Multi-element analysis of Baijiu (Chinese liquors) by ICP-MS and their classification according to geographical origin

Multi-element analysis of Baijiu (Chinese liquors) by ICP-MS and their classification according... Objectives: Investigating the element profiles of Be, Al, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Ag, Cd, Cs, La, Pr, Nd, Sm, Eu, Gd, Pb, Th, and U of Baijiu (Chinese liquors), and setting up their classification according to geographical origin. Materials and Methods: Twenty-eight Chinese liquors from Shandong, Guizhou, and Sichuan provinces were analyzed by inductively coupled plasma mass spectrometry with the aid of matrix matching, online internal calibration, and direct injection to determine the concentrations of the aforementioned 26 elements. Multivariate statistical analysis, based on the contents of elements in the liquors, was applied to differentiate the liquors from different origins. Results: Both the cluster analysis based on 11 elements and the discriminant analysis based on 5 elements can separate the liquors of Shandong Province from others. A leave-one-out cross test of the discriminant analysis data resulted in 100 per cent accuracy regarding the recognition ability and prediction ability for the liquors from Shandong Province, and an overall 75.0 per cent accuracy of its prediction for all the 28 total liquors. Limitations: The liquors of Guizhou and Sichuan provinces can not be differentiated successfully. Conclusions: The liquors produced in Shandong Province can be differentiated in a great extent from Guizhou and Sichuan provinces based on the multivariate statistical analysis of the concentration of elements in liquors, while those of Guizhou and Sichuan provinces can not be differentiated successfully due to their geographical adjacency. Key words: ICP-MS; Baijiu; Chinese liquor; element; multivariate statistics. cultivated fermented starters (or commonly called Daqu and/ Introduction or Xiaoqu that contain numerous microorganisms). At present, Chinese liquors (or commonly called Baijiu in China) are produced Chinese liquors are classified into 12 aroma types, including the mainly from sorghum or a mixture of sorghum, rice, wheat, corn, sauce, strong, light, miscellaneous, medicine, sesame, rice, Feng (or glutinous rice, and barley, as the raw materials, plus naturally The Author(s) 2018. Published by Oxford University Press on behalf of Zhejiang University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact jour- nals.permissions@oup.com Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 44 X. Song et al., 2018, Vol. 2, No. 1 phoenix), Te, Chi, Laobaigan, and Fuyu aroma types (Yu, 2009; Materials and methods Zheng and Han, 2016). Apparatus The strong aroma type accounts for the largest proportions of The major analytical instrument was an inductively coupled plasma the commercial markets of Baijiu in China, and customers in favour mass spectrometer (iCAP Q, Thermo Scientific, USA) equipped of the Baijiu with strong aroma type are mainly living in the region with a concentric nebulizer, quartz glass cyclonic spray chamber, of the Huai River and the south side of the Yellow River, whereas nickel sampling cone, and skimmer cone. The optimized condi- the soy sauce aroma-type Baijiu is more popular in the south side tions for analysis of liquors were with the following parameters of the Yangtze River (Xin et al., 2014). In addition, the provinces of (Hou et al., 2017): the internal diameter of the quartz injector was Shandong, Guizhou, and Sichuan are the main areas for Baijiu pro- 1.0 mm (because of the instability of plasma and measuring with ductions in China. For instance, Shandong Province is well known 50 per cent ethanol, the size of the quartz injector was changed for its production of the roasted sesame aroma Baijiu. In compari- from 2.5 to 1.0  mm in order to reduce the injection volume of son, Sichuan Province is famous for its production of the Baijiu with samples), sampling depth 5, torch horizontal position 0.17, torch strong and soy sauce aroma types. vertical position −1.7; RF power 1550 W, spray chamber tempera- Moreover, many elements can be introduced into alcoholic bever- ture 3°C, auxiliary gas flow rate 0.8 l/minute, cool gas flow rate ages during the processing, which can affect not only human’s health 14 l/minute, nebulizer gas flow rate 0.7 l/minute, dwell time 100 but also the taste of alcoholic beverages (Pohl, 2007; Banovic et al., milliseconds, peristaltic pump speed at 20.0 rpm, Q cell gas (He) 2009; Tariba, 2011; Pozo-Bayón et  al., 2012). Elements that affect rate 4.6 ml/minute, and KED voltage was at 3 V. Ultrapure water the taste, the smell, and the colour of distilled spirits can originate was prepared by a GenPure UV-TOC Xcad plus system (Thermo from raw materials and utensils used for processing. Some research- Scientific, USA). ers have evaluated the possible impact of trace elements in distilled spirits (Szymczycha-Madeja et al., 2015). The presence of Cu in most Reagents parts of distilleries results in the trace amount of unpleasant, vola- High purity multi-element standard solution (IV-ICPMS-71A, tile, and S-containing compounds, but have better taste and aroma Inorganic Ventures, Inc., USA) in 10  mg/l was used as a stock so- of the distilled spirits (Reche et  al., 2007). However, an elevated lution for all the subsequent quantitative analyses. A solution of an level of Cu probably poses a threat to human health, mostly due to internal standard, i.e. 6020ISS (Inorganic Ventures, Inc., USA) in increased catalytic formation of acetaldehyde and other aldehydes, 10  mg/l, was also prepared as a stock solution for analytical cali- and ethyl carbamate (Penteado and Masini, 2009). In addition, Cu bration. HPLC grade ethanol and optima grade nitric acid were pur- can catalyze the formation of carcinogenic ethyl carbamate (Neves chased from Thermo Scientific. High purity water with a maximum et  al., 2007). Due to air oxidation, additional Fe compounds can resistivity of 18.20 MΩ·cm prepared from GenPure UV-TOC Xcad be formed, changing the colour of spirits to yellow or even brown plus system was used throughout the experiment. (Flores et  al., 2009). The presence of elevated amounts of minor and trace elements in the distilled spirits may have certain sensory implications. Certain elements, such as Al, Cu, Fe, and Zn, affect Liquor samples stability, colour, aroma, clarity, and other organoleptic properties Twenty-eight Chinese liquors from the aforementioned three prov- (e.g. by imparting a bitter taste to distilled spirits) (Bonic et  al., inces were purchased from local markets in Beijing, China. The con- 2013). Therefore, measurement of the elemental concentrations is of tents of alcohol of the liquors were in a range from 34 to 53 per cent, great importance to consumers and Baijiu producers. At present, the which are the most commonly sold and desirable alcoholic degrees # # important methods for elemental detection include atomic fluores- of Baijiu in China. The liquors labelled with 1 –11 were produced # # cence spectrometry (AFS) (Karadjova et  al., 2007), atomic absorp- in Shandong Province, 12 –17 produced in Guizhou Province, and # # tion spectrometry (AAS) (Boschetti et  al., 2013; Rodríguez-Solana 18 –28 produced in Sichuan Province (Table  1). In addition, the et  al., 2014), inductively coupled plasma emission spectrometry locations of the three provinces in a Chinese map are shown in (ICP-OES) (Pan et al., 2013), and inductively coupled plasma mass Supplementary Figure (see online supplementary material for a col- spectrometry (ICP-MS) (Fiket et al., 2011). Particularly, the ICP-MS our version of this figure). has been most frequently used for the detection of elemental concen- trations and can analyze the multi-elements at the same time. The Preparation of standard solutions most often analyzed elements in liquors include Na, Mg, Al, K, Ca, Preparation of the multi-element standard solutions: 5 ml of ethanol V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, As, Se, Rb, Sr, Pb, Ag, Cd, Cs, Ba, was respectively transferred into 7 volumetric flasks of 10 ml, named and Hg (Zhang et al., 2011; Zhang et al., 2013; Zhang et al., 2014; as A1–A7. The A1 and A2 flasks were added with 1 and 0.5 ml of Ma et al., 2015). the aforementioned multi-element stock solution (see Reagents), In this context, the aims of the present research were as follows: whereas the A3 to A6 flasks were poured with 2, 1, 0.5, and 0.1 ml 1. to apply ICP-MS to investigate the mineral profiles, in terms of Be, of the diluted stock solution of 100 μg/l and then diluted with 2 Al, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Ag, Cd, Cs, La, Pr, per cent HNO solution to a fixed volume, of which the contents of Nd, Sm, Eu, Gd, Pb, Th, and U of 28 Baijiu (Chinese liquors) from multi-elements were in six concentrations, i.e. 1.0, 5.0, 10.0, 20.0, Shandong, Guizhou, and Sichuan provinces; 2.  