Characterization of light-induced aroma components in green tea beverages using aroma extract dilution analysis with gas chromatography–olfactometry

Abstract

Most green tea beverages are sold in transparent polyethylene terephthalate bottles and are displayed under lighting on the shelves in convenience stores and supermarkets. The exposure of green tea beverages to light sometimes results in off-flavors. The key aroma components contributing to the off-flavors of these beverages are unknown. We identified key aroma components contributing to the off-flavors induced by light exposure using aroma extract dilution analysis with gas chromatography–olfactometry. In addition, odor-active values of these components were calculated based on the threshold and quantitative values. Seven aroma-active components namely, 1-octen-3-one, (Z)-1,5-octadien-3-one, (E,E)-2,4-heptadienal, (E,Z)-2,6-nonadienal, 3-methylnonane-2,4-dione, (E,E)-2,4-decadienal, and trans-4,5-epoxy-(E)-2-decenal, were identified as the key aroma components contributing to the off-flavors induced by light exposure of green tea beverages. These seven components were lipid-derived aroma components, and we believe they are produced from linoleic and linolenic acids in green tea beverages upon exposure to light.

Introduction

Among beverages, the production of green tea beverages in Japan is 2.92 million kL, second only to coffee (3.06 million kL)ⁱ⁾. Most green tea beverages are sold in transparent polyethylene terephthalate (PET) bottles and are displayed on the shelves of convenience stores and supermarkets. With technical advancements in the processing of foods and beverages that have enabled expiration dates of the products to be extended, the light exposure of products in stores has increased. Consequently, concerns regarding the influence of increased light exposure on food and beverages has also increased. Light is known to enhance oxidation of food components, thus causing deterioration of fats and vitamins and changes in color and flavor (Bekőblet, 1990). For example, green tea beverages have been shown to develop off-flavors with strong metallic-like odors after light exposure (Tsuji et al., 2019).

Light-induced changes in fruit juices, dairy products, and drinks have been reported in previous studies. Stark and Forss (1967) identified the odorant contributing to a metallic off-flavor in oxidized butter as 1-octen-3-one, which is probably derived from linoleic acid or arachidonic acids. Umekawa (2000) found metallic off-flavors in grapefruit juice stored in a PET bottle and identified 1-octen-3-one, 3-methoxy-1-butanol, anisaldehyde, and α-cardiol as volatile components. A strong metallic off-flavor in apple juice is formed by photooxidation and a major contributor to this off-flavor was 1-octen-3-one (Hashizume et al., 2007). Exposure of tea leaves to light leads to the production of pentanal, (E)-2-alkenals, C₈-alcohols, bovolides, and dihydrobovolides (Horita, 1987). These aroma components induced by light exposure are suspected to be the photooxidative degradation products of fatty acids and carotenoids (Horita, 1987). However, the contribution of these aroma components to the light-induced off-flavor and metallic-like odor of tea beverages remains unclear.

To the best of our knowledge, a comprehensive study on the light-induced aroma components in green tea beverages has not yet been performed. Consequently, the key aroma components contributing to the off-flavors of these beverages are unknown. Thus, this study aimed to identify the key odorants induced by light exposure in green tea beverages using sensory-directed flavor analyses, such as aroma extract dilution analysis (AEDA) performed using gas chromatography–olfactometry (GC–O). Subsequently, the odor-active values (OAVs) of the light-induced aroma components were calculated based on the threshold and quantitative values.

Materials and Methods

Chemicals Prior to use, all of the solvents were distilled using a three-ball Snyder column concentrator. To identify and quantify the aroma components, 1-octen-3-one, 2-acetyl-1-pyrroline, (Z)-1,5-octadien-3-one, 4-mercapto-4-methyl-2-pentanone, 3-methylnonane-2,4-dione, and trans-4,5-epoxy-(E)-2-decenal were synthesized using the method mentioned in a previous study (Mizukami, 2018). Hotrienol was synthesized from linalyl acylate as described previously (Yuasa and Kato, 2003). β-Damascenone was obtained from Merck Co., Ltd. The other aromatic components listed in Table 1 were obtained from Tokyo Chemical Industry Co., Ltd. N-alkanes used to calculate the retention index of the aroma components were also obtained from Tokyo Chemical Industry Co., Ltd. For quantification of (Z)-1,5-octadien-3-one in tea beverages, an isotopically labeled internal standard [5,6-²H₂]-(Z)-1,5-octadien-3-one was synthesized as described previously (Guth and Grosch, 1990), but the oxidation reaction was performed using N-tert-butylbenzenesulfonamide as a catalyst with N-chlorosuccinimide (Matsuo et al., 2003).

Table 1. Aroma components contributing to the off-flavor identified in the aroma extract isolated from green tea beverage stored in the dark and fluorescent light.

