- Email: [email protected]
Comparison of espresso coffee brewing techniques
Journal of Food Engineering 121 (2014) 112–117
Contents lists available at ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
Comparison of espresso coffee brewing techniques Alessandro Parenti a,⇑, Lorenzo Guerrini a, Piernicola Masella a, Silvia Spinelli a, Luca Calamai b,c, Paolo Spugnoli a a
Dipartimento di Gestione dei Sistemi Agrari, Alimentari e Forestali (GESAAF), Università degli Studi di Firenze, Piazzale delle Cascine 15, 50144 Firenze, Italy Dipartimento di Scienze delle Produzioni Agroalimentari e dell’Ambiente (DISPAA), Università degli Studi di Firenze, Piazzale delle Cascine 18, 50144 Firenze, Italy c Centro di servizi di Spettrometria di Massa (CISM), Università degli Studi di Firenze, Via U. Schiff 6, 50019 Sesto Fiorentino (FI), Italy b
a r t i c l e
i n f o
Article history: Received 26 March 2013 Received in revised form 27 June 2013 Accepted 12 August 2013 Available online 28 August 2013 Keywords: Coffee machines Capsule Foam Espresso reproducibility
a b s t r a c t Several brewing techniques are used to make espresso coffee. Among them, the most widespread are bar machines and single-dose capsules, designed in large numbers because of their commercial popularity. As none of the current literature compares the effects of these different brewing techniques on espresso quality, this paper looks at two capsule methods and the traditional bar method. The methods were evaluated on the basis of the physico-chemical parameters and aromatic profile of nine espresso coffees prepared using the different techniques. Our results showed that with the same batch of roasted coffee, the same water and the same operative settings, the three different techniques can be distinguished by a principal component analysis. Furthermore, in terms of product reproducibility, the best results are provided by the two capsule systems. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction There are a large number of devices and methods to produce espresso coffee (EC). EC is defined as ‘‘a brew obtained by percolation of hot water under pressure through compacted cake of roasted ground coffee, where the energy of the water pressure is spent within the cake’’ (Illy et al., 2005). In this process flavors are extracted from the coffee by means of hot water. EC consists of about 35 ml of dark beverage, usually served in a small cup with a brown foam layer called crema covering the liquid. Crema is a distinctive feature of EC, as it is absent in other coffee brews and is required for consumer acceptance. EC is conventionally brewed using bar machines (BM), which consist of a rotating pump, a heat exchanger and an extraction chamber (Illy et al., 2005). The water pressure provided by the pump strongly affects the physical and sensory properties of the brew (Andueza et al., 2002) and maximal EC quality seems to correspond to an optimal water pressure of nine bars. ECs prepared at higher pressure have negative sensorial qualities as they are excessively bitter, astringent and contain more key odorants. In conventional EC preparation pressurized water reaches the heat exchanger where its temperature rises. In this type of machine Andueza et al. (2003) found the best key odorant profile, flavor notes and highest overall acceptability at 92 °C. Coffees brewed at lower temperatures had less odor, flavor and body intensity
⇑ Corresponding author. Tel.: +39 553288319; fax: +39 55 3288316. E-mail address: alessandro.parenti@unifi.it (A. Parenti). 0260-8774/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jfoodeng.2013.08.031
(they are generally called ‘‘under extracted’’), whereas coffees brewed at higher temperatures had a burnt/roast flavor and a higher content of negative key odorants (they are generally called ‘‘over extracted’’). The extraction chamber consists of a filter where the ground coffee is placed and compacted. The filter provides the hydraulic resistance required to produce EC. During this procedure there are many variables that cannot be controlled by the extraction device (e.g. the coffee powder particle size, powder compression); these have a high impact on the properties of the final brew (Illy et al., 2005) and depend on the ability of the barman. Among the other methods developed to make EC, pod and capsule systems have recently gained market share because they are user-friendly. They also make it easy to prepare good-quality coffees through the reduction of uncontrolled preparation variables. Furthermore, these systems preserve the quality of the ground coffee by protecting it against moisture and oxidation processes (Vanni, 2009). For these reasons, in 2005, pod and capsule sales added up to 14 billion units (Tozzi, 2007). As a result of their popularity several kinds of capsule have been developed. The simplest consists of a chamber to hold the coffee and a film that provides the needed resistance when water is added. More sophisticated capsules are equipped with devices that should lead to the production of top-quality EC. However, the environmental impact of these approaches is significantly higher than other preparation methods. This is mainly due to the production of disposable capsules that cause significant greenhouse gas emissions (Brommer et al., 2011). In fine EC quality is strongly affected by the operative conditions of the extraction, which differ depending on the device.
113
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
To the best of the authors’ knowledge, there are no studies that compare different preparation techniques (e.g. BM and capsule) in terms of the sensory and chemical characteristics of the resulting EC. In this work we evaluate and compare the differences in terms of quality between EC made using three different extraction procedures. The differences are assessed in terms of sensory characteristics, physical parameters, and extracted volatile key compounds. The selected methods were: the bar method (the traditional way to make EC), a simple commercial capsule method (Illy; I-Espresso System), and a more advanced commercial capsule method (Illy; HyperEspresso), primarily designed to increase the sensory attributes related to the colloidal state of the beverage (Navarini et al., 2008). 2. Materials and methods 2.1. Experimental design The extraction methods were compared through the preparation of three ECs per day over a period of 3 days for each device, resulting in a total of 27 ECs. The order of preparation for each day was completely random. 2.2. Extraction devices 2.2.1. Bar machine (BM) A conventional bar machine (the Alina model manufactured by Cimbali S.p.A. Italy) was used. The machine was designed to make two ECs at the same time in the same extraction chamber by splitting the exit flow equally. Therefore, the ECs were prepared with double the amount of ground coffee (14.5 ± 0.2 g). Chemical and physical analyses were only performed on one of the two ECs. The extraction parameters were: water temperature 92 °C, water pressure 9 bar, and 25–30 s of percolation time, assuming an optimal flow rate of about 1 ml s 1 (Illy et al., 2005). 2.2.2. Hyper Espresso method (HIP) The capsules were brewed using the Good News coffee machine (manufactured by D.P.I. Service SNC, Italy), designed for Hyper Espresso capsules (produced by illycaffè S.p.A, Italy) at a pressure of 12 bar and an extraction temperature of 92 °C. The HIP capsules contained 6.7 ± 0.1 g of ground coffee and consisted of five parts: a cover, an upper and a lower internal filter, an infusion chamber and a flow conveyor. This design only allows the EC to flow out of the capsule when a fixed pressure is reached. During EC preparation the upper side of the capsule (the cover) is punched and the water is added (pre-infusion phase); then the water compresses the gas in the capsule and the pressure rises (infusion phase) until the capsule film bursts; once the pressure to burst the capsule is reached, the coffee flows out through a micro-hole (emulsion phase). 2.2.3. I-Espresso System (IT) The capsules were brewed using the Mitaca machine (manufactured by illycaffè S.p.A, Italy). The capsules contained 6.9 ± 0.1 g of ground coffee and consisted of a plastic cylinder covered by a plastic film. Hot water at 92 °C is introduced into the capsule. The bottom of the capsule has a central hole allowing EC outflow when a given pressure is reached.