to use multivariate 50.0, and 100.0 μg/l. The A7 flask was diluted only with 2 per cent statistics, including the non-parametric test, cluster analysis (CA), HNO solution without addition of multi-elements solution, which and discriminant analysis (DA), to differentiate the Chinese liquors will be used as a blank control. by their original geographical regions depending on their elemental Preparation of an internal standard solution: a volume of 0.5 ml composition; and 3.  to extract and discuss the effective elemental of the IS stock solution of 10  mg/l was transferred into a 500  ml indicators for the liquor classification based on their geographical volumetric flask and then diluted with ultrapure water to the final regions, which were further validated by a leave-one-out cross test to concentration of 10 μg/l. judge the obtained DA data and prove the accuracy of its prediction. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Multi-element analysis of Baijiu and its classification, 2018, Vol. 2, No. 1 45 Table  1. Quantitative mean values, standard deviations, and concentration ranges of 26 elements in 28 Chinese liquors produced in Shandong, Guizhou, and Sichuan provinces and values of LOD for 26 elements. −1 # # # # # # Element Shandong Province (1 –11 ) Guizhou Province (12 –17 ) Sichuan Province (18 –28 ) LOD/μg∙l Mean ± SD Range Mean ± SD Range Mean ± SD Range Al 0.0957 1813.98 ± 1896.38 103.79–5792.49 190.0 ± 212.11 3.90–554.41 382.85 ± 413.82 17.12–1146.92 Mn 0.0072 73.32 ± 97.07 4.75–334.05 27.87 ± 31.42 0.82–71.18 15.0 ± 13.47 0.80–46.69 Fe 0.2189 624.59 ± 445.14 262.78–1786.66 250.41 ± 366.80 10.69–987.91 103.29 ± 83.52 13.54–257.57 Cu 0.0145 93.61 ± 160.25 11.00–505.46 4.14 ± 5.24 0.0–11.54 3.66 ± 4.89 0.32–16.68 Zn 0.0465 154.85 ± 195.98 19.77–698.20 11.95 ± 10.80 1.52–31.45 17.86 ± 22.54 2.82–80.49 Be 0.0126 0.54 ± 0.08 0.45–0.72 0.56 ± 0.17 0.44–0.90 0.51 ± 0.04 0.47–0.62 V 0.0040 6.47 ± 14.78 0.39–50.28 1.02 ± 0.88 0.23–2.69 4.36 ± 4.40 0.27–14.79 Cr 0.0021 14.85 ± 22.79 2.93–80.32 4.96 ± 3.63 1.93–11.67 11.62 ± 12.46 1.47–33.26 Co 0.0011 0.89 ± 0.44 0.37–1.67 0.51 ± 0.27 0.30–1.02 0.62 ± 0.28 0.33–1.26 Ni 0.0091 10.91 ± 8.95 3.92–35.11 6.27 ± 7.03 0.38–19.51 8.84 ± 6.57 1.57–18.93 As 0.0007 2.91 ± 2.70 0.89–8.14 3.83 ± 3.34 0.28–10.03 4.66 ± 2.48 0.80–46.69 Se 0.0243 0.64 ± 0.44 0.25–1.41 0.32 ± 0.10 0.23–0.49 0.31 ± 0.08 0.22–0.47 Rb 0.0030 1.07 ± 0.85 0.26–3.18 1.44 ± 1.63 0.15–3.98 1.15 ± 0.81 0.25–2.73 Sr 0.0009 15.97 ± 13.01 1.67–38.42 8.88 ± 9.22 0.47–23.69 8.72 ± 8.13 0.93–27.19 Ag 0.0027 2.32 ± 0.02 2.31–2.38 2.31 ± 0.01 2.30–2.32 2.31 ± 0.01 2.30–2.33 Cd 0.0013 0.31 ± 0.07 0.23–0.47 0.30 ± 0.11 0.22–0.50 0.71 ± 0.93 0.24–3.08 Cs 0.0008 0.23 ± 0.02 0.20–0.28 0.22 ± 0.02 0.21–0.27 0.24 ± 0.04 0.20–0.30 La 0.0003 0.57 ± 0.42 0.31–1.68 0.33 ± 0.08 0.27–0.48 0.36 ± 0.08 0.27–0.52 Pr 0.0001 0.41 ± 0.11 0.34–0.70 0.35 ± 0.02 0.33–0.38 0.35 ± 0.02 0.33–0.40 Nd 0.0017 0.51 ± 0.46 0.20–1.75 0.23 ± 0.08 0.17–0.39 0.25 ± 0.08 0.17–0.40 Sm 0.0010 0.28 ± 0.10 0.21–0.54 0.22 ± 0.02 0.20–0.26 0.22 ± 0.01 0.20–0.24 Eu 0.0004 0.28 ± 0.02 0.27–0.33 0.27 ± 0.01 0.26–0.28 0.27 ± 0.01 0.26–0.31 Gd 0.0009 0.30 ± 0.10 0.22–0.56 0.24 ± 0.03 0.22–0.30 0.23 ± 0.01 0.22–0.25 Pb 0.0033 2.79 ± 5.82 0.08–19.64 1.19 ± 2.24 0.0–5.71 0.97 ± 2.13 0.0–6.92 Th 0.0002 1.34 ± 0.03 1.33–1.42 1.33 ± 0.01 1.33–1.34 1.34 ± 0.02 1.32–1.37 U 0.0004 0.96 ± 0.08 0.90–1.16 0.92 ± 0.03 0.89–0.97 0.93 ± 0.05 0.89–1.00 Sample analysis by ICP-MS Non-parametric test The sample solutions of the blank, multi-element standards, and Since the concentrations of 26 elements did not follow a normal 28 liquor samples (without dilution) were respectively analyzed by distribution, thus, the non-parametric test was used to carry out an ICP-MS under aforementioned optimized conditions (Hou et  al., inference statistical analysis. In this study, the Jonckheere–Terpstra 2017), with an online internal calibration of 10 μg/l of Sc, Y, In, Rh, non-parametric test, rather than a Wald–Wolfowitz non-parametric Bi, and Tb. Optimization of the apparatus conditions was described test that involves two independent samples (Paneque et  al., 2010), in Apparatus. was used to check the discriminant capacity of each element in the liquors from three provinces. If the asymptotic significance of an ele- ment of liquors is less than 0.05, then the element is considered a Multivariate statistical analysis discriminant variable. As a result, the following 11 elements, includ- The statistical software SPSS 18.0 was used for the non-parametric ing Al, Mn, Fe, Cu, Zn, Se, Nd, Sm, Gd, Pb, and Th, were determined test, CA, and DA based on the concentrations of elements in order to to be the discriminant elements of the liquors, which were used in differentiate 28 liquors according to its origin. subsequent multivariate analyses. Results and discussion Cluster analysis Analysis of multi-elements by ICP-MS The relevant reports for Chinese liquors are rare to use the informa- tion of elements in liquors to trace back their geographical origins, ICP-MS was used for the determination of concentrations of 26 ele- although there are a lot of similar reports for wines. Back in 1979, it ments, including Be, Al, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, was found that wines from different regions contained characteristic Sr, Ag, Cd, Cs, La, Pr, Nd, Sm, Eu, Gd, Pb, Th, and U, in 28 Chinese elements in terms of types and contents (Kwan et al., 1979). Later, liquors, which are listed in Table 1 with their quantitative mean val- multivariate statistical methods, such as principal component analy- ues, standard deviations, and concentration ranges. Table 1 also lists sis (PCA) and canonical variate analysis, were applied to successfully the limit of detection (LOD) of 26 elements, which depend on the classify the white wines (Moret et  al., 1994). This case prompted stability of the blank measurement. In addition, the elements Al, Fe, other researchers using the multivariate statistical methods to ana- Zn, Se, and Gd were found to have significant differences (P < 0.05) lyze the origin of wines, based on their elemental concentrations, and between the liquors from Shandong and Sichuan provinces, and Al those in the soil of the vineyard (Kment et al., 2005; Coetzee et al., and Se also had significant differences between the liquors from 2014; Vystavna et  al., 2014). The previous studies suggested that Shandong and Guizhou provinces, after the independent-samples there were several possible sources of minerals in wines, either from t-test according to the mean values of elements. Besides, there was no the natural source such as soil and rock, or from wine-making prac- significant difference between the liquors produced from Guizhou tices, or from extraneous pollutions in agricultural practice (Kment and Sichuan provinces. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 46 X. Song et al., 2018, Vol. 2, No. 1 Figure 1. Dendrogram of cluster analysis of 28 liquors produced in Shandong, Guizhou, and Sichuan provinces. et  al., 2005; Kruzlicova et  al., 2013). These factors could not only dendrogram according to the standardized data. Figure 1 shows the significantly affect the elemental compositions of the final products, results of 28 liquors according to the contents of 11 elements by a but also render the correlation between the wine products and their hierarchical cluster, in which the first column of data is the serial elemental compositions to be more complex, because not all the ele- number of liquors and the second column on the far left represents ments are reliable for authentication of wines, and only some of the the manufacturing provinces of the liquors. stable elements in wines can be used for product classification (Thiel Figure 1 shows that 28 liquors can be divided into two groups. The et al., 2004). Based on the previous studies, the common elements to first group includes the liquors produced from the Guizhou and Sichuan classify the geographical origin of wines included Fe, Mg, Mn, and provinces, which cannot be clearly distinguished from each other. Zn (Coetzee et  al., 2005; Razic et  al., 2007; Paneque et  al., 2010; However, on the other hand, it also reflects the similar geographical Geana et  al., 2013; Selih et  al., 2014). It was reported that wines location and climate of these two neighbouring provinces. The second from different regions could be classified according to the contents group combined the liquors from the Shandong Province, which can of their inherent elements (Paneque et  al., 2010). Therefore, hier- completely be separated from the liquors from Sichuan and Guizhou archical cluster, which is one of the common methods of CA and provinces, except the No. 15 liquor. The results demonstrated that the illustrated by a dendrogram, was applied in an effort to differentiate CA method was efficient and suitable for the classification of liquors. the liquors based on the contents of elements. Discriminant analysis The contents of 11 elements of 28 liquors were analyzed by z-scores standardized processing, and then the Ward’s hierarchical DA is also a common statistical method to deal with the classifica- method and squared Euclidean distance were used to obtain the tion issues (Paneque et al., 2010). Particularly, it is better to use it for Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Multi-element analysis of Baijiu and its classification, 2018, Vol. 2, No. 1 47 the observed samples under the conditions that the classifications of have obviously provided a better classification of the liquors. The the research objects are known. For example, it was used to classify DA also shows that liquors from the Guizhou and Sichuan prov- 40 wines from three different regions according to the variables of inces have many similarities that are difficult to be separated from 20 elements (Coetzee et al., 2005). In order to further discriminate each other, or classified. the analyzed liquors, a discrimination model, reflected by discrimi- Tables 2 and 3 list the results of classification of 28 liquors nation functions, was established through stepwise DA according from three provinces, and recognition ability and prediction abil- to the contents of 26 elements, followed by filtration of the effec- ity obtained from the leave-one-out cross validation technique tive variables, which enable us to classify the liquors according to based on five variables (i.e. Se, Ag, Al, Th, and Cu). As shown their manufacturing regions. The ability of classification (also called in Table  2, the recognition ability of liquors from Shandong, the recognition ability) was performed by comparing the number Guizhou, and Sichuan provinces was 100, 83.3, and 63.6 per of correctly classified samples to the total number of samples. In cent, respectively, with an average value of recognition ability in addition, leave-one-out cross validation test was carried out in order 82.1 per cent calculated from all 28 liquors. Furthermore, using to provide the prediction ability to judge the discrimination model. the leave-one-out cross test to validate the aforementioned results However, the prediction ability is usually lower than the recognition of the recognition ability (Table  3) resulted in a prediction abil- ability which is performed by the discriminant model. ity of liquors from Shandong Province to an accuracy of 100 per In this research, five effective variables, including Se, Ag, Al, Th, cent, which means that the model based on the variables of Se, and Cu, were selected (P  <  0.05) for the discrimination of the liq- Ag, Al, Th, and Cu was sufficient to differentiate the liquors of uors from three provinces. In addition, two discriminant functions Shandong Province from other liquors, whereas the low predic- (shown below) were obtained based on the five variables, tion values of 66.7 and 54.5 per cent for liquors from the Sichuan and Guizhou provinces, respectively, meant the similar characters F4 = 00 .. 00 () Al -+ 013() Cu 8.295() Se in terms of the element content in the liquors from these two provinces. The result was in consistence with those obtained from       +- 80.. 417 () Ag 97 642(Th))-62.055, the CA mentioned above. It was reported that the elements in wines were significantly F1 = -- 00 .. 4C () u 00 328() Se ++ 23.. 16() Ag 44 620() Th -112.886, affected and closely related to soil of the vineyard (Kment et  al., 2005). Similarly, liquors from different geographical origins have where F means the discriminant function 1 and F means the discri- a strong link with the local environment, such as water and soil, 1 2 minant function 2. as well as their processing methods. The soil in Shandong Province Figure 2 profiles the scatter plot reflected by the DA of liquors is mainly composed of brown soil and cinnamon soil, whereas the from three provinces based on the two functions, where its X-axis soil in Guizhou and Sichuan provinces is mainly composed of red is labelled as Dist 1 that represents the discriminant score from soil and yellow soil. Such kinds of differences in soil types have function 1, and its Y-axis is labelled as Dist 2 that represents the remarkably affected the type and the amounts of metal elements in discriminant score from function 2. As shown in Figure 2, the liq- Chinese liquors. In addition, the Baijiu processing method, such as uors from Shandong Province can completely be separated from the difference of raw materials, fermentation methods, and fermen- those produced in the Guizhou and Sichuan provinces according to tation equipment, can significantly influence the elements in liquors. the Dist 1 (or discriminant function 1). However, the liquors from Moreover, all commodity liquors need to be blended with the base the Guizhou and Sichuan provinces, to a large extent, have over- liquors and local water before their sales in markets, which may laps in the figure, which means that they cannot be separated from bring some characteristic elements too. Of course, the most impor- each other according to the discriminant functions 1 and 2.  The tant factor is ascribed to the soil in different geographical loca- results were consistent with the CA results mentioned above, but tions that contain different types and levels of elements. Shandong Figure 2. The scatter plot of discriminant analysis of 28 liquors from the Shandong, Guizhou, and Sichuan provinces based on the two functions. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 48 X. Song et al., 2018, Vol. 2, No. 1 Table 2. Classification of 28 Chinese liquors and recognition ability based on five elemental variables (i.e. Se, Ag, Al, Th, and Cu). Variety Recognition ability, % 1 (Shandong Province) 2 (Guizhou Province) 3 (Sichuan Province) 1 (Shandong Province) 100.0 11 0 0 2 (Guizhou Province) 83.3 0 5 1 3 (Sichuan Province) 63.6 0 4 7 Total 82.1 Table  3. Classification of 28 Chinese liquors and prediction ability using leave-one-out cross validation test based on five elemental variables (i.e. Se, Ag, Al, Th, and Cu). Variety Prediction ability, % 1 (Shandong Province) 2 (Guizhou Province) 3 (Sichuan Province) 1 (Shandong Province) 100.0 11 0 0 2 (Guizhou Province) 66.7 0 4 2 3 (Sichuan Province) 54.5 0 5 6 Total 75.0 Province is located in the eastern coast line of China, whereas References Guizhou Province is located in the inner southwest of China, adja- Banovic, M., Kirin, J., Curko, N., Ganic, K. (2009). Influence of vintage on cent to Sichuan Province that is located in the inner west of China. Cu, Fe, Zn and Pb content in some Croatian red wines. Czech Journal of Obviously, the geographical difference determined the characteristic Food Sciences, 27: 1–3. elements in Chinese Baijiu. Bonic, M., Tesevic, V., Nikicevic, N., Cvejic, J., Milosavljevic, S., Vajs, V. (2013). The content of heavy metals in Serbian old plum brandies. Journal of the Serbian Chemical Society, 78: 933–945. Conclusions Boschetti, W., et al. (2013). Detection of the origin of