Aroma Components Analysis and Origin Differentiation of Black Tea Based on ATD-GC-MS and E-Nose

Abstract

Black tea (Fuyun 6) samples collected from three regions, Youxi, Fu’an, and Datian, were analysed by automatic thermal desorption-gas chromatography–mass spectrometry (ATD-GC–MS) combined with the electronic nose (E-nose) technique to investigate the aroma composition differences between black teas from different regions. The response surface methodology was used to optimize the ATD conditions for extracting the aroma components from the black tea. The results revealed that the optimal conditions for aroma component accumulation from black tea samples included a sample weight of 2.8 g, an adsorption time of 39 min, an adsorption temperature of 75 °C, and a cold trap temperature of −30 °C. The ATD-GC–MS analyses identified a total of 71 aroma components in the black tea samples, of which 31 were utilized to differentiate the origins of the black teas. Additional aroma activity analyses indicated that benzyl alcohol, linalool, hexanal, octanal, and nonanal had odour activity values (OAVs) greater than 10. Additionally, the OAV of decanal exceeded 100, indicating its significant contribution to the aroma profile of Fuyun 6 black tea. The E-nose results demonstrated the ability to differentiate the black tea samples from the three different origins. This study successfully identified the specific aroma substances associated with different tea origins, providing valuable insights into the aroma characteristics of black teas from various regions.

1. Introduction

Black tea originated in Fujian and is the main tea exported from China and consumed globally [1,2,3,4]. With the enhancement of peoples’ concepts of health care, black tea has been consumed all over the world. Tea aromas vary based on the region of origin, the season, and the cultivars, which affect the contents of the aroma substances present, the component ratio, and the characteristics of the glycoside hydrolases; there are significant differences in the aromas of black teas when they are fully expressed [5,6,7]. Black teas present characteristics such as floral, fruity, sweet, and potato aromas [8,9]. The aroma substances of teas have been separated and identified, including aldehydes, alcohols, ketones, esters, acids, and 10 other categories of compounds [10]. It is crucial to determine a tea’s quality grade based on its aroma profile, which is also the most important factor in capturing and cultivating consumer loyalty, and is important in determining tea prices on the market [11]. Therefore, the identification and analysis of the aroma characteristics of black teas benefit production and sales and protect the origin and the construction of regional brands of tea.

At present, the main means of detecting and evaluating the aroma characteristics of tea are through experimental instruments, such as the electronic nose (E-nose) and gas chromatography–mass spectrometry (GC–MS) [7]. E-nose technology is fast and efficient but does not address quantitative questions [12]. GC–MS technology is qualitatively and quantitatively accurate, and its combination with the E-nose takes better advantage of the two instruments [13,14]. Automatic thermal desorption (ATD) has been applied to qualitative and quantitative analyses of food volatiles [15,16,17]. However, it is rarely used in the field of tea detection. Additionally, due to differences in the origins, processes, and aroma substances associated with different black teas, and even differences in the GC–MS conditions and chromatographic columns, the optimal parameters for extracting samples are also different, so method optimization is needed before using ATD-GC–MS [18,19]. Xu optimized their parameters through single-factor experiments to determine the aroma compounds of ripened and raw pu-erh tea [3]. Yang determined the best parameters for green tea aroma identification and determined the important aroma substances in green tea through single-factor experiments, orthogonal experiments, and the chemometric method [17]. Therefore, in this study, the conditions for the ATD detection of black teas are optimized, and the teas are identified and analysed by GC–MS, then verified and distinguished with the E-nose in order to better realize the use of the ATD-GC–MS detection method with black teas.

In this study, black teas (Fuyun 6) collected from Youxi (25°50′–26°26′ N, 117°48′–118°39′ E), Fu’an (26°41′–27°24′ N, 119°23′–119′ E), and Datian (24°06′ N, 116°19′ E) in the Fujian Province were used as the research samples (samples YF, FF, and DF). ATD-GC–MS, combined with E-nose stoichiometric studies, was used to study the aromas of and identify the different metabolites in black teas and provide technical support for the construction of a black tea public brand and quality evaluation.

2. Materials and Methods

2.1. Materials

In this study, 27 black tea samples were collected from the three production areas of Youxi, Fu’an, and Datian. The experimental tea samples were black teas provided by the Youxi Guangxing Tea Group. The rest of the experimental Fuyun 6 tea samples were produced by the Guangxing Tea Group (abbreviation: Guangxing), Yunfu Tea Group (abbreviation: Jinyunfu), Shenxi Tea Professional Cooperative (abbreviation: Minxin), Fujian New Tanyang Tea Group (abbreviation: New tanyang), The Fu’an Tea Bureau of Fu’an (abbreviation: Tea bureau) and the Fu’an Nongken Group (abbreviation: Nongken). As shown in Supplementary Table S1, the teas were prepared according to Chinese national standard GB/T 35810-2018 [20].

2.2. Pharmaceuticals and Instruments

A gas chromatograph–mass spectrometer (Shimadzu Corporation, Kyoto, Japan), automatic thermal desorption-desorption instrument with an adsorption tube (Chengdu Kelin Analysis Technology Co., Ltd., Chengdu China), a QC-1S Atmospheric Sampler (Beijing Kean Labour Protection Technology Co., Ltd., Beijing China), an E-nose system (iNose, Shanghai Angshen Intelligent Technology Co., Ltd., Shanghai China), ether (Sinopharm Group Chemical Reagent Co., Ltd., Shanghai China), and ethyl caprate (Sinopharm Group Chemical Reagent Co., Ltd., Shanghai China) were used in this study.

