DSpace at VNU: Analysis of fuel oxygenates in the environment

E-mail: torsten.schmidt@eawag.ch; Fax: +41 1 823 5210; Tel: +41 1 823 5076 b Center for Environmental Chemistry CEC, Vietnam National University, 334 Nguyen Trai Street, Hanoi, Vietnam R

www.rsc.org/analyst

Analysis of fuel oxygenates in the environment

Torsten C Schmidt,*aHong-Anh Duong,bMichael Bergaand Stefan B Haderleina

a Swiss Federal Institute for Environmental Science and Technology (EAWAG) and Swiss

Federal Institute of Technology (ETH), Ueberlandstr 133, P.O Box 611, CH-8600,

Duebendorf, Switzerland E-mail: torsten.schmidt@eawag.ch; Fax: +41 1 823 5210;

Tel: +41 1 823 5076

b Center for Environmental Chemistry (CEC), Vietnam National University, 334 Nguyen Trai

Street, Hanoi, Vietnam

Received 19th October 2000, Accepted 23rd January 2001

First published as an Advance Article on the web 20th February 2001

Summary of Contents

1 Introduction

2 Occurrence in water and air

3 Analytical methods

3.1 Sampling and enrichment: water

3.1.1 Water sampling

3.1.2 Direct aqueous injection

3.1.3 Headspace analysis

3.1.4 Purge and trap enrichment

3.1.5 Solid phase microextraction

3.2 Sampling and enrichment: air

3.3 Separation

3.4 Preparation of standards and calibration

3.5 Detection

3.5.1 Flame ionisation detection

3.5.2 Photoionisation detection

3.5.3 Mass spectrometry

3.5.4 Atomic emission detection

3.5.5 Fourier transform infrared spectroscopy

4 Conclusions and outlook

5 Acknowledgements

6 Appendix: Abbreviations

7 References

1 Introduction

Fuel oxygenates are oxygen-containing substances, mainly dialkyl ethers and alcohols, used as blending components in order to increase the octane number of gasoline.1The demand for fuel oxygenates has increased rapidly due to the phase-out of tetraalkyl lead compounds as octane enhancers and the regulation of gasoline composition during the 1990s in order to improve air quality In the USA, the 1990 Amendments to the Clean Air Act require a minimum oxygen content of 2.7% (w/ w) for oxyfuels and 2.0% (w/w) for reformulated gasoline in CO and ozone non-attainment areas, respectively In the European Union, there is no minimum requirement but, depending on the type of fuel oxygenate, the addition of up to 15% (v/v) is allowed

Table 1 lists the common fuel oxygenates, their abbreviations used throughout the text and the physicochemical properties relevant for their analysis or environmental behaviour Because the variability of the available data often is quite large, we critically evaluated the data and reported the most plausible

values For instance, the Henry’s law constants of methyl

tert-butyl ether (MTBE) range between 5.3 3 1024and 3 3 1023

atm m3mol21.2,3The chemical structures of the fuel oxygenates are shown in Fig 1

All fuel oxygenates in Table 1 are Organisation for Economic Cooperation and Development (OECD) High Production Vol-ume Chemicals.4The most important fuel oxygenates today are MTBE and ethanol The production amounts of MTBE and fuel ethanol in the USA in 1999 were 9.3 and 4.4 million tons, respectively.5 tert-Butyl alcohol (TBA) is also of importance

because it is the major degradation product of MTBE in aqueous

systems Other dialkyl ethers, such as ethyl tert-butyl ether (ETBE), tert-amyl methyl ether (TAME) and diisopropyl ether

(DIPE), are currently considered as substitutes for MTBE However, apart from TAME in Finland and ETBE in France, these substances are not yet used in large amounts because they cannot compete economically with MTBE Alcohols other than ethanol are at the moment not considered as fuel oxygenates on

a large scale Methanol-based fuels [e.g M85 with 85% (v/v)

methanol and 15% (v/v) conventional gasoline6] are still in use but their market share is very limited

The physicochemical properties in Table 1 imply that fuel oxygenates released into the environment will predominantly reside in air and water compartments rather than in soil and biota Therefore, the focus of this review is the analysis of fuel oxygenates in air and water phases Compared with classical fuel-related contaminants, such as benzene and other aromatics, alcohols and ethers have higher water solubilities, lower Henry’s law constants and lower sorption constants These

Torsten C Schmidt obtained his PhD in analytical chemistry at

Philipps-University Marburg, Germany, in 1997 The subjects

of his dissertation were the development of new analytical

methods for the determination of aromatic amines in water and

site investigations at former ammunition plants After a

postdoctoral stay with the mass spectrometry unit at the same

university, he has been working as a Research Associate at the

Swiss Federal Institute for Environmental Science and

Technology (EAWAG) His cur-rent research interests are environmental forensics, the environmental assessment of polar fuel constituents such as MTBE and the study of the long-term behaviour of persis-tent organic compounds in the subsurface, in particular, with single compound isotope ratio mass spectrometry (see also the URL http://www.eawag.ch/

~ schmidto).

properties make them difficult to enrich from aqueous samples,

rendering their analysis in water at trace levels (µg L21range)

particularly demanding Because the monitoring of fuel

oxygen-ates in surface and groundwater will be frequently required in

the future, methods for water analysis are emphasised Although

the literature concentrates mainly on MTBE, most analytical

methods are also applicable to other dialkyl ethers Methods for

the trace analysis of alcohols in water are generally scarce, and

only a few might be suitable for the simultaneous analysis of

ethers and alcohols

With this review, we intend: (i) to provide an overview of the

analytical methods available today; (ii) to evaluate critically the

advantages and disadvantages of the different methods; and (iii)

to point out needs for future developments in the environmental

analysis of fuel oxygenates

2 Occurrence in water and air

MTBE is by far the most important fuel oxygenate and its use

has been a matter of controversy for the last few years

Numerous studies on environmental behaviour,2,7–9

toxic-ity3,7,10 and human exposure11,12have therefore been carried

out for MTBE A comprehensive overview on the

environ-mental impact of MTBE was prepared at the University of

California at Davis in 1998.7Other important reports on fuel

oxygenates in water have been published by US agencies.13,14

However, these studies almost exclusively deal with MTBE, for

which a regularly updated information resource (498 entries by

December 2000) on water quality issues is maintained by the

US Geological Survey (USGS).15Apart from MTBE, there is

only little information on the environmental occurrence of fuel

oxygenates

Because the concentrations of fuel oxygenates differ by

orders of magnitude from environmental background to sites

affected by point sources, different analytical strategies are

required Leaking underground storage tanks are frequent point

sources of MTBE into groundwater, which have led to

numerous research and remediation efforts, particularly in the

USA.7,13,14 Possible sources of non-point or diffuse input of

fuel oxygenates include precipitation, stormwater, runoff water

and small watercraft.16–18Recent studies on the occurrence of

volatile organic compounds (VOCs) in groundwater have

shown that MTBE is one of the most frequently detected

substances today.19,20 In samples only affected by non-point

sources, concentrations of MTBE in groundwater and surface

water are in the low µg L21range.18–20In the vicinity of point

sources, high mg L21 levels may be reached in

ground-water.21–25 Schirmer and Barker26 and Landmeyer et al.22 reported that, after a fuel release, the highest concentration of MTBE (and benzene) was not found near the water table but at deeper sampling points The authors attributed this to recharge

on top of the aquifer by rainwater Their results imply that samples should be taken at several depths when investigating contaminated sites

