Results and Discussion

3.1. The influence of mixing procedure (Run1)

For each of NBS28 and SWy-1, three conditions were compared (Fig. 2): the samples enveloped with PTFE in silver capsules (Fig. 2(a), (b)) show a CO peak with half the peak height of the reference gas peak. The small peaks around 170–180 s between the reference gas and CO peaks are the peaks of atmospheric N2 remaining in the silver capsule. Comparing the CO gas generation from non-silicates (NBS18, IAEA-601, and -602) and silicates (NBS28 and SWy-1), the latter showed peaks with lower intensity and tailing on the right side, suggesting that thermal decomposition did not occur instantaneously. The tailed peaks took 40–50 min to return to background levels, resulting in a measurement time of approximately 60 min per sample. Next, whether using Ni/C instead of graphite would improve the reactivity was tested (Fig. 2(c), (d)). The shape of the CO peaks became sharper, and the peak height increased distinctly, especially in SWy-1. Furthermore, the homogenization of Ni/C and PTFE with the sample before weighing and encapsulation resulted in sharper and higher peaks (Fig. 2(e), (f)). The peak height was close to that of non-silicate minerals. This is considered to be due to the closer distance between the sample and the F and C sources, in addition to the homogenization process, which crushed the grains of NBS28 (several hundred micrometers in diameter) into finer grains. However, the tailings of the peaks were still not completely resolved. It took 40–50 min to return to the background level. It is possible that some oxygen was retained in the residue and then gradually released, even though the majority of the samples broke down. The CO peak area, which is directly proportional to the oxygen output, and the isotope ratio could not be calculated since the peaks’ long tailings made it impossible to establish their end positions.