Accurate determination of 2,4,6-trichloroanisole in wines at low parts per trillion by solid-phase microextraction followed by GC-ECD

Article abstract of DOI:10.1021/jf0211682

A headspace solid-phase microextraction (HS-SPME) procedure at 30 °C with a 100 μm PDMS fiber of a saturated NaCl solution stirred at 1100 rpm combined to GC-ECD for the 2,4,6-trichloroanisol (TCA) determination in wines has been developed. Due to the matrix complexity and ethanol absorption into the fiber, the internal standard selection was crucial to obtain unbiased results. Thus, matrix effects were observed when analyzing different types of Spanish wines (white, early, and vintage red wines) spiked with TCA at low concentration levels (i.e., < 40 ng L^-1). In contrast, the use of 2,4,6-tribromoanisole (TBA) as internal standard overcame these matrix effects, whereas the use of 2,4,6-trichlorophenyl ethyl ether led to inconsistent results. The developed HS-SPME-GC-ECD methodology reaches a limit of quantitation for TCA in wine within 2.9-18 ng L^-1, with a relative standard deviation of 2.5-13.4%, depending on the TCA concentration level and wine characteristics. This analytical method is comparable to the existing methodologies based on HS-SPME followed by GC-MS in terms of accuracy, precision, length of determination, and length of quantification; however, analysis cost is reduced.

Full text of DOI:10.1021/jf0211682

J. Agric. Food Chem. 2003, 51, 3509−3514 3509
Accurate Determination of 2,4,6-Trichloroanisole in Wines at
Low Parts Per Trillion by Solid-Phase Microextraction Followed
by GC-ECD
ROBERTO ALZAGA,*,† LAURA ORTIZ,† FRANCISCO SAÄ NCHEZ-BAEZA,‡
†
M.-PILAR MARCO,‡ AND JOSEP MARIA BAYONA
Environmental Chemistry Department and Organic Biological Department, IIQAB-CSIC,
08034-Barcelona, Spain
A headspace solid-phase microextraction (HS-SPME) procedure at 30 °C with a 100 µm PDMS fiber
of a saturated NaCl solution stirred at 1100 rpm combined to GC-ECD for the 2,4,6-trichloroanisol
(TCA) determination in wines has been developed. Due to the matrix complexity and ethanol absorption
into the fiber, the internal standard selection was crucial to obtain unbiased results. Thus, matrix
effects were observed when analyzing different types of Spanish wines (white, early, and vintage
red wines) spiked with TCA at low concentration levels (i.e., <40 ng L-1). In contrast, the use of
2,4,6-tribromoanisole (TBA) as internal standard overcame these matrix effects, whereas the use of
2,4,6-trichlorophenyl ethyl ether led to inconsistent results. The developed HS-SPME-GC-ECD
methodology reaches a limit of quantitation for TCA in wine within 2.9-18 ng L-1, with a relative
standard deviation of 2.5-13.4%, depending on the TCA concentration level and wine characteristics.
This analytical method is comparable to the existing methodologies based on HS-SPME followed by
GC-MS in terms of accuracy, precision, length of determination, and length of quantification; however,
analysis cost is reduced.
KEYWORDS: SPME-headspace; TCA; GC-ECD; internal standard selection; wine; matrix effect; musty
odor
INTRODUCTION
show that there is a great deal of uncertainty surrounding the
setting of sensory limit values. In addition, not only TCA can
contribute to give an unpleasant odor/flavor to wine. The
presence of other compounds, such as guaiacol, 1-octen-3-ona,
1-octen-3-ol, geosmine, and 2-methylisoborneol can also pro-
duce T&O problems to wine, but TCA is one of the most
common occurrences (2).
A serious and highly costly problem related to a moldy-musty
off-flavor called “cork taint” or “moldy taint” is raising the
attention of wine and cork industries. Buser et al. (1) were the
first authors to report the correlation of cork taint in wine with
the presence of 2,4,6-trichloroanisole (TCA). However, not all
the TCA present in wine can arise from cork, other factors, such
us transportation, storage, and handling can increase the TCA
concentration in the final product (2). Moreover, not all TCA
present in the cork can contaminate wine, because only a fraction
of the total TCA is releasable (3).
