You are on page 1of 15

Finite size effects on aluminum/Teflon reaction channels under combustive environment: A RiceRamspergerKasselMarcus and transition state theory study

of fluorination
Martin Losada and Santanu Chaudhuri Citation: J. Chem. Phys. 133, 134305 (2010); doi: 10.1063/1.3480020 View online: http://dx.doi.org/10.1063/1.3480020 View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v133/i13 Published by the American Institute of Physics.

Related Articles
Scaling of light emission from detonating bare Composition B, 2,4,6-trinitrotoluene [C7H5(NO2)3], and PE4 plastic explosive charges J. Appl. Phys. 110, 084905 (2011) Impact-induced initiation and energy release behavior of reactive materials J. Appl. Phys. 110, 074904 (2011) Nitrogen/argon diluted acetylene and ethylene blue flames under infrared CO2 laser irradiation AIP Advances 1, 032142 (2011) Electromagnetically induced localized ignition in secondary high explosives: Experiments and numerical verification J. Appl. Phys. 110, 034902 (2011) Control of noise and exhaust emissions from dual fuel engines J. Renewable Sustainable Energy 3, 043103 (2011)

Additional information on J. Chem. Phys.


Journal Homepage: http://jcp.aip.org/ Journal Information: http://jcp.aip.org/about/about_the_journal Top downloads: http://jcp.aip.org/features/most_downloaded Information for Authors: http://jcp.aip.org/authors

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

THE JOURNAL OF CHEMICAL PHYSICS 133, 134305 2010

Finite size effects on aluminum/Teon reaction channels under combustive environment: A RiceRamspergerKasselMarcus and transition state theory study of uorination
Martin Losada and Santanu Chaudhuria
ISP/Applied Sciences Laboratory, Washington State University, Spokane, Washington 99210-1495, USA

Received 8 November 2009; accepted 28 July 2010; published online 6 October 2010 The effect of particle size on combustion efciency is an important factor in combustion research. Gas-phase aluminum clusters in oxidizing environment constitute a relatively simple and extensively studied system. In an attempt to underscore the correlation between electronic structure, nite size effect, and reactivity in small aluminum clusters, reactions between aluminum, Al13 cluster, and Teon decomposition fragments were studied using theoretical calculations at the density functional theoretical level. The unimolecular rate constants calculated using transition state and RiceRamspergerKasselMarcus theory show that reactions with COF and CF2 species with aluminum are faster than those involving CF3 and COF2. The results show that the kinetic barriers along different exothermic reaction channels correlate with the trends in HOMO R HOMO TS HOMO denotes highest occupied molecular orbital energy gap and related shifts of the HOMO levels of reactants. Overall reactions involving carbonyl uoride species COF and COF2 lead to CO elimination and uorination of the Al cluster. The CF3 / CF2 fragments lead to stable multicenter AlC bond formation on the uorinated Al cluster surface. Temperature-, energy-, and pressure-dependent rate constants are provided for extrapolating the expected reaction kinetics to conditions similar to known combustion reactions. 2010 American Institute of Physics. doi:10.1063/1.3480020
I. INTRODUCTION

There is considerable interest in the structure and reactivity of nanoscale metal clusters as a building block for energetic materials and propellants. Extremely small clusters, particularly those with a closed shell electronic structure, are relatively stable and are well suited for exploring gas-phase reaction chemistry of small metal clusters. These reactions also provide insight into the condensed phase effects of solid propellants. For example, nanocomposites consisting of metal nanoparticles in an oxidizing medium can be engineered to provide much higher propagation rates and ignition sensitivity than their microscale counterparts.1 One particularly interesting reactive material is composed of Al nanoparticles embedded in a Teon oxidizer. Improved combustion efciency is achieved using aluminum nanoparticles in place of micron scale particles in these thermites.2 Furthermore, it has been shown experimentally3 that the combustion behavior of highly energetic thermites is strongly dependent on aluminum particle size; this indicates that the incorporation of nanoaluminum particles 1 signicantly reduces ignition temperatures and 2 produces a unique reaction behavior that can be attributed to a different chemical kinetic mechanism than observed with microaluminum particles. Propagation of a self-sustaining reaction depends on many factors, particularly particle diameter, reactive medium, rate of individual chemical reactions, surface
a

Author to whom correspondence should be addressed. Electronic mail: chaudhuri@wsu.edu.

catalytic effects, bulk diffusion of reactants inside the particle, pressure, etc.4 In order to accurately describe and predict combustion processes, it is imperative that we understand the role of these factors. Because it is sometimes difcult to ascribe a particular observation to a distinct factor and/or property in a combustion process, a rst-principles approach to nanoscale reactivity is important. Recently, we published the results of a preliminary study to elucidate the fundamentals of decomposition reactions between aluminum atoms and Teon fragments.5 This work follows up on the previous study5 by investigating the reactions of the stable Al13 cluster as a model system to shed light on the issues of cluster size versus reactivity discussed above. The smallest cluster that can serve as a model for understanding nite size effects in these systems is Al13. Experimental studies of the ionization potentials,6 dissociation,7,8 and reactivity9 of aluminum clusters have identied both neutral Al13 and anionic Al13 as particularly stable structures; they have been supported by theoretical work.10 Their stability has been attributed to the fact that Al13 has 39 valence electrons and therefore is one electron short of being closed shell in the jellium model.11 In fact, the Al13 anion, with 40 valence electrons, is a closed shell and is more stable than Al13 by 68.0 kcal/ mol.12 Hettich13 reported the corresponding mass spectra for Aln anions with n = 3 50. The study showed that clusters with odd-numbered atoms were more intense than those with an even-number of atoms, with Al13 observed to be the most abundant. The Al13 system is the smallest cluster in which an Al atom can have a coordination number of 12 of a close-packed bulk
2010 American Institute of Physics

