Analytica Chimica Acta 583 (2007) 1522

Experimental design for extraction and quantication of phenolic

compounds and organic acids in white Vinho Verde grapes
M.S. Dopico-Garca, P. Valent ao, L. Guerra, P.B. Andrade, R.M. Seabra

REQUIMTEServi co de Farmacognosia, Faculdade de Farm acia, Universidade do Porto, R. Anbal Cunha 164, 4050-047 Porto, Portugal
Received 25 July 2006; received in revised form 25 September 2006; accepted 29 September 2006
Available online 4 October 2006
Abstract
An experimental design was applied for the optimization of extraction and clean-up processes of phenolic compounds and organic acids from
white Vinho Verde grapes. The developed analytical method consisted in two steps: rst a solidliquid extraction of both phenolic compounds and
organic acids and then a clean-up step using solid-phase extraction (SPE). Afterwards, phenolic compounds and organic acids were determined by
high-performance liquid chromatography (HPLC) coupled to a diode array detector (DAD) and HPLCUV, respectively. PlackettBurman design
was carried out to select the signicant experimental parameters affecting both the extraction and the clean-up steps. The identied and quantied
phenolic compounds were: quercetin-3-O-glucoside, quercetin-3-O-rutinoside, kaempferol-3-O-rutinoside, isorhamnetin-3-O-glucoside, quercetin,
kaempferol and epicatechin. The determined organic acids were oxalic, citric, tartaric, malic, shikimic and fumaric acids. The obtained results
showed that the most important variables were the temperature (40

C) and the solvent (acid water at pH 2 with 5% methanol) for the extraction
step and the type of sorbent (C18 non end-capped) for the clean-up step.
2006 Elsevier B.V. All rights reserved.
Keywords: White grapes; Experimental design; PlackettBurman; Phenolic compounds; Organic acids
1. Introduction
The content of phenolic compounds and organic acids is of
great importance for the organoleptic characteristics of grapes
and therefore, in the wine obtained from them [12], and it
is related with the degree of grape ripening [1,3]. Moreover,
avonoids and many other phenolic compounds have been
shown to possess antioxidant activity and benecial effects for
health in a great number of studies [45]. Vinho Verde grapes
and wines have been studied in order to determine monoter-
penic compounds [6] and anthocyanin content [7] respectively,
but, as far as we know, their content in non-coloured phe-
nolic compounds and organic acids has not been determined
yet.
Phenolic compounds and organic acids determination have
been reported for other types of grapes or winemaking by-
products and different extraction methods have been devel-
oped. For example, Palma and Taylor [8] used supercritical

Corresponding author. Tel.: +351 222078934; fax: +351 222003977.

E-mail address: rseabra@ff.up.pt (R.M. Seabra).
uid extraction to determine phenolic compounds in grape
seeds, while Palma et al. [9] and Pi neiro et al. [10] employed
pressurized-uid extraction in grapes [9] and in grapes seeds
[10], obtaining the best recoveries of phenolic compounds using
methanol instead of other organic solvents or water. Extrac-
tion of phenolic compounds from grapes or winemaking by-
products has however been generally performed with different
solvents, by ultrasonication, stirring or shaking. Different sol-
vents or solvents mixtures, generally an organic solvent and
water, have been used, such as ethylacetate:water for red grape
marcs extracts [11] and methanol:water for seeds and skins from
grapes [12]. Acidied solvents or mixtures of solvents have also
been used like methanol [13] and ethanol:water [14] for pomace
from red and white grapes, or methanol:water for grapes com-
ponents (skin, seed and pulp) [15].
The determination of organic acids in grapes is less usual.
Palma and Barroso [3] developed an analytical method to
extract tartaric and malic acids fromgrapes and winemaking by-
products using ultrasound-assisted extraction, obtaining the best
results employing water at 70

