354
M. Manley and P. J. Williams
15.10 Coffee
The initial work on coffee, aimed at process control, illustrated the ability to distinguish between regular and decaffeinated coffee [57]. While this is no longer the focus
of current work, at the time it showed the great potential of NIR spectroscopy. Today,
the majority of commercially produced coffee is either Arabica or Robusta, with the
former highly regarded for its improved sensory attributes. Since replacing the one
with the other, or mixing/blending Arabica with Robusta is considered adulteration,
detecting and quantifying this are important. Downey et al. [58] illustrated the capability of NIR spectroscopy to discriminate between pure and blends of Arabica and
Robusta coffees. The coffee samples were either green or roasted and whole or ground
beans, and a classification accuracy of 96.2% was achieved for the pure whole bean
coffees. This was attributed to the caffeine content; it is well known that Robusta
has a higher concentration than Arabica. When 50:50 blends were included in the
model, lower accuracies of between 82.9 and 87.6% were, respectively, obtained for
20 green and 20 roast samples. A handheld device was successfully used for Arabica
coffee grading, detecting the presence of peel/sticks, maize and Robusta coffee [59].
15.11 Wine and Distilled Alcoholic Beverages
There are numerous and diverse applications of NIR spectroscopy for wine analysis.
Cozzolino et al. [7] predicted a number of phenolic compounds, simultaneously, in
fermenting must and red wine. Most of the applications on wine focused on characteristics such as alcohol content, sensory and aromatic attributes and fermentation [4].
Cozzolino et al. [7] reviewed additional properties, such as the measurement of grape
composition. Good to excellent RPD values were reported for total soluble solids
(4.0), total anthocyanins (4.2), acidity and pH (2.8). In addition, measures such as
alcoholic degree (5.7), total acidity (2.27), pH (2.4), glycerol (4.0), reducing sugars
(10.3) and total sulphur dioxide (1.8) were reported for wine composition. Dambergs
et al. [7] and Cozzolino et al. [7] were able to predict wine sensory quality, demonstrating the versatility of NIR spectroscopy. A problem often encountered with wine
analysis and specifically when observing the fermentation process is the fact that the
sample changes with time. In another study, Manley et al. [7] used NIR spectroscopy
to measure sugar in grape must and to distinguish between samples based on their
free amino nitrogen (FAN) content. Furthermore, the authors distinguished between
Chardonnay wines and tables wines, based on their malolactic fermentation status
and ethyl carbamate content, respectively.
Wine authenticity received considerable attention in the past [60]. Manley et al.
[61] categorised four classes of rebate brandy; whereas, Pontes et al. [62] proposed
a strategy to detect adulteration in whiskeys, brandies, rums and vodkas.
Since its first application for grape compositional analysis, there has always been
a need to take the instrument to the sample, enabling analysis of grapes on the vine.
M. Manley and P. J. Williams
15.10 Coffee
The initial work on coffee, aimed at process control, illustrated the ability to distinguish between regular and decaffeinated coffee [57]. While this is no longer the focus
of current work, at the time it showed the great potential of NIR spectroscopy. Today,
the majority of commercially produced coffee is either Arabica or Robusta, with the
former highly regarded for its improved sensory attributes. Since replacing the one
with the other, or mixing/blending Arabica with Robusta is considered adulteration,
detecting and quantifying this are important. Downey et al. [58] illustrated the capability of NIR spectroscopy to discriminate between pure and blends of Arabica and
Robusta coffees. The coffee samples were either green or roasted and whole or ground
beans, and a classification accuracy of 96.2% was achieved for the pure whole bean
coffees. This was attributed to the caffeine content; it is well known that Robusta
has a higher concentration than Arabica. When 50:50 blends were included in the
model, lower accuracies of between 82.9 and 87.6% were, respectively, obtained for
20 green and 20 roast samples. A handheld device was successfully used for Arabica
coffee grading, detecting the presence of peel/sticks, maize and Robusta coffee [59].
15.11 Wine and Distilled Alcoholic Beverages
There are numerous and diverse applications of NIR spectroscopy for wine analysis.
Cozzolino et al. [7] predicted a number of phenolic compounds, simultaneously, in
fermenting must and red wine. Most of the applications on wine focused on characteristics such as alcohol content, sensory and aromatic attributes and fermentation [4].
Cozzolino et al. [7] reviewed additional properties, such as the measurement of grape
composition. Good to excellent RPD values were reported for total soluble solids
(4.0), total anthocyanins (4.2), acidity and pH (2.8). In addition, measures such as
alcoholic degree (5.7), total acidity (2.27), pH (2.4), glycerol (4.0), reducing sugars
(10.3) and total sulphur dioxide (1.8) were reported for wine composition. Dambergs
et al. [7] and Cozzolino et al. [7] were able to predict wine sensory quality, demonstrating the versatility of NIR spectroscopy. A problem often encountered with wine
analysis and specifically when observing the fermentation process is the fact that the
sample changes with time. In another study, Manley et al. [7] used NIR spectroscopy
to measure sugar in grape must and to distinguish between samples based on their
free amino nitrogen (FAN) content. Furthermore, the authors distinguished between
Chardonnay wines and tables wines, based on their malolactic fermentation status
and ethyl carbamate content, respectively.
Wine authenticity received considerable attention in the past [60]. Manley et al.
[61] categorised four classes of rebate brandy; whereas, Pontes et al. [62] proposed
a strategy to detect adulteration in whiskeys, brandies, rums and vodkas.
Since its first application for grape compositional analysis, there has always been
a need to take the instrument to the sample, enabling analysis of grapes on the vine.