TCA possess very low taste and odor (T&O) thresholds in
wine, but the concentration to determine a defect is dependent
on wine characteristics and composition. Although the percep-
tion level for TCA is in the range of 0.03-10.0 ng L-1 (4, 7),
the TCA concentration considered to produce a defect in wine
is higher, ranging from 10 to 40 ng L-1 (2). This sensory
threshold can be affected by different factors, such as wine type
(white, red, etc.), and it is closely related to the sensor capacity
of the expert taster. Moreover, the number of organoleptic assays
that the expert taster can do per day is limited. All these factors
Different analytical methodologies have been developed to
determine TCA presence in corks and/or wine (4). Because of
its volatility, gas chromatography (GC) is the most suitable
technique for TCA determination, coupled to a mass spectrom-
eter detector (MSD) at selective ion monitoring mode (7-9,
13-15) or to an electron capture detector (ECD) (16). However,
due to low target TCA concentration level in wine, a precon-
centration step is required. Liquid-liquid extraction (LLE) (5,
17) and solid-phase extraction (SPE) with different sorbents,
such as C18 (11), Ambersorb 512 (18), silica gel/ENVI-Carb
(19), have been tested. The drawbacks of these methodologies
are the use of solvents, matrix dependent recoveries in cases of
complex matrixes such wines, and poor selectivity. Solid-phase
microextraction (SPME), because is a solventless, cost-effective,
and provides high sensitivity, can be an alternative to conven-
tional extraction procedures. Although headspace solid-phase
microextraction (HS-SPME) has been successfully applied to
TCA analysis in water and wines (7-9, 13-15), a detailed study
* To whom correspondence should be addressed. E-mail: ramqam@
iiqab.csic.es. Fax: (34) 93 2045904. Tel. (34) 93 4006100 ext 242.
† Environmental Chemistry Department.
‡ Biological Organic Department.
10.1021/jf0211682 CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/07/2003

3510 J. Agric. Food Chem., Vol. 51, No. 12, 2003
Alzaga et al.
(Teknokroma, Sant Cugat, Spain) column was used. The gas chro-
matographic conditions were as follows: the initial oven temperature
was 70 °C for 2.0 min, then programmed from 70 to 250 °C at 10 °C
min-1, to a final holding time of 1.0 min. The detector temperature
was 280 °C. The injector temperature was 260 °C at the splitless mode
(2 min), and desorption time for the SPME fiber was 5 min. Ethanol
extraction profile was determined using a GC-FID (split ratio 1/20).
Data were acquired by a Nelson-PE interface with a sampling frequency
of 100 Hz and processed with a PC computer using PE 2600 software.
SPME Procedure. Before the initial analysis, the PDMS fiber (100
µm) was conditioned for 60 min at 250 °C. After the conditioning
process, a fiber blank was run to confirm fiber cleanness. Samples for
method development were prepared by adding 25 mL of wine or
ethanol/water (12% v/v) into a 40-mL glass vial, sealed with a PTFE
septum, and 270 mm × 90 mm magnetic stirring bar at 1100 rpm were
used. The extraction temperature was controlled by a water bath system,
maintained at constant temperature (30 °C). Microliter volume of
working standard solution of analytes was spiked into a vial to obtain
on the matrix effects in the TCA determination in wines has
not been carried out yet, being of primary interest for its
application in routine basis.
The objective of this work has been to develop a cost-
effective, accurate, and fast analytical methodology for the
determination of TCA in wines at low ng L-1 concentration
levels. Therefore, a HS-SPME followed by GC-ECD procedure
has been developed, with emphasis on the variables affecting
the matrix effects. Different internal standards (IS) have been
evaluated to match the TCA behavior in the HS-SPME from
ethanol/water (12% v/v) and real matrixes (i.e., white wine,
vintage red wine, and early red wine). Temperature and pH as
key variables affecting matrix-analyte interaction have been
tested. Furthermore, the experimental distribution constants of
the selected IS and TCA were measured in ethanol/water (12%)
to evaluate their affinity for the PDMS fiber.
the following concentrations: TCPEE 18.8 ng L-1, TBA 23.5 ng L-1
,
and TCA 0.1-150 ng L-1. HS-SPME was performed, avoiding any
direct contact with the sample. The sorption time profile was performed,
exposing the fiber in the HS sample for 1, 5, 10, 15, 20, 30, and 40
MATERIALS AND METHODS
Materials. The following reagents were obtained from Sigma-
Aldrich Chemie (Steinheim, Germany): 2,4,6-trichlorophenol (98%);
2,4,6-trichloroanisole (TCA, 99%); 2,4,6-tribromoanisole (TBA, 99%);
and 1,3,5-trichlorobenzene (TCB, 99%). The 2,4,6-trichlorophenylethyl
ether (TCPEE) was prepared as described below. Water (HPLC grade),
toluene, ethyl acetate, hexane, acetone, and methanol were obtained
from Merck for trace organic grade (Darmstadt, Germany). Sodium
chloride (for analysis) and absolute ethanol were obtained from Carlo
Erba (Milan, Italy). K2CO3 anhydride was obtained from Panreac
Qu´ımica (Barcelona, Spain). Ethyl iodide (>99%) and 2,3,6-trichlo-
rotoluene (TCT, 87.6%) were obtained from Fluka Chemie GmbH
(Buchs, Switzerland). Silica gel was obtained from SDS (Marseille,
France). Poly(dimethylsiloxane) (PDMS, 100 µm) fiber was obtained
from Supelco (Bellefonte, PA).
min (TCB ) 250 ng L-1, TCT ) 189.6 ng L-1, TCA ) 133.3 ng L-1
,
TCPEE ) 60.19 ng L-1, TBA ) 75.3 ng L-1).