Table 1 The physico-chemical characteristics of mineral water as listed on the bottle’s label. Analytical parameter
Values
pH Electrical conductivity (20 °C) Total dissolved solids Hardness Kubel oxydability Free carbon dioxide Calcium (Ca2+) Magnesium (Mg2+) Sodium (Na+) Potassium (K+) Hydrogen carbonate (HCO3 )
8.1 249 lS/cm 148 mg/l 14 °F 0.6 mg/l 3.3 mg/l 30.1 mg/l 15.0 mg/l 1.4 mg/l 0.5 mg/l 157 mg/l 10.7 mg/l
Sulfate (SO24 ) Nitrate (NO3 ) Chloride (Cl ) Fluoride (F ) Silicon dioxide (SiO2)
5.0 mg/l 1.5 mg/l 0.06 mg/l 6.6 mg/l
the HIP and IT capsules. The remaining roasted beans were used for the BM trials. These beans were ground immediately before preparation, using a professional coffee grinder (KE640 model manufactured by Ditting Maschinen AG, Switzerland). The resulting particle size distribution was: 29% >500 lm; 250 lm < 47.4% < 500 lm; 125 lm < 22.2% < 250 lm; and 1.4% < 125 lm. 2.3.2. Water According to Navarini and Rivetti (2010), water quality plays a key role in EC quality. Consequently, all tests were performed using the same commercial brand of mineral water. The physical and chemical characteristics of this water, according to the manufacturer’s specification, are shown in Table 1. 2.4. Measurements and analyses All brewed coffee samples were immediately collected at the outflow of the machine in a glass weighing bottle (75 ml volume, 53 mm internal diameter, 34 mm high) with a ground glass lid and equipped with two valves specifically designed for the sampling of the headspace above the coffee (described in more detail later). In order to obtain homogeneous samples, the same weight of percolated liquid was collected, regardless of flow rate or percolation time. Thus, a digital scale (max capacity 300.0 g; precision 0.1 g; manufactured by D-Mail S.R.L., Italy) was placed under the vessel and a preselected weight of 25 g of brewed coffee was collected. The resulting final brew weight was 25.7 ± 0.6 g averaged over all the samples. The temperature of the outflowing coffee was measured directly under the liquid flow with a digital thermometer (HD2107.1, manufactured by Delta OHM S.R.L., Italy). The following parameters were analyzed and evaluated for all samples. 2.4.1. Foam Index and persistency The foam index is defined as the ratio between the foam and liquid volume (vol vol 1%) measured 30 s after extraction (the geometry of the sampling vessel is given above). Persistency is defined as the time (in minutes) before the foam breaks up, leaving an uncovered black spot on the surface of the beverage (Petracco, 2001).
2.3. Espresso coffee preparation 2.3.1. Coffee All the ECs were prepared from the same batch of roasted coffee beans, provided by illycaffè S.p.A. (Italy). Some of the roasted beans were ground (Colombini Icoperfexand grinder) and used to prepare
2.4.2. Density, pH, and viscosity Before taking these measurements samples, were cooled to 20 °C. Density was measured with a 25 ml pycnometer. A digital pH meter (GLP 21, manufactured by CRISON INSTRUMENTS, S.A. Spain) was used to determine the pH of the ECs. Viscosity was
114
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
measured with a capillary viscometer (Ostwald) fitted with an automatic optical reader (Viscoclock, manufactured by SCHOTT Instruments GmbH, Germany) and expressed as mN s m 2. 2.4.3. Refractive index The refractive index was measured with a portable digital refractometer (Refracto 30PX, manufactured by Mettler Toledo S.p.A., Italy), using the total internal reflection method, a light source wavelength of k = 589.3 nm and automatic temperature compensation set at 20 °C. 2.4.4. Total solids, extraction, concentration Total solids (expressed as mg ml 1) were determined by ovendrying 15.4 ± 1.4 ml of EC until a constant weight was reached (18 h, 104 ± 2 °C). The extraction percentage was defined as the percentage by weight of total solids with respect to the dosage of ground roast coffee. Concentration was defined as percentage ratio between total solids and the volume of EC (w vol 1%). 2.4.5. Lipids The total lipid concentration (mg ml 1) was determined by liquid–liquid extraction using n-hexane as the solvent. About 10 ml of EC was extracted using 5 ml of n-hexane three times and centrifuged at 3500 rpm for 3 min. Each time the hexane phase was pipetted out and placed in a crystallizer. The total weight of lipids was determined after the solvent had completely evaporated. 2.4.6. Caffeine, trigonelline and chlorogenic acids Extraction, sample preparation and high-performance liquid chromatography (HPLC) conditions were carried out as described by Maeztu et al. (2001b). The HPLC system consisted of an Agilent 1200 Rapid Resolution unit equipped with a diode array detector and fitted to an Agilent 6410 QQQ mass spectrometer operating in scan mode. An alternating positive and negative ionization switching (every 500 ms) was set up for optimal ionization of basic (caffeine and trigonelline) and acidic (chlorogenic acids) molecules. A Varian (2 mm id, 15 cm length, particle size 3 lm, porosity 100 Å, temperature 60 °C) Polaris HPLC column was used for the separation of compounds. The mobile phases were: water with 0.1% formic acid and acetonitrile. The flow rate was constant at 0.4 ml/min and the eluent composition was water/acetonitrile (95:5 v/v) for 1 min, gradient 5–60% of acetonitrile for 20 min, gradient 60–90% of acetonitrile for 3 min, water/acetonitrile (10:90 v/ v) for 1 min, gradient 90–5% of acetonitrile for 2 min, and 10 min equilibration at the initial conditions. Quantitative data was obtained through the construction of calibration lines with authentic standards both in UV-DAD and MS chromatograms in the 1–20 mg ml 1 range. Under analysis conditions the retention times for trigonelline, caffeine, 4-caffeoilquinic acid, and 3,4-dicaffeoilquinic acid were respectively 1.4, 8.8, 5.8, and 14.5 min. 2.4.7. Volatile compound determination It is known that perceived EC aroma is only partially related to the total amount of aromatic compounds extracted during preparation. It also depends upon conditions such as the temperature of the EC, and the presence/absence of foam. Consequently, a new method was developed for the evaluation of the volatile compounds escaping from the top of the cup and perceived by consumers. Immediately after EC preparation, the sampling vessel was hermetically sealed using the ground glass lid and placed in a thermostatic bath at 60 °C. Then, a 60 ml syringe fitted with a Tenax cartridge trap for TDU sampling (Thermal Desorption Unit, manufactured by Gerstel, Germany), was plugged to one of the lid valves, and 50 ml of head space vapors were aspirated for 30 s allowing
the absorption of volatile compounds in the vessel headspace. During this step the second valve was opened, allowing air to flow into the vessel in order to offset the outflowing headspace gases. The cartridges were then stored in a sealed plastic tube and analyzed the same day using the thermal desorption unit/gas chromatography/mass spectrometry (TDU/GC/MS) technique. The Tenax tubes were placed in a sampling system (Gerstel MPS2XL) equipped with a TDU mounted over a Programmable Temperature Vaporizer (PTV) injector (MOD CIS4, cooled with liquid carbon dioxide). The TDU temperature was ramped to 300 °C after cooling the CIS4 at 40 °C allowing the compound cryofocalization in the PTV liner. Then the temperature was increased to 300 °C at a rate of 12 °C s 1. The GC/MS analysis was performed with an Agilent 7890 gas chromatograph and a 5975C selective mass detector. A J&W INNOVAX (30 m, 0.25 mm id, 0.5 lm df) column was used for compound separation. An initial compound identification was carried out by comparing the mass spectra of the separated compounds and their retention indices with those reported in Nist05 and Wyley07 spectral libraries following dynamic background compensation with Target view software (ALMSCO, United Kingdom). 2.5. Statistical analysis Principal component analysis (PCA) and linear discriminant analysis (LDA) were applied to identify patterns and structures in the analyzed data. Moreover, conventional variance analysis (ANOVA) was applied to evaluate the effect of the different extraction methods on the selected compounds. Where the univariate F-test was significant at the p < 0.05 level, multiple comparison tests of pairs of means were checked for significance using the Tukey Honest Significance Difference (HSD) post hoc test (p < 0.05). 