Direct quantification of aromatic hydrocarbons in geochemical fluids with a mid-infrared attenuated total reflection sensor

Organic Geochemistry 55 (2013) 63–71

Contents lists available at SciVerse ScienceDirect

Organic Geochemistry

journal homepage: www.elsevier .com/locate /orggeochem

Direct quantification of aromatic hydrocarbons in geochemical fluids with amid-infrared attenuated total reflection sensor

Bobby Pejcic a,⇑, Leigh Boyd b, Matthew Myers a, Andrew Ross a, Yosef Raichlin c, Abraham Katzir c,Rui Lu d,e, Boris Mizaikoff d

a CSIRO, Earth Science and Resource Engineering, Wealth from Oceans Flagship, Kensington, WA 6151, Australiab National Measurement Institute, Perth, WA 6151, Australiac Tel Aviv University, School of Physics and Astronomy, Tel Aviv, Israeld University of Ulm, Institute of Analytical and Bioanalytical Chemistry, Albert-Einstein-Allee 11, 89081 Ulm, Germanye University of Science and Technology of China, Department of Chemistry, Hefei 230026, PR China

a r t i c l e i n f o

Article history:Received 8 June 2012Received in revised form 7 November 2012Accepted 25 November 2012Available online 5 December 2012

0146-6380/$ - see front matter Crown Copyright � 2http://dx.doi.org/10.1016/j.orggeochem.2012.11.011

⇑ Corresponding author. Tel.: +61 8 6436 8814; faxE-mail address: [email protected] (B. Pejcic).

a b s t r a c t

An improved analytical method for directly and rapidly quantifying various monocyclic and polycyclicaromatic molecules in oil–water mixtures has been developed. This study reports on the application ofa sensor based on attenuated total reflection Fourier transform infrared spectroscopy (ATR–FTIR) fordetermining the concentration of a number of volatile organic compounds (VOCs) and polycyclic aro-matic hydrocarbons (PAHs) in water. The mid-infrared (MIR) sensor consists of a zinc selenide (ZnSe)waveguide with the surface modified by a thin poly(isobutylene) (PIB) coating. The sensitivity was inves-tigated at different polymer film thicknesses and molecular weights. The analytical performance of thesensor was validated in the laboratory against a standard analytical technique for analyzing petroleumbased samples. We have shown that the sensor may accurately quantify the benzene, toluene, ethylben-zene, xylenes and naphthalene (BTEXN) concentration in an oil–water mixture. The ATR–FTIR methodovercomes the limitations associated with sampling/sample preparation and has the advantage that itmay easily discriminate between meta- and para-xylene, which is difficult with conventional analyticaltechniques (i.e., GC). In addition, this technology may potentially be deployed in the field for geochemicalmapping and to monitor in situ the concentration profile of a number of hydrocarbons in geological for-mations (e.g., petroleum systems). To demonstrate this, a miniaturized prototype of an ATR–IR sensorsystem taking advantage of planar-tapered silver halide fibers has been developed and tested.

Crown Copyright � 2012 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Considerable research effort has been directed towards devel-oping analytical tools for identifying the type of organic moleculespresent and quantifying their concentration in a range of petro-leum based systems (Rodgers and McKenna, 2011). Recent ad-vances in gas chromatography (GC) and mass spectrometry (MS)have been able to provide invaluable compositional informationand in most cases the concentration of many different hydro-carbon components in complex samples can be easily obtained(Ballesteros-Gomez and Rubio, 2011). The determination of hydro-carbons in crude oils typically involves several steps such as thecollection of a representative sample followed by extraction/separation prior to molecular characterization and quantificationusing GC–MS (Audino et al., 2004). Sampling/sample handling isone of the major contributors to uncertainty, and the challenges

012 Published by Elsevier Ltd. All r

: +61 8 6436 8555.

associated with quantifying volatile organic compounds and tracelevels of hydrocarbons in geochemical and environmental samplesis well established (Demeestere et al., 2007). Consequently, there isa demand for more reliable methods and analytical techniques forcharacterizing oil matrices in their natural unperturbed state. Overthe past several decades there has been significant interest in usingchemical sensors to address these and other environmental moni-toring issues (Lawrence, 2006; Lieberzeit and Dickert, 2007).Although many sensors are suitable for measuring hydrocarbonconcentrations at very low levels (i.e., ppb to ppm or lg/l to mg/l), the majority of these devices struggle to separate the varioussignal contributions during the analysis of multi-componenthydrocarbon mixtures (Ueyama et al., 2002; Pejcic et al., 2011b).Analyte selectivity remains a major challenge impeding theacceptance of most sensors for assaying complex geochemicaland environmental samples.

The sensor based on attenuated total reflection Fourier trans-form infrared (ATR–FTIR) spectroscopy is one promising analyticaltool that is capable of achieving reliable hydrocarbon identification

ights reserved.

64 B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71

and quantification in geochemical fluids (Mizaikoff, 2003; Pejcicet al., 2009b; Gonzalvez et al., 2011). Due to its inherent selectivitythe ATR–FTIR has been successfully used to detect a wide variety oforganic compounds such as phenols (Yang and Cheng, 2001), caf-feine (Alcudia-Leon et al., 2008), organophosphonates (Bryantet al., 2007), adamantane (Luzinova et al., 2009), alkyl halides(Acha et al., 2000; Yang and Ramesh, 2005), aromatic hydrocarbons(Karlowatz et al., 2004; Pejcic et al., 2009a; Young et al., 2011), andfor the detection of oil-in-water as well as water-in-oil (Luzinovaet al., 2012a,b). The sensor consists of an IR light source, a MIRtransparent waveguide to propagate the signal, and an opticaltransducer/detector. The waveguide is usually either a trapezoidalshaped zinc selenide (ZnSe) crystal or an optical fiber based on asilver halide or chalcogenide material; the surface of the wave-guide is in many cases coated with a thin polymer film. The analytemolecules that are present in solution partition and enrich into thepolymer membrane, and the generated evanescent field interactsdirectly with the extracted compounds. A hydrophobic coating al-lows non-polar hydrocarbon molecules to partition into the mem-brane, while water and other interfering components are excludedfrom the analytical volume probed by the evanescent field extend-ing a couple of micrometers into the adjacent medium. This meth-od is based on the principles of solid phase extraction and it hasbeen shown that detection limits of low ppb levels may be ob-tained under certain conditions (Roy and Mielczarski, 2002;Karlowatz et al., 2004; Lin and Li, 2010). Recent engineeringadvances in IR light sources, waveguides, and detectors/spectrometers have led to the development of miniaturized MIRsensing systems that are capable of being deployed directly inthe field (Charlton et al., 2006; Kim et al., 2008; Wang et al.,2012). A portable MIR sensor has been remotely and successfullyoperated in marine systems during environmental monitoring ofvarious chlorinated and aromatic hydrocarbons (Mizaikoff, 1999;Kraft and Mizaikoff, 2000; Kraft et al., 2003). These studiesrevealed that the sensor is unaffected by salinity, turbidity andnatural organic matter/organic pollutants.