Desorption times were evaluated at 2, 5, and 15 min. NaCl addition
was evaluated at 0, 25, 50, and 100% of the saturated concentration at
25 °C. The linearity was evaluated from 0.1 to 150 ppt (ng L-1) for
TCA. Detection (LOD) and quantitation (LOQ) limits were calculated
from low concentration value of the calibration curves, by considering
the peak area corresponding from three (LOD ) 3σ) to 10 (LOQ )
10σ) times the signal-to-noise ratio of a procedural blank.
According to previous reports, the selected fiber was PDMS 100
µm (7, 8, 16). Salt addition, high stirring rate, and mild temperatures
(20-30 °C) are relevant variables affecting to the TCA extraction (8)
in wine. Thus, the HS-SPME extraction conditions adopted were as
follows: stirring rate 1100 rpm, saturated with NaCl, extraction
temperature 30 °C, sample volume 25 mL (40 mL of vial sample),
extraction time 20 min, and desorption time 2 min at splitless mode.
The sample was allowed five minutes of equilibration time before the
SPME analysis. No carry-over was detected, and low concentration
levels (very low ng L-1) were able to be determined.
TCA, TCPEE, TCT, TCB, and TBA stock standard solutions were
diluted in toluene to prepare a working standard solution of 1883 µg
g-1, 1725 µg g-1, 2344 µg g-1, 2000 µg g-1, and 1490 µg g-1
,
respectively. Standard working solutions of TCA (8.4 and 42.2 ng g-1),
TCPEE (24.5 ng g-1), and TBA (29.8 ng g-1) were prepared by diluting
stock standard solution in methanol. Stock and working solutions were
stored at 4 °C in the dark. Different (white, red early, and red vintage)
commercial Spanish wines (13 samples) were analyzed.
2,4,6-Trichlorophenylethyl Ether (TCPEE) Synthesis. To a
suspension of anhydrous K2CO3 (700 mg) in dry acetone (5 mL), 2,4,6-
trichlorophenol (200 mg, 1.01 mmol) and iodoethane (160 µL, 312
mg, 2 mmol) were sequentially added. After stirring the mixture for
24 h at room temperature (c.a. 25 °C), the solid was filtered off, and
the filtrate was evaporated under slight vacuum. The residue (310 mg)
was purified by column chromatography over silica gel (30 g, 40-63
µm, 60 Å), eluting with hexane/ethyl ether (80:20) to obtain the target
compound as a slightly yellow solid (155 mg, 68% of the theoretical
yield) with purity better than 99.9% by GC-ECD. In addition, the
compound identity was confirmed by proton 1H NMR (δ, CDCl3) 7.30
(s, 2H), 4.09 (q, J ) 7 Hz,2H), 1.46 (t, J ) 7 Hz,3H) ppm and 13C
NMR (δ, CDCl3) 150.65 (s), 130.18 (s), 129.24 (s), 128,68 (d), 69.81
(t), 15.42 (q) ppm. For the synthesis of the internal standard, thin-
layer chromatography (TLC) was performed on 0.25-mm, precoated
silica gel 60 F254, aluminum sheets (Merck, Darmstadt, Germany).
1H and 13C NMR spectra were obtained with a Varian Unity-300 (Varian
RESULTS AND DISCUSSION
Internal Standard Evaluation. The use of IS for SPME is
strongly recommended to improve the accuracy and precision
of the analytical procedure (20). Obviously, perdeuterated
surrogates are the best option, because physicochemical proper-
ties are closely related to those of the analytes. However,
deuterated surrogates are very expensive, and MS is mandatory
when deuterated IS’s are employed. In the case of TCA, its
synthesis (7) or its purchase is demanded. MS is a highly
sensitive and selective detector, especially when selective ion
monitoring (SIM) mode is employed, but ECD can compete or
even improve MS in terms of sensitivity and cost-effective
analysis. IS selection for ECD has to be a halogenated compound
with similar behavior to TCA. In this work, four different
halogenated compounds as IS have been tested: TCB, TCT,
TBA, and TCPEE (see Table 1). These compounds exhibit high
response factors in the ECD; however, their behavior by HS-
SPME-ECD, using the PDMS 100 µm fiber, was different,
regarding the matrix effect. As shown in Figure 1, TCB and
TCT reach the equilibrium conditions in 10-15 min, but when
the extraction time is increased, the TCB and TCT extracted
amount decreases. This effect can be explained by the chemical
properties of TCB/TCT and the matrix evaluated. TCB and TCT
are nonpolar compounds and possess a high affinity for the
PDMS fiber. However, because of the high amount of ethanol
present in wine (i.e., 11-14%), a small fraction of ethanol
1
Inc., Palo Alto, CA) spectrometer (300 MHz for H and 75 MHz for
13C). Theoretical calculations regarding pKa values were carried out
using the ACD/pKa 1.2 software (Advanced Chemistry Development
Inc., Toronto, ON) at the Department of Analytical Chemistry
(University of Lund, Sweden).
