NIES CRM No. 28 Urban Aerosols, collected from the filters of a central ventilating system of a building located in Beijing city center, was produced to evaluate the analytical accuracy of determining the mass fraction of selected elements (18 certified and 14 reference values) (Mori et al., 2008). The certified value of Sr is 469 ± 16 mg/kg (certificate is available on https://www.nies.go.jp/labo/crm-e/hrfba300000ble6p-att/No.28_E.pdf). The Hg isotopic composition was also determined as an information value for the CRM (Yamakawa et al., 2020).

Sr isotopic measurements of the secondary reference, JA-2 and JB-1b, were performed using the same methods to manage the analytical accuracy of our method. These geological samples were produced at the Geological Survey of Japan in the National Institute of Advanced Industrial Science and Technology. The ⁸⁷Sr/⁸⁶Sr ratios of JA-2 and JB-1b are reported in Miyazaki and Shuto (1998) and Yuhara et al. (2000), respectively.

NIST SRM 987 was prepared at each laboratory by the following methods. At NIES and Kumamoto University, strontium carbonate powder was dissolved in dilute HNO₃ to make a stock solution. The solution was then adjusted to 300 ng g⁻¹ in 2% HNO₃ for the isotope measurement. At Okayama University, a ∼1 μg ml⁻¹ stock solution of NIST SRM 987 was prepared by dissolving strontium carbonate powder in dilute HCl.

Three bottles (bottle No. 044, 375, and 597) were randomly selected to assess the between- and within-bottle homogeneity of Sr isotope. The three subsamples were taken from each bottle and decomposed on a hotplate with a concentrated acid mixture of HNO₃/HClO₄/HF. Approximately 100 mg of powdered samples of the CRM, JA-2 and JB-1b were weighed and decomposed overnight in 5 ml of HNO₃ at 140°C. Then, 2 ml of 7 M HClO₄ was added and heated overnight at 200°C and dried to 1 ml. A mixture of 2 ml of 13 M HNO₃, 1 ml of 7 M HClO₄, and 1 ml 30 M HF was added and heated for 2 h. The decomposed samples were heated to dryness at 200°C by step heating. The resulting sample cake was redissolved in 1 ml of 3 M HNO₃, and the insoluble fraction was removed by centrifuging. This procedure was repeated several times, and the supernatant was subjected to the following chemical separation.

At Okayama University, the Sr isotope ratios were investigated using different acid decomposition and Sr separation methods. To obtain a representative sample, approximately 200 mg of powdered sample (bottle No. 35) was digested. The sample was dissolved in an HF/HNO₃ mixture using Teflon capsules sealed in stainless steel bombs for 72 h at 190°C. Once complete digestion was achieved, the sample was transferred to a 15 ml Teflon vial and 0.2 ml of HClO₄ was added before drying down at 120°C–200°C. The evaporated sample was further treated in 0.2 ml of HClO₄ to avoid fluoride precipitation. The sample was subsequently dissolved in HCl and ∼25% split was taken for Sr isotopic analysis.

Sr was separated from the other elements, particularly Rb, in the digest using Sr spec resin (Eichrom Technologies, US) at NIES. One milliliter of Sr spec resin was packed in a size S polypropylene column (Muromachi Chemical Inc., Japan). The resin was cleaned by passing 3 ml of 3 M HCl, 18 ml of 0.05 M HNO₃, and 3 ml of 3 M HNO₃. The sample dissolved in 1 ml of 3 M HNO₃ was loaded onto the column after conditioning the resin using 1 ml of 3 M HNO₃. The fraction containing Rb (and Ca, K, Mg, Ba, etc.), which began eluting immediately, was discarded by passing an additional 3 ml of 3 M HNO_3, 4 ml of 6 M HNO₃, and 1 ml of 3 M HNO₃. The Sr fraction was obtained by passing 5 ml of 0.05 M HNO₃ through a filter. During sample decomposition and chemical separation, the recovery yield was >95%. The total procedural blank for Sr was <1 ng, which was negligible compared to the sample size used in this study. For the isotopic measurement, the final solutions were diluted to a Sr concentration of 300 μg g⁻¹ using 2% HNO₃.

At Okayama University, Sr was separated by passing it through cation exchange resin in 2 M HCl (1 ml of AG50 × 12, 200–400 mesh, packed in a size S polypropylene column (Muromachi Chemical Inc.)). To achieve complete separation of Rb, the Sr separation was repeated twice. The total procedural blank was ∼30 pg, which was insignificant relative to the amount of Sr extracted.

The MC-ICP-MS used in this study was a Nu Plasma II (Nu Instruments, UK) at NIES. Although Sr was isolated from the matrix components and Rb by column chemistry using Sr spec resin, the signal for ⁸⁵Rb was simultaneously measured for the isobaric correction. As krypton (Kr) is present as a contaminant in the Ar plasma gas, the contribution of Kr had to be corrected. The mass numbers of 88 (Sr), 87 (Sr, Rb), 86 (Sr, Kr), 85 (Rb), 84 (Sr, Kr), 83 (Kr), and 82 (Kr) were detected using individual Faraday cups. The preamplifier gains associated with each Faraday cup were calibrated daily. The operating conditions (e.g., the torch position, Ar gas flow rates, and lens settings) were adjusted to maximize the signal intensity of ⁸⁸Sr (sensitivity of ⁸⁸Sr in 300 ng g⁻¹ was typically ∼10 x 10⁻¹¹ A). Details of the operation are summarized in Table 1. ⁸³Kr and ⁸⁵Rb were monitored for the isobaric correction of ⁸⁶Kr contribution to ⁸⁶Sr (⁸⁶Kr/⁸³Kr = 1.503), ⁸⁴Kr to ⁸⁴Sr (⁸⁴Kr/⁸³Kr = 4.955), and ⁸⁷Rb to ⁸⁷Sr (⁸⁷Rb/⁸⁵Rb = 0.3856). These isobaric interferences were corrected by blank subtraction. The ⁸⁷Sr/⁸⁶Sr ratios were corrected for mass fractionation using the exponential law relative to ⁸⁶Sr/⁸⁸Sr = 0.1194 (Steiger and Jäger, 1977). Because of the instability of the Ar gas flow, cone and slit degradation, and/or cup aging, Sr isotopic ratios may drift during a daylong analysis. To overcome these problems, a sample-standard bracketing technique was used. The ⁸⁷Sr/⁸⁶Sr values of the samples were adjusted using the NIST SRM 987 value of 0.710248 (McArthur et al., 2001).

The Sr isotopic measurements were performed using two TIMS instruments: the TRITON (Thermo Fisher Scientific, Germany) at Kumamoto University and the Finnigan MAT 262 (Thermo Fisher Scientific) at Okayama University. Approximately 200–500 ng Sr was loaded onto a degassed W single filament along with Ta-H₃PO₄ activator, and the ⁸⁷Sr/⁸⁶Sr ratio was measured at ⁸⁸Sr intensity 3–4 × 10⁻¹¹ A. The ⁸⁷Sr/⁸⁶Sr ratios were normalized to ⁸⁶Sr/⁸⁸Sr = 0.1194 (Steiger and Jä;ger, 1977) using an exponential law. To make an interlaboratory comparison of the data possible, NIST SRM 987 was analyzed at both laboratories, and the final ⁸⁷Sr/⁸⁶Sr ratios of the samples were adjusted using the recommended NIST SRM 987 value of 0.710248 (McArthur et al., 2001).

The reproducibility of the Sr isotopic compositions of the NIST SRM 987 was monitored during the study period to validate the analytical stability of our operating conditions. The ⁸⁷Sr/⁸⁶Sr ratios of 0.710243 ± 0.000016 (2SD, n = 74), 0.710254 ± 0.000012 (2SD, n = 15), and 0.710234 ± 0.000022 (2SD, n = 16) were obtained during the period of analysis at NIES, Kumamoto University, and Okayama University, respectively. To manage the analytical accuracy of our method, two geological standards, JA-2 and JB-1b, were analyzed at least twice on different days to monitor instrument stability at NIES. The values for the ⁸⁷Sr/⁸⁶Sr isotopic ratios of JA-2 and JB-1b were 0.706315 ± 0.000022 (2SD, n = 7) and 0.704093 ± 0.000023 (2SD, n = 8), respectively (Table 2). These values were identical within an acceptable error to their literature counterparts (JA-2: 0.706331 ± 0.000013 (2SD, n = 5) in Miyazaki and Shuto, 1998; JB-1b: 0.704095 ± 0.000012 (2SD, n = 13) in Yuhara et al., 2000). The ⁸⁷Sr/⁸⁶Sr isotopic values of JA-2 and JB-1b, same aliquots prepared at NIES, showed 0.706311 ± 0.000007 and 0.706325 ± 0.000007 for JA-2, and 0.704083 ± 0.000007 and 0.704084 ± 0.000006 for JB-1b at Kumamoto University (the analytical error is described in 2SE, and 150 ratios were taken in a single measurement) (Table 2). The Sr isotopic composition of JA-2 was also measured at Okayama University, showing the ⁸⁷Sr/⁸⁶Sr ratios of 0.706295 ± 0.000024 (2SD, n = 9) (Table 2). These results indicate that our technique was robust enough to measure ⁸⁷Sr/⁸⁶Sr ratios.

Table 3 shows the Sr isotopic ratio of NIES CRM No. 28. Within- and between-bottle were evaluated using MC-ICP-MS at NIES, yielding ⁸⁷Sr/⁸⁶Sr isotopic ratios of 0.710227 ± 0.000019 (2SD, n = 18). The uncertainty of the Sr isotopic values is an expanded uncertainty determined using a coverage factor k = 2, which corresponded to a confidence interval of ∼95%. To investigate the homogeneity of the isotopic results in CRM (Table 4), the ⁸⁷Sr/⁸⁶Sr isotopic ratios were tested using a one-way analysis of variance (ANOVA). The between-bottle variation was not statistically significant (p > 0.05 and F _{calculated value} < F _{critical value} ) as evaluated by one-way ANOVA. Therefore, when applied to the Sr isotopic ratios presented in this study, the CRM was homogeneous.

For the MC-ICP-MS vs. TIMS comparison, the Sr isotopic ratios of the three CRM samples (bottle No. 044, 375, and 597), which had been digested and processed for Sr separation at NIES, were measured by TIMS at the Kumamoto University to confirm the consistency of the ⁸⁷Sr/⁸⁶Sr isotopic ratios due to different analytical instruments: NIES was 0.710227 ± 0.000019 (2SD, n = 18) and Kumamoto University was 0.710229 ± 0.000011 (2SD, n = 12) (Figure 1; Table 3).

For sample decomposition and Sr separation, different methods were performed to evaluate bias during sample digestion and Sr separation at NIES and Okayama University. During the decomposition of the HNO₃/HClO₄/HF mixture using a hotplate at NIES, undissolved residues were observed, whereas a high-pressure bomb with HNO₃/HF/HClO₄ at Okayama University achieved complete digestion. The ⁸⁷Sr/⁸⁶Sr isotopic ratio obtained at Okayama University was 0.710226 ± 0.000019 (2SD, n = 12) (Figure 1; Table 3). Despite the incomplete dissolution at NIES, the ⁸⁷Sr/⁸⁶Sr isotopic ratio was consistent with the latter. Thus, the NIES CRM No. 28 Urban Aerosols was homogeneous enough for the Sr isotopic measurement of 100–200 mg subsamples using the methods described in Sample Preparation . This CRM will be of great value for the analytical quality assurance of environmental monitoring studies of PM.