Direct Dynamics Simulations of the Reaction O+(4S) + HCN at

Oct 16, 2009 - (4S) + HCN reaction at hyperthermal energies were performed based on the ground state PM3 potential energy surface, and the results hav...
1 downloads 0 Views 2MB Size
J. Phys. Chem. C 2010, 114, 5263–5275

5263

Direct Dynamics Simulations of the Reaction O+(4S) + HCN at Hyperthermal Collision Energies† Lipeng Sun‡ and George C. Schatz* Department of Chemistry, Northwestern UniVersity, EVanston, Illinois 60208-3113 ReceiVed: April 6, 2009; ReVised Manuscript ReceiVed: May 10, 2009

Direct dynamics classical trajectory simulations of the O+(4S) + HCN reaction at hyperthermal energies were performed based on the ground state PM3 potential energy surface, and the results have been used to interpret guided ion-beam experiments. This simulation exemplifies the complex chemistry and dynamics that can happen in hyperthermal ion-molecule reactions. Calculations were carried out at various collision energies ranging from 2 to 15 eV with emphasis on energies higher than 5 eV. All the energetically accessible reaction channels were found, including the HCO+/HOC+, NO+, and CO+ ions reported in the experiment. OCN+ and CN+ were found to be negligible at high energies. Relatively lightweight ions, NH+, CH+, and OH+, which are energetically accessible on both quartet and doublet potential surfaces but which were not reported in the experiment, are found to have cross sections below 1 Å2. The calculated excitation functions for NO+, HCO+, and CO+ agree with experiment qualitatively with quantitative agreement for NO+. Cross sections for HCO+ and CO+ are consistently overestimated at high energies, which is attributed to the poor description of the repulsive part of the potential energy surface by the PM3 Hamiltonian. Product energy disposal and angular distribution reveal complex mechanisms for elimination, abstraction, and exit-channel charge transfer reactions. The angular distributions at high energies (>8 eV) are characterized by dominant forward scattering as a result of direct reaction mechanisms. Backward and sideways scattering at lower energies is due to the formation of very stable intermediate molecular ions. The agreement between simulation and experiment for cross sections and angular distributions, especially their dependence on the collision energy, suggests that the quartet ground state dynamics are dominant in high energy collisions. I. Introduction Motivated by the need to understand chemistry in extreme conditions, recently there has been considerable interest in the study of hyperthermal reactions.1,2 In contrast to thermal reactions, the large amount of energy in the colliding species enables the system to follow reaction pathways far away from minimum energy paths on the potential energy hypersurface (PES), leading to product branching ratios, energy transfer and partitioning, and reaction mechanisms that are difficult to predict or interpret. Reactions involving the O+ cation are important in atmospheric and interstellar chemistry. O+ is the most abundant ion in low earth orbit (LEO), and it is also present in the atmospheres of planets such as Mars and Venus.3,4 O+ also plays an important role in oxygen/air plasmas that are important in ignition and control systems in high performance aircraft engines.5 Experimentally, the O+ reactions with CH3N2, NH3, H2S, CH3OH, H2CO, O2, H2O, CH4, CO2, CO, H2, and N2 has been studied by Smith, Adams, and Miller.6 Dressler and coworkers investigated O+ collisions with small hydrocarbons (e.g., CH4, C2H6, C3H8, etc.) and methyl-hydrazine,7 while Morris et al.8 studied the reactions with fluorohydrocarbons. There has been very little theory (other than simple models and a few stationary point energy calculations) done for the collisions of O+ (in any electronic state) with any molecule with the †

Part of the “Barbara J. Garrison Festschrift”. * To whom correspondence should be addressed. E-mail: schatz@chem. northwestern.edu. ‡ Current address: Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC 27695-7907.

exception of Morokuma’s extensive electronic structure studies of the O+(4S) + C2H2,9 and [O-H-C-N]+10 potential surfaces and our own work on O+ + CH411,12 that we discuss below. In the work reported here, we consider the O+ + HCN reaction at hyperthermal energies (2-15 eV) as a followup to guided ion-beam experiments that were carried out by Bastian et al.13 over a wide range of energies (0.25-25 eV). The following reactions were considered in the experimental work, corresponding to exothermic reactions on the quartet PES

O+(4S) + HCN f O(3P) + HCN+(2II)

-0.018 eV

(1) f HCO+(1Σ+) + N(4S)

-4.18 eV

(2)

f COH+(1Σ+) + N(4S)

-2.76 eV

(3)

f NO+(1Σ+) + CH(a,4Σ-) f CO+(2Π) + NH(4Σ-)

-0.53 eV - 0.84 eV

(4) (5)

As in the O+ + CH4 reaction, a long-range resonant charge transfer process is dominant at high collision energies. In addition, the HCO+/HOC+, NO+, and CO+ product ions were all found in the experiment. Experimental evidence suggests that the reaction proceeds via an ion-molecule mediated mechanism at low energy and evolves into a direct reaction

10.1021/jp903177a  2010 American Chemical Society Published on Web 10/16/2009

5264

J. Phys. Chem. C, Vol. 114, No. 12, 2010

Sun and Schatz

TABLE 1: Relative Energies for Reactants and Products on the Ground State Quartet Potential Energy Surfacea chemical species +

