Evaluation on the Thermal Stability and Hazards Behaviors of ADVN Using Green Thermal Analysis Approach

Transcription

1 Journal of Civil Engineering and Architecture 1 (216) doi: / / D DAVID PUBLISHING Evaluation on the Thermal Stability and Hazards Behaviors of ADVN Using Green Thermal Analysis Approach Chi-Min Shu 1, Yi-Hong Chang 1 and Chen-Wei Chiu 2 1. Department of Safety, Health, and Environmental Engineering, National Yunlin University of Science and Technology, Douliou, Yunlin 642, Taiwan, R.O.C. 2. Department of Fire Safety, National Taiwan Police College, Taipei 11696, Taiwan, R.O.C. Abstract: ADVN (2,2 -Azobis (2,4-dimethyl) valeronitrile), a free radical initiator, is widely applied for the polymerization reaction of polymers in the chemical industries. When ADVN releases free radical during the decomposition process, it can accompany abundant heat and huge pressure to increase the possibility of thermal runaway and hazard, causing unacceptable thermal explosion or fire accidents. To develop an inherently safer process for ADVN, the thermal stability parameters of ADVN were obtained to investigate thermal decomposition characteristics using a DSC (differential scanning calorimetry) and TG (thermogravimetry). We used various kinetic models to completely depict the kinetic behavior and determine the thermal safety parameters for ADVN. The green thermal analysis approach could be used to substitute for complicated procedures and large-scale experiments of traditional thermal analysis methods, avoiding environmental pollution and energy depletion. Key words: ADVN (2,2 -Azobis (2,4-dimethyl) valeronitrile), green thermal analysis approach, kinetic models, thermal runaway and hazard, thermal safety parameters. Nomenclature A Pre-exponential factor (s 1 ) A Ratio of mass loss from TG for different partitioned regions (dimensionless) dα/dt Conversion rate (s 1 ) E a Apparent activation energy (kj mol 1 ) f(α) Function of degree of conversion (dimensionless) IPDT Integral procedural decomposition temperature ( C) k Reaction rate constant (s 1 ) K Ratio of mass loss from TG (thermogravimetry) for different partitioned regions (dimensionless) r Correlation coefficient, 1~1 (dimensionless) R Gas constant (J K 1 mol 1 ) R 2 Coefficient of determination, ~1 (dimensionless) S1 Thermogravimetric area under TG curve (dimensionless) S2 Thermogravimetric area of non mass loss (dimensionless) S3 Thermogravimetric area above TG curve (dimensionless) T Absolute temperature (K) Corresponding author: Chi-Min Shu, Ph.D., professor, research fields: process safety, explosion prevention and quantitative risk assessment. T T i T f T max T p t α Apparent exothermic onset temperature ( C) Mass loss onset temperature ( C) Final temperature ( C) Maximum reaction temperature ( C) Peak temperature ( C) time (s) Degree of conversion (dimensionless) β heating rate ( C min 1 ) H d Heat of decomposition (J g 1 ) ( H d ) t Heat of decomposition with time (J g 1 ) H endo Heat of decomposition of endothermic reaction (J g 1 ) H exo Heat of decomposition for exothermic reaction (J g 1 ) 1. Introduction ADVN (2,2 -Azobis (2,4-dimethyl) valeronitrile), an azo compound, is generic initiator for the polymerization reaction of polyvinyl chloride, polyacrylonitrile, or poly (vinyl) alcohol. Through a special chemical structure, N-N bonds, ADVN can rapidly and abundantly provide nitrogen free radicals (N ) to interconnect two or several monomeric

