3.2. Determination of Caffeine and Volatile Compounds in Coffee by Chromatography
Caffeine is a methylxanthine that contributes to the bitter taste of coffee and whose quantity varies depending on the species of coffee [
38]. A liquid chromatographic analysis was carried out for the determination of caffeine in samples of Geisha specialty coffee roasted in natural and washed processing.
This caffeine determination yielded similar values for both post-harvest processing methods (dry and wet) of Geisha coffee. The results are shown in
Table 3.
Gas chromatography was used to identify VOC in dry and infused samples. The identification of the VOC of specialty Geisha ground coffee with three roasting levels and two elaboration processes, natural and washed, as well as their respective infusions, was carried out. On the other hand, the VOC of a commercial roasted ground coffee of the robusta variety were identified to compare its volatile profile with the specialty Geisha coffee. The results obtained are shown in
Table 4.
A total of 57 different VOC were identified among all the analyzed samples. Common compounds found in all of them included Acetic acid, 2-Furancarboxaldehyde (with the characteristic odors of toast, caramel, and coffee), 2-Furanmethanol, acetate (with a fruity banana odor), and 1H-Pyrrole, 1-(2-furanylmethyl)- (with vegetable, potato, or mushroom odors). Acetic acid is noteworthy in coffee; besides its characteristic vinegar taste, at low levels, it can give rise to pleasant lime flavors, and at high concentrations, it emits a fermentative odor.
However, some compounds were only found in the commercial robusta coffee, contributing distinctive aromas. These compounds include 2,3-Butanediol (with creamy and buttery aromas), Phenol, 2-methoxy- and 4-ethyl guaiacol (imparting smoky and woody odors), Pyrazine, (1-methylethenyl)- (with chocolate and nutty notes), 5H-5-Methyl-6,7-dihydrocyclopentapyrazine (evoking the aroma of roasted coffee beans), Pyrazine, 3,5-diethyl-2-methyl- (with a green nutty scent), Pyridine (featuring an ammoniacal and sour fishy aroma), and Indole (exhibiting a fecal and musty odor). Among these mentioned compounds, some have unpleasant aromas that are not found in the specialty Geisha coffee; therefore, their presence and aromas could raise doubts about the purity of the analyzed Geisha coffee. The results of the VOC of Geisha coffee are shown in
Figure 2.
The presence of acidic compounds in ground coffee diminishes as the roast level increases, with a higher roast level resulting in fewer acids. Additionally, depending on the post-harvest treatment, their quantity is higher in natural Geisha than in washed Geisha. Furthermore, the robusta variety contains fewer acidic compounds than the arabica variety. Once the coffee is brewed, these quantities decrease drastically in the headspace, possibly due to the dilution effect of the water. Although these compounds are associated with sour, rancid aromas and vinegar or cheese odors, in small quantities they give Geisha coffee aromas of fresh, lime-like coffee.
Regarding aldehydes, they are compounds of intermediate polarity and are only found in infused Geisha coffee, with chocolate, nutty, and spicy aromas. However, the higher polarity group of alcohols only appears in ground coffee with herbal fragrances, possibly because they remain soluble in water and evaporate less.
The furans in Geisha coffee increase with the degree of roasting and contribute aromas of toasted bread, almonds, walnuts, and caramel, among others. These results agree with authors who stated that furans are the major compounds in roasted coffee [
39]. With respect to post-harvest treatment, they are found in greater proportion in washed Geisha than in natural coffee. These volatiles also appear in robusta coffee and, in much smaller quantities, in Geisha infusion.
Ketones with sweet, herbal, floral, and licorice-like aromas, as well as phenols with medicinal, woody, and smoky scents, do not appear to be significantly influenced by post-harvest treatment or roasting degree. Instead, their presence depends on the coffee variety, with notably higher levels in the robusta variety compared to arabica Geisha. Ketones such as phenols only appear in ground coffee and are not present when Geisha coffee is prepared as an infusion.
Pyrazines, which contribute roasted, chocolate, and sometimes musty and potato-like aromas, are significantly influenced by the post-harvest process, with notably higher levels in the dry process. They are also influenced to a lesser degree by the coffee variety or degree of roasting. These compounds do not appear in the infusion of arabica coffee.
The pyridine compound with sour fish and ammoniacal aromas is exclusive to the robusta coffee sample and is not detected in the studied arabica variety, whether in ground or infusion. Therefore, it can be regarded as a key compound for identifying the presence of robusta coffee mixed with arabica specialty coffee.
On the other hand, the number of pyrroles with burnt, smoked, and cooked potato aromas is not modified by variety, post-harvest treatment, or roasting degree. However, their quantity decreases greatly in infused coffee.
Terpenes with floral, herbal, citrus, and fresh aromas appear in much more significant quantities in Geisha coffee infusions and do not change in quantity depending on the processing or degree of roasting. These compounds are not found in the robusta variety.
3.3. Sensory Analysis of Coffee
Sensory analysis of ground and brewed coffee was carried out following the SCA protocol. The attributes evaluated included fragrance, aroma, flavor, aftertaste, residual flavor, acidity, body, and sweetness. Perceived descriptors were used to score these attributes. Most descriptors were related to the fragrance and aroma of Geisha Natural coffee: NL fragrance—caramel, barley, lemon, and NL aroma—honey, chocolate cereal; for the NM fragrance, fruity, biscuit and malt, and NM aroma—red fruits, cereal, citrus; ND fragrance—toasted bread and chocolate; and ND aroma—roasted, biscuit, citrus. In the case of Geisha Washed coffee, the descriptors identified were WL fragrance—caramel, barley, cereal, and WL aroma—roasted, malt, citrus; WM fragrance—bread, malt, lemon, and WM aroma—roasted bread, chocolate;, WD fragrance—fruity, cereal, biscuit, and WD aroma-roasted, floral, caramel.
