Sie sind auf Seite 1von 10

Food Chemistry 234 (2017) 121–130

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Effect of different drying techniques on bioactive components, fatty acid


composition, and volatile profile of robusta coffee beans
Wenjiang Dong, Rongsuo Hu, Zhong Chu, Jianping Zhao, Lehe Tan ⇑
Spice and Beverage Research Institute, Chinese Academy of Tropical Agricultural Sciences (CATAS), Wanning, Hainan 571533, China
National Center of Important Tropical Crops Engineering and Technology Research, Wanning, Hainan 571533, China

a r t i c l e i n f o a b s t r a c t

Article history: This study investigated the effect of different drying techniques, namely, room-temperature drying (RTD),
Received 14 December 2016 solar drying (SD), heat-pump drying (HPD), hot-air drying (HAD), and freeze drying (FD), on bioactive
Received in revised form 3 April 2017 components, fatty acid composition, and the volatile compound profile of robusta coffee beans. The data
Accepted 25 April 2017
showed that FD was an effective method to preserve fat, organic acids, and monounsaturated fatty acids.
Available online 26 April 2017
In contrast, HAD was ideal for retaining polyunsaturated fatty acids and amino acids. Sixty-two volatile
compounds were identified in the differently dried coffee beans, representing 90% of the volatile com-
Keywords:
pounds. HPD of the coffee beans produced the largest number of volatiles, whereas FD resulted in the
Robusta coffee beans
Drying techniques
highest volatile content. A principal component analysis demonstrated a close relationship between
Phytochemical compounds the HPD, SD, and RTD methods whereas the FD and HAD methods were significantly different. Overall,
Fatty acids the results provide a basis for potential application to other similar thermal sensitive materials.
Volatile profile Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Currently, coffee is drunk by millions of people every day, and


the demand for high-quality specialty coffee is also increasing.
Coffee is one of the most popular beverages worldwide with an However, the harvest and supply periods for fresh coffee fruit are
extremely complex flavor; moreover, coffee is a valuable agricul- relatively short. Fresh coffee fruit contain 75%–85% water and is
tural commodity and an important export in many developing susceptible to microbial spoilage and chemical deterioration,
tropical countries (Alves et al., 2016). Coffee is cultivated in over which can affect coffee’s characteristics, such as its body, flavor,
sixty countries located in regions lying between the latitudes of and aroma (Batista et al., 2009). Currently, coffee dehydration is
Cancer (23°260 north of the equator) and Capricorn (23°260 south the main processing procedure used to inhibit microbial spoilage
of the equator) (Carter, Yates, & Tinggi, 2015). The most important and prevent chemical deterioration. However, the drying process
commercial coffee species are arabica (Coffea arabica L.) and may be accompanied with changes in the physical, chemical, and
robusta (Coffea canephora Pierre ex A. Froehner) (Garrett, organoleptic properties of heat-sensitive components, including
Rezende, & Ifa, 2016). In China, coffee is mainly grown in the Yun- color degradation, crust formation, decrease of sensory quality,
nan and Hainan provinces. The arabica and robusta species are inactivation of bacteria and enzymes, and loss of nutrients and
planted in those two districts, respectively. Recently, phytochemi- aroma (Naidu et al., 2016). Thus, the selection and control of the
cal studies have shown that coffee consumption can reduce the drying methods used for fresh coffee fruit are very important. At
risk of type 2 diabetes (Ding, Bhupathiraju, Chen, van Dam, & Hu, present, research into new drying techniques is a rapidly develop-
2014), colorectal cancer (Li, Ma, Zhang, Zheng, & Wang, 2013), ing area, caused by growing consumer demand, the need for lower
and cardiovascular diseases (Yang et al., 2016). In recent years, cof- production costs, and a desire to lessen environmental impact.
fee has attracted considerable attention among consumers world- Drying is one of the oldest and most efficient preservation tech-
wide mainly because of its health benefits, specific aroma, and niques used in the food industry. In most coffee-producing coun-
characteristic bitter taste. tries, room-temperature drying (RTD) is the simplest and
cheapest method used to dry fresh coffee fruit; however, RTD is
often susceptible to contamination with foreign materials, such
as dust and litter, and the coffee fruit can be exposed to fungi,
⇑ Corresponding author at: Spice and Beverage Research Institute, Chinese
birds, insects, and rodents, which can cause microorganism growth
Academy of Tropical Agricultural Sciences, Wanning 571533, China.
E-mail addresses: dongwenjiang.123@163.com (W. Dong), tlh3687@163.com
and deterioration of flavor quality (Fudholi, Sopian, Othman, &
(L. Tan). Ruslan, 2014). Solar drying (SD) is one attractive alternative

http://dx.doi.org/10.1016/j.foodchem.2017.04.156
0308-8146/Ó 2017 Elsevier Ltd. All rights reserved.
122 W. Dong et al. / Food Chemistry 234 (2017) 121–130

