JFS S: Sensory and Food Quality
Sensory Characteristics and Consumer Acceptability of Decaffeinated Green Teas S.M. LEE, H.-S. LEE, K.-H. KIM, AND K.-O. KIM
Introduction
G
reen tea, one of the most preferred beverages in the world known to be processed from Camellia sinensis plant, is gaining more popularity, and recently, the market has been further expanded by introduction of canned green teas (Kim 1996), flavored green teas, and decaffeinated green teas. One of the main reasons for its attractiveness is due to its health beneficial aspect, which is however controversial. While catechins in green tea are well known for their antioxidative, anti-inflammatory, anticancer, antibiotic, and antiviral effects (Frankel 2007), caffeine in green teas has been reported to have negative effects on the cardiac vascular system (Hosenpud and others 1995) and human behaviors (Smith 2002), and cause sleep deprivation (Hindmarch and others 2002), miscarriages (Giannelli and others 2003; Rasch 2003), hypersensitivity (Bernhisel-Broadbent 1999), and calcium absorption that results in osteoporosis (Heaney and Recker 1982; Massey and Wise 1989; Hasling and others 1992). Ironically, this negative aspect of caffeine brought a new market in the green tea industry for those who want to avoid caffeine ingestion such as pregnant females, infants, and children (Park and others 2007a), who have a relatively slow detoxification rate of caffeines (Nawrot and others 2003). Decaffeination technique applying supercritical carbon dioxide fluid extraction (SC-CO 2 ) method developed by Zosel (1974), which was used mostly for decaffeinated coffee, has obtained increased interest in food processing. This technique has great advantages such as convenience, nonexplosiveness, nontoxicity (Lack
MS 20080895 Submitted 11/11/2008, Accepted 1/8/2009. Authors S.M. Lee, H.-S. Lee, and K.-O. Kim are with Dept. of Food Science and Engineering, Ewha Womans Univ., Seoul 120-750, South Korea. Author K.-H. Kim is with Div. of Food Bioscience and Technology, College of Life Science and Biotechnology, Korea Univ., Seoul 136-713, South Korea. Direct inquiries to author K.-O. Kim (E-mail:
[email protected]).
R Institute of Food Technologists doi: 10.1111/j.1750-3841.2009.01100.x
C 2009
Further reproduction without permission is prohibited
and Seidlitz 1992), and selective extraction (Moyler 1993) compared to conventional techniques that used organic solvents including trichloroethylene and dichloromethane known or suspected as carcinogens (Lavin and others 2000; Sherratt and others 2002; Park and others 2007a,b). Considering the potentiality of SC-CO 2 technique in the green tea market, however, information on sensory characteristic is lacking. The researches on decaffeinated green teas applying SC-CO 2 method are mainly limited to instrumental analyses (Cao and others 2000; Chang and others 2000; Ra and others 2001; PervaUzunalic and others 2006; Lee and others 2007; Park and others 2007a,b). Therefore, the main purpose of this study was to investigate the effect of SC-CO 2 decaffeination technique on sensory properties of green teas and also on Korean young consumers’ sensory acceptability. Ultimately, this study was aimed to examine the potentiality of applying SC-CO 2 technique to decaffeinated green tea products.
Materials and Methods Materials Three types of decaffeinated green tea samples (C60, C35, and C10), each with different content of caffeine, and an original green tea (C100), which was not treated by the decaffeinating process, were investigated in this experiment. The 4 samples were infused in 2 different periods (1 and 2 min), resulting in 8 different samples. The descriptions and designation for those 8 samples were given in Table 1. For green tea samples, green tea leaves (2nd crop) cultivated in Bosung (Jeollanamdo, South Korea) in 2004 was used. They were roasted 3 times at 250 ◦ C, rolled twice, and then dried. They were ground and then filtered through a sieve to gain particle size in Vol. 74, Nr. 3, 2009—JOURNAL OF FOOD SCIENCE
S135
S: Sensory & Food Quality
ABSTRACT: Green tea has been widely consumed for its mild flavors and its health benefits, yet caffeine in green tea has been a limitation for those who want to avoid it. The limitation brought increase in need for decaffeinated products in the green tea market. Most of the conventional decaffeination techniques applied in food use organic solvents. However, supercritical carbon dioxide fluid extraction (SC-CO 2 ) method is gaining its intension as one of the future decaffeination methods that overcomes the problems of conventional methods. The purpose of this study was to identify sensory characteristics of decaffeinated green teas applied with SC-CO 2 method and to observe the relationship with consumer acceptability to elucidate the potentiality of applying SC-CO 2 technique in decaffeinated green tea market. Descriptive analysis was performed on 8 samples: green teas containing 4 caffeine levels (10%, 35%, 60%, and 100%) infused at 2 infusing periods (1 or 2 min). It was found that the SC-CO 2 process not only reduced caffeine but also decreased some important features of original tea flavors. Two groups were recruited for consumer acceptability test: one (GP I, N = 52), consuming all types of green teas including hot/cold canned teas; and the other (GP II, N = 40), only consuming the loose type. While GP II liked original green tea the most, GP I liked highly decaffeinated green teas. Although the SC-CO 2 method had limitations of losing complex flavors of green teas, it appeared to have future potential in the decaffeinated green tea market within or without the addition of desirable flavors. Keywords: consumer acceptability, descriptive analysis, PLSR, SC-CO 2 decaffeinated green tea, sensory characteristics
Sensory evaluation of decaffeinated green teas . . .