Brazilian wines based on The concentrations of 26 elements in 28 Chinese liquors pro- the determination of only four elements using high-resolution continuum duced from the Shandong, Guizhou, and Sichuan provinces were source flame AAS. Talanta, 111: 147–155. determined by ICP-MS and subsequently analyzed by multivariate Coetzee, P. P., Steffens, F. E., Eiselen, R. J., Augustyn, O. P., Balcaen, L., Vanhaecke, F. (2005). Multi-element analysis of South African wines by statistical analysis in an effort to make classifications and/or authen- ICP-MS and their classification according to geographical origin. Journal tication. The CA method based on the concentrations of 11 most of Agricultural and Food Chemistry, 53: 5060–5066. discriminant variables (i.e. Al, Mn, Fe, Cu, Zn, Se, Nd, Sm, Gd, Pb, Coetzee, P. P., van Jaarsveld, F. P., Vanhaecke, F. (2014). Intraregional clas- and Th) was carried out for 28 liquors. The CA results showed that sification of wine via ICP-MS elemental fingerprinting. Food Chemistry, liquors produced in Shandong Province can be separated, in a great 164: 485–492. extent, from those produced in Guizhou and Sichuan provinces, but Fiket, Ž., Mikac, N., Kniewald, G. (2011). Arsenic and other trace elements in liquors produced in the Guizhou and Sichuan provinces could not wines of eastern Croatia. Food Chemistry, 126: 941–947. be separated from each other. In comparison, the DA has provided Flores, C., Figueroa, J., Wrobel, K., Wrobel, K. (2009). ICP-MS multi-element a better result. The DA based on five elements, including Se, Ag, Al, profiles and HPLC determination of furanic compounds in commercial Th, and Cu, could effectively differentiate the liquors in Shandong tequila. European Food Research and Technology, 228: 951–958. Geana, I., Iordache, A., Ionete, R., Marinescu, A., Ranca, A., Culea, M. (2013). Province from those produced in Guizhou and Sichuan provinces, Geographical origin identification of Romanian wines by ICP-MS elemen- which was further confirmed by a leave-one-out cross test with over - tal analysis. Food Chemistry, 138: 1125–1134. all values of recognition ability and prediction ability of liquors on Hou, M., Li, Z., Sun, X., Ma, Y., Zheng, F., Sun, B. (2017). Determination 100 per cent. However, the indices for other liquors from other two of 24 kinds of metal elements in Baijiu (Chinese Liquors) by ICP-MS provinces need to be improved in regards of liquor classification. using direct sampling. Journal of Chinese Institute of Food Science and In conclusion, a technique combining the ICP-MS and multivariate Technology, 17: 239–246. statistics could be a useful method to classify food products, based Karadjova, I. B., Lampugnani, L., D’Ulivo, A., Onor, M., Tsalev, D. L. (2007). on some of their characteristic markers, such as mineral elements Determination of lead in wine by hydride generation atomic fluorescence and aroma. spectrometry in the presence of hexacyanoferrate(III). Analytical and Bioanalytical Chemistry, 388: 801–807. Kment, P., Mihaljevič, M., Ettler, V., Šebek, O., Strnad, L., Rohlová, L. (2005). Supplementary material Differentiation of Czech wines using multielement composition—a com- Supplementary material is available at Food Quality and Safety parison with vineyard soil. Food Chemistry, 91: 157–165. Kruzlicova, D., Fiket, Ž., Kniewald, G. (2013). Classification of Croatian wine online. varieties using multivariate analysis of data obtained by high resolution ICP-MS analysis. Food Research International, 54: 621–626. Kwan, W., Kowalski, B., Skogerboe, R. (1979). Pattern recognition analy- Funding sis of elemental data. Wines of Vitis vinifera cv. Pinot Noir from France This work was financially supported by the National Key R&D and the United States. Journal of Agricultural and Food Chemistry, 27: Program of China (2016YFD0400500), the National Natural 1321–1326. Science Foundation of China (31601556), and Beijing Natural Ma, X., Liu, L., Shi, Q., Ke, R., Yang, K., Chi, Y., Yin, J., Song, Q. (2015). Science Foundation Program and Scientific Research Key Program of Determination of 39 trace elements in Chinese liquor by ICP-MS. Liquor Making, 42: 45–48. Beijing Municipal Commission of Education (KZ201410011015). Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Multi-element analysis of Baijiu and its classification, 2018, Vol. 2, No. 1 49 Moret, I., Scarponi, G., Cescon, P. (1994). Chemometric characterization and Selih, V. S., Sala, M., Drgan, V. (2014). Multi-element analysis of wines by classification of five Venetian white wines. Journal of Agricultural and ICP-MS and ICP-OES and their classification according to geographical Food Chemistry, 42: 1143–1153. origin in Slovenia. Food Chemistry, 153: 414–423. Neves, E., Oliveira, A., Fernandes, A., Nobrega, J. (2007). Simple and efficient Szymczycha-Madeja, A., Welna, M., Jamroz, P., Lesniewicz, A., Pohl, P. (2015). elimination of Cu (II) in sugar-cane spirits. Food Chemistry, 101: 33–36. Advances in assessing the elemental composition of distilled spirits using Pan, X., Tang, J., Chen, Q., Wu, P., Han, J. (2013). Evaluation of direct sam- atomic spectrometry. Trends in Analytical Chemistry, 64: 127–135. pling method for trace elements analysis in Chinese rice wine by ICP–OES. Tariba, B. (2011). Metals in wine–impact on wine quality and health out- European Food Research and Technology, 236: 531–535. comes. Biological Trace Element Research, 144: 143–156. Paneque, P., Álvarez-Sotomayor, M., Clavijo, A., Gómez, I. (2010). Metal con- Thiel, G., Geisler, G., Blechschmidt, I., Danzer, K. (2004). Determination of tent in southern Spain wines and their classification according to origin trace elements in wines and classification according to their provenance. and ageing. Microchemical Journal, 94: 175–179. Analytical and Bioanalytical Chemistry, 378: 1630–1636. Penteado, J., Masini, J. (2009). Multivariate analysis for the classification dif- Vystavna, Y., Rushenko, L., Diadin, D., Klymenko, O., Klymenko, M. (2014). ferentiation of Brazilian sugarcane spirits by analysis of organic and inor- Trace metals in wine and vineyard environment in Southern Ukraine. Food ganic compounds. Analytical Letters, 42: 2747–2757. Chemistry, 146: 339–344. Pohl, P. (2007). What do metals tell us about wine? Trends in Analytical Xin, C., Ma, D., Wang, W., Dong, Q., Zhao, J. (2014). Discussion on liquor Chemistry, 26: 1–9. making. Liquor Making, 41: 32–38. Pozo-Bayón, M. Á., Monagas, M., Bartolomé, B., Moreno-Arribas, M. V. Yu, Q. (2009). “Traditional Liquor Brewing Technology”. Chinese Light (2012). Wine features related to safety and consumer health: an integrated Industry Press, Beijing, China. perspective. Critical Reviews in Food Science and Nutrition, 52: 31–54. Zhang, Y., Xia, Y., Yang, Y., He, K., Cheng, T. (2011). Determination of five Razic, S., Cokesa, D., Sremac, S. (2007). Multivariate data visualization meth- heavy metals in Chinese liquor with application of ICP-MS online internal ods based on elemental analysis of wines by atomic absorption spectrom- calibration. China Brewing, 229: 169–170. etry. Journal of the Serbian Chemical Society, 72: 1487–1492. Zhang, J., Tian, Z., Lu, Y., Chen, X., Sun, Z. (2013). Application of induc- Reche, R. V., Neto, A. F., Silva, A. A., Galinaro, C. A., Osti, R. Z., Franco, D. W. tively coupled plasma mass spectrometry (ICP-MS) for determination of (2007). Influence of type of distillation apparatus on chemical profiles of 28 trace elements in Chinese Liquor. Food Science, 34: 257–260. Brazilian cachaças. Journal of Agricultural and Food Chemistry, 55: 6603–6608. Zhang, Y., Li, S., Shi, B., Gao, X. (2014). Determination of trace elements in Rodríguez-Solana, R., Salgado, J., Domínguez, J., Cortés, S. (2014). Assessment liquor with application of ICP-MS. Liquor Making, 41: 87–89. of minerals in aged grape marc distillates by FAAS/FAES and ICP-MS. Zheng, X., Han, B. (2016). Baijiu, Chinese liquor: history, classification and Characterization and safety evaluation. Food Control, 35: 49–55. manufacture. Journal of Ethnic Foods, 3: 19–25. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Food Quality and Safety Oxford University Press

Multi-element analysis of Baijiu (Chinese liquors) by ICP-MS and their classification according to geographical origin

Free
7 pages

Loading next page...
