Multielement Principal Component Analysis and Origin Traceability of Rice Based on ICP-MS/MS

In this experiment, inductively coupled plasma tandem mass spectrometry (ICP-MS/MS) was used to determine the content of 30 elements in rice from six places of production and to explore the relationship between the multielement content in rice and the producing area. The contents of Ca, P, S, Zn, Cu, Fe, Mn, K, Mg, Na, Ge, Sb, Ba, Ti, V, Se, As, Sr, Mo, Ni, Co, Cr, Al, Li, Cs, Pb, Cd, B, In, and Sn in rice were determined by ICP-MS/MS in the SQ and MS/MS mode. By passing H 2 , O 2 , He, and NH 3 /He reaction gas into the ICP-MS/MS, respectively, the interference was eliminated by means of in situ mass spectrometry and mass transfer. The detection limit of each element was 0.0000662–0.144mg/kg, and the limit of quantiﬁcation was in the range of 0.000221–0.479mg/kg, the linear correlation coeﬃcient was greater or equal to 0.9987 ( R 2 ≥ 0.9987), and the detection results had low detection limit and great linear regression. Recovery of the method was in the range of 80.6% to 110.5% with spike levels of 0.10–100.00mg/kg, and relative standard deviations were lower than 10%. For the multielement content of rice from diﬀerent producing areas, the principal component factor analysis can get six principal component factors, 87.878% cumulative contribution rate, and the distribution of the principal component scores of each element and diﬀerent producing areas. Based on the multielement content and cluster analysis, the samples were accurately divided into two major categories and six subcategories according to the places of production, which proved that there was a signiﬁcant correlation between the multielement content in rice and the place of production, so that the place of rice origin can be traced.


Introduction
Rice is the main staple food of our country, which contains sugar, protein, fat and dietary fiber, and other main nutrition elements and also contains a lot of necessary trace elements, such as Ca, Fe, Zn, Se, and other mineral elements [1]. Heavy and toxic metals, especially As and Cd, present due to environmental pollution are taken up by the rice plant [2][3][4][5]. In China, rice varieties are rich and diverse, with large planting area span and large quality difference. China is a vast country with diverse climatic and geographical conditions, and the crops have different biological characteristics and physical and chemical indexes. erefore, it is valuable to analyze and compare the differences of multielement contents in rice from south to north China and to provide theoretical basis and technical support for distinguishing rice from different places of origin.

Sample Collection and Preparation.
e samples were collected from six rice-producing areas in Anhui Province, Guangxi Province, Guangdong Province, Jilin Province, Heilongjiang Province, and Inner Mongolia. We purchased common local rice samples with large planting areas in each rice market, a total of 18 batches. ree independent packages were purchased for each batch, and mixed samples were taken to ensure uniformity. e rice samples of each batch were hulled, ground, crushed, and stored in a sealed, low-temperature, and dark place.
In a PTFE digestion tank, each rice sample which weighs 0.3-0.5 g (accurate to 0.001 g) was added to 4 mL HNO 3 and 1 mL 30% H 2 O 2 and soaked for 3-4 h or overnight, the upper cap was screwed, and it was digested with the microwave digestion instrument (CEM MARS6, CEM, Matthews, USA). e conditions of the microwave digestion instrument are shown in Table 1.
en, they were placed on the temperature-controlled electric heating plate (BHW-09C, Shanghai Botong Chemical Technology Co., Ltd., Shanghai, China) and heated at 100°C for 20-30 min for degassing. After cooling, the samples were diluted to 50 mL with deionized water and shook well for later use. For each group of samples, blanks (deionized water and reagents) and reference materials were included throughout the entire sample preparation and analytical process.

Inductively Coupled Plasma Tandem Mass Spectrometry
Analysis.
is experiment was carried out by tandem mass spectrometry. e concentration of 30 isotopes ( 7 Li, 23 Na, 24 Mg, 27 Al, 11 137 Ba, and 208 Pb) in rice was determined by inductively coupled plasma tandem mass spectrometry (Agilent 8900 Series, Agilent, USA). 1 μg/mL mixed solution of Li, Y, Co, Tl, Ce, and Mg was used as the tuning solution, and 0.10 rps speed of the peristaltic pump was used to continuously feed the solution.
rough the tuning program, the conditions of no gas, H 2 , O 2 , He, and NH 3 /He multimode analysis methods were optimized. In the no-gas mode, the monitored ions were 7, 89, and 205. In the He mode, the monitored ions were 59, 89, and 205. In the H 2 mode, the monitored ions were Q 1 � Q 2 � 59, 89, and 205. In the O 2 mode, the monitored ions were Q 1 � Q 2 � 59, Q 1 � 89/Q 2 � 105, and Q 1 � Q 2 � 205. In the NH 3 /He mode, the monitored ions were Q 1 � Q 2 � 59, Q 1 � 89/Q 2 � 191, and Q 1 � Q 2 � 205. Under different modes, RF power was 1550 W, auxiliary gas was 0.90 L/min, plasma gas was 15.0 L/ min, sampling depth was 8.0 mm, and extraction lens was −7.6 V. e instrument's other conditions of ICP-MS/MS are shown in Table 2.
For the selection of element determination mode and reagent gas, this method involves two modes: the SQ (single quadrupole) standard mode and MS/MS tandem mode.
ere are He and no-gas reagent gas modes in the SQ mode and He, NH 3 /He, H 2 , O 2 , and no-gas reagent gas modes in the MS/MS mode. e elements are measured in all modes, and the mode with the lowest detection limit of each element is determined as the best measurement mode. rough the measured experimental conditions and methods, the elements Sc, Y, Rh, and Bi were used as the internal standard elements. Analyzing the experimental data can get a linear fitting standard curve with the X-axis as the concentration point and the Y-axis as the response value. rough this standard curve, the detection limit and background equivalent concentration of the analysis element can be obtained by calculating the element standard deviation. e linear correlation coefficient and range, internal standard elements, limit of detection (LOD), and limit of quantification (LOQ) are shown in Table 3.
At the same time, the content of each element in rice reference materials (GBW10043, GBW10044, and GBW10045) was determined, the standard value was compared, and the recovery rate was calculated to prove the accuracy and reliability of the method, and the recovery experiment was conducted.

Statistical Analysis.
All analyses were conducted in triplicate.
e results reported were the average of these three replicates. Each sample was considered as an assembly of 30 variables represented by the data of chemical information. e analysis data and the fitted linear regression curve were analyzed by Agilent Mass Hunter software (Agilent Inc., USA). A normal distribution test of multielements, principal component analysis, and clustering analysis were performed with SPSS 25.0 software (SPSS, IBM Corp., USA).

Mass Spectrometry Mode Selection and Interference
Elimination. In this experiment, the SQ (single quadrupole) standard mode and MS/MS tandem mode were used to simultaneously determine the concentration of multielement. e elements were measured in different modes and different reaction gas modes, and the element detection limit was used as the criterion to determine the best measurement mode for each element. e results are shown in Table 4.    e interference was eliminated by making full use of the collision mode between the element and the reaction gas. In the SQ mode, the mass ions of 63 Cu, 111 Cd, 118 Sn, 121 Sb, and 208 Pb had the characteristics of high abundance value and less interference. e corresponding Q 2 mass number was the only one that needs to be set during the determination.
In the MS/MS mode, the NH 3 /He mixture gas collided with 7 Li, 24 Mg, 44 Ca, 60 Ni, 95 Mo, and 137 Ba ions in the reaction cell, H 2 collided with 23 Na, 27 Al, 55 Mn, 66 Zn, 72 Ge, 78 Se, 88 Sr, and 115 In ions, and O 2 collided with 39 K and 133 Cs ions, respectively. e interference was eliminated by in situ mass spectrometry, which means the elements only collide with the reaction gas and do not combine with each other. erefore, the mass number of the front and after tetrodes to be set remains unchanged (Q 1 � Q 2 ). However, the system will still have the same amount of heterotopic number signal superposition interference and double charge ion interference; for example, ions Ni ++ , SiH, CO, and NO may interfere with 31 P; ions Zn ++ , NO, and OO may interfere with 32 S; ions CAr and ArO interfere with 52 Cr; ions ArCl, CaCl, and CoO interfere with 75 As; ions ArO and MnH interfere with 56 Fe; and ions Sn ++ , NiH, and MgCl interfere with 59 Co.   Journal of Food Quality 5 erefore, in the determination of some specific elements, if the reactant gas and the element collide with each other to generate ions with a new mass number, the abovementioned interferences can be better avoided. In addition, when the gas collided with the analysis element, new mass ions were formed in the reaction, that is, mass transfer ( Figure 1). In the experiment, NH 3 /He mixture gas can react with 11

Standard Material Determination and Precision.
Multielement determination was performed on the standard materials GBW10043, GBW10044, and GBW10045, and the       Table 5. e average recoveries of element content of reference materials were in the range between 82.9% and 115%. e recoveries of analytes were evaluated by adding the standard solutions with three different concentration levels to the known amounts of samples. e data of recovery and precision are given in Table 6, and the average recoveries of element content in rice were in the range between 80.6% and 110.5%. e RSDs were in the range of 0.4%-8.9%. e measurement results show that this method has high accuracy and meets the requirements of analysis and measurement.