Odorant	Identification					Flavor dilution factor
	Odor quality	Confimation	Fractiona	RIb		Stored in the dark	Exposed to light
				Pure-WAX	DB-5MS
(Z)-4-heptenal	Green, oily	RI, MS, Odor	Rf 0.7–0.9	1245	901	10	10
1-octen-3-one	Mashroom like	RI, MS, Odor	Rf 0.7–0.9	1300	976	10	100
2-acetyl-1-pyrroline	Popcorn like	RI, MS, Odor	Basic	1327	923	100	100
(Z)-1,5-octadien-3-one	Metalic	RI, MS, Odor	Rf 0.7–0.9	1366	981	100	1000
4-mercapto-4-methyl-2-pentanone	Meaty	RI, MS, Odor	Acidic	1372	975	100	100
(Z)-3-hexenol	Green	RI, MS, Odor	Rf 0.3–0.7	1391	858	n.d.d	10
trimethyl pyrazine	Roasty	RI, MS, Odor	Basic	1415	1005	10	10
unknownc	Coffee			1423		10	10
unknownc	Earthy	RI, Odor		1427		10	10
2-ethyl-3,6-dimethylpyrazine	Roasty	RI, MS, Odor	Basic	1438	1082	10	10
methional	Potato	RI, MS, Odor	Acidic	1448	909	100	100
2-ethyl-3,5-dimethylpyrazine	Roasty	RI, MS, Odor	Basic	1453	1086	1000	1000
(E, E)-2,4-heptadienal	fatty	RI, MS, Odor	Rf 0.7–0.9	1474	1013	10	10
2,3-diethyl-5-methylpyrazine	Roasty,earty	RI, MS, Odor	Basic	1487	1160	100	100
unknownc	Earthy			1497			10
unknownc	Fruity			1510		10	10
2-isobutyl-3-methoxypyrazine	Earthy	RI, MS, Odor	Rf 0.3–0.7	1519	1175	10	10
(E)-2-nonenal	Green	RI, MS, Odor	Rf 0.9–1.0	1526	1158	10	100
linalool	Floral	RI, MS, Odor	Rf 0.3–0.7	1540	1101	10	10
(E,Z)-2,6-nonadienal	Green	RI, MS, Odor	Rf 0.7–0.9	1577	1150	100	1000
hotrienol	Earthy, floral	RI, MS, Odor	Rf 0.3–0.7	1601	1100	10	10
2-acetylpyrazine	Roasty,earthy	RI, MS, Odor	Acidic	1604	1019	10	10
phenylacetaldehyde	Floral	RI, MS, Odor	Rf 0.7–0.9	1628	1049	10	10
(E,E)-2,4-nonadienal	Green	RI, MS, Odor	Rf 0.7–0.9	1690	1217	10	10
3-methylnonane-2,4-dione	Green	RI, MS, Odor	Rf 0.3–0.7	1710	1256	1000	1000
2-acetyl-2-thiazoline	Roasty	RI, MS, Odor	Basic	1729	1104	10	10
(E,E)-2,4-decadienal	Fruity	RI, MS, Odor	Rf 0.3–0.7	1799	1319	10	100
β-damascenone	Sweet	RI, MS, Odor	Rf 0.3–0.7	1812	1387	10	10
3-mercapto-1-hexanol	Fruity	RI, MS, Odor	Acidic	1832	1122	100	100
geraniol	Floral	RI, MS, Odor		1836	1250	10	10
α -ionone	Floral, Green	RI, MS, Odor		1858	1422	10	10
maltol	Sweet, Roasty	RI, MS, Odor		1946	1011	100	100
unknownc	Sweet, Grassy			1959		10	10
trans-4,5-epoxy-(E)-2-decenal	Sweet	RI, MS, Odor	Rf 0.3–0.7	1989	1394	100	100
4-hydroxy-2,5-dimethyl-3(H)-furanone	Sweet	RI, MS, Odor		2017	1065	1000	1000
5-ethyl-4-hydroxy-2-methyl-3(2H)-furanone	Sweet	RI, MS, Odor	Rf 0.0–0.3	2034	1159	10	10
unknownc	Spicy			2062		10	10
p -cresol	Spicy	RI, MS, Odor		2074	1063	10	100
eugenol	Spicy	RI, MS, Odor		2148	1353	10	10
2-methoxy-4-vinylphenol	Spicy	RI, MS, Odor		2161	1306	100	100
3-hydroxy-4,5-dimethylfuran-2(5H)-one	Spicy	RI, MS, Odor		2172	1112	100	100
unknownc	Spicy,sweet			2201		10	10
indole	Animal	RI, MS, Odor		2427	1289	100	100
coumarin	Sweet	RI, MS, Odor		2433	1438	10	10
skatole	Animal	RI, MS, Odor		2476	1393	10	10
vanillin	Sweet	RI, MS, Odor		2547	1419	100	100

^a  Acidic, basic, and neutral fractions were isolated from the aroma extract by performing liquid-liquid extraction. The neutral fraction was applied to a column (94 × 16 mm internal diameter [i.d.], 19 mL) containing silica gel (10 g; particle size, 100 µm; TCI). Samples were eluted in one of the following solvents: hexane (60 mL), hexane/dichloromethane (48 mL/12 mL), hexane/dichloromethane (36 mL/24 mL), hexane/dichloromethane (24 mL/36 mL), hexane/dichloromethane (12 mL/48 mL), dichloromethane (60 mL), and ethyl acetate (60 mL). A total of 80 fractions (6 mL) were obtained. Rf (rate of flow) values were determined using thin-layer chromatography with dichloromethane as the solvent.

^b  Retention index (RI) was calculated using a GC-flame ionization detector.

^c  Unidentifed odorant.

^d  “n.d.” indicates that the odorant was not detected using gas chromatography–olfactometry.

Preparation of the green tea beverage Japanese green tea leaves used were mainly from Shizuoka prefecture and were stored at room temperature below 28 °C for approximately 7 to 9 months before use. The moisture content of tea leaves was less than 5 % (w/w). Green tea leaves (80 g) were added to 2400 g of hot water (70 °C) and extracted for 6 min, followed by solid–liquid separation to prepare the green tea extract. After adding 4 g of ascorbic acid to the green tea extract, the pH was adjusted to 5.5, and ion-exchanged water was added to a constant volume to prepare 10 L of green tea beverage. The prepared green tea beverage was sterilized by ultra-heat treatment (130 °C, 0.5 min), filled into 500-mL transparent PET bottles, and subjected to storage experiments. The green tea beverages were prepared three times on three different days. Twenty bottles of tea beverages (10 L total) obtained from one preparation was used for AEDA experiment, while the tea beverages obtained from the other two preparations were used for quantitative analysis.