C. Organic acids have been more

studied in grape juices and wines [2], for which Keremet al. [16]
proposed a reversed-phase liquid chromatographyultraviolet
0003-2670/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2006.09.056
16 M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522
method to simultaneously determine organic acids and phenolic
compounds in red wine and must.
Instead of using a traditional strategy of optimizing each vari-
able separately, experimental design is often used to select the
best conditions for extraction. This approach was applied in nat-
ural matrixes for extracting phenolic compounds from milled
berries [17], Citrus bergamia juice [18] and almond hulls and
pine sawdust [19]. In what concerns grapes, Pinelo et al. [20]
used it on extracting polyphenolic compounds (gallic acid, cat-
echin, epicatechin and quercetin) from their residue and Palma
and Barroso [3] employed it on the determination of tartaric and
malic acids.
The aim of this work was to develop an analytical methodol-
ogy that allowed the chemical characterization of white grapes,
namely Vinho Verde ones, by simultaneously determining
their most important phenolic compounds and organic acids.
The developed method combines solidliquid extraction of
both phenolic compounds and organic acids from grapes and
a subsequent clean-up step using solid-phase extraction (SPE).
Later, the studied compounds were determined by using high-
performance liquid chromatography (HPLC) coupled to a diode
array detector (DAD) for phenolic compounds and to an UV
detector for organic acids. PlackettBurman design was used as
a screening method in order to select the variables that have inu-
ence on the system, both in the extraction and in the clean-up
steps.
2. Experimental
2.1. Samples
Samples of Pedern a white Vinho Verde grapes were col-
lected fromQuinta da Facha, Ponte de Lima (North Portugal), in
September 2005, and immediately stored at 20

C and freeze-
dried in a Labconco 4.5 apparatus (Kansas City, MO).
2.2. Reagents and solvents
Methanol, ethanol and formic acid were obtained fromMerck
(Darmstadt, Germany), hydrochloric acid was obtained from
Pronalab (Lisboa, Portugal) and sulphuric acid was obtained
from Fluka Sigma-Aldrich (Seetze, Germany). The water was
treated in a Milli-Q water purication system (Millipore, Bed-
ford, MA). Ultrasonic bath was Bondelin Sonorex equipment.
The studied phenolic compounds and the organic acids were
obtained from the following sources: oxalic, citric, fumaric,
l()malic, ()shikimic and dl-tartaric acids, quercetin and
quercetin-3-O-glucoside from Sigma-Aldrich (Steinheim, Ger-
many), ()epicatechin, kaempferol, kaempferol-3-O-rutinoside
and quercetin-3-O-rutinoside from Extrasynth ese (Genay,
France).
2.3. Solid-phase extraction columns
The chromabond C18 non end-capped (NEC) and end-
capped (EC) columns (50 m particle size, 60

A porosity; 10 g
sorbent mass/70 mL reservoir volume) were purchased from
Macherey-Nagel (D uren, Germany). C18 EC columns present
a sorbent structure of C18 silane and a trimethyl silyl group,
covalently bonded to the surface of the silica particle, which
reduce the polar secondary interactions associated with surface
silanol groups. These columns exhibit non-polar retention mech-
anism. C18 NEC columns present a sorbent structure only of
C18 silane covalently bonded to the surface of the silica parti-
cle, which provide additional polar interactions associated with
surface silanol groups. These columns exhibit non-polar, polar
and cation exchange retention mechanisms.
2.4. Solidliquid extraction and solid-phase extraction
One gram of sample was accurately weighed and mixed
with solvent (n 50 mL) at controlled temperature in ultrasonic
bath or using stirring. The extracts were ltered and collected
together. The organic acids fractions were obtained according
to a described procedure [21], with some changes: when the
solvent used was pure methanol it was concentrated to dryness
under reduced pressure (30

C) and the residue was redissolved

in acid water (pH 2 with HCl) (50 mL); when the solvent
was acid water or a mixture of acid water and methanol, it was
used directly. The aqueous solution was then passed through a
Chromabond C18 NEC or EC column, previously conditioned
with 30 mL of methanol and 70 mL of acid water (pH 2 with
HCl). The aqueous extract, containing the organic acids, was
evaporated until dryness under reduced pressure (30

C) and
redissolved in 0.01 Nsulphuric acid (1 mL) for HPLCUVanal-
ysis (20 L).
After the elution of organic acids and other polar com-
pounds with the aqueous solvent, the retained phenolic fraction
was eluted with an organic solvent (methanol or ethanol). The
extracts were concentrated until dryness under reduced pressure
(30

C) and redissolved in 0.01 N methanol (1 mL), of which

20 L were analysed by HPLCDAD.
2.5. HPLCUV analysis of organic acids
The chromatographic experiments were performedonanana-
lytical HPLC unit (Gilson), as previously reported in [22]. The
analytes were separated using an ion exclusion Nucleogel Ion
300 OA (300 mm7.7 mm) column, in conjunction with a col-
umn heating device at 30

