87.

Ribeiro J. S., Salva T., Ferreira, M. M. C., "Exploratory data analysis of commercial coffees with different roasted degrees and process using diffuse reflection mid-infrared spectroscopy". Águas de Lindóia, SP, Brazil, 10-15/09/2006: 10th International Conference on Chemometrics in Analytical Chemistry (CAC-2006, CAC-X), Book of Abstracts (2006) P028. Poster 028.


10th International Conference on Chemometrics in Analytical Chemistry P028

Exploratory data analysis of commercial coffees with different
roasted degrees and process using diffuse reflection mid-
infrared spectroscopy

Ribeiro, J. S.1,2*, Salva, T.2 and Ferreira, M. M. C1   jribeiro@iqm.unicamp.br

1. Universidade Estadual de Campinas - UNICAMP - Campinas - SP - Brazil;
2. Instituto Agronômico de Campinas - IAC - Campinas - SP - Brazil;

Keywords: Roosted Coffee, mid-infrared and chemometrics
_____________________________________________________________________________________
 

    Coffee is one of the most popular drinks around the world. Strategies used to authenticate certain  of  its
characteristics   have  been  traditionally  relied  on  wet  chemistry.   These  methods,  however,  are  time-
consuming  and expensive,   and  consequently,   demand  for  rapid and  inexpensive controls  is  growing.
Infrared spectroscopy  is  regarded  as a  quick and reliable mean  of  investigating food quality  and safety.
    The main advantages  of  mid-infrared  (MIR)  spectroscopy include  its  speed of analysis and  potential
selectivity,   when coupled to chemometric data analysis techniques.
    Despite the  amount  of  research conducted to classify varieties  of the same food product,   there  is no
published work  applying  mid-infrared  spectroscopy  to  discriminate different  roasted  degrees  of  coffee
samples.
    In the  present  work,   Brazilian  commercial  coffees  of  different  roasting  degrees  were  analyzed  by
diffuse reflectance  mid-infrared  spectroscopy.
    Samples.  Twenty one samples  of  Brazilian  commercial coffee were analyzed in triplicates  (7 of  them
medium roasting and decaffeinated;  7 of normal roasting;  and the remainder  of  extra roasting  –  all  from
different production batches).
    Data acquisition.    MIR  spectra  of  each   solid  sample  were  recorded   at   room   temperature  and
transformed  to  absorbance  units  with  a   background  spectrum  of  ground  KBr.   Each  spectrum   was
obtained by scanning the  400 – 4000 cm-1  region  256  times at a nominal resolution of  4  cm-1.   Spectral
data were smoothed  by  a moving average filter with  a  window size  n = 25,  followed  by  a  multiplicative
signal correction (MSC).   Principal Component  analysis  of  meancentered  pre-treated data  (Fig. 1)  was
carried out with Pirouette 3.11 chemometrics software.



    Variable selection was performed to  remove  those  regions  of  low discriminating and modeling power.
Therefore, the spectral regions  4000 – 3500,  2500 – 2300  and  100 - 400 cm-1  were excluded from data
set.
    Principal  component  analysis   was   applied  to  the  selected  variables.   PC1xPC2xPC3  scores  plot
describing 84.92 % of the total variance (Fig. 2) shows that three clusters of coffee could  be distinguished
as follows: the upper group of decaffeinated coffee  (green color)  is  clearly discriminated  from  the  other
two probably due to the roasting and caffeine extraction processes.  The other two groups at the bottom of
the plot correspond to the other roasting processes - traditional (red color) and extra roasting  (violet).
    This exploratory data analysis clearly shows that diffuse reflectance mid-infrared spectroscopy  coupled
to chemometrics has potential  advantages  to  differentiate among commercial coffee samples  of  various
roasting degrees.

Acknowledgment. Capes, FAPESP  and Agronomic Institute of Campinas
_____________________________________________________________________________________