2 281 compounds to yield polymers [1, 2]. However, when decomposing, ADVN has a potential thermal hazard because of low apparent onset temperature, high enthalpy, and the releasing of large amounts of flammable gases. These hazardous characteristics may incur violent runaway reaction and/or explosion to cause equipment damage, financial losses, or heavy casualties [3]. A serious accident and social outcry may unavoidably ensue. Therefore, the thermal hazards characteristics of ADVN should be evaluated to ensure that ADVN can be safely manufactured or used. In the past, many related researches have completely explored the thermal hazards of many substances. However, the thermal hazard characteristics and kinetic behaviors of ADVN were rarely mentioned in the open literatures. In this study, we developed a methodology to investigate these characteristics for ADVN through calorimetric technology to obtain a variety of thermal stability parameters, such as ΔH d, T, T f and T p. Furthermore, various kinetic models, ASTM-E698-5 and isoconversional method, can be implemented to acquire E a (apparent activation energy), and A (pre-exponential factor) and elucidate the kinetic behaviors for the inherently safer design of ADVN [4]. Calorimetric technology combined with thermokinetic models, an advanced green analytical method, could be used to provide related information of thermal hazards of ADVN and to substitute for the complicated procedures and large-scale experiments of traditional thermal analysis methods to avoid environmental pollution and energy depletion. 2. Experimental and Methods 2.1 Sample ADVN, a white solid powder, was purchased from ACE Chemical Corp. in Taiwan with 98% purity and packed in 1 mg commercial package. This product is thermally sensitive and instable, so that it has to be stored in refrigerator at 4 C. For clarity, the chemical structures of ADVN are presented in Fig. 1 [5]. 2.2 DSC (Differential Scanning Calorimetry) DSC, a common thermal analysis technique, often is used to understand various physical or chemical characteristics via observing exothermic or endothermic reaction of a chemical, such as thermal decomposition reaction, curing reaction, crystallization, or phase change. Thus, it can be a safety assessment methodology to provide the thermal hazard information for ADVN. The type of DSC is selected in the Mettler DSC 821 e. To investigate thermal characteristics of ADVN, ADVN was heated from 3 to 3 C at different heating rates.5, 1, 2, 4 and 8 C min 1 using the DSC test. The sample quantity for each experiment was approximately 1.5~5. mg and the sample was sealed in a gold crucible [6]. 2.3 Evaluation of E a by ASTM-E698-5 ASTM-E698-5, from ASTM (American Society for Testing and Materials), is a popular and reliable way to determine E a via calorimetric experiments, by the Arrhenius equation [7, 8]: d Ea k exp f ( ) (1) dt RT E a, k o, R and T are apparent activation energy, frequency factor, gas constant and absolute temperature. N H 3 C N C H 2 C N CH CH 3 H 2 C C N CH 3 Fig. 1 Chemical structure of ADVN [1, 2]. CH 3 CH CH 3 CH 3

3 282 f(α) is the kinetic model depended on different reaction mechanisms, i.e., nth order or autocatalysis reaction. In addition, α, the degree of conversion for decomposition of ADVN, can be given in Eq. (2): H d t (2) H where, (ΔH d ) t is heat of decomposition with time t, and ΔH d is total heat of decomposition. We substitute β as dt/dt, and Eq. (1) can be rearranged to Eq. (3): dt Ea A f exp (3) d RT Eq. (4) can be obtained by taking the natural logarithm of Eq. (3): dt Ea ln ln A f d RT (4) 2.3 Evaluation of E a for Different Conversions by Friedman Method In general, a chemical reaction is rather intriguing, and it can be changed with temperature or time under non-isothermal conditions. It means that E a is variable depending on the reaction progress (α). However, E a is usually denoted as a fixed value derived from the Arrhenius equation. To completely investigate the kinetic behavior of ADVN, the Friedman method can be adopted to reckon the value of E a at various conversions. The Friedman method can be derived as follows [9]: Eq. (1) can be arranged as Eq. (5): d E Aexp a f (5) dt RT We consider that E a and A are varied at variable T under non-isothermal test. Then Eq. (5) should be switched to Eq. (6): d E A exp a f (6) dt RT The natural logarithm of Eq. (6) leads to Eq. (7): d d E ln ln A f (7) dt RT t Therefore, if we draw the diagram of ln(dα/dt) versus 1/T(t) to obtain the line of slope and intercept, the E(α) and ln[a(α)f(α)] can be presented at different conversions [1]. 2.4 TG (Thermogravimetry) TG can be used to observe the relationship between temperature and mass loss for the decomposition or combustion reaction for chemicals. Therefore, the thermal stability of chemicals can be evaluated in a heating system. TG analysis coupled with mass spectrometry purchased from PerkinElmer Clarus 68 was utilized to analyze products of ADVN after the thermal decomposition. The TG experiments were performed from ambient temperature to 6 C with a heating rate of 5, 1, 15 and 2 C min 1 in nitrogen gas purged at a flow rate of 1 ml min 1 [11, 12]. 2.5 Evaluation of IPDT (Integral Procedural Decomposition Temperature) IPDT is a useful parameter for assessing the overall thermal stability for chemicals during the decomposition process. Because the thermal stability of chemicals is considering three factors: (1) initial reaction; (2) end reaction; (3) ratio for mass loss. Doyle created IPDT according to thermogravimetric regions to be a thermal stability parameter via the TG experimental curve. The equations of IPDT are as follows [13-15]: IPDT A K T f Ti T (8) i A S1 S2 / S1 S2 S3 (9) K S1 S2 / S1 (1) where, A and K are the ratio of mass loss from TG experimental curve for different partitioned regions. T i is mass loss onset temperature and T f is mass loss final temperature. S1, S2, and S3 are different partitioned regions from TG plots. S1 is thermogravimetric area under TG curve; S2 is thermogravimetric area of non mass loss; S3 is thermogravimetric area above TG curve. The schematic diagram of IPDT as delineated in Fig. 2.