C. Elution was carried out at a solvent

ow rate of 0.2 mLmin
1
, isocratically, with 0.01 N sulphuric
acid as the mobile phase. Detection was performed with a UV
detector set at 214 nm. Each compound was identied by com-
parison of its retention time with the corresponding peak in the
standard solution. Chromatogram quantication was achieved
using a calibration plot of external standard.
2.6. HPLCDAD analysis of phenolic compounds
The extracts were analyzed on an analytical HPLC unit
(Gilson), using a Spherisorb ODS2 column (25.0 cm0.46 cm;
5 m particle size Waters, Milford, MA, USA) [22]. The sol-
vent system used was a gradient of water/formic acid (19:1)
(A) and methanol (B), starting with 5% methanol and installing
M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522 17
Fig. 1. HPLCchromatograms of the phenolic compounds extracted froma sam-
ple of white Vinho Verde grapes at 280 and 350 nm. Identity of compounds
as in Table 3.
a gradient to obtain 15% B at 3 min, 25% B at 13 min, 30%
B at 25 min, 35% B at 35 min, 45% B at 39 min, 45% B
at 42 min, 50% B at 44 min, 55% B at 47 min, 70% B at
50 min, 75% B at 56 min and 100% B at 60 min, at a sol-
vent ow rate of 0.9 mLmin
1
. Detection was achieved with
a Gilson diode array detector. The compounds in each sam-
ple were identied by comparison of their retention times
and UVvis spectra in the 200400 nm range with individ-
ual standards. Phenolic compounds quantication was achieved
using a calibration plot of external standard, at 350 nm for all
compounds except for epicatechin (280 nm). Quercetin-3-O-
glucoside and quercetin-3-O-rutinoside were quantied together
as quercetin-3-O-rutinoside, and Kaempferol-3-O-rutinoside
and isorhamnetin-3-O-glucoside were quantied together as
kaempferol-3-O-rutinoside.
2.7. Statistical treatment
Experimental design and statistical treatment of the results
were performed using Statgraphics Plus for Windows V. 4.0.
3. Results and discussion
3.1. Identication of compounds
The HPLCchromatograms obtained for phenolic compounds
and organic acids are shown in Figs. 1 and 2, respectively. Seven
phenolic compounds were identied: quercetin-3-O-glucoside
and quercetin-3-O-rutinoside, kaempferol-3-O-rutinoside and
isorhamnetin-3-glucoside, quercetin, kaempferol and epicate-
chin (Fig. 1). Other three phenolic compounds with a UVvis
Fig. 2. HPLC chromatogram of the organic acids extracted from a sample of
white Vinho Verde grapes at 214 nm. Identity of compounds as in Table 3.
spectra of phenolic acids, with retention time of 12.2, 19.9 and
26.0 min, were detected but could not be identied.
The HPLCUV analysis allowed the identication of six
organic acids: oxalic, citric, tartaric, malic, shikimic and fumaric
acids. The chromatographic peak between tartaric and malic
acids was identied as fructose (Fig. 2).
3.2. Experimental design
Several analytical methods previously used for determining
phenolic and organic acids in different natural matrixes, such as
chantarelle mushrooms [23], Porto wine grapes [24] and quince
[25], were the basis for the study of the best extractive con-
ditions of these compounds in white Vinho Verde grapes.
The optimization of the phenolic compounds and organic acids
extraction from grapes, by solidliquid extraction and separa-
tion by SPE, was achieved by the application of the experimental
design.
PlackettBurman design was used as a screening method, in
order to select the variables that have inuence on the system.
Initially, eight variables were selected as potentially affecting
the extraction efciency (Table 1), with ve variables relative to
the extraction step: solvent (methanol or acidic water at pH 2),
volume of the extractive solvent, extraction time, extraction tem-
perature and type of extraction (stirring or ultrasonic energy).
Three variables relative to purication step by SPE were also
considered: type of adsorbent (C18 EC and NEC), elution sol-
Table 1
Factors and levels used in the experimental PlackettBurman design
Optimizing factors () (+)
Extractive solvent
a
Acid water (pH 2) MeOH
Extraction volume (mL) 50 250
Extraction time (min) 5 20
Temperature (