O + HCN HCN+ + O HCO+ + N HOC+ + N OCN+ + H NO+ + CH CH+ + NO CO+ + NH NH+ + CO OH+ + CN CN+ + OH CO+ + N + H

PM3 0.0 -0.490 (-0.474) -4.222 (-4.205) -2.257 (-2.239) -2.809 (-2.643) -0.822 (-0.786) 0.174 (0.269) -1.634 (-1.598) -2.755 (-2.711) 1.082 (1.114) 1.660 (1.685) 2.103 (2.379)

ROHF-MP2/aug-cc-pVTZ

B3LYP/6-311G(d,p)b

0.0 -0.522 (-0.497) -4.014 (-4.068) -2.092 (-2.074) -2.008 (-1.843) -0.549 (-0.502) 1.448 (1.534) -0.771 (-0.732) -1.348 (-1.267) 0.779 (0.851) 2.147 (2.194) 2.467 (2.718)

0.0 -0.577 -4.306 -2.693 -2.800 -0.360 0.889 -1.253 -1.665 0.130 1.136 2.257

exptc 0.0 -0.02 -4.159 (-2.728) (-1.99) -0.53d -0.84e -1.56f 0.347 (1.435) 2.43d

a Energies are in eV. Numbers in parentheses are energies without ZPE correction. b DFT results are taken from ref 10. c Experiments are taken from ref 23. The numbers in parentheses are not well-established data. d Experimental value is taken from ref 13. e Reference 24. f Value is taken from NIST Web site http://webbook.nist.gov.

mechanism at high energies. Luna, Mebel, and Morokuma examined the O+ + HCN reaction PES and reported an extensive investigation of the global PES for the [H,C,N,O]+ system in both quartet and doublet states. However, dynamics calculations were not reported. Theoretically, hyperthermal reactions involving O+ can be complex to describe due to the possibility of charge transfer in both entrance and exit channels, intersystem crossing that could lead to the formation of doublet states in addition to the initially produced quartet, multiple isomerization paths for stable molecular ions, electronically excited products, and so forth. Accurate dynamics simulation of this reaction could require knowledge of both ground and excited electronic state PESs for all reactions and a proper treatment of the nonadiabatic dynamics. Such calculations would require considerable computational resources in addition to human effort and are not practical. In this work, we simplify the description by assuming that internal conversion is efficient, and intersystem crossing is weak so that only ground state dynamics are needed to understand the dynamics. In addition, we assume that a simple semiempirical electronic structure method (PM3) can be used to determine the reactive force field. Despite these serious assumptions, we show that at least for hyperthermal energies this treatment works well, and as a result we can provide a simple but meaningful interpretation of the experiments. This work follows up our previous studies of the O+ + CH4 reaction using direct dynamics simulations with the PM3 electronic ground state Hamiltonian.11,12 Although the PM3 potential surface is only qualitative (off by about 0.5 eV for many processes), the good agreement between the calculated and experimental cross sections and the product angular distribution rationalizes the significant assumptions involved. Indeed, these calculations predicted product ions that were missing in the original experiments6,14 and which were later confirmed.15 However, the general applicability of ground state dynamics simulations for O+ cation reactions and the validation of using only a qualitatively correct PM3 Hamiltonian to describe this complex reaction need to be further tested, thus providing a motivation for the present study. A key reason for our choice of PM3, which admittedly is a low-level electronic structure method, is that it is considerably more robust than higher level methods with respect to convergence of the selfconsistent field equations. This makes it feasible to use, while other choices that we considered were not. The paper is structured as follows: in Section II, the computational details are provided; in Section III, the results are described and discussed; and Section IV summarizes the conclusions.

II. Computational Details II.A. PM3 Potential Energy Surface. Luna, Mebel, and Morokuma10 characterized the global PES of [O,H,C,N]+ using B3LYP/6-311G(d,p), that is, the Becke three-parameter nonlocal exchange functional16,17 with the nonlocal correlation functional of Lee, Yang, and Parr.18 They showed that the B3LYP/6311G(d,p) energies are in overall good agreement with MP2/ 6-311G(d,p), G2 and experiments for ionization potentials (IPs), reaction enthalpies, and proton affinities. However, the quality of DFT for the transition state energies was not characterized by the authors. In our previous studies of O+ + CH4 system, it is found that B3LYP consistently underestimates the reaction barriers and overbinds the reaction intermediates. This is a wellknown limitation in DFT methods which is due to the insufficient treatment of the fractional charge occupation in the density functionals.19 Although high level ab initio calculations can be applied to achieve “chemical accuracy” (ca. < 1 kcal/ mol), this requires significant computational effort and is beyond the scope of this paper. Alternatively, second-order Møller-Plesset perturbation theory (MP2) with large basis set requires moderate computing power and has been shown to give good agreement with CCSD(T)20 calculations for the O + + CH4 reaction. Therefore, in order to provide a more quantitative estimation of the barrier heights, we further characterize the PES using MP2 with an aug-cc-pVTZ basis set.21 All the electronic structure calculations were performed using the GAMESS program package.22 Table 1 presents the ab initio, PM3, and experimental enthalpies23,24 for reactions 1-5 and additional possible reaction channels on the quartet ground state PES. Although not reported in the experiment, the OCN+, CH+, NH and CN+ and OH+ ions are all energetically accessible under the experimental conditions and the OCN+ + H and NH+ + CO products are even thermodynamically quite favorable. In addition, the experimentally observed CO+ can also be produced by secondary fragmentation of the HCO+ and OCN+ ions. This reaction channel is endothermic with a threshold of above 2.4 eV. The average absolute mean deviation from experiment for PM3 is 0.50 eV while those for the MP2 and DFT calculations are 0.24 and 0.25 eV, respectively. Here, MP2 calculations give the best agreement with experiment except for the CN+ + OH channel which is overestimated by 1 eV. The PM3 PES is generally more exothermic than experiment, except for the HOC+ + N, OH+ + CN, and OH + CN+ reactions. Reaction intermediates, transition states, and reaction paths are depicted in Figure 1. Energies calculated by the PM3, DFT,

