Using JCP format - KU Leuven

3 downloads 0 Views 218KB Size Report
May 8, 2001 - The cyclic anionic species constitutes a weak complex between N2 and O ... precursor to produce nitric oxide anion in its triplet ground.
JOURNAL OF CHEMICAL PHYSICS

VOLUME 114, NUMBER 18

8 MAY 2001

Another look at the electron attachment to nitrous oxide Eugene S. Kryachko, Chris Vinckier, and Minh Tho Nguyena) Department of Chemistry, University of Leuven, Celestijnenlaan 200F, B-3001 Leuven, Belgium

共Received 6 December 2000; accepted 22 February 2001兲 Ab initio molecular orbital calculations up to the coupled-cluster level with the aug-cc-pVQZ basis set allowed us to have a new look at the electron affinity of nitrous oxide (N2O) resulting in a detection of a new N2O⫺ entity, and thereby a novel mechanism for the dissociative electron attachment process, N2O⫹e ⫺ →N2⫹O⫺. Addition of an electron to the linear N2O ground state (X 1 ⌺ ⫹ ) leads first to an open-chain bound anion which lies 25 kJ/mol above the neutral. Upon a cyclization of the open anion with an additional energy barrier of 25 kJ/mol, a cyclic anionic species is formed which is more stable than the open isomer and lies now, at most, 3 kJ/mol above the neutral ground state 共the transition structure for cyclization being 50 kJ/mol above neutral N2O). The cyclic anionic species constitutes a weak complex between N2 and O⫺ characterized by a binding energy of only 16 kJ/mol. The electronic structure of the anion complex is analyzed, a number of earlier experimental results are clarified and a resolution for the long-standing disagreement between experiment and theory around the electron affinity of N2O is proposed. © 2001 American Institute of Physics. 关DOI: 10.1063/1.1364679兴

I. INTRODUCTION

shown to generate both heterogeneous and homogeneous cluster anions.19 The catalytic decomposition of nitrous oxide on various oxide surfaces such as Mo/SiO2 has also been shown to produce adsorbed oxygen anion via electron transfer.20 The authors of this study stated that a knowledge of the N2O⫺ potential energy surface and its relationship to that for N2O is highly demanded to understand the mechanism of the electron transfer induced processes. A common and major difficulty encountered in a quantitative treatment of such reactions and processes was that it boiled down to the value of the electron affinity of N2O, which turns out to be poorly established. Let us summarize the main reported results that disagreed with each other on this quantity. Experimental evaluations in the early 1970s from collisional dissociation or collisional ionization spectra suggested a negative value with a lower limit of ⫺0.15 ⫾0.1 eV for the electron affinity.21 The positive value of EA共N2O兲⫽0.22 eV currently recommended in, among other compilations, the CRC-Handbook22 was taken from a 1976 evaluation by Hopper et al.23 which was based on results of molecular beam experiments using a charge transfer technique. A negative vertical electron affinity EAvert共N2O兲 ⫽⫺2.23 eV and a dissociation limit of 0.42 eV of N2O⫺ relative to N2⫹O⫺ were also derived in this study. This implies that the anionic system gains a stabilization of 2.45 eV following geometry relaxation. The latter authors23 also carried out ab initio computations at the MCSCF–Cl level of MO theory using a double-zeta plus diffuse functions basis set to determine the geometry of the N2O⫺ anion and thereby the adiabatic EA共N2O兲. Nevertheless, these calculations, quite extensive by the standard of that time, indicated that both vertical and adiabatic electron affinities of the linear molecule N2O (X 1 ⌺ ⫹ ) are negative by ⫺4.1 and ⫺1.7 eV, respectively, even though the difference between both vertical and adiabatic values seem to be comparable with the