The results of the sensory analysis of Geisha Natural and Washed coffee at the three roasting levels are shown graphically in
Figure 3.
The graph shows different sensory profiles for the six Geisha coffee samples. The highest acidity values correspond to the dark roasts. In terms of flavor, the highest score was for the WL, while the values were similar for the rest of the samples. Regarding residual flavor, the highest score was obtained for ND coffee. The score for fragrance and aroma was highest for the NC and WD samples.
To determine the significant differences between roasting levels within the same processing, a one-factor ANOVA was applied. To see the significant differences between roasting levels within each type of processing, a Student’s
t-test was used (
Table 5).
When comparing the three roast levels of Natural coffee, there were significant differences in fragrance between light and dark roasts, but not between light and medium roast levels. Regarding aroma, significant differences were found between light roast and medium and dark roasts, but not between medium and dark roasts. Flavor, residual flavor, and body did not exhibit significant differences. For acidity and sweetness, there were only significant differences between light and dark roast.
For the three roast levels of Washed coffee, significant differences were found in fragrance between light roast and medium and dark roast, but not between medium and dark roast. For aroma, there are significant differences between light and dark roast. There are no significant differences for the other attributes flavor, residual flavor, acidity, body, and sweetness.
If we compare the three roasting levels of the two processes, Natural and Washed, the only significant differences between the two processes are in fragrance between light and dark roasts and in aroma at all three levels. There were no significant differences between flavor, residual flavor, acidity, body, and sweetness.
In addition, sensory analysis of the fragrance of pure samples of specialty Geisha coffee and commercial robusta coffee was conducted. Furthermore, Geisha coffee was adulterated with 10%, 20%, and 50% robusta coffee to assess whether the cupping panel and EN could differentiate between them. The ANOVA results of the sensory analysis of pure and blended samples are shown in
Table 6.
In the sensory analysis, the ANOVA results for fragrance show no significant difference between N100 and N90R10, but there are significant differences with the other proportions. Therefore, the cupping panel was only able to detect the addition of 20% or more coffee to Geisha coffee.
Robusta coffee is known for its very strong fragrance. However, it may not be readily perceived when blended in small concentrations. Nevertheless, as the concentration of robusta coffee increases, its strong fragrance can overpower that of Geisha Natural, especially in the 50% blend.
3.4. Evaluation of the Olfactory Pattern of Coffee with an Electronic Nose
The olfactory pattern was evaluated by the EN using three samples each of light, medium, and dark roasted natural dry-ground Geisha coffee, as well as three samples each of light, medium, and dark roasted wet-ground or washed Geisha coffee. Additionally, the olfactory pattern of these infused coffee samples was evaluated. The data and graphs of the sensor signals whose processed data give rise to the graphs in
Figure 4,
Figure 5,
Figure 6 and
Figure 7 have been added to the
Supplementary Materials section.
The data obtained from the EN measurements were subjected to exploratory principal component analysis to group the samples according to their olfactory pattern for different roast levels.
Figure 4 shows the graphs of the first two principal components. Graphs (a) and (b) correspond to Natural and Washed roasted ground coffee.
The results showed that the EN could differentiate three groups of coffee with good resolution in separation according to their roast level, with the medium roast in the center. In (a) and (b), the total variance of the data was explained by 85.8% and 90.1%, respectively.
In
Figure 5, graphs (a) and (b) correspond to infused coffee. The PCA results for the brewed coffee, like those for ground coffee, exhibited good resolution, with data from the various coffee roasts clustering similarly. In (a) and (b), the total variance of the data was explained by 91.43% and 85.66%, respectively.
On the other hand,
Figure 6 shows the data according to the type of brewing. A clear resolution is observed for Geisha Natural and Washed coffees. Graph (a) shows the values for ground coffee, and graph (b) for brewed coffee. The first two principal components explain 82.9% and 94.9% of the total variance of the data in (a) and (b), respectively.
In addition, the olfactory pattern of blended arabica and robusta coffee was evaluated with the EN.
Figure 7 shows in graph (a) a sample of specialty Geisha coffee to which different portions of a commercial robusta coffee were added: 0% (G100), 10% (G90R10), 20% (G80R20), 50% (G50R50), and pure robusta (R100). It can be seen how the pure Geisha coffee, depicted in green, forms an isolated group that is distinct from the robusta variety coffees. The robusta coffee variety is also clustered separately, represented in red, and is closely situated to the 50% blend, represented by the pink color.
The same samples were also subjected to a PLS analysis to evaluate the prediction of adulteration of Geisha coffee with robusta portions with the EN.
Figure 7b shows a good correlation between the actual values obtained from the cupping panel and the values predicted by the EN.
Furthermore,
Figure 8 shows a radial graph where the different olfactory patterns can be observed according to the proportions of the coffee blends.
The sensors exhibited the highest response amplitudes for Geisha coffee, depicted in green. In contrast, for robusta coffee, the responses were the lowest among most of the sensors. The blends displayed intermediate response amplitudes between the two coffee varieties.