Instruments and Apparatus. The SPME holding device was
purchased from Supelco (Barcelona, Spain). GC-ECD determination
of TCA was carried out using an HRGC 5300 Mega series chromato-
graph (Carlo Erba Instruments, Milan, Italy) with an ECD 800 (Fisons
Instruments, Milan, Italy) at 310 °C of body temperature, pulse
amplitude of 50 V, current of 1 nA, and pulse width of 1µs. A Tracer
Meta X5 (30 m, 0.25 mm i.d.) coated with a 0.25 µm film thickness

TCA Determination in Wines by SPME-GC-ECD
J. Agric. Food Chem., Vol. 51, No. 12, 2003 3511
Table 1. Physicochemical Properties of the Compounds Evaluated
solubilitya
(g L-1)b
log k′
(RSD%)c
compound
abbrev
log pa
2,4,6-trichloroanisole
2,4,6-trichlorophenyl
ethyl ether
2,4,6-tribromoanisole
1,3,5-trichlorobenzene
2,3,6-trichlorotoluene
TCA
TCPEE
5.4 × 10-3
22 × 10-3
3.97 ± 0.34
4.5 ± 0.34
3.93 (3.5)
4.33 (2.4)
TBA
TCB
TCT
32 × 10-3
16 × 10-3
4.8 × 10-3
4.14 ± 0.49
4.04 ± 0.31
4.5 ± 0.32
4.03 (2.3)
d
ND
Figure 2. HS-SPME-GC-ECD chromatograms of wine matrixes (A: blank)
with low/high (B/C) TCA interaction. Spiked level 20 ng L-1 of TCA (1),
ND
-1
40 ng L-1 TCPEE (2) and 25 ng L of TBA (3).
a Calculated at the Analytical Chemistry Department (Lund, Sweden) Advanced
Chemistry Development log P program. b Solubility at 25 °C in water. c n ) 8,
experimental condition in the text. d ND ) not determined.
is possible to calculate the distribution constant (K) of the TCA
and TCPEE at the experimental condition established, with the
following equation (20):
nVs
K′ )
Vf(CoVs - n)
where the fiber volume (Vf) is 3.57 × 10-4 mL, the initial
concentration (Co) 50 ng L-1, sample volume (Vs) is 25 mL
(ethanol/water 12% v/v, NaCl saturated) and the amount
extracted into the fiber (n) was determined by direct injection
(GC-ECD) of the calibration curve standards of TCA, TBA,
and TCPEE. The log experimental K′ (logK′
)
TCA,TBA&TCPEE
values were 3.93 with an RSD of 3.5% (n ) 8) for TCA, 4.03
with an RSD of 2.3% (n ) 8) for TBA and 4.33 with an
RSD of 2.4% (n ) 8) for TCPEE (see Table 1). The
logK′TCA,TBA&TCPEE values show a high affinity of these ana-
lytes for the PDMS 100 µm fiber. More than 15% of the total
TCA, TCB, and TCPEE present in the sample were extracted
into the fiber, allowing very low trace level TCA analysis in
wine. TCPEE shows a higher experimental K′ value than TCA.
The introduction of a methylene group in the TCPEE decreases
the polarity of this compound compared to TCA, and this effect
is reflected on its log K′ value.
Variables Affected by the Matrix Effect. TCPEE and TBA
exhibited a similar HS-SPME behavior to that of TCA in
ethanol/water (12%v/v). However, a broad spectrum of com-
pounds occurring in wine can interact with TCA. To test TCPEE
and TBA as internal standard candidates for TCA analysis in
wines, 13 different Spanish wines, free of TCA, were spiked at
the 20 ng L-1 level and analyzed using the external calibration
(EC) and the internal calibration (IC) methods. The TCA
calibration curve was prepared in ethanol/water (12% v/v to
mimic the wine matrix). From the results shown in Figure 3, it
can be observed that every wine exhibited a different behavior
during TCA analysis, independent of the wine nature (white,
early red, and vintage red wine). Each wine is unique, and this
effect is reflected in the results shown in Figure 3. The use of
internal standard is mandatory to improve accuracy and preci-
sion. Thus, when the EC method was applied, low TCA
recoveries were determined, assuming all the samples analyzed
as a whole (4.0 ng L-1, RSD 80.2%). Due to the intrinsic TCA
properties, we assumed that at low concentrations, most of the
Figure 1. Extraction time profile of TCA, TCB, TCT, TCPEE, TBA and
ethanol in ethanol/water (12% v/v). PDMS fiber (100 µm), extraction
temperature 30 °C, NaCl saturated, sample volume 20 mL, vial volume
40 mL, and stirring rate 1100 rpm.