4 TT i i Mass loss/% Mass loss ( C) 6 4 S1 T f f S3 2 S Fig. 2 Schematic diagram for conception of IPDT. Temperature/ ( C) o C 3. Results and Discussion 3.1 DSC Tests A DSC experiment can accurately and quickly indicate the decomposition reaction of ADVN. Fig. 3 shows the temperature versus heat flow curve from DSC data, recorded from room temperature to 3 C at heating rates of.5, 1, 2, 4 and 8 C min 1 for thermal decomposition reaction of ADVN. A small endothermic curve occurs at 55 C and ΔH endo is ca. 85 J g 1. Because ADVN is a solid chemical, it has to be melted to transform the phase from solid to liquid prior to decomposing. According to the literature [16], the melting point of ADVN is from 4 to 7 C, matching with the experimental results. The exothermic curve of ADVN is corresponding to average T, T p, and T f of 69, 94 and 117 C, respectively, at five heating rates. We observed that when the heating rate was higher, the exothermic curve could be shifted to higher reaction temperature and the maximum exothermic peak was increased. Above data present an excellent relationship for DSC experimental results. In addition, the average ΔH exo is ca. 535 J g 1. According to CCPS (Center for Chemical Process Safety) [17], when the H d of the chemical is arranged between 25 J g 1 and 3 J g 1, it could be defined as potentially hazardous materials because of the degree H d which could rise the adiabatic temperature of 1 C to 2 C. Moreover, if H d exceeds 1, J g 1 or 3, J g 1, it can cause detonation or deflagration, respectively. Thus, a strong heat of decomposition may tend to serious runaway reaction, even causing fire or explosion. The various thermal stability parameters of ADVN are listed in Table ASTM-E698-5 Tests To investigate the reaction mechanism and predict the kinetic behavior for ADVN, a kinetic analysis was conducted based on DSC data. Figs. 4 and 5 demonstrate the calculation results of endothermic and exothermic curves of ADVN using ASTM-E698-5, accompanying with linear regression method, respectively. The plot of lnβ versus 1/T diagram determined

5 284 Exo Heat flow/mw Exo Heat flow/mw o C min -1 8 C min 1 4 o C min -1 4 C min 1 2 o C min C min o C min C min C min o C min Temperature/ o C Temperature ( C) Fig. 3 DSC thermal curves of 98 mass% ADVN at heating rates of.5, 1, 2, 4 and 8 C min 1. Table 1 Calorimetric data from dynamic scanning experiments of 98 mass% ADVN using DSC technique. β/ C min 1 T / C (endo) T / C (exo) T p / C (endo) T p / C (exo) T f / C (endo) T f / C (exo) ΔΗ endo /J g 1 ΔΗ exo /J g the slope and intercept, which represented E a /R and A(dα/dT)f(α), respectively. We found that the calculation of E a was independent of consideration of f(α), thus we could obtain reliable E a value by ASTM-E From the results, the value of E a of endothermic curve is ca. 1,4 kj mol 1 and the value of E a of exothermic curve is ca. 11 kj mol 1, and R 2 of both of them is ca..99. Kakoly and Clark [18] presented in 1999 that if the number of sample size is greater than five heating rates and the value of R 2 is greater than.95, it ought to has high significance. Moreover, there was extreme different in value of E a between phase change and decomposition reaction, indicating that the decomposition reaction could be more easily triggered than a phase change reaction and the reaction rate is lower with higher E a (fundamentals of Arrhenius equation). Therefore, the partial phase change reaction and decomposition reaction may be occurring at the same time, resulting in the actual ΔH exo of ADVN which was higher than DSC test results. Thus, the degree of thermal hazard of ADVN can be inevitably increased. Furthermore, value of E a of decomposition reaction is corresponding to the normal property of azo compounds, that is, ca. 8~2 kj mol 1 [19]. Both the high R 2 and reasonable E a values of the model demonstrated that the model entirely fitted the non-isothermal data. This study used an appropriate

