International Journal of Analytical Chemistry

International Journal of Analytical Chemistry / 2012 / Article

Research Article | Open Access

Volume 2012 |Article ID 840528 |

Regina F. R. Pereira, Carla B. Vidal, Ari C. A. de Lima, Diego Q. Melo, Allan N. S. Dantas, Gisele S. Lopes, Ronaldo F. do Nascimento, Clerton L. Gomes, Maria Nataniela da Silva, "Chemometric Characterization of Alembic and Industrial Sugar Cane Spirits from Cape Verde and Ceará, Brazil", International Journal of Analytical Chemistry, vol. 2012, Article ID 840528, 6 pages, 2012.

Chemometric Characterization of Alembic and Industrial Sugar Cane Spirits from Cape Verde and Ceará, Brazil

Academic Editor: Richard G. Brereton
Received13 Jun 2012
Accepted22 Aug 2012
Published25 Nov 2012


Sugar cane spirits are some of the most popular alcoholic beverages consumed in Cape Verde. The sugar cane spirit industry in Cape Verde is based mainly on archaic practices that operate without supervision and without efficient control of the production process. The objective of this work was to evaluate samples of industrial and alembic sugar cane spirits from Cape Verde and Ceará, Brazil using principal component analysis. Thirty-two samples of spirits were analyzed, twenty from regions of the islands of Cape Verde and twelve from Ceará, Brazil. Of the samples obtained from Ceará, Brazil seven are alembic and five are industrial spirits. The components analyzed in these studies included the following: volatile organic compounds (n-propanol, isobutanol, isoamylic, higher alcohols, alcoholic grade, acetaldehyde, acetic acid, acetate); copper; and sulfates.

1. Introduction

Sugar cane spirit, or cachaça, is a typical Brazilian distilled beverage [1, 2]. This spirit is the second most consumed alcoholic beverage in the country and the third most consumed in the world. There are almost 30,000 cachaça producers in Brazil and over 5,000 cachaça brands available on the market [3].

In Cape Verde, the sugar cane spirit, grogue, is a drink produced mainly by archaic practices, without supervision and without efficient control of the production process. To ensure the quality of grogue, it is necessary to correct deviations over the course of the whole production process. For Cape Verde, because there is still no local legislation, the results are evaluated according to Brazilian legislation.

Mathematical and statistical methods of analysis can be used for diverse scientific purposes, such as selecting the measurements and procedures best suited to chemical experiments or obtaining a more accurate analysis of the resulting information. According to the needs of any particular study, chemometrics can be used for analytical signal processing, experimental planning and optimization, pattern recognition, data classification, multivariate calibration, and/or monitoring and modeling of various processes, among other applications [47].

One of the first steps of chemometric analysis is to plot data in a multidimensional space, grouping the data with similar characteristics to demonstrate that there is some natural relationships between these data points. Thus, groups with distinct characteristics will be differentiated.

Exploratory multivariate analysis is performed in a matrix, and the data are organized in a spreadsheet where “ ” samples with “ ” variables results in a matrix [8, 9].

Two methods were applied: the supervised method, which is applied when the classes of the samples are known and these data are used to build the model; and the unsupervised method, which is applied when there is no previous knowledge of sample classification and the samples are grouped according to the analysis of the acquired data.

For pattern recognition, the chemometric method can be applied via hierarchical cluster analysis (HCA) or via principal component analysis (PCA). These techniques are based on the assumption that similar samples will be plotted closer together in multidimensional space than dissimilar samples [810].

Principal component analysis was originally described by Pearson in 1901 and was subsequently applied by Hotelling in various scientific areas [810].

Principal component analysis consists of redefining the coordinates of an axis system to make it more convenient for the analysis of a given set of samples. That is, original variables generate, through their linear combinations, principal components, which are obtained in decreasing order of maximum variance. Thus, the first principal component has more statistical information than the second principal component, which, in turn, has more statistical information than the third principal component, and so on. This method allows reduction of the dimensionality of the representative points of the samples [9, 11].

The graph of principal component 1 versus principal component 2 provides window for the observation of data points in -dimensional space. Principal component analysis can also be used to judge the importance of the original variables chosen. The original variables receive the highest weighting, or loading, in the linear combination of the first principal components and are thus the most important from a statistical point of view. The task of the professional who works with multivariate statistics is to interpret the point distributions on the graphs of the principal components and identify the original variables with the most weight in a linear combination of the most important principal components [11, 12].

The technique of hierarchical clustering draws connections between the samples, producing a dendrogram in which similar samples are grouped together, and this similarity is a function of the distance between the points.

The goal of this work was to develop analytical methods for the chemical characterization of sugar cane spirits and to use multivariate analysis to distinguish between the spirits according to their origin, quality, variety, and type.

2. Experimental

2.1. Instrumentation

A Shimadzu gas chromatograph (model GC-17A-FID) was used for the determination of volatile organic compounds present in samples of sugar cane spirits. The chromatographic column used was a DB-5 (30mx 0.25mm ID, 0.25mm thick film stationary phase of 5% phenyl and 95% dimethylsiloxane).

A Dionex ion chromatograph (ICS-3000) was used for the determination of anionic species. The chromatographic column used was an AS18 column (25cm 4.0mm with an ASRS suppressor—Ultra II—4mm). Other components used included a conductivity detector, a 25μL loop, and a potassium hydroxide (KOH) eluent generator.

Copper analysis was performed using a simultaneous inductively coupled plasma optical emission spectrometer with axial and radial views (Optima 4300 series, Perkin Elmer).

Acid digestions were performed in a digestion block (Tecnal, Piracicaba, SP, Brazil) equipped with Teflon vessels.

2.2. Reagents, Solutions, and Samples

All reagents used were of analytical grade, and the water (resistivity of 18.2MΩcm) was purified in a Milli-Q system (Millipore, Bedford, MA, USA).

The samples were collected from three islands of Cape Verde (Santiago, Santo Antão, and Brava) and also from Brazil (Ceará). Table 1 presents the sample symbols used in this work and identifies the samples according to the region and the manufacturing process used.

Cape VerdeSymbol

S. AntãoV3—Industrial
S. AntãoV12—Industrial
S. AntãoV18—Alembic
Ceará B1—Industrial

Analytical grade solvents were used. Ethanol (VETEC, Rio de Janeiro, Brazil) had a purity of 99.5%. The analytical standards used, which were all 99.9% pure unless otherwise noted, were as follows: acetaldehyde (VETEC, Rio de Janeiro, Brazil); n-propanol (Merck, USA); ethyl acetate (Merck, USA); isobutanol (Merck, USA); butanol (Merck, USA) and isoamyl alcohol (Merck, USA).

Stock aqueous solutions of copper (1000 mgL−1) were purchased from Acros Organics (New Jersey, USA). The analytical calibration curve was prepared using 1.0, 5.0, 10, and 12 mgL−1 solutions of copper in 1.4 molL−1 HNO3.

2.3. Procedures

For analysis of the volatile compounds, the oven’s temperature program was as follows: an initial temperature of 40°C for 3 min which was increased at a rate of 5°Cmin−1 to 65°C, followed by a rate of 50°Cmin−1 to a final temperature of 200°C. The temperatures of the injector and the detector were 200°C and 300°C, respectively. Direct injection of samples was performed in split mode, with 1 mL of sample injected for gas chromatographic analysis.

Sulfate analysis was performed using an elution program described as follows: 7.2 mmol between 0 and 10 min, increased to 22 mmol by 13.2 min, increased to 44 mmol by 13.44 min, and maintained at 44 mmol until the 15.0 min. The column temperature was 40°C. The current was 40 mA, and the conductivity was 0.6μS.

Copper was investigated after digestion of the samples. Aliquots of 2.0 mL sugar cane spirits and 2.0 mL HNO3 (68% w/w) were transferred to a Teflon vessel. The heating program of the digestion block was adjusted to increase the temperature to 120°C for 3 h.

HCA was implemented in R using a stats package [14, 15] and was performed using a variation of Ward’s linkage method [16], which adopted Euclidean distances as measures of dissimilarity. PCA was implemented using the FactoMineR (Factor analysis and data mining with R) package [17]. The approach used to implement PCA in FactoMineR is described in detail elsewhere [18]. In FactoMineR, the solution which maximizes the variance of the projected points is selected. Although rotational algorithms can help to improve the interpretability of the principal components, the FactoMineR method does not apply any rotation to keep its optimal property, the maximization of the variance of the projected points.

3. Results and Discussion

The results of the chromatographic separation of the components present in the spirits from Cape Verde are shown in Figure 1. It was observed that total separation of the components occurred at a runtime of 4 min. Among the higher alcohols, the area of the isoamyl alcohol peak is remarkable not only for being the largest but also because it reveals a mixture of isomers (n-amyl and isoamyl alcohols).

The concentrations of each component were evaluated according to Brazilian legislation using the values obtained from the chromatographic analysis. The results are shown in Table 2.


Higher alcohols
Alcoholic grade
Acetic acid

*AA: Absolute alcohol; BA: Brazilian alembic; BI: Brazilian industrial spirits; CVA: Cape Verde alembic; CVI: Cape Verde industrial spirits.

The acetaldehyde content is an important parameter in controlling the processing quality of spirits. In Cape Verde spirits, the average acetaldehyde contents were 20 and 15.2 mg/100 mL AA for industrial and alembic spirits, respectively. For the Brazilian spirits, the average acetaldehyde contents were 21.8 and 11.9 mg/100 mL AA for the alembic and industrial, respectively (Table 4). Some spirit samples from Cape Verde (V3, V5, V7, V13, V19) and Brazil (B10 and B18) exceeded the limit of 30 mg/100 mL AA stipulated by Brazilian legislation.

According to Almeida and Barreto, there is a correlation between the level of n-propanol and the quality of the spirits, such that the presence of higher levels of n-propanol occurs in lower-quality spirits [19]. In this work, the highest levels of n-propanol were found in the alembic spirits of Cape Verde.

Among the alcohols identified in Brazilian spirits, isoamyl alcohol is predominant, with average concentrations of 50.2 mg/100 mL AA and 120.3 mg/100mL AA for industrial and alembic spirits, respectively. In Cape Verde spirits, the averages were 100.2 mg/100 mL AA and 164.4 mg/100 mL AA for alembic and industrial spirits, respectively.

The sum total of the higher alcohol concentrations is low, with the exception of some Cape Verde samples (C8, C17, and C20), which exceeded the limit permitted by Brazilian legislation [20, 21].

Spirits produced in copper stills are in high demand because of their unique taste [22]. However, the amount of copper in these spirits has been of concern as health problems can result when this element occurs in high concentrations [23]. The Ministry of Agriculture, Livestock and Supply (MAPA) has set a maximum limit of 5 mg L−1 for the distillation process [21].

The copper concentrations of the samples show that more than fifty percent of the Cape Verde alembic spirits have levels above the maximum level of 5.0 mg L−1. Among the Brazilian spirits, only two samples exceeded the maximum level.

One proposal to lower the copper content in the Cape Verde spirits is to replace the copper pot stills with stainless steel ones, thus eliminating the presence of copper salts available to be dissolved by the acid vapors during distillation and carried over to the final distillate [24].

Other studies with Brazilian spirits have shown that the concentration of higher alcohols may vary depending on the region (Table 3).

Spirits/regionComponentsValues (mg/100 mL AA)Reference

Minas Gerais; BrazilAcetic acid81.93 [7]
Higher alcohols179.13

São Paulo; Brazil (copper)Acetic acid2.07 [13]
Higher alcohols135.65

São Paulo; Brazil (steel)Acetic acid2.15 [13]
Higher alcohols187.42

CE; Brazil (steel)Acetic acid17.5 This work
Higher alcohols235.2


Alcoholic grade0.6044−0.47890.1078−0.0229
Acetic acid0.5419−0.49540.09320.4859
Higher alcohols0.62570.73440.1686−0.0009

Nascimento et al. studied the components of spirits from Sao Paulo and obtained concentrations of 135.62 and 187.42 mg/100 mL AA in copper pot stills and stainless steel stills, respectively. In this work an average of 235.2 mg/100mL AA for alembic spirits from Ceará (Table 3) was obtained [13, 25].

Figure 2 shows the dendrogram of sugar cane spirits based on the chosen variables. It was possible to observe the formation of four main groups that separated the Brazilian sugar cane spirits from those of Cape Verde. It was noted that alembic sugar cane spirits from Cape Verde, as well as some Brazilian ones, were concentrated more in the first group (first branch) and that the Brazilian industrial spirits have a distinct profile from that of Cape Verde industrial spirits.

According to Muñoz et al., the vectors farthest from zero correspond to the variables with the greatest influence on the principal component value, while vectors closer to zero indicate variables with little influence on the principal component [26].

In this study, it was revealed that the variables with the greatest influence on the principal components 1 and 2 were sulfates and the higher-level alcohols, isobutanol and isoamylic alcohol (Table 4).

In the loadings plot (Figure 3), the principal components 1 and 2 contain negative and positive values. Principal component 1 shows that higher-level alcohols, isobutanol and isoamylic alcohol, were present in high levels in the sugar cane spirits from Cape Verde when compared to those from Brazil. Both principal components 1 and 2 presented high levels of higher alcohols in sugar cane spirits from Brazil.

In the graph of the “scores” (Figure 4), it is possible to observe that principal component 1 separates Brazilian (B) sugar cane spirits from those originating in Cape Verde (V) and discriminates between Brazilian industry-produced and alembic spirits. These observed subgroups are due to differences in quality control, and it was observed that industrial spirits from Brazil have a very narrow profile while industrial spirits from Cape Verde exhibit higher dispersion.

PCA is a useful technique for finding patterns in high-dimensional data and for plotting those data in a way that highlights the similarities and differences among the points. The first component accounts for most of the variability in the data, and each succeeding component accounts for the highest variance possible under the constraint that it is orthogonal to (i.e., uncorrelated with) the preceding components [27].

In PCA, the data matrix is decomposed into scores and loading matrices. The scores vectors describe the relationship among the samples in the model subspace, and the loadings vectors describe the importance of each descriptor within the model.

For this work, principal component 1 describes 24.49% of the original information and principal component 2 describes 23.75%. The cumulative percentage of principal components 1 and 2 was 48.24%. Four principal components were required to describe 70.78% of the original data.

4. Conclusions

Chemometric analysis enabled the extraction of more information for data analysis, promoting interpretation of the results with greater security and reliability.

The results showed that both hierarchical cluster and principal component analysis were able to separate sugar cane spirits from Cape Verde from Brazilian sugar cane spirits, thus highlighting the differences in production of sugar cane spirits between the two countries.

For alembic and industrial samples, PCA analysis indicated that sugar cane spirits from Cape Verde are disparate, while sugar cane spirits from Brazil are more closely grouped, indicative of the greater control over sugar cane spirit manufacturing in Brazil.


The authors are grateful to the CNPq, Capes, and FUNCAP for the financial support and for the A. N. S. Dantas, C. B. Vidal, A. C. A. Lima, D. Q. Melo fellowships.