Extensive information on gasoline spill sites as point sources

is available from the American Petroleum Institute.27 In groundwater, MTBE is more recalcitrant to degradation than other fuel constituents of environmental concern, in particular under anaerobic conditions Although laboratory studies have shown that microorganisms can degrade MTBE to TBA under various conditions,9,23–30the transformation rates are low and thus difficult to measure in the field The long-term behaviour of MTBE in groundwater therefore remains one of the most important areas for further research

Compared with groundwater data, there are rather few studies

on fuel oxygenates in air Background concentrations of MTBE range from 0.15 to 1 ppbv,31,32and for ethanol and methanol up

to two times higher.33In urban areas, MTBE concentrations can

be substantially higher, ranging from 0.5 to 7 ppbv.31,34,35For methanol and ethanol, concentrations in urban air of 17 and 4–12 ppbv, respectively, have been reported.32,34 ETBE, TAME, DIPE and TBA were not detected in urban air in the only study available.31

Close to emission sources (i.e gasoline service stations), Vainiotalo et al.36found MTBE concentrations ranging from 69

to 370 ppbv at the centre of the pump island and 0.14 to 34 ppbv

at 50 m distance

MTBE partitions readily into atmospheric water, which may lead to the diffuse input of MTBE with rain into soil and

Table 1 Environmentally relevant physicochemical parameters of fuel oxygenatesa

Methyl

tert-butyl ether (MTBE, 1634-04-4)

Ethyl

tert-butyl ether (ETBE, 637-92-3)

tert-Amyl

methyl ether (TAME, 994-05-8)

Diisopropyl ether (DIPE, 108-20-3)

Methanol (MeOH, 67-56-1)

Ethanol (EtOH, 64-17-5)

Isopropyl alcohol (IPA, 67-63-0)

Isobutyl alcohol (IBA, 78-83-1)

tert-Butyl

alcohol (TBA, 75-65-0)

Water solubility, csat

Henry’s law constant, KH /atm m 3

mol 21

5.9 3 10 24 2.7 3 10 23 1.3 3 10 23 4.77 3 10 23 4.6 3 10 26 5.2 3 10 26 7.9 3 10 26 1.2 3 10 25 1.4 3 10 25

Max IR absorption frequency/cm 21 1205–1213 1199–1207 1185–1193 1122–1130 1055–1063 1052–1060 1141–1149 1037–1045 1207–1215

a Values are at 25 °C unless otherwise stated Data sources (references given therein): Houben, 1Zogorski et al.,13 US EPA, 14Diehl et al.,79Sablji´c et al.81 Environmental Fate Database by Syracuse Research Corporation (http://esc_plaza.syrres.com/efdb.htm) bAbbreviation and CAS no given in parentheses.

Fig 1 Chemical structures of fuel oxygenates (except methanol and ethanol).

groundwater The partitioning is strongly temperature

depend-ent (see Fig 2) Pankow et al.17have shown that, for shallow

aquifers, only a few years might be required for groundwater to

equilibrate with atmospheric MTBE Experimental data on

rainwater concentrations of MTBE are scarce

3 Analytical methods

3.1 Sampling and enrichment: water

Several enrichment and injection techniques are described in the

literature, including purge and trap (P&T), headspace analysis

(HS), direct aqueous injection (DAI) and solid phase

micro-extraction (SPME) An overview of the enrichment and

injection techniques described below is given in Table 2 A

compilation of reported method detection limits (MDLs)

achieved with the discussed methods is given in Table 3 All of

the enrichment techniques are exclusively combined with gas

chromatography (GC) as described in Section 3.3

3.1.1 Water sampling When sampling water for subsequent

analysis of the fuel oxygenates, the same precautions as for

sampling other volatile compounds have to be made.37This

includes the gentle filling of sample bottles until overflow to

prevent volatilisation during sampling and storage The choice

of sample bottles depends on the enrichment technique used

Field blanks with analyte-free water passed through the whole

sampling procedure are recommended

For MTBE analysis, aqueous samples do not have to be

preserved as biodegradation is very slow Cool storage and

analysis within a week is generally recommended for all fuel

oxygenates Koester et al.38recovered 80% of MTBE after 28

days of storage at 4 °C Special precautions are necessary in highly reactive media, such as bacterial cultures, and for the

analysis of the atmospheric oxidation product of MTBE,

tert-butyl formate (TBF), which rapidly hydrolyses under acidic or alkaline conditions.39For TBF, the use of acid as preservative, common in VOC analysis, must therefore be avoided

Some standard preservation methods may also interfere with

analytical techniques, for instance with DAI, where acids (e.g formate or hydrochloric acid) and/or involatiles (e.g mercury

chloride) in rather high concentrations may damage the separation column

3.1.2 Direct aqueous injection DAI allows an aqueous

sample aliquot to be injected directly onto the chromatographic column of a GC This technique is nowadays a common approach in water analysis with GC-flame ionisation detection (GC-FID) and GC-electron capture detection (GC-ECD) A prerequisite in all DAI applications is the use of a wide-bore precolumn that has to be shortened and exchanged regularly The combination of DAI with GC-mass spectrometry (GC-MS) is more difficult because of the large amount of water vapour generated upon aqueous injection: due to the small molar volume of water, evaporation leads to five to eight times higher vapour volumes than for typical organic solvents Very efficient pumps are therefore required to maintain a stable vacuum in the ion source Otherwise, a breakdown of the vacuum causes a shutdown of the instrument

Advantages of the DAI approach are the simplicity and speed

of analysis as essentially no sample handling is necessary apart from the dilution of highly polluted samples Duplicate or triplicate analyses of the same sample can be easily carried out, and very low sample volumes are required The lack of sample preparation minimises the possibility of analyte losses The high polarity of all fuel oxygenates makes their enrichment from the aqueous phase or partitioning to the gas phase difficult Thus, these compounds are ideal candidates for DAI, which may allow the simultaneous analysis of alcohols and ethers in water

Despite these advantages, there are only a few reports on the use of DAI for fuel oxygenate analysis Reasons for this include the above-mentioned necessity for appropriate GC-MS systems and the fact that chemical bonding of stationary phases has only recently been improved to an extent that tolerates multiple water injections without phase deterioration Furthermore, sensitivity may not be sufficient for ultratrace (ng L21range) analysis and precolumns need to be replaced frequently due to the accumula-tion of non-volatiles

Potter40described the use of DAI-GC-FID for the analysis of water-soluble fractions from fuels, including a series of alcohols from methanol to hexanol and MTBE He used hot on-column

Fig 2 Temperature dependence of MTBE partitioning between air and

rainwater.