present in the matrix is absorbed into the PDMS affecting the
TCB and TCT extraction time profiles and diminishing the
amount extracted with time (see Figure 1). The influence of
ethanol concentration on the SPME extraction of TCA has been
evaluated previously (16). When ethanol concentration is
increased, TCA recovery decreases. The use of a suitable internal
standard as TBA or TCPEE compensates this effect; however,
neither TCT nor TCB is not able to.
This behavior prompted us on a replacement of TCB and
TCT as IS, because of their effect of worsening the precision
of the analytical methodology. Thus, we had proven that the
extraction time profile of TCPEE and TBA was very similar to
TCA. Both compounds reached the equilibrium conditions at
15 min, and the extraction time profile was not affected by the
presence of ethanol in the matrix (see Figure 1). Moreover, as
is shown in Figure 2, a high chromatographic resolution was
obtained, and the total GC run can be accomplished in less than
20 min. Therefore, TCPEE and TBA appeared as suitable
internal standards for TCA determination by HS-SPME-GC
combined either to ECD or MS even for samples with high
ethanol content such as wines.
Determination of Experimental K′TCA, K′
and K’
TBA
TCPEE
Values. At equilibrium conditions with PDMS 100 µm fiber, it

3512 J. Agric. Food Chem., Vol. 51, No. 12, 2003
Alzaga et al.
Table 2. Accuracy and Precision Results from the Analysis of Two
Wine Samples (high/low matrix interaction) Spiked at Different TCA
Levels Using Two Different Internal Standards (n ) 5 at each level)
wine matrix
interaction
internal
standard
spiked level measured mean recovery RSD
(ng L-1)
value (ng L-1)
(%)
(%)
low
TCPEE
TBA
10.1
76.0
20.0
66.7
10.1
73.5
22.5
69.3
100
97
113
104
1.4
2.5
2.5
1.9
high
TCPEE
TBA
20.0
66.7
200
20.0
66.7
200
5.6
48.4
191.8
16.6
71.8
208.5
28
72
96
83
108
104
11.5
3.8
1.5
13.4
3.3
3.1
Figure 3. TCA determination of different commercial Spanish wines (white,
red early, and red vintage wines) at 20 ng L-1 spiked level by EC and
IC, employing TCPEE and TBA as IS. TCA mean value concentration,
assuming the total samples analyzed as wine matrixes by EC and IC.
The analyses were performed in triplicate.
TCA is strongly interacting with the matrix (wine) components.
It could explain the low recoveries achieved by the HS-SPME
of TCB and TCT, probably because they do not interact with
the wine matrix, as TCA does. Thus, when TCB and TCT were
used as IS, the TCA determination results were similar to those
obtained by using the EC method. By using TCPEE as IS in
the IC method, the accuracy was improved, but the experimental
value obtained still underestimated the target (14 ng L-1, RSD
37%). In contrast, the use of TBA as IS afforded the best results
(21 ng L-1, RSD 10%). Although initially TCPEE seemed to
be an ideal candidate as IS in HS-SPME-GC-ECD due to the
slight physicochemical and structural differences (methoxy vs
ethoxy group) between TCA and TCPEE (see Table 1), such a
subtle variation is the only argument to explain the disparity
observed in the results obtained. Thus, TBA with a methoxy
group in the molecular structure is a much better IS candidate.
We can speculate that TCPEE interacts with the matrix to a
lower extent than TCA, leading to a drop of the method accuracy
during the TCA determination. TCPEE could be used as IS,
but calibration curves had to be prepared using the same wine
as a solvent or to perform the standard addition; however, it is
evident that analysis time would be increased.
Employing an adequate IS, matrix effect can be compensated,
but it cannot be avoided. Different attempts have been carried
out to reduce the TCA-wine interactions and to increase the
TCA recoveries. Changes in pH and extraction temperature
could potentially avoid the TCA matrix effect. However, our
results were not satisfactory. Increase of the extraction temper-
ature (i.e. 40 °C) dramatically worsened the LOQ, because of
a decrease in the K′ value. Changes in pH improved the TCA
determination (pH range 7-9) in some wines analyzed but failed
in others. At low pHs (i.e., 1-2) it was observed that the TCA-
matrix interaction was stronger than at neutral-basic pH,
probably due to an increase in matrix hydrophobicity due to
the protonated forms of acidic substances. Matrix effect was
always more evident at low TCA concentration values. Wine 4
and wine 5 (Figure 3), which were characteristic of wines
exhibiting low and high matrix effects, respectively, were spiked
at different TCA concentration levels, as shown in Table 2. At
higher TCA concentration values, the method accuracy im-
proved when TCPEE was used as IS, demonstrating a clear
evidence of matrix effect.
