Characterization of flavor volatile compounds in industrial stir‐frying mutton sao zi by GC‐MS, E‐nose, and physicochemical analysis

Abstract This study aims to investigate the flavor changes of industrial stir‐frying mutton sao zi, a mutton product popular in the northwest of China, at different stir‐frying stages. Electronic nose (E‐nose) was used to recognize mutton sao zi odors at different processing time points, and the individual volatile compounds were further identified by the solid‐phase microextraction (SPME) combined with gas chromatography–mass spectrometry (GC‐MS). A total of 105 volatile compounds were detected by GC‐MS, of which 51 were major volatile compounds. Additionally, GC‐MS and E‐nose data of the samples were also correlated with the fatty acids, crude composition (moisture, fat, protein), and amino acids. The stir‐frying time and temperature may be the critical contributors to different flavors of industrial stir‐frying mutton sao zi. The signal intensities of W1S, W1W, W2S, W2W, and W3S sensors positively correlate with protein, fat, and 18 amino acids, but negatively with SFA and moisture. Hence, this study explored the flavor changes of industrial stir‐frying mutton sao zi by E‐nose and SPME‐GC‐MS for the first time, providing an insight into the industrial production and flavor control stir‐frying machine of stir‐frying mutton products with household flavor.

are no stir-fried mutton products. Mutton sao zi manufactured by traditional stir-frying technologies is particularly popular among northwest Chinese for its unique characteristic flavors and rich nutrition.
E-nose and GC-MS are two alternative approaches to determine the odors. E-nose makes significant contributions to determination of flavors, offering a comprehensive and fast alternative to assess meat quality . GC-MS is one of the main methods to identify traditional meat volatile compounds, such as sauce spareribs , cold-smoked Spanish mackerel (Huang et al., 2019), beef (Aaslyng & Meinert, 2017), and so on. Solidphase microextraction (SPME) is widely applied to collect volatiles from meat, as it is environment-friendly, fast, and simple operation (Dominguez et al., 2019). Although they used GC-MS and E-nose to study the composition and correlation of the volatile compounds, they ignored the role of amino acid, fatty acid, or crude composition in meat products at different processing stages. In another word, the flavor changes of the mutton during the stir-frying process have been less explored.
Hence, in this study, a systematic analysis on the changes of volatile compounds at different operating stages of traditional industrial stir-frying mutton sao zi was performed. The extracted and identified volatile components from combination of SPME and GC-MS might contribute to evaluate the flavor formation at different processing time points of stir-frying mutton sao zi. The associations between content of volatile compounds and the intensities of E-nose sensors, amino acid, crude composition, and fatty acids were elucidated. This work may provide a reference for controlling the mutton sao zi of stir-frying production chain and a technical guidance for industrial production, to improve the flavor quality.

| Preparation of samples
The mutton sao zi samples was taken from a commercial meat company in Yinchuan (Ningxia, China) and prepared according to the traditional method of Hui people in Ningxia, China. Firstly, 130kg of fresh Tan sheep hind leg meat with a fat-lean ratio of 3:7 was cleaned and dice into 1 cm × 1 cm × 1 cm (length × width ×thickness) cubes.
Secondly, cooking wine and soy sauce were added to marinate the meat for 10 min. Then, a small amount of sesame oil was heated in the pot to 80°C and poured into the marinated mutton. Finally, the meat was stir-fried for 35 min according to the stir-frying process. The stirring rate of stir-frying mutton sao zi was 30 times/min. Samples at predetermined time points (0, 5, 10, 15, 20, 25, 30, and 35 min) were stored at −20°C for following experimental analysis.

| Determination of crude composition, fatty acid and free amino acids
Moisture, protein, and fat were determined as described by Association of Official Agricultural Chemists (AOAC, 2012). The moisture content was calculated by a percentage of the weight loss of the samples before and after drying, the crude protein content quantified by the automatic Kjeldahl analyzer (KDN-520, Lvbo) and the fat content extracted by the Soxhlet method which calculates the percentage of weight loss before and after extraction.
The fatty acids compositions were determined by GC analysis with a slight adjustment to the method described .
Briefly, fat was saponified in a water bath (HWS-12) at 70°C for 1 hr and then methylated by reacting with 7ml 15% boron trifluoride methanol. The saturated NaCl solution and n-hexane (C11) were added and then stratified. The gas chromatography (Shimadzu GC-2010 plus) was equipped with a flame ionization detector. The upper solution was injected into a GC and separated on a SP-2560 column (100 m × 0.25 mm × 0.25 mm, Supelco). The fatty acid was identified by comparison with standard certificate (37 types of fatty acids, Sigma-Aldrich). Results were expressed as a percentage of its peak area to the total peak area. The compounds were identified by comparing and matching their mass spectra of each component with NIST 14 mass spectra database.
The analysis of 16 amino acids was performed with LA8080 amino acid analyzer (Hitachi, Japan). 200 mg of the sample was homogenized by sterile homogenizer (HX-11L) and put in a hydrolytic tube. 1 0ml hydrochloric acid of 6 mol/L (0.5% mercaptoacetic acid) was added to the hydrolytic tube and frozen in a refrigerant for 3 min. The process to vacuum (close to 0 Pa) and fill with high purity nitrogen was repeated for 3 times. Then, the sealed hydrolytic tube was placed in drying oven of a constant temperature (110 ± 2) °C with nitrogen filled or screw cap tightened. Following the hydrolysis for 22 hr, it was taken out for cooling. After the filtrate was filtered, the hydrolytic tube was washed with deionized water for several times. The vacuum dryer was dried at 40 ~ 50 degrees, and the residue was dissolved by 0.02moL/L hydrochloric acid to dilute the volume. After filtration, the solution was analyzed by a 0.45 μm filtration membrane.
The content of cystine was slightly different from that of 16 amino acids by LA8080 amino acid analyzer. After vacuum drying and filtering, the cystine was added with 3ml sodium citrate buffer solution to a constant volume.
The analysis of tryptophan was performed with LC-20 A high-performance liquid chromatography (SHIMADZU). 100 mg of the sample was homogenized by sterile homogenizer (HX-11L, China) and put in a polytetrafluoroethylene tube. 1.5 ml of 4mol/L lithium hydroxide solution was added, filled with nitrogen for 1 min, sealed and put into a drying oven at a constant temperature of 110 ± 2°C for hydrolysis for 20 hr. After cooling to room temperature, the hydrolysate was quantitatively transferred to a 25ml volumetric flask with 0.0085 mol/L sodium acetate buffer solution for constant volume.
The 0.45 μm filtration membrane was used for analysis.

| Electronic nose detection
The PEN 3.5 electronic nose (Airsense) is an array of chemical gas sensors for measurement of volatile compounds within the headspace over stir-frying mutton sao zi samples. The sensor array is composed of 10 gas sensors with metal oxide semiconductors of different selectivity and sensitivity to volatile compounds, with certain specificity. The sensors include W1C (aromatic compounds), W1S (methane, broad range of compounds), W1W (sulfur compounds, terpenes), W2S (broad range, alcohols), W2W (aromatics and organic sulfur compounds), W3C (ammonia, aromatic compounds), W3S (methane and aliphatic compounds), W5C (alkanes and aromatics), W5S (nitrogen oxides), and W6S (hydrocarbons). Before the analysis, 5 g of minced meat samples was put into 20 ml airtight vials and incubated in the water bath at 25°C for 20 min. The chamber was flushed with clean air until the sensor signal returned to the baseline before testing new samples. The adsorbed SPME fiber was inserted into the injection port of GC-MS after extraction, and desorbed at 250°C for 5 min. Helium served as the carrier gas with a flow rate of 2.0 ml/min (constant flow). The temperature gradient in GC oven was as follows: initial 40°C for 3 min, 40°C to 200°C under a 5°C/min rate, and 200°C to 230°C at which holding for 3 min. Electron impact (EI) mode was used at 70 eV, with a full scan range from 50 to 350 amu. The interface temperature 250°C, ion source temperature 230°C, and a solvent delay time of 2.5 min were adopted. The mass spectra of volatiles detected from samples were compared and matched with the mass spectra from NIST 14.0 and the standard compound retention index (retention index, RI). Semi-quantitative determinations were obtained by using 2-methyl-3-heptanone as an internal standard. The volatile compounds acquired were identified for the reverse match factor (similarity > 700) and retention index (RI). The C7 ~ C30 n-alkanes were employed to calculate linear RI of volatile compounds.

| Crude composition of Mutton sao zi
The chemical composition (moisture, protein, and fat) characteristics of mutton sao zi were showed great variations among different processing time points as shown in Table 1

TA B L E 1
Crude composition (%) of mutton sao zi from different processing time (means, ±SD) TA B L E 2 Fatty acids profile (g/100g of total fatty acids) in mutton sao zi from different processing time (means,

TA B L E 3
The 18 amino acid composition (g/100g of total amino acids) in mutton sao zi from different processing time (means,

Stir-frying 20min
Stir-frying 25min processing time. With the extension of processing time of the average moisture content of chicken nuggets decreased during frying at 175°C and 190°C (Bansal et al., 2015). Moisture content ranged from 69.79% for stir-frying 0 min samples to 47.11% for stir-frying 35 min samples. This could be attributed to the difference in stir-frying time.
At the same temperature, longer time stir-frying may cause greater moisture loss.
The protein undergoes Maillard and thermal degradation reaction during heating, and further has some contributions to the quality and flavor of meat products (Wall et al., 2019). After the raw meat was stir-fried for 35 min, the protein content increased from 18.11% to 32.24%. In our research, the protein content did not differ be- Fat is a major precursor of cooked meat flavor, and lipid oxidation is the leading reaction during cooking which helps uncooked meat with bloody taste and little aroma develop the meat flavor (Khan et al., 2015). The raw meat stir-fried for 20 min had the fat content gradually increased (to 9.39%, 9.98%, 11.86%, 12.67%, 14.52%, and 15.47%, respectively), which may be attributable to the moisture loss during the stir-frying process. Boiled samples contained a lower content of protein and fat than the dry heat (baked and fried) treated ones, probably due to a dilution effect of the major water retention (Tavares et al., 2018). The moisture, protein, and fat values did not show significant differences (p > .05).

| Fatty acids profiling of Mutton sao zi
Fatty acids in industrially produced mutton sao zi showed significant difference in content at different processing stages as shown in

| Amino acid analysis of mutton sao zi
Cooking causes protein denaturation and degradation in meat, releasing some free amino acids to participate in flavor formation (Zhao et al., 2017). Amino acids are nutrient components of mutton and precursors for aroma compounds. They directly contribute to the flavor of meat and can be used to evaluate the quality of meat products (Khan et al., 2015). The composition of 18 amino acids in stir-fried mutton sao zi is presented in Table 3. The data showed that amino acid content in mutton increased significantly (p < .05) after stir-frying, due to moisture loss (  (Purchas et al., 2009). Glutamine is the most abundant amino acid in the body of mammals, comprising nearly 60% of the free intracellular amino acids in skeletal muscle (Lopes et al., 2015). In contrast, HIS, Trp, and Cys were recorded at the lowest contents (<1 g/100 g), which might be due to that Trp and Cys were destroyed by thermal hydrolysis (Moughan, 2003). However, in this study, cysteine was completely lost when the muscle tissue was hydrolyzed before any acid oxidation. Most of the analyzed 18 amino acids changed significantly as a result of stir-frying, namely, stir-frying can change the content of amino acids in mutton sao zi. Over the past 15 years, increasing evidence suggests that 6 out of the 15 amino acids measured in samples of beef changed significantly with cooking (Lopes et al., 2015).

| Response signals of E-nose for stir-frying mutton sao zi samples
Electronic noses are devices able to characterize and differentiate the aroma profiles of various food, especially meat and meat products. As seen in Figure 1, the response signals of E-nose for stir-frying mutton sao zi sample at each processing stage and the odor contour curves were different. Stir-frying processing lowered the signal response values of W1C for 0 min sample, suggesting that stir-frying processing could reduce aromatics as toluene (Appendix S1: Table S1) Hence, PCA was further used in the study.

| PCA of E-nose data
As shown in Figure 2, the accumulative variance contribution rate of the first two principal components is > 90%, indicated that the

F I G U R E 3
Clustering heat map of the concentration of volatile compounds in stir-frying mutton sao zi at each processing time

| Volatile compounds analysis stir-frying mutton sao zi
As we all know, the changes in flavor and surface color of meat are due to the Maillard reaction and thermal degradation of lipid, as well as the interaction between the two reaction pathways (Mottram, 1998). To have a better understanding of the flavor of the mutton sao zi, GC-MS was utilized to analyze the volatile compounds at different stir-frying time points. The results showed that 105 volatile compounds were identified (Appendix S1: Table S1) For their low perception thresholds, aldehydes often presented samples with special aroma even at trace amounts (Cai et al., 2016).
The levels of aldehydes were slightly higher in the stir-frying 15 min sample (about 70°C), possibly due to formation of these compounds with the rising temperature (Dermiki et al., 2013). Most aldehydes were derived from lipid oxidation: hexanal, nonanal, octanal, heptanal, pentanal, 2-hexenal, and benzaldehyde (Rasinska et al., 2019), which suggested that the most aldehydes in industrial stir-frying mutton sao zi were produced by lipid oxidation. These compounds also detected in chicken (Man et al., 2018), beef , and pork of black-pig (Zhao et al., 2017) have been reported in the scientific literature. It is interesting to note that the mutton sao zi samples stir-fried in the last 10 min presented the highest amounts of hexanal, which could be the main reason for the high flavor quality. As the main aldehyde produced during lipid oxidation of meat (Shahidi & Zhong, 2010), hexanal increased, maybe attributable to lipid oxidation of unsaturated fatty acids at high temperature (30% fat in the sample). Benzaldehyde (cherry-like odor), the one supposed to provide mutton sao zi special overall aroma, was originated from the Strecker degradation of phenylalanine. Report suggested that benzaldehyde produced almond and burnt sugar special flavor during processing (Cai et al., 2016;Pham et al., 2008). The results for volatile aldehydes from this study reveal that heating promotes rapid oxidation of polyunsaturated fatty acids, which in turn produces more free radicals capable of attacking other fatty acids less prone to oxidize, for example oleic acid, promoting formation of heptanal, octanal, and nonanal among other aldehydes (Roldan et al., 2014).
Statistical analysis showed that alcohol content of mutton sao zi was affected by thermal treatment (Appendix S1: Table S1). As shown in Figures 3 1-octen-3-ol and 1-pentanol from decomposition of lipids were detected in all the samples, while 4-nonanol and 2,3-dimethyl-2,3-butanediol were only observed in the 0-10 min sample stir-fried at a low temperature stage (Appendix S1: Table   S1 and Figure S1). Considering higher threshold value of alcohols, short straight chain alcohols might not contribute any flavor to the product. However, long straight chain alcohols like 1-pentanol, 1-hexanol, 1-dodecanol, and 4-nonanol may contribute aroma to the miso products as reported with relatively lower threshold value (Giri et al., 2010). In addition, branched-chain alcohols like 1-octen-3-ol, (E)-2-octen-1-ol, and 1-nonen-3-ol might contribute significant aroma to the mutton sao zi as they're having lower threshold values.
Aliphatic alcohols might contribute aroma to meat flavor through unsaturated alcohols, the threshold values of which are lower than those of saturated ones (Giri et al., 2010). For example, 1-octen-3-ol F I G U R E 4 Principal component analysis of volatile compounds (GC-MS data) of stir-frying mutton sao zi contributing to a mushroom odor was detected in cooked and grilled lamb at different temperature-time combinations in substantial amounts (Bueno et al., 2014;Wang et al., 2019). However, alcohols with higher odor thresholds, generally, are not deemed as important flavor contributors to meat products (Huang et al., 2019).
Ketones are considered to have a great impact on the aroma of meat and meat products as they give off a peculiar odor and appear in large amount in food (Man et al., 2018). 2,3-octanedione could be counted as one of the important compounds distinguishing stir-frying stages, because it differed significantly in content, so much that detected at some stir-frying stages but not others (Appendix S1: Table S1). Although acetoin (sweet, buttery odor) was not a major odor influencer in grilled beef (Kilgannon et al., 2020), it presented at a high concentration in mutton sao zi and might have a subtle effect on the overall odor perception.
The esters usually have sweet and typical fruity odors generated by esterification of acids and alcohols in meat products . Only ten esters were detected by GC-MS in mutton sao zi samples at different stir-frying time points (Appendix S1: Table S1), including two long-chain esters (benzoic acid, 2-ethylhexyl ester, and Isopropyl myristate), and three lactones (delta-nonalactone, 5-ethyldihydro-2(3H)-furanone, and gamma-dodecalactone). Among them, isopropyl myristate was at the highest level and detected in all the samples, significantly up as the temperature rose in 0-20 min and down as the temperature dropped within 25-35 min. In particular, the lactones could come from the intramolecular esterification of the hydroxy acids .
Of course, furan and pyrazine compounds, such as 2-furanmethanol, 2(5H)-furanone, 2-pentyl-furan, methyl-pyrazine, and 2,3-dimethyl-pyrazine, come from Maillard reactions and (Continues) Strecker degradation (Giri et al., 2010), were all detected during this research. The 2-pentyl-furan from linoleic acid oxidation was found attributable to fatty and meaty aroma, and its formation was normally connected with heat (Man et al., 2018). The last but not the least, some hydrocarbons were detected in mutton sao zi. Hydrocarbons showed a high proportion in the volatile compounds though, they were not considered as a major contributor to the overall flavor of samples for their high odor thresholds (Lu et al., 2008).
In this study, 3-thiophenecarboxaldehyde was detected in late stir-frying period, and its structure revealed that it's originated from free Cys rather than Cys-containing peptides (Zou et al., 2019).
D-Limonene was imparted by flavorings and spice (Z. Wang et al., 2020). Volatile compounds, such as alcohols, aldehydes, ketones, acids, esters, and hydrocarbons, together were closely related to the unique flavor of mutton sao zi.

| Correlation between E-nose and GC-MS
Based on GC-MS analysis, highly abundant volatile components were selected to correlate with E-nose signals, and moisture, protein, fat, fatty acids, and amino acid ( Figure 5).
The highly abundant volatile components selected to correlate with E-nose signals are shown in Figure 5a. The results indicated that the E-nose signal intensities of W1S, W1W, W2S, W2W, and W3S sensors were in positive correlations with hexanal, 1-hexanol,  In contrast, W1C, W3C, and W5C signals had negative correlation with the abundances of hexanal, 1-hexanol, 1-octen-3-ol, 1-pentanol, 4-ethylcyclohexanol, and 2,3-octanedione. The results showed that W1S, W1W, W2S, W2W, and W3S sensors were sensitive to the volatile compounds at different stir-frying time points of industrially produced mutton sao zi. This showed that E-nose can discriminate the different stages of stir-frying by responding specifically to volatile compounds originated from stir-frying mutton sao zi.

| Correlation between E-nose and fatty acid, crude composition, and amino acids
The E-nose analysis of the industrial stir-frying mutton sao zi may be  analysis of GC-MS and E-nose confirmed that E-nose sensors were sensitive to volatile flavor compounds, which further validated the E-nose data. Therefore, E-nose can be used to discriminate industrial stir-frying mutton sao zi from different processing time points.

| CON CLUS ION
In addition, this study also emphasized the feasibility of evaluating flavor of industrial stir-frying mutton sao zi using SPME-GC-MS and E-nose.

ACK N OWLED G M ENTS
This study was financially supported by the State Key Research and Development Plan "Modern Food Processing and Food Storage and Transportation Technology and Equipment" (2018YFD0400101).

CO N FLI C T S O F I NTE R E S T
No conflict of interest exits in the submission of this manuscript, and manuscript is approved by all authors for publication. I would like to declare on behalf of my co-authors that the work described was original research that has not been published previously, and not under consideration for publication elsewhere. We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.