a Gaetane Lespes,a Valerie Desauziersb and Martine Potin-Gautiera

aLaboratoire de Chimie Analytique Bio-Inorganique et Environnement (L.C.A.B.I.E.), UMR5034 CNRS - Universite de Pau et des Pays de l’Adour, Avenue de l’Universite, 64000 Pau,France

bLaboratoire de Genie de l’Environnement Industriel (L.G.E.I.), Ecole des Mines d’Ales, 6Avenue de Clavieres, 30319 Ales cedex, France

Received 12th December 2000, Accepted 9th January 2001First published as an Advance Article on the web 13th February 2001

The performances of four specific detectors used for the speciation of butyl- and phenyltin compounds after

solid phase microextraction (SPME) and gas chromatography (GC) separation are evaluated. A flame

photometric detector (FPD), a pulsed flame photometric detector (PFPD), a microwave induced plasma atomic

emission spectrometer (MIP-AES) and an inductively coupled plasma mass spectrometer (ICP-MS) were used.

The principle of PFPD, a new generation of FPD, is presented. The original transfer line used between GC and

ICP-MS is detailed. The high SPME preconcentration allows very low limits of detection (LOD) to be reached

(less than 500 pg l21 Sn for all the detectors). Sensitivity, linearity and selectivity of the different detectors are

also discussed. As expected, ICP-MS is the most sensitive (LOD ranged from 0.6 to 20 pg l21 Sn) but the

cheapest PFPD is also of significant interest. The analytical procedure is applied to the determination of

organotins in two different reference materials: a sediment (PACS 2) and a fish tissue (NIES 11). These

different examples show that the detection of ultra-trace tin species is now possible in natural samples using a

combination of SPME and GC with a specific detector.

Tri-substituted butyl- and phenyltins (TBT and TPhT) havebeen extensively used as agrochemical products and biocides inlots of applications. Mono- and di-substituted butyltincompounds are also widely employed as PVC stabilizers,catalysts or wood preservatives for example.1 These numerousapplications give rise directly or indirectly to an importantdiffusion of free organotins in the environment. So, asignificant contamination of marine2,3 and freshwater4–6

ecosystems has been reported in the literature. Organotinspecies coming from domestic activities are also present inwastewater and sewage sludge.7

The toxicity of organotins depends on their chemical formsand varies according to the living organisms. Several studiesreport the toxicity of TBT to sensitive aquatic organisms froma few ng Sn l21 of tin.8,9 Therefore, speciation of organotincompounds has become necessary.

In order to control tin pollution, many analytical procedureshave been developed. A speciation procedure involvingethylation with sodium tetraethylborate (NaBEt4) followedby a liquid–liquid extaction in hexane or isooctane before GCseparation has been demonstrated to be the most suitablemethod for organotin determination.10–14 Moreover, solidphase microextraction (SPME) used during the derivatizationstep has recently been proposed, in order to increase theanalytical potential of the procedure. This cheap and solvent-free technique gives a ‘‘quasi on-line’’ process from the samplepreparation to the analysis. Since 1992, SPME has beensuccessfully applied to the extraction of numerous tracepollutants such as phenols,15 organophosphorus pesticides16

or organic forms of lead,17–19 mercury17–20 and tin.17,21–23 Thehigh preconcentration offered by this new tool facilitates theanalysis of organometallics at trace and ultra-trace levels, asrequired in environmental controls. So, the SPME procedurefollowed by GC with specific detection has been developed in

previous works22,23 for the speciation of butyl- and phenyltinsin various environmental samples.

Most of the developed analytical procedures are based ongas chromatographic separation hyphenated to element-specific detection systems, such as atomic absorption spectro-metry (AAS),24,25 microwave-induced plasma atomic emissionspectrometry (MIP-AES),26–28 flame photometric detection(FPD)10–14 or a specific mass spectrometric detector.29 Morerecently, two other specific detectors have been used. First, theapplication of ICP-MS as a detector for GC has led toconsiderable interest owing to its high sensitivity andselectivity.17,30,31 However, the technique involves a self-constructed transfer line thus limiting its extensive use.Second, a new generation of FPD based on a pulsed flame,the PFPD, has been developed by Amirav and Jing and offereda new alternative for routine analysis.32

The evaluation of performances of four specific detectors(FPD, PFPD, MIP-AES, ICP-MS) in terms of sensitivity,selectivity, linearity and operational cost are presented in thispaper in order: to determine and compare the real respectiveperformances of these four detectors (such a comparisonhaving never been reported in the literature); and to proposethe best compromise(s) for a sensitive and accurate speciationof organotins in the environment for present and futureanalytical needs.

Experimental

All organotin concentrations reported in this paper areexpressed as the mass of Sn per mass or volume unit.

