Journal of Chromatography A, 1065 (2005) 207–210

Separation and quantification of beer carbohydrates by high-performance

liquid chromatography with evaporative light scattering detection
Luciana C. Nogueiraa , Filipe Silvab , Isabel M.P.L.V.O. Ferreirab,∗ , L.C. Trugoa
a Laboratório de Bioquı́mica Nutricional e de Alimentos, Instituto de Quı́mica, Universidade Federal do Rio de Janeiro, Brazil
b REQUIMTE, Serviço de Bromatologia, Faculdade de Farmácia, Universidade do Porto, Rua Anı́bal Cunha, 4050-047 Porto, Portugal

Received 19 November 2004; received in revised form 17 December 2004; accepted 29 December 2004
Available online 18 January 2005

Abstract

An HPLC method with an evaporative light scattering detector was optimized and validated for quantification of carbohydrates in beer. The
chromatographic separation was achieved using a Spherisorb NH2 , 5 ␮m chromatographic column and gradient elution with acetonitrile/water.
The determinations were performed in the linear range of 0.05–5.0 g/L for fructose, 0.05–5.0 g/L for glucose, 0.05–15.0 g/L for maltose,
0.05–10.0 g/L for maltotriose, and 0.05–5.0 g/L for maltotetraose. The detection limits were 0.005 g/L for fructose, 0.008 g/L for glucose, and
0.01 g/L for maltose, maltotriose, and maltotetraose. The reliability of the method in terms of precision and accuracy was evaluated in three
beer matrices, low alcohol beer, 6% alcohol beer, and beer made with part of adjuncts (4.5% alcohol). Relative standard deviations (RSDs)
ranged between 1.59 and 5.95% (n = 10), and recoveries ranged between 94 and 98.4%.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Beer; Carbohydrates

1. Introduction contemplated [5]. In this respect it is important to control not

only the total amount of fermentable carbohydrates formed
Beer is a fermented beverage made from malted grains but also the relative amounts of the different sugars.
(usually barley), hops, yeast, and water [1,2]. It has a com- The analysis of carbohydrates is generally carried out
plex composition, containing a vast number of compounds. by high-performance liquid chromatography (HPLC), which
Major beer components are water, ethanol and carbohydrates can provide not only the qualitative analysis but also the quan-
comprising fermentable sugars (i.e., fructose, glucose, mal- titative determination [6]. The main chromatographic sys-
tose, and maltotriose) as well as glucose oligosaccharides tems used for the separation of underivatized carbohydrates
[3]. Fermentable sugars directly contribute to the sweetness can be generalized as anion-exchange column with water
of beer, whereas carbohydrates with more than four glycosyl containing bases or salts as the eluent [7]; cation-exchange
units can be beneficial to the perception of beer in that they column with water as the eluent [8]; alkyl-bonded silica gel
contribute to body or mouthfeel [4]. column with water as the eluent [9] and amine-bonded silica
Quantitative evaluation of malto-oligosaccharides pro- gel column with water–acetonitrile as the eluent [8,10–14].
vides a useful control of the complex enzymatic system in Of these systems, an amine-bonded silica gel column is the
beer brewing, particularly when changes in procedure are one mostly used.
Refractive index (RI) [8,9,12–15] measurement is the
most popular detection method for carbohydrates. How-
ever, it has many disadvantages, such as lacking sensitiv-
∗ ity, temperature and flow-rate dependent, and incompatibility
Corresponding author. Tel.: +351 222078929; fax: +351 222003977.
E-mail addresses: [email protected] (L.C. Nogueira), is- with gradient elution. Evaporative light scattering detection
[email protected] (I.M.P.L.V.O. Ferreira). (ELSD) [16] is widely used as a semi-universal mass detector

0021-9673/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2004.12.074
208 L.C. Nogueira et al. / J. Chromatogr. A 1065 (2005) 207–210

for HPLC. It is based on the detection of solute molecules Gradient elution was carried out with a mixture of two sol-
by light scattering after nebulization and evaporation of the vents. Solvent A consisted of acetonitrile and solvent B con-
mobile phase, so it is suitable to detect the nonvolatile com- sisted of water. Carbohydrates were eluted increasing the pro-
pounds such as lipids [17,18] and carbohydrates [8,10,11]. portion of solvent B from 19 to 25% over 40 min: 0–19 min,
However, some disadvantages of this detector are described, 19% B; 20–40 min, 25% B. The flow-rate was 1 mL/min.
because its response is proportional to the compound mass The temperature of the heated drift tube was 45 ◦ C, the gas
not being limited by their spectral feature, some authors find pressure was 3.0 bar, and gain 5.
that quantification by ELSD may represent a problem, since
it is described that response factor is linear at very low or 2.5. Statistical analysis
high levels of analytes [19].
In this work, a HPLC method with an evaporative light Data are presented as the mean ± SD. The results were
scattering detector was optimized and validated for quantifi- statistically analyzed by analysis of variance (ANOVA). Dif-
cation of carbohydrates in beer. ferences were considered significant for p < 0.05. Statistical
analyses were all performed with SPSS for Windows version
11.5 (SPSS, Chicago, IL).
2. Experimental

2.1. Sampling 3. Results and discussion

Three different samples of pilsner type, were analyzed. 3.1. Separation and quantification of carbohydrates
The first two samples were prepared from the same wort,
100% malt, and fermented during 24 h (sample 1: low alco- The optimization of the HPLC procedure was focused on
hol beer, sample 2: 6% alcohol beer, added with alcohol of the chromatographic separation as well as on the detector op-
cereals). The third sample was added with part of adjuncts erational parameters. The stationary phase selected required
(sample 3: standard pilsner, added with high maltose syrup, the use of acetonitrile–water mixture for peak separation.
4.5% alcohol). Different gradient conditions were assayed at a flow rate
of 1 mL/min, with the aim of obtain a well-resolved chro-
2.2. Sample preparation matogram.
The three instrumental parameters affecting sensitivity
The beer samples were degassed for 15 min in an ultra- were: temperature of the detector, the nebulizing gas pressure
sonic bath model Bandelin Sonorex RK (Bandelin, Berlin, and the gain. The temperature of the detector was studied in
Germany), diluted (1:2) in acetonitrile and filtered through the interval 40–60 ◦ C. A temperature of 45 ◦ C was enough
a 25 mm organic syringe filter (0.2 ␮m pore size). All the to allow complete solvent evaporation and therefore giving
samples were stored at 10 ◦ C. negligible noise. A flow-rate (pressure) of air set at 3 bar and
a gain of 5 provided a good sensitivity and adequate signal-
2.3. Reagents and carbohydrate standards to-noise ratio. Fig. 1 shows a typical chromatogram for sep-

All reagents used were of analytical grade purity. Sol-

vents for HPLC were filtered trough 0.22 ␮m NL 17 fil-
ters and degassed under vacuum for at least 15 min before
use. Maltotriose, maltotetraose and fructose were supplied
by Sigma (St. Louis, MO, USA), glucose was supplied by
Merck (Darmstadt, Germany) and maltose was supplied by
Fluka. Standard solutions were prepared in a mixture of
water–acetonitrile (50:50, v/v).

2.4. Apparatus

The chromatographic analysis was carried out in an analyt-

ical HPLC unit (Jasco, Japan), equipped with a low pressure
quaternary pump (PU-1580), an evaporative light scattering
Fig. 1. Typical chromatogram for separation of five carbohydrates (chro-
detector (LSD, Sedex 75, France) and a type 7125 Rheo-
matographic conditions described in the text): (1) fructose (tR 8.76 min),
dyne injector with a 10 ␮L loop. A Borwin Controller Soft- (2) glucose (tR 9.71 min), (3) maltose (tR 15.45 min), (4) maltotriose (tR
ware (JMBS Developments) was also used. The column was 22.21 min), (5) maltotetraose (tR 29.41 min). The concentration of standards
a Spherisorb NH2 , 5 ␮m, 250 mm × 4.6 mm i.d. in the mixture was 2 g/L.
 L.C. Nogueira et al. / J. Chromatogr. A 1065 (2005) 207–210 209

Table 1
Calibration curves determined by the external standard method
Carbohydrates Concentration range (g/L) na Slopeb (area counts/g) Interceptb (area counts) rc
Fructose 0.50–2.5 5 2728.9 −339.39 0.9988
Glucose 0.50–5.0 5 2514.9 −111.31 0.9932
Maltose 0.50–15.0 5 2324.2 −200.2 0.9997
Maltotriose 0.50–10.0 5 2827.4 −800.09 0.9967
Maltotetraose 0.50–5.0 4 1979.6 −294.95 0.9943
a Number of points considered for the regression. Each point represents the average of three injections of each standard solution.
b Standard deviation in the slope.
c Correlation coefficient.

aration of the five carbohydrates used in the optimization although the matrix composition is complex, it does not cause
process. interference effects. The presence of alcohol and other com-
The external standard method was used to calibrate the pounds from fermentation does not affect the accuracy, be-
chromatographic system for carbohydrates quantification. cause no significant differences were found between samples
For this purpose, sugars standard solutions with different 1 and 2, with similar composition except alcohol content, and
concentrations (0.05–5.0 g/L for fructose, 0.05–5.0 g/L for between sample 3 that suffered higher fermentation and addi-
glucose, 0.05–15.0 g/L for maltose, 0.05–10.0 g/L for mal- tion of adjuncts. A statistical one-way ANOVA test was used,
totriose, and 0.05–5.0 g/L for maltotetraose) were used, ac- in order to verify whether the average recoveries obtained, for
cording to the quantity of these compounds in the beer matrix. each variable (fructose, glucose, maltose, maltotriose, mal-
Each solution was analyzed in triplicate. totetraose) could be considered different or not, for the three
Calibration curves between peak areas and the mass of an- samples. This turned to be possible as recovery values, for
alyte injected were linear for the five carbohydrates following each sample, had a normal distribution (Shapiro–Wilk Test)
the equation Y = aX + b. The values of the slope, intercept and and homocedasticity of variances (Levene Test). Results ob-
correlation coefficient are given in Table 1. These results are tained for ANOVA showed that, with 95% confidence, that
in good agreement with those obtained by other authors [20]. there were no significant differences between recovery val-
Identification of the carbohydrates in beers was performed ues obtained for the three samples analyzed concerning the
by comparison with the retention times of the standards. The carbohydrates in study.
detection limit values were estimated as the concentration
providing a signal three times higher than the standard devia- 3.3. Chromatograms and results for beer samples
tion of the background noise. Thus, successive dilutions were
performed to find the smallest concentrations that could be Figs. 2 and 3 show the typical chromatograms for samples
measured without confusion with background noise. Detec- 1 and 3, respectively.
tion limits were 0.005 g/L for fructose, 0.008 g/L for glucose, Table 2 presents the concentrations of fructose, glucose,
and 0.01 g/L for maltose, maltotriose, and maltotetraose, re- maltose, maltotriose, and maltotetraose in beer samples. As
spectively. expected samples 1 and 2 provided similar quantitative carbo-
hydrate profile, these samples were prepared from the same
wort and suffered short fermentation, thus, great part of fruc-
3.2. Validity of the method tose, glucose and maltose remained in beer. Sample 3 pre-

The reliability of the method in terms of precision and

accuracy was evaluated in the three beer matrices, low alcohol
beer, 6% alcohol beer, and beer made with part of adjuncts
(4.5% alcohol).
The precision of this method was evaluated taking into
account its relative standard deviation (RSD) for 10 analy-
ses of each beer sample. RSDs ranged between 2.78–2.89%,
3.02–3.11%, 1.59–4.77%, 2.89–4.79%, and 4.12–5.95%, re-
spectively, for fructose, glucose, maltose, maltotriose, and
maltotetraose.
Recovery studies were carried out to determine the accu-
racy of the method. Samples were analyzed before and after
the addition of known amounts of fructose, glucose, maltose,
maltotriose, and maltotetraose it was found that recoveries Fig. 2. Typical chromatogram for beer sample 1. The numbers correspond
ranged between 94 and 98.4%. These results confirmed that to the numbers in Fig. 1 with respect to peak identification.
210 L.C. Nogueira et al. / J. Chromatogr. A 1065 (2005) 207–210

