Volatile Profiling in Wine Using GasChromatography Mass Spectrometrywith Thermal Desorption

Authors

Kaushik Banerjee, Narayan Kamble,

and Sagar Utture

National Research Centre for Grapes,

Pune, India

Chandrasekar Kandaswamy

Agilent Technologies,

Bangalore, India

Application Note

Food sensory

Abstract

Wine aroma is an important characteristic and may be related to certain specific

parameters such as raw material, production process, and so forth, that impart spe-

cific aromas to the wines. Gas chromatography-mass spectrometry with thermal

desorption (GC/MS-TD) parameters were optimized to profile volatile compounds

from wines. Sample preparation involved extraction of an 8-mL wine sample into a

headspace vial with 2 g sodium chloride and internal standard (IS). A preconditioned

SPE cartridge was used to adsorb released volatile compounds while heating the

vial on a magnetic stirrer for 40 minutes at 80 °C. A SPE cartridge loaded with

volatiles was inserted into a TD tube, and the tube was kept in an autosampler for

analysis. The GC/MS-TD parameters, desorption time, desorption temperature, low

and high trap temperatures, and so forth, were optimized. The profile of 225 volatile

compounds of Indian wines was qualitatively analyzed by GC/MS-TD. The targeted

deconvolution was used to identify a large number of volatile compounds in wine

samples in a shorter time. Wine contains esters (67), alcohols (61), aldehydes (19),

terpenes (19), organic acids (18), ketones (14), ethers (7), phenols (5), lactones (4),

Pyrazines (3), and others (8). Based on the wine aroma profile of 15 Indian wines,

these wines could be differentiated into three groups. Thirteen of the wines could

be placed into a single group, whereas Cinsaut and Gewurztraminer showed signifi-

cantly different concentrations of volatile compounds, and formed individual groups

of wines.

2

Introduction

The wine aroma profile is important as it contributes to thequality of the final product. Aroma compounds are closelyconnected with sensory attributes, which are crucial in deter-mining consumer acceptability. In terms of volatile com-pounds, wine is one of the most complex beverages [1]. Morethan 800 volatile compounds such as alcohols, acids, esters,ethers, ketones, terpenes, aldehydes, and so forth, are foundin wine and identified with wide range of concentrations. Howmany, and what types of volatile compounds are presentdepends on many factors such as the vineyard’s geographicalsite, which is related to soil and climate characteristics [5],grape variety [6], yeast strain, and technical conditions duringwine making [2].

Usually, volatile compounds in wine are present in differentconcentrations ranging from mg/L down to a few ng/L, andseveral sampling techniques are used for analysis (isolation)of these volatile compounds. Of the many methods of extract-ing volatiles from wine, liquid-liquid extraction (LLE) [3] is pri-marily used for the fractionation of free and bound volatilecompounds [4]. Some other modern volatile extraction techniques used for volatiles analysis of grapes and winesinclude:

• Static headspace [5]

• Stir bar sorptive extraction (SBSE) [6]

• Solid phase microextraction (SPME) [7]

• Purge and trap/dynamic headspace, and so forth

SBSE is a volatile extraction technique developed in 1999 [8].It is a highly sensitive technique for trace and ultratraceanalyses, and uses a magnetic stir bar coated with polydi-methylsiloxane (PDMS). This application note describes theoptimization of the SBSE technique using a SPE cartridge(SPE-td cartridge) in a thermal desorption system coupledwith GC-MS for screening wine samples for large numbers of

volatile compounds. This optimized method was applied toscreen volatiles from different wine samples. This applicationnote demonstrates the use and need of software-based identification in large-scale screening methods.