6 y = 128x R 2 =.991 E a = 1,4 kj mol 1 y = -128x E a = 14 kj mol -1 R 2 =.99 ln lnβ Equation y = a + b*x Adj. R-Square Value Standard Error B Intercept B Slope /T 1,/T/K p /K -1 1 Fig. 4 Evaluation of E a of phase change stage for 98 mass% ADVN using ASTM-E y = -13x + 32 E a = 11 kj mol -1 y = 13x + 32 R 2 =.997 E a = 11 kj mol 1 R 2 = ln lnβ Equation y = a + b*x Adj. R-Square Value Standard Error B Intercept B Slope /T p /K -1 1,/T/K 1 Fig. 5 Evaluation of E a of decomposition stage for 98 mass% ADVN using ASTM-E698-5.

7 286 Table 2 Evaluation of E a for 98 mass% ADVN at different reaction stage using ASTM-E Situations E a /kj mol 1 Coefficient of determination/r 2 Phase change stage 1,4.99 Thermal decomposition stage C min o C min C min o C min C min o C min C min o C min C min o C min 1-1 ln(ddt) ln(dα/dt) Fig /T/k 1,/T/K -1 1 Evaluation of E(α) and lna(α) using Friedman method. kinetic model to evaluate kinetic parameters for ADVN. Therefore, the results are able to be applied to further studies. The calculation results of E a are presented in Table Friedman Method Tests Fig. 6 shows the linear regression method of plotting to ln(dα/dt) versus 1/T by Friedman method, and Fig. 7 is the results of E(α) and ln[a(α)f(α)] with different conversions α (~1). We found that the E(α) and ln[a(α)f(α)] is ca. 22 kj mol 1 and 65 s 1 at initial decomposition stage of ADVN, respectively. The value of E a by Friedman method is higher than ASTM method s results. Therefore, the prediction of E a may be underestimated at initial decomposition stage of ADVN by ASTM-E698-5 methods. However, the average E a value is ca. 111 kj mol 1 with α from.2 to.9, and the majority of E a equals 11 kj mol 1 within the range; the error of correlation is less than 1% (negative correlation), indicating that there are the same simulation results with evaluation of E a by ASTM methods. Therefore, these results demonstrated that the range of α was the main decomposition stage of ADVN. In addition, when α increased from.2 to.9, the change of E a was obvious. According to Ref. [2], the change of E a is highly dependent on the reaction mechanism. Therefore, when E a cannot be changed with α, it can be identified as simple nth order reaction (A B). Based on the above results, the Friedman method could be applied to completely predict E(α) or ln[a(α)f(α)] and explore the reaction characteristics for ADVN; and when more kinetic models were used to evaluate kinetic parameters, the results could be at least more reliable and accepted. The value of E(α), ln[a(α)f(α)], and their correlation coefficient are displayed in Table 3. Furthermore, we could use the results of the Friedman model to predict the kinetic behavior for ADVN by substituting E(α) and ln[a(α)f(α)] into

