Sunteți pe pagina 1din 9

Experimental study and theoretical analysis on decomposition mechanism of benzoyl peroxide

Jiayu Lv, Wanghua Chen *, Liping Chen, Haisu Gao, Yingtao Tian, Xin Sun Dept. of Safety Eng., Nanjing University of Science and Technology, Nanjing, Jiangsu, 210094, China Abstract Benzoyl peroxide (BPO), a source of radical used as initiator, is an important member in the organic peroxide family. To master its thermal characters, experimental study and theoretical analysis were employed on studying its decomposition mechanism. The autocatalytic process, which was deduced from dynamic experimental data by differential scanning calorimeter (DSC), was testified by isothermal DSC tests. Thermal scanning unit (TSU) was taken to predict pressure hazard of BPO decomposition and accelerating rate calorimeter (ARC) tests were conducted to investigate the thermal behavior of BPO in adiabatic condition. The high pressure and fast pressure release rate indicated a high possibility of accompanied pressure hazard once the runaway decomposition was trigged. Combined with experimental results, quantum chemistry method was used to locate the most possible reaction path and calculate thermodynamic energies of BPO molecule in gas phase. The calculated total released heat during the whole decomposition was in good accordance with the values tested by thermal analysis instruments. And bond dissociation enthalpy (BDE) obtained at different theory levels provided a comparable value in runaway possibility analysis. Keywords: Benzoyl peroxide; Micro-calorimeter; Autocatalysis; Thermal analysis; Quantum chemistry 1. Introduction Benzoyl peroxide (BPO) is widely used in fine chemistry industry as initiator because of its special structure. The free radicals provided by peroxy group (-O-O-) in BPO play a crucial role in chemical transformations. Nevertheless, like a double-edged sword, thermal instability and explosion property were also brought [1,2]. Many countries have suffered accidents due to BPOs high sensitivities to heat, shock, impact and friction [3]. BPO decomposed with strong exothermic reaction and its kinetics was sensitive to temperature: not detectable at ambient temperature, but as soon as the temperature rises, the decomposition rate can quickly become important. Thus, the follow-up of such reactions seems to be practically impossible in calorimetric reactions because of the energy released during its realization [4]. Although several studies [5-8] have been done to detect the thermal behavior of BPO, most of them were focused on the usual kinetic methods which were based on nth reaction, or presented the experimental results without further mechanism analysis. Besides, papers which applied quantum chemistry method to calculate thermal ability of organic peroxide were less common. This paper tried to work out a comprehensive thermal evaluation of BPO. Differential scanning calorimeter (DSC), thermal scanning unit (TSU) and accelerating rate calorimeter (ARC) were employed to track the temperature histories and violent pressure during BPOs decomposition, testing under dynamic, isothermal and adiabatic condition respectively. Thermal kinetic parameters such as activation energy Ea and frequency factor A were calculated.
*

Corresponding author. Tel.: 0086-025-84315526-8005

E-mail address: chenwh_nust@163.com (Prof. Chen)


1/9

Based on quantum chemistry, the optimized structures of BPO and its corresponding radicals were obtained using density-functional method (DFT) B3LYP and B3PW91 with the 6-31G(d) basis set. After analyzing related energies of the assumptive intermediates, we pointed out the most possible reaction route and presented overall reaction path to study the reaction mechanism. Then, peroxy groups gas-phase bond dissociation enthalpy BDE was calculated. 2. Experimental 2.1. Dynamic and isothermal modes by DSC BPO, with a purity of 98%, was provided by Shanghai Lingfeng Chemical reagent co., LTD. DSC (DSC1, Mettler Toledo) conducted tests both in dynamic and isothermal modes with extra pure nitrogen purging (30mLmin-1). Stainless steel high pressure crucibles were used for the experiment. Sample masses in this work were around 2mg. Fig. 1 illustrated the heat flow versus temperature of BPO decomposition at various heating rate (1, 2, 4 and 8 Cmin-1) at the range of 30-200C. A sudden temperature rise appeared just following with a sharp endothermic peak due to fusion of BPO, which meant the decomposition of BPO overlapped with its fusion in dynamic mode. Because of this, the exothermic onset temperature T0 was unable to accurately determine. However, the maximum temperature Tp could be obtained directly by DSC dynamic scanning tests [9] and the heat of decomposition H was calculated by DSC analysis software STARe, showed in Fig.1. And with the increase of , the value of H added except 8Cmin-1, which might due to stronger endothermic reaction during BPOs fusion at a quick forced heating.

Fig. 1 Heat flow during DSC experiments on BPO at various heating rates

To estimate kinetic parameters, Kissinger method [10], presented as the following equation, was chosen.
ln = ( i Tpi2 ) ln AR Ea Ea RTpi (i=1,2,,6)

(1)

Where is the heating rate, Tp is the temperature maximum, A is the frequency factor, Ea is the activation energy of the reaction, and R is the gas constant. By plotting of ln(i/T2pi) vs. 1/T pi (=1, 2, 4 Cmin-1 were chosen), which was expected to be a straight line, Ea and lnA were calculated as 637kJmol-1 and 204s-1 with a correlation coefficient value of 0.9995 (Fig. 2).

2/9

Fig. 2 Estimated Ea for BPO by Kissingers method

The Ea mentioned above was apparent activation energy. For Ea between 220 and 1000kJmol-1, the decompositions were of autocatalytic nature [11]. These types of reactions required our special attention and should be clearly distinguished from nth order reactions because of their high hazards. The high values of steepness of the exothermal peak in Fig. 1 also indicated the great probability that this decomposition reaction was autocatalytic. Considering isothermal measurements were the most reliable way to detect and characterize an autocatalytic decomposition [12], we showed the thermal profiles of BPO in isothermal mode DSC in Fig. 3. Temperatures were set at 96C, 94C, 92C, 90C and 88C.

Fig. 3 BPO decomposition in the isothermal condition

In Fig. 3, the heat flow passed through a maximum exothermic peak after an induction period and then decreased again. And the higher the isothermal temperature was, the shorter the was gained. The corresponding were 17min, 30min, 55min, 100min and 180min respectively. Fig. 3 also indicated that as the temperature increased, the exothermic peak became sharper and had a poorer symmetry. At 88 and 90C, the DSC curves were symmetrical and identified as bell-shaped curves, which confirmed that BPO decomposes in autocatalytic mechanism. From 92C, the curve symmetry decreased gradually because heat released in exothermic reaction was covered by decalescence in the beginning of reaction. From this phenomenon, we deduced that the fusion point of BPO was approximately from 90 to 92C. To estimate isothermal kinetic parameters of BPO, we employed the Prout-Tompkins model [13]:
k A + B 2B

with rA = dC A dt = k C A CB

(2)

Where r is reaction rate, k is constant of reaction rate, t is time and C is concentration. In isothermal condition, r is proportional to C of the product. So the rate equation can express as a function of conversion :
d dt =kC A0 1-

(3)

After integral, function below can be get:

ln = kt + c (1- )
3/9

(4)

Where c is constant. By inserting the k values calculated in different temperatures into Arrhenius equation, Ea and A was revealed as 191kJmol-1 and 4.18E25s-1 (Fig. 4).

Fig. 4 Kinetic parameters calculated with P-T model

2.2. Thermal scanning by TSU DSC tests have presented a heat release details of BPOs decomposition and pointed out its autocatalytic property. To understand both temperature and pressure histories during decomposition, 0.4g sample was added into a 316ss test bomb (8mL) for TSU (HEL in UK) scanning, and the ramp rate was 0.5Cmin-1. Temperature and pressure recorded was drawn in Fig. 5.

Fig. 5 Temperature and pressure histories of BPO by TSU

After an initial delay due to thermal lagging effects, the sample temperature was found to follow the oven ramp at the same rate with a slight offset. To get the exact T0 value, we used iQ software to reduce the raw data and calculate dT/dt value, presented in Fig. 6. At 90C, dT/dt started to drift from 0. The maximum dT/dt was above 220Cmin-1, and high pressure presented by TSU indicated that failures during manufacture or transportation might give rise to a pressure hazard and more attention should be paid.

