Characterization of volatile compounds of Pixian Douban fermented in closed system of gradient steady‐state temperature field

Abstract As an essential flavor condiment in Sichuan cuisine, Pixian Douban (PXDB) is usually produced by open fermentation process in strip pools or ceramic vats. In this study, an experiment of PXDB fermentation was conducted for 90 days in a closed system of gradient steady‐state temperature field (GSTF). To investigate the characterization of volatile compounds of PXDB in the closed system, the volatiles in three kinds of samples including samples of GSTF (SGT), samples of constant temperature (SCT), and samples of traditional fermentation (STF) were analyzed. The results showed that 75, 67, and 68 volatile compounds were detected in SGT, SCT, and STF, respectively. Compared with the traditional fermentation, the process in the closed system of GSTF was conducive to produce more kinds of esters and alcohols. A total of 22 major aroma active compounds were identified in three samples by combination analyses of gas chromatography‐olfactometry (GC‐O) and odor activity value (OAV). The appearance, smell, texture, and taste of the three different samples had shown different changes, but the sensory characteristics of the SGT were more similar to those of the STF by quantitative descriptive analysis (QDA) and principal component analysis (PCA). This study indicated that the closed system of GSTF could be applied in PXDB fermentation to obtain higher quality products, which brought a bright prospect of replacing the traditional fermentation process to realize the controllable industrialized production of PXDB.


| INTRODUC TI ON
As a traditional fermented food produced from a small town named Pixian County in Sichuan Province of China, Pixian Douban (PXDB) is an essential condiment in preparing Sichuan cuisine, which is honored as the spirit of Sichuan Cuisine and famous for its fascinating flavor globally. The production process of PXDB employed in most factories is traditional process, which has a history of hundreds of years. However, this process has many drawbacks, such as lower mechanization and automatization, lower productivity, higher labor and production cost, less quality and higher risks of food safety, which have seriously hindered the development of the industry (Ding et al., 2020). Therefore, with the development of fermentation technologies, it has become an urgent requirement to upgrade this industry by using advanced technologies and equipment.
In recent two years, the production of PXDB in tank fermenter has been proposed and carried out in some factories. However, compared with traditional fermentation, this new process which just changed the equipment from strip pool or ceramic vat to tank fermenter without any other improvement is still dependent on the weather heavily ( Figure 1). Moreover, the new process was not verified by the strict tests before it was amplified, which caused that the defects of the traditional process has not been overcome although the occupancy area of the factory had been greatly reduced by the new process (Ding et al., 2020). Therefore, as a new process with the most potential to replace the traditional fermentation, the PXDB process of tank fermenter should be further studied deeply to overcome the defects of traditional process and obtain high-quality products before it was used in the factory.
Besides, aroma is a key indicator in evaluating the product quality of PXDB as the same other fermented foods such as soy sauces and vinegars (Al-Dalali et al., 2019;Feng et al., 2014). Therefore, the new process should be developed based on the study of traditional fermentation process to obtain high-quality product with good aroma. In the literatures, many studies about the volatile compositions of PXDB have been conducted, and many valuable results were obtained (Li et al., 2016(Li et al., , 2018Lin et al., 2019).
In Lu's study, 22 key volatile compounds were identified in PXDB fermented with traditional process (Lu et al., 2019). The microbial communities, flavors, and their relationships in PXDB were studied in Liu's study, which revealed that the flavor compounds would accumulate significantly with the prolonging of the fermentation period, especially for umami-taste amino acids, organic acids, and volatile compounds . In Lin's study, 21 aroma active compounds were detected with flavor dilution factor ranging from 2 to 16 in PXDB (Lin et al., 2019). Those researches were benefit of revealing the aroma of the traditional fermentation products and provided the foundation for the studies of new process.
In this study, a closed system of PXDB fermentation under gradient steady-state temperature field (GSTF) was constructed based on the results of our previous research (Ding et al., 2020).
Then, three experiments of PXDB fermentation were conducted in the closed system of GSTF, closed system of constant F I G U R E 1 Process diagram of PXDB production temperature, and traditional fermentation system, respectively.
The volatiles of the product fermented in the new system were characterized by comparing with those of the products from the closed system of constant temperature and the traditional fermentation system in order to clarify the volatile compounds of the three products and improve the new process to produce highquality products.

| Procedures of three different fermentation processes
The schematic apparatus of fermentation processes of PXDB is shown in Figure 2. As shown in Figure 2, the apparatus of the new closed system was mainly constructed from two tank fermenters of 50 L, an air supply system, and a thermostat. The fermentation of PXDB in closed system was conducted with the fermenters of a filling coefficient of 0.8 at constant temperature of 30°C and at GSTF, respectively. It should be noted that the GSTF in the fermenter was designed and achieved by keeping the temperature of jacket at 27°C and the propeller temperature between 45°C and 52°C. The traditional fermentation was carried out in an open ceramic vat of 30 L following the traditional operation. The meju and red peppers purchased from Sichuan PXDB Co., LTD. were first mixed at a ratio of 1:3 before starting the fermentation and then transfer into the fermenters and the ceramic vats. The PXDB in the tank fermenters was stirred once every 6 hr at a rate of 35 rpm for 3 min and ventilated every 4 hr at a rate of 5 L/min for 5 min. The experiment was carried out for 90 days from August to October.

| Extraction of volatile compounds
The HS-SPME sampling was carried out according to previously described method with some modifications (Lu et al., 2019). Sample was mashed into homogenized paste, and 5.0-g mashed sample was put into a 15-ml solid-phase micro-extraction (SPME) vial, where 10 μl of 1,2-dichlorobenzene (10 μg/ml in methanol) was used as the internal standard. Then, the vial was sealed and equilibrated at 55°C for 30 min. Afterwards, a carboxen-polydimethylsiloxane fused silica (CAR/PDMS, 75 μm)-coating fiber (Anpel Inc.) was exposed to the headspace of the sample to absorb the volatiles at 55°C for 40 min and the coating fiber was quickly inserted into a GC injection port and desorbed at 250°C for 5 min.

| Gas chromatography-mass spectrometer analysis
GC-MS analysis was performed on a Shimadzu gas chromatograph (Shimadzu). Separation of volatile components was performed on a DB-5MS column (Agilent). Helium was used as a carrier gas at a constant flow rate of 1 ml/min. Oven temperature was maintained at 40°C for 1 min, programmed at 5°C/min to 100°C and held for 1 min, programmed at 7°C/min to 150°C and held for 4 min, thereafter programmed at 5°C/min to 185°C and held for 5 min, finally programmed at 10°C/min to 200°C. The mass spectrometer was operated in electron impact mode with the electron energy set at 70 eV and a scan range of 35-500 m/z. The temperature of MS source was set at 200°C. A mixture of n-alkanes (C6-C20) was injected directly into GC-MS under the same condition as that for the samples to calculate the retention indices (RIs). Each volatile compound was F I G U R E 2 Schematic apparatus of three fermentation systems for PXDB. (1) pump; (2) solenoid valve; (3) air dryer; (4) air filter; (5) gas rotameter; (6) constant temperature fermenter; (7) temperature control unit; (8) GSTF fermenter; (9) temperature detected unit; (10) ceramic vat; (11) timing socket; (12) intelligent temperature controller identified using the National Institute of Standards and Technology (NIST17) library. The content of each compound is obtained by comparing it with the internal standard.

| Gas chromatography-olfactometry analysis
The samples were analyzed by using a gas chromatograph equipped with an olfactory detector (Shimadzu). The sample was injected in the splitless mode at 200°C for 5 min. The oven temperature was adjusted to the same condition as that of GC-MS. Retention times and descriptions of aromas were recorded by three trained assessors (replaced at 10 min intervals) after the sample injection. Each trained assessor sniffed each sample in three replicates.

| Qualification and quantification of volatile compounds
The qualification of the volatile compounds was done of the following methods. The compounds were tentatively identified by comparing their mass spectra with those which were found in the National Institute of Standards and Technology library (NIST17), and by comparing their RIs with those reported in the previous literature. Quantitative analysis of volatile compounds was conducted under the same condition as that of GC-MS. The chemical quantities were calculated based on the relative peak area to the area of 1,2-dichlorobenzene, which was used as the internal standard.

| Calculation of odor activity values
Although GC-O analysis is an effective method for odorant identification, it could not indicate a final importance of the odorant to the overall aroma (Gao et al., 2014). Therefore, the ultimate contribution of a particular compound to the overall aroma of PXDB mainly determined its odor threshold and the odor activity value (OAV) (Giri et al., 2010). If the OAV of the compound calculated as the ratio of its concentration to odor threshold was larger than or equal to 1, it could be considered to contribute to the overall aroma .

| Establishment of flavor descriptors
The sensory characteristics of the three kinds of samples were evaluated by the quantitative descriptive analysis (QDA).
According to the methods previously described (Chen et al., 2013;Gao et al., 2018;Zhao et al., 2020), the sensory characteristics of samples were characterized by the sensory team, which was composed of 10 students (five males and five females, aged 23-26) with rich sensory evaluation experience from Xihua University.
Firstly, the evaluators discussed the aroma characteristics and put forward the descriptors of PXDB. Then, a flavor description 2.7.2 | Evaluation method 10 g PXDB was taken into a 30-ml plastic cup and coded randomly.
The intensity range of aromas were 0-9 scale (0 represented none, 9 represented very strong). The sensory evaluation of each sample was tested three times and the results were averaged.

| Statistical analysis
All data were presented as means ± SD for at least three replicates.
The graph presentations were generated using Origin version 8.5 (OriginLab Inc.). Principal component analysis (PCA) was performed using the Simca version 14.1 (Simca Inc.).

| Identification and quantification of volatile compounds by GC-MS
The volatile compounds in the samples of gradient temperature of steady-state (SGT), samples of constant temperature (SCT), and samples of traditional fermentation (STF) were identified by GC-MS.
As a result, a total of 103 volatiles were identified in the three samples (Table 1), which could be categorized into 10 different groups, including alcohols, aldehydes, acids, esters, hydrocarbons, ketones, phenols, heterocyclics, and others. As shown in Figure 3, alcohols, acids, esters, and phenols were the largest four groups in SGT accounting for approximately 6.1%, 2.9%, 4.4%, and 4.0% of the total volatiles, respectively. For SCT, alcohols, esters, and phenols were the three largest groups accounting for approximately 10.0%, 8.6%, and 7.1% of the total volatiles, respectively. For STF, alcohols, aldehydes, acids, esters, and phenols were the dominant classes, and accounted for 12.8%, 11.3%, 8.2%, and 8.4% of the total volatiles, respectively.
As shown in Figure 4, there were 75, 67, and 68 kinds of volatile compounds in SGT, SCT, and STF, respectively. Compared with traditional fermentation, the number of esters in SGT and SCT was larger while the number of aldehydes was smaller, which implied that SGT and SCT both had an advantage of producing esters rather than aldehydes. Besides, more kinds of esters and acids were included TA B L E 1 Identification and quantification of volatile compounds in three samples  Esters are highly important indicators for the quality grade of fermentation products (Moy et al., 2012), which might mainly be formed by alcohol fermentation or esterification between acids and alcohols during the aging technology (Charles et al., 2000;Li et al., 2018). In this work, ester was also the largest group of all the volatiles as reported in other studies of fermented foods, in which 38 volatile esters were identified in a concentration ranging from 7.5 to 672 µg/kg (Table 1). Totally, 33, 33, and 23 volatile esters were contained in SGT, SCT, and STF, respectively, indicating that the fermentation processes of closed system were beneficial to the formation of ester species. Nineteen of the 38 identified volatile esters in this study were also found previously in other fermented soybean products as marked in Note: Values expressed as average (n = 3) ± standard deviation.
Abbreviation: GSTF, gradient steady-state temperature field; nd, not detected; SCT, samples of constant temperature; STF, samples of traditional fermentation; SGT, samples including samples of GSTF. a Relative peak area to that of internal standard (10 μl of 10 μg/ml 1,2-dichlorobenzene in methanol) using DB-  Fourteen volatile aldehydes were obtained in the three samples becoming the third largest group of the identified volatiles, which might be resulted from alcohol oxidation (Chinnici et al., 2009).
Among the identified aldehydes, 13 were found in STF much larger than 6 in SGT and SCT. Five of the 14 volatile aldehydes identified in this work were reported in the literatures including phenylacetaldehyde, benzaldehyde, furfural, 1-nonanal, and decanal. Besides, as shown in Table 1

F I G U R E 4 Number of volatile compounds in three samples
were detected in STF, which concentrations were ranged from 121.1 µg/kg to 2,299.6 µg/kg. P-cresol was only found both in SGT and SCT, but the other three phenols were identified in all the three samples. According to the reports, the phenolic compounds might mainly come from the lignin degradation (Lu et al., 2019;NATERA et al., 2003). Besides, guaiacol and 4-ethylphenol also existed naturally in the PXDB (Lu et al., 2019). 4, 3, and 1 heterocyclic compounds were identified in SGT, SCT, and STF, respectively, of which the concentration of 2-acetyl pyrrole was the highest in the three samples.
Furthermore, the concentration of 2,6-dimethylpyrazine in SGT was twice as that in SCT. 2-acetylfuran and 2,6-dimethylpyrazine were found both in SGT and SCT, while 2-propionylfuran only appeared in SGT.
For hydrocarbons, five compounds were obtained in SCT and STF but three in SGT. n-hexadecane was found in all the three samples, while the other hydrocarbons were detected only in one or two samples. Therefore, compared with the traditional fermentation, more kinds of esters and alcohols were easily formed in the SGT, which might account for the aroma differences between these two samples. To characterize the aroma of the SGT, it was necessary to screen the aroma active compounds of the three kinds of samples by GC-O.

| Characterization of the aroma active compounds by GC-O
In order to further characterize the important volatile components in the three kinds of samples, the aroma extracts obtained by HS-SPME were subjected to GC-O analysis. As shown in Figures 5 and 6, alcohols, esters, and aldehydes were the three largest major volatile compounds in the three samples, which were consistent with the results of the previous study (Wu et al., 2018). As shown in Table 2, a total of 27 active compounds detected by HS-SPME/GC-O could be grouped into sweet, burnt, sour, honey, bread, fruity, floral, green leaves, almonds, pickles, baked potatoes, sweat, mint, and nuts, which were associated with different chemical groups such as alcohols, aldehydes, acids, esters, phenols, pyrroles, and pyrazines. Among the active aromas, the compound of the highest concentration was phenethyl alcohol (honey-like) in the SGT and SCT but 4-Ethyl-2-methoxyphenol (burnt, spicy) in the STF. Besides, there were 20 aroma active volatiles that were shared in the three samples, while the other volatiles were found only in one or two samples. For example, the aroma-active compounds including alpha-terpineol (oil, anise, and mint), 1-octen-3-ol (earthy), decanal (floral, sweet), and ethyl linoleate (grease) were considered to be the unique aroma compounds of the STF.
As reported in the literatures, the production of fruit aroma is related to ester compounds (Al-Dalali et al., 2019). Isovaleric acid was found in different foods with a sweaty, strong pungent, and cheesy taste (Zhou et al., 2017). Moreover, the studies had shown that the spiciness was caused by some phenolic compounds such as 4-ethylphenol and 4-ethyl-2-methoxyphenol in PXDB (Zhang et al., 2020).

F I G U R E 5
Concentrations of each class of aroma-active compounds in three samples

| OAV analysis of aroma active compounds
As shown in Table 3 b Odor description at the olfactory detection port.
3 Triqui and Reineccius (1995). 4 Lee and Noble (2003). 5 Schnermann and Schieberle (1997). As reported in the literatures, phenylacetaldehyde known as main volatiles were found to come from free amino acids formed during fermentation (Lin et al., 2019). 3-methylthiopropanal was a sulfur-containing compound with a cooked potato, which was also found in other fermented products, such as Korean soy sauce, high-salt soy sauce, and yeast extraction (Zhao et al., 2011).
4-ethyl-2-methoxyphenol with burnt and spicy aroma was considered as a dominant odor impression of PXDB, which had been reported to be related to the metabolic activity of yeast (Lin et al., 2019). Moreover, esters also played an important role in the aroma profile of PXDB due to their low odor threshold   Figure 7, the SGT was closer to the STF on the coordinate axis, so the three samples could be divided into two groups in which SCT was divided into a group but SGT and STF were divided into another group indicating that the odor characteristics of the SGT were similar to that of the STF. This PCA results were most partly in accordance with those of aromaactive compounds analysis and OAV analysis.

| Sensory analysis of three kinds of PXDB products
The sensory quality of the three samples was identified by QDA.
The description categories were divided into appearance (reddish brown, moisture, graininess, and pepper size), smell (soy sauce-like, mellow, pungent, sour, musty, coordination, durability, and intensity), texture (hardness, viscosity, elasticity, and adhesiveness), and taste (salty, delicate taste, spicy, and citric acid-like). It could be seen from Figure 8, for the appearance, the reddish brown and the moisture score of SGT were closer to those of the STF than those of the SCT, which probably caused by the variable temperature in the GSTF that was closer to that in the traditional fermentation process, resulting in some similarity in color and moisture between the two products. In terms of smell, soy sauce-like odor scored the highest, indicating that soy sauce-like odor was the characteristic smell of PXDB. Compared with the SCT, the results showed that the mellow, soy sauce-like and sour odor of the SGT was closer to the STF, which could be related to some alcohols, aldehydes, and acids, such as 3-(methylthio)propionaldehyde, isovaleric acid, butyric acid, and phenylethanol. In addition, the scores of coordination, durability, and intensity of the three kinds of PXDB were similar. In terms of texture, the score of the SGT was more similar to the STF, which might be caused by the differences of moisture. In terms of taste, the evaluation results of salty taste, delicate taste, and citric acidlike taste in the SGT were more similar to those of the STF. In conclusion, compared with the SCT, the sensory characteristics of SGT were more similar to those of the STF, which were consistent with the analyses of OAV and PCA.
The sensory scores for the three samples were analyzed by the PCA as shown in Figure 9. The sensory scores for the three samples given by the 10 evaluators were relatively concentrated, indicating that the consistency and repeatability of the evaluation results were good and the results could truly reflect the sensory characteristics of the three types of products. According to the positions of the F I G U R E 7 Principal component analysis of major aroma active compounds in three samples F I G U R E 6 Number of aroma active compounds in three samples three samples, they could be divided into two groups, in which SCT was divided into a group but STF and SGT were divided into another group. The results were consistent with those in Figure 8 and the sensory quality of the SGT was closer to the STF than that of the SCT.

| CON CLUS ION
This study focused on the characteristics of the volatile compounds in PXDB produced from different three fermentation processes. A total of 103 volatile compounds were detected in the three samples, of which 75, 67, and 68 volatiles were detected in SGT, SCT, and STF, respectively. Compared with STF, more kinds of esters and alcohols were obtained in SGT by analyzing with GC-MS although the total concentrations of volatiles in the SGT were smaller, which suggested that the process in the closed system of GSTF was conducive to produce more kinds of esters and alcohols compared with traditional fermentation. A total of 27 active compounds including eight alcohols, six aldehydes, two acids, seven esters, two phenols, and two heterocyclics were detected in the three kinds of samples by GC-O analysis, of which the three largest major compounds were alcohols, esters, and aldehydes. A total of 22 major aroma-active compounds were identified in the three samples by the combination analysis with GC-O and OAV. The PCA results of 22 major aromaactive compounds had shown that SGT and STF could be divided into a group indicating that the odor characteristics of the SGT were similar to those of the STF. This PCA results were most partly in accordance with those of aroma-active compounds analysis and OAV analysis. The appearance (reddish brown and moisture), smell (soy-sauce-like mellow and sour), texture, and taste (salty, delicate, and citric acid) of sensory index in the SGT exhibited a more similar profile with the STF by the sensory evaluation of QDA and PCA, which had shown that the sensory characteristics of the SGT were more similar to those of the STF. The closed system of GSTF could be applied in PXDB fermentation to obtain higher quality products, which brought a bright prospect of replacing the traditional fermentation process to realize the controllable industrialized production of PXDB.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are not shared.