C) 40 50
Extraction type 1 (ultrasonic energy) +1 (stirring)
Sorbent type 1 (EC) +1 (NEC)
Elution solvent
a
1 (EtOH) +1 (MeOH)
Elution volume (mL) 20 150
Dummy 1 +1
a
MeOH (methanol), EtOH (ethanol).
18 M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522
Table 2
PlackettBurman design matrix
Extractive
solvent
a
Extraction
volume (mL)
Extraction
time (min)
Temperature
(

C)
Extraction type Sorbent type Elution
solvent
a
Elution
volume (mL)
Dummy
1 MeOH 50 20 40 Ultrasonic EC MeOH 150 +1
2 MeOH 250 5 50 Ultrasonic EC EtOH 150 +1
3 Acid water 250 20 40 Stirring EC EtOH 20 +1
4 MeOH 50 20 50 Ultrasonic NEC EtOH 20 1
5 MeOH 250 5 50 Stirring EC MeOH 20 1
6 MeOH 250 20 40 Stirring NEC EtOH 150 1
7 Acid water 250 20 50 Ultrasonic NEC MeOH 20 +1
8 Acid water 50 20 50 Stirring EC MeOH 150 1
9 Acid water 50 5 50 Stirring NEC EtOH 150 +1
10 MeOH 50 5 40 Stirring NEC MeOH 20 +1
11 Acid water 250 5 40 Ultrasonic NEC MeOH 150 1
12 Acid water 50 5 40 Ultrasonic EC EtOH 20 1
a
MeOH (methanol), EtOH (ethanol).
vent and volume of elution solvent. One dummy variable was
introduced to control the experimental error [26].
Although alcoholic solvents have allowed better recoveries
than pure water to extract polyphenols from grapes and by-
products [27], acid solvents such as methanol acidied with HCl
0.1%[13], 80%ethanol acidied by 0.5%0.1 NHCl [14] or 80%
methanol in 6 N HCl [15] have been generally used. So, in this
work, the type of solvent, methanol or acid water was tested.
The chosen interval of temperature was very narrow because,
according to literature, despite the fact that high temperature
allows high recoveries in the extraction of phenolic compounds,
at the same time these compounds can be not stable above 50

C
[20]. The type of sorbent was chosen as a variable because in
a previous work [25] it was shown to be an important factor to
clean-up the extract of phenolic compounds. Because of matrix
properties, characterized by a very high content of sugars, size of
particle was not considered, but the sample was carefully ground
to guarantee the highest possible surface contact between solid
sample and extracting liquid [20]. Instead of considering the
liquidsolid ratio, amount of sample was xed as 1 g and vol-
ume of solvent was considered as a factor to test.
Afactorial design PlackettBurman 2