Experimental

InstrumentationTD system series 2 UNITY assembled with autosampler series2 ULTRA (Markes International, LLANTRISANT, RCT, UK. TheTD tube containing the SPE cartridge (Markes p/n C-SPTD)(loaded with volatile compounds) was kept in the autosam-pler. The desorption of volatiles was performed at 180 °C for20 minutes. All volatiles and semivolatiles were passed fromthe autosampler to the TD trap unit using a transfer line(maintained at 200 °C). All volatiles were trapped by a coldtrap, which was kept at –20 °C using an electronic coolingsystem (Pelletier cooler). After that, the cold trap was heatedto 275 °C at the rate of 60 °C/sec, and maintained at thattemperature for 5 minutes. The desorbed vapours were transferred to the analytical column through a transfer linemaintained at 200 °C, and analyzed using GC/MS.

Table 1. Instrumental Conditions for the Analysis of Volatiles in Wine

GC conditions

Column HP-INNOWAX 60 m × 0.25 mm, 0.25 µm(p/n 19091N-136)

Flow 1.3 mL/min

Oven ramp 40 °C (1 minute hold) –5 °C/min to 250 °C (24 minutes hold)

Total run time 67 minutes

MS conditions

Ion source temperature 230 °C

Electron voltage –70 eV

Quadrupole temperature 150 °C

EM gain

EM volts 2035

Scan range 30 m/z – 300 m/z

Tuning

Interface temperature 240 °C

Thermal desorption conditions

Autosampler

Desorption time 20 minutes

Desorption temperature 180 °C

Low trap temperature –20°C

High trap temperature 275 °C

Trap hold 5 minutes

3

Sample extractionThe 8-mL wine sample was drawn into a 20-mL glass head-space vial. A 3-octanol internal standard (150 µg/L), and 2 gof NaCl were added. A SPE cartridge (for adsorbing volatilecompounds) and magnetic stirrer (for shaking) were placed inthe mixture. After crimping the vial, it was heated for 40 min-utes at 80 °C, followed by cooling for 30 minutes at room tem-perature. The SPE cartridge with the volatile compounds waswiped with a lint-free tissue paper, and transferred to anempty glass TD tube prior to analysis.

Figure 1. Effect of desorption time for sample tube on increase in the response of the volatilecompounds from wine.

Table 2. Optimization Parameters to Get Better Abundance and Repeatability

S. no Parameter in optimization Function Condition Optimized

Thermal desorption system optimization

1 Desorption time Time required to desorb volatiles from Spe-TD 5, 10, 15, 20, 25 minutes 20 minutes

2 Cold trap temperature Concentrating volatiles from Spe-Td +10, 0, –10, –20, –30 –20 °C

3 High trap temperature Temperature to transfer concentrated volatiles to column 250, 275, 300, 325, 350, 375 °C 250 °C

Extraction optimization

4 Volume of wine for extraction Get better abundance 4, 6, 8,10 mL 6 mL

5 Spe-TD soaking time with stirring Spe-TD exposure time to wine to adsorb 10, 20, 30, 40, 50, 60, 70 minutes 40 minutes

6 Sodium chloride addition Increase absorption by ionization 0.5, 1, 1.5, 2, 2.5, 3.0 g 2 g

7 Heating cartridge while extraction To increase absorption capability 40, 50, 60, 70, 80, 90 °C 80 °C

3-Pentan

ol, 2,4-

dimeth

yl-

Ethyl

2-hyd

roxy

isova

lerate

1-Octa

nol, 2-m

ethyl-

2,3-B

utanedione

Meth

yl dim

ethoxy

acetat

e

Phenyleth

yl alc

ohol

Hexanal

Camphene

Benzene, (e

thoxy

methyl)

-

Furan

, 2-pentyl

-

3-Hexe

n-1-ol, (

Z)-

4-Hexe

n-1-ol, (

E)-

2-Hexe

n-1-ol, (

Z)-

1-Octe

n-3-ol

1-Hexa

nol, 2-eth

yl-

2-Nonenal,

(Z)-

Benzaldehyd

e

1-Octa

nol

1-Nonan

ol

Acetic

acid

D-Limonene

Isopro

pyl myri

state

1-Dodeca

nol

Pentadeca

noic ac

id

5 minutes

10 minutes

15 minutes

20 minutes

25 minutes

0

50

100

150

200

250

300

350

400

450

% R

epon

se

4

Figure 2. Effect of different cold trap temperatures on adsorption.