Figure 4. External (A) and Internal (B) calibration curves of TCA in different
matrixes (ethanol/water (12% v/v), white wine, red early wine, and red
vintage wine). TBA is used as IS.
evaluated. TCA external calibration curves of each matrix are
shown in Figure 4A. The calibration curves (with a correlation
coefficient of r ) 0.98-0.99) reflect a different sensitivity
(slope) for each matrix evaluated for TCA determination. The
sensitivity decreased from ethanol/water (12% v/v) to vintage
red wine. This trend is closely related to the complexity of the
matrix. Ethanol/water (12% v/v) was the simplest matrix
evaluated, and no interaction with TCA was expected if
compared to wine samples. White and early red wines showed
similar matrix effect based on their TCA-HS-SPME behavior;
however, small sensitivity differences were appreciated. Vintage
red wine was the most complex matrix evaluated, showing the
highest matrix effect dependence for TCA determination. This
effect shows the need to perform a calibration curve depending
on the matrix evaluated, when external calibration is carried
out. This drawback is critical, when different wine samples are
analyzed during TCA quality control, increasing the analysis
cost and time. TCA determination by HS-SPME is a matrix-
effect-dependent analytical procedure, but the use of a suitable
internal standard can minimize or avoid this effect. In Figure
4B is shown the internal calibration curve for all the matrixes
evaluated (r > 0.99). Thus, a single calibration curve can be
used, avoiding the necessity to perform a new calibration curve
each time a new matrix is evaluated. Employing an adequate
internal standard such as TBA, the matrix effect is minimized
in TCA-HS-SPME. Some authors have suggested building a
calibration curve by using a mixture of different red wines from
TCA determination by HS-SPME-GC-ECD at different
matrixes (water/ethanol and three different wines) has been

TCA Determination in Wines by SPME-GC-ECD
J. Agric. Food Chem., Vol. 51, No. 12, 2003 3513
Table 3. LOD and LOQ for TCA Determination (ng L-1) with the
Methodology Developed and Comparison with Other Analytical
Methodologies
different origins and ethanol contents, to have a “representative”
wine matrix sample when internal calibration (using TCT as
IS) is developed (16, 21). However, to our knowledge and
according to our experience, each wine sample is unique, and
therefore, analysis reliability could be questioned.
analytical
matrix
methodology
LOD
0.15
LOQ
0.5
RSD%
1.5
ref
Partitioning of volatile substances such as TCA between
liquid and gas phase is mainly governed by the aroma compound
volatility and solubility. However, these physicochemical
properties could be influenced by other wine constituents present
in the medium, such as polysaccharides, proteins and polyphe-
nols. Wine phenolic compounds originating from grapes en-
compass several structural groups, such as anthocyanins, pro-
anthocyanidins, etc. (22). Moreover, aging in oak barrels pro-
motes the extraction of low molecular weight phenolic com-
pounds, mainly ellagitannins, into wine (23). Wine polyphenols
have attracted much attention, because of their ability to interact
with proteins as well as aroma substances such as isoamyl
acetate, ethyl acetate, ethyl hexanoate, and benzaldehyde (22).
Processes involving polyphenol aggregation could lead to a
significant loss of aroma compounds through intermolecular
interactions. Traditionally, wine samples are divided into two
well differentiated matrixes, white and red, when TCA deter-
minations have been developed by HS-SPME (16). White wine
is assumed to be a matrix with low TCA interaction, and red
vintage wine is asumed to be a matrix with high matrix effect.
Following the same approach, red early wine is an intermediate
matrix. This wine matrix classification is based on their poly-
phenol contents. However, our results show that the approach
regarding wines as whites or reds (low/high polyphenol contents)
might be too simple. Both white and early red wines can exhibit
a higher TCA-matrix interaction (low TCA recoveries) than
red vintage wine, as shown in Figure 3. Although the content
of red wine polyphenolic substances is higher than that for white
wines, their concentration level is at the same range (0.5-3.0
g/L), considering that it is 108 times higher than the expected
TCA concentration levels found in wines. Thus, it is probably
more the nature of these polyphenolic compounds that affects
TCA recoveries than their concentration in wine.
ethanol/water
12% v/v
HS-SPME-ECD
present
work
wine
(low interaction)
wine
(high interaction)
white varietal
wine
HS-SPME-ECD
HS-SPME-ECD
HS-SPME-MSD
LLE, GC-MSD
1.0
5.4
2.9
2.5
present
work
present
work
7
18.0
5.0
13.4
13
wine not
2−5
5−8
24
specified
white wine
red wine
HS-SPME-ECD
HS-SPME-ECD
1.0
1.0
4.0
8.0
7.8
6.1
16
16
water (12% v/v) can be used to determine the releasable TCA
in cork, instead of wine, and then to apply the analytical
methodology developed, improving their determination.
Two well-characterized wine samples were spiked with TCA
at two concentration levels (20 and 67 ng L-1) and, applying
the analytical procedure developed, the within-day assay repeat-
ability RSD value (n ) 5) found was 2.5 and 1.9% for the wine
with low TCA-matrix interaction and 13.4 and 3.3% for the
one with high TCA-matrix interaction. The interday repeat-
ability RSD (n ) 10) values were at the same range. TCA
determination by HS-SPME-GC-ECD improves the analytical
parameters in terms of precision (RSD%), LOD, and LOQ
compared to their equivalent methodology employing MSD (7)
as is shown in Table 3. HS-SPME-GC-ECD avoids exhaustive
extraction, cleanup step, solvents, etc., significantly diminishing
the total analysis time, compared to other conventional methods
(1, 12, 24). The same PDMS 100 µm fiber was used to perform
more than 100 analyses without any significant damage.
In summary, HS-SPME-GC is a suitable technique for TCA
determination in wines, provided that an appropriate internal
standard is used. Among the compounds evaluated, only TBA
exhibited a satisfactory performance as internal standard,
because its HS-SPME behavior is similar to that of TCA in the
wine samples evaluated. LOQ and RSD depend on the wine
evaluated (high/low TCA-matrix interaction) and are indepen-
dent of wine class (white/red wine), with a range of 2.9-18 ng
L-1 and 2.5-13.4%, respectively. Although mass spectrometry
is the detector selected for most of the TCA determination in
wines, our work demonstrate that a selective, low cost, and
reliable detector such as ECD can be used, providing similar
accuracy, precision, LOD’s, and LOQ’s, but significantly
reducing the total analysis cost.
Method Evaluation. Once the HS-SPME-GC-ECD proce-
dure for TCA determination was optimized, the analytical
parameters (LOD, LOQ, accuracy, precision, linearity, repro-
ducibility, and repeatability) were evaluated. The linearity of
the response was checked in the range of 0.1 to 150 ng L-1 for
TCA, showing a linearity range (internal standard calibration)
with an excellent correlation coefficient (r > 0.995). LOD, LOQ,
and precision values are affected, depending on the wine
characteristic (high/low TCA interaction, low/high TCA recov-
eries) analyzed. Figure 2 shows a chromatogram with different
TCA responses at two very different wine samples (low/high
matrix interaction). Table 3 shows the LOQ and LOD for TCA
in ethanol/water (12% v/v) and in two well-characterized wines,
depending on TCA-matrix effect (the lowest and the highest
TCA-matrix interaction among the 13 commercial Spanish
wines evaluated). All matrixes evaluated were used as solvent
for calibration curves, and TCA no trace was found in the
procedural blank. However, noise level is increased when wine
samples are evaluated (matrix complexity) and different TCA
levels are obtained (Table 3). This fact is reflected on the LOD’s
and LOQ’s reached.
LITERATURE CITED
(1) Buser, H.-R.; Zanier, C.; Tanner, H. Identification of 2,4,6-
Trichloroanisole as a Potent compound causing cork taint in
wine. J. Agric. Food Chem. 1982, 30, 359-362.
(2) Silva, P. C.; Figuereido, M., J. J.; San Roma˜o, M. V. Cork taint
in wine: Scientific knowledge and public perception-A critical
review. Crit. ReV. Microbiol. 2000, 26, 147-162.
(3) Herve´, E.; Price, S.; Burns, G.; Weber, P. ASEV Annual Meeting
1999, Reno, NV.
(4) Ribe´reau-Gayon, P.; Glories, Y.; Maujean, A.; Dubourdieu, D.
Traite´ d’Oneologie. 2-Chimie du Vin. Stabilisation et traitements;
Dunod: Paris, 1998; p 519.
(5) Bayonove, C.; Leroy, F. Detection of chlorophenols in wines
and corks. Ind. BeVande 1994, 23, 231-37, 242.
Ethanol/water (12% v/v) matrix allows the lowest LOD (0.5
ng L-1) among the matrixes evaluated. Releasable TCA is
defined as the concentration of TCA in a cork soak after it
reaches equilibrium. Only a small portion of the cork’s TCA is
transferred to a soak solution (less than 1%). Therefore, ethanol/

3514 J. Agric. Food Chem., Vol. 51, No. 12, 2003
Alzaga et al.
(6) Sponholz, W. R.; Grossmann, M. K.; Muno, H.; Hoffmann, A.
The distribution of chlorophenols and chloroanisoles in cork and
microbiological method to prevent their formation. Ind. BeVande
1997, 26, 602-607.
