Content uploaded by Adriana S Franca
Author content
All content in this area was uploaded by Adriana S Franca on Oct 26, 2015
Content may be subject to copyright.
In: Food Chemistry Research Developments ISBN: 978-1-60456-303-0
Editor: Ernst N. Koeffer, pp. 105-138 © 2008 Nova Science Publishers, Inc.
Chapter 4
CHEMISTRY OF DEFECTIVE COFFEE BEANS
Adriana S. Franca* and Leandro S. Oliveira
Programa de Pós-Graduação em Ciência de Alimentos/UFMG
Av. Antônio Carlos, 6627 – 31270-901 – Belo Horizonte, MG – Brasil
ABSTRACT
The term coffee is usually employed in reference to the consumable beverage
obtained by extracting roasted coffee with hot water, but it actually comprises a wide
range of intermediate products, starting from the freshly harvested fruit (coffee cherries),
then to green beans and to the final product of consumption (roasted coffee). Green
coffee beans are the main item of international trade, and their quality is evaluated
according to a wide variety of criteria, including bean size, color, and shape, processing
method, crop year and flavor (cup quality). Among these, flavor is the most important
criterion, and it is directly affected by the presence of the so-called defective coffee
beans. The presence of defective beans is usually a consequence of problems that occur
during harvesting and pre-processing operations. The most important defects are black,
sour or brown, immature, bored or insect-damaged, and broken beans. Both black and
sour defects are associated with bean fermentation and play a major role in downgrading
coffee flavor. Immature beans (from immature fruits) contribute to beverage astringency.
Such defective beans are usually present in the coffee produced in Brazil, due to the strip-
picking harvesting and processing practices adopted by the coffee producers. They are
separated (color sorting) from the non-defective beans prior to commercialization in
external markets, and the majority of these beans are dumped on the Brazilian internal
market. Thus, the roasting industry in Brazil has been using these defective beans in
blends with healthy ones, and, overall, a low-grade roasted coffee is consumed in the
country. Currently there are no analytical methodologies that allow for detection and
quantification of defective beans in roasted coffee, and thus an assessment of chemical
attributes that could provide differentiation between defective and healthy coffee beans is
of relevance. Thus, a review on physical and chemical attributes of defective coffee beans
in comparison to healthy ones is herein provided, for both green and roasted coffees.
Physical attributes include bean size, volume, density, color and water activity. Chemical
attributes include proximate composition, acidity, pH, sucrose levels, caffeine,
*E-mail:adrianafranca@pesquisador.cnpq.br.Tel:+55-31-34096910.Fax:+55-31-34096989
Adriana S. Franca and Leandro S. Oliveira
106
trigonelline, chlorogenic acids, amines and volatile substances. The evaluation of such
attributes indicates that, in the case of green coffee, it is possible to differentiate defective
and non-defective (healthy) beans by color, size, acidity levels, sucrose levels, and the
presence of histamine. In the case of roasted coffee, only an evaluation of the volatile
profile will effectively provide the means for differentiation.
1. INTRODUCTION
The quality of coffee used for beverages is strictly related to the chemical composition of
the roasted beans, which in turn is affected by the composition of the green beans and post-
harvesting processing conditions (handling, drying, storage, roasting and grinding). The
criteria commonly used to evaluate the quality of coffee beans include bean size, color, shape,
roast potential, processing method, crop year, flavor or cup quality and presence of defects
(Banks et al., 1999). Among those, the last two are the most important criteria and are
employed worldwide in coffee trading. Flavor is the main and most important criterion for
evaluating coffee quality. Brazilian coffees are officially categorized by reference to the
following flavor scale (Clarke and Macrae, 1987; Banks et al., 1999):
1) Strictly soft: low acidity, mellow sweetness, pleasant feel of the mouth easiness;
2) Soft: same characteristics as strictly soft, only less accentuated;
3) Softish: same characteristics as soft, only less accentuated;
4) Hard: lacks sweetness and softness;
5) Rioysh: iodine, inky flavor from microbe-tainted beans; and
Rio: same characteristics as Rioysh, only more accentuated
This classification, also known as ‘cup quality’, is assessed using the brewing method of
steeping (10 g of roast and ground coffee per 150 ml boiling water left for 5 min), with a
medium roast coffee prepared in a sample roaster and with a relatively fine grind. The term
defect is used in commercial practice in reference to the presence of defective (black, sour or
brown, immature, insect-damaged or bored) beans and also of extraneous matter (twigs,
husks, skins and stones) in a given coffee sample. The New York Coffee and Sugar Exchange
devised a type-classification system in which a rank is assigned to the coffee type based on
the number of defects counted in a sample (table 1). A system, known as the ‘‘black bean
count basis’’, was also devised in which all defects, intrinsic and foreign, are accounted for in
terms of equivalence to black beans (table 2). For example, five insect-damaged beans
correspond to one black bean, which is equivalent to one defect.
Table 1. Summary description of the NY type classification system (Franca et al., 2005a)
Type Number Maximum allowable number of defects per 300 g sample
NY 2 6
NY 3 13
NY 4 30
NY 5 60
NY 6 120
NY 7 240
NY 8 450
Chemistry of Defective Coffee Beans 107
Table 2. Equivalence ratings (number of defects) for type classification according
to the count of defective beans in a sample (Freitas, 2001)
Count Intrinsic defect Equivalence rating (number of defects)
1 Black bean 1
2 Brown or sour 1
2 to 5 Insect-damaged 1
3 Shells 1
5 Immature 1
5 Broken 1
5 Malformed 1
Count Foreign defect Equivalence rating (number of defects)
1 Large rock or stick 5
1 Medium rock or stick 2
1 Small rock or stick 1
1 Dried cherry 1
2 Floater 1
1 Large skin or husk 1
3 Medium skin or husk 1
5 Small skin or husk 1
The presence of defects is quite relevant in establishing coffee quality, since they could
be associated with specific problems during harvesting and post-harvest processing operations
(figure 1). Black beans result from dead beans within the coffee cherries (Clarke and Macrae,
1987) or from beans that fall naturally on the ground by action of rain or over-ripening
(Mazzafera, 1999). The presence of sour beans can be associated with ‘overfermentation’
during wet processing (Clarke and Macrae, 1987) and with improper drying or picking of
overripe cherries (Sivetz and Desrosier, 1979). Immature beans come from immature fruits.
Immature-black beans are those that fall on the ground while immature, remaining in contact
with the soil and thus subject to fermentation (Mazzafera, 1999). Typical defects, such as
black, sour and immature beans, are known to negatively affect beverage quality. According
to Clarke and Macrae (1987), the black bean is usually associated with a heavy flavor; sour
beans contribute to sour and oniony tastes, while immature beans will impart astringency to
the beverage.
Figure 1. Coffee beans: (a) black; (b) sour; (c) immature; (d) non-defective.
Adriana S. Franca and Leandro S. Oliveira
108
Brazil is the largest coffee producer in the world and, currently, about 15 to 20% of its
production consists of defective beans, which upon roasting decrease beverage quality
(Oliveira et al., 2006). Regardless of growth, harvest and processing conditions, some defects
are intrinsic in nature and will always be present and black, brown and immature beans are
the ones that affect beverage quality the most. Usually, these beans are mechanically
separated from the non defective beans prior to commercialization in international markets.
However, since to coffee producers they represent an investment in growing, harvesting, and
handling, these defective beans are commercialized both in the international trade market and
internal market in Brazil, where the roasting industries uses them in blends with non defective
beans.Thus, the overall quality of the roasted coffee consumed worldwide is depreciated.
After separation from the exportable portion, defective beans may be representing more than
50% of the coffee consumed in Brazil (Oliveira et al., 2008). In order to eliminate these
defective beans from the trade market, thus improving the quality of the beverage consumed
worldwide, there is a need for proposals of more attractive alternative uses for them. In view
of that, an assessment of chemical attributes that could allow for differentiation between
defective and non-defective beans prior and after processing is of relevance (Oliveira et al.,
2006). Only a few studies have addressed physical and chemical attributes of defective coffee
beans in comparison to non-defective ones and it is the aim of this work to present a
compilation of the literature reports so far.
2. PHYSICAL ATTRIBUTES
Physical attributes such as size, density and color of coffee beans are influenced by
botanical species and variety, growth and post-harvest processing conditions and, to some
extent, are employed for grading and separation between defective and non-defective coffee
beans. Furthermore, such parameters present significant variations during roasting and
specifically bean color and weight loss are employed for determination of the roasting degree
of the final product. Thus, discussion of each of these parameters will be first presented for
green (crude) coffee, in relation to discrimination between defective and non-defective beans
prior to roasting. Then, the effect of roasting will be discussed.
2.1. Size/Shape
The coffee bean shape can be described as oval with one flat face or, in terms of a
geometric form, as half a triaxial ellipsoid (Dutra et al., 2001), as shown in figure 2 (a). Some
beans, called peaberries, do not present the flat size because they come from coffee cherries
bearing only one seed and are usually smaller than the common beans (figure 2 (b)). The
coffee bean size is usually assessed in order to ensure uniformity and when a size is specified
for a given coffee sample or lot, it means that over 85% of the beans should be in the same
size range (Sivetz and Desrozier, 1979). For commercialization purposes, coffee beans size
and shape are commonly assessed by sieving. Screen apertures are either round (for regularly
shaped beans) or oblong (for peaberries), with diameters varying from 12/64 to 18/64 inches
Chemistry of Defective Coffee Beans 109
(round) or from 8/64 to 13/64 inches (oblong) at increments 1/64 inches (Vincent, 1987;
Freitas, 2001).
Coffee bean sizes can also be evaluated in terms of measurements of major, minor and
intermediate diameters of individual beans (see figure 3). A compilation of volume
measurements data for defective and non-defective coffee beans from the same origin is
displayed in table 3. Such results show that Arabica beans are larger than Robusta and that all
types of defective beans are smaller than non-defective ones, for both coffee species. Also,
the smaller beans are the ones that underwent fermentation (sour and black). These results
confirm that Arabica and Robusta coffees can be separated by size, as previously reported in
the literature (Martín et al., 1999; Casal et al., 2000) and also indicate that sieving could be
employed for separation of non-defective and defective coffee beans of a given coffee
species. Sieving is also efficient for separation of broken and cut beans (Ramalakshmi et al.,
2007).
Figure 2. Coffee beans photograph: (a) regular; (b) peaberries.
Figure 3. Coffee bean dimensions - 2a, 2b and 2c correspond to major, intermediate and minor diameter
measurements, respectively.
Adriana S. Franca and Leandro S. Oliveira
110
Table 3. Green (crude) coffee beans volumes and densities – Compilation of
measurement data from coffees from Minas Gerais State, Brazil
(Franca et al., 2005a; Mendonça, 2006)
Volume x 109 (m3) Density (kg m-3)
Arabica Robusta Arabica Robusta
Non-defective 111-119 72-76 1210-1310 1246-1250
Immature 94-105 50-54 1234-1297 1272-1278
Light Sour 90-105 69-71 1231-1267 1295-1299
Dark Sour 87-93 62-70 1230-1274 1261-1269
Black 62-79 40-45 1133-1236 1182-1192
Coffee bean size, and in consequence its volume, increases during roasting as a result of
the softening of the cellulose bean structure coupled to the increase in pressure from the
release of pyrolysis products. Franca et al. (2005a) presented volume measurements for
defective and non-defective beans prior to and after roasting. An evaluation of the percentage
increase in volume after roasting showed that defective beans presented lower degrees of
swelling (35-41%) in comparison to non-defective coffee beans (65%). Such results can be
associated to the fact that beans that swell less should roast more slowly (Clarke, 1987a), fact
later corroborated by weight loss data presented by Franca et al. (2005a). Such results are an
indication that defective beans will attain a lighter roasting degree than non-defective ones
under the same roasting conditions. Also, aside from their inherent physico-chemical
characteristics affecting their degree of roast, Cristo et al. (2006), in their study of flow
regimes in rotating coffee roasters, discussed that smaller beans will tend to migrate to the
low velocity core of the bed transverse section. Thus, due to poor mixing (low heat transfer
rates) in that region these beans will necessarily roast to a lesser degree, which would be the
case of smaller black and sour beans in a mixture with larger non-defective beans.