93/128 resolution III

was chosen to screen the relative inuence of the above men-
tioned factors in the experimental domain. The effects of the
selected nine factors were studied in 12 runs shown in Table 2.
The experiments were carried out in duplicate and the mean of
each run was considered. Responses were expressed in area per
gram of sample. The mean response (peak area divided by the
sample amount) is shown in Table 3.
Data analysis was performed with the statistical package Stat-
graphics Plus for Windows V. 4.0. The analysis of these results
produced the standardized main effect Pareto charts (P=95%)
(Fig. 3). When the effect of the least signicant variables was dis-
carded and the effect of the signicant variables and interactions
between three more signicant variables was again evaluated
[28], the standardized Pareto charts (P=95%) shown in Fig. 4
were obtained.
The obtained results showed that the most inuential vari-
ables were temperature and solvent in the extraction step, and
type of sorbent in the clean-up step. The dummy variable did
not appear as a main parameter, so the method was validated.
The most inuential effect of each variable was as follows.
Fig. 3. Pareto charts for the standardized effects in the PlackettBurman design. The vertical lines dene the 95% condence interval.
M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522 19
Table 3
PlackettBurman design results (area/amount of sample (g))
Phenolic compounds Organic acids
1 2 3 4 5 6 7 8 9 10 11
1 1.83E+8 6.98E+7 9.35E+6 3.02E+6 9.49E+7 nd 1.20E+4 6.24E+5 2.43E+5 3.32E+5 1.76E+5
2 3.39E+8 5.31E+7 2.35E+6 8.39E+5 1.39E+8 nd 1.67E+4 8.27E+5 3.13E+5 3.61E+5 2.91E+4
3 2.94E+8 9.67E+7 1.55E+7 3.82E+6 1.18E+8 5.63E+5 4.90E+4 3.29E+6 4.40E+5 4.36E+5 5.42E+4
4 3.81E+8 6.45E+7 2.09E+6 nd 1.68E+8 nd 2.75E+4 8.33E+5 4.81E+5 3.92E+5 2.12E+4
5 2.88E+8 5.23E+7 3.14E+6 7.89E+5 9.28E+7 nd 2.13E+4 8.21E+5 3.50E+5 3.72E+5 nd
6 3.07E+8 5.34E+7 2.64E+6 1.69E+6 1.41E+8 nd 4.56E+4 1.06E+6 5.02E+5 3.83E+5 2.61E+4
7 2.49E+8 4.03E+7 1.68E+6 3.91E+5 1.68E+6 3.70E+5 5.70E+4 3.17E+6 7.19E+5 4.40E+5 1.97E+5
8 2.63E+8 3.76E+7 9.41E+5 2.54E+5 8.76E+7 3.88E+5 5.20E+3 3.44E+6 4.29E+5 3.32E+5 3.72E+4
9 1.69E+8 2.10E+7 nd nd 7.70E+7 2.45E+5 4.66E+4 2.89E+6 6.20E+5 3.05E+5 4.11E+4
10 1.62E+8 2.65E+7 nd nd 7.65E+7 nd 1.87E+4 5.74E+5 5.47E+5 2.87E+5 1.28E+4
11 3.15E+8 7.40E+7 5.54E+6 1.56E+6 1.41E+8 4.00E+5 1.02E+5 3.18E+6 4.12E+5 4.56E+5 1.05E+5
12 1.47E+8 7.14E+7 2.46E+7 5.87E+6 6.88E+7 6.73E+5 nd 2.46E+6 2.60E+5 2.83E+5 nd
1: quercetin-3-O-glucoside +quercetin-3-O-rutinoside; 2: kaempferol-3-O-rutinoside +isorhamnetin-3-O-glucoside; 3: quercetin; 4: kaempferol; 5: epicatechin; 6:
oxalic acid; 7: citric acid, 8: tartaric acid; 9: malic acid; 10: shikimic acid; 11: fumaric acid; nd: not detected.
3.2.1. Extraction step
3.2.1.1. Temperature. As can be seen (Fig. 3) the tem-
perature was statistically signicant for the extraction of
quercetin (p =0.049), the pair kaempferol-3-O-rutinoside
and isorhamnetin-3-O-glucoside (p =0.024), and kaempferol
(p =0.018) (Fig. 4). The amount of extracted analyte increases
while temperature extraction decreases (negative sign), as was
expected due to degradation of phenolic compounds at high
Fig. 4. Second-order effect combined Pareto charts in the PlackettBurman design (P=95%).
20 M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522
temperatures. For the other phenolics and all organic acids, tem-
perature was not a statistically signicant factor. So, temperature
was xed at 40