S: Sensory & Food Quality
the range of 425 to 710 μm. Grinding was performed to enhance oped proper attributes for the samples to be investigated in the exthe decaffeinating process. Original green tea, which was not de- periment. Panelists also defined and selected reference samples for caffeinated, was also ground to have equivalent shape. each of the descriptor. Evaluation procedure was also developed by panelists during the training sessions. Panelists were trained until Decaffeination of green teas consensus agreement was obtained among panelists. When a panGreen teas were decaffeinated using Pilot plant-scale SC-CO 2 elist showed inconsistency with the rest of the panelists, additional extraction system (Thar Technologies, Pittsburgh, Pa., U.S.A.) training was provided. Training sessions were held for 4 d/wk for equipped with an extraction vessel of an internal volume of 12 L, 4 wk and each session took approximately 1 h. following the same method described in Lee and others (2007). Green tea sample preparation. Aliquots (13 g) of green tea Green teas were firstly soaked in 95% (w/w) ethanol (Duksan Pure leaves were put into an Erlenmeyer flask (1 L) and then sealed. The Chemical Co. Ltd., Ansan, Gyeonggido, South Korea), which was outer surface of the flask was heated for 5 s in a water bath (Changused as a cosolvent in the decaffeinating process, and were loaded Shin Scientific Co., Seoul, South Korea) set at 70 ◦ C, which was the into the vessel when the extraction system reached the designated fixed infusing temperature of green teas in this experiment. Adetemperature. To obtain decaffeinated sample C60, the amount of quate temperatures for green teas are known to be 60 to 70 ◦ C in cosolvent used for the process was 2.3 kg of 95% ethanol per Korea (Byun and Han 2004) and 80 to 100 ◦ C in other Southeast 100 kg of CO 2 and was mixed with 1.88 kg of ground green tea Asia countries (Sharma and others 2005), however, our preliminary leaves. To obtain decaffeinated green tea samples C35 and C10, study had shown that the tea prepared at 80 ◦ C was too bitter for 7 kg of 95% ethanol was used. After loading the tea leaves, the vessel original (nondecaffeinated) green tea samples and the tea prepared was closed and the liquid CO 2 contained in the siphoned cylinder at 60 ◦ C was too weak in flavor for decaffeinated green tea samwas pumped through the water bath to maintain the temperature ples, especially for the decaffeinated tea with only 10% of caffeine constantly. When the SC-CO 2 extraction vessel was fully pressured content. After prewarming the surface of the flask, 1 L of filtered to a designated pressure within 1 min, CO 2 was pumped into the (Ceramic Filter System, Supercape, Dalton, Fariey Industrial CeSC-CO 2 extraction vessel, and then the leaves were gone through ramics Ltd., London, U.K.) tap water was poured into the flask, and the decaffeination process. The extraction condition was: 300 bar sealed again. During the infusion of green tea samples, the flask was of pressure; at 70 ◦ C; for 51 min per each cycle; with 1.25 kg/min of put in the water bath to minimize the possible temperature loss. CO 2 flow rate. The extraction time differed as 204 min for green tea Four teas with different content of caffeine were infused for 1 and C60 and C35, and 714 min for green tea C10, which had the low- 2 min, resulting in 8 samples and they were designated as follows: est level of caffeine. These conditions were monitored and auto- C100—1, C100—2, C60—1, C60—2, C35—1, C35—2, C10—1, and matically controlled in the SC-CO 2 extraction system. After each C10—2, respectively (Table 1). After infusion, the green tea samples extraction batch, the samples were removed