  1. F. W. B. De Aquino, S. Rodrigues, R. F. Do Nascimento, and A. R. S. Casimiro, “Simultaneous determination of aging markers in sugar cane spirits,” Food Chemistry, vol. 98, no. 3, pp. 569–574, 2006. View at: Publisher Site | Google Scholar
  2. F. W. B. Aquino, L. M. Boso, D. R. Cardoso, and D. W. Franco, “Amino acids profile of sugar cane spirit (cachaça), rum, and whisky,” Food Chemistry, vol. 108, no. 2, pp. 784–793, 2008. View at: Publisher Site | Google Scholar
  3. S. A. V. Tfouni, R. M. D. Machado, M. C. R. Camargo, S. H. P. Vitorino, E. Vicente, and M. C. F. Toledo, “Determination of polycyclic aromatic hydrocarbons in cachaça by HPLC with fluorescence detection,” Food Chemistry, vol. 101, no. 1, pp. 334–338, 2006. View at: Publisher Site | Google Scholar
  4. C. A. Galinaro, D. R. Cardoso, and D. W. Franco, “Profiles of polycyclic aromatic hydrocarbons in Brazilian sugar cane spirits: discrimination between cachaças produced from nonburned and burned sugar cane crops,” Journal of Agricultural and Food Chemistry, vol. 55, no. 8, pp. 3141–3147, 2007. View at: Publisher Site | Google Scholar
  5. A. Ilin and T. Raiko, “Practical approaches to principal component analysis in the presence of missing values,” Journal of Machine Learning Research, vol. 11, pp. 1957–2000, 2010. View at: Google Scholar
  6. W. Z. Lu, H. D. He, and L. Y. Dong, “Performance assessment of air quality monitoring networks using principal component analysis and cluster analysis,” Building and Environment, vol. 46, no. 3, pp. 577–583, 2011. View at: Publisher Site | Google Scholar
  7. A. P. Fernandes, M. C. Santos, S. G. Lemos, M. M. C. Ferreira, A. R. A. Nogueira, and J. A. Nóbrega, “Pattern recognition applied to mineral characterization of Brazilian coffees and sugar-cane spirits,” Spectrochimica Acta—Part B, vol. 60, no. 5, pp. 717–724, 2005. View at: Publisher Site | Google Scholar
  8. F. Husson, J. Josse, S. Lê, and J. Mazet, FactoMineR: Multivariate Exploratory Data Analysis and Data Mining with R. R Package Version 1. 16, 2011,
  9. A. C. Rencher, Methods of Multivariate Analysis, Wiley-Interscience, New York, NY, USA, 2002.
  10. U. Contreras, O. Barbosa-García, J. L. Pichardo-Molina et al., “Screening method for identification of adulterate and fake tequilas by using UV-VIS spectroscopy and chemometrics,” Food Research International, vol. 43, no. 10, pp. 2356–2362, 2010. View at: Publisher Site | Google Scholar
  11. Y. Fu, L.-T. Lim, and Y. Kakuda, “Chemometric analysis of gas chromatographic data-investigation of enological parameters of a bag-in-box white wine as affected by storage time and temperature,” Journal of Chemometrics, vol. 25, no. 11, pp. 610–619, 2011. View at: Publisher Site | Google Scholar
  12. J. Kumirska, N. Migowska, M. Caban, A. Plenis, and P. Stepnowski, “Chemometric analysis for optimizing derivatization in gas chromatography-based procedures,” Journal of Chemometrics, vol. 25, no. 12, pp. 636–643, 2011. View at: Publisher Site | Google Scholar
  13. R. F. Nascimento, D. R. Cardoso, B. S. Lima Neto, and D. W. Franco, “Influência do material do alambique na composição química das aguardentes de cana-de-açúcar,” Química Nova, vol. 21, pp. 735–739, 1998. View at: Google Scholar
  14. R Development Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2007,
  15. R Development Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2011,
  16. J. H. Ward, “Hierarchical grouping to optimize an objective function,” JAMA, vol. 58, pp. 236–254, 1963. View at: Google Scholar
  17. F. Husson, J. Josse, S. Lê, and J. Mazet, FactoMineR: Factor Analysis and Data Mining with R. R Package Version 1.12, 2009,
  18. S. Lê, J. Josse, and F. Husson, “FactoMineR: an R package for multivariate analysis,” Journal of Statistical Software, vol. 25, no. 1, pp. 1–18, 2008. View at: Google Scholar
  19. M. E. W. Almeida and H. H. C. Barreto, “superiores em aguardente de cana por cromatografia em fase gasosa,” Revista do Instituto Adolfo Lutz, vol. 31, pp. 117–124, 1971. View at: Google Scholar
  20. Brasil. Ministério da Agricultura, Pecuária e Abastecimento, Decreto n. 2314 de de 04/09/1997.
  21. Brasil. Ministério da Agricultura Pecuária e Abastecimento, Instrução Normativa N 13 de 29 de junho de 2005. Brasília: Diário Oficial da União.
  22. D. R. Cardoso, B. S. Lima-Neto, and D. W. Franco, “Influência do material do destilador na composição química das aguardentes de cana. Parte II,” Química Nova, vol. 26, pp. 165–169, 2003. View at: Google Scholar
  23. A. J. B. Lima, M. G. Cardoso, M. C. Guerreiro, and F. A. Pimentel, “Emprego do carvão ativado para remoção de cobre em cachaça,” Química Nova, vol. 29, pp. 247–250, 2006. View at: Google Scholar
  24. J. V. Weber, “Ethylcarbamate in foods and beverages,” Environmental Chemistry Letters Rewiew, vol. 7, pp. 233–247, 2008. View at: Google Scholar
  25. R. F. Nascimento, J. C. Marques, B. S. Lima Neto, D. De Keukeleire, and D. W. Franco, “Qualitative and quantitative high-performance liquid chromatographic analysis of aldehydes in Brazilian sugar cane spirits and other distilled alcoholic beverages,” Journal of Chromatography A, vol. 782, no. 1, pp. 13–23, 1997. View at: Publisher Site | Google Scholar
  26. A. M. Muñoz, G. V. Civille, and B. T. Carr, Sensory Evaluation in Quality Control, Van Nostrand Reinhold, New York, NY, USA, 1992.
  27. N. Mashal and A. Kasirer, “Principal component analysis study of visual and verbal metaphoric comprehension in children with autism and learning disabilities,” Research in Developmental Disabilities, vol. 33, no. 1, pp. 274–282, 2012. View at: Publisher Site | Google Scholar

Copyright © 2012 Regina F. R. Pereira et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

More related articles

 PDF Download Citation Citation
 Download other formatsMore
 Order printed copiesOrder

Related articles

Article of the Year Award: Outstanding research contributions of 2020, as selected by our Chief Editors. Read the winning articles.