application that arises from solar energy systems and presents a (analytical grade) were obtained from Damao Chemical Reagent
great potential for use in drying (Fudholi, Sopian, Ruslan, Co., Ltd. (Tianjin, China). Caffeine, trigonelline, C7–C30 n-alkanes,
Alghoul, & Sulaiman, 2010). Hot-air drying (HAD) and vacuum and amino acid standards, as well as oxalic, tartaric, quininic,
freeze drying (FD) are also two methods commonly used to dehy- formic, malic, citric, and succinic acids were obtained from
drate agricultural products. HAD is the most conventional drying Sigma-Aldrich Company (St. Louis, MO, USA). A solid-phase
method, and is relatively cheap; however, HAD can cause chemical microextraction (SPME) fiber assembly composed of carboxen/
degradation of coffee fruit. FD can yield high-quality products, as polydimethylsiloxane (CAR/PDMS, 75 mm) was obtained from
well as maintaining the nutrient content, color, flavor, and texture; Supelco (Bellefonte, PA, USA). HPLC-grade water prepared by a
however, FD requires high-energy consumption and investment Mater-s-plus UVF system (Hetai, Shanghai, China) was employed
costs (Wojdylo et al., 2016). Heat-pump drying (HPD) is a novel throughout the experiments and additionally filtered through a
and promising drying technique with the evident advantages of 0.22 mm membrane immediately before use.
low energy consumption, reduced quality loss, independent con-
trol of operation parameters, and suitability for drying heat- 2.2. Coffee samples
sensitive products (Deng, Luo, Wang, & Zhao, 2015; Deng et al.,
2011). Fresh mature coffee fruit (Coffea robusta, RY1 cultivar) was har-
Fats, proteins, and free amino acids are important contributors vested in January 2015 from the coffee research farm of the Spice
to beverage quality because they serve as precursors to coffee and Beverage Research Institute of Chinese Academy of Tropical
aroma and flavors, which are formed during the roasting process Agricultural Sciences, Wanning, Hainan, China. The raw coffee fruit
through predominantly Strecker and Maillard reactions (Czech was washed in water to remove dirt and sand, followed by blotting
et al., 2016). Caffeine has a bitter taste, which is an important to remove excess water. The average initial moisture content of the
determinant of coffee flavor formation; in contrast, trigonellines coffee samples was expressed in g/g dry weight (DW). Samples
are known to contribute to the formation of pleasing flavors during were stored in a refrigerator at 4 °C and 90% relative humidity.
coffee roasting (Perrone, Donangelo, & Farah, 2008). The acidity Samples were removed from the refrigerator and left to stand for
and sourness of coffee brews are important because of their sen- 2.5 h to reach room temperature before the drying experiments.
sory quality; most of the acidity is generated during roasting using The raw coffee fruit was divided into five equal portions of 300 g
organic acids as corresponding precursors (Rodrigues et al., 2007). for each drying method. The dried coffee fruit was mechanically
Although the volatile changes in roasted coffee beans have been hulled to obtain green coffee beans.
well investigated, information is lacking on the influence of drying
procedures on the volatile compounds in green coffee beans. Sev-
2.3. Drying methods
eral studies on dried coffee beans, whole fresh coffee fruit, wet cof-
fee beans, and coffee powder have been conducted. Thus, different
The fresh coffee fruit (1500 g) was divided into five equal por-
drying techniques that can affect the composition of particular
tions, of 300 g each, and these five portions were applied to five dif-
characteristic phytochemicals in coffee beans should be performed.
ferent drying methods, respectively. Samples were dried using one
Suzihaque and Driscoll (2016) studied the effects of solar radiation,
of the five different methods until the final moisture content was
buoyancy of air flow, and optimization of coffee drying using a heat
approximately 11.0 ± 1.0 g/100 g DW. All drying experiments were
recovery dryer and found that the tray area, density of coffee
performed in triplicate. After drying, samples were preserved in a
beans, and the price of labor exhibited the most significant influ-
desiccator for three days to achieve a dynamic water equilibrium.
ence on profit. Kleinwächter and Selmar (2010) reported the effects
The dried fruit was then hulled and the green coffee beans were
of drying on the content of sugars in wet processed green Arabica
ground using a high-speed grinder (QE-100, Yili Instrument Co.,
coffee. They found that the fructose and glucose contents
Ltd., Wuyi, China) and passed over a 60-mesh sieve. The powdered
decreased significantly within the first day of drying because of
samples were sealed and stored in plastic bags at 4 °C until further
metabolic processes. A coffee FD process has been designed using
analysis.
mathematical modeling to maximize energy efficiency and retain
the effective ingredients of coffee (Fissore, Pisano, & Barresi,
2.3.1. Room-temperature drying (RTD)
2014). To our knowledge, no comprehensive study has simultane-
The first portion of fresh coffee fruit was placed onto large bam-
ously investigated the effects of drying methods on the color,
boo plates to form a layer (0.5 cm thick). The plates were placed in
bioactive components, free amino acids and volatile compounds
a ventilated open area, which allowed natural air flow to dry the
of coffee.
samples at an ambient room temperature of 20 °C to 30 °C and
Therefore, this study investigated the influence of different dry-
70% relative humidity for seven days. Drying continued until the
ing methods (namely, RTD, SD, HAD, HPD, and FD) on the caffeine,
final moisture content of samples reached 11.0 ± 1.0 g/100 g DW.
trigonelline, fat, protein, amino acid, and fatty acid composition as
well as the volatile profile of coffee beans while considering the
conservation of coffee quality. The results reported here should 2.3.2. Solar drying (SD)
serve as a guide to produce high-quality green coffee beans and The second portion of fresh coffee fruit was placed onto three
improve the chemical components involved in coffee conservation. stainless-steel trays and placed in the SD chamber with ambient
air heated by the solar collectors through recirculation of the dry-
ing air. The SD experiment was completed within five consecutive
2. Materials and methods days during sunlight hours until the moisture content reached
11.0 ± 1.0 g/100 g DW.
2.1. Chemicals and reagents
2.3.3. Heat-pump drying (HPD)
High-performance liquid chromatography (HPLC)-grade metha- The third portion of fresh coffee fruit was evenly arranged on
nol, acetic acid, and phosphoric acid were purchased from Merck several specialized sieve trays to form a layer (0.5 cm thick) and
(Darmstadt, Germany). Saturated fatty acids (SFAs) and unsatu- then dehydrated using a laboratory-scale heat-pump dehumidified
rated fatty acids (UNSFAs) were purchased from Yuanye Biotech- dryer (RF-25II, Guangzhou Quanneng Energy Technology Co., Ltd.,
nology Co., Ltd. (Shanghai, China). Petroleum ether and ethanol Guangzhou, China) at 50 °C for 72 h at an air flow rate of 2 m/s
W. Dong et al. / Food Chemistry 234 (2017) 121–130 123