Reaction O+(4S) + HCN at Hyperthermal Collision Energies

J. Phys. Chem. C, Vol. 114, No. 12, 2010 5265

Figure 1. Diagram showing the primary reaction channels. Energies are in eV with the top, middle, and bottom numbers for each structure giving the zero point energy corrected PM3, B3LYP, and MP2 energies, respectively. The zero energy reference is for O+ + HCN. Reaction intermediates are arbitrarily labeled as IMn with n ) 1-10 denoting the 10 complexes. (a) Diagram for HCO+, OCN+, HOC+, NO+, and CH+ product channels. (b) Diagram for CO+, NH+, OH+, and CN+ channels.

and MP2 methods are listed along with each of the structures. For simplicity, the reaction intermediates are randomly labeled as IMn with the subscript n ) 1-10 accounting for the total 10 intermediate structures. Similar to what was found for the O+ + CH4 system, the B3LYP hybrid density functional calculations for the intermediates and saddle points have systematically lower energies than MP2. Quantitatively, the difference between the DFT and MP2 energies is about 1 eV and can be nearly 2 eV for some structures (e.g., IM1TIM4 isomerization barrier). The PM3 energies are higher than B3LYP and lower than MP2, but more similar to the B3LYP results.

Except the barrier for the IM4-IM10 transition which was not found on the MP2 PES, all other ion-molecule reaction intermediates, transition states, and reaction paths from both MP2 and DFT calculations were also found for the PM3 PES. This indicates that the PM3 Hamiltonian is qualitatively valid for the O+ + HCN reaction system. However, the PM3 results are not expected to be quantitative at low collision energies due to the large (>1 eV) deviation from the MP2 PES. The reaction intermediates can be further be categorized in two types. One is the pre- and postreaction complexes that are formed by long-range electrostatic interactions, such as the

5266

J. Phys. Chem. C, Vol. 114, No. 12, 2010

Sun and Schatz

Figure 2. Snapshots of representative trajectories for different reaction products. (a) Trajectory snapshots for HCO+, HOC+, and OCN+ channels. (b) Trajectory snapshots for NO+ and CH+ channels. (c) Trajectory snapshots for OH+ and CN+ channels. (d) Trajectory snapshots for CO+ and NH+ channels.

prereaction complexes [O---HCN]+ (IM1) and [O---NCH]+ (IM2), and the other involves stable molecular cations such as [HOCN]+ (IM4) and [HCNO]+ (IM5). There exist two prereaction complexes (i.e., IM1, IM2) at the entrance channel, either of which can lead to all products. However, IM2 is a much more stable structure with lower energy barriers leading to the products. A distinct feature of the O+ + HCN+ system, as shown in Figures 1 and 2, is that it is strongly mediated by stable molecular ions (e.g., IM3-IM5 and IM8). Indeed the same products can be formed by different reaction paths following

multiple isomerization reactions. For example, both IM3fIM8 and IM3fIM4fIM8 can lead to the CO+/CO products. Except for the IM1-IM4-IM10 isomerization, other transition states have lower energy than the reactants. There are also other intermediate isomerization steps that require both isomers and transition states with higher energies than the reactants (e.g., Figure 8 and 9 in ref 10). However, these paths are expected to be both kinetically and dynamically unfavorable and will not be discussed here. Indeed, these pathways were not found in the dynamics simulations.

