- Email: [email protected]
Statistical optimization of supercritical carbon dioxide antisolvent process for preparation of HMX nanoparticles
J. of Supercritical Fluids 72 (2012) 248–254
Contents lists available at SciVerse ScienceDirect
The Journal of Supercritical Fluids journal homepage: www.elsevier.com/locate/supflu
Statistical optimization of supercritical carbon dioxide antisolvent process for preparation of HMX nanoparticles Yadollah Bayat a , Seied Mahdi Pourmortazavi b,∗ , Hatef Iravani a , Hamideh Ahadi a a b
Department of Chemistry and Chemical Engineering, Malek Ashtar University of Technology, Tehran, Iran Faculty of Material and Manufacturing Technologies, Malek Ashtar University of Technology, Tehran, Iran
a r t i c l e
i n f o
Article history: Received 29 February 2012 Received in revised form 24 September 2012 Accepted 26 September 2012 Keywords: HMX Micronization SAS process Energetic nanoparticles ANOVA Taguchi robust design Statistical optimization
a b s t r a c t Taguchi robust design was employed as a statistical optimization method to optimize parameters of SAS process in order to prepare HMX nanoparticles. The current application of the Taguchi statistical optimization was highly efficient in optimizing of experimental variables which are effective in HMX particle size produced by the SAS technique. The effect of operation conditions such as: pressure, temperature, HMX concentration, solution flow rate, solvent identification, and flow rate of CO2 on the size of micronized HMX particles under various levels was investigated. The effects of these variables on the particle size of produced HMX were quantitatively evaluated by the analysis of variance (ANOVA). The results showed that the size of HMX particles prepared by SAS technique could be tuned effectively by controlling the main parameters under their optimum level. Finally, the optimum conditions for preparation of HMX nanoparticles via SAS method were proposed. The results of ANOVA showed that 3.5 mol/l HMX concentration, 3 ml/min solution flow rate, cyclohexanone as solvent, and 70 ml/min flow rate of CO2 are optimum conditions for producing HMX nanoparticles. Experimental observations showed that under optimum conditions of SAS process, the particle size of produced HMX is about 56 nm. © 2012 Elsevier B.V. All rights reserved.
1. Introduction HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine) is a high explosive which is widely used in various formulations of plastic bonded explosives (PBX), double base propellants and composite propellants due to its high calorific potential, high density and smokeless combustion products [1,2]. HMX could be prepared in four different polymorphs that commonly known as ␣ (orthorhombic),  (monoclinic), ␥ (monoclinic) and ␦ (hexagonal) phases. Among these various phases, -HMX has the most stability one and possesses the highest explosion power which arises from its crystal phase and its high density. Therefore, preparation of HMX in -phase would be exclusively interested [3]. The size and shape of energetic material particles have a significant influence on the different properties of propellants and explosive formulations [4–6]. Several properties of these compounds such as sensitivity, mechanical properties, content of total solid, density, homogeneity of formulation, and burning rate are dependent to the particle size of energetic materials [7,8]. The particle size reduction of energetic compounds could be lead to the reduction defects, voids and inclusions of crystal and hence
∗ Corresponding author at: P.O. Box 16765-3454, Tehran, Iran. Tel.: +98 2122952285; fax: +98 2122936578. E-mail address: [email protected] (S.M. Pourmortazavi). 0896-8446/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.supflu.2012.09.010
reduction the sensitivity of the formulations and enhancing their safety [9,10]. In order to tailoring fine particles, various conventional methods including milling, liquid crystallization and spray drying could be employed [11–13]. But, these techniques are not suitable for treating energetic materials due to some drawbacks such as difficulty of controlling particle size and particle size distribution, high residual solvent, high shear forces, and unexpected problems during their size reduction process containing thermal explosion of energetic materials [14,15]. Supercritical fluids (SCFs) techniques may be useful to overcome these drawbacks of classical micronization processes and have been applied by researchers to size reduction of explosives [16]. The supercritical fluid is characterized with diffusivities considerably higher than those of liquid, which make it suitable for fast supersaturation of solute and then precipitation of small particles of targeting material. Until today, various techniques for particle formation via supercritical fluid technology including two general processes have been used: supercritical antisolvent (SAS) process [17,18] and rapid expansion from supercritical solutions (RESS) process [19,20]. The SCFs based techniques have many advantages rather than classical micronization processes such as preventing thermal degradation of the target compounds, no mechanical damage, and no residual solvent problem [21]. The main work for an explosive particle formation via SCFs processes such SAS recrystallization and RESS process is preparation of fine particles with various sizes and shapes
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
249
Fig. 1. (a) Schematic of the experimental setup used for SAS process and (b) and the cross section of used nozzle. Taken from Ref. [18] with permission.
by controlling the supersaturation and nucleation rates of the explosive in the SCF media using variation in solvent strength. Especially, the SAS process as a recrystallization technique is a promising process for production of solvent-free explosive particles with fine sizes [22]. The particle size of compounds which prepared by the aid of SAS process is dependent on the various operation parameters and procedure conditions; thus, optimization is an important step in developing SAS technique for size reduction of various materials [23]. Generally, two systematic procedures including simultaneous and sequential methods are used in optimization of the experiments [24]. The main disadvantages of sequential methods are slow convergence on complex response surface and difficulty in dealing with response surface with high dimensionality; while, simultaneous optimization techniques do not suffer from such problems [25]. On the other hand, an obvious disadvantage of the full factorial designs is the number of experimental trials required which increases geometrically by increasing number of variables. Fortunately, this required number of experimental trials is minimized by use of fractional factorial experiments, such as orthogonal array (OA) designs [26]. Since statistical experimental design methods could be carefully explored the experimental space while studying several variables via a small number of observations; Taguchi robust design as a statistical experiment optimization has been widely used in various procedures optimization [27–29]. Therefore, various control factors could be simultaneously investigated and optimized by the aid of Taguchi statistical design. The aim of this study was applying the Taguchi robust design to the optimization of various parameters affecting on particle size of HMX prepared via SAS process and evaluating the effect of these parameters on the size of produced HMX particles. Our goal in this study was optimization of conditions to reduce the size of produced HMX particles and preparation of HMX nanoparticles. To the best of our knowledge, various reports could be found on the micronization of HMX by various techniques [1,30–34]; but, until today there is no report on the preparation of HMX nanoparticles via SAS process. Therefore, optimization of SAS process parameters affecting the size of HMX particles and preparation nanoparticles of this explosive is interested. 2. Experimental Cyclotetramethylene tetranitramine (Octogen, HMX) was used to investigate the possibility for formation of its nanoparticles using compressed SAS process. It should be noted that HMX is a high explosive. Thus, during its operation much care is required to
prevent unexpected hazard such as accidental explosion. Fortunately, in the SAS process which used in this study, the solid of explosive is dissolved in a liquid, and then a supercritical fluid, which is miscible with the liquid but in lower solvent power to the explosive, is added to recrystallize the HMX. This wet operation process is safer for some difficult-to-handle high explosive such as HMX. Analytical grade acetone, cyclohexanone, and ␥-butyrolactone as solvents were used as received from Merck Company (Germany). Carbon dioxide (99.99% purity), contained in a cylinder with an eductor tube, was obtained from Sabalan Co. (Tehran, Iran). The scheme of the used SAS apparatus is shown in Fig. 1. This SAS system is consisting of a precipitation chamber and a gas–liquid separation chamber. As could be seen in Fig. 1a, the CO2 is cooled by a cooler (4) before being compressed via a liquid pump (8) and the pressure is controlled with a backpressure-regulating valve. After pre-heating of CO2 in a heat exchanger (13), CO2 enters to the precipitation chamber (18). Simultaneously, the solution containing target compound is pumped, heated and fed to the precipitation chamber (1000 ml) through a stainless steel nozzle (16). Meanwhile, Fig. 1b shows a close up with the details of the nozzle, which is a laser-drilled orifice (150 m inner diameter). This nozzle is placed at the top of the precipitation chamber; which is located in a distinct inlet point from the CO2 . A stainless steel filter (17) with 200 nm pore sizes was put into the precipitation chamber to collect the micronized particles and to let the SC-CO2 /organic solvent mixture pass through. The flow rate of the mixture that leaves the precipitator is controlled via valve (21), which is located between the precipitation chamber and the gas–liquid separation chamber (22), where the mixture suffers a decompression (pressure <5 MPa) to induce separation of the CO2 from the organic solvent. The prepared HMX samples via SAS process under various experimental conditions were characterized by scanning electron microscopy (SEM). Scanning electron micrographs were recorded using on a Philips XL30 series instrument using a gold film for loading the dried particles on the instrument. Gold films were prepared by a sputter coater model SCD005 made by BAL-TEC (Switzerland). To optimize experimental conditions of SAS process, for the micronization of HMX, an experimental design approach was followed. Various operation parameters including pressure, temperature, HMX concentration, solution flow rate, nature of the solvent, and flow rate of CO2 were investigated in three levels using a L18 array proposed by Taguchi robust design as shown in Table 1. As could be seen in this table, each of variables was studied under three different levels. These levels were selected based on the
250
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
Table 1 Assignment of the factors and levels of the experiments by using an OA18 (36 ) matrix and average particle size of produced HMX as response. Average particle size (nm)
Flow rate of CO2 (ml/min)
Solvent nature
Solution flow rate (ml/min)
HMX concentration (ml/l)
Temperature (◦ C)
Pressure (bar)
Trial no.
421 330 196 230 410 428 265 307 265 89 185 671 180 224 160 285 195 278
40 70 100 100 40 70 70 100 40 70 100 40 100 40 70 40 70 100
Acetone Cyclohexanone ␥-Butyrolactone Cyclohexanone ␥-Butyrolactone Acetone ␥-Butyrolactone Acetone Cyclohexanone Cyclohexanone ␥-Butyrolactone Acetone Acetone Cyclohexanone ␥-Butyrolactone ␥-Butyrolactone Acetone Cyclohexanone
1.5 2.5 3.5 1.5 2.5 3.5 2.5 3.5 1.5 3.5 1.5 2.5 2.5 3.5 1.5 3.5 1.5 2.5
1 2 3 2 3 1 1 2 3 3 1 2 3 1 2 2 3 1
35 60 85 35 60 85 35 60 85 35 60 85 35 60 85 35 60 85
80 80 80 180 180 180 280 280 280 80 80 80 180 180 180 280 280 280
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
previous studies [32–34] and the experimental conditions which were accessible. 3. Results and discussion Simultaneous optimization techniques, such as Taguchi robust design are included a previously planned array experiments; then, collection the results and determination of the optimum condition via constructing a response surface or by retention mapping [35]. In orthogonal array designs, the designed arrays are used to assigning some factors to a series of experiment combinations whose results could then be analyzed via a common mathematical procedure [36]. Meanwhile, the significant effects of each factor are independently determined. The main performing of the experiments via orthogonal arrays is possibility for the separation of different effects [37]. 3.1. Optimization of parameters for particle size reduction of HMX by SAS procedure In this research, HMX explosive was micronized by SAS procedure. HMX was dissolved in various solvents and micronization process was performed via SAS technique by the supercritical carbon dioxide under different conditions of operation in accordance with Table 1. The purpose of this study was to investigate the effect of various procedure parameters on the size of produced HMX particles. The factors included in this study were pressure, temperature, HMX concentration, solution flow rate, nature of solvent, and flow rate of CO2 . Factors and their tested levels are shown in Table 1. Fig. 2 shows the SEM images of four HMX samples obtained by SAS procedure under various conditions of Table 1. Also, the average size of produced HMX particles at each experiment is given in this table as the response. In Taguchi robust design, analysis of the collected data while there is no interaction between variables three steps includes following steps: determination of the optimum condition for the studied procedure; assignment the individual effect of each parameter on the result which in our study is the average size of prepared HMX particles; and finally, prediction the process performance under the proposed optimum conditions. The averages values correspond to effect of the each parameter under any level were computed according to assignment of the designed experiments. Therefore, calculation for determining the influence of each parameter at any investigated level on the size of HMX particles (prepared via SAS procedure) was carried out by pooling the results of all
experiments in which a factor was set at same level and thus dividing by the number of trials used this level. As an example, the average effect of supercritical fluid pressure at level 1 (80 bar) was determined by pooling the average particle size of HMX obtained via all experiments in which pressure of the supercritical fluid was set at level 1 (trials 1, 2, 3, 10, 11, and 12; while here the number of experiments is six) as shown in Table 1. On the other hand, Fig. 2 shows the curves correspond to the variation in the particle size of HMX due to changes in the level of any factor. These graphs show variations in average particle size of HMX due to the effect of each variable when the level of the variable is changed. Table 2 presents the results for analysis of variance (ANOVA) for the obtained data (particle size of produced HMX via SAS process). In this study, the effects of various parameters including pressure, temperature, HMX concentration, solution flow rate, solvent identification, and flow rate of CO2 on the average particle size of produced HMX by SAS technique at three different levels were investigated. The results of ANOVA for HMX micronization experiments via SAS confirm that (at 90% confidence level) except pressure and temperature, all other variables have significant effects on the size of HMX particles produced via SAS process. In this work, there is no interaction between the variables. It was found that pressure and temperature are not significant parameters for the controlling size of HMX particles during its micronization via SAS method. Also, the effect of HMX concentration and its solution feeding flow rate on the size of HMX particles was studied. Three different HMX concentrations (1, 2, and 3 mol/l) and solution feeding flow rates (1.5, 2.5, and 3.5 ml/min) were investigated and our finding showed that both parameters have significant effects in tuning the size of produced HMX particles (Fig. 3). Other investigated parameters for determining their significance on the size of HMX particles were nature of solvent, and flow rate of CO2 . The choice of solvent is known to affect particle size and morphology of produced particles in SAS technique [16]. In SAS process as a micronization technique, the target compound should be dissolved in the solvent to form a homogeny solution. In ideal conditions, the target is completely insoluble in the produced mixture of antisolvent/solvent (SC-CO2 /liquid solvent) during the SAS process; while the solvent is totally miscible with the antisolvent (SC-CO2 ). Therefore, injection of the solution into the SC-CO2 causes a high supersaturation of the injected solution and hence, nucleation of the target solute happened. Therefore, the nature of the solvent via its solvating power, interaction of solvent–solute and its miscibility with SCCO2 could be effective in the morphology of the formed particles
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
251
Fig. 2. SEM images correspond to HMX particles obtained by different runs of OA18 (Table 1) via SAS process: (a) run 4, (b) run 6, and (c) run 9.