and 70% relative humidity. The drying temperature was deter- The filtrate was filtered through a 0.22 mm regenerated cellulose
mined based on previous experiments. membrane prior to HPLC analysis. Caffeine and trigonelline in the
coffee extracts were quantified by HPLC based on the method
2.3.4. Hot-air drying (HAD) developed by Smrke, Kroslakova, Gloess, and Yeretzian (2015),
The fourth portion of fresh coffee fruit was uniformly dis- with minor modifications, using an Agilent 1290 instrument (Agi-
tributed over several stainless steel trays to form a layer (0.5 cm lent Technologies Inc., Palo Alto, CA, USA). Extracted coffee bean
thick) and then dried in an electro-thermal blowing dry box extracts were separated using a reverse-phase C-18 column
(DHG-9030A, Shanghai Yiheng Laboratory Instrument Co., Ltd., (100 mm  2.1 mm, i.d., 1.7 mm particle size). The mobile phase
Shanghai, China) at 50 °C for 108 h. The drying temperature was consisted of two components: methanol (A) and a 0.1% (v/v) glacial
determined based on our previous experiments. acetic acid aqueous solution (C). Gradient elution conditions were
as follows: 0–30 min, 5%–100% A; 30–32 min, 100%–100% A. The
2.3.5. Freeze drying (FD) mobile phase flow rate was 0.2 mL/min. The column temperature
The fifth portion of fresh coffee fruit was first frozen at 40 °C was 25 °C, and the injection volume was 1.5 mL. The detector was
for 10 h. The coffee fruit was then placed horizontally onto drying set at 275 nm for caffeine and 268 nm for trigonelline. All samples
plates with a layer thickness of 0.5 cm and dried in a freeze dryer were analyzed in triplicate.
(JDG-0.2, Lanzhou Kejin Freeze Drying Instruments Co., Ltd., Lanz-
hou, China) under 40 Pa vacuum pressure for 25 h. The tempera- 2.7. Organic acid determination
ture in the drying chamber was 40 °C, whereas those in the
heating plate and cold trap were 40 °C and 50 °C, respectively. Up to 5 g of ground coffee sample was transferred to a 100 mL
round-bottomed flask and 60 mL of deionized water was then
added. The reaction mixture was extracted three times at 70 °C
2.4. Color measurement
for 2.5 h in reflux extraction mode. The water extract was filtered
through a membrane (Whatman No. 4) and concentrated using a
The color of coffee samples dried under the different drying
rotary evaporator. The filtrate (volume 100 mL) was then filtered
conditions was measured with a colorimeter (X-rite, SP62, Grand
again through a 0.22 mm regenerated cellulose membrane prior
Rapids, MI, USA) using the CIELab method. The procedure con-
to organic acid analysis. Organic acids were determined using the
formed to the standard test method (ASTM, 2013). Color parame-
method reported by Niu et al. (2012) with minor modifications.
ters included L⁄ (L⁄ = 100 means white, L⁄ = 0 means black), a⁄
The analysis was performed using the HPLC system equipped with
[redness (+) and greenness ()], and b⁄ [yellowness (+) and blue-
a diode array detector (DAD) at 210 nm with a Zorbox SB-Aq
ness ()]. Color coordinates were determined in the CIE L⁄ a⁄ b⁄
(4.6 mm  250 mm, 5 mm particle). The mobile phases consisted
space, and the illuminant D65/10° was used. Total color differences
of a mixture of methanol and 0.01 mol/L NaH2PO4 buffer solution
(DE), chroma (C⁄), and hue (H⁄) were calculated using Eqs. (1) and
(pH = 2.00 ± 0.02 adjusted with o-phosphoric acid). The isocratic
(2). Each measurement was determined using five replicates. The
elution program was performed on mobile phases at a flow rate
average values of the color parameters with standard deviations
of 0.5 mL/min. Quantification of organic acids was achieved using
were then obtained.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi a standard curve and data were expressed as g/100 g DW of coffee
  2 beans.
DE  ¼ ðLo  L Þ þ ðao  a Þ2 þ ðbo  b Þ
2
ð1Þ
2.8. Fatty acid analysis
2 1=2
C  ¼ ða2 þ b Þ ð2Þ
Fatty acid composition was determined by gas chromatography
 
 1 b (GC) after trans-methylation with 2 M methanolic potassium
H ¼ tan ð3Þ hydroxide in accordance with ISO 12966-2 (2011). An Agilent
a
7890A GC equipped with a flame ionization detector and an
auto-sampler was used for the analysis. Fatty acid separation
2.5. Total crude protein and fat contents was performed on a fused silica cyanopropyl HP-88 column (length
100 m, i.d. 0.25 mm, film thickness 0.20 mm, J&W Scientific, USA).
Protein content was determined using the Dumas combustion The column temperature was programmed as follows: 100 °C ini-
method and an adaptation of the AOAC Official Method (AOAC, tial temperature, 2 min hold, then at a rate of 20 °C min1 to
1995) using a Dumas analyzer (NDA 701, Velp Scientifica, Milan, 170 °C, followed by a rate of 2 °C min1 to 200 °C, 10 min hold,
Italy). Up to 100 mg of coffee powder was accurately weighed in and then to 250 °C, and finally a hold for 5 min. Helium was used
tin foil and analyzed using the Dumas analyzer with ethylenedi- as the carrier gas at a flow rate of 1.0 mL/min and injector split
aminetetraacetic acid as a standard. The nitrogen content was mul- ratio of 1:50. The injector volume was 1 mL. A reference standard
tiplied by a factor of 6.5 to calculate the crude protein content. Fat FAME mix (Supelco Inc., Sigma-Aldrich) was analyzed under the
content analysis was measured using the Soxhlet extraction same conditions as the coffee extracts to identify the chromato-
method (No. 991.36, AOAC, 1996). Total fat was extracted from graphic peaks by comparing their retention times and data were
the dried coffee powder using petroleum ether in a Soxhlet appa- expressed as percentage of total fatty acids.
ratus. The extraction was performed at 50 °C for 16–18 h until
the petroleum ether was colorless. All analyses were conducted 2.9. Amino acid analysis
in triplicate.
Coffee samples were first hydrolyzed to convert proteins into
2.6. Determination of caffeine and trigonelline by HPLC methods their constituent amino acids. The samples were acid digested
(6 mol/L HCl) under vacuum in sealed glass tubes at 110 °C for
For caffeine and trigonelline analyses, an aliquot of 0.2 g ground 24 h. Amino acids were then analyzed using an automated SYKAM
coffee sample was extracted three times with deionized water at Amino Acid Analyzer S-433D (Sykam GmbH; Kleinostheim,
95 °C for 60 min, filtered through a membrane (Whatman No. 4) Germany) with a polyetheretherketone column (4.6 mm  150 mm,
under a vacuum, and then diluted to 100 mL with deionized water. 7 mm) equipped with diode array detector (DAD). The amino acid
124 W. Dong et al. / Food Chemistry 234 (2017) 121–130