0021-9606/2010/133 13 /134305/14/$30.00

133, 134305-1

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-2

M. Losada and S. Chaudhuri

J. Chem. Phys. 133, 134305 2010

crystal and the surface atoms mimic the triangular Al arrangement on an Al 111 surface termination. Thus, Al13 is a reasonably well understood system with experimental validation of ground and excited state structures. Systematic experimental investigation of kinetics in energetic material is invaluable in developing validated models for initiation condition, oxidizer-metal ratio, and other condensed matter effects beyond particle size. Recently, Dlott and co-workers14 investigated chemical reactions initiated by ash-heating of a nanoenergetic material consisting of Al nanoparticle fuel in a TeflonAF oxidizer. They used ultrafast midinfrared spectroscopy to monitor the IR absorption transients of vibrations in CF3, CF2, and COF. They found that reactions of Al with COF are more than ten times faster than reactions with CF3 or CF2. This led them to propose that Al-TeflonAF chemistry can be described by two processes: a slower one involving Al+ CF3 or CF2 and a faster process involving Al+ COF. The formation of AlF3 as a nal reaction product of aluminum and Teon is calculated to be highly exothermic 258 kcal/mol .5 Thus, in principle, aluminum makes an excellent high-energy fuel and is of considerable interest in understanding the kinetics of aluminum oxidation, especially in Al nanoparticles. To understand the fundamental process that leads to the initiation of a self-sustaining combustion reaction for metal particles in an oxidizing atmosphere, the intermediate products of individual reaction steps on the metal surface must be identied. Because of varying surface properties, however, it is difcult to study local microscopic reactivity on macroscopic metal surfaces. Aluminum clusters, on the other hand, are well-dened molecular model compounds and are well suited to such investigations. In our previous investigation of the elementary steps in Al-Teon fragment reactions,5 we found that carbonyl uoride COF2 and COF species generally react faster than uoromethane CF3 or uoromethylene CF2 species. We also found that Teon fragment addition, followed by unimolecular elimination of AlF species, is predicted to be faster than reactions proceeding through direct abstraction of atomic uorine. In addition, the kinetic model used in the investigation predicted CF3 / CF2 + Al and COF2 / COF+ Al to be competitive under the same thermal conditions. However, a detailed analysis of the elementary steps also requires an understanding of size effects and surface mediated processes. In this respect, we give a full account of density functional theory DFT electronic structure calculations of the reaction of Al13 clusters with Teon fragments CF3, CF2, COF2, and COF . All reactions initially produce the Al13 CF3, Al13 CF2, Al13 COF2, and Al13 COF complexes. To calculate then effective unimolecular rate constants at different temperature and pressure conditions, we have to analyze on a microcanonical level the interplay of activation and isomerization for the present complexes on a single well and also account for the energy transfer effects. A rigorous way of predicting the kinetics of these complexes is to solve the master equation. Detailed analysis and solutions of the master equation at low pressures are given by Troe.15 Here, we present rate coefcients, using transition state theory, for the rst uorination step upon formation of the initial Al13 -Teon fragment complexes. The calculated

values of the energy barriers and the reactant and transition state vibrational frequencies were taken as the basis for RiceRamspergerKasselMacus RRKM modeling of the pressure dependence of reaction rates. Here, the RRKM Ref. 16 modeling denotes a procedure consisting of the calculation of the microcanonical rate constants, k E , coupling of these k E s with a collision energy transfer model by means of the master equation. Thereafter, pressuredependent rates were obtained by solving the master equation.

II. THEORETICAL METHODS A. Ab initio calculations and unimolecular canonical rate constants

Structures of reactants, products, and transition states were determined using DFT with the PBE0 functional17 and the 6-311G d basis set.18 Vibrational frequencies were determined at the same theoretical level in the harmonic approximation and are provided in Table S1 of the supporting information.19 The relative energies were then obtained by performing single-point calculations of the optimized PBE0/ 6-311G d geometries using the hybrid meta-DFT functional, M05 Ref. 20 with 6-311+ G 3d2f basis set. We paid special attention to spin contamination, as indicated by the deviation of the value of the spin operator S2 from its ideal value for a pure spin state. For doublet states, the value was less than 0.760. All quantum mechanical calculations were performed using the GAUSSIAN 03 suite of programs.21 Temperature-dependent canonical rate constants, k T , at the high-pressure limit can be determined using transition state theory TST as follows: kT = kBT Q# E /k T e 0 B h Q 1

with Q# and Q being the respective partition functions of the transition state and reactants, respectively, and is the reaction path symmetry number. The reaction path degeneracy is set to 1 for each of the RRKM and TST rate constant calculations, as no symmetry constrains were imposed in the optimization of either reactants or transition state structures of the four reaction channels. Highly accurate prediction of transmission coefcients that includes many degrees of freedom is a difcult quantum mechanical problem. A simplifying approximation is to consider tunneling only in the degree of freedom corresponding to the reaction coordinate. Within this one-dimensional formalism, various levels of approximation are available. We rst treat the tunneling effects using the Wigner22 approximation that takes the following form: T =1+ 1 h 24 kBT
2

where is the imaginary frequency at the saddle point. The kBT. In our Wigner correction works well provided that h case, this condition is fullled. Thus, this simplication should not pose a problem, especially in the temperature range of 500900 K considered here.

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-3

RRKM study of Al/Teon reaction

J. Chem. Phys. 133, 134305 2010

A more robust approximation to T has been provided by Skodje and Truhlar.23 In this approximation T = / sin / + exp V , 3

using the RRKM theory16 by solving the master equation. Vibrational frequencies, moments of inertia, and energy barriers from the ab initio calculations were used as input data. For a unimolecular reaction involving a single well, the master equation can be expressed in the following form: gi E = t Pi E,E gi E P E ,E gi E
E0

where = 2 / h , = 1 / kBT, is the imaginary frequency at the saddle point, h is the Planck constant, and V is the zero-point corrected energy difference between the TS structure and the reactants. The tunneling corrections originating from both approximations are listed in the supplementary information19 Table S2 for the rst uorination step. The corrections conrm that the tunneling effect is not signicant at the temperature range considered in this work. The Wigner and the SkodjeTruhlar corrections are very close to each other. For example, for the rst step in the CF3 channel, k T is calculated to be 1.55 1012 exp 20 600/ RT using the Wigner method, compared to 1.535 1012 exp 20 600/ RT using the SkodjeTruhlar method. Thus, we decided to adopt the less expensive Wigner methodology in the calculation of the tunneling corrections for all the rate constants reported here.
B. Calculation of energy and pressure-dependent microcanonical rate constants

ki E gi E ,

The basis for our RRKM Ref. 16 calculations is the expression for energy-dependent unimolecular rate constant, kE = W# E E0 , E h 4

where is the reaction path symmetry number here set to 1 , h is the Planck constant, and E represents the density of states of the reactant, and W# E E0 is the total number of states at the transition state with energy less than or equal to E, where E is the total internal energy available to the system. To evaluate Eq. 4 , we must know a the threshold energy E0 and b the vibrational frequencies of the reactant and transition state species. All of these information was extracted from the present ab initio calculations. W# E E0 and E were evaluated by direct count of the harmonic vibrational states using the QCPE program24 a general RRKM program . For the calculation of Eq. 4 involvement of rotations k E , J , and rotational-vibrational couplings were not taken into account. The pressure-dependent rate constants for the rst uorination reaction from the initial Al13 CF3, Al13 -CF2, Al13 COF2, and Al13 COF clusters were determined

where gi E is the population of energy level E in well i, is the collision frequency of the reactant with the bath gas, Pi E , E is the transition probability for a molecule in well i with energy E to be excited to another vibrational state with energy E via collisions, and k E is the RRKM rate constant. Equation 5 is a simplied version of the master equation obtained by neglecting angular momentum contributions. For the present work, we employed different Lennard-Jones parameters and energy-dependent expressions for E down for each complex. Two different models for the collisional energy transfer function were employed: the exponential-down model25 and the model suggested by Luther and co-workers.26 The energy increment was xed to 10 cm1 in all the sum of states and the density of states calculations. The frequency of collisions was derived from Lennard-Jones parameters. The Lennard-Jones collision diameters, AB, and well depths, AB K , for the initial clusters were estimated from AB = A + B / 2 and AB = A K B K 1/2, where A and A K are for the Al13 and B and B K are for the corresponding CF3, CF2, COF2, and COF moieties in each cluster. The A, 2.35 , and A K , 13 345.1 K parameters for Al13 and the B and B K for CF3 and CF2 are taken from the literature.27 The Lennard-Jones parameters for the COF2 molecule were derived from the critical data and the acentric factor tabulated in the Matheson Gas Data Book.28 Finally, the parameters for the COF molecule were adopted from the values of COF2. Pressure-dependent rate constants were calculated using the MULTIWELL program of Barker.29
III. REACTION CHANNELS

In this study, we attempt to build a minimal reaction set for the uorination of the Al13 cluster in a Teon environment by considering reactions involving CF2, CF3, COF, and COF2 fragments. We do not include the CF species that were the least reactive species in our previous work.5 Thus, the following reaction channels and their elementary steps were investigated:

k1,I

k2,I

k3,I

Al13 CF3 F Al13 CF2 2F Al13 CF 3F Al13 C


k1,II k2,II

I ,

Al13 CF2 F Al13 CF 2F Al13 C


k0,III k1,III

II ,
k2,III k3,III

Al13 + COF2 Al13 COF2 F- Al13 -COF F- Al13 COF A 2F Al13 + CO


k1,IV

III ,

Al13 COF F Al13 + CO

IV .