C.
3.2.1.2. Type of solvent. Type of solvent was statistically sig-
nicant for oxalic (p =0.004) and tartaric (p =0.003) acids (neg-
ative sign). The best results were achieved using acid water (pH
2) as extractive solvent (Fig. 3). This factor was however also
statistically signicant for epicatechin (p =0.036, positive sign)
(Fig. 4) obtaining the best results with methanol. Therefore, this
factor was not xed at this time. Previous research carried out
in the laboratory showed that the addition of a small quantity
of methanol increased the yields of phenolic compounds in the
SPE step [25], so, new assays were carried out to determine the
most suitable type of solvent.
3.2.1.3. Type of extraction, volume of extractive solvent and
extraction time. These factors were not statistically signicant
for any compound. Only when the least important variables
were discarded, extractive volume was a statistically signicant
factor for citric (p =0.035) and shikimic (p =0.016) acids (pos-
itive sign) (Fig. 4) and it was in the limit of being statistically
signicant for the pair quercetin-3-O-glucoside and quercetin-
3-O-rutinoside (p =0.051, positive sign) (Fig. 4). The selection
of the suitable value was made according to the sign of the esti-
mated effect associated with it. So, 250 mLof extractive solvent,
20 min as extraction time and ultrasonic energy as type of extrac-
tion were chosen.
3.2.2. Separation and clean-up step by SPE
3.2.2.1. Type of sorbent. Due to the fact that phenolics are
acidic compounds, our rst choice was a C18 column, since
these sorbents provide enhanced retention of this kind of com-
pounds. Besides, they also allow the simultaneous extraction of
both organic acids and phenolics. The sorbent type was the most
inuential variable for malic acid (p =0.017) (Fig. 3) and, when
the effect of the least signicant variables was discarded, it was
also statistically signicant for citric acid (p =0.032), quercetin
(p =0.036) and kaempferol (p =0.044) (Fig. 4). Although for
organic acids the most suitable cartridge was NEC (positive
sign), for phenolic compounds the most suitable was EC (nega-
tive sign). The C18 EC cartridge has fewer silanol groups, and
therefore less polar and secondary interactions associated with
the surface than the NEC one. This would improve the retention
of phenolic compounds in C18 EC cartridge, reducing the elu-
tion with the acidied water. In what concerns organic acids they
were more retained by the C18 ECcolumn than by the C18 NEC
one, which reduces its content in the acidied water, leading to
lower recovering amounts. As the importance in the clean-up
step of the type of sorbent was greater for malic acid than for
phenolic compounds, the NEC cartridge, which improves the
elution of organic acids, was selected as sorbent.
3.2.2.2. Type and volume of elution solvent. Only the results for
phenolic compounds were considered because the organic acids
are not retained in the cartridge. Elution solvent was at the limit
to be statistically signicant for epicatechin (p =0.049, positive
sign) (Fig. 4) and the best results were obtained with ethanol. As
type of sorbent was previously xed as NEC cartridge, interac-
tions plots between type of sorbent, type of solvent and elution
volume were considered in addition to the sign of each factor,
to x the most suitable nal value. So, ethanol was chosen as
elution solvent and a 150 mL volume was selected to guarantee
the complete elution of all compounds.
The results obtained with PlackettBurman design showed
that the most suitable conditions for extracting phenolic and
organic acids fromgrapes samples were extraction by ultrasonic
energy at 40