1,4-B

utanediol

Isoam

yl ac

etate

Isoam

yl pro

pionate

2-Pro

penal

4-Penten-2-

ol

1-Pentan

ol, 3-m

ethyl-

1-Pro

panol, 3

-ethoxy

-

Diethyl

maleate

Hexanal

Decanal

2-Butenal

Butryolac

etone

Cyclopentan

one

Octanal

Ethyl

methoxy

acetat

e

Ethyl

9-dece

noate

Furan

, 2-eth

yl-

3-Hexe

n-1-ol, (

Z)-

Acetal

dehyde

2-Hexy

l-1-o

ctanol

Nonanal

1-Octe

n-3-ol

D-Limonene

Dodecanoic

acid, e

thyl

ester

Pentadeca

noic ac

id

Octadeca

noic ac

id

Diethyl

methyls

uccinate

Heptanoic

acid

Phenyleth

yl octa

noate

Butanoic

acid

n-Pentad

ecanol

n-Trideca

n-1-ol

4-Meth

yl-2-o

xova

leric ac

id

Trap temperature 0 °C

Trap temperature 10 °C

Trap temperature -10 °C

Trap temperature -20 °C

Trap temperature -30 °C

0

50

100

150

200

250

% R

espo

nse

trans-2

-Meth

yl-4-h

exen-3-

ol

Propan

oic ac

id, 2-m

ethyl-

, eth

yl este

r

2,3-B

utanedione

Butanoic

acid, 3

-meth

yl-, 2

-meth

ylpro

pyl este

r

1-Octa

nol, 2-m

ethyl-

1-Butan

ol

1-Butan

ol, 3-m

ethyl-

, ace

tate

Butanal,

2-meth

yl-

2-Butenedioic

acid (Z

)-, dieth

yl este

r

Octanal

4-Hexe

n-1-ol, (

E)-

Heptanoic

acid, e

thyl

ester

3-Hexe

n-1-ol, (

Z)-

1-Octe

n-3-ol

1-Octe

n-1-ol, (

E)-

1-Octa

nol

1-Nonan

ol

1-Deca

nol

Hexanoic

acid

2-Hexe

noic ac

id

Octanoic

acid

Butanoic

acid

Nonanoic

acid, e

thyl

ester

2-Pro

penal

Ethyl

hydro

gen succ

inate

Benzaldehyd

e, 4-m

ethoxy

-

2-Butan

one

Acetic

acid, d

iethoxy

-, eth

yl este

r

Santal

ol

2,4,6-

Octatri

ene, 2,6-

dimeth

yl-, (E

,Z)-

1,2-B

utanediol, 3

,3-dim

ethyl-

Temperature 250 °C

Temperature 275 °C

Temperature 300 °C

Temperature 325 °C

Temperature 350 °C

Temperature 375 °C

Temperature 400 °C

0

20

40

60

80

100

120

140

Con

cent

rati

on (p

pb)

High trap temperature optimization

Figure 3. Effect of different cold trap temperatures on desorption.

5

Figure 4. Effect of percentage of NaCl addition and compound response.

Figure 5. Effect of shaking up time for SPE-td during extraction.