3. Results and discussion The chemical characterization of ECs is shown in Table 2, which compares the three preparation methods. The infusion temperature is generally recognized as an important factor in EC preparation. It is generally related to extraction parameters (namely total solids, concentration and extraction%), and lower temperatures usually lead to less extracted EC (Andueza et al., 2003). In this experiment the input water temperature was the same in all tests, while the output temperature of the percolated coffee flow (measured immediately after extraction) was significantly higher in the HIP system than the other two methods. This result could be explained by better thermal insulation of the infusion chamber (i.e. the capsule main body) provided by the HIP apparatus. Consistent with output temperature, ECs prepared using the HIP system showed significantly higher values for extraction parameters (total solids, extraction%, concentration; see Table 2) than the other two systems. The trend was consistent; the BM, which had the lowest output temperature corresponded to the lowest values for concentration, extraction% and total solids. For the capsule systems, our results are higher than those found in the literature. For example, Illy et al. (2005) proposed average values of 20–60 mg ml 1, while for extraction%, Lingle (1996) proposed a threshold of 22% as the limit above which tasters perceive an unpleasant astringency and excess bitterness. Among the methods tested here, only the BM was consistent with these results (Illy et al., 2005; Lingle, 1996). Another sensorial attribute widely recognized as important to the taste of EC is sourness (or acidity). The measured pH values ranged from 5.2 to 5.8, consistent with those found in the literature. EC prepared using the BM method had a higher pH than that obtained using capsules, but this resulted in an unusually high perception of acidity, as other work has found only a loose correlation
115
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
Table 2 Average value ± standard deviation, and coefficient of variation of measured physico-chemical parameters. Different letters indicate statistically significant differences (p > 0.05 Tukey’s test). BM Output temperature (°C) pH Density at 20 °C (g ml 1) Viscosity at 20 °C (mN s m 2) Refractive index (nD) Foam index (%) Foam persistency (min) Total solids (mg ml 1) Extraction (%) Concentration (%) Total lipids (mg ml 1) Caffeine (mg ml 1) Trigonelline (mg ml 1) 4-Caffeoilquinic acid (mg ml 1) 3-Caffeoilquinic + 5-caffeoilquinic acids (mg ml 3,4-Dicaffeoilquinic acid (mg ml 1) 4,5-Dicaffeoilquinic acid (mg ml 1) Total caffeoilquinic acids (mg ml 1) Total dicaffeoilquinic acids (mg ml 1)
1
)
HIP
Average values ± sd 79.6 ± 3.0 b 5.32 ± 0.09 a 1.029 ± 0.016 a 1.36 ± 0.05 a 1.355 ± 0.022 a 32.4 ± 7.3 b 3.88 ± 0.93 b 59.48 ± 1.78 c 21.1 ± 0.6 b 5.9 ± 0.2 c 2.66 ± 0.75 b 2.22 ± 0.55 a 1.32 ± 0.27 a 1.05 ± 0.35 a 2.27 ± 0.72 a 0.14 ± 0.04 a 0.10 ± 0.04 a 3.77 ± 1.05 a 0.23 ± 0.07 a
between pH and perceived acidity (Clarke and Vitzthum, 2001). According to Bähre (1996) perceived acidity mainly depends upon total acidity, which is determined by chlorogenic acids and caffeine. These latter compounds also contribute to astringency and bitterness, thus modifying the typical bitterness-acidity balance of EC. As our tests of the three systems showed no significant difference in the concentration of these compounds it may be assumed that only small differences could be perceived on the sourness profile. The density and viscosity of the EC was also measured. The density of EC prepared using the IT system was significantly lower (less body) than EC prepared using the BM system, while the BM and HIP system did not show a significant difference. Similarly, no significant difference was found in viscosity, although there were some significant differences in total lipid concentration. These two parameters, which affect perceived creaminess, are generally related to each other. Raising the total lipid concentration results in an increase in viscosity (on a logarithmic scale; Kilcast and Clegg, 2002). In the present experiment only total lipids were significantly affected, with higher concentrations found in the capsule systems (IT and HIP) and lower concentrations in the traditional system (BM). It is also known that lipid concentration strongly affects the foam phase. In this case the HIP system, which is designed to enhance this parameter, shows a foam index considerably higher than the reference value of 10% (Illy et al., 2005) and very high persistency (about 234 min on average). Both of these parameters are significantly higher for the HIP system than the IT or BM system. There were no significant differences between the IT and BM systems despite differences in lipid content. Although the BM and IT systems had high foam indices, persistency was very short
87.6 ± 1.8 a 5.15 ± 0.10 b 1.012 ± 0.014 ab 1.40 ± 0.04 a 1.331 ± 0.019 ab 61.3 ± 10.1 a 234.27 ± 9.34 a 69.69 ± 1.78 a 25.1 ± 0.7 a 7.0 ± 0.2 a 4.49 ± 0.73 a 2.31 ± 0.19 a 1.22 ± 0.12 a 0.96 ± 0.08 a 2.59 ± 0.22 a 0.11 ± 0.01 ab 0.09 ± 0.01 a 3.54 ± 0.29 a 0.20 ± 0.02 a
IT
BM
HIP
IT
82.5 ± 4.3 b 5.05 ± 0.06 b 0.991 ± 0.018 b 1.33 ± 0.06 a 1.303 ± 0.023 b 39.7 ± 7.7 b 2.52 ± 0.49 b 66.14 ± 1.83 b 24.4 ± 0.9 a 6.6 ± 0.2 b 3.94 ± 0.79 a 2.14 ± 0.38 a 1.23 ± 0.13 a 0.93 ± 0.11 a 2.50 ± 0.24 a 0.10 ± 0.01 b 0.08 ± 0.01 a 3.43 ± 0.35 a 0.18 ± 0.04 a
CV (%) 3.72 1.62 1.56 3.89 1.62 22.44 23.96 2.99 2.69 2.99 28.18 24.99 20.73 33.42 26.43 31.05 38.37 27.78 31.95
2.04 1.98 1.38 3.14 1.43 16.5 3.99 2.55 2.85 2.55 16.21 8.17 9.8 8.27 8.31 8.89 10.95 8.25 9.32
5.16 1.19 1.82 4.44 1.76 19.52 19.44 2.77 3.7 2.77 19.94 17.6 10.86 11.28 9.75 12.47 15.14 10.05 22.22
(about 2 min), which is close to the threshold of acceptability (Illy et al., 2005). The aromatic profile was evaluated on the basis of the key compounds identified in previous work (Grosch, 1998; Maeztu et al., 2001b). Eight of these key compounds showed peak areas with significant differences between preparation methods (Table 3). The main differences in aromatic profile related to the BM system, which showed higher amounts of all eight key compounds, indicating a more marked profile than the other preparation methods. Among the key compounds, 2-methylbutanal and 3methylbutanal along with pyrazines are responsible for the malted and toasted aroma (Semmelroch and Grosch, 1995). Particularly, the three key odorants belonging to the class of pyrazines (2ethyl-3,5-dimethyilpyrazine, 2-ethyilpyrazine and 2-ethyl-6methylpyrazine) are linked to a toasted and burned flavor (Maarse and Visscher, 1996; Maeztu et al., 2001a). The BM method resulted in significantly higher amounts of these compounds than the other methods. Diacetyl and 2,3-pentanedione are often associated with a buttery taste (Blank et al., 1991; Grosch, 1998) and, once more, the BM method resulted in EC with the highest concentration. A statistically significant difference was also found for diacetyl between the two capsule methods, where IT showed higher levels than HIP. The guaiacol peak area (smoke note) was higher in the BM and HIP methods than IT. Thus, based on these key compounds, the EC made with the BM seems to have a more intense flavor than the capsule systems, while significant differences were found between the latter in two of the nine compounds examined. To confirm this finding, the comparison was extended to all identified volatiles. As a general rule, the trend observed for key compounds held. For almost all the chemical compounds examined, the BM
Table 3 Peak area ± standard deviation 106 of EC key compounds that show significant differences between methods. Different letters indicate statistically significant differences (p > 0.05 Tukey’s test). Key compounds
Bar
Hyper Espresso
I-Espresso System
2-Methylbutanal + 3-methylbutanal Diacetyl 2,3-Pentanedione 2-Ethylpyrazine 2-Ethyl-6-methylpyrazine 2-Ethyl-3,5-dimethylpyrazine Guaiacol
60.09 ± 12.34 a 143.04 ± 29.8 a 97.46 ± 18.18 a 51.93 ± 6.32 a 43.82 ± 5.27 a 14.48 ± 1.83 a 6.19 ± 0.73 a
42.20 ± 5.21 82.3 ± 9.34 56.34 ± 7.01 36.41 ± 7.18 31.74 ± 6.35 11.42 ± 2.48 5.45 ± 0.96
40.93 ± 5.81 b 109.53 ± 17.43 b 68.58 ± 9.66 b 33.00 ± 4.41 b 25.55 ± 3.85 b 8.72 ± 1.79 b 3.85 ± 0.61 b
b c b b b b a
116
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
Fig. 1. The graph on the left represents the dispersion plot of correlation coefficients among principal components and variables. The graph on the right is the dispersion plot of factorial scores on the first and second axes.
