The direct slurry sampling electrothermal vaporization (DSS-ETV) system was fabricated in our laboratory, and its schematic diagram is shown in Figure 1. The DSS-ETV mainly consists of a graphite furnace as a vaporizer, a temperature sensor and control system, a cooling system, a gas line system, an interface with ICP-MS and a power supply. Graphite tube (28 × 7.4 mm, ɸ5.9 mm, Beijing Xiangchenghe Photoelectric Technology Co., Ltd., Beijing, China) was employed as a sampling boat in the graphite furnace to heat (longitudinally with graphite electrodes) a slurry sample; as well, it was coated with pyrolytic carbon via heating in methane and ammonia according to the previous study (26). To monitor the heating temperature of ETV, photoelectric sensors (S1226-44BQ, Beijing Hamamatsu Photoelectric Technology Co., Ltd., Beijing, China) were used for room temperature to 1,200°C, and infrared sensors (SA120BSK, Wuxi Shiao Technology Co., Wuxi, China) for 1,000 to 2,600°C. The ETV temperature from room temperature to 2,600°C was controlled by a proportional integral derivative (PID) algorithm based on a computer and power supply (AC, 50 Hz). The cooling system consists of water and air cooling lines, in which cooling water was fed into the copper electrode holders and cooling gas flowed between the graphite tube and the graphite furnace. The gas line system was controlled by two gas mass flow meters (GMFC) and a manual valve for carrier gas, gas turbulator and protection gases, respectively. The ETV was connected to an ICP-MS (SUPEC 7000, Hangzhou Puyu Technology Development Co., Ltd., Hangzhou, China) by a cyclone nebulizer (internal capacity ~40 mL) made of polyvinylidene fluoride (PFA), which acts as an external signal delay device to improve the analysis precision. The detailed parameters of the ETV-ICP-MS instrumentation are shown in Supplementary Tables S1, S2.

To validate the proposed method, ICP-MS was used for the determination of Se, Cd, As and Pb in food samples after microwave digestion. HNO₃ and H₂O₂ (guaranteed grade) were purchased from Sinopharm Chemical Regent Co., Ltd. (Shanghai, China). The detailed operating conditions of the microwave digestion system (790S, Hangzhou Puyu Technology Development Co., Ltd., Hangzhou, China) and ICP-MS (SUPEC 7000, Hangzhou Puyu Technology Development Co., Ltd., Hangzhou, China) are shown in Supplementary Table S3.

Standard solutions (1,000 mg L⁻¹) of Se, Cd, As and Pb were purchased from the National Center for Analysis and Testing of Nonferrous Metals and Electronic Materials (Shanghai, China). Deionized (DI) water (15 MΩ) was prepared by a Milli-Q Elix Essential purification system (Millipore, Burlington, MA, United States). Working standards were obtained by diluting the stock solutions. The certified reference materials (CRMs) of rice (GBW10010a and GBW10045a), soybean (GBW10013), and tea (GBW10016a and GBW10052a) were purchased from the National Research Center for Reference Materials (Beijing, China). HNO₃ and H₂O₂ (guaranteed grade) were used for sample digestion.

The analytical procedures for the determination of Se, Cd, As and Pb in food samples by the proposed ETV-ICP-MS method are described as follows: (1) Dehydration: 10 μL of slurry sample was taken into the graphite tube and kept at 200°C for 15 s to remove moisture in sample, (2) Ashing: ETV was heated to 450°C within 20 s and held for 20 s under 100 mL min⁻¹ carrier gas and 300 mL min⁻¹ gas turbulator, respectively, to remove organic substances in sample outside via an outlet, (3) Vaporization: After shutting down the outlet, the carrier and gas turbulatores were adjusted to 500 mL min⁻¹ and 600 mL min⁻¹, respectively and kept for 5 s to form a stable working atmosphere. Then, the ETV was heated to 2,300°C within 1 s and then held for 5 s to vaporize Se, Cd, As and Pb analytes, (4) Detection: The vaporized analytes were transported into the SDD and then mixed into the ICP source for the quantitative analysis by the quadrupole mass spectrometer, (5) Cleaning: 2500°C for 7 s was performed under 500 mL min⁻¹ carrier gas and 800 mL min⁻¹ gas turbulator to remove possible residues in the instrumental system.

According to the Chinese national standard of GB 5009.268 (27), food samples were measured for the presences of Se, Cd, As and Pb to verify the proposed method. A 1 g food sample was weighed and then mixed with 1 mL DI water, 2 mL HNO₃ and 1 mL H₂O₂ in modified tetrafluoroethylene (TFM) digestion tube. After the microwave digestion as shown in Supplementary Table S3, the sample solution was cooled to room temperature in the end. The final sample solution was transferred to a 50 mL centrifuge tube and diluted to 15 mL with 2% HNO₃ (v:v). The digested solution was detected by ICP-MS using He collision cell mode as shown in Supplementary Table S4.

Celery, spinach, and carrot samples were purchased from local supermarkets (Hangzhou, China), and then washed and aired at ambient temperature; subsequently, cut into pieces and dehydrated for 48 h at ~50°C in an oven (101-3B, Shanghai Lianjing Electronic Technology Co., Ltd., Shanghai, China). The dried samples were ground into powder by a grinder (BJ-800A, Deqing Baijie Electric Co., Ltd., Huzhou, China) and then sieved through 60 mesh. According to the previous studies (28, 29), 0.05 g celery, spinach, carrot or tea sample was placed in a crushing tube with 8 zirconium balls (ɸ2.0 mm) and mixed with 1 mL of 0.5% (v:v) Triton X-100 (PerkinElmer, Waltham, MA, United States) solution; for grain sample, ɸ1.5 mm zirconium balls were employed. After sealing, the crushing tube was put into a grinder (DS-100, Hubei Xinzongke Virus Engineering Technology Co., Ltd., Wuhan, China) and shaken for 3 times with 15 s each time at 5500 rpm. After grinding, the sample will be shaken again for 1 min using a vortex mixer, and measured by the proposed method as soon as possible. The prepared slurry does not need to be separated from the zirconium beads and can be sampled directly using a pipette gun. The slurry sample preparation time was controlled within 5 min, and grinder can process dozens of samples at one time in 1 min, enabling high throughput sample treatment.

Statistical analysis of experimental data was performed using Origin 2021 and Microsoft^R Excel 2019. Significant differences were assessed by t-test, in which p-values lower than 0.05 (p < 0.05) support statistical significance, t-test is two-sided. The “green” performance of the method was evaluated using the software “AGREE.”

The graphite furnace was first used as an ETV by L’vov in the 1950s in coupling with AAS. Nowadays, graphite furnaces are also the most common devices for ETV-ICP-MS. However, the commercialized graphite furnace for AAS is not professionally designed for ICP-MS, so the hyphenated gas line system and analyte transportation must be modified and improved for ICP-MS.

The composition of food matrix is complicated including water, carbonhydrate, protein, lipid and so on, which are always a puzzle for the direct slurry sampling elemental analysis. In theory, these potential interferents mentioned above should be removed as possible prior to vaporization, as well as avoiding the loss of the targeted analytes of Se, Cd, As and Pb. As usual, the dehydration and ashing via heating must be indispensable to remove moisture and organic substances to separate with targeted elements. To investigate the heating conditions, a soybean CRM (GBW10013) sample was measured under different temperatures of ETV. As shown in Figure 4, from 500°C, As and Se start to be vaporized and finish before 900°C; Cd is completely vaporized from 700 to 1,000°C; and Pb ranges from 800 to 1700°C, which are almost consistent with the previous studies (33, 34). To avoid the loss of As and Se, the highest temperature of dehydration and ashing must be controlled below 500°C. Herein, 450°C was proved effective for ashing food samples including grain, oilseed, tea, and celery, spinach, carrot. At the same time, to avoid the boiling-over loss of slurry sample caused by overheating, the dehydration and ashing should be performed with a mild heating process. As a compromise, the dehydration process was completed from room temperature to 200°C within 20 s and held for 15 s, and then continued to heat to 450°C within 20 s and then held for 20 s for sample ashing. In addition, the carrier and gas turbulatores were set as 100 mL min⁻¹ and 300 mL min⁻¹ , respectively, to blow the moisture, gaseous interferents and microparticles out of the ETV and prevent them from entering the SDD and ICP-MS.

After ashing, the targeted elements remained in sample residue and got ready for the following vaporization under high temperature. From Figure 4, 1800°C heating fulfilled the complete vaporization of Se, Cd, As and Pb. Thus, to ensure the complete vaporization of all targeted elements in various food matrices, the highest vaporization temperature was set as 2,300°C and the heating process was finished within 1 s under the optimized power supply condition. The following results also proved 2,300°C was competent for the complete vaporization of Se, Cd, As and Pb from various plant food samples.

Also, the gas line system plays a key role in the transportation of analyte aerosol to the following SDD and ICP-MS, including carrier gas and gas turbulator herein. To investigate the effect of carrier gas and gas turbulator, a rice CRM sample was measured under different gas flow rates, and the results are shown in Supplementary Figure S4. The signal intensities of Se, Cd, As and Pb reached the highest values under 500 mL min⁻¹ carrier gas and 600 mL min⁻¹ gas turbulator with favorable RSDs in the range of 1.1–2.6%. As well, the total flow rate of carrier and gas turbulatores was 1,100 mL min⁻¹ and is able to replace the nebulizing gas for ICP-MS pneumatic nebulization. However, excessive gas flow could dilute the analyte aerosol to thereby reduce the signal intensities. Therefore, the argon carrier and gas turbulatores were chosen as 500 mL min⁻¹ and 600 mL min⁻¹ , respectively.

To further investigate the possible interference of inorganic elements, 10 μL of solutions containing Na, Mg, K, Fe, Ca, Co, Mo, etc. were separately mixed with 10 μL of a mixture of Se, Cd, As and Pb (20 ng mL⁻¹ ) for measurement. Since these elements that may cause interference due to their high content in the matrix of food samples. As shown in Table 1, the average recoveries of Se, Cd, As and Pb ranged from 91.5 to 112.5%, indicating free of significant interference at normal elemental levels in food matrices.

However, as shown in Supplementary Figure S5, the slope of the calibration curve using standard additions is significantly higher than the slope of the standard solution calibration. So, when introducing real food samples such as rice, tea and spinach, there is still some matrix interference, which are possibly caused by the “carry on” effect of microparticles accompanied with targeted elements (35, 36). These matrix interferents are detrimental to the analytical accuracy of the proposed ETV-ICP-MS method. To solve this problem, the calibration strategy of standard addition method was used to calibrate Se, Cd, As, and Pb in food samples (37). Under the above conditions, the linearity R² of each element was at least 0.999, proving that standard addition method is competent here considering the results of linearity.

Under optimized conditions, the limits of detection (LOD) were calculated for 11 assays of Se, Cd, As and Pb measurements in 10 μL Triton X-100 (0.5%, v:v) using an equation 3 σ/m (σ is the standard deviation and m is the slope of the Se, Cd, As and Pb calibration curves); and the limits of quantification (LOQ) using an equation 10 σ/m (38). The method LODs were 0.5 ng g⁻¹, 0.3 ng g⁻¹, 0.3 ng g⁻¹ and 0.6 ng g⁻¹ for Se, Cd, As and Pb, respectively; the corresponding LOQs were 1.7 ng g⁻¹, 1.0 ng g⁻¹, 1.0 ng g⁻¹ and 1.9 ng g⁻¹, respectively; and the linearity R² > 0.999 was reached in the range of 1 to 4,000 ng g⁻¹ using the standard addition method.

To verify the accuracy of the method, real food samples including CRM like rice, tea and soybeans (n = 6) were measured using this proposed method and microwave digestion ICP-MS method or directly compare the certified values. As shown in the Table 2, the results were in agreement with the certified value of CRMs and that of microwave digestion ICP-MS; the recoveries (the recovery was calculated as “the value measured by this method/the value measured by the standard method or the certified value “× 100%) of multi-elements in food samples were 86–118%, and the RSDs (n = 6) of repeated measurements were in the range of 1.2–8.9%. These results demonstrate a favorable analytical precision and accuracy of Se, Cd, As and Pb in typical food sample matrices mentioned above; as well, the analysis can be completed within 3 min only using a simple sample preparation, providing a fast, green and robust instrumental method. To compare the analytical performances, the related ETV-ICP-MS methods for food analysis previously reported (17, 32, 34, 38, 40–43)are summarized in Supplementary Table S6. The LODs of Se, Cd, As and Pb in this work was lower than or comparable to that of other ETV-ICP-MS methods previously reported; as well, the analytical precision is better. Furthermore, this proposed method is applicable to more food samples including carbonhydrate, protein, lipid, fibre-rich matrices. So, the novel disgn of SDD with gas turbulator is more beneficial for ETV-ICP-MS application.

To compare this proposed method with the standard method, their analysis conditions and cost were listed as shown in Supplementary Table S7. This proposed method is free of strong acid, and hazardous chemicals and only consumes 8 min for one aliquot including sample preparation vs. > 8 h using the standard method of wet digestion coupled with ICP-MS. At the same time, the “green” performance of the method was evaluated using the new metric software “AGREE” (39). Twelve parameters such as sample processing, sample size, and energy consumption were evaluated and a score was obtained. From Supplementary Table S7, the score of the standard method is 0.47 and the score of this method is 0.68. Therefore, this method is relatively greener compared with the standard method. By the way, some problems in this work should be solved in the future, eg. using the solution standard calibration to replace the standard addition method, and detecting more elements in more food sample matrices.