(7) Evans, T. J.; Butzke, C. E.; Ebeler, S. E. Analysis of 2,4,6-
trichloroanisole in wines using solid-phase microextraction
coupled to gas chromatography-mass spectrometry. J. Chro-
matogr. A 1997, 786, 293-298.
(8) Fischer, C.; Fischer, U. Analysis of Cork Taint in Wine and Cork
Material at Olfactory Subthreshold Levels by Solid-Phase
Microextraction. J. Agric. Food Chem. 1997, 45, 1995-1997.
(9) Butzke, C. E.; Evans, T. J.; Ebeler, S. E. Detection of cork
taint in wine using automated solid-phase microextraction in
combination with GC/MS-SIM. ACS Symp. Ser. 1998, 714, 208-
216.
extraction and gas chromatography-electron-capture detection.
J. Chromatogr. A 2002, 977, 1-8.
(17) Lin, M. S.; Him, V.; Mazur, Y. L. A liquid/liquid extraction
procedure for taste and odor compound analysis. Proc. Water
Quality Technol. Conf. 1997, 2A1/1-2A1/12.
(18) Palmentier, J. P. F. P.; Taguchi, V. Y. The determination of six
taste and odour compounds in water using Ambersorb 572 and
high-resolution mass spectrometry. Analyst (Cambridge, U.K.)
2001, 126, 840-845.
(19) Fuehrer, U.; Deissler, A.; Ballschmiter, K. Determination of
biogenic halogenated methyl-phenyl ethers (halogenated anisoles)
in the picogram m-3 range in air. Fresenius’ J. Anal. Chem. 1996,
354, 333-43.
(20) Pawliszyn, J. Solid-Phase Microextraction. Theory and Practice;
Wiley-VCH: New York, 1997.
(10) Taylor, M. K.; Young, T. M.; Butzke, C. E.; Ebeler, S. E.
Supercritical Fluid Extraction of 2,4,6-Trichloroanisole from
Cork Stoppers. J. Agric. Food Chem. 2000, 48, 2208-2211.
(11) Soleas, G. J.; Yan, J.; Seaver, T.; Goldberg, D. M. Method for
the Gas Chromatographic assay with Mass Selective detection
of trichloro compounds in corks and wines applied to elucidate
the potential cause of cork taint. J. Agric. Food Chem. 2002,
50, 1032-1039.
(12) Juanola, R.; Subira´, D.; Salvado´, V.; Garc´ıa Regueiro, J. A.;
Antico´, E. Evaluation of an extraction method in the determi-
nation of the 2,4,6-trichloroanisole content of tainted cork. J.
Chromatogr. A 2002, 953, 207-214.
(21) Mestres, M.; Busto, O.; Guasch, J. Application of headspace
solid-phase microextraction to the determination of sulphur
compounds with low volatility in wines. J. Chromatogr. A 2002,
945, 211-219.
(22) Dufour, C.; Bayonove, C. Interactions between wine polyphenols
and aroma substances. An insight at the molecular level. J. Agric.
Food Chem. 1999, 47, 678-684.
(23) Moutounet, M.; Rabier, P.; Puech, J. L.; Verette, E.; Barrillere,
J. M. Analysis by HPLC of extractable substances in oak wood.
Application to a Chardonnay wine. Sci. Aliments 1989, 9, 35-
51.
(24) Whitfield, F. B.; Shaw, K. J.; Nguyen, T. H. L. Simultaneous
determination of 2,4,6-trichloroanisole, 2,3,4,6-tetrachloroanisole
and pentachloroanisole in foods and packaging materials by high-
resolution gas chromatography-multiple ion monitoring-mass
spectrometry. J. Sci. Food Agric. 1986, 37, 85-96.
(13) Giardina, M.; Olesik, S. V. Application of low-temperature glassy
carbon films in solid-phase microextraction. Anal. Chem. 2001,
73, 5841-5851.
(14) Li, Y.; George, J. E., III; Thoma, J. J.; Hansen, E. M. A broad
spectrum taste and odor analysis using solid-phase microextrac-
tion (SPME) and ion trap detection with selected ion storage
(SIS). Proc. Water Quality Technol. Conf. 2000, 690-700.
(15) Graham, D.; Hayes, K. A new method for analyzing off flavors
in drinking water. Water (Artarmon, Aust.) 1998, 25, 24.
(16) Riu, M.; Mestres, M.; Busto, O.; Guasch, J. Determination of
2,4,6-trichloroanisole in wines by headspace solid-phase micro-
Received for review November 27, 2002. Revised manuscript received
March 10, 2003. Accepted March 14, 2003. Research funding was
obtained from the I.N.I.A of the Spanish Ministry of Science and
Technology (VIN00-053-C3-2).
JF0211682