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-4

M. Losada and S. Chaudhuri

J. Chem. Phys. 133, 134305 2010

FIG. 1. Energy level diagram for reaction channel I. Dotted lines connect the transition states with the reactants and the products in each step. The reaction barriers Ea and reaction enthalpies H are provided for each step. The value in green is for single-point calculation at the M05/ 6-311+ G 3d2f level for the rst uorination step.

The predicted structures for intermediates and transition state species along with their energetics are shown in Fig. 1 for reaction channel I, Fig. 2 for channel II, and Fig. 3 for channels III and IV.
IV. RESULTS AND DISCUSSION

orbitals in the Al13 -Teon adducts as well as analysis of their correlation with the calculated gas-phase kinetics.

A. Reaction mechanisms 1. Channel I

A limited number of experimental data on Al-Teon system have been published. The rationale behind the choice of temperature and pressure range for presenting our calculated rate constant is guided by a few experimental observation. First, combustion reactions between aluminum and Teon fragments require decomposition of Teon into fragments. The products of the thermal degradation of Teon at temperatures ranging from 583 to 793 K are CF2, CF3, C2F4, and COF2.5 On the other hand, Jones et al.,30 using differential scanning calorimetry on the nanoaluminum and Teon, reported two distinct exothermic reactions between 800 and 900 K. They postulated that the rst exotherm was caused by the interaction between Al and Teon and the second was due to the oxidation of Al in air and combustion of Teon. Additionally, in the study of oxidation versus uorination of nano- and micrometer sized aluminum by Watson et al.2 it was found that the use of Teon as an oxidizing agent in aluminum reactions yields higher peak pressures of about 11 MPa. In this section we present the reaction mechanisms and the temperature-, energy-, and pressure-dependent rate constants for the rst uorination step of each reaction channel. The results are augmented with an analysis of the frontier

In the rst step cf. Fig. 1 , oxidation of the Al13 cluster surface proceeds to form the Al13 CF3 complex that is a local minimum energy structure on the potential energy surface PES . The energy gain after the formation of the complex is calculated to be 50 kcal/mol. The reaction is followed by the transfer of a uorine atom, which leads to the formation of the F Al13 CF2 complex. This complex is 12.7 kcal/mol more stable than the initial Al13 CF3 structure. Subsequent uorination reactions are presented in Fig. 1. Because the H is negative for all of the elementary reactions across the reaction channel, they are energetically favorable. Kinetically, the rst step is predicted to be the least favorable because its energy barrier is calculated to be 20 kcal/mol. This barrier is then followed by much lower barriers of the second and third reaction steps in channel I. Note that the calculations predict the formation of AlC bonds on the Al13 cluster surface in the nal stage of the reaction. These predicted aluminum carbide AlC bonds are consistent with the x-ray photoelectron spectroscopy investigation of the interaction of aluminum with Teon-AF by Ding et al.31 The results of that investigation show that new chemical bonds are formed at the Al/Teon-AF interface including graphitelike CC, AlOC, AlC, and AlF bonds.

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-5

RRKM study of Al/Teon reaction

J. Chem. Phys. 133, 134305 2010

FIG. 2. Energy level diagram for reaction channel II. Dotted lines connect the transition states with the reactants and the products in each step. The reaction barriers Ea and reaction enthalpies H are provided for each step. The value in green is for single-point calculation at the M05/ 6-311+ G 3d2f level for the rst uorination step.

The multistep process for the uorination of Al13 from 179 CF3 fragments is calculated to be highly exothermic kcal/mol .
2. Channel II

The uorination reactions for this channel are shown in Fig. 2. The initial energy gain of 34 kcal/mol comes from

the formation of the Al13 CF2 complex. For this channel, we found that the rst uorination step is not only more exothermic than its CF3 counterpart, but the associated barrier is 3.3 times smaller 6 kcal/mol compared to 20 kcal/ mol . This activation energy indicates that a faster uorination reaction can be expected from CF2 than from CF3 radicals. Interestingly, the second uorination step of this

FIG. 3. Energy level diagram for reaction channels III and IV. Dotted lines connect the transition states with the reactants and the products in each step. The reaction barriers Ea and reaction enthalpies H are provided for each step. The value in green is for single-point calculation at the M05/ 6-311 + G 3d2f level for the rst uorination step.

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-6

M. Losada and S. Chaudhuri

J. Chem. Phys. 133, 134305 2010

channel is calculated to have a higher energy barrier 17 kcal/mol than the second uorination in the reaction with CF3 3.4 kcal/mol . Based on both, the assumption of the present approximation TST that intermediates and transition state structures are in thermal equilibrium and the fact that 17 kcal/mol is the highest barrier in this channel, the second uorination reaction in this reaction channel can be seen as a kinetic bottleneck and thus becoming a ratedetermining step for complete uorination of the Al cluster from CF2 fragments at lower temperatures. The formation of aluminum carbide bonds is also predicted in the last step in this reaction mechanism. The total reaction enthalpy is also calculated to be exothermic by 128 kcal/mol.
3. Channels III and IV

The results of the carbonyl uoride calculations indicate that COF and COF2 react differently with the Al13 cluster. The formation of the Al13 COF2 complex is kinetically hindered with a barrier of 8.4 kcal/mol, while the formation of the initial Al13 COF complex is predicted to be barrierless. In the case of the COF fragments, the reaction proceeds via the carbon atom to form the Al13 COF complex with an energy gain of 52 kcal/mol, whereas COF2 reacts via both the carbon and the oxygen atoms, as shown in Fig. 3. Unlike channels I and II, the COF and COF2 reaction channels are predicted to be less exothermic, with calculated reaction enthalpies of 74 and 65.1 kcal/mol, respectively. In terms of energy barriers, the rst uorination step in channel III is predicted to be 4.2 kcal/mol, compared to 9.0 kcal/ mol in the case of channel IV. In addition, the adduct formation between Al13 and COF2 has a kinetic barrier of 8.4 kcal/mol. This shows that the uorination of the Al13 cluster will not likely be much faster compared to COF channel if we consider the barrier before adduct formation. To look for possible carbidelike compounds on the PES of reaction channels III and IV, the formation of a second AlF bond from the F- Al13 COF complex in channel III was promoted by articially breaking the CF bond, followed by relaxation of the system; this resulted in the formation of the 2F Al13 CO carbidelike complex with an enthalpy of 36.9 kcal/mol, as shown in Fig. 3. By the same procedure, the F Al13 CO complex was obtained from the Al13 COF structure in channel IV. Note that the formation of the carbidelike structures, compared to the mechanism in which CO is eliminated, is also predicted to be exothermic cf. Fig. 3 . Because no TS structures were localized for these reactions, we cannot conclude whether the CO elimination or the carbidelike bond formation reaction is kinetically most favorable. By analyzing the exothermicity and the energy barriers across the reaction channels, we found, on the one hand, that reaction channel III is kinetically the most favorable; on the other hand, channel I is the most exothermic of the four reaction channels studied. A complete mechanistic description of the Al13 -Teon uorination reactions leading to the formation of AlF3 that includes all subsequent branching and side reactions becomes very complex. Here, we restrict ourselves to the initiation reaction: essentially Al13 clusters reacting with Teon fragments to give Al13 CF3 / CF2 / COF/ COF2