via SAS process. In this study, three different solvents containing acetone, cyclohexane and ␥-butirolactone were used for dissolving of HMX. The results of ANOVA showed that the nature of used solvent is a significant parameter for the controlling the particle size of produced HMX. Also, significance of CO2 flow rate under three levels of 40, 70, and 100 ml/min was studied for its effect on the particle size of product. Our finding showed that CO2 flow rate is a significant parameter for the controlling the size of HMX particles. By considering the results obtained by ANOVA it could be seen that 3.5 ml/min is the optimum flow rate for HMX solution during its micronization via SAS process. Also, by increasing concentration of the HMX in the initial solution from 1 to 2%, increases the size of precipitated HMX particles; while, enhancing concentration from 2 to 3% causes a significant decrease in the size of prepared HMX particles. Beyond a certain point, as the solution concentration is increased, nucleation process dominates particle growth, which results in larger particles. Thus, 3% was selected as optimum concentration for HMX. Among various solvents investigated cyclohexanone showed the best performance in size reduction of HMX by SAS technique;
because this solvent enhances dissolution and miscibility of HMX solution in SC-CO2 . On the other hand, 70 ml/min was chosen as optimum CO2 flow rate during SAS process for production nanosized HMX. 3.2. Preparation of HMX nanoparticles under optimum condition Under the optimized conditions which proposed by analysis of the data obtained via designed OA18 matrix for optimization of variables, the particle size of HMX recrystallized via SAS process could be predicted according to the following expression [36,37]: Yopt =
T T + Fsol − N N +
FCO2 −
T N
+ CHMX −
T N
+ SNature −
T N
where T/N is average particle size of HMX + contribution of Fsol , CHMX , SNature and FCO2 in their minimum values which calculated from average effect of each factor; while, T is the grand total of
Table 2 ANOVA results for micronization procedure of HMX via SAS procedure using OA18 (36 ) matrix. Factor
DOFa
Sb
Vc
DOF
S Pooledf
F d
P e
Pressure (bar) Temperature (◦ C) HMX concentration (ml/l) Solution flow rate (ml/min) Solvent nature Flow rate of CO2 (ml/min) Error
2 2 2 2 2 2 5
16792.2 10097.4 35152.4 60408.7 35883.5 99780.1 25535.9
8396.1 5048.7 17576.2 30204.4 17941.7 49890.1 5107.2
– – 2 2 2 2 9
– – 23502.3 48758.6 24233.3 88130.0 52425.5
– – 3.02 5.18 3.08 8.56 –
– – 8.29 17.19 8.54 31.07 34.91
a b c d e f
Degree of freedom for each variable; DOF is degree of freedom for each variable after pooling of insignificant parameters. Standard deviation (mean squares) for each parameter; S is pure standard deviation for each parameter after pooling of insignificant parameters. Mean square (variance) value for each parameter; V is the variance value for each parameter after pooling of insignificant parameters. The variance ratio or F statistic (the ratio of variance due to the effect of a parameter and variance value due to the error term). Percent contribution of each parameter or error term in the results of process. The critical value was at 90% confidence level; pooled error results from pooling insignificant parameters [25].
252
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
Fig. 3. Average particle size of HMX obtained via SAS process correspond to the effect of each level for various factors: (a) pressure; (b) temperature; (c) concentration of HMX; (d) flow rate of solution; (e) nature of solvent, and (f) CO2 flow rate.
particle size for all runs, N is total number of results, Yopt is particle size of HMX under optimum condition, Fsol , CHMX , SNature and FCO2 are average particle size of HMX at optimum levels of HMX feed solution, concentration of HMX in the solution, nature of the solvent and flow rate of supercritical carbon dioxide, respectively. On the other hand, the confidence interval (C.I.) for the predicted
Fig. 4. SEM image of prepared HMX under optimum condition.
particle size of HMX under the optimum conditions could be calculated from the following expression [36]:
CI = ±
F˛ (f1 , f2 ) · Ve Ne
Fig. 5. Size distribution for HMX nanoparticles prepared under optimum condition of SAS process.
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
253
Fig. 6. (a) XRD pattern for the prepared HMX nanoparticles obtained by SAS process under optimum conditions and (b) standard XRD pattern for  form of HMX obtained from PC-APD, diffraction software.
where F˛ (f1 , f2 ) is the critical value for F that could be obtained from the F table at the degrees of freedom (DOF), f1 and f2 at the level of significance ˛ (in this work, ˛ = 90%), f1 = DOF of mean (which always equals 1), f2 = DOF of the pooled error term and Ne could be calculated via the following equation: Ne =
Number of trials DOF of mean(always1) + DOF of all factors used in estimation
Computations for estimation the particle size of HMX under optimum condition and CI for this predicted particle size show that at 90% confidence level, the size of HMX particles under optimum conditions of SAS process will be about 75 ± 20 nm. In this investigation, the effects of various parameters including pressure, temperature, HMX concentration, solution flow rate, nature of solvent, and flow rate of CO2 on the particle size of HMX produced via SAS process at three different levels were investigated and our finding showed that 3% HMX concentration, 3.5 ml/min solution flow rate, cyclohexanone as solvent, and 70 ml/min as flow rate of CO2 are optimum condition for preparation of HMX nanoparticles. Meanwhile, ANOVA results for these experiments indicated that (at 90% confidence level) pressure and temperature of the supercritical carbon dioxide in SAS process have no significant effects on the size of produced HMX particles. In the next step of this study, HMX nanoparticles were prepared under optimum conditions proposed by ANOVA. Fig. 4 shows the SEM image for prepared HMX nanoparticles under optimum conditions of SAS process. Meanwhile, Figs. 5 and 6 show the size distribution graph for HMX particles obtained under these optimum conditions. As could be seen in this graph, the average particle size of HMX particles is about 56 nm.
