characterization of incense smoke by solid phase microextraction—comprehensive two-dimensional gas...

ARTICLE IN PRESS

1352-2310/$ - se

doi:10.1016/j.at

�Correspondfax: +613 9925

E-mail addr

Atmospheric Environment 41 (2007) 5756–5768

www.elsevier.com/locate/atmosenv

Characterization of incense smoke by solid phasemicroextraction—Comprehensive two-dimensional

gas chromatography (GC�GC)

Tin C. Tran, Philip J. Marriott�

Australian Centre for Research on Separation Science, School of Applied Sciences, RMIT University,

GPO Box 2476V, Melbourne 3001, Australia

Received 20 November 2006; received in revised form 21 February 2007; accepted 22 February 2007

Abstract

Comprehensive two-dimensional gas chromatography in tandem with flame ionization detection (GC�GC-FID) was used

for the qualitative fingerprint characterisation of four different types of powdered incense headspace (H/S), and incense smoke.

Volatile organic compounds (VOCs) in the incense powder and smoke were extracted by using solid phase microextraction

(SPME) with a polydimethylsiloxane/divinylbenzene (PDMS/DVB) 65mm fiber. Low-polarity/polar, and polar/non-polar

phase combinations were tested to contrast the GC�GC separation of components in these two column sets.

A total of 324 compounds were tentatively identified, with more than 100 compounds in incense powders and more than

200 compounds in the incense smoke, by using GC coupled to quadrupole mass spectrometric detection. Identification

required at least 90% match with the NIST library; otherwise they were considered as unidentified. The smoke stream

comprised compounds originating from the incense powder, and combustion products such as PAH, N-heterocyclics, and

furans. However, GC�GC was able to separate many more volatile compounds (possibly hundreds more) present in the

complex smoke samples, many of which cannot be separated by conventional 1D-GC; this is a direct consequence of the

high-resolution power of GC�GC. GC�GC fingerprint comparison of powder H/S with smoke allows facile subtraction

of the former from the latter to assist identification of compounds generated from burning incense.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Incense; Smoke; Headspace; Solid phase microextraction; Comprehensive two-dimensional gas chromatography

1. Introduction

Incense is used to mask odour, for aromatherapy,and plays an important role in many religionsaround the world, especially in eastern religions.The Hindu, Buddhist, Taoist and Shinto religions

e front matter r 2007 Elsevier Ltd. All rights reserved

mosenv.2007.02.030

ing author. Tel.: +61 3 99252632;

3747.

ess: [email protected] (P.J. Marriott).

all burn incense in festivals, processions and manydaily rituals, such as paying respect to ancestors andvarious gods.

There are two broad types of incense. Westernincenses are normally burned to produce pleasantfragrance inside home, shopping centres, shops andpublic places. These incenses are largely made fromthe gum resins in tree bark (such as frankincense),aromatic woods, flowers, essential oils and syntheticsubstitute perfume chemicals. Eastern incenses are

.

www.elsevier.com/locate/atmosenv

dx.doi.org/10.1016/j.atmosenv.2007.02.030

mailto:[email protected]

ARTICLE IN PRESST.C. Tran, P.J. Marriott / Atmospheric Environment 41 (2007) 5756–5768 5757

processed from plants such as sandalwood, patch-ouli, agarwood and vetiver, and are used for cere-monial practices. Incenses are available in variousforms including joss stick, cones, coils and rocks.

Incense burning is a long, slow, incompletecombustion process, resulting in the generation ofa continuous smoke stream, which is an importantsource of indoor air pollution due the emission ofPM10 and PM2.5 (particulate matter less than 10 and2.5 mm in aerodynamic diameter), carbon monoxide(CO) and volatile organic compounds (VOCs)(Lee and Wang, 2004). These researchers measuredthe emission rates of 10 different incense typesvarying in combustible volume from 0.29 to 2.8 cm3

and found that emission rates for PM2.5 and PM10

ranged between 9.8–2160.3 and 10.8–2536.6mg h�1,respectively (emission rates did not correlate withincense volume; burn times varied from 25 to50min). Lung et al. (2003) showed a significantcontribution from incense burning to indoor PM10

and particulate polycyclic aromatic hydrocarbon(PAH), which comprises a large group of organiccontaminants generated by incomplete combustionof organic material. One incense stick usually lastsabout 45–60min depending on the length andtightness of the powder packing. The greater themass of material burned, the greater the particulatemass that is generated (Mannix et al., 1996). Otherstudies indicate that the particulate burden fromburning one incense stick is 1.2–2.9 times of that ofburning one cigarette (Chen and Lee, 1996; Mannixet al., 1996). Lung and Hu (2003) showed thatincense burning produced about 28.3–30.5 mg PAHsper stick.

The aerosol particulate matter emitted fromincense burning was found to be mutagenic in theAmes Salmonella test (Lofroth et al., 1991). Incensesticks were found to contain compounds causingallergic contact dermatitis and photosensitization,such as musk ambrette, musk ketone and muskxylene (Roveri et al., 1998).

Solid phase microextraction (SPME) is a solvent-free sampling technique, introduced by Pawliszyn inthe 1990s, and is a popular alternative samplepreparation and extraction method to classicalextraction methods. The transport of analytes fromthe matrix into the fiber-supported polymer coatingcommences when the fiber is placed in contact withthe sample. In the headspace (H/S) mode, theanalytes are transported through the air before theyare sorbed into or onto the coating. SPME has beensuccessfully applied in many applications, including

wound-induced VOCs from plants (Perera et al.,2002), analysis of PAHs in atmospheric parti-culates (Vaz, 2003), volatile terpenes in olibanum(Hamm et al., 2005), aroma compounds in palmsugar (Ho et al., 2006) and methoxypyrazines inwine (Ryan et al., 2005).

The hyphenated GC technique of comprehensivetwo-dimensional gas chromatography (GC�GC)was reported in 1991 by Phillips (Liu and Phillips,1991; Phillips and Beens, 1999), comprising twoserially-coupled GC columns containing differentstationary phases with a modulator near theirjunction. The modulator serves to periodically andrapidly sample effluent from the first column, as aseries of concentrated zones, which are thenreinjected as narrow peaks into the second column.The second column generates a chromatogram of afew seconds (s) duration. Therefore, the entiresample is submitted to two ‘‘orthogonal’’ separa-tions. The process results in significantly higherpeak capacity with substances that co-elute on theprimary column ideally resolved on the secondarycolumn. GC�GC has been successfully applied toa wide range of samples including sedimentarypetroleum hydrocarbons (Reddy et al., 2002),volatile compounds in strawberry (Williams et al.,2005), natural fats and oils (Mondello et al., 2003),and urban aerosols (Kallio et al., 2003).

Much previous work on incense smokes em-ployed conventional extraction and GC methods;no reports of analysis of incense smoke have usedSPME with GC�GC. Previous studies mainlyfocused on the emission rates of the smoke fromincense burning, however, the compounds that werereleased from the smoke were not well characterized(and chromatograms often not shown in publishedliterature). Therefore, it is important to develop ananalytical method that can be used for the chemicalcharacterization of compounds present in theparticulate phase especially where the complexityis significant. Notwithstanding the quantitationlimitations of SPME, the present study seeks toexamine the feasibility of using SPME-GC�GC forsampling and identifying volatile compoundsemitted by incense powder and in the smoke stream,and to generate a fingerprint chromatogram fordifferent types of incense. Two different SPMEextraction setups were used for the analysis ofincense smoke—direct from the smoke stream, andthe side stream. Different GC�GC column confi-gurations were studied to contrast the separation ofcompounds.

ARTICLE IN PRESST.C. Tran, P.J. Marriott / Atmospheric Environment 41 (2007) 5756–57685758

2. Materials and methods

2.1. Samples

Four different types of incense in various forms,including sticks and cones, were used in this study.Incenses were chosen based on their different uses,and all differed in appearance, such as length, mass,colour and fragrance; lotus aromatic incense (Europe)was claimed to have aromatherapy use; Tibetan redincense (Hong Kong) was for ceremonial practices fordeity worshipping; brown smokeless incense (China)was claimed to be environmentally friendly, produ-cing less smoke with reduced air pollution; medicineherb incense (Vietnam) was promoted as effec-tive, non-toxic and non-addictive antidote for stressor depression. All incenses were purchased fromlocal gift shops and shopping centres (Melbourne,Australia).

2.2. SPME

2.2.1. Crushed incense

One stick of each incense sample was powderedand transferred to a 4mL glass vial with a screw capand PTFE/silicone septum, in order to sample thefragrance/essential oil H/S. Extraction was per-formed manually with an SPME holder (SupelcoInc., Bellefonte, PA, USA). Prior to extraction, thepolydimethylsiloxane/divinylbenzene (PDMS/DVB)coated fiber (65 mm; Stableflex, Supelco Inc.), wasconditioned according to the manufacturer’s in-struction by heating in the GC injection port at250 1C for 30min. The PDMS/DVB fiber waschosen because it is suitable for extraction of abroad range of volatile compounds, with bothabsorption by the PDMS polymer, and adsorptionby DVB particles (Pawliszyn, 1997). Samplingparameters for H/S SPME analysis included effectof extraction time (10–120min), and extractiontemperature (room temperature �100 1C) for eachsample. Extraction at room temperature for 60minwas found to adequately trap volatile compoundsonto the fiber; extraction for longer time at hightemperature may result in overloading the capillarycolumn for the major component(s). Therefore, H/Sextraction was performed by exposing the 65 mmPDMS/DVB fibers inside the 4mL glass vial, 1 cmabove the sample for 60min at room temperature.The fiber was then transferred to the GC, andthermally desorbed for 3.5min into the glass liner ofthe GC injector at 250 1C. Fiber blanks were run

between each sample injection to ensure no con-tamination or carry over from previous samples.Duplicate runs of each analysis were carried out.Good repeatability and reproducibility for SPMEextraction were obtained with RSDs of 2.22% and3.05%, respectively, based on the total area of themost intense peaks from the chromatograms.Reproducibility of retention time of the mostintense peaks gave RSD of less than 0.50%.

2.2.2. Incense smoke

Sampling of smoke volatiles emitted from burningincense was performed in two ways. Fig. 1a showsthe set up for extraction of smoke volatiles, where theSPME fiber is directly exposed to the smoke streamfrom the incense stick burning inside the invertedglass funnel. This experiment allows sorption ofsmoke volatiles and also potentially particulates fromthe smoke onto the fiber. The fiber turned a brownishcolour over time with increased sampling events.Sampling the smoke stream over the duration ofburning of a whole stick, half a stick, and finally for20min showed that 20min sampling time wassufficient. Therefore, for both side stream and directsmoke, the SPME fiber was exposed for 20min.

Fig. 1b shows the SPME set up for side streamextraction, intended to collect only the aroma andvolatile compounds from the incense smoke withoutcollection of particulates onto the fiber. A T-piecewas attached to the glass funnel, with the incenseburnt in the bottom portion of the inverted funnel,and the SPME fiber was inserted into the side armof the T-piece. The mainstream smoke was ventedthrough the funnel neck, and volatile compoundsdiffuse into the side arm of the T-piece for SPMEsampling. The side arm was sealed to avoid backflush from outside air, and to prevent free smokepassage into the side arm. Visually, no smoke wasseen in the arm, although this is not evidence thatno particulates enter this region. A blank extractionwas performed to test for any laboratory environ-ment residual compounds that might lead toextraneous peaks in the GC. The fiber was exposedusing the same apparatus and in the same mode for60min, but without incense, to collect backgroundair sample. No compounds of note were found inthe laboratory environment.

2.3. GC-flame ionization detection (FID)

A Shimadzu GC-17A GC (Shimadzu, Rydalmere,Australia) configured with a split/splitless injector,

ARTICLE IN PRESS

SPME

holder

Joss stick

Smoke

Exposed fibre to collect

volatiles and particulates

Glass funnel

Clamps

Stand

Support

Clamps

Stand

SPMEExposed fibre to

collect volatiles

Joss stick

Support

Glass funnel

SmokeSeal to prevent

back flush

Fig. 1. Diagram of SPME set up for sampling of smoke by (a) direct sampling, and (b) side stream sampling.

T.C. Tran, P.J. Marriott / Atmospheric Environment 41 (2007) 5756–5768 5759

and FID was used for conventional 1D-GC analyses.Splitless injections were performed by introducingthe SPME fiber into the hot injection port (250 1C).Hydrogen carrier gas was set at a constant flow of1.5mLmin�1, with a non-polar (5% phenyl poly-silphenylene siloxane) phase (BPX5) column (30m�250mm id � 0.25mm df (film thickness); SGE Inter-national, Ringwood, Australia). The GC tempera-ture programme commenced at 40 1C (2min),heated to 130 1C at 25 1Cmin�1, then to 260 1C at15 1Cmin�1 (10min).

2.4. GC-quadrupole mass spectrometry (qMS)

GC-qMS was performed using an Agilent model6890 GC (Agilent Technologies, Burwood, Austra-lia) configured with a split/splitless injector, coupledto an Agilent model HP5973 Mass Selective Detector(qMS) through a heated transfer line (250 1C). ABPX5 capillary column (30m� 250mm id � 0.25mmdf) and helium carrier gas (0.7mLmin�1) was used.Manual splitless injections at 250 1C were made byexposing the SPME fiber, using a narrow SPME inletliner. The GC temperature was initially set at 40 1C(2min), then heated to 280 1C at 5 1Cmin�1. TheqMS source temperature was 230 1C, with a massscan range of 40–340u, and acquisition rate of8.62 scans s�1. Generally, matched spectra with the

NIST library were only considered if the matchmetric was 490% according to the MS searchcriteria.

2.5. GC�GC-FID

GC�GC was performed using an Agilent 6890GC under Chemstation software control and dataacquisition, and was configured for GC�GC byfitting a longitudinally-modulated cryogenic system(LMCS; Chromatography Concepts, Doncaster,Australia). Two different column sets were used inthis study. Column set 1: the first dimension column(1D) comprised a low-polarity phase (BPX5; 30m�250 mm id � 0.25 mm df), coupled to a polar(polyethylene glycol) phase-coated capillary (BP20;1.0m� 100 mm id � 0.10 mm df) for the seconddimension (2D). Column set 2: a polar 1D column(50% phenyl polysilphenylene-siloxane) phase(BPX50; 30m� 250 mm id � 0.25 mm df) wascoupled to a non-polar 2D column (100% di-methyl-siloxane) phase (BP1; 1m� 100 mm i.d.�0.10 mm df). All columns were supplied by SGEInternational. Hydrogen carrier gas was used at1.5mLmin�1. Manual SPME injections were made,in splitless mode. The injector temperature was250 1C with FID temperature at 250 1C. The GCoven programme was from 40 1C (2min), heated to


260 1C at 5 1Cmin�1 for column set 1, and at3 1Cmin�1 for column set 2. Modulator cryotraptemperature was set at �20 1C, commencing at10min for column set 1 and 4min for column set 2.The modulation period (PM) was 6 s for column set1 and 5 s for column set 2.

3. Results

The first analytical step in this work wasconventional 1D-GC analysis of incense powderH/S and smoke stream volatiles, with a low-polaritycolumn (BPX5). Due to the large number ofanalytes trapped on the fiber, or the elevatedconcentrations of some compounds trapped on thefiber, the analytical column may either haveinsufficient resolving power, or may be easilyoverloaded by larger abundance peaks resulting inserious peak asymmetry. In the latter case, peakswith lower concentration may be merged in the tailof an intense peak. Study of extraction time andextraction temperature assisted in selection ofadequate SPME conditions to avoid overloadingof the column. All extractions were therefore carriedout for a pre-set time (most likely not at equili-brium) with extraction conditions described in theexperimental section. This study was intended toprovide qualitative comparisons of the composi-tions of the samples investigated. Side sampling ofincense smoke as described above was conducted,however, the results obtained were very similar tothose obtained for the direct smoke sampling.Results obtained for the side sampling experimentwill not be separately discussed here.

3.1. Qualitative comparison of GC-FID with

GC�GC-FID

Fig. 2a and 2b chromatograms illustrate thecomplexity of both powder H/S volatiles and smokevolatile composition of orchid-scented incensepowder and smoke, respectively. The latter demon-strates perhaps hundreds of overlapping peaks in1D-GC. A few peaks have been asterisked toindicate they co-exist in each trace (based on tR(retention time) similarity). The added complexityof the smoke sample reduces the likelihood that anyof these peaks will be single, pure peaks. Theirrelative intensities are different for each sampleshowing that specific volatile compounds in bothpowder H/S and smoke may differ in relativeamounts. The chromatogram of smoke (Fig. 2b)

shows that many additional compounds of lowermolar mass (eluting at lower tR) arise from thecombustion process. Fig. 2c by contrast shows theGC�GC contour plot of the SPME extract ofsmoke volatiles; the significant increase in resolutionreveals many more peaks. Separations in GC�GCwere achieved by using two stationary phases withdifferent polarity, thus compounds not separated onthe first column are likely to be separated on theother. Here, the first column is a polar stationaryphase, providing a polarity/boiling point-basedseparation, with the second column a non-polarstationary phase. Thus a boiling point-based se-paration mechanism defines the separation alongthe second axis. Each compound forms a zone onthe plane defined by retention times on the twocolumns. It is clear that many peaks in the 1D resultwill be made up of more than a single component.

3.2. Identification of components of incense powder

and smoke via SPME GC-qMS

A total of 324 compounds were tentativelyidentified using GC-qMS. A nominal total of 123,139, 152 and 118 compounds for incense powderH/S extraction, and 220, 226, 218 and 189compounds in the smoke, were detected for lotus-scented, red Tibetan, medicine herb and brownsmokeless incense respectively, using conventionalGC-qMS. These component numbers are ratherarbitrary given that the composition of overlappedpeaks is not readily quantified. However, onlyhalf—or even less—of the compounds were tenta-tively identified by matching peak apex mass spectrawith mass spectra in the library. Identified com-pounds had matching quality of at least 90%between the mass spectra of the compoundsobtained from the samples to the mass spectra fromthe NIST library; compounds with matching qualityless than this were recorded as unidentified. Theinability to identify compounds is largely ascribedto their overlapping with other compounds, leadingto poor quality spectra for those compounds, whichreduces the match quality with the MS library. Also,compounds at trace level have inadequate sensitiv-ity, with the MS detector giving poor signal-to-noiseratio (S/N) and poor library matching.

A wide range of different compound classes wereidentified in the H/S of incense powder and in thesmoke, including essential oil-type components,aldehydes, alcohols, pyrazines, ketones, pyrans,acids, mono aromatics and PAHs. An abbreviated

ARTICLE IN PRESS

*

**

*

*

*

*

* *

**

*

*

**

*

*

*

*

* *

**

*

*

**

*

*

*

*

* *

**

*

*

**

*

*

*

*

* *

**

*

2D

Rete

ntion tim

e (

s)

0 10 20 30

1D Retention time (min)1D

40 50 60 70

4.0

3.0

2.0

1.0

10 20 30 40 50 60 70

Retention time (min)

Fig. 2. GC-FID and GC�GC-FID chromatograms of crushed orchid-scented incense after SPME extraction; separation on column set 2

(polar–non-polar) (a) 1D-GC of powder extracted at ambient temperature for 60min (b) 1D-GC smoke and (c) GC�GC of smoke (b and

c extracted for 20min).


list of peaks assignments can be found in Table 1(refer to Supplementary Information for a completetabulated data set), along with an indication of theirpresence or absence in the powder H/S, and in thesmoke. Classes of compounds such as hydrocar-bons, carbonyl, hydroxy and phenols were found inboth H/S and smoke in all incense types. Supple-mentary Data should be referred to for furtherinformation.

In the powder H/S extraction, allergenic nitro-musk compounds such as musk ketone (3,5-dinitro-2,6-dimethyl-4-tert-butylacetophenone) were foundto be present in lotus-scented, red Tibetan andmedicine incenses and musk xylene (2,4,6-trinitro-3,5-dimethyl-1-tert-butylbenzene) was found in allincenses, which is in agreement with previous results(Roveri et al., 1998). Other compounds, such asbenzaldehyde, phenol, benzyl alcohol, phenylethyl

alcohol, 2H-1-benzopyran-2-one, diethyl phthalate,benzyl benzoate, cinnamaldehyde and vanillin(4-hydroxy-3-methoxy benzaldehyde) were foundin all incense powders. Vanillin has been previouslyreported (Nevell and Zeronian, 1985) as a degrada-tion product of lignin, which has an oil-fruity andfloral citrus smell. Compounds that are usuallypresent in essential oils such as terpenes, linalool,eugenol, copaene, thujopsene, lilial, patchoulene,methyl ionone and guinene were also detected inincense powders. This is logical since they are burntto produce aromas, arising from the essential oilsthat are added as an ingredient in the incensemanufacture. Different essential oil compoundswere found in different incense types, according tothe specific ingredients which were used to make theincense to be used for different purposes. Thus eachGC trace is rather specific to each of the incenses.

ARTICLE IN PRESS

Table 1

Abbreviated table of indicative compounds identified in the powder (H/S) and smoke (S) of four different incense types by using 1D

GCMS

No. Compounds Incense types

Lotus-scented Red Tibetan Medicine herb Brown

H/S S H/S S H/S S H/S S

1 Pyrazine �a |b

� � � | � �

2 Pyridine � | � � � � � |3 2,4-Pentadienenitrile � | � � � | � �

4 Furfural � | � | � | � |5 2-Furanmethanol � | � | � | � |6 Styrene � � | | | | | |7 5-Methyl-2-furancarboxaldehyde � | � | | | � |8 Benzaldehyde | | | | | | | |9 Phenol | | | | | | | |10 Benzyl Alcohol | | | | | � | |11 Linalool | � | | | � � �

12 Phenylethyl Alcohol | | | | | � | �

13 Camphor � � | � � � � �

14 Naphthalene � | � | � | � |15 Cinnamaldehyde (E) | | | | | | | |16 Indole � � � | � | � |17 2-Methoxy-4-vinylphenol � | � | � | � |18 Piperonal � | | | | � | |19 Biphenyl � | � | � | � |20 Vanillin | � | � | � | �

21 2H-1-Benzopyran-2-one | | | | | | | |22 Acenaphthylene � | � | � � � |23 .alpha.-Calacorene | � | | | | | �

24 Diethyl Phthalate | | | | | | | |25 8.beta.H-Cedran-8-ol | | | | | | | |26 Benzophenone | | � � � � � �

27 Patchouli Alcohol � � | | | � | |28 9H-Fluoren-9-one � | � � � | � |29 1-Octadecene � | � | � | � |30 Benzyl Benzoate | | | | | | | |31 Octadecane | | � � � � � |32 Phenanthrene � | � | � | � |33 Musk Xylene | | | | | � | |34 Dibutyl Phthalate | | | | � � � �

35 Fluoranthene � � � � � � � |

Refer to supplementary information for a complete listing.a� compound not found in the sample.

b|compound found in the sample.

T.C. Tran, P.J. Marriott / Atmospheric Environment 41 (2007) 5756–57685762

Obviously, there were many more peaks gener-ated when the incenses were burnt; many of thesecompounds were not observed in the incensepowder. The manufacturing process of incense,quality control of raw materials, and actualcomposition of the fragrances used is unknown.Contamination may be a concern. However, bycomparing the H/S composition with the smoke, itis certainly possible to deduce which compoundsarise from the combustion process.

However, aroma compounds are also present inthe combustion stream. When the incenses wereburnt, incense powder may be either completely orpartly combusted to produce new compounds.Compounds that were generated in the smoke weremainly di-aromatics, benzenes, ketones, alkenes andsome nitrogen containing compounds such as N-heterocyclics (pyrazine, pyridine), benzenecarboni-trile and 2,4-pentadienenitrile. Some componentswere found to be present in both incense powder


and smoke; in some instances they were higher inconcentration in the smoke, with higher response inboth GC-MS and GC-FID. Their higher concentra-tion is probably due to their volatilization caused bythe heated tip of the incense stick. Thus specificcompounds from the aroma constituents may beeither absent, attenuated or amplified in the smokestream compared with the powder H/S.

3.3. Comparison of GC�GC-FID analysis of

incense powders and smokes

The GC�GC result in Fig. 3a represents smokevolatiles; Fig. 3b and 3c are the lotus-scentedincense powder H/S and the subtracted chromato-gram representing (smoke–powder volatiles), re-spectively. Compounds are spread over most of theGC�GC plane, between 1 and 6 s in the D2

retention time, with the benefits of the secondaryseparation stage apparent. The subtracted result in

0 20 40

10

Second d

imensio

n r

ete

ntion tim

e (

s)

Second d

imensio

n r

ete

ntion tim

e (

s)

5.0

4.0

3.0

1.0

0.0

10

First dimension retention time (min)

30 50

5.0

4.0

3.0

2.0

1.0

0.0

0 20 3

First dimension rete

2.0

Fig. 3. GC�GC-FID chromatograms of lotus-scented incense separ

chromatogram obtained by subtraction of powder H/S chromatogram

Fig. 3c is intended to reveal only compounds thatare emitted from or generated by the burningprocess, present in the smoke stream. Each darkspot on the chromatogram represents an individualcompound, and the concentration of the compoundin the sample is represented to the first approxima-tion by the response signal of the spot; largerspot�higher amount of compound. Note that in thesubtracted chromatogram, if a spot in the powderH/S is located at the same point as a spot in thesmoke, the peak is subtracted from the smoke resultto give a difference chromatogram, regardless of therelative spot sizes—they only have to be at the sameposition.

There are several peaks that appearo1 s in thesecond dimension (which approximates the unre-tained peak time). This is due to their high polarity,thus they are strongly retained on the polarsecondary column to the extent that they elutein a subsequent modulation event after being


0 40

Second d

imensio

n r

ete

ntion tim

e (

s)

10

5.0

4.0

3.0

2.0

1.0

0.0

20 30 50

0 40 50

ntion time (min)

ated on column set 1 (a) smoke (b) powder and (c) subtracted

from smoke chromatogram (a and b).


transferred onto the second column. This effectcalled ‘wrap around’, can be avoided by setting themodulation period (PM) to exceed the D2 retentiontime of the strongest retained analyte. Wrap aroundcan cause overlapping with compounds in neigh-bouring modulation cycles, and affect identificationand/or quantification. Ideally, wrap around shouldbe avoided, however, it can be accepted if thecompounds showing wrap-around do not co-eluteor interfere with analytes from the next modulation.With this column configuration, PM�6 s is sufficientbecause only little wrap around arises; however,there is no co-elution with other analytes from thenext modulation.

The chromatogram of lotus-scented incensepowder, revealed about 40–60 peaks visualized onthe 2D contour plot, with most of them wellseparated in both dimensions. The actual numberof peaks is much higher, due to the ‘threshold’ ofpeak presentation. A high-signal intensity thresholdwill exclude the smaller peaks from the 2D plot. Thechromatogram of the incense powder (Fig. 3b)contained many fewer peaks compared to thechromatogram of smoke volatiles (Fig. 3a) (atequivalent threshold). GC-qMS indicates that, whenthe incenses are burned, they generate a lot ofpollutants, with many toxic compounds bound tothe atmospheric particulate or in the volatile phase.

Contrary to the structured retentions obtained ine.g. petrochemical samples (where zones of differentcompound classes are readily apparent), the smokeconstituents did not exhibit such order, in agree-ment with previous GC�GC studies on air samples(Hamilton et al., 2004; Xu et al., 2003). The mainreason for this will be the many different grouptypes, with their inherent widely varying boilingpoint and polarity properties, present in these sortsof samples. Structure may indeed be present withingiven subsets of compounds, but the compositionalheterogeneity of the sample simply does not revealthis when an FID detector is used here.

Responses of many compounds that were ob-tained from the H/S SPME of incense powder wereincreased significantly during burning. Some ingre-dients used in incense do not have any noticeableodour unless they are burned (and so are liberatedupon burning). The subtracted chromatogramreveals especially the new compounds that wereproduced during the burning process, showingmany compounds present in the smoke which areproduced by the long and slow incomplete combus-tion process of incense burning.

3.4. Comparison of different incense smokes

The GC�GC contour plots of incense smokesobtained from four different types of incense usingcolumn set 1 are depicted in Fig. 4a–4d. Visually,different patterns were obtained for each incensetype, with lotus-scented incense showing a largenumber of compounds located around 1.5–6.0 sretention time (2tR) in the second dimension and10–45min retention time in the first dimension (1tR).Many of the components apparently are abundanthighly polar peaks. There are also some stronglyretained compounds that underwent ‘wrap around’,however, this did not appear to cause too muchpeak overlap.

Some components in the red Tibetan incense alsowrap around (Fig. 4b) and their peak positionscoincide with (and so are most probably the sameas) peaks in the lotus-scented smoke; however, itseems the number of polar compounds is not asgreat as in the lotus-scented incense; red Tibetansmoke produces fewer polar compounds. Therewere some compounds that eluted later in the firstdimension (retention beyond 40min) in red Tibetanincense which were not observed in the lotus-scented product.

Medicine herb incense smoke had the leastcompounds of all samples, with only a few of themore polar compounds present. Naphthalene andbiphenyl can be seen in all incense smokes (Table 1),with biphenylene in the medicine herb smoke(Supplementary Data).

Brown smokeless incense (Fig. 4d) was claimed tobe environmentally friendly, producing less or nosmoke. However, the 2D contour plot indicated thatit does in fact emit a lot of compounds, althoughnot as much as lotus-scented and red Tibetanincenses. There is a cluster of peaks that cannot beeasily resolved, located around 35–40min in firstdimension. This was also observed in 1D-GC, wherethe baseline rises due to overlapping peaks. Brownincense smoke was found to contain a number ofPAHs, such as fluoranthene and pyrene that werenot found in other incenses.

The results indicate that different aroma materi-als were used for different incenses, and theircombustion will emit different compounds into theatmosphere. However, PAHs such as naphthaleneand biphenyl were found to be present in all theincense types, because these compounds are com-monly present in smoke of combusted or partiallycombusted products.

ARTICLE IN PRESS

0

Second d

imensio

n r

ete

ntion t

ime (

s)

Second d

imensio

n r

ete

ntion t

ime (

s)

5.0

4.0

3.0

2.0

1.0

0.0

10 20


30 40 50

5.0

4.0

3.0

2.0

1.0

0.0

0 10


20 30 40 50


0 10 20 30 40 50

Second d

imensio

n r

ete

ntion t

ime (

s)

Second d

imensio

n r

ete

ntion t

ime (

s)

5.0

4.0

3.0

2.0

1.0

0.0


5.0

4.0

3.0

2.0

1.0

0.0

0 10 20 30 40 50 55

Fig. 4. GC�GC-FID chromatograms of smoke obtained from four different incense samples separated on column set 1

(non-polar–polar) (a) lotus-scented, (b) red Tibetan, (c) medicine herb, and (d) brown (smokeless).


3.5. Comparison of column sets used in GC� GC-FID

analyses

A low-polarity—polar arrangement (column set 1)is often used for GC�GC analysis, however, the‘inverted’ column phase configuration of polar—non polar (column set 2) has been applied to aselection of samples, such as middle-distillatesamples (Vendeuvre et al., 2005), roasted coffeebeans (Ryan et al., 2004) and crude oils (Tran et al.,2006). Therefore, this type of column set was alsotested in this study in an attempt to maximize theresolution of compounds, and observe if wraparound can be lessened. Different column phasesare required in order to achieve orthogonality of

separation (Venkatramani et al., 1996), which ariseswhen separations performed on each column areessentially independent of each other. Orthogonalseparation in GC�GC implies that elution timesfor each dimension can be treated as statisticallyindependent (Schoenmakers et al., 2003). Ryanet al. (2005) concluded that when 1D and 2D columnphases are most dissimilar, the most differentiation inthe second dimension for compounds of differentchemical classes can be obtained.

All incense samples were analysed by using bothcolumn sets 1 and 2. Fig. 5A1 and B1 show theGC�GC contour plots of red Tibetan incensepowder and smoke on column set 1, and Fig. 5A2and B2 the respective results on column set 2. Visual

ARTICLE IN PRESS

A1

A2 B2

B15.0

4.0

3.0

2.0

1.0

0.00 10


20 30 40 50

Second d

imensio

n r

ete

ntion tim

e (

s)

Second d

imensio

n r

ete

ntion tim

e (

s)

4.0

3.0

2.0

1.0

0.0

0 13 25


38 50 63 75 0 13


25 38 50 63 75

4.0

3.0

2.0

1.0

0.0

Second d

imensio

n r

ete

ntion t

ime (

s)

5.0

4.0

3.0

2.0

1.0

0.0

0 10


Smoke, set 1

20 30 40 50 55

Powder, set 1

Powder, set 2 Smoke, set 2Second d

imensio

n r

ete

ntion tim

e (

s)

Fig. 5. GC�GC-FID chromatograms of red Tibetan powder (A) and smoke (B) separated on column set 1 (non-polar–polar, 1) and set 2

(polar–non-polar, 2).

T.C. Tran, P.J. Marriott / Atmospheric Environment 41 (2007) 5756–57685766

comparison of the contour plots shows that bothcolumn sets provided sufficient separation ofcomponents, in which compounds are well spreadout in the primary retention axis. The effect onsecond dimension spread of peaks is dependent onthe polar/non-polar nature of the sample compo-nents. Consequently, when column set 1 is used,peaks are widely distributed (some very stronglyretained) along the secondary retention axis,which suggests that having a polar second columnretains compounds over a large retention window.This results in good separation space use in thesecond dimension. Because some components areretained a little too strongly on the polar secondcolumn, ‘wrap-around’ arises. This accompaniespeak broadening in the second dimension.

With column set 2, dense peak clusteringoccurred between 1.5 and 3.5 s on the second

retention axis, so the 2D retention window is lessthan for column set 1. The compounds were welldistributed on the polar primary column. The polarnature of smoke components implies that polarsolutes will be retained to higher elution tempera-ture (Te) on a polar first column, and then poorlyretained on the non-polar secondary column phase.This causes rapid elution, narrow peak widths, andreduced retention and possibly increased probabilityof component overlap. This results in the com-pounds being somewhat clustered together in thesecond dimension. Notwithstanding this observa-tion, separation of components on the seconddimension was still acceptable. By inference, com-pounds from both incense powder and smoke werepredominantly polar in nature. Both column setsare suitable for the separation of sample compo-nents present in incense powder and smoke, even


though both have their specific advantages anddisadvantages. Generally both column sets providedsufficient resolution of components, and may beused for a study of this type.

4. Conclusions

A rapid and simple SPME extraction methodol-ogy along with the powerful separation technique ofGC�GC has been developed for qualitativecharacterization and comparison of compounds inincense powder H/S and smoke. Compounds ofdifferent classes such as alcohols, benzaldehydes,pyrans, furan, benzene and PAHs were present infour different incense types, and were tentativelyidentified using conventional GC-qMS. In additionto the nitromusk reported by Roveri et al. (1998),others musks were also identified in all four incensetypes. More than half of the compounds identifiedwere generated in incense smokes, which producedvery complex chromatograms. Different com-pounds were emitted during the burning processwhen different incenses were burnt, presumablydue to the varied ingredients used to make theincenses.

Both ‘normal and inverted-phase’ GC�GCcolumn sets can be used to provide good separationfor a complex sample such as this, producingmany more resolved peaks than possible in theconventional 1D-GC experiment. Many minorcomponents can be readily seen in the 2D plots.However, different separation qualities of thecomponents in the 2D-GC space were observedwhen different column sets were used. It wasdeduced that most components were polar innature. The method developed appears suitable foruse of GC�GC fingerprinting for ready compar-ison of different incense types, comparison ofsmoke with powder H/S, and allows facile subtrac-tion of the H/S compounds from the total smoketo reveal new/unique compounds generated bythe burning process. The results obtained are pro-mising for future characterization and identifi-cation of compounds emitted from combustionprocesses.

Acknowledgements

T.C. Tran acknowledges support via ARCLinkage Grant (LP0455515). The authors thankMr Paul Morrison for his technical assistance.

Appendix A. Supplementary data

Supplementary data associated with this articlecan be found in the online version at doi:10.1016/j.atmosenv.2007.02.030.

References

Chen, C.-C., Lee, H., 1996. Genotoxicity and DNA adduct

formation of incense smoke condensates: comparison with

environmental tobacco smoke condensates. Genetic Toxico-

logy 367, 105–114.

Hamilton, J., Webb, P., Lewis, A., Hopkins, J., Smith, S., Davy,

P., 2004. Partially oxidised organic components in urban

aerosol using GC�GC-TOF/MS. Atmospheric Chemistry

and Physics Discussions 4, 1393–1423.

Hamm, S., Bleton, J., Connan, J., Tchapla, A., 2005. A chemical

investigation by headspace SPME and GC-MS of volatile and

semi-volatile terpenes in various olibanum samples. Phyto-

chemistry 66, 1499–1514.

Ho, C.W., Wan Aida, W.M., Maskat, M.Y., Osman, H., 2006.

Optimization of headspace solid phase microextraction (HS-

SPME) for gas chromatography mass spectrometry (GC-MS)

analysis of aroma compound in palm sugar (Arenga pinnata).

Journal of Food Composition and Analysis 19, 822–830.

Kallio, M., Hyotylainen, T., Lehtonen, M., Jussila, M.,

Hartonen, K., Shimmo, M., Riekkola, M.-L., 2003. Compre-

hensive two-dimensional gas chromatography in the analysis

of urban aerosols. Journal of Chromatography A 1019,

251–260.

Lee, S.-C., Wang, B., 2004. Characteristics of emissions of air

pollutants from burning of incense in a large environmental

chamber. Atmospheric Environment 38, 941–951.

Liu, Z., Phillips, J.B., 1991. Comprehensive two-dimensional gas

chromatography using an on-column thermal modulator

interface. Journal of Chromatography Science 29, 227–231.

Lofroth, G., Stensman, C., Brandhorst-Satzkorn, M., 1991.

Indoor sources of mutagenic aerosol particulate matter:

smoking, cooking and incense burning. Mutation Research

261, 21–28.

Lung, S.-C.C., Hu, S.-C., 2003. Generation rates and emission

factors of particulate matter and particle-bound polycyclic

aromatic hydrocarbons of incense sticks. Chemosphere 50,

673–679.

Lung, S.-C.C., Kao, M.-C., Hu, S.-C., 2003. Contribution of

incense burning to indoor PM10 and particle-bound polycyclic

aromatic hydrocarbons under two ventilation conditions.

Indoor Air 13, 194–199.

Mannix, R.C., Nguyen, K.P., Tan, E.W., Ho, E.E., Phalen, R.F.,

1996. Physical characterisation of incense aerosols. The

Science of the Total Environment 193, 149–158.

Mondello, L., Casilli, A., Tranchida, P.Q., Dugo, P., Dugo, G.,

2003. Detailed analysis and group-type separation of natural

fats and oils using comprehensive two-dimensional gas

chromatography. Journal of Chromatography A 1019,

187–196.

Nevell, T.P., Zeronian, S.H. (Eds.), 1985. Cellulose chemistry

fundamentals. Cellulose Chemistry and its Applications. Ellis

Horwood Ltd, New York, pp. 15–29.




Pawliszyn, J., 1997. SPME method development. In: Pawliszyn,

J. (Ed.), Solid Phase Microextraction: Theory and Practice.

Wiley-VCH, New York, pp. 97–140.

Perera, R.M.M., Marriott, P.J., Galbally, I.E., 2002. Headspace

solid-phase microextraction—comprehensive two-dimen-

sional gas chromatography of wound induced plant

volatile organic compound emissions. The Analyst 127,

1601–1607.

Phillips, J.B., Beens, J., 1999. Comprehensive two-dimensional

gas chromatography: a hyphenated method with strong

coupling between the two dimensions. Journal of Chromato-

graphy A 856, 331–347.

Reddy, C.M., Eglinton, T.I., Hounshell, A., White, H.K., Xu, L.,

Gaines, R.B., Frysinger, G.S., 2002. The West Falmouth oil

spill after thirty years: the persistence of petroleum hydro-

carbons in Marsh sediments. Environmental Science and

Technology 36 (22), 4754–4760.

Roveri, P., Andrisano, V., Di Pietra, A.M., Cavrini, V., 1998.

GC-MS analysis of incenses for possible presence of allergenic

nitromusks. Journal of Pharmaceutical and Biomedical

Analysis 17, 393–398.

Ryan, D., Watkins, P., Smith, J., Allen, M., Marriott, P., 2005.

Analysis of methoxypyrazines in wine using headspace solid

phase microextraction with isotope dilution and comprehen-

sive two-dimensional gas chromatography. Journal of Se-

paration Science 28, 1075–1082.

Ryan, D., Morrison, P., Marriott, P., 2005. Orthogonality

considerations in comprehensive two-dimensional gas chro-

matography. Journal of Chromatography A 1071, 47–53.

Ryan, D., Shellie, R., Tranchida, P., Cassili, A., Mondello, L.,

Marriott, P., 2004. Analyis of roasted coffee bean volatiles by

using comprehensive two-dimensional gas chromatography-

time-of-flight mass spectrometry. Journal of Chromatography

A 1054, 57–65.

Schoenmakers, P., Marriott, P., Beens, J., 2003. Nomenclature

and conventions in comprehensive multidimensional chroma-

tography. LC GC Europe 16, 1–4.

Tran, T.C., Logan, G.A., Grosjean, E., Harynuk, J., Ryan, D.,

Marriott, P., 2006. Comparison of column phase configura-

tions for comprehensive two-dimensional gas chromato-

graphic analysis of crude oil and bitumen. Organic

Geochemistry 37, 1190–1194.

Vaz, J.M., 2003. Screening direct analysis of PAHs in atmo-

spheric particulate matter with SPME. Talanta 60, 687–693.

Vendeuvre, C., Ruiz-Guerrero, R., Bertoncini, F., Duval, J.L.,

Thiebaut, D., Hennion, M-C., 2005. Characterization of

middle-distillates by comprehensive two-dimensional gas

chromatography (GC�GC): a powerful alternative for

performing various standard analysis of middle-distillates.

Journal of Chromatography A 1086, 21–28.

Venkatramani, C.J., Xu, J., Phillips, J.B., 1996. Separation

orthogonality in temperature-programmed comprehensive

two-dimensional gas chromatography. Analytical Chemistry

68, 1486–1492.

Williams, A., Ryan, D., Guasca, A.O., Marriott, P., Pang, E., 2005.

Analysis of strawberry volatiles using comprehensive two-

dimensional gas chromatography with headspace solid-phase

microextraction. Journal of Chromatography B 817, 97–107.

characterization of incense smoke by solid phase microextraction—comprehensive two-dimensional gas...

Documents