Diisopro

pylmeth

anol

Acetal

dehyde

Ethyl

isobutyr

ate

Acetic

acid, b

utyl este

r

1-Pro

panol

Benzeneac

etaldehyd

e

Hexanal

1,4-B

utanediol

Isoam

yl ac

etate

2-Pro

penal

4-Penten-2-

ol

Camphene

Octanal

1-Pentan

ol, 3-m

ethyl

Meth

ane, is

othiocy

anato

-

3-Hexe

n-1-ol, (

Z)-

4-Hexe

n-1-ol, (

E)-

2-Hexe

n-1-ol, (

Z)-

1-Heptan

ol

Acetic

acid

Ethyl

3-hyd

roxy

butanoate

Benzaldehyd

e

1-Octa

nol

Propan

oic ac

id, 2-m

ethyl-

Benzaldehyd

e, 4-m

ethyl-

Isoam

yl ca

prylate

1-Nonan

ol

Butanoic

acid, 3

-meth

yl-

6-Octe

n-1-ol, 3

,7-dim

ethyl-

2-Pro

penal, 3-

phenyl-

Hexanoic

acid

Benzyl a

lcohol

Phenyleth

yl alc

ohol

Heptanoic

acid

Phenol

Phenol, 4-eth

yl-

Meth

yl tri

decanoate

No NaCl

0.5 g NaCl

1 g NaCl

2 g NaCl

3 g NaCl

0

50

100

150

200

250

300

350

Without heating Heating

400

450

% R

espo

nse

Isopro

pyl iso

butyrate

Acetic

acid, b

utyl este

r

Cyclopentan

one

2-Meth

ylheptan

oic ac

id

2,3 - B

utanedione

Allyl p

ropionate

Butanal,

3-hyd

roxy

-

1-Pentan

ol

1-Pentan

ol, 4-m

ethyl-

2-Pro

penal

1-Pro

panol, 3

-ethoxy

-

4-Hexe

n-1-ol, (

E)-

2-Hexe

n-1-ol, (

Z)-

1-Heptan

ol

Furfu

ral

Benzaldehyd

e

Linalo

ol

Butyrolac

tone

Decanoic

acid, e

thyl

ester

1-Pro

panol, 3

-(meth

ylthio)-

2-Pro

penal, 3-

phenyl-

Hexanoic

acid

Butanoic

acid, b

utyl este

r

Octanoic

acid

Ethyl

4-eth

oxybenzo

ate

Ethyl

oleate

2-Hexa

decanol

Phenol, 4-eth

yl

Benzyl a

lcohol

Eugenol

10 minutes

20 minutes

30 minutes

40 minutes

50 minutes

60 minutes

70 minutes

0

50

100

150

200

250

300

350

400

450

500

% R

espo

nse

Data analysisThe optimized method for analysis of volatile compoundscould generate a very complex chromatogram with a largenumber of compounds (225), eluting in a shorter run time of67 minutes. Manual integration of these samples and theiridentification could limit up to 70 compounds because identi-fication of complex coelutions was ambiguous. The major dis-advantage of manual integration is that integration of onesuch sample requires more than 8 hours, and analysis ofbatches of samples will be a herculean task. Using targeteddeconvolution, the integration of one sample could beprocessed within 10 minutes with identification of the com-pounds against the targeted library of 370 compounds. Thesame processing could also be done by AMDIS. This tooluses the NIST11 spectral database for identification, whichcontains thousands of mass spectra of compounds other thantarget compounds. During the processing of the samples, it ispossible to have a library hit for a nontarget or a false identifi-cation of a compound, therefore, manual review of the peaksis essential. In an analysis using targeted deconvolution software (Agilent MassHunter quan. Tool B.06), only targetcompounds are identified, because the processing is doneagainst the target library, which is more reliable.

6

0

20

40

60

80

100

120

Diisopro

pylmeth

anol

Acetaldehyd

e

Ethyl

isobuty

rate

Acetic acid, b

utyl e

ster

1-Pro

panol

Hexanal

1,4-B

utanediol

Isoamyl

acetate

2-Pro

penal

4-Pente

n-2-o

l

Camphene

Octanal

1-Penta

nol, 3-m

ethyl-

Meth

ane,…

3-Hexe

n-1-ol, (

Z)-

4-Hexe

n-1-ol, (

E)-

2-Hexe

n-1-ol, (

Z)-

1-Hepta

nol

Acetic acid

Ethyl

3-…

Benzaldehyd

e

1-Octa

nol

Propanoic

acid, 2-…

Benzaldehyd

e, 4-m

ethyl-

Isoamyl

caprylate

1-Nonanol

Butanoic

acid, 3-m

ethyl-

2-Pro

penal, 3-p

henyl-

Hexanoic

acid

Benzyl A

lcohol

2 mL

4 mL

6 mL

8 mL

10 mL

Figure 6. Effect of volume optimization.