Table 2
Results obtained in the monitoring of carbohydrates in beersa
Samples Fructose Glucose Maltose Maltotriose Maltotetraose
1 2.6 ± 0.04 a 4.6 ± 0.5 a 38.5 ± 0.6 a 8.1 ± 0.2 a 1.4 ± 0.1 a
2 2.4 ± 0.07 a 4.2 ± 0.2 a 38.3 ± 0.4 a 7.4 ± 0.3 a 1.1 ± 0.1 a
3 nd b nd b 0.35 ± 0.00 b 1.3 ± 0.1 b 2.9 ± 0.1 b
Values are expressed as mean ± SD of two determinations (g of carbohydrate/L). (a, b) means columns without common superscripts are significantly
a

different (p < 0.05).

References

[1] D. De Keukeleire, Quim. Nova 23 (2000) 108.

[2] I.F. Duarte, M. Godejohann, U. Braumann, M. Spraul, A.M. Gil, J.
Agric. Food Chem. 51 (2003) 4847.
[3] V. Brandolini, E. Menziani, D. Mazzotta, P. Cabras, B. Tosi, G. Lodi,
J. Food Composition Anal. 8 (1995) 336.
[4] C. Corradini, G. Canali, I. Nicoletti, Sem. Food Anal. 2 (1997) 99.
[5] A. Plaga, J. Stumpfel, H. Fiedler, Appl. Microbiol. Biotechnol. 32
(1989) 45.
[6] Y. Wei, M. Ding, J. Chromatogr. A 904 (2000) 113.
[7] N. Torto, T. Buttler, L. Gorton, G. Marko-Varga, H. Stalbrand, F.
Tjerneld, Anal. Chim. Acta 313 (1995) 15.
Fig. 3. Typical chromatogram for beer sample 3. The numbers correspond [8] A. Clement, D. Young, C. Brechet, J. Liq. Chromatogr. 15 (1992)
to the numbers in Fig. 1 with respect to peak identification. 805.
[9] E. Rajakyla, J. Chromatogr. 353 (1986) 1.
[10] R. Macrae, J. Dick, J. Chromatogr. 210 (1981) 138.
sented significantly lower sugar content owing to an extended [11] R. Macrae, L.C. Trugo, J. Dick, Chromatographia 15 (1982) 476.
[12] Z.L. Nikolov, M.M. Meagher, P.J. Reilly, J. Chromatogr. 319 (1985)
fermentation process. 51.
[13] I.M.P.L.V.O. Ferreira, M.A. Ferreira, J. Liq. Chromatogr. Rel. Tech-
nol. 20 (1997) 3419.
4. Conclusions [14] I.M.P.L.V.O. Ferreira, A.M.P. Gomes, M.A. Ferreira, Carbohydr.
Polym. 37 (1998) 225.
[15] M. Boumahraz, V.Y. Davydov, A.V. Kiselev, Chromatographia 15
Beer monosaccharides and malto-oligosaccharides can be (1982) 751.
separated easily using a simple gradient elution profile. No [16] Y. Mengerink, H.C.J.D. Man, S. van der Wal, J. Chromatogr. 552
chemical manipulation is involved and no derivatization is (1991) 593.
needed. Sensitivity and stability of baseline is better than [17] B. Marcato, G. Cecchin, J. Chromatogr. A 730 (1996) 83.
that achieved using RI detection frequently described in lit- [18] C.D. Chang, D.J. Harris, J. Liq. Chromatogr. Rel. Technol. 21 (1998)
1119.
erature. Appropriate accuracy and precision was obtained in [19] J.S. Perona, L.J.R. Barrón, V. Ruiz-Gutiérrez, J. Chromatogr. A 706
the analyses of beer samples with variable alcohol content (2001) 173.
and different composition. [20] Y. Wei, M.Y. Ding, J. Chromatogr. A 904 (2000) 113.