Reaction O+(4S) + HCN at Hyperthermal Collision Energies

J. Phys. Chem. C, Vol. 114, No. 12, 2010 5267

The relative barrier heights among most of the transition states for the PM3 surface are qualitatively consistent with DFT/MP2 predictions. However, the barrier for the IM3-IM8 isomerization is higher than that for IM4-IM8 on DFT/MP2 PES, while on PM3 PES, this relationship is reversed. Under thermal conditions, such disagreements may affect the product branching ratios calculated on the DFT/MP2 and PM3 surfaces. However, it may not be important in hyperthermal collisions where a direct reaction mechanism is dominant. Although HCO+/HOC+, OCN+, CO+, NH+, and NO+ ions are all exothermic reaction products that can be formed following reaction paths with low energy barriers, only HCO+/ HOC+, NO+, and CO+ were detected in the previous experiments. It may be argued that this could be due to significant quartet-doublet intersystem crossing for thermal conditions. However, both the NH+ and OCN+ reaction channels are also exothermic on the doublet surface. Thus, the OCN+ and NH+ products are kinetically and energetically feasible on both quartet and doublet surfaces. The importance of the triplet-singlet intersystem crossing for reactions involving O(3P) with small molecules such as H2 and C2H4 has been investigated theoretically by Schatz and co-workers.25-27 For O + H2, which is a direction reaction, intersystem crossing plays only a small role. For O + C2H4, which involves intermediate complex formation, intersystem crossing was found to be important at thermal energies, but is much less significant at hyperthermal energies due to the weak spin-orbit coupling and the short lifetime of any intermediate complexes that are formed. Most likely this situation also applies to O+ + HCN. In our previous study of O+ + CH4, some of the possible product ions predicted by theory were missing from early experimental studies, but were found in the most recent experiments.15 Therefore, further experimental work will be valuable in resolving this discrepancy. Figure 1 shows that the CH+, OH+, and CN+ product ions can be formed in hyperthermal collisions. No transition state structures were found for the CH+, CN+, and CO+ products (indicated by dashed lines in Figures 1). Although hard-to-find high energy transition states might exist on the quartet potential energy surface, these ions can also be formed from unimolecular dissociation of IM5, IM4, and IM8, respectively, or by charge transfer reactions with the CH, CN, NH products in the exit channel. Secondary fragmentation (with respect to the products in table 1) of HCO+ and OCN+ can also happen given the large exothermicity of the primary reaction channels. The dissociation energy of HCO+ f H + CO+ is 6.58 eV by PM3, which is in good agreement with the 6.73 eV value from recent high level ab initio calculations28 (CR-CC(2,3)/cc-pVQZ).29,30 In addition, the 6.78 eV dissociation energy from MP2 is in excellent agreement with the CC(2,3)/cc-pVQZ results.25 For the OCN+ f CO+ + N reaction, dissociation energies of 4.93 and 4.47 eV are predicted by PM3 and MP2, respectively. PM3 overestimates the dissociation threshold by about half an eV. II.B. Direct Dynamics Simulations. Direct dynamics simulations in which the electronic structure calculations are performed “on the fly” during the integration of classical trajectories have widely been used for chemical dynamics simulations. In the work reported here, analytic energy gradients are directly computed from the PM3 Hamiltonian as coded in the GAMESS program.22 Quasiclassical sampling was used for selecting trajectory initial conditions with the center-of-mass relative translational energy ranging from 2 to 15 eV. The impact parameter b is randomly selected according to b ) bmax(ξ)1/2, where ξ is a random number ranging from 0 to 1 and bmax is

selected to be 4 Å, which is large enough that there are no chemical reactions at larger impact parameters. We have not considered large enough impact parameters to converge the charge transfer cross section, so this will not be reported. The charge transfer dynamics, which is difficult to study due to nonadiabatic effects, is not coupled to the reactive channels at large impact parameters so it is not needed for the present study. The trajectory integration step-size is chosen to be 0.1 fs. Trajectories are integrated for 500 fs, except at 2 eV collision energy where a longer 1.5 ps integration was used. Some intermediate complexes were not dissociated at the end of the trajectory duration; however, in no case does this influence the calculated cross sections by more than their uncertainty. For all the results reported, spin contamination is found to be small (change in 〈S2〉 is less than 8%. For some products where secondary fragmentation happens, spin contamination can be large. This is expected and the trajectories are still included in the resulting analysis. Five thousand trajectories are calculated for each collision energy. III. Results and Discussion III.A. Reaction Products. Before we discuss our calculations in detail we will first present some qualitative results. In agreement with experiment, HCO+/HOC+, NO+, and CO+ (the mass-to-charge ratio, m/e, ranging from 28 to 30u) were all found in the simulation. In addition, other product ions including lighter ions such as CH+ (13u), NH+(15u), and OH+(17u) and heavier ions such as CN+(26u) and OCN+ (42u) also appeared in the dynamics simulations, although with smaller cross sections. As shown in Section IIA and in ref 10, the latter ions are all energetically accessible on both quartet and doublet potential surfaces but were not reported in the experiment. To make sure these ions were not falsely detected due to the short trajectory integration time, we calculated the internal energy of these ions. The results show that the fraction of the ions that have enough energy to undergo further fragmentation never exceeds 3%. As expected from Table 1 and Figure 1, only the CH+, CN+, and OH+ ions have nonzero threshold energy. The trajectory simulations show that at hyperthermal energies a direct reaction mechanism becomes more important. We also found that there are almost no trajectories that follow exactly the reaction paths illustrated in Figure 1. At a collision energy of 2 eV, however, reaction intermediate mediated mechanisms are still important for some products such as NO+. The molecular ion products (identified by when the bond lengths are less than ∼1.8 times their equilibrium values), mainly [HCNO]+ and [OHCN]+, were found to contribute about 6% to the total reaction cross section after integrating the trajectories for 500 fs at 5 eV, so the short integration time is not a major factor in determining product branching. Snapshots of representative trajectories for the product ions are depicted in Figure 2a-d at 5 eV collision energy. As clearly shown in Figure 2, none of the trajectories is trapped in a prereaction potential well, that is, the IM1 and IM2 structures in Figure 1. The HCO+, HOC+, and OCN+ products are elimination reactions with different reaction mechanisms. HCO+ is produced from the direct formation of IM3 followed by N-C bond rupture within one rotational period, while HOC+ and OCN+ formation involves C-H bond breaking by O insertion. The NO+ + CH channel was found to be the major reaction channel for NO+ formation. The NO+ + C + H channel can also occur at higher energies since the threshold for this reaction is ∼2.21 eV, however its contribution is small (60%) of the HCO+ product ions having enough internal energy to further dissociate into CO+ + H given a longer time than 500 fs. We have assigned these trajectories to the CO+ channel in making this plot. As is shown in Figure 3a, the shape of the excitation functions for HCO+ and CO+ agree well with experiment. Quantitatively, the HCO+ cross sections agree well with experiment, but the CO+ cross sections are consistently overestimated. Although CO+ can also come from OCN+ dissociation, the contribution of this process to the CO+ cross section is small, ranging from 5% at 5 eV to 25% at 15 eV. Therefore, the overestimation of the CO+ excitation function is coupled to the HCO+ channel and the good agreement of the HCO+ cross section with experiment is somewhat accidental, However, the underestimation of the CO+ cross section in the 2-5 eV range is not clear. Bastian et al. suggest that electronically excited CO+ (A, 2Π) + NH(3Σ-), which is 1.74 eV endothermic, is also possible on the quartet surface.13 However, the ground state dynamics simulation performed here is not capable of producing this reaction channel. Therefore, the branching ratio between CO+ (A, 2Π) and CO+ (X, 2Π) cannot be determined. Analysis of the trajectories results shows that HCO+ is mainly formed by direct C-O bond formation and C-N bond dissociation. Since the barrier for C-O bond formation (the IM2fIM3 transition state) is lower than the reactant energy, there will not be much resistance for O directly approaching the C atom without forming a prereaction complex, that is, the IM2 structure. Given the high kinetic energy, the newly formed C-O bond can be significantly compressed, and thus the repulsive part of the potential energy surface will be an important factor governing the magnitude of the reaction cross section. With this picture, an excessively repulsive potential surface that prefers rebounding of the reactants will lead to underestimation of the cross section, while a too attractive surface that leads to excessive translation to internal (T-I) energy transfer will overestimate the reaction probability. In Figure 4, we plot the potential energy as a function of the C-O bond distance with the H-C-N fixed at the transition state configuration for both MP2 and PM3 potentials. We see that the MP2 and PM3 energies increase sharply when then C-O distance is shorter than ∼1.2 Å, but the MP2 surface is much more repulsive than PM3. The average C-O bond length at the trajectory turning point is ∼1.02 Å at 5 eV at which point the MP2 surface is about 4 eV more repulsive than PM3. Thus it is expected that PM3 will overestimate the cross sections and this is what is found in the simulation. Once the potential barrier is passed, the post-transition state dynamics is characterized by the release of potential energy to relative translation since the reaction coordinate in the exit channel is largely determined by C-N bond breaking. The decrease of the excitation functions with increasing energy in Figure 3a is determined by the reduction in impact parameter. The calculated average impact parameters for the NO+ product decrease from 1.8 to 0.6 Å going from 2 to 15 eV. The average impact parameter for HCO+ decreases from 1.3 to 0.7 Å. The average impact parameters for the HCO+

