A total of 41 sample types were used in this investigation; all reagents were sourced from Wako Pure Chemical Industries, Ltd. The samples were chosen from substances having the characteristic structures listed in Table 1. We focused on substances that may react with stainless steel. There may be other materials that react with stainless steel, in addition to those listed here. For convenience, these reagents are hereafter referred to by their abbreviationslisted in Table 1.

Two types of glass vessels were used in this study: a borosilicate glass (Pyrex) capillary vessel (GC) and a Pyrex ampule (GA). The glass capillary vessel (GC) had an outside diameter of 1.7 mm, an inside diameter of 1.4 mm, and a length of 11

mm. After filling this vessel with 0.5 - 1 mg of the sample, the upper part of the vessel (positioned about 11 mm from the bottom) was sealed using a burner, while the lower part was cooled by liquid nitrogen; the ratio of the vessel volume to the sample mass varied between 10 and 27 μl/mg. The capillary was then inserted into an aluminum holder and set horizontally in a differential scanning calorimeter (EXSTAR DSC7020, manufactured by SII Nano Technology Inc.), as shown in Figure 1. The sample was then heated to 500˚C at 5˚C/min.

The glass ampule vessels (GA) had an inside diameter of 3.4 mm, an outside diameter of 5 mm, and a length of 20 mm. These were filled with 10 - 20 mg of sample, and then, sealed with a burner while cooling the opposite end with liquid nitrogen, giving a vessel volume to sample mass ratio of between 12 and 20 μl/mg. The ampules were then tested using a high-pressure DSC device manufactured by Palmetrics Co., Ltd., which was slightly modified to estimate the thermal behavior using a glass ampule. Finally, the ampules were inserted into an aluminum holder (Figure 2) and set in the high-pressure DSC device. Each sample was heated to 400˚C at 2˚C/min.

Evaluations with the high-pressure stainless-steel vessel (SS) were performed to compare with the results of the glass vessels. The stainless-steel vessel was manufactured by SII Nano Technology Inc. and the sample amount was about 1 mg. The vessel volume to sample mass ratio was 15 μl/mg. The measurements were performed while heating to 500˚C at 5˚C/min.

All devices were calibrated using 10 mg of four to five types of a dedicated standard metal (biphenyl, In, Sn, Pb, Zn, and Al) under each condition. The device condition was also confirmed by measuring these standard substances in the measurement mode. According to the ASTM (E537-07) [15] [16] , Q_DSC data can be considered reliable if the standard deviation is within 3.5%, and the reproducible standard deviation for a given sample is within 4.7%. However, there is a problem if these values exceed 9.7%. For each condition, the Q_DSC obtained for the standard material was in the permissible range of the fluctuation.

As for nitro compounds, there is a tendency for the Q_DSC value obtained in a glass vessel to be smaller than that obtained in a stainless-steel vessel in p-ni- trotoluene and a mononitro compound. However, we cannot ignore the influence of a different factor on these substances. Figure 4 shows the effect of varying the amount of oNP relative to the volume of the glass capillary vessel (GC) on the Q_DSC values obtained. It is evident that varying the relative amount of oNP affects the peak shape, with a similar tendency being observed for the stainless-steel vessel. As the quantity of the sample increases, the peaks at around 290˚C and 380˚C become shoulders of the mainpeak and that at around 313˚C increases in intensity. oNP is volatile at room temperature, which is attributed to the sublimation characteristics of the sample. Changing the amount of solid material should affect the ratio between the gas-phase reaction and the condensation-layer reaction. In the case of oNP, it is also confirmed that the peak shape changes according to the physical state of the sample in the vessel (fluffy or compressed), despite the quantity of the sample being identical. In addition to

chemical reactions, other factors can also influence the Q_DSC value, making the evaluation of the calorific value difficult. Microphotographs of the vessel surface taken after calorific evaluation in a stainless-steel vessel do not indicate surface modification. Figure 5 shows a DSC curve for a mixture of an oNP and a stainless-steel foil (SUS304, 1 mm × 6 mm × 0.01 mm), where the amount of oNP is about 0.30 mg/10 μL. This is the smallest quantity of sample used to fill the vessel shown in Figure 4. Figure 6 shows a microscope photograph of a glass capillary vessel (GC) with added SUS304 foil before and after the measurement. The addition of the SUS foil results in a change in the peak shape; the peak at around 290˚C increases in intensity and that at around 380˚C decreases. It is also confirmed that the peak shape is affected by the sample quantity of oNP. It is not

possible to comment on the calorific value associated with this effect, because the influence on energy calibration by the addition of the SUS foil is unknown. It is thought that the change in the peak shape is caused by not only the volatile nature of the sample but also the reaction with the stainless steel. A similar tendency is also confirmed in the case of pNP. On the other hand, although pNT is also volatile at room temperature, the change in the thermal behavior is not confirmed by the addition of the stainless-steel foil. It is supposed that the decrease in Q_DSC is caused by the volatility of the sample only in the case of pNT.

In any case, mono-nitrophenol might react with stainless steel. If the acidity of the phenols is involved, the acidity is higher in trinitrophenol. However, there is no difference between the Q_DSC obtained with the glass vessels (GC/GA) and the stainless-steel vessel for TNP. Further research is necessary for elucidation of the reason.

When evaluating some substances having particular characteristic structures, such as functional groups containing a halogen or sulfur, the Q_DSC value obtained in a glass vessel is notably smaller than that obtained in a stainless-steel vessel. It is well known that stainless steel reacts with a halogen, and the results obtained in this study certainly support this fact. For example, the DSC curve of 3-chloroperbenzoic acid (mCPBA) shown in Figure 7 clearly shows that the peak becomes much sharper when a stainless-steel vessel is used. The peak temperature is also shifted into the low temperature range, and takes on a more complex shape at higher temperatures. Plainly, the use of a stainless-steel vessel to evaluate the thermal reactivity of halogen-containing substances is problematic, and thus, is best avoided.

On the other hand, Figure 8 shows the DSC curves of TSC as a notable example of sulfur-containing substances. Furthermore, microphotographs of the vessel taken during the experiment are shown in Figure 9 for the temperatures (A)

240, (B) 400, and (C) 650˚C in Figure 5. Figure 10 also shows microphotographs of the glass capillary vessel before and after the DSC measurement. In the glass capillary vessel (GC) experiment, there is a single peak in the DSC curve. After measurement, an orange liquid residue is found. In the stainless-steel vessel, a small amount of corrosion is confirmed at 400˚C, and at 650˚C, there is a notable increase in the surface roughness of the vessel. However, there is no observable change in the vessel surface at 240˚C. Furthermore, an orange liquid residue is confirmed, like the residue observed with a glass capillary vessel (GC). We expect that the same phenomenon occurs in the glass capillary vessel at 240˚C. Thirty-four fragments are detected at temperatures of 240˚C or greater in

the investigation of TG-MS measurement, and the rotten-egg odor of hydrogen sulfide is also evident. In addition, SOx gas is detected in another substance containing sulfur and oxygen in its structure (64 of the fragments). In this way, it is thought that these corrosive gases deteriorate stainless steel in the case of substances containing sulfur. On the other hand, as for substances having the thiosemicarbazide group in their structure, if the molecular weight increases, e.g., for 1-phenyl-3-thiosemicarbazide, the influence on the vessel material diminishes.

In the case of a substance containing a hydrazine bond, two types of substances react with the stainless steel: DFH and APH. Here, as for 4-phenylhydrazine, there are differences between the Q_DSC values obtained with the two types of glass vessels. When pNPH fills a glass capillary vessel, it unavoidably attaches to the wall surface of the container. The sample that is attached to the upper part of the glass vessel is dissolved partly by heating (Figure 11(c)) when the glass capillary vessel is sealed. A black substance adheres to the container in an extreme case, as shown in Figure 11(d), and this sample probably decomposes rather than melting. Adhesion to the wall surface is observed especially in this sample. The glass capillary vessel is observed with a microscope after filling with the sample, and this is not used for the measurement if a black substance is confirmed. Suitable examples are shown in Figure 11(a) and Figure 11(b).

The DSC curve for DFH is presented in Figure 12. The heat behaviors for the stainless-steel vessel and the glass capillary vessel (GC) begin to deviate in the low temperature range, unlike those for TSC. Figure 13 shows a microphotograph of the stainless-steel vessel at 260˚C, where corrosion is evident. The generation of HCN (27 of the fragments) gas is confirmed by TG-MS. On the other hand, the generation of gases that might cause corrosion is not confirmed for

1-acetyl-phenylhydrazine. The decomposition product, which reacts with stainless steel, may be formed, not in the vapor phase but in the condensation phase. The product in the condensation phase has not yet been investigated. In the future, we will investigate whether there is a particular structural group in substances that react with the stainless-steel vessel.

In peroxides and other materials, there is no significant difference between the Q_DSC obtained with a glass vessel and that with a stainless-steel vessel. However, a difference is confirmed in the peak shape, as for TBH. Figure 14 shows the

DSC curve obtained using the glass capillary vessel (GC) and the stainless-steel vessel (SS) with TBH. Here, the peak shapes obtained with the two types of glass vessels (GC and GA) are consistent despite the difference in the sample scale. Reproducibility is achieved for the results with the glass vessel. If the peak shapes are different, the calorific behavior for the different phenomena is evaluated, even if there is no difference in the calorific value. This phenomenon is confirmed only for TBH in this study. TBH is also volatile at room temperature, like oNP and pNT. It is also confirmed that varying the amount of TBH in the vessel affects the peak shape. However, in the case of TBH, the results are found to change in relation to the composition of the lid and body of the stainless-steel vessel. Figure 15 shows the DSC curves obtained with various lid compositions of the stainless-steel vessels. Here, the body composition of the vessel is the same. Figure 16 shows the compositions of the lids shown in Figure 15. It is confirmed that the lid compositions influence the peak shapes and Q_DSC values. On the other hand, the influence of the body composition on the DSC curve is less than that of the lids. This suggests that the decomposition behavior of TBH is influenced more by a gas-phase reaction, and the generated gas appears to react with stainless steel. In this sample, it is thought that the change in the peak shape is caused not only by the volatility of the sample but also by the reaction with the stainless steel. Further research is needed regarding this phenomenon.

Based on the above results, it is proposed that in all substances containing sulfur or halogen, if thermal evaluation is performed with a stainless-steel vessel,

then Q_DSC will be affected by the vessel material. For such substances, a stainless-steel vessel is inappropriate and an inert material should be used.

For other substances, we intend to clarify whether there is a particular structural group in the substance that reacts with stainless steel. We plan to evaluate the Q_DSC of more materials.

Where Q_DSC is affected by the stainless steel, the Q_DSC value obtained in a glass vessel is smaller than that obtained in a stainless-steel vessel in almost all cases. The thermal hazard is overestimated when a stainless-steel vessel is used; thus, it might be safer to use this in screening tests for explosive chemicals by the United Nations. However, the most suitable vessel materials should be chosen in order to evaluate the true reactivity of a substance. Care should be taken when evaluating, because on the contrary, the glass vessel might also react with the substance being tested.