Multielement Analysis of Samples.
ere are obvious differences in the content of Ba, Ge, Co, Cu, Cr, Ti, S, Ca, Mg, Na, Li, and other elements in rice from different producing areas in north and south China. In southern China, there are differences in the content of Na, Mg, K, Ca, V, Ge, Cs, Ba, and other elements in rice produced in Anhui Province, Guangxi Province, and Guangdong Province. However, in northern China, there are obvious differences in the content of B, Na, Ca, P, Cr, Mn, Ni, Co, Zn, Sr, Mo, Cs, and other elements in batches of rice in Jilin Province, Heilongjiang Province, and Inner Mongolia (Table 7). e contents of Al, In, and Sn were not detected.
We conducted further statistical analysis on the abovementioned experimental data, by calculating the standard deviation of each element and judging the difference of each element in different regions according to the degree of dispersion of the value of each element. As shown in Figure 2, the standard deviations of S, P, K, Cd, Mg, and other elements were large, and the degree of dispersion was relatively higher than that of other elements, which can be initially used as indicative elements for traceability.

Multielement Normal Distribution Test.
e Kolmogorov-Smirnov test was conducted on the content of 30 elements in rice from different origins. e asymptotic significance (bilateral) value was calculated. e content data of 24 elements obeyed normal distribution.

Principal Component Analysis.
Principal component analysis (PCA) is a multivariate statistical analysis method that analyses a few variables which can reveal the internal structure sufficiently by studying the relationship between multiple original variables. According to the rule that the characteristic value is greater than 1 and the cumulative variance contribution rate is greater than 80%, six principal component factors were obtained through rotation and extraction factors, and the total contribution rate was 87.878%, indicating that the experimental data can fully reflect the original information (Table 8).
e first principal component is mainly composed of S, Ti, Ni, Cu, Co, Ge, Mo, Cd, Cs, Ba, Zn, and Se elements. e second principal component is mainly composed of Li, B, Mg, K, Ca, P, V, Pb, Fe, and As elements. e third principal component is mainly composed of Mn and Sb elements (Table 9). e first principal component, the second principal component, and the third principal component were used to analyze the contribution of the principal components of samples from different origins (Figure 3). e contribution scores of the principal components of samples from the same origin were concentrated, while the distribution of different origins is scattered. On the whole, rice samples from the north and south of China have a large difference in the contribution scores of the principal components which can be clearly distinguished. is result has certain guiding significance for the distinction of rice from different production places.

Cluster Analysis.
e contents of multielements in rice from different areas were analyzed by cluster analysis. e samples were successfully divided into two categories (the north and south of China) and six subcategories (six riceproducing areas) by the method of intergroup connection ( Figure 4). e results show that there were obvious differences in the contents of multielements in rice from different producing areas, and they had certain regional characteristics. erefore, by measuring the multielement content of rice, it is possible to accurately classify the samples according to the place of origin and finally realize the traceability of the production place of the rice.

Conclusions
In this experiment, the ICP-MS/MS method was developed to determine the content of 30 elements in rice from different production areas. e determination mode and reaction gas conditions were optimized, and the optimal determination conditions were selected for each element in five determination modes of no gas, H 2 , O 2 , He, and NH 3 / He. In addition, in situ mass spectrometry and mass transfer technology were used to eliminate the interference and reduce the detection limit. To achieve the determination of ultratrace elements, we established a complete detection method, which provided a method basis for rice origin traceability. rough the principal component analysis of the multielement content of 18 batches of samples from different origins, the distribution of the six principal components of the samples and the characteristic elements of each principal component were determined. rough cluster analysis, the samples were accurately classified according to the place of production based on the multielement content, which proved that there was a significant correlation between the content of multielement in rice and the place of production, providing technical support and research direction for the traceability of the origin of rice.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare no conflicts of interest regarding the publication of this article.