Storage of green tea beverages Half of the samples (10 bottles) of the prepared tea beverages in PET bottles were stored under continuous exposure to fluorescent light (3000 lx) at 10 °C for 14 days. The others were wrapped with a light-shielding film and allowed to stand in the dark (20 °C) for 14 days.

Isolation of the aroma concentrate of green tea beverages Isolation of the aroma concentrate was performed as described previously (Mizukami, 2022). For the AEDA, aroma components in 10 bottles (5.0 L) of the tea beverage were extracted by vacuum distillation (40 °C, 2.0 kPa, 90 min). The aroma components remaining after vacuum distillation were reextracted with dichloromethane, followed by high-vacuum distillation using a solvent-assisted flavor evaporation unit (SAFE, 35 °C, 10⁻³ Pa). The aromatic extract was concentrated to 1.0 mL with a Snyder column at 45 °C, followed by a nitrogen stream to 0.1 mL. For quantification of the aroma components, 2-acetylpyridine (2.58 mg/L, 0.02 mL) and [5,6-²H₂]-(Z)-1,5-octadien-3-one (26.4 mg/L, 0.1 mL) were added as internal standards to 1.0 L of the green tea beverage after confirming that the tea beverage used in this experiment did not contain 2-acetylpyridine. Subsequently, the aroma components were isolated by vacuum distillation, followed by high-vacuum distillation using the SAFE unit using the same method as described above.

GC–O The GC–O analyses were performed using a gas chromatograph type GC-4000 (GL sciences, Tokyo, Japan) with two types of fused-silica capillary columns namely, InertCap PureWAX (30 m × 0.25 mm internal diameter [i.d.], 0.25-µm film thickness; GL sciences) and DB-5MS (30 m × 0.25 mm i.d., 0.25 µm film thickness; J & W Scientific, Folsom, CA). The concentrate (1 µL) was injected into the column in the splitless injection mode, and the injector temperature was maintained at 230 °C. The oven temperature was maintained at 40 °C for 1 min, increased by 4 °C/min to 240 °C, and maintained at this temperature for 20 min. Helium (over 99.999 %) was used as the carrier gas under a flow rate of 1.0 mL/min (40 °C). For the GC–O, the effluent was split 1:1 (by volume) evenly at the end of the column using two deactivated fused-silica capillaries (1.2 m × 0.25 mm i.d.). One part was directed to a flame ionization detector (FID) maintained at 230 °C, and the other was directed to a sniffing port (OP275; GL science, Tokyo, Japan) maintained at 230 °C.

AEDA AEDA is a useful tool for screening the contribution of odorants to the overall aroma. Therefore, concentrated aroma extracts were diluted stepwise with dichloromethane. Sensory evaluation of each diluted solution was performed using GC–O by an experienced panelist until no odorant was detected at the sniffing port. The panelist was trained before the AEDA, as described previously (Vene et al., 2013). The original odor concentrate was diluted stepwise to 1:10, 1:100, 1:1000, and 1:10000 with dichloromethane. Next, 1-µL aliquots of each fraction were analyzed using the InertCap PureWAX capillary column for determining flavor dilution (FD) factors. By definition, the FD factor obtained for each odorant in the AEDA was equal to the highest dilution at which the odorant was perceived at the sniffing port. The GC–O analysis was repeated three times, and detection of the aroma component twice or more was considered to indicate that it contributed to the overall aroma.

Gas chromatography–mass spectrometry (GC–MS) For the identification and quantification of aroma components, mass spectra were generated using a mass spectrometer (JMS-Q1500GC; Jeol) in the electron impact ionization mode at 70 eV. The source temperature was set to 200 °C. A scan range of m/z 30–400 at 2.05 scans·s⁻¹ was employed. Two types of fused-silica capillary columns (InertCap PureWAX and DB-5MS) were used for identification. Additionally, a fused-silica capillary column, DB-225MS (30 m × 0.25 mm i.d., 0.25 µm film thickness; J & W Scientific, Folsom, CA), was used quantify aroma compounds. Helium (> 99.999 %) was used as the carrier gas at a constant flow rate of 1.0 mL/min (constant). The oven program and injector were identical to those described in the GC–O section.

Enrichment of aroma components for identification After the AEDA, all the concentrates and dilutions used for AEDA were combined and then, concentrated to 2 mL using a three-ball Snyder column concentrator. Basic volatiles were isolated from the concentrate using 1 M hydrochloric acid (3 × 2 mL). The aqueous acidic extract (6 mL) was neutralized with aqueous sodium hydroxide and extracted with dichloromethane (3 × 6 mL). This extract was washed with brine (3 × 18 mL), dried over anhydrous sodium sulfate, and finally, concentrated to 0.1 mL (basic fraction, B) using a purified nitrogen stream. Acidic volatiles were isolated from the organic phase using 1 M sodium hydroxide (3 × 2 mL), neutralized with aqueous hydrochloric acid, and extracted with dichloromethane (3 × 6 mL). This extract was washed with brine (3 × 18 mL), dried over anhydrous sodium sulfate, and finally, concentrated to 0.1 mL (acidic fraction, A). The residual organic phase was washed with brine (3 × 2 mL), dried over anhydrous sodium sulfate, and finally, concentrated to 0.1 mL (neutral fraction). The neutral fraction was added to a column (94 × 16 mm i.d.) containing silica gel (10 g; particle size, 100 µm; TCI). Elution was performed using hexane (60 mL), hexane/dichloromethane (48 mL/12 mL), hexane/dichloromethane (36 mL/24 mL), hexane/dichloromethane (24 mL/36 mL), hexane/dichloromethane (12 mL/48 mL), dichloromethane (60 mL), and ethyl acetate (60 mL). The eluate from the column was fractionated into 6-mL aliquots to obtain a total of 80 fractions. Each fraction was concentrated to approximately 0.5 mL in a nitrogen stream, and the retention factor (Rf) value of each fraction was calculated using thin-layer chromatography (silica gel; film thickness, 0.25 mm; Merck) with dichloromethane as the solvent. The fractions with similar Rf values were mixed and concentrated to approximately 0.1 mL in a stream of nitrogen. Each component was identified by comparing its Kovats GC retention index (RI), mass spectrum, and odor quality with those of an authentic component.