J. Phys. Chem. C, Vol. 114, No. 12, 2010 5269

Figure 4. Potential energy as a function of C-O distance with H-C-N fixed at the IM2TIM3 transition barrier. The zero energy reference is O+ + HCN.

product channel is consistently smaller than the NO+ channel, except at 15 eV. This is consistent with the slower decreasing rate for HCO+ in Figure 3a. Detailed examination of the trajectory animations shows that C-O bond formation is concerted with C-N bond elongation during the course of the reaction. This indicates very efficient coupling between incoming O translation and antisymmetric stretching of the O-C-N, thereby enhancing the reaction probability. This direct approach to a late transition state by the O atom will favor a small impact parameter. This explains the smaller impact parameter for the HCO+ channel. For the NO+ + CH products, however, the O atom approaches the N atom in the direction normal to the H-C-N axis with a larger impact parameter (e.g., the average incident angle is 85° at 5 eV and the corresponding average impact parameter is 1.8 Å). This is the characteristic of an abstraction reaction. Compared to the mechanism for HCO+ product formation, the O atom approaches the N atom in the HCN molecule at a ∠ONC angle about 90° that is not efficient in promoting the C-N bond stretching. Therefore, the reaction cross section is smaller than for HCO+ and decreases more rapidly with collision energy. The excitation functions for product ions not reported by the experiment, but found in the simulation, are plotted in Figure 3b. OCN+ +H is an energetically and kinetically favorable product channel on both quartet and doublet potential surfaces. The barrier along the reaction path for OCN+ + H is much higher than that for the HCO+ + N channel, thus making OCN+ an unimportant product for thermal conditions. At hyperthermal energies, the OCN+ + H product can be formed via a direct H-elimination mechanism. We find that the OCN+ cross section quickly decreases above 5 eV due to OCN+ fragmentation. Figure 3b shows that the OH + CN+ and OH+ + CN products have similar threshold energies. The more endothermic CN+ is a minor product and vanishes quickly beyond 12 eV. Both CH+ + NO and CH + NO+ can be formed from [HCNO]+ dissociation, however, the cross section for the CH+ channel increases with collision energy and becomes more important than the CH + NO+ channel at high energies, implying a different reaction mechanism. This point will be considered later. Interestingly, NH+ + CO is a more exothermic channel than NH + CO+ and both ions can be produced by [HNCO]+ dissociation, but NH+ was not reported in the experiment. The

5270

J. Phys. Chem. C, Vol. 114, No. 12, 2010

smaller cross section for the NH+ product at 2 eV indicates that NH+ may not be an important product at lower energies. As shown in Figure 1 this reaction competes with that for the HCO+ channel which has a much lower energy barrier making it unfavorable under thermal conditions. This is consistent with the experimental observation that no NH+ was found under near thermal conditions. At hyperthermal energies, however, PM3 shows that the NH+ product is not negligible. III.C. Product Energy and Angular Momentum. The mechanism for ion-molecule reactions under thermal conditions usually can be interpreted as involving an ion-molecule complex in which the long-range electrostatic attraction leads to an orbiting prereaction complex by overcoming a centrifugal energy barrier. In this case the products are formed through internal energy redistribution. On the basis of this model, the efficiency of initial kinetic to complex rotational energy transfer (T-R) and the subsequent energy randomization (IVR) governs both reactivity and the product energy partitioning. Increasing the collision energy requires closer impact of the reactants in order to form a prereaction complex, which leads to a smaller cross section. At hyperthemal energies, direct reaction without forming the prereaction intermediate becomes progressively more important. Clearly, this transition of reaction mechanisms will result in different product energy partitioning. In the experiment, the TOF spectra were transformed to the product ion velocity (V′1p, along the incident ion beam direction) distribution f(V′1p). A velocity larger than that of the center-ofmass indicates forward scattering in the laboratory frame and vice versa. To compare with experiment, we calculated the velocity distribution from our simulation in which the centerof-mass velocity of the product ion was projected along the incident O atom direction and then was transformed into the laboratory coordinate system. The velocity distribution was calculated for 5 eV collision energy. The velocity component normal to the ion beam also contributes to the scattering angle distribution. We thus also calculated the angular distribution functions in the center-ofmass frame. The angular distribution is defined in terms of the normalized differential cross section (DCS, (1/σ)(dσ/d(cos (k · k′)), where σ is the integral cross section and k and k′ are the initial and final velocity unit vectors) as a function of the scattering angle cos(k · k′). Unless specified, the angular distributions at 5 and 8 eV were plotted together for comparison. HCO+(HOC+) + N/OCN+ + H. HCO+, HOC+, and OCN+ are product ions from elimination reactions so we discuss these products together in this section. For the HCO+ channel, the large amount of kinetic energy in the incident oxygen atom that is deposited into the C-O bond efficiently excites C-N bond stretching lead to N atom elimination. As shown in Figure 5, most of the energy goes into the relative translation for high energy collisions, while vibrational excitation is more important at 2 eV. The more rapid increase of the fraction in product translation beyond 5 eV indicates increased preference for energy deposition into C-N stretching motion. The HCO+ axial velocity distribution and angular distribution are plotted in Figure 6a,b, respectively. Consistent with experiment, Figure 6a shows a peak at a velocity larger than that of the center-of-mass that indicates a forward scattering mechanism at hyperthermal energies. However, a fraction of the HCO+ distribution function is produced with axial velocities smaller than the center-of-mass velocity, meaning that scattering in the backward hemisphere is non-negligible in the 5 eV trajectory simulations. This is in contrast to the experimental observation at 6.2 eV, but similar to that at 2.2 eV. As discussed in the

Sun and Schatz

Figure 5. Partitioning of the total energy in CHO+ vibration, CHO+ rotation, and CHO+-N relative translation for HCO+ + N products.

Figure 6. (a) Calculated HCO+ axial velocity distribution in laboratory reference system. The vertical line corresponds to the center-of-mass velocity in experiment with 5 eV collision energy. (b) HCO+ angular distribution functions in the center-of-mass coordinate system.

previous sections, the PM3 potential surface is too “soft” at short C-O (see Figure 4) so translation to internal energy transfer is overestimated, therefore favoring backward scattering. Nevertheless, the strong forward peak indicates that at 5 eV the direct reaction mechanism is dominant. This product scattering pattern also agrees with the angular distribution function shown in Figure 5b. The stronger scattering in the forward hemisphere at 8 eV suggests a progressively more important direct reaction mechanism with increasing collision energy. This trend also qualitatively agrees with the experimental observations. The formation of HOC+ involves O-atom insertion into the C-H bond and C-N bond breaking when eliminating the N

Reaction O+(4S) + HCN at Hyperthermal Collision Energies

J. Phys. Chem. C, Vol. 114, No. 12, 2010 5271

atom. These events require more translation-to-vibration energy transfer. Compared to HCO+, the trajectory simulation shows that HOC+ vibration picks up an additional ∼15% of the total energy. Since HOC+ is a minor product that accounts for less than 10% of the HCO+ cross section for all energies investigated, this product will not be further discussed. The OCN+ cation is produced by H-elimination. OCN+ is only important at low energies and quickly becomes negligible beyond 5 eV. The trajectory simulation shows that the total energy is predominantly partitioned into the product internal degrees of freedom with less than 18% going to relative translation at either 2 or 5 eV. Since H elimination to give the OCN+ channel also involves C-O bond formation, it is likely that the OCN+ product is also overestimated due to the excessively “soft” repulsive part of the C---O interaction. This makes OCN+ a less important product above 2 eV. NO+ + CH/NO + CH+. As described in Section IIIA, NO+ is formed with the O atom approaching the N atom side in a direction nearly normal to the HCN axis (see Figure 1a, for example). The high velocity impact of O generates a large torque and rips the N atom off from HCN+. Following this mechanism, much of the total energy is expected to be partitioned into relative translation with appreciable NO+ rotational excitation. This picture is consistent with what is shown in Figure 7, where about 50% of the total energy is deposited in relative translation and about 20% is in NO+ rotation. CH vibration and rotation receive similar amounts of energy but with a much smaller fraction than for NO+ rotation. Comparing with H abstraction in O+ + CH4 f OH+ + CH3 which is a typical H + HL f HL + H system,11 the energy partitioning for N-abstraction differs from O+ + CH4 in that there is less product translational energy and higher rotational energy partitioning, and a spectratorstripping model cannot be used. Energy disposal for the NO + CH+ channel is also included in Figure 7 for comparison. Here we see that the energy partitioning for NO+ and NO rotations, and CH+ and CH vibrations and rotations are very similar. However, a significant difference is found for relative translation and NO+ vibration with a much larger fraction of the total energy redistributed from translation to NO vibration beyond 5 eV for the NO + CH+ channel. As shown in Figure 1a, both NO+ + CH and NO + CH+ are [HCNO]+ dissociation products with the latter being a higher energy channel. On the basis of this picture, the excitation functions for the NO+ and CH+ channels should have a similar dependence on collision energy with the NO+ channel having larger cross sections. However, the CH+ product increases with collision energy and becomes more important above 10 eV. This behavior of the NO + CH+ excitation function can be explained by an exit-channel endothermic charge transfer mechanism with appropriate internal excitation. The product vibrational energy partitioning in Figure 8 and the NO/ NO+ vibrational energy distribution in Figure 8 clearly shows that there is significant NO vibrational excitation. Further analysis of the trajectories shows that NO vibration is promoted by abstraction at smaller impact parameters, for example, 1.2 Å at 5 eV. The calculated velocity and angular distributions for the NO+ + CH and NO + CH+ channels are shown Figure 9. Figure 9a shows that the NO+ product is forward scattered at 5 eV with a peak at 5500-6000 ms-1 in the laboratory reference system. There is also a small band (∼2000-3600 ms-1) in the backward scattering region. This distribution agrees with the experimental measurement at 6.2 eV very well suggesting that the PM3 Hamiltonian is valid for describing the NO+ dynamics. The