Crude oil is a complex mixture containing hundreds of com-pounds with a wide range of molecular weights. The hydrocarbondistribution of a crude oil varies considerably and the factors influ-encing the oil composition are discussed in greater detail else-where (Tissot and Welte, 1984). Understanding how oil is alteredby chemical and/or physical processes is of significant researchinterest and a number of papers have investigated the effect ofoil–water interactions on the hydrocarbon distribution within acrude oil (Lafargue and LeThiez, 1996; Barman Skaare et al.,2007). Others have characterized the formation/produced waterfrom the North Sea and showed that it contains dispersed oil alongwith a variety of organic compounds (Faksness et al., 2004).Detecting petroleum contaminants in water arising from oil spillsand natural oil seeps are important issues in environmental science(Readman et al., 1992) and petroleum exploration (Kvenvoldenand Cooper, 2003). A robust and portable analytical technique thatmay directly measure the concentration of volatile organic com-pounds (VOCs) and polycyclic aromatic hydrocarbons (PAHs)would be extremely valuable for understanding oil–water interac-tions and issues relating to environmental geochemistry. Theobjective of this paper is to evaluate the suitability of ATR–FTIRspectroscopy for quantifying aromatic hydrocarbon compoundsin complex multi-component solutions. An ATR–FTIR sensor basedon a polyisobutylene (PIB) coated zinc selenide (ZnSe) waveguidewas investigated for the detection of benzene (B), toluene (T), eth-ylbenzene (E), xylenes (X) and naphthalene (N) in water containingdissolved crude oil. A number of polymeric based materials havebeen previously used for extracting different types of hydrocarboncompounds (Regan et al., 2003; Yang and Ramesh, 2005; Flavinet al., 2006). However, PIB is one of a few polymers that is appro-

priate for BTEXN detection in terms of having no or very littleinterfering absorption bands in the mid-infrared region of interest.The potential of the mid-infrared spectroscopic sensor for deter-mining hydrocarbons in petroleum based systems (e.g., formationwater) will be demonstrated by comparing its analytical perfor-mance to a standard method. Furthermore, first tests using a min-iaturized mid-infrared sensor prototype providing similarsensitivity will be shown.

2. Materials and methods

2.1. Materials and reagents

Benzene (99%), toluene (99.8%), p-xylene (99%), m-xylene (99%),o-xylene (97%), ethylbenzene (99.8%), naphthalene (99%), and n-hexane were all obtained from Sigma–Aldrich and were used aspurchased. All organic solvents were AR grade and used withoutfurther purification. Poly(isobutylene) (PIB) was supplied by Scien-tific Polymer Products, Inc. (Ontario, NY, USA). The molecularweight (Mw) of PIB was determined using gel permeation chroma-tography (polystyrene standards, chloroform solvent) and is400,000. The glass transition temperature (Tg) was measured usinga Mettler Toledo DSC and is �64 �C. Studies were also performedusing PIB at different Mw. Ethylene–propylene copolymer (60:40)was purchased from Sigma–Aldrich and used as provided.

Two different types of waveguides were used in this study. Atrapezoidal (4 mm thick, 80 mm long, angle of incidence 45�) ZnSewas used as a multi-reflection waveguide. The ZnSe was mountedinto a trough plate ATR unit and stainless steel flow cell, which wasobtained from Pike Technologies (Madison, WI, USA). Polycrystal-line silver halide waveguides were prepared by the research teamat Tel-Aviv University for transmitting IR radiation in the spectralrange of 25,000–400 cm�1 (0.4–25 lm). The waveguides usedherein are of the composition AgCl0.4Br0.6 and are characterizedby typical attenuation losses in the range of 0.5 dB/m at 10.6 lm(Saar et al., 1986). Silver halide waveguides with a planarized ac-tive sensing region were created via mechanical press-tapering ofunclad cylindrical fiber segments with an initial diameter of700 lm, yet providing attached cylindrical coupling segmentsfacilitating coupling of IR radiation emanating from an FTIR spec-trometer into the waveguide. The flat-tapered segment has a thick-ness of approximately 150 lm, a width of 4 mm and a length of40 mm in the middle (see Fig. 1).

2.2. Film preparation

Prior to film deposition the surface of the ZnSe waveguide wascleaned with acetone soaked Whatman lens cleaning tissue (What-man International Ltd., Maidstone, England) followed by rinsingwith acetone and drying by flowing high purity nitrogen gas forseveral minutes. A thin polymer film was deposited onto the ZnSewaveguide by placing 500 ll of a chloroform solution that con-tained PIB. The PIB solution concentration was varied between0.5% and 1% w/v in order to study the effect of film thickness (from4.5 to 8.4 lm) on the hydrocarbon sensitivity. The thickness of thelayer was determined by differential weighing using the methodreported elsewhere (Murphy and McLoughlin, 2003). In all cases,the PIB coated waveguide was allowed to dry overnight and IRspectra were collected to ensure that the solvent evaporated com-pletely. All of the films were reproducible in thickness (<10% vari-ation) and uniform in terms of quality and integrity.

For coating silver halide waveguides with a thin layer ofethylene–propylene copolymer (E/P-co), a 1% (w/v) coating solu-tion was prepared by dissolving 0.5 g of granular polymer underreflux in 50 ml of n-hexane. Approximately 100 ll of clear hot

Fig. 1. Experimental set up of fiber optically coupled planar-tapered silver halide waveguide sensor (PM = planar mirror, OAPM = off-axis parabolic mirror, WG = waveguide,MCT = mercury-cadmium-telluride).

B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71 65

solution was then applied to each side of the planar segment of aflat-tapered silver halide waveguide using an Eppendorf pipette.Thus obtained coated silver halide sensing elements were kept atroom temperature for at least 2 h, thereby ensuring evaporationof the solvent and reconstitution of a homogenous polymer filmwith a thickness of approximately 5 lm.