method had significantly higher peak area values than HIP and IT (the data is not included here). However, in view of these results, some crucial points should be taken into account. Firstly, according to Dold et al. (2011), the foamy layer that lies on the liquid exerts a selective barrier against escaping volatiles. Hence, it is possible that wide differences in foam thickness could modify the amount of volatile compounds in the cup headspace. From this point of view, BM has a very low foam thickness compared to the other two systems and, consistently, the highest amount of volatile compounds was detected. Consequently, the chosen sampling method for determining volatile compounds could have exaggerated the differences between the profiles (especially between BM and HIP), although it should be noted that this method may well mimic consumer perceptions during smelling. Furthermore, a very large amount of volatile compounds does not always imply greater quality or sensorial acceptance. The aromatic profile is the result of a complex balance between volatiles and the compounds discussed here cannot entirely describe the perceived aroma. It is possible that the abundance of some compounds could even result in an unpleasant burned taste and/or a so-called over-extracted EC. Finally, consideration must be given to the reproducibility of the three EC preparation methods. Patents concerning EC preparation using capsules or cartridges sometimes cite the consistency of extraction conditions as a particular feature of such preparation methods, but no estimate of this parameter has been found in the literature. The results of the present experiment (as reported in Table 2), shows dissimilarities between the preparation methods in terms of variability in parameters. The three preparation methods can be easily compared by computing variation coefficients (CV%) for the different parameters (Table 2). The capsule systems show noticeably lower CV% compared to the BM method in 17 and 14 out of the 19 physico-chemical parameters for HIP and IT, with the only exceptions being pH and extraction% for HIP. This could mean that these methods produced more consistent EC than the conventional BM method, where the uncertainty due to the ability of the operator is greater. A principal component analysis (PCA) summarized the differences related to the physical parameters and aromatic profile of the three extraction techniques (Fig. 1). As much as 74.1% of the total variance is explained by the first (54.5%) and second (19.6%) components. Multivariate analysis was able to separate the three machines, showing that different extraction techniques produce different EC, which highlights the importance and effectiveness
Table 4 Results of classification of brewing technique using linear discriminant analysis. Predicted Real
BM
HIP
IT
BM HIP IP
8 0 0
0 9 0
1 0 9
of improvements to these devices. To confirm the class separation of the brewing methods, data was analyzed using linear discriminant analysis (LDA) and the classification model was validated using a leave-one-out cross-validation, which showed a 96.3% correct classification (Table 4).
4. Conclusion The comparison of these three methods provides information about different EC characteristics obtained by different brewing techniques. The three techniques produce different EC from the same ground coffee. The EC produced with the BM method was different to the others in terms of its physico-chemical characteristics and higher levels of volatile compounds detected directly above the cup. Moreover, differences were found between the two capsule systems. EC produced with the HIP method has thicker foam and more total extracted solids than that produced using the IT method. Both capsule methods make the preparation of EC easier, through the reduction of the number of variables involved in the brewing process and lead to a more consistent brew than the bar machine. Acknowledgements The authors would like to express their gratitude to Why? S.R.L. and Francesco Illy, for sponsoring this research. References Andueza, S., Maeztu, L., Dean, B., de Peña, M.P., Bello, J., Cid, C., 2002. Influence of water pressure on the final quality of arabica espresso coffee. Application of multivariate analysis. Journal of Agricultural and Food Chemistry 50 (25), 7426– 7431.
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117 Andueza, S., Maeztu, L., Pascual, L., Ibanez, C., de Pena, M.P., Cid, C., 2003. Influence of extraction temperature on the final quality of espresso coffee. Journal of the Science of Food and Agriculture 83 (3), 240–248. Bähre, F., 1996. Electrophoretic clean-up of organic acids from coffee for the GC/MS analysis. Fresenius’ Journal of Analytical Chemistry 355 (2), 190–193. Blank, I., Sen, A., Grosch, W., 1991. Aroma impact compounds of Arabica and Robusta coffee. Qualitative and Quantitative investigations. In: ASIC 14th Colloque, San Francisco, CA, pp. 117–129. Brommer, E., Stratmann, B., Quack, D., 2011. Environmental impacts of different methods of coffee preparation. International Journal of Consumer Studies 35 (2), 212–220. Clarke, R.J., Vitzthum, O.G., 2001. Coffee Recent Development. Blackwell Science, p. 266. Dold, S., Lindinger, C., Kolodziejczyk, E., Pollien, P., Ali, S., Germain, J.C., Perin, S.G., Pineau, N., Folmer, B., Engel, K.H., Barron, D., Hartmann, C., 2011. Influence of foam structure on the release kinetics of volatiles from espresso coffee prior to consumption. Journal of Agricultural and Food Chemistry 59 (20), 11196– 11203. Grosch, W., 1998. Flavour of coffee. A review. Nahrung 42 (6). Illy, A., Viani, R., Suggi Liverani, F., 2005. Espresso Coffee: The Science of Quality. Elsevier Academic Press, pp. 16–19. Kilcast, D., Clegg, S., 2002. Sensory perception of creaminess and its relationship with food structure. Food Quality and Preference 13, 609–623. Lingle, T.R., 1996. Coffee brewing control chart. In: Lingle, T.R. (Ed.), The Coffee Brewing Handbook. A Systematic Guide to Coffee Preparation. Speciality Coffee Association of America, Long Beach, CA, pp. 3–5.
117
Maarse, A., Visscher C.A., 1996. Volatile Compounds in Food. Qualitative and Quantitative Data. TNO Food Analysis Institute, The Nederlands: 7th Zeist. Maeztu, L., Andueza, S., Ibañez, C., Paz de Peña, M., Bello, J., Cid, C., 2001a. Multivariate methods for characterization and classification of espresso coffees from different botanical varieties and types of roast by foam, taste, and mouthfeel. Journal of Agricultural and Food Chemistry 49 (10), 4743– 4747. Maeztu, L., Sanz, C., Andueza, S., De Peña, M.P., Bello, J., Cid, C., 2001b. Characterization of espresso coffee aroma by static headspace GC–MS and sensory flavor profile. Journal of Agricultural and Food Chemistry 49 (11), 5437–5444. Navarini, L., Colomban, S., Lonzarich, V., Rivetti, D., Brollo G., Suggi-Liverani F., 2008. In: Proceeding 22th colloquie ASIC. Navarini, L., Rivetti, D., 2010. Water quality for espresso coffee. Food Chemistry. Elsevier Ltd., vol. 122(2), pp. 424–428. Petracco, M., 2001. Technology IV: beverage preparation: brewing trends for the new millennium. In: Coffee Recent Development Ed by ClarkeRJ and VitzthumOG, Blackwell Science, Oxford, pp. 140–164. Semmelroch, P., Grosch, W., 1995. Analysis of roasted coffee powders and brews by gas chromatography – olfactometry of headspace samples. LebensmittelWissenschaft und -Technologie 28 (2), 310–313. Tozzi, F., 2007. Dalla moka alla cialda. Largo Consumo 7, 26–28. Vanni, A., 2009. Sealed capsule for the preparation of a beverage. In particular espresso coffee European Patent 1 886 942 b1 (12).