Quantification of aroma components The aroma components contributing to the off-flavor caused by light were quantified from the extracted ion peak areas obtained using mass chromatography. GC–MS was operated in the selected ion mode (SIM). The extracted ions were monitored in the ranges listed in Table 2. We used an internal standard to calibrate a mixture of aroma compounds using the same method utilized for tea beverages, taking into account the extraction rate and volatility. The calibration factors were determined in a mixture of an aroma compound and an internal standard compound mixed in equal amounts by weight and were calculated as the ratio of the extracted ion peak area of the internal standard to the extracted ion peak area of the aroma compounds (Table 2). The peak intensity of (Z)-1,5-octadien-3-one at m/z 124 was weak, thus indicating a risk of erroneous peak detection. Therefore, the stable isotope [5,6-²H₂]-(Z)-1,5-octadien-3-one was used as an internal standard for quantification.

Table 2. Method validation and analytical quality control based on the SANTE/11813/2017 guidelines for quantifying aroma components in tea beverages.

Component	Selected iona m/z	Linearity		Additive recovery rateb (%)	Rangec (µg/L)	LOQd (µg/L)	Truenesse RSD (%)	Repeatabilityf RSD (%)	Intermediate precisiong RSD (%)
		Calibration factor	Correlation
(Z)-1,5-octadien-3-one	124	0.981	0.999	90–116	0.05–0.35	0.05	102	4	7
(E,E)-2,4-heptadienal	110	1.58	0.990	84–118	2.06–22.1	2.06	101	8	9
(E, Z)-2,6-nonadienal	138	31.3	0.993	74–116	0.02–1.02	0.02	94	14	15
(E,E)-2,4-decadienal	152	5.26	0.995	71–114	0.05–1.05	0.05	98	8	11
trans -4,5-epoxy-(E)-2-decenal	68	0.331	0.989	75–118	0.07–1.07	0.07	94	9	14
1-octen-3-one	70	0.546	0.993	79–116	0.01–0.41	0.01	100	4	12
3-methy lnonane-2,4-dione	170	22.2	0.989	81–117	0.02–0.42	0.02	99	9	12
(Z)-3-hexenol	82	1.96	0.996	78–112	0.15–1.15	0.15	101	9	12
(E)-2-nonenal	70	1.51	0.999	76–113	0.02–0.22	0.02	95	10	11
p-cresol	108	0.388	0.999	79–109	0.04–1.04	0.04	93	7	10
(E,E)-2,4-nonadienal	81	0.218	0.986	80–113	0.0005–0.10	0.0005	95	8	10
[5,6-2H2]-(Z)-1,5-octadien-3-oneh	126
2-acetylpyridineI	121

^a  Selected ion chromatography yield peak that did not overlap with other peaks.

^b  Additive recovery rate obtained by adding aroma components to tea beverages at five concentrations for each aroma component, followed by analysis of five replicates per level.

^c  Range satisfied the condition for which trueness was confirmed to be in the additive recovery range of 70–120 % with a precision RSD < 20 % based on the additive recovery test.

^d  Limit of quantification (LOQ) selected as the lowest level for which trueness was confirmed to be 70–120 % and precision RSD < 20 % based on the additive recovery test.

^e  Trueness: the average recovery rate for the additive recovery test.

^f  Five replicates per day

^g  Five replicates per day on three different days

^h  Internal standard for quantification of (Z)-l,5-octadien-3-one

^I  Internal standard for quantification of aroma compounds, excluding (Z)-l,5-octadien-3-one

The validity of quantitative analysis of aroma components in the tea beverages was evaluated according to SANTE/1183/2017 guidelines based on linearity, additive recovery rate, range, limit of quantification, trueness, repeatability, and intermediate precision (Table 2). The light-exposure treatment was repeated twice. The isolation of aroma extracts from green tea beverages was repeated three times for each treatment, and the average value and standard deviation obtained from six analyses were determined. The OAV of the aroma components was calculated using the quantitative value/odor-threshold value. The odor-threshold values (Table 3) were obtained from the studies by Flaig et al. (2022) and Czerny et al. (2008).

Table 3. Odor-threshold and odor-activity values of selected odorants in green tea beverage stored in the dark and exposed to fluorescent light.

Odorant	Odor threshold in water µg/L	Odor activity value
		Stored in the dark	Exposed to light
(Z)-1,5-octadien-3-one*	0.00034	164.	695.
(E,E)-2,4-heptadienal*	0.032	105.	300.
(E, Z)-2,6-nonadienal**	0.0045	15.2	26.7
(E, E)-2,4-decadienal**	0.027	2.01	9.59
trans -4,5-epoxy-(E)-2-decenal**	0.038	1.97	6.13
1-octen-3-one**	0.016	0.93	2.38
3-methylnonane-2,4-dione*	0.046	1.11	2.13
(Z)-3-hexenol**	0.39	0.89	0.70
(E)-2-nonenal**	0.19	0.22	0.47
p -cresol**	3.90	0.01	0.03
(E, E)-2,4-nonadienal**	0.062	0.01	0.01

The odor-threshold value in water was obtained from the studies by *Flaig et al. (2020) and **Czerny et al. (2008).

Quantification of linoleic and linolenic acids Fatty acids, including linoleic and linolenic acids, were extracted from 25 mL of tea beverages using 20 mL of hexane (twice) and concentrated. Glutaric acid was added as an internal standard, followed n-butanol derivation, according to the method described by Shirai (2019) and analyzed using GC–MS. GC–MS was operated in SIM. The extracted ions were monitored at m/z values of 115 (internal standard), 261, and 263. The calibration factors were determined in a mixture of equal amounts by weight of the fatty acid and internal standard compound and were calculated as the ratio of the extracted ion peak area of the internal standard to the extracted ion peak area of the fatty acids. The light-exposure treatment was repeated twice. Isolation of linoleic and linolenic acids from green tea beverages was repeated three times for each treatment, and the average value and standard deviation obtained from six analyses were determined.

Statistical analysis To determine the differences in mean values between the beverages stored in light and the dark, the significance probability value was calculated using Welch’s t test with Microsoft Excel (Microsoft 365 MSO).

Results and Discussion

After 14 days of exposure to light, the green tea beverages not only had a strong metallic odor but also had fatty, fishy, and plastic-like odors. These findings confirmed that the aroma of the tea deteriorated after exposure to light. However, the flavors of green tea beverages stored in the dark were comparable to those of green tea beverages immediately after production. Therefore, the prepared tea beverages were sufficient to address the purpose of this study.

The distillate of the volatile fraction isolated from the tea beverages elicited the overall aroma of the beverage when a drop of the distillate was evaluated on a filter paper, suggesting that the extraction procedure used was suitable for isolating the entire fraction of odor-active components. Screening of the concentrated aroma distillate using GC–O followed by AEDA revealed 46 odor-active components in the FD factor range of 10–1 000. The first indication of the odorant eliciting the odor-detected area was obtained by comparing the linear retention indices and perceived odor qualities with data from an in-house database of more than 700 food aroma components identified in several previous studies. Although the mass spectra were obtained in the MS-EI mode, these data alone can be misleading because the odorant may be present in trace amounts (but can be detected by sniffing) while the mass spectrum of a coeluting major compound is generated. Thus, the distillate of the volatile fraction was fractionated by liquid–liquid extraction, followed by chromatography using silica gel. To identify the key aroma components, the retention indices, odor quality, and mass spectra were compared with those of the authentic reference components. This process revealed 39 odor-active components. The odor-active components listed in Table 1 are the key components that contribute to the aroma of teas (Mizukami, 2020). Thus, the off-flavor of green tea beverages induced by light exposure is caused by changes in the balance of well-known components.

The three components with the highest FD factors in dilutions isolated from green tea beverages stored in the dark were 2-ethyl-3,5-dimethylpyrazine, 3-methylnonane-2,4-dione, and 4-hydroxy-2,5-dimethyl-3(H)-furanone (Table 1). In addition to these three compounds, (Z)-1,5-octadien-3-one and (E,Z)-2,6-nonadienal were also identified as having high FD factors in the distillate isolated from green tea beverages stored under light. These five components have been previously identified as key odorants in distillates isolated from green tea leaves and infusions (Mizukami, 2020). Six components, 1-octen-3-one, (Z)-1,5-octadien-3-one, (E)-2-nonenal, (E,Z)-2,6-nonadienal, (E,E)-2,4-decadienal, and p-cresol, were identified as key odorants with high FD factors of 100 or higher in the dilutions isolated from tea beverages stored under exposure to light. In addition, the FD factors of these six components in the tea beverages stored under exposure to light were higher than those of the components in tea beverages stored in the dark.

The aroma components resulting from the breakdown of citral were α, p-dimethylstyrene, p-cymen-8-ol, p-cymene, p-methylacetophenone, o- and p-cresol via acid cyclization (Schieberle, 1988). Although α, p-dimethylstyrene, p-cymen-8-ol were not detected, p-methyl acetophenone and o-cresol were detected in the aroma extract isolated from tea beverages (pH 5.5) stored under light. Therefore, it is possible that p-cresol was produced due to the breakdown of citral in light.

In contrast, the other five components namely, 1-octen-3-one (Ho, 2015), (Z)-1,5-octadien-3-one (Ullrich, 1988), (E)-2-nonenal (Kuroda, 2003), (E,Z)-2,6-nonadienal (Ullrich, 1988), and (E,E)-2,4-decadienal (Nielsen G.S, 2004), are well known lipid-derived flavor-contributing components. This finding implied that lipid-derived components of tea beverages are produced upon exposure to light. In addition to these five components, the lipid-derived flavor-contributing components (Z)-3-hexenol (Ho, 2015), (E,E)-2,4-heptadienal (Nielsen, 2004), (E,E)-2,4-nonadienal (Warner, 2001), 3-methylnonane-2,4-dione (Guth, 1989), and trans-4,5-epoxy-(E)-2-decenal (Kumazawa, 2006) were also detected using the GC–O analysis. However, the FD factors of these components, which indicate their degree of contribution to the aroma, were comparable between the dark-stored and fluorescent light-exposed tea beverages. Thus, the levels of these five components were also increased by fluorescent light exposure, and although they showed no difference in the FD factors, they could conceivably contribute to the off-flavors induced by light exposure.

Although dilution to odor-threshold techniques, such as AEDA, are useful screening methods for selecting aroma components that potentially contribute to a food aroma, these data are not directly related to the food aroma itself because, during GC–O, the entire amount of each aroma component is completely volatilized. In AEDA using GC–O, the volatility of aroma components from the matrix, such as water, is not taken into account (Schuh, 2006). Furthermore, although the FD factor is useful as a reference value, as the FD factor has a large error margin, it is not an absolute value and is used as a screening analysis. Therefore, quantitative measurements and correlations with the odor threshold are necessary for further experiments to link analytical data with the overall tea aroma perception. Thus, based on the results of AEDA, this study quantitatively analyzed 11 components that were considered to contribute to the off-flavor of green tea beverages induced by exposure to light. A total of 11 components, including six components (1-octen-3-one, (Z)-1,5-octadien-3-one, (E)-2-nonenal, (E,Z)-2,6-nonadienal, (E,E)-2,4-decadienal, and p-cresol), had high FD factors. The other five lipid-derived flavor-contributing components were (Z)-3-hexenol, (E,E)-2,4-heptadienal, (E,E)-2,4-nonadienal, 3-methylnonane-2,4-dione, and trans-4,5-epoxy-(E)-2-decenal.

The volatility, extraction rate, and solubility of individual aroma components vary greatly. Additionally, aroma components tend to be lost during extraction, concentration, dilution, and GC analysis. Moreover, simultaneous analysis of aroma components tends to have large errors. Therefore, it is desirable to use a stable isotope as an internal standard for quantitative analysis. However, given the difficulty in obtaining stable isotopes for all of the aroma components in the current study, 2-acetylpyridine was applied as the internal standard for quantification. Subsequently, the validity of the quantitative analysis was evaluated according to the SANTE/11813/2017 guidelines. Table 2 presents the linearity, range, limit of quantification, trueness, repeatability, and intermediate precision of the study data. The validation confirmed that the data achieved an additive recovery of 70–120 % and a precision with a relative standard deviation (RSD) < 20 %. Moreover, comparing the data obtained in Table 2 and Fig. 1, the quantitative values for the aroma components exceeded the limit of quantification (LOQ). Collectively, the quantitative analysis of aroma components in this study satisfied the SANTE/11813/2017 guidelines.

Fig. 1

Comparison of the levels of selected odorants in tea beverage stored in the dark and exposed to fluorescent light. The asterisk indicates a significant difference with *p < 0.05 and **p < 0.01 (Welch’s t test).

Quantification analyses revealed that the levels of nine components, except (Z)-3-hexenol and (E,E)-2,4-nonadienal, in green tea beverages were increased by exposure to light (Fig. 1). Among them, green tea beverages contained light-induced (Z)-1,5-octadien-3-one and (E,E)-2,4-decadienal more than four times compared to those in tea stored in the dark. The OAV of the aromatic components were also calculated (Table 3). The OAVs of (E)-2-nonenal and p-cresol, which increased with light exposure, were less than 1, thereby suggesting a minor contribution of these components to the aroma of the tea beverages. In contrast, the OAVs of 1-octen-3-one, (Z)-1,5-octadien-3-one, (E,E),2,4-heptadienal, (E,Z)-2,6-nonadienal, 3-methylnonane-2,4-dione, (E,E)-2,4-decadienal, and trans-4,5-epoxy-(E)-2-decenal were greater than 1, thereby suggesting that they strongly contribute to the light-induced off-flavor of tea beverages. Among these, the OAVs of (Z)-1,5-octadien-3-one and (E,E),2,4-heptadienal were much higher than those of the other aroma components. Therefore, these two lipid-derived aroma components considerably contribute to the deterioration of odor induced by the exposure of green tea beverages to light.

Green tea leaves contain palmitic, stearic, oleic, linoleic, and linolenic fatty acids (Shirai, 2019). Although these fatty acids are poorly soluble in water, they may be present in small amounts in green tea beverages. Therefore, the fatty acids contained in tea beverages were analyzed, and only linoleic and linolenic acids were detected. Quantitative analysis of these two fatty acids confirmed that linoleic acid and linolenic acid levels in tea beverages decreased upon exposure to light (Fig. 2). Linoleic and linolenic acids dissolved in a small amount of ethanol (equivalent to 1 mg/L) were added to the green tea beverages. The tea beverages were then stored and exposed to light. The (E,E)-2,4-heptadienal level increased with the addition of both linoleic and linolenic acids. The addition of linoleic acid to tea tended to increase (E)-2-nonenal, (E,E)-2,4-decadienal, and trans-4,5-(E)-2-decenal levels upon exposure to light. Moreover, the addition of linolenic acid to the tea beverages tended to increase the levels of 1-octen-3-one, (Z)-1,5-octadien-3-one, (E,Z)-2,6-nonadienal, and 3-methylnonane-2,4-dione (data not shown). These results imply that 1-octen-3-one, (Z)-1,5-octadien-3-one, (E,E)-2,4-heptadienal, (E,Z)-2,6-nonadienal, 3-methylnonane-2,4-dione, (E,E)-2,4-decadienal, and trans-4,5-epoxy-(E)-2-decenal in green tea beverages are mainly formed from fatty acids upon exposure to light. These seven aroma components contribute to the off-flavors induced by exposure to light. Among these, (Z)-1,5-octadien-3-one and (E,E)-2,4-heptadienal contribute significantly to the off-flavors. Fluorescent light was used for this experiment. Thus, these aroma components are expected to be generated via oxidation of fatty acids, which is promoted by ultraviolet radiation-induced free radical generation as well as oxidation reaction resulting from photoexcitation of chlorophyll (Rontani, 2022). Therefore, to maintain green tea beverage aroma, it is important to suppress the deterioration of fatty acids.

Fig. 2

Comparison of the levels of linoleic acid and linolenic acid in green tea beverage stored in the dark and exposed to fluorescent light. The double asterisks indicate a significant difference with p < 0.01 (Welch’s t test).

(Z)-1,5-Octadien-3-one is present in green, oolong, and black teas. It has been identified as a significant odorant in all types of tea (Mizukami 2020, Ho et al., 2015). Its levels increase with the maturation of tea shoots and with covering culture of tea plants (Mizukami, 2020). (Z)-1,5-Octadien-3-one in tea shoots is believed to be derived from linolenic acid (Ho et al., 2015). Our findings also suggested that it was generated from linolenic acid due to the exposure of green tea beverages to light. (Z)-1,5-Octadien-3-one has a strong metallic odor. Evaluation of green tea using GC–O with the original aroma simultaneously inputted into the sniffing port confirmed that adding (Z)-1,5-Octadien-3-one at high concentrations intensified the metallic odor of green tea (Hattori et al., 2005). Therefore, (Z)-1,5-Octadien-3-one is considered to strongly influence the metallic flavor induced by exposure of green tea beverage to light. (E,E)-2,4-heptadienal produces fatty and fishy odors. It enhances the fishy off-flavors of milk (Venkateshwarlu et al., 2004). Fresh tea leaves contain high amounts of (E,E)-2,4-heptadienal. However, its levels decrease during the green tea manufacturing process (Flaig et al., 2020). It is believed to be the main aroma component responsible for the fishy odor of green tea, which is significantly reduced by roasting (Mizukami, 2015). However, its levels increase more easily during storage than those of the other lipid-derived aroma components because of lipid oxidation (Horita, 1987). Therefore, upon exposure of green tea beverages to light, the levels of (E,E)-2,4-heptadienal increased significantly more than those of the other lipid-derived aroma components, exhibiting a high OAV, and contributed to the fishy odor.

In conclusion, with the growing popularity of green tea beverages, most of these beverages are sold in transparent PET bottles and displayed under lighting on shelves in convenience stores and supermarkets. However, light exposure can cause deterioration of the aroma of green tea beverages. This study used AEDA with GC–O to identify the aroma compounds contributing to the off-flavors induced by exposure of green tea beverages to light. Subsequent quantitative analysis was performed to determine the OAVs based on the threshold values. This study revealed that lipid-derived aroma components in green tea beverages contribute to the off-flavors induced by exposure to light. Suppressing the decomposition of fatty acids is important to prevent the deterioration of green tea beverages as a result of light exposure.

Funding This research was funded by the Kirin Beverage Co., Ltd. and the National Agricultural and Food Research Organization.

Conflict of interest There are no conflicts of interest to declare.

Abbreviations

AEDA

Aroma extract dilution analysis

FD

Flavor dilution

FID

Flame ionization detector

GC–O

Gas chromatography–olfactometry

GC–MS

Gas chromatography–mass spectrometry

OAVs

Odor-active values

PET

Polyethylene terephthalate

Rf

Retention factor

RI

Retention index

SAFE

Solvent-assisted flavor evaporation

i.d.

internal diameter

SIM

Selected ion mode

References

•  Bekőblet,  M. (1990). Light Effects on Food. J. Food Prot., 53, 430–440. doi: 10.4315/0362-028X-53.5.430
•  Czerny,  M.,  Christlbauer,  M.,  Christlbauer,  M.,  Fischer,  A.,  Granvogl,  M.,  Hammer,  M.,  Hartl,  C.,  Hernandez,  N.M., and  Schieberle,  P. (2008). Re-investigation on odour thresholds of key food aroma compounds and development of an aroma language based on odour qualities of defined aqueous odorant solutions. Eur. Food Res. Technol., 228, 265–273. doi: 10.1007/s00217-008-0931-x
• European Commission (2017). Guidance document on analytical quality control and validation procedures for pesticide residues analysis in food and feed. Directorate General for Health and Food Safety, SANTE/11813/2017.
•  Flaig,  M.,  Qi,  S.,  Wei,  G.,  Yang,  X., and  Schieberle,  P. (2020). Characterization of the key odorants in a high-grade Chinese green tea beverage (Camellia sinensis; Jingshan cha) by means of the sensomics approach and elucidation of odorant changes in tea leaves caused by the tea manufacturing process. J. Agric. Food Chem., 68, 5168–5179. doi: 10.1021/acs.jafc.0c01300
•  Guth,  H., and  Grosch,  W. (1989). 3-Methylnonane-2,4-dione – an intense odour compound formed during flavour reversion of soya-bean oil. Lipid/Fett, 91, 225–230. doi: 10.1002/lipi.19890910604
•  Guth,  H., and  Grosch,  W. (1990). Deterioration of soy–bean oil: quantification of primary flavor compound using a stable isotope dilution assay. Lebensim. Wiss. Technol., 23, 513–522.
•  Kumazawa,  K.,  Wada,  Y., and  Masuda,  H. (2006). Characterization of epoxydecenal isomers as potent odorants in black tea (Dimbula) infusion. J. Agric. Food Chem., 54, 4795–4801. doi: 10.1021/jf0603127
•  Kuroda,  H.,  Furusho,  S.,  Maeba,  H., and  Takashio,  M. (2003). Characterization of factors involved in the production of 2 (E)-nonenal during mashing. Biosci. Biotechol. Biochem., 67, 691–697. doi: 10.1271/bbb.67.691
•  Hashizume,  M.,  Gordon,  M.H., and  Mottram,  D.S. (2007). Light–Induced off–flavor development in cloudy apple juice. J. Agric. Food Chem., 55, 9177–9182. doi: 10.1021/jf0715727
•  Hattori,  S.,  Takagaki,  H., and  Fujimori,  T. (2005). Identification of volatile compounds which enhance odor notes in Japanese green tea using the OASIS (Original aroma simultaneously input to the sniffing port) method. Food Sci. Tehcnol. Res., 11, 171–174. doi: 10.3136/fstr.11.171
•  Ho,  C–T.,  Zheng,  X., and  Li,  S. (2015). Tea aroma formation. Food Sci. Huma. Well., 4, 9–27. doi: 10.1016/j.fshw.2015.04.001
•  Horita,  H. (1987). Off-flavor components of green tea during preservation. JARQ, 21, 192–197.
•  Matsuo,  J.,  Iida,  D.,  Yamanaka,  H., and  Mukaiyama,  T. (2003). N-tert-butylbenzenesulfenamide–catalyzed oxidation of alcohols to the corresponding carbonyl compounds with N-chlorosuccinimide. Tetrahedron, 59, 6739–6750. doi: 10.1016/S0040-4020(03)00479-4
•  Mizukami,  Y. (2015). Changes in key odorants in Sen-cha caused by roasting. Tea Res. J. (Chyagyo Kenkyu Houkoku), 119, 13–20. (in Japanese) doi: 10.5979/cha.2015.119_13
•  Mizukmai,  Y. (2018). Characteristic aroma of sencha produced in the mountainous area. Tea Res. J. (Chyagyo Kenkyu Houkoku), 126, 9–15. (in Japanese) doi: 10.5979/cha.2018.126_9
•  Mizukami,  Y. (2020). Key odorants in tea detected by a gas chromatography-olfactometry. Tea Res. J. (Chyagyo Kenkyu Houkoku), 130, 1–16. (in Japanese) doi: 10.5979/cha.2020.130_1
•  Mizukami,  Y. (2022). Analytical methods for isolating aroma compounds from tea. Tea Res. J. (Chyagyo Kenkyu Houkoku), 134, 45–57. (in Japanese)
•  Nielsen,  G.S.,  Larsen,  L.M., and  Poll,  L. (2004). Formation of volatile compounds in model experiments with crude leek (Allium ampeloprasum Var. Lancelot) enzyme extract and linoleic acid or linolenic acid. J. Agric. Food Chem., 52, 2315–2321. doi: 10.1021/jf030600s
•  Rontani,  J.-F. (2022). Use of gas chromatography-mass spectrometry techniques (GC-MS, GC-MS/MS and GC-QTOF) for the characterization of photooxidation and autooxidation products of lipid of autotrophic organisms in environmental samples. Molecules, 27, 1629. doi.org/10.3390/molecules27051629
•  Schuh,  C., and  Schieberle,  P. (2006). Characterization of the key aroma compounds in the beverage prepared from Darjeeling black tea: quantitative differences between tea leaves and infusion, J. Agric. Food Chem., 54, 916–924. doi: 10.1021/jf052495n
•  Shirai,  N. (2019). Gas chromatographic analysis of organic acid in Japanese green tea leaves. J. Oleo Sci., 68, 1271–1277. doi: 10.5650/jos.ess19208
•  Schieberle,  P. and  Grosch,  W. (1988). Identification of potent flavor compounds formed in an aqueous lemon oil/citric acid emulsion. J. Agric. Food Chem., 36, 797–800, doi: 10.1021/jf00082a031
•  Stark,  W. and  Forss,  D.A. (1962). A compound responsible for metallic flavor in dairy products. I. isolation and identification. J. Dairy. Res., 29, 173–180. doi: 10.1017/S0022029900017787
•  Tsuji,  K.,  Karasawa,  A.,  Yoshida,  S.,  Koizumi,  H., and  Wasano,  N. (2019). Packaged tea beverage with suppressed photo deterioration. Japan Tokkyo Koho, Patent No. 6576202, Aug. 30. (in Japanese)
•  Ullrich,  F. and  Grosch,  W. (1988). Identification of the most intense odor compounds formed during autoxidation of methyl linolenate at room temperature. J. Am. Oil Chem. Soc., 65, 1313–1317. doi: 10.1007/BF02542413
•  Umekawa,  T. (2000). Identification of the compounds which are the cause of light-induced off-flavor in grapefruit drink and estimation of their precursors. Soft Drink Technical Report 132; Japan Soft Drink Association: Tokyo, Japan, pp. 257–262. (in Japanese)
•  Vene,  K.,  Seisonen,  S.,  Koppel,  K.,  Leitner,  E., and  Palme,  T. (2013). A method for GC-olfactometry panel training. Chem. Percept., 6, 179–189. doi: 10.1007/s12078-013-9156-x
•  Venkateshwarlu,  G.,  Let,  M.B.,  Meyer,  A.S., and  Jacobsen,  C. (2004). Modeling the sensory impact of defined combination of volatile lipid oxidation product on fishy and metallic off-flavors. J. Agric. Food Chem., 52, 1635–1641. doi: 10.1021/jf0351321
•  Warner,  K.,  Neff,  W.E.,  Byrdwell,  C., and  Gardner,  H.W. (2001). Effect of oleic and linoleic acids on the production of deep-fried odor in heated triolein and trilinolein. J. Agric. Food Chem., 49, 899–905. doi: 10.1021/jf000822f
•  Yuasa,  Y., and  Kato,  Y. (2003). A practical and convenient synthesis of hotrienol, an excellent fruity smelling compound. J. Agric. Food. Chem., 51, 4036–4036. doi: 10.1021/jf030031o
• i)  http://www.j-sda.or.jp/images_j/stories/con05_about_jsda/2021jsda_databook.pdf (Aug 9, 2023) (in Japanese)