2.3. Instrumentation and sensor calibration

Sensor measurements were made using a Bruker Vertex 70 Fou-rier transform infrared spectrometer equipped with a liquid nitro-gen cooled mercury-cadmium-telluride (MCT) detector coupled toboth waveguides, conventional ATR crystals and flat-tapered silverhalide sensing elements. Analyte solutions were prepared in deion-ized water and pumped over the polymer coated ZnSe surfaceusing an Ismatec peristaltic pump (IDEX Corporation) at a constantflow rate of 1.5 ml/min. Data were recorded in the 4000–650 cm�1

range using a spectral resolution of 2 cm�1 and a total of 32 scanswere averaged for each spectrum. All experiments were performedat room temperature (21 ± 1 �C).

Single analyte calibration curves were established by dissolvingthe BTEX compounds in deionized water in the concentrationrange between 0 and 100 ppm. Naphthalene standards (0–20 ppm) were made by serial dilution of a saturated solution,which was prepared by weighing naphthalene (0.2 g, AR grade)into a glass Schott bottle that comprised deionized water(1000 ml) followed by constant stirring (�500 rpm) at room tem-perature (�21 �C) for 3 days. The solution was filtered through a0.45 lm filter paper and the naphthalene concentration was veri-fied by GC–MS (20.5 ± 0.5 ppm, experiment repeated on three sep-arate occasions). All analytical calibration curves were generatedusing at least four standard solutions. Since the VOC concentrationin water may vary considerably during solution preparation/stor-age, it is recommended that the calibration standards are preparedin sealed glass containers with very little headspace and the aque-ous solutions analyzed almost immediately. Although organic sol-vents (i.e., methanol) have been used to disperse and stabilize theVOC during the preparation of stock standard solutions (Kraft andMizaikoff, 2000; Lima et al., 2011), we avoided their use in thisstudy in order to minimize potential issues that may arise fromsolvent/matrix effects during the measurement of complex sam-ples. Previous GC–MS studies by us have shown that the loss dur-ing the preparation of the BTEX standards directly in deionizedwater can vary between 20% and 60%, and this depended on thetype of hydrocarbon, the concentration and the method of prepara-tion (i.e., vigorous shaking versus ultrasonication, storage timeprior to analysis) (Pejcic et al., 2011b). To limit losses by volatilityand degradation, all standards were prepared freshly and ATR–FTIRstudies were performed within 5 h. All of the standards used in theconstruction of the calibration curves in the ATR–FTIR studies havebeen verified using GC–MS.

2.4. Spectral data processing

The infrared absorption features were evaluated by peak areaanalysis using the software package OPUS (Bruker Optics). At leastduplicate measurements were performed and the peak areas fromeach spectral region were averaged.

2.5. Total petroleum hydrocarbon solution preparation

Total petroleum hydrocarbons (TPHs) dissolved in water wereprepared according to a procedure outlined elsewhere (Tsvetnenkoand Evans, 2002; Qi et al., 2011). Briefly, this involved placingNorth West Shelf (Australia) crude oil (70 ml) into a glass flask thatcontained deionized water (900 ml) and allowing the mixture toequilibrate for a period of 3 days at room temperature (�21 �C).The flask was sealed and the solution constantly stirred using amagnetic stirrer bar at a rotation speed of �500 rpm.

2.6. Gas chromatography–mass spectrometry (GC–MS)

The concentration of the various solutions used during the ATR–FTIR study was analyzed using GC–MS. The BTEX compounds weredetermined on a HP 6890 GC-Agilent 5973 MSD with an EST Enconpurge-and-trap concentrator. Helium was bubbled through thewater sample and the vapor was swept through a sorbent trap (Vo-carb 3000). Naphthalene was liquid/liquid extracted with dichloro-methane and analyzed on a HP 6890 GC-Agilent 5973 N MSD.Concentrations were determined by comparison with standardsusing electronic integration.

3. Results and discussion

3.1. Calibration and sensitivity determination

PIB was found to be suitable for the detection of benzene, tolu-ene, ethylbenzene, o-xylene, m-xylene, p-xylene and naphthalenesince it has no significant spectral interferences in the region be-tween 800 and 650 cm�1. Although it has been revealed thatpoly(ethylene-co-propylene) may be more sensitive (Karlowatzet al., 2004), some studies suggest that these films are more sus-ceptible to delamination compared to PIB (Beardslee et al., 2010).We have shown previously that the hydrocarbon sensitivity ofthe PIB film does not change considerably with water exposuretime (Pejcic et al., 2011a), and this is an important issue to considerwith respect to continuous long term monitoring of hydrocarbonsin geochemical fluids. ATR–FTIR measurements were made inaqueous solutions comprising different levels of hydrocarbonsand Fig. 2 shows the infrared spectra of the PIB coated waveguidewhen exposed to the BTEX compounds. The spectra were collectedafter an enrichment time of 40 min and the IR absorption peaks forthe various hydrocarbons have distinct molecular signatures,

(a)

670680690700710720730740750

0.00

0.05

0.10

0.15

0.20

0.25 0 ppm

5 ppm10 ppm

26 ppm

52 ppm

(b)

0 ppm

6 ppm12 ppm

30 ppm

60 ppm

655660665670675680685690695

Wavenumber (cm-1) Wavenumber (cm-1)

Wavenumber (cm-1) Wavenumber (cm-1)

Wavenumber (cm-1) Wavenumber (cm-1)

0.00

0.05

0.10

0.15

0.20

Abso

rban

ce (A

U)

Abso

rban

ce (A

U)

Abso

rban

ce (A

U)

Abso

rban

ce (A

U)

Abso

rban

ce (A

U)

Abso

rban

ce (A

U)

710720730740750760770

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 0 ppm

9 ppm18 ppm

46 ppm

92 ppm

(d)

660680700720740760780800

0.0

0.1

0.2

0.3

0.4

0 ppm

8 ppm16 ppm

41 ppm

82 ppm

(e)

0 ppm

7 ppm13 ppm

33 ppm

66 ppm

760770780790800810820830

0.0

0.1

0.2

0.3

0.4

0.5

0.6 (f)

660680700720740760780800820

0.0

0.1

0.2

0.3

0.4

0.5

0 ppm

6 ppm12 ppm

31 ppm

61 ppm

(c)

Fig. 2. Infrared spectra of a PIB-coated ZnSe waveguide exposed to (a) benzene, (b) toluene, (c) ethylbenzene, (d) ortho-xylene, (e) meta-xylene, and (f) para-xylene atdifferent concentrations dissolved in deionized water. Note that the polymer film thickness was 8.4 ± 0.5 lm. All spectra were recorded after the PIB film had been exposed toeach solution for 40 min.

66 B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71

which correspond to C–H bending vibrations (out of plane bend-ing). The out of plane bending vibrations of the aromatic com-pounds are strong and characteristic of the number of hydrogensin the ring. The following band assignments were obtained fromthe single component experiments: benzene at 674 cm�1, tolueneat 727 cm�1, ethylbenzene at 696 cm�1, o-xylene at 741 cm�1,m-xylene at 767 cm�1, and p-xylene at 794 cm�1, which are consis-tent with values published elsewhere (Karlowatz et al., 2004).There is a slight overlap of the ethylbenzene peak at 696 cm�1 withtoluene (peak at 693 cm�1) and m-xylene (peak at 690 cm�1) sug-gesting that the presence of these compounds may interfere withthe determination of ethylbenzene. In addition, ethylbenzene hasabsorption bands present at 745, 771 and 788 cm�1, which mayinterfere with the peaks due to o-xylene, m-xylene and p-xylene.Although these ethylbenzene absorption bands are of much lowerintensity, they are likely to cause interference problems during thequantification of the xylene isomers.

The application of ATR–FTIR for determining PAH concentrationhas received much less attention in the literature and the analyti-cal performance of a PIB coated ZnSe waveguide was investigated.Fig. 3a and b shows the respective infrared spectra and diffusioncurves of the PIB coating when exposed to naphthalene at differentconcentrations. The characteristic absorption peak for naphthaleneoccurs at 781 cm�1 and this does not overlap with any of the BTEXcompounds. However, the diffusion of naphthalene was relativelyslow and after �40 min the enrichment process reached equilib-rium. Unsurprisingly, the response time varied with concentrationand the time taken to achieve a steady state was typically longer athigher concentrations. This is probably related to a polymer swell-ing process at high analyte levels which increases the free volumeand has been documented elsewhere (Pejcic et al., 2011a). In addi-tion, it took more than 60 min for the IR signal to return to baselineafter it had been exposed to 21 ppm naphthalene. By contrast, thediffusion process of the PIB film in benzene and toluene solutions

0 ppm

1 ppm

5 ppm

10 ppm

21 ppm

750760770780790800810

Wavenumber (cm-1)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Abso

rban

ce (A

U)

0

0.5

1

1.5

2

2.5

0 50 100 150 200 250

Time (min)

Peak

Are

a (A

U)

1 ppm

5 ppm

10 ppm

21 ppm

(a)

(b)

Fig. 3. (a) Infrared spectra of a PIB coated ZnSe waveguide exposed to naphthaleneat different concentrations dissolved in deionized water. All spectra were recordedafter the PIB film had been exposed to each solution for 40 min. (b) Transientresponse of the PIB-coated ZnSe waveguide when exposed to different concentra-tions of naphthalene dissolved in deionized water. Note that the polymer filmthickness was 8.4 ± 0.5 lm.

0

0.5

1

1.5

2

2.5

3

3.5

4

0 20 40 60 80 100

Concentration (ppm)

Inte

grat

ed A

bsor

banc

e (A

U) Benzene

TolueneEthylbenzenep-Xylenem-Xyleneo-XyleneNaphthalene

Fig. 4. ATR–FTIR calibration graphs of a PIB coated ZnSe waveguide exposed tovarious hydrocarbons dissolved in deionized water. Note that the polymer filmthickness was 8.4 ± 0.5 lm. All absorbance readings were plotted after the PIB filmhad been exposed to each solution for 40 min.

Table 1ATR–FTIR calibration data for various hydrocarbons studied over the 0–40 ppmconcentration range (with the exception of naphthalene which was between 0 and20 ppm). Note that the film thickness was determined gravimetrically to be8.4 ± 0.5 lm.

Molecule Wavenumber (cm�1) Response (AU/ppm) R2

Benzene 674 y = 0.0062x + 0.0021 0.9991Toluene 727 y = 0.0164x + 0.0047 0.9984Ethylbenzene 696 y = 0.018x + 0.0001 0.9994o-Xylene 741 y = 0.034x + 0.0095 0.9995m-Xylene 767 y = 0.0278x + 0.0012 0.9995p-Xylene 794 y = 0.0218x � 0.0065 0.9989Naphthalene 781 y = 0.0954x + 0.0055 0.9997

B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71 67

were much more rapid and the time taken to reach equilibriumwas <20 min (not shown). Generally, longer response times wereobserved for naphthalene, xylenes and ethylbenzene compared totoluene and benzene, and this is consistent with the diffusion coef-ficient being smaller for the larger molecules (Flavin et al., 2006).Reversibility studies revealed that the time needed to completelyremove the analyte from the PIB film during water washing wasmuch longer (typically > 30 min) and this depended on the typeof analyte and its concentration. The different equilibration timesfor the various hydrocarbon analytes can present problems whenevaluating the IR response in multi-component mixtures. To pro-vide maximum sensitivity and reliability, all calibration readingsand sensor measurements were recorded after the PIB film hadbeen exposed to solution for at least 40 min. It is suggested thatcomplete analyte enrichment is not necessary for signal evaluation,since the slope of the response curve after several minutes ofenrichment can be used (Mizaikoff et al., 1998).

It is evident that the intensity of the peaks in Figs. 2 and 3a in-crease with increasing concentration and this is consistent withthe hydrocarbon molecules partitioning from the water phaseinto the polymer film. Calibration curves were generated byintegrating the absorbance peaks and Fig. 4 shows a plot of thepeak area versus concentration of the PIB coating exposed tothe individual BTEXN compounds dissolved in deionized water.The absorbance increases linearly with increasing concentrationfor all hydrocarbons and this is in agreement with the Beer–Lambert law. In general, the PIB responded linearly over the0–80 ppm concentration range with a correlation coefficient (r2)

of greater than 0.97 for the BTEX analytes. However, the r2 valuewas much larger (r2 > 0.998) in the range between 0 and 40 ppm,and the calibration curves were used in this region for the quanti-fication of BTEX compounds in the multi-component hydrocarbonmixtures. For naphthalene quantification, the calibration curve inthe range of 0–20.5 ppm was used. Table 1 summarizes thecalibration data obtained on the change in peak area as a functionof concentration. Interestingly, the slope of the curve increaseswith hydrocarbon molecular weight and a possible reason for thismay be due to the much higher analyte amount in the PIB filmwith increasing hydrocarbon hydrophobicity. This is consistentwith recent reports which show that the hydrocarbon–polymerpartition process increases in the following order: benzene <toluene < xylenes � ethylbenzene < naphthalene (Pejcic et al.,2012). However, the observed relationship between the slopeand hydrocarbon molecular weight is complicated by factors otherthan just concentration changes in the PIB film that arise fromdifferences in the partition coefficient. Variations also occur inthe absorption extinction coefficient and the evanescent wavepenetration depth/sample pathlength which depend on the natureof the analyte and wavelength, respectively. These factors wouldneed to be considered before any reliable relationship betweensensitivity and molecular weight can be established.

3.2. Effect of PIB film thickness and molecular weight

Some of the factors that affect the partitioning behavior and thediffusion properties were studied and Fig. 5a shows the enrich-ment curves for toluene as a function of polymer film thickness.The coating thickness on the ZnSe waveguide was controlled byvarying the concentration of PIB in the solvent. It is evident that

0.000

0.003

0.006

0.009

0.012

0.015

Molecular weight (g/mol)

Sens

itivi

ty (A

U/p

pm)

0.000

0.020

0.040

0.060

0.080

0.100

0.120

85,000 400,000 1,912,000

0 5 10 15 20 25 30 35 40

Time (min)

Peak

Are

a (A

U)

4.5 µm 7.4 µm 8.4 µm(a)

(b)

Fig. 5. (a) The effect of film thickness on the enrichment curve for toluene into a PIBcoated ZnSe waveguide. Note that the toluene solution concentration was 5 ppm.(b) The effect of PIB molecular weight on the ATR–FTIR sensitivity for toluene. Theerror bars show the variation obtained on duplicate measurements.

68 B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71

the peak intensity arising from toluene increases with PIB thick-ness and this is indicative of an increase in the amount of toluenethat partitions into the film. The absorbance signal at equilibriumwas highest for the 8.4 lm film, whereas the peak intensity waslower and less reproducible at 4.5 lm. However, the time takenfor toluene to reach equilibrium was much longer for the 8.4 lm(�10–20 min) film compared to the 4.5 lm (�5–10 min). The in-creased response time with film thickness is consistent with tolu-ene taking longer to diffuse through the polymer membrane intothe region of the evanescent wave. Similar absorption and diffusionprofiles were observed for the other aromatic hydrocarbons. Basedon the results obtained in Fig. 5a, a film thickness of 8.4 lm waschosen for further studies since it displayed the most optimumanalytical performance in terms of sensitivity and reproducibility.Various reports recommend that the film thickness needs to beapproximately three times the infrared penetration depth to en-sure optimum performance (Pejcic et al., 2009b; Lin and Li, 2010).

Fig. 5b shows the variation in toluene sensitivity as a function ofPIB molecular weight. It is evident that the amount of toluene thatpartitions into the PIB film does not vary significantly with poly-mer molecular weight. However, it was found that the time forthe IR signal to reach equilibrium was slightly longer for the highermolecular weight polymer. Interestingly, the amount of water ab-sorbed by PIB was a factor of �3 lower for the high molecularweight (i.e., 1,912,000 g/mol) film compared to the two other films(not shown). This may be either due to PIB adhesion onto the ZnSesubstrate being greater for the high molecular weight film and/orthe water diffusion processes being slower. Despite some differ-ences in the rate of molecular diffusion, this study confirms thatthe amount of hydrocarbon absorbed is independent of the poly-

mer molecular weight and that this cannot be used to control thesensor sensitivity. However, previous studies have shown using apolyethylene coating that the ATR–FTIR sensitivity depends signif-icantly on the density of the polymer (Heinrich et al., 1990).

3.3. Analysis of total petroleum hydrocarbon solution

To test the applicability of the established method for an un-known sample, a crude oil–water mixture was prepared and thewater extract was analyzed using a 8.4 lm thick PIB coated ATR–FTIR sensor. Characterization of the oil–water mixture was alsoperformed using gas chromatography and the TPH was determinedto be 19.0 ± 0.6 ppm, noting that only the hydrocarbons present inthe water soluble fraction were of interest in this study. Fig. 6shows a typical chromatogram of the water dissolved fraction.More than 65% of the components in the TPH solution compriseC6 to C9 compounds (i.e., benzene, toluene, ethylbenzene, xylenes,alkylated benzenes, and methyl phenols). This is not surprisingconsidering that these compounds are generally more soluble/mis-cible in water compared to some of the other larger and less polarhydrocarbon molecules (i.e., >C10) that are present in the oil. Anumber of reports have found that the light aromatic and lightend hydrocarbon components are preferentially solubilized inwater (Lafargue and LeThiez, 1996; Faksness et al., 2004). Table 2displays the GC determined concentrations and it is evident thatthe BTEXN compounds are present at low ppm levels. ATR–FTIRmeasurements were made on the TPH solution and the IR spectracollected with time are shown in Fig. 7. The PIB was equilibratedin deionized water for at least 1 day prior to the analysis of theTPH solution. It is apparent that a number of hydrocarbons are ex-tracted into the PIB film and the peak intensities gradually increasewith time. After �40 min the extraction process reached equilib-rium for most of the hydrocarbon compounds. The evolution ofadditional IR peaks was also observed in the region between 800and 900 cm�1 and this is consistent with other hydrocarbons par-titioning from the TPH solution. It is not the purpose of the presentstudy to identify all hydrocarbon components by infrared spectros-copy; however, Fig. 7 illustrates the potential of using the ATR–FTIR sensor to quantify many different compounds other than justBTEXN. Compared to the standard GC technique (see Fig. 6), theATR–FTIR sensor method can easily distinguish between the meta-and para-xylene compounds.

The classical least squares and linear regression analysis in con-junction with a single analyte calibration in standard solutions wasused here for the quantitative analysis of BTEXN in the TPH solu-tion. Multivariate analytical methods such as principal componentanalysis (PCA) or artificial neural networks (ANNs) are normallyemployed to determine the hydrocarbon concentration in a mul-ti-component complex mixture (Silva et al., 2009). However, wefound that interferences and interaction effects arising from thepresence of many hydrocarbon components in the TPH solutiondo not drastically affect the quantification of BTEXN under ourexperimental conditions. Table 2 compares the actual concentra-tions determined using GC to those measured with the ATR–FTIRsensor, and it is evident that the sensor and GC results are almostidentical. The relative error (i.e., the difference between the ATR–FTIR determined concentration and the actual concentration ob-tained by GC) was generally <30% and the difference appears tobe partly due to interferences from other compounds present inthe TPH solution along with losses during solution storage/han-dling. The overestimation of ethylbenzene is most likely attributedto interference by toluene and/or m-xylene. However, the m-xylene appears to produce a much smaller interference effect rela-tive to toluene given that the peak overlap with ethylbenzene andthe m-xylene concentration in the TPH solution was much lower.The problem with overlapping peaks cannot be completely

mV

-2.000.002.004.006.008.00

10.0012.0014.0016.0018.0020.0022.0024.0026.0028.0030.0032.0034.0036.00

Minutes0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20

Benzene

Toluene

Ethylbenzene

m, p-Xylene

o-XyleneNaphthalene

MethylNaphthalenes

Time

Res

pons

e AlkylatedBenzenes

MethylPhenols

-2.000.002.004.006.008.00

10.0012.0014.0016.0018.0020.0022.0024.0026.0028.0030.0032.0034.0036.0038.00

1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00Time (Min)

Res

pons

e (m

V)

Fig. 6. Chromatogram of the dissolved North West Shelf crude oil in deionized water. An expanded chromatogram has been included in the inset.

Table 2Comparison between the actual hydrocarbon concentrations in the TPH solutiondetermined by GC and that measured using a PIB-coated ATR (film was 8.4 ± 0.5 lmthick). The results are the average of three separate measurements on different daysand using freshly prepared TPH solutions. The relative error was determined bycalculating the difference of the ATR–FTIR measured concentration and the GCobtained concentration and dividing the difference by the GC concentration value.

Molecule ATR-FTIR (ppm) GC-FID/GC–MS (ppm) Relative error (%)

Benzene 2.5 ± 0.6 2.7 ± 0.3 7.4Toluene 5.7 ± 0.5 4.8 ± 0.9 18.7Ethylbenzene 0.5 ± 0.2 0.4 ± 0.1 25.0o-Xylene 0.8 ± 0.2 0.8 ± 0.1 0.1m-Xylene 2.2 ± 0.2

2.0 ± 0.130.0

p-Xylene 0.4 ± 0.1Naphthalene 0.25 ± 0.05 0.25 ± 0.05 0.1

0 min

5 min

30 min

60 min

900 850 800 750 700 650

Wavenumber (cm-1)

Abs

orba

nce

(AU

)

Benzene

Ethylbenzene

Toluene

o-Xy

lene

p-Xy

lene

m-X

ylen

e

Nap

htha

lene

other hydrocarbons

Fig. 7. Infrared spectra of a PIB coated ZnSe waveguide exposed to the TPH solutionas a function of time. Note that the polymer film thickness was 8.4 ± 0.5 lm. Thepeak area at 40 min was integrated for each analyte and used to determine thecorresponding hydrocarbon concentration.

B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71 69

overcome, but the error associated with the quantification of eth-ylbenzene may be minimized by determining the relative sensitiv-ity factors for interfering peaks and fitting more than one peak tothe curve at 696 cm�1. In addition, the presence of other hydrocar-bon compounds in the TPH solution (i.e., 1,2,3-trimethylbenzene)can also interfere with the determination of m-xylene, consideringthat 1,2,3-trimethylbenzene has a characteristic IR absorption peakat�676 cm�1 which overlaps with m-xylene. Although the concen-tration of 1,2,3-trimethylbenzene was found by GC to be relativelylow (i.e., <0.2 ppm), it can still present a challenge when determin-ing m-xylene in complex geochemical samples. By contrast, Reganet al. (2003) observed that the infrared diffusion and enrichmentprofiles of certain hydrocarbon compounds are affected by thepresence of more than one component in the BTEX mixture. Itwas shown that the xylene isomers inhibit ethylbenzene fromentering a PVC based film (Regan et al., 2003). However, they per-formed studies on a plasticized PVC film in solutions that con-tained BTEX compounds at a concentration of 100 ppm.Notwithstanding, analytical errors are likely to occur for the PIBcoated ATR–FTIR sensor particularly when analyzing solutionsthat comprise some of these hydrocarbons at much higher

1000 950 900 850 800 750 700 650

-0.01

0.00

0.01

0.02

0.03

0.04

0.05

benz

ene@

676

tolu

ene@

690

tolu

ene@

727

o-xy

lene

@74

0

m-x

ylen

e@76

7

p-xy

lene

@79

5

Abs

orba

nce

(a.u

.)

Wavenumber (cm-1)

BTX mixture

Fig. 8. IR absorption signal obtained at planar silver halide waveguide sensor of aBTX sample mixture in aqueous solution after enrichment into an E/P-co layer(enrichment time: 60 min; concentration: 300 ppb (v/v) for each constituent).

70 B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71

concentrations. It is important to note that no corrections were ap-plied to the data presented in Table 2; however, the quantificationerrors were minimized by integrating over a narrow peak range.Although chemometrics was not employed in this study, it is pos-sible that multivariate statistical analysis may help reduce the ana-lytical errors and provide a better understanding of all thecomponents that could potentially interfere during BTEXNquantification.

In future, such sensing systems may be significantly reduced insize yet maintaining their analytical performance and molecularselectivity. This is shown for the example of a miniaturized IR sen-sor concept taking advantage of planar-tapered silver halide sens-ing elements serving as the active transducer coated with an E/P-comembrane. Since many of the hydrocarbons of interest are likely tobe present at very low concentrations in the geochemical fluids(i.e., ppb levels), the E/P-co coating would be more appropriatefor field measurements in terms of having a higher sensitivity com-pared to PIB. Fig. 8 shows the IR spectrum of a mixture containingbenzene, toluene, and the xylene isomers at a sample concentra-tion of 300 ppb (v/v), which was obtained after an enrichment timeof 60 min into the E/P-co membrane. The corresponding absorp-tion peaks have been labeled for clarity (benzene at 676 cm�1, tol-uene at 690 and 727 cm�1, o-xylene at 740 cm�1, m-xylene at767 cm�1, p-xylene at 795 cm�1) (Karlowatz et al., 2004). Similarto the measurements using the conventional ZnSe waveguide, eachanalyte remains clearly discernible via distinctive absorption fea-tures using this prototype miniaturized IR sensing setup. Furtherreduction of the system dimensions will be achieved in future byreplacing the FTIR spectrometer with broadly tuneable quantumcascade lasers covering the spectral range of interest (Younget al., 2009).

4. Conclusions

This study has shown that the ATR–FTIR sensor may be used todistinguish and quantify benzene, toluene, ethylbenzene, xyleneisomers and naphthalene in a complex oil–water mixture. Theamount of hydrocarbon extracted into the PIB film was propor-tional to the concentration in solution and linear relationships(r2 > 0.998) between absorption peak areas and concentration wereobtained in the range 0–40 ppm for BTEX and 0–20 ppm for naph-thalene. A simple calibration method using the classical least

squares and linear regression analysis was used to perform quan-titative analysis of BTEXN in the multi-component mixture. Itwas found that the hydrocarbon–polymer partition process forthe single analytes was independent of the amount and type ofhydrocarbons present in the aqueous solution. Further studiesare being planned to determine if the sensor can distinguish be-tween different sources and types of crude oil in water. In addition,experiments are being planned to optimize the response (i.e., timeand sensitivity) and conditions necessary for directly assaying realsamples from the field. It is expected that this work will lead to thedevelopment of a miniaturized field deployable device for moni-toring hydrocarbons in geological formations (i.e., oil reservoirsand produced water). Potential miniaturization of such sensingconcepts is demonstrated here with a prototype sensing systemwhich replaces the ATR crystal with a planar-tapered silver halidewaveguide achieving comparable sensitivity.

Acknowledgments

We greatly acknowledge the funding provided by the CSIROWealth from Oceans Flagship and the capability development fundat CESRE. RL and BM acknowledge support of parts of this study bythe China Scholarship Council (2011634033) and by the Young Sci-entist Exchange Program (YSEP) within the BMBF–MOST German–Israeli Water Technology Research Initiative (Project #YSEP63). Weare grateful to the three anonymous reviewers for constructivecomments which helped improve the quality of the manuscript.

Associate Editor–Simon George

References

Acha, V., Meurens, M., Naveau, H., Agathos, S.N., 2000. ATR–FTIR sensordevelopment for continuous on-line monitoring of chlorinated aliphatichydrocarbons in a fixed-bed bioreactor. Biotechnology and Bioengineering 68,473–487.

Alcudia-Leon, M.C., Lucena, R., Cardenas, S., Valcarcel, M., 2008. Characterization ofan attenuated total reflection-based sensor for integrated solid-phaseextraction and infrared detection. Analytical Chemistry 80, 1146–1151.

Audino, M., Grice, K., Alexander, R., Kagi, R., 2004. Macrocyclic alkanes in crude oilsand sediment extracts: enrichment using molecular sieves. OrganicGeochemistry 35, 661–663.

Ballesteros-Gomez, A., Rubio, S., 2011. Recent advances in environmental analysis.Analytical Chemistry 83, 4579–4613.

Barman Skaare, B., Wilkes, H., Vieth, A., Rein, E., Barth, T., 2007. Alteration of crudeoils from the Troll area by biodegradation: analysis of oil and water samples.Organic Geochemistry 38, 1865–1883.

Beardslee, L.A., Demirci, K.S., Luzinova, Y., Mizaikoff, B., Heinrich, S.M., Josse, F.,Brand, O., 2010. Liquid-phase chemical sensing using lateral mode resonantcantilevers. Analytical Chemistry 82, 7542–7549.

Bryant, C.K., LaPuma, P.T., Hook, G.L., Houser, E.J., 2007. Chemical agentidentification by field-based attenuated total reflectance infrared detectionand solid-phase microextraction. Analytical Chemistry 79, 2334–2340.

Charlton, C., Giovannini, M., Faist, J., Mizaikoff, B., 2006. Fabrication andcharacterization of molecular beam epitaxy grown thin-film GaAs waveguidesfor mid-infrared evanescent field chemical sensing. Analytical Chemistry 78,4224–4227.

Demeestere, K., Dewulf, J., De Witte, B., Van Langenhove, H., 2007. Samplepreparation for the analysis of volatile organic compounds in air and watermatrices. Journal of Chromatography A 1153, 130–144.

Faksness, L.G., Grini, P.G., Daling, P.S., 2004. Partitioning of semi-soluble organiccompounds between the water phase and oil droplets in produced water.Marine Pollution Bulletin 48, 731–742.

Flavin, K., Hughes, H., Dobbyn, V., Kirwan, P., Murphy, K., Steiner, H., Mizaikoff, B.,McLoughlin, P., 2006. A comparison of polymeric materials as pre-concentratingmedia for use with ATR/FTIR sensing. International Journal of EnvironmentalAnalytical Chemistry 86, 401–415.

Gonzalvez, A., Garrigues, S., de la Guardia, M., Armenta, S., 2011. The ways to thetrace level analysis in infrared spectroscopy. Analytical Methods 3, 43–52.

Heinrich, P., Wyzgol, R., Schrader, B., Hatzilazaru, A., Lubbers, D.W., 1990.Determination of organic compounds by IR/ATR spectroscopy with polymer-coated internal reflection elements. Applied Spectroscopy 44, 1641–1646.

Karlowatz, M., Kraft, M., Mizaikoff, B., 2004. Simultaneous quantitativedetermination of benzene, toluene, and xylenes in water using mid-infraredevanescent field spectroscopy. Analytical Chemistry 76, 2643–2648.

B. Pejcic et al. / Organic Geochemistry 55 (2013) 63–71 71

Kim, S.S., Young, C., Mizaikoff, B., 2008. Miniaturized mid-infrared sensortechnologies. Analytical and Bioanalytical Chemistry 390, 231–237.

Kraft, M., Jakusch, M., Karlowatz, M., Katzir, A., Mizaikoff, B., 2003. New frontiers formid-infrared sensors: towards deep sea monitoring with a submarine FT-IRsensor system. Applied Spectroscopy 57, 591–599.

Kraft, M., Mizaikoff, B., 2000. A mid-infrared sensor for monitoring of chlorinatedhydrocarbons in the marine environment. International Journal ofEnvironmental Analytical Chemistry 78, 367–383.

Kvenvolden, K.A., Cooper, C.K., 2003. Natural seepage of crude oil into the marineenvironment. Geo-Marine Letters 23, 140–146.

Lafargue, E., LeThiez, P., 1996. Effect of waterwashing on light ends compositionalheterogeneity. Organic Geochemistry 24, 1141–1150.

Lawrence, N.S., 2006. Analytical detection methodologies for methane and relatedhydrocarbons. Talanta 69, 385–392.

Lieberzeit, P.A., Dickert, F.L., 2007. Sensor technology and its application inenvironmental analysis. Analytical & Bioanalytical Chemistry 387, 237–247.

Lima, K.M.G., Raimundo Jr., I.M., Pimentel, M.F., 2011. Simultaneous determinationof BTX and total hydrocarbons in water employing near infrared spectroscopyand multivariate calibration. Sensors and Actuators B – Chemical 160, 691–697.

Lin, W.D., Li, Z.J., 2010. Detection and quantification of trace organic contaminantsin water using the FT-IR–attenuated total reflectance technique. AnalyticalChemistry 82, 505–515.

Luzinova, Y., Dobbs, G.T., Sassen, R., Mizaikoff, B., 2009. Quantification ofadamantane in organic media via infrared attenuated total reflectionspectroscopy. Organic Geochemistry 40, 1143–1150.

Luzinova, Y., Zdyrko, B., Luzinov, I., Mizaikoff, B., 2012a. Detecting trace amounts ofwater in hydrocarbon matrices with infrared fiberoptic evanescent fieldsensors. Analyst 137, 333–341.

Luzinova, Y., Zdyrko, B., Luzinov, I., Mizaikoff, B., 2012b. In situ trace analysis of oilin water with mid-infrared fiberoptic chemical sensors. Analytical Chemistry84, 1274–1280.

Mizaikoff, B., 1999. Mid-infrared evanescent wave sensors – a novel approach forsubsea monitoring. Measurement Science & Technology 10, 1185–1194.

Mizaikoff, B., 2003. Mid-IR fiber-optic sensors. Analytical Chemistry 75, 258A–267A.Mizaikoff, B., Kraft, M., Jakusch, M., 1998. Water monitoring using infrared fiber

optic sensors. In: Oceans Conference Record (IEEE), 3, Proceedings of the 1998Oceans Conference. Part 1(of 3), pp. 1401–1404. IEEE, Nice, Fr.

Murphy, B., McLoughlin, P., 2003. Determination of chlorinated hydrocarbonspecies in aqueous solution using Teflon coated ATR waveguide/FTIRspectroscopy. International Journal of Environmental Analytical Chemistry 83,653–662.

Pejcic, B., Crooke, E., Boyd, L., Doherty, C.M., Hill, A.J., Myers, M.B., White, C., 2012.Using plasticizers to control the hydrocarbon selectivity of a poly(methylmethacrylate)-coated quartz crystal microbalance sensor. Analytical Chemistry84, 8564–8570.

Pejcic, B., Crooke, E., Doherty, C.M., Hill, A.J., Myers, M., Qi, X.B., Ross, A., 2011a. Theimpact of water and hydrocarbon concentration on the sensitivity of a polymer-based quartz crystal microbalance sensor for organic compounds. AnalyticaChimica Acta 703, 70–79.

Pejcic, B., Myers, M., Crooke, E., Ross, A., Baker, M., 2009a. Improvements to ATR–FTIR based chemical sensors for the detection of organic contaminants

dissolved in water. Proceedings of IEEE Sensors, 299–303. Article number5398158.

Pejcic, B., Myers, M., Ranwala, N., Boyd, L., Baker, M., Ross, A., 2011b. Modifying theresponse of a polymer-based quartz crystal microbalance hydrocarbon sensorwith functionalized carbon nanotubes. Talanta 85, 1648–1657.

Pejcic, B., Myers, M., Ross, A., 2009b. Mid-infrared sensing of organic pollutants inaqueous environments. Sensors 9, 6232–6253.

Qi, X.B., Crooke, E., Ross, A., Bastow, T.P., Stalvies, C., 2011. Revealing the propertiesof oils from their dissolved hydrocarbon compounds in water with anintegrated sensor array system. Analyst 136, 3731–3738.

Readman, J.W., Fowler, S.W., Villeneuve, J.P., Cattini, C., Oregioni, B., Mee, L.D., 1992.Oil and combustion-product contamination of the Gulf marine environmentfollowing the war. Nature 358, 662–665.

Regan, F., Walsh, F., Walsh, J., 2003. Development of plasticised PVC sensing filmsfor the determination of BTEX compounds in aqueous samples. InternationalJournal of Environmental Analytical Chemistry 83, 621–631.

Rodgers, R.P., McKenna, A.M., 2011. Petroleum analysis. Analytical Chemistry 83,4665–4687.

Roy, G., Mielczarski, J.A., 2002. Infrared detection of chlorinated hydrocarbons inwater at ppb levels of concentrations. Water Research 36, 1902–1908.

Saar, A., Moser, F., Akselrod, S., Katzir, A., 1986. Infrared optical properties ofpolycrystalline silver halide fibers. Applied Physics Letters 49, 305–307.

Silva, A.M.S., Pimentel, M.F., Raimundo, I.M., Almeida, Y.M.B., 2009. Effect ofplasticizers on a PVC sensing phase for evaluation of water contamination byaromatic hydrocarbons and fuels using infrared spectroscopy. Sensors andActuators B – Chemical 139, 222–230.

Tissot, B.H., Welte, D.H., 1984. Petroleum Formation and Occurrence. Springer-Verlag, Berlin.

Tsvetnenko, Y., Evans, L., 2002. Improved approaches to ecotoxicity testing ofpetroleum products. Marine Pollution Bulletin 45, 148–156.

Ueyama, S., Hijikata, K., Hirotsuji, J., 2002. Water monitoring system for oilcontamination using polymer-coated quartz crystal microbalance chemicalsensor. Water Science & Technology 45, 175–180.

Wang, X., Kim, S.-S., Rossbach, R., Jetter, M., Michler, P., Mizaikoff, B., 2012. Ultra-sensitive mid-infrared evanescent field sensors combining thin-film stripwaveguides with quantum cascade lasers. Analyst 137, 2322–2327.

Yang, J., Cheng, M.L., 2001. Development of an SPME/ATR–IR chemical sensor fordetection of phenol type compounds in aqueous solutions. Analyst 126, 881–886.

Yang, J., Ramesh, A., 2005. Membrane-introduced infrared spectroscopic chemicalsensing method for the detection of volatile organic compounds in aqueoussolutions. Analyst 130, 397–403.

Young, C., Kim, S.S., Luzinova, Y., Weida, M., Arnone, D., Takeuchi, E., Day, T.,Mizaikoff, B., 2009. External cavity widely tunable quantum cascade laser basedhollow waveguide gas sensors for multianalyte detection. Sensors andActuators B – Chemical 140, 24–28.

Young, C.R., Menegazzo, N., Riley, A.E., Brons, C.H., DiSanzo, F.P., Givens, J.L., Martin,J.L., Disko, M.M., Mizaikoff, B., 2011. Infrared hollow waveguide sensors forsimultaneous gas phase detection of benzene, toluene, and xylenes in fieldenvironments. Analytical Chemistry 83, 6141–6147.