 
/lp/ou_press/multi-element-analysis-of-baijiu-chinese-liquors-by-icp-ms-and-their-F9kN0oIOxs
Publisher
Zhejiang University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Zhejiang University Press.
ISSN
2399-1399
eISSN
2399-1402
D.O.I.
10.1093/fqsafe/fyx030
Publisher site
See Article on Publisher Site

Abstract

Objectives: Investigating the element profiles of Be, Al, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Ag, Cd, Cs, La, Pr, Nd, Sm, Eu, Gd, Pb, Th, and U of Baijiu (Chinese liquors), and setting up their classification according to geographical origin. Materials and Methods: Twenty-eight Chinese liquors from Shandong, Guizhou, and Sichuan provinces were analyzed by inductively coupled plasma mass spectrometry with the aid of matrix matching, online internal calibration, and direct injection to determine the concentrations of the aforementioned 26 elements. Multivariate statistical analysis, based on the contents of elements in the liquors, was applied to differentiate the liquors from different origins. Results: Both the cluster analysis based on 11 elements and the discriminant analysis based on 5 elements can separate the liquors of Shandong Province from others. A leave-one-out cross test of the discriminant analysis data resulted in 100 per cent accuracy regarding the recognition ability and prediction ability for the liquors from Shandong Province, and an overall 75.0 per cent accuracy of its prediction for all the 28 total liquors. Limitations: The liquors of Guizhou and Sichuan provinces can not be differentiated successfully. Conclusions: The liquors produced in Shandong Province can be differentiated in a great extent from Guizhou and Sichuan provinces based on the multivariate statistical analysis of the concentration of elements in liquors, while those of Guizhou and Sichuan provinces can not be differentiated successfully due to their geographical adjacency. Key words: ICP-MS; Baijiu; Chinese liquor; element; multivariate statistics. cultivated fermented starters (or commonly called Daqu and/ Introduction or Xiaoqu that contain numerous microorganisms). At present, Chinese liquors (or commonly called Baijiu in China) are produced Chinese liquors are classified into 12 aroma types, including the mainly from sorghum or a mixture of sorghum, rice, wheat, corn, sauce, strong, light, miscellaneous, medicine, sesame, rice, Feng (or glutinous rice, and barley, as the raw materials, plus naturally The Author(s) 2018. Published by Oxford University Press on behalf of Zhejiang University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact jour- nals.permissions@oup.com Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 44 X. Song et al., 2018, Vol. 2, No. 1 phoenix), Te, Chi, Laobaigan, and Fuyu aroma types (Yu, 2009; Materials and methods Zheng and Han, 2016). Apparatus The strong aroma type accounts for the largest proportions of The major analytical instrument was an inductively coupled plasma the commercial markets of Baijiu in China, and customers in favour mass spectrometer (iCAP Q, Thermo Scientific, USA) equipped of the Baijiu with strong aroma type are mainly living in the region with a concentric nebulizer, quartz glass cyclonic spray chamber, of the Huai River and the south side of the Yellow River, whereas nickel sampling cone, and skimmer cone. The optimized condi- the soy sauce aroma-type Baijiu is more popular in the south side tions for analysis of liquors were with the following parameters of the Yangtze River (Xin et al., 2014). In addition, the provinces of (Hou et al., 2017): the internal diameter of the quartz injector was Shandong, Guizhou, and Sichuan are the main areas for Baijiu pro- 1.0 mm (because of the instability of plasma and measuring with ductions in China. For instance, Shandong Province is well known 50 per cent ethanol, the size of the quartz injector was changed for its production of the roasted sesame aroma Baijiu. In compari- from 2.5 to 1.0  mm in order to reduce the injection volume of son, Sichuan Province is famous for its production of the Baijiu with samples), sampling depth 5, torch horizontal position 0.17, torch strong and soy sauce aroma types. vertical position −1.7; RF power 1550 W, spray chamber tempera- Moreover, many elements can be introduced into alcoholic bever- ture 3°C, auxiliary gas flow rate 0.8 l/minute, cool gas flow rate ages during the processing, which can affect not only human’s health 14 l/minute, nebulizer gas flow rate 0.7 l/minute, dwell time 100 but also the taste of alcoholic beverages (Pohl, 2007; Banovic et al., milliseconds, peristaltic pump speed at 20.0 rpm, Q cell gas (He) 2009; Tariba, 2011; Pozo-Bayón et  al., 2012). Elements that affect rate 4.6 ml/minute, and KED voltage was at 3 V. Ultrapure water the taste, the smell, and the colour of distilled spirits can originate was prepared by a GenPure UV-TOC Xcad plus system (Thermo from raw materials and utensils used for processing. Some research- Scientific, USA). ers have evaluated the possible impact of trace elements in distilled spirits (Szymczycha-Madeja et al., 2015). The presence of Cu in most Reagents parts of distilleries results in the trace amount of unpleasant, vola- High purity multi-element standard solution (IV-ICPMS-71A, tile, and S-containing compounds, but have better taste and aroma Inorganic Ventures, Inc., USA) in 10  mg/l was used as a stock so- of the distilled spirits (Reche et  al., 2007). However, an elevated lution for all the subsequent quantitative analyses. A solution of an level of Cu probably poses a threat to human health, mostly due to internal standard, i.e. 6020ISS (Inorganic Ventures, Inc., USA) in increased catalytic formation of acetaldehyde and other aldehydes, 10  mg/l, was also prepared as a stock solution for analytical cali- and ethyl carbamate (Penteado and Masini, 2009). In addition, Cu bration. HPLC grade ethanol and optima grade nitric acid were pur- can catalyze the formation of carcinogenic ethyl carbamate (Neves chased from Thermo Scientific. High purity water with a maximum et  al., 2007). Due to air oxidation, additional Fe compounds can resistivity of 18.20 MΩ·cm prepared from GenPure UV-TOC Xcad be formed, changing the colour of spirits to yellow or even brown plus system was used throughout the experiment. (Flores et  al., 2009). The presence of elevated amounts of minor and trace elements in the distilled spirits may have certain sensory implications. Certain elements, such as Al, Cu, Fe, and Zn, affect Liquor samples stability, colour, aroma, clarity, and other organoleptic properties Twenty-eight Chinese liquors from the aforementioned three prov- (e.g. by imparting a bitter taste to distilled spirits) (Bonic et  al., inces were purchased from local markets in Beijing, China. The con- 2013). Therefore, measurement of the elemental concentrations is of tents of alcohol of the liquors were in a range from 34 to 53 per cent, great importance to consumers and Baijiu producers. At present, the which are the most commonly sold and desirable alcoholic degrees # # important methods for elemental detection include atomic fluores- of Baijiu in China. The liquors labelled with 1 –11 were produced # # cence spectrometry (AFS) (Karadjova et  al., 2007), atomic absorp- in Shandong Province, 12 –17 produced in Guizhou Province, and # # tion spectrometry (AAS) (Boschetti et  al., 2013; Rodríguez-Solana 18 –28 produced in Sichuan Province (Table  1). In addition, the et  al., 2014), inductively coupled plasma emission spectrometry locations of the three provinces in a Chinese map are shown in (ICP-OES) (Pan et al., 2013), and inductively coupled plasma mass Supplementary Figure (see online supplementary material for a col- spectrometry (ICP-MS) (Fiket et al., 2011). Particularly, the ICP-MS our version of this figure). has been most frequently used for the detection of elemental concen- trations and can analyze the multi-elements at the same time. The Preparation of standard solutions most often analyzed elements in liquors include Na, Mg, Al, K, Ca, Preparation of the multi-element standard solutions: 5 ml of ethanol V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, As, Se, Rb, Sr, Pb, Ag, Cd, Cs, Ba, was respectively transferred into 7 volumetric flasks of 10 ml, named and Hg (Zhang et al., 2011; Zhang et al., 2013; Zhang et al., 2014; as A1–A7. The A1 and A2 flasks were added with 1 and 0.5 ml of Ma et al., 2015). the aforementioned multi-element stock solution (see Reagents), In this context, the aims of the present research were as follows: whereas the A3 to A6 flasks were poured with 2, 1, 0.5, and 0.1 ml 1. to apply ICP-MS to investigate the mineral profiles, in terms of Be, of the diluted stock solution of 100 μg/l and then diluted with 2 Al, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Ag, Cd, Cs, La, Pr, per cent HNO solution to a fixed volume, of which the contents of Nd, Sm, Eu, Gd, Pb, Th, and U of 28 Baijiu (Chinese liquors) from multi-elements were in six concentrations, i.e. 1.0, 5.0, 10.0, 20.0, Shandong, Guizhou, and Sichuan provinces; 2.  