Reactions of nitrous oxide (N2O) have generated in the past considerable experimental and theoretical interest. This largely concerns its reactions with metal atoms 共M兲 in the gas phase. Numerous studies in flames1 and molecular beam experiments2,3 have shown that, due to the high exothermicity of these reactions, metal oxides were formed at various levels of electronic excitation. In view of the efficiency of these reactions, the addition of metals to combustion systems has been recently considered as a possible way of reducing the emission of the greenhouse N2O gas.4 The chemistry of nitrous oxide seems to provide an important way of controlling the ozone concentration in the stratospheric ozone layers.5 Studies of the chemical dynamics of the M⫹N2O reactions remain a challenging issue due to the possible transfer of a negative charge from the metal atom towards the N2O molecule. In fact, it was assumed that these reactions proceed along the formation of a certain ionic transition state (M⫹N2O⫺). 6,7 Empirical correlations linking the observed reaction rate constants and activation energies with the ionization energies of the metal atoms and the electron affinity 共EA兲 of N2O have been tested extensively.8 The processes involved in the direct and dissociative attachment of an electron to the nitrous oxide molecule have been widely investigated in the past.9–15 In addition, these reactions have often been used in mass spectrometry as a precursor to produce nitric oxide anion in its triplet ground state.16–18 Formation of the atomic oxygen anion has been postulated as an intermediate step in this procedure, namely: 共i兲 N2O⫹e ⫺ →N2⫹O⫺, 共ii兲 O⫺⫹N2O→3NO⫺⫹NO. Electron attachment to clusters of nitrous oxide was recently a兲

Electronic mail: [email protected]

0021-9606/2001/114(18)/7911/7/$18.00

7911

© 2001 American Institute of Physics

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

7912

J. Chem. Phys., Vol. 114, No. 18, 8 May 2001

experimental estimate. According to these calculations, the N2O⫺ anion has a bent form ( 2 A ⬘ ) with a NNO bond angle of 132.7° and the two N–O and N–N distances are significantly stretched as compared with the corresponding ones in the neutral species. In a 1983 theoretical study, Yarkony24 reported that the 2 A ⬘ state of N2O⫺ is actually stable by 0.39 eV with respect to N2O (X 1 ⌺ ⫹ )⫹e ⫺ . Such a positive EA came no doubt from the fact that only Hartree–Fock calculations with a double-zeta basis set were used in Ref. 24 for both N2O and N2O⫺ species. Hopper et al.23 already noted the fortuitous results of the SCF treatment which gave in this case positive electron affinities. The electron affinity of N2O was further evaluated in three theoretical works.25–27 Yu et al.25 considered in 1992 this quantity for a series of triatomic species and paid a special attention to the N2O/N2O⫺ case. Electronic energies of ions and neutrals were calculated by the G2 approach28 using geometry optimization at the 共U兲MP2/6-31G共d兲 level. With the N–N and N–O bond lengths of 1.128 and 1.674 Å, respectively, and a NNO bond angle of 134.6°, a negative value of ⫺0.13 eV for the adiabatic electron affinity EA共N2O兲 was derived. The N–O distance of 1.674 Å, which is rather long with respect to standard N–O distances, was no doubt due to the absence of diffuse functions in the basis sets inherent in the geometry optimization step of the G2 procedure. In 1997, a density functional theory 共DFT兲 study using different functions by Tschumper and Schaefer26 indicated that nitrous oxide has positive electron affinity ranging from ⫹0.3 to ⫹0.8 eV. Nevertheless, the negative EA value was recently confirmed in a 1999 study by McCarthy et al.27 using essentially quadratic configuration interaction QCISD and QCISD共T兲 methods with aug-cc-pVDZ and aug-ccpVTZ basis sets.29,30 Indeed, these calculations led to values for EA共N2O兲 in the range of ⫺0.0205 and ⫺0.149 eV. Yu et al.25 also attempted to figure out the reason for the disagreement between experiment and theory and suggested that it apparently came from the presence of an energy barrier of about 0.15 eV to the decomposition of N2O⫺ into N2⫹O⫺ which was not taken into account by Hopper et al.23 in the interpretation of their experimental results for the threshold energy of the collisionally induced reactions of the anions. In other words, it was related to the actual value for the 关 NN–O兴 ⫺ dissociation energy which was used in the experimental evaluation of electron affinity. In view of such a long-standing controversy on a fundamental quantity, EA共N2O兲, we set out to take a new look at the N2O⫺ paradigm by reinvestigating its potential energy surface using higher level ab initio MO calculations. This led us to discovering a novel structure of N2O⫺ which could shed some light into the mechanism of the dissociative electron attachment to nitrous oxide. II. DETAILS OF CALCULATION

Ab initio calculations were carried out with the aid of the set of programs.31 The potential energy surface was initially explored using the unrestricted second-order Møller–Plesset perturbation theory 共UMP2兲 and the 6-311 ⫹G共d兲 basis set. The harmonic vibrational frequencies were subsequently calculated at this level in order to characterize GAUSSIAN 94

Kryachko, Vinckier, and Nguyen

FIG. 1. Optimized geometries of the N2O⫺ anionic system at two levels of theory: upper values, UMP2/6-13⫹G共d兲; and lower values, UQCISD共T兲/ aug-cc-pVDZ. Bond lengths are given in angstrom and bond angles in degrees.

the stationary points and to determine their zero-point energies 共ZPE兲. The geometries of the relevant structures were further refined using the quadratic configuration interaction methods including single and double excitations plus a perturbative correction for the triple excitations, QCISD共T兲, in conjunction with the correlation consistent aug-cc-pVDZ basis set.29 This level was previously employed in Ref. 27 for geometry optimizations. The relative energies were further refined by single-point electronic energy computations at QCISD共T兲/aug-cc-pVDZ geometries making use of the coupled-cluster theory CCSD共T兲 with the larger aug-ccpVTZ and aug-cc-pVQZ basis functions. The spin contamination in the unrestricted Hartree–Fock references of the doublet anionic systems is not large. The expectation values of the 具 S 2 典 operators on UHF wave functions are not larger than 0.78. While the geometrical parameters of anionic structures are displayed in Fig. 1, calculated relative energies are summarized in Table I. Figure 2 shows a schematic potential energy diagram illustrating, among other things, the dissociative electron attachment to nitrous oxide. In general, the results obtained by CCSD共T兲 with both basis sets are similar; except for one case, the variations in both relative energies are within 3 kJ/mol. Unless otherwise noted, we use in what follows the values obtained from CCSD共T兲/aug-cc-

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

Electron attachment to N2O

J. Chem. Phys., Vol. 114, No. 18, 8 May 2001

7913

TABLE I. Relative energiesa between different neutral N2O and anionic N2O⫺ structures calculated at various levels of theory.b Structure

I

II

III

IV

V

VI

N1 N2 NTS N2¿O(3P) N2¿O(1D) A1 A2 ATS ATS* N2¿OÀ

0 296 398 173 457 107 35 149 50 51

0 301 405 201 478 112 48 159 64 65

0 305 409 211 483 111 51 162 67 67

0 253 319 120 339 21 ⫺12 36 0 4

0 258 330 145 358 26 0 47 12 17

0 262 335 158 364 25 3 50 14 19

a

Relative energies given in kJ/mol, including zero-point vibrational energies 共ZPE兲, were based on QCISD共T兲/aug-cc-pVDZ optimized geometries. b The levels are I, MP2/aug-cc-pVDZ; II, MP2/aug-cc-pVTZ; III, MP2/augcc-pVQZ; IV, QCISD共T兲/aug-cc-pVDZ; V, CCSD共T兲/aug-cc-pVTZ; and VI, CCSD共T兲/aug-cc-pVQZ.

pVQZ⫹ZPE computations 共Fig. 2兲. To further improve the calculated electron affinity of nitrous oxide, a set of diffuse sp functions has also been added into the aug-cc-pVQZ basis set whose details will be given in a following section. As for a convention, the electron affinity, ionization energy and other transition energies are given hereafter in eV whereas the relative energies between different points on the energy surface are expressed in kJ/mol. III. RESULTS AND DISCUSSION

Geometries and energies of nitrous oxide (N2O) have been the subject of a large body of experimental and theoretical studies. For a discussion of earlier results we would mention a recent paper by Hwang and Mebel 共referred to hereafter as HM兲32 which reported a detailed ab initio study on the dissociation mechanism of the neutral N2O in both lowest-lying singlet and triplet states. As for a calibration, the accuracy of the results obtained in this work for some well-established quantities such as the binding energy and ionization energy of N2O, was evaluated. Because the dissociative electron attachment involves the oxygen anion, let us first consider the accuracy of the electron affinity of the oxygen atom. The present calculated value EA(O, 3 P)⫽1.40 eV can be compared with the experimental estimate of 1.46 eV.33 It is known that its further improvement requires a substantial extension of the oneparticle basis functions.34 Regarding the O( 1 D)←O( 3 P) energy separation, the present value of 2.19 eV is quite close to the HM value of 2.22 eV32 but differs by 0.2 eV from the available experimental result of 1.958 eV.35 The NN–O dissociation energy of the nitrous oxide N1 corresponds to the heat of the spin-forbidden reaction NNO 3 ( 1 ⌺ ⫹ )→N2( 1 ⌺ ⫹ g )⫹O( P) whose formal connection of N1 and the dissociation asymptote is shown in Fig. 2. For a detailed discussion of the intersystem crossing between both the singlet and triplet states of the neutral nitrous oxide, we refer to the paper by HM.32 Use of the available experimental heats of formation 共at 0 K兲22,36 of 246 kJ/mol for the oxygen atom and 85 kJ/mol for N2O leads to an experimental value of 161 kJ/mol. Our calculated value of 158 kJ/mol is in

