Boscolo M., Ramos L. A., Lima-Neto B. S., Franco D. W., Ferreira M. M. C. “Spetrophotometric determination of caramel content in spirit aged in oak casks”. Araraquara, SP, Brazil, 01-03/12/1999: II Brazilian Meeting on Chemistry of Food and Beverages, Book of Abstracts, 4 (1999). Poster.

SPECTROPHOTOMETRIC   DETERMINATION   OF   CARAMEL  CONTENT   IN
SPIRITS AGED IN OAK CASKS

M. Boscolo¹,  L. A. Ramos¹,  B. S. Lima-Neto¹,  D. W. Franco¹*   and   M. M. C. Ferreira^2
1Instituto de Química de São Carlos,  Universidade de São Paulo,  CP 780,  13560-970 São
Carlos,  SP,  Brazil.    E-mail   address:    douglas@iqsc.sc.usp.br;   ²Instituto  de  Química,
Universidade Estadual de Campinas 13083-970 Campinas, SP Brazil

          Based  on  the  differnces  between  the  electronic  spectra  of  oak   hydro-alcoholic
extracts  and  caramel  hydro-alcoholic  solutions  a   methodology  has  been  developed  to
caramel analysis  in  spirits  aged in oak casks.  The data treatment  has  been carried out by
two different approaches. The simplest is based on plot of caramel concentration versus  the
ratio of absorbances at  210  and  282  nm.  The another procedure is based on partial  least
squares  (PLS)  calibration model  using  the  first  derivative of  the  spectral data.  A  good
correlation was observed between  the  caramel concentration for the same sample obtained
by the  two  methods  (y = ax + b;  a = 0.90;  b = 0.02;  r = 0.97).  The  estimated  caramel
detection limit was about 0.10 g/L.  The uncertainty following the  PLS  and the absorbance
ratio method are respectively 0.01 and 0.03 g/L  for a sample containing  0.45 g/L.
          These  methodologies  have been applied  for  the analysis of ninety five  aged  spirit
samples.  The sugar cane spirits  and  American  whiskeys exhibit  a  mean caramel content
smaller than Scottish whiskies and other brandies from many countries.
          The  treatment  of the  experimental  data  by  pattern recognition  using  hierarchical
cluster  analysis  (HCA)  and  principal  component analysis  (PCA)  allows  to  discriminate
the  samples  as  a  function of  their  caramel content.  The dendogram from  HCA  shows
clearly  two  different  groups  (A  and  B).  The  samples  with  spectral  similarities  to oak
hydro-alcoholic  solutions  added with caramel,  belong  to  the  group A;  the  others which
exhibited  similarity  to  oak  hydro-alcoholic solutions without  caramel addition,  belong  to
the group B.
          A  PCA  scores  plot  from  the  two  first  principal components  has  separated  the
samples  as  a  function of the  PC1.  The samples  with  spectral  similarity  to  oak  hydro-
alcoholic  solutions  without  caramel addition have negative  PC1  values  and  the samples
with spectral  similarity  to  oak  hydro-alcoholic solutions added  of  caramel  have positive
PC1 values.

Acknowledgements: CAPES, FAPESP and Indústria Müller de Bebidas Ltda.