Fig.6 Temperature rate vs. temperature of BPO by TSU


4/9

2.3. Adiabatic test in ARC Considering the case of an adiabatic runaway, test was further conducted by esARC (THT. UK). The start temperature was 70C based on results above and temperature step was 3C. Temperature rate sensitivity was set 0.02Cmin-1. Sample mass was 0.32g. Relationship between temperature, pressure and time was displayed in Fig. 7. For better analyzing, the temperature and pressure history during exothermal was zoomed in into Fig. 8.

Fig. 7 BPO decomposition in adiabatic condition

Fig. 8 Temperature and pressure of BPOs decomposition during exothermal

T0 of this decomposition was obtained as 94.3C, and the rises of temperature and pressure were 38C and 7bar respectively. The maximum value of dT/dt and dT/dt tested by ARC were above 150Cmin-1 and 16barmin-1 (Fig. 9). Released heat after thermal inertia correction was 1120Jg-1.

Fig. 9 Temperature and pressure rate vs. temperature of BPO by ARC

Within the tested temperature range, BPO had only one exothermic curve. The temperature remained stable during the induction period and suddenly increased sharply, which was different from the case of nth order reactions (dash line in Fig. 8). This was due to the fact that BPOs autocatalytic reaction initially shows no or only a small heat release (with an onset temperature rate of 0.089Cmin-1). As stated by Manfred A. Bohn [14], the steeper the slope of the self-heat rate is the greater the part of autocatalytic decomposition, decomposition process of BPO showed a high autocatalytic feature from ARC test. Whilst, the delay of temperature increase leaded a later detection of runaway reaction until the reaction rate became sufficiently fast. If a temperature alarm was set as a design of emergency
5/9

measure [15], it was not effective because there was no time left to take measures due to the sharp temperature increase. 3. Theoretical study 3.1. Computational details Density-functional quantum mechanical method, B3LYP [16] and B3PW91 [17,18] were employed to calculate the thermal energies for hemolytic cleavage of bonds which may break during BPOs decomposition, with a basis set 6-31G(d) [19]. After that, two more basis sets and one more quantum chemical method were designated to calculate the O-O bond dissociation energy BDE(O-O) based on the molecular structures optimized with B3LYP/6-31G(d). The calculations were performed with the Gaussian 03 program package [20]. 3.2. Design of reaction path Although for most organic peroxides, the decomposition reactions were regarded as taking up from the break of peroxy bond, the existence of benzoxy radical was controversial [21,22]. In this study, the reaction mechanism of BPO decomposition in gas phase is described below:
1 2(C6 H 5CO)O (C6 H 5CO)OO(COC6 H 5 ) IM1 2 (C6 H 5CO)OOCO + C6 H 5 (C6 H 5CO)OO(COC6 H 5 ) IM2 IM5 3 (C6 H 5CO)OO + C6 H 5 CO (C6 H 5CO)OO(COC6 H 5 ) IM3 IM4

3.3. Results and discussion Geometries of the parent molecules and the corresponding radical (noted as IM1, IM2, Im3, IM4 and IM5) have been optimized at the theory levels, followed by a frequency calculation. To confirm that stationary points were minimum energy structures, vibration frequency was analyzed. Table 1 showed calculated energies E of the five structures, and the lower E values obtained by B3LYP illustrated that this method level gave more stable structures, presented in Fig. 10.
Table 1 Calculated energies of the five structures (Hartree) BPO B3LYP/6-31G(d) B3PW91/6-31G(d) -840.3435 -840.0171 IM1 -420.1517 -419.9878 IM2 -608.7362 -608.4948 IM3 -495.2962 -495.1032 IM4 -344.9206 -344.7850 IM5 -231.5613 -231.4719

6/9

Fig. 10 B3LYP/6-31G(d) optimized structures (Bond distances in , bond angle in )