FIG. 2. Schematic energy profiles illustrating the rearrangement and dissociation pathways of both neutral N2O and anionic N2O⫺ systems. Relative energies are obtained from 共U兲CCSD共T兲/aug-cc-pVQZ⫹ZPE based on 共U兲QCISD共T兲/aug-cc-pVDZ optimized geometries. The zero-point energies were evaluated from 共U兲MP2/6-31⫹G共d兲 harmonic vibrational frequencies and uniformly scaled down by 0.95.

a somewhat better agreement with the experimental one than the value of 155 kJ/mol derived by HM.32 With respect to the N2⫹O( 1 D) limit, the NN–O dissociation energy of nitrous oxide N1 is calculated equal to 364 kJ/mol. The first adiabatic ionization energy of nitrous oxide, IEa 共N2O兲 ⫽12.87 eV, derived from our computations is quite close to the experimental one of 12.89 eV.36 Nitrous oxide is known to have a linear structure N1 in its global energy minimum. Besides N1, there is a higherlying cyclic isomer N2 that is placed, according to our estimates, by 262 kJ/mol above the ground state linear N2O. Both the linear N1 and the cyclic N2 isomers are separated from each other by an energy barrier of 335 kJ/mol relative to the linear form. The latter values are comparable to those of 269 and 335 kJ/mol, respectively, reported by HM.32 The cyclic isomer is connected to the N2⫹O( 1 D) dissociation asymptote without requiring an additional activation energy. It is thus remarkable that the transition structure NTS linking both isomers N1 and N2 of nitrous oxide and shown in Fig. 2 actually lies about 29 kJ/mol below the dissociation limit N2⫹O( 1 D). This implies that, in the absence of a singlet– triplet intersystem crossing,32,37 an oxygen elimination from

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

7914

J. Chem. Phys., Vol. 114, No. 18, 8 May 2001

the linear nitrous oxide N1 eventually occurs throughout the intermediacy of the cyclic N2. We also note that the dipole moments computed at the HF/aug-cc-pVQZ level for both isomers are the following: 0.7 D for N1 and 1.3 D for N2. Due to the large variation in geometries of both neutral and anionic equilibrium structures, the difference between both vertical and adiabatic electron affinities of N2O is expected to be large. An experimental estimate of the vertical detachment energy of about 1.5 eV was derived from the maximum position in the photoelectron spectrum.38 Overall, the present results obtained using the CCSD共T兲/aug-cc-pVQZ⫹ZPE level of theory compare quite well with either the established experimental estimates or the most recent theoretical ones. They differ from each other by an average deviation of, at most, ⫾5 kJ/mol. We now turn to the anionic system whose main features are summarized in Figs. 1 and 2. Starting from the linear neutral form N1, an electron attachment leads to the open

Kryachko, Vinckier, and Nguyen

anion structure A1 which is characterized by a bent shape (NNO⫽130° – 140°, depending on the method employed兲 and a stretched N–O distance 共up to 0.12 Å relative to that in N1兲. This anion has been located in the earlier theoretical studies.25–27 Our best estimates 共cf. Fig. 2兲 place the anion A1 by 25 kJ/mol above the neutral form N1. While this is qualitatively in line with the most recent evaluation of McCarthy et al.27 suggesting a negative electron affinity, the absolute value somewhat differs from their estimate of ⫺14 kJ/mol. Similar to the neutral system, we have also located the cyclic anion structure A2 which has not been found in earlier studies. It is characterized by the following properties. First of all, by a rather long N–O distances of 2.7–2.8 Å 共Fig. 1兲 which indicates that A2 can be regarded as a transient complex between the molecular nitrogen and the oxygen anion. A Mulliken population analysis based on UHF/aug-cc-pVQZ wave functions shows in fact that the negative charge almost

FIG. 3. 共Color兲 The shape 共contour spacing of 0.07兲 of the highest occupied 共HOMO兲, lowest unoccupied 共LUMO兲 orbitals of the cyclic nitrous oxide N2 and the singly occupied 共SOMO兲, HOMO, HOMO-1, and LUMO of the cyclic anion A2 using UHF/aug-cc-pVQZ wave functions. ⑀ stands for orbital energy.