Contents lists available at ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
Comparison of espresso coffee brewing techniques Alessandro Parenti a,⇑, Lorenzo Guerrini a, Piernicola Masella a, Silvia Spinelli a, Luca Calamai b,c, Paolo Spugnoli a a
Dipartimento di Gestione dei Sistemi Agrari, Alimentari e Forestali (GESAAF), Università degli Studi di Firenze, Piazzale delle Cascine 15, 50144 Firenze, Italy Dipartimento di Scienze delle Produzioni Agroalimentari e dell’Ambiente (DISPAA), Università degli Studi di Firenze, Piazzale delle Cascine 18, 50144 Firenze, Italy c Centro di servizi di Spettrometria di Massa (CISM), Università degli Studi di Firenze, Via U. Schiff 6, 50019 Sesto Fiorentino (FI), Italy b
a r t i c l e
i n f o
Article history: Received 26 March 2013 Received in revised form 27 June 2013 Accepted 12 August 2013 Available online 28 August 2013 Keywords: Coffee machines Capsule Foam Espresso reproducibility
a b s t r a c t Several brewing techniques are used to make espresso coffee. Among them, the most widespread are bar machines and single-dose capsules, designed in large numbers because of their commercial popularity. As none of the current literature compares the effects of these different brewing techniques on espresso quality, this paper looks at two capsule methods and the traditional bar method. The methods were evaluated on the basis of the physico-chemical parameters and aromatic profile of nine espresso coffees prepared using the different techniques. Our results showed that with the same batch of roasted coffee, the same water and the same operative settings, the three different techniques can be distinguished by a principal component analysis. Furthermore, in terms of product reproducibility, the best results are provided by the two capsule systems. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction There are a large number of devices and methods to produce espresso coffee (EC). EC is defined as ‘‘a brew obtained by percolation of hot water under pressure through compacted cake of roasted ground coffee, where the energy of the water pressure is spent within the cake’’ (Illy et al., 2005). In this process flavors are extracted from the coffee by means of hot water. EC consists of about 35 ml of dark beverage, usually served in a small cup with a brown foam layer called crema covering the liquid. Crema is a distinctive feature of EC, as it is absent in other coffee brews and is required for consumer acceptance. EC is conventionally brewed using bar machines (BM), which consist of a rotating pump, a heat exchanger and an extraction chamber (Illy et al., 2005). The water pressure provided by the pump strongly affects the physical and sensory properties of the brew (Andueza et al., 2002) and maximal EC quality seems to correspond to an optimal water pressure of nine bars. ECs prepared at higher pressure have negative sensorial qualities as they are excessively bitter, astringent and contain more key odorants. In conventional EC preparation pressurized water reaches the heat exchanger where its temperature rises. In this type of machine Andueza et al. (2003) found the best key odorant profile, flavor notes and highest overall acceptability at 92 °C. Coffees brewed at lower temperatures had less odor, flavor and body intensity
⇑ Corresponding author. Tel.: +39 553288319; fax: +39 55 3288316. E-mail address: alessandro.parenti@unifi.it (A. Parenti). 0260-8774/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jfoodeng.2013.08.031
(they are generally called ‘‘under extracted’’), whereas coffees brewed at higher temperatures had a burnt/roast flavor and a higher content of negative key odorants (they are generally called ‘‘over extracted’’). The extraction chamber consists of a filter where the ground coffee is placed and compacted. The filter provides the hydraulic resistance required to produce EC. During this procedure there are many variables that cannot be controlled by the extraction device (e.g. the coffee powder particle size, powder compression); these have a high impact on the properties of the final brew (Illy et al., 2005) and depend on the ability of the barman. Among the other methods developed to make EC, pod and capsule systems have recently gained market share because they are user-friendly. They also make it easy to prepare good-quality coffees through the reduction of uncontrolled preparation variables. Furthermore, these systems preserve the quality of the ground coffee by protecting it against moisture and oxidation processes (Vanni, 2009). For these reasons, in 2005, pod and capsule sales added up to 14 billion units (Tozzi, 2007). As a result of their popularity several kinds of capsule have been developed. The simplest consists of a chamber to hold the coffee and a film that provides the needed resistance when water is added. More sophisticated capsules are equipped with devices that should lead to the production of top-quality EC. However, the environmental impact of these approaches is significantly higher than other preparation methods. This is mainly due to the production of disposable capsules that cause significant greenhouse gas emissions (Brommer et al., 2011). In fine EC quality is strongly affected by the operative conditions of the extraction, which differ depending on the device.
113
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
To the best of the authors’ knowledge, there are no studies that compare different preparation techniques (e.g. BM and capsule) in terms of the sensory and chemical characteristics of the resulting EC. In this work we evaluate and compare the differences in terms of quality between EC made using three different extraction procedures. The differences are assessed in terms of sensory characteristics, physical parameters, and extracted volatile key compounds. The selected methods were: the bar method (the traditional way to make EC), a simple commercial capsule method (Illy; I-Espresso System), and a more advanced commercial capsule method (Illy; HyperEspresso), primarily designed to increase the sensory attributes related to the colloidal state of the beverage (Navarini et al., 2008). 2. Materials and methods 2.1. Experimental design The extraction methods were compared through the preparation of three ECs per day over a period of 3 days for each device, resulting in a total of 27 ECs. The order of preparation for each day was completely random. 2.2. Extraction devices 2.2.1. Bar machine (BM) A conventional bar machine (the Alina model manufactured by Cimbali S.p.A. Italy) was used. The machine was designed to make two ECs at the same time in the same extraction chamber by splitting the exit flow equally. Therefore, the ECs were prepared with double the amount of ground coffee (14.5 ± 0.2 g). Chemical and physical analyses were only performed on one of the two ECs. The extraction parameters were: water temperature 92 °C, water pressure 9 bar, and 25–30 s of percolation time, assuming an optimal flow rate of about 1 ml s 1 (Illy et al., 2005). 2.2.2. Hyper Espresso method (HIP) The capsules were brewed using the Good News coffee machine (manufactured by D.P.I. Service SNC, Italy), designed for Hyper Espresso capsules (produced by illycaffè S.p.A, Italy) at a pressure of 12 bar and an extraction temperature of 92 °C. The HIP capsules contained 6.7 ± 0.1 g of ground coffee and consisted of five parts: a cover, an upper and a lower internal filter, an infusion chamber and a flow conveyor. This design only allows the EC to flow out of the capsule when a fixed pressure is reached. During EC preparation the upper side of the capsule (the cover) is punched and the water is added (pre-infusion phase); then the water compresses the gas in the capsule and the pressure rises (infusion phase) until the capsule film bursts; once the pressure to burst the capsule is reached, the coffee flows out through a micro-hole (emulsion phase). 2.2.3. I-Espresso System (IT) The capsules were brewed using the Mitaca machine (manufactured by illycaffè S.p.A, Italy). The capsules contained 6.9 ± 0.1 g of ground coffee and consisted of a plastic cylinder covered by a plastic film. Hot water at 92 °C is introduced into the capsule. The bottom of the capsule has a central hole allowing EC outflow when a given pressure is reached.
