Accepted Manuscript Analytical methods Detection of roasted and ground coffee adulteration by HPLC by amperometric and by post-column derivatization UV-Vis detection Diego S. Domingues, Elis D. Pauli, Julia E.M. de Abreu, Francys W. Massura, Valderi Cristiano, Maria J. Santos, Suzana L. Nixdorf PII: DOI: Reference:
S0308-8146(13)01331-9 http://dx.doi.org/10.1016/j.foodchem.2013.09.066 FOCH 14694
To appear in:
Food Chemistry
Received Date: Revised Date: Accepted Date:
9 October 2011 9 September 2013 11 September 2013
Please cite this article as: Domingues, D.S., Pauli, E.D., de Abreu, J.E.M., Massura, F.W., Cristiano, V., Santos, M.J., Nixdorf, S.L., Detection of roasted and ground coffee adulteration by HPLC by amperometric and by postcolumn derivatization UV-Vis detection, Food Chemistry (2013), doi: http://dx.doi.org/10.1016/j.foodchem. 2013.09.066
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Detection of roasted and ground coffee adulteration by HPLC by amperometric
2
and by post-column derivatization UV-Vis detection
3 4 5 6
Diego S. Domingues, Elis D. Pauli, Julia E. M. de Abreu, Francys W. Massura, Valderi
7
Cristiano, Maria J. Santos, Suzana L. Nixdorf*
8 9
State University of Londrina
10
Center for Exact Sciences – Department of Chemistry
11
Rodovia Celso Garcia Cid, km 380
12
Jardim Portal de Versalhes I
13
Campus Universitário, LAB DIA or 342
14
Mail Box: 6001
15
Zip code: 86055-900
16
Londrina – Paraná – Brazil
17 18 19 20 21 22
*Corresponding author: Phone/Fax: +55-(43)33714836/+55-(43)33714286; [email protected]
23 24 25
1
26
ABSTRACT
27
Coffee is one of the most consumed beverages in the world. Due to its
28
commercial importance, the detection of impurities and foreign matters has been a
29
constant concern in fraud verification, especially because it is difficult to percept
30
adulterations with the naked eye in samples of roasted and ground coffee. In Brazil, the
31
most common additions are roasted materials, such as husks, sticks, corn, wheat
32
middling, soybean, and more recently – acai palm seeds.
33
The performance and correlation of two chromatographic methods, HPLC-
34
HPAEC-PAD and post-column derivatization HPLC-UV-Vis, were compared for
35
carbohydrate analysis in coffee samples.
36
To verify the correlation between the two methods, the principal component
37
analysis for the same mix of triticale and acai seeds in different proportions with coffee
38
was employed. The performance for detecting adulterations in roasted and ground
39
coffee of the two methods was compared.
40
Keywords: coffee, frauds, carbohydrates, electrochemical detection, authenticity.
41 42 43 44 45 46 47 48 49 50
2
51
1. INTRODUCTION
52
Coffee is one of the most valuable basic products, constituting the second major
53
commodity just after oil (ICO, 2011; Nabais, Carrott, Carrot, Luz, Ortiz, 2008).
54
According to the International Coffee Organization – ICO (2011), the total coffee
55
production in crop year 2011/2012 was about 131,3 million bags (each bag weighing 60
56
kg), with approximately 33,1 % produced in Brazil.
57
The economic importance of coffee makes it clear that studies related to its
58
composition, quality evaluation, and fraud detection appear to be of greatest
59
significance (Moreira, Trugo, Maria, 1997), setting up marketing requirements,
60
especially in an increasingly globalized market, which demands effective product
61
quality control (ISO, 1995; ABIC, 2011a).
62
The detection of impurities and mixes in coffee is a constant concern, especially
63
in relation to the product quality assurance. A mix, intentional or not, of foreign
64
materials to the product, usually of low-cost, which alter the product quality and can
65
cause damages to consumers, particularly those of economic nature, is considered fraud
66
(Assad, Sano, Correa, Rodrigues, Cunha, 2002). According to the ISO 3509: Coffee and
67
its products – vocabulary - The International Organization for Standardization, defines
68
"impurities" as any foreign matter, which may be found in coffee like: wood, twigs,
69
husks (or straw), and whole cherries (ISO, 1989). In Brazil, the most frequently
70
substances reported by the literature, added to coffee are: husks and sticks, corn, barley,
71
wheat middling, brown sugar, and soybean (Assad, Sano, Correa, Rodrigues, Cunha,
72
2002); rye, triticale, and acai may also be added to this list (ABIC, 2011b). According to
73
Bernal, Toribio, Del Alamo and Del Nozal (1996), the individual determination of
74
carbohydrates has gained significant importance not only for providing compositional
75
information on samples, but also for assisting in the identification of adulterants. 3
76
The carbohydrate profile studies, carried out by Blanc, Davis, Viani and Parchet
77
(1989) for hundreds of samples of commercial soluble coffees using HPLC with UV-
78
Vis detection, enabled to verify the addition of coffee husk extracts at concentrations
79
above 25%. In this studies, the concentration of free and total carbohydrates made it
80
possible to evidence frauds by the determination of intentional contamination with
81
coffee husk and ligneous material (sticks) that had caused an increase in the content of
82
mannitol, xylose, glucose, and fructose, as well as to distinguish pure products from
83
adulterated ones by verifying the adulterant nature (Nogueira, Lago, 2009). For roasted
84
and ground coffee the total carbohydrates content are still scarce in the literature
85
(Garcia, Pauli, Cristiano, Camara, Scarminio, Nixdorf, 2009).
86
Methods for the liquid chromatographic analysis of carbohydrates often have
87
employed columns of amino-bonded silica-based or of metal-loaded cation-exchange
88
polymer-based. These columns have the advantage of not requiring regeneration after
89
every run. However, columns of metal-loaded cation-exchange require heating
90
presenting low resolution, with restrictions on pH range and in use of organic solvents
91
(Dionex, 2012). Although, according to Lanças (2004) mobile phases of liquid
92
chromatography represent a powerful tool for manipulation the analyte retention and
93
selectivity, but in this case usually precludes the use of gradients and often requires
94
stringent sample cleanup prior to injection.
95
For the ion exchange mechanism the main factors influencing the separation are:
96
selectivity of counter ion, solvent, pH and flow (Lanças, 2004). As the pH is main
97
factor for the analyte become completely ionized, it should be adjusted to two units
98
above the pKa (for the acid) and two units below the pKa (for the basis). If the analyte is
99
in its ionized form, it will be retained by the strongly anion (SAX) stationary phase.
100
Elution is then done, by adjusting the pH of the mobile phase at two units above the
4
101
pKa, which will increase the unionized form and will allows the elution of the exchange
102
stationary phase, promoting regeneration of the column (Lanças, 2004).
103
Analyses of carbohydrates are also difficult to do, due to their structural
104
diversities. The hydroxyl groups of carbohydrates are partially ionized under highly
105
alkaline conditions to form oxyanions, and thus can be separated by the anion-exchange
106
mechanism (Inoue, Kitahara, Aikawa, Arai, Masuda-Hanada, 2011). Currently, the high
107
performance anion-exchange chromatography (HPAEC), takes advantage of the weakly
108
acidic nature of carbohydrates to give highly selective separations at high pH, using a
109
strong anion-exchange stationary phase with electrochemical detection (ED), as a high
110
sensitive detection method for carbohydrates, without the need for prior derivatization
111
(Dionex, 2012). However, a limited number of sorbents are commercially available: on
112
the electrostatically latex-coated pellicular polymeric-based anion-exchange, and in
113
macroporous poly(styrene-divinylbenzene) with trimetylammonium group. An anion-
114
exchange stationary phase prepared from polystyrene-based copolymer and diamine has
115
been reported for separation of aldopentoses and aldohexoses (Inoue, Kitahara, Aikawa,
116
Arai, Masuda-Hanada, 2011).
117
According to Inoue, Kitahara, Aikawa, Arai and Masuda-Hanada (2011)
118
separation of D-aldopentoses (D-arabinose and D-xylose – Fig. 1.) and D-aldohexoses
119
(D-glucose; D-manose and D-galactose) gradually increased in an almost linear manner
120
with the decreasing concentration of the NaOH eluent from 100 to 30 mmol L-1, and
121
below 30 mmol L-1, the retention time ratios steeply increased around 20 mmol L-1
122
NaOH (pH = 12.3), corresponding to the pKa values of the aldoses. These results
123
indicate that the dissociated aldoses strongly interact with the quaternary nitrogen atom
124
of the stationary phase, than the competitive hydroxide ions in the eluent. In contrast, at
5
125
low NaOH concentrations (from 30 to 10 mmol l-1), were reasonably retained as
126
follows: D-mannose (pKa 12.08) > D-glucose (pKa 12.28) > D-galactose (pKa 12.35).
127
It is well known that the anomeric hydroxy group of the pyranose form is more
128
acid, that the other hydroxy groups. However, the ionization of the hydroxy groups
129
other than the anomeric one is possible. Koizumi, Kubota, Ozaki, Shigenobu, Fukuda
130
and Tanimoto (1992) concluded in his study of positional isomers of methyl ethers of
131
D-glucose, that the acidity of the monosaccharide is in the following order: 1-OH> 2-
132
OH> 6-OH> 3-OH> 4-OH. Since the individual hydroxy groups of the
133
monosaccharides reveal the different pKa values, the ionization of the hydroxy groups
134
other than anomeric one, probably play important roles during elution. So, besides the
135
pKa values additional factors for the elution characteristics of carbohydrates should be
136
considered. The aldoses exist as an equilibrium between the pyranoses and furanoses;
137
the percentage composition of the cyclic forms of monosaccharides is given in Table 1.
138
Usually, in aqueous solutions, aldopentoses and aldohexoses exist primarily in the
139
six-membered pyranose form. But, it is noteworthy that aldoses possessing higher
140
percentage of furanose composition are retained strongly at low NaOH concentrations.
141
This suggested that strong binding ability of fructose with an anion exchange column
142
may be due to their furanose form. These results suggest that the elution behavior of the
143
aldoses, would probably correlate not only with the pKa values, but also with the
144
furanoses forms (Inoue, Kitahara, Aikawa, Arai, Masuda-Hanada, 2011).
145
In addition, refractive index (RI) and low-wavelength UV detection methods are
146
sensitive to eluent and sample matrix components. These analytical methods require
147
attention to sample solubility and sample concentration (Dionex, 2012).
148
Post-column derivatization is required in HPLC-UV-Vis systems for generating
149
necessary photometrically-active derivatives, since carbohydrates do not possess any
6
150
conjugated π-bonds, and therefore, they are not directly detectable (Pauli, Cristiano,
151
Nixdorf, 2011). Despite its simplicity, and considering that in most laboratories HPLC
152
is coupled with UV-Vis detection, the UV-Vis technique has the disadvantage of non-
153
detection of mannitol and the difficulty in quantifying xylose due to its low
154
concentration in coffee (Coutinho, 2003). However, this technique has demonstrated its
155
applicability as a method for initial screening to identify possible adulterants for coffee,
156
despite their low resolution, according to reference values established by AFCASOLE
157
(Pauli, Cristiano, Nixdorf, 2011).
158
Unlike the HPLC-UV-Vis method, the ion-exchange chromatographic method,
159
using a strong anion-exchange column coupled with an electrochemical detector and
160
applying pulsed amperometry – High-Performance Anion-Exchange Chromatography
161
with Pulsed Amperometric Detection (HPAEC-PAD) - has become the ISO 11292
162
standardized methodology (ISO, 1995) for the determination of free and total
163
carbohydrates found in soluble coffees.
164
Pulsed amperometry permits detection of carbohydrates with excellent signal-to-
165
noise ratios down to approximately 10 picomol without requiring derivatization.
166
Carbohydrates are detected by measuring the electrical current generated by their
167
oxidation at the surface of a gold electrode. At high pH, carbohydrates are
168
electrocatalytically oxidized at the surface of the gold electrode by application of a
169
positive potential. The current generated is proportional to the carbohydrate
170
concentration, and therefore carbohydrates can be detected and quantified. The products
171
of this oxidation reaction also poison the surface of the electrode, which means that it
172
has to be cleaned between measurements. This is accomplished by first raising the
173
potential to a level sufficient to oxidize the gold surface. This cause desorption of the
7
174
carbohydrate oxidation products. The electrode potential is then lowered to reduce the
175
electrode surface back to gold (Dionex, 2012).
176
The association of analytical techniques using experimental design, with principal
177
component analysis (Barros Neto, Scarminio, Bruns, 2003), has been increasingly
178
applied, facilitating the establishing of correlations between various raw materials,
179
based on their chromatographic profiles (Garcia, Pauli, Cristiano, Camara, Scarminio,
180
Nixdorf, 2009).
181
This study aims to evaluate the performance and correlation between two different
182
chromatographic systems: HPLC-HPAEC-PAD and post-column derivatization HPLC-
183
UV-Vis, applied for carbohydrate determination (method ISO 11292), following
184
simplex-centroid design, to verify the ability to distinguish a mixture of triticale and
185
acai in arabica coffee.
186
2. MATERIALS AND METHODS
187
2.1.
Preparation of Samples of Arabica Coffee and Adulterants (Triticale and Acai)
188
The samples of arabica coffee, triticale, and acai seeds were provided by the
189
Agronomic Institute of Parana (Londrina, Parana State, Brazil). The samples were
190
roasted and ground to achieve a color similar to that of commercial roasted and ground
191
coffee, presenting a medium roast.
192
For the adulterant study, sampling followed the simplex-centroid experimental design,
193
represented by an equilateral triangle, with a total of 10 different compositions coded
194
from 1 to 10. The vertices of which, corresponded to the pure matrices. The edges
195
corresponded to the binary mixes of the same proportion; the central point – to the
196
ternary mix with equal proportions; and the three axial points – to the proportions 4:1:1,
197
1:4:1, 1:1:4. All samples of arabica coffee-triticale-acai were prepared in duplicate for 8
198
both systems, except for the central point, that samples were prepared in triplicate. The
199
preparation was given by weighing different proportions of the matrices in order to
200
always reach on a dry weight basis 0.3000 g for the analysis by HPLC-HPAEC-PAD,
201
and 0.2000 g for the analysis by post-column derivatization reaction HPLC-UV-Vis. In
202
sequence, samples with the respective weights, according to each method, were
203
hydrolyzed, by transferring to a 500 mL erlenmeyer with screw-cap, with adding 50 mL
204
of 1.00 mol L-1 hydrochloric acid, and by placing in a water bath thermostated at 85 °C
205
for 180 min, stirring every 30 min manually. After, the solution was cooled down with
206
tap water until room temperature, filtered with a blue-stripe pleated paper into a 100 mL
207
volumetric flask that was completing up to the mark with ultrapure water. An aliquot of
208
10.0 mL of the solution was passed through a C18 cartridge (Sep Pak, Waters)
209
preconditioned with methanol and water, and in a 0.22 µm nylon membrane (Millipore),
210
collecting the filtrate in vials that were injected into the respective chromatographic
211
systems. For the HPLC-UV-Vis system the extracted samples were neutralized, prior to
212
the injection, due to the narrow range of pH tolerance of the Aminex column. Each
213
aliquot of the mixture of samples (weighed and extracted) in duplicate or triplicate, were
214
then randomly analyzed, by the HPAEC-PAD and by HPLC-UV-Vis (mean values are
215
shown in Table 2). The standards and samples were injected randomly to avoid any
216
tendency of systematic error in the data throughout the day.
217
For the principal component analysis, the SPSS 18 software (Softonic, Spain) was
218
used.
219
2.2.
Reagents and Standards
220
Sodium hydroxide (50% solution; Fisher, USA and Isosol, Brazil) and
221
hydrochloric acid (p.a. grade; F. MAIA, Brazil) were used as solvents for the mobile
9
222
phase extraction and preparation steps. All water used for the preparation of standards
223
and solutions was purified and filtered with a Milli-Q® system (Millipore, Milford, MA,
224
USA). The mobile phases were degassed with nitrogen prior to use (99.99973% purity
225
cylinder from LINDE, Brazil, with 2nd-stage regulator from Inpagás).
226
The standards used were: D (-)– mannitol, D (-)– arabinose, D (+)– galactose, D
227
(+)– glucose, D (+)– xylose, D (+)– mannose, D (-)– fructose, all from Merck
228
(Darmstadt, Germany). Due to high hygroscopicity of carbohydrates, the standards were
229
stored in a glass desiccator under vacuum over phosphorus pentoxide (Merck,
230
Darmstadt, Germany) and utilized only after one week desiccation.
231
2.3. Standard Solutions
232
2.3.1. HPLC-HPAEC-PAD Analysis
233
For the preparation of the carbohydrate standard stock mix solution, 0.0030 g of
234
mannitol, 0.0300 g of arabinose, 0.1200 g of galactose, 0.0450 g of glucose, 0.0120 g of
235
xylose, 0.0900 g of mannose, and 0.0450 g of fructose were weighed, added to a 100.0
236
mL volumetric flask and made up to the mark with ultrapure water. The solution was
237
sonicated in an ultrasonic bath for 10 min (Garcia, Pauli, Cristiano, Camara, Scarminio,
238
Nixdorf, 2009).
239
The identification and quantification of the carbohydrates were performed on the
240
basis of retention times of components eluted from the column, comparing them the
241
retention times of the components with known concentrations of individual external
242
standards, and by co-chromatography.
243
For the carbohydrate quantification in the samples, a 10% (v/v) mix of analytical
244
standards was injected into ultrapure water. This standard mix corresponded to the
245
following concentrations in relation to 0.3000 g of sample: 0.10% (w/w) of mannitol, 10
246
1.00% (w/w) of arabinose, 4.00% (w/w) of galactose, 1.50% (w/w) of glucose, 0.40%
247
(w/w) of xylose, 3.00% (w/w) of mannose, and 1.50% (w/w) of fructose.
248
2.3.2. Post-Column Derivatization Reaction HPLC-UV-Vis Analysis
249
For the preparation of the carbohydrate standard stock mix solution, 0.0300 g of
250
glucose, 0.0200 g of xylose, 0.1100 g of galactose, 0.0400 g of arabinose, and 0.0600 g
251
of mannose were weighed, transferred to a 100.00 mL volumetric flask and made up to
252
the mark with ultrapure water. The solution was sonicated in an ultrasonic bath for 5
253
min (Pauli, Cristiano, Nixdorf, 2011). The standard was stored in a refrigerator at ~4°C.
254
This stock solution was diluted to obtain a 25% (v/v) analytical standard, which was
255
injected each quantification day.
256
2.4. Chromatographic Systems
257
2.4.1. HPLC-HPAEC-PAD
258
For the chromatographic analysis, an instrumental system was used, which
259
consisted of a PEEK inert liquid chromatograph composed of two Nalgene® bottles for
260
the mobile phase storage; a LC-10Ai inert high-pressure pump (Shimadzu®, Tokyo,
261
Japan); a Research-1367-72 3-way solenoid low-pressure valve with a "lab-made"
262
solenoid valve activation external circuit for eluent exchange; a SIL-Prominence 20A
263
automatic injector (Shimadzu®, Tokyo, Japan); a CarboPac PA1 pre-column and a
264
CarboPac PA1 polystyrene-divinylbenzene-based high-performance anion-exchange
265
resin column (10 µm, 250 mm × 4 mm; Dionex, Sunnyvale, CA, USA); a Waters
266
thermostatic column oven coupled with a 650 CHX temperature controller (Pickering
267
Laboratories); a RS570 current amplifier (Stanford Research System); a dual supply of
268
±12 V used for activation of an external circuit for electrical signal transmission; an ED11
269
50 electrochemical cell (Dionex, Sunnyvale, CA, USA); an Autolab PGSTAT 30
270
potentiostat (Eco-Chemie, Utrecht, Netherlands), an Autolab interface (Eco Chemie,
271
Utrecht, Netherlands); a data acquisition and processing system on the basis of Pentium
272
IV working with the GPES (General Purpose Electrochemical System)–Eco software.
273
For the integration of chromatograms, the INTEGRA software, developed by
274
Professor Dr. Carlos Alberto Paulinetti da Camara (DIA Group, Department of
275
Chemistry, State University of Londrina, Brazil), was employed. It makes use of Eq. (1)
276
to quantify the carbohydrate content (ω), expressed as a percentage of mass, by
277
implementing external standardization:
278
279
Am0V x100 A0 mVst
280
where,
281
A – peak area for the individual carbohydrate in the sample;
282
A0 – peak area for the individual carbohydrate in the standard solution;
283
m – mass of the sample aliquot, expressed on a dry weight basis, in g;
284
m0 – mass of the carbohydrate in the standard solution, in g,
285
V – volume of the sample, in mL;
286
Vst – volume of the standard solution, in mL.
287
2.4.2. Post-Column Derivatization Reaction HPLC-UV-Vis
(1)
288
The instrumental system used consisted of a high-performance liquid
289
chromatograph composed of two amber bottles for the storage of the mobile phase
290
eluent and the post-column reagent; two Waters 515 high-pressure pumps (Milford,
291
MA, USA); a Waters 717 Plus automatic injector; a SP-1010P pre-column (6 x 50 mm;
292
Shodex, NY, USA); an Aminex HPX-87P cation-exchange column in the Pb2+ form (9 12
293
µm, 7.8 x 300 mm; Bio-Rad, Richmond, CA, USA); a tee mixer; a thermostatic column
294
oven coupled with a Waters Module II temperature controller; a CRX 390 post-column
295
reactor (Pickering Laboratories, Mountain View, CA, USA) containing a stainless-steel
296
reaction coil inside the controlled temperature reactor; a pulse dampener with back
297
pressure of 100 psi connected to the detector output, a Waters 2487 UV-Vis detector;
298
and the Empower Build 1154 software for data acquisition and processing with a
299
microcomputer.
300
2.5. Chromatographic Analyses
301
For the extraction of the coffee carbohydrates, triticale and acai, the ISO 11292
302
standardized method (ISO, 1995) was implemented according to the adaptation
303
described above.
304
The chromatographic conditions used for the two validated HPLC systems are:
305
electrochemical detection HPAEC-PAD: pre-column: CarboPac PA1; column:
306
CarboPac PA1: 250 mm x 4 mm, 10 μm; mobile phase composition: eluent and
307
equilibrium: 1.4 mmol L-1 of NaOH and regeneration: 300.0 mmol L-1 of NaOH; flow
308
rate: 1.0 mL min-1; detection: determination potential:+0.20 V (400 ms), oxidation
309
potential: +0.65 V (200 ms), reduction potential: -0.20 V (400 ms); injection volume:
310
20.0 µL; column temperature: 28ºC and chromatographic run-time: 72.60 min (Garcia
311
Pauli, Cristiano, Camara, Scarminio,
312
derivatization: pre-column: SP-1010P; column: Aminex HPX-87P; mobile phase: eluent
313
composition (pump 1) - ultrapure water, post-column (Pump 2) – ABH + NaOH; flow:
314
pump 1: 0.5 mL min-1, pump 2: 0.6 mL min-1; detection: 410 nm; injection volume: 20.0
315
µL, column temperature: 85ºC, post-column reactor temperature: 100ºC and
316
chromatographic run-time: 25 min (Pauli, Cristiano, Nixdorf, 2011).
Nixdorf, 2009). UV-Vis post-column
13
317
The accuracy for both methods previously cited (HPLC-HPAEC-PAD and HPLC-
318
UV-Vis) was calculated by the recovery rate of analyte, which was done in triplicate, by
319
adding into the sample in proportion of 1:1 (v/v) of standard in low concentration level
320
(50%), medium (100%) and high (150%), according to calibration curve in the dynamic
321
range, calculated by Eq. (2).
C C2 rec (%) 1 x 100 C3
322 323
where,
324
rec (%) = percentage of recovery;
325
C1 = concentration of analyte in the spiked sample with standard addition;
326
C2 = concentration of analyte in the original sample without spiked standard;
327
C3 = concentration of the analyte standard added to the sample spiked.
328
(2)
Results were expressed as mean recoveries from the low, medium and high
329
concentrations levels.
330
3. RESULTS AND DISCUSSION
331
3.1.
Evaluation of Separation of Carbohydrates
332
For separation system employing HPLC-UV-Vis post-column derivatization, after
333
testing three columns, we chose to use a divalent cation lead - Aminex HPX-87P, as it
334
had the highest resolution compared to the other two - a divalent column of calcium and
335
the other a monovalent of hydrogen. By being cationic, their use required a higher
336
temperature (85oC) which discourages the interaction, as can be observed by rapidly
337
eluting peaks, impairing resolution (Fig. 3).
338
The variation of - solvent, flow, pH and ionic strength, to improve the selectivity
339
(Lanças, 2004) were not feasible in these experiments, since the strength of the mobile
14
340
phase couldn´t be varied; by the fact of Aminex column does not allow the use of
341
organic solvents. The flow rate could not be reduced to increase interaction, since was
342
already low (0.5 mL min-1). Adding salt for change the ionic strength favored the
343
competition with the active sites disadvantaging the interaction between the counter-ion
344
of the stationary phase and the carbohydrates, resulting in a worsening in the resolution
345
between the peaks. In this case also, it was not possible ionize the sample, using a pH
346
two points above of the pKa of the carbohydrates (12.08 - 12.35 - Table 1), as
347
recommended by Lanças (2004), since the pH range of this column is restricted to 5.0-
348
9.0. This justify that the hydrolyzed sample should be neutralize before analysis. Thus,
349
not as many remaining alternatives to the use of different mobile phases, the ultrapure
350
water were adopted. The neutral medium favors broadening and flattening of
351
chromatographic peaks in the ion exchange mode, helping in the inadequate resolution,
352
observed in the chromatogram of the UV-Vis system with a partial co-elution of
353
arabinose and mannose (peaks 4 and 5 – Fig. 3).
354
Adding to this, for sure the major factor contributing in proportion to the low
355
resolution is the way in which the chemical structures of carbohydrates (aldoses) are in
356
the aqueous medium. Table 1 permits us to see that there are higher proportions of
357
pyranose, compared to furanose, once the cycle of six members is thermodynamically
358
more stable in aqueous medium (Inoue, Kitahara, Aikawa, Arai, Masuda-Hanada,
359
2011). Mute rotations of the anomeric carbon also in Table 1, shown that there is a
360
predominance of the alpha pyranose form. This is justified by the hydroxyl group in the
361
alpha configuration is pointing down, while in beta form hydroxyl is pointing upwards
362
(Fig. 1), so that the aligning two heteroatoms partially suffer repulsion. It was also
363
observed, from the data of Table 1 that in equilibrium in the aqueous medium, there is
364
predominance to the β form for glucose (62%), xylose (63%) and galactose (64%),
15
365
while is a predominance of α form for arabinose (60%) and mannose (64%), the form
366
more stable and retained in chromatography. Considering the aldopentoses, the
367
arabinose has a superior retention than the xylose, since it has a higher proportion of
368
furanose (2.5%) against (
S0308-8146(13)01331-9 http://dx.doi.org/10.1016/j.foodchem.2013.09.066 FOCH 14694
To appear in:
Food Chemistry
Received Date: Revised Date: Accepted Date:
9 October 2011 9 September 2013 11 September 2013
Please cite this article as: Domingues, D.S., Pauli, E.D., de Abreu, J.E.M., Massura, F.W., Cristiano, V., Santos, M.J., Nixdorf, S.L., Detection of roasted and ground coffee adulteration by HPLC by amperometric and by postcolumn derivatization UV-Vis detection, Food Chemistry (2013), doi: http://dx.doi.org/10.1016/j.foodchem. 2013.09.066
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Detection of roasted and ground coffee adulteration by HPLC by amperometric
2
and by post-column derivatization UV-Vis detection
3 4 5 6
Diego S. Domingues, Elis D. Pauli, Julia E. M. de Abreu, Francys W. Massura, Valderi
7
Cristiano, Maria J. Santos, Suzana L. Nixdorf*
8 9
State University of Londrina
10
Center for Exact Sciences – Department of Chemistry
11
Rodovia Celso Garcia Cid, km 380
12
Jardim Portal de Versalhes I
13
Campus Universitário, LAB DIA or 342
14
Mail Box: 6001
15
Zip code: 86055-900
16
Londrina – Paraná – Brazil
17 18 19 20 21 22
*Corresponding author: Phone/Fax: +55-(43)33714836/+55-(43)33714286; [email protected]
23 24 25
1
26
ABSTRACT
27
Coffee is one of the most consumed beverages in the world. Due to its
28
commercial importance, the detection of impurities and foreign matters has been a
29
constant concern in fraud verification, especially because it is difficult to percept
30
adulterations with the naked eye in samples of roasted and ground coffee. In Brazil, the
31
most common additions are roasted materials, such as husks, sticks, corn, wheat
32
middling, soybean, and more recently – acai palm seeds.
33
The performance and correlation of two chromatographic methods, HPLC-
34
HPAEC-PAD and post-column derivatization HPLC-UV-Vis, were compared for
35
carbohydrate analysis in coffee samples.
36
To verify the correlation between the two methods, the principal component
37
analysis for the same mix of triticale and acai seeds in different proportions with coffee
38
was employed. The performance for detecting adulterations in roasted and ground
39
coffee of the two methods was compared.
40
Keywords: coffee, frauds, carbohydrates, electrochemical detection, authenticity.
41 42 43 44 45 46 47 48 49 50
2
51
1. INTRODUCTION
52
Coffee is one of the most valuable basic products, constituting the second major
53
commodity just after oil (ICO, 2011; Nabais, Carrott, Carrot, Luz, Ortiz, 2008).
54
According to the International Coffee Organization – ICO (2011), the total coffee
55
production in crop year 2011/2012 was about 131,3 million bags (each bag weighing 60
56
kg), with approximately 33,1 % produced in Brazil.
57
The economic importance of coffee makes it clear that studies related to its
58
composition, quality evaluation, and fraud detection appear to be of greatest
59
significance (Moreira, Trugo, Maria, 1997), setting up marketing requirements,
60
especially in an increasingly globalized market, which demands effective product
61
quality control (ISO, 1995; ABIC, 2011a).
62
The detection of impurities and mixes in coffee is a constant concern, especially
63
in relation to the product quality assurance. A mix, intentional or not, of foreign
64
materials to the product, usually of low-cost, which alter the product quality and can
65
cause damages to consumers, particularly those of economic nature, is considered fraud
66
(Assad, Sano, Correa, Rodrigues, Cunha, 2002). According to the ISO 3509: Coffee and
67
its products – vocabulary - The International Organization for Standardization, defines
68
"impurities" as any foreign matter, which may be found in coffee like: wood, twigs,
69
husks (or straw), and whole cherries (ISO, 1989). In Brazil, the most frequently
70
substances reported by the literature, added to coffee are: husks and sticks, corn, barley,
71
wheat middling, brown sugar, and soybean (Assad, Sano, Correa, Rodrigues, Cunha,
72
2002); rye, triticale, and acai may also be added to this list (ABIC, 2011b). According to
73
Bernal, Toribio, Del Alamo and Del Nozal (1996), the individual determination of
74
carbohydrates has gained significant importance not only for providing compositional
75
information on samples, but also for assisting in the identification of adulterants. 3
76
The carbohydrate profile studies, carried out by Blanc, Davis, Viani and Parchet
77
(1989) for hundreds of samples of commercial soluble coffees using HPLC with UV-
78
Vis detection, enabled to verify the addition of coffee husk extracts at concentrations
79
above 25%. In this studies, the concentration of free and total carbohydrates made it
80
possible to evidence frauds by the determination of intentional contamination with
81
coffee husk and ligneous material (sticks) that had caused an increase in the content of
82
mannitol, xylose, glucose, and fructose, as well as to distinguish pure products from
83
adulterated ones by verifying the adulterant nature (Nogueira, Lago, 2009). For roasted
84
and ground coffee the total carbohydrates content are still scarce in the literature
85
(Garcia, Pauli, Cristiano, Camara, Scarminio, Nixdorf, 2009).
86
Methods for the liquid chromatographic analysis of carbohydrates often have
87
employed columns of amino-bonded silica-based or of metal-loaded cation-exchange
88
polymer-based. These columns have the advantage of not requiring regeneration after
89
every run. However, columns of metal-loaded cation-exchange require heating
90
presenting low resolution, with restrictions on pH range and in use of organic solvents
91
(Dionex, 2012). Although, according to Lanças (2004) mobile phases of liquid
92
chromatography represent a powerful tool for manipulation the analyte retention and
93
selectivity, but in this case usually precludes the use of gradients and often requires
94
stringent sample cleanup prior to injection.
95
For the ion exchange mechanism the main factors influencing the separation are:
96
selectivity of counter ion, solvent, pH and flow (Lanças, 2004). As the pH is main
97
factor for the analyte become completely ionized, it should be adjusted to two units
98
above the pKa (for the acid) and two units below the pKa (for the basis). If the analyte is
99
in its ionized form, it will be retained by the strongly anion (SAX) stationary phase.
100
Elution is then done, by adjusting the pH of the mobile phase at two units above the
4
101
pKa, which will increase the unionized form and will allows the elution of the exchange
102
stationary phase, promoting regeneration of the column (Lanças, 2004).
103
Analyses of carbohydrates are also difficult to do, due to their structural
104
diversities. The hydroxyl groups of carbohydrates are partially ionized under highly
105
alkaline conditions to form oxyanions, and thus can be separated by the anion-exchange
106
mechanism (Inoue, Kitahara, Aikawa, Arai, Masuda-Hanada, 2011). Currently, the high
107
performance anion-exchange chromatography (HPAEC), takes advantage of the weakly
108
acidic nature of carbohydrates to give highly selective separations at high pH, using a
109
strong anion-exchange stationary phase with electrochemical detection (ED), as a high
110
sensitive detection method for carbohydrates, without the need for prior derivatization
111
(Dionex, 2012). However, a limited number of sorbents are commercially available: on
112
the electrostatically latex-coated pellicular polymeric-based anion-exchange, and in
113
macroporous poly(styrene-divinylbenzene) with trimetylammonium group. An anion-
114
exchange stationary phase prepared from polystyrene-based copolymer and diamine has
115
been reported for separation of aldopentoses and aldohexoses (Inoue, Kitahara, Aikawa,
116
Arai, Masuda-Hanada, 2011).
117
According to Inoue, Kitahara, Aikawa, Arai and Masuda-Hanada (2011)
118
separation of D-aldopentoses (D-arabinose and D-xylose – Fig. 1.) and D-aldohexoses
119
(D-glucose; D-manose and D-galactose) gradually increased in an almost linear manner
120
with the decreasing concentration of the NaOH eluent from 100 to 30 mmol L-1, and
121
below 30 mmol L-1, the retention time ratios steeply increased around 20 mmol L-1
122
NaOH (pH = 12.3), corresponding to the pKa values of the aldoses. These results
123
indicate that the dissociated aldoses strongly interact with the quaternary nitrogen atom
124
of the stationary phase, than the competitive hydroxide ions in the eluent. In contrast, at
5
125
low NaOH concentrations (from 30 to 10 mmol l-1), were reasonably retained as
126
follows: D-mannose (pKa 12.08) > D-glucose (pKa 12.28) > D-galactose (pKa 12.35).
127
It is well known that the anomeric hydroxy group of the pyranose form is more
128
acid, that the other hydroxy groups. However, the ionization of the hydroxy groups
129
other than the anomeric one is possible. Koizumi, Kubota, Ozaki, Shigenobu, Fukuda
130
and Tanimoto (1992) concluded in his study of positional isomers of methyl ethers of
131
D-glucose, that the acidity of the monosaccharide is in the following order: 1-OH> 2-
132
OH> 6-OH> 3-OH> 4-OH. Since the individual hydroxy groups of the
133
monosaccharides reveal the different pKa values, the ionization of the hydroxy groups
134
other than anomeric one, probably play important roles during elution. So, besides the
135
pKa values additional factors for the elution characteristics of carbohydrates should be
136
considered. The aldoses exist as an equilibrium between the pyranoses and furanoses;
137
the percentage composition of the cyclic forms of monosaccharides is given in Table 1.
138
Usually, in aqueous solutions, aldopentoses and aldohexoses exist primarily in the
139
six-membered pyranose form. But, it is noteworthy that aldoses possessing higher
140
percentage of furanose composition are retained strongly at low NaOH concentrations.
141
This suggested that strong binding ability of fructose with an anion exchange column
142
may be due to their furanose form. These results suggest that the elution behavior of the
143
aldoses, would probably correlate not only with the pKa values, but also with the
144
furanoses forms (Inoue, Kitahara, Aikawa, Arai, Masuda-Hanada, 2011).
145
In addition, refractive index (RI) and low-wavelength UV detection methods are
146
sensitive to eluent and sample matrix components. These analytical methods require
147
attention to sample solubility and sample concentration (Dionex, 2012).
148
Post-column derivatization is required in HPLC-UV-Vis systems for generating
149
necessary photometrically-active derivatives, since carbohydrates do not possess any
6
150
conjugated π-bonds, and therefore, they are not directly detectable (Pauli, Cristiano,
151
Nixdorf, 2011). Despite its simplicity, and considering that in most laboratories HPLC
152
is coupled with UV-Vis detection, the UV-Vis technique has the disadvantage of non-
153
detection of mannitol and the difficulty in quantifying xylose due to its low
154
concentration in coffee (Coutinho, 2003). However, this technique has demonstrated its
155
applicability as a method for initial screening to identify possible adulterants for coffee,
156
despite their low resolution, according to reference values established by AFCASOLE
157
(Pauli, Cristiano, Nixdorf, 2011).
158
Unlike the HPLC-UV-Vis method, the ion-exchange chromatographic method,
159
using a strong anion-exchange column coupled with an electrochemical detector and
160
applying pulsed amperometry – High-Performance Anion-Exchange Chromatography
161
with Pulsed Amperometric Detection (HPAEC-PAD) - has become the ISO 11292
162
standardized methodology (ISO, 1995) for the determination of free and total
163
carbohydrates found in soluble coffees.
164
Pulsed amperometry permits detection of carbohydrates with excellent signal-to-
165
noise ratios down to approximately 10 picomol without requiring derivatization.
166
Carbohydrates are detected by measuring the electrical current generated by their
167
oxidation at the surface of a gold electrode. At high pH, carbohydrates are
168
electrocatalytically oxidized at the surface of the gold electrode by application of a
169
positive potential. The current generated is proportional to the carbohydrate
170
concentration, and therefore carbohydrates can be detected and quantified. The products
171
of this oxidation reaction also poison the surface of the electrode, which means that it
172
has to be cleaned between measurements. This is accomplished by first raising the
173
potential to a level sufficient to oxidize the gold surface. This cause desorption of the
7
174
carbohydrate oxidation products. The electrode potential is then lowered to reduce the
175
electrode surface back to gold (Dionex, 2012).
176
The association of analytical techniques using experimental design, with principal
177
component analysis (Barros Neto, Scarminio, Bruns, 2003), has been increasingly
178
applied, facilitating the establishing of correlations between various raw materials,
179
based on their chromatographic profiles (Garcia, Pauli, Cristiano, Camara, Scarminio,
180
Nixdorf, 2009).
181
This study aims to evaluate the performance and correlation between two different
182
chromatographic systems: HPLC-HPAEC-PAD and post-column derivatization HPLC-
183
UV-Vis, applied for carbohydrate determination (method ISO 11292), following
184
simplex-centroid design, to verify the ability to distinguish a mixture of triticale and
185
acai in arabica coffee.
186
2. MATERIALS AND METHODS
187
2.1.
Preparation of Samples of Arabica Coffee and Adulterants (Triticale and Acai)
188
The samples of arabica coffee, triticale, and acai seeds were provided by the
189
Agronomic Institute of Parana (Londrina, Parana State, Brazil). The samples were
190
roasted and ground to achieve a color similar to that of commercial roasted and ground
191
coffee, presenting a medium roast.
192
For the adulterant study, sampling followed the simplex-centroid experimental design,
193
represented by an equilateral triangle, with a total of 10 different compositions coded
194
from 1 to 10. The vertices of which, corresponded to the pure matrices. The edges
195
corresponded to the binary mixes of the same proportion; the central point – to the
196
ternary mix with equal proportions; and the three axial points – to the proportions 4:1:1,
197
1:4:1, 1:1:4. All samples of arabica coffee-triticale-acai were prepared in duplicate for 8
198
both systems, except for the central point, that samples were prepared in triplicate. The
199
preparation was given by weighing different proportions of the matrices in order to
200
always reach on a dry weight basis 0.3000 g for the analysis by HPLC-HPAEC-PAD,
201
and 0.2000 g for the analysis by post-column derivatization reaction HPLC-UV-Vis. In
202
sequence, samples with the respective weights, according to each method, were
203
hydrolyzed, by transferring to a 500 mL erlenmeyer with screw-cap, with adding 50 mL
204
of 1.00 mol L-1 hydrochloric acid, and by placing in a water bath thermostated at 85 °C
205
for 180 min, stirring every 30 min manually. After, the solution was cooled down with
206
tap water until room temperature, filtered with a blue-stripe pleated paper into a 100 mL
207
volumetric flask that was completing up to the mark with ultrapure water. An aliquot of
208
10.0 mL of the solution was passed through a C18 cartridge (Sep Pak, Waters)
209
preconditioned with methanol and water, and in a 0.22 µm nylon membrane (Millipore),
210
collecting the filtrate in vials that were injected into the respective chromatographic
211
systems. For the HPLC-UV-Vis system the extracted samples were neutralized, prior to
212
the injection, due to the narrow range of pH tolerance of the Aminex column. Each
213
aliquot of the mixture of samples (weighed and extracted) in duplicate or triplicate, were
214
then randomly analyzed, by the HPAEC-PAD and by HPLC-UV-Vis (mean values are
215
shown in Table 2). The standards and samples were injected randomly to avoid any
216
tendency of systematic error in the data throughout the day.
217
For the principal component analysis, the SPSS 18 software (Softonic, Spain) was
218
used.
219
2.2.
Reagents and Standards
220
Sodium hydroxide (50% solution; Fisher, USA and Isosol, Brazil) and
221
hydrochloric acid (p.a. grade; F. MAIA, Brazil) were used as solvents for the mobile
9
222
phase extraction and preparation steps. All water used for the preparation of standards
223
and solutions was purified and filtered with a Milli-Q® system (Millipore, Milford, MA,
224
USA). The mobile phases were degassed with nitrogen prior to use (99.99973% purity
225
cylinder from LINDE, Brazil, with 2nd-stage regulator from Inpagás).
226
The standards used were: D (-)– mannitol, D (-)– arabinose, D (+)– galactose, D
227
(+)– glucose, D (+)– xylose, D (+)– mannose, D (-)– fructose, all from Merck
228
(Darmstadt, Germany). Due to high hygroscopicity of carbohydrates, the standards were
229
stored in a glass desiccator under vacuum over phosphorus pentoxide (Merck,
230
Darmstadt, Germany) and utilized only after one week desiccation.
231
2.3. Standard Solutions
232
2.3.1. HPLC-HPAEC-PAD Analysis
233
For the preparation of the carbohydrate standard stock mix solution, 0.0030 g of
234
mannitol, 0.0300 g of arabinose, 0.1200 g of galactose, 0.0450 g of glucose, 0.0120 g of
235
xylose, 0.0900 g of mannose, and 0.0450 g of fructose were weighed, added to a 100.0
236
mL volumetric flask and made up to the mark with ultrapure water. The solution was
237
sonicated in an ultrasonic bath for 10 min (Garcia, Pauli, Cristiano, Camara, Scarminio,
238
Nixdorf, 2009).
239
The identification and quantification of the carbohydrates were performed on the
240
basis of retention times of components eluted from the column, comparing them the
241
retention times of the components with known concentrations of individual external
242
standards, and by co-chromatography.
243
For the carbohydrate quantification in the samples, a 10% (v/v) mix of analytical
244
standards was injected into ultrapure water. This standard mix corresponded to the
245
following concentrations in relation to 0.3000 g of sample: 0.10% (w/w) of mannitol, 10
246
1.00% (w/w) of arabinose, 4.00% (w/w) of galactose, 1.50% (w/w) of glucose, 0.40%
247
(w/w) of xylose, 3.00% (w/w) of mannose, and 1.50% (w/w) of fructose.
248
2.3.2. Post-Column Derivatization Reaction HPLC-UV-Vis Analysis
249
For the preparation of the carbohydrate standard stock mix solution, 0.0300 g of
250
glucose, 0.0200 g of xylose, 0.1100 g of galactose, 0.0400 g of arabinose, and 0.0600 g
251
of mannose were weighed, transferred to a 100.00 mL volumetric flask and made up to
252
the mark with ultrapure water. The solution was sonicated in an ultrasonic bath for 5
253
min (Pauli, Cristiano, Nixdorf, 2011). The standard was stored in a refrigerator at ~4°C.
254
This stock solution was diluted to obtain a 25% (v/v) analytical standard, which was
255
injected each quantification day.
256
2.4. Chromatographic Systems
257
2.4.1. HPLC-HPAEC-PAD
258
For the chromatographic analysis, an instrumental system was used, which
259
consisted of a PEEK inert liquid chromatograph composed of two Nalgene® bottles for
260
the mobile phase storage; a LC-10Ai inert high-pressure pump (Shimadzu®, Tokyo,
261
Japan); a Research-1367-72 3-way solenoid low-pressure valve with a "lab-made"
262
solenoid valve activation external circuit for eluent exchange; a SIL-Prominence 20A
263
automatic injector (Shimadzu®, Tokyo, Japan); a CarboPac PA1 pre-column and a
264
CarboPac PA1 polystyrene-divinylbenzene-based high-performance anion-exchange
265
resin column (10 µm, 250 mm × 4 mm; Dionex, Sunnyvale, CA, USA); a Waters
266
thermostatic column oven coupled with a 650 CHX temperature controller (Pickering
267
Laboratories); a RS570 current amplifier (Stanford Research System); a dual supply of
268
±12 V used for activation of an external circuit for electrical signal transmission; an ED11
269
50 electrochemical cell (Dionex, Sunnyvale, CA, USA); an Autolab PGSTAT 30
270
potentiostat (Eco-Chemie, Utrecht, Netherlands), an Autolab interface (Eco Chemie,
271
Utrecht, Netherlands); a data acquisition and processing system on the basis of Pentium
272
IV working with the GPES (General Purpose Electrochemical System)–Eco software.
273
For the integration of chromatograms, the INTEGRA software, developed by
274
Professor Dr. Carlos Alberto Paulinetti da Camara (DIA Group, Department of
275
Chemistry, State University of Londrina, Brazil), was employed. It makes use of Eq. (1)
276
to quantify the carbohydrate content (ω), expressed as a percentage of mass, by
277
implementing external standardization:
278
279
Am0V x100 A0 mVst
280
where,
281
A – peak area for the individual carbohydrate in the sample;
282
A0 – peak area for the individual carbohydrate in the standard solution;
283
m – mass of the sample aliquot, expressed on a dry weight basis, in g;
284
m0 – mass of the carbohydrate in the standard solution, in g,
285
V – volume of the sample, in mL;
286
Vst – volume of the standard solution, in mL.
287
2.4.2. Post-Column Derivatization Reaction HPLC-UV-Vis
(1)
288
The instrumental system used consisted of a high-performance liquid
289
chromatograph composed of two amber bottles for the storage of the mobile phase
290
eluent and the post-column reagent; two Waters 515 high-pressure pumps (Milford,
291
MA, USA); a Waters 717 Plus automatic injector; a SP-1010P pre-column (6 x 50 mm;
292
Shodex, NY, USA); an Aminex HPX-87P cation-exchange column in the Pb2+ form (9 12
293
µm, 7.8 x 300 mm; Bio-Rad, Richmond, CA, USA); a tee mixer; a thermostatic column
294
oven coupled with a Waters Module II temperature controller; a CRX 390 post-column
295
reactor (Pickering Laboratories, Mountain View, CA, USA) containing a stainless-steel
296
reaction coil inside the controlled temperature reactor; a pulse dampener with back
297
pressure of 100 psi connected to the detector output, a Waters 2487 UV-Vis detector;
298
and the Empower Build 1154 software for data acquisition and processing with a
299
microcomputer.
300
2.5. Chromatographic Analyses
301
For the extraction of the coffee carbohydrates, triticale and acai, the ISO 11292
302
standardized method (ISO, 1995) was implemented according to the adaptation
303
described above.
304
The chromatographic conditions used for the two validated HPLC systems are:
305
electrochemical detection HPAEC-PAD: pre-column: CarboPac PA1; column:
306
CarboPac PA1: 250 mm x 4 mm, 10 μm; mobile phase composition: eluent and
307
equilibrium: 1.4 mmol L-1 of NaOH and regeneration: 300.0 mmol L-1 of NaOH; flow
308
rate: 1.0 mL min-1; detection: determination potential:+0.20 V (400 ms), oxidation
309
potential: +0.65 V (200 ms), reduction potential: -0.20 V (400 ms); injection volume:
310
20.0 µL; column temperature: 28ºC and chromatographic run-time: 72.60 min (Garcia
311
Pauli, Cristiano, Camara, Scarminio,
312
derivatization: pre-column: SP-1010P; column: Aminex HPX-87P; mobile phase: eluent
313
composition (pump 1) - ultrapure water, post-column (Pump 2) – ABH + NaOH; flow:
314
pump 1: 0.5 mL min-1, pump 2: 0.6 mL min-1; detection: 410 nm; injection volume: 20.0
315
µL, column temperature: 85ºC, post-column reactor temperature: 100ºC and
316
chromatographic run-time: 25 min (Pauli, Cristiano, Nixdorf, 2011).
Nixdorf, 2009). UV-Vis post-column
13
317
The accuracy for both methods previously cited (HPLC-HPAEC-PAD and HPLC-
318
UV-Vis) was calculated by the recovery rate of analyte, which was done in triplicate, by
319
adding into the sample in proportion of 1:1 (v/v) of standard in low concentration level
320
(50%), medium (100%) and high (150%), according to calibration curve in the dynamic
321
range, calculated by Eq. (2).
C C2 rec (%) 1 x 100 C3
322 323
where,
324
rec (%) = percentage of recovery;
325
C1 = concentration of analyte in the spiked sample with standard addition;
326
C2 = concentration of analyte in the original sample without spiked standard;
327
C3 = concentration of the analyte standard added to the sample spiked.
328
(2)
Results were expressed as mean recoveries from the low, medium and high
329
concentrations levels.
330
3. RESULTS AND DISCUSSION
331
3.1.
Evaluation of Separation of Carbohydrates
332
For separation system employing HPLC-UV-Vis post-column derivatization, after
333
testing three columns, we chose to use a divalent cation lead - Aminex HPX-87P, as it
334
had the highest resolution compared to the other two - a divalent column of calcium and
335
the other a monovalent of hydrogen. By being cationic, their use required a higher
336
temperature (85oC) which discourages the interaction, as can be observed by rapidly
337
eluting peaks, impairing resolution (Fig. 3).
338
The variation of - solvent, flow, pH and ionic strength, to improve the selectivity
339
(Lanças, 2004) were not feasible in these experiments, since the strength of the mobile
14
340
phase couldn´t be varied; by the fact of Aminex column does not allow the use of
341
organic solvents. The flow rate could not be reduced to increase interaction, since was
342
already low (0.5 mL min-1). Adding salt for change the ionic strength favored the
343
competition with the active sites disadvantaging the interaction between the counter-ion
344
of the stationary phase and the carbohydrates, resulting in a worsening in the resolution
345
between the peaks. In this case also, it was not possible ionize the sample, using a pH
346
two points above of the pKa of the carbohydrates (12.08 - 12.35 - Table 1), as
347
recommended by Lanças (2004), since the pH range of this column is restricted to 5.0-
348
9.0. This justify that the hydrolyzed sample should be neutralize before analysis. Thus,
349
not as many remaining alternatives to the use of different mobile phases, the ultrapure
350
water were adopted. The neutral medium favors broadening and flattening of
351
chromatographic peaks in the ion exchange mode, helping in the inadequate resolution,
352
observed in the chromatogram of the UV-Vis system with a partial co-elution of
353
arabinose and mannose (peaks 4 and 5 – Fig. 3).
354
Adding to this, for sure the major factor contributing in proportion to the low
355
resolution is the way in which the chemical structures of carbohydrates (aldoses) are in
356
the aqueous medium. Table 1 permits us to see that there are higher proportions of
357
pyranose, compared to furanose, once the cycle of six members is thermodynamically
358
more stable in aqueous medium (Inoue, Kitahara, Aikawa, Arai, Masuda-Hanada,
359
2011). Mute rotations of the anomeric carbon also in Table 1, shown that there is a
360
predominance of the alpha pyranose form. This is justified by the hydroxyl group in the
361
alpha configuration is pointing down, while in beta form hydroxyl is pointing upwards
362
(Fig. 1), so that the aligning two heteroatoms partially suffer repulsion. It was also
363
observed, from the data of Table 1 that in equilibrium in the aqueous medium, there is
364
predominance to the β form for glucose (62%), xylose (63%) and galactose (64%),
15
365
while is a predominance of α form for arabinose (60%) and mannose (64%), the form
366
more stable and retained in chromatography. Considering the aldopentoses, the
367
arabinose has a superior retention than the xylose, since it has a higher proportion of
368
furanose (2.5%) against (