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 114, No. 18, 8 May 2001

FIG. 3. 共Continued.兲

entirely resides on the oxygen atom 关 q共O兲⫽⫺0.95 e 兴 . For the sake of comparison, Fig. 3 displays the singly occupied 共SOMO兲, highest occupied 共HOMO and HOMO-1兲, and lowest unoccupied 共LUMO兲 orbitals of the cyclic form in both the neutral and anionic states. It can be seen that while the LUMO of N2O(N1) is the ␲ * 共NN兲 orbital, the SOMO共N2O⫺兲 becomes the in-plane orbital centered at the oxygen atom. This implies that the electronic relaxation and reorganization following the electron attachment is important in this system. The cyclic anion A2 turns out to be more stable than the bent-shaped counterpart A1 by 22 kJ/mol lying now only marginally, 3 kJ/mol, above the linear neutral from N1. Due to the fact that A2 is more stable than A1, it seems legitimate to consider the former as the adiabatic anion. In this view, the electron affinity of nitrous oxide evaluated as the difference between the total energies of A2 and N1 amounts to only EA共N2O兲⫽⫺0.03 eV at the CCSD共T兲/aug-cc-pVQZ ⫹ZPE level. It is well known that the accuracy of calculated electron affinities is strongly dependent on the massive presence of diffuse functions in the atomic basis set. The diffuse charac-

Electron attachment to N2O

7915

ter of the orbital describing a loosely bound electron necessitates the use of additional diffuse functions having very low orbital exponents. Such basis set are more able not only to describe the static charge distribution of the neutral host but also to allow for the polarization and dispersion stabilization of the anion. To select the orbital exponents of the additional diffuse functions, we followed the procedure recently suggested by Simons and co-workers.39,40 Overall, each of the aug-cc-pVTZ and aug-cc-pVQZ basis for every atom was supplemented with a set of two s and two p diffuse functions. However there are two main differences: first, we achieved the lowest exponent at the level of 10⫺3 a.u. for each symmetry instead of that of 10⫺5 a.u. as suggested by Simons et al.39,40 Second, the extra diffuse set was added to every atom of the molecule rather than just one set for the whole molecule. Using both resulting aug-cc-pVTZ⫹2s2p and aug-cc-pVQZ⫹2s2 p basis sets, the electron affinity of nitrous oxide derived from coupled-cluster CCSD共T兲 calculations was lowered by 0.5 kJ/mol 共0.005 eV兲 relative to the corresponding values obtained without additional diffuse functions. Spin-contamination effect can certainly be ruled out because the expectation value for 具 S2 典 of A2 is 0.7504. We note that in some of their studies, Simons et al.39,40 used the smaller aug-cc-pVDZ basis set. Therefore the lowering became somewhat larger. It appears that the polarization and diffuse spaces in the aug-cc-pVTZ and aug-cc-pVQZ sets are already large enough and from there the convergence on the calculated electron affinity becomes extremely slow. A similar observation was previously made for the electron affinity of the oxygen atom.34 It can be expected that further extensions of the polarization and diffuse spaces in the atomic functions could result in a lowering of the electron affinity under focus, but the resulting lowering could hardly be larger than 1 or 2 kJ/mol. In other words, the calculated electron affinity of nitrous oxide remains marginally negative, EA共N2O兲⫽⫺0.01 eV which is far smaller than the expected margin error of the computational methods employed here which is up to ⫾0.1 eV. Such a ‘‘too close to call’’ situation does not allow us to definitely rule out the possibility of a small but nevertheless positive electron affinity of nitrous oxide. However, even in the case of a positive EA, the available experimental positive electron affinity of 0.22 eV 共Ref. 23兲 is, in our opinion, overestimated. It is worth noticing that the density functional theory with the currently popular functionals such as the BLYP, BP86, B3LYP, B3P8, etc., tends to overestimate the anion stability resulting in the large positive electron affinity for nitrous oxide.26 The cyclic anion A2 is rather a weak complex having the binding energy of only 16 kJ/mol relative to the N2⫹O⫺ fragments. Both the bent and cyclic anionic forms are connected with each other by the transition structure ATS which is placed by 25 and 50 kJ/mol above the anion A1 and the neutral N1, respectively. The structure ATS shows the open bonding angle of 130°–135° and it is also characterized by the stretched N–O distance equal to 1.6–1.7 Å. Although the ATS transition structure has earlier been located by Yu et al. in their theoretical study25 共being placed by 31 and 15 kJ/mol above A1 and the N2⫹O⫺ fragments, respectively, at the G2

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

7916

J. Chem. Phys., Vol. 114, No. 18, 8 May 2001

level兲, it has in contrast been interpreted as the transition structure governing the dissociation of the open anion A1 rather than the structure connecting the two lowest-energy anionic forms. An analysis of the vibrational mode associated with the imaginary frequency of ATS confirms that it is a transition structure which links both the linear A1 and cyclic A2 isomers. As stated above, the cyclic complex A2 has not been found in Ref. 25. These authors suggested in addition that the position of ATS should be taken into account in evaluating the dissociation energy of A1 and in reinterpreting the experimental results of Ref. 23. With respect to the cyclic anion A2, the cyclic neutral N2 is associated with a large electron affinity EA共cyclic-N2O兲⫽2.68 eV. The linear structure ATS* shown in Figs. 1 and 2 and featuring a long N–O distance is characterized by a degenerate negative frequency. The normal vibrational mode of each of these two frequencies corresponds to a cyclization yielding the cyclic anionic form A2. It is obvious that such a motion from a linear form is possible in two distinct directions. ATS* is calculated to lie by 14 kJ/mol above N1, by 11 kJ/mol above A2 but only by 5 kJ/mol below the dissociated fragments N2⫹O⫺. Proceeding in the opposite direction starting from the fragments N2⫹O⫺, ATS* represents somehow a ‘‘bump’’ point on the potential energy surface which connects the fragments at the early stage of the condensation and when it is passed along the reaction coordinate, a bifurcation of the reaction routes occurs leading both to the cyclic anion A2. The section of the potential energy surface schematically illustrated in Fig. 2 suggests that following the electron attachment, the nitrous oxide system undergoes a rather facile dissociation process throughout the cyclic isomer to the final fragments, N1→A1→ATS→A2→N2¿OÀ, and characterized by a threshold energy of 50 kJ/mol achieved at ATS. This reaction path thus constitutes the new mechanism of the dissociative electron attachment reaction N2O⫹e ⫺ →N2⫹O⫺. It should also be pointed out that this transition structure ATS with the threshold of 50 kJ/mol qualitatively explains the activation energy of 43.5⫾1.7 kJ/mol observed for this reaction.11 In a more recent experimental study41 on the same dissociative process using crossed electron/ molecular beams in conjunction with quadrupole mass spectrometric techniques at different temperatures, a first threshold for O⫺ formation was seen at 53 kJ/mol (0.55 ⫾0.1 eV). This was assigned to an electron attachment to the linear 2 ⌺ state of N2O⫺. However, our calculated results shown in Fig. 2 rather indicate that this threshold is related to the energy barrier at the energy level of the transition structure ATS. A second energy threshold for O⫺ formation was detected at 231 kJ/mol 共2.4 eV兲41 and tentatively assigned to the 2 ⌸ state of the anion. In spite of several attempts, we were not successful in locating a structure corresponding to this energy threshold. Due to the low energy threshold at ATS, the dissociation is likely to be a facile process in many experimental gas phase conditions, especially in low pressure experiments. Of course, the lifetime of the nitrous oxide anion depends not only on the position of ATS but also on how its higher vibrational levels become populated at higher

Kryachko, Vinckier, and Nguyen

temperatures.27 It was also noted from ion–molecule reaction experiments that the thermal detachment reaction O⫺⫹N2→N2O⫹e ⫺ has a very low rate constant at room temperature with an upper limit of ⬍1⫻10⫺12 cm3/molecule•s. 14 This could be explained at that time by adopting the experimentally derived value of 1.2 eV10 for the D0共N2 –O兲 dissociation energy resulting in an endothermic detachment reaction. Our present calculations do not only confirm the recently accepted value of 1.6 eV for the D0共N2O兲 dissociation energy yielding on the contrary a reaction exothermicity of 19 kJ/mol, but also explain the very low value of the rate constant at 300 K by the presence of an energy barrier of 31 kJ/mol. We also note that in the transformation of N2O as a ligand at the Mo/SiO2 surfaces into the adsorbed oxygen anion, an activation energy of 20⫾4 kJ/mol has been measured in the 323–393 K range.20 Thus, it is apparent that the Mo/SiO2 surfaces exert a catalytic effect in reducing the activation energy of the N2O⫹e ⫺ →N2⫹O⫺ process. Having confirmed the discrepancy between experiment and theory, we now attempt to explain the reason for it. In their study, Hopper et al.23 used the following thermodynamic cycle to evaluate the adiabatic electron affinity of N2O: EA共N2O兲⫽⫺D 0 共NN–O兲⫹EA共O兲⫹D 0 共 NN⫺O⫺ 兲 in which the dissociation energy D 0 共NN–O兲⫽1.68 eV, the electron affinity of the oxygen atom EA共O兲⫽1.47 eV and the dissociation energy D 0 共NN–O⫺ )⫽0.43⫾0.1 eV. While the first two quantities are well established as shown above: the last one was determined by these authors from fitting of the experimental reaction cross sections derived in their mass spectrometric experiments23 and using some assumptions in the interpretation of data that we could not usefully comment on. The authors however admitted that the determination of D 0 共NN–O⫺兲 was made with the largest uncertainties. A comparison with our results displayed in Fig. 2 clearly shows that the latter quantity was overestimated by up to 0.2 eV. This in turn overestimates the stability of the anion N2O⫺ by placing it well below the neutral counterpart and, as a consequence, implying a positive electron affinity for nitrous oxide. It is beyond any doubt that the long-standing discrepancy on the electron affinity of nitrous oxide likely arises from an ill determination of the dissociation energy D 0 共NN–O⫺兲. It is also obvious that an overestimation by 0.2 eV on the dissociation energy due to a rather complicated method for its determination which used mass spectrometric techniques, was quite reasonable in the 1970s and well in the range of expected error bars. Nevertheless, in the present case which involved a small absolute value, it was large enough to cause a qualitative change and thereby an intriguing disagreement. IV. CONCLUDING REMARKS

In the present theoretical study, we have revisited the process of the electron attachment to nitrous oxide. We have found for the first time the cyclic anionic structure which is more stable than the bent-shaped isomer and which is likely to be involved in the dissociative attachment mechanism.

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

Electron attachment to N2O

J. Chem. Phys., Vol. 114, No. 18, 8 May 2001

This leads to a new value for the electron affinity, EA共N2O兲⫽⫺0.03⫾0.1 eV. We have also identified the reason for the apparent discrepancy between theory and experiment on the quantity under consideration. This arises from an overestimation by about 0.2 eV of the dissociation energy of the transient anion. Such a deviation is not unreasonable due to the complicated character of the experiment. However, it is large enough to cause a perplexing and longstanding disagreement. We note that even nowadays, a similar discrepancy is still frequently occurred. For example, a recent negative ion photoelectron spectroscopic study42 suggested a triplet ground state for di-iodo-carbene (Cl2) whereas the high level ab initio calculations43,44 consistently indicated a large triplet–singlet energy gap in favor of the singlet state. This case illustrates again the need to put a further effort on both experimental and theoretical sides, better in concert, in the determination of thermochemical parameters of unstable species. ACKNOWLEDGMENTS

The authors thank the Fund for Scientific Research FWO-Vlaanderen and the KU-Leuven Research Council for financial support. 1

O. E. Kashireninov, G. B. Manelis, and R. F. Repka, Russ. J. Phys. Chem. 56, 630 共1982兲. 2 J. W. Cox and P. J. Dagdigian, J. Phys. Chem. 86, 3738 共1982兲. 3 C. D. Jonah, R. N. Zare, and C. Ottinger, J. Chem. Phys. 56, 263 共1972兲. 4 R. A. Perry and J. A. Miller, Int. J. Chem. Kinet. 28, 217 共1996兲. 5 W. Seiler and R. Conrad, J. Air Pollut. Control Assn. 31, 767 共1981兲. 6 D. Ritter and J. C. Weisshaar, J. Phys. Chem. 94, 4907 共1990兲. 7 J. M. J. Plane, C. F. Nien, and B. Rajasekhar, J. Phys. Chem. 96, 1296 共1992兲. 8 D. P. Belyung, P. M. Futerko, and A. Fontijn, J. Chem. Phys. 102, 155 共1995兲. 9 P. Chantry, J. Chem. Phys. 51, 3369 共1969兲; 51, 3380 共1969兲. 10 G. J. Schulz, J. Chem. Phys. 34, 1778 共1961兲. 11 W. E. Wentworth, E. Chen, and R. Freeman, J. Chem. Phys. 55, 2075 共1971兲. 12 E. E. Ferguson, F. C. Fehsenfeld, and A. G. Schmeltekopf, J. Chem. Phys. 47, 3085 共1967兲. 13 J. F. Paulson, Adv. Chem. Ser. 58, 28 共1966兲. 14 F. C. Fehsenfeld, E. E. Ferguson, and A. L. Schmeltekopf, J. Chem. Phys. 45, 1844 共1966兲.