from the vessel and were filtered through a cloth made of polyethylene and polypropysoaked with the same amount of cosolvent used previously. When lene complex fibers (Dashibag, T&C Electronics, Yiwnag, Gyeongthe extraction was over, aliquots (100 g) of samples were dried in gido, South Korea). The samples were poured into a thermos (1 L, a vacuum oven (IsotempModel 280A, Fisher Scientific, N.J., U.S.A) One touch Apollo, Apollo Corp., Yangju, Geoynggido, South Korea) at 50 ◦ C for 3 h to remove possible residual ethanol in the sample to maintain the temperature before evaluation. and at the same time to minimize the possible loss of tea flavors. Sample presentation. For tasting to evaluate flavor and mouthThe contents of caffeine for C60, C35, and C10 were 61.42%, 35.23%, feel, 125 mL of green tea samples were poured into a smaller inand 12.59%, respectively, after the extraction; for convenience, the dividual thermos (200 mL, IB-020TPY, Sejongisoli Corp., Daegu, contents of caffeine were approximated in the samples designation Korea) from each 1 L thermos right before the evaluation and pre(Table 1). sented simultaneously to each panelist with an individual tasting beaker (50 mL). The presentation order of the sample followed 8 × 8 Descriptive analysis modified William’s Latin square design (Williams 1949) and each The sensory properties of green tea samples were measured sample were coded with 3-digit random numbers. The tempera R using a variation of the quantitative descriptive analysis (QDA ) ture of the green teas at the time of evaluation was 50 ± 2 ◦ C. Filmethod (Stone and Sidel 1992). Panels of 8 judges (22 to 35 years of tered (Ceramic Filter System, Supercape, Dalton, Fariey Industrial age, female) were recruited among graduate students in the Dept. Ceramics Ltd., London, U.K.) and heated tap water (45 ± 2 ◦ C) of Food Science and Engineering at Ewha Womans Univ. (Seoul, was also provided for rinsing the mouth. For evaluation of appearSouth Korea). These panelists were experienced in at least 4 de- ance, all 8 green tea samples were presented in glass cups (diameter scriptive analyses of various food products and had participated in 3 cm, height 6 cm) simultaneously under a light box (Superlightdescriptive analysis of green tea previously. The panelists first de- III, Boteck, Siheong, Gyeonggido) of daylight mode (D65). Samples veloped a list of terms describing the sensory attributes of green used for evaluation of appearance were coded with new 3-digit rantea using various commercial green tea samples. Then they devel- dom numbers different from the number used in evaluation of flavor and mouthfeel. Table 1 --- Descriptions of green tea samples used in the Evaluation procedure. Intensity of each attribute of 8 samples experiment. was evaluated using a 15-point category scale labeled with “weak” Sample Caffeine Infusing period on the point 1 and “strong” on the point 15. The evaluation was identification content (%) (min) carried out over 4 sessions based on a randomized complete block C100—1 100.00 1 design. Within 1 session, first, panelists evaluated flavor and C60—1 61.42 1 mouthfeel attributes in monadic way, one by one and then evalC35—1 35.23 1 uated appearance attributes simultaneously with all the samples. C10—1 12.59 1 During the evaluation of flavor and mouthfeel attributes, a 10-min C100—2 100.00 2 C60—2 61.42 2 break was given in between accessing the first 4 samples and the C35—2 35.23 2 remaining 4 samples to prevent the adaptation and fatigue bias. A C10—1 12.59 2 total of 13 descriptors were used for descriptive analysis and their S136
JOURNAL OF FOOD SCIENCE—Vol. 74, Nr. 3, 2009
Sensory evaluation of decaffeinated green teas . . .
