Dynamic Changes in Qidan Aroma during Roasting: Characterization of Aroma Compounds and Their Kinetic Fitting

Qidan is one of the most famous varieties of Wuyi Rock tea and has a strong aroma. The aroma-active compounds in Qidan subject to different roasting times were analyzed using solid-phase microextraction two-dimensional gas chromatography–olfactometry–mass spectrometry (SPME-GC×GC-O-MS), and a total of 92 aroma-active compounds were detected. Multivariate statistical analysis showed that the roasting time had a significant effect on the aroma characteristics of Qidan, and that the key products in the Maillard reaction accumulated with the extension of the roasting time; these key products were screened out according to the calculation of the odor activity values (OAVs), from which kinetic equations were established. It was found that the levels of 2-methylbutanal, 3-methylbutanal, 2-methylpyrazine, 2-ethyl-5-methylpyrazine, and benzaldehyde increased with time, while the contents of benzeneacetaldehyde showed a tendency to first increase and then decrease. This study provides a theoretical basis for flavor quality control during Qidan processing.


Introduction
Tea is an important beverage in Chinese culture, playing an important role in cultural exchange and economic development.Wuyi Rock tea (WRT), a type of oolong tea, has a unique 'rock flavor' due to various factors, such as the variety, climate, and production process [1].According to the Chinese national standard, Dahongpao is the most representative variety of WRT and is considered to be one of the four famous traditional teas [2].Qidan has an elegant aroma, with a slightly spicy and mellow sweet taste, making it a treasure of WRT [3].
Aroma is a crucial factor in evaluating the quality of tea.More than 700 volatile compounds have been identified to come from tea, including aldehydes, alcohols, ketones, esters, and heterocyclic compounds [4].Yue et al. [5] studied the aroma characteristics of 16 oolong tea varieties and identified 368 volatile compounds using the headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC/MS) technique.The volatile compounds with a floral aroma have the highest r-OAVs, such as β-ionone, geraniol, benzene acetaldehyde, benzoic acid methyl ester, and indole.
Roasting is the most crucial factor affecting the aroma of tea, causing the cleavage and isomerization of some insoluble substances, which improves its mellow taste and pure aroma [6].According to Zhang et al. [7], oolong tea can enhance the flavor of summer tea by controlling the moisture content.This combination of teas has been found to effectively remove the stale taste, improve the aroma, and reduce the bitterness and astringency.Roasting is a key process that is linked to the quality of oolong tea [8].Appropriate measures must be taken during the drying process to ensure smooth thermochemical, decomposition, and Maillard reactions, resulting in the production of more aromatic substances.During the roasting process, most of the aromatic substances produced during pre-fermentation should be retained as much as possible, while the main aromatic substances that determine the aroma of WRT should be selectively fixed [8].
The Maillard reaction, also known as the carbonyl ammonia reaction [9,10], is the main non-enzymatic browning reaction that occurs during baking in food technology.The Amadori rearrangement products produced during this process can react with ammonia and undergo cyclization to form pyrazines and pyrroles.In addition, they can be directly cyclized to form aromatic compounds, such as furans and their derivatives.The Amadori products can also go on to react with amino acids to produce Strecker degradation products, such as Strecker aldehydes and sulfur-containing compounds [11].
In summary, the Maillard reaction is important in exploring the basic scientific issues related to the roasting process of WRT.While many reports have investigated the effect of temperature on the aroma of WRT, few have focused on the effect of time.Additionally, most studies have used Dahongpao samples rather than pure-bred Dahongpao Qidan samples.Therefore, in this study, the aims were as follows: (1) to evaluate the aroma profiles of Qidan under different roasting time conditions by using an electronic nose system; (2) to identify the differences in the aroma compounds and their respective characteristics in samples roasted for different lengths of time by using SPME and comprehensive twodimensional gas chromatography-olfactometry-mass spectrometry (GC×GC-O-MS); and (3) to establish the Maillard reaction kinetic equation for the main aroma compounds.Therefore, this work is expected to reveal the effects of roasting time on the aroma of Qidan and to provide scientific guidance for tea production.

Preparation of Tea Samples
Qidan is produced in the Bishiyan area of the Wuyi Mountains, located in Fujian Province, with the geomagnetic coordinates 117.57• E and 27.42 • N.All samples were provided by Qiwei Tea Co., Ltd.(Wuyishan, China) and were roasted according to GB/T 18745-2006 [15], with the roasting temperature controlled at 110 ± 5 • C. The samples (500 g) were taken from three roasting cages every 2 h.The final samples of Qidan were obtained at 0 h, 2 h, 4 h, 6 h, 8 h, 10 h, and 12 h.The samples were sealed in an aluminum foil bag and placed in a refrigerator at 4 • C for subsequent analysis.

GC×GC-O-MS Analysis
The GC×GC-O-MS instrumentation utilized in this study included an Agilent 8890 GC-5977B MS, equipped with a primary polar DB-WAX capillary column (30 m × 0.25 mm, 0.25 µm film thickness; Agilent Technologies, Bejing, China) and a secondary mediumpolarity DB-17 MS column (1.85 m × 0.18 mm, 0.18 µm film thickness; Agilent Technologies).The method used was based on the one described by Yang [17], with appropriate modifications.The temperature was initially set at 40 • C and held for 3 min, before being increased to 230 • C at a rate of 4 • C/min and held for 5 min.The inlet temperature was 230 • C and helium was used as the carrier gas, with a flow rate of 1 mL/min.The mass spectrometry conditions were as follows: the ion source temperature was 230 • C, the transmission line temperature was 280 • C, and the temperature of the quadrupole was 150 • C. The electron bombardment ionization mode was employed, with a scanning range of m/z 29~500 and an electron energy of 70 eV.Additionally, a sniffing port (Sniffer 9000; Brechbühler, Schlieren, Switzerland) was installed between the GC and MS to transport aroma-active compounds to the human nose using an ultra-high-purity helium carrier gas at a flow rate of 1 mL/min.The olfactometry part was composed of three trained people, who described the aroma and recorded the time that it flowed out.Each sample was analyzed three times.

Aroma Extraction Dilution Analysis (AEDA)
The dilution of the aroma-active compounds was achieved by controlling the volume of gas purged into the sniffing device, through adjusting the split ratio of the carrier gas during the purge process (splitless or 1:1, 3:1, 7:1, or 15:1).The diluted extracts were then analyzed under the same experimental conditions until the maximum dilution at which a particular odor could be smelled was reached.The maximum dilution multiple was the dilution factor of the compound corresponding to that odor, i.e., the flavor dilution (FD) factor.Each sample was tested in three parallel experiments.

Odor Activity Value (OAV)
The OAV is calculated from the ratio of the concentration of the aroma-active compound to the threshold value.The threshold values for different aroma-active compounds are referenced in a previous study [18].In general, aroma-active compounds with an OAV greater than or equal to 1 are considered to contribute to the overall odor of the sample.

Qualitative and Quantitative Analysis
The aroma-active compounds (FD ≥ 8) were quantified using gas chromatographytriple quadrupole mass spectrometry (GC-QqQ, Agilent 7890 A GC-7000 MS) in selective ion mode (SIM), according to the results of AEDA.The external standard method with internal standard correction [17] was employed.To reduce errors, the standard compounds were prepared and diluted with n-hexane.The mixed standard solution was configured based on the difference in the retention time or peak area between the compounds, to ensure that each solution contained a similar level of compounds for the GC.Nine appropriate concentration gradients were used, and 1 µL of 2-methyl-3-heptanone (0.816 µg/µL) was added to each as an internal standard compound.The standard curve was drawn according to the ratio of the GC peak area of both the standard compound and the internal standard substance to the added concentration.Then, the peak area ratio of each compound in the sample to the internal standard substance was determined and, with the concentration of the internal standard substance, was used to calculate the compound concentration [17].To ensure the accuracy of the results, the analysis of both the standard compound solution and the sample was taken as the average of three results.

Kinetic Studies Analysis
With the increase in the osmosis time, the concentrations of the main aroma-active compounds in tea changed by different degrees.To investigate the formation pattern of these characteristic compounds, their kinetics were further investigated with reference to previous experimental methods [19,20].The kinetic formulae are as follows: The letters A, t, K, and n in the formula represent the content of volatile compounds, the reaction time, the reaction rate constant, and the reaction energy level, respectively.By integrating kinetic Equation (1), we obtained the following linear Equation (2): A 0 represents the initial concentration.The first-order kinetic equation is as follows: Foods 2024, 13, 1611 5 of 15 The fit of the regression curve between the measured values and model calculations is represented by R 2 , where the closer the value of R 2 is to 1, the closer the model is to the measured values.

Statistical Analysis
All the results are expressed as the mean ± standard deviation (average ± SD).The mean and ANOVA of all the aromatic compounds were collated and analyzed using Excel 2019 and SPSS Statistics 27 software.Statistical significance was determined at the p ≤ 0.05 level.Figures were created using OriginPro 2023 b software.Partial least square discriminant analysis (PLS-DA) and principal component analysis (PCA) were performed using the SIMCA-P14.1 software.

Results of E-Nose Analysis
Figure 1A displays a radargram that was derived from the e-nose data.The seven samples exhibit similar spindle-shaped profiles, but the ten types of metal oxide sensors differ in intensity.The seven samples responded most strongly to the W5S receptor, which is very sensitive to nitrogen oxides, suggesting that this sensor is the most crucial indicator for distinguishing between these samples.Moreover, nitrogen oxides are linked to Maillardrelated products.After 12 h of roasting, the response value of the W5S sensor for Qidan was significantly higher than that of the other samples.This may be due to the accumulation of nitrogen oxides produced by the Maillard reaction, as the roasting time is increased [17].In Figure 1B, the X and Y axes represent the contribution rates of PC1 (99.03%) and PC2 (0.57%), respectively.The larger the contribution rate, the better the PC performance and the better it reflects the original information.Therefore, it is evident that samples roasted for different lengths of time can be distinguished from one another, with slight but noticeable variations between the 0 h, 2 h, and 4 h samples, as well as smaller yet distinguishable differences between the 8 h, 10 h, and 12 h samples.In conclusion, the electronic nose results indicate that different roasting times produce samples with distinct flavor profiles.The variation in odor may be attributed to the nitrogen oxides generated by the Maillard reaction.

Analysis of Aroma-Active Compounds in Qidan
In this study, 92 aroma-active compounds were detected in seven samples that were analyzed using SPME-GC×GC-O-MS.Table 1 shows the semi-quantitative concentrations

Analysis of Aroma-Active Compounds in Qidan
In this study, 92 aroma-active compounds were detected in seven samples that were analyzed using SPME-GC×GC-O-MS.Table 1 shows the semi-quantitative concentrations of each sample.The compounds were classified into one of the following seven groups: aldehydes (26), alcohols (20), heterocycles (15), esters (12), ketones (9), acids (6), and alkenes (4).Table 1 shows that with an increase in the roasting time, the number of heterocyclic compounds gradually increases, while the number of other compound types remains relatively constant.During the manufacturing process, the Maillard reaction produces a significant amount of heterocyclic compounds, such as furans, pyrroles, thiophenes, and their derivatives, in addition to Strecker degradation products, sulfur-containing compounds, and so on [17].According to Guo et al. [3], 3,5-dimethyl-2-ethylpyrazine may contribute to the aroma of Narcissus after roasting.The difference in flavor between Dahongpao and Jieding Oolong after roasting has been attributed to the higher content of both 2,5dimethylpyrazine and trimethylpyrazine in Dahongpao [21].Additionally, Yuan et al. [22] found that 2-methylpyrazine is an important metabolite for Shuixian, allowing it to present a woody fragrance."-": not detected. 1Aroma perception of each aroma-active compound that was detected at the sniffing port. 2 Retention index (RI) on DB-WAX polar column. 3MS, mass spectrum; RI, retention index; O, olfactory; STD, standard compounds.Identification methods with no STD, considered as 'tentatively identified'.
The type and content of aldehydes also affect the aroma of tea.Peter Schieberle's team [23] used molecular sensory science techniques to identify 2/3-methylbutyraldehyde as the key aroma-active compound in Darjeeling black tea.The above studies have confirmed the important contribution of Maillard reaction products to the characteristic aroma of WRT.However, the key aroma-active compounds related to the Maillard reaction in Qidan are unknown, and the effects of different roasting times on these compounds have rarely been reported.

Analysis of Maillard Reaction-Related Products in Qidan
A total of 20 key aroma-active compounds (FD ≥ 8) related to the Maillard reaction were screened in the Qidan samples and were quantified (Table 2).
As shown in Table 2, the amount and concentration of aroma-active compounds related to the Maillard reaction increased with the increase in roasting time, with the lowest number being at 0 h, but increasing at 12 h.Moreover, 2-Ethyl-6-methylpyrazine, 2-ethyl-3-methylpyrazine, and ethenylpyrazine started to be produced after 8 h of roasting; 2-ethylpyrazine and 2-ethyl-3,6-dimethylpyrazine were identified after 10 h; whereas 2ethyl-3,5-dimethylpyrazine was only identified at 12 h.It was demonstrated that pyrazine production increases significantly at 70 • C and reaches its maximum rate at 120 • C.However, when heated at 100 • C, it takes 4 h to reach its maximum content [24].These findings indicate that the duration of the roasting process can impact the production of pyrazines; as the roasting time is increased at a certain temperature, the number of pyrazine species will increase, and the content will accumulate.In order to further investigate the changes in the flavor-active compounds of Qidan during roasting, multivariate statistical analyses were performed based on the concentrations of 20 key aroma-active compounds in the Maillard reaction (Figure 2).The results of the principal component analysis (PCA) are shown in Figure 2A.Samples with different roasting times are categorized into the following three categories: the first category includes 0 h, 2 h, 4 h, and 6 h (green); the second category includes 8 and 10 h (blue); and the third category includes 12 h (red).This indicates that there are differences in the aroma characteristics of Qidan at different roasting times.Figure 2B is a biplot with different elliptical areas indicating the 95% confidence intervals.As the roasting time increases, the concentrations of most aroma-active compounds reach their maximum value, and the VIP value indicates their relevance.Figure 2C details the aromatic compounds with VIP values greater than 1, including 2-ethyl-6-methylpyrazine, 2-ethyl-3,5-dimethylpyrazine, furfural, 2-ethylpyrazine, benzeneacetaldehyde, 2-ethyl-3,6-dimethylpyrazine, and ethenylpyrazine.These substances are the main reason for the differences in the aroma of Qidan at different roasting times.These compounds have all been reported in studies on aroma-active substances after the roasting of WRT [7,25].The replacement test in Figure 2D shows that the model was successfully validated and that both Q 2 and R 2 were higher than 0.5.The samples in these three categories had significant differences that could be differentiated from each other, which indicated that the roasting process had a more significant effect on the aroma composition and release of Qidan, which is in line with the findings of Wang et al. and Zhan et al. [2,26].

Modeling the Generation of Characteristic Differential Aroma Compounds
Kinetic equations were developed to investigate the release pattern of key aromaactive compounds resulting from the Maillard reaction at different roasting times.Ge [19] selected glucose and lysine to establish kinetic models to simulate the Maillard reaction in the processing system of brown sugar, in order to study the effects of the processing time and temperature conditions on the main flavor-active compounds, such as aldehydes, pyrazines, and furans, as well as to provide a theoretical basis for their actual production through the establishment of kinetic equations.Yang [17] explored the generation of aromaactive substances in each system under medium-fire roasting conditions by modeling glucose and theanine, as well as glucose and glutamic acid.The glutamic acid system produced more pyrazines, including 2,5-dimethylpyrazine, 5-methyl-2-ethylpyrazine, 2,3,5trimethylpyrazine, and 2,5-dimethyl-3-ethylpyrazine, compared to the theanine system.Most previous studies have simplified the model of the Maillard reaction by selecting key sugars and amino acids in order to establish it.However, the Maillard reaction is an extremely complex process, and the simplified model does not simulate the change rule of the precursor substances and intermediates in the real products.Therefore, in this paper, we selected Qidan as the research object to study its key aroma-active compounds.The most important aroma-active compounds in the Maillard reaction process were screened out through the OAV calculation (Table 3), resulting in six aroma-active compounds with average OAV values of ≥1, which were 2-methylbutanal, 3-methylbutanal, 2-methylpyrazine, 2-ethyl-5-methylpyrazine, benzaldehyde, and benzeneacetaldehyde.The production and generation patterns of these compounds were fitted, as shown in Table 4.In Figure 3, the concentrations of the key differential aroma-active compounds changed by varying degrees in all simulated systems, as the roasting time was extended.Moreover, 2-Methylbutanal, 3-methylbutanal, 2-methylpyrazine, 2-ethyl-5methylpyrazinee, and benzaldehyde exhibited a similar generation pattern during the roasting process, with their concentrations increasing as the roasting time was extended, reaching their maximum values after 12 h.The correlation coefficients obtained by fitting the first-order kinetic equations were significantly higher, with R 2 values close to 1, based on the concentration of the compounds as a function of time.These key aroma-active compounds are closely related to the Maillard reaction and present a roasted aroma.Guo et al. [13] found that 5-methyl-2-ethylpyrazine is the key aroma-active compound in WRT.In addition, 2-Methylbutanal and 3-methylbutanal are also essential compounds in the aroma profile of WRT [23].The contribution of benzaldehyde to the aroma of WRT has also been emphasized in the research by Wang et al. [2] as well.Benzeneacetaldehyde is an important contributor to the floral aroma of Qidan, and its content showed a tendency to increase and then decrease during the roasting process, reaching the maximum at 10 h.The formation of benzeneacetaldehyde is also closely related to lipid oxidation; Zamora [27] found a similar pattern of change in their study investigating the effect of the roasting time on the aroma of cinnamon, concluding that benzeneacetaldehyde is an important phenylalanine volatile compound in tea leaves.When phenylalanine and lipid oxidation products were heated together, benzaldehyde and benzeneacetaldehyde were produced, the amount of which depended on the lipid oxidation products involved.contributor to the floral aroma of Qidan, and its content showed a tendency to increase and then decrease during the roasting process, reaching the maximum at 10 h.The formation of benzeneacetaldehyde is also closely related to lipid oxidation; Zamora [27] found a similar pattern of change in their study investigating the effect of the roasting time on the aroma of cinnamon, concluding that benzeneacetaldehyde is an important phenylalanine volatile compound in tea leaves.When phenylalanine and lipid oxidation products were heated together, benzaldehyde and benzeneacetaldehyde were produced, the amount of which depended on the lipid oxidation products involved.The study above shows that the Maillard reaction's key aroma-active compounds accumulate as the roasting time extends, except for benzeneacetaldehyde, which reaches its maximum amount of content at 10 h.To ensure the harmony of Qidan's aroma, controlling the roasting time during production is crucial.It is important to note that this study is limited by its sample size.Therefore, further studies are needed to expand the sample size and verify the effect of roasting time on the flavor of Qidan samples.Additionally, a kinetic model should be established to validate the above conclusions.

Conclusions
In conclusion, this study identified the characteristics of aroma-active compounds during the roasting process of Qidan through molecular sensory analysis.Differences in the odor of samples with different roasting times were identified using an electronic nose.The samples were analyzed by clustering through PCA and OPLS-DA.The OAVs were calculated to identify the key aroma-active compounds in the Maillard reaction, with kinetic equations being developed for the six key compounds.The results showed that the The study above shows that the Maillard reaction's key aroma-active compounds accumulate as the roasting time extends, except for benzeneacetaldehyde, which reaches its maximum amount of content at 10 h.To ensure the harmony of Qidan's aroma, controlling the roasting time during production is crucial.It is important to note that this study is limited by its sample size.Therefore, further studies are needed to expand the sample size and verify the effect of roasting time on the flavor of Qidan samples.Additionally, a kinetic model should be established to validate the above conclusions.

Conclusions
In conclusion, this study identified the characteristics of aroma-active compounds during the roasting process of Qidan through molecular sensory analysis.Differences in the odor of samples with different roasting times were identified using an electronic nose.The samples were analyzed by clustering through PCA and OPLS-DA.The OAVs were calculated to identify the key aroma-active compounds in the Maillard reaction, with kinetic equations being developed for the six key compounds.The results showed that the roasting process had a more positive effect on the aroma-active components of Qidan, resulting in a richer overall flavor of the samples.The concentrations of 2-methylbutyraldehyde, 3-methylbutyraldehyde, 2-methylpyrazine, 5-methyl-2-ethylpyrazine, and benzaldehyde increased with the increase in roasting time.In contrast, the concentration of phenylacetaldehyde showed a trend of first increasing and then decreasing, reaching its maximum concentration after 10 h.Therefore, it is particularly important to control the roasting time in the production process, to ensure the richness of the aroma.In order to ensure the harmony of the Qidan aroma, the roasting time should not exceed 10 h.An in-depth study of Qidan samples with different roasting times was carried out to eliminate empiricism in the traditional tea roasting process and to provide a theoretical basis for this purpose.

Foods 2024, 13 , 1611 6 of 15 Figure 1 .
Figure 1.Radargrams and PCA analyses of the e-nose.(A) Radargrams of e-nose responses for different types of volatile compounds at different roasting times; (B) e-nose score plots (PCA) at different roasting times.

Figure 1 .
Figure 1.Radargrams and PCA analyses of the e-nose.(A) Radargrams of e-nose responses for different types of volatile compounds at different roasting times; (B) e-nose score plots (PCA) at different roasting times.

Figure 2 .
Figure 2. OPLS-DA plots based on SPME-GC×GC-MS data results: (A) is a scatter plot of the predicted scores, the green, blue, and red signals represent the Qidan samples with different roasting times, respectively; (B) is a biplot, with different elliptical areas indicating 95% confidence intervals; (C) represents the variable importance in projection (VIP); and (D) represents the validated model.

925 Figure 3 .
Figure 3. Variation pattern of the concentration of key aroma-active compounds with time.

Figure 3 .
Figure 3. Variation pattern of the concentration of key aroma-active compounds with time.

Table 1 .
The aroma-active compounds and their FD factors detected in Qidan using SPME-GC×GC-O-MS.

Table 2 .
Quantitative results of 20 aroma-active compounds related to Maillard reaction products at different roasting times.Ions for SIM mode quantitative analysis are selected as coarsened to represent the mother ion with the largest molecular weight.Values in the same line followed by different superscript letters are significantly different (p < 0.05). *

Table 3 .
OAVs of 20 aroma-active compounds related to Maillard reaction products at different roasting times.

Table 4 .
Fitted curves of key aroma-active compounds with different roasting times.

Table 4 .
Fitted curves of key aroma-active compounds with different roasting times.