8 287 E(α) (kj mol 1 ) E() (kj mol -1 ) E() under different conversions E(α) under different conversions Fig. 7 ln[a(α)f(α)] (s 1 ) ln[a()f()] (s -1 ) 8 ln[a(α)f(α)] under different conversions ln[a()f()] under different conversions α Calculation results of E(α) and ln A(α) using Friedman method as function of conversion. Table 3 Evaluation of E a and ln[a(α)f(α)] for 98 mass% ADNN using Friedman method with various conversions. α (%) E a (α) ln[a(α)f(α)] Correlation coefficient (r) Initial stage Eq. (1) by determining α and temperature. A kinetic model (dα/dt) of ADVN at five heating rates of.5, 1, 2, 4 and 8 C min 1 could be established by the Friedman model. Fig. 8 shows the comparison of simulation and experimental results, presenting that the simulation curve is in good agreement with experimental curves, so the Friedman model can provide a trustworthy model fitting for depicting the kinetic behaviors of thermal decomposition of ADVN. 3.4 TG Tests Fig. 9 demonstrates mass loss versus temperature diagram from TG test at four heating rates 5, 1, 15 and 2 C. Under the N 2 environment, ADVN is stable up to 8 C and then starts to form two decomposition stages. In first stage, there is a vertical thermo gravimetric curve and the mass loss is ca. 4%. In addition, the curve shape and T of

9 288 Heat Heat flow/mw Exo Exo Exp. 8 o C min C -1 1 Exp. 4 o C min C -1 1 Exp. 2 o C min C -1 1 Exp. 1 o C min C -1 1 Exp..5 o C C min -1 1 Sim..5 o C C min -1 1 Sim. 1 o C min C -1 1 Sim. 2 o C min C -1 1 Sim. 4 o C min C -1 1 Sim. 8 o C min C Fig. 8 Temperature/ ( C) o C Comparison of experimental data and simulation results obtained from Friedman method. Mass loss (%) Mass loss/% C min C min- 1 1 C min C min C min C min- 1 2 C min C min- 1 Fig. 9 Mass loss rate/% min TG thermal curve of ADVN C min C min- 1 1 C min C min C min C min- 1 2 C min C min Temperature/ ( C) o C DTG (differential thermal gravity) of ADVN are very matched to DSC thermal curve of melting stage. Therefore, when melting, ADVN may accompany with decomposition reaction to release lighter molecular to atmosphere, resulting in the chemical structure of ADVN changed and thermal stability decreased. In second stage, the decomposition reaction starts from 8 C to complete reaction.

10 C min o C min 1-1 IPDT: = 152 C o C IPDT/ C o C y = 1.64x R 2 =.99 1 o C min C 1-1 IPDT: = 135 o C C 15 o C min C min 1 IPDT: 142 o C IPDT = 142 C C min o C 1-1 IPDT: = 127 o CC Fig Heating rate/ C min Heating rate/ o C min -1 1 Linear dependence of IPDT with different heating rates. 3.5 IPDT Calculations The thermal stability of ADVN could be assessed by IPDT. The value of IPDT is increased from 127 to 152 C with increasing heating rate because of thermal delay phenomenon of TG tests. Moreover, we plotted the IPDT versus heating rate curve that there was extreme R 2 from linear regression. IPDT will be used as basic and rudimental and dependable thermal stability parameter for future study. The linear dependence of IPDT with different heating rates is shown in Fig Conclusions Based on DSC tests, ADVN has two reaction stages: phase change and decomposition reaction. And it also has high heat of decomposition and low T, thus it can be identified as a potentially hazardous chemical. We used ASTM-E698-5, Friedman method, and IPDT to predict reliable kinetic parameters by non-isothermal test, and completely established the kinetic model of ADVN. Through green thermal analysis approach, we could use a small scale experiment to substitute for complicated and extensive explosion tests, and obtain the thermal hazard information of ADVN to prevent thermal runaway accident during storage, transportation, and manufacturing. References [1] Werber, J., Wang, Y. J., Milligan, M., Li, X., and Ji, J. A Analysis of 2,2 -Azobis (2-Amidinopropane) Dihydrochloride Degradation and Hydrolysis in Aqueous Solutions. Journal of Pharmaceutical Sciences 1: [2] Yang, B., Ahotupa, M., Määttä, P., and Kallio, H Composition and Antioxidative Activities of Supercritical CO 2 -Extracted Oils from Seeds and Soft Parts of Northern Berries. Food Research International 44: [3] Lu, K. M., Lee, W. J., Chen, W. H., and Lin, T. C Thermogravimetric Analysis and Kinetics of Co-pyrolysis of Raw/Torrefied Wood and Coal Blends. Applied Energy 15: [4] Tsai, Y. T., You, M. L., Qian, X. M., and Shu, C. M Calorimetric Techniques Combined with Various Thermokinetic Models to Evaluate Incompatible Hazard of Tert-Butyl Peroxy-2-Ethyl Hexanoate Mixed with Metal Ions. Industrial & Engineering Chemistry Research 52: [5] Nowicka, B., Gruszka, J., and Kruk, J Function of Plastochromanol and Other Biological Prenyllipids in the Inhibition of Lipid Peroxidation A Comparative Study

11 29 in Model Systems. BBA (Biochimica et Biophysica Acta)-Biomembranes 1828: [6] Ramírez, C., Rico, M., Torres, A., Barral, L., López, J., and Montero, M. 28. Macromolecular Nanotechnology-Review Epoxy/POSS Organic-Inorganic Hybrids: ATR-FTIR and DSC Studies. European Polymer Journal 44: [7] Bacosca, I., Hamciuc, E., Cristea, M., Lisa, G., and Bruma, M Poly (ether imide)s Containing Cyano Substituents and Thin Films Made from Them. Thermochim Acta 124: [8] Sovizi, M. R. 21. Thermal Behavior of Drugs: Investigation on Decomposition Kinetic of Naproxen and Celecoxib. Journal of Thermal Analysis and Calorimetry 12: [9] Alonso, M. V., Oliet, M., Garcia, J., Rodriguez, F., and Echeverria, J. 26. Gelation and Isoconversional Kinetic Analysis of Lignin Phenol Formaldehyde Resol Resins Cure. Chemical Engineering Journal 122: [1] Pérez, J. M., Oliet, M., Alonso, M. V., and Rodriguez, F. 29. Cure Kinetics of Lignin Novolac Resins Studied by Isoconversional Methods. Thermochim Acta 487: [11] Park, S. J., and Jin, F. L. 24. Thermal Stabilities and Dynamic Mechanical Properties of Sulfone-Containing Epoxy Resin Cured with Anhydride. Polymer Degradation and Stability 86: [12] Guidechem ,2 -Azobis (2,4-Dimethyl) Valeronitrile (CAS No ). Guidechem. Accessed August 9, [13] Doyle, C. D Estimating Thermal Stability of Experimental Polymers by Empirical Thermogravimetric Analysis. Analytical Chemistry 22: [14] Chiu, Y. C., and Tsai, H. C Thermal and Morphology Properties of Various Silica Contents in Sulfone Epoxy Nanocomposites. Journal of Applied Polymer Science 125: E523-E31. [15] Jin, F. L., and Park, S. J Thermal Properties of Epoxy Resin/Filler Hybrid Composites. Polymer Degradation and Stability 97: [16] Hou, H. Y., Liao, T. S., Duh, Y. S., and Shu, C. M. 26. Thermal Hazard Studies for Dicumyl Peroxide by DSC and TAM. Journal of Thermal Analysis and Calorimetry 1: [17] Sato, Y., Okada, K., Akiyoshi, M., Murayama, S., and Matsunaga, T Diphenylmethane Diisocyanate Self-polymerization: Thermal Hazard Evaluation and Proof of Runaway Reaction in Gram Scale. Journal of Loss Prevention in the Process Industries 24: [18] Kakoly, R. F., and Clark, R. N Spectroscopic Determination of Leaf Biochemistry Using Band-Depth Analysis of Absorption Features and Stepwise Multiple Linear Regression. Remote Sensing of Environment 67: [19] Naranjo, R. A., Conesa, J. A., Pedretti, E. F., and Romero, O. R Kinetic Analysis: Simultaneous Modelling of Pyrolysis and Combustion Processes of Dichrostachys Cinerea. Biomass & Bioenergy 36: [2] Omrani, A., Simon, L. C., Rostami, A. A., and Ghaemy, M. 28. Cure Kinetics, Dynamic Mechanical and Morphological Properties of Epoxy Resin Im 6 NiBr 2 System. European Polymer Journal 44: