water solution

water solution

Journal of Loss Prevention in the Process Industries 16 (2003) 41–53 www.elsevier.com/locate/jlp Study on decomposition of hydroxylamine/water soluti...

192KB Sizes 0 Downloads 166 Views

Journal of Loss Prevention in the Process Industries 16 (2003) 41–53 www.elsevier.com/locate/jlp

Study on decomposition of hydroxylamine/water solution Yusaku Iwata a,∗, Hiroshi Koseki a, Fumio Hosoya b a

National Research Institute of Fire and Disaster, 14-1, Nakahara 3-Chome, Mitaka, Tokyo 181-8633, Japan b Hosoya Kako Co. Ltd., 1847, Sugao, Akiruno, Tokyo 197-0801, Japan

Abstract The risk evaluation of decomposition of hydroxylamine(HA)/water solution was studied experimentally. The thermal property of HA/water solution was studied from the calorimetric data obtained using the differential thermal analysis (DTA). The intensity of decomposition was studied on the basis of the results of the mini closed pressure vessel test (MCPVT) and the pressure vessel test (PVT) in addition to the steel tube test. The thermal property of HA/water solution was evaluated on the basis of results of the DTA. The heat-release onset temperatures using the no-treated stainless steel cells were more than 70 K below those measured using the GSC. This result implies that the heat-release onset temperature depends on the materials of sample cell. On the other hand, the heat of reaction did not depend on the materials of sample cell. The intensity of the thermal decomposition was investigated on the basis of results of the MCPVT, the PVT and the steel tube test. The intensity of the thermal decomposition increased as the HA concentration increased in the MCPVT. The intensity of the thermal decomposition increased greatly when the HA concentration was beyond 80wt.% in the MCPVT. It was elucidated that the thermal decomposition of HA 70wt.%/water solutions was very violent in the PVT. In addition, HA/water solutions of more than 80wt.% concentration could detonate in the steel tube test. HA 80 wt.% water/solution was easily detonated by a detonator without RDX in the steel tube test. In addition, the decomposition hazard of HA/water solution by the metal ion and the iron powder was studied in this paper. The thermal stability of HA85%/water solution with the iron ion or the iron powder was discussed on the basis of the heat-release onset temperature by the DTA. The heat-release onset temperatures decreased when the concentration of the iron ion or the iron powder increased in the DTA measurements. The reactiveness of HA/water solution with the metal ion of iron, manganese, nickel, chromium and copper was examined by measuring the mass loss of HA/water solution after the metal ion was added to HA/water solution at room temperature. The reactiveness of HA/water solution with the iron powder was also studied in this paper. The ferrous ion, the ferric ion and the iron powder reacted with HA/water solution. Ignition automatically began when the 0.2wt.% ferric ion solution was added to HA85wt.%/water solution. The mass loss rate depended on the HA concentration greater than the iron ion concentration. The mass loss rate increased when an amount of the iron powder increased. On the other hand, the decomposition reaction of HA85wt.%/water solution with Cu2+ was calm compared to that of the iron ion. HA/water solution did not react with Mn2+, Ni2+ and Cr3+ at room temperature.  2003 Elsevier Science Ltd. All rights reserved. Keywords: Hydroxylamine; Risk evaluation; Thermal decomposition; Differential thermal analysis; Mini closed pressure vessel test; Pressure vessel test; Steel tube test; Iron ion

1. Introduction Recently two tragic accidents occurred by hydroxylamine (HA)/water solution, one was in the United States of America in February 1999 and the other

Corresponding author. Tel.: +81-422-44-8331; fax: +81-422-427719. E-mail address: [email protected] (Y. Iwata). ∗

was in Japan in June 2000. The demand of HA is expanding because it is required as a flaking off agent, a metallic surface treatment agent and an agent for organic synthesis. A chemical formula of HA is NH2OH. HA100wt.% is a colorless crystal at room temperature. The density of HA100wt.% (solid) and HA50wt.%/water solution are 1.20 and 1.12 g/cm3, respectively. It is well known that HA decomposes by heat and a metal ion (Sax & Lewis, 1989). The hazardous properties of HA/water solution are associated with violent

0950-4230/03/$ - see front matter  2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0950-4230(02)00072-4

42

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

decomposition. There are some papers regarding the thermal decomposition of HA/water solution. The results of the thermal analysis regarding HA/water solution were reported (Hazardous Materials Safety Techniques Association, 2001; Uchida & Wakakura, 2000). The heat-release onset temperature, the heat-release rate and the activation energy of HA50wt.%/water solution were measured in the adiabatic condition (Cisneros, Rogers, & Mannan, 2001). However, there is little information about the thermal property and the intensity of the decomposition of HA/water solution. It is important to obtain more knowledge of the decomposition of HA/water solution for the purpose of safe handling, use and storage of HA/water solution. This paper presents the decomposition hazard of HA/water solution with three aspects. First one was the thermal property of HA/water solution with various concentrations, which was evaluated from the calorimetric data obtained using the differential thermal analysis (DTA). The thermal stability regarding the thermal property was studied on the basis of the heat-release onset temperature and the heat of reaction measured by the DTA. The no-treated stainless steel cells and the GSCs were used to investigate the effects of the materials of sample cell. In addition, the accelerating rate calorimeter (ARC) was used to investigate influence of titanium sample cell on the thermal decomposition. Second, was the intensity of the thermal decomposition of HA/water solution with various concentrations, which was investigated from the results of the mini closed pressure vessel test (MCPVT) and the pressure vessel test (PVT). In relation to investigating the intensity of the thermal decomposition, the steel tube test was done to determine whether HA/water solution actually detonated. In addition, the shock sensibility of HA/water solution was studied in the steel tube test. Third was the decomposition hazard of HA/water solution with the metal ion or the iron powder. It is well known that decomposition of HA/water solution occurred by the metal ion in addition to heat. Iron, manganese, nickel and chromium are involved in the stainless steel. Copper widely exists in the natural world. Therefore these ions and the iron powder were used in the experiments. The thermal stability of HA85%/water solution with the iron ion or the iron powder was discussed on the basis of the heat-release onset temperature by the DTA experiment. The reactiveness of HA/water solution with the metal ion such as iron, manganese, nickel, chromium and copper was examined by measuring the mass loss after the metal ion was added to HA/water solution. The reactiveness of HA/water solution with the iron powder was also studied in this paper.

2. Summary of accident of HA in Japan An accident occurred on June 10, 2000 in a factory where HA/water solution was manufactured in Gunma prefecture, Japan (Hazardous Materials Safety Techniques Association, 2001). The residential houses around the factory were damaged by the blast of the explosion to the range of a 1500 m radius around this factory. The panes of a convenience store and a restaurant, which were about 100 m away from the factory, were broken by the blast of the explosion. In addition to the explosion, a fire occurred in this accident. Some of the facilities in the factory were burnt down by the fire due to HA/water solution and hydroxylamine salts, for example hydroxylamine chloride and hydroxylamine sulphate. A huge amount of smoke occurred in the fire and prevented fire fighting activity. The explosion killed four people, injured 58 people, and destroyed the facilities. The accident happened in the process of the re-distillation. A re-distillation tower was completely destroyed by the explosion. In this redistillation tower, the refined HA50wt.%/water solution was produced by the re-distillation of HA85wt.%/water solution. The refined HA50wt.%/water solution was a commercial product. Impurities such as a metal ion were hardly included in this refined HA50wt.%/water solution. HA85wt.%/water solution was not distributed in the market and existed only at this factory. A vessel containing HA85wt.%/water solution in the re-distillation tower was completely destroyed by the explosion. It was predicted that initial detonation occurred in HA85wt.%/water solution of 750–800 L, which existed in the vessel and the piping connected to the vessel.

3. Experiment The following seven kinds of experiments were conducted; 1. Thermal analysis of HA/water solution using the differential thermal analysis (DTA) 2. Accelerating rate calorimeter (ARC) 3. Mini closed pressure vessel test (MCPVT) 4. Pressure vessel test (PVT) 5. Steel tube test 6. Thermal analysis of HA/water solution added the metal ion or the iron powder using the DTA 7. Mass loss of HA/water solution added the metal ion or the iron powder 3.1. Samples HA90wt.%/water solution and HA50wt.%/water solution by weight were supplied by the HA manufacturer.

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

They contained a kind of stabilizer. HA/water solutions with other concentrations used in the experiments were made by diluting HA90wt.%/water solution. The HA concentrations of HA90wt.%/water solution and HA50wt.%/water solution were checked by oxidationreduction titration. Sources of ferric ion (Fe3+), ferrous ion (Fe2+), copper ion (Cu2+), manganese ion (Mn2+), nickel ion (Ni2+) and chromium ion (Cr3+) are ammonium ferric sulfate [Fe(III)(NH4)(SO4)2·12H2O], ammonium ferrous sulfate [Fe(II)(NH4)2(SO4)2·6H2O], copper sulfate pentahydrate [Cu(II)(SO4)·5H2O], ammonium manganese sulfate hexahydrate [(Mn(II)(NH4)2 (SO4)2·6H2O), ammonium nickel sulfate hexahydrate [(NH4)2Ni(II) (SO4)2·6H2O] and chromium potassium sulfate dodecahydrate [Cr(III)K(SO4)2·12H2O], respectively. These compounds including a metallic element were supplied by Wako Pure Chemical Industries, Ltd. Averaged diameter of the iron powder was approximately 250 µm. The iron powder was supplied by Kanto Kagaku. 3.2. DTA experiment The old stainless steel cell (OSC), the new stainless steel cell (NSC) and the gold-coated stainless steel cell (GSC) were used in the experiments. The stainless steel cells were made of SUS 304 of Japanese Industrial Standard (JIS G 4304). The stainless steel of SUS 304 contained manganese, nickel and chromium in addition to iron of the main ingredient. Three kinds of cells were dealt with tempering and cleaning. The OSCs were made more than five years ago. The NSCs were used in the DTA experiments within a month after they were made. All sample cells were sealed and were able to endure a pressure until 5 MPa. TAS-100, which was made by Rigaku Corporation, was used as the TG-DTA apparatus. The equipment constant for calculating heat of reaction was calibrated by In, Sn and Pb. HA/water solutions with various concentrations were used as samples in the DTA experiment. A sample was placed into a sample cell using disposable plastic pipettes. The sample was weighted in a sample cell. The sample weights were 2.2 mg (±0.3 mg). The number in parenthesis is a standard deviation of the sample mass weighted. The sample cells were covered with caps and sealed by a cell sealer in air. Aluminum oxide (Al2O3) was used as the reference material. Heating rate was 10 K/min in the DTA. The weight change of the samples was measured by thermogravimetry to check leakage from the sample cells during the measurements. The number of test was more than five for each condition. The DTA data were excluded when samples leaked during the measurements or the baselines of the DTA curves were not stabilized.

43

3.3. ARC experiment The ARC is an adiabatic calorimeter, which measures exothermic behavior in the adiabatic condition for the hazard evaluation of reactive chemicals. The activation energy of chemical substance can be obtained by measuring the relation of temperature vs the heat-release rate in the adiabatic condition. The ARC apparatus was made by Columbia Scientific Industries. Sample cell was made of titanium. The measurement started from room temperature. HA50wt.%/water solution was used as a sample in the ARC experiment. The amount of sample was 0.9 g. 3.4. MCPVT The main components of the MCPVT apparatus are an electrical furnace, a pressure vessel, a temperature control unit, a data recorder and analysis software. Details of the MCPVT apparatus are shown in the literature (Liu & Hasegawa, 2000). The volume of the pressure vessel was 6 cm3. The heating rate was 10 K/min in the MCPVT. HA/water solution with various concentrations were used as samples in the MCPVT. The weight of sample was 0.5 g. A sample vessel to contain the sample was made of glass. The sample vessel was placed inside the pressure vessel. The pressure rise inside the pressure vessel was measured by the pressure transducer and recorded in the data acquisition system. 3.5. PVT The PVT was conducted according to the test method prescribed by the Japanese Fire Service Law. The pressure vessel of the PVT was basically similar to those of USA and Dutch from the point of using an open pressure vessel with almost the same volume. The volume of the pressure vessel was approximately 200 cm3. The heating rate was 40 K/min in the PVT. HA50wt.%/water solution (HA50wt.%/water solution used in the PVT contained no stabilizers, which was supplied by Waco Pure Chemical Industries, Ltd.) and HA70wt.%/water solution were used as samples. The weight of sample was 5 g. The experiment condition is shown in Table 1. Two types of sample vessel were used to contain the sample. One was an aluminum sample vessel, which was prescribed in the Japanese Fire Service Law. The other was a glass sample vessel. The aluminum sample vessel was used in the PVT with HA50wt.%/water solution and HA70wt.%/water solution. The glass sample vessel was used in the PVT with HA70wt.%/water solution. A stainless plate of 2 mm in thickness with a 1.0 mm or 9.0 mm orifice in diameter was fitted with the pressure vessel. A rupture disc, which could endure 600 kPa, was

44

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

Table 1 Results of the pressure vessel tests No.

Sample cell

HA concentration (wt.%)

Orifice diameter (mm)

Number of trials

Number of burst discs

1

Aluminium

50

2

Aluminium

70

3

Glass

70

9.0 1.0 9.0 1.0 9.0 1.0

1 10 5 – 3 –

0 2 5 – 3 –

set on the pressure vessel. The measurements were finished when the rupture disc burst or all of the samples were evaporated and disappeared. If the rupture disc operates five times or more in ten trials, it is decided that the sample has the hazardous property regarding the intensity of the thermal decomposition in the Japanese Fire Service Law. 3.6. Steel tube test A glass tube to contain the sample was placed inside a steel tube. The dimensions of the glass tube were 24 mm in outside diameter, 150 mm in length and 1.3 mm in thickness. The dimensions of the steel tube were 25 mm in inside diameter, 200 mm in length and 4.5 mm in thickness. The volume of the sample was 40 cm3. The sample was poured into the glass tube. A cap made of silicone rubber was put on the glass tube. The glass tube was placed inside the steel tube. Trimethylenetrinitroamine (RDX) was wrapped with the electric detonator by a transparent thin film. RDX was used as a booster in the measurements except for No. 2 and No. 4 test (see Table 2). The electric detonator with RDX was attached Table 2 Results of the steel tube tests No.

HA concentration (wt.%)

Igniting method

1a 2a 3a 4a 5a 6a 7a 8b 9a

90 85 85 80 80 75 70 50 0

Detonator+RDX Detonator Detonator+RDX Detonator Detonator+RDX Detonator+RDX Detonator+RDX Detonator+RDX Detonator+RDX (blank test)

Result

10 g 2g 10 10 10 50 10

g g g g g

Detonation Detonation Detonation Detonation Detonation No Detonation No Detonation No Detonation No Detonation

a Stainless tube of 25 mm in diameter and 200 mm in length. Sample volume was 40 cm3. Each HA sample was put into a glass tube. RDX was used as a booster except for No. 2 test and No. 4 test. b Stainless tube of 50 mm in diameter and 500 mm in length. Sample volume was 980 cm3. HA sample was put into a tube made of polyethylene. Compressed RDX was used as a booster.

to the outside of the glass tube. Initiation was conducted using the electric detonator No.6. The experimental condition is described in detail elsewhere (Ogawa, Miyake, Hosoya, Hatano, & Takishita, 2001). In addition, larger size of the steel tube was used in the measurement (No. 8 in Table 2). The dimension of the steel tube was 50.7 mm in inside diameter, 500 mm in length and 4.9 mm in thickness. This steel tube test was conducted according to BAM 50/60 steel tube test, which was developed in the Germany. The volume of sample was 980 cm3. The sample was poured into a tube made of polyethylene (sample of No. 8 test was the same as HA50wt.%/water solution used in the PVT). The polyethylene tube was placed inside the steel tube. Compressed RDX, which contained 5wt.% wax, was used as a booster. Initiation was conducted using an electric detonator No.6. 3.7. Thermal stability of HA/water solution added iron ion solution or iron powder 3.7.1. Iron ion solution When HA85wt.%/water solution (9.8 g) and the iron ion (Fe3+) solution (0.2 g) of 500 ppm were mixed, HA83wt.%/water solution which contained the iron ion of 10 ppm was made. When HA85wt.%/water solution (9.7 g) and the iron ion solution (0.3 g) of 1000 ppm were mixed, HA82wt.%/water solution which contained the iron ion of 30 ppm was made. The resulting solution was stirred immediately when the iron ion solution was added to HA/water solution. HA85%wt./water solution added the iron ion solution was measured by the DTA. The GSCs were used. Heating rate was 10 K/min. 3.7.2. Iron powder HA 85wt.%/water solution was poured on the iron powder in the GSC using in the DTA. All of the iron powder were contacted with HA85wt.%/water solution. The amounts of the iron powder were 0.01 mg, 0.13 mg or 0.62 mg in HA85wt.%/water solution. The amount of HA85wt.%/water solution was approximately 2.5 mg. The mixture of HA85%wt./water solution and the iron powder was measured by the DTA. Heating rate was 10 K/min.

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

45

3.8. Reaction of HA/water solution with metal ion or iron powder

Compounds including the metal element were dissolved perfectly by stirs in water to make the metal ion solution except for ammonium ferrous sulfate. Ammonium ferrous sulfate was not dissolved perfectly. Ammonium ferrous sulfate which was insoluble, was uniformly diffused in water by stirs. The metal ion/water solution or the iron powder was added to HA/water solution at room temperature. 3.8.1. Mn2+, Ni2+, Cr3+ and Cu2+ HA50wt.%/water solution and HA85wt.%/water solution were used in the experiments. The concentration of the added metal ion was 0.5wt.% (5000 ppm). The metal ion solution was added to HA/water solution in a glass vessel without stirs. The dimensions of the glass vessel were 40 mm in outside diameter, 110 mm in length and 2 mm in thickness. The amount of the metal ion solution was 1 g. The amount of HA/water solution was 9 g. The concentration of the metal ion became 0.05wt.% (500 ppm) in the resulting solution. HA85wt.%/water solution and HA50wt.%/water solution were used in the experiments. The mass loss was measured after the addition of the metal ion solution.

3.8.2. Iron ion and iron powder The mass loss rate was measured regarding additions of the ferrous ion (Fe2+), the ferric ion (Fe3+) and the iron powder. The mass loss rate was obtained from the curve of the mass loss. The weight of solution was measured by an electrical balance. The limit of the electrical balance was 0.01 g in the mass loss measurements. The weight data were stored in a personal computer every one second. The iron ion or the iron powder was added to HA/water solution in the glass vessel. The dimensions of the glass vessel were the same as one used in the mass loss experiment of HA/water solution including the metal ion. When the iron ion solution was added to HA/water solution of more than 50wt.%, bubbles were generated very violently. The iron ion solution was diffused uniformly in HA/water solution. When the iron ion was added to the HA30wt.%/water solution, bubbles were not generated very violently. The resulting solution was stirred so that the iron ion was diffused in HA/water solution. On the other hand, the resulting mixture was stirred after the iron powder was added to HA/water solution. It took approximately one minute from the addition of the iron powder to the onset of the weight measurement because stirs were conducted.

Fig. 1. DTA curves of HA20wt.%/water solution, HA30wt.%/water solution, HA40wt.%/water solution, HA50wt.%/water solution, HA75wt.%/water solution and HA85wt.%/water solution measured using the new stainless steel cell.

4. Results and discussion 4.1. DTA 4.1.1. DTA curve The DTA measurements of the HA/water solution with various concentrations (HA 20wt.%–HA 85wt.%) were conducted using the NSC and the GSC. The DTA curves of HA/water solution with various HA concentrations measured using the NSC and the GSC are shown in Figs. 1 and 2. The horizontal axis of the DTA curve indicates the sample temperature. The vertical axis of the DTA curve indicates electromotive force by temperature

Fig. 2. DTA curves of HA20wt.%/water solution, HA30wt.%/water solution, HA50wt.%/water solution, HA75wt.%/water solution and HA85wt.%/water solution measured using the gold-coated stainless steel cell.

46

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

difference between the reference material and the sample. An intersection point of the baseline and the maximum slope of the peak in the DTA is commonly used as the heat-release onset temperature (Tonset). The temperature rising from the baseline (Ta) was also obtained and discussed in this paper. Both of Tonset and Ta were used as the onset temperature of the thermal decomposition. The low Tonset indicates the thermal instability as well as the low Ta. The thermal decomposition reaction began at Ta. The thermal decomposition reaction, which gave the peak with large heat of reaction in the DTA curve, began at Tonset. The DTA curves measured using the NSC had one peak in HA20wt.%/water solution and HA30wt.%/water solution (Fig. 1). They had two peaks in more than 40wt.% concentrations of HA. On the other hand, The DTA curves measured using the GSC had only one peak in all of the HA concentrations (Fig. 2). Tonset of the peak on the lower temperature side was obtained to study the thermal stability when two peaks were measured on the DTA curve. The DTA curves of HA50wt.%/water solution and HA85wt.%/water solution were also measured using the OSC. They had two peaks in both of the HA concentrations. The configuration of the DTA curves in both of the HA concentrations measured using the OSC was similar to that of the DTA curves in more than 40wt.% concentration measured using the NSC. However, the first peak measured using the OSC was on the lower temperature side than the first peak measured using the NSC. 4.1.2. Heat-release onset temperature (Tonset) and heat of reaction (HR) The thermal stability was studied on the basis of the heat-release onset temperature and the heat of reaction measured by the DTA. The low heat-release onset temperature indicates the thermal instability. The relationship between the HA concentration and Tonset is shown in Fig. 3. All data of Tonset using the GSC were almost constant for different HA concentrations (136±5 °C). When the NSC were used, Tonset decreased with an increasing of the HA concentration. When the HA concentration became more than 50wt.%, Tonset became almost constant (63±3 °C). The high heat of reaction indicates that there is the possibility that intense decomposition occurs. The relationship between the HA concentration and the heat of reaction (HR) is shown in Fig. 4. HR increased proportionally with the HA concentration for the NSC and the GSC. HR obtained by the NSC were almost the same as those of the GSC. This result implied that the chemical equation of the overall decomposition of HA might be the same in the NSC and the GSC. HR of HA 100 wt.% was 4.3–4.6 kJ/g assuming a pro-

Fig. 3. Relationship between HA concentration and the heat-release onset temperature (Tonset) and relationship HA concentration and the temperature rising from the baseline (Ta). NSC: new stainless steel cell, GSC: gold coated-stainless steel cell, OSC: old stainless steel cell.

Fig. 4. Relationship between HA concentration and the heat of reaction (HR). The vertical bars indicate the maximum value and the minimum value in the measurements of NSC. NSC: new stainless steel cell, GSC: gold coated-stainless steel cell, OSC: old stainless steel cell.

portional relationship between the HA concentration and HR in Fig. 4. The heat of decomposition in trinitrotoluene (TNT) is 5.1 kJ/g (Bretherick, 1985). Assuming the heat of reaction can be considered as an index of the intensity of the heat decomposition, there is a possibility that HA involves a great hazard similar to an explosive such as TNT. However, it is necessary to conduct the MCPVT and the PVT in order to examine the intensity of the thermal decomposition. 4.1.3. Influence of sample cell on Tonset, Ta and HR The process of the thermal decomposition reaction in the NSC was different from that of the GSC because the configuration of two DTA curves was different (see Figs.

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

1 and 2). Tonset of HA 50wt.% solution measured using the NSC and the GSC were 66 °C and 141 °C, respectively (see Fig. 3). Tonset measured using the NSC was lower than Tonset measured using the GSC except for Tonset of HA20wt.%/water solution. The materials of sample cell influenced the thermal decomposition of HA/water solution. The relationship between the HA concentration and Ta was shown in Fig. 3. Ta measured using the NSC decreased with the increase of the HA concentration. This tendency was similar to the results of Tonset. Ta measured using the GSC was almost the same and approximately 90 °C when the HA concentration was less than 50wt.%. Ta of HA75wt.%/water solution and HA85wt.%/water solution measured using the GSC was approximately 50 °C. Ta of HA75wt.%/water solution and HA85wt.%/water solution measured using the GSC was lower than Ta of HA/water solution below 50wt.%. The thermal stability of HA/water with the concentration of more than 75wt.% was reduced compared to that of HA/water with the low concentrations the from the results of Ta. This tendency was similar to the results obtained by the DSC (differential scanning calorimeter) with a glass capillary cell in the other report (The National Institute of Industrial Safety, 2001). Ta of HA85wt.%/water solution was lower than Ta of HA50wt.%/water solution even when using the GSC and the glass capillary cell. On the other hand, HR values measured using the NSC were almost the same as HR values measured using the GSC (see Fig. 4). The DTA curve measured using the NSC had two peaks. The heat of reaction of the peak on the low temperature side below 130 °C measured using the NSC is shown in Fig. 4. The vertical bar in Fig. 4 indicates the maximum value and the minimum value for measurement values. The heat of reaction of the first peak on the low temperature side was almost one third of the total HR measured using the NSC. It was variable even when HA/water solutions of the same concentration were measured. This indicated that the configuration of the DTA curve was not fixed when the HA concentration was more than 40wt.%. The DTA curves measured using the GSC had one peak. It is known that HA 100wt.% (solid) starts decomposing violently at a temperature higher than approximately 130 °C (Uehara, 1994). It was considered that this peak measured with the GSC was assigned to the thermal decomposition of HA without the effect of metal such as iron. The DTA curves measured using the NSC had two peaks in more than 40wt.% concentrations of HA (see Fig. 1). The peak on the high temperature side over 130 °C might be assigned to the thermal decomposition without the effect of metal such as iron. The peak on the low temperature side below 130 °C might be

47

assigned to the thermal decomposition with the effect of metal such as iron. The ARC was used to measure the activation energy of HA/water solution in the thermal decomposition. Influence of titanium sample cell used in the ARC on the thermal decomposition of HA/water solution was investigated. The activation energy of HA50wt.% solution measured by the ARC using titanium sample cell was 2.2 kJ/g. The thermal decomposition started at room temperature in the ARC experiment. The activation energy of HA50wt.%/water solution measured by the Adiabatic Pressure Tracking Accelerating Calorimeter (APTAC) using the glass sample cell was 3.7–4.3 kJ/g (Cisneros, Rogers & Mannan, 2000, 2001). The heatrelease onset temperature was 133–136 °C in the APTAC experiment. The titanium sample cell reduced the activation energy by the catalytic effect of the surface and accelerated the thermal decomposition. Tonset measured using the OSC was lower than Tonset measured using the NSC (see Fig. 3). Tonset of HA50wt.%/water solution measured using the OSC was 44 °C. The followings might be predicted as one of the reasons for the Tonset difference between the NSC and the OSC. Originally, the stainless steel surface reacted with the HA. The surface of the stainless steel has been oxidized since the stainless steel cell was made. The surface of the stainless steel might have iron oxide and turn rough. As a result, the decomposition rate of HA might become larger because of the wider surface area, which was in contact with HA/water solution. In addition, the oxides of the sample cell surface might accelerate the decomposition of the HA. The surface situation of the sample cell might influence the thermal decomposition of HA/water solution in addition to the materials of sample cell. HR measured using the OSC was smaller than HR measured using the other cells (see Fig. 4). The generation of heat had already started from room temperature in the DTA curves of the OSC. It was difficult that the baselines of the DTA curves were stabilized when the OSC were used. There was a possibility that the decomposition of HA started before the measurements. Therefore the peak area measured using the OSC was estimated to be small in the DTA curve. 4.1.4. Decomposition reaction It is known that HA decomposes into ammonia and water and so on very fast with heat or the presence of a metal ion. Products of decomposed HA depend on the experimental conditions. The following chemical equation was made referring to chemical equations of the literature (Cisneros, Rogers & Mannan, 2000): 3NH2OH(cryst)→NH3(gas) ⫹ N2 ⫹ 3H2O(gas) HR ⫽ 4.3(kJ / g)

(1)

The HR value of the chemical Eq. (1) was close to

48

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

the HR value of HA 100wt.% estimated from the DTA experiment (see Fig. 4). The chemical reaction in the chemical Eq. (1) might occur in the NSC and the GSC. The chemical equation of the overall decomposition did not depend on the catalytic effect. It was considered that final products in the NSC were the same as those in the GSC when HA decomposed completely. However, the process of the thermal decomposition reaction depended on the materials of sample cell because the DTA curves of HA/water solution obtained using the NSC were different from those obtained using the GSC (see Figs. 1 and 2). 4.2. MCPVT The intensity of the thermal decomposition of HA can be estimated by the (dP/dt)max value in the MCPVT. The (dP/dt)max value is the maximum of the pressure rise rate. Relationship between the (dP/dt)max value and HA concentration is shown in Fig. 5. The (dP/dt)max value increased with the HA concentration increasing. The rupture disc operated in the measurements of HA90wt.%/water solution. The rupture disc of the MCPVT was fitted as a safety valve. The rupture disc fitted with the pressure vessel of the MCPVT could endure with the pressure of 35 MPa. Generally, the rupture disc did not operate in the measurements for organic peroxides when the sample mass of organic peroxides was approximately 1 g. The (dP/dt)max values of organic peroxides were generally less than 100 MPa/s when the sample of 1 g was used in the MCPVT (Liu & Hasegawa, 2001). The thermal decomposition of HA90wt.%/water solution was very violent compared with those of organic peroxides. The (dP/dt)max value of HA90wt.%/water solution was approximately 500 times than that of HA80wt.%/water solution.

4.3. PVT The PVT results are shown in Table 1. HA50wt.%/water solution exploded twice in ten trials when a 1 mm diameter orifice was used with the aluminum sample vessel. Therefore it was decided that the thermal decomposition of HA50wt.%/water solution did not have the hazardous property according to the criterion. When HA50%/water solution was applied to the PVT as a sample, HA50%/water solution was concentrated by heating and the HA concentration became higher because water evaporated easier than HA. In the measurements of HA70wt.%/water solution, the rapture disc operated five times in five trials when a 9 mm diameter orifice was used with the aluminum sample vessel. Therefore it was decided that the thermal decomposition of HA70wt.%/water solution had the hazardous property according to the criterion (see Table 1). The shape of the aluminum sample vessels changed and they were broken to several pieces by the thermal decomposition twice in the five times when the rapture discs operated. It was elucidated that the thermal decomposition of HA 70wt.%/water solution was very violent. When HA/water solution made contact with the metal surface, it was surmised that the intensity of decomposition might become larger because of the catalytic effect of metal. In the measurements of HA70wt.%/water solution, the glass sample vessel was used to examine the influence of the materials of sample cell. The rapture disc operated three times in three trials when 9 mm diameter orifice was used with the glass sample vessel (see Table 1). The difference in the materials of sample cell might not give results of the PVT. Because there was possibility that the thermal decomposition occurred more violently on the basis of the MCPVT results, no PVT measurements were conducted using HA80wt.%/water solutions. 4.4. Steel tube test

Fig. 5. Relationship between HA concentration and the intensity of the thermal decomposition measured using the MCPVT.

The hazard of HA/water solution was evaluated according to the appearance of the destruction of the steel tube. It was examined to see whether HA/water solution could detonate. In addition, the shock sensibility was evaluated with changing detonator in the steel tube tests. The experiment result is shown in Table 2. When the steel tube fragmented into six or more small pieces, it was decided that detonation occurred. It was investigated to see whether a large amount of HA50wt.%/water solution detonated (see Table 2 No.8). HA50wt.%/water solution did not detonate. It was clear that HA/water solutions with more than 80wt.% concentration could detonate. Especially, HA80wt.%/water solution was easily

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

Fig. 6. DTA curves of HA85wt.%/water solution added iron ion solution.

detonated by a detonator without RDX (see Table 2 No.4). It is necessary to examine whether HA80wt.%/water solution detonate by smaller energy or not. 4.5. Thermal stability of HA/water solution added iron ion or iron powder 4.5.1. Iron ion The DTA curves of HA/water solution added the iron ion solution are shown in Fig. 6. Tonset and HR of HA/water solution added the iron ion solution are shown in Table 3. Ta could not be obtained because heat release was measured at room temperature. The DTA curves of HA/water solution added the iron ion solution had large one peak. They had a small shoulder of heat release beyond the large peak. Tonset shifted to the low temperature side when the iron ion concentration increased. When the iron ion concentration was more than 40 ppm, great heat release began at room temperature in the sample vessel. The DTA measurement could not be conducted when the iron ion concentration was more than 40 ppm. Tonset of HA83wt.%/water solution which contained

49

the iron ion of 10 ppm was lower approximately 40 K than that of HA/water solution without the iron ion (see Table 3). Tonset of HA82wt.%/water solution which contained the iron ion of 30 ppm was lower approximately 60 K than that of HA/water solution without the iron ion (see Table 3). The thermal stability of HA/water solution of more than 80wt.%, which contained the iron ion of 10 ppm, was largely reduced. HR of HA/water solution contained the iron ion of 10 ppm was almost the same as HR of HA/water without the iron ion. HR of HA/water solution containing the iron ion of 30 ppm was smaller than HR of HA/water solution containing the iron ion of 10 ppm and without the iron ion (see Table 3). Heat release had already started from room temperature in the DTA curves of HA/water solution containing the iron ion of 30 ppm. The decomposition started before the measurements. Therefore HR of HA/water solution containing the iron ion of 30 ppm became smaller than the other HR values. 4.5.2. Iron powder The DTA curves of HA85wt.%/water solution added the iron powder are shown in Fig. 7. Tonset and HR of them are shown in Table 4. Ta could not be obtained because heat release was measured at room temperature. Tonset decreased when an amount of the iron powder increased. The DTA curves of HA/water solution added the iron powder had a small and broad peak beyond a sharp and large peak. HR values were the same though amounts of the iron powder were different, except for HR of HA85wt.%/water solution containing the iron powder of 0.62 mg. HR for the iron powder of 0.62 mg was smaller than the other HR values because the decomposition started before the measurement.

Table 3 Heat-release onset temperature (Tonset) and heat of reaction (HR) for HA 85wt.%/water solution added iron ion measured using the goldcoated stainless steel cell No

HA concentration (wt.%)

Iron ion (Fe3+) concentration (ppm)

Tonset (°C)

HR (kJ/g)

1 2 3

85 83 82

0 10 30

141 101 84

3.8 3.6 3.2

Fig. 7. DTA curves of HA85wt.%/water solution with various amounts of iron powder measured using the gold-coated stainless steel cell.

50

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

Table 4 Heat-release onset temperature (Tonset) and heat of reaction (HR) for HA/water solution with various amounts of iron powder measured using the gold-coated stainless steel cell No

HA concentration (wt.%)

Weight of iron powder (mg)

Tonset (°C)

HR (kJ/g)

1 2 3 4

85 85 85 85

0 0.01 0.13 0.62

141 102 70 45

3.8 3.8 3.8 3.2

4.6. Reaction of HA/water solution with metal ion or iron powder 4.6.1. Mn2+, Ni2+, Cr3+ and Cu2+ Manganese ion (Mn2+), nickel ion (Ni2+) and chromium ion (Cr3+) did not react with HA85wt.%/water and HA50wt.%/water solution at room temperature. There were no mass losses by decomposition in mixtures of these ions and HA/water solutions. Copper ion (Cu2+) slightly reacted with HA85wt.%/water solution and HA50wt.%/water solution. The mass loss of HA50wt.%/water solution was almost the same as that of HA85wt.%/water solution when the Cu2+ 0.5wt.%/water solution was added. The mass loss of HA50wt.%/water solution was 0.13 g per hour. 4.6.2. Iron ion and iron powder 4.6.2.1. Iron ion When the iron ion (Fe3+) solution was added to HA/water solution, bubbles were quickly generated and gas was emitted. The mass loss could be measured within two minutes even when solution was added to Fe3+0.05wt.%/water HA30%wt.%/water solution. The reaction ended at tens of seconds. Especially, ignition automatically began when the iron ion (Fe3+) of more than 0.2wt.% was added to HA85%/water solution. The HA concentration of the resulting solution was 0.02wt.%. The iron ion very violently reacted with HA/water solution, depending on the concentrations of HA and the iron ion. It was difficult to measure heat release using the thermal analysis equipment because the reaction was very intense and much gas was emitted. Therefore, in order to analyze the decomposition reaction, the mass loss rate was used. It was implied that the mass loss rate was corresponded to the reaction rate of the HA decomposition. The reactiveness of the HA decomposition was examined by the mass loss rate. The slope of the time-weight curve was used as the mass loss rate. The slope was obtained by drawing a tangential line to the time-weight curve. However, when Fe3+0.5wt.%/water solution or Fe3+0.2wt.%/water solution was added to HA85wt.%/water solution, the mass loss could not be measured accurately in the experiments. The one reason for this was that the iron

ion/HA/water solution bubbled so violently that the balance could not indicate the accurate value. The other reason was that drops of HA/water solution popped besides the vessel. Examples of the mass loss of the iron ion/HA/water solution are shown in Fig. 8. The mass loss rate(g/s) on the onset of the mass loss can be shown by the following equation: Mass loss rate ⫽ k × [HA]m × [Fe3+]n.

(2)

[HA] (wt.%) and [Fe3+] (wt.%) are the HA concentration and the ferric ion concentration, respectively. [HA] and [Fe3+] are concentrations in the resulting solution after the addition of the ferric ion solution. k, n and m are constants. The mass loss rate was the value obtained from the slope of the time-weight curve on the onset of the mass loss. When the HA concentration was constant, the iron ion concentrations were changed to obtain m. When the iron ion concentration was constant, the HA concentrations were changed to obtain n. Log k was calculated from the intercept of log (mass loss rate) line with the ordinate. The results are as follows: Mass loss rate ⫽ 10⫺6.2 × [HA]4.3 × [Fe3+]1.2. (27wt.%ⱕ[HA]ⱕ77wt.%,0.005wt.%ⱕ[Fe3+]ⱕ0.05wt.%)

(3) The relationship between the HA concentration and the mass loss rate of HA/water solution added Fe3+/water solution with various concentrations is shown in Fig. 9. The calculation values and the experiment values were almost corresponding (see Fig. 9). The decomposition reaction was the first reaction to the iron ion concentration and the fourth reaction to the HA concentration in the concentration range of [HA] and [Fe3+] in Eq. (3). The HA concentration had a larger

Fig. 8. Time history of the mass loss of iron ion/HA/water solution after iron ions of various concentrations were added to HA75wt.%/water solution.

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

Fig. 9. Relationship between HA concentration and the mass loss rate of HA/water solution added iron ion (Fe3+)/water solution with various concentrations.

influence on the mass loss rate than the iron ion concentration. This experimental result might indicate that the iron ion behaved like catalysis in the decomposition reaction of HA. The mass loss rate was measured when Fe2+/water solution was added to HA85wt.%/water solution and HA50wt.%/water solution. Ignition automatically began when Fe2+0.5wt.%/water solution was added to HA85%/water solution. The curve of the mass loss of Fe2+/HA/water solution was similar to that of Fe3+/HA/water solution. The mass loss rate of Fe2+/HA/water solution was almost the same as that of Fe3+/HA/water solution when Fe2+0.5wt.%/water solution was added to HA50wt.%/water solution. It was implied that the reactiveness with Fe2+/water solution and HA/water was similar to that of Fe3+/water solution. 4.6.2.2. Iron powder When the iron powder was added to HA85wt.%/water solution, the quick mass loss did not happen like the addition of the iron ion. Mist and gas was suddenly emitted after time elapsed. All of HA/water solutions disappeared at the end of the reaction. Even when the iron powder of 4 g was added to HA85wt.%/water solution, HA/water solution was not ignited. The drops of HA/water solution did not pop beside the vessel like the addition of the iron ion. Examples of the mass loss of the iron powder/HA/water solution are shown in Fig. 10. The time 0 in Fig. 10 is arbitrary. There were two stages in the mass loss process. The mass loss rate was slow in the first stage. The mass loss rate increased gradually as time elapsed. The mass loss rate of the first stage became large very quickly in the second stage. The period prior to the increase of the mass loss rate by more than three times was regarded as the first stage. The period posterior to the increase of the mass loss rate by more than three was regarded as the second stage.

51

Fig. 10. Time history of the mass loss of iron powder/HA/water solution after various amounts of iron powder were added to HA85wt.%/water solution. Time 0 is arbitary.

The results of the addition of the iron powder to HA85wt.%/water solution are shown in Table 5. The mass loss rate was very small in the initial period of the first stage. Therefore the (mass loss rate)init was calculated by dividing the mass loss of 0.1 g from the weight measurement by the corresponding period for an expedient in Table 5. The (mass loss rate)init increased when an added amount of the iron powder increased. In addition, when an amount of the iron powder increased, the period from the addition of the iron powder to the end of the first stage was shortened. The increase of the iron powder signified that the surface area of the iron powder increased. The mass loss of HA/water solution from the onset of the weight measurement to the end of the first stage was 16% to the initial weight of HA/water solution on an average regarding various amounts of the iron powder (see Table 5). It was implied that HA of nearly 16% decomposed by the iron powder in the first stage. HA85%wt.%/water solution was not ignited when the iron powder was added. One reason was that HA was consumed in the first stage before the second stage with the large mass loss rate began.

5. Conclusions The hazard evaluation of decomposition of HA/water solution was conducted on the basis of the results of the experiments with the DTA, the MCPVT, the PVT, and the steel tube test. In addition, the decomposition hazard of HA/water solution by the metal ion and the iron powder was studied experimentally. The thermal stability of HA/water solution with the iron ion or the iron powder was discussed on the basis of the results of the DTA. The reactiveness of HA/water solution with the metal

52

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

Table 5 Results of the addition of iron powder to HA85wt.%/water solution No

W(HA)a (g)

Weight of iron powder (g)

(Mass loss rate)initb×104 (g/s)

⌬tc

⌬W(HA)d/W(HA)×100 (%)

1 2 3 4 5

9.5 9.6 10.4 9.8 8.9

0.020 0.20 1.0 2.0 4.0

0.057 0.50 1.9 4.7 8.1

15.5 hr 2 hr 10 min 5 min 3 min

23 14 12 14 21

a

W(HA): Weight of HA85wt.%/water solution (g). (Mass loss rate)init: Mass loss rate calculated by dividing the mass loss of 0.1g from the onset of the weight measurement by the corresponding period. c ⌬t: Period from the addition of the iron powder to the time when the mass loss rate increased more than three times. d ⌬W(HA): Mass loss of HA85wt.%/water solution at the time when the mass loss rate increased more than three times (g). b

ion or the iron powder was investigated by measuring the mass loss of HA/water solution after the metal ion or the iron powder was added to HA/water solution. The following conclusions were made: 1. The thermal property of HA/water solution was studied by the DTA. The heat-release onset temperature depended on the materials of sample cell. The heatrelease onset temperatures measured using the NSC were approximately 70 K lower than those measured using the GSC. The heat of reaction did not depend on the materials of sample cell. 2. The intensity of the thermal decomposition was studied by the MCPVT, the PVT and the steel tube test. The intensity of the thermal decomposition of HA increased as the HA concentration increased in the MCPVT. The intensity of the thermal decomposition of HA increased greatly when the HA concentration was beyond 80wt.% in the MCPVT. It was elucidated that the thermal decomposition of HA70wt.%/water solution was very violent in the PVT. It was clear that HA/water solutions with more than 80wt.% concentration could detonate. HA 80wt.% solution was easily detonated by a detonator without RDX in the steel tube test. 3. The heat-release onset temperature of HA82wt.%/water solution containing the iron ion of 30 ppm was approximately 60 K lower than that of HA85wt.%/water solution without the iron ion in the DTA. The heat-release onset temperature of HA85wt.%/water solution containing the iron powder of 0.62 mg was approximately 90 K lower than that of HA85wt.%/water solution without the iron powder in the DTA. The heat-release onset temperatures decreased when the concentration of the iron ion or the iron powder increased. 4. The ferrous ion, the ferric ion and the iron powder reacted with HA/water solution at room temperature. Ignition automatically began when the ferric ion solution of 0.2wt.% was added to HA85wt.%/water sol-

ution. The mass loss rate depended on the HA concentration greater than the concentration of the iron ion. The mass loss rate increased when an amount of the iron powder increased. On the other hand, the decomposition reaction of HA85wt.%/water solution with Cu2+ was calm compared to that of the iron ion. HA/water solution did not react with other metal ions such as Mn2+, Ni2+ and Cr3+ at room temperature.

Acknowledgements The authors wish to express their thanks to Professor Masamitsu Tamura of the University of Tokyo for many helpful discussions and advice in this study. The authors wish to express their thanks to Professor Terushige Ogawa of Yokohama National University, Dr. Hidenori Matsui of the National Institute of Industrial Safety and Mr. Masahide Wakakura of the Kanagawa Industrial Technology Research Institute for their advice and cooperation in the steel tube test. The authors also wish to thank to Professor M. Sam Mannan of Texas A&M University for personal information regarding the thermal decomposition of hydroxylamine/water solution.

References Bretherick, L. (1985). Bretherick’s handbook of reactive chemical hazards (3rd ed.). London: Butterworth. Cisneros, L., Rogers, W.J., & Mannan, M.S. (2000). Thermal decomposition study of hydroxylamine. In: Proceeding of the Symposium at the Mary Kay O’Connor Process Safety Center, Texas A&M University, TX (p.140). College Station. Cisneros, L., Rogers, W. J., & Mannan, M. S. (2001). Adiabatic calorimetric decomposition studies of 50wt.% hydroxylamine/water. Journal of Hazardous Materials, 82, 13–24. Hazardous Materials Safety Techniques Association (2001). Report of investigation of explosive fire of hydroxylamine at a chemical plant in gunma prefecture (in Japanese). Liu, X., & Hasegawa, K. (2000). On the measurement of violence in the thermal decomposition of self-reactive substances by the mini-

Y. Iwata et al. / Journal of Loss Prevention in the Process Industries 16 (2003) 41–53

closed pressure vessel test. OECD-IGUS-EOS Ad hoc Meeting, Japan. Liu, X., & Hasegawa, K. (2001). Personal Communications (in Japanese). Ogawa, T., Miyake, A., Hosoya, F., Hatano, H., & Takishita, Y. (2001). An explosion at a hydroxylamine plant (in Japanese). Investigation of Disaster, 32, 233–242. Sax, N. I., & Lewis, R. J. (1989). Dangerous properties of industrial materials volume III (7th ed.). New York: Van Nostrand Reinhold.

53

The National Institute of Industrial Safety (2001). Guide for explosion hazards of hydroxylamines and their safe handling, NIIS-SG-NO.1 (in Japanese). Uchida, T., & Wakakura, M. (2000). Hazardous evaluation of hydroxylamine. In: Proceeding of 33rd Meeting of Japan Society for Safety Engineering (in Japanese) (p.141). Tokyo. Uehara, Y. (1994). Data handbook for safety of chemical substances (in Japanese). Tokyo: Ohmsha.