Apparatus

The manual SPME device was obtained from Supelco(Supelco, Bellefonte, USA). The fibres used were coated with

DOI: 10.1039/b008223f J. Anal. At. Spectrom., 2001, 16, 263–269 263

This journal is # The Royal Society of Chemistry 2001

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online / Journal Homepage / Table of Contents for this issue

an apolar stationary phase (PDMS 100 mm). This phase waspreviously found to give the best results.22 An elliptic table KS250 basic (Prolabo, Fontenay Sous Bois, France) was used forstirring during the derivatization step, because it gives betterextraction yields than the classical magnetic stirring.23

Different instruments were used for the speciation proce-dure. A Varian 3300 gas chromatograph (Palo Alto, CA, USA)equipped with a conventional flame photometric detector(FPD) and a Varian 1075 split/splitless injector was used. TheFPD operating conditions were as previously described.11 A610 nm optical filter corresponding to Sn–H emission was used.

A Varian 3800 gas chromatograph (Walnut Creek, CA,USA) equipped with a pulsed flame photometric detector(PFPD) and a Varian 1079 split/splitless on -column injectorwas also used. The principle of this new generation of FPD wasdescribed elsewhere.33 It is based on a discontinuous air–hydrogen flame in which each species formed has its own timephotometric emission profile.32 So, the period of emissionwhich has to be integrated to generate the detector signal can benow selected by adjusting the start (gate delay) and theduration (gate width) of the detection according to the emissionprofile of organotin compounds. This operation widelydecreases the detection of the main potential interferents ontin, i.e., sulfur and phosphorus compounds. Thanks to this highselectivity, an optical filter at 390 nm (corresponding to Sn–Cemission) was preferred to the classical 610 nm interferencefilter (corresponding to Sn–H emission), because Sn–Cemission is up to 100–1000 times more important. A firststudy has demonstrated that no sulfur or phosphorusinterferent appears in sediments and oysters analysed withthe PFPD system at 390 nm.33 The gate delay was fixed to3.0 ms and the gate width to 2.0 ms, according to PFPD tinemission time.

A HP 6890 gas chromatograph (Hewlett-Packard, Wilming-ton, DE, USA) equipped with a split/splitless injection port wasused. Detection was achieved with an HP model G2350Amicrowave induced plasma atomic emission spectrometer(MIP-AES).

A HP 5890 series II plus gas chromatograph (Hewlett-Packard, Wilmington, DE, USA) equipped with a split/splitlessinjection port was used. The ICP-MS was an HP Model 4500plus system (Yokogawa Analytical Systems, Japan). The

interface between GC and ICP-MS is elaborated by HP anddescribed below.

The separation was carried out on capillary columnsDB1 (30 m60.25 mm id60.25 mm film thickness) and HP5(30 m60.32 mm id60.25 mm film thickness). Both theseapolar or low-polar columns give similar separation andhave not any influence on the detection.23,27 Nitrogen or heliumwas used as the carrier gas, according to recommendations ofeach manufacturer, at flow rates between 1 and 2 ml min21.Details for each chromatograph are given in Table 1. Thefollowing temperature program was chosen in order to allow asatisfactory separation of organotin compounds The columntemperature was held at 80 ‡C for the first minute, increased to180 ‡C at the rate of 30 ‡C min21 and then increased to 270 ‡Cat 10 ‡C min21 for both columns

ICP-MS transfer line. The prototype used in this work isdescribed in Fig. 1. A deactivated silica capillary tube(1.5 m60.32 mm) allows the transfer of the analytes fromthe end of the capillary column to the ICP-MS torch. The mainobjective of this transfer line is to avoid peak broadeningcaused by condensation of the GC effluent in the interfacebefore reaching the plasma. So, this capillary is surroundedwith a 1/16 inch stainless-steel tube. Argon, used as auxiliarygas, is passed through this tube, previously heated in the oven,at a high flow rate of about 1 l min21. Heated argon ensures asufficient and constant temperature all along the transfer line.The high flow make-up also allows the analytes to be pushedfrom the end of the capillary column to the plasma without anyloss and condensation. The flexible part of the transfer line isalso resistively heated at 250 ‡C and thermally insulated. Xenongas (100 ppm in argon) was added to the argon auxiliary gas ata constant flow rate (5 ml min21) regulated with a massflowmeter. 128Xe was constantly measured in order to correctinstrumental instabilities.

Instrumental operating conditions. The operating conditionsof separation and detection are summarized in Table 1. ForFPD, MIP-AES and PFPD, the optimum conditions weredetermined in previous works.12,26,33 Concerning the newcoupling between GC and ICP-MS, the ICP-MS operatingconditions were determined using the signal of the 128Xe

Table 1 Optimal operating conditions of GC and detectors

Conditions FPD PFPD MIP-AED ICP-MS

Gas chromatographyCarrier gas N2 N2 He HeFlow rate 1–2 ml min21 1–2 ml min21 1.4 ml min21 2 ml min21

Column DB1 DB1 HP5 HP5

Transfer lineTemperature 250 ‡C 250 ‡C

Specific detectorWavelength 610 nm 610 nm/390 nm 303 nmH2 flow rate/pressure 185 ml min21 26.5 ml min21 20 psiAir flow rate 285 ml min21 Air 1a 28.0 ml min21

Air 2b 20.5 ml min21

O2 pressure 20 psiHelium make-up flow 240 ml min21

Detector temperature 290 ‡C 350 ‡C 250 ‡C

Isotopes 120Sn, 118Sn, 128XeRf power 1250 WSampling depth 7–8 mmRf matching voltage 2.1 VSampler and skimmer cones NiCarrier gas flow rate 0.75 l min21

Auxiliary gas flow rate 1 l min21

Plasma gas flow rate 15 l min21

Xe flow rate (100 ppm in argon) 5 ml min21

aIgnition chamber. bCombustion chamber.

264 J. Anal. At. Spectrom., 2001, 16, 263–269

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online

isotope. 118Sn and 120Sn were also chosen in order to evaluatethe background. The position of the capillary in the torch, thesampling depth, the horizontal and vertical position of thetorch and the flow of the carrier gas were adjusted in order toobtain the best sensitivity for the xenon signal and the lowestrelative standard deviation (RSD) on the background for tinisotope signals. These conditions were verified every day.Moreover, the 128Xe signal is monitored during all the analysesin order to control the stability of the plasma and the massspectrometer.

Reagents

Monobutyltin trichloride (MBT, 95%), dibutyltin dichloride(DBT, 97%), tributyltin chloride (TBT, 96%), tetrabutyltin(TeBT, 93%), monophenyltin trichloride (MPhT, 98%),diphenyltin dichloride (DPhT, 96%), triphenyltin chloride(TPhT, 95%), tricyclohexyltin (TCHexT, 90%) were purchasedfrom Aldrich (Milwaukee, WI, USA). Tripropyltin chloride(TPrT, 98%) was obtained from Strem Chemicals (Bischeim,France). Trioctyltin (TOcT, w99%) came from Fluka. Theother octyltin compounds [mono (MocT) and di (DOcT)] weresynthesized in the laboratory. The organotin stock solutionscontaining 1000 mg l21 as tin were prepared in methanol.Stored at z4 ‡C in the dark, they were stable for one year.12

Working standards were obtained by dilution in water weeklyfor solutions of 10 mg l21 and daily for 100 mg l21. They werestored in the dark at z4 ‡C.

Methanol and sodium acetate were purchased from Prolabo(France). Hydrochloric, nitric and ethanoic acids wereobtained from Merck (Darmstadt, Germany), and isooctanefrom Fluka (Buchs, Switzerland). The deionized water usedwas 18 MV (Millipore, Bedford, MA, USA).

Sodium tetraethylborate (NaBEt4) was obtained from StremChemicals (France). The working solution was made daily bydissolving 0.02 g in 2 ml of deionized water and stored atz4 ‡C in the dark. NaBEt4 is a pyrophoric and air sensitivereagent, so special precautions are required when handling.

Glassware was rinsed with deionized water, decontaminatedovernight in 10% (v/v) nitric acid solution and then rinsedagain.

Samples and reference materials

The comparison of the different analytical procedures wasperformed using two certified reference materials: PACS 2, asediment from National Research Council of Canada (NRCC)certified for its DBT and TBT contents (an indicative value isgiven for MBT); and NIES 11, a freeze-dried fish prepared bythe National Institute of Environmental Studies (Japan), whichis certified for its TBT content and with an indicative valuegiven for TPhT.

Analytical procedure

Extraction from solid samples. Detailed operating conditionswere carried out and are precisely described elsewhere.12,34

Acidic extraction was performed before the derivatizationstep. For sediment, 1 g of freeze-dried sediment was extractedin 20 ml of glacial ethanoic acid by mechanical stirring for 12 h,with 50 ml of a 10 mg ml21 TPrT solution used as internalstandard.12 For biological material, 0.25 g of a freeze-driedmaterial was humidified in 2.5 ml of methanol by mechanicalstirring for 2 h. Then, 12.5 ml of HCl in methanol(0.12 mol l21) were added and the mixture was stirred sonicallyfor 1 h.34

Derivatization and analysis. The glass reactors were obtainedfrom Prolabo (France). The reactor volume is about 250 mland the neck (19/26) is adapted to make the SPME needlestable during a vigorous agitation. 0.5 – 2 ml of centrifugedacidic extract were directly introduced into the derivatizationreactor in the presence of a 100 ml sodium ethanoate–ethanoicacid buffer. Ethylation was carried out using NaBEt4 (0.3 –0.5 ml of the 2% solution). The buffer composition wasadjusted by adding a sufficient mass of sodium hydroxide inorder to obtain a pH of 4.8 after adding acidic extract andNaBEt4. Therefore the same pH conditions are obtainedwhatever the sample. Then, the PDMS SPME fibre was directlyplunged into the sample. An equilibrium was establishedbetween the sample solution and the stationary phase coatedon the fibre, analyte sorption depending essentially on itsaffinity with the stationary phase and the thickness of thecoating material. Shaking the solution allows the equilibriumof sorption to be reached more quickly. So, the reactor wasplaced on the elliptical table previously described.23 In all cases,the continuous contact between the fibre and the solution wascarefully verified. Previously, this type of shaking has beendemonstrated to be particularly good.23 After the sorption step,the fibre was directly placed into the injection port of the GCwhere the compounds were thermally desorbed. The preciseoperating conditions, previously optimised by experimentaldesign methodology, are the following:22,23 sorption time,40 min; sample volume, 100 ml; injection temperature, 270 ‡C;and desorption time, 1 min.

Quantitation. Tripropyltin was used as internal standard.The use of relative responses avoids errors due to the variationsof the detector sensitivity. For each sample, the tripropyltin-relative responses of butyl- and phenyltin compounds werecalculated by standard additions. This procedure was appliedon three – five aliquots of 100 ml freshwater or three portionsof acidic extract from solid material. Standard additionsmethodology decreases the matrix effects as much as possible.

Results and discussion

Comparison of analytical performances

The figures of merit were evaluated according to the IUPACrecommendations. Limits of detection (LOD), evaluated withstandard solutions, were determined as three times thestandard deviation of the noise. Repeatability (relativestandard deviation, RSD) of the whole procedure (i.e.ethylation, SPME, GC, specific detection) was calculatedfrom 6 replicates. Linearity of the different detectors wasstudied between the LOD and 400 ng l21 of tin. Table 2summarises the performances obtained.

The relative limits of detection are very low whatever thedetector used (v200 pg l21 of Sn for all the organotins exceptTPhT). The coupling with ICP-MS is logically found to be themost sensitive. However, the PFPD has also a very highsensitivity; the performance of the SPME-GC-PFPD is

Fig. 1 Prototype of GC-ICP-MS interface constructed by Hewlett-Packard. 1, SPME device; 2, capillary column; 3, stainless steel tube; 4,heated transfer line (250 ‡C–1.5 m); 5, ICP-MS torch.

J. Anal. At. Spectrom., 2001, 16, 263–269 265

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online

evaluated and presented in this paper for the first time. TheLOD given in the literature concern only the butyltins; theyrespectively ranged over 0.34–1.1 ng l21 (headspace SPME-GC- ICP-MS) and 900–1000 ng l21 (headspace SPME-GC-FID).17,35 The LOD for butyltins are about 3 to 100 timeslower, especially for di- and tributyltins. Concerning thephenyltins, the LOD obtained using the most sensitive ICP-MSand PFPD are generally 10 times higher than for butyltin. Thelimit of detection of TPhT remains the highest, probably due tothe low affinity of TPhT for the PDMS SPME fibre.23

Nevertheless, all these LOD are lower than those presentedin the literature. This improvement can be attributed to SPMEand the high efficiency of the mechanical stirring used duringliquid/solid extraction, as has been demonstrated previously.23

The absolute limits of detection ranged over 4–10 pg (FPD),0.2–1.0 pg (AES), 0.07–0.38 pg (PFPD) and 0.05–0.08 pg (ICP-MS). As expected, ICP-MS is the most sensitive detector.However, the new PFPD appears really interesting, its LODbeing only about 2 to 10 times higher than for ICP-MS . Thesimplicity, the low consumption of gases and more generallythe low operating cost offered by PFPD are also importantadvantages considering the routine control of tin pollution.

The linearity was studied for up to 400 ng Sn l21 whichcorresponds to the highest concentration usually found inenvironmental water samples. It was noted that, over this valueof concentration, problems of contamination appear duringanalysis, even using the most sensitive PFPD and ICP-MSapparatus. Moreover, when water contains more than 200 and300 ng l21, respectively, of butyltins and MPhT, the pulsedflame photometric detection overloads and the peaks split. So,it is necessary to dilute the aqueous sample or reduce thevolume of extract taken for analysis. This is not really ananalytical problem but this observation confirms the well-known narrower range of linearity of the flame photometricsystem.

The repeatability was evaluated on absolute chromato-graphic peak areas with 100 ng l21 standard solutions using thesame fibre for the successive extractions. The values obtained inthe present work are generally lower than 10%, (except whenICP-MS is used), which is of the same order of magnitude asthose usually reported in the literature.21 This result can beexpected because repeatability is evaluated on the wholeprocess, from ethylation–SPME to detection. For TPhT, theprocedure appears less repeatable due to the extraction stepand probably the competition for sorption between thedifferent phenyltins, as was previously described.23

Fig. 2 presents a typical GC-ICP-MS chromatogram ofPACS 2 spiked with phenyltins. With regard to the wholechromatogram, the resolution obtained is excellent. However,when a zoom is used on low signal intensity, a little peakdistortion can be observed. This could be due to a very low

condensation of the species in the transfer line. Moreover, thebaseline obtained with ICP-MS is submitted to large variationsand the repeatability is very bad. Thus, the signal can decreaseby 60% during a working day. Between two consecutiveanalyses, variations from about 3 to 15% are possible. So the128Xe isotope used for the optimisation of analytical conditionsis also used as ‘‘internal standard’’ in order to correct theinstrumental instabilities, such as signal drift in the massspectrometer during the GC run. The corrected repeatabilityobtained is always higher than that obtained by photometric oratomic emission detectors. However, the corrected GC-ICP-MS repeatability becomes acceptable and shows that theinternal standard quantitation method (i.e., use of relativechromatographic peaks) is absolutely required to obtainrepeatable ICP-MS analyses. This problem has already beenmentioned in the literature for a similar coupling between GCand ICP-MS.36 Uncorrected repeatability reported using GC-

Table 2 Comparison of analytical performances

MBT DBT TBT MPhT DPhT TPhT

LOD/pg Sn l21) FPDa 31 7 6 114 167 583AED 42 11 9 53 58 415PFPDb 4 1 1 8 13 200ICP-MS 2 0.7 0.6 4 6 20

Linearity from LOD to.../ng l21 FPDa 400 400 400 400 400 400AED 400 400 400 400 400 400PFPDb 200 200 200 300 400 400ICP-MS 400 400 400 400 400 400

Repeatability (%) FPDa 3 3 5 9 8 16AED 5 4 6 5 8 18PFPDb 4 5 7 5 9 18ICP-MSc 21 17 25 18 47 62ICP-MSd 8 9 16 14 8 25

aSn–H emission. bSn–C emission. cUncorrected repeatability. dCorrected repeatability (using Xe signal).

Fig. 2 (a) Chromatogram of PACS 2 spiked with phenyltins by GC-ICP-MS (120Sn). (b) Zoom on peaks of low intensity from (1). SnIV, (2)MBT, (3) TPrT, (4) DBT, (5) MPhT, (6) TBT, (7) DPhT, (8) TPhT;(a*–f*) unidentified peaks.

266 J. Anal. At. Spectrom., 2001, 16, 263–269

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online

ICP-MS ranged over 12–25% for organolead compounds,which is similar to our results.

Application to environmental samples

The analytical method using the four different detectors wasevaluated with certified reference materials (NIES 11 andPACS 2).

The SPME-GC-FPD process was validated for the analysisof water samples in a previous work,23 giving results in goodagreement with those obtained after liquid–liquid extraction.However, when matrices are more complex, the wholederivatization step (ethylation with NaBEt4 and/or extraction)can be disturbed. This can induce a decrease in organotinresponse, especially for TPhT. The matrix effects during thesorption step on the fibre were reduced as much as possible byadjusting the operating conditions.23 Unfortunately, with theless sensitive and selective FPD, the TPhT signal can sometimesdisappear when very complex samples (such as a sewage sludgeor a biological tissue) are analysed.23 So a comparison of thepotential of the more sensitive and specific detectors was madein order to propose a suitable procedure for organotinspeciation in various matrices. With MIP-AES, PFPD orICP-MS, the detection of all the species, and particularlyTPhT, appears possible in environmental samples such assediment or biological materials, as the different applicationspresented later on show.

The analysis of the certified sediment PACS 2 was performedby GC-PFPD, GC-MIP-AES and GC-ICP-MS. This materialis highly polluted by butyltin compounds and in our case adilution of the acidic extract was necessary before the GC-PFPD analysis due to the narrower linearity range of thisdetector for butyltins. The results are presented in Table 3. Allthe experimental values are in good agreement with both thecertified values. For MBT, the values found are systematicallyhigher than the indicative one. This fact has already been notedby several authors,31,35 who, respectively, found 510 ng g21

(ethylation/isooctane extraction) and 800 ng g21 (ethylation/headspace SPME) of tin, whereas the indicative value is300 ng g21. Moreover, a previous study using SPME-GC-FPDprocedure has shown that the use of SPME instead of anorganic solvent could increase the extraction yield of somecompounds from complex matrices.23 This observation couldexplain why the MBT concentrations found are so high.

However, these MBT values are quite different according tothe detector used. For GC-PFPD analysis, the high dilution ofthe acidic extract (500 times) could explain the highestconcentration obtained. The ethylation/extraction yield ofMBT is probably dependent on the concentration of otherproducts (such as organic matter) and so varies according tothe different operating conditions used during derivatization(i.e., high dilution reduces the organic matter effect andincreases ethylation/extraction yield). Despite the high selec-tivity of PFPD, disturbance of Sn–C emission has also to be

considered if a significant amount of organic interferent iscoeluted simultaneously with MBT.

According to the different results, the GC-MIP-AES and theGC-ICP-MS procedures present the lowest relative standarddeviations. The use of internal standard quantitation allowssatisfactory and repeatable results to be obtained by ICP-MS,even if instrumental instabilities are observed.

The analytical method was applied to the analysis of thecertified reference material NIES 11. The results are presentedin Table 4. The fish tissue was analyzed under routineconditions by GC-MIP-AES and GC-PFPD. It is certifiedfor TBT and an indicative value is given for TPhT content. Thismaterial is interesting because a value of concentration isavailable for phenyltin species. However, numerous problemsof degradation of the tri-substitued compounds in this materialhave already been noted in the literature,34 due to transportand/or storage. Degradation products (MBT, DBT and MPhT)were then quantified at significant levels. Nevertheless, theconcentrations obtained (Table 4) for the sums of butyltinswith both instruments are in good agreement with the TBTcertified value. These results confirm the decompositionphenomena of TBT. The quantitation of TPhT was morecritical, as expected, because of the low intensity of thecorresponding spectrometric signal, as Fig. 3 clearly illustrates.This is probably due to the wide presence of organic matter,inducing matrix effects such as those previously described.These effects are more important when the less sensitive MIP-AES is used. In this case a larger volume of acidic extract (2 ml)must be taken for analysis whereas only 0.5 ml was necessaryusing GC-PFPD. These results show the interest in highlysensitive detection such as PFPD, which also appeared morereliable since the sum of phenyltins is in agreement with theTPhT indicative value given. No interfering peak appearedduring analysis, demonstrating that PFPD at 390 nm is highlyselective and convenient, even in the case of complex matricesanalysis.

Selectivity/specificity

The very high sensitivity offered by the combination of SPMEwith GC-ICP-MS or GC-PFPD allows the detection ofunknown peaks with low intensity in some natural samples.The analysis of PACS 2 showed the presence of compoundswhich were not reported during the certification exercise. Inorder to confirm the effective presence of other tin forms, a fewstudies were carried out with both these detectors.

First, the sediment was analysed by GC-PFPD at the twowavelengths: 390 nm (Sn–C emission) and 610 nm (Sn–Hemission, potentially more selective than the first one). Among

Table 3 Determination of organotin compounds in a certified referencematerial (freshwater sediment PACS 2) by GC-MIP-AES, GC-ICP-MSand GC-PFPD

Concentration¡standard deviationin ng (Sn) g21 (RSD in %)

MBT DBT TBT

Certified value 300a 1090¡150 980¡130GC/MIP-AES 566¡36 (6) 1013¡89 (9) 964¡64 (7)GC/ICP-MS 1301¡27 (3) 981¡73 (7) 931¡153 (16)GC/PFPD 2000¡480 (24) 1158¡148 (13) 892¡214 (24)aIndicative value.

Fig. 3 Typical chromatograms obtained from NIES 11 with MIP-AES(a) and PFPD (b); 2 ml of extract used in (a) and 0.5 ml of extract usedin (b). (1) MBT, (2) TPrT (I.S.), (3) DBT, (4) MPhT, (5) TBT, (6)TPhT.

J. Anal. At. Spectrom., 2001, 16, 263–269 267

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online

the detection methods studied in this paper, flame photometricdetection is expected to be of the lowest selectivity because anoptical filter is used instead of a monochromator. However,considering the principle of the PFPD, this potential insuffi-ciency should be solved, as has been demonstrated in a recentstudy.33 The use of two different filters also allows the possibleinterferences to be checked.37 A chromatogram of PACS 2obtained using the two wavelengths is presented in Fig. 4,where a zoom on the lowest intensity signals was used. Nointerfering species disturbed the analyses and quantitation.Only a few compounds are detected at short retention times(v2 min) when Sn–C emission is used. These peaks havealready been identified as organic compounds of sulfur (such asdimethylsulfur and diethyldisulfur).37 Some other unidentifiedforms can also be detected with this emission. By comparingrelative retention times from sample and standard solutions,most of these peaks can be attributed to tin compounds(Fig. 4). Only one peak at 6.7 min (indicated by *) has not beenidentified. As it is detected with both emissions, it can beconsidered as a probable tin species, such as an intermediatedegradation product of phenylated, octylated or hexylatedforms. However, except for butyltins, the other tin compoundsfound in PACS 2 have probably no significant environmentalimpact since their concentrations are sub-ng g21.

ICP-MS has also been used to confirm the presence of othertin species in this sediment. A chromatogram of PACS 2 spikedwith butyl- and phenyltins is presented in Fig. 2. Numerousunknown little peaks appear between 3 and 8 min. Some ofthem are also detected in standard solutions, because thesestandards are not sufficiently pure and could contain isomericforms. The peaks, which also appear in standards, are notnoted on the chromatogram. The isotopic ratio 118Sn/120Sn wasalso evaluated for unknown chromatographic peaks in thenatural sample. The natural isotopic ratio of 118Sn/120Sn is0.729 a 10% relative standard deviation being arbitrarilyconsidered as acceptable. Under these conditions, all theunknown compounds noted (a*,b*,c*,d*,e*,f*) have isotopic

ratios corresponding to tin forms. The compounds noted (e*)and (f*) can be attributed to octyl species (certainly MOcT andDOcT) according to their retention times. As in the previousGC-PFPD analysis, the other unknown species (a*,b*,c*,d*)are probably mixed methyl-alkylated or -arylated organiccompounds of tin. Such forms have already been determined indifferent works by Amouroux et al., or Edelmann et al.,38,39

who reported the formation of methylated forms of butyltinderivatives due to bacterial activity in anoxic sediments(BuSnMe3, Bu2SnMe2, Bu3SnMe). In order to identify thesetin forms, standards have to be synthesized in the laboratory.The other unknown peaks (respective retention times: 5.30,6.18, 6.70, 7.15, 7.65 min) cannot be attributed to tin forms andcould be interferents. However, their low intensity does notdisturb the analysis.

Conclusions

SPME combined to GC-specific detection offers new perspec-tives for the speciation of organotins in environmental samples.Very low detection limits (less than 500 pg l21 for all thecompounds) can be reached with the four detectors studied(FPD, MIP-AES, PFPD and ICP-MS).

GC-FPD and GC-MIP-AES are today usually used in mostlaboratories for the control of organotin pollution. SPMEcombined with these methods offers interesting performancesfor the analysis of water samples. In complex solid samples,organic matter can widely decrease ethylation yields and/orextraction yields on the fibre, particularly for TPhT. Ifsufficiently sensitive detection is used, the correspondingdecrease of signal is not critical and TPhT can be detectedand quantified as tin at the 10 ng Sn g21 level.

The performance of SPME-GC-PFPD is, for the first time,evaluated in this paper. The highly selective and sensitivePFPD appears as the best choice for analysis, especially inroutine use, considering its low cost and the reducedconsumption of gases. It suffers from a very narrow range oflinearity only. The most sensitive SPME-GC-ICP-MS appa-ratus also offers lot of advantages, even if the transfer linetested during the study does not completely avoid the problemsof condensation. Thus, the possibilities of multielementanalysis and the evaluation of isotopic ratios in order toconfirm the presence of tin compounds at very low concentra-tions are very interesting. Considering the different complexmatrices studied and the satisfactory results obtained, theSPME-GC hyphenated with the most sensitive PFPD or ICP-MS appears very suitable for the analysis of environmentalsamples. So, a larger survey of organotins becomes possibleand new forms of tin can be detected at low concentrations.Finally, this new analytical tool could contribute to a betterevaluation of the hydrobiological cycle of organotins.

Acknowledgements

The authors thank Yokogawa Analytical Systems for theloan of the GC-ICP-MS interface and Hewlett-Packard,France, for the loan of the ICP-MS spectrometer. They alsothank Varian Analytical Instruments for the loan of the PFPD.

Table 4 Determination of organotin compounds in a biological material: the certified fish tissue (NIES 11)

Concentration¡standard deviation in ng (Sn) g21

MBT DBT TBT MPhT TPhT S butyl- S phenyl-

Certified value 474¡37 1940a

GC/MIP-AED 215¡34 27¡4 316¡21 502¡31 769¡58 558¡50 1271¡66GC/PFPD 138¡7 22¡2 261¡25 618¡42 1466¡347 421¡26 2085¡350aIndicative value.

Fig. 4 Chromatograms of PACS 2 obtained by GC-PFPD with twodifferent optical filters, Sn–C and Sn–H: (1) SnIV; (2) MBT; (3) TPrT(IS); (4) DBT; (5) TBT; (a) MPhT; (b) MOcT; (c) TeBT; (d) DPhT; (e)TPhT; (*) unidentified peak.

268 J. Anal. At. Spectrom., 2001, 16, 263–269

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online

References

1 K. Fent, Crit. Rev. Toxicol., 1996, 26, 1.2 M. J. Waldock, M. E. Waite and J. E. Thain, Environ. Technol.

Lett., 1988, 9, 999.3 P. M. Sarradin, A. Astruc, R. Sabrier and M. Astruc, Mar. Pollut.

Bull., 1994, 28, 621.4 L. Schebek and M. O. Andreae, Environ. Sci . Technol, 1991, 25,

871.5 K. Fent and J. Hunn, Environ. Sci. Technol., 1991, 25, 5.6 K. Becker, L. Merlini, N. de Bertrand, L. F. De Alencastro and

J. Tarradellas, Bull. Environ. Contam. Toxicol., 1992, 48, 37.7 K. Fent and M. D. Muller, Environ. Sci. Technol., 1991, 25, 489.8 P. E. Gibbs, P. L. Pascoe and G. R. Burt, J. Mar. Biol. Ass. UK,

1988, 68, 715.9 D. Lapota, D. E. Rosenberger, M. F. Platter Rieger and

P. F. Seligman, Mar. Biol., 1993, 115, 413.10 P. Michel and B. Averty, Appl. Organomet. Chem., 1991, 5, 393.11 C. Carlier-Pinasseau, G. Lespes and M. Astruc, Appl. Organomet.

Chem., 1996, 10, 505.12 C. Carlier-Pinasseau, G. Lespes and M. Astruc, Environ. Technol.,

1997, 18, 1179.13 C. Carlier-Pinasseau, A. Astruc, G. Lespes and M. Astruc,

J. Chromatogr., 1996, 750, 317.14 G. Lespes, C. Carlier-Pinasseau, M. Potin-Gautier and M. Astruc,

Analyst, 1996, 121, 1969.15 K. D. Buchholz and J. Pawliszyn, Environ. Sci. Technol., 1993, 27,

2844.16 I. Valor, J. C. Molto, D. Apraiz and G. J. Font, J. Chromatogr. A,

1997, 767, 195.17 L. Moens, T. De Smaele, R. Dams, P. Van Den Broek and

P. Sandra, Anal. Chem., 1997, 69, 1604.18 T. Gorecki and J. Pawliszyn, Anal. Chem., 1996, 68, 3008.19 X. Yu, H. Yuan, T. Gorecki and J. Pawliszyn, Anal. Chem., 1999,

71, 2998.20 J. P. Snell, W. Fresh and Y. Thomassen, Analyst, 1996, 121, 1055.21 S. Tutschku, S. Mothes and R. Weinrich, Fresenius’ J. Anal.

Chem., 1996, 354, 587.

22 G. Lespes, V. Desauziers, C. Montigny and M. Potin-Gautier,J. Chromatogr. A, 1998, 826, 67.

23 S. Aguerre, C. Bancon-Montigny, G. Lespes and M. Potin-Gautier, Analyst, 2000, 125, 263.

24 Y. Cai, S. Rapsomanikis and M. Andreae, Environ. Sci. Technol.,1992, 23, 615.

25 V. Desauziers, F. Leguille, R. Lavigne, M. Astruc and R. Pinel,Appl. Organomet. Chem., 1989, 3, 469.

26 M. Ceulemans, W. M. R. Dirks, R. Lobinski, F. C. Adams andJ. Szpunar-Lobinska, Int. J. Environ. Anal. Chem., 1993, 52,113.

27 J. Szpunar, V. Schmitt, R. Lobinski and J. L. Monod, J. Anal. At.Spectrom., 1996, 11, 193.

28 Y. K. Chau, F. Yang and M. Brown, Anal. Chim. Acta, 1997, 338,51.

29 L. Dunemann, H. Hajimiragha and J. Begerow, Fresenius’ J. Anal.Chem., 1999, 363, 466.

30 T. De Smaele, L. Moens, R. Dams and P. Sandra, LC-GC, 1996,138.

31 I. Rodriguez, S. Mounicou, R. Lobinski, V. Sidelnikov,Y. Patrushev and M. Yamanaka, Anal. Chem., 1999, 71, 4534.

32 A. Amirav and H. Jing, Anal. Chem., 1995, 67, 3305.33 C. Bancon-Montigny, G. Lespes and M. Potin-Gautier, J. Chrom-

atogr. A, 2000, 896, 149.34 F. Pannier, A. Astruc, M. Astruc and R. Morabito, Anal. Chim.

Acta, 1996, 327, 287.35 E. Millan and J. Pawliszyn, Instrumental Methods of Analysis ’99,

Chalkidiki, Greece, 1999, p. 239.36 M. Heisterkamp, T. De Smaele, J. P. Candelone, L. Moens,

R. Dams and F. C. Adams, J. Anal. At. Spectrom, 1999, 12,1077.

37 C. Montigny, G. Lespes and M. Potin-Gautier, J. Chromatogr. A,1998, 819, 221.

38 D. Amouroux, E. Tessier and O. F. X. Donard, Environ. Sci.Technol., 2000, 34, 988.

39 D. Adelmann, K. R. Hinga and M. E. Q. Pilson, Environ. Sci.Technol., 1990, 24, 1027.

J. Anal. At. Spectrom., 2001, 16, 263–269 269

Publ

ishe

d on

13

Febr

uary

200

1. D

ownl

oade

d by

Mou

nt A

lliso

n U

nive

rsity

on

22/0

5/20

13 0

4:17

:14.

View Article Online