molecular species, followed by dissociative chemisorption. Understanding AlC and AlF bond formation is an important step in understanding the reactivity of the Al-Teon system under combustive environments. As predicted in our previous study,5 the formation of AlF is important for the production of AlF3, which is the nal product of the aluminum uorination reactions. With a predicted enthalpy reaction of 258 kcal/mol,5 the formation of AlF3 is believed to account for the high exothermicity of the Al-Teon reaction. For this reason, it is important to include all possible steps in the formation of AlF3 from Al clusters. In this case, the process probably involves the breakdown of the Al13 cluster into smaller units, as documented in the theoretical and experimental study by Burgert et al.32 for the chlorination of the Al13 cluster. The reaction path presented in the work of Burgert et al. for reactions of Aln with chlorine resulting in the release of AlCl as the main product offers a plausible model for the corresponding formation of AlF from the reaction of aluminum metal and uorine.
B. Rate constant calculations 1. Temperature-dependent canonical rate constants

The results of the ab initio calculations have been used to determine unimolecular canonical rate constants, k T , over the 500900 K temperature range using Eq. 1 . Arrhenius plots and the predicted energy of activation Ea for the rst uorination step of channels IIV are displayed in Fig. 4. Regression analysis of the corresponding Arrhenius plots yielded the following rate constants: k1,I T = 1.55 k1,II T = 2.38 k1,III T = 9.69 k1,IV T = 2.63 1012 exp 20 600/RT , 1012 exp 6500/RT , 1011 exp 4400/RT , 1012 exp 9650/RT .

The energies of activation are in cal/mol. The predicted Arrhenius expressions show the following: a among the rst uorination step reactions, the one from channel I is the slowest, while the reaction from channel IV is the fastest over the 500900 K range; b channels II and III are predicted to be competitive for temperatures greater than 700 K; and c frequency factors are predicted to be similar, as was the case in our previous study.5 Because the values of the frequency factors are relatively close, activation energy is the only criterion we can use to predict the reactivity in these cluster systems. Higher temperature reactions become competitive with fragmentation of the Al13 cluster. Nevertheless, the plot for the rate constants over the 10002000 K range is shown in Fig. S1 of the supporting information.19
2. Energy-dependent rate constants

In order to discuss the results of the kinetic calculations of the four reaction channels, it is important to consider the energetic stability of the Al13 cluster to provide a relevant scale for the combustion reactions. To address this, we re-

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-7

RRKM study of Al/Teon reaction

J. Chem. Phys. 133, 134305 2010

FIG. 5. Calculated energy-dependent RRKM rate constants, k E , for the rst uorination reaction of channels IIV.

FIG. 4. Arrhenius plots for the rst uorination step in reaction channels IIV within the 500900 K temperature range. Note that COF2 adduct formation is kinetically hindered Fig. 3 and thus may not be the main reaction channel driving the uorination reaction.

viewed the work of Li and Truhlar,33 in which the association of Al atoms with Aln clusters and the unimolecular dissociation reactions of Aln clusters were studied using classical trajectory simulations. From their work, we were able to establish a cluster dissociation energy threshold of 80 kcal/mol for Al13. Dissociation energies of aluminum clusters Aln+ with n = 7 17 were determined by Ray et al.7 By comparing the measured lifetimes with predictions of a RRKM model, they determined dissociation energy for the Al13 cluster of about 87 kcal/mol. These experimental and theoretical observations prompted us to restrict the RRKM rate constant calculations up to 80 kcal/mol. The Al13 cluster dissociation energy threshold is important to our calculations because it ensures that the predicted kinetic data presented in this study correspond to this cluster size. The energy-dependent RRKM rate constants for the rst uorination step of each reaction channel are plotted in Fig. 5 and the numerical values for all of the individual steps of each reaction channel are given in Table S3 of Ref. 19. Rate calculations were carried out for energies between the reaction threshold energy barrier and the Al13 dissociation energy of 80 kcal/mol. Figure 5 shows that the rst uorination of channel III, that is, Al13 COF2 F Al13 COF, is calculated to have the largest k E , and hence it is the fastest reaction up to excitation energies of 20 kcal/mol. At 20 kcal/ mol, the predicted k E rate for

the rst uorination of channel II overtakes that of channel III. Channel II retains the largest k E up to the highest excitation energy calculated. For excitation energies close to 60 kcal/ mol, the predicted k E of the rst uorination of channel IV, that is, Al13 COF F Al13 + CO, overtakes that of channel III. Thereafter, the calculated k E for channels IIIV are very close to each other. Note that for the rst uorination reaction of channel I, the calculated k E is the smallest over the entire excitation energy range, which clearly shows that it is kinetically the least favorable among the four reaction channels studied. The results of the energydependent RRKM rates are qualitatively consistent with our previous ndings5 and with recent experimental results14 where COF radical species reacts faster with aluminum than do CF3 fragments.
3. Pressure-dependent rate constants

In solving the master equation the choice of the downward collisional energy transfer parameter, E down, is an important factor. In most cases the assumed expression for E down satisfactorily reproduced experimental observations for the pressure dependence of a particular reaction. Unfortunately, there is no experimental pressure-dependent kinetic data for Al-Teon adducts that can guide us in the choice of E down. For our particular case, the closest reaction reported in literature is that for the thermal decomposition of CF3, CF3 + M CF2 + F + M, studied by Srinivasan et al.34 However, the temperature-dependent form of E down. used in this study was found to be not suitable for the Al-Teon system. This is not surprising, as Al13 -CF3 and related species would behave differently as colliders. As an alternative, we used two different models to solve the master equation as described below. The two models differ in the way they treat collisions using variants of energy-dependent expression for E down. For all of the calculations, the canonical rate constants at the high-pressure limit as discussed previously were used as a guide to the upper limit for the fall-off curves at a particular temperature. a. Model 1. In this model the master equation calculations used the exponential-down model,

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-8

M. Losada and S. Chaudhuri

J. Chem. Phys. 133, 134305 2010

P E,E = C exp

E E E down

where E down is represented with an energy-dependent function derived following the recommendations of Barker et al.35 It is assumed that E down takes the form E down E = c0 + c1 E, where c0 40 cm1 and c1 = E 40 / E0, with E0 being the reaction threshold energy and E is the average energy transferred in downward collisions. To estimate E, the procedure of Harrison et al.36 was adopted because it is well suited to treat dissociative reactions of molecular adsorbates on metal surfaces, a process similar to uorination of Al clusters described in this work. It is initially assumed that the average energy transfer between the complexes and the surrounding bath gas is mediated by the phonon modes of the Al13 -X adducts studied in this work. Thus, E is calculated to be the mean phonon energy for each complex. For example, the Al13 CF3 complex has 45 vibrational modes and its mean phonon energy is calculated to be 300.5 cm1. The above procedure led to the following expression for this complex, E down E = 40 + 0.037 245 E. Similar procedure was followed for the Al13 CF2, Al13 COF2, and Al13 COF clusters. b. Model 2. In this model a generalized exponential function26 for the collision model described by the following expression is used: P E,E = C exp E E E down . 7

For E down, we used the same energy-dependent expressions described in model 1. In the absence of specic information, the parameter was tested for values between 0.5 and 1.5, as suggested in literature. A value of 1.2 seems to provide a consistent behavior across all channels at different temperatures. This value is in line with the published values for polyatomic colliders. The results of the calculations using the two models for each reaction channel are displayed in Fig. 6. The fall-off curves calculated using models 1 and 2 are plotted using solid and dotted traces, respectively. The model 1 fall-off curves followed the expected behavior for pressure dependence of unimolecular rate constants at different temperatures for all four channels. In all cases, the fall-off curves level off before reaching the high-pressure limit canonical rate constants at these temperatures presented by an arrow in Fig. 6. It is also evident that with increasing temperatures, higher pressures are needed to reach the level-off or the plateau regions of the fall-off curves. The unimolecular rate constant for Al13 CF3 channel reaches the plateau at lowest pressures. Model 1, however, leaves a particularly wide gap from the high-pressure limits at 800 and 900 K for Al13 CF2 and Al13 COF channels see Figs. 6 b and 6 c , respectively . If we compare the pressure dependence of all the channels at 900 K, we nd the following trend: a the Al13 CF3 channel levels off at lower pressures of around 0.90 MPa, b the Al13 COF, Al13 CF2, and Al13 COF2 channels reaches the high-pressure plateau at 320, 2000, and 2400 MPa, respectively, and c the unimo-

lecular rate constants approach the high-pressure limit consistently in most cases except for the Al13 COF channel at 800 and 900 K. Model 2 values of unimolecular rate constants plotted in Fig. 6 dotted line are an improvement over those generated using model 1 if we consider lowering of the gap between high-pressure limit and the plateau of the fall-off curves as a measure. In general, it can be seen that it has a smaller effect at lower temperatures in comparison to model 1, but it shows some noticeable effects as temperature increases. The most noticeable improvement is observed for the reaction rates in the Al13 COF F Al13 + CO and Al13 CF2 F Al13 CF channels at 800 and 900 K with respect to model 1. Note that now the curves almost reach the highpressure limit at these temperatures Figs. 6 b and 6 c . Although the experimentally reported combustion regime2,30 for Al-Teon reactions is 500900 K and a peak pressure of 11 MPa, the calculated data in Figure 6 show a broader region of pressure dependence for some channels with pressures as high as 3000 MPa. A true comparison with experimental observations will require further experiments to understand the temperature dependence. Also, improved theoretical models using a multichannel treatment of the AlTeon reactions for obtaining net pressure dependence curves including the side reactions such as Al oxidation are currently being pursued. Note that although the rate constants for the uorination of the Al13 COF2 F Al13 COF reaction are the highest among the four reactions considered in Fig. 6, the kinetically limited formation of the Al13 COF2 cluster, compared to the formation of the other clusters, may render the uorination from this reaction less signicant at relatively low temperatures. The other channels have negligible barriers according to our DFT level calculations and thus are kinetically more favorable in comparison. The combined picture that emerge out of the rate constants calculated using different assumptions for Al cluster reaction with Teon fragments can be summarized as follows: a the rate constants k T , k E , and kuni P are consistent in predicting favorable kinetics for the CF2, COF, and COF2 channels; b the unimolecular rate from COF2 channel although high, the bimolecular reaction between Al cluster and COF2 is kinetically hindered; c the reaction of Al13 with CF3 fragments are predicted to be the more exothermic one 179 kcal/mol , followed by the reaction with CF2 128 kcal/mol , COF 74 kcal/mol , and COF2 65 kcal/mol fragments; d CF2, COF, and COF2 reaction channels exhibit strong pressure dependence at high temperatures. In contrast, CF3 reaction channel is almost pressure independent, as its high-pressure limit is reach at very low pressures; and e the rate constants are valid only between certain limits as Al cluster is prone to fragmentation at energies supplied either via collision or thermal routes. As the cluster is fragmented, our previous studies for Al atoms using higher level theories become more relevant for gasphase reactions under extreme conditions.

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-9

RRKM study of Al/Teon reaction

J. Chem. Phys. 133, 134305 2010

FIG. 6. Calculated pressure-dependent rates for a Al13 CF3 F Al13 CF2; b Al13 CF2 F Al13 CF; c Al13 COF F Al13 + CO; d Al13 COF2 F Al13 COF using model 1, solid trace, and model 2 with = 1.2, dashed-dot trace, for the master equation calculations see text . High-pressure limit rates are indicated by arrows with the corresponding numerical value. The formation of the initial complex, Al13 COF2, has an energy barrier of 8.4 kcal/mol. Although the rates plotted in panel d suggest that the Al13 COF2 F Al13 COF reaction is the fastest, the kinetically hindered formation of the Al13 COF2 initial complex may render this reaction comparable to the COF channel.

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-10

TABLE I. The calculated atomic charges from CHeLpG population analysis at the PBE0/6-311G d level of theory. Teon fragments Atom C
F1 F2 F3 O

CF3 0.9880
0.3363 0.3258 0.3258

CF2 0.2400
0.1200 0.1200

COF 0.6504
0.2947

COF2 M. Losada and S. Chaudhuri 1.4289


0.3970 0.3970 0.6348

0.3556 Clusters Al13

Al13 CF3 1.6425 0.5734 0.5567 0.5540

F Al13 CF2 1.2100 0.5175 0.5043 0.4317 0.4422 0.1500 0.2217 0.1705 0.0662 0.2732 0.1496 0.2687 0.1750 0.2173 0.2260 0.2784 1.7503

2F Al13 CF 0.7470 0.4481 0.4307 0.4304 0.1245 0.1300 0.2708 0.3854 0.0880 0.2646 0.1747 0.0935 0.3066 0.1490 0.1278 0.1684 1.4830

3F Al13 C 0.0459 0.4600 0.4622 0.4427 0.2081 0.1525 0.2045 0.2844 0.2556 0.3001 0.2560 0.2889 0.2501 0.2071 0.1578 0.2527 1.4792

Al13 CF2 1.3277 0.5766 0.5765

F Al13 CF 0.8135 0.5014 0.4236

2F Al13 C 0.1733 0.4564 0.4803

Al13 COF 1.2258 0.5802

Al13 F

Al13 COF2 1.5970 0.6320 0.6319 0.4604 0.3448 0.1635 0.2953 0.2279 0.2894 0.2332 0.2844 0.0668 0.2245 0.2443 0.2252 0.1683 1.8936

F Al13 COF 1.1685 0.4818 0.5570 0.3165 0.5091 0.1328 0.2716 0.2667 0.2975 0.3383 0.2965 0.2221 0.2924 0.2364 0.1934 0.2116 1.8607

F Al13 COF A 1.1933 0.4976 0.5377 0.7003 0.3996 0.1403 0.3328 0.2500 0.4556 0.2244 0.1637 0.1643 0.1782 0.3093 0.2380 0.1586 1.6463

Al13 2F

C F1 F2 F3 O Al1 Al2 Al3 Al4 Al5 Al6 Al7 Al8 Al9 Al10 Al11 Al12 Al13

0.5458

0.5222 0.5734

0.2747 0.2673 0.2657 0.2657 0.2673 0.2673 0.2657 0.2657 0.2673 0.2747 0.2744 0.2744 2.2304

0.3557 0.2245 0.2336 0.2598 0.2476 0.2632 0.2381 0.2246 0.2555 0.2231 0.2588 0.2190 2.0458

0.4007 0.1868 0.3023 0.2607 0.2605 0.2600 0.2600 0.3949 0.1894 0.3010 0.1810 0.1766 1.9998

0.3791 0.2373 0.2455 0.3857 0.1244 0.1278 0.0531 0.3836 0.1694 0.2455 0.2494 0.1743 1.7764

0.2392 0.2740 0.1673 0.3178 0.2184 0.2474 0.3390 0.3887 0.1850 0.1767 0.2264 0.1997 1.5868

0.7129 0.2917 0.2630 0.0782 0.2531 0.2765 0.2529 0.2673 0.2406 0.2489 0.2451 0.2404 0.2020 2.0855

0.2187 0.2222 0.3058 0.5107 0.3008 0.2190 0.2503 0.2117 0.3212 0.2981 0.2165 0.3050 1.9005

0.3585 0.2084 0.1684 0.3027 0.6130 0.0513 0.4940 0.2633 0.4171 0.0938 0.3239 0.0938 1.2703

J. Chem. Phys. 133, 134305 2010

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-11

RRKM study of Al/Teon reaction

J. Chem. Phys. 133, 134305 2010

2 FIG. 7. Plot of the molecular orbital density , HOMO, LUMO, and LUMO+ 1 of the Al13 , Al13 -CF3, Al13 -CF2, Al13 COF, and Al13 COF2 clusters calculated at the PBE0/6-311G d level.

C. Electronic structure and reactivity

In the present study, atomic charges, given in Table I, and dipole moment, displayed in Fig. S2 of Ref. 19, for all of the intermediate structures were calculated using the CHeLpG scheme37 as implemented in GAUSSIAN 03.21 Atomic radii for carbon 0.69 , aluminum 1.35 , uorine 0.67 , and oxygen 0.65 were obtained from the optimized cluster geometries and used for the population analysis calculations. Figure S2 of Ref. 19 shows the atomic charge density of the intermediate structures of reaction channels IIV. From the gure, it can be summarized that as the uorination reaction proceeds, the atomic charge of the core Al decreases and that of the carbon atom becomes more negative. This charge redistribution changes not only the orientation of the dipole moment but also increases its magnitude from 2.32 D in the Al13 CF3 cluster to 4.95 D in the 3F Al13 C cluster. Similar trends can be observed for the clusters involved in the other reaction channels. Taken together, the observed charge transfer within the clusters from the core Al atom to the carbon is probably involved in the dissociation reaction of the Teon fragments. For instance, once CF3 dissociates, the carbon atom moves to bond to a second, third, and nally a fourth Al atom with a net charge transfer of 1.596 to the carbon atom. This tetravalent behavior of the carbon arises due to the charge transfer from the core Al to the C atom. Mulliken population analysis was also carried out. The corresponding atomic charge values and densities of the four reaction channels are given in Table S4 and Fig. S3 of Ref. 19, respectively. Although Mulliken population analysis shows the same charge transfer effect to the carbon atom, the predicted atomic charges and dipole moments are quite different from those predicted by the CHeLpG scheme. Such difference is mostly due to the fact that the Mullikens as-

signment of half the overlap probability density to each atomic center is rather arbitrary and may lead to unphysical results. Although the calculated charge densities could not fully illuminate the reactive pathway of channels III and IV, they were nevertheless instructive and suggest a different chemistry between aluminum and Teon fragments when oxygen is involved. Because the reaction between Al and Teon ultimately produces CO, CO2, and graphitic phases, they may be produced from different reaction channels than those suggested by our calculations. Carbon adsorbed on the aluminum surface results from oxygen lean combustion conditions where CF3 and CF2 dominate the reaction chemistry. It is evident from the energy barriers for the rst uorination Figs. 13 that Al13 CF2 and Al13 COF2 are more reactive toward dissociation than the Al13 CF3 and Al13 COF clusters. Indeed, it has been shown that cluster reactivity is governed by the electronic and geometric shell closures.9 Both Al13 CF3 and Al13 COF have an icosahedral geometry, that is, it is not distorted with respect to the geometry of the bare Al13 cluster that is known to be highly stable. On the other hand, Al13 CF2 and Al13 COF2 have distorted geometries when compared to the icosahedral Al13 . In these terms, the reactivity of the Al13 CF2 and Al13 COF2 clusters toward dissociation is, in part, explained by this geometric factor. The explanation for the difference in reactivity also has an electronic origin, which makes it difcult to delineate between different pathways. The rst factor often used to describe cluster reactivity is the gap between the highest occupied molecular orbital HOMO and the lowest unoccupied molecular orbital LUMO , the HOMO LUMO gap. A smaller HOMO LUMO gap can be responsible for higher reactivity in metal clusters.38 Next, we examine if the HOMO LUMO gap could be responsible for the reactivity

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-12

M. Losada and S. Chaudhuri

J. Chem. Phys. 133, 134305 2010

TABLE II. The calculated HOMO LUMO, HOMO R HOMO TS , HOMO R HOMO P gaps, and the respective energy barriers along four reactions channels R = reactant, P = product, TS= transition state . HOMO LUMO gap eV 3.0 2.69 2.10 2.22 2.30 1.60 1.86 1.58 2.09 2.17 1.61 1.68 1.54 2.42 2.44 2.31 2.27 2.53 2.45 2.07 2.66 2.32 2.87 0.01 0.15 6.11 17.0 + 0.04 0.22 HOMOR HOMOTS eV Energy barrier kcal/mol HOMOR HOMOP eV

Reaction channel Channel I H = 179.0 kcal/ mol


a

Cluster Al13 Al13 CF3 TS1 F Al13 CF2 TS2 2F Al13 CF TS3 3F Al13 C
a

0.72 0.07 0.10

20.0 3.40 5.80

0.23 0.11 0.34

Channel II

H = 128.0 kcal/ mol

Al13 CF2 TS1 F Al13 CF TS2 2F Al13 C

Channel III

H = 65.1 kcal/ mol

Al13 COF2 TS1 F Al13 COF TS2 F Al13 COF A TS3 2F- Al13 Al13 COF TS F- Al13

0.12 + 0.12 0.20

4.20 2.70 6.60

0.073 + 0.24 0.28b

Channel IV

H = 74.1 kcal/ mol

0.40

9.00

+ 0.07

Total channel exothermicity. Taken only the 2F Al13 and F Al13

structures as products for reaction in channels III and IV, respectively cf. Fig. 3 .

and if the rules proposed by Chrtien et al.39 for the binding sites and relative energies for binding of propene to gold clusters could be extended to the present case. In all calculations in this work, the HOMO LUMO gap of starting AlTeon adduct decreases in the following order: ECF3 ECOF ECOF2 ECF2, as shown in Fig. 7. Although this trend correlates well with the order of reactivity of the rst step in the uorination dissociative reactions in each channel cf. Figs. 13 , the same cannot be said for the subsequent reaction steps. During dissociative uorination, CF bonds are broken and new AlC/AlF bonds are formed across the channels. We believe that this process should be more correlated with the difference in HOMO energy levels in reactant R versus transition state TS . By calculating such values, we nd that the HOMO R HOMO TS enRTS RTS RTS ECOF ECOF ergy gap follows the trend: ECF 3 2 RTS ECF see Table II . By following this trend across the 2 reaction channels, a correlation can be established between the HOMO R HOMO TS energy difference and the energy barrier of each uorination step. So, for instance, in channel I, the HOMO R HOMO TS of the rst step has an energy of 0.72 eV, which decreases to 0.07 eV, and then increases again to 0.10 eV for the second and third steps, respectively. This decreasing and increasing trend correlates well with the corresponding energy barriers of 20, 3.40, and 5.80 kcal/mol,

respectively, as shown in Table II. Note that this correlation is also observed in reaction channel II, where the HOMO R HOMO TS energy difference and the corresponding energy barriers follows the same trend. However, there is an exception, note that the HOMO R HOMO TS gap between the second intermediate F Al13 COF and the transition state TS2 of reaction channel III is calculated to have an energy of +0.12 eV, in contrast to the 0.12 eV calculated for the corresponding gap between the Al13 COF2 and the TS1 structures. Thus, it is found that one of the lowest barriers 2.7 kcal/mol among all the reactions studied does not follow the trend mentioned above. To further discuss this, we rst summarize the ndings as follows: a the transition state HOMO orbital is energetically less stable than the HOMO orbital of the reactant, except for that of second step in channel III; b individual energy gaps vary and are similar to the rules of Chrtien et al.,39 we nd that the barriers shown in column 5 of Table II follow the trend in HOMO R HOMO TS energy gaps. The exception found for the second step of channel III is primarily due to the fact that in this step, the molecular system undergoes a structural reorganization, where only the CO bond of the AlOCF moiety breaks to form subsequently a COF fragment bonded to the Al center. This is not similar to the other rearrangements that lead to the direct formation of a more stable AlF bond by transfer-

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-13

RRKM study of Al/Teon reaction

J. Chem. Phys. 133, 134305 2010

TABLE III. Role of nite size as evident from the comparison of trends in temperature-dependent rate constants performed at CCSD T /aug-cc-pVTZ//MP2/ aug-cc-pVDZ and DFT/PBE0/6-311G d levels for reactions involving Al atom and Al clusters, respectively. Activation energies are in cal/mol. 500900 K Reaction I CF3 fragments II CF2 fragments III COF2 fragments IV COF fragments Al atom 1.36 10 exp 4010/ RT 1.07 1013 exp 12 150/ RT 1.50 1012 exp 8520/ RT 5.58 1012 exp 1850/ RT
13

10002000 K Al13
12

cluster

Al atom 1.44 10 exp 4020/ RT 1.25 1013 exp 12 900/ RT 1.71 1012 exp 8750/ RT 5.63 1012 exp 1870/ RT
13

Al13
12

cluster

1.55 10 exp 20 600/ RT 2.38 1012 exp 6500/ RT 9.69 1011 exp 4400/ RT 2.63 1012 exp 9650/ RT

1.91 10 exp 21 020/ RT 2.63 1012 exp 6620/ RT 1.05 1012 exp 4510/ RT 2.95 1012 exp 9860/ RT

ring an F atom from the C atom. But rather, it leads to the formation of an intermediate structure F Al13 COF A in Fig. 3 from where a subsequent uorination step can take place towards the nal product formation 2F- Al13 + CO . We have also estimated the net change in the HOMO energies between reactants R and products P HOMO R HOMO P , column 6 in Table II . In a similar trend, only for the step where the HOMO R HOMO TS gap is positive, we nd that the corresponding HOMO R HOMO P gap, between the F Al13 COF and F Al13 COF A structures, is also calculated to be positive by +0.24 eV. This calculated positive energy gap correlates well with the corresponding HOMO R HOMO TS gap of the reaction step and further supports that the molecular system undergoes a structural and electronic reorganization before any further uorination reaction can take place.
V. CONCLUSIONS AND OUTLOOK

We have performed DFT based quantum chemistry calculations to study the reaction between the Al13 cluster and four Teon fragments: CF3, CF2, COF, and COF2. Analysis of the PES, which includes zero-point energy corrections, shows that all four reaction channels are exothermic, and that adsorption, followed by dissociative uorination of the Al cluster, is the reaction mechanism for each of the molecular systems studied. Barrier heights were predicted for individual steps of each reaction channel. We intended to study from rst-principles the kinetics of reactions taking place on the Al13 -Teon fragment PES. This was accomplished by determining temperaturedependent canonical rate constants using ab initio calculations by means of TST and accounting for pressure- and energy-dependent reaction rates through the solution of the master equations. Two different collision models that included energy-dependent expression for E down improved the unimolecular rate constants and behavior of the fall-off curves across different temperature and pressure regions. However, experimental pressure-dependent kinetic data are needed under these conditions for these empirical models to provide more robust description of collisional energy transfer in Al-Teon combustion reactions. We have shown that both geometric and electronic factors are important in explaining the reactivity of these systems. It was found that the rules proposed by Chrtien et al.39 for the reactivity of Au and Ag clusters partially explain the reactivity of the Al13 -Teon fragments molecular system towards dissociation. It may need an additional descriptor such as HOMO R

LUMO TS energy gap to explain the exceptions. In addition, the present rate constants for the Al cluster can be compared with our previous results at the CCSD T / aug-cc-pVTZ//MP2/aug-cc-pVDZ level for atomic Al.5 The comparison of rate constants is tabulated in Table III between 500900 and 10002000 K temperature ranges. Furthermore, only the rate constants of the rst uorination step of the cluster reactions are considered. It should be noted that DFT results are not exactly comparable, at a numerical level, with the higher level theories. However, the trends are worth mentioning. First, atomic level calculations and cluster calculations predict a comparable reactivity for all the reaction channels. The differences are mainly for CF3 and COF2 channels. The reactions with CF3 are slowed down at the cluster level. The COF2 reaction, although fast, has a kinetic barrier before adduct formation, which will make it slower than COF reactions. The CF2 seems to the ubiquitous intermediate in oxygen lean conditions in both length scales that will drive most of the exothermic yield from the uorination reactions. This is because although COF reactions can be comparable in terms of kinetics with CF2, the net exothermicity is lower 128 kcal/mol versus 74 kcal/mol . So any Teon fragmentation pathway that favors higher CF2 formation might produce the fastest uorination rate for Al based fuels and propellants. Our ongoing calculations for larger Al clusters and Al surface reactions for the same reactions will be able to identify if these trends hold across different length scales and cluster geometries.
ACKNOWLEDGMENTS

The authors want to thank Dr. Y. M. Gupta for suggesting this problem and for subsequent discussions. This work was supported by the Ofce of Naval Research under Grant Nos. N00014-04-1-0688 and N00014-06-1-0315.
1 2

J. J. Granier and M. L. Pantoya, Combust. Flame 138, 373 2004 . K. W. Watson, M. L. Pantoya, and V. I. Levitas, Combust. Flame 155, 619 2008 . 3 M. L. Pantoya and J. J. Granier, Propellants, Explos., Pyrotech. 30, 53 2005 . 4 K. Park, D. Lee, A. Rai, D. Mukherjee, and M. R. Zachariah, J. Phys. Chem. B 109, 7290 2005 ; A. Rai, K. Park, L. Zhou, and M. R. Zachariah, Combust. Theory Modell. 10, 843 2006 ; T. Bazyn, H. Krier, and N. Glumac, Combust. Flame 145, 703 2006 . 5 M. Losada and S. Chaudhuri, J. Phys. Chem. A 113, 5933 2009 . 6 K. Schriver, J. L. Persson, E. C. Honea, and R. L. Whetten, Phys. Rev. Lett. 64, 2539 1990 . 7 U. Ray, M. F. Jarrold, J. E. Bower, and J. S. Kraus, J. Chem. Phys. 91, 2912 1989 . 8 M. F. Jarrold, J. E. Bower, and J. S. Kraus, J. Chem. Phys. 86, 3876 1987 .

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

134305-14
9

M. Losada and S. Chaudhuri

J. Chem. Phys. 133, 134305 2010 E.01, Gaussian, Inc., Wallingford, 2004. E. P. Wigner, J. Chem. Phys. 5, 720 1937 . 23 R. T. Skodje and D. G. Truhlar, J. Phys. Chem. 85, 624 1981 . 24 W. L. Hase and L. Zhu, QCPE: A general RRKM program, Wayne State University, 1993. 25 D. C. Tardy and B. S. Rabinovitch, J. Chem. Phys. 45, 3720 1966 . 26 T. Lenzer, K. Luther, K. Reihs, and A. C. Symonds, J. Chem. Phys. 112, 4090 2000 . 27 G. H. Peslherbe and W. L. Hase, J. Chem. Phys. 105, 7432 1996 ; C. J. Cobos, A. E. Croce, K. Luther, and J. Troe, J. Phys. Chem. A 114, 4755 2010 . 28 C. L. Yaws, Matheson Gas Data Book, 7th ed. McGraw-Hill, New York, 2001 . 29 J. R. Barker, Int. J. Chem. Kinet. 33, 232 2001 ; 41, 748 2009 . 30 D. E. G. Jones, R. Turcotte, R. C. Fouchard, Q. S. M. Kwok, A.-M. Turcotte, and Z. Abdel-Qader, Propellants, Explos., Pyrotech. 28, 120 2003 . 31 S.-J. Ding, V. Zaporojtchenko, J. Kruse, J. Zekonyte, and F. Faupel, Appl. Phys. A: Mater. Sci. Process. 76, 851 2003 . 32 R. Burgert, H. Schnckel, M. Olzmann, and K. H. Bowen, Jr., Angew. Chem., Int. Ed. 45, 1476 2006 . 33 Z. H. Li and D. G. Truhlar, J. Phys. Chem. C 112, 11109 2008 . 34 N. K. Srinivasan, M. C. Su, J. W. Michael, A. W. Jasper, J. S. Klippenstein, and L. B. Harding, J. Phys. Chem. A 112, 31 2008 . 35 J. R. Barker, L. M. Yoder, and K. D. King, J. Phys. Chem. A 105, 796 2001 . 36 A. Bukoski, D. Blumling, and I. Harrison, J. Chem. Phys. 118, 843 2003 . 37 C. M. Breneman and K. B. Wiberg, J. Comput. Chem. 11, 361 1990 . 38 S. N. Khanna, B. K. Rao, and P. Jena, Phys. Rev. B 65, 125105 2002 . 39 S. Chrtien, M. S. Gordon, and H. Metiu, J. Chem. Phys. 121, 3756 2004 ; 121, 9925 2004 ; 121, 9931 2004 .
22

R. E. Leuchtner, A. C. Harms, and A. W. Castleman, J. Chem. Phys. 91, 2753 1989 ; 94, 1093 1991 . 10 C. W. Bauschlicher and L. G. M. Pettersson, J. Chem. Phys. 84, 2226 1986 ; H. P. Cheng, R. S. Berry, and R. L. Whetten, Phys. Rev. B 43, 10647 1991 ; S. H. Yang, D. A. Drabold, J. B. Adams, and A. Sachdev, ibid. 47, 1567 1993 . 11 W. D. Knight, K. Clemenger, W. A. de Heer, W. A. Saunders, M. Y. Chou, and M. L. Cohen, Phys. Rev. Lett. 52, 2141 1984 ; W. A. de Heer, W. D. Knight, M. Y. Chou, and M. L. Cohen, Solid State Phys. 40, 93 1987 . 12 B. K. Rao and P. Jena, J. Chem. Phys. 111, 1890 1999 . 13 R. L. Hettich, J. Am. Chem. Soc. 111, 8582 1989 . 14 M. A. Zamkov, R. W. Conner, and D. D. Dlott, J. Phys. Chem. C 111, 10278 2007 . 15 J. Troe, J. Chem. Phys. 66, 4745 1977 ; 66, 4758 1977 . 16 J. I. Steinfeld, J. S. Francisco, and W. L. Hase, Chemical Kinetics and Dynamics, 2nd ed. Prentice-Hall, Upper Saddle River, NJ, 1998 ; W. L. Hase and T. Baer, Unimolecular Reaction Dynamics: Theory and Experiments Oxford University Press, New York, 1996 . 17 C. Adamo, M. Cossi, and V. Barone, J. Mol. Struct.: THEOCHEM 493, 145 1999 . 18 R. Krishnan, J. S. Binkley, R. Seeger, and J. A. Pople, J. Chem. Phys. 72, 650 1980 . 19 See supplementary material at http://dx.doi.org/10.1063/1.3480020 for harmonic vibrational frequencies, tabulated energy-dependent RRKM rate constants, Mulliken atomic charges, Arrhenius plots for the 10002000 K range, CHeLpG and Mulliken charge density plots, and Cartesian coordinates for the optimized geometries calculated at the PBE0 / 6-311G level for reactants, products, and transition state structures of each reaction channel. 20 Y. Zhao, N. E. Shultz, and D. G. Truhlar, J. Chem. Phys. 123, 161103 2005 . 21 A. Frisch, M. J. Frisch, and G. W. Trucks et al., GAUSSIAN 03, Revision

Downloaded 19 Dec 2011 to 134.148.177.155. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

You might also like