angular distribution function in Figure 9b is consistent with the velocity distribution in that the NO+ reaction is mostly forward scattered at high energies and progressively increases with increasing collision energy. As shown in Figure 9a, the NO neutral product also favors forward scattering but with a more significant intensity close to the center-of-mass velocity compared to NO+. The increased intensity around the system center-of-mass velocity indicates more efficient T-I energy transfer which is consistent with the observation that NO is more vibrationally excited than NO+. Indeed, the NO angular distribution in Figure 9c shows more backward and sideways scattering. Again, increasing the collision energy will lead to more forward scattering as shown in both Figure 9c. OH+ + CN/OH + CN+. For reasons similar to those discussed in the previous section, we discuss the OH+ + CN and OH + CN+ channels together. The energy partitioning scheme for product rotation is found to be similar for the two channels, which is similar to what we found for the NO+/NO channels, and therefore will not be further discussed. Instead, we examine the distributions of relative translation, CN/CN+ vibration, and CH/CH+ vibration, as plotted in Figure 10a. In Figure 10b, we only plotted the angular distribution for OH+/ OH at 8 eV due to the small cross section at 5 eV. For the OH+ + CN products, the total available energy is dominantly partitioned into relative translation and becomes relatively constant above 10 eV. The average impact parameter at 10 eV is 2.2 Å. The large impact parameter, forward scattering, and relatively large fraction of the product translational energy partitioning are characteristics of the high energy abstraction reaction for the H + LH f HL + H type. For the OH + CN+ channel, the largest fraction of the total energy is also channeled to relative translation for and is progressively more important with increasing collision energy. However, the fraction for CN vibration is significantly different from CN+ where CN+ receives 10∼20% more of the total energy. This vibrational excitation in the higher energy charge transfer state is exactly the same as what was found for NO + CH+ indicating that exit channel charge transfer is associated with vibrational excitation. Interestingly, the CN and NO+ products, the charge transfer counterparts to CN+ and NO, have lower vibrational frequencies than OH+ and CH, respectively. Therefore, it seems that excitation of the lower vibrational modes facilitates the resonance charge transfer process. In addition, Figure 10a shows that the fraction of CN vibration energy decreases with increasing energy suggesting less vibrational excitation. Thus exit channel charge transfer is less favored, which is again consistent with the above-described charge transfer model and the calculated CN+ excitation function in Figure 3. Considering the fact that the C-N bond has to be excited while abstraction is happening on the H atom side, it not efficient to excite the C-N bond. Therefore, the very small cross sections for CN+ formation are understandable. As shown in Figure 10b, both OH+ and OH show strong forward scattering at 8 eV indicating a direct reaction mechanism. CO+ + N + H/CO + NH+. For the CO+ channel, most of the total energy (59-69%) is imparted to product translation, while CO vibration and rotation receive similar amounts of energy. The dependence of the energy partitioning scheme on collision energy is weak and changes by about 10% from 2 to 15 eV. The velocity distributions at 5 eV are plotted in Figure 11a. The velocity distribution at 5 eV shows a peak centered at the center-of-mass velocity with a smaller shoulder in the forward scattering region. In contrast, the velocity transformed

5272

J. Phys. Chem. C, Vol. 114, No. 12, 2010

Sun and Schatz

Figure 7. The energy partitioning in product relative translation, vibration, and rotation for NO+ + CH channel (solid lines, filled symbols) and NO + CH+ channel (dashed lines, open symbols) at 5 eV collision energy.

TOF spectra at 6.2 eV in experiment shows a forward scattering peak slightly larger than the center-of-mass velocity (Figure 6b in ref 10) and has a larger intensity in the backward scattering region. Since CO+ is from the secondary dissociation of the HCO+ and OCN+ ions, we expect this disagreement is mainly from deficiencies in the PM3 Hamiltonian and will become less

significant at higher energies. Indeed, the distribution function at 8 eV, also plotted in Figure 11 a, reflects all the major features of the experimental distribution at 6.2 eV. The angular distribution functions in Figure 11b show broad scattering that is predominantly sideways at 5 eV and then forward at 8 eV with only a small portion of backward scattering. These results are

Reaction O+(4S) + HCN at Hyperthermal Collision Energies

J. Phys. Chem. C, Vol. 114, No. 12, 2010 5273

Figure 8. NO+ and NO vibrational energy distribution at 5 eV for NO+ + CH and NO + CH+ channels, respectively.

Figure 10. (a) Product energy partitioning for OH+ + CN channel (solid line, filled symbols) and OH + CN+ channel (dashed line, open symbols) at 5 eV, respectively. Not included in the figure are the product rotational energy partitionings. (b) Angular distribution expressed as normalized differential cross section for OH+ and OH products at 5 eV.

orbital angular momentum transfer to CO rotation. Indeed trajectory animations show that this correlates with a decreasing O-C-N incident angle from 93 to 50° when the collision energy goes from 5 to 15 eV. The mechanism in which the dissociating H atom swings toward the departing N atom and forms NH+ was found to be still valid at these energies IV. Summary

Figure 9. (a) Calculated NO+ and NO axial velocity distribution in laboratory reference system. The vertical lines correspond to the centerof-mass velocity in the experiment at 5 and 8 eV collision energy. (b) Angular distribution expressed as a normalized differential cross section for NO+ product at 5 and 8 eV, respectively. (c) Angular distribution expressed as normalized differential cross section for NO product at 5 and 8 eV, respectively.

consistent with what is shown in Figure 11a and also agrees with experiment qualitatively. Unlike the OH+ + CN/OH + CN+ and NO+ +CH/NO + CH+ channels that can be correlated through charge transfer processes, the CO neutral molecule is generated with quite a different mechanism at high energies from that for CO+. As shown in Figure 12, about one-third of the total energy goes to translation and about a quarter goes as the NH+ internal energy. Interestingly, the CO rotational energy quickly rises when the collision energy is above 5 eV, while the energy fraction in CO vibration decreases correspondingly. This suggests large

We have performed direct dynamics simulations of O+ + HCN collisions on the ground electronic quartet potential energy surface with the PM3 Hamiltonian. The quality of the PM3 potential surface was calibrated through comparisons with B3LYP density functional and MP2 methods and with experiment. We found that the PM3 reaction energies show overall qualitative agreement with experiment. But, compared to higher level MP2 calculations, the PM3 energies are in most cases lower by about 1 eV for transition states and reaction intermediates, although most of the reaction paths are the same. Despite the large difference between the PM3 and MP2 potential surfaces, we find that the PM3 direct dynamics results for hyperthermal energies are in at least qualitative agreement with experiment, based on comparisons with reaction cross sections, and product velocity and angular distributions. The trajectory results showed rich chemistry and complicated dynamics for O+ + HCN reactions at hyperthermal energies. In agreement with experiment HCO+, NO+, and CO+ cations with m/e ratios ranging from 28-30 were found in the simulations. In addition, other ions, NH+, CH+, OH+ (13-17u), CN+ (26u), and OCN+ (42u), which are accessible both kinetically and energetically from 2-15 eV, were also found

5274

J. Phys. Chem. C, Vol. 114, No. 12, 2010

Figure 11. (a) Calculated CO+ axial velocity distribution at 5 and 8 eV in laboratory reference system. The vertical lines mark the centerof-mass velocities in experiment. (b) Angular distribution expressed as normalized differential cross section for CO+ product at 5 and 8 eV, respectively.

Sun and Schatz Å2). Lighter ions, NH+, CH+, OH+ (13-17u), however, have appreciable cross sections at high energies. The overall shape of the calculated excitation functions for HCO+/HOC+, NO+, and CO+ agrees with experiment especially at energies higher than 5 eV and nearly quantitative agreement is achieved for NO+ which suggests that the dynamics can largely be characterized using the ground state quartet potential surface. The HCO+/HOC+ and CO+ cross sections are consistently overestimated by the simulation that we conclude is an artifact due to deficiency of the PM3 Hamiltonian in repulsive portions of the potential energy surface. Although the high energy O+ + HCN collisions generally involve direct collisions, many different reaction mechanisms are possible, leading to different reactivity, energy transfer and partitioning even for the same type of reactions. The elimination mechanisms and energy transfer are distinct for HCO+, HOC+, and OCN+. Both NO+ + CH and OH+ + CN are abstraction reactions with large impact parameter, but the energy partitioning and product angular distributions are different. Exit channel charge transfer was found for the NO + CH+ and OH + CN+ products that is facilitated by vibrational excitation. Despite the qualitative agreement with experiment for product ions, cross sections, and angular distributions, quantitative comparisons for the HCO+/CO+ products show inaccuracies due to the quality of the PM3 surface. Thus dynamics simulations on a more accurate potential surface are desirable in future work to achieve a quantitative description of the reaction dynamics. Recent dynamics simulations have shown that the specific reaction parameters (SRP) approach for improving the semiempirical Hamiltonian is a promising direction both for accuracy and computational efficiency requirements.32-35 In addition, further experimental work will be extremely valuable in confirming predictions about the ion products and dynamics found in current study. Acknowledgment. This work was supported by AFOSR Grant FA955D-07-1-0095 References and Notes

Figure 12. (a) Product energy partitioning for CO + NH+ channel. (b) Angular distribution expressed as normalized differential cross section for CO product at 5 and 8 eV, respectively.

in the simulations. The heavy ion OCN+ is only stable in the 2-5 eV range. Although CN+ has a m/e ratio close to HCO+, NO+, and CO+, it is a negligible product (cross section