content was quantified by comparing the retention times and peak Fig. S1a–e (Supplementary material). For analytical purposes, the
areas with those of the standard curves and expressed in milligram RTD sample was selected as the reference sample. Drying caused
of amino acid per 100 g DW. an increase in the L⁄, H⁄, and DE values compared with those in
the RTD sample; in contrast, the value of a⁄ decreased when all
2.10. Volatile compound composition analysis by headspace solid- drying methods were used to prepare coffee beans. Maximum
phase microextraction/gas chromatography mass spectrometry (HS- color lightness was found for FD (64.02 ± 0.40), followed by HAD
SPME/GC–MS) (62.10 ± 0.16), and then HPD (60.76 ± 0.22). No significant differ-
ence existed between the HPD and SD samples. The increases in
Up to 2 g of ground coffee was added to a 20 mL vial. The sam- the L⁄ value indicated color brightening of the dried coffee beans.
ple vial was closely capped with a polytetrafluoroethylene-silicon This result indicated a small amount of reddish components. For
stopper. The sample vial was then pre-equilibrated for 20 min at a⁄, the highest and lowest values were found in samples dried by
350 rpm and 60 °C, followed by extraction for 30 min using a car- the RTD and FD methods, respectively. The value of the blue-
boxen/polydimethylsiloxane (CAR/PDMS, 75 mm) SPME fiber in the yellow coordinate parameter ranged from 19.62 ± 1.00 (FD) to
headspace. Volatile compound analyses were performed using an 23.00 ± 0.20. Changes in the b⁄ value can be estimated by the sam-
Agilent 7890A GC coupled to a 5975C quadrupole mass spectrom- ples’ loss or increase in yellowness. These value variations of L⁄, a⁄,
eter with electron ionization at 70 eV. Compounds were separated and b⁄ may be attributed to degradation of pigments or non-
using a polar DB-WAX capillary column (300 mm  0.25 mm, enzymatic Maillard browning (Dadali, Demirhan, & Özbek, 2007).
0.25 mm film thickness) (J&W Agilent) using the following temper- The total color change of coffee beans (DE) was evaluated for each
ature program: initially 40 °C for 2 min, increased to 130 °C at method and compared to the RTD sample. The highest DE was
1.5 °C/min, and then at 4 °C/min up to 220 °C and then held for found in the FD samples, whereas the lowest DE occurred in sam-
5 min. The ionization source and quadrupole temperatures were ples dried by RTD. This result can be mainly ascribed to changes in
set at 230 °C and 150 °C, respectively. Identification of volatile the L⁄ and a⁄ values of the FD samples. However, no significant dif-
compounds was achieved by comparison with linear retention ference was observed between the DE values for the SD and HPD
indices, derived from C7-C40 alkane standards analyzed under methods. These data demonstrate that darker coffee beans, with
the same chromatographic conditions; semi-quantification of vola- more red and yellow coloration, were obtained through drying
tiles was obtained using gas chromatography-flame ionization with either the RTD or SD methods. In contrast, lighter coffee beans
detector GC-FID peak areas (Toci & Farah, 2014). Other tentative were obtained if they were dried using the FD, HAD, or HPD meth-
identifications were performed by comparing the mass spectra ods; the coffee bean coloration after HPD being intermediate
obtained with those mass spectra in the NIST08 library as well as between FD and HAD. In summary, the changes in color were sub-
related literature (Iamanaka et al., 2014; Lee et al., 2017). The rel- tle, but were difficult to detect using the human eye.
ative amounts of volatile compounds were expressed as percent-
ages of the peak area of the total ion chromatogram. 3.2. Lipid and protein contents of dried coffee beans

2.11. Statistical analysis Coffee lipid contributes to the texture and feeling of the bever-
age in the mouth because they carry flavors and fat-soluble vita-
Experimental data were analyzed using SPSS 20.0 (SPSS Inc., mins. The effect of drying on the fat and protein content of coffee
Chicago, IL, USA) and Matlab R2010a (The MathWorks, Inc., Natick, beans is shown in Fig. 1a and b. The total lipid content of coffee
MA, United States) software. Statistical significance was assessed beans dried using the RTD method was 6.63 ± 0.65 g/100 g DW.
using one-way ANOVA (SPSS Program version 20.0). Significant dif- In comparison, all the other drying methods, with the exception
ferences were evaluated by Duncan’s test (p < 0.05). Principal com- of HPD, resulted in a significant increase in lipid content, with
ponent analysis (PCA) and hierarchical cluster analysis were beans prepared using FD yielding the highest lipid content, fol-
performed using Matlab R2010a software. All analyses were con- lowed by SD (8.74 ± 0.06 g/100 g DW), HAD (8.31 ± 0.42 g/100 g
ducted in triplicate. Data were expressed as mean ± standard DW), and HPD (7.47 ± 1.02 g/100 g DW). The increases in lipid con-
deviation. tent in the SD and HAD dried beans were statistically significant
(p < 0.05) compared to the RTD samples. These findings are in
3. Results and discussion agreement with previous reports showing that the lipid content
of green coffee beans ranged from 7.2% to 16.3% (Ramalakshimi,
3.1. Color parameters of dried coffee beans Kubra, & Rao, 2007). Protein is also an important flavor precursor
in the coffee bean roasting process. The total protein contents of
Color plays an important role in food quality assessment by the differently dried coffee beans are shown in Fig. 1b. Beans dried
consumers; it can also be used to optimize the drying process using SD had the highest total protein compared to the RTD coffee
and reduce the degradation of bioactive compounds. The color beans. Significant increases in protein were observed with the
parameters of dried coffee beans under the different drying meth- other drying methods. When the protein content of all the dried
ods are summarized in Table 1. Photographs of the dried coffee coffee bean samples was compared, samples prepared using HAD
beans prepared using the different drying methods are shown in and FD had 14.83 ± 0.03 g/100 g DW and 14.71 ± 0.32 g/100 g

Table 1
Effect of drying methods on the color parameters in dried coffee beans.

Drying methods L* a* b* C* H*
d a bc b
RTD 57.02 ± 0.51 3.60 ± 0.03 21.91 ± 0.14 22.20 ± 0.15 80.85 ± 0.01d 2.40 ± 0.55c
SD 60.07 ± 0.36c 3.58 ± 0.04a 22.43 ± 0.25ab 22.71 ± 0.25ab 83.19 ± 0.03b 2.77 ± 0.35b
HPD 60.76 ± 0.22c 2.28 ± 0.22b 20.45 ± 0.02d 20.58 ± 0.46c 81.97 ± 0.04c 2.69 ± 0.16b
HAD 62.10 ± 0.16b 3.41 ± 0.07a 23.00 ± 0.20a 23.25 ± 0.20a 83.42 ± 0.04b 7.94 ± 0.43a
FD 64.02 ± 0.40a 1.53 ± 0.21c 19.62 ± 1.00e 19.68 ± 1.00d 84.87 ± 0.15a 8.08 ± 0.17a

Results are expressed as mean ± standard error. Different capital letters in the same column indicate significant differences at p < 0.05.
W. Dong et al. / Food Chemistry 234 (2017) 121–130 125

Fig. 1. Average values and the standard deviation of fat (a), protein (b), caffeine (c), and trigonelline (d) in dried coffee samples obtained by the different drying methods.
ANOVA analysis between groups are indicated by letters.

DW, respectively. Therefore, we conclude that changes in the dry- trigonelline content in beans processed using the wet method was
ing process can lead to significant relative increases or decreases in higher than that in beans processed using the semi-dry method.
the levels of fat and protein, it may be related to the fatty acid com- Based on the data generated here, we conclude that the SD method
position or the combination mode of the protein. of drying has the ability to preserve caffeine and trigonelline con-
tent, but it takes a longer time for beans to dry. In contrast, the
3.3. Caffeine and trigonelline contents of dried coffee beans HPD and HAD methods can each preserve these two important
bioactive compounds and have the advantage of being quick to
The caffeine and trigonelline contents of the differently dried cof- complete.
fee beans are shown in Fig. 1c and d. The caffeine content was the
highest in beans dried using the SD method (2.47 ± 0.20 g/100 g 3.4. Organic acid composition
DW) and lowest in beans dried using the RTD method
(2.30 ± 0.02 g/100 g DW). The caffeine content was similar among Acidity is an important attribute of coffee quality along with
the SD and FD methods (2.40 ± 0.08 g/100 g DW), as well as between sweetness, bitterness, and aromatic profile. The acidity of coffee
the RTD and HAD methods (2.31 ± 0.05 g/100 g DW), ranging from is mainly related to organic acids, which have a significant effect
2.30 ± 0.02 g/100 g DW to 2.47 ± 0.20 g/100 g DW. With the excep- on the flavor balance, as well as influencing pH and coffee quality.
tion of the RTD and FD samples, there was no significant difference As shown in Table 2, a total of seven organic acids were measures
in the caffeine content in dried coffee beans prepared by the differ- in dried coffee beans. Coffee beans prepared by the FD method had
ent drying methods (p > 0.05). This result might be explained by the the highest total organic acid content, whereas beans prepared
relatively stable structure of caffeine. The levels of caffeine reported with the RTD method had the lowest content. The most abundant
here are in agreement with previous studies that indicated that the common organic acid was malic acid, followed by tartaric and
caffeine content of robusta coffee beans ranged from 1.2% to 4.0% citric acids with oxalic acid being the lowest of all the common
(Richo, Leitão, Ramalho, de Alvarenga, & Lidon, 2013; Esquivel & organic acids. Of interest, quinic acid was only detected in beans
Jiménez, 2012). The effects of the five different drying conditions prepared using the FD method. This result suggests that quinic acid
on trigonelline content were almost in accordance with those on is degraded under the other drying conditions. Formic and succinic
the caffeine content. The trigonelline content was ranked in the acids accounted for approximately 10% and 7% the total organic
following order: SD > HAD > FD > RTD > HPD. The trigonelline acids present in the dried coffee beans, respectively. The tartaric
contents of beans dried using the FD, RTD, or HPD methods were acid content in the dried coffee beans fluctuated contents depend-
0.88 ± 0.03 g/100 g DW, 0.87 ± 0.02 g/100 g DW, and ing on the drying method; beans prepared using FD had the highest
0.86 ± 0.01 g/100 g DW, respectively. Duarte, Pereira, and Farah content, whereas beans prepared by the RTD method had the low-
(2010) have reported the effects of wet and semi-dry methods on est. Jham, Fernandes, Garcia, and da Silva (2002) have compared
the trigonelline level in post-harvested coffee seeds, in which the the use of GC and HPLC methods for the quantification of organic
126 W. Dong et al. / Food Chemistry 234 (2017) 121–130

Table 2
Organic acid composition of dried coffee beans under different drying conditions (g/g DW).

Organic acids RTD SD HPD HAD FD


a a a b
Oxalic acid 41.57 ± 2.47 39.75 ± 0.22 42.86 ± 0.09 31.23 ± 0.01 31.46 ± 0.64b
Tartaric acid 1358.38 ± 28.06d 1861.92 ± 22.50b 1685.0 ± 17.91c 1821.35 ± 25.80b 3168.98 ± 97.49a
Quinic acid ND ND ND ND 690.98 ± 30.95a
Formic acid 1204.78 ± 34.52c 1177.71 ± 2.19d 1161.96 ± 2.98d 1392.33 ± 13.60a 1330.74 ± 7.46b
Malic acid 4464.68 ± 119.42b 4235.16 ± 8.10d 4304.69 ± 5.48c 5525.58 ± 45.81a 4170.15 ± 5.73d
Citric acid 1708.06 ± 8.80b 1580.02 ± 0.45c 1688.40 ± 0.17b 1905.66 ± 0.50a 1445.24 ± 104.25d
Succinic acid 578.67 ± 24.57d 725.38 ± 4.12c 560.54 ± 0.76d 1011.34 ± 8.38b 1799.42 ± 49.01a

ND: not detected.


Results are expressed as mean ± standard error. Different capital letters in the same row indicate significant differences at p < 0.05.

acids in coffee samples. In their study, the major acids were found Ghribi et al., 2015). Total free amino acid content was significantly
to be citric, quinic, and malic acids. These results were similar to different in coffee beans prepared by the different methods
the findings reported by Alcázar, Fernández-Cáceres, Martín, (p < 0.05) and had a range of 4.81–7.79 g/100 g DW. For essential
Pablos, and González (2003). The variation in organic acids amino acids (EAAs), the beans prepared using HAD had significant
reported by previous researchers may be ascribed to the different amounts of threonine, valine, methionine, isoleucine, as well as
extraction and detection methods, or to the different coffee culti- other EAAs. Across the different coffee bean drying methods, the
vars used in those studies. total essential amino acids (TEAA) were ranked in the order of
HAD > FD > RTD > SD > HPD. The high amount of EAA present in
3.5. Fatty acid composition HAD-prepared coffee beans might arise as a result of the proteoly-
sis that likely occurred as a result of heating during the HAD pro-
The fatty acid profile (% weight of methyl esters); the sum of cess. The most predominant EEAs in the dried coffee beans were
total SFA, monounsaturated (MUFA), and polyunsaturated (PUFA) leucine, lysine, and phenylalanine. Ten non-essential amino acids
fatty acids; and the UNSFA/SFA ratios are presented in Table S1 were also detected in dried coffee bean samples. The FD-
(Supplementary material). Thirteen different kinds of fatty acids, prepared coffee beans had a significantly higher total amino acid
including nine SFAs, two MUFAs, and two PUFAs, were character- (TAA) content compared to coffee beans prepared using the RTD,
ized in the differently dried coffee beans. Eleven of them were SD, and HPD methods. This result shows that FD is a highly effi-
detected in all five dried coffee samples. C21:0 and C23:0 were cient drying method that can maintain the levels of amino acids
not detected in beans prepared using the FD method; C23:0 was important for taste. Murkovic and Derler (2006) have reported that
also not found in HAD prepared beans. The major fatty acid in alanine is the amino acid with the highest concentration in robusta
the coffee samples was C18:2, accounting for 42.10% (FD sample) coffee beans, followed by asparagine. They also reported that the
to 44.53% (HAD sample) of total fatty acids, followed by C16:0 amino acid content was higher in robusta beans than in arabica
(34.11%–35.90%), C18:1 (8.53%–10.42%), C18:0 (5.98%–6.83%), beans. These finding differed from our study and may be attributed
and C20:0 (2.91%–3.86%). Other fatty acids were detected at per- to the differences in coffee variety, geographical origin, or analyti-
centages of less than 0.5%. However, the content of most individual cal methods used. In conclusion the current study showed that cof-
fatty acids in the coffee samples differed significantly from each fee beans prepared using the HAD method had a much higher
other. The total SFA content ranged from 45.01% to 45.62%, with content of total amino acids compared to beans prepared by the
the highest percentage in RTD-prepared beans and lowest in other drying methods; this phenomenon may be attributed to
HAD-prepared beans. Significant differences existed in the MUFA the higher degradation rate that occurs during the other drying
and PUFA contents of the dried coffee samples. The total MUFA processes.
content in HAD, SD, RTD, HPD, and FD processed beans increased
successively, ranging from 8.92% to 10.88% of the total fatty acid
content. The fatty acid composition of coffee has been reported 3.7. Volatile compounds of dried coffee beans
by Romano et al. (2014) who demonstrated that coffee beans con-
tained more UNSFAs. This finding was in accordance with our The profile of volatile compounds in roasted coffee is one of the
experimental results. Oliveira, Franca, Mendonca, and Barros- most important factors influencing coffee quality. The GC–MS
Júnior (2006) also reported that a high free fatty acid content chromatograms of one sample extract, examples of the corre-
was found in pressed coffee oil: 4.9 ± 0.4 g oleic acid/100 g oil. This sponding methods, and the volatile composition of dried coffee
high free fatty acid content may be ascribed to the pressing beans are shown in Fig. 2a–c. Some compounds present in green
method use for extracting the oil. We conclude that HAD was an coffee are also found in roasted coffee. A total of 62 volatile com-
ideal drying method for retaining polyunsaturated fatty acid and pounds were identified in the headspace of green coffee powder
amino acid content. In addition, beans prepared by HPD had a (Table S3, Supplementary material). In the case of coffee powder
higher content of saturated fatty acids and a higher ratio of unsat- prepared using HAD, the majority of the volatiles were esters
urated fatty acids to saturated fatty acids. (4.55%), aldehydes (25.08%), alcohols (16.24%), acids (39.84%),
hydrocarbons (6.83%), and ketones (5.49%); these have previously
3.6. Amino acid composition been reported in other coffee types (Bertrand et al., 2012;
Moreno et al., 2015). A clear quantitative difference in the chro-
Coffee beans contain a variety of amino acids. The amino acid matographic profile of GC–MS is shown in Fig. 2b. However, no sig-
content and composition of the coffee beans prepared by the vari- nificant difference was found in the qualitative profiles of the
ous drying methods are shown Table S2 (Supplementary material). volatile compounds from coffee beans prepared using the different
Amino acids have been reported to be susceptible to the processing drying methods. Using an ANOVA analysis to compare the total
conditions, depending on the source, the amino acid itself, and the peak area of volatile compounds, significant differences were
processing method (Boye, Wijesinha-Bettoni, & Burlingame, 2012; observed among most of these compounds.
W. Dong et al. / Food Chemistry 234 (2017) 121–130 127

Fig. 2. GC–MS chromatograms of one sample extract (a), examples of the corresponding drying methods for dried coffee beans (b), the relative percentage of volatile
compounds (c).
128 W. Dong et al. / Food Chemistry 234 (2017) 121–130

The different drying methods led to changes in the profile of


volatile compounds found in the coffee beans. The concentration
of acids was the highest in beans prepared by the FD method
whereas in contrast, the lowest concentration of acids was found
in beans prepared by the HPD method, suggesting that FD method
was efficient in preserving acidic volatile compounds. In contrast,
the levels of either hydrocarbons or ketones did not differ when
comparing coffee beans prepared by the different drying methods.
The principal volatile compounds found in the dried coffee samples
were found to be acids, aldehydes, and alcohols, and 2,3-
butanediol, benzyl alcohol, hexanal, benzaldehyde, and acids (such
as acetic, 3-methyl-butanoic, and 3-methyl-2-butenoic acids) were
particularly prominent. A total of 39 and 40 volatile compounds
were detected in RTD and SD dried coffees, respectively In coffee
beans prepared using the HAD process, a total of 40 compounds
were detected, probably because heating during drying accelerated
the release of volatile compounds; surface moisture evaporation
likely also accelerated the loss of volatile compounds. Interestingly,
pyridine, which contributes to the smoky flavor of coffee, was only
detected in the coffee beans prepared using the SD method. More-
over, 1,3-diazine, which has an earthy taste, was found in the SD,
HPD, and FD dried coffee bean samples. Of note, pentylfuran, which
has a caramel flavor, was also absent in the HAD samples. Accord-
ing to the reports of Sunarharum, Wiliams, and Smyth (2014) and
Burdock (2010), volatile furans, which exhibit malty and sweet
roasted aromas, have a relatively high sensory threshold compared
with other coffee volatiles. In conclusion, the HAD and HPD drying
methods resulted in alterations in the volatile compounds found in
dried coffee beans likely because of the high temperatures and
intensive ventilation used in these methods. In particular, the
HPD method promotes the production of a number of dried coffee
bean volatile compounds leading to the largest number of volatiles.
Thus, the HPD method could be used to enhance the aroma and fla-
vor of processed coffee beans.

3.8. Principal component analysis

In order to observe the overall effects of the different drying


methods on the physicochemical composition of dried coffee Fig. 3. Score (a) and loading (b) plots from the principal component analysis of
beans, a principal component analysis (PCA) was performed on dried coffee samples showing the relationship among phytochemical profiles and
the total data set (15 samples  24 variables). The twenty-four different drying methods.
variables were fat, protein, caffeine and trigonelline contents,
levels of organic acids (oxalic, tartaric, quinic, formic, malic, citric,
and succinic acids), levels of amino acids (TEAA, TNEAA, and TAA), such as quinic, tartaric, succinic, and citric acids, were closely
levels of fatty acids (SFA, MUFA, PUFA, and UNSFA/SFA), and levels related to PC2.
of volatile chemical groups (esters, aldehydes, alcohols, acids,
hydrocarbons, and ketones). PC1 and PC2 were retained on the
basis of the cumulative percentage of total variation (95.7%). The 3.9. Hierarchical cluster analysis
first two principal components accounted for 91.2% and 4.5% of
the total variance, respectively. Fig. 3a and b shows the score and Each of the phytochemical variables was normalized. Moreover,
loading bioplots defined by the first two PCs obtained from the a heat map was produced to display how the differently prepared
PCA. Coffee bean samples prepared by the different drying meth- dried coffee beans varied in composition, using Euclidean distance
ods were clearly distinguished from each other by their PCA score as a similarity measure (Fig. 4). The clustering that resulted was
plots. The HAD and FD prepared samples were located in the pos- similar to that in the PCA bioplot shown in Fig. 3a and b. In the heat
itive PC1 direction. Whereas, in contrast, the HPD, RTD, and SD pre- map, the similarity measure was Euclidean distance. The Ward’s
pared samples were close to each other and in the negative PC1 linkage method was used as the clustering algorithm to minimize
direction, indicating that they differed from HAD and FD prepared the sum of squares of all clusters. All samples were grouped into
samples. A relatively low content of compounds was positively rel- five clusters, via cluster analysis, in accordance with the different
evant to this axis. Samples prepared using the HAD method were drying methods as indicated in Fig. 4. The clustering of phytochem-
located at the far end of the right axis (Fig. 3a and b). In addition, ical compounds demonstrated that the RTD, SD, and HPD dried
samples prepared using the FD method exhibited high positive samples grouped closely with high contents of alcohols, caffeine,
scores along the PC2 axis. Formic acid, TAA, TEAA, TNEAA (in the oxalic acid, aldehydes, SFA, and trigonelline. The HAD dried
positive direction), and, caffeine and oxalic acid (in the negative samples were separated from others in the dendrogram because
direction) contributed to a large extent to PC1. Ketones and acids, they were characterized by TEAA, TAA, TNEAA, malic acid, and
W. Dong et al. / Food Chemistry 234 (2017) 121–130 129

Fig. 4. Heat map and clustering of the total data set for coffee samples dried with different methods. Values were normalized between the maximum and minimum values for
each variable.

hydrocarbons. Interestingly, the FD samples formed a single group acids. The HAD method was ideal for retaining polyunsaturated
that was particularly rich in quinic, tartaric, and succinic acids. fatty acids and amino acids, whereas the HPD method produced
the largest number of volatiles as well as the highest content of
4. Conclusions saturated fatty acids and ratio of unsaturated fatty acids to satu-
rated fatty acids. Our findings demonstrated that the conventional
Five different drying methods based on color, phytochemical drying methods can be effectively replaced by HAD or HPD meth-
compounds, and volatile profiles of robusta coffee beans were ods in the future. Considering its drying efficiency, good quality
examined. On the basis of the current results, we can conclude that retention and production cost, the HPD has the broadest market
the drying and processing conditions exert profound effects on the prospect for potential application. In the meantime, more in-
flavor of dried agricultural products. A comparison between the depth studies should be performed to confirm this conclusion.
different drying methods indicated a difference in the quality of
the dried coffee beans. Significant differences were found in the Acknowledgments
contents of organic acids, total UNSFAs, and amino acids across
the different drying methods. However, the influence of the drying This work was supported by the China National Natural Science
method on other parameters such as SFA, caffeine, and trigonelline Foundation (31501404) and the Central Public-interest Scientific
content was very small. The HAD and HPD drying methods Institution Basal Research Fund of Chinese Academy of Tropical
resulted in a significant alteration in the dried coffee bean volatile Agricultural Sciences (CATAS) (1630142017005, 1630142016012).
compound profile likely because of the high temperatures and
intensive ventilation treatments used. Overall, the different meta- Appendix A. Supplementary data
bolic responses to the drying methods resulted in the multiple
metabolic association and clustering on the planes of PCA and heat Supplementary data associated with this article can be found, in
map. The FD method was a good drying method that can be uti- the online version, at http://dx.doi.org/10.1016/j.foodchem.2017.
lized to preserve fat, organic acids, and monounsaturated fatty 04.156.
130 W. Dong et al. / Food Chemistry 234 (2017) 121–130

References influence sensory quality of coffee beverage. Food Research International, 64,
166–170.
ISO 12966-2 (2011). Animal and vegetable fats and oils – gas chromatography of fatty
Alcázar, A., Fernández-Cáceres, P. L., Martín, M. J., Pablos, F., & González, A. G.
acids methyl esters – Part 2: Preparation of methyl esters of fatty acids.
(2003). Ion chromatographic determination of some organic acids, chloride and
Jham, G. N., Fernandes, S. A., Garcia, C. F., & da Silva, A. A. (2002). Comparison of GC
phosphate in coffee and tea. Talanta, 61, 95–101.
and HPLC for the quantification of organic acids in coffee. Phytochemical
Alves, L. C., De Magalhães, D. M., Labate, M. T. V., Guidetti-Gonzalez, S., Labate, C. A.,
Analysis, 13, 99–104.
Domingues, D. S., et al. (2016). Differentially accumulated proteins in Coffea
Kleinwächter, M., & Selmar, D. (2010). Influence of drying on the content of sugars
arabica seeds during perisperm tissue development and their relationship to
in wet processed green Arabica coffees. Food Chemistry, 119(2), 500–504.
coffee grain size. Journal of Agricultural and Food Chemistry, 64(7), 1635–1647.
Lee, L. W., Tay, G. Y., Cheong, M. W., Curran, P., Yu, B., & Liu, S. Q. (2017). Modulation
Batista, L. R., Chalfoun, S. M., Silva, C. F., Cirillo, M., Varga, E. A., & Schwan, R. F.
of the volatile and non-volatile profiles of coffee fermented with Yarrowia
(2009). Ochratoxin A in coffee beans (Coffea arabica L.) processed by dry and
lipolytica: I. Green coffee. LWT - Food Science and Technology, 77, 225–232.
wet methods. Food Control, 20(9), 784–790.
Li, G. W., Ma, D. F., Zhang, Y. M., Zheng, W., & Wang, P. Y. (2013). Coffee
Bertrand, B., Boulanger, R., Dussert, S., Ribeyre, F., Berthiot, L., Descroix, F., et al.
consumption and risk of colorectal cancer: A meta-analysis of observational
(2012). Climatic factors directly impact the volatile organic compound
studies. Public Health Nutrition, 16(2), 346–357.
fingerprint in green Arabica coffee bean as well as coffee beverage quality.
Moreno, F. L., Quintanilla-Carvajal, M. X., Sotelo, L. I., Osorio, C., Raventós, M.,
Food Chemistry, 135, 2575–2583.
Hernández, E., et al. (2015). Volatile compounds, sensory quality and ice
Boye, J., Wijesinha-Bettoni, R., & Burlingame, B. (2012). Protein quality evaluation
morphology in falling-filum and block freeze concentration of coffee extract.
twenty years after the introduction of the protein digestibility corrected amino
Journal of Food Engineering, 166, 64–71.
acid score method. British Journal of Nutrition, 108(S2), S183–S211.
Murkovic, M., & Derler, K. (2006). Analysis of amino acids and carbohydrates in
Burdock, G. A. (2010). Fenaroli0 s handbook of flavor ingredients (6th ed.). Boca Raton:
green coffee. Journal of Biochemical and Biophysical Methods, 69(1–2), 25–32.
CRC Press/Taylor & Francis.
Naidu, M. M., Vedashree, M., Satapathy, P., Khanum, H., Ramsamy, R., & Hebbar, H.
Carter, J. F., Yates, H. S. A., & Tinggi, U. (2015). Isotopic and elemental composition of
U. (2016). Effect of drying methods on the quality characteristics of dill
roasted coffee as a guide to authenticity and origin. Journal of Agricultural and
(Anethum graveolens) greens. Food Chemistry, 192, 849–856.
Food Chemistry, 63(24), 5771–5779.
Niu, Y. W., Zhang, X. M., Xiao, Z. B., Song, S. Q., Jia, C. S., Yu, H. Y., et al. (2012).
Czech, H., Schepler, C., Klingbeil, S., Ehlert, S., Howell, J., & Zimmermann, R. (2016).
Characterization of taste-active compounds of various cherry wines and their
Resolving coffee roasting-degree phases based on the analysis of volatile
correlation with sensory attributes. Journal of Chromatography B, 902, 56–60.
compounds in the roasting off-gas by photoionization time-of-flight mass
Oliveira, L. S., Franca, A. S., Mendonca, J. C. F., & Barros-Júnior, M. C. (2006).
spectrometry (PI-TOFMS) and statistical data analysis: Toward a PI-TOFMS
Proximate composition and fatty acids profile of green and roasted defective
roasting model. Journal of Agricultural and Food Chemistry, 64(25), 5223–5231.
coffee beans. LWT-Food Science and Technology, 39, 235–239.
Dadali, G., Demirhan, E., & Özbek, B. (2007). Color change kinetics of spinach
Perrone, D., Donangelo, C. M., & Farah, A. (2008). Fast simultaneous analysis of
undergoing microwave drying. Drying Technology, 25, 1713–1723.
caffeine, trigonelline, nicotinic acid and sucrose in coffee by liquid
Deng, Y., Luo, Y. L., Wang, Y. G., & Zhao, Y. Y. (2015). Effect of different drying
chromatography-mass spectroscopy. Food Chemistry, 110(4), 1030–1035.
methods on the myosin structure, amino acid composition, protein digestibility
Ramalakshimi, K., Kubra, I. R., & Rao, L. J. M. (2007). Physicochemical characteristics
and volatile profile of squid fillets. Food Chemistry, 171, 168–176.
of green coffee: Comparison of graded and defective beans. Journal of Food
Deng, Y., Wu, J., Su, S. Q., Liu, Z. D., Ren, L., & Zhang, Y. L. (2011). Effect of far-infrared
Science, 72, 333–337.
assisted heat pump drying on water status and moisture sorption isotherm of
Richo, N. C., Leitão, A. E., Ramalho, J. C., de Alvarenga, N. B., & Lidon, F. C. (2013).
squid (Illex illecebrosus) fillets. Drying Technology, 29(13), 1580–1586.
Identification of chemical clusters discriminators of Arabica and Robusta green
Ding, M., Bhupathiraju, S. N., Chen, M., van Dam, R. M., & Hu, F. B. (2014).
coffee. International Journal of Food Properties, 16(4), 895–904.
Caffeinated and decaffeinated coffee consumption and risk of type 2 diabetes: A
Rodrigues, C. I., Marta, L., Maia, R., Miranda, M., Ribeirinho, M., & Máguas, C. (2007).
systematic review and a dose-response meta-analysis. Diabetes Care, 37(2),
Application of solid-phase extraction to brewed coffee caffeine and organic acid
569–586.
determination by UV/HPLC. Journal of Food Composition and Analysis, 20(5),
Duarte, G. S., Pereira, A. A., & Farah, A. (2010). Chlorogenic acids and other relevant
440–448.
compounds in Brazilian coffees processed by semi-dry and wet post-harvesting
Romano, R., Santini, A., Le Grottaglie, L., Manzo, N., Visconti, A., & Ritieni, A. (2014).
methods. Food Chemistry, 118, 851–855.
Identification markers based on fatty acid composition to differentiate between
Esquivel, P., & Jiménez, V. M. (2012). Functional properties of coffee and coffee by-
roasted Arabica and Canephora (Robusta) coffee varieties in mixtures. Journal of
products. Functional properties of coffee and coffee by-products. Food Research
Food Composition and Analysis, 35, 1–9.
International, 46(2), 488–495.
Smrke, S., Kroslakova, I., Gloess, A. N., & Yeretzian, C. (2015). Differentiation of
Fissore, D., Pisano, R., & Barresi, A. A. (2014). Applying quality-by-design to develop
degrees of ripeness of Catuai and Tipica green coffee by chromatographical and
a coffee freeze-drying process. Journal of Food Engineering, 123, 179–187.
statistical techniques. Food Chemistry, 174, 637–642.
Fudholi, A., Sopian, K., Othman, M. Y., & Ruslan, M. H. (2014). Energy and exergy
Sunarharum, W. B., Wiliams, D. J., & Smyth, H. E. (2014). Complexity of coffee flavor:
analyses of solar drying system of red seaweed. Energy and Buildings, 68,
A compositional and sensory. Food Research International, 62, 315–325.
121–129.
Suzihaque, M. U. H., & Driscoll, R. (2016). Effects of solar radiation, buoyancy of air
Fudholi, A., Sopian, K., Ruslan, M. H., Alghoul, M. A., & Sulaiman, M. Y. (2010).
flow and optimization study of coffee drying in a heat recovery dryer. Procedia
Review of solar dryers for agricultural and marine products. Renewable and
Engineering, 148, 812–822.
Sustainable Energy Reviews, 14(1), 1–30.
Toci, A. T., & Farah, A. (2014). Volatile fingerprint of Brazilian detective coffee seeds:
Garrett, R., Rezende, C. M., & Ifa, D. R. (2016). Revealing the spatial distribution of
Corroboration of potential marker compounds and identification of new low
chlorogenic acids and sucrose across coffee bean endosperm by desorption
quality indicators. Food Chemistry, 153, 298–314.
electrospray ionization-mass spectrometry imaging. LWT - Food Science and
Wojdylo, A., Figiel, A., Legua, P., Lech, K., Carbonell-Barrachina, A. A., & Hernández, F.
Technology, 65, 711–717.
(2016). Chemical composition, antioxidant capacity, and sensory quality of
Ghribi, A. M., Gafsi, I. M., Blecker, C., Danthine, S., Attia, H., & Besbes, S. (2015). Effect
dried jujube fruits as affected by cultivar and drying method. Food Chemistry,
of drying methods on physico-chemical and functional properties of chickpea
207, 170–179.
protein concentrates. Journal of Food Engineering, 165, 179–188.
Yang, N., Liu, C. J., Liu, X. K., Degn, T. K., Munchow, M., & Fisk, I. (2016).
Iamanaka, B. T., Teixeira, A. A., Teixeira, A. R. R., Vicente, E., Frisvad, J. C., Taniwaki,
Determination of volatile marker compounds of common coffee roast defects.
M. H., et al. (2014). Potential of volatile compounds produced by fungi to
Food Chemistry, 211, 206–214.

Das könnte Ihnen auch gefallen