Cwith 5 50 mL of extractive solvent and 20 min

of extraction time for each fraction of solvent. For the clean-up
step by SPE, the best results were obtained using a C18 NEC
column and 150 mL of ethanol as elution solvent to recover the
phenolic compounds.
3.3. Fine tuning for solvent
Considering the different effect that the extractive solvent had
in epicatechin and oxalic and tartaric acids, new assays were
carried out to decide the most suitable type of solvent to extract
these compounds fromgrapes. Since we were dealing with large
volume samples, the addition of methanol to acid water was
required in order to maintain an active sorbent column surface.
So, the effect of the addition of a small percentage of methanol
to acid water maintaining the pH at 2 was tested, for 0, 5, 10
and 25% methanol (n =2). The other conditions used were the
above mentioned values xed according to PlackettBurman
design. As can be seen in Fig. 5, the best results for epicatechin
Fig. 5. Recoveries (area/amount of sample (g)) obtained using acid water (pH
2) with different percentages of methanol as extractive solvent.
M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522 21
Table 4
Repeatability (n =6), detection and quantication limits of the analytical method developed for the sample of white Vinho Verde grapes
mg kg
1
lyophilized
sample
R.S.D.
a
LOD
b
(mg kg
1
lyophilized sample)
LOQ
c
(mg kg
1
lyophilized sample)
Phenolic compounds
Quercetin-3-O-glucoside +quercetin-3-O-rutinoside 288 2.8 2.5 8.4
Kaempferol-3-O-rutinoside +Isorhamnetin-3-O-glucoside 145 15 1.8 5.9
Quercetin 9.5 8.9 1.4 4.5
Kaempferol 1.0 13 0.55 1.8
Epicatechin 289 6.2 3.5 12
Organic acids
Oxalic acid 422 6.7 62 207
Citric acid 271 8.4 86 286
Tartaric acid 4733 2.2 27 90
Malic acid 1857 5.6 52 175
Shikimic acid 37 3.1 4.8 16
Fumaric acid 4.4 9.1 1.5 5.1
a
R.S.D.: relative standard deviation.
b
LOD: detection limit.
c
LOQ: quantication limit.
were obtained with 5% of methanol, with which the extractive
efciency is better and no losses during the SPE procedure are
observed. Therefore, acid water at pH2 (HCl) with 5%methanol
was xed as extractive solvent.
3.4. Final extraction and clean-up conditions
The following optimal values were adopted for the extrac-
tion of the considered organic acids and phenolic compounds:
extraction by ultrasonic energy, acid water at pH 2 (HCl)
with 5% methanol as extractive solvent with volume of
5 50 mL and 20 min of extraction time for each fraction of
solvent.
Each extract was ltered at vacuum, gathered and passed
through a C18 NEC column using 150 mL of ethanol as elution
solvent to recover the phenolic compounds.
In order to check if the extraction of phenolics from the
matrix was complete, a sixth extraction, obtained with 50 mL
of solvent, was collected, and subjected to the NaOH test. The
absence of a yellowcolour revealed that the extraction was com-
pleted. This test was also performed after SPE column elution.
Besides, to access the completeness of the extraction of phe-
nolic compounds and organic acids, after performance of the
all extractive method, the extracted matrix was subjected to
another full extraction procedure and neither of the metabolites
were detected by HPLC, which guarantees that the matrix was
depleted of those compounds.
Table 4 shows the results of precision study, and limits
of detection (LOD) and quantication (LOQ) for the opti-
mized analytical method. A repeatability study was carried
out by repeating six experiments. For calculating detection
and quantication limits of the developed analytical method,
detection and quantication limits for the chromatographic
method were rst calculated according to Miller and Miller
[29]. Calibration graphs were used so that LOD=y
B
+3 S
B
and LOQ=y
B
+10 S
B
, in which y
B
(blank signal) =a (inter-
cept of the calibration graph) and S
B
(S.D. of the blank) =S
y/x
.
These calibration graphs were built using very diluted solutions
of the compounds object of this study, corresponding to concen-
trations lower than those found in samples.
Limits of detection of the developed analytical method
were between 1.586 mg kg
1
for organic acids and 0.55
3.5 mg kg
1
for phenolic compounds, while limits of quantica-
tion ranged between 5.1286 mg kg
1
and 1.812 mg kg
1
for
organic acids and phenolic compounds, respectively.
The developed analytical method was precise since relative
standard deviation (R.S.D.) was 2.29.1% for organic acids and
2.815% for phenolic compounds (Table 4). On comparing the
concentration levels of organic acids and phenolic compounds
with the quantication limits obtained for the developed analyt-
ical method, it can be seen that the concentrations of oxalic acid,
fumaric acid and kaempferol are slightly lower than the quanti-
cation levels of the method. This explains the higher variability
values obtained for these compounds.
In order to study the recovery of the procedure, a sample
was added to known quantities of the studied analytes and was
analysed in triplicate before and after the additions. Recovery
values were between 72 and 95% for phenolic compounds and
between 72 and 103% for organic acids.
The performed analytical method enables simultaneous
extraction, from a natural matrix, of two types of compounds,
organic acids and phenolic compounds, using a simple method-
ology. A great number of variables that could be potentially
important, both in the extraction and in the separation/clean-
up steps, were tested using experimental design. Temperature
and type of solvent were the factors that showed the high-
est inuence and were xed by looking for the best value
for the two classes of compounds. Temperature and type of
solvent have also been the most inuential factors in the opti-
mization of extraction of tartaric and malic acids from grapes
carried out by Palma and Barroso [3]. In that research how-
ever, the temperature and type of solvent had been xed at
70

C and pure water, because only these two compounds were

determined.
22 M.S. Dopico-Garca et al. / Analytica Chimica Acta 583 (2007) 1522
4. Conclusions
The following conclusions can be drawn from the present
study: the optimized conditions that were xed according to
an experimental design enabled the simultaneously extraction
of two types of analytes from white grapes, phenolic com-
pounds and organic acids, with acceptable precision (lower than
15%). In addition, the most inuential variables in the developed
extraction method were the type of solvent and temperature.
Using 5% methanol at pH 2 and 40

C the best recovering rates

for the considered analytes were obtained.
Acknowledgments
The authors thank Eng. Jos e Manuel Afonso, Eng.