Figure 7. Overlay chromatogram of heating and without heating.

10 20 30 40 50 60 70 80 90 1000

2

4

6

8

10

12

14

16

18HeatingNo heating

×107A

bund

ance

7

Figure 8. Target deconvolution tool for wine volatile screening.

Figure 9. Compound at a glance for filtering out qualifying peaks.

The Compounds at a Glance feature of Agilent MassHunterSoftware (Figure 9) quickly differentiates matching the com-pounds analyzed in the respective batch, based on qualifier,purity, and retention time.

8

Practical application of the method to the realwine samplesThe method was applied to different Indian wine samples,and gave satisfactory performance for volatiles analysis.Tentative identification and quantification of volatile com-pounds was done using targeted deconvolution software byputting a match factor minimum of 50 % against a self-gener-ated in-house library for 370 volatile compounds. Tentativeconcentration of the identified compounds was calculatedagainst the concentration of 3-octanol.

Conclusion

The volatiles profiling of Indian wines was done usingGC/MS-TD. In addition, for the first time, step-by-step opti-mization of extraction and thermal desorption GC/MS para-meters was accomplished and reported for the analysis oflarge numbers of volatile compounds including compoundswith low abundance, such as pyrazines. With the help of thisoptimized method and targeted deconvolution software,225 compounds could be tentatively identified from 15 differ-ent wine varieties. The method seems to be more efficientthan headspace and HS-SPME analysis, in terms of number ofcompounds identified and the comparatively lower cost of thetechnique. With the help of target deconvolution software,compounds with complex coelutions could also be satisfacto-rily identified. Thus, the method described in this paper is abetter option than other currently used methods. Based onthe analysis of wines for volatiles profiling and statisticalanalysis of the results, different wines could be easily distinguished from each other.

Figure 10. Radar diagram representing composition of volatile compound classes in different Indian wines.

Sauvignon blanc

Chenin blanc

Cabernet sauvignon

Shiraz

Viognier

Merlot

Acids

Alcohol

Aldehydes

Esters

Ethers

Ketones

Phenols

Pyran

Sulfur

Terpenes

Cabernet and shiraz blend

ZinfandelRiesling

Chardonnay

Blush

Gewurztraminer

Cinsuat

Colombard

Muscat

9

References

1. M. P. Martiä, et al. Agric. Food Chem. 51, 7861 (2003).

2. M. Esti, P. Tamborra. Analytica Chimica Acta 563, 173(2006).

3. M. A. Gonzalez-Viiias, et al. Food Chemistry 56, 399 (1996).

4. Z. Piñeiro, M. Palma, C. G. Barroso. Analytica ChimicaActa 513, 209 (2004).

5. M. Ortega-Heras, M. L. González-SanJosé, S. Beltrán.Analytica Chimica Acta 458, 85 (2002).

6. A Zalacain, et al. Talanta 71, 1610 (2007).

7. E. Sánchez-Palomo, M. C. Díaz-Maroto, M. S. Pérez-Coello.Talanta 66, 1152 (2005).

8. M. E. O. Mamede, G. M. Pastore. Food Chemistry 96, 586(2006).

For More Information

These data represent typical results. For more information onour products and services, visit our Web site atwww.agilent.com/chem.

www.agilent.com/chem

Agilent shall not be liable for errors contained herein or for incidental or consequentialdamages in connection with the furnishing, performance, or use of this material.

Information, descriptions, and specifications in this publication are subject to changewithout notice.

© Agilent Technologies, Inc., 2015Printed in the USAOctober 29, 20155991-5680EN