Method validation for determination of thallium by inductively coupled plasma mass spectrometry and monitoring of various foods in South Korea

Tl was determined using an Agilent 7700x (Agilent Technologies, Santa Clara, CA, USA) ICP-MS with ion monitoring at mass-to-charge ratios (m/z) 203 and 205 to gather the data. The specific instrument parameters and analytical conditions are described in Table S9.

The reagents used for this study were of analytical grade, and the water was deion-ized and purified below 18.2 MΩ using a YL WPS System (Young Lin Instrument Co., Anyang-si, Kyunggi-do, Korea). As a standard solution for Tl, the periodic table mix 1 for the ICP solution (Sigma-Aldrich, St. Louis, MO, USA) was diluted with 2.5% nitric acid to 100 μg kg⁻¹ of stock solution. Analytical grade nitric acid (Chemitop, Jin-cheon, Cheongbuk-do, Korea) and hydrogen peroxide (Chemitop, Jincheon, Cheongbuk-do, Korea) were used. BCR-679 (white cabbage) certified reference materials (CRM) for method validation and quality assurance were purchased from the European Commission (European Com-mission, Joint Research Centre, Institute for Reference Materials and Measurements, Geel, Belgium).

According to KNHANES, monitored samples were chosen to represent commonly consumed foods. Additionally, we referred to a report on the reassessment of toxic metal(loid)s criteria in South Korean food published by the Korea Ministry of Food and Drug Safety (MFDS) [54,56]. For representativeness, agricultural (113 samples), animal (12 samples), and fishery products (50 samples) were purchased from more than three distinct locations. For processed foods (129 samples), more than three products were purchased per manufacturer. Before homogenizing, the samples were washed with deionized water, and the surface moisture and non-edible components (e.g., peels, seeds, and bones) were removed. Fruits and vegetables were dried at 65 °C for 18 h before homogenization with a lab blender for 60 s and sealed at −18 °C. For infusion teas, 1.2 g of tea sachet was immersed in 100 mL of distilled water at 80 °C for 2 min. We used the infused water as the samples.

The current study’s sample preparation procedure was based on the elemental analysis manual of the US Food and Drug Administration (US FDA) [57] and guidelines for the analysis of toxic metal(loid)s in food by MFDS [58]. We classified foods into six categories based on their energy (i.e., calories) and matrix. The calorific value of each food was determined using the MFDS food nutrient database [59]. The classification included non-fatty solids, hydrated solids, fatty solids, salty solids, non-fatty liquids, and fatty liquids. Table 5 lists the sample weights for various sample classifications. The samples were pre-decomposed on a hot plate with 4 mL of nitric acid (70 vol%) and 1 mL of hydrogen peroxide (30 vol%) in a polytetrafluoroethylene (PTFE) vessel. After cooling at 25 °C, 3 mL of nitric acid (70 vol%) was added. PTFE vessels were sealed and placed in a microwave digestion system (ETHOS Easy, Milestone, Italy). After digestion, the samples were cooled at 25 °C and diluted to 20.0 g (in the case of salty foods, they were diluted to 80.0 g) with distilled water. All samples were analyzed in triplicates. The hydrated solid foods were converted to wet-based concentrations by considering the moisture content before drying.

Food comprises various organic components. In the method validation for determining toxic metal(loid)s, the matrix effect is one of the critical factors in ICP-MS analysis. We selected six representative foods for each category, taking into account the effects of matrices, such as salts and fats, and validated the method for determining Tl. We confirmed the factors for method validation in terms of linearity, trueness, accuracy, and precision. The linearity for external standard calibration curves was measured in stock solution diluted to 0.025, 0.05, 0.1, 0.25, 0.5, 1.0, and 2.0 μg kg⁻¹. The trueness of Tl determination was verified using CRM Tl mass fractions. The measured values were compared with the certification values seven times. The accuracy and precision were calculated intraday and interday. The standard solution was spiked into each representative sample, resulting in the final concentrations of 0.1 μg kg⁻¹ (low concentration), 0.5 μg kg⁻¹ (medium concentration), and 1.0 μg kg⁻¹ (high concentration) in seven replicates, respectively. The measurement uncertainty was calculated for the rice matrix spiked with a medium concentration of Tl standard solution. The sources of measurement uncertainty were expressed in the form of a fishbone diagram (Figure S1).

The MLOD and MLOQ were determined using the standard addition method, which involved adding five different Tl concentrations (0.05, 0.1, 0.25, 0.5, and 1.0 μg kg⁻¹) to representative foods for calibration curve with seven repetitions. The equations are as follows:

where 3.14 = student t factor for seven replicates, S = slope mean, and σ = standard deviation of intercept. For quality control, CRM was analyzed for every batch of samples, and recovery was checked within 80–120%.

The analysis data were processed using SPSS 21 (IBM Corporation, New York, NY, USA) to compare the levels among the food groups. Significant differences were defined using t-tests or one-way ANOVA. Statistical significance was set at p < 0.05. For statistical calculation, values below MLOD were replaced with half the MLOD values, and its natural logarithm was used.