Oscar
Pereira and Eng. Le ao, from Direcc ao Regional de Agricul-
tura de Entre-Douro-e-Minho, for supplying samples. Sonia
Dopico is indebted to Xunta de Galicia Govern (IN 809 A
2005) and to the Fundac ao para a Ci encia e a Tecnologia
(SFRH/BPD/21757/2005) for the grants.
References
[1] S. P erez-Magari no, M.L. Gonz alez-San Jos e, J. Agric. Food Chem. 52
(2004) 1181.
[2] I. Mato, S. Su arez-Luque, J.F. Huidobro, Food Res. Int. 38 (2005) 1175.
[3] M. Palma, C.G. Barroso, Anal. Chim. Acta 458 (2002) 119.
[4] B. Halliwell, in: C. Rice-Evans (Ed.), Wake up to avonoids, International
Congress andSymposiumSeries 226, The Royal Societyof Medicine Press,
UK, 2000, p. 13.
[5] P. Zafrilla, J. Morillas, J. Mulero, J.M. Cayuela, A. Martnez-Cach a, F.
Pardo, J.M. L opez Nicol as, J. Agric. Food Chem. 51 (2003) 4694.
[6] J.M. Oliveira, I.M. Ara ujo, O.M. Pereira, J.S. Maia, A.J. Amaral, M.O.
Maia, Anal. Chim. Acta 513 (2004) 269.
[7] J.J. Castillo-S anchez, J.C. Mejuto, J. Garrido, S. Garca-Falc on, Food
Chem. 97 (2006) 130.
[8] M. Palma, L.T. Taylor, J. Chromatogr. A 849 (1999) 117.
[9] M. Palma, Z. Pi neiro, C.G. Barroso, J. Chromatogr. A 968 (2002) 1.
[10] Z. Pi neiro, M. Palma, C.G. Barroso, J. Chromatogr. A 1026 (2004) 19.
[11] F. Bonilla, M. Mayen, J. Merida, M. Medina, Food Chem. 66 (1999) 209.
[12] Y. Yilmaz, R.T. Toledo, J. Agric. Food Chem. 52 (2004) 255.
[13] D. Kammerer, A. Claus, R. Carle, A. Schieber, J. Agric. Food Chem. 52
(2004) 4360.
[14] C. Negro, L. Tommasi, A. Miceli, Bioresour. Technol. 87 (2003) 41.
[15] E. Pastrana-Bonilla, C.C. Akoh, S. Selleppan, G. Krewer, J. Agric. Food
Chem. 51 (2003) 5497.
[16] Z. Kerem, B. Bravdo, O. Shoseyov, Y. Tugendhaft, J. Chromatogr. A 1052
(2004) 211.
[17] J.E. Cacace, G. Mazza, J. Food Eng. 59 (2003) 379.
[18] M. Pinelo, M. Rubilar, J. Sineiro, M.J. N u nez, Food Chem. 85 (2004) 267.
[19] M.L. Calabr o, V. Galtieri, P. Cutroneo, S. Tommasini, P. Ficarra, R. Ficarra,
J. Pharm. Biomed. Anal. 35 (2004) 349.
[20] M. Pinelo, P. Del Fabbro, L. Manzocco, M.J. N unhez, M.C. Nicoli, Food
Chem. 92 (2005) 109.
[21] B.M. Silva, P.B. Andrade, G.C. Mendes, R.M. Seabra, M.A. Ferreira, J.
Agric. Food Chem. 50 (2002) 2313.
[22] P. Valent ao, R.M. Seabra, G. Lopes, L.R. Silva, V. Martins, M.E. Trujillo,
E. Vel azquez, P.B. Andrade, Food Chem. 100 (2007) 64.
[23] P. Valent ao, P.B. Andrade, J. Rangel, B. Ribeiro, B.M. Silva, P. Baptista,
R.M. Seabra, J. Agric. Food Chem. 53 (2005) 4925.
[24] P.B. Andrade, G. Mendes, V. Falco, P. Valent ao, R.M. Seabra, Food Chem.
73 (2001) 397.
[25] B.M. Silva, P.B. Andrade, R.M. Seabra, M.A. Ferreira, J. Liq. Chromatogr.
Rel. Technol. 24 (2001) 2861.
[26] J. Moreda-Pi neiro, C. Moscoso-P erez, P. L opez-Maha, S. Muniategui-
Lorenzo, E. Fern andez-Fern andez, D. Prada-Rodrguez, Talanta 53 (2001)
871.
[27] B. Lapornik, M. Prosek, A.G. Wondra, J. Food Eng. 71 (2005) 214.
[28] P. Bermejo-Barrera, A. Moreda-Pi neiro, A. Bermejo-Barrera, J. Anal. At.
Spectrom. 15 (2000) 121.
[29] J.C. Miller, J.N. Miller, Estadstica para qumica analtica, Addison-Wesley
Iberoamericana, Wilmington, DE, USA, 1993, p. 87.