analysis of flavor and perfume using an internally cooled coated fiber device

J. Sep. Sci. 2007, 30, 1037 –1043 Y. Chen et al. 1037

Yong Chen1

Fr�d�ric Begnaud2

Alain Chaintreau2*Janusz Pawliszyn1

1 Department of Chemistry,University of Waterloo,Waterloo, Ontario, Canada

2 Firmenich SA, Corporate R & DDivision, Analysis, PhysicalChemistry and HumanBioresponses, Geneva,Switzerland

Original Paper

Analysis of flavor and perfume using an internallycooled coated fiber device

A miniaturized internally cooled coated fiber device was applied for the analysis offlavors and fragrances from various matrices. Its integration with a CTC CombiPALautosampler enabled high throughput for the analysis of analytes in complex matri-ces that required simultaneous heating of the matrices and cooling of the fiber coat-ing to achieve high extraction efficiency. It was found that up to ten times increaseof extraction efficiencies was observed when the device was used to extract flavorcompounds in water, even when limited sample temperatures were used to preservethe integrity of target compounds. The extraction of the flavor compounds in waterwith the device was reproducible, with RSD not larger than 15%. The lower limits ofthe linear ranges were in the low ppb range, which was about one order of magni-tude smaller than those obtained with the commercialized 100 lm PDMS fibers.Exhaustive extraction of some perfume ingredients from a complex matrix (sham-poo) was realized. All achieved recoveries were not less than 80%. The repeatabilityof the extraction of the perfume compounds from shampoo was better than 10%.The linear ranges were about 1–3000 lg/g, and the LOD was about 0.2–1 lg/g. Theautomated internally cooled coated fiber device was demonstrated to be a powerfulsample preparation tool in flavor and fragrance analysis.

Keywords: Automation / Flavor / Internally cooled coated fiber device / Perfume / Shampoo /

Received: August 26, 2006; revised: December 29, 2006; accepted: December 29, 2006

DOI 10.1002/jssc.200600333

1 Introduction

Sample preparation for reliable quantitative analysis offlavor and perfume is critical due to their large variety ofconstituents and low concentration in a complex nonvo-latile matrix. The most often used sample preparationmethods in flavor and perfume analysis are those requir-ing little sample preparation, like direct headspace sam-pling and sorbent trapping [1, 2]. Direct headspace sam-pling is the most attractive method because the sam-pling process is quite close to that of human odor percep-tion, especially when olfactory detection is employed.But this method generally suffers from low sensitivity foranalytes with low volatility. Sorbent trapping is believedto have higher sensitivity because it could extract somespecific analytes exhaustively, depending on their affi-nity for the extraction media. However, interactions

with matrices often jeopardizes efficient extraction oftarget compounds.

Solid-phase microextraction (SPME), utilizing a thinlayer of polymer dispersed on a fine fused silica or ametal wire as the extraction phase, has been extensivelyused for flavor and fragrance analysis [2–7]. Since it is asolvent-less technique, chromatographic interferencesfrom solvent and its impurities are essentially absent,which is one of the most attractive features for trace anal-ysis. Moreover, sample preparation, sampling, and sam-ple introduction are integrated into one step and com-pletely automated. Traditional SPME is a nonexhaustiveextraction technique, which removes a negligible por-tion of the analyte from the sample due to the smallvolume of the extraction phase. To enhance the sensitiv-ity of SPME, an internally cooled coated fiber device wasdeveloped [8]. Miniaturization and automation of thedevice were recently realized [9]. Utilizing this tech-nique, concomitant heating of the sample and cooling ofthe fiber coating greatly elevated the extraction effi-ciency, and may result in exhaustive extraction under

Correspondence: Dr. Yong Chen, Department of Chemistry,University of Waterloo, Waterloo, Ontario, N2L 3G1, CanadaE-mail: [email protected]: (+1)-519-746-0435

Abbreviation: SPME, solid-phase microextraction

i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

* Dr. A. Chaintreau, Fax: (+41)-22-780-33-34; E-mail: [email protected]

1038 Y. Chen et al. J. Sep. Sci. 2007, 30, 1037 – 1043

optimized conditions. Application of the automatedinternally cooled coated fiber device for environmentalanalysis demonstrated the efficacy of the technique [10].The objective of this work was to demonstrate the effi-ciency of the miniaturized and automated internallycooled SPME device for quantitative analysis of flavorcompounds in water and perfume compounds in sham-poo.

2 Experimental

2.1 Chemicals and materials

Benzyl acetate, geraniol (3,7-dimethyl-2,6-octadien-1-ol),Cetaloxm ((l)-8,12-epoxy-13,14,15,16-tetranorlabdane), fla-vor model components (hexanal, butyl acetate, (E)-2-hex-enal, isoamyl acetate, isobutyl isobutyrate, hexyl acetate,and heptyl acetate), and ethanol were obtained from Fir-menich (Geneva, Switzerland). Galaxolidem (1,3,4,6,7,8-hexahydro-4,6,6,7,8,8-hexamethyl-cyclopenta[G]isochro-mene) was purchased from IFF (New York, NY, USA).Unperfumed shampoo bases and perfumed shampoosamples were obtained from Firmenich. This includedsodium lauryl sulfate based conditioning shampoo,ammonium lauryl sulfate based conditioning shampoo,sodium lauryl sulfate based simple shampoo and a mar-ket product sodium lauryl sulfate based benchmark con-ditioning shampoo. Water was collected from a Milliporepurifying system (Billerica, MA, USA).

SPME fibers were purchased from Supelco (Bellefonte,PA, USA). Ten and twenty milliliter sample vials wereused for automated analysis with magnetic crimp capsand PTFE coated silicone septa (Chromacol, Welwyn Gar-den City, UK).

2.2 Instrumentation

GC was performed on a Varian 3800 GC (gas chromato-graph) coupled with FID (Flame Ionization Detection)using a Star Chromatography Workstation (ver. 5.51).Automated analysis was performed using a CTC Combi-PAL autosampler (Zwingen, Switzerland) with the asso-ciated Cycle Composer software (ver. 1.4.0). The PAL wasequipped with a SPME fiber/syringe holder and a tem-perature controlled six-vial agitator tray. Separation wasperformed using a 30 m60.25 mm id, 0.25 lm filmthickness DB-1 fused silica column (Supelco) installed inthe Varian GC. Helium was used as carrier gas at a flowrate of 1 mL/min. The Varian FID heated to 2508C withgas flows for hydrogen, high purity air, and make-up gas(nitrogen) set at 30, 300, and 25 mL/min, respectively.

For the analysis of the flavor ingredients, the columnwas initially set at 458C for 1 min and then ramped at108C/min to 2508C. The injector temperature was set to2508C. For analysis of perfume compounds the column

was initially set at 458C for 1 min and then ramped at58C/min to 2708C. The injector was set at a temperatureof 2708C.

2.3 Extraction of flavor and perfume ingredientsfrom aqueous media

The extraction conditions were carried out as follows,unless otherwise specified. Optimization of extractionconditions can be found elsewhere [11].

The extraction of flavor ingredients from water(9.3 lg/mL) was performed in the headspace of 8 mL ofaqueous solutions in 20 mL vials. The samples were incu-bated at 30, 45, and 608C, with and without agitation, fordifferent times as indicated in the discussion.

The extraction of perfume ingredients from shampoowas performed in 20 and 2 mL vials containing 10–200 lL of 1% aqueous shampoo solutions. The sampleswere incubated at 608C. The extraction time was 45 min.

When the fiber was cooled with CO2, the coating tem-perature was about 1 l 58C for all extractions.

3 Results and discussion

The detailed description of the miniaturized and auto-mated internally cooled device can be found in the litera-ture [9]. As a result of the miniaturization, the fiber couldbe accommodated in a 18-gauge stainless steel tubingwhich was integrated with the barrel of a commercia-lized 100 lL syringe. This design allowed the device to beused routinely, in a similar way as a traditional SPMEdevice. The miniaturized device was then ready to beused on a CTC CombiPAL autosampler with a minor mod-ification of the syringe holder. To control the tempera-ture of the fiber coating, a temperature controller, a sole-noid valve, and tubing of different inner diameter wereused. With this system, the temperature of the fiber coat-ing could be controlled to within 58C of the preset value.Full automation of the process was realized by couplingthe external temperature control system with the auto-sampler through a logic circuit built in the temperaturecontroller (Fig. 1). This allowed the controller to beturned on or off as required. The temperature of the fibercoating was thus well controlled, and the extractionwith the device was more reproducible. The automationmade high throughput analysis, and exhaustive extrac-tion possible, providing reduced analytical time and sim-plified procedures of quantification.

3.1 Extraction of flavor ingredients from water

The flavor ingredients used in this study are organic com-pounds with a wide range of volatility and polarity.Extraction of these compounds from the headspace of


J. Sep. Sci. 2007, 30, 1037 –1043 Gas Chromatography 1039

aqueous matrices is challenging because most of thesecompounds are moderately to highly polar compounds.Moreover their concentrations in the headspace are verylow due to their small Henry constants. Heating thematrices would increase the concentrations of thesecompounds in the headspace and extraction of theheated samples using the internally cooled fiber wouldfurther increase the extraction efficiency.

Figure 2 shows the extraction time profiles of butylacetate and heptyl acetate. The internally cooled fiberdevice was used in two ways: cooling the fiber to 18C andwithout cooling the fiber. Both results were comparedwith the extraction time profiles obtained with a com-mercialized 100 lm PDMS fiber. The extraction time pro-files of the device, when it was not cooled, were similarto those of the commercialized 100 lm PDMS fiber.When the internally cooled fiber was cooled to about18C, the amounts of analytes extracted in the fiber weresignificantly increased. Table 1 summarizes the increaseof extraction efficiency for each component of the flavorand the extraction reproducibility. Reasonable extrac-tion reproducibility was observed for all the components(3–9%). It is noteworthy that heating the sample andsimultaneously cooling the fiber affected the extractionefficiency of different compounds to different degrees.The increase in extraction efficiency varied from abouttwice for hexyl acetate and heptyl acetate to about 11times for butyl acetate and (E)-2-hexenal. Three main rea-sons explain these differences. Firstly, the extractionequilibrium for some compounds was not reached underthe experimental conditions; secondly, the heat capaci-ties for different compounds are different; thirdly, thechange of Henry constants with temperature was differ-ent for different compounds.

Figure 2 also demonstrates that the extraction equili-brium was reached in about 20 min for butyl acetate, butit was not reached even after 1 h for heptyl acetate. Vola-

tile compounds tend to accumulate into the headspace,leading to higher concentrations in the gaseous phase.As their distribution constants are small, only smallamounts of analytes are required to be transferred to thefiber to establish the extraction equilibrium, which can


Figure 1. Scheme of automation of the internally cooledfiber.

Figure 2. Extraction time profiles of (a) butyl acetate and (b)heptyl acetate from aqueous solution. Samples were incu-bated at 458C. The concentrations of butyl acetate and hep-tyl acetate were 371 and 279 ng/g, respectively.

Table 1. Comparison of the extraction efficiencies of theinternally cooled fiber for flavor compounds in water with andwithout cooling the fiber

Increase ofthe extrac-tion effi-ciencywhen cool-ing the fi-ber (times)

RSD of ex-tractiona)

(%, n = 7,fiber withcooling)

RSD of ex-tractiona)

(%, n = 7,fiber with-out cooling)

Hexanal 9 9 5Butyl acetate 11 9 5(E)-2-Hexenal 11 7 5Isoamyl acetate 5 5 5Isobutyl isobutyrate 6 7 5Hexyl acetate 2 4 2Heptyl acetate 2 4 3

a) RSD (%) of the amounts of analytes extracted in thefiber.


be reached quickly. On the contrary, the concentrationsof less volatile compounds in the headspace are low, andtheir distribution coefficients are large. Larger amountof compounds need to be transferred from the aqueoussolution to its headspace then to the coating, in order toreach extraction equilibrium, which requires a longertime. Increase of temperature and/or agitation will accel-erate mass transfer, and thus shorten the equilibrationtime.

This is illustrated by the behavior of heptyl acetate. Ina typical experiment, when the internally cooled fiberwas cooled to about 18C during extraction, and the tem-perature of the sample was increased from 30 to 45, andthen to 608C, the amount of heptyl acetate extracted inthe internally cooled fiber increased (Fig. 3). This couldbe ascribed to two reasons. Firstly, the temperatureincrease accelerated the mass transfer rate, which couldincrease the extraction efficiency for pre-equilibriumextraction. The second reason is that the increase of tem-perature gap between the internally cooled fiber and thesample increased the distribution coefficient, and subse-quently the amount of analyte extracted in the intern-ally cooled fiber.

Due to the limited thermal stability of fragrance andflavor analytes and of the matrices, the maximum tem-perature investigated was 608C. As a consequence, agita-tion was used to improve the extraction. Thus a micromagnetic stirrer was installed under the agitator, and amagnetic stir bar was put in samples prior to sealing thevials. The extraction of butyl acetate, which was mostlyconducted from the headspace, was not affected by theagitation (Fig. 4). In contrast, the agitation significantlyaffected the extraction of less volatile compounds suchas heptyl acetate, because it accelerated their matrix-to-air mass transfer of less volatile compounds. Agitation isefficient only for less volatile compounds in liquidmatrices with small viscosity when the sampling tem-

peratures are low. With elevated sampling temperatures,the mass transfer rate is increased by fast diffusion andconvection is induced by simultaneously heating thesamples and cooling the internally cooled fiber. Whenpossible, increasing the sampling temperatures could bemore efficient than agitation, for instance, for matricessuch as oils.

To further explore the advantages of the internallycooled fiber, calibration curves were constructed andcompared with those obtained from a commercialized100 lm PDMS fiber. All standards were agitated with amagnetic stir bar during extraction. The slopes of thecalibration curves using the internally cooled fiber weremuch higher than those of the commercial fiber, whichimplied that the internally cooled fiber was more sensi-


Figure 3. Temperature effect on the extraction of heptylacetate from the headspace of the aqueous solutions usingthe internally cooled fiber cooled to about 18C. The concen-tration of heptyl acetate was 279 ng/g.

Figure 4. Comparison of the extraction of (a) butyl acetateand (b) heptyl acetate from the headspace of aqueous solu-tion with and without agitation. The internally cooled fiberwas cooled to 18C during the extraction. Samples were incu-bated at 308C. The concentrations of butyl acetate and hep-tyl acetate were 371 and 279 ng/g, respectively.


tive to the concentration variation due to its higheradsorption capacity (Table 2). Both fibers maintainedwide linear ranges of at least two orders of magnitude.The LOQ using the internally cooled fiber were muchlower, from 2 to 10 times, than those using the commer-cialized fiber. However, the extraction with the commer-cialized fiber was more reproducible than that with theinternally cooled fiber. This might be due to the largertemperature variation of the internally cooled fiber(€58C) compared to that of the commercialized fiber (lessthan €0.58C).

3.2 Extraction of perfume ingredients fromshampoo

Traditional SPME under equilibrium extraction impliesthat only a small portion of analytes is extracted in thefiber coating. The disadvantage is that the change ofmatrix composition influences the free concentrationsof analytes, and subsequently changes the amount ofanalytes extracted in the fiber. Using the internallycooled fiber under optimized conditions to achieveexhaustive extraction could overcome the drawbacksdue to the matrix effects.

Equation (1) describes the extraction of analytes froma headspace sample using SPME under equilibrium [12],

nf ¼KfsVf

KfsVf þ KhsVh þ Vsn0 ð1Þ

where nf is the amount of analyte in the fiber, n0 is thetotal amount of analyte in the sampling system, Kfs is thefiber-to-matrix distribution coefficient, Khs is the head-space-to-matrix distribution coefficient, Vf is the volumeof fiber coating, Vs is the volume of sample matrix, and Vh

is the volume of headspace. To maximize the analyterecovery (nf/n0), Kfs, or Vf should be increased, or Khs, Vh, orVs should be decreased. The most convenient way is todecrease the sample volume Vs and the volume of theheadspace Vh. Utilizing a small size vial and a smallamount of sample would significantly increase the recov-ery of analytes in the fiber.

Figure 5 shows the effects of sample volume on therecovery of perfume ingredients with the internallycooled fiber. When the volume of the sample was as lowas 10 lL, benzyl acetate and geraniol were completelyrecovered. Full recovery of Cetalox was not achieved forall cases, because the extraction was kinetically con-trolled, and a longer extraction time would be required.

According to Eq. (1), the decrease of the vial volumewould increase the recovery. However, Eq. (1) onlydescribes equilibrium extraction. In practice, on decreas-ing the vial volume from 20 to 2 mL, the recovery of ben-zyl acetate was improved as it reached equilibrium dur-ing the extraction. Conversely, the recoveries of geranioland Cetalox, for which equilibrium was not reached dur-ing the extraction time, were significantly decreased(Fig. 6). As the mass transport in the sample matrix andthrough the interface to its headspace was the rate-limit-ing process, the same amount of sample would possess ahigher surface-to-volume ratio in 20 mL vials than in2 mL vials, which would facilitate higher mass transferrate.

Varying the sample temperature from 25 to 608C(Fig. 7) increased the extraction efficiency for all perfume


Table 2. Summary of calibration curves for the extraction offlavor ingredients from water by the use of the internallycooled SPME fiber and a 100 lm PDMS fiber

Y = aX + b Linearrange(ppb)

RSD (%,n = 5)a)

a b R2

Internally cooled SPME fiberHexenal 392 – 1 813 0.9979 3 – 400 9Butyl acetate 1131 682 0.9999 2 – 100 12(E)-2-Hexanal 1561 3 366 0.9982 4 – 100 13Isoamyl acetate 4380 9 067 0.9996 4 – 500 7Isobutyl isobutyrate 6682 68 252 0.9959 4 – 200 9Hexyl acetate 5913 47 852 0.9989 2 – 1000 6Heptyl acetate 7727 9 036 0.9995 1 – 300 11100 lm PDMS fiberHexenal 27 1 833 0.9986 35 – 400 5Butyl acetate 49 1 010 0.9999 25 – 2800 1(E)-2-Hexanal 60 1 571 0.9949 20 – 200 6Isoamyl acetate 192 13 226 0.9999 40 – 17 000 1Isobutyl isobutyrate 1009 7 037 0.9999 4 – 1900 3Hexyl acetate 872 197 823 0.9934 100 – 1200 1Heptyl acetate 3065 8 187 0.9983 5 – 700 4

a) The samples were incubated at 308C, and the extractiontime was 20 min. There were at least seven concentra-tion levels for each calibration curve. RSD (%) was deter-mined based on the amounts of analytes extracted inthe fiber and at close to up-limit range of the calibrationcurve.

Figure 5. Sample volume effect on the extraction of perfumecompounds. Extraction of small amounts of samples usingthe internally cooled fiber in 20 mL vials. The concentrationof benzyl acetate, geraniol, and cetalox in the aqueousshampoo solutions was 3.0, 4.7, and 0.4 lg/g.


compounds to different degrees. For benzyl acetate, theaugmentation was not significant, because the recoverywas also high (about 88%) at room temperature (l258C).

For geraniol and Cetalox, the increase was more signifi-cant due to the accelerated mass transfer rates and theincrease of distribution coefficient. The fiber coating was“zero sink” to Cetalox, meaning that the concentrationof this compound at the fiber coating-headspace inter-face can be considered as zero, so the extraction effi-ciency was maintained even when the fiber coating wasnot cooled.

Table 3 summarizes the calibration details for theextraction of perfume ingredients from 1% shampooaqueous solution using the internally cooled fiber. Themethod maintained large linear ranges in terms of con-centrations of perfume ingredients in shampoo.Although the volume of the samples was significantlysmaller, the sensitivity of the method was not jeopar-dized due to the higher recoveries and higher concentra-tions of perfume ingredients in standards with dilutionof shampoo to only 1%.

4 Concluding remarks

The internally cooled fiber device was successfully usedto quantify flavor compounds in water and perfumecompounds in shampoo. A significant increase of extrac-tion efficiency for the extraction of flavor from waterwas demonstrated, together with a lower LOQ andhigher sensitivity. Quasi-exhaustive extraction of per-fume compounds from shampoo using the device wasrealized under optimized extraction conditions. Owingto the automation, high throughput analysis of complexflavor and fragrance samples would be possible. In addi-tion, the capacity of the internally cooled fiber devicewould be further extended if the temperature gapbetween the fiber coating and samples could beincreased.

5 References

[1] Dewulf, J., Langenhove, H. V., Wittmann, G., TrAC, Trends Anal.Chem. 2002, 21, 637 – 646.

[2] Kafkas, E., Cabaroglu, T., Selli, S., Bozdogan, A., et al., Flavor Fragr.J. 2006, 21, 68 – 71.

[3] Dong, C. Z., Wang, W. F., Anal. Chim. ACTA. 2006, 562, 23 – 29.


Table 3. Calibration for the extraction of perfume compounds from 1% aqueous shampoo solutions using the internally cooledfiber

Y(mass, ng) = aX(concentration, lg/g) + b Linear range(lg/g)a)

LOD(lg/g)a)

Repeatabilityof extraction(%, n = 7)

Recovery(%)

a (mass/(lg/g)) b (mass) R2

Benzyl acetate 0.4497 –9.0077 0.9937 6–2000 0.6 8 87Geraniol 0.4606 –8.3185 0.9989 8–3000 1 9 90Cetalox 0.4288 –1.0011 0.9952 1–300 0.2 9 83

a) Concentrations were expressed as the concentrations of perfume ingredients in shampoo.

Figure 6. Vial volume effect on the extraction of perfumecompounds. Extraction of 50 lL of 1% shampoo standardaqueous solutions using the internally cooled fiber. Theextraction time was 45 min. The concentration of benzylacetate, geraniol, and Cetalox in the aqueous shampoo solu-tions was 3.0, 4.7, and 0.4 lg/g.

Figure 7. Effect of sampling temperature on the extraction ofperfume compounds from 50 lL 1% aqueous shampoo solu-tions using the internally cooled fiber that was cooled to 18Cduring extraction. The concentrations of benzyl acetate, ger-aniol, and Cetalox in the aqueous shampoo solutions were3.0, 4.7, and 0.4 lg/g.


[4] Tranchida, P. Q., Lo Presti, M., Costa, R., Dugo, P., et al., J. Chroma-togr. A 2006, 1103, 162 – 165.

[5] Coleman, W. M., Dube, M. F., J. Sci. Food Agric. 2005, 85, 2645 –2654.

[6] Martinez-Urunuela, A., Gonzalez-Saiz, J. M., Pizarro, C., J. Chroma-togr. A 2005, 1089, 31 – 38.

[7] Hawthorne, S. B., Miller, D. J., Pawliszyn, J., Arthur, C. L., J. Chro-matogr. A 1992, 603, 185 – 191.

[8] Zhang, Z., Pawliszyn, J., Anal. Chem. 1995, 67, 34 – 43.

[9] Chen, Y., Pawliszyn, J., Anal. Chem. 2006, 78, 5222 – 5226.

[10] Ghiasvand, A. R., Hosseinzadeh, S., Pawliszyn, J., J. Chromatogr. A2006, 1124, 35 – 42.

[11] Chen, Y., Begnaud, F., Chaintreau, A., Pawliszyn, J., Flavor Fragr. J.2006, 21, 822 – 832.

[12] Pawliszyn, J., Solid Phase Microextraction – Theory and Practice,Wiley VCH, New York 1997.


analysis of flavor and perfume using an internally cooled coated fiber device

Documents