In Fig. 10, the bond distances R and bond angle A were partly marked. For BPO, the experimental R(4,5) and R(1,2) were 1.45 and 1.54 [23], which proved the creditability of our calculated results. Fig.11 showed that the R, A and symmetry of IM1 to IM4 differed with these of BPO, which had a high symmetrical characteristic. Actually, advantages brought by optimized structures to study the properties of materials were more than this. For instance, dihedral angles D of atoms illustrated the positional relation. In BPO, the calculated values of D(3,2,4,5), D(9,1,2,4) and D(2,4,5,6) were 0.0001, 179.9994 and -179.9503, which meant these atoms were almost coplane. To confirm the most possible reaction mechanism of BPO, the changed thermodynamic values between resultants and reactants were listed in Table 2. Equations used were showed as follows [24]: E0 = Eelec + ZPE

E = E0 + Evib + Erot + Etrans H = E + RT Where E0 is original energy, Eelec is electronic energy, ZPE is zero-point energy, E is thermal energy,
Evib is vibrational energy, Erot is rotational energy, Etrans is translation energy and H is thermal enthalpy.
Table 2 Thermodynamic calculations of the three reaction paths (298K, kJ/mol) Path B3LYP/6-31G(d) E0 E H B3PW91/6-31G(d) E0 E H 1 91.3359 89.1357 91.6142 94.1084 92.0422 94.5206 2 100.4438 103.1244 105.6029 110.3261 109.2024 111.6809
7/9

3 318.3944 317.2260 319.7045 323.8423 322.7396 325.2181

4 17.2653 17.1340 19.6125 16.1600 25.9788 18.4573

5 -455.4087 -457.2124 -459.6909 -457.0287 -458.7903 -461.2688

6 -329.5423 -333.8087 -328.8518 -330.6003 -334.7906 -329.8337

Table 2 displayed that the change of thermodynamics parameters were in good consistency by both methods, and values of E0, E and H were sequenced as path1 < path2 < path3. Thinking of this in thermodynamic way, path1 was the easiest reaction path for BPOs decomposition. H. Fischer [25] described the NMR diagram when BPO decomposed in cyclohexanone, and the absorption spectrum of benzene was obtained. According to this, the reaction mechanism followed with path1 could be composed as:
4 (C6 H 5CO)O C6 H 5 +CO 2 5 2 C6 H 5 C6 H 5 C6 H 5

Therefore, the overall reaction path of decomposition reaction of BPO was concluded below: 6 (C6 H 5CO)OO(COC6 H 5 ) C6 H 5C6 H 5 +2CO 2 The related energies were superadded to Table 2. Negative values listed in Table 2 illustrated that reaction path5, generation of biphenyl, was heat generating reaction. And the total heat released during the decomposition was calculated to be 333.8 and 334.8 kJmol-1 using the two different methods, equal to 1378 and 1382Jg-1, which provided good convenience in dealing with hazardous evaluation of dangerous substances due to its approximate heat data compared to calorimetry tests. Table 3 were BDE(O-O) values under different methods and basis sets. It was shown that with the increase of basis set, the BDE had a tendency of diminution. Limited by the hardware, higher basis sets were abandoned. However, due to the dipole and - affection in BPO molecule, polarization should be included. If considering the same basis set, B3LYP theory level was more accurate than MP2 method.
Table 3 BDE of O-O calculated with different methods (298K, kJ/mol) B3LYP/6-31G(d) BDE 105.2826 B3LYP/6-31+G(d) 94.5180 B3LYP/6-311G(d,p) 89.7921 MP2/6-31+G(d) 119.1977

4. Conclusions The hazards of heat release and pressure generation were predicted by three micro-calorimeters under dynamic, isothermal and adiabatic modes. In dynamic DSC measurements, an endothermic process precedes the exothermic decomposition of BPO, which hindered the T0 capture. By analyzing kinetic parameters using Kissinger method, we found this conversion affected method was not suitable for BPOs autocatalytic decomposition. Therefore, based on isothermal results, Ea and A was revealed as 191kJmol-1 and 4.18E25s-1 by P-T model and the fusion point of BPO was deduced approximately from 90 to 92C. T0 obtained by TSU and ARC was 90 and 94.3C, meanwhile, heat release rates and pressure generation observed in these experiments showed high possibility of thermal runaway. By applying quantum chemistry method on further mechanism study, the most possible reaction path of BPOs decomposition was O-O bond dissociated into benzoxy radical according to calculation by Gaussian 03 software package. The total released heat was in good accordance between experimental calorimetry and theoretical results. BDE(O-O) was calculated at B3LYP theory level and MP2 level with three different basis sets, and B3LYP method had a more accurate values. References
[1] L. Jiayu, W. Shuiai, C. Wanghua, C. Gufeng, C. Liping, Adv. Materials Research, 550-553 (2012) 2782-2785. [2] L. Jiayu, C. Wanghua, C. Liping, T. Yingtao, S. Xin, Procedia Engineering, 43 (2012) 312-317. [3] W. Guanwei, S. Qimin, Thermal hazard evaluation of benzoyl peroxide, National Yunlin University of Science
8/9

and Technology, Taiwan, 2005. [4] I. Ben Talouba, L. Balland, N. Mouhab, M.A. Abdelghani-Idrissi, J. Loss Prevent. Process Ind. 24 (2011) 391-396. [5] Z. Fan, X. Chuanxin, G. Jing, Chem. Engineer, 1 (2010) 23-26. [6] Lu K T, Chen T C, Hu K H, J. Hazardous Materials, 161 (1) (2009) 246-256. [7] L. Xinrui, H. Koseki, Thermochimica Acta, 403 (2005) 113116. [8] F. Zaman, Beezer A E, Mitchell J C, International J. Phamaceutics, 227 (1-2) (2001) 133-137. [9] Z. Hai, X. Zhiming, G. Pengjiang, H. Rongzu, G. Shengli, N. Binke, F. Yan, S. Qizhen, L. Rong, J. Hazardous Materials, A94 (2002) 205-210. [10] Kissinger H E. Anal, Chem., 29 (11) (1957) 1702-1706. [11] Leila Bou-Diab, Hans Fierz, J. Hazardous Materials, 93 (2002) 137-146. [12] Francis Stoessel. C.Wanghua, P. Jinhua, C. Liping, tran, Thermal Safety of Chemical Process: Risk Assessment and Process Design, Science Press, Beijing, China, 2009, 270-282. [13] Michael E. Brown, Beverley D. Glass, J. Pharma., 190 (1999) 129-137. [14] Manfred A. Bohn, Heike Pontius. ICT. 57 (2012) 1-40. [15] J. M. Dien, H. Fierz, F. Stoessel, G. Kille, Chimia, 48 (1994) 542. [16] A.D. Becke, J. Chem. Phys.98 (1993) 5648. [17] J.P. Perdew, K.Burke, Y. Wang, Phys. Rev. B 54 (1996) 16533. [18] Z. Xiulin, C. Wanghua, L. Jiacong, K. Jinlin, J. Molecular Structure, 810 (2007) 47-51. [19] A.D. McLean, G.S. Chandler, J. Chem. Phys. 72 (1980) 5639. [20] M. J. Frisch, G. W. Trucks, H. B. Schlegel, et.al, Gaussian, Inc., Pittsburgh PA, 2003. [21] W. Qihua, S. Yuxiang, Z. Yaoxing, Introductory theory of organic reaction mechanism, Higher Education Press, Beijing, China, 1991. [22] Q. Muge, Explosion reaction mechanism and identification study of organic peroxides, Beijing institute of technology, 2004. [23] Y. Yunbin, X. Tao, G. Yingmin, Handbook of Chemistry and physics, Shanghai Scientific and Technical Publishers, Shanghai, China, 1985, 200-202. [24] E. Frisch, Michael J. Frisch, Gary W. Trucks, Gaussian 03 help, 2003. [25] H. Lin, Reactive intermediates, Higher Education Press, Beijing, China, 1990, 115-125.

9/9

S-ar putea să vă placă și