4. Conclusion In this study, HMX nanoparticles were successfully prepared via SAS process. Taguchi robust design method was applied to optimize the operation conditions of HMX micronization by this method with aim of preparation HMX nanoparticles. Several parameters affecting on the size of HMX particles during SAS procedure were investigated and optimized. Our result showed that pressure and temperature of the supercritical carbon dioxide have no significant effects on the particle size of produced HMX during SAS process; while, other studied variables including HMX concentration, solution flow rate, solvent identification, and flow rate of CO2 have a considerable role in determination of product particles in this
procedure. Finally, under optimized conditions of SAS process, HMX nanoparticles with average particle size of 56 nm were prepared. References [1] Z. Yongxu, L. Dabin, L. Chunxu, Preparation and characterization of reticular nano-HMX, Propellants, Explosives, Pyrotechnics 30 (2005) 438–441. [2] M. Fathollahi, S.M. Pourmortazavi, S.G. Hosseini, Particle size effects on thermal decomposition of energetic material, Journal of Energetic Materials 26 (2008) 52–69. [3] M. Herrmann, W. Engel, N. Eisenreich, Phase transitions of HMX and their significance for the sensitivity of explosives, in: Proceedings of the Technical Meeting of Specialists MWDDEA AF-71-F-G-7304-Physics of Explosives, 1990. [4] U. Teipel, H. Krober, H. Krause, Formation of energetic materials using supercritical fluids, Journal of Supercritical Fluids 26 (2001) 168–173. [5] Y. Bayat, V. Zeynali, Preparation, Characterization of nano-CL-20 explosive, Journal of Energetic Materials 29 (2011) 281–291. [6] A. Okmar, R.P. Nafday, L.W. Brandon, H. Janson, Patterning high explosive at the nanoscale, Propellants, Explosives, Pyrotechnics 31 (2006) 376. [7] R. Sivabalan, G.M. Gore, U.R. Nair, A. Saikia, S. Venugopalan, B.R. Gandhe, Study on ultrasound assisted precipitation of CL-20 and its effect on morphology and sensitivity, Journal of Hazardous Materials 139 (2007) 199–203. [8] B. Wang, L.L.M. Yu-cun, Study on the influence of particle size on the impact sensitivity of HMX, Journal of North China Institute of Technology 26 (2005) 35. [9] X. Song, Y. Wang, C. An, X. Guo, F. Li, Dependence of particle morphology and size on the mechanical sensitivity and thermal stability of HMX, Journal of Hazardous Materials 159 (2008) 222–229. [10] Z. Shuhai, Z. Jinglin, W. Jingyu, Preparation and properties of sub-micron nitramine explosives, Fraunhofer-Institut fur Chemische Technologie (ICT) (2004) 100–104. [11] S. Jain, S. Nandagopal, P.P. Singh, K.K. Radhakrishnan, B. Bhattacharya, Size and shape of AP and their influence on properties of composite propellant, Journal of Defense Science 59 (2009) 294–299. [12] S.M. Pourmortazavi, M. Rahimi-Nasrabadi, A.A. Davoudi-Dehaghani, A. Javidan, M.M. Zahedi, S.S. Hajimirsadeghi, Statistical optimization of experimental parameters for synthesis of manganese carbonate and manganese oxide nanoparticles, Materials Research Bulletin 47 (2012) 1045–1050. [13] G. Yang, F. Nie, T. Li, Q. Guo, Z. Qiao, Preparation and characterization of nanoNTO explosive, Journal of Energetic Materials 25 (2007) 35–47. [14] M.M. Joshi, C.R. Dayanandan, A.G.S. Pillai, B.R. Gandhe, J.S. Karir, Studies on the effects of RDX particle size on the burning rate of gun propellants, Journal of Defense Science 48 (1998) 317–321. [15] C.L.Z. Pei, Q. Luo, G. Yang, F. Nie, H. Huang, Preparation, Characterization of nano-structured HMX, Journal of Nanoscience 11 (2006) 234. [16] S. Dowy, A. Braeuer, K. Reinhold-Lopez, A. Leipertz, Laser analyses of mixture formation and the influence of solute on particle precipitation in the SAS process, Journal of Supercritical Fluids 50 (2009) 265–275. [17] F. Mattea, A. Martin, A. Matias-Gago, M.J. Cocero, Supercritical antisolvent precipitation from an emulsion: -carotene nanoparticle formation, Journal of Supercritical Fluids 51 (2009) 238–247. [18] X. Zhao, Y. Zu, Q. Li, M. Wang, B. Zu, X. Zhang, R. Jiang, C. Zu, Preparation and characterization of camptothecin powder micronized by a supercritical antisolvent (SAS) process, Journal of Supercritical Fluids 51 (2010) 412–419. [19] B.-M. Lee, D.S. Kim, Y.-H. Lee, B.-C. Lee, H.-S. Kim, H. Kim, Y.-W. Lee, Preparation of submicron-sized RDX particles by rapid expansion of solution using compressed liquid dimethyl ether, Journal of Supercritical Fluids 57 (2011) 251–258.
254
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
[20] Y.-M. Chen, P.-C. Lin, M. Tang, Y.-P. Chen, Solid solubility of antilipemic agents and micronization of gemfibrozil in supercritical carbon dioxide, Journal of Supercritical Fluids 52 (2010) 175–182. [21] C. Zhao, L. Wang, Y. Zu, C. Li, S. Liu, L. Yang, X. Zhao, B. Zu, Micronization of Ginkgo biloba extract using supercritical antisolvent process, Powder Technology 209 (2011) 73–80. [22] A. Braeuer, S. Dowy, E. Torino, M. Rossmann, S.K. Luther, E. Schlueckerd, A. Leipertz, E. Reverchon, Analysis of the supercritical antisolvent mechanisms governing particles precipitation and morphology by in situ laser scattering techniques, Chemical Engineering Journal 173 (2011) 258–266. [23] I. De Marco, E. Reverchon, Influence of pressure, temperature and concentration on the mechanisms of particle precipitation in supercritical antisolvent micronization, Journal of Supercritical Fluids 58 (2011) 295–302. [24] G. Taguchi, Systems of Experimental Design, vols. 1–2, Kraus, New York, 1987. [25] R.K. Roy, A Primer on the Taguchi Method, Van Nostrand Reinhold, NewYork, 1990. [26] S.M. Pourmortazavi, S.S. Hajimirsadeghi, M. Rahimi-Nasrabadi, Statistical optimization of condition for synthesis lead sulfide nanoparticles, Materials and Manufacturing Processes 24 (2009) 524–528. [27] P.J. Ross, Taguchi Techniques for Quality Engineering, McGraw-Hill, New York, 1988. [28] S.G. Hosseini, S.M. Pourmortazavi, M. Fathollahi, Orthogonal array design for the optimization of silver recovery from waste photographic paper, Separation Science and Technology 39 (2004) 1953–1963. [29] S.M. Pourmortazavi, S.S. Hajimirsadeghi, I. Kohsari, S.G. Hosseini, Orthogonal array design for the optimization of supercritical carbon dioxide extraction of
[30]
[31] [32]
[33]
[34]
[35] [36]
[37]
different metals from a solid matrix with cyanex 301 as a ligand, Journal of Chemical Engineering Data 49 (2004) 1530–1534. Y. Bayat, M. Eghdamtalab, V. Zeynali, Control of the particle size of submicron HMX explosive by spraying in non-solvent, Journal of Energetic Materials 28 (2010) 273–284. G.P. Sutton, O. Biblarz, Rocket Propulsion Elements, seventh edition, 2000, pp. 503–505. Z. Shuhai, Z. Jiglin, W. Jingyu, Preparaion, Properties of sub-micron nitramine explosives, Fraunhofer-Institut fur Chemische Technologie (ICT) 100 (2004) 1–4. S.M. Pourmortazavi, S.S. Hajimirsadeghi, Application of supercritical carbon dioxide in energetic materials processes: a review, Industrial Engineering Chemistry Research 44 (2005) 6523–6533. U. Teipel, H. Krober, H. Krause, Formation of energetic materials using supercritical fluids, Propellants, Explosives, Pyrotechnics 26 (2001) 168–173. C. Carino, Structural layout assessment by orthogonal array based simulation, Mechanic Research Communication 33 (2006) 292–301. Y. Bayat, S.S. Hajimirsadeghi, S.M. Pourmortazavi, Statistical optimization of reaction parameters for the synthesis of 2,4,6,8,10,12-hexanitro-2,4,6,8,10,12hexaaza isowurtzitane, Organic Process Research and Development 15 (2011) 810–816. P. Sharma, A. Vermab, R.K. Sidhub, O.P. Pandey, Effect of processing parameters on the magnetic properties of strontium ferrite sintered magnets using Taguchi orthogonal array design, Journal of Magnetic and Magnetism Materials 307 (2006) 157–164.
Contents lists available at SciVerse ScienceDirect
The Journal of Supercritical Fluids journal homepage: www.elsevier.com/locate/supflu
Statistical optimization of supercritical carbon dioxide antisolvent process for preparation of HMX nanoparticles Yadollah Bayat a , Seied Mahdi Pourmortazavi b,∗ , Hatef Iravani a , Hamideh Ahadi a a b
Department of Chemistry and Chemical Engineering, Malek Ashtar University of Technology, Tehran, Iran Faculty of Material and Manufacturing Technologies, Malek Ashtar University of Technology, Tehran, Iran
a r t i c l e
i n f o
Article history: Received 29 February 2012 Received in revised form 24 September 2012 Accepted 26 September 2012 Keywords: HMX Micronization SAS process Energetic nanoparticles ANOVA Taguchi robust design Statistical optimization
a b s t r a c t Taguchi robust design was employed as a statistical optimization method to optimize parameters of SAS process in order to prepare HMX nanoparticles. The current application of the Taguchi statistical optimization was highly efficient in optimizing of experimental variables which are effective in HMX particle size produced by the SAS technique. The effect of operation conditions such as: pressure, temperature, HMX concentration, solution flow rate, solvent identification, and flow rate of CO2 on the size of micronized HMX particles under various levels was investigated. The effects of these variables on the particle size of produced HMX were quantitatively evaluated by the analysis of variance (ANOVA). The results showed that the size of HMX particles prepared by SAS technique could be tuned effectively by controlling the main parameters under their optimum level. Finally, the optimum conditions for preparation of HMX nanoparticles via SAS method were proposed. The results of ANOVA showed that 3.5 mol/l HMX concentration, 3 ml/min solution flow rate, cyclohexanone as solvent, and 70 ml/min flow rate of CO2 are optimum conditions for producing HMX nanoparticles. Experimental observations showed that under optimum conditions of SAS process, the particle size of produced HMX is about 56 nm. © 2012 Elsevier B.V. All rights reserved.
1. Introduction HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine) is a high explosive which is widely used in various formulations of plastic bonded explosives (PBX), double base propellants and composite propellants due to its high calorific potential, high density and smokeless combustion products [1,2]. HMX could be prepared in four different polymorphs that commonly known as ␣ (orthorhombic),  (monoclinic), ␥ (monoclinic) and ␦ (hexagonal) phases. Among these various phases, -HMX has the most stability one and possesses the highest explosion power which arises from its crystal phase and its high density. Therefore, preparation of HMX in -phase would be exclusively interested [3]. The size and shape of energetic material particles have a significant influence on the different properties of propellants and explosive formulations [4–6]. Several properties of these compounds such as sensitivity, mechanical properties, content of total solid, density, homogeneity of formulation, and burning rate are dependent to the particle size of energetic materials [7,8]. The particle size reduction of energetic compounds could be lead to the reduction defects, voids and inclusions of crystal and hence
∗ Corresponding author at: P.O. Box 16765-3454, Tehran, Iran. Tel.: +98 2122952285; fax: +98 2122936578. E-mail address: [email protected] (S.M. Pourmortazavi). 0896-8446/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.supflu.2012.09.010
reduction the sensitivity of the formulations and enhancing their safety [9,10]. In order to tailoring fine particles, various conventional methods including milling, liquid crystallization and spray drying could be employed [11–13]. But, these techniques are not suitable for treating energetic materials due to some drawbacks such as difficulty of controlling particle size and particle size distribution, high residual solvent, high shear forces, and unexpected problems during their size reduction process containing thermal explosion of energetic materials [14,15]. Supercritical fluids (SCFs) techniques may be useful to overcome these drawbacks of classical micronization processes and have been applied by researchers to size reduction of explosives [16]. The supercritical fluid is characterized with diffusivities considerably higher than those of liquid, which make it suitable for fast supersaturation of solute and then precipitation of small particles of targeting material. Until today, various techniques for particle formation via supercritical fluid technology including two general processes have been used: supercritical antisolvent (SAS) process [17,18] and rapid expansion from supercritical solutions (RESS) process [19,20]. The SCFs based techniques have many advantages rather than classical micronization processes such as preventing thermal degradation of the target compounds, no mechanical damage, and no residual solvent problem [21]. The main work for an explosive particle formation via SCFs processes such SAS recrystallization and RESS process is preparation of fine particles with various sizes and shapes
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
249
Fig. 1. (a) Schematic of the experimental setup used for SAS process and (b) and the cross section of used nozzle. Taken from Ref. [18] with permission.
by controlling the supersaturation and nucleation rates of the explosive in the SCF media using variation in solvent strength. Especially, the SAS process as a recrystallization technique is a promising process for production of solvent-free explosive particles with fine sizes [22]. The particle size of compounds which prepared by the aid of SAS process is dependent on the various operation parameters and procedure conditions; thus, optimization is an important step in developing SAS technique for size reduction of various materials [23]. Generally, two systematic procedures including simultaneous and sequential methods are used in optimization of the experiments [24]. The main disadvantages of sequential methods are slow convergence on complex response surface and difficulty in dealing with response surface with high dimensionality; while, simultaneous optimization techniques do not suffer from such problems [25]. On the other hand, an obvious disadvantage of the full factorial designs is the number of experimental trials required which increases geometrically by increasing number of variables. Fortunately, this required number of experimental trials is minimized by use of fractional factorial experiments, such as orthogonal array (OA) designs [26]. Since statistical experimental design methods could be carefully explored the experimental space while studying several variables via a small number of observations; Taguchi robust design as a statistical experiment optimization has been widely used in various procedures optimization [27–29]. Therefore, various control factors could be simultaneously investigated and optimized by the aid of Taguchi statistical design. The aim of this study was applying the Taguchi robust design to the optimization of various parameters affecting on particle size of HMX prepared via SAS process and evaluating the effect of these parameters on the size of produced HMX particles. Our goal in this study was optimization of conditions to reduce the size of produced HMX particles and preparation of HMX nanoparticles. To the best of our knowledge, various reports could be found on the micronization of HMX by various techniques [1,30–34]; but, until today there is no report on the preparation of HMX nanoparticles via SAS process. Therefore, optimization of SAS process parameters affecting the size of HMX particles and preparation nanoparticles of this explosive is interested. 2. Experimental Cyclotetramethylene tetranitramine (Octogen, HMX) was used to investigate the possibility for formation of its nanoparticles using compressed SAS process. It should be noted that HMX is a high explosive. Thus, during its operation much care is required to
prevent unexpected hazard such as accidental explosion. Fortunately, in the SAS process which used in this study, the solid of explosive is dissolved in a liquid, and then a supercritical fluid, which is miscible with the liquid but in lower solvent power to the explosive, is added to recrystallize the HMX. This wet operation process is safer for some difficult-to-handle high explosive such as HMX. Analytical grade acetone, cyclohexanone, and ␥-butyrolactone as solvents were used as received from Merck Company (Germany). Carbon dioxide (99.99% purity), contained in a cylinder with an eductor tube, was obtained from Sabalan Co. (Tehran, Iran). The scheme of the used SAS apparatus is shown in Fig. 1. This SAS system is consisting of a precipitation chamber and a gas–liquid separation chamber. As could be seen in Fig. 1a, the CO2 is cooled by a cooler (4) before being compressed via a liquid pump (8) and the pressure is controlled with a backpressure-regulating valve. After pre-heating of CO2 in a heat exchanger (13), CO2 enters to the precipitation chamber (18). Simultaneously, the solution containing target compound is pumped, heated and fed to the precipitation chamber (1000 ml) through a stainless steel nozzle (16). Meanwhile, Fig. 1b shows a close up with the details of the nozzle, which is a laser-drilled orifice (150 m inner diameter). This nozzle is placed at the top of the precipitation chamber; which is located in a distinct inlet point from the CO2 . A stainless steel filter (17) with 200 nm pore sizes was put into the precipitation chamber to collect the micronized particles and to let the SC-CO2 /organic solvent mixture pass through. The flow rate of the mixture that leaves the precipitator is controlled via valve (21), which is located between the precipitation chamber and the gas–liquid separation chamber (22), where the mixture suffers a decompression (pressure <5 MPa) to induce separation of the CO2 from the organic solvent. The prepared HMX samples via SAS process under various experimental conditions were characterized by scanning electron microscopy (SEM). Scanning electron micrographs were recorded using on a Philips XL30 series instrument using a gold film for loading the dried particles on the instrument. Gold films were prepared by a sputter coater model SCD005 made by BAL-TEC (Switzerland). To optimize experimental conditions of SAS process, for the micronization of HMX, an experimental design approach was followed. Various operation parameters including pressure, temperature, HMX concentration, solution flow rate, nature of the solvent, and flow rate of CO2 were investigated in three levels using a L18 array proposed by Taguchi robust design as shown in Table 1. As could be seen in this table, each of variables was studied under three different levels. These levels were selected based on the
250
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
Table 1 Assignment of the factors and levels of the experiments by using an OA18 (36 ) matrix and average particle size of produced HMX as response. Average particle size (nm)
Flow rate of CO2 (ml/min)
Solvent nature
Solution flow rate (ml/min)
HMX concentration (ml/l)
Temperature (◦ C)
Pressure (bar)
Trial no.
421 330 196 230 410 428 265 307 265 89 185 671 180 224 160 285 195 278
40 70 100 100 40 70 70 100 40 70 100 40 100 40 70 40 70 100
Acetone Cyclohexanone ␥-Butyrolactone Cyclohexanone ␥-Butyrolactone Acetone ␥-Butyrolactone Acetone Cyclohexanone Cyclohexanone ␥-Butyrolactone Acetone Acetone Cyclohexanone ␥-Butyrolactone ␥-Butyrolactone Acetone Cyclohexanone
1.5 2.5 3.5 1.5 2.5 3.5 2.5 3.5 1.5 3.5 1.5 2.5 2.5 3.5 1.5 3.5 1.5 2.5
1 2 3 2 3 1 1 2 3 3 1 2 3 1 2 2 3 1
35 60 85 35 60 85 35 60 85 35 60 85 35 60 85 35 60 85
80 80 80 180 180 180 280 280 280 80 80 80 180 180 180 280 280 280
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
previous studies [32–34] and the experimental conditions which were accessible. 3. Results and discussion Simultaneous optimization techniques, such as Taguchi robust design are included a previously planned array experiments; then, collection the results and determination of the optimum condition via constructing a response surface or by retention mapping [35]. In orthogonal array designs, the designed arrays are used to assigning some factors to a series of experiment combinations whose results could then be analyzed via a common mathematical procedure [36]. Meanwhile, the significant effects of each factor are independently determined. The main performing of the experiments via orthogonal arrays is possibility for the separation of different effects [37]. 3.1. Optimization of parameters for particle size reduction of HMX by SAS procedure In this research, HMX explosive was micronized by SAS procedure. HMX was dissolved in various solvents and micronization process was performed via SAS technique by the supercritical carbon dioxide under different conditions of operation in accordance with Table 1. The purpose of this study was to investigate the effect of various procedure parameters on the size of produced HMX particles. The factors included in this study were pressure, temperature, HMX concentration, solution flow rate, nature of solvent, and flow rate of CO2 . Factors and their tested levels are shown in Table 1. Fig. 2 shows the SEM images of four HMX samples obtained by SAS procedure under various conditions of Table 1. Also, the average size of produced HMX particles at each experiment is given in this table as the response. In Taguchi robust design, analysis of the collected data while there is no interaction between variables three steps includes following steps: determination of the optimum condition for the studied procedure; assignment the individual effect of each parameter on the result which in our study is the average size of prepared HMX particles; and finally, prediction the process performance under the proposed optimum conditions. The averages values correspond to effect of the each parameter under any level were computed according to assignment of the designed experiments. Therefore, calculation for determining the influence of each parameter at any investigated level on the size of HMX particles (prepared via SAS procedure) was carried out by pooling the results of all
experiments in which a factor was set at same level and thus dividing by the number of trials used this level. As an example, the average effect of supercritical fluid pressure at level 1 (80 bar) was determined by pooling the average particle size of HMX obtained via all experiments in which pressure of the supercritical fluid was set at level 1 (trials 1, 2, 3, 10, 11, and 12; while here the number of experiments is six) as shown in Table 1. On the other hand, Fig. 2 shows the curves correspond to the variation in the particle size of HMX due to changes in the level of any factor. These graphs show variations in average particle size of HMX due to the effect of each variable when the level of the variable is changed. Table 2 presents the results for analysis of variance (ANOVA) for the obtained data (particle size of produced HMX via SAS process). In this study, the effects of various parameters including pressure, temperature, HMX concentration, solution flow rate, solvent identification, and flow rate of CO2 on the average particle size of produced HMX by SAS technique at three different levels were investigated. The results of ANOVA for HMX micronization experiments via SAS confirm that (at 90% confidence level) except pressure and temperature, all other variables have significant effects on the size of HMX particles produced via SAS process. In this work, there is no interaction between the variables. It was found that pressure and temperature are not significant parameters for the controlling size of HMX particles during its micronization via SAS method. Also, the effect of HMX concentration and its solution feeding flow rate on the size of HMX particles was studied. Three different HMX concentrations (1, 2, and 3 mol/l) and solution feeding flow rates (1.5, 2.5, and 3.5 ml/min) were investigated and our finding showed that both parameters have significant effects in tuning the size of produced HMX particles (Fig. 3). Other investigated parameters for determining their significance on the size of HMX particles were nature of solvent, and flow rate of CO2 . The choice of solvent is known to affect particle size and morphology of produced particles in SAS technique [16]. In SAS process as a micronization technique, the target compound should be dissolved in the solvent to form a homogeny solution. In ideal conditions, the target is completely insoluble in the produced mixture of antisolvent/solvent (SC-CO2 /liquid solvent) during the SAS process; while the solvent is totally miscible with the antisolvent (SC-CO2 ). Therefore, injection of the solution into the SC-CO2 causes a high supersaturation of the injected solution and hence, nucleation of the target solute happened. Therefore, the nature of the solvent via its solvating power, interaction of solvent–solute and its miscibility with SCCO2 could be effective in the morphology of the formed particles
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
251
Fig. 2. SEM images correspond to HMX particles obtained by different runs of OA18 (Table 1) via SAS process: (a) run 4, (b) run 6, and (c) run 9.
via SAS process. In this study, three different solvents containing acetone, cyclohexane and ␥-butirolactone were used for dissolving of HMX. The results of ANOVA showed that the nature of used solvent is a significant parameter for the controlling the particle size of produced HMX. Also, significance of CO2 flow rate under three levels of 40, 70, and 100 ml/min was studied for its effect on the particle size of product. Our finding showed that CO2 flow rate is a significant parameter for the controlling the size of HMX particles. By considering the results obtained by ANOVA it could be seen that 3.5 ml/min is the optimum flow rate for HMX solution during its micronization via SAS process. Also, by increasing concentration of the HMX in the initial solution from 1 to 2%, increases the size of precipitated HMX particles; while, enhancing concentration from 2 to 3% causes a significant decrease in the size of prepared HMX particles. Beyond a certain point, as the solution concentration is increased, nucleation process dominates particle growth, which results in larger particles. Thus, 3% was selected as optimum concentration for HMX. Among various solvents investigated cyclohexanone showed the best performance in size reduction of HMX by SAS technique;
because this solvent enhances dissolution and miscibility of HMX solution in SC-CO2 . On the other hand, 70 ml/min was chosen as optimum CO2 flow rate during SAS process for production nanosized HMX. 3.2. Preparation of HMX nanoparticles under optimum condition Under the optimized conditions which proposed by analysis of the data obtained via designed OA18 matrix for optimization of variables, the particle size of HMX recrystallized via SAS process could be predicted according to the following expression [36,37]: Yopt =
T T + Fsol − N N +
FCO2 −
T N
+ CHMX −
T N
+ SNature −
T N
where T/N is average particle size of HMX + contribution of Fsol , CHMX , SNature and FCO2 in their minimum values which calculated from average effect of each factor; while, T is the grand total of
Table 2 ANOVA results for micronization procedure of HMX via SAS procedure using OA18 (36 ) matrix. Factor
DOFa
Sb
Vc
DOF
S Pooledf
F d
P e
Pressure (bar) Temperature (◦ C) HMX concentration (ml/l) Solution flow rate (ml/min) Solvent nature Flow rate of CO2 (ml/min) Error
2 2 2 2 2 2 5
16792.2 10097.4 35152.4 60408.7 35883.5 99780.1 25535.9
8396.1 5048.7 17576.2 30204.4 17941.7 49890.1 5107.2
– – 2 2 2 2 9
– – 23502.3 48758.6 24233.3 88130.0 52425.5
– – 3.02 5.18 3.08 8.56 –
– – 8.29 17.19 8.54 31.07 34.91
a b c d e f
Degree of freedom for each variable; DOF is degree of freedom for each variable after pooling of insignificant parameters. Standard deviation (mean squares) for each parameter; S is pure standard deviation for each parameter after pooling of insignificant parameters. Mean square (variance) value for each parameter; V is the variance value for each parameter after pooling of insignificant parameters. The variance ratio or F statistic (the ratio of variance due to the effect of a parameter and variance value due to the error term). Percent contribution of each parameter or error term in the results of process. The critical value was at 90% confidence level; pooled error results from pooling insignificant parameters [25].
252
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
Fig. 3. Average particle size of HMX obtained via SAS process correspond to the effect of each level for various factors: (a) pressure; (b) temperature; (c) concentration of HMX; (d) flow rate of solution; (e) nature of solvent, and (f) CO2 flow rate.
particle size for all runs, N is total number of results, Yopt is particle size of HMX under optimum condition, Fsol , CHMX , SNature and FCO2 are average particle size of HMX at optimum levels of HMX feed solution, concentration of HMX in the solution, nature of the solvent and flow rate of supercritical carbon dioxide, respectively. On the other hand, the confidence interval (C.I.) for the predicted
Fig. 4. SEM image of prepared HMX under optimum condition.
particle size of HMX under the optimum conditions could be calculated from the following expression [36]:
CI = ±
F˛ (f1 , f2 ) · Ve Ne
Fig. 5. Size distribution for HMX nanoparticles prepared under optimum condition of SAS process.
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
253
Fig. 6. (a) XRD pattern for the prepared HMX nanoparticles obtained by SAS process under optimum conditions and (b) standard XRD pattern for  form of HMX obtained from PC-APD, diffraction software.
where F˛ (f1 , f2 ) is the critical value for F that could be obtained from the F table at the degrees of freedom (DOF), f1 and f2 at the level of significance ˛ (in this work, ˛ = 90%), f1 = DOF of mean (which always equals 1), f2 = DOF of the pooled error term and Ne could be calculated via the following equation: Ne =
Number of trials DOF of mean(always1) + DOF of all factors used in estimation
Computations for estimation the particle size of HMX under optimum condition and CI for this predicted particle size show that at 90% confidence level, the size of HMX particles under optimum conditions of SAS process will be about 75 ± 20 nm. In this investigation, the effects of various parameters including pressure, temperature, HMX concentration, solution flow rate, nature of solvent, and flow rate of CO2 on the particle size of HMX produced via SAS process at three different levels were investigated and our finding showed that 3% HMX concentration, 3.5 ml/min solution flow rate, cyclohexanone as solvent, and 70 ml/min as flow rate of CO2 are optimum condition for preparation of HMX nanoparticles. Meanwhile, ANOVA results for these experiments indicated that (at 90% confidence level) pressure and temperature of the supercritical carbon dioxide in SAS process have no significant effects on the size of produced HMX particles. In the next step of this study, HMX nanoparticles were prepared under optimum conditions proposed by ANOVA. Fig. 4 shows the SEM image for prepared HMX nanoparticles under optimum conditions of SAS process. Meanwhile, Figs. 5 and 6 show the size distribution graph for HMX particles obtained under these optimum conditions. As could be seen in this graph, the average particle size of HMX particles is about 56 nm.
4. Conclusion In this study, HMX nanoparticles were successfully prepared via SAS process. Taguchi robust design method was applied to optimize the operation conditions of HMX micronization by this method with aim of preparation HMX nanoparticles. Several parameters affecting on the size of HMX particles during SAS procedure were investigated and optimized. Our result showed that pressure and temperature of the supercritical carbon dioxide have no significant effects on the particle size of produced HMX during SAS process; while, other studied variables including HMX concentration, solution flow rate, solvent identification, and flow rate of CO2 have a considerable role in determination of product particles in this
procedure. Finally, under optimized conditions of SAS process, HMX nanoparticles with average particle size of 56 nm were prepared. References [1] Z. Yongxu, L. Dabin, L. Chunxu, Preparation and characterization of reticular nano-HMX, Propellants, Explosives, Pyrotechnics 30 (2005) 438–441. [2] M. Fathollahi, S.M. Pourmortazavi, S.G. Hosseini, Particle size effects on thermal decomposition of energetic material, Journal of Energetic Materials 26 (2008) 52–69. [3] M. Herrmann, W. Engel, N. Eisenreich, Phase transitions of HMX and their significance for the sensitivity of explosives, in: Proceedings of the Technical Meeting of Specialists MWDDEA AF-71-F-G-7304-Physics of Explosives, 1990. [4] U. Teipel, H. Krober, H. Krause, Formation of energetic materials using supercritical fluids, Journal of Supercritical Fluids 26 (2001) 168–173. [5] Y. Bayat, V. Zeynali, Preparation, Characterization of nano-CL-20 explosive, Journal of Energetic Materials 29 (2011) 281–291. [6] A. Okmar, R.P. Nafday, L.W. Brandon, H. Janson, Patterning high explosive at the nanoscale, Propellants, Explosives, Pyrotechnics 31 (2006) 376. [7] R. Sivabalan, G.M. Gore, U.R. Nair, A. Saikia, S. Venugopalan, B.R. Gandhe, Study on ultrasound assisted precipitation of CL-20 and its effect on morphology and sensitivity, Journal of Hazardous Materials 139 (2007) 199–203. [8] B. Wang, L.L.M. Yu-cun, Study on the influence of particle size on the impact sensitivity of HMX, Journal of North China Institute of Technology 26 (2005) 35. [9] X. Song, Y. Wang, C. An, X. Guo, F. Li, Dependence of particle morphology and size on the mechanical sensitivity and thermal stability of HMX, Journal of Hazardous Materials 159 (2008) 222–229. [10] Z. Shuhai, Z. Jinglin, W. Jingyu, Preparation and properties of sub-micron nitramine explosives, Fraunhofer-Institut fur Chemische Technologie (ICT) (2004) 100–104. [11] S. Jain, S. Nandagopal, P.P. Singh, K.K. Radhakrishnan, B. Bhattacharya, Size and shape of AP and their influence on properties of composite propellant, Journal of Defense Science 59 (2009) 294–299. [12] S.M. Pourmortazavi, M. Rahimi-Nasrabadi, A.A. Davoudi-Dehaghani, A. Javidan, M.M. Zahedi, S.S. Hajimirsadeghi, Statistical optimization of experimental parameters for synthesis of manganese carbonate and manganese oxide nanoparticles, Materials Research Bulletin 47 (2012) 1045–1050. [13] G. Yang, F. Nie, T. Li, Q. Guo, Z. Qiao, Preparation and characterization of nanoNTO explosive, Journal of Energetic Materials 25 (2007) 35–47. [14] M.M. Joshi, C.R. Dayanandan, A.G.S. Pillai, B.R. Gandhe, J.S. Karir, Studies on the effects of RDX particle size on the burning rate of gun propellants, Journal of Defense Science 48 (1998) 317–321. [15] C.L.Z. Pei, Q. Luo, G. Yang, F. Nie, H. Huang, Preparation, Characterization of nano-structured HMX, Journal of Nanoscience 11 (2006) 234. [16] S. Dowy, A. Braeuer, K. Reinhold-Lopez, A. Leipertz, Laser analyses of mixture formation and the influence of solute on particle precipitation in the SAS process, Journal of Supercritical Fluids 50 (2009) 265–275. [17] F. Mattea, A. Martin, A. Matias-Gago, M.J. Cocero, Supercritical antisolvent precipitation from an emulsion: -carotene nanoparticle formation, Journal of Supercritical Fluids 51 (2009) 238–247. [18] X. Zhao, Y. Zu, Q. Li, M. Wang, B. Zu, X. Zhang, R. Jiang, C. Zu, Preparation and characterization of camptothecin powder micronized by a supercritical antisolvent (SAS) process, Journal of Supercritical Fluids 51 (2010) 412–419. [19] B.-M. Lee, D.S. Kim, Y.-H. Lee, B.-C. Lee, H.-S. Kim, H. Kim, Y.-W. Lee, Preparation of submicron-sized RDX particles by rapid expansion of solution using compressed liquid dimethyl ether, Journal of Supercritical Fluids 57 (2011) 251–258.
254
Y. Bayat et al. / J. of Supercritical Fluids 72 (2012) 248–254
[20] Y.-M. Chen, P.-C. Lin, M. Tang, Y.-P. Chen, Solid solubility of antilipemic agents and micronization of gemfibrozil in supercritical carbon dioxide, Journal of Supercritical Fluids 52 (2010) 175–182. [21] C. Zhao, L. Wang, Y. Zu, C. Li, S. Liu, L. Yang, X. Zhao, B. Zu, Micronization of Ginkgo biloba extract using supercritical antisolvent process, Powder Technology 209 (2011) 73–80. [22] A. Braeuer, S. Dowy, E. Torino, M. Rossmann, S.K. Luther, E. Schlueckerd, A. Leipertz, E. Reverchon, Analysis of the supercritical antisolvent mechanisms governing particles precipitation and morphology by in situ laser scattering techniques, Chemical Engineering Journal 173 (2011) 258–266. [23] I. De Marco, E. Reverchon, Influence of pressure, temperature and concentration on the mechanisms of particle precipitation in supercritical antisolvent micronization, Journal of Supercritical Fluids 58 (2011) 295–302. [24] G. Taguchi, Systems of Experimental Design, vols. 1–2, Kraus, New York, 1987. [25] R.K. Roy, A Primer on the Taguchi Method, Van Nostrand Reinhold, NewYork, 1990. [26] S.M. Pourmortazavi, S.S. Hajimirsadeghi, M. Rahimi-Nasrabadi, Statistical optimization of condition for synthesis lead sulfide nanoparticles, Materials and Manufacturing Processes 24 (2009) 524–528. [27] P.J. Ross, Taguchi Techniques for Quality Engineering, McGraw-Hill, New York, 1988. [28] S.G. Hosseini, S.M. Pourmortazavi, M. Fathollahi, Orthogonal array design for the optimization of silver recovery from waste photographic paper, Separation Science and Technology 39 (2004) 1953–1963. [29] S.M. Pourmortazavi, S.S. Hajimirsadeghi, I. Kohsari, S.G. Hosseini, Orthogonal array design for the optimization of supercritical carbon dioxide extraction of
[30]
[31] [32]
[33]
[34]
[35] [36]
[37]
different metals from a solid matrix with cyanex 301 as a ligand, Journal of Chemical Engineering Data 49 (2004) 1530–1534. Y. Bayat, M. Eghdamtalab, V. Zeynali, Control of the particle size of submicron HMX explosive by spraying in non-solvent, Journal of Energetic Materials 28 (2010) 273–284. G.P. Sutton, O. Biblarz, Rocket Propulsion Elements, seventh edition, 2000, pp. 503–505. Z. Shuhai, Z. Jiglin, W. Jingyu, Preparaion, Properties of sub-micron nitramine explosives, Fraunhofer-Institut fur Chemische Technologie (ICT) 100 (2004) 1–4. S.M. Pourmortazavi, S.S. Hajimirsadeghi, Application of supercritical carbon dioxide in energetic materials processes: a review, Industrial Engineering Chemistry Research 44 (2005) 6523–6533. U. Teipel, H. Krober, H. Krause, Formation of energetic materials using supercritical fluids, Propellants, Explosives, Pyrotechnics 26 (2001) 168–173. C. Carino, Structural layout assessment by orthogonal array based simulation, Mechanic Research Communication 33 (2006) 292–301. Y. Bayat, S.S. Hajimirsadeghi, S.M. Pourmortazavi, Statistical optimization of reaction parameters for the synthesis of 2,4,6,8,10,12-hexanitro-2,4,6,8,10,12hexaaza isowurtzitane, Organic Process Research and Development 15 (2011) 810–816. P. Sharma, A. Vermab, R.K. Sidhub, O.P. Pandey, Effect of processing parameters on the magnetic properties of strontium ferrite sintered magnets using Taguchi orthogonal array design, Journal of Magnetic and Magnetism Materials 307 (2006) 157–164.