to use multivariate 50.0, and 100.0 μg/l. The A7 flask was diluted only with 2 per cent statistics, including the non-parametric test, cluster analysis (CA), HNO solution without addition of multi-elements solution, which and discriminant analysis (DA), to differentiate the Chinese liquors will be used as a blank control. by their original geographical regions depending on their elemental Preparation of an internal standard solution: a volume of 0.5 ml composition; and 3.  to extract and discuss the effective elemental of the IS stock solution of 10  mg/l was transferred into a 500  ml indicators for the liquor classification based on their geographical volumetric flask and then diluted with ultrapure water to the final regions, which were further validated by a leave-one-out cross test to concentration of 10 μg/l. judge the obtained DA data and prove the accuracy of its prediction. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Multi-element analysis of Baijiu and its classification, 2018, Vol. 2, No. 1 45 Table  1. Quantitative mean values, standard deviations, and concentration ranges of 26 elements in 28 Chinese liquors produced in Shandong, Guizhou, and Sichuan provinces and values of LOD for 26 elements. −1 # # # # # # Element Shandong Province (1 –11 ) Guizhou Province (12 –17 ) Sichuan Province (18 –28 ) LOD/μg∙l Mean ± SD Range Mean ± SD Range Mean ± SD Range Al 0.0957 1813.98 ± 1896.38 103.79–5792.49 190.0 ± 212.11 3.90–554.41 382.85 ± 413.82 17.12–1146.92 Mn 0.0072 73.32 ± 97.07 4.75–334.05 27.87 ± 31.42 0.82–71.18 15.0 ± 13.47 0.80–46.69 Fe 0.2189 624.59 ± 445.14 262.78–1786.66 250.41 ± 366.80 10.69–987.91 103.29 ± 83.52 13.54–257.57 Cu 0.0145 93.61 ± 160.25 11.00–505.46 4.14 ± 5.24 0.0–11.54 3.66 ± 4.89 0.32–16.68 Zn 0.0465 154.85 ± 195.98 19.77–698.20 11.95 ± 10.80 1.52–31.45 17.86 ± 22.54 2.82–80.49 Be 0.0126 0.54 ± 0.08 0.45–0.72 0.56 ± 0.17 0.44–0.90 0.51 ± 0.04 0.47–0.62 V 0.0040 6.47 ± 14.78 0.39–50.28 1.02 ± 0.88 0.23–2.69 4.36 ± 4.40 0.27–14.79 Cr 0.0021 14.85 ± 22.79 2.93–80.32 4.96 ± 3.63 1.93–11.67 11.62 ± 12.46 1.47–33.26 Co 0.0011 0.89 ± 0.44 0.37–1.67 0.51 ± 0.27 0.30–1.02 0.62 ± 0.28 0.33–1.26 Ni 0.0091 10.91 ± 8.95 3.92–35.11 6.27 ± 7.03 0.38–19.51 8.84 ± 6.57 1.57–18.93 As 0.0007 2.91 ± 2.70 0.89–8.14 3.83 ± 3.34 0.28–10.03 4.66 ± 2.48 0.80–46.69 Se 0.0243 0.64 ± 0.44 0.25–1.41 0.32 ± 0.10 0.23–0.49 0.31 ± 0.08 0.22–0.47 Rb 0.0030 1.07 ± 0.85 0.26–3.18 1.44 ± 1.63 0.15–3.98 1.15 ± 0.81 0.25–2.73 Sr 0.0009 15.97 ± 13.01 1.67–38.42 8.88 ± 9.22 0.47–23.69 8.72 ± 8.13 0.93–27.19 Ag 0.0027 2.32 ± 0.02 2.31–2.38 2.31 ± 0.01 2.30–2.32 2.31 ± 0.01 2.30–2.33 Cd 0.0013 0.31 ± 0.07 0.23–0.47 0.30 ± 0.11 0.22–0.50 0.71 ± 0.93 0.24–3.08 Cs 0.0008 0.23 ± 0.02 0.20–0.28 0.22 ± 0.02 0.21–0.27 0.24 ± 0.04 0.20–0.30 La 0.0003 0.57 ± 0.42 0.31–1.68 0.33 ± 0.08 0.27–0.48 0.36 ± 0.08 0.27–0.52 Pr 0.0001 0.41 ± 0.11 0.34–0.70 0.35 ± 0.02 0.33–0.38 0.35 ± 0.02 0.33–0.40 Nd 0.0017 0.51 ± 0.46 0.20–1.75 0.23 ± 0.08 0.17–0.39 0.25 ± 0.08 0.17–0.40 Sm 0.0010 0.28 ± 0.10 0.21–0.54 0.22 ± 0.02 0.20–0.26 0.22 ± 0.01 0.20–0.24 Eu 0.0004 0.28 ± 0.02 0.27–0.33 0.27 ± 0.01 0.26–0.28 0.27 ± 0.01 0.26–0.31 Gd 0.0009 0.30 ± 0.10 0.22–0.56 0.24 ± 0.03 0.22–0.30 0.23 ± 0.01 0.22–0.25 Pb 0.0033 2.79 ± 5.82 0.08–19.64 1.19 ± 2.24 0.0–5.71 0.97 ± 2.13 0.0–6.92 Th 0.0002 1.34 ± 0.03 1.33–1.42 1.33 ± 0.01 1.33–1.34 1.34 ± 0.02 1.32–1.37 U 0.0004 0.96 ± 0.08 0.90–1.16 0.92 ± 0.03 0.89–0.97 0.93 ± 0.05 0.89–1.00 Sample analysis by ICP-MS Non-parametric test The sample solutions of the blank, multi-element standards, and Since the concentrations of 26 elements did not follow a normal 28 liquor samples (without dilution) were respectively analyzed by distribution, thus, the non-parametric test was used to carry out an ICP-MS under aforementioned optimized conditions (Hou et  al., inference statistical analysis. In this study, the Jonckheere–Terpstra 2017), with an online internal calibration of 10 μg/l of Sc, Y, In, Rh, non-parametric test, rather than a Wald–Wolfowitz non-parametric Bi, and Tb. Optimization of the apparatus conditions was described test that involves two independent samples (Paneque et  al., 2010), in Apparatus. was used to check the discriminant capacity of each element in the liquors from three provinces. If the asymptotic significance of an ele- ment of liquors is less than 0.05, then the element is considered a Multivariate statistical analysis discriminant variable. As a result, the following 11 elements, includ- The statistical software SPSS 18.0 was used for the non-parametric ing Al, Mn, Fe, Cu, Zn, Se, Nd, Sm, Gd, Pb, and Th, were determined test, CA, and DA based on the concentrations of elements in order to to be the discriminant elements of the liquors, which were used in differentiate 28 liquors according to its origin. subsequent multivariate analyses. Results and discussion Cluster analysis Analysis of multi-elements by ICP-MS The relevant reports for Chinese liquors are rare to use the informa- tion of elements in liquors to trace back their geographical origins, ICP-MS was used for the determination of concentrations of 26 ele- although there are a lot of similar reports for wines. Back in 1979, it ments, including Be, Al, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, was found that wines from different regions contained characteristic Sr, Ag, Cd, Cs, La, Pr, Nd, Sm, Eu, Gd, Pb, Th, and U, in 28 Chinese elements in terms of types and contents (Kwan et al., 1979). Later, liquors, which are listed in Table 1 with their quantitative mean val- multivariate statistical methods, such as principal component analy- ues, standard deviations, and concentration ranges. Table 1 also lists sis (PCA) and canonical variate analysis, were applied to successfully the limit of detection (LOD) of 26 elements, which depend on the classify the white wines (Moret et  al., 1994). This case prompted stability of the blank measurement. In addition, the elements Al, Fe, other researchers using the multivariate statistical methods to ana- Zn, Se, and Gd were found to have significant differences (P < 0.05) lyze the origin of wines, based on their elemental concentrations, and between the liquors from Shandong and Sichuan provinces, and Al those in the soil of the vineyard (Kment et al., 2005; Coetzee et al., and Se also had significant differences between the liquors from 2014; Vystavna et  al., 2014). The previous studies suggested that Shandong and Guizhou provinces, after the independent-samples there were several possible sources of minerals in wines, either from t-test according to the mean values of elements. Besides, there was no the natural source such as soil and rock, or from wine-making prac- significant difference between the liquors produced from Guizhou tices, or from extraneous pollutions in agricultural practice (Kment and Sichuan provinces. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 46 X. Song et al., 2018, Vol. 2, No. 1 Figure 1. Dendrogram of cluster analysis of 28 liquors produced in Shandong, Guizhou, and Sichuan provinces. et  al., 2005; Kruzlicova et  al., 2013). These factors could not only dendrogram according to the standardized data. Figure 1 shows the significantly affect the elemental compositions of the final products, results of 28 liquors according to the contents of 11 elements by a but also render the correlation between the wine products and their hierarchical cluster, in which the first column of data is the serial elemental compositions to be more complex, because not all the ele- number of liquors and the second column on the far left represents ments are reliable for authentication of wines, and only some of the the manufacturing provinces of the liquors. stable elements in wines can be used for product classification (Thiel Figure 1 shows that 28 liquors can be divided into two groups. The et al., 2004). Based on the previous studies, the common elements to first group includes the liquors produced from the Guizhou and Sichuan classify the geographical origin of wines included Fe, Mg, Mn, and provinces, which cannot be clearly distinguished from each other. Zn (Coetzee et  al., 2005; Razic et  al., 2007; Paneque et  al., 2010; However, on the other hand, it also reflects the similar geographical Geana et  al., 2013; Selih et  al., 2014). It was reported that wines location and climate of these two neighbouring provinces. The second from different regions could be classified according to the contents group combined the liquors from the Shandong Province, which can of their inherent elements (Paneque et  al., 2010). Therefore, hier- completely be separated from the liquors from Sichuan and Guizhou archical cluster, which is one of the common methods of CA and provinces, except the No. 15 liquor. The results demonstrated that the illustrated by a dendrogram, was applied in an effort to differentiate CA method was efficient and suitable for the classification of liquors. the liquors based on the contents of elements. Discriminant analysis The contents of 11 elements of 28 liquors were analyzed by z-scores standardized processing, and then the Ward’s hierarchical DA is also a common statistical method to deal with the classifica- method and squared Euclidean distance were used to obtain the tion issues (Paneque et al., 2010). Particularly, it is better to use it for Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Multi-element analysis of Baijiu and its classification, 2018, Vol. 2, No. 1 47 the observed samples under the conditions that the classifications of have obviously provided a better classification of the liquors. The the research objects are known. For example, it was used to classify DA also shows that liquors from the Guizhou and Sichuan prov- 40 wines from three different regions according to the variables of inces have many similarities that are difficult to be separated from 20 elements (Coetzee et al., 2005). In order to further discriminate each other, or classified. the analyzed liquors, a discrimination model, reflected by discrimi- Tables 2 and 3 list the results of classification of 28 liquors nation functions, was established through stepwise DA according from three provinces, and recognition ability and prediction abil- to the contents of 26 elements, followed by filtration of the effec- ity obtained from the leave-one-out cross validation technique tive variables, which enable us to classify the liquors according to based on five variables (i.e. Se, Ag, Al, Th, and Cu). As shown their manufacturing regions. The ability of classification (also called in Table  2, the recognition ability of liquors from Shandong, the recognition ability) was performed by comparing the number Guizhou, and Sichuan provinces was 100, 83.3, and 63.6 per of correctly classified samples to the total number of samples. In cent, respectively, with an average value of recognition ability in addition, leave-one-out cross validation test was carried out in order 82.1 per cent calculated from all 28 liquors. Furthermore, using to provide the prediction ability to judge the discrimination model. the leave-one-out cross test to validate the aforementioned results However, the prediction ability is usually lower than the recognition of the recognition ability (Table  3) resulted in a prediction abil- ability which is performed by the discriminant model. ity of liquors from Shandong Province to an accuracy of 100 per In this research, five effective variables, including Se, Ag, Al, Th, cent, which means that the model based on the variables of Se, and Cu, were selected (P  <  0.05) for the discrimination of the liq- Ag, Al, Th, and Cu was sufficient to differentiate the liquors of uors from three provinces. In addition, two discriminant functions Shandong Province from other liquors, whereas the low predic- (shown below) were obtained based on the five variables, tion values of 66.7 and 54.5 per cent for liquors from the Sichuan and Guizhou provinces, respectively, meant the similar characters F4 = 00 .. 00 () Al -+ 013() Cu 8.295() Se in terms of the element content in the liquors from these two provinces. The result was in consistence with those obtained from       +- 80.. 417 () Ag 97 642(Th))-62.055, the CA mentioned above. It was reported that the elements in wines were significantly F1 = -- 00 .. 4C () u 00 328() Se ++ 23.. 16() Ag 44 620() Th -112.886, affected and closely related to soil of the vineyard (Kment et  al., 2005). Similarly, liquors from different geographical origins have where F means the discriminant function 1 and F means the discri- a strong link with the local environment, such as water and soil, 1 2 minant function 2. as well as their processing methods. The soil in Shandong Province Figure 2 profiles the scatter plot reflected by the DA of liquors is mainly composed of brown soil and cinnamon soil, whereas the from three provinces based on the two functions, where its X-axis soil in Guizhou and Sichuan provinces is mainly composed of red is labelled as Dist 1 that represents the discriminant score from soil and yellow soil. Such kinds of differences in soil types have function 1, and its Y-axis is labelled as Dist 2 that represents the remarkably affected the type and the amounts of metal elements in discriminant score from function 2. As shown in Figure 2, the liq- Chinese liquors. In addition, the Baijiu processing method, such as uors from Shandong Province can completely be separated from the difference of raw materials, fermentation methods, and fermen- those produced in the Guizhou and Sichuan provinces according to tation equipment, can significantly influence the elements in liquors. the Dist 1 (or discriminant function 1). However, the liquors from Moreover, all commodity liquors need to be blended with the base the Guizhou and Sichuan provinces, to a large extent, have over- liquors and local water before their sales in markets, which may laps in the figure, which means that they cannot be separated from bring some characteristic elements too. Of course, the most impor- each other according to the discriminant functions 1 and 2.  The tant factor is ascribed to the soil in different geographical loca- results were consistent with the CA results mentioned above, but tions that contain different types and levels of elements. Shandong Figure 2. The scatter plot of discriminant analysis of 28 liquors from the Shandong, Guizhou, and Sichuan provinces based on the two functions. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 48 X. Song et al., 2018, Vol. 2, No. 1 Table 2. Classification of 28 Chinese liquors and recognition ability based on five elemental variables (i.e. Se, Ag, Al, Th, and Cu). Variety Recognition ability, % 1 (Shandong Province) 2 (Guizhou Province) 3 (Sichuan Province) 1 (Shandong Province) 100.0 11 0 0 2 (Guizhou Province) 83.3 0 5 1 3 (Sichuan Province) 63.6 0 4 7 Total 82.1 Table  3. Classification of 28 Chinese liquors and prediction ability using leave-one-out cross validation test based on five elemental variables (i.e. Se, Ag, Al, Th, and Cu). Variety Prediction ability, % 1 (Shandong Province) 2 (Guizhou Province) 3 (Sichuan Province) 1 (Shandong Province) 100.0 11 0 0 2 (Guizhou Province) 66.7 0 4 2 3 (Sichuan Province) 54.5 0 5 6 Total 75.0 Province is located in the eastern coast line of China, whereas References Guizhou Province is located in the inner southwest of China, adja- Banovic, M., Kirin, J., Curko, N., Ganic, K. (2009). Influence of vintage on cent to Sichuan Province that is located in the inner west of China. Cu, Fe, Zn and Pb content in some Croatian red wines. Czech Journal of Obviously, the geographical difference determined the characteristic Food Sciences, 27: 1–3. elements in Chinese Baijiu. Bonic, M., Tesevic, V., Nikicevic, N., Cvejic, J., Milosavljevic, S., Vajs, V. (2013). The content of heavy metals in Serbian old plum brandies. Journal of the Serbian Chemical Society, 78: 933–945. Conclusions Boschetti, W., et al. (2013). Detection of the origin of Brazilian wines based on The concentrations of 26 elements in 28 Chinese liquors pro- the determination of only four elements using high-resolution continuum duced from the Shandong, Guizhou, and Sichuan provinces were source flame AAS. Talanta, 111: 147–155. determined by ICP-MS and subsequently analyzed by multivariate Coetzee, P. P., Steffens, F. E., Eiselen, R. J., Augustyn, O. P., Balcaen, L., Vanhaecke, F. (2005). Multi-element analysis of South African wines by statistical analysis in an effort to make classifications and/or authen- ICP-MS and their classification according to geographical origin. Journal tication. The CA method based on the concentrations of 11 most of Agricultural and Food Chemistry, 53: 5060–5066. discriminant variables (i.e. Al, Mn, Fe, Cu, Zn, Se, Nd, Sm, Gd, Pb, Coetzee, P. P., van Jaarsveld, F. P., Vanhaecke, F. (2014). Intraregional clas- and Th) was carried out for 28 liquors. The CA results showed that sification of wine via ICP-MS elemental fingerprinting. Food Chemistry, liquors produced in Shandong Province can be separated, in a great 164: 485–492. extent, from those produced in Guizhou and Sichuan provinces, but Fiket, Ž., Mikac, N., Kniewald, G. (2011). Arsenic and other trace elements in liquors produced in the Guizhou and Sichuan provinces could not wines of eastern Croatia. Food Chemistry, 126: 941–947. be separated from each other. In comparison, the DA has provided Flores, C., Figueroa, J., Wrobel, K., Wrobel, K. (2009). ICP-MS multi-element a better result. The DA based on five elements, including Se, Ag, Al, profiles and HPLC determination of furanic compounds in commercial Th, and Cu, could effectively differentiate the liquors in Shandong tequila. European Food Research and Technology, 228: 951–958. Geana, I., Iordache, A., Ionete, R., Marinescu, A., Ranca, A., Culea, M. (2013). Province from those produced in Guizhou and Sichuan provinces, Geographical origin identification of Romanian wines by ICP-MS elemen- which was further confirmed by a leave-one-out cross test with over - tal analysis. Food Chemistry, 138: 1125–1134. all values of recognition ability and prediction ability of liquors on Hou, M., Li, Z., Sun, X., Ma, Y., Zheng, F., Sun, B. (2017). Determination 100 per cent. However, the indices for other liquors from other two of 24 kinds of metal elements in Baijiu (Chinese Liquors) by ICP-MS provinces need to be improved in regards of liquor classification. using direct sampling. Journal of Chinese Institute of Food Science and In conclusion, a technique combining the ICP-MS and multivariate Technology, 17: 239–246. statistics could be a useful method to classify food products, based Karadjova, I. B., Lampugnani, L., D’Ulivo, A., Onor, M., Tsalev, D. L. (2007). on some of their characteristic markers, such as mineral elements Determination of lead in wine by hydride generation atomic fluorescence and aroma. spectrometry in the presence of hexacyanoferrate(III). Analytical and Bioanalytical Chemistry, 388: 801–807. Kment, P., Mihaljevič, M., Ettler, V., Šebek, O., Strnad, L., Rohlová, L. (2005). Supplementary material Differentiation of Czech wines using multielement composition—a com- Supplementary material is available at Food Quality and Safety parison with vineyard soil. Food Chemistry, 91: 157–165. Kruzlicova, D., Fiket, Ž., Kniewald, G. (2013). Classification of Croatian wine online. varieties using multivariate analysis of data obtained by high resolution ICP-MS analysis. Food Research International, 54: 621–626. Kwan, W., Kowalski, B., Skogerboe, R. (1979). Pattern recognition analy- Funding sis of elemental data. Wines of Vitis vinifera cv. Pinot Noir from France This work was financially supported by the National Key R&D and the United States. Journal of Agricultural and Food Chemistry, 27: Program of China (2016YFD0400500), the National Natural 1321–1326. Science Foundation of China (31601556), and Beijing Natural Ma, X., Liu, L., Shi, Q., Ke, R., Yang, K., Chi, Y., Yin, J., Song, Q. (2015). Science Foundation Program and Scientific Research Key Program of Determination of 39 trace elements in Chinese liquor by ICP-MS. Liquor Making, 42: 45–48. Beijing Municipal Commission of Education (KZ201410011015). Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Multi-element analysis of Baijiu and its classification, 2018, Vol. 2, No. 1 49 Moret, I., Scarponi, G., Cescon, P. (1994). Chemometric characterization and Selih, V. S., Sala, M., Drgan, V. (2014). Multi-element analysis of wines by classification of five Venetian white wines. Journal of Agricultural and ICP-MS and ICP-OES and their classification according to geographical Food Chemistry, 42: 1143–1153. origin in Slovenia. Food Chemistry, 153: 414–423. Neves, E., Oliveira, A., Fernandes, A., Nobrega, J. (2007). Simple and efficient Szymczycha-Madeja, A., Welna, M., Jamroz, P., Lesniewicz, A., Pohl, P. (2015). elimination of Cu (II) in sugar-cane spirits. Food Chemistry, 101: 33–36. Advances in assessing the elemental composition of distilled spirits using Pan, X., Tang, J., Chen, Q., Wu, P., Han, J. (2013). Evaluation of direct sam- atomic spectrometry. Trends in Analytical Chemistry, 64: 127–135. pling method for trace elements analysis in Chinese rice wine by ICP–OES. Tariba, B. (2011). Metals in wine–impact on wine quality and health out- European Food Research and Technology, 236: 531–535. comes. Biological Trace Element Research, 144: 143–156. Paneque, P., Álvarez-Sotomayor, M., Clavijo, A., Gómez, I. (2010). Metal con- Thiel, G., Geisler, G., Blechschmidt, I., Danzer, K. (2004). Determination of tent in southern Spain wines and their classification according to origin trace elements in wines and classification according to their provenance. and ageing. Microchemical Journal, 94: 175–179. Analytical and Bioanalytical Chemistry, 378: 1630–1636. Penteado, J., Masini, J. (2009). Multivariate analysis for the classification dif- Vystavna, Y., Rushenko, L., Diadin, D., Klymenko, O., Klymenko, M. (2014). ferentiation of Brazilian sugarcane spirits by analysis of organic and inor- Trace metals in wine and vineyard environment in Southern Ukraine. Food ganic compounds. Analytical Letters, 42: 2747–2757. Chemistry, 146: 339–344. Pohl, P. (2007). What do metals tell us about wine? Trends in Analytical Xin, C., Ma, D., Wang, W., Dong, Q., Zhao, J. (2014). Discussion on liquor Chemistry, 26: 1–9. making. Liquor Making, 41: 32–38. Pozo-Bayón, M. Á., Monagas, M., Bartolomé, B., Moreno-Arribas, M. V. Yu, Q. (2009). “Traditional Liquor Brewing Technology”. Chinese Light (2012). Wine features related to safety and consumer health: an integrated Industry Press, Beijing, China. perspective. Critical Reviews in Food Science and Nutrition, 52: 31–54. Zhang, Y., Xia, Y., Yang, Y., He, K., Cheng, T. (2011). Determination of five Razic, S., Cokesa, D., Sremac, S. (2007). Multivariate data visualization meth- heavy metals in Chinese liquor with application of ICP-MS online internal ods based on elemental analysis of wines by atomic absorption spectrom- calibration. China Brewing, 229: 169–170. etry. Journal of the Serbian Chemical Society, 72: 1487–1492. Zhang, J., Tian, Z., Lu, Y., Chen, X., Sun, Z. (2013). Application of induc- Reche, R. V., Neto, A. F., Silva, A. A., Galinaro, C. A., Osti, R. Z., Franco, D. W. tively coupled plasma mass spectrometry (ICP-MS) for determination of (2007). Influence of type of distillation apparatus on chemical profiles of 28 trace elements in Chinese Liquor. Food Science, 34: 257–260. Brazilian cachaças. Journal of Agricultural and Food Chemistry, 55: 6603–6608. Zhang, Y., Li, S., Shi, B., Gao, X. (2014). Determination of trace elements in Rodríguez-Solana, R., Salgado, J., Domínguez, J., Cortés, S. (2014). Assessment liquor with application of ICP-MS. Liquor Making, 41: 87–89. of minerals in aged grape marc distillates by FAAS/FAES and ICP-MS. Zheng, X., Han, B. (2016). Baijiu, Chinese liquor: history, classification and Characterization and safety evaluation. Food Control, 35: 49–55. manufacture. Journal of Ethnic Foods, 3: 19–25. Downloaded from https://academic.oup.com/fqs/article-abstract/2/1/43/4823048 by Ed 'DeepDyve' Gillespie user on 16 March 2018

Journal

Food Quality and SafetyOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off