Table 2 Comparison of injection and enrichment techniques for water analysis

Sensitivity: alcohol/

ethera

Selectivity criteria Purgeable compounds High Henry’s constant Elution earlier than

water (polar column)

Medium to high Henry’s constant, sorption to polymer

Sorption to polymer

Matrix effects Contamination of

apparatus at high concentrations of a volatile compound critical

Applicable to all kinds

of water samples;

addition of high concentrations of salts (salting out) compensates for different matrices

Involatiles or aggressive media may lead to contamination, use of guard column essential

Applicable to all kinds

of water samples;

addition of high concentrations of salts (salting out) compensates for different matrices

Addition of high concentrations of salts (salting out) compensates for different matrices; lifetime of fibre limited

Cost:a

automated/non-automated

a Very good/inexpensive (++), good/inexpensive (+), fair (0), poor/expensive ( 2), very poor/expensive (22) b With the use of an autosampler, the next sample can usually be processed during the current chromatographic run

injection of 1–5 µL at 165 °C and found a much better

performance than with split or splitless injection Using a

DB-624 (J&W Scientific) as a stationary phase, water is eluted

before the analytes MDLs were only 5–100 µg L21, but were

sufficient for the analysis of aqueous extracts of fuels

MDLs as low as 0.1 µg L21were reported by Church et al.41

using a mass spectrometer with a large vacuum capacity

(Finnigan 4000) This study was the first to include the major

dialkyl ethers, alcohols and three carbonyl compounds found as

atmospheric oxidation products in one analytical method A

polar polyethylene glycol column was used for the separation of

the analytes This column retains water more strongly than the

analytes, which opens up a retention window of several minutes

between injection and the water vapour flush An important

feature of this method was the protrusion of the glass wool

packed liner from the hot splitless injector (130 °C) into the cold

oven (30 °C) This approach was used for soil column studies,41

as well as for field investigations.22,26

Hong et al.42have recently shown that a benchtop GC-MS

(GC HP 5890, MS HP 5971A) can also be used for the DAI

technique, and they extended the range of analytes to small

organic acids proposed as oxidation products of MTBE in the

atmosphere In order to avoid contamination of the column by

the high salt content of the samples, a reverse-cup liner filled

with Carbofrit (Restek) was used in a split/splitless injector and

replaced every 30 injections A nitroterephthalic acid-modified

polyethylene glycol (FFAP) stationary phase was found to be

superior to a 6% cyanopropylphenyl type column (DB-624

equivalent) The system was not optimised for sensitivity; the

MDLs ranged from 30 to 100 µg L21

Our laboratory also developed a DAI method using a

benchtop GC-MS (Fisons GC 8000, MD800, 250 L pumps) for

the investigation of groundwater at fuel spill sites.43Using a

polar polyethylene glycol column [Stabilwax, 60 m 3 0.25 mm,

1 µm film (Restek)], the separation of fuel oxygenates, their

major degradation products and BTEX (benzene, toluene,

ethylbenzene, xylenes) was achieved in a single

chromato-graphic run The 60 m column results in a better separation of

MTBE and ETBE than reported by Church et al.41and retains

water sufficiently to prevent a breakdown of the vacuum in the

ion source

3.1.3 Headspace analysis In this section, direct sampling of

the headspace is discussed, whereas sampling of the headspace

with SPME is covered in Section 3.1.5 HS utilises the partitioning of compounds from water to air in a closed system

As in P&T analysis, sufficiently high Henry’s law constants of the analytes are necessary HS is a rather robust technique which can easily be automated, requires little sample preparation and can be used with all kinds of water sample Elevated temperatures and addition of salt (‘salting out’) are used to enhance partitioning of the solutes into the gas phase HS is a static technique and not very sensitive, especially for the

alcohols (compare KH values in Table 1) Thus, HS is well suited for the analysis of highly polluted samples which otherwise might cause matrix and carryover problems

Robbins et al.44used HS to determine Henry’s law constants for BTEX and MTBE; 200 mL of the headspace were withdrawn from the vials and injected; a DB-1 column (J&W Scientific) was used for separation and a flame ionisation detector for detection, although only single analyte samples were used

Nouri et al.45 reported an HS method for MTBE in environmental samples using GC-FID Although the transfer of MTBE to the gas phase was enhanced by thermostatic control of the samples to 60 °C, a ten times higher detection limit than

reported by Robbins et al.44 was obtained However, this method was specifically developed for screening aqueous samples at a contaminated site after experiencing problems with highly contaminated samples in a P&T system The authors reported less interferences in HS analysis of the original samples than in P&T analysis after several dilution steps Shaffer and Uchrin46used HS-GC-FID with an HP-1 column (Hewlett Packard, dimensions not given) for the analysis of MTBE in soil adsorption studies, which were performed with 15

g of soil and 30 mL of water in 60 mL vials suitable for HS After addition of 1 µL MTBE to the system, the concentration

of MTBE was measured by consecutively analysing 50 µL gas phase over a 72 h period

3.1.4 Purge and trap enrichment P&T enrichment can be

used for analytes having a sufficiently high Henry’s law constant (see Table 1) enabling an efficient stripping from the aqueous phase For the dialkyl ethers, P&T is very sensitive and can be used for analysis in the ng L21to µg L21range in water Cryofocusing of the analytes on top of the GC column is essential in order to obtain good chromatographic separation One of the disadvantages of P&T is its susceptibility to contamination from highly polluted samples, in particular when using an autosampler If contamination of the system occurs, it often takes a long time until acceptable baseline levels are achieved again Some laboratories therefore never use P&T for the analysis of unknown samples without previously checking

the VOC content with a less sensitive method (e.g HS) Other

points to consider in the choice of a suitable enrichment method are the complexity of P&T systems compared with other methods, and the low sensitivity for alcohols

A standard analytical method for purgeable organic com-pounds by a P&T technique is US Environmental Protection Agency (EPA) method 524.2.47 When using MS detection, more than 60 VOCs can be analysed simultaneously with this method, but MTBE is the only fuel oxygenate included The EPA method is fully validated and often used in routine VOC monitoring of groundwater and drinking water In this method,

a 5 or 25 mL aliquot of water is introduced into the purge vessel The sample is purged with helium for 11 min at a flow rate of

40 mL min21 Purged compounds are collected on a three-stage trap containing Tenax®, silica gel and charcoal The trap is then heated rapidly to 180 °C The desorbed compounds are cryofocused on the head of the GC capillary column at 210 °C The chromatographic separation is performed on a semi-polar capillary column (6% cyanopropylphenyl, 75 m 3 0.53 mm id,

3 µm film thickness) A purging efficiency of 74% and an MDL

of 0.09 µg L21for MTBE in water samples were reported

Table 3 Method detection limits for trace analysis of fuel oxygenates in

water ( mg L 21 )a

a Different approaches were used to calculate MDLs, which also depend on

the equipment actually being employed Therefore these values have to be

validated in each laboratory bThis method is also suitable for TAME and

ethanol with MDLs of 0.038 and 15 µg L 21 , respectively c This method

is also suitable for TAME, tert-amyl alcohol and TBF with MDLs of 0.1, 0.1

and 5 µg L 21 , respectively d MDL for TAME: 0.02 µg L 21 e High

resolution mass spectrometry

Other EPA approved methods for VOCs, often adapted to

fuel oxygenates, are method 8021B with photoionisation

detection (PID)48and method 8260B with MS detection.49

Raese et al.50 and Reuter et al.18 used a trap based on

Carbopack B/Carboxen 1000 and 1001 sorbents (VOCARB

3000, Supelco) Desorption was carried out at 250 °C,

cryofocusing at 220 °C The MDL of this method for MTBE

was 0.06 µg L21, evaluated by the USGS The time dependence

(1 day to 1 year) of the method performance was investigated on

three different GC-MS instruments using low concentration

spike samples of MTBE (0.1–5 µg L21) in distilled water

Typical results showed a recovery of 98% and a relative

standard deviation of 8–13%

Although traps containing a mixture of different sorbents

appear to be more effective than those with Tenax® as sole

sorbent, the use of Tenax® traps was described in the

literature.45,51,52Nouri et al.45reported a recovery of 90% for

MTBE at room temperature under optimised conditions (purge

volume, 440 mL; purge flow rate, 40 mL min21; purge time, 11

min) and desorption at 200 °C for 2.5 min followed by

cryofocusing at 260 °C

TBA analysis using P&T has been reported for blood and

urine analysis,51,52but not yet for environmental samples Lee

and Weisel52 described a method for the determination of

MTBE and TBA in a urine matrix with a non-commercial P&T

system At a purging temperature of 90 °C, recoveries of 97.0%

(TBA) and 97.6% (MTBE) were achieved using a high flow rate

of 140 mL min21for 15 min A condensation trap was placed in

an ice bath between the purge and the adsorbent tube in order to

remove the large amount of water vapour in the hot purge

gas

Bonin et al.51 reported improved recovery, reproducibility

and sensitivity of MTBE and TBA determination in blood and

urine matrices by using an isotope dilution method with

[2H12]methyl tert-butyl ether and [2H9]tert-butyl alcohol and a

double focusing magnetic sector mass spectrometer The

recoveries of 87–100% for MTBE and 94–107% for TBA were

obtained with conditions of the P&T system similar to the EPA

method 524.2 Due to the use of an internal standard (4-[2H3

]-2-butanone), the MDLs in blood matrix were reduced from 0.05

and 0.25 µg L21to 0.01 and 0.06 µg L21for MTBE and TBA,

respectively The costs of instrumentation and deuterated

standards prevent the adoption of this interesting method for

routine analysis

In closed loop stripping analysis (CLSA),53a method related

to P&T, a closed cycle system is used during the purge interval

Activated carbon filters (20 µg) are used to trap the analytes,

which are subsequently desorbed manually with an appropriate

solvent For MTBE, very low MDLs (20 ng L21) were achieved

even with a flame ionisation detector,54 thus providing the

lowest MDLs of all reported enrichment techniques However,

CLSA has several drawbacks: it is very time consuming,

difficult to automatise and the organic solvent used for filter

extraction often is carbon disulfide, which is highly toxic

3.1.5 Solid phase microextraction SPME has become

increasingly popular for the analysis of volatile and

semi-volatile compounds, whereas attempts to extract polar,

non-volatile compounds from aqueous environments have only

recently been reported.55 The advantages of SPME are its

simplicity, low costs, ease of automation and the commercial

availability of a wide range of fibre coatings with different

properties SPME is either performed directly in aqueous

solution or in the headspace above the sample In the direct

mode, partitioning of the analytes between the fibre coating and

water determines the extraction efficiency In the headspace

mode, two-phase transition processes (partitioning between

water and air and between air and the fibre coating) are

involved

The disadvantages of SPME are the often poor reproducibil-ity due to sensitivreproducibil-ity to matrix effects, which requires the use of appropriate internal standards for quantification, the slow phase transfer kinetics and the limited lifetime of the fibre, in particular in the direct mode

In both modes, and regardless of the type of coating, salting out is used to enhance the partitioning from water either to the coating or to air Mostly used for that purpose is sodium chloride at concentrations of 25–35% (w/v) Adding salt improved the amount extracted with SPME for MTBE56and alcohols55by a factor of three Kadokami et al.57studied the effect of different salts on extraction efficiencies for alcohols from water with a 85 mm polyacrylate phase (Supelco) They reported a 50 times higher peak area for TBA from a saturated potassium carbonate solution (approximately 8 mol L21) in comparison with a saturated sodium chloride solution (approx-imately 6 mol L21) without providing a rationale for this surprising finding The recovery and the MDL for TBA in water samples were excellent for their method (90–104% and 0.63 µg

L21, respectively)

The most common polymer coating in SPME is poly-dimethylsiloxane (PDMS) Sjöberg54 used a 100 mm PDMS coating (Supelco) for MTBE analysis MTBE was extracted in the headspace of a water sample for 10 min at room temperature and subsequently desorbed in a splitless injector for 2 min at 190

°C The limit of quantification for MTBE was approximately 0.7 µg L21with a relative standard deviation of 13.3%

Carboxen/PDMS (Supelco) is a relatively new coating used for SPME of low molecular weight analytes, which has shown

a ten times higher affinity for MTBE than PDMS56and a higher efficiency than PDMS/divinyl benzene (PDMS/DVB) and Carbowax/DVB (both from Supelco).58Achten and Püttmann58 used direct extraction for 60 min and a Carboxen/PDMS fibre at

5 °C With this set-up they achieved an MDL of 10 ng L21for MTBE, which is even lower than with most P&T systems

Cassada et al.59 used SPME with DVB/Carboxen/PDMS (Supelco) for the analysis of MTBE, ETBE, TAME, ethanol and TBA and achieved an MDL of 8 ng L21for MTBE For the other analytes, MDLs were 25 and 38 ng L21for the ethers and

15 and 1.8 µg L21for the alcohols, respectively (see also Table 3)

Górecki et al.55compared the extraction efficiency of several fibre coatings (polyacrylate, PDMS/DVB, Carbowax/DVB, all from Supelco, and Nafion custom-made fibres) for polar analytes, including TBA and isopropyl alcohol (IPA) Of the investigated coatings, the 65 µm PDMS/DVB provided the highest affinity for the adsorption of alcohols and ketones from water Estimated MDLs for TBA and IPA in water were 5 and

2 µg L21, respectively Combining an uncoated (deactivated) precolumn with a DB-WAX analytical column (J&W Scien-tific) improved the peak shape of the alcohols With the polyacrylate phase, insufficient enrichment was found Some of the analytes could not be detected even at 500 µg L21, which

contradicts the results of Kadokami et al.57 When analysing mixtures of polar and non-polar analytes, competitive displacement of polar analytes by less polar compounds with higher affinities for the fibre but smaller diffusion constants occurs with time Short extraction times under vigorous stirring conditions or extraction of non-stirred samples under non-equilibrium conditions regarding phase transfer are recommended to cope with this problem.55 However, shortened extraction times also lower the sensitivity

of the method

3.2 Sampling and enrichment: air

Trace analysis for fuel oxygenates in air generally requires the preconcentration of analytes from large sample volumes (1–10

L air) Using solid sorbent traps,31,36,60,61 this can be done

directly at the sampling site Otherwise, air samples are

collected in stainless steel canisters32,34,62–64or Tedlar®bags65

and concentrated in the laboratory, where cryotrapping34,62–64

or solid sorbents32,65 are used for analyte enrichment Solid

sorbents (on-site or in the laboratory) can be extracted with a

solvent or, more commonly, thermally desorbed at 220–360

°C

Brymer et al.64 carried out a thorough study of sample

collection and storage of VOCs in polished steel canisters They

found stable concentrations of methanol, ethanol, IPA and

MTBE after a storage time of 30 days In contrast, Kelly et al.32

reported stable concentrations 4 days after sampling, but a

significant loss after 12 days During air sampling, humidity is

an important parameter and should be measured

simultane-ously A broader discussion of the different sampling methods

can be found in Pankow et al.31and references cited therein

The US National Institute for Occupational Safety and Health

(NIOSH) provides sampling guidelines for regulated substances

at workplaces, which include the alcohols listed in Table 1 and

MTBE.60 The NIOSH methods are summarised in Table 4

Although these guidelines provide important information, their

focus is on occupational hygiene with threshold limit values in

the mg m23range, and therefore they may not necessarily be

suited for trace analysis Ethanol and methanol are also included

in the NIOSH screening method 2549 for VOCs, which uses a

Carbopack/Carboxen (Supelco) multibed sorbent tube for

enrichment, followed by thermodesorption and GC-MS Harper

et al.61described the use of passive sampling for MTBE, ETBE

and TAME Pankow et al.31used traps filled with Carbotrap B

and Carboxen 1000 (Supelco) and concentrated analytes on-site

from 5 L air

A rather new approach is the use of SPME fibres as passive

samplers for air analysis A non-polar PDMS coating was used

by Quigley et al.66for the sampling of MTBE and other gasoline

vapour constituents in the gas phase It remains unclear at the

moment whether this method is also suited for trace analysis

Nguyen et al.67recently described a method for air analysis

of alcohols that uses the sampling of 200 mL air in glass bottles

and subsequent derivatisation with nitrogen dioxide After 30

min, the alkyl nitrites formed were analysed with GC-ECD

With an injection volume of 500 µL of air, MDLs for methanol,

ethanol and IPA were 0.9, 0.7 and 1.8 ppbv, respectively Due

to the derivatisation step, this approach is limited to alcohols

3.3 Separation

GC is the principal method employed to separate fuel

oxygenates as other chromatographic techniques require

pre-column derivatisation of the analytes for efficient separation A

few exceptions are spectroscopic methods, in particular Fourier

transform infrared (FTIR), which do not necessitate any

separation step (see below) For the separation with GC, a wide

variety of columns may be used, ranging from non-polar PDMS

to polar polyethylene glycol phases The appropriate choice depends on the enrichment and injection technique used as well

as on the sample matrix

For P&T analysis, a semi-polar megabore column (DB-624 like) with a thick film is often used The use of such a column

is described in EPA method 524.2 (see above) and allows the separation of a wide variety of volatile compounds Non-polar columns (DB-1 and DB-5 like) are specified in most HS and SPME applications Polar columns (DB-WAX like, DB-FFAP like) are sometimes used with DAI to retain water strongly For air analysis, long non-polar columns (DB-1 like) with a thick film are used GC ovens are often held at subambient temperatures in order to improve the retention of volatiles Fuel oxygenates in environmental matrices may have to be analysed in the presence of other fuel-related compounds of similar volatility that tend to coelute A comprehensive list of retention indices on a non-polar DB-1 column (J&W Scientific)

is available,65from which possible interfering compounds can

be identified These are n-butane and trans-but-2-ene (coeluting

with methanol), 3-methylbut-1-ene (ethanol), and 2,3-dime-thylbutane and 2-methylpentane (MTBE) J&W Scientific recently introduced a special column for MTBE analysis, called DB-MTBE, which is claimed to be less polar than a 100% PDMS, and was shown to resolve MTBE and 2-methylpen-tane.68On a 75 m 3 0.53 mm DB-624 column (J&W Scientific) with a 3 µm film as used in the P&T EPA method 524.2, MTBE

coelutes with trans-1,2-dichloroethene.47

Lacy et al.69suggested sequential purging and HS of aqueous samples to remove interfering alkanes with higher Henry’s law constants than the fuel oxygenates After six sequential purge steps, HS was carried out and the response for MTBE was back-calculated to the theoretical response without purge steps The results agreed well with P&T measurements In contrast, the headspace measurements without prior purging always overes-timated MTBE concentrations However, the method is rather

time consuming and not very sensitive Gaines et al.56obtained baseline separation of MTBE from both 2,3-dimethylbutane and 2-methylpentane with thermally modulated two-dimen-sional chromatography using the following columns: 2 m 3 0.1

mm, 5 µm Quadrex 007-2 (DB-5 equivalent) and (as second column) 1 m 3 0.1 mm, 0.14 µm Quadrex 007-1701 (DB-1701 equivalent) Another two-dimensional system without thermal

modulation was used by Poore et al.35 for ambient air measurements with a DB-WAX precolumn and a DB-1 (both J&W Scientific) analytical column

3.4 Preparation of standards and calibration

Because some fuel oxygenates are rather sensitive to volatilisa-tion losses, stock soluvolatilisa-tions have to be prepared with special care Typically, a volumetric glass flask is 90% filled with the

Table 4 NIOSH methods for fuel oxygenates in air

Sorbent material Solid sorbent tube: silica gel (50/100

mga)

Solid sorbent tube: coconut shell charcoal 50/100 mg

Solid sorbent tube: two charcoal tubes in series, front

400 mg, back 200 mg

TBA: 1–10

2–96

a At high humidity, 700 mg sorbent should be used b For desorption of IBA, 1% isopropanol in CS 2 is suggested in NIOSH method 1401

solvent (e.g methanol or acetonitrile) A known amount of pure

liquid compound is then added with a microsyringe below the

surface of the solvent, the flask is filled up to the mark and

immediately stoppered The mass concentration is calculated

from the compound’s density

Some researchers44,55,56have used a gravimetric–volumetric

method for the preparation of stock solutions A stoppered flask

is weighed after solvent addition and 10 min to allow drying of

the wetted flask neck After addition of the analyte, the flask is

stoppered and weighed again prior to being filled up to the

mark The mass concentration is calculated from the mass

difference However, according to our experience, the

measure-ment of a small difference between two rather large masses may

not be very accurate and this method is therefore not used in our

laboratory

Calibration standards are prepared by dilution of an aliquot of

the stock solution in double distilled or Nanopure water In our

experience, it is of the utmost importance to check the purity of

the water used for the preparation of standards as MTBE

contamination is frequently observed in Nanopure water Water

exposed to laboratory air also becomes rapidly contaminated

with MTBE.38Stock solutions in organic solvents can be used

up to 1 month;69diluted calibration solutions in water should be

replaced daily.56

For the analysis of fuel oxygenates in air, calibration

standards are often prepared in the solvent used to elute the trap

(see above) Trapping and desorption efficiencies are

deter-mined separately However, calibration with gaseous standards

is generally necessary The preparation of gaseous standard

mixtures has been described for fuel oxygenates32and for an

internal standard mixture used in fuel oxygenate analysis.31

External calibration is common in both water and air analysis

However, addition of an internal standard considerably

im-proves the data reliability The physicochemical behaviour of

the internal standard should closely resemble that of the

analytes Thus the best choice is an isotopically labelled analyte

if the detector can resolve the two isomers (usually possible

only with a mass spectrometer) For methanol, ethanol, IPA,

TBA and MTBE, deuterated and/or 13C-labelled isomers are

commercially available, e.g from Fluka (Buchs, Switzerland),

Cambridge Isotope Laboratories (Woburn, MA, USA) and

Isotec (Miamisburg, OH, USA) The use of MTBE-d358 and

MTBE-d1238has been described in the literature

If labelled compounds cannot be used, one should choose a

physically and chemically similar ether (e.g tert-amyl ethyl

ether) or alcohol (e.g tert-amyl alcohol), which is absent in the

target samples, rather than internal standards common in VOC

analysis (fluorobenzene31,47,50or 1,2-dichlorobenzene-d431,50)

In this case, complete chromatographic separation of the

internal standard from all analytes must be achieved and often

a combined internal/external external calibration is used

3.5 Detection

3.5.1 Flame ionisation detection Next to MS detection, FID

is most frequently used for fuel oxygenate analysis FID is

inexpensive, easy to use, and all fuel oxygenates are detected

with a similar response; however, it lacks selectivity and

sensitivity Therefore, FID is the ideal choice for laboratory

studies with a limited number of compounds (coelution usually

not a problem) and rather high concentrations (high sensitivity

not necessary) For example, in the majority of laboratory

studies on MTBE phase partitioning,70sorption46and

degrada-tion,28,71,72 FID was used For environmental trace analysis,

FID is often not sufficient

3.5.2 Photoionisation detection In PID, the column effluent

is irradiated with intense UV light in the 100–150 nm region,

which causes the ionisation of molecules The ions are

accelerated in an electric field and the resulting ion current amplified PID allows a more selective detection of fuel oxygenates in the presence of alkanes which have a much weaker response.73Although it might be a low price alternative

to MS detection, the use of PID in the reviewed literature has rarely been described.64,69

3.5.3 Mass spectrometry MS is the universal detection

principle in GC Correspondingly, it is also the predominant detector in fuel oxygenate analysis Its use is mandatory in P&T analysis according to EPA method 524.2 The selectivity of MS

is very good, and thus interferences due to other fuel constituents are unlikely The on-line acquisition of mass spectra in combination with retention times allows the un-equivocal identification of compounds In scan mode the sensitivity is comparable to FID in many cases If the target compounds are known, selected ion monitoring (SIM) can be used, which provides 100–1000 times lower MDLs than FID, thus rendering the mass spectrometer the most sensitive GC detector for fuel oxygenates (see also Table 3) Electron impact ionisation is the only ionisation mode described Due to the rather high energy transfer in this ionisation mode, fuel oxygenates, except for methanol, do not yield molecular ions Instead, after a-cleavage, (M2CH3)+or (M2C2H5)+fragments are obtained as base peaks in the mass spectra In Table 5, the main fragment ions of the fuel oxygenates, which are commonly used for quantification and confirmation in SIM, are given Drawbacks of MS are its cost, both in acquisition and maintenance, and limitations in sample size due to the vacuum requirements (see discussion above for DAI-GC-MS)

3.5.4 Atomic emission detection The use of atomic

emission detection (AED) has not yet been described for environmental analysis of fuel oxygenates However, there are two reports on gasoline analysis with GC-AED, which have both utilised atomic oxygen emission at 777.3 nm.74,75At this wavelength, a selectivity of about 5000 : 1 over carbon is obtained, but the low sensitivity does not allow trace analysis In general, trace oxygen measurements are complicated by the background signal from air oxygen diffusing into the de-tector

3.5.5 Fourier transform infrared spectroscopy FTIR

spectroscopy has frequently been used in laboratory studies for monitoring the degradation of fuel oxygenates in air.76,77 In these cases, very long absorption path lengths are used (up to 60 m) in order to enhance sensitivity Remote sensing with FTIR has recently been described for the determination of methanol in the ppmv range.78 Comparison of this range with typical ambient air concentrations (see above) shows that the sensitivity

Table 5 Ions used in GC-MS-SIM with corresponding fragments Compound Quantification

ion (m/z)

Corresponding fragment ion Confirmation

ion (m/z)

MTBE 73 (CH 3 ) 2 COCH 3+a-cleavage 43, 57 ETBE 59 (CH 3 ) 2 COH + a-cleavage,

onium reaction

57, 87 TAME 73 (CH 3 ) 2 COCH 3+a-cleavage 43, 55 DIPE 45 CH 3 CHOH + a-cleavage,

onium reaction

43, 87 Methanol 31 CH 2 OH + hydrogen radical

loss

32

IPA 45 CH 3 CHOH + a-cleavage 59, 43 IBA 43 (CH 3 ) 2 CH + a-cleavage 74, 31 TBA 59 (CH 3 ) 2 COH + a-cleavage 41 TBF 59 (CH 3 ) 2 COH + a-cleavage, CO

loss

57, 41

still needs to be improved by at least three orders of

magnitude

In aqueous systems, FTIR detection has not yet been used

However, FTIR has been applied to the determination of ethers

and alcohols in gasolines.79,80 Reconstructed IR absorption

frequencies from Diehl et al.79are reported in Table 1 Due to

the overlap of analyte signals, Choquette et al.80 employed

multivariate calibration to obtain single compound

concentra-tions This technique may also be helpful for the resolution of

other overlapping signals In both cases, determination of fuel

oxygenates in gasoline was limited to the per mille range, and it

seems rather difficult to improve the sensitivity to

environmen-tally relevant concentrations

4 Conclusions and outlook

A wide variety of analytical methods have been published on

the analysis of fuel oxygenates in environmental matrices The

choice of an appropriate method depends on the compounds and

matrix to be investigated, the concentration ranges to be

analysed, the available laboratory equipment and compliance

with regulations In general, the enrichment of fuel oxygenates

is the crucial step in all methods, whereas the separation and

detection of fuel oxygenates and their degradation products are

less critical

For the analysis of fuel oxygenates in water, it is necessary to

distinguish between alcohol and ether analysis Enrichment

techniques that allow the sensitive analysis of low molecular

weight alcohols in water are still lacking For this purpose,

SPME with the recently introduced coatings Carboxen/PDMS

or DVB/Carboxen/PDMS may be useful, as shown for

ethanol.59DAI also seems to be promising Both methods need

to be further evaluated for a variety of alcohols

For ether analysis, all enrichment techniques described are

principally suited The concentration range to be analysed will

often determine the choice of a particular method For

laboratory studies, HS-GC-FID will mostly be sufficient At

spill sites, HS, SPME or DAI might be used, preferably with

GC-MS DAI allows the simultaneous determination of alcohol

degradation products of dialkyl ethers For measurements of

background concentrations, P&T-GC-MS is the only approved

method However, new developments in SPME have

sig-nificantly improved MDLs and it may thus complement P&T

methods in ultratrace analysis (ng L21range) of ethers

To date, methods for the in situ determination of alcohols and

ethers in aqueous systems at the trace level are not available

The development of such methods should be further pursued

because they would allow real-time measurements at a high

spatial resolution

For the analysis of fuel oxygenates in air, validated sampling

procedures exist, which involve trapping and either solvent or

thermal desorption of the analytes FTIR allows real-time, in

situ measurements of most analytes, but is still limited to

laboratory systems due to poor sensitivity However, with the

advent of new lasers in the IR range, it may become a versatile

tool in air monitoring

5 Acknowledgements

We gratefully acknowledge financial support by Compagnie

Générale des Eaux (CGE) and the Swiss Agency for

Environ-ment, Forests and Landscape (BUWAL) H.-A D

acknowl-edges the Swiss Agency for Development and Cooperation

(SDC) for support of her stay at EAWAG We thank Annette

Johnson and Luc Zwank for helpful discussions

6 Appendix: Abbreviations

Abbreviations used for the fuel oxygenates are given in Table 1

AED atomic emission detection BTEX benzene, toluene, ethylbenzene, xylenes CLSA closed loop stripping analysis

DAI direct aqueous injection EPAUS Environmental Protection Agency FID flame ionisation detection FTIR Fourier transform infrared

HS headspace analysis MDL method detection limit

P&T purge and trap PID photoionisation detection SPME solid phase microextraction VOC volatile organic compounds

7 References

1 M C M Houben, Motor Gasoline, ed E L Marshall, Royal Society

of Chemistry, Cambridge, 1995, pp 45–71.

2 P J Squillace, J F Pankow, N E Korte and J S Zogorski, Environ.

Toxicol Chem., 1997, 16, 1836.

3 World Health Organization, Methyl tertiary-Butyl Ether,

Environ-mental Health Criteria 206, World Health Organization, Geneva,

1998.

4 Organization for Economic Cooperation and Development (OECD),

The 1997 OECD List of High Production Volume Chemicals, OECD

Environment Directorate, Paris, 1997.

5 Energy Information Administration (EIA), Monthly Oxygenate

Telephone Report, Form EIA-819M, EIA, Washington, DC, 1999.

6 W G Rixey, Proceedings of the NWWA/API Conference on

Petroleum Hydrocarbons and Organic Chemicals in Ground Water,

National Water Well Association, Houston, 1994, p 75.

7 A Keller, J Froines, C Koshland, J Reuter, I Suffet and J Last,

Health and Environmental Assessment of MTBE, University of

California, Davis, 1998 Available at http://tsrtp.ucdavis.edu/ mtberpt/

8 M Schirmer, Grundwasser, 1999, 3, 95.

9 R C Prince, Crit Rev Microbiol., 2000, 26, 163.

10 L Caprino and G I Togna, Environ Health Perspect., 1998, 106,

115.

11 J M Davidson, Proceedings of the NWWA/API Conference on

Petroleum Hydrocarbons and Organic Chemicals in Ground Water,

ed A Stanley, National Water Well Association, Houston, 1995, p 285.

12 S L Brown, Regul Toxicol Pharmacol., 1997, 25, 256.

13 J S Zogorski, A Morduchowitz, A L Baehr, B J Bauman, D L Conrad, R T Drew, N E Korte, W W Lapham, J F Pankow and

E R Washington, Interagency Assessment of Oxygenated Fuels,

Office of Science and Technology Policy, Washington, DC, 1996, ch 2.Available at http://www.epa.gov/otaq/regs/fuels/ostp-2.pdf

14 US EPA, Oxygenates in Water: Critical Information and Research

Needs, EPA-600-R-98/048, Office of Research and Development,

Washington, DC, 1998.

Geological Survey, Denver, CO, 2000 Available at http:/ /wwwsd.cr.usgs.gov/nawqa/vocns/mtbe/bib/

16 P J Squillace, J S Zogorski, W G Wilber and C V Price, Environ.

Sci Technol., 1996, 30, 1721.

17 J F Pankow, N R Thomson, R L Johnson, A L Baehr and J S.

Zogorski, Environ Sci Technol., 1997, 31, 2821.

18 J E Reuter, B Allen, R C Richards, J F Pankow, C R Goldman,

R L Scholl and J S Seyfried, Environ Sci Technol., 1998, 32,

3666.

19 T J Lopes and D A Bender, Environ Pollut., 1998, 101, 221.

20 A L Baehr, P E Stackelberg and R J Baker, Water Resour Res.,

1999, 35, 127.

21 P Garrett, M Moreau and J D Lowry, Proceedings of the NWWA/

API Conference on Petroleum Hydrocarbons and Organic Chemicals

in Ground Water, National Water Well Association, Houston, TX,

1986, p 227.

22 J E Landmeyer, F H Chapelle, P M Bradley, J F Pankow, C D.

Church and P G Tratnyek, Ground Water Monitor Rev., 1998, 18,

93.

23 D P Lince, L R Wilson and G A Carlson, Bull Environ Contam.

Toxicol., 1998, 61, 484.

24 J W Weaver, J E Haas and J T Wilson, Non-Aqueous Phase

Liquids (NAPLs) in the Subsurface Environment: Assessment and

Remediation, ed L N Reddi, ASCE, Washington, DC, 1996, p.

707.

25 R C Borden, R A Daniel, L E LeBrun IV and C W Davis, Water

Resour Res., 1997, 33, 1105.

26 M Schirmer and J F Barker, Ground Water Monitor Rev., 1998, 18,

113.

27 American Petroleum Institute, MTBE Resource Page, 1999

Availa-ble at http://www.api.org/ehs/mtbelink.htm

28 M R Mormile, S Liu and J M Suflita, Environ Sci Technol., 1994,

28, 1727.

29 R J Steffan, K McClay, S Vainberg, C W Condee and D L.

Zhang, Appl Environ Microbiol., 1997, 63, 4216.

30 J R Hanson, C E Ackerman and K M Scow, Appl Environ.

Microbiol., 1999, 65, 4788.

31 J F Pankow, W Luo, L M Isabelle, D A Bender and R J Baker,

Anal Chem., 1998, 70, 5213.

32 T J Kelly, P J Callahan, J Plell and G F Evans, Environ Sci.

Technol., 1993, 27, 1146.

33 E Leibrock and J Slemr, Atmos Environ., 1997, 31, 3329.

34 E Grosjean, D Grosjean, R Gunawardena and R A Rasmussen,

Environ Sci Technol., 1998, 32, 736.

35 M Poore, B Chang, F Niyati and S Madden, ACS Div Environ.

Chem Prep Ext Abstr., 1997, 37, 407.

36 S Vainiotalo, Y Peltonen and P Pfaffli, Atmos Environ., 1998, 32,

3503.

37 M T Koterba, S D Wilde and W W Lapham, Ground-Water Data

Collection Protocols and Procedures for the National Water-Quality

Assessment Program: Collection and Documentation of

Water-Quality Samples and Related Data, Open File Report 95-399, US

Geological Survey, Denver, CO, 1995.

38 C J Koester, M L Davisson and J E Moran, Proceedings of the

American Water Works Association Water Quality Conference, San

Diego, 1998.

39 C D Church, J F Pankow and P G Tratnyek, Environ Toxicol.

Chem., 1999, 18, 2789.

40 T L Potter, Ground Water Monitor Rev., 1996, 16, 157.

41 C D Church, L M Isabelle, J F Pankow, D L Rose and P G.

Tratnyek, Environ Sci Technol., 1997, 31, 3723.Additional

informa-tion available at http://cgr.ese.ogi.edu/MTBE/

42 S Hong, C M Duttweiler and A T Lemley, J Chromatogr A, 1999,

857, 205.

43 L Zwank, M Berg, S B Haderlein and T C Schmidt, in

preparation.

44 G A Robbins, S Wang and J D Stuart, Anal Chem., 1993, 65,

3113.

45 B Nouri, B Fouillet, G Toussaint, R Chambon and P Chambon, J.

Chromatogr A, 1996, 726, 153.

46 K L Shaffer and C G Uchrin, Bull Environ Contam Toxicol.,

1997, 59, 744.

47 J W Munch and J W Eichelberger, J Chromatogr Sci., 1992, 30,

471.

48 US EPA, Method 8021B, Aromatic and Halogenated Volatiles by Gas

Chromatography Using Photoionization and/or Electrolytic

Con-ductivity Detectors, SW-846, 2nd Revision, Office of Solid Waste,

Washington, DC, 1996 Available at http://www.epa.gov/epaoswer/

hazwaste/test/8xxx.htm

49 US EPA, Method 8260B, Volatile Organics by Gas Chromatography/

Mass Spectrometry (GC/MS), SW-846, 2nd Revision, Office of Solid

Waste, Washington, DC, 1996 Available at http://www.epa.gov/ epaoswer/hazwaste/test/8xxx.htm

50 J W Raese, M W Sandstrom and D L Rose, US Geological Survey

Laboratory Method for Methyl tert-Butyl Ether and Other Fuel Oxygenates, Fact Sheet 219-95, US Geological Survey, Denver, CO,

1995.

51 M A Bonin, D L Ashley, F L Cardinali, J M McGraw and J V.

Wooten, J Anal Toxicol., 1995, 19, 187.

52 C.-W Lee and C P Weisel, J Anal Toxicol., 1998, 22, 1.

53 K Grob and F Zürcher, J Chromatogr., 1976, 117, 285.

54 A Sjöberg, MS Thesis, ETH Zürich/EAWAG, Department of Microbiology, Dübendorf, 1994.

55 T Górecki, P Martos and J Pawliszyn, Anal Chem., 1998, 70, 19.

56 R B Gaines, E B Ledford, Jr and J D Stuart, J Microcolumn Sep.,

1998, 10, 597.

57 K Kadokami, K Sato, T Iwamura and Y Hanada, Bunseki Kagaku,

1996, 45, 1013.

58 C Achten and W Püttman, Environ Sci Technol., 2000, 34, 1359.

59 D A Cassada, Y Zhang, D D Snow and R F Spalding, Anal.

Chem., 2000, 72, 4654.

60 M E Cassinelli and P F O’Connor, NIOSH Manual of Analytical

Methods, National Institute for Occupational Safety and Health

(NIOSH), Washington, DC, 4th edn., 1994 Available at http:/ /www.cdc.gov/niosh/nmam/nmammenu.html

61 M Harper, C O’Lear and A A Fiore, Analyst, 1996, 121, 1265.

Supplementary material available on the SKC Inc website at http:/ /www.skcinc.com/reports.html

62 T W Kirchstetter, B C Singer, R A Harley, G R Kendall and W.

Chan, Environ Sci Technol., 1996, 30, 661.

63 E Grosjean, R A Rasmussen and D Grosjean, Atmos Environ.,

1998, 32, 3371.

64 D A Brymer, L D Ogle, C J Jones and D L Lewis, Environ Sci.

Technol., 1996, 30, 188.

65 S K Hoekman, Environ Sci Technol., 1992, 26, 1206.

66 C J Quigley, D T Allen and R L Corsi, ACS Div Environ Chem.

Prep Ext Abstr., 1997, 37, 384.

67 H T.-H Ngyuyen, Y Fujio, N Takenaka, H Bandow and Y Maeda,

Anal Chim Acta, 1999, 402, 233.

68 C George and A K Vickers, Evaluation of a New Capillary Column

for the Positive Identification of Methyl-tert-Butyl Ether (MTBE),

J&W Scientific, Folsom, CA, 1999

69 M J Lacy, G A Robbins, S Wang and J D Stuart, J Hazard.

Mater., 1995, 43, 31.

70 M Poulsen, L Lemon and J F Barker, Environ Sci Technol., 1992,

26, 2483.

71 J P Salanitro, L A Diaz, M P Williams and H L Wisniewski,

Appl Environ Microbiol., 1994, 60, 2593.

72 C K Yeh and J T Novak, Water Environ Res., 1995, 67, 828.

73 I A L Rhodes and A W Verstuyft, Selecting Analytical Methods for

the Determination of Oxygenates in Environmental Samples and Gasoline, American Petroleum Institute, Washington, DC, 1999.

Available at http://www.api.org/ehs/oxyMTBE10.doc

74 J W Diehl, J W Finkbeiner and F P DiSanzo, J High Resolut.

Chromatogr., 1995, 18, 108.

75 S R Goode and C L Thomas, J Anal At Spectrom., 1994, 9, 73.

76 T J Wallington and S M Japar, Environ Sci Technol., 1991, 25,

410.

77 E C Tuazon, W P L Carter, S M Aschmann and R Atkinson, Int.

J Chem Kinet., 1991, 23, 1003.

78 L Ingling and T L Isenhour, Field Anal Chem Technol., 1999, 3,

105.

79 J W Diehl, J W Finkbeiner and F P DiSanzo, Anal Chem., 1992,

64, 3203.

80 S J Choquette, S N Chesler, D L Duewer, S Wang and T C.

O’Haver, Anal Chem., 1996, 68, 3525.

81 A Sablji´c, H Güsten, H Verhaar and J Hermens, Chemosphere,

1995, 31, 4489.