Table 1 The physico-chemical characteristics of mineral water as listed on the bottle’s label. Analytical parameter
Values
pH Electrical conductivity (20 °C) Total dissolved solids Hardness Kubel oxydability Free carbon dioxide Calcium (Ca2+) Magnesium (Mg2+) Sodium (Na+) Potassium (K+) Hydrogen carbonate (HCO3 )
8.1 249 lS/cm 148 mg/l 14 °F 0.6 mg/l 3.3 mg/l 30.1 mg/l 15.0 mg/l 1.4 mg/l 0.5 mg/l 157 mg/l 10.7 mg/l
Sulfate (SO24 ) Nitrate (NO3 ) Chloride (Cl ) Fluoride (F ) Silicon dioxide (SiO2)
5.0 mg/l 1.5 mg/l 0.06 mg/l 6.6 mg/l
the HIP and IT capsules. The remaining roasted beans were used for the BM trials. These beans were ground immediately before preparation, using a professional coffee grinder (KE640 model manufactured by Ditting Maschinen AG, Switzerland). The resulting particle size distribution was: 29% >500 lm; 250 lm < 47.4% < 500 lm; 125 lm < 22.2% < 250 lm; and 1.4% < 125 lm. 2.3.2. Water According to Navarini and Rivetti (2010), water quality plays a key role in EC quality. Consequently, all tests were performed using the same commercial brand of mineral water. The physical and chemical characteristics of this water, according to the manufacturer’s specification, are shown in Table 1. 2.4. Measurements and analyses All brewed coffee samples were immediately collected at the outflow of the machine in a glass weighing bottle (75 ml volume, 53 mm internal diameter, 34 mm high) with a ground glass lid and equipped with two valves specifically designed for the sampling of the headspace above the coffee (described in more detail later). In order to obtain homogeneous samples, the same weight of percolated liquid was collected, regardless of flow rate or percolation time. Thus, a digital scale (max capacity 300.0 g; precision 0.1 g; manufactured by D-Mail S.R.L., Italy) was placed under the vessel and a preselected weight of 25 g of brewed coffee was collected. The resulting final brew weight was 25.7 ± 0.6 g averaged over all the samples. The temperature of the outflowing coffee was measured directly under the liquid flow with a digital thermometer (HD2107.1, manufactured by Delta OHM S.R.L., Italy). The following parameters were analyzed and evaluated for all samples. 2.4.1. Foam Index and persistency The foam index is defined as the ratio between the foam and liquid volume (vol vol 1%) measured 30 s after extraction (the geometry of the sampling vessel is given above). Persistency is defined as the time (in minutes) before the foam breaks up, leaving an uncovered black spot on the surface of the beverage (Petracco, 2001).
2.3. Espresso coffee preparation 2.3.1. Coffee All the ECs were prepared from the same batch of roasted coffee beans, provided by illycaffè S.p.A. (Italy). Some of the roasted beans were ground (Colombini Icoperfexand grinder) and used to prepare
2.4.2. Density, pH, and viscosity Before taking these measurements samples, were cooled to 20 °C. Density was measured with a 25 ml pycnometer. A digital pH meter (GLP 21, manufactured by CRISON INSTRUMENTS, S.A. Spain) was used to determine the pH of the ECs. Viscosity was
114
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
measured with a capillary viscometer (Ostwald) fitted with an automatic optical reader (Viscoclock, manufactured by SCHOTT Instruments GmbH, Germany) and expressed as mN s m 2. 2.4.3. Refractive index The refractive index was measured with a portable digital refractometer (Refracto 30PX, manufactured by Mettler Toledo S.p.A., Italy), using the total internal reflection method, a light source wavelength of k = 589.3 nm and automatic temperature compensation set at 20 °C. 2.4.4. Total solids, extraction, concentration Total solids (expressed as mg ml 1) were determined by ovendrying 15.4 ± 1.4 ml of EC until a constant weight was reached (18 h, 104 ± 2 °C). The extraction percentage was defined as the percentage by weight of total solids with respect to the dosage of ground roast coffee. Concentration was defined as percentage ratio between total solids and the volume of EC (w vol 1%). 2.4.5. Lipids The total lipid concentration (mg ml 1) was determined by liquid–liquid extraction using n-hexane as the solvent. About 10 ml of EC was extracted using 5 ml of n-hexane three times and centrifuged at 3500 rpm for 3 min. Each time the hexane phase was pipetted out and placed in a crystallizer. The total weight of lipids was determined after the solvent had completely evaporated. 2.4.6. Caffeine, trigonelline and chlorogenic acids Extraction, sample preparation and high-performance liquid chromatography (HPLC) conditions were carried out as described by Maeztu et al. (2001b). The HPLC system consisted of an Agilent 1200 Rapid Resolution unit equipped with a diode array detector and fitted to an Agilent 6410 QQQ mass spectrometer operating in scan mode. An alternating positive and negative ionization switching (every 500 ms) was set up for optimal ionization of basic (caffeine and trigonelline) and acidic (chlorogenic acids) molecules. A Varian (2 mm id, 15 cm length, particle size 3 lm, porosity 100 Å, temperature 60 °C) Polaris HPLC column was used for the separation of compounds. The mobile phases were: water with 0.1% formic acid and acetonitrile. The flow rate was constant at 0.4 ml/min and the eluent composition was water/acetonitrile (95:5 v/v) for 1 min, gradient 5–60% of acetonitrile for 20 min, gradient 60–90% of acetonitrile for 3 min, water/acetonitrile (10:90 v/ v) for 1 min, gradient 90–5% of acetonitrile for 2 min, and 10 min equilibration at the initial conditions. Quantitative data was obtained through the construction of calibration lines with authentic standards both in UV-DAD and MS chromatograms in the 1–20 mg ml 1 range. Under analysis conditions the retention times for trigonelline, caffeine, 4-caffeoilquinic acid, and 3,4-dicaffeoilquinic acid were respectively 1.4, 8.8, 5.8, and 14.5 min. 2.4.7. Volatile compound determination It is known that perceived EC aroma is only partially related to the total amount of aromatic compounds extracted during preparation. It also depends upon conditions such as the temperature of the EC, and the presence/absence of foam. Consequently, a new method was developed for the evaluation of the volatile compounds escaping from the top of the cup and perceived by consumers. Immediately after EC preparation, the sampling vessel was hermetically sealed using the ground glass lid and placed in a thermostatic bath at 60 °C. Then, a 60 ml syringe fitted with a Tenax cartridge trap for TDU sampling (Thermal Desorption Unit, manufactured by Gerstel, Germany), was plugged to one of the lid valves, and 50 ml of head space vapors were aspirated for 30 s allowing
the absorption of volatile compounds in the vessel headspace. During this step the second valve was opened, allowing air to flow into the vessel in order to offset the outflowing headspace gases. The cartridges were then stored in a sealed plastic tube and analyzed the same day using the thermal desorption unit/gas chromatography/mass spectrometry (TDU/GC/MS) technique. The Tenax tubes were placed in a sampling system (Gerstel MPS2XL) equipped with a TDU mounted over a Programmable Temperature Vaporizer (PTV) injector (MOD CIS4, cooled with liquid carbon dioxide). The TDU temperature was ramped to 300 °C after cooling the CIS4 at 40 °C allowing the compound cryofocalization in the PTV liner. Then the temperature was increased to 300 °C at a rate of 12 °C s 1. The GC/MS analysis was performed with an Agilent 7890 gas chromatograph and a 5975C selective mass detector. A J&W INNOVAX (30 m, 0.25 mm id, 0.5 lm df) column was used for compound separation. An initial compound identification was carried out by comparing the mass spectra of the separated compounds and their retention indices with those reported in Nist05 and Wyley07 spectral libraries following dynamic background compensation with Target view software (ALMSCO, United Kingdom). 2.5. Statistical analysis Principal component analysis (PCA) and linear discriminant analysis (LDA) were applied to identify patterns and structures in the analyzed data. Moreover, conventional variance analysis (ANOVA) was applied to evaluate the effect of the different extraction methods on the selected compounds. Where the univariate F-test was significant at the p < 0.05 level, multiple comparison tests of pairs of means were checked for significance using the Tukey Honest Significance Difference (HSD) post hoc test (p < 0.05). 3. Results and discussion The chemical characterization of ECs is shown in Table 2, which compares the three preparation methods. The infusion temperature is generally recognized as an important factor in EC preparation. It is generally related to extraction parameters (namely total solids, concentration and extraction%), and lower temperatures usually lead to less extracted EC (Andueza et al., 2003). In this experiment the input water temperature was the same in all tests, while the output temperature of the percolated coffee flow (measured immediately after extraction) was significantly higher in the HIP system than the other two methods. This result could be explained by better thermal insulation of the infusion chamber (i.e. the capsule main body) provided by the HIP apparatus. Consistent with output temperature, ECs prepared using the HIP system showed significantly higher values for extraction parameters (total solids, extraction%, concentration; see Table 2) than the other two systems. The trend was consistent; the BM, which had the lowest output temperature corresponded to the lowest values for concentration, extraction% and total solids. For the capsule systems, our results are higher than those found in the literature. For example, Illy et al. (2005) proposed average values of 20–60 mg ml 1, while for extraction%, Lingle (1996) proposed a threshold of 22% as the limit above which tasters perceive an unpleasant astringency and excess bitterness. Among the methods tested here, only the BM was consistent with these results (Illy et al., 2005; Lingle, 1996). Another sensorial attribute widely recognized as important to the taste of EC is sourness (or acidity). The measured pH values ranged from 5.2 to 5.8, consistent with those found in the literature. EC prepared using the BM method had a higher pH than that obtained using capsules, but this resulted in an unusually high perception of acidity, as other work has found only a loose correlation
115
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
Table 2 Average value ± standard deviation, and coefficient of variation of measured physico-chemical parameters. Different letters indicate statistically significant differences (p > 0.05 Tukey’s test). BM Output temperature (°C) pH Density at 20 °C (g ml 1) Viscosity at 20 °C (mN s m 2) Refractive index (nD) Foam index (%) Foam persistency (min) Total solids (mg ml 1) Extraction (%) Concentration (%) Total lipids (mg ml 1) Caffeine (mg ml 1) Trigonelline (mg ml 1) 4-Caffeoilquinic acid (mg ml 1) 3-Caffeoilquinic + 5-caffeoilquinic acids (mg ml 3,4-Dicaffeoilquinic acid (mg ml 1) 4,5-Dicaffeoilquinic acid (mg ml 1) Total caffeoilquinic acids (mg ml 1) Total dicaffeoilquinic acids (mg ml 1)
1
)
HIP
Average values ± sd 79.6 ± 3.0 b 5.32 ± 0.09 a 1.029 ± 0.016 a 1.36 ± 0.05 a 1.355 ± 0.022 a 32.4 ± 7.3 b 3.88 ± 0.93 b 59.48 ± 1.78 c 21.1 ± 0.6 b 5.9 ± 0.2 c 2.66 ± 0.75 b 2.22 ± 0.55 a 1.32 ± 0.27 a 1.05 ± 0.35 a 2.27 ± 0.72 a 0.14 ± 0.04 a 0.10 ± 0.04 a 3.77 ± 1.05 a 0.23 ± 0.07 a
between pH and perceived acidity (Clarke and Vitzthum, 2001). According to Bähre (1996) perceived acidity mainly depends upon total acidity, which is determined by chlorogenic acids and caffeine. These latter compounds also contribute to astringency and bitterness, thus modifying the typical bitterness-acidity balance of EC. As our tests of the three systems showed no significant difference in the concentration of these compounds it may be assumed that only small differences could be perceived on the sourness profile. The density and viscosity of the EC was also measured. The density of EC prepared using the IT system was significantly lower (less body) than EC prepared using the BM system, while the BM and HIP system did not show a significant difference. Similarly, no significant difference was found in viscosity, although there were some significant differences in total lipid concentration. These two parameters, which affect perceived creaminess, are generally related to each other. Raising the total lipid concentration results in an increase in viscosity (on a logarithmic scale; Kilcast and Clegg, 2002). In the present experiment only total lipids were significantly affected, with higher concentrations found in the capsule systems (IT and HIP) and lower concentrations in the traditional system (BM). It is also known that lipid concentration strongly affects the foam phase. In this case the HIP system, which is designed to enhance this parameter, shows a foam index considerably higher than the reference value of 10% (Illy et al., 2005) and very high persistency (about 234 min on average). Both of these parameters are significantly higher for the HIP system than the IT or BM system. There were no significant differences between the IT and BM systems despite differences in lipid content. Although the BM and IT systems had high foam indices, persistency was very short
87.6 ± 1.8 a 5.15 ± 0.10 b 1.012 ± 0.014 ab 1.40 ± 0.04 a 1.331 ± 0.019 ab 61.3 ± 10.1 a 234.27 ± 9.34 a 69.69 ± 1.78 a 25.1 ± 0.7 a 7.0 ± 0.2 a 4.49 ± 0.73 a 2.31 ± 0.19 a 1.22 ± 0.12 a 0.96 ± 0.08 a 2.59 ± 0.22 a 0.11 ± 0.01 ab 0.09 ± 0.01 a 3.54 ± 0.29 a 0.20 ± 0.02 a
IT
BM
HIP
IT
82.5 ± 4.3 b 5.05 ± 0.06 b 0.991 ± 0.018 b 1.33 ± 0.06 a 1.303 ± 0.023 b 39.7 ± 7.7 b 2.52 ± 0.49 b 66.14 ± 1.83 b 24.4 ± 0.9 a 6.6 ± 0.2 b 3.94 ± 0.79 a 2.14 ± 0.38 a 1.23 ± 0.13 a 0.93 ± 0.11 a 2.50 ± 0.24 a 0.10 ± 0.01 b 0.08 ± 0.01 a 3.43 ± 0.35 a 0.18 ± 0.04 a
CV (%) 3.72 1.62 1.56 3.89 1.62 22.44 23.96 2.99 2.69 2.99 28.18 24.99 20.73 33.42 26.43 31.05 38.37 27.78 31.95
2.04 1.98 1.38 3.14 1.43 16.5 3.99 2.55 2.85 2.55 16.21 8.17 9.8 8.27 8.31 8.89 10.95 8.25 9.32
5.16 1.19 1.82 4.44 1.76 19.52 19.44 2.77 3.7 2.77 19.94 17.6 10.86 11.28 9.75 12.47 15.14 10.05 22.22
(about 2 min), which is close to the threshold of acceptability (Illy et al., 2005). The aromatic profile was evaluated on the basis of the key compounds identified in previous work (Grosch, 1998; Maeztu et al., 2001b). Eight of these key compounds showed peak areas with significant differences between preparation methods (Table 3). The main differences in aromatic profile related to the BM system, which showed higher amounts of all eight key compounds, indicating a more marked profile than the other preparation methods. Among the key compounds, 2-methylbutanal and 3methylbutanal along with pyrazines are responsible for the malted and toasted aroma (Semmelroch and Grosch, 1995). Particularly, the three key odorants belonging to the class of pyrazines (2ethyl-3,5-dimethyilpyrazine, 2-ethyilpyrazine and 2-ethyl-6methylpyrazine) are linked to a toasted and burned flavor (Maarse and Visscher, 1996; Maeztu et al., 2001a). The BM method resulted in significantly higher amounts of these compounds than the other methods. Diacetyl and 2,3-pentanedione are often associated with a buttery taste (Blank et al., 1991; Grosch, 1998) and, once more, the BM method resulted in EC with the highest concentration. A statistically significant difference was also found for diacetyl between the two capsule methods, where IT showed higher levels than HIP. The guaiacol peak area (smoke note) was higher in the BM and HIP methods than IT. Thus, based on these key compounds, the EC made with the BM seems to have a more intense flavor than the capsule systems, while significant differences were found between the latter in two of the nine compounds examined. To confirm this finding, the comparison was extended to all identified volatiles. As a general rule, the trend observed for key compounds held. For almost all the chemical compounds examined, the BM
Table 3 Peak area ± standard deviation 106 of EC key compounds that show significant differences between methods. Different letters indicate statistically significant differences (p > 0.05 Tukey’s test). Key compounds
Bar
Hyper Espresso
I-Espresso System
2-Methylbutanal + 3-methylbutanal Diacetyl 2,3-Pentanedione 2-Ethylpyrazine 2-Ethyl-6-methylpyrazine 2-Ethyl-3,5-dimethylpyrazine Guaiacol
60.09 ± 12.34 a 143.04 ± 29.8 a 97.46 ± 18.18 a 51.93 ± 6.32 a 43.82 ± 5.27 a 14.48 ± 1.83 a 6.19 ± 0.73 a
42.20 ± 5.21 82.3 ± 9.34 56.34 ± 7.01 36.41 ± 7.18 31.74 ± 6.35 11.42 ± 2.48 5.45 ± 0.96
40.93 ± 5.81 b 109.53 ± 17.43 b 68.58 ± 9.66 b 33.00 ± 4.41 b 25.55 ± 3.85 b 8.72 ± 1.79 b 3.85 ± 0.61 b
b c b b b b a
116
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117
Fig. 1. The graph on the left represents the dispersion plot of correlation coefficients among principal components and variables. The graph on the right is the dispersion plot of factorial scores on the first and second axes.
method had significantly higher peak area values than HIP and IT (the data is not included here). However, in view of these results, some crucial points should be taken into account. Firstly, according to Dold et al. (2011), the foamy layer that lies on the liquid exerts a selective barrier against escaping volatiles. Hence, it is possible that wide differences in foam thickness could modify the amount of volatile compounds in the cup headspace. From this point of view, BM has a very low foam thickness compared to the other two systems and, consistently, the highest amount of volatile compounds was detected. Consequently, the chosen sampling method for determining volatile compounds could have exaggerated the differences between the profiles (especially between BM and HIP), although it should be noted that this method may well mimic consumer perceptions during smelling. Furthermore, a very large amount of volatile compounds does not always imply greater quality or sensorial acceptance. The aromatic profile is the result of a complex balance between volatiles and the compounds discussed here cannot entirely describe the perceived aroma. It is possible that the abundance of some compounds could even result in an unpleasant burned taste and/or a so-called over-extracted EC. Finally, consideration must be given to the reproducibility of the three EC preparation methods. Patents concerning EC preparation using capsules or cartridges sometimes cite the consistency of extraction conditions as a particular feature of such preparation methods, but no estimate of this parameter has been found in the literature. The results of the present experiment (as reported in Table 2), shows dissimilarities between the preparation methods in terms of variability in parameters. The three preparation methods can be easily compared by computing variation coefficients (CV%) for the different parameters (Table 2). The capsule systems show noticeably lower CV% compared to the BM method in 17 and 14 out of the 19 physico-chemical parameters for HIP and IT, with the only exceptions being pH and extraction% for HIP. This could mean that these methods produced more consistent EC than the conventional BM method, where the uncertainty due to the ability of the operator is greater. A principal component analysis (PCA) summarized the differences related to the physical parameters and aromatic profile of the three extraction techniques (Fig. 1). As much as 74.1% of the total variance is explained by the first (54.5%) and second (19.6%) components. Multivariate analysis was able to separate the three machines, showing that different extraction techniques produce different EC, which highlights the importance and effectiveness
Table 4 Results of classification of brewing technique using linear discriminant analysis. Predicted Real
BM
HIP
IT
BM HIP IP
8 0 0
0 9 0
1 0 9
of improvements to these devices. To confirm the class separation of the brewing methods, data was analyzed using linear discriminant analysis (LDA) and the classification model was validated using a leave-one-out cross-validation, which showed a 96.3% correct classification (Table 4).
4. Conclusion The comparison of these three methods provides information about different EC characteristics obtained by different brewing techniques. The three techniques produce different EC from the same ground coffee. The EC produced with the BM method was different to the others in terms of its physico-chemical characteristics and higher levels of volatile compounds detected directly above the cup. Moreover, differences were found between the two capsule systems. EC produced with the HIP method has thicker foam and more total extracted solids than that produced using the IT method. Both capsule methods make the preparation of EC easier, through the reduction of the number of variables involved in the brewing process and lead to a more consistent brew than the bar machine. Acknowledgements The authors would like to express their gratitude to Why? S.R.L. and Francesco Illy, for sponsoring this research. References Andueza, S., Maeztu, L., Dean, B., de Peña, M.P., Bello, J., Cid, C., 2002. Influence of water pressure on the final quality of arabica espresso coffee. Application of multivariate analysis. Journal of Agricultural and Food Chemistry 50 (25), 7426– 7431.
A. Parenti et al. / Journal of Food Engineering 121 (2014) 112–117 Andueza, S., Maeztu, L., Pascual, L., Ibanez, C., de Pena, M.P., Cid, C., 2003. Influence of extraction temperature on the final quality of espresso coffee. Journal of the Science of Food and Agriculture 83 (3), 240–248. Bähre, F., 1996. Electrophoretic clean-up of organic acids from coffee for the GC/MS analysis. Fresenius’ Journal of Analytical Chemistry 355 (2), 190–193. Blank, I., Sen, A., Grosch, W., 1991. Aroma impact compounds of Arabica and Robusta coffee. Qualitative and Quantitative investigations. In: ASIC 14th Colloque, San Francisco, CA, pp. 117–129. Brommer, E., Stratmann, B., Quack, D., 2011. Environmental impacts of different methods of coffee preparation. International Journal of Consumer Studies 35 (2), 212–220. Clarke, R.J., Vitzthum, O.G., 2001. Coffee Recent Development. Blackwell Science, p. 266. Dold, S., Lindinger, C., Kolodziejczyk, E., Pollien, P., Ali, S., Germain, J.C., Perin, S.G., Pineau, N., Folmer, B., Engel, K.H., Barron, D., Hartmann, C., 2011. Influence of foam structure on the release kinetics of volatiles from espresso coffee prior to consumption. Journal of Agricultural and Food Chemistry 59 (20), 11196– 11203. Grosch, W., 1998. Flavour of coffee. A review. Nahrung 42 (6). Illy, A., Viani, R., Suggi Liverani, F., 2005. Espresso Coffee: The Science of Quality. Elsevier Academic Press, pp. 16–19. Kilcast, D., Clegg, S., 2002. Sensory perception of creaminess and its relationship with food structure. Food Quality and Preference 13, 609–623. Lingle, T.R., 1996. Coffee brewing control chart. In: Lingle, T.R. (Ed.), The Coffee Brewing Handbook. A Systematic Guide to Coffee Preparation. Speciality Coffee Association of America, Long Beach, CA, pp. 3–5.
117
Maarse, A., Visscher C.A., 1996. Volatile Compounds in Food. Qualitative and Quantitative Data. TNO Food Analysis Institute, The Nederlands: 7th Zeist. Maeztu, L., Andueza, S., Ibañez, C., Paz de Peña, M., Bello, J., Cid, C., 2001a. Multivariate methods for characterization and classification of espresso coffees from different botanical varieties and types of roast by foam, taste, and mouthfeel. Journal of Agricultural and Food Chemistry 49 (10), 4743– 4747. Maeztu, L., Sanz, C., Andueza, S., De Peña, M.P., Bello, J., Cid, C., 2001b. Characterization of espresso coffee aroma by static headspace GC–MS and sensory flavor profile. Journal of Agricultural and Food Chemistry 49 (11), 5437–5444. Navarini, L., Colomban, S., Lonzarich, V., Rivetti, D., Brollo G., Suggi-Liverani F., 2008. In: Proceeding 22th colloquie ASIC. Navarini, L., Rivetti, D., 2010. Water quality for espresso coffee. Food Chemistry. Elsevier Ltd., vol. 122(2), pp. 424–428. Petracco, M., 2001. Technology IV: beverage preparation: brewing trends for the new millennium. In: Coffee Recent Development Ed by ClarkeRJ and VitzthumOG, Blackwell Science, Oxford, pp. 140–164. Semmelroch, P., Grosch, W., 1995. Analysis of roasted coffee powders and brews by gas chromatography – olfactometry of headspace samples. LebensmittelWissenschaft und -Technologie 28 (2), 310–313. Tozzi, F., 2007. Dalla moka alla cialda. Largo Consumo 7, 26–28. Vanni, A., 2009. Sealed capsule for the preparation of a beverage. In particular espresso coffee European Patent 1 886 942 b1 (12).