Consumer acceptability test Ninety-two female consumers, who were recruited from the campus of Ewha Womans Univ., participated in consumer test. All participants were frequent green tea consumers who are consuming green tea at least 3 times a week. They were divided into 2 different groups according to their preference of green tea drinks; one (GP I, N = 52) consuming all types of green tea products including hot/cold canned teas and the other (GP II, N = 40) only consuming loose-type green teas infused with hot water. Green tea samples were prepared and presented in the same way as in the descriptive analysis. A bipolar 15-point category scale, labeled with words “dislike very much,” and “like very much” on the left and right end of the scale and “neither nor” in the middle, was used to rate the overall acceptability of the samples. For evaluation, consumers were instructed to pour approximately 35 mL of each sample from the thermos to the individual tasting cup and they were allowed to retaste and change their previous scores. Before the test began, consumers were carefully instructed about the test protocol, use of scale, tasting method, and rinsing procedure.
Statistical analyses Analysis of variance (ANOVA) was applied to the descriptive and affective acceptability ratings. For multiple comparisons between samples for each sensory attribute, Duncan’s multiple range test was used with the significance level of P < 0.05. For descriptive
data, multivariate analysis of variance (MANOVA) was preceded before ANOVAs, confirming the significant differences among the samples (P < 0.001). Principal component analysis (PCA) was performed (using the covariance matrix extraction method) on the mean values of descriptive ratings of attributes for each sample to effectively summarize large number of variable to examine the relationship among sensory attributes and differences among the green tea samples in design. Partial least square-regression (PLSR) was conducted to understand the relationship between descriptive sensory data and consumer acceptability data. When performing PLSR, mean values of descriptive ratings were used as X data set and mean values of consumer acceptability ratings were used as Y data set. All the statistical analyses were conducted using SPSS software (version 12.0 for Windows, SPSS, Chicago, Ill., U.S.A.) and XLSTAT software (version 2007 for Windows, XLSTAT, Addinsoft, Brooklyn, N.Y., U.S.A.).
Results and Discussion Sensory characteristics of decaffeinated green tea The F-values from ANOVA indicating the effects of experimental variables (caffeine content, infusing period, and interaction between the two) on each sensory attributes across the 8 green tea samples for 8 judges were summarized in Table 3. Caffeine content had the strongest effects on all the sensory attributes while infusing period had significant effect only on the “bitter taste,” “metallic,” and appearance attributes including “yellowness,” “turbidity.” The interaction between the 2 variables had a significant effect only on the “yellowness” attribute. The mean scores with the results of the multiple comparisons at 5% significance level for all the samples are given in Table 4. It shows that for “fermented tea,” “alcohol,” “burnt leaf,” and “metallic,” the decaffeinated green tea samples, C60—2, C30—2, C60—1, and C30—1 had significantly higher mean scores than the original green tea samples (C100—1 and C100—2). For the rest of the sensory attributes, the original green tea sample
Table 2 --- Definitions and reference samples for the sensory attributes of green teas. Sensory attributes Yellowness Turbidity Bitter taste Floral
Definitions Intensity of yellowness of green tea Turbidity of green tea Fundamental taste sensation of which caffeine and quinine is typical Aromatics associated with flower such as jasmine
Grassy Fermented tea
Aromatics associated with cut grass Aromatics associated with fermented teas such as oolong tea or black tea
Roasted grain
Aromatics associated with roasted grain
Chestnut shell
Aromatics associated with chestnut shell
Dried straw Alcohol
Aromatics associated with dried straw Aromatics associated with alcohol
Burnt leaf Metallic Astringency
Aromatics associated with burnt leaf Aromatics associated with metals, tinny, and iron The feeling which shrivels the tongue associated with tannins
a
Reference samples
0.05% caffeine (Sigma-Aldrich, Inc., St.Louis, Mo., U.S.A.) solution 10 g jasmine tea (1 tea bag [Chian Jasmine, Amorepacific Co. Jinchun, Chungcheongbukdo, Korea] infused with 200 mL boiling water for 2 min) mixed with 60 g green teaa 10 g Sedum sarmentosum 25 g Canned oolong tea (Dong Suh Oolong tea, Dong Suh Food Co., Ltd., Jinchun, Chungcheongbukdo, Korea) mixed with 45 g green teaa 25 g roasted-corn tea (2 g roasted corn [roasted-corn tea, Dong-suh Food Co., Ltd., Jinchun, Chungdheongbukdo, Korea] infused with 200 mL boiling water for 2 min) mixed with 45 g green teaa 30 g chestnut shell (from Gonju, Supermarket, Youngdongpo, Seoul, Korea) boiled with 500 mL water at low heat for 10 min 10 g dried straw 7 g vodka (Smirnoff vodka red, Smirnoff, London) mixed with 100 g green teaa Burnt leaves Stainless steel spoon 0.1% Tannic acid (Ducksan Pure Chemical Co. Ltd., Ansan, Gyeonggido, Korea) solution
One green tea bag (Shulloc Cha Jinhyang, Amorepacific Co. Jinchun, Chungcheongbukdo, Korea) was infused with 200 mL boiling water for 2 min.
Vol. 74, Nr. 3, 2009—JOURNAL OF FOOD SCIENCE
S137
S: Sensory & Food Quality
definitions and reference samples used in the experiment are listed in Table 2. The tasting was conducted in an isolated sensory evaluation booth under dim red light to avoid possible bias from the differences of appearance among samples. Panelists were instructed to pour approximately 10 mL of samples from the individual thermos into the individual beakers and to sip the whole amount. Panelists rinsed their mouth once before and in between tasting each sample. Appearance attributes were evaluated simultaneously. Each session lasted approximately 30 min.
Sensory evaluation of decaffeinated green teas . . . (C100—1 and C100—2) tended to have a higher mean score than other decaffeinated green tea samples. C100—2 and C100—1 tended to have very similar sensory profile, but C100—2 had higher scores than C100—1 for “bitter taste,” “floral,” and “grassy.” The matrix of mean scores for 13 sensory attributes across 8 judges was analyzed by principal component analysis (PCA) with no rotation. The first 2 principal components (PCs) explained 69.4% and 28.3% of the total variance, respectively (Figure 1). The PC 1 dimension was mainly defined by “roasted grain,” “grassy,” “turbidity,” “floral,” “dried straw,” and “yellowness.” To the lesser
extent, “chestnut shell,” “bitter taste,” “astringency” affected on the PC 1 positively and “alcohol” affected negatively. The original green tea samples (C100—1 and C100—2) were separated from the decaffeinated samples along the PC1 indicating that these decaffeinated samples, which were all plotted on the negative side of PC 1, had much reduced flavor of the previously mentioned sensory attributes than the original. Among decaffeinated samples, highly decaffeinated green teas (C10—1, C10—2) were located the most distant from the original green teas, indicating that the PC 1 dimension separated samples according to the treatment of the decaffeination process and there was significant loss of color and flavor during the decaffeinating process (Table 4, Figure 1). The relatively Table 3 --- F -valuesa for the sensory attributes of green higher “alcohol” intensities of decaffeinated samples have shown teas (8 judges). that there might be some residual of cosolvent (ethanol) from the Caffeine Infusing Caffeine × decaffeination process. Sensory attributes content period infusing period On the other hand, the PC2 dimension was mainly affected by Yellowness 216.27∗∗∗ 71.86∗∗∗ 6.55∗∗∗ “burnt leaf” and also affected by “astringency,” “bitter taste,” “ferTurbidity 71.33∗∗∗ 11.18∗∗∗ 1.70 mented tea,” “metallic,” “chestnut shell,” and “alcohol” to the lesser Bitter taste 73.42∗∗∗ 8.40∗∗ 2.57 extent. The original green tea samples (C100—1 and C100—2) were ∗∗∗ Floral 23.88 2.11 0.06 the most close to the origin of PC2 axis and the highly decaffeinated Grassy 185.10∗∗∗ 1.77 1.21 Fermented tea 187.49∗∗∗ 1.05 1.18 green teas (C10—1 and C10—2) were located on the negative side Roasted grain 31.84∗∗∗ 3.28 1.23 of PC2 different from the semidecaffeinated green teas (C35—1, ∗∗∗ 0.89 1.90 Chestnut shell 166.35 C35—2, C60—1, and C60—2), which were located on the positive Dried straw 44.72∗∗∗ 3.60 1.06 ∗∗∗ side of PC2. The attribute “burnt leaf” was significantly correlated Alcohol 39.18 1.44 0.10 with “fermented tea” (r = 0.70, P < 0.01) and most strong in semideBurnt leaf 44.72∗∗∗ 0.12 0.10 Metallic 18.33∗∗∗ 8.41∗∗ 0.35 caffeinated green tea samples. It was thought that the reduced Astringency 19.81∗∗∗ 0.66 2.31 flavor attributes such as “floral” and “grassy,” which sometimes a∗ P < 0.05, ∗∗ P