2.2. Density
A compilation of density measurements data for defective and non-defective coffee beans
from the same origin is also displayed in table 3. No significant variations in bean density
were observed, indicating that this parameter alone does not provide the means for separation
between defective and non-defective coffees, prior to roasting. Ramalakshmi et al. (2007)
evaluated samples of Arabica and Robusta coffees from India and reported that good quality
coffee beans were heavier than low quality coffees (mixture of defects) from the same origin.
However, such differences were attributed to the presence of broken beans and were not
associated with the presence of specific defects such as black, sour or immature beans. Coffee
bean density decreases considerably during roasting, due to the simultaneous increase in
volume and decrease in weight associated to the loss of water and volatile components (Dutra
et al., 2001; Rodrigues et al., 2003). The percent density reduction will depend on the degree
of roast, with average reported values of 45 and 60% for light to medium and medium to dark
roasts, respectively (Sivetz and Desrozier, 1979; Clarke, 1987a; Dutra et al., 2001). Franca et
al. (2005a) reported that black beans were the only defects that presented higher density
values in comparison to non-defective ones, after roasting under the same conditions. This
can be attributed to the fact that such defects present lower weight loss values and swell less
Chemistry of Defective Coffee Beans 111
than non-defective beans during roasting, i.e., roast to a lighter degree. As a consequence, the
beans will be heavier and smaller than non-defective ones, and therefore will present higher
density.
2.3. Weight Loss
Another important physical aspect that is commonly monitored during coffee roasting is
weight loss. This parameter can be employed to assess the degree of roast (Clarke, 1987a).
According to the literature on weight loss variation during roasting, the weight loss curve can
be represented by two lines of different slopes, as shown in figure 4 (a) (Sivetz and Desrosier,
1979; Dutra et al., 2001; Rodrigues et al., 2003; Oliveira et al., 2005). The smaller slope of
the first line representing weight loss is attributed to the slow release of water and green
coffee volatile components and it is known as the “drying phase” of the roasting process. The
increase in weight loss that occurs afterwards can be attributed to the intensive release of
organic compounds and CO2 associated to the pyrolysis reactions, and therefore this is the so
called “pyrolysis phase”. The onset of pyrolysis is then associated with the transition between
the two slopes, which has been reported to occur in the range of 7 to 12% weight loss,
depending on processing conditions (Sivetz and Desrosier, 1979; Rodrigues et al., 2003;
Oliveira et al., 2005). It is noteworthy to mention that this behavior seems to be related to
roasting temperature. A recent study (Mancha Agresti et al., 2008) reported a logarithmic
model fit (figure 4 (b)) for weight loss behavior when roasting coffee in a convective oven at
200oC. However, when temperature was raised to 300oC (unpublished data), the weight loss
behavior changed to the one presented in figure 4 (a). Such variation in weight loss behavior
with roasting temperature can be perceived as an indication that the roasting temperature is
affecting the production of volatile components by favoring different reactions and removal
pathways at different processing temperatures.
Figure 4. Weight loss behavior during coffee roasting at (a) high and (b) low temperatures. VL = very
light roast, L=light roast, M=medium roast, MD=medium/dark roast, D=dark roast.
Adriana S. Franca and Leandro S. Oliveira
112
Since the weight loss is a direct measure of the roasting degree, the evolution of this
parameter during roasting was investigated for both non-defective and defective coffee beans
(Franca et al., 2005a; Vasconcelos et al., 2007; Mancha Agresti et al., 2008). The data
presented in those studies showed that non-defective beans presented higher weight loss
values than defective ones, in the following order: non-defective > immature > sour > black.
Also, a comparison of the logarithmic (Mancha Agresti et al., 2008) or linear model
coefficients (Franca et al., 2005; Vasconcelos et al., 2007) indicates that non-defective coffee
beans roast slightly faster than defective coffee beans and that, among the defects, black
beans are the ones that roast slower. Such results confirm that defective beans roast to a lesser
degree than non-defective ones, under the same processing conditions.
2.4. Color
Bean color is an important attribute for classification of green coffee, due to its direct
impact on appearance. It will vary considerably depending on coffee species, origin and
processing conditions such as dry vs. wet processing, drying time, exposure to light, storage
conditions, etc. Variations on the external color of the coffee beans can be also be associated
to aging (see table 4) and non uniform coloration of a given coffee lot is usually taken as a
sign of non-uniform quality. For Arabica coffees, blue to gray green is the most desirable
color, based on correlation of visual inspection and cupping results (Sivetz and Desrozier,
1979; Freitas, 2001). Good quality Robusta coffees, on the other hand, present a brownish
color, whereas for Arabica, brown coloration is usually associated with aging, fermentation or
improper drying.
Table 4. Association between aging and external bean color
for Arabica coffees (Freitas, 2001)
Coffee bean age Color (based on visual inspection)
Recent green, blue to gray green (wet processed)
1-2 years greenish
3-4 years whitish
5-10 years yellowish
+ 10 years brownish
Color sorting is the major procedure employed for separation of defective and non-
defective coffee beans prior to roasting. In Brazil, manual sorting is usually employed for
type classification and electronic sorting is employed in farms and cooperatives of producers
for the actual removal of defective beans. In the electronic sorters, coffee beans pass, one by
one, by an electronic eye or camera system, and depending on wavelength measurements, the
bean is either allowed to pass or it is shot with a puff of air into a reject pile. The sorting
systems employ either monochromatic or bichromatic light measurements, and can be
adjusted to eliminate more or less defects. Monochromatic machines will be able to remove
black, grey and some beans attacked by insects and the more specific bichromatic color
sorting machines detect and eliminate coffee beans that are white, immature, broken, insect
damaged, brown (sour) and black (Sivetz and Desrosier, 1979; Clarke, 1987b). Even though
color sorting is the procedure employed in the farms, cooperatives and industries for
Chemistry of Defective Coffee Beans 113
separation of defective beans, there are only a few literature data on color measurements for
defective or low quality coffees (Franca et al., 2005a; Vasconcelos, 2005; Mendonça, 2006;
Ramalakshmi et al., 2007). Most reported data on measurements of color (L*a*b*) of coffee
beans prior to or after roasting are either restricted to good quality coffee or do not make
reference to coffee quality (Mendes et al., 2001; Cirilo et al., 2003; Marcone, 2004; Gokmen
and Senyuva, 2006; Summa et al., 2007).
Franca et al. (2005a) presented color measurements data for both whole and ground
defective and non-defective coffee beans, prior to and after roasting. Regarding whole green
beans, it was observed that luminosity values were higher for non-defective and immature
beans in comparison to sour and black defects, with the lowest luminosity values measured
for black beans. Such results raise the question to whether monochromatic separation will be
effective in separating immature and sour beans from non-defective ones. Actually, in order
make sure that such defects are effectively removed from a specific coffee lot, color sorting
machines are usually set up to allow some non-defective coffees to be also removed if their
color is similar to that of sour or immature beans. As a consequence of this, PVA mixtures of
Arabica coffees usually present a high percentage of non-defective coffees, as pointed out in
studies employing machine sorted PVA mixtures or low quality Arabica coffees from
different origins and crops (Franca et al., 2005ab; Farah et al., 2006; Vasconcelos et al.,
2007). Franca et al. (2005a) also reported that luminosity values were higher for ground beans
compared to whole ones (surface measurements), indicating that the bean surface is darker
than its interior, and that differences in luminosity between surface and interior were more
significant for defective beans in comparison to non-defective ones. Regarding chroma (color
saturation) and hue angle (color hue) measurements for whole beans, only black beans
presented significant differences in comparison to other coffee samples. Black beans
presented the highest hue value, corresponding to a dark green, whereas the other beans
presented a yellowish green hue. After grinding, no significant differences in chroma and hue
angle were detected.
Mendonça (2006) compared defective and non-defective Arabica and Robusta green
(crude) coffees from the same origin and crop in terms of L*a*b* color measurements. Sour
coffees were separated into two categories, light and dark, based on manual separation by
visual inspection. These authors reported no significant differences between Arabica and
Robusta coffees, in terms of luminosity measurements. However, black and dark sour beans
presented lower luminosity values than the other samples, for both Arabica and Robusta
species. It was also reported that, for a specific type of bean, healthy or defective, Robusta
coffees presented higher chroma values than Arabica. Also, black beans presented the lowest
chroma values among the defects, for both Arabica and Robusta species. These authors did
not find significant variations in color hue (hue angle) between Arabica/Robusta and
defective/non-defective coffees. This was attributed to aging of the beans, since external bean
color tends to become yellowish/brownish with time. This was observed even for the
immature beans. Based on color difference calculations, which takes into account all color
measurements in one parameter, it was concluded that, in the case of Robusta coffees, color
separation between defective (black, immature, dark sour and light sour) and non-defective
coffees can be safely accomplished. However, the results obtained for Arabica coffee indicate
that separation will not be effective in the case of immature and light sour beans. This study
also concluded that color parameters can be employed as a criterion for separation between
Arabica and Robusta coffees, only in the case of good quality samples, since no significant
Adriana S. Franca and Leandro S. Oliveira
114
color differences between the two species were observed for low quality (defective) beans.
An evaluation of color reflectance results presented by Ramalakshmi et al. (2007) for green
Arabica and Robusta coffees from India, of good (graded) and low (mixture of defects)
qualities corroborates such conclusions.
The most obvious physical change that takes place during roasting is the variation of the
bean color from green to yellow to brown to dark brown as roasting becomes more intense.
The effect of roasting on color parameters of defective and non-defective coffee beans is
discussed by Franca et al. (2005a). Luminosity values decreased as expected and black/sour
beans still presented lower luminosity values than non-defective and immature beans (surface
measurements). However, for measurements taken after grinding the coffee beans, luminosity
values for black and sour beans increased in comparison to surface measurements, and this
was not observed for non-defective and immature beans. These results reinforced the
conclusion that black/sour beans were roasted to a lighter degree than the others, since
uniformity of color between surface and interior is expected as roasting progresses. The bean
surface was still darker than its interior after roasting, thus indicating that these beans were
roasted to a lighter degree than the others. Non-defective and immature beans did not present
significant differences in luminosity values between whole and ground beans, indicating that
the beans attained a uniform color after roasting. Chroma values diminished after roasting,
except for black beans. Also, whole black beans still presented the highest values of hue angle
after roasting, maintaining a greenish hue, whereas the other beans became brown. After
grinding, no significant differences in hue angle were detected.
Similar measurements and conclusions were found in another study (Vasconcelos, 2005)
employing Arabica coffees from the same origin but a later crop (See table 5). In this case,
the effect of roasting on color was further investigated, and measurements were taken at
different degrees of roast. As expected, luminosity decreases as roasting progresses.
However, a comparison of luminosity values for defective and non-defective beans at a
specific degree of roast shows that, after roasting, defective beans are actually lighter than
non-defective ones, since they roast less than non-defective ones. An evaluation of the data
presented in table 5 indicates that, after roasting, even though there are differences in color
attributes between defective and non-defective beans from a statistical point of view, there is
no difference in attributes between the PVA mixture and non-defective beans, except for
luminosity of green coffee. These results indicate that color separation is only feasible for
green coffee (i.e., prior to roasting) and other parameters should be sought for discrimination
between defective and non-defective coffee after roasting.
Chemistry of Defective Coffee Beans 115
Table 5. CIE L*a*b parameters for crude (green) and roasted
coffees (ground coffees) (Vasconcelos, 2005)
Black Immature Sour PVA mixture Non-defective
Green
L* 41,59±1,59d,w 66,05±1,01a,w 60,92
±
0,76c,w 63,22
±
0,16
b
,w 67,52±1,63a,w
a* - 6,32±0,95
b
,z -2,82±1,88a,z -2,6
±
1,37a,z -1,77
±
1,09a,z -3,16±1,52a,z
b* 35,56±0,44a,x 30,66±1,33abc,x 32,33
±
2,39ab,x 28,81
±
2,25c,x 30,37±1,92
b
c,x
c* 36,18±0,49a,w 30,87±1,50
b
c,x 32,76
±
2,42
b
,x 28,91
±
2,30c,x 30,61±1,87
b
c,w
hab 99,86±1,49w 95,13±3,32w 94,32
±
2,25w 93,22
±
1,99w 95,97± 2,80w
Light Roast
L* 36,55±1,12
b
,x 35,65±0,22
b
,x 38,70
±
0,94a,x 32,21
±
1,24c,x 30,03±1,41d,x
a* 3,47±1,18
b
,wx 10,05±0,13c,w 8,12
±
1,57a,w 8,76
±
0,82a,w 7,91±1,39a,w
b* 37,69±1,81
b
,w 38,09± 0,44
b
,w 42,8
±
1,63a,w 33,43
±
1,91c,w 30,06±2,14d,w
c* 38,15 ±1,84
b
,w 39,94 ± 0,45
b
,w 44,53
±
1,39a,w 34,65
±
1,69c,w 31,10±1,90d,w
hab 84,20 ±2.96a,x 75,54±1,68
b
c,x 79,74
±
2,12
b
,x 74,03
±
1,85c,x 75,36±2.76
b
c,y
Medium Roast
L* 33,06±1,46a,
y
30,89±1,93a,
y
32,77
±
2,11a,y 26,67
±
0,68
b
,y 25,46±0,76
b
,y
a* 4,53±0,84c,w 8,72±0,65a,w 8,40
±
0,98a,w 6,72
±
0,61
b
,x 5,26±0,99c,x
b* 33,23±1,87a,x 30,55±2,65a,x 33,53
±
3,27a,x 24,35
±
1,18
b
,y 22,61±1,30
b
,x
c* 33,85±1,86a,x 31,87±2.53a,x 34,77
±
3,21a,x 24,89
±
1,36
b
,y 22,43 ±1,47
b
,x
hab 82,21±1,69a,
y
74,38± 2,73
b
,x 75,93
±
2,74
b
,x 75,42
±
1,22
b
,x 77,35±1,82
b
,x
y
Dark roast
L* 29,35±0,45a,z 26,70±0,09
b
c,z 27,30
±
1,24
b
,z 24,97
±
0,70cd,z 23,97±0,23d,y
a* 2,76±0,07c,x 6,22 ± 0,71a,x 5,80
±
0,46ab,x 4,85
±
0,92
b
,y 3,54 ± 0,63c,x
b* 28,20±0,96a,z 24,73±0,22
b
,
y
25,75
±
2,21
b
,y 21,89
±
1,19c,y 19,93±0,51c,x
c* 28,19±0,78a,
y
25,07± 0,34
b
,y 25,81
±
2,21
b
,y 21,51
±
1,34c,z 19,30 ±0,63c,
y
hab 84,19±0,60a,xy 75.49 ± 1,10c,x 77,49
±
1,05c,x 76,97
±
1,79c,x 79.87± 0,94
b
,x
Average value±Standard Deviation. Values followed by the same letter in the same line (a,b,c,d) or for
a specific parameter in the same column (w,x,y,z) do not differ significantly by the Duncan test at
5% probability. Values without letters in the same line do not differ significantly by variance
analysis at 5% probability.
3. CHEMICAL ATTRIBUTES
Green coffee is devoid of the pleasant aroma and flavor appreciated worldwide in
association with the coffee beverage. Such desired aroma and flavor are developed in the
roasting process, where the beans undergo a series of reactions leading to the corresponding
changes in chemical composition. These changes will affect sensory attributes and therefore
the way the consumer perceives coffee quality. They are affected by roasting conditions, and
are dependent on the chemical composition of green coffee beans, which is in turn influenced
by several factors such as species, variety, growth and post-harvest processing conditions, and
of course the presence of defective coffee beans. At this point it is clear that separation of
defective and non-defective coffee beans prior to roasting can be satisfactorily accomplished
by sieving and/or color sorting, in the case of black and sour coffees. Separation of immature
beans by means of physical attributes does not seem feasible. Furthermore, detection and
quantification of defective coffee beans in roasted coffee still poses a problem, and in this
case an evaluation of chemical attributes, which are directly related to the chemical changes
Adriana S. Franca and Leandro S. Oliveira
116
taking place during roasting, could provide the means for that. Therefore, a discussion of
chemical attributes of defective and non-defective coffee beans is herein presented, for both
green and roasted coffees.
Average data on the proximate composition of green and roasted coffee is displayed in
table 6. The observed variations are mostly related to variations in species/variety and to a
lesser extent to other factors including origin, agricultural practices, growth and storage
conditions and maturation degree (Clarke and Macrae, 1985). There are extensive changes in
chemical composition with roasting, as a result of pyrolysis reactions. However, the
proximate composition remains approximately the same, since changes occur within a
specific class of compounds. For example, protein results are based on the determination of
crude nitrogen and multiplication by the factor 6.25, so they include all nitrogenous
compounds. In view of the high temperatures attained during roasting, a decrease in total
protein levels is expected. However, the degradation into polypeptides and the formation of
volatile (pyrroles, pyridines, pyrazines, etc) and non-volatile (pigments, such as melanoidins,
and others) nitrogenous components during roasting (Homma, 2001) also contributes to
protein nitrogen determination by Kjeldahl and therefore the protein results for roasted coffee
remain practically unaltered on an as is basis.
Table 6. Proximate composition of green and roasted Arabica and
Robusta coffee beans (g/100 g dry basis green coffee)a
Arabica Robusta
Green Roasted Green Roasted
Water 8-13 1-3 12-13 1-2
Protein 11-17 12-15 11-13 13-15
Lipids 9-18 15-20 9-13 11-16
Minerals 4-5 4-5 4-5 5
Carbohydrate
b
60-76 40-79 69-76 64-71
a Compilation of data presented by Sivetz and Desrosier, 1979; Clarke and Macrae, 1985; Clarke and
Macrae, 1987; Speer and Kölling-Speer, 2001; Reh et al., 2006; Mendonça et al., 2007.
b Determined by difference.
Average data on the proximate composition of green and roasted, defective and non-
defective Arabica coffee from Brazil are displayed in table 7. Reported values for defective
and non-defective coffee beans are in the same range reported in the coffee literature in
general (See table 6). Also, there are no significant variations in proximate composition of
defective and non-defective coffee beans, both green and roasted. Only small variations have
been reported. For example, non-defective beans presented slightly lower moisture contents
than healthy ones, prior to roasting (Mazzafera, 1999; Oliveira et al., 2006; Vasconcelos et
al., 2007). Slightly higher average protein levels are reported for black beans in comparison to
the other samples (Oliveira et al., 2006; Vasconcelos et al., 2007). However, this was actually
due to caffeine and trigonelline contribution to protein determination based on crude nitrogen.
No differences were detected between defective and non-defective coffee beans after protein
levels were corrected by subtracting both caffeine and trigonelline contents. Healthy coffee
beans presented slightly higher oil contents than both black and sour beans whereas ash
contents were slightly lower (Mazzafera, 1999; Oliveira et al., 2006; Vasconcelos et al.,
2007). The proximate composition did not vary considerably after roasting, with the obvious
Chemistry of Defective Coffee Beans 117
exception of moisture content. In view of that, and of the fact that there were no significant
variations between defective and non-defective coffees in terms of the general classes of
compounds, a discussion on specific substances in each class is presented as follows.
Table 7. Proximate composition of green and roasted coffee
beans (g/100 g dry basis green coffee)a
Black Immature Sour Non-defective
Green
Water 9-13 9-10 8-10 9-12
Protein 16-17 15 15-16 14-15
Lipids 8-10 10 8-9 10-11
Ash 6 5-6 5-6 5
Carbohydrate
b
59-70 62-70 61-71 60-72
Roasted
Water 1 1-2 1 1
Protein 14-16 13-15 14-15 13-14
Lipids 10-12 9-11 10-12 9-11
Ash 6 5 5 4-5
Carbohydrate
b
66-69 68-71 67-71 68-73
a Compilation of data presented by Mazzafera, 1999; Oliveira et al., 2006; Vasconcelos et al., 2007.
b Determined by difference.
3.1. Water Content
Water content is a critical parameter for evaluation of green coffee quality, in the sense
that it affects mold growth, mycotoxin production, fermentation, physical, chemical and
sensory parameters. As water is quite cheap compared to coffee, its amount within the beans
is also of interest from a commercial point of view (Mendonça et al., 2007). Its ammount is
usually controlled by processing conditions, and, for green coffee, it is kept at about 12% to
allow for safe transportation and storage (Clarke, 1985; Reh et al., 2006). Immediately after
roasting, coffee moisture content can be as low as zero moisture, especially for dark roasts
(Clarke, 1985) when water quenching is not employed. However, measured values can
increase up to 3% as the coffee beans tend to absorb water from the surrounding air.
Reported data on moisture content of defective vs. non-defective coffee beans indicated
that non-defective green coffees presented slightly higher moisture contents than defective
ones (Mazzafera, 1999; Oliveira et al., 2006; Vasconcelos et al., 2007). The same behavior
was reported in terms of water activity measurements in green coffee, with average values of
0.44 and 0.48 for defective and non-defective coffees, respectively (Vasconcelos et al., 2007).
Such differences could be attributed to defective beans presenting smaller water diffusion
coefficients, but further studies are needed in order to validate such hypothesis. Both moisture
content and water activity decreased with roasting, without significant differences between
defective and healthy coffee beans.
Adriana S. Franca and Leandro S. Oliveira
118
3.2. Nitrogen-Based Compounds
Such class of substances will be discussed in terms of protein contents and also with
respect to other nitrogenous substances of relevance in terms of coffee and/or defective coffee
beans.
3.2.1. Protein
The reported values of green coffee protein contents are usually based on the
determination of crude nitrogen and multiplication by the factor 6.25 (Macrae, 1985) and are
in the range of 11 to 17% dry basis, without significant differences between Arabica and
Robusta. True protein values are in the range of 8 – 9%, with the difference being mainly
attributed to caffeine and trigonelline nitrogen.
Protein data on defective coffees are based on crude nitrogen (Oliveira et al., 2006;
Vasconcelos et al., 2007) or NaOH extraction (Mazzafera, 1999; Ramalakshmi et al., 2007).
Mazzafera (1999) reported that the total protein content was higher for black beans in
comparison to immature and immature-black. However, this could not be correlated to amino
acid content, which was higher for immature beans in comparison to black and immature-
black coffees, with the main amino acids in immature beans being asparagine, followed by
tyrosine, alanine, lysine, serine and glutamic acid. Although immature-black and black
coffees presented similar contents of total amino acids, qualitative changes were observed.
Lysine, tyrosine and alanine were the main amino acids for immature-black coffee, as
opposed to glutamic acid, asparagine, alanine, aspartic acid and lysine for black beans.
Oliveira et al. (2006) and Vasconcelos et al. (2007) presented data on Arabica coffees from
the same origin but subsequent crops and both reported higher protein contents for black
beans in comparison to other defects and to non-defective coffee. However, such higher
contents were actually due to caffeine-nitrogen. Ramalakshmi et al. (2007) reported slightly
lower protein contents for low quality coffees.
3.2.2. Caffeine
This compound has been historically accredited with most of the physiological effects of
coffee and therefore it is the single most frequently determined compound in coffee (Macrae,
1985). The caffeine content of green coffees varies considerably, especially with respect to
species and also within the same species. Robusta coffees have approximately twice the
amount of caffeine presented in Arabica coffees, with reported average values being 1.2% db
for Arabica (0.6–1.9% range) and 2.2% db for Robusta.
Comparison of caffeine levels between defective coffees have been reported by
Mazzafera (1999), for black, immature-black and immature coffees, and by Franca et al.
(2005a), for black, sour and immature defects. No significant differences in caffeine levels
were found among the defective beans, with an average of 1.3% caffeine in defective coffees
reported by both studies. However, Franca et al. (2005a) found that non-defective beans
presented lower caffeine contents (0.9%) than the defective ones. A few recent studies have
also compared caffeine levels for samples of different sensory qualities (Franca et al., 2005b;
Farah et al., 2006). According to Franca et al. (2005b), the highest (1.1%) and lowest (0.7%)
caffeine levels corresponded to the highest (soft) and lowest (rio) quality samples,
respectively. Farah et al. (2006) also reported that the highest caffeine level (1.2%) was
determined for the highest quality sample (soft), but found the lowest caffeine level (1.0%)
Chemistry of Defective Coffee Beans 119
for a good quality sample (hard), that consisted of 97% non-defective beans. Ramalakshmi et
al. (2007) also reported slightly lower caffeine levels for low quality coffees. These results
seem to be in contradiction with the fact that higher caffeine levels have been associated to
defective or low quality coffees (Mazzafera, 1999; Franca et al., 2005a). However, the studies
by Franca et al. (2005b), Farah et al. (2006) and Ramalakshmi et al. (2007) did not employ
coffees from the same origin/crop, thus the highest caffeine levels in good quality coffees
may be associated to variations in growth and processing conditions.
Roasting was reported to cause approximate reductions of 30% (Franca et al., 2005a,b) in
caffeine content. Since the solubility of this compound in water increases with temperature,
the caffeine loss was attributed to a drag by water vapor released during roasting, and this
compound has been detected in the exhaust gas from roasting (Dutra et al., 2001). Roasting
did not significantly affect caffeine levels of either black or sour beans (Franca et al., 2005a).
The fact that the caffeine content remained relatively constant for both black and sour beans
upon roasting was interpreted as an indication that black and sour beans were roasted to a
lesser extent than the other types of coffee beans (immature and non-defective).
3.2.3. Trigonelline
Together with caffeine, this compound has received considerable attention in coffee
chemistry research, given that its thermal degradation products are important in terms of
sensory and nutritional aspects (Macrae, 1985). Nicotinic acid is the major non-volatile
component resulting from demethylation of trigonelline during roasting. The amount of
nicotinic acid present in roasted coffee (10-40 mg/100g) coupled to the fact that it is readily
available for biological conversion indicates that roasted coffee should be considered a
significant source of this vitamin (Macrae, 1985). The sensory importance of trigonelline is
associated to the fact that approximately half of the volatile substances generated during its
thermal degradation consist of pyridines and pyrroles. Both classes of compounds have been
detected in roasted coffee. Pyrroles have been associated to medicine and cereal flavours
(Mancha Agresti et al., 2008) whereas pyridines are associated with adstringent/hazelnut (2-
methylpyridine) and buttery/caramel (3-ethylpyridine) aromas (Dart and Nursten, 1985).
Evaluations of trigonelline levels correlation with coffee quality are available from
studies presented by Franca et al. (2005a) (defective vs non-defective coffee), Franca et al.
(2005b) and Farah et al. (2006) (low vs high quality coffees). Franca et al. (2005a) reported
trigonelline levels of approximately 1% in non-defective, immature and sour coffee beans and
lower values (~0.8%) for black beans. Franca et al. (2005b) did not find significant
differences in trigonelline levels between high and low quality coffees, whereas Farah et al.
(2006) reported a decrease in trigonelline levels as coffee quality worsened, and also verified
a strong negative correlation of trigonelline levels with poor quality and with the Rio-off-
flavor. Trigonelline levels were significantly reduced by roasting, with such reduction
attaining higher values for darker roasts. After roasting, trigonelline levels did not show
significant correlations with cup quality or the presence of defective coffee beans. Slightly
higher trigonelline levels were observed for black/sour beans and lower quality samples
(rio/rioysh/rio zona), in association with such beans attaining lighter roasting degrees in
comparison to high quality ones.
Adriana S. Franca and Leandro S. Oliveira
120
3.2.4. Amines
Amines are organic bases encountered in both raw and processed food products. They
can be formed either by natural metabolic pathways or by action of decarboxylase-positive
microorganisms (Fernandes and Ferreira, 2000; Oliveira et al., 2005). Amine levels and
distribution will vary considerably among different foodstuffs and also for a specific food
product (Kalač and Krausová, 2005). In non-fermented foods, the presence of biogenic
amines above a certain level has been associated with undesired microbial activity.
Furthermore, several amines are known to be vasoactive and psychoactive, and ingestion of
high quantities has been associated to food poisoning cases. Thus, amine levels could be
indicative of microbial spoilage and have been employed as a criterion for quality evaluation
and spoilage detection in a wide variety of food products (Bardócz, 1995; Santos, 1996). In
view of this, it could also be considered a quality parameter for coffee and thus the studies
that present an evaluation of amines in green and roasted coffees are discussed below.
Only a few studies have addressed the presence of amines in green and roasted coffee
(Amorim et al., 1977; Casal et al., 2004; Cirilo et al., 2003; Oliveira et al., 2005), and only the
study by Vasconcelos et al. (2007) dealt with the presence of amines in defective coffee
beans. A brief review on the major findings from these studies is presented as follows.
Amorim et al. (1977) investigated the presence of polyamines in coffee, before and after
roasting at 240 oC for 10 (light roast) and 12 (dark roast) min. Their experiments were
conducted with Arabica coffees of the same variety (Coffea Arabica L. var. Mundo Novo)
harvested in the same year, but classified as of different cup qualities. They compared high
quality samples (mild flavor, cup classification as soft) to low quality ones (strong medicinal
flavor, cup classification as rio). Prior to roasting, putrescine was the prevailing amine in
green coffee (64%), followed by spermidine (23%) and spermine (12%), regardless of sample
quality. These authors reported that only putrescine was present after a light roast and no
amines were detected in the samples submitted to a dark roast. Given that no significant
differences were found in amine levels between the high and low quality coffee samples, it
was concluded that polyamine content did not present correlation with the Rio off-flavor.
Cirilo et al. (2003) investigated the presence of amines in coffees submitted to two
degrees of roasting, herein designated as light (American roast, 6 min roasting at 300oC) and
dark (French roast, 12 min roasting at 300oC). These authors encountered serotonin,
putrescine, spermine and spermidine in green coffee and the reported amine levels were quite
different from the ones presented by Amorim et al. (1977). Cirilo et al. (2003) found both
serotonin (35%) and putrescine (32%) as the prevailing amines, followed by spermidine
(19%) and spermine (14%). Roasting caused a significant decrease in amine content.
However, total amine contents after the dark roast were twice the ones observed after the light
roast. According to this study, both putrescine and spermine were not present after roasting,
whereas spermidine and serotonine were still detected in low concentrations. Agmatine was
only detected for the sample that was submitted to dark roasting and the authors suggested
that its formation could be associated to the decarboxylation of arginine or to the release of
this amine from conjugated forms.
Casal et al. (2004) evaluated the levels of biogenic amines (putrescine, cadaverine,
serotonin, tyramine, spermidine, and spermine) in Robusta and Arabica coffees. These
authors reported putrescine as the most abundant amine in both species, followed by
spermidine, spermine and serotonin (see figure 5). They also detected small amounts of
cadaverine and tyramine. Their results indicated that putrescine levels could be used as a
Chemistry of Defective Coffee Beans 121
criterion for the discrimination between Arabica and Robusta coffees. These authors also
mentioned that amines could be used for discrimination between green coffees subjected to
different postharvest processes (wet vs. dry). They reported variations in biogenic amine
levels after roast, but the statistical significance for species discrimination was reduced.
Figure 5. Percent mass distribution of biogenic amines in Arabica and Robusta green coffees based on
the data from Casal et al. (2004). PUT = putrescine; CAD = cadaverine; SER = serotonin; TYR =
tyramine; SPD = spermidine; SPM = spermine.
Oliveira et al. (2005) evaluated the effect of roasting on the levels of amines in high and
low quality coffees, classified by cup as soft and rio, respectively. Results with respect to
amines distribution in green coffee were similar to those presented by Amorim et al. (1977)
and Casal et al. (2004), given that putrescine was the prevailing amine (75%), followed by
spermine (12%) and spermidine (12%). These authors also reported that putrescine levels
were significantly higher for the low quality sample and that both histamine and tryptamine
were only present in the low quality sample. Their results indicated that both amine levels and
profiles could be related to coffee quality. In this study, amine levels were also monitored
during roasting and a significant decrease in total amine content occurred with an increase in
roasting time. Based on weight loss monitoring during roasting and comparison to other
literature reports (Amorim et al., 1977; Cirilo et al., 2003), these authors concluded that
amine degradation takes place mainly during the drying stage of the roasting process, which
precedes the pyrolysis stage. Thus, even after a mild roast, total amine levels are expected to
be quite small.
Vasconcelos et al. (2007) investigated the presence of amines in defective and non-
defective coffee beans, green and submitted to three different levels of roasting at 200oC:
light (30 min roasting), medium (60 min roasting), and dark (120 min roasting). Seven of the
ten evaluated amines were detected in green coffee: putrescine, cadaverine, histamine,
serotonin, spermidine, spermine and tryptamine. Among those, only putrescine, the prevailing
amine in all samples, was detected in all samples above trace levels. Prevalence of putrescine
followed by spermine and spermidine was observed for all coffee samples with the exception
of black beans, for which histamine levels corresponded to approximately 50% of putrescine
levels, and both spermine and spermidine were present in smaller quantities. Histamine was
detected, above trace levels, only in the samples that consisted purely of defective coffee
beans. This corroborated the study by Oliveira et al. (2005), who reported that histamine was
detected only in low sensory quality coffee. The results obtained in both studies are an
indication that the detection of histamine in green coffee could be associated with the
Adriana S. Franca and Leandro S. Oliveira
122
presence of defective beans. After roasting to a light degree, only traces of serotonin were
detected and no amines were detected after roasting to medium and dark degrees, also in
accordance with the data by Oliveira et al. (2005) regarding amine degradation before
pyrolysis reactions.
Based on the reviewed literature on amines in green coffee, it can be concluded that
putrescine is the prevailing amine, regardless of variety, quality or growth and processing
conditions. However, its level does not seem to be directly related to quality. Oliveira et al.
(2005) reported higher putrescine levels for the lowest quality coffee whereas Vasconcelos et
al. (2007) reported the opposite when comparing low and high quality coffees from the same
origin. Since the samples employed by Oliveira (2005) did not come from the same crop, one
could argue that the differences could also be attributed to other factors associated to growth
and processing conditions. A compilation of results regarding total amine levels in green
coffees, taking into consideration only the studies that specified sample quality (Amorim et
al., 1977; Oliveira et al., 2005; Vasconcelos et al., 2007) is displayed in figure 6. From these
results it is clear that only black coffee beans presented a significantly lower amount of
amines in comparison to other coffees. Since such defective beans are present in much lower
amounts than the other types of defects, regardless of coffee quality, the variations in total
amine contents between coffee samples will probably be due to variations in crop and
processing conditions rather than be related to sample quality.
Figure 6. Total amine levels in Arabica green coffees of different qualities, based on the data published
by Amorim et al. (1974), Oliveira et al. (2005) and Vasconcelos et al. (2007).
The results obtained by Oliveira et al. (2005) and Vasconcelos et al. (2007) indicate that
histamine is a potential marker for detection of defective beans in green coffee, since this
amine was detected either in defective beans or low quality coffee. However, histamine levels
in coffee are quite low compared to other amines, ranging from 0.1 to 1.0 mg/100g. Also,
Chemistry of Defective Coffee Beans 123
histamine levels found in coffee do not pose a concern in terms of intoxication. The
toxicological levels of amines are not easily established, because they depend on individual
characteristics of each amine, as well as on interactions among amines and also with the food
substract. Nevertheless, the highest histamine value reported in the literature for green coffee
(Vasconcelos et al., 2007) corresponds to approximately 10% of the suggested limit for
histamine toxicity (Halász et al., 1994; Santos, 1996). Furthermore, histamine was not
detected after roasting of the low quality samples (Oliveira et al., 2005; Vasconcelos et al.,
2007).
3.3. Lipids
The lipids (fat) of coffee beans are comprised of coffee oil, mainly present in the
endosperm, and of coffee wax, located in the outer layers of the beans in small amounts
(Folstar, 1985). The chemical composition of both green and roasted coffee oils is similar to
those of edible oils (Sivetz and Desrosier, 1979) with the non-glyceride fraction, which can
amount up to 25% (w/w) of the oil, being comprised mainly of free and esterified diterpene
alcohols and sterols (Speer and Kölling-Speer, 2001). Considering the bulk of the literature
data on coffee oil, the total oil content in green coffee beans, determined by a diversity of
methodologies, falls in the range of 9 to 15% (w/w) db for arabicas and in the range of 7 to
12% (w/w) db for robustas (Folstar, 1985; Nikolova-Damyanova et al., 1998; Speer and
Kölling-Speer, 2001; Turatti, 2001; Vila et al., 2005; Oliveira et al., 2006). The values for oil
content in roasted coffees are higher than those for green beans due to the beans dry matter
loss during roasting, which in turn varies with the degree of roast (Speer and Kölling-Speer,
2001). Comprehensive reviews on coffee lipids content and composition were presented by
Folstar (1985) and Speer and Kölling-Speer (2001). However, specific information on the
content and composition of the lipids in defective coffee beans is scarce and is discussed next.
Mazzafera (1999) reported a study on the chemical composition of defective coffee beans
in which the total lipid content was determined by Soxhlet extraction with hexane. In this
study, black beans presented the highest average value for oil content (~13%) and immature
beans presented no significant differences in oil content when compared to immature-black
beans. Oliveira et al. (2006) determined the lipid contents of Arabica coffees in their study of
the proximate composition of green and roasted defective coffee beans and the respective
values (dry green basis) are 9.2±0.6 g/100g for sour, 9.6±0.7 g/100g for immature, 10.0±0.3
for black and 10.8±0.3 g/100g for non-defective beans. The method employed was that of
Soxhlet extraction with petroleum ether. The lipid contents were found to be significantly
different among non-defective and defective beans according to the Duncan test at 5%
probability. Non-defective mature coffee beans presented higher oil contents than those for
defective beans. Amongst the defective beans, black beans presented higher oil contents than
those for immature and sour beans, corroborating the results presented by Mazzafera (1999).
There were no significant differences among the oil contents for roasted non-defective and
sour beans. Regarding the effect of the roasting process on the oil content, there was a slight
decrease (in dry green basis) in the oil content of non-defective mature beans while the
content of the other types remained constant. This fact corroborates the observation that
defective beans are roasted to a lesser degree than non-defective mature beans for the same
Adriana S. Franca and Leandro S. Oliveira
124
processing conditions (Franca et al., 2005a). The lipid contents determined by Oliveira et al.
(2006), be it for non-defective or defective beans, fall within the range encountered in the
literature (9–15%) for unspecified types of arabica coffee (regarding the amount of defects)
(Folstar, 1985; Speer and Kölling-Speer, 2001; Turatti, 2001; Vila et al., 2005). The solvent
extracted coffee oils obtained for each separate sample of black, immature, sour and non-
defective beans, and the screw-pressed coffee oil obtained from a mixture (PVA) containing
31.2% non-defective, 40.5% sour, 21.1% immature and 3.2% black beans were analyzed for
the fatty acids profile. There were no significant differences in the amounts of all fatty acids
for green and roasted non-defective and defective beans, based on the Duncan test at 5%
probability. Linoleic and palmitic acids were the predominant fatty acids, with averages of
44% and 34%, respectively. The oil samples were comprised of moderate quantities of oleic
(9%) and estearic (7%) acids and low quantities of araquidic (3%), linolenic (1.5%), behenic
(0.7%) and eicosenoic (0.3%) acids. Miristic and palmitoleic acids were present in trace
amounts.
The factors that are usually employed to define the quality of edible oils were determined
by Oliveira et al. (2006) for coffee oils obtained by screw pressing of the mixture of defective
beans (PVA). These factors are represented by the following parameters: saponification
value, unsaponifiable matter, free fatty acids, free acidity, iodine value, and refractive index.
The unsaponifiable matter was determined for both screw-pressed and solvent-extracted PVA
oil in order to verify the effects of the extraction method on this quantity. The unsaponifiable
matter content determined for the solvent-extracted oil was 9.2 g/100 g oil, a value
significantly smaller than that for the screw-pressed oil (12.8 g/100 g oil). This was somewhat
expected, since the screw-pressing is not a selective process, thus extracting with the oil other
matter than exclusively the lipidic fraction. The value obtained by Oliveira et al. (2006) for
the pressed oil is within the range published in the literature of 9–13 g/100 g oil for green
(crude) coffee oil (Ravindranath et al., 1972; Turatti, 2001).
Oliveira et al. (2006) also obtained high values for free fatty acids contents of screw-
pressed coffee oils: 4.9±4 g oleic acid/100 g oil. High values for free fatty acids content can
be attributed either to inadequate storage conditions or long periods of storage of the beans
(Vila et al., 2005) or to the fact that the oil was extracted by more than one pressing cycle
(Rossell and Pritchard, 1991). Spiz et al. (1989) also found high values for free fatty acids in
coffee oil (1.9–6.7 g oleic acid/100 g oil). The calculated value for free acidity (9.7±0.8%
w/w expressed as oleic acid) was also high since it is based on the value for free fatty acids.
The saponification value determined for the green coffee oil presented a value of 192.0±1.4
mg KOH/g oil that lies within the range encountered in the literature which is 180–200 mg
KOH/g oil for coffee oil (Lago, 2001) and also within standards established for the trading of
other edible oils, such as virgin olive oil (184–196 mg KOH/g oil). The iodine value (Wijs)
was determined to be 95.5±5.6 and also falls within the range published in the literature for
coffee oil (84.6–98.5; Lago, 2001). The refractive index determined for the coffee oil in this
study was 1.468±0.001 and falls outside the range determined by Lago (2001), which was
1.458–1.462. This can be attributed to the facts that the crude oil used in the study by Oliveira
et al. (2006) presented a high value of unsaponifiable matter and also contained fine
particulate matter in it.
In their study of the evaluation of the potential of coffee oil as a feedstock for biodiesel
production, Oliveira et al. (2008) used oils extracted from a batch of exclusively non-
Chemistry of Defective Coffee Beans 125
defective beans and also from a mixture of defective (34% w/w) and non-defective beans
(66% w/w). The extraction was in a Soxhlet apparatus (20 kg solids capacity) with hexane for
16 h. The compositions of the oils from the non-defective and the PVA mixture were
determined by a thermogravimetric procedure, adapted from that proposed by Goodrum and
Geller (2002), to be comprised of 81 and 76% (w/w) of triglycerides, respectively. The
titrated acidities of the oils of non-defective and defective beans were 2.62±0.29 and
10.04±0.03% (w/w), respectively. The contribution of the defective beans to the acidity of the
oil was fairly high, since these beans were partially comprised of fermented beans (black and
brown, ~20% w/w).
3.4. Ash (Minerals)
Mineral content represents 4-5% of coffee weight on a dry basis (Clarke, 1985).
Potassium is the main constituent, representing approximately 40% of total mineral content,
followed by calcium and magnesium in much smaller amounts (Clarke, 1985). Other minerals
are present in small quantities (iron, cesium, manganese, sodium, vanadium and aluminum;
Santos and Oliveira, 2001; Vega-Carrillo et al., 2002). Regarding mineral content in roasted
coffee, literature data are scarce and show a small increase associated to weight loss during
roasting (Ferreira et al., 1971; Clarke, 1985). There are evidences that both the mineral
content and profile in coffee could be affected by growth conditions (e.g. fertilizers, maturity
state), pre-processing operations (dry vs. wet processing) and fermentation (Clarke, 1985).
Since the mineral content of coffee is not affected by roasting and it is known to vary with
bean maturity state, elemental analysis could be a potential candidate for differentiation
between defective and non-defective beans in roasted coffee.
Oliveira et al. (2006) and Vasconcelos et al. (2007) evaluated the total mineral content of
defective coffee beans in comparison to non-defective ones. Both studies reported higher
mineral contents for defective beans (~6% d.b.) in comparison to non-defective ones (~5%
d.b.), prior to and after roasting. Among the defects, black beans presented the highest
mineral content. The only reported data on the mineral profile of defective and non-defective
coffees, based on atomic absorption spectrometry, was presented by Custódio et al. (2005)
and are displayed in tables 8 and 9.
Table 8. Mineral profile in green coffee beans (g/100 g dry basis)
Calcium Magnesium Potassium Zinc
Non-defective 0.448±0.093a 0.257±0.005a 1.335±0.047c 0.0027±0.0004a
Immature 0.307±0.016
b
0.261±0.035a 1.545±0.065
b
0.0001±0.0001a
Black 0.335±0.022
b
0.274±0.010a 1.838±0.074a 0.0028±0.0026a
Sour 0.301±0.025
b
0.258±0.017a 1.635±0.110
b
0.0028±0.0040a
Average value±Standard deviation. Mean values with the same letter in the same column do not differ
significantly by the Duncan test at 5% probability
Adriana S. Franca and Leandro S. Oliveira
126
Table 9. Mineral profile in roasted coffee beans (g/100 g dry green basis)
Calcium Magnesium Potassium Zinc
Non-defective 0.227±0.024
b
0.246±0.005
b
c 1.240±0.025
b
ND
Immature 0.250±0.012ab 0.254±0.010ab 1.360±0.042
b
ND
Black 0.267±0.008a 0.268±0.006a 1.645±0.094a ND
Sour 0.263±0.016a 0.232±0.020c 1.563±0.084a ND
Average value±Standard deviation. Mean values with the same letter in the same column do not differ
significantly by the Duncan test at 5% probability. ND=non detected.
Potassium was found to be the main constituent of the ashed mineral content, in
agreement with literature data on coffee mineral content (Clarke, 1985). Significant
differences were detected among some samples. For green beans, potassium levels were the
highest for black beans and the lowest for non-defective ones. Intermediate values were
obtained for both the immature and sour beans. A small decrease in potassium levels was
observed after roasting. The highest levels were found in both black and sour samples,
whereas the lowest were observed for non-defective and immature beans. Calcium levels,
about 80% lower than potassium, were higher for the non-defective coffee beans in
comparison to the defective ones, prior to roasting. Calcium levels decreased approximately
30% after roasting, with the highest levels presented by black and sour beans and the lowest
by non-defective ones. Such behavior could be attributed to the fact that both black and sour
coffee beans are known to roast to a lesser degree than non-defective ones (Franca et al.,
2005a). No significant differences were found in magnesium and zinc levels, prior to and
after roasting. Also, a comparison of mineral profiles among both coffees from different
crops, green and roasted did not show significant variations (see table 10). The obtained data
showed that high potassium levels could be perceived as a marker for black and sour coffees,
even after roasting. However, given the expressive amounts of non-defective beans usually
found in PVA mixtures, such attribute can not be employed for separation of defective/non-
defective coffees after roasting. These results suggest that a more sensitive analysis should be
performed for minerals that are present in low quantities, in order to find specific markers for
defective coffees.
Table 10. Mineral profile of the PVA mixture (g/100 g dry green basis)
Calcium Magnesium Potassium
Crop 1 Green 0.212±0.015 0.238±0.004 1.605±0.049
Roasted 0.233±0.012 0.241±0.003 1.725±0.034
Crop 2 Green 0.210±0.014 0.217±0.002 1.541±0.051
Roasted 0.213±0.007 0.230±0.015 1.719±0.136
Average value±Standard deviation. Values did not differ significantly by Analysis of Variance at 5%
probability.
3.5. Carbohydrates
Green coffee contains a wide range of carbohydrates, including poly-, tri-, di- and
monosaccharides. Such class of substances is quite important for flavor development and
several studies confirm that carbohydrate degradation is directly linked to the formation of
Chemistry of Defective Coffee Beans 127
several substances that affect coffee flavor and aroma (De Maria et al., 1994 ; De Maria et al.,
1996; Alcázar et al., 2005; Murkovic & Derler, 2006). According to Oosterveld et al. (2003),
approximately 20 to 40% of the total carbohydrates in coffee are degraded during roasting.
The roasting process is responsible for opening the cell-wall matrix resulting in the
solubilization of polysaccharides upon extraction. Then, the hydrolysis of the polysaccharides
results in a release of oligosaccharides and monosaccharides, which in turn are converted into
their degradation products through Maillard and Strecker reactions (Oosterveld et al., 2003).
Such products, including furans, aldehydes, aliphatic and carboxylic acids, are aroma
precursors and also play a major role on sensory perception in association with acidity (Ginz
et al., 2000; Farah et al., 2006). Reactions between sucrose and aminoacids produce pyrazines
and carbonilic substances that also affect coffee flavors (De Maria et al., 1996).
Sucrose is the main free sugar present in green coffee, in amounts varying according to
species, maturity, origin and growth conditions, with averages of 6-9% db and 4-6% db for
Arabica and Robusta, respectively (Trugo, 1985; Murkovic & Derler, 2006). Sucrose levels
are expected to increase with coffee maturation and, since this is the main free sugar
avalaible, its levels will naturally decrease if fermentation occurs. Therefore, this parameter is
a potential candidate for separation of defective and non-defective beans and some reports are
available with respect to correlation with the presence of defects (Mazzafera, 1999; Bradbury,
2001; Vasconcelos et al., 2007).
Mazzafera (1999) compared the amount of reducing sugars and sucrose of immature,
black and immature-black coffees, and encountered lower sugar and sucrose contents for
immature beans, so it was concluded that the amount of sugar is primarily associated with the
developmental stage. It was also observed that the ratio between sucrose and reducing sugars
increased almost four times from immature-black to black beans. According to this study,
such difference can be attributed to the fact that, during seed filling, most of the carbohydrate
metabolism is directed to accumulation of storage polysaccharides in the coffee seed.
According to Bradbury (2001), the occurrence of black beans was associated to low
carbohydrate accumulation during coffee bean development, which resulted in approximately
50% and 75% lower sucrose levels in black beans compared to non-defective coffees of the
Arabica (from India) and Robusta (from Vietnam) species, respectively. Vasconcelos et al.
(2007) compared sucrose levels of non-defective, immature, sour and black beans. Average
reported levels were 1, 5 and 8% dry basis for black, immature/sour and non-defective,
respectively. A comparison of these results with the ones presented by Mazzafera (1999)
indicates that sucrose levels will be affected by both bean maturation stage and fermentation
and therefore it can not be accertained which factor will be predominant. Only traces of
sucrose were detected by Vasconcelos et al. (2007) in the light roasted coffee samples and no
sucrose was detected for the other roasting degrees. This behavior is in agreement with
literature data on sucrose degradation during roasting, ranging from 97% to 100% from light
to dark roasts (Trugo, 1985).
The literature data on the correlation between sugar levels and coffee quality is still
scarce. Chagas (1994) reported a positive association between coffee quality and content of
reducing and non-reducing sugars in coffee beans from Brazil. Farah et al. (2006) did not find
a direct correlation between sucrose levels and coffee quality and reported both the highest
(~8 % db) and lowest (~5 % db) sucrose levels for low quality samples. However, the sample
with the highest sucrose content contained only 8% defective beans whereas the amount of
defective beans in the sample with less sucrose was almost 40%. Ramalakshmi et al. (2007)
Adriana S. Franca and Leandro S. Oliveira
128
reported lower sucrose level for a mixture of defective beans in comparison to non-defective
ones. These results confirm that the presence of defective coffees will have a direct impact on
lowering sucrose levels of a given coffee sample.
3.6. Chlorogenic Acids
Chlorogenic acids (CGA) is the denomination given to a group of phenolic esters that
correspond to approximately 6 to 12% mass of green coffee. They are the main phenolic
compounds present in coffee and comprise esters of trans-cinnamic acids, such as caffeic,
ferulic and p-coumaric acids, with (−)-quinic acid (Farah et al., 2005). Such substances are
known to be responsible for coffee pigmentation, aroma formation, astringency and bitterness
(Clifford, 1985; De Maria et al., 1995; Farah et al., 2005; Farah et al., 2006). The major CGA
subgroups in coffee are the caffeoylquinic acids (CQA), feruloylquinic acids (FQA) and
dicaffeoylquinic acids (diCQA) (Clifford, 1985). Caffeoylquinic acids, specifically 3- 4- and
5-CQA, correspond to approximately 85% of the total CGA in green coffee, with 5-CQA
representing approximately 75% of the total CCA.
According to Farah et al. (2005) such class of compounds (CGA) in coffee has received
much attention lately due to various pharmacological activities observed in vitro and in vivo
studies. Examples include vascular activity (Bonita et al., 2007), antioxidant activity (Castilho
et al., 2002), inhibition of the mutagenicity of carcinogenic compounds (Stich, 1982), glucose
reduction in type 2 diabetes pacients (Van Dam, 2006), and others. Robusta coffees are
reported to have higher CGA contents (7-10%) than Arabica (5-8%) (Clifford, 1985; Perrone
et al., 2008). Martín et al. (1998) reported that the ratio between chlorogenic acids and
caffeine levels provides a good separation between Arabica and Robusta coffees of different
origins. According to Clifford (1985), there is an increase in total CGA content with maturity,
with the higher levels being attained approximately four weeks before full maturity. Also,
diCQA contents are reported to decrease with bean maturity.
CGA play an important role in the formation of roasted coffee flavor (Perrone et al.,
2008). CGA are partially degraded during roasting, producing phenolic lactones and other
derivatives. According to Clifford (1985) a decrease of approximately 8-10% of the original
CGA content is observed for every 1% weight loss (dry matter basis). They are first
hydrolyzed to quinic and caffeic acids, which then form phenolic volatiles during pyrolysis
(Homma, 2001). Given the fact that such compounds affect coffee flavor, in association with
their antioxidant and other physiological effects, some studies have investigated the relation
of such class of compounds with coffee quality (Franca et al., 2005b; Farah et al., 2006;
Ramalakshmi et al., 2007) and defective beans (Mazzafera, 1999; Franca et al., 2005a).
Mazzafera (1999) compared the levels of 5-CQA (by HPLC) and soluble phenols
(colorimetricaly) of immature, black and immature-black green coffees and reported that the
contents of soluble phenols and 5-CQA were approximately 35% higher in immature coffees
compared to black and immature-black. Franca et al. (2005a) evaluated 5-CQA levels of non-
defective, black, immature and sour coffee beans. It was observed that black beans had
significantly lower levels of 5-CQA (~1.8% db) compared to the other samples (~3.2% db).
Also, no significant differences in 5-CQA levels were observed for non-defective, immature
and sour beans. It is noteworthy to point out that 5-CQA levels reported in this study were on
the lower limit of the range reported by Clifford (1985): 3-5%. It was also reported that
Chemistry of Defective Coffee Beans 129
degradation of 5-CQA occurred to a smaller extent for black beans, which was attributed to
the lighter roasting degree attained by these defects as previously discussed in terms of other
attributes. The same research group (Franca et al., 2005b) evaluated 5-CQA levels of Arabica
coffees of different cup qualities (soft, hard, rioysh and rio), prior to and after roasting. The
highest (~2.5% db) and lowest (~2.3% db) 5-CQA contents were observed for the rioysh and
rio samples, respectively, which represented the low cup quality samples. However, it was
pointed out that the methodology employed in this study for evaluation of 5-CQA levels
needed revision, since average values were quite low compared to coffee literature data on 5-
CQA (Clifford, 1985; Farah, 2004; Farah et al., 2005). It is noteworthy to mention that the
sample with the highest 5-CQA levels was also the one presenting the highest amount of
immature beans. After roasting, 5-CQA levels presented slight variations among the samples,
without correlation with cup quality.
Farah et al. (2005) identified eight CGA in Brazilian coffees of different cup qualities: 3-
CQA, 4- CQA, 5-CQA, 3,4-diCQA, 3,5- diCQA, 4,5-diCQA, 4-FQA and 5-FQA, and
reported that the CQA accounted for approximately 83% of the total CGA in green beans. In
terms of total CGA, they reported the highest (~7.0% db) and lowest (~5.8% db) contents for
the worst and best quality samples, respectively. It was also observed that 5-CQA and 5-FQA
levels correlated strongly with poor cup quality, and 4-CQA, 5-CQA, 4-FQA and 5-FQA
levels correlated strongly with the Rio-off-flavor. Furthermore, the amount of immature beans
also correlated with poor cup quality and Rio-off-flavor. It was also reported that immature
and immature-black defective beans contained significantly higher levels of all CGA, but
mostly CQA and FQA, comparing to healthy and black defective beans (data not shown).
Such results are in agreement with the high percentage of immature beans in the low quality
samples employed in that study. The average loss of total CGA during roasting was 93%.
Correlations between CQA and FQA levels and cup quality became less significant with
roasting. Ramalakshmi et al. (2007) reported slightly higher CGA levels in low quality
coffees (mixture of defective beans) in comparison to high quality ones (non-defective
beans).
An evaluation of the published works on CGA in relation to coffee quality indicates that
high CGA levels, mainly 5-CQA, seem to be associated to the presence of immature beans.
However, since 5-CQA levels decrease as roasting progresses, such correlation is not
significant for roasted coffees.
3.7. Acidity and pH
The acidity of coffee brews has been considered an important sensory attribute, and a
certain level of acidity, that is quickly removed from the palate, is usually associated with
good cup quality (Freitas, 2001). Lightly roasted good quality coffees should present a
“clean” and “fine” acidity whereas such characteristic is not well perceived for dark roasts
(Woodman, 1985). Acidity, in its strict sense, is determined by the hydrogen ion
concentration related to the ionization degree of a given acid (or acids) present in an aqueous
solution. Presence of acids will affect flavor, either through the acid taste in the tongue
associated with the hydrogen ions, or through aroma perception if the undissociated acid is
volatile enough (Woodman, 1985). Adstringency effects on the tongue may be also
encountered in association with high acidity, and it will be clearly perceived if the coffee is
Adriana S. Franca and Leandro S. Oliveira
130
not properly roasted. Furthermore, the term sour is employed in coffee tasting in association
with an undesirable form of acidity. Given the significant effect of acidity on coffee flavor
and quality, and the fact that both pH and titrable acidity can be correlated to perceived
acidity (sensory), such parameters have also been investigated in terms of defective coffee
beans (Mazzafera, 1999; Vasconcelos et al., 2007) and cup quality (Franca et al., 2005b).
Mazzafera (1999) presented pH and acidity measurements for immature, black and
immature-black green (crude) coffees. Although significant differences were not reported for
pH measurements (~5.9), the lowest pH value corresponded to the highest acidity found in
immature beans (~3.5 mL NaOH g-1). Both black and immature-black beans presented similar
acidity (~2.5 mL NaOH g-1). Franca et al. (2005b) evaluated both pH and total acidity levels
of Arabica coffees of different cup qualities (soft, hard, rioysh and rio). It was reported that
acidity increased (and pH decreased) as cup quality worsened, i.e., as both the amount of
immature and sour beans increased. Ramalakshmi et al. (2007) also reported slightly higher
acidity levels for a mixture of defective beans in comparison to non-defective beans, and
attributed this to the presence of sour beans. High acidity of sour beans can be associated with
bean fermentation. Acidity levels presented a significant decrease with roasting (50-60%
reduction), and no significant differences were detected among the samples, even though the
tendency for increase in acidity with decrease in cup quality could still be observed. pH
values increased after roasting, with the highest values corresponding to the lowest quality
samples.
Vasconcelos et al. (2007) presented both acidity and pH measurements for defective
beans in comparison to non-defective ones, prior to and after light, medium and dark roasts.
Before roasting, acidity values were the highest for sour beans and the lowest for black beans.
The lowest acidity values observed for black beans were attributed to loss of acids during soil
contact. pH values were higher for defective beans in comparison to non-defective ones, with
the highest pH values encountered for black beans (~6). Acidity values decreased after
roasting, without differences among samples. Regarding pH measurements, values remained
higher for defective beans in comparison to non-defective ones, after roasting.
3.8. Volatile Components
A considerable amount of research has been done on the characterization of flavor related
substances in coffee (Grosch, 2001; Rocha et al., 2003). The most common classes of
compounds reported include acids, alcohols, aldehydes, alkanes, alkenes, benzenic
compounds, esters, furans, ketones, lactones, oxazoles, phenolic compounds, pyridines,
pyrazines, pyrroles, sulfur compounds, thiazoles, and thiophenes (Dart & Nursten, 1985; Sanz
et al., 2001; Yeretzian et al., 2003; Mondello et al., 2004; Ryan et al., 2004). Given the
direct impact of such components on flavor characteristics of roasted coffee, some recent
studies have focused on the comparison of the volatile profiles of roasted Arabica and
Robusta coffees and also of roasted coffees from different origins (Costa Freitas & Mosca,
1999; Costa Freitas et al. 2001; Rocha et al., 2003; Zambonin et al., 2005). It was concluded
that the volatile profiles provided the means for separation of Arabica and Robusta coffees
after roasting. Also, recent studies by Costa Freitas et al. (2001) and Zambonin et al. (2005)
showed that gas chromatography-mass spectrometry analysis (GC-MS) of the volatile
components of the coffee headspace concentrated by solid-phase micro-extraction (SPME) in
Chemistry of Defective Coffee Beans 131
combination with principal component analysis (PCA) also allowed for discrimination of
Arabica/Robusta blends having different geographical origins.
The successful application of chemometrics for discrimination between coffee varieties
and geographical origins indicated that this tool could also be evaluated for discrimination
between defective and healthy coffee beans and, thus, a recent study was presented to that
objective (Mancha Agresti et al., 2008). The coffee headspace volatiles profile was evaluated
for defective coffee beans (black, immature and sour) in comparison to non-defective ones,
after roasting to light, medium and dark degrees (roasting at 200oC for 30, 60 and 120 min,
respectively). Volatiles extraction and concentration were performed by solid phase micro-
extraction (SPME) of the roasted coffee headspace and analysis of the volatile profiles was
performed by GC-MS. Detection of approximately 250 volatile compounds was reported,
with only 2% (5 substances) being detected in all defects and not in the healthy coffee
samples. Sour beans presented the highest amount of substances not detected in healthy
coffee, followed by black and immature beans. Pyrazines and pyrroles were detected in
greater quantities in sour/black coffee beans in comparison to immature ones. Principal
component and hierarchical cluster analysis showed that the data could be separated into two
groups, one containing immature and black beans and the other containing healthy and sour
coffee beans. Such results are an indication that black and sour beans can be associated to
fermentation of immature and non-defective coffees, respectively. Some separation between
fermented/non fermented samples and also between roasting degrees was also observed.
4. CONCLUSION
Coffee is deemed as a commodity ranking second only to petroleum in terms of currency
(usually US dollars) traded worldwide (Illy, 2002). Brazil is the largest coffee producer and
exporter in the world and is the second largest consumer. As such this commodity is quite
relevant to the country economy. This is also true for other countries which are relevant
producers/exporters, such as Vietnam, Colombia, Ethiopia, Indonesia, Mexico and India. As
an agricultural product, the produced coffee is prone to present a parcel that does not fit to
established quality parameters and this parcel is denominated “defects”. Currently, in Brazil,
this category of beans amounts to 15 to 20% of the total production. Since such defects
negatively affect the beverage quality, it is desirable not to have them in the trading and
roasting markets. However, since these defects are intrinsic to production, harvesting and
processing of coffee, although separated from non-defective coffee in farms and cooperatives,
they are necessarily commercialized, because they represent an investment for the producers
and, also, because currently there are no alternative uses for them. In order to propose
alternative uses and to provide a means for the consumer awareness of quality, it is necessary
to understand what are the underlying physical-chemical differences between defective and
non-defective beans, green or roasted. Although coffee can be assuredly considered one of the
most studied food products in the last two centuries, only recently the physical-chemical
characterization of defective beans has been addressed for the purpose of
discrimination/differentiation.
Regarding physical attributes, it can be concluded that crude black and sour beans can be
satisfactorily discriminated from non-defective beans by size and color, while immature beans
Adriana S. Franca and Leandro S. Oliveira
132
cannot. Upon roasting, it was demonstrated by several studies that defective beans will roast
to a lesser degree than non defective ones. As a consequence, only black beans, which when
crude are usually smaller than other beans, may be discriminated from the others, since they
swell less and thus present higher density values than those for other beans under the same
roasting conditions. Hence, as a general conclusion, discrimination of immature beans by
means of physical attributes does not seem feasible and discrimination by color is only
feasible for green coffee.
As for chemical attributes, no significant differences in proximate compositions were
determined for defective and non defective coffee beans, both green and roasted. Again, black
beans were the ones that presented the most pronounced differences in specific chemical
components (caffeine, trigonelline, potassium) or classes of compounds (amines, lipids).
However, since this type of beans is usually encountered in small amounts within a sample
(2-3% w/w) it makes it difficult to detect and quantify it by means of specific chemical
markers. Histamine was deemed a potential marker for defective beans in green coffee since
it was detected only in those types of beans. However, since it was present in very small
amounts, its detection and quantification in a large sample of a mixture of defective and non
defective beans cannot be considered practical. A high CGA content was associated to the
presence of immature beans in mixtures with other beans and, thus, with further studies,
considering different crops and geographic regions, it could be considered as a potential
marker for the detection of this type of defect in green coffee samples. However, since both
CGA and histamine contents significantly decrease upon roasting, the use of these chemical
markers was demonstrated not to be feasible for discrimination of defective and non defective
beans in roasted coffees. The only promising alternative for discrimination of defective and
non defective beans in roasted coffees was the multivariate statistical analysis of the volatile
components of coffee headspace that allowed for separation between defective and non
defective coffees together with roasting degrees. There is still a lot to be done in terms of
research in regard to being capable of both qualitatively and quantitatively discriminating
defective from non defective coffee beans, preferably after roasting. In regard to qualitative
discrimination, geographic origin and crop year are parameters to be studied since they were
demonstrated to affect the chemical profile of coffees. As for quantitative discrimination, the
identification of chemical markers for each specific type of defect is imperative.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge financial support from the following Brazilian
Government Agencies: CNPq and FAPEMIG.
REFERENCES
Alcázar, A., Jurado, J.M., Martín, M. J., Pablos, F. and González, A.G. (2005) Enzymatic-
spectrophotometric determination of sucrose in coffee beans. Talanta, 67, 760-766.
Amorim, H.V., Basso, L.C., Crocomo, O.J. and Teixeira, A.A. (1977) Polyamines in green
and roasted coffee, Journal of Agricultural and Food Chemistry, 25, 957–958.
Chemistry of Defective Coffee Beans 133
Banks, M., McFadden, C. and Atkinson, C. (1999) The World Encyclopaedia of coffee ,
London: Anness Publishing Limited.
Bardócz, S. (1995) Polyamines in food and their consequences for food quality and human
health, Trends in Food Science and Technology, 6, 341–346.
Bonita, J.S., Mandarano, M., Shuta, D. and Vinson, J. (2007). Coffee and cardiovascular
disease: In vitro, cellular, animal, and human studies, Pharmacological Research, 55,
187-198.
Bradbury, A.G.W. (2001) Carbohydrates – Chemistry I: Non-volatile compounds. In
R.J.Clarke e O.G.Vitzthum (Eds.), Coffee Recent Developments (pp.1-17), London:
Elsevier Applied Science.
Casal, S., Mendes, E., Alves, M.R., Alves, R.C., Oliveira, M.B.P.P. and Ferreira, M.A. (2004)
Free and conjugated biogenic amines in green and roasted coffee beans, Journal of
Agricultural and Food Chemistry, 52, 6188–6192
Casal, S., Oliveira, M.B. and Ferreira, M.A. (2000) HPLC/diode array applied to the thermal
degradation of trigonelline, nicotinic acid and caffeine in coffee, Food Chemistry, 68,
481-485.
Castillo, D.M., Ames, J.M. and Gordon, M.H. (2002). Effect of Roasting on the Antioxidant
Activity of Coffee Brew, Journal of Agricultural and Food Chemistry, 50, 3698 – 3703.
Chagas, S. J. R. (1994). Caracterização química e qualitativa de cafés de alguns municípios
de três regiões produtoras de Minas Gerais (Chemical and qualitative characterization of
coffees from three regions in Minas Gerais State) MSc thesis, Universidade Federal de
Lavras (in portuguese).
Cirilo, M.P.G., Coelho, A.F.S., Araujo, C.M., Goncalves, F.R.B., Nogueira, F.D. and Gloria,
M.B.A. (2003). Profile and levels of bioactive amines in green and roasted coffee, Food
Chemistry, 82, 397-402.
Clarke, R.J. (1987a). Roasting and Grinding. In R.J. Clarke and R. Macrae (Eds.), Coffee Vol.
2: Technology (pp. 73- 107), London: Elsevier Applied Science.
Clarke, R. J. (1987b). Grading, Storage, Pre-treatments and Blending. In R.J. Clarke and R.
Macrae (Eds.), Coffee Vol. 2: Technology (pp. 35-58.), London: Elsevier Applied
Science.
Clarke, R. J. (1985). Water and mineral contents. In R.J. Clarke and R. Macrae (Eds.), Coffee
Vol. 1: Chemistry (pp. 42–82), London: Elsevier Applied Science.
Clarke, R. J. and Macrae, R. (1987). Coffee Vol. 2: Technology. London: Elsevier Applied
Science.
Clarke, R. J. and Macrae, R. (1985) Coffee Vol. 1: Chemistry. London: Elsevier Applied
Science.
Clifford, M.N. (1985) Chlorogenic Acids. In R.J. Clarke and R. Macrae (Eds.), Coffee Vol. 1:
Chemistry (pp. 153-202), London: Elsevier Applied Science.
Costa Freitas, A. M., Parreira, C. and Vilas-Boas, L. (2001). The Use of an Electronic Aroma-
sensing Device to Assess Coffee Differentiation - Comparison with SPME Gas
Chromatography-Mass Spectrometry Aroma Patterns, Journal of Food Composition and
Analysis, 14, 513-522.
Costa Freitas, A.M. and Mosca, A. I. (1999). Coffee geographic origin - an aid to coffee
differentiation, Food Research International, 32, 565-573.
Cristo, H.P., Martins, M.A., Oliveira, L.S. and Franca, A.S. (2006). Transverse flow of coffee
beans in rotating roasters, Journal of Food Engineering, 75, 142-148.
Adriana S. Franca and Leandro S. Oliveira
134
Custódio, B. M. O., Magalhães, F. S., Mansur, M. B., Franca, A. S. and Oliveira, L.S. (2005).
Elemental analysis of defective and non-defective beans in green and roasted coffee. In:
2005 ASAE Annual International Meeting, 2005, Tampa, FL. ASAE Paper no. 056085,
8p.
Dart, S.K. and Nursten, H.E. (1985) Volatile components. In R.J. Clarke and R. Macrae
(Eds.), Coffee Vol. 1: Chemistry (pp. 223-265), London: Elsevier Applied Science.
De Maria, C.A.B., Trugo. L.C., Neto, F.R.A., Moreira, R.F.A. and Alviano, C.S. (1996).
Composition of green coffee water-soluble fractions and identification of volatiles
formed during roasting. Food Chemistry, 55, 203-7.
De Maria, C.A.B., Trugo, L.C., Moreira, R.F.A. and Petracco, M. (1995) Simultaneous
determination of total chlorogenic acids, trigonelline and caffeine in green coffee samples
by high performance gel filtration chromatography, Food Chemistry 52, 447–449.
De Maria, C.A.B., Trugo, L.C., Moreira, R.F.A. and Werneck, C.C. (1994) Composition of
green coffee fractions and their contribution to the volatile profile formed during
roasting. Food Chemistry, 50, 141-145.
Dutra, E.R., Oliveira, L.S., Franca, A.S., Ferraz, V.P. and Afonso, R.J.C. (2001). A
preliminary study on the feasibility of using the composition of coffee roasting exhaust
gas for the determination of the degree of roast. Journal of Food Engineering, 47, 241–
246.
Farah, A., Monteiro, M.C., Calado, V., Franca, A.S. and Trugo, L.C. (2006). Correlation
between cup quality and chemical attributes of Brazilian coffee. Food chemistry, 98, 373-
380.
Farah, A., De Paulis, T., Trugo, L.C. and Martin, P.R. (2005). Effect of roasting on the
formation and degradation of chlorogenic acid lactones in roasted coffee, Journal of
Agricultural and Food Chemistry, 53,1505–1513.
Farah, A. (2004). Distribuição nos grãos, importância na qualidade da bebida e
biodisponibilidade dos ácidos clorogênicos do café. (Distribution in coffee beans,
importance in beverage quality and biodisponibility of chlorogenic acids in coffee) Ph.D.
Thesis, UFRJ (in Portuguese).
Fernandes, J. O. and Ferreira, M. A. (2000) Combined ion-pair extraction and gas
chromatography-mass spectrometry for the simultaneous determination of diamines,
polyamines and aromatic amines in Port wine and grape juice, Journal of
Chromatography A, 886, 183-195.
Ferreira, L.A.B., Fragoso, M.A.C., Peratta, M.F., Silvo, M.C.C. and Rebeto, M.C. (1971)
Constituintes minerais dos cafés de Angola. (Mineral constituents of Angola coffees).
Proc. 5th International Scientific Colloquium on Green and Roasted Coffee Chemistry,
159-173.
Folstar, P. (1985) Lipids. In R.J. Clarke and R. Macrae (Eds.), Coffee Vol. 1: Chemistry (pp.
203–222), London: Elsevier Applied Science.
Franca, A.S., Oliveira, L.S., Mendonça, J.C.F. and Silva, X.A. (2005a) Physical and chemical
attributes of defective crude and roasted coffee beans, Food Chemistry, 90, 84–89.
Franca, A.S., Mendonca, J.C.F. and Oliveira, S.D. (2005b) Composition of green and roasted
coffees of different cup qualities, LWT - Food Science and Technology, 38, 709-715.
Freitas, M.L.G. 2001. A última impressão é a que fica - Introdução à classificação e
degustação de café (The last impression stays... Introduction to coffee classification and
cupping) Sindicafé MG, 74 p. (in Portuguese).
Chemistry of Defective Coffee Beans 135
Ginz, M., Balzer, H.H., Bradbury, A.G.W. and Maier, H.G. (2000) Formation of aliphatic
acids by carbohydrate degradation during roasting of coffee. European Food Research
and Technology, 211, 404-410.
Gokmen, V. and Senyuva, H.Z. (2006). Study of colour and acrylamide formation in coffee,
wheat flour and potato chips during heating, Food Chemistry, 99, 238-243.
Goodrum, J.W. and Geller, D.P. (2002). Rapid thermogravimetric measurements of boiling
points and vapor pressure of saturated medium- and long-chain triglycerides. Bioresource
Technology, 84, 75–80.
Grosch, W. (2001). Chemistry III: Volatile compounds. In: Clarke, R.J. and Vitzthum, O.Z.
Coffee: Recent Developments, (pp. 68-89), Oxford: Blackwell Science.
Halász, A., Baráth, A., Simon-Sarkadi, L. and Holzapfel, W. (1994) Biogenic amines and
their production by microorganisms in food, Trends in Food Science and Technology, 5,
42–49.
Homma, S. (2001) Chemistry II: Non-volatile compounds, Part II. In R.J.Clarke e
O.G.Vitzthum (Eds.), Coffee Recent Developments (pp.50-67), London: Elsevier Applied
Science.
Illy, E. (2002) The complexity of coffee, Scientific American, June, 86-91.
Kalač, P. and Krausová, P. (2005) A review of dietary polyamines: formation, implications
for growth and health and occurrence in foods, Food Chemistry, 90, 219–230.
Lago, R.C.A. (2001) Lipídios em grãos de café (Lipids in coffee beans), B. CEPPA (Centro
de Pesquisa e Processamento de Alimentos), 19, 319–341 (in portuguese).
Macrae, R. (1985) Nitrogenous compounds. In R.J. Clarke and R. Macrae (Eds.), Coffee Vol.
1: Chemistry (pp. 115–152) London: Elsevier Applied Science.
Mancha Agresti, P.D.C., Franca, A.S., Oliveira, L.S. and Augusti, R. (2008) Discrimination
between defective and non-defective Brazilian coffee beans by their volatile profile, Food
Chemistry, 106, 787-796.
Marcone, M.F. (2004) Composition and properties of Indonesian palm civet coffee (Kopi
Luwak) and Ethiopian civet coffee, Food Research International, 37, 901-912].
Martín, M.J., Pablos, F. and Gonzalez, A.G. (1998) Discrimination between arabica and
robusta green coffee varieties according to their chemical composition, Talanta, 46, 1259-
1264.
Martin, M.J., Pablos, F. and Gonzalez, A.G. (1999) Characterization of Arabica and Robusta
varieties and misture resolution according to their metal content. Food Chemistry, 66,
365-370.
Mazzafera, P. (1999) Chemical composition of defective coffee beans, Food Chemistry, 64,
547–554.
Mendes, L.C., Menezes, H.C., Aparecida, M. and Silva, A.P. (2001) Optimization of the
roasting of robusta coffee (C. canephora conillon) using acceptability tests and RSM,
Food Quality and Preference, 12, 153-162.
Mendonça, J.C.F.(2006) Avaliação do perfil de carboidratos em grãos de café defeituosos e
sadios das espécies Coffea arabica e Coffea canephora (Evaluation of the carbohydrate
profiles of defective and non-defective coffees of Coffea arabica and Coffea canephora
species). Doctoral Proposal, Universidade Federal de Minas Gerais, Belo Horizonte,
MG, Brazil, 114 pp. (in Portuguese).
Adriana S. Franca and Leandro S. Oliveira
136
Mendonça, J.C.F., Franca, A. S. and Oliveira, L.S. (2007) A comparative evaluation of
methodologies for water content determination in green coffee, LWT - Food Science and
Technology, 40, 1300-1303.
Mondello, L., Casilli, A., Tranchida, P.Q., Dugo, P., Costa, R., Festa , S. and Dugo, G.
(2004). Comprehensive multidimensional GC for the characterization of roasted coffee
beans, Journal of Separation Science, 27, 442-450.
Murkovic, M. and Derler, K. (2006). Analysis of amino acids and carbohydrates in green
coffee, Journal of Biochemical and Biophysical Methods, 69, 25-32.
Nikolova-Damyanova, B., Velikova, R. and Gulab, N. J. (1998). Lipid classes, fatty acid
composition and triacylglycerol molecular species in crude coffee beans harvested in
Brazil, Food Research International, 31, 479–486.
Oliveira, L.S., Franca, A.S., Camargos, R.R.S. and Ferraz, V.P. (2008) Coffee oil as a
potential feedstock for biodiesel production, Bioresource Technology, 99, 3244-3250.
Oliveira, L.S., Franca, A. S., Mendonça, J.C.F. and Barros-Júnior, M.C. (2006). Proximate
composition and fatty acids profile of green and roasted defective coffee beans, LWT -
Food Science and Technology, 39, 235–239.
Oliveira, S.D., Franca, A. S., Gloria , M.B.A. and Borges, M.L.A. (2005). The effect of
roasting on the presence of bioactive amines in coffees of different qualities, Food
Chemistry, 90, 287-291.
Oosterveld, A., Voragen, A.G.J. and Schols, H.A. (2003). Effect of roasting on the
carbohydrate composition of coffea arabica beans, Carbohydrate Polymers, 54, 183-192.
Perrone, D., Farah, A., Donangelo, C.M., Tomas De Paulis, T. and Martin, P.R. (2008)
Comprehensive analysis of major and minor chlorogenic acids and lactones in
economically relevant Brazilian coffee cultivars, Food Chemistry, 106, 859-867.
Ramalakshmi, K., Kubra, I.R. and Rao, L.J.M. (2007). Physicochemical Characteristics of
Green Coffee:Comparison of Graded and Defective Beans, Journal of Food Science, 72,
S333-S337.
Ravindranath, R., Yousuf Ali Khan, R. and Reddy, O.T. (1972) Composition and
characteristics of Indian coffee bean, spent grounds and oil. Journal of Food and
Agriculture Science, 23, 307–310.
Reh, C.T., Gerber, A., Prodolliet, J. and Vuataz, G. (2006) Water content determination in
green coffee - Method comparison to study specificity and accuracy, Food Chemistry, 96,
Issue 3 423-430.
Rocha, S., Maeztu, L., Barros, A., Cid, C. and Coimbra, M.A. (2003). Screening and
distinction of coffee brews based on headspace solid phase microextraction / gas
chromatography / principal component analysis, Journal of the Science of Food and
Agriculture, 84, 43-51.
Rodrigues, M.A.A., Borges, M. L. A., Franca, A. S. , Oliveira, L.S. and Correa, P.C. (2003)
Evaluation of physical properties of coffee during roasting. E-journal. International
Commission of Agricultural Engineering, 5, Manuscript FP 03004, 12 p.
Rossell, J. B., and Pritchard, J. L. R. (1991). Analysis of oilseeds, fats and fatty foods.
London: Elsevier Applied Science.
Ryan, D., Shellie, R., Tranchida, P., Casilli, A., Mondello, L. and Marriott, P. (2004).
Analysis of roasted coffee bean volatiles by using comprehensive two-dimensional gas
chromatography-time-of-flight mass spectrometry, Journal of Chromatography A, 1054,
57-65.
Chemistry of Defective Coffee Beans 137
Santos, M.H.S. (1996). Biogenic amines: their importance in foods, International Journal of
Food Microbiology, 29, 213–231.
Santos, E.J. and Oliveira, E. (2001). Determination of mineral nutrients and toxic elements in
brazilian soluble coffee by ICP-AES, Journal of Food Composition and Analysis, 14,
523-531.
Sanz, C., Ansorena, D., Bello, J. and Cid, C. (2001). Optimizing headspace temperature and
time sampling for identification of volatile compounds in ground roasted Arabica coffee,
Journal of Agricultural and Food Chemistry, 49,1364-1369.
Sivetz, M. and Desrosier, N.W. (1979). Coffee Technology. Westport ,CO: Avi Publishing
Co.
Speer, K. and Kölling-Speer, I. (2001). Non-volatile compounds – lipids. In R.J. Clarke and
O.G. Vitzhum (Eds.), Coffee: Recent Developments (pp. 33–49), Oxford: Blackwell
Science.
Spiz, R.R., Jablonka, F.H. and Pereira, D.A. (1989). Composição em ácidos graxos:
Triacilgliceróis, monoésteres de diterpenos e óleo de grãos de café. Congresso Brasileiro
de Ciência e Tecnologia de Alimentos (p. 151). Resumos, Rio de Janeiro: SBCTA (in
portuguese).
Stich, H. F., Rosin, M. P. and Bryson, L. (1982) Inhibition of mutagenicity of a model
nitrosation reaction by naturally occurring phenolics, coffee and tea. Mutation Research,
95, 119-128.
Summa, C.A., De la Calle, B., Brohee, M., Stadler, R.H. and Anklam, E. (2007) Impact of the
roasting degree of coffee on the in vitro radical scavenging capacity and content of
acrylamide, LWT - Food Science and Technology, 40, 1849-1854.
Trugo, L.C. (1985) Carbohydrates In R.J. Clarke and R. Macrae (Eds.), Coffee Vol. 1:
Chemistry (pp. 83-114), London: Elsevier Applied Science.
Turatti, J. M. (2001). Extração e caracterização de óleo de café (Extraction and
characterization of coffee oil), Proceedings of Simpósio de Pesquisa dos Cafés do Brasil,
Vitória, Brazil, 2, 1533–1539 (in portuguese).
Van Dam, R.M. (2006). Coffee and type 2 diabetes: From beans to beta-cells, Nutrition,
Metabolism and Cardiovascular Diseases, 16, 69-77.
Vasconcelos, A.L. 2005. Caracterização de grãos defeituosos de café quanto a aspectos físico-
químicos e teores de aminas, para diversos graus de torração (Characterization of
defective coffee beans with respect to physical and chemical attributes and amine leves,
for several roasting degrees) M.Sc. Thesis, Universidade Federal de Minas Gerais, Belo
Horizonte, MG, Brazil, 101 pp. (in Portuguese).
Vasconcelos, A.L.S., Franca, A.S., Glória, M.B.A. and Mendonça, J.C.F. (2007) A
comparative study of chemical attributes and levels of amines in defective green and
roasted coffee beans. Food Chemistry, 101, 26-32.
Vega-Carrillo, H.R., Iskander, F.Y. and Manzanares-Acuña, E.(2002). Elemental content in
ground and soluble/instant coffee, Journal of Radioanalytical and Nuclear Chemistry,
252,75-80.
Vila, M.A., Andueza, S., de Peña, M.P. and Cid, C. (2005). Fatty acid evolution during the
storage of ground, roasted coffees, JAOCS, 82, 639-646.
Vincent, J.C. (1987) Green Coffee Processing. In R.J. Clarke and R. Macrae (Eds.), Coffee
Vol. 2: Technology (pp. 1 -58), London: Elsevier Applied Science.
Adriana S. Franca and Leandro S. Oliveira
138
Woodman, J.S. (1985) Carboxylic Acids In R.J. Clarke and R. Macrae (Eds.), Coffee Vol. 1:
Chemistry (pp. 266-289), London: Elsevier Applied Science.
Yeretzian, C., Jordan, A. and Lindinger, W. (2003). Analysing the headspace of coffee by
proton-transfer-reaction mass-spectrometry,International Journal of Mass Spectrometry,
223-224,115-139.
Zambonin, C.G., Balest, L., De Benedetto, G.E. and Palmisano, F. (2005). Solid-phase
microextraction-gas chromatography mass spectrometry and multivariate analysis for the
characterization of roasted coffees, Talanta, 66, 261-265.