2.3. Method

2.3.1. Electronic Nose Detection Method

The E-nose method was based on Yan’s method and was slightly modified, and each sample was tested three times [7]. Based on the preliminary tests and comprehensive considerations, 3 g of the tea sample was placed in a 60 mL headspace bottle and held in an oven at 50 °C for 50 min; the E-nose gas flow rate was 800 mL/min; the sampling time was 5 min; the waiting time was 10 s; and the sample washing time was 120 s. The humidity of the room was controlled at 36% ± 1%, and the temperature was controlled at 25 °C ± 1 °C.

2.3.2. ATD-GC–MS Detection Method

The samples were ground and passed through a 40-mesh sieve, and 15 μL of ethyl caprate with a concentration of 100 ppm was added to 2.5 g of the sample, sealed, and stored in a dark, dry place for later use.

The GC–MS method was based on Wang’s method with slight modifications [21]. The GC column was an Elite-FFAP column (30 m × 0.25 mm × 0.25 μm, Perkin Elmer, MA, USA); the GC heating conditions were 50 °C for 5 min, 3 °C/min to 125 °C, hold for 2 min, and 5 °C/min to 180 °C, hold for 3 min, 15 °C/min to 230 °C, hold for 5 min. High purity He (99.999%) was used as the carrier gas with a flow rate of 1.0 mL/min. The split ratio was 1:40. The inlet temperature was 240 °C. The ionization mode of the MS was electron impact (EI). The temperatures of the interface and ion source were 280 °C and 230 °C, respectively. The acquisition mode was full scan.

The MS conditions included an electron bombardment (EI) ion source, an electron energy of 70 eV, an ion source temperature 230 °C, a mass spectrometry transmission line temperature 250 °C, and a mass scanning range of m/z 45–500.

2.3.3. Qualitative and Quantitative Analyses of the Aroma Substances

The GC–MS total ion chromatograms and mass spectra of the aroma components in the black teas were measured with the GC–MS conditions noted above. Using the NIST standard spectral library, each mass spectrum in the total ion flux map was searched based on the retention time, the characteristic ions, the relative abundances, etc. [22]. The identifications were based on matching degrees greater than 80%. The CAS number for each aroma component was referred to and then compared with the data in the original literature to finally identify the compound. The volatile contents of different origins were calculated with the internal standard method, and the units were 10 μg/kg [23]. The volatile contents of the optimization experiments were calculated by the peak area method [17].

2.3.4. Single-Factor Experimental Optimizations

Adsorption Grams of Tea Samples

The adsorption times were 30 min, the adsorption temperature was 70 °C, and the other conditions remained unchanged. With 0.5, 1.5, 2.5, 3.5, and 4.5 g of tea samples, the total peak areas and the types of aroma compounds were detected, and the effects of the tea sample sizes on detection of the black teas by GC–MS were investigated.

Optimization of the Adsorption Time

The tea sample weighed 2.5 g, the adsorption temperature was 70 °C, and the other conditions remained unchanged. The total peak areas and the types of aroma compounds were detected with adsorption times of 10, 20, 30, 40, and 50 min.

Optimization of the Adsorption Temperature

The tea samples weighed 2.5 g, the adsorption time was 30 min, and the other conditions remained unchanged. The total peak areas and the aroma compounds were detected with adsorption temperatures of 40, 50, 60, 70, and 80 °C.

Optimization of the Cold Trap Temperature

The tea samples weighed 2.5 g, the adsorption time was 30 min, the adsorption temperature was 70 °C, and the other conditions remained unchanged. The sum of the peak areas and the types of aroma compounds were detected with temperatures of −30, −20, and −10 °C for the cold trap.

2.3.5. Response Surface Optimization Experiment

The response surface optimization experiments were based on Wang’s method and slightly modified [24]. Based on the single factor experiments, the software design response surface experiment was designed, and the factors and levels are shown in

Table 1. Factors and levels of response surface experiment.

Factor	Level
	−1	0	1
A/Tea samples (g)	1.5	2.5	3.5
B/Adsorption time (min)	20	30	40
C/Adsorption temperature (°C)	60	70	80

. The optimized ATD-GC–MS conditions were selected, and the reproducibilities of the black tea samples were determined with repeatability tests.

2.4. Statistical Analyses

Excel 2016 was used for data sorting, to calculate the means and standard deviations, to generate the bar charts and line charts, and to line analyse the graphs; SPSS 20.0 software was used for the significant difference analyses; Design Expert 8.0.6 statistical software was used for the response surface experiments; and https://www.genescloud.cn/ (accessed on 6 July 2023) was used for the radar chart analysis. The heatmap analyses were performed with https://www.metaboanalyst.ca/ (accessed on 6 July 2023). SIMCA 14 was used for opls-da analyses, and the Chinese medicine chromatographic fingerprint similarity evaluation system was used for fingerprint comparisons.

3. Results and Discussion

3.1. One-Factor Optimization of the ATD Conditions

Optimization of the black tea sample amounts was crucial for achieving the optimal conditions for aroma component accumulation. Figure 1A illustrates that a sample amount of 0.5 g resulted in low contents of the aroma components being adsorbed through ATD and a limited variety of compounds, so it did not reflect the true aromas of the samples. As the sample amount was increased, both the total peak areas and the number of aroma compounds exhibited significant increases, demonstrating increasing adsorption of the aroma compounds. When the sample amount reached 2.5 g, the total peak areas and the number of aroma compounds reached their maximum values. However, beyond 3.5 g, the total peak areas started to decline, and the number of aroma compounds reached saturation. This decrease in the adsorption efficiency was attributed to the reduced headspace volume, which altered the equilibrium coefficients for the thermal desorption and headspace. In conclusion, the optimal sample amount was determined to be 2.5 g.

The adsorption efficiencies of the aroma components were positively correlated with the adsorption times [25]. As shown in Figure 1B, the total amount of aroma compounds increased rapidly with an increase in the adsorption time from 10 to 50 min. The maximum adsorption of the aroma varieties was attained at 20 and 40 min, but the long-term adsorption-induced chemical reactions among the components [26] were detrimental to the thermal desorption coating. In conclusion, the optimal adsorption time was 20 min.

The aroma characteristics of black teas depend on the compositions of the volatile and semivolatile compounds. These compounds are easily decomposed and modified during the adsorption process. Traditional adsorption procedures were undergone in a temperature range from 60 °C to 70 °C [17,25]. As shown in Figure 1C, when the adsorption temperature was 40 °C, the total amounts and varieties of the aroma compounds were the smallest. The temperature was too low to extract the aroma compounds from the matrix [27,28]. With increases in the adsorption temperature, the total amount and varieties of the aroma compounds increased rapidly due to accelerated volatilization and enhanced adsorption efficiencies of the aroma compounds [29]. In conclusion, the optimal adsorption temperature was 60 °C.

The cold trap temperature was another key parameter for thermal desorption. The cold trap temperature and desorption at high temperature reduced the systematic errors and improved the sensitivity of the analyses [30]. As shown in Figure 1D, the cold trap temperature had a minor impact on the total amounts and varieties of the aroma compounds. The optimal cold trap temperature was −30 °C because the low temperature reduced the errors, minimized band broadening, and improved the sensitivity [27].

3.2. Response Surface Optimization of the ATD Conditions

Response surface optimization was carried out based on the one-factor-at-a-time results. The sample amounts were 1.5, 2.5, and 3.5 g, the adsorption times were 20, 30, and 40 min, and the adsorption temperatures were 60, 70, and 80 °C. The cold trap temperature was −30 °C in the response surface tests. The results are shown in

Table 2. Response surface test design and results.

Test Number	Factor			Total Peak Area (Y1)	Number of Aroma Compounds (Y2)
	Tea Samples	Adsorption Time	Adsorption Temperature
	(A)/g	(B)/min	(C)°C
1	1.5	20	70	80.98	80
2	3.5	20	70	98.09	86
3	1.5	40	70	99.67	84
4	3.5	40	70	99.97	92
5	1.5	30	60	80.65	82
6	3.5	30	60	86.9	84
7	1.5	30	80	86.5	84
8	3.5	30	80	102.18	92
9	2.5	20	60	86.48	85
10	2.5	40	60	100.79	85
11	2.5	20	80	90.75	87
12	2.5	40	80	100.58	93
13	2.5	30	70	100.97	88
14	2.5	30	70	98.08	90
15	2.5	30	70	102.34	95
16	2.5	30	70	102.34	93
17	2.5	30	70	106.9	93

.

Multiple regression fitting was applied to the test results in Table 2, and the sum of the black tea peak areas (Y₁), the types of aroma compounds (Y₂), and the tea sample (A), adsorption times (B), and adsorption temperatures were obtained (C). The quadratic model for the factors is shown in S2.

The coefficients and polynomials of the regression equation were tested for significance, and the results are shown in

Table 3. Analysis of variance for Box–Behnken test results based on sum of peak areas.

Source	Degrees of Freedom	Sum of Squares	Mean Square	F Value	p Value	Significance
Model	1011.65	9	112.41	9.7	0.0034	**
A	193.45	1	193.45	16.69	0.0047	**
B	249.87	1	249.87	21.55	0.0024	**
C	79.32	1	79.32	6.84	0.0346	*
AB	70.64	1	70.64	6.09	0.0429	*
AC	22.23	1	22.23	1.92	0.2087
BC	5.02	1	5.02	0.43	0.5317
A2	179.02	1	179.02	15.44	0.0057	**
B2	3.63	1	3.63	0.31	0.5934
C2	180.53	1	180.53	4.24	0.0056	**
Residual	81.16	7	11.59
Lack of fit	40.57	3	13.52	1.33	0.3814
Pure error	40.59	4	10.15
Total dispersion	1092.81	16
	R2	0.925		R2Adj	0.8302

Note: * indicates that the significant level is above 95%, ** indicates that the significant level is above 99%.

and

Table 4. Analysis of variance for Box–Behnken test results based on the number of aroma compounds.

Source	Degrees of Freedom	Sum of Squares	Mean Square	F Value	p Value	Significance
Model	295.67	9	32.85	6.61	0.0105	*
A	72	1	72	14.48	0.0067	**
B	32	1	32	6.44	0.0388	*
C	50	1	50	10.06	0.0157	*
AB	1	1	1	0.2	0.6673
AC	9	1	9	1.81	0.2204
BC	9	1	9	1.81	0.2204
A2	72.52	1	72.52	14.59	0.0065	**
B2	19.46	1	19.46	3.92	0.0884
C2	19.46	1	19.46	3.92	0.0884
Residual	34.8	7	4.97
Lack of fit	4	3	1.33	0.17	0.9093
Pure error	30.8	4	7.7
Total dispersion	330.47	16
	R2	0.8947		R2Adj	0.7593

Note: * indicates that the significant level is above 95%, ** indicates that the significant level is above 99%.

.

Table 3 shows that the quadratic equation Y₁ was extremely significant (p < 0.01). In the primary term, the sample amount A and the adsorption time B reached very significant levels, and the adsorption time C reached a significant level (p < 0.05). In the quadratic term, the sample amount A and the adsorption time C reached very significant levels, while the adsorption time B was not significant. In the interaction term, AB reached a significant level, but BC and AC were not significant, indicating that the influence of the three factors on the sum of the peak areas exhibited a simple linear relationship. It can be seen from the regression equation that the total coefficient of determination for the model was R² = 0.925, the adjusted R² = 0.8302, and the predicted value was correlated with the experimental value, indicating that the model fit the experimental data well and that the experimental design was reliable. Using this equation to simulate the real three factors and the three levels of analysis was feasible [30]. The lack of a fit term was not significant (p = 0.3814), which proved that the regression equation (Y₁) fit the experimental data well, and that the regression model can be used to replace the experimental points to analyse and predict the experimental results [31].

Table 4 shows that the quadratic equation model of Y₂ reached significance (p < 0.05). In the first-order term, the first-order terms for the adsorption time B and the adsorption temperature C reached significant levels, and the first-order term for the sample size A reached a very significant level (p < 0.01). In the quadratic term, the tea sample A reached a very significant level, but the adsorption time B and the adsorption temperature C were not significant. Among the interaction terms, AB, BC, and AC were not significant, indicating that the effects of the three factors on the types of aroma compounds did not follow a simple linear relationship. It can be seen from the regression equation that the total coefficient of determination for the model was R² = 0.8947, the R²_Adj = 0.7593, and the predicted value was correlated with the experimental value, indicating that the model fully fit the experimental data and that the experimental design was reliable. This equation was used to simulate the real three factors, and the three-level analysis was feasible [30]. The lack of fit was not significant (p = 0.9093), which proved that the regression equation (Y₂) fit the experimental data well, and that the regression model can be used to replace the real experimental points and analyse and predict the experimental results [31].

Combined with the quadratic polynomial equations Y₁ and Y₂, one of the three factors, tea samples A, the adsorption time B, and the adsorption time C, were fixed at the centre levels, and the corresponding surface graph of the two factors was drawn [32]. The response surface diagram intuitively showed the effects of the interactions between various factors on the total peak area and the types of aroma compounds.

The results of the response surface and contour line, optimized for the two quadratic regression equations Y₁ and Y₂, are shown in Figure 2 and Figure 3. The openings of the corresponding surface maps for the six two-factor factors were all facing down, and the response surface and contour lines were all facing downwards. The lines were all elliptical, there was a pole in the selected range, the slope was steep, and the centre of the minimum ellipse represented the maximum value and was near the selected range of the figure, indicating that the response value may have a maximum value within the horizontal design range [10].

Figure 2 shows that the total peak area increased with an increasing adsorption time, and that the tea samples and adsorption temperature increased first and then decreased. The curve for the corresponding surface graph of the AC factor was steep, the contour line was elliptical, the interaction between the two factors was strong, and the impact was significant [33]. Figure 3 shows that, with increases in the tea sample sizes, the adsorption times, and the adsorption temperatures, the aroma compounds that were detected first increased and then decreased. Consistent with the Y₁ quadratic regression equation, the curve for the corresponding surface graph of the AC factor was steeper than the curves for AB and BC. The comprehensive analysis showed that the adsorption temperature was the most significant factor that affected the detection of black teas by ATD-GC–MS, and the contour line was steeper, followed by the tea sample.

The ranges of the three factors were set within the factor level range of the experiment: tea sample sizes were 1.5–3.5 g, the adsorption times were 20–40 min, and the adsorption temperatures were 60–80 °C. The sum of the peak areas and the types of aroma compounds were maximized. The optimal conditions for ATD-GC–MS detection of the black teas were as follows: the tea samples weighed 2.85 g, the adsorption times were 39.25 min, and the adsorption temperatures were 75.19 °C. With these conditions, the sum of the peak areas and the types of aroma compounds that were detected reached 105.82 × 10⁶ and 94.23, respectively. Based on the feasibility for actual operation, the conditions for ATD-GC–MS detection of the black teas were revised as follows: the tea samples weighed 2.8 g, the adsorption times were 39 min, and the adsorption temperatures were 75 °C. With the modified detection conditions, experimental verification showed that the actual measured peak area sum and the number of aroma compounds were 95.48 × 10⁶ and 96, and that the measured values were 90.23% and 101.87% of the theoretical values. This showed that the regression equation was reasonable and feasible for analysing and predicting the aroma components of the black teas extracted by ATD-GC–MS.

3.3. Analysis of Aroma Characteristics of Black Tea from Different Regions

A total of 71 volatile aroma compounds in the black tea samples from different regions were analysed by ATD-GC–MS, including 19 alcohols, 12 aldehydes, 5 esters, 24 hydrocarbons, 5 ketones, a phenol, a furan, a nitrogen compound, an acid, and 2 ethers. As shown in

Table 5. The aroma components of black tea from different regions.

No	Time	CAS	Compound	Aroma Substance Content (10 μg/kg)			VIP
				FF	YF	DF
1	2.55	590-86-3	Butanal, 3-methyl-	77.18 ± 35.75 b	84.03 ± 36.6 ab	45.31 ± 17.29 a	1.18
2	2.63	96-17-3	Butanal, 2-methyl-	103.24 ± 54.57 b	87.38 ± 37.43 ab	47.69 ± 11.62 a	1.03
3	2.85	616-25-1	1-Penten-3-ol	39.34 ± 11.44 b	20.9 ± 14.83 a	13.33 ± 4.47 a	1.31
4	2.93	142-82-5	Heptane	1.76 ± 2.95 a	6.39 ± 6.26 a	4.73 ± 2.84 a	0.81
5	3	110-62-3	Pentanal	24.99 ± 15.61 a	17.15 ± 12.25 a	24.42 ± 19.92 a	0.61
6	3.25	109-60-4	Propyl acetate	0.71 ± 1.11 a	0.03 ± 0.08 a	2.75 ± 7.42 a	1.06
7	3.72	624-92-0	Dimethyl disulfide	0.92 ± 0.85 a	1.23 ± 0.57 a	3.05 ± 4.66 a	1.07
8	3.95	105-46-4	sec-Butyl acetate	13.84 ± 11.01 a	6.69 ± 8.48 a	9.04 ± 7.04 a	0.64
9	4.11	108-88-3	Toluene	20.21 ± 13.75 b	8.05 ± 6.65 a	6.90 ± 3.95 a	1
10	4.26	71-41-0	1-Pentanol	13.67 ± 4.73 b	4.56 ± 3.44 a	4.57 ± 1.91 a	1.39
11	4.38	1576-96-1	(Z)-2-Penten-1-ol	25.44 ± 13.84 b	8.80 ± 6.71 a	9.16 ± 5.81 a	1.04
12	4.86	66-25-1	Hexanal	123.74 ± 80.19 b	53.1 ± 31.17 a	51.42 ± 24.88 a	1.15
13	5.28	123-86-4	Acetic acid, butyl ester	12.55 ± 11.62 a	14.43 ± 28.63 a	1.88 ± 1.44 a	0.99
14	5.6	7580-85-0	2-Tert-Butoxyethanol	2.61 ± 1.10 a	2.09 ± 1.10 a	2.65 ± 4.20 a	0.93
15	6.05	98-01-1	Furfural	2.15 ± 2.38 a	1.36 ± 1.35 a	0.97 ± 0.92 a	0.48
16	6.45	928-96-1	(Z)-3-Hexen-1-ol	82.91 ± 97.73 a	70.31 ± 51.05 a	41.7 ± 23.43 a	0.4
17	6.71	95-47-6	o-Xylene	45.47 ± 18.83 a	55.17 ± 65.59 a	32.79 ± 18.10 a	0.92
18	6.78	106-42-3	p-Xylene	23.92 ± 8.92 b	19.35 ± 17.84 ab	9.13 ± 5.42 a	1.05
19	6.83	928-95-0	(E)-2-Hexen-1-ol	10.26 ± 10.88 a	4.55 ± 4.53 a	2.82 ± 1.87 a	0.84
20	6.91	111-27-3	1-Hexanol	34.55 ± 16.29 b	14.14 ± 6.72 a	10.86 ± 3.71 a	1.29
21	7.32	104-76-7	1-Hexanol, 2-ethyl-	190.48 ± 125.07 ab	295.21 ± 153.66 b	57.42 ± 49.37 a	1.36
22	7.33	143-08-8	1-Nonanol	2.69 ± 1.31 b	0.09 ± 0.19 a	1.63 ± 2.91 ab	1
23	7.6	108-94-1	Cyclohexanone	17.27 ± 6.79 b	15.62 ± 10.9 b	4.97 ± 3.30 a	1.15
24	7.86	111-71-7	Heptanal	18.52 ± 6.30 a	12.21 ± 12.31 a	22.09 ± 26.31 a	1.06
25	8.61	2437-95-8	alpha-Pinene	3.26 ± 2.02 a	2.23 ± 1.02 a	3.41 ± 2.45 a	0.91
26	10.07	127-91-3	beta-Pinene	5.41 ± 5.40 a	2.49 ± 3.14 a	3.09 ± 3.73 a	1.00
27	10.39	13475-82-6	Heptane, 2,2,4,6,6-pentamethyl-	40.15 ± 18.69 a	30.96 ± 15.88 a	20.65 ± 19.18 a	0.96
28	10.62	3777-69-3	Furan, 2-pentyl-	14.29 ± 5.22 b	7.29 ± 4.89 a	3.78 ± 4.60 a	1.29
29	10.73	108-67-8	Mesitylene	4.92 ± 2.7 b	2.68 ± 1.36 a	0.66 ± 0.99 a	1.26
30	10.85	1120-21-4	Undecane	5.78 ± 2.19 a	3.93 ± 1.92 a	3.70 ± 3.70 a	0.89
31	11.13	124-13-0	Octanal	17.95 ± 15.67 a	10.06 ± 6.08 a	9.00 ± 4.47 a	1.02
32	11.51	142-62-1	Hexanoic acid	7.59 ± 10.36 a	13.73 ± 8.01 a	5.87 ± 6.46 a	0.64
33	11.64	62183-79-3	2,2,4,4-Tetramethyloctane	5.23 ± 2.37 a	3.66 ± 1.32 a	3.43 ± 1.63 a	0.97
34	11.88	5989-27-5	D-Limonene	29.06 ± 51.06 a	14.01 ± 15.64 a	29.33 ± 52.02 a	0.41
35	12.5	100-51-6	Benzyl alcohol	108.98 ± 75.17 a	133.16 ± 98.66 a	49.37 ± 33.99 a	0.85
36	12.69	2167-14-8	1-Ethylpyrrole-2-carbaldehyde	30.97 ± 25.21 a	62.23 ± 31.96 a	32.13 ± 22.57 a	1.06
37	12.78	62185-53-9	Nonane, 5-(2-methylpropyl)-	4.91 ± 7.35 a	3.69 ± 3.33 a	1.93 ± 3.36 a	0.24
38	12.94	78-59-1	Isophorone	122.25 ± 38.81 b	201.69 ± 87.86 a	48.20 ± 23.02 a	1.59
39	13.28	5989-33-3	(Z)-Linalool oxide (furanoid)	65.54 ± 40.73 a	29.73 ± 16.38 a	45.03 ± 40.41 a	0.84
40	13.51	111-87-5	1-Octanol	7.43 ± 7.86 a	7.76 ± 9.63 a	2.10 ± 2.10 a	0.73
41	13.56	471-01-2	3-Cyclohexen-1-one, 3,5,5-trimethyl-	9.47 ± 4.38 a	9.61 ± 8.11 a	9.20 ± 4.96 a	0.75
42	13.82	34995-77-2	(E)-Linalool oxide (furanoid)	155.8 ± 64.67 b	86.82 ± 50.34 a	60.56 ± 35.55 a	1.15
43	14.28	78-70-6	Linalool	130.1 ± 73.42 b	53.94 ± 29.87 a	48.03 ± 24.44 a	1.09
44	14.41	124-19-6	Nonanal	59.29 ± 22.65 a	49.85 ± 14.86 a	47.23 ± 24.92 a	0.91
45	14.58	5337-72-4	Cyclohexanol, 2,6-dimethyl-	39.08 ± 23.49 a	35.09 ± 18.39 a	36.53 ± 17.13 a	0.9
46	14.71	99-87-6	p-Cymene	3.52 ± 3.21 a	1.72 ± 1.31 a	1.1 ± 0.98 a	0.96
47	14.87	1960/12/8	Phenylethyl Alcohol	63.13 ± 31.9 b	55.44 ± 27.21 ab	30.66 ± 12.51 a	0.89
48	15.76	464-49-3	(+)-2-Bornanone	52.32 ± 55.15 a	77.60 ± 69.51 a	29.13 ± 24.17 a	0.79
49	16.48	112-54-9	Dodecanal	3.90 ± 3.71 a	2.65 ± 1.91 a	6.89 ± 4.94 a	1.15
50	16.54	39028-58-5	(E)-Linalool oxide (pyranoid)	9.94 ± 9.06 ab	28.36 ± 29.04 b	7.27 ± 4.31 a	1.06
51	16.71	39028-58-5	Linalool oxide	29.27 ± 21.08 b	2.88 ± 8.63 a	6.71 ± 6.98 a	1.1
52	16.76	15356-70-4	Menthol	6.97 ± 13.83 a	26.6 ± 29.22 a	13.01 ± 15.25 a	0.7
53	16.92	53398-84-8	Butanoic acid, 3-hexenyl ester, (E)-	7.45 ± 4.77 b	1.35 ± 2.26 a	0.77 ± 1.67 a	1.17
54	17.02	275-51-4	Azulene	10.92 ± 5.54 b	10.69 ± 3.65 b	1.63 ± 3.84 a	1.27
55	17.24	12-40-3	Dodecane	13.19 ± 6.90 a	16.67 ± 5.88 a	21.58 ± 9.14 a	0.94
56	17.29	119-36-8	Methyl salicylate	87.8 ± 51.2 b	38.01 ± 18 a	37.91 ± 39.09 a	1.05
57	17.44	116-26-7	Safranal	10.60 ± 7.48 a	9.20 ± 4.27 a	10.99 ± 5.50 a	0.77
58	17.57	112-31-2	Decanal	36.46 ± 25.77 a	29.67 ± 15.70 a	55.78 ± 36.06 a	1.13
59	18.03	432-25-7	beta-Cyclocitral	19.89 ± 9.46 a	16.67 ± 7.12 a	15.08 ± 8.58 a	0.9
60	19.04	106-24-1	Geraniol	294.94 ± 269.96 a	260.02 ± 117.84 a	194.93 ± 135.83 a	0.58
61	22.65	4506-36-9	1,5,8-Trimethyl-1,2-dihydronaphthalene	0.59 ± 1.77 a	2.80 ± 2.37 ab	5.04 ± 1.87 b	1.15
62	23.52	17699-14-8	alpha-Cubebene	0.50 ± 0.54 a	0.62 ± 0.87 a	0.80 ± 0.79 a	0.12
63	24.79	475-20-7	Longifolene	5.20 ± 2.56 a	9.00 ± 4.12 b	1.93 ± 1.35 a	1.56
64	24.96	469-61-4	Cedrene	21.03 ± 10.23 a	47.08 ± 20.71 b	30.8 ± 13.45 ab	1.36
65	25.03	87-44-5	Caryophyllene	6.80 ± 5.47 a	6.04 ± 2.12 a	8.46 ± 5.30 a	0.83
66	25.26	79120-98-2	beta-Cedrene	5.64 ± 8.51 a	15.52 ± 5.95 b	14.44 ± 4.13 b	1.11
67	25.34	470-40-6	cis-Thujopsene	1.17 ± 1.52 a	3.25 ± 1.95 a	2.38 ± 2.34 a	1.03
68	26.12	14901-07-6	beta-Ionone	34.46 ± 19.15 a	28.14 ± 20.71 a	36.68 ± 15.08 a	0.58
69	26.4	128-37-0	Butylated Hydroxytoluene	5.95 ± 5.26 a	6.23 ± 4.31 a	10.99 ± 8.55 a	0.85
70	26.47	16982-00-6	Cuparene	0.19 ± 0.58 a	3.84 ± 2.18 b	6.86 ± 4.78 b	1.3
71	26.84	1560-78-7	2-Methyltetracosane	1.63 ± 3.09 a	2.96 ± 2.11 a	2.51 ± 3.25 a	0.66

Note: The values in the table are the mean ± standard deviation. Different lowercase letters after the data in the same column indicate significant differences.

, 31 aroma compounds (VIP > 1) played significant roles in the differentiation of the black tea samples.

Seventy-one common aroma components were used as dependent variables, and the aroma compounds from different regions were used as independent variables for OPLS-DA. As shown in Figure 4A, the samples from DF were distributed in the third quadrant, the samples from FF were distributed in the fourth quadrant, and the samples from YF were mainly distributed in the first quadrant. Only two samples from YF were in the first quadrant. These results show that the black tea samples from different regions were effectively distinguished. A cross-validation model with 200 permutations and tests (Figure 4B) was used to prove the rationality of the OPLS-DA of the black tea samples from FF, YF, and DF. The fitting parameters for the independent variable fitting index R²X were 0.523, the dependent variable fitting index R²Y was 0.881, and the model prediction index Q² was 0.704. This showed that the OPLS-DA discriminant model was reliable. The intercepts of the Q² regression line relative to the y axis were all less than 0, indicating that the OPLS-DA discriminant model was not overfitted and that the model was reliable (R² = 0.461, Q² = −0.341). All squared points (Q²) were below the original point of Q² on the right side of Figure 4B, and the intersection of the Q² regression line with the vertical line occurred below zero, indicating the validity of the model.

The GC–MS ion chromatographic data of a total of 27 samples were imported into the Chinese Medicine Chromatographic Fingerprint Similarity Evaluation System (2012). Taking the chromatographic peaks of samples DF1, FF1, and YF1 as reference peaks corresponding to their origin, the time width was set to 0.1. The total control fingerprint R (Figure 5) was generated after automatically correcting and matching the retention times of multiple common chromatographic peaks.

A heatmap was used to comprehensively analyse the different aroma compounds in the black teas from the three regions (Figure 6). Sample FF was a separate category, and samples DF and YF were classified into one category. This indicated that samples from DF and YF had the closest similarity with each other, which is consistent with their geographic proximity [34]. As shown in Figure 6, area A contained 25 prominent aroma components from sample FF, including 1-nonanol, linalool oxide, beta-pinene, (Z)-linalool oxide, sec-butyl acetate, nonanal, (E)-linalool oxide, furfural, furan, 2-pentyl-hexanal, methyl salicylate, linalool, etc. Area B showed higher contents of 15 aroma components in sample YF, including (E)-linalool oxide, beta-cedrene, cedrene, menthol, cis-thujopsene, etc. The dominant aroma components in sample DF, including dodecanal, caryophyllene, decanal, dodecanal, and propyl acetate, are shown in Area C.

Since the flavour dilution factor is not a direct measure of the odour potency of the aroma compounds in foods, the odour activity values (OAVs), based on accurate quantitative data, are needed to assess the contributions of the flavour compounds [35]. Identified odours with high OAVs can be used as objective metrics to assess differentiation in the flavours of food. If the OAV of a single aroma component is greater than one, it is considered to have an influence on the tea aroma. If the OAV is greater than 10, it is considered to make a significant contribution to the overall aroma of the tea [36,37]. The OAVs of the black teas from the three regions were calculated, and the results are summarized in

Table 6. OAV analysis of black tea from different origins.

Compound	Aroma Threshold (μg/kg)	OAV			Aroma Description
		FF	YF	DF
1-Penten-3-ol	400	0.10 ± 0.03 b	0.05 ± 0.04 a	0.03 ± 0.01 a	fruity, green
Benzyl alcohol	2.54	42.91 ± 29.59 a	52.43 ± 38.84 a	19.44 ± 13.38 a	sweet, floral
Phenylethyl Alcohol	390	0.16 ± 0.08 b	0.14 ± 0.07 ab	0.08 ± 0.03 a	floral
Linalool	3.8	34.24 ± 19.32 b	14.19 ± 7.86 a	12.64 ± 6.43 a	Lavender
Geraniol	20	14.75 ± 13.50 a	13.00 ± 5.89 a	9.75 ± 6.79 a	rose scent
beta-Cyclocitral	5	3.98 ± 1.89 a	3.33 ± 1.42 a	3.02 ± 1.72 a	fruity
Pentanal	12	2.08 ± 1.30 a	1.43 ± 1.02 a	2.04 ± 1.66 a	Almond, Malt
Butanal, 3-methyl-	9	8.58 ± 3.97 b	9.34 ± 4.07 ab	5.03 ± 1.92 a	apple like
Hexanal	4.5	27.50 ± 17.82 b	11.80 ± 6.93 a	11.43 ± 5.53 a	green, herbal
Heptanal	10	1.85 ± 0.63 a	1.22 ± 1.23 a	2.21 ± 2.63 a	sweet herbal scent
Octanal	0.5	35.9 ± 31.34 a	20.12 ± 12.16 a	18.00 ± 8.94 a	Sweet orange, honey
Nonanal	1	59.29 ± 22.65 a	49.85 ± 14.86 a	47.23 ± 24.92 a	rose, citrus scent
Decanal	0.1	364.60 ± 257.70 a	296.70 ± 157.00 a	557.80 ± 360.60 a	Fruity
Methyl salicylate	40	2.20 ± 1.28 b	0.95 ± 0.45 a	0.95 ± 0.98 a	Wintergreen like
D-Limonene	10	2.91 ± 5.11 a	1.40 ± 1.56 a	2.93 ± 5.20 a	Lemony
Toluene	5	4.04 ± 2.75 b	1.61 ± 1.33 a	1.38 ± 0.79 a	Sweet floral
alpha-Pinene	140	0.02 ± 0.01 a	0.02 ± 0.01 a	0.02 ± 0.02 a	Pine oil
beta-Ionone	3.5	9.85 ± 5.47 a	8.04 ± 5.92 a	10.48 ± 4.31 a	floral

Note: The values in the table are the mean ± standard deviation. The aroma description and aroma threshold of volatile compounds in black tea were obtained from the related literature [44,45,49]. Different lowercase letters after the data in the same column indicate significant differences.

. Fourteen common aroma compounds with OAV values larger than one were identified, among which seven compounds showed much higher OAVs, including benzyl alcohol (OAV > 10), linalool (OAV > 10), hexanal (OAV > 10), octanal (OAV > 10), nonanal (OAV > 10), and decanal (OAV > 100). The main fragrance-contributing substances in sample FF included decanal, nonanal, benzyl alcohol, octanal, and linalool. For sample YF, decanal, benzyl alcohol, nonanal, octanal, and linalool were the main flavouring substances. Decanal, nonanal, benzyl alcohol, octanal, and linalool were the dominant flavouring substances of sample DF. Alcohols are particularly common aroma components in teas, and they are relatively abundant [7,38]. Linalool and geraniol are monoterpene alcohol compounds that are formed by the hydrolyses of glycosides, and they are important indicators for evaluating the quality of a tea aroma and endow black tea with a sweet floral flavour and elegant aroma. The addition of β-glucosidase during black tea processing was found to increase the contents of the volatile monoterpene alcohol compounds, such as linalool and geraniol, in the black teas [38]. The Maillard reactions of tea leaves occurring during the drying process also increased the contents of linalool and its oxides, enhancing the aromas of the black teas [39]. Benzyl alcohol and phenylethyl alcohol were formed by the hydrolysis of benzyl β-D-glucoside with phenylethyl-β-D-glucopyranoside [38,40]. Fermentation is the most critical step in the formation of black tea aroma, as during fermentation the alcohols are oxidised to aldehydes and the tea aroma is transformed from grassy to floral and fruity, or sweet and roasted. The process also converts benzyl alcohol and phenethyl alcohol into benzaldehyde and phenylacetaldehyde [41]. Decanal was found to possess the highest OAV value in the black tea that was analysed, indicating that it contributed the most to the black tea’s aroma. The content of decanal in sample DF was much higher than the contents in samples YF and FF, and close to twice the content in sample YF. Fuyun 6 black tea, as a cultivar of traditional black tea, has floral and fruity aromas and sweet and roasted aromas. Xu believed that the main ingredient that contributes to the fragrance in Fuyun 6 was decanal [42], which is consistent with our results. The OAVs of hexanal, phenylethyl alcohol, geraniol, methyl salicylate, and alpha-pinene were smaller than that of decanal due to the high aroma threshold. Aldehydes with from five to nine carbon atoms have oily fragrances, aldehydes with higher molecular weights have an orange peel flavour, and branched-chain aldehydes have pleasantly sweet or fruity flavour profiles [43]. In tea leaves, pentanal, butanal, 3-methyl-, hexanal, heptanal, octanal, nonanal, decanal, and other fatty aldehydes produced floral and fruity aromas [44,45]. A fruity aroma was also provided by beta-cyclocitral [46]. These aldehydes give black teas their floral and fruity aromas. Volatile esters are mainly related to tea fermentation and fatty acid metabolism, and most esters are long-lasting and have strong fruit aromas [43]. Methyl salicylate presents wintergreen and mint flavours, and is a major component in black tea aromas [47]. Beta-ionone is a carotenoid cleavage product with a violet flavour [48]. It is believed that these compounds may be the most important aroma compounds of Fuyun 6 black tea.

3.4. Analyses with the Electronic Nose for Black Teas from Different Origins

As shown in Figure 7A, the samples from FF were mainly distributed in the third quadrant, samples from YF were mainly distributed in the first and fourth quadrants, and samples from DF were mainly distributed in the second quadrant. Sample DF-4 was in the third quadrant, and DF-2 was distributed in the first quadrant. It is shown that the E-nose sensor can distinguish black teas from different origins. Figure 7B is a cross-validation model that was permutation tested 200 times. This proved the rationality of OPLS-DA in determining the aroma compounds in samples FF, YF, and DF. The fitting parameters of the independent variable fitting index R²X were 0.82, the dependent variable fitting index R²Y was 0.675, and the model prediction index Q² was 0.573. This showed that the OPLS-DA discriminant model was reliable. The intercepts of the Q² regression line relative to the y axis were all less than zero, indicating that the OPLS-DA discriminant model was not overfitted and that the model was reliable (R² = 0.126, Q² = −0.383). All squared points (Q²) were below the original point on the right of Figure 7B, and the intersection of the regression line for Q² with the vertical line was less than zero, validating the model. The variable weight values (VIP) of the electronic nose sensor are shown in Figure 7C. Six of these sensors (S1, S2, S3, S6, S7, and S10) had VIPs > 1.0, indicating major contributions to the discrimination of these varieties. Using the response values of the electronic nose sensor, the odour radar fingerprint was drawn (Figure 7D). The electronic nose proved capable of effectively distinguishing black teas from different regions and can provide valuable assistance in their utilization.

4. Conclusions

We utilized a postoptimized method to analyse the aroma compositions of black teas from various origins, revealing the presence of 71 volatile aroma compounds. While the types and concentrations of the aroma components in black tea from the different regions were largely similar, some variations were still observed. Sample FF exhibited substantial amounts of 1-nonanol, linalool oxide, beta-pinene, furfural, hexanal, linalool, and methyl salicylate. Sample YF contained heptane, (E)-linalool oxide, hexanoic acid, and others. Sample DF, on the other hand, showed substantial amounts of decanal and dodecane. Based on the OAVs, the combined effects of compounds such as benzyl alcohol, linalool, geraniol, hexanal, octanal, nonanal, decanal, and beta-ionone contributed to the overall aroma profiles of the black teas. These findings provide valuable data for further research on the quality and standardization of black teas from different origins.