7917

J. N. Bardsley, J. Chem. Phys. 51, 3384 共1969兲. F. Bru¨ning, S. Matejcik, I. Illenberger, Y. Chu, G. Senn, D. Muigg, G. Denifl, and T. D. Ma¨rk, Chem. Phys. Lett. 292, 177 共1998兲. 17 M. J. Travers, D. C. Cowles, and G. B. Ellision, Chem. Phys. Lett. 164, 449 共1989兲. 18 G. A. Janaway and J. A. Brauman, J. Phys. Chem. 104, 1795 共2000兲. 19 E. Leber, S. Barsotti, J. Bommels, J. M. Weber, I. I. Fabrikant, M.-W. Ruf, and H. Hotop, Chem. Phys. Lett. 325, 345 共2000兲. 20 Z. Sojka and M. Che, J. Phys. Chem. 100, 14 776 共1996兲. 21 S. J. Nalley, R. N. Compton, H. C. Schweinler, and V. E. Anderson, J. Chem. Phys. 59, 4125 共1973兲, and references therein. 22 CRC-Handbook of Chemistry and Physics, Web Edition 1999, 10–150. 23 D. G. Hopper, A. C. Wahl, R. L. C. Wu, and T. O. Tiernan, J. Chem. Phys. 65, 5474 共1976兲. 24 D. R. Yarkony, J. Chem. Phys. 78, 6763 共1983兲. 25 D. Yu, A. Rauk, and D. A. Armstrong, J. Phys. Chem. 96, 6013 共1992兲. 26 G. S. Tschumper and H. F. Schaefer, J. Chem. Phys. 107, 2529 共1997兲. 27 M. C. McCarthy, J. W. R. Allington, and K. O. Sullivan, Mol. Phys. 96, 1735 共1999兲. 28 L. A. Curtiss, K. Raghavachari, G. W. Trucks, and J. A. Pople, J. Chem. Phys. 94, 7221 共1991兲. 29 T. H. Dunning, Jr., J. Chem. Phys. 90, 1007 共1989兲. 30 R. A. Kendall, T. H. Dunning, Jr., and R. J. Harrisson, J. Chem. Phys. 96, 6796 共1992兲. 31 M. J. Frisch, G. W. Trucks, H. B. Schlegel et al., GAUSSIAN 94, Revision B.2, Gaussian, Inc., Pittsburgh, PA, 1995. 32 D. Y. Hwang and A. M. Mebel, Chem. Phys. 259, 89 共2000兲. 33 D. M. Neumark, K. R. Lykke, T. Anderson, and W. C. Lineberger, Phys. Rev. A 32, 1890 共1985兲. 34 G. De Oliveira, J. M. L. Martin, F. De Proft, and P. Geerlings, Phys. Rev. A 60, 1034 共1999兲. 35 C. Moore, Atomic Energy Levels as Derived from the Analyses of Optical Spectra 共NIST, Washington, 1971兲. 36 S. G. Lias, J. E. Bartmess, J. F. Liebman, J. L. Holmes, R. D. Levin, and W. G. Mallard, J. Phys. Chem. Ref. Data Suppl. 17, 1 共1988兲. 37 A. H. H. Chang and D. R. Yarkony, J. Chem. Phys. 99, 6824 共1993兲. 38 J. V. Coe, J. T. Snodgrass, C. B. Freidhoff, K. M. McHugh, and K. H. Bowen, Chem. Phys. Lett. 124, 274 共1986兲; J. Chem. Phys. 87, 4302 共1987兲. 39 P. Skurski, M. Gutowski, and J. Simons, Int. J. Quantum Chem. 76, 197 共2000兲; 80, 1024 共2000兲. 40 P. Skurski, J. Simons, X. B. Wang, and L. S. Wang, J. Am. Chem. Soc. 122, 4499 共2000兲. 41 F. Bruning, S. Matejcik, E. Illenberger, Y. Chu, G. Senn, D. Muigg, G. Denifl, and T. D. Mark, Chem. Phys. Lett. 292, 177 共1998兲. 42 R. L. Schwartz, G. E. Davico, T. M. Ramont, and W. C. Lineberger, J. Phys. Chem. A 103, 8213 共1999兲. 43 M. Schwartz and P. Marshall, J. Phys. Chem. A 103, 7900 共1999兲. 44 B. Hajgato, H. M. T. Nguyen, T. Veszpremi, and M. T. Nguyen, Phys. Chem. Chem. Phys. 2, 5041 共2000兲. 15 16

Downloaded 15 Aug 2001 to 140.109.112.248. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp