Effect of water on gas explosions: combined ReaxFF and ab initio MD calculations

Abstract

In order to understand the intrinsic effect mechanism of water addition on gas explosions, the methane explosion systems with water addition of different mole fractions were systematically studied by reactive force field and first-principles molecular dynamics (MD) simulations. The results show that the effects of water addition on a gas explosion process greatly depend on the system temperature at different reaction stages. Although the water can effectively suppress the methane oxidation process at the initial reaction stage, the same amount of water addition will obviously promote the gas explosion at the later reaction process. The ab initio MD simulations reveal that the water molecules can induce the reactions between ˙HO₂ and ˙H with ˙OH radicals at the initial reaction stage. These reactions consume the reactive radicals, causing the reaction activity of the methane oxidation system to decrease. However, at a higher temperature (about 3000 K), water molecules react with ˙O and ˙H radicals to form extra ˙OH free radicals, and these ˙OH free radicals can be transferred rapidly to interact with the methane molecules by the water molecules. All these processes lead to a better reactive performance at the later reaction stage. These results not only identify the intrinsic interaction mechanism of water addition on the gas explosion system, but also provide a significant theoretical guide for the development of a highly efficient suppression method for gas explosions.

---

1. Introduction

Gas explosions are a severe threat to the production safety in the petrochemical and coal mine industries.^1,2 It is urgently required to develop an efficient and reliable suppression scheme to mitigate or inhibit gas explosions. However, a gas explosion is a rather rapid and complex chemical reaction process initiated by fire sources or high temperatures. This process can be automatically accelerated by the accumulation of free radicals. Furthermore, the concentrations of reactants and free radicals are widely thought to affect the methane branch explosion.^3,4 It is a significant challenge to efficiently suppress gas explosions because of the complexity of the explosion process.

In recent years, the reaction kinetics of gas explosions has become the focus of gas explosion research.⁵ Many reaction kinetics models have been proposed to illustrate the reaction properties during the methane oxidation process.^6–11 For example, the reduction kinetic software SAFEKINEX, proposed by Griffiths et al.,^6,11,12 was successfully used to describe the combustion behavior of hydrocarbons. The whole reaction process of a gas explosion, such as the ignition process, flame behavior and burning velocities of methane–air mixture, are widely examined.^13–16 Some suppression methods for gas explosions have been proposed in order to isolate the gas explosion flame or inhibit the shock wave propagation.^1,2,17–20 Concretely, the porous materials inside the explosion channel can significantly attenuate the gas explosion pressure and shock wave intensity, and they reduce the flame propagation speed.^1,20,21 The water mist sprayed on the flame front can efficiently cool down the temperature and isolate the gas flame front.^{17–19,22–24}

As an economic, convenient and non-pollution resource, the water mist technology has attracted special attention in gas explosion mitigation.²⁴ Zeng and Liang¹⁹ discussed the effect of water addition on the gas explosion, and they found that the induced explosion time was prolonged by feeding additional water. Xu et al.²³ studied the mitigation of methane/coal dust mixture explosions with ultrasonic water mist in the presence of obstacles, and they found that when the volume flux of water mist was larger than a certain amount, the explosions could be completely mitigated. Zhang et al.²⁴ investigated the mitigation of methane/air explosion by ultrafine water fog in a closed vessel. They suggested that the ultrasonic water fog was a promising technology in gas explosion mitigation. Furthermore, the maximum flame propagation velocity of a methane explosion would be reduced significantly by increasing the spraying time.

However, most of the previous studies mainly focus on the macroscopic properties of gas explosions such as flame propagation velocities and explosion pressures. The detailed reaction kinetic properties of a gas explosion process are still missing. It is rather difficult to characterize these detailed properties using experimental methods because the explosion process is rather complex and drastic. Although some previous numerical calculations reported the relevant main reaction species involved in the explosion process, these works mainly investigated intermediate species and free radicals fraction changes. The intrinsic chemical kinetics mechanism of water addition on the gas explosion process is not well understood.

In this work, both reactive force field and ab initio molecular dynamics (MD) techniques were employed to study the effects of water addition on the methane explosion process. The completely different effect mechanisms of water addition on the gas explosion process at different temperatures were discussed at the atomic level. The results show that water addition can significantly decrease the reaction activity at the initial period. However, with a higher temperature at the later reaction stage, the water addition displays a promoting effect on the gas explosion process. The ab initio MD (AIMD) calculations indicate that the consumption of ˙HO₂, ˙H and ˙OH free radicals induced by water molecules results in a poor reaction activity. The formation of extra ˙OH free radicals by the reaction between water molecules with ˙O and ˙H radicals at a high reaction temperature is the primary reason for a higher reaction activity. The rapid transfer of ˙OH free radicals by water molecules can also efficiently promote the methane oxidation process. Such results will not only help to understand the intrinsic interaction mechanism of water in the gas explosion process but also shed light on the development of the suppression method for gas explosions.

2. Computational methods

The effect of water addition on a gas explosion is first studied by the reactive force field molecular dynamics (MD) simulation method. The effect of water on the reaction activity of a gas explosion system is examined under several conditions. Different amount of water molecules with mole fractions from 10 to 40 mol% are added into the system at different reaction stages. The intrinsic reaction mechanism between the water molecules and the main free radicals involved in the gas explosion process is further explored based on the ab initio MD calculation method.

2.1. Reactive molecular dynamics simulation

The initial simulation model, containing 50CH₄ and 100O₂ molecules, was used in all the reactive MD simulations, which was the same conformation as the one explored in our previous work.²⁵ The LAMMPS package was employed for MD simulations with the reactive force field, ReaxFF.^26,27 The parameters for C, H, and O have been identified accurately for such kind of an oxidation process, as shown in previous studies.^26–28 First, the initial system was equilibrated at 298 K for 50 ps to obtain a reasonable configuration. After the equilibrium, the system temperature was further raised to 2000 K from 298 K within 50 ps at a uniform rate. The time step of 0.1 fs was used for all the calculations. The temperature damping constant of 100 time-steps was used for the temperature rising period. After these two stages, the micro-canonical ensemble (NVE) was used to explore the detailed reaction process of the gas explosion system. As soon as the reaction occurs, the system releases a huge amount of energy. Furthermore, the simulated system cannot transfer this energy to the environment or any other zones. Thus, the temperature of the simulated system may be too high to mimic the real situation at the final stage. To resolve this problem, we used the T-rescaling method to control the temperature within 3000 K at the final stage of the simulation to avoid any unrealistic results. Once the temperature reaches over 3000 K, the rescale starts. The whole MD-NVE calculations were run for 2 ns. The more detailed computational setups were the same as the previous work.²⁵

In order to study the effects of water addition on the methane oxidation process, water molecules were equilibrated at 298 K first using MD-NVT (canonical ensemble) method to ensure their random distribution, and then added into the methane–oxygen system in the Z-axis direction at 700, 800, and 900 ps of the MD-NVE simulation process with mole fractions ranging from 10 to 40 mol%. Fig. 1 shows the configuration of the methane oxidation system with 20 mol% water added at 700 ps. The upper part divided by a thin blue line in the picture denotes the additional water molecules. Methane and oxygen molecules are listed on the right section of the figure. Furthermore, the other small fragments represent the intermediates of the methane oxidation reaction. The other configurations of the methane oxidation system with 10, 30, and 40 mol% water addition are listed in the supporting materials (see Fig. S1–S3†). Obviously, the side lengths of the Z-axis direction of these simulation boxes were extended to keep the density of the reaction system constant. Fig. S4–S5† denote the configurations of the methane oxidation system with 20 mol% water added in at 800 and 900 ps, respectively.

Fig. 1 Configuration of methane–oxygen system with 20 mol% water addition (white, gray and red balls denote hydrogen atoms, carbon atoms and oxygen atoms, respectively).

2.2. Ab initio molecular dynamics calculation

The more intrinsic effect mechanism of water addition on the gas explosion process was further studied using the ab initio molecular dynamics method. The interactions between the water molecules and main free radicals form a substantial part of this process. A 3-dimensional periodic model, containing 20H₂O, 10O₂, 5CH₄, 5˙OH, 3˙H and 2˙HO₂, was typically used in this work. In order to ensure the reliability of our results, three different initial configurations (defined as models A, B and C, see Fig. 2) with the same components as listed above, were investigated. Furthermore, several different reaction temperatures ranging from 1000 to 3000 K were explored for all the configurations. The PBE²⁹ exchange-correlation functional was employed for the DFT calculations, combined with the basis set of Goedecker, Teter and Hutter (GTH) pseudopotentials³⁰ for all the atoms. The CP2K/Quickstep³¹ code was employed, which uses a hybrid Gaussian and plane wave basis set. The valence orbitals were expanded in the Gaussian functions at the optimized double-ζ basis set level, including the diffusion terms. A plane wave cutoff of 320 Ry was used in the plane wave expansion of the electron density. The ab initio MD simulations were performed with the canonical ensemble (NVT) following the Born–Oppenheimer (BO) implementation.³¹ The equations of motion were integrated using the velocity Verlet algorithm with a time step of 1.0 fs. The accuracy of such computational setups have been fully tested in the previous works.^32–34 All the calculation processes were run for 2 ps.

Fig. 2 Different initial configurations: A, B, and C of water and main species involved in the gas explosion process (white, cyan and red balls denote hydrogen atoms, carbon atoms and oxygen atoms, respectively).

3. Results and discussion

3.1. Reactive MD simulations of effect of water on gas explosions

In this section, the effects of water addition on a gas explosion at different reaction stages are discussed at first. In such simulations, water molecules with a mole fraction of 20 mol% are added into the gas explosion systems at 700, 800, and 900 ps during the MD-NVE process. When the water molecules with an initial velocity of 298 K are added in the reaction system, the systemic temperature decreases significantly (see Fig. S6 in the ESI†). This would definitely yield a poor reaction activity. Fig. 3 shows the change of the number of reactants. Obviously, the molecular numbers of the methane and oxygen reactants decrease more slowly with the simulation time when the water molecules are added in the system at 700 ps and 800 ps. This indicates that the reaction activity of the methane–oxygen system is efficiently reduced by water addition. Moreover, the passive effect is more apparent at the condition of 700 ps, which has the lowest temperature rising rate at the later reaction stage. However, when the water molecules are added in the explosion system at 900 ps, after a short period of about 40 ps, the number of the reactants decrease even more quickly than that of the system without water addition (see Fig. 3). A promoting effect of water addition on the methane oxidation process is observed, unexpectedly.

Fig. 3 Changes of the concentration of reactants (a) methane and (b) oxygen when water is added at the different reaction stages (700, 800, and 900 ps).

The number changes of the main free radicals (˙H, ˙OH, ˙HO₂, ˙CH₃, CH₃O˙) involved in the gas explosion process are shown in Fig. 4. The maximum values of the number of the main free radicals appear at 1024, 976 and 810 ps, when the water molecules are added at 700, 800 and 900 ps during the MD-NVE simulation process, respectively. The initial times of the maxima of the free radical numbers are apparently postponed when the water molecules are added at 700 and 800 ps. The reaction activity of the explosion system greatly depends on the concentrations of the free radicals.²⁵ The times when the systems reach the greatest reaction activity are considerably delayed at the conditions of 700 and 800 ps, indicating that the water addition reduces the reaction activity of the methane oxidation process. However, when the water is added into the system at 900 ps, the gas explosion system already reaches its highest reaction activity (at 810 ps).²⁵ There is no distinct inhibition effect of water addition on the later reaction process; i.e., when the gas explosion has been fully activated, the same amount of additional water cannot efficiently inhibit the reaction process.

Fig. 4 Changes of the concentration of main free radicals when water molecules are added at the different reaction stages: 700, 800 and 900 ps.

The effects of water addition on the gas explosion with different mole fractions (from 10 to 40 mol%) are further discussed in this section. Fig. 5 displays the changes of the reactant numbers when different numbers of water molecules are added into the system at 700 ps. The results clearly show that the molecular numbers of the reactants decrease slowly versus the simulation time (see Fig. 5). The passive effects on the gas explosion are more apparent at the initial reaction stage (during 700–1100 ps). The suppression effect of the water addition with the increasing mole fraction becomes more and more significant, which can also be illustrated clearly by comparing the temperature changes shown in Fig. S7 in the ESI.† The reaction temperatures decrease more significantly with the increasing water mole fractions. The temperatures of all the different systems have reached 3000 K at about 1100 ps. Then, the systems with water addition attain better reaction activities, especially for the conditions of 30 and 40 mol%. The number of methane molecules decreases from 23 to 9 for the condition of 30 mol% and from 30 to 3 for the condition of 40 mol% during the period of 1100–1600 ps. Both these conditions have a faster oxidation rate than the condition without water addition, which only has 8 oxidized methane molecules. The condition of 40 mol% has an even larger conversion rate than the condition without water addition (see Fig. 5(a)).

Fig. 5 Changes of the concentration of (a) methane and (b) oxygen reactants with the addition of different water mole fractions from 0 to 40 mol%.

This phenomenon can be further understood by analyzing the maximum numbers of the main free radicals involved in the gas explosion process. In Fig. 6, the maximum values of the numbers of the main free radicals appear at 810, 958, 1024, 1096 and 1132 ps, corresponding to the water mole fractions of 0, 10, 20, 30 and 40 mol%, respectively. The initial time of the maximum numbers of free radicals is significantly postponed with the increasing of the water mole fractions. This indicates that the water addition apparently leads to a lower reaction activity of the gas explosion system at the initial stage. The maximum values of free radical numbers do not change considerably among these conditions (see Fig. 6), which indicates that these systems have a very approximate maximum reaction activity. However, the free radical numbers of the conditions of 30 and 40 mol% are noticeably higher than those of the other three conditions during the period from 1100 to 1600 ps. This is responsible for the relatively higher reaction activities of these conditions at the later simulation process.

Fig. 6 Changes of the concentration of main free radicals with the addition of different water mole fractions from 0 to 40 mol%.

During all these processes mentioned above, water, which acts as the additional species, can interact with many important species involved in the gas explosion system. The primary reactions are listed below (see eqn (1)–(7)). Obviously, some reactions consume the free radicals such as the reaction between ˙OH and ˙HO₂. However, the others yield extra free radicals, like the reactions between ˙O, ˙H and H₂O. Moreover, the ˙OH and ˙H can be rapidly transferred by the water molecules (see eqn (4) and (5)). The higher concentration of free radicals denotes a better reaction activity; therefore, the water addition can efficiently affect the gas explosion process by consuming or forming free radicals. However, the main reactions, which exactly affect the reaction activity of the explosion system at different reaction stages, could not be clearly confirmed by simply analyzing these reactions occurring in the MD trajectories because of the change of the reaction temperature and the fast metabolic process of free radicals during the methane oxidation process.

H₂O +˙O → 2˙OH (2)

H₂O + ˙H → ˙OH + H₂ (3)

H₂O + ˙H → ˙H + H₂O (4)

H₂O + ˙OH → ˙OH + H₂O (5)

H₂O + O₂ → ˙OH + ˙HO₂ (6)

H₂O + ˙HO₂ → ˙OH + H₂O₂ (7)

From the discussion above, the water addition displays both suppression and promoting effects on the methane oxidation process at different reaction stages. Note that a similar phenomenon was also observed in the experiment carried out by Zhang et al.³⁵ At the initial period after water was added into the gas explosion system, a larger water mole fraction can more obviously suppress the gas explosion. Furthermore, when the water molecules are added to the reaction system at an earlier stage of methane oxidation, a more efficient mitigation effect would be observed. However, when the gas explosion develops completely, the same amount of water addition will act as a promoter for methane oxidation.

3.2. Ab initio MD calculations for the intrinsic interaction between water and methane oxidation system

The effects of water addition on the gas explosion process were discussed in Section 3.1. Although some similar phenomena were already investigated by Liang et al., there is no exact explanation for these phenomena at the atomic level. Moreover, the reaction temperature dependence of the different water effects on a gas explosion is not discussed yet. It should be noted that water addition definitely decreases the temperature of the gas explosion system, causing a low reaction activity. With the reaction further proceeding, however, the temperature of the methane oxidation system rises, resulting in water molecules to work as a promoter for methane oxidation. Obviously, the reaction temperature will significantly affect the substantial interaction between water molecules and the main species involved in the reaction system.

As shown above, the water addition has different effects on the methane oxidation rate. In order to certify the accuracy of the reactive MD and explore the effect of reaction temperature, AIMD calculations were carried out. The intrinsic mechanism between the water molecules and the main species at different reaction temperatures will also be discussed. The initial structures of the AIMD simulations begin from three distinct configurations (see Fig. 2), and several simulations for each structures were run at different temperatures from 1000 to 3000 K. The primary reactions among the water molecules and the main free radicals involved in these processes are listed in Fig. 7. The main reactions involved in the radical metabolism process are consistent with the results in the reactive MD simulation. The detailed reactions involved in each model at different reaction temperatures are listed in Fig. 7.

Fig. 7 Reactions among water molecule and main free radicals.

The suppression effect of the water addition to a gas explosion can be typically illustrated by reaction I (see Fig. 7, especially for the condition of 1000 K). This is a typical chain termination reaction, which can effectively decrease the number of ˙HO₂ and ˙OH free radicals. The concrete process of this reaction is shown in Fig. 8. The snapshots are drawn from AIMD simulations of model A at 1000 K. The initial reaction occurs with abstracting a hydrogen atom from the ˙HO₂ free radical by water molecules, resulting in the formation of a ˙H₃O free radical. ˙H₃O free radicals can also be formed by directly colliding ˙H and H₂O, which is observed in model B. A ˙H₃O free radical is a metastable structure, which can continuously exist under low temperatures. Like the Grotthuss exchange mechanism,³⁶ although the hydrogen atom of the ˙H₃O free radical can be transferred among the water molecules (see Fig. 8(b)), it has no significance in methane oxidation, because no interaction between methane molecules and ˙H₃O free radical is observed in these calculations. The hydrogen atom is finally transferred to the oxygen containing species,³⁷ ˙OH radical, in this work. These processes mentioned above cause a consumption of the main free radicals, ˙HO₂, ˙H and ˙OH. It is the crucial reason for the suppression effect of water molecules to gas explosions at low reaction temperatures.

Fig. 8 Typical suppression mechanism of the water addition on gas explosion. The snapshots are selected from the AIMD simulations with model A at 1000 K ((a) 880 fs, (b) 950 fs, (c) 1125 fs); (white and red balls denote hydrogen and oxygen atoms, respectively).

As shown in Fig. 7, when the systemic temperature rises, some new reactions between the water molecules and free radicals are observed. All the three new reactions (II–IV) will cause a formation of ˙OH free radicals, which is the main free radical carrier during the methane oxidation process.²⁵ More importantly, reactions III and IV are the main chain branch reactions, which are observed in the condition of 3000 K. These reactions lead to form extra free radicals. This is beneficial to the accumulation of free radicals, corresponding to a high reaction activity at a high reaction temperature.

The other reason for the promoting effect of water addition on the gas explosion would be attributed to the hydroxyl transfer, a similar phenomenon as proton transfer. As shown in Fig. 9, the hydroxyl O can easily interact with the hydrogen atoms of water, forming an adsorption structure (see Fig. 9(b)). Then, the ˙OH abstracts a hydrogen atom from the water molecule, forming a new ˙OH free radical. Finally, the transferred ˙OH interacts with the methane molecule in the vicinity to generate a ˙CH₃ radical and water molecule (see Fig. 9(c)). The rapid transfer of ˙OH radicals between the water molecules can efficiently increase the collision ratio between the ˙OH radical and methane molecules, resulting in an acceleration of the methane oxidation reaction. Furthermore, the hydroxyl transfer process can be apparently enhanced with increasing reaction temperature. Fig. 10 displays the relationship between temperature and the hydroxyl transfer rate. The y-axis denotes the total transfer numbers of the three models discussed above within 2 ps. The transfer rate almost linearly increases versus the reaction temperature, indicating a more efficiently promoting effect of water addition on the methane oxidation process at high temperatures.

Fig. 9 Acceleration mechanism of water addition on a gas explosion. The snapshots are selected from the AIMD simulations with model B at 3000 K ((a) 1205 fs, (b) 1215 fs, (c) 1228 fs); (white, cyan and red balls denote hydrogen atoms, carbon atoms and oxygen atoms, respectively).

Fig. 10 Transfer rate of hydroxyl between water molecules.

As discussed above, when the temperature is about 1000 K, the hydronium radical, ˙H₃O, can easily be formed by interaction between ˙H and ˙HO₂ with water molecules. Compared with ˙H and ˙HO₂ radicals, ˙H₃O has a relatively low reaction activity for methane oxidation, which only reacts with the ˙OH free radical in these systems. This process causes the consumption of the key free radicals, such as ˙HO₂, ˙H, and ˙OH, for the gas explosion process. Consequently, the reaction activity of gas explosion system decreases because of the lack of active centers. The process almost exists during the whole reaction period, especially for temperatures below 2000 K (see Fig. 7). On the other hand, a higher temperature (3000 K) would promote the interaction between the water molecules with ˙H and ˙O free radicals. An extra ˙OH radical is formed, which is the most important free radical carrier for the whole methane oxidation process. Furthermore, the ˙OH radical is transferred rapidly among the water molecules, resulting in a higher collision frequency between ˙OH and methane molecule. The extra formation of radicals and transfer of ˙OH free radicals are the primary reasons for the promoting effect of water molecules on the gas explosion process at relatively high temperatures.

4. Conclusion

The substantial effect of water addition on the gas explosion process was studied based on the combined reactive molecular dynamics and ab initio molecular dynamics calculations. Water addition can efficiently suppress the methane oxidation at the initial reaction process. On the contrary, the same amount of water addition may promote the methane oxidation process at the relative high temperature, when the gas explosion process is completely activated. Water addition can help to consume the main free radicals by inducing the reactions between ˙HO₂, ˙H and ˙OH radicals at the initial stage of the methane oxidation process, causing a poor reaction activity. However, when the water molecules are added at the later reaction process with a high reaction temperature (about 3000 K), the water molecules will react with the free radicals, such as ˙H and ˙O, to form ˙OH free radicals, which can rapidly transfer among the water molecules. This process will enhance the collision frequency between ˙OH and methane molecule, and accelerate the methane oxidation process. Therefore, a single water spray could not efficiently suppress the reaction process when the gas explosion happens completely. Thus, some novel methods should be considered based on our results.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (no. 51222212, 11102194), the CAEP foundation (Grant no. 2012B0302052), the MOST of China (973 Project, Grant no. 2011CB922200).

References

1. B. Nie, X. He, R. Zhang, W. Chen and J. Zhang, J. Hazard. Mater., 2011, 192, 741–747.
2. R. Zhang, B. Nie, X. He, C. Wang, C. Zhao, L. Dai, Q. Li, X. Liu and H. Li, Procedia Eng., 2011, 26, 1467–1472.
3. T. Hirano, J. Loss Prev. Process Ind., 2006, 19, 245–249.
4. T. Hirano, J. Loss Prev. Process Ind., 2001, 14, 553–557.
5. G. Ferrara, A. Di Benedetto, E. Salzano and G. Russo, J. Hazard. Mater., 2006, 137, 654–665.
6. B. F. Gray, J. F. Griffiths and G. A. Foulds, Stud. Surf. Sci. Catal., 1994, 81, 13–24.
7. M. V. Petrova and F. A. Williams, Combust. Flame, 2006, 144, 526–544.
8. K. Gosiewski, A. Pawlaczyk, K. Warmuzinski and M. Jaschik, Chem. Eng. J., 2009, 154, 9–16.
9. R. l. Quiceno, F. Chejne and A. Hill, Energy Fuels, 2002, 16, 536–542.
10. T. Mendiara, M. U. Alzueta, A. Millera and R. Bilbao, Energy Fuels, 2004, 18, 619–627.
11. http://www.morechemistry.com/SAFEKINEX/index.html.
12. J. F. Griffiths, K. J. Hughes and R. Porter, Proc. Combust. Inst., 2005, 30, 1083–1091.
13. G. Lacaze, E. Richardson and T. Poinsot, Combust. Flame, 2009, 156, 1993–2009.
14. M. Bidi, R. Hosseini and M. R. H. Nobari, Energy Convers. Manage., 2008, 49, 3634–3647.
15. E. L. Wang, X. C. Xu and M. C. Zhang, Ind. Eng. Chem. Res., 2011, 50, 9834–9838.
16. J.-H. C. Djamila Brahimi, P. S. Youn, Y.-W. Jeon, S. D. Kim and Ho-J. Ryu, Energy Fuels, 2012, 26, 1441–1448.
17. H. Shimizu, M. Tsuzuki, Y. Yamazaki and A. K. Hayashi, J. Loss Prev. Process Ind., 2001, 14, 603–608.
18. T. Parra, F. Castro, C. Méndez, J. M. Villafruela and M. A. Rodríguez, Fire Saf. J., 2004, 39, 581–600.
19. Y. Liang and W. Zeng, J. Hazard. Mater., 2010, 174, 386–392.
20. J. Sun, Y. Zhao, C. Wei, S. Xie and D. Huang, Procedia Eng., 2011, 26, 954–960.
21. Y. Yang, X. He, G. Luo and H. Wang, Procedia Eng., 2011, 26, 70–74.
22. D. A. Schwer and K. Kailasanath, Proc. Combust. Inst., 2007, 31, 2361–2369.
23. H. Xu, Y. Li, P. Zhu, X. Wang and H. Zhang, J. Loss Prev. Process Ind., 2013, 26, 815–820.
24. P. Zhang, Y. Zhou, X. Cao, X. Gao and M. Bi, Saf. Sci., 2014, 62, 1–7.
25. Z.-H. He, X.-B. Li, L.-M. Liu and W. Zhu, Fuel, 2014, 124, 85–90.
26. K. Chenoweth, A. C. T. van Duin and W. A. Goddard, J. Phys. Chem. A, 2008, 112, 1040–1053.
27. A. C. T. van Duin, S. Dasgupta, F. Lorant and W. A. Goddard, J. Phys. Chem. A, 2001, 105, 9396–9409.
28. J. P. Alister and M. Behdad, J. Phys. Chem. A, 2009, 113, 1539–1547.
29. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
30. S. Goedecker, M. Teter and J. Hutter, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 1703–1710.
31. J. VandeVondele, M. Krack, F. Mohamed, M. Parrinello, T. Chassaing and J. Hutter, Comput. Phys. Commun., 2005, 167, 103–128.
32. L.-M. Liu, M. Krack and A. Michaelides, J. Chem. Phys., 2009, 130, 234702.
33. L.-M. Liu, A. Laio and A. Michaelides, Phys. Chem. Chem. Phys., 2011, 13, 13162–13166.
34. L. Liu, M. Krack and A. Michaelides, J. Am. Chem. Soc., 2008, 130, 8572–8573.
35. P. Zhang, Y. Zhou, X. Cao, X. Gao and M. Bi, J. Loss Prev. Process Ind., 2014, 29, 313–318.
36. D. Marx, ChemPhysChem, 2006, 7, 1848–1870.
37. L.-M. Liu, R. Car, A. Selloni, D. M. Dabbs, I. A. Aksay and R. A. Yetter, J. Am. Chem. Soc., 2012, 134, 19011–19016.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra04178j

---