Planar nitrogen-doped aluminum clusters Al x N x=3 5

Transcription

1 THE JOURNAL OF CHEMICAL PHYSICS 25, Planar nitrogen-doped aluminum lusters Al x N x=3 5 Boris B. Averkiev and Alexander I. Boldyrev a Department of Chemistry and Biohemistry, Utah State University, 0300 Old Main Hill, Logan, Utah Xi Li b and Lai-Sheng Wang Department of Physis, Washington State University, 270 University Drive, Rihland, Washington and Chemial Sienes Division, Paifi Northwest National Laboratory, MS K8-88, P.O. Box 999, Rihland, Washington Reeived 28 April 2006; aepted 4 July 2006; published online 22 September 2006 The eletroni and geometrial strutures of three nitrogen-doped aluminum lusters, Al x N x=3 5, are investigated using photoeletron spetrosopy and ab initio alulations. Well-resolved photoeletron spetra have been obtained for the nitrogen-doped aluminum lusters at four photon energies 532, 355, 266, and 93 nm. Global minimum struture searhes for Al x N x=3 5 and their orresponding neutrals are performed using several theoretial methods. Vertial eletron detahment energies are alulated using three different methods for the lowest energy strutures and low-lying isomers are ompared with the experimental observations. Planar strutures have been established for all the three Al x N x=3 5 anions from the joint experimental and theoretial studies. For Al 5 N, a low-lying nonplanar isomer is also found to ontribute to the experimental spetra, signifying the onset of two-dimensional to three-dimensional transition in nitrogen-doped aluminum lusters. The hemial bonding in all the planar lusters has been eluidated on the basis of moleular orbital and natural bond analyses Amerian Institute of Physis. DOI: 0.063/ I. INTRODUCTION By doping pure metal lusters with one or more impurity atoms one an generate novel hemial speies and manipulate their physial and hemial properties. Understanding how impurities affet the hemial bonding in doped lusters an provide valuable information in understanding nanomaterials and nanostruture interfaes and may be an important step in developing a robust hemial bonding model whih ould be used in the rational design of the smallest eletroni devies based on nanolusters. Aluminum is widely used as ondutor in eletroni devies and aluminum nitride AlN is an important semiondutor material. Several prior experimental and theoretial studies have been reported on small aluminum nitride Al n N m lusters. In partiular, Nayak et al. have reported a ombined experimental and density funtional study on Al 3 N and Al 4 N. 4 Reently, Li and Wang reported an extensive set of photoeletron spetra of Al x N lusters from x=2 22 at 93 nm and ompared them to those of pure Al x lusters. 0 The present ontribution fouses on a detailed investigation of the eletroni struture and hemial bonding in three small aluminum lusters doped with one impurity N atom, Al x N and Al x N x=3 5. An extensive set of photoeletron spetra has been obtained for eah anion at four detahment photon energies 532, 355, 266, and 93 nm. Wellresolved spetral features were interpreted using ab initio a Eletroni mail: boldyrev@.usu.edu b Present address: Rowland Institute at Harvard, Harvard University, 00 Edwin H. Land Blvd., Cambridge, MA Eletroni mail: ls.wang@pnl.gov theoretial alulations at several levels of theory, whih allow us to unequivoally eluidate the eletroni and geometrial strutures, stability, and low-lying isomers of these three anions and their orresponding neutrals. The ground states of all three speies are found to be planar. Moleular orbital and natural bond analyses have been arried out to understand the hemial bonding in the planar lusters. We also found a nonplanar isomer of Al 5 N, whih is very lose in energy to the planar ground state isomer and is present in the moleular beam, suggesting the onset of two-dimensional 2D to three-dimensional 3D strutural transitions. II. EXPERIMENTAL METHOD The experiment was performed using a magneti-bottle photoeletron spetrosopy PES apparatus with a laser vaporization luster soure. Details of this apparatus have been published elsewhere. 2 Briefly, the Al x N lusters were produed by laser vaporization of either a pure Al or an AlN alloy disk target with a 5% N 2 /He arrier gas. The anion lusters of interest were size seleted and photoeletron spetra were taken at several different photon energies. The eletron energy resolution was better than 30 mev for ev eletrons. We have measured photodetahment spetra of Al x N for x=2 45 at 93 nm and reported those data up to x=22 reently. 0 In the urrent study, we obtained the photoeletron spetra of Al x N with x=3 5 at four photon energies, 532 nm 2.33 ev, 355 nm ev, 266 nm 4.66 ev, and 93 nm ev, as shown in Figs. 3, respetively /2006/25 2 /24305/2/$ , Amerian Institute of Physis Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

2 Averkiev et al. J. Chem. Phys. 25, FIG.. Photoeletron spetra of Al 3 N at four photon energies: 532 nm 2.33 ev, 355 nm ev, 266 nm 4.66 ev, and 93 nm ev. III. THEORETICAL METHODS The initial searh for the most stable strutures was performed using our gradient embedded geneti algorithm GEGA program written by Alexandrova. 3,4 We used a hybrid method known in the literature 5 7 as B3LYP with the small split-valene basis sets 8 3-2G for energy, gradient, and fore alulations. The lowest few strutures in every system were realulated using the B3LYP, a seond order Moller-Plesset perturbation theory MP2, 9 and oupledluster method with single, double, and noniterative triple exitations CCSD T based on the unrestrited Hartree-Fok UHF formalism for open-shell systems and all with the polarized split-valene basis sets 6-3+G * Total energies of these strutures were also alulated using the extended 6-3+G 2df basis sets. In order to test the validity of the one-eletron approximation, single point alulations were performed using the multionfiguration self-onsistent-field method 26,27 CASSCF X,Y with X ative eletrons and Y ative moleular orbitals. The vertial eletron detahment energies were alulated using R U CCSD T /6-3+G 2df and the outer valene Green s funtion method R U OVGF/6-3 FIG. 2. Photoeletron spetra of Al 4 N at four photon energies: 532 nm 2.33 ev, 355 nm ev, 266 nm 4.66 ev, and 93 nm ev. +G 2df at the CCSD T /6-3+G * geometries, as well as at the time-dependent density-funtional theory DFT method 33,34 TD-B3LYP/6-3+G 2df at the B3LYP/6-3+G * geometries. Core eletrons were frozen in treating the eletron orrelation at the R U CCSD T and R U OVGF levels of theory. The B3LYP, MP2, R U CCSD T, and R U OVGF ab initio alulations were performed using the GAUSSIAN 98 and GAUSSIAN 03 programs. 35,36 Moleular orbital visualization has been done using the MOLDEN3.4 program. 37 IV. EXPERIMENTAL RESULTS The 93 nm spetrum of Al 3 N Fig. reveals six spetral bands below 5.8 ev, labeled as X, A, B, C, D, and E, respetively. The features X around.9 ev and E around 5.2 ev are very broad. The onset of the first two transitions at.9 and 2.78 ev represents a large energy gap, suggesting that the neutral Al 3 N luster is likely to be losed shell with a large highest oupied moleular orbital HOMO -lowest unoupied moleular orbital LUMO gap. The broad nature of the X band indiates that there is a signifiant geometry hange between the ground states of the anion and neutral. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

3 Nitrogen-doped aluminum lusters J. Chem. Phys. 25, learly at lower photon energies, espeially at 532 nm. There are also weak signals around 2.7 ev. The intensities of these features depend on the soure onditions slightly but ould not be eliminated. All these weak features are most likely due to exited states or different isomers of Al 4 N. The spetra of Al 5 N are quite ompliated with numerous well-defined features X, A, B, C, D, E, and F at 266 nm Fig. 3, and more features are also revealed at higher binding energies in 93 nm. A strong feature around 2.0 ev x at both 532 and 355 nm is resolved, whih appears to be merged with the X band. The ompliated spetra of Al 5 N suggest the possible population of losely lying isomers, as born out from our theoretial alulations. The VDEs of the main spetral features for Al x N x =3, 4, and 5 are all given in Table I and are ompared with ab initio results in Tables V VII, respetively vide infra. V. THEORETICAL RESULTS We performed extensive searhs for the global minimum strutures of Al x N x=3 5 and their neutrals using our GEGA program at the B3LYP/3-2G level of theory. FIG. 3. Photoeletron spetra of Al 5 N at four photon energies: 532 nm 2.33 ev, 355 nm ev, 266 nm 4.66 ev, and 93 nm ev. The spetral features are better resolved in the lower photon energy spetra and the obtained vertial detahment energies VDEs are given in Table I. The spetra of Al 4 N Fig. 2 are surprisingly simple with only two intense bands observed in the 93 nm spetrum. The X band at 2.32 ev in the 355 and 266 nm spetra is very sharp with no indiation of any vibrational struture, whereas the 3.4 ev feature A is broad with a partially resolved vibrational progression, whih yields a vibrational frequeny of 80±50 m. A shoulder on the higher binding energy side of the A band is identified as another eletroni transition, labeled as B. A weak band C is observed in the 93 nm at very high binding energies. There are also other weak features present in the spetra of Al 4 N. The features, ourring at around.76 x and.93 ev, are resolved more A. Al 3 N and Al 3 N The planar D 3h A,a 2 e 4 2a 2 2 2a 2 2e 4 3e 0 struture I Fig. 4 was found to be the global minimum struture for Al 3 N from prior ab initio alulations 3 6 and from infrared matrix investigations. 6 Andrews et al. assigned two sharp bands at and m to the e antisymmetri strething mode of Al 3 N split by interation in nitrogen matrix. The major peak at 773. m in argon matrix was assigned to the same mode. Our alulations of the planar D 3h A geometri struture for Al 3 N Table II at our highest level of theory CCSD T /6-3+G * reveal 3 e =776 m, in exellent agreement with the matrix experiment. For the Al 3 N anion we performed GEGA searh at the B3LYP/3-2G level of theory. The two lowest below 20 kal/mol doublet strutures are presented in Figure 4. The planar T-shaped C 2v 2 B 2,a 2 b 2 2 2a 2 b 2 3a 2 2b 2 2 4a 2 3b 2 struture II is predited by GEGA to be the global minimum struture, similar to that reported previously by Nayak et al. 4 using DFT alulations. The seond C 2v 2 A onfiguration originated from the a 2 e 4 2a 2 2 2a 2 2e 4 3e oupation is a first order saddle point on the intramoleular rearrangement of the anion from one global minimum struture into another. We performed single point alulations at the CASSCF 7,8 /6-3+G *, CASSCF 7,0 /6-3+G *, and CASSCF 3,2 /6-3+G * and found that the Hartree- Fok wave funtion was dominant C HF =0.949, C HF =0.943, TABLE I. Experimental vertial detahment energies in ev for the Al 3 N,Al 4 N, and Al 5 N anions from the photoeletron spetra. The number in parentheses represents the unertainty of the last digit. X A B C D E F Al 3 N Al 4 N Al 5 N Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

4 Averkiev et al. J. Chem. Phys. 25, FIG. 4. The lowest isomers for Al 3 N and Al 3 N. Relative energies are presented at CCSD T /6-3+G 2df //B3LYP/6-3+G * and at B3LYP/6-3+G * in brakets. N imag is the number of imaginary frequenies alulated at B3LYP/6-3+G *. and C HF =0.945, respetively in the CASSCF expansion, thus onfirming the appliability of MP2 and CCSD T theoretial methods. However, at our highest level of theory CCSD T /6-3+G * the planar T-shaped C 2v 2 B 2 struture is a saddle point with the 3 b =37i m imaginary frequeny Table II. Geometry optimization following the imaginary frequeny leads to a slightly nonplanar struture C s 2 A with the nitrogen atom being out of plane by 0.04 Å. The energy differene between the planar and nonplanar strutures is only kal/mol and that value is signifiantly lower than the differene in zero point energy ZPE orretions kal/mol for the two strutures. Thus, the vibrationally averaged Al 3 N struture is atually planar and for all pratial purpose we will onsider Al 3 N as being planar in the following disussion. The lowest alternative struture III C 2v, 2 B,a 2 2a 2 b 2 2 b 2 3a 2 2b 2 2 4a 2 2b orresponds to a loal minimum, whih is substantially higher in energy and will not be further disussed. The T shape for the isoeletroni Al 3 O luster was predited before by Boldyrev and Shleyer 38 and Sun et al. 39 with quite similar moleular parameters. The CAlSi 2 luster, another 5 valene eletron tetra-atomi moleule, was, however, found to have a Y-type struture by Boldyrev et al. 40 When an additional eletron is added to Al 3 N, the resulting 6 valene eletroni Al 3 N 2 luster was found to have the T shape again, whih has been found to be the ground state for all other studied isoeletroni speies suh as BSi 3, CAlSi 2,CSi 3,NSi 3 +,NAl 2 Si,Al 3 O, and Al 3 F. 40 B. Al 4 N and Al 4 N For Al 4 N the GEGA searh led to a global minimum planar D 4h 2 B 2g,a 2 g e 4 u a 2 2u 2a 2 g b 2 g 2e 4 u b 2g struture IV Fig. 5, in agreement with several previously reported theoretial studies. 3 5,7,9 At our highest level of theory TABLE II. The moleular properties of the global minimum Al 3 N and Al 3 N strutures. Moleular parameter Al 3 N C 2v, 2 B 2 Al 3 N C s, 2 A B3LYP/6-3+G * MP2/6-3+G * CCSD T /6-3+G * CCSD T /6-3+G *a E a.u b R N Al Å R N Al 2,3 Å Al NAl 2,3 deg a m a m a m b m i 46 4 b 2 m b 2 m Al 3 N D 3h, A Moleular parameter B3LYP/6-3+G * MP2/6-3+G * CCSD T /6-3+G * E a.u R N Al Å a m a a m e m e m a E tot = a.u.. b E tot = a.u. all at CCSD T /6-3+G 2df / /CCSD T /6-3+G *. Nitrogen atom omes out of the Al 3 plane by Å. Values in parentheses represent relative absorbane intensities in the IR spetrum. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

5 Nitrogen-doped aluminum lusters J. Chem. Phys. 25, CCSD T /6-3+G *, the D 4h 2 B 2g struture is a saddle point Table III. Distortion along the b 2u mode of imaginary frequeny leads to a slightly nonplanar butterfly distorted struture D 2d 2 B. However, the energy differene between D 2d 2 B and D 4h 2 B 2g is only 0.92 kal/mol, whih is smaller than the differene in ZPE kal/mol. Therefore, the vibrationally averaged struture is atually planar and in the following disussion we will onsider the Al 4 N luster to be planar. The similar small deviation from planarity was previously reported for the valene isoeletroni Al 4 C anion. 4 The planar struture was also found to be a global minimum struture for other 8 valene eletron penta-atomi speies Al 4 O,,38,39 Al 2 Si 2 C,,42 Al 2 Ge 2 C,,42 Al 3 SiC, 43 and Al 3 GeC, 43 as well as for other 7 valene eletron penta-atomi speies Al 4 C, 4 Al 3 SiC, 43 and Al 3 GeC. 43 Hene, the 8 and 7 eletron rule,38,42 45 for planarity of penta-atomi lusters is a general rule for these speies. For the neutral Al 4 N luster, in addition to the global minimum D 4h 2 B 2g struture IV, we also found two lowlying isomers C 2v 2 A struture V in Fig. 5 and C 2v 2 B struture VI in Fig. 5, whih are 5.4 and 0.9 kal/mol above the global minimum at the CCSD T /6-3 +G 2df / /B3LYP/6-3+G * level of theory. The planar D 4h A g struture for Al 4 N was first omputationally predited by Shleyer and Boldyrev on the basis of moleular orbital analysis for the five-atomi 8 valene eletron systems and was onfirmed in follow up alulations. 2,4,5,7 In order to onfirm these results for the Al 4 N anion, we run GEGA alulations for both singlet and triplet states of the Al 4 N anion. The lowest less than about 20 kal/mol singlet and triplet strutures found by GEGA are presented in Fig. 5. The GEGA searh found the planar D 4h A g,a 2 g e 4 u a 2 2u 2a 2 g b 2 g 2e 4 u b 2 2g struture VII to be the global minimum, in agreement with previously reported theoretial results.,2,4,5,7 We performed single point alulations at the CASSCF 8,8 /6-3+G *, CASSCF 2,2 /6-3+G *, and CASSCF 8,4 /6-3+G * and found that the Hartree-Fok wave funtion was dominant C HF =0.980, C HF =0.969, and C HF =0.958, respetively in the CASSCF expansion. We found that the next lowest isomer is a triplet C 2v 3 B,a 2 2a 2 b 2 2 3a 2 b 2 4a 2 2b 2 2 5a 2 6a 2b struture VIII 20.6 kal/mol higher at CCSD T /6-3 +G 2df / /B3LYP/6-3+G *, whih is similar to previously reported triplet struture by Nayak et al. 4 using DFT alulations, though they did not speify the spetrosopi state of their triplet isomer. We also found one singlet struture IX C s, A,a 2 2a 2 3a 2 4a 2 a 2 5a 2 6a 2 7a 2 8a 2, whih is more than 20 kal/mol higher. FIG. 5. The lowest isomers for Al 4 N and Al 4 N. Relative energies are presented at CCSD T /6-3+G 2df //B3LYP/6-3+G * and at B3LYP/6-3+G * in brakets. N imag is the number of imaginary frequenies alulated at B3LYP/6-3+G *. C. Al 5 N and Al 5 N For Al 5 N the GEGA B3LYP/3-2G searh found many strutures Fig. 6, X XXI with the planar C 2v A,a 2 2a 2 b 2 2 3a 2 b 2 4a 2 2b 2 2 3b 2 2 5a 2 6a 2 struture being the global minimum. The refinement at B3LYP/6-3+G * and at CCSD T /6-3+G 2df / /B3LYP/6-3+G * onfirmed our GEGA results. However, at MP2/6-3+G * and CCSD T /6-3+G *, it has one imaginary frequeny Table IV. The most stable struture C s A at the last two levels of theory is only slightly distorted from the C 2v symmetry and after ZPE orretions it is effetively C 2v symmetry. The C 2v A struture X was also reported to be the global minimum struture by Guo and Wu 9 who used two B3LYP/6-3+G * and SVWN/6-3+G * theoretial methods. Aording to Ling et al. 8 the global minimum of Al 5 N orresponds to the struture XVIII at the full-potential linearmuffin-tin-orbital moleular dynamis FP-LMTO-MD method. We found that this struture is a seond order saddle point, whih is 7.3 kal/ mol higher in energy at B3LYP/6-3+G * 3.5 kal/mol at CCSD T /6-3 +G 2df / /B3LYP/6-3+G *. Geometry optimization of the struture XVIII following the imaginary frequeny led initially to the struture XII, and geometry optimization following the imaginary in the struture XII led eventually to the global minimum struture X. Thus, the struture XVIII an be safely exluded from being a global minimum struture of Al 5 N. Nayak et al. 3 reported that the struture XI is the global minimum struture for the Al 5 N luster using the BPW9/6-3G ** level of theory. The same global minimum struture was also reported by Leskiw et al. 7 Aording to our alulations, the struture XI is a first order saddle point at B3LYP/6-3+G *. Geometry optimization of the struture XI following the imaginary frequeny led to a slightly distorted struture C 2 A in whih the top Al atom still loated on the C 2 axis with the beneath tetrahedral-type Al 4 luster being slightly distorted. The energy differene between the C 2v A struture XI and slightly distorted loal minimum struture is only 0.02 kal/ mol, whih is signifiantly lower than the differene in ZPE orretions kal/mol for the two strutures. At the MP2/6-3 +G * and CCSD T /6-3+G * levels of theory, the C 2v A struture was found to be a minimum. Thus, for all pratial purpose, we will onsider the struture XI of Al 5 N as being C 2v symmetry. The struture XI is 6.8 kal/mol higher in Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

6 Averkiev et al. J. Chem. Phys. 25, TABLE III. The moleular properties of the Al 4 N and Al 4 N speies. Values in parentheses represent relative absorbane intensities in the IR spetrum km/mol. Moleular parameter Al 4 N D 4h, A Al 4 N D 4h, 2 B 2g Al 4 N D 2d, 2 B B3LYP/6-3+G * MP2/6-3+G * CCSD T /6-3+G *a B3LYP/6-3+G * MP2/6-3+G * CCSD T /6-3+G *b CCSD T /6-3+G *b E a.u R N Al Å a g m a 2u m b g m b 2g m b 2u m i 47 6 e u m e u m a E tot = a.u. b E tot = a.u. all at CCSD T /6-3+G 2df / /CCSD T /6-3+G *. energy than the struture X at B3LYP/6-3+G *, but this differene is only.3 kal/mol at CCSD T /6-3 +G 2df / /B3LYP/6-3+G * and.3 kal/mol at CCSD T /6-3+G 2df / /CCSD T /6-3+G *. Thus, these two strutures are almost degenerate at our highest level of theory. Two strutures XIV and XVI were found to have two and one imaginary frequenies at B3LYP/ 6-3 +G * level of theory: however, they have effetive C 2v symmetry after ZPE orretions. Other strutures identified in our alulations, within 20 kal/ mol above the ground state, are summarized in Fig. 6. The struture of the Al 5 N anion has been previously studied by Leskiw et al., 7 who reported that the global minimum struture XXIII is similar to the struture XI for Al 5 N neutral. Our GEGA searh at B3LYP/3-2G for Al 5 N found the planar C 2v 2 B,a 2 2a 2 b 2 2 3a 2 b 2 4a 2 2b 2 2 3b 2 2 5a 2 6a 2 2b struture XXII to be the global minimum, whih is similar to the global minimum of the neutral struture X. We also performed single point alulations at CASSCF 3, /6-3+G *, CASSCF, /6-3+G *, and CASSCF,2 /6-3+G * and found that the Hartree- Fok wave funtion was dominant C HF =0.932, C HF =0.925, and C HF =0.92, respetively in the CASSCF expansion. At MP2/6-3+G * and CCSD T /6-3+G * it has one and two imaginary frequenies, respetively Table IV. However, the vibrationally averaged global minimum struture an be onsidered to have the C 2v symmetry. The seond lowest energy struture C s 2 A in our alulations at B3LYP/3-2G, B3LYP/6-3+G *, and MP2/6-3+G * is similar to the global minimum struture XXIII proposed by Leskiw et al., 7 though in our ase it has lower symmetry C s TABLE IV. The moleular properties of the Al 5 N and Al 5 N speies. Moleular parameter Al 5 N C 2v, 2 B Al 5 N C 2v, A B3LYP/6-3+G * MP2/6-3+G * CCSD T /6-3+G *a B3LYP/6-3+G * MP2/6-3+G * CCSD T /6-3+G *b E a.u R N Al Å R N Al 2,3 Å R N Al 4,5 Å Al NAl 2,3 deg Al NAl 4,4 deg a m a m a m a m a m a m i b m b m i 9 62i i 5 32i 9 b 2 m b 2 m b 2 m b 2 m a E tot = a.u. b E tot = a.u. all at CCSD T /6-3+G 2df / /CCSD T /6-3+G *. Values in parentheses represent relative absorbane intensities in the IR spetrum km/mol. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

7 Nitrogen-doped aluminum lusters J. Chem. Phys. 25, instead of C 2v. Its relative energy is 5.4 kal/mol at B3LYP/6-3+G * and 3.0 kal/mol at CCSD T /6-3 +G 2df / /B3LYP/6-3+G *. Geometry optimization at the CCSD T /6-3+G * level of theory revealed that C s 2 A struture ollapsed into the C 2v 2 A struture XXIII. The struture XXIII is 2.6 kal/mol CCSD T /6-3 +G 2df / /CCSD T /6-3+G * higher in energy than the global minimum. The third lowest energy struture XXIV C s 2 A lies 6.2 kal/mol at B3LYP/6-3+G *, 4.2 kal/mol at CCSD T /6-3+G 2df / /B3LYP/6-3 +G *, and 4.0 kal/mol at CCSD T /6-3 +G 2df / /CCSD T /6-3+G * higher in energy than the global minimum. In addition, two more strutures XXV and XXVI were found to be about 6 8 kal/mol CCSD T /6-3+G 2df / /CCSD T /6-3+G * above the global minimum. Other strutures identified in our alulations, within 20 kal/mol above the global minimum, are also shown in Fig. 6. FIG. 6. The lowest isomers for Al 5 N and Al 5 N. Relative energies are presented at CCSD T /6-3+G 2df //B3LYP/6-3+G * and at B3LYP/6-3+G * in brakets. N imag is the number of imaginary frequenies alulated at B3LYP/6-3+G *. VI. COMPARISON OF CALCULATED VDEs WITH EXPERIMENT A. Al 3 N The ab initio VDEs alulated at the TD-B3LYP/ 6-3+G 2df, ROVGF/6-3+G 2df, and CCSD T /6-3+G 2df levels for Al 3 N are ompared with the experimental data in Table V, and good agreement is obtained among the different levels of theory and between the theory and experiment. The global minimum of Al 3 N was found to be the planar struture II C 2v, 2 B 2 with the valene eletroni onfiguration a 2 b 2 2 2a 2 b 2 3a 2 2b 2 2 4a 2 3b 2. As given in Table V, our alulated VDE for removal of an eletron from the HOMO of the global minimum is.06 ev at the UCCSD T /6-3+G 2df level of theory,.6 ev at the UOVGF/6-3+G 2df level of theory, and.5 ev at the TD-B3LYP/6-3+G 2df level of theory. The pole strength UOVGF was found to be 0.90, indiating that the detahment hannel an be primarily desribed by a oneeletron detahment proess. The alulated first VDE for the struture II C 2v, 2 B 2 is in exellent agreement with the measured VDE of.9±0.04 ev for this feature Table V. The 3b 2 -HOMO of Al 3 N is a bonding orbital within the triangular wing Al Al Al in the global minimum struture Fig. 7. Detahment of the eletron from this orbital results in a signifiant geometry relaxation from C 2v in the anion to D 3h in the neutral. The large geometry hange is onsistent with the broad PES bandwidth observed for this transition Fig.. The alulated adiabati eletron detahment energy ADE is 0.73 ev CCSD T 6-3+G 2df / /CCSD T /6-3+G * +ZPE/ /CCSD T /6-3+G *, in good agreement with the experimental threshold 0.7 ev for the X band. Beause of the large geometry hanges and the lak of vibrational resolution, the ADE annot be aurately determined from the experimental PES spetra. Eletron detahment from the doubly oupied MOs ould result in either triplet or singlet final states at the C 2v global minimum struture. The two triplet 3 B 2 a 2 b 2 2 2a 2 b 2 3a 2 2b 2 2 4a 3b 2 and 3 A a 2 b 2 2 2a 2 b 2 3a 2 2b 2 4a 2 3b 2 states are assigned to the sharp features A and B, respetively Table V, and the orresponding singlet B 2 a 2 b 2 2 2a 2 b 2 3a 2 2b 2 2 4a 3b 2 and A a 2 b 2 2 2a 2 b 2 3a 2 2b 2 4a 2 3b 2 states are assigned to the relatively weak features C and D Table V. The 4a and 2b 2 MOs are essentially nonbonding Fig. 7, onsistent with the relatively sharp spetral features observed for bands A, B, C, and D. Finally, the broad feature E most probably orresponds to several transitions involving the two triplet 3 B 2 a 2 b 2 2 2a 2 b 2 3a 2b 2 2 4a 2 3b 2 and 3 A 2 a 2 b 2 2 2a 2 b 3a 2 2b 2 2 4a 2 3b 2 states Table V. The 3a and b orbitals are strongly bonding MOs Fig. 7, onsistent with the broad E band. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

8 Averkiev et al. J. Chem. Phys. 25, TABLE V. Comparison of the experimental VDEs to alulated VDEs for struture II of the Al 3 N anion. Feature VDE Expt. ev Final state and eletroni onfiguration VDE Theor. ev TD-B3LYP a OVGF a,b CCSD T a X A B C D E A,b 2 3a 2 2b 2 2 4a 2 3b B 2,b 2 3a 2 2b 2 2 4a 3b 2 3 A,b 2 3a 2 2b 2 4a 2 3b 2 B 2,b 2 3a 2 2b 2 2 4a 3b 2 A,b 2 3a 2 2b 2 4a 2 3b 2 3 B 2,b 2 3a 2b 2 2 4a 2 3b 2 3 A 2,b 3a 2 2b 2 2 4a 2 3b a a a a a 6-3+G 2df basis set. b Values in parentheses represent the pole strength. B. Al 4 N The alulated VDEs for the global minimum struture VII, as well as for the low-lying isomer VIII of Al 4 N, are ompared with the experimental data in Table VI. The alulated VDEs from the b 2g HOMO of the planar square D 4h struture VII at three levels of theory are 2.6 ev TD- B3LYP/6-3+G 2df, 2.26 ev ROVGF/6-3 +G 2df, and 2.29 ev CCSD T /6-3+G 2df, agreeing very well with the experimental value of 2.32±0.03 ev Table VI. The alulated adiabati eletron detahment energy ADE is 2.28 ev CCSD T /6-3+G 2df / /CCSD T /6-3+ G * + ZPE/ /CCSD T /6-3+G *, in good agreement with the experimental value of 2.29 ev. 4 The broadband A and B in the experimental spetra Fig. 2 are due to detahment from HOMO- 2e u and HOMO-2 b g Table VI, whih are very lose to eah other, resulting in the overlap of the two detahment bands. Beause the order of the 2 E u and 2 B g states is different at TD-B3LYP and OVGF, we annot be sure with ertainty whih spetrosopi state is atually lower in energy. The next rather weak feature C ould be assigned to detahment from HOMO-3 2a g. Aording to our alulations there should be one detahment hannel from HOMO-4 a 2u around 6 ev Table VI, whih may be underestimated beause no major detahment band was observed in the higher binding energy side. Overall, the alulated VDEs from the planar D 4h global minimum are in good agreement with the experiment, in partiular, for the first three detahment hannels. The weak features, observed in the low energy part 2.3 ev of the spetrum and in between the intense peaks X and A Fig. 2, annot be explained by the global minimum D 4h struture and they should belong to either alternative isomers or to impurities. Nayak et al. 4 explained these small features by ontributions from a triplet isomer. Aording to our alulations there are two lowest isomers: a triplet C 2v 3 B struture VIII 9.8 kal/mol higher at CCSD T /6-3+G 2df / /B3LYP/6-3+G * and singlet a C s struture IX 2.8 kal/mol higher at CCSD T /6-3 +G 2df / /B3LYP/6-3+G * Fig. 5. Indeed the triplet C 2v 3 B struture VIII gives alulated VDEs, whih fall in the right energy ranges for the weak features in the PES spetra of Al 4 N Table VI. Our result agrees with the previous assignment by Nayak et al., 4 whose DFT alulations suggest that the triplet isomer is 0.97 ev 22.3 kal/mol above the D 4h ground state. It is very surprising that suh a high energy isomer an be populated at all in the experiment. The explanation suggested by Nayak et al. that suh a high energy isomer is spin proteted seems reasonable, i.e., the triplet isomer one formed is prevented from being relaxed to the D 4h ground state beause it is spin forbidden. Although this is unusual, we have inreasingly observed in several luster system population of high energy triplet isomers, for example, in B 7 and B 3. 46,47 C. Al 5 N We omputed the VDEs from the three low-lying isomers for Al 5 N and ompared them with the experimental data in Tables VII IX. Clearly, at least two isomers are needed to interpret the observed spetra of Al 5 N. The OVGF and TD-DFT methods give a similar first VDE for all three isomers and annot be used to distinguish them. However, our most aurate CCST T method gives a first VDE of 2.0 ev for struture XXII, whih is in good agreement with that of feature x 2.0 ev, whereas the first VDE for struture XXIII from CCST T is.83 ev in very good agreement with that of feature X.89 ev, suggesting that the observed first two detahment bands ome from two different isomers of Al 5 N. The fat that the two features have similar intensities suggests that the two isomers are most likely degenerate, onsistent with the lose energies of the two isomers Fig. 6. The alulated seond VDE from CCST T for the planar C 2v struture XXII of Al 5 N is 2.63 ev, in good agreement with that of band B 2.66 ev. The seond CCST T VDE for the C 2v struture XXIII is 2.23 ev, in exellent agreement with that of band A 2.29 ev. The next CCST T detahment hannel for both isomers is at muh higher energies, thus preventing us from making more definitive assignments for the higher binding energies features. However, our TD-DFT data suggest that eah of the higher binding energy feature may ontain ontributions from both isomers. We also omputed the VDEs for isomer XXIV Table IX, whih seem to be all similar to those of the planar isomer XXII. Sine this isomer is higher in energy, we suspet that it may not be signifiantly populated. Its minor ontribution to the observed spetra is likely to be obsured. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

9 Nitrogen-doped aluminum lusters J. Chem. Phys. 25, FIG. 7. Valene moleular orbitals for the struture II of Al 3 N UHF/6-3+G *. Al 5 N is a rare ase, where two or more isomers are nearly degenerate and seem to be equally populated experimentally. The two isomers of Al 5 N are very different, one planar 2D struture XXII and the other 3D struture XXIII. Al 3 N,Al 3 N, Al 4 N, and Al 4 N are all overwhelmingly stable as 2D strutures. However, the 2D and 3D strutures for Al 5 N are nearly degenerate, signifying the onset of 2D to 3D transitions. VII. CHEMICAL BONDING ANALYSES A. Al 3 N and Al 3 N The peuliar T shape for Al 3 N an be understood on the basis of MO analysis, as shown in Fig. 7. The four lowest valene MOs HOMO-7, HOMO-6, HOMO-5, and HOMO-4 are primarily formed from 2s HOMO-7, 2p x HOMO-4, 2p y HOMO-6, and 2p z HOMO-5 atomi orbitals AOs of N. The next three MOs HOMO-, HOMO-2, and HOMO-3 orresponding to three lone pairs are formed primarily by 3s AOs of Al. When only these MOs are oupied as that in Al 3 N, then the resulting struture is a perfet triangle with N being at the enter and formal harge distribution is lose to ioni Al + 3 N 3, i.e., Al is ating as a valene + atom. Indeed, our alulated natural bond orbital NBO effetive harges q N = 2.46 e and q Al = e B3LYP/6-3+G * support this simple ioni piture. The entral N atom has already a full otet of valene eletrons and it should not form any additional bonds. Thus one would expet that the Al 3 N anion may not be an eletronially stable speies. That is indeed the ase for NH 3, whih does not bind an aess eletron. 48 However, the Al 3 N anion is quite eletronially stable with a VDE of.9±0.04 ev. The stability of the Al 3 N anion omes from the extra eletron oupying 3b 2 -HOMO, whih is a pure ligand Al Al bonding orbital Fig. 7. Calulated NBO effetive harges q N = 2.32 e, q Al a = e, and q Al e = e B3LYP/6-3+G * support this desription. The bonding harater of the 3b 2 -HOMO is responsible for the eletroni stability of this anion and for its T shape. The lower symmetry of Al 3 N an also be understood as a Jahn-Teller distortion of the initial D 3h struture of Al 3 N when an additional eletron oupies its double degenerate LUMO 3e. B. Al 4 N and Al 4 N The planar square struture of Al 4 N an also be understood on the basis of the MO analysis, as shown in Fig. 8. The four lowest valene MOs HOMO-6, HOMO-5, HOMO-5, and HOMO-4 are again primarily formed from 2s HOMO-6, 2p x HOMO-5, 2p y HOMO-5, and 2p z HOMO-4 AOs of N. The next four MOs HOMO-, HOMO-2, HOMO-2, and HOMO-3 orresponding to four lone pairs are formed primarily by 3s AOs of aluminum. When only these MOs are oupied as in Al 4 N +, the resulting struture is a tetrahedron with N being at the enter and a formal harge distribution lose to ioni, similar to the isoeletroni Al 4 C moleule. 4,44 The tetrahedral Al 4 C moleule has the following eletron onfiguration a 2 t 6 2 2a 2 2t 6 2 e 0. When one or two eletrons oupy one of the doubly degenerate e-lumo, the resulting anions Al 4 C and Al 4 C 2 undergo Jahn-Teller distortion toward the planar D 4h struture. 4,45 Similarly, oupation of e-lumo in the tetrahedral Al 4 N + should result in the geometri distortion towards D 4h struture in Al 4 N and Al 4 N.InAl 4 N + the entral atom N has a full otet and formally the hyperstoihiometri moleules Al 4 N and Al 4 N should not be stable. However, the ligand-ligand bonding HOMO in Al 4 N and Al 4 N is responsible for the eletron and geometri stability TABLE VI. Comparison of the experimental VDEs to alulated VDEs for Al 4 N. Feature VDE Expt. ev Final state and eletroni onfiguration VDE Theor. ev TD-B3LYP a OVGF a,b CCSD T a X A B C Struture VII 2 B 2g,a 2 2u 2a 2 g b 2 g 2e 4 u b 2g 2 E u,a 2 2u 2a 2 g b 2 g 2e 3 2 u b 2g 2 B g,a 2 2u 2a 2 g b g 2e 4 2 u b 2g 2 A g,a 2 2u 2a g b 2 g 2e 4 2 u b 2g 2 A 2u,a 2u 2a 2 g b 2 g 2e 4 2 u b 2g Struture VIII 2 A,b 2 4a 2 2b 2 2 5a 2 6a 2 B,b 2 4a 2 2b 2 2 5a 2 2b 4 B,b 2 4a 2 2b 2 2 5a 6a 2b a 6-3+G 2df basis set. b Values in parentheses represent the pole strength. The pole strength is too low. This value is not reliable. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

10 Averkiev et al. J. Chem. Phys. 25, TABLE VII. Comparison of the experimental VDEs to alulated VDEs for the struture XXII of Al 5 N. Feature VDE Expt. ev Final state and eletroni onfiguration VDE Theor. ev TD-B3LYP a UOVGF a,b CCSD T a X x A B C D E F A,b 2 4a 2 2b 2 2 3b 2 2 5a 2 6a 2 2b 2 3 B,b 2 4a 2 2b 2 2 3b 2 2 5a 2 6a 2b B,b 2 4a 2 2b 2 2 3b 2 2 5a 2 6a 2b 3 B,b 2 4a 2 2b 2 2 3b 2 2 5a 6a 2 2b B,b 2 4a 2 2b 2 2 3b 2 2 5a 6a 2 2b A 2,b 2 4a 2 2b 2 2 3b 2 5a 2 6a 2 2b 3 A 2,b 2 4a 2 2b 2 2 3b 2 5a 2 6a 2 2b 3 A 2,b 2 4a 2b 2 3b 2 2 5a 2 6a 2 2b A 2,b 2 4a 2 2b 2 3b 2 2 5a 2 6a 2 2b 3 B,b 2 4a 2b 2 2 3b 2 2 5a 2 6a 2 2b d a 6-3+G 2df basis set. b Values in parentheses represent the pole strength. This value annot be alulated at the this level of theory. d The pole strength is too low. This value is not reliable. of these speies. Al 4 N and Al 4 N are analogs of the first experimentally disovered penta-atomi tetraoordinate planar arbon moleules, Al 4 C and Al 4 C C. Al 5 N and Al 5 N The geometries of the strutures X of Al 5 N and XXII of Al 5 N Fig. 6 hint that they ould be formally onsidered as Al + or Al oordinated to the edge of the planar Al 4 N anioni struture VII Fig. 5, respetively. Calulated NBO harges B3LYP/6-3+G * on the entral nitrogen atom 2.3 and 2.33 e are almost the same in Al 4 N VII, Al 5 N X, and Al 5 N XXII. The NBO harge on the apex Al atom in Al 5 N is e, whih is lower than ioni limit of +.0 e, but that is qualitatively onsistent with the formal Al 4 N Al + formulation, indiating the strutural stability of the planar tetraoordinate Al 4 N. In the anioni Al 5 N luster, a signifiant portion of the additional eletron goes to the apex Al atom, whih now has a NBO harge of 0.25 e. The moleular orbital pitures for Al 5 N and Al 5 N Fig. 9 are also onsistent with the hemial bonding desribed above. One an see that moleular orbitals HOMO-, HOMO-2, HOMO-3, HOMO-4, HOMO-5, HOMO-6, HOMO-8, HOMO-9, and HOMO-0 in Al 5 N an be approximately orrelated to the moleular orbitals HOMO, HOMO-2, HOMO-, HOMO-2, HOMO-3, HOMO-4, HOMO-5, HOMO-5, and HOMO-6 in Al 4 N, respetively. The HOMO and HOMO-7 are responsible for the - and -ovalent bondings between the apex Al atom and the Al 4 N luster. The seond lowest isomer of Al 5 N XI ould be formally onsidered as AlN 2 oordinated to the edge of the tetrahedral Al 4 2+ diation. It was previously shown 38 that the Al 4 2+ diation has a tetrahedral struture. Calulated NBO TABLE VIII. Comparison of the experimental VDEs to alulated VDEs for the struture XXIII of Al 5 N. Feature VDE Expt. ev Final state and eletroni onfiguration VDE Theor. ev TD-B3LYP a OVGF a,b CCSD T a X x A B C D E F A,2b 2 4a 2 2b 2 2 5a 2 6a 2 7a 0 3 A,2b 2 4a 2 2b 2 2 5a 2 6a 7a A,2b 2 4a 2 2b 2 2 5a 2 6a 7a 3 A,2b 2 4a 2 2b 2 2 5a 6a 2 7a A,2b 2 4a 2 2b 2 2 5a 6a 2 7a 3 B 2,2b 2 4a 2 2b 2 5a 2 6a 2 7a B 2,2b 2 4a 2 2b 2 5a 2 6a 2 7a 3 A,2b 2 4a 2b 2 2 5a 2 6a 2 7a 3 B,2b 4a 2 2b 2 2 5a 2 6a 2 7a A,2b 2 4a 2b 2 2 5a 2 6a 2 7a B,2b 4a 2 2b 2 2 5a 2 6a 2 7a d a 6-3+G 2df basis set. b Values in parentheses represent the pole strength. This value annot be alulated at the this level of theory. d The pole strength is too low. This value is not reliable. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

11 Nitrogen-doped aluminum lusters J. Chem. Phys. 25, TABLE IX. Comparison of the experimental VDEs to alulated VDEs for the struture XXIV of Al 5 N. Feature VDE Expt. ev Final state and eletroni onfiguration VDE Theor. ev TD-B3LYP a OVGF a,b CCSD T a X x A B C D E F A, 4a 2 5a 2 6a 2 2a 2 7a 2 3a 2 8a A, 4a 2 5a 2 6a 2 2a 2 7a 2 3a 8a A, 4a 2 5a 2 6a 2 2a 2 7a 2 3a 8a A, 4a 2 5a 2 6a 2 2a 2 7a 3a 2 8a A, 4a 2 5a 2 6a 2 2a 2 7a 3a 2 8a A, 4a 2 5a 2 6a 2a 2 7a 2 3a 2 8a A, 4a 2 5a 2 6a 2 2a 7a 2 3a 2 8a A, 4a 2 5a 2 6a 2 2a 7a 2 3a 2 8a A, 4a 2 5a 6a 2 2a 2 7a 2 3a 2 8a A, 4a 2 5a 6a 2 2a 2 7a 2 3a 2 8a 4.37 A, 4a 4a 2 6a 2 2a 2 7a 2 3a 2 8a 4.8 a 6-3+G 2df basis set. b Values in parentheses represent the pole strength. This value annot be alulated at the this level of theory. harge B3LYP/6-3+G * on the AlN group is.42 e and that is qualitatively onsistent with the formal Al 4 2+ NAl 2 formulation. Aording to our NBO analysis of Al 5 N XXIII, an additional eletron in Al 5 N goes to the Al 4 unit. VIII. CONCLUSIONS Well-resolved photoeletron spetra were obtained for three nitrogen-doped aluminum lusters Al 3 N,Al 4 N, and Al 5 N at four photon energies 532, 355, 266, and 93 nm and ompared with theoretial alulations to eluidate their eletroni struture and hemial bonding. Global minimum strutures of Al 3 N,Al 4 N, and Al 5 N were identified first by using gradient embedded geneti algorithm B3LYP/3-2G followed by B3LYP/6-3+G *, MP2/6-3+G *, and CCSD T /6-3+G * geometry and frequeny alulations. By omparing the theoretial VDEs with the experimental data, we established that NAl 3 is D 3h and Al 3 N has a T-shaped struture II C 2v 2 B 2. The ground states of Al 4 N and Al 4 N are both square planar in agreement with previously reported results. For the Al 5 N anion we found two quasidegenerate strutures XXII C 2v, 2 B and XXIII C 2v, 2 A, whih are almost equally populated experimentally. We also omputationally identified three other low energy strutures: XXIV C s, 2 A, XXV C s, 2 A, and XXVI C 2v, 2 A, whih lie with 6 8 kal/mol above the ground state. The low-lying isomers of the neutral Al 5 N luster are found to be similar to those of its anion. Chemial bonding analysis revealed that Al 3 N an be desribed as an ioni luster Al + 3 N 3. In the planar C 2v struture of Al 3 N, an additional eletron oupies the peripheral ligand-ligand bonding HOMO, resulting in the strutural distortion to the T shape. Both Al 4 N and Al 4 N have FIG. 8. Valene moleular orbitals for the struture VII of Al 4 N RHF/6-3+G *. FIG. 9. Valene moleular orbitals for the struture XXII of Al 5 N UHF/6-3+G *. Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see

12 Averkiev et al. J. Chem. Phys. 25, planar struture with a entral nitrogen atom. The planarity of this struture is due to the singly or doubly oupied b 2g -HOMO, whih is a peripheral four-enter ligand-ligand bonding orbital, similar to the tetraoordinate planar arbon moleules, CAl 4 and CAl 4 2. Finally, for the Al 5 N and Al 5 N speies, two nearly degenerate strutures ompeting for the ground state, a planar one and a 3D one. The planar struture an be viewed as an Al oordinated to the planar Al 4 N : Al 4 N Al + or Al 4 N Al, whereas the 3D struture an be viewed as an AlN unit interating with a tetrahedral Al 4 motif: Al 4 2+ NAl 2 or Al 4 + NAl 2. ACKNOWLEDGMENTS The theoretial work done at Utah State University was supported by The Petroleum Researh Fund ACS-PRF No. 430-AC6, administered by the Amerian Chemial Soiety and by the National Siene Foundation CHE Computer time from the Center for High Performane Computing at Utah State University is gratefully aknowledged. The omputational resoure, the Uinta luster superomputer, was provided through the National Siene Foundation under Grant No. CTS with mathing funds provided by Utah State University. The experimental work done at Washington State University was supported by the National Siene Foundation DMR and was performed at the W. R. Wiley Environmental Moleular Sienes Laboratory, a national sientifi user faility sponsored by DOE s Offie of Biologial and Environmental Researh and loated at Paifi Northwest National Laboratory, whih is operated for DOE by Battelle. P. v. R. Shleyer and A. I. Boldyrev, J. Chem. So., Chem. Commun. 99, V. G. Zakrzewski, W. v. Niessen, A. I. Boldyrev, and P. v. R. Shleyer, Chem. Phys. 74, S. K. Nayak, S. N. Khana, and P. Jena, Phys. Rev. B 57, S. K. Nayak, B. K. Rao, P. Jena, X. Li, and L. S. Wang, Chem. Lett. 30, B. H. Boo and Z. Liu, J. Phys. Chem. A 03, L. Andrews, M. Zhou, G. V. Chertihin, W. D. Bare, and Y. Hannahi, J. Phys. A 04, B. R. Leskiw, A. W. Castleman, Jr., C. Ashman, and S. N. Khanna, J. Chem. Phys. 4, L. Ling, B. Song, and P.-L. Cao, J. Mol. Strut.: THEOCHEM 728, L. Gou and H.-S. Wu, Int. J. Quantum Chem. 06, X. Li and L. S. Wang, Eur. Phys. J. D 34, G. Meloni, S. M. Sheehan, B. F. Parsons, and D. M. Neumark, J. Phys. Chem. A 0, L. S. Wang, H. S. Cheng, and J. Fan, J. Chem. Phys. 02, A. N. Alexandrova, A. I. Boldyrev, Y.-J. Fu, X.-B. Wang, and L. S. Wang, J. Chem. Phys. 2, A. N. Alexandrova and A. I. Boldyrev, J. Chem. Theory Comput., A. D. Beke, J. Chem. Phys. 98, S. H. Vosko, L. Wilk, and M. Nusair, Can. J. Phys. 58, C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B 37, J. S. Binkley, J. A. Pople, and W. J. Hehre, J. Am. Chem. So. 02, ; M. S. Gordon, J. S. Binkley, J. A. Pople, W. J. Pietro, and W. J. Hehre, ibid. 04, ; W. J. Pietro, M. M. Franl, W. J. Hehre, D. J. Defrees, J. A. Pople, and J. S. Binkley, ibid. 04, M. Head-Gordon, J. A. Pople, and M. J. Frish, Chem. Phys. Lett. 53, J. Cizek, Adv. Chem. Phys. 4, P. J. Knowles, C. Hampel, and H.-J. Werner, J. Chem. Phys. 99, K. Raghavahari, G. W. Truks, J. A. Pople, and M. Head-Gordon, Chem. Phys. Lett. 57, A. D. MLean and G. S. Chandler, J. Chem. Phys. 72, T. Clark, J. Chandrasekhar, G. W. Spitznagel, and P. v. R. Shleyer, J. Comput. Chem. 4, M. J. Frish, J. A. Pople, and J. S. Binkley, J. Chem. Phys. 80, F. Bernardi, A. Bottini, J. J. W. MDougall, M. A. Robb, and H. B. Shlegel, Faraday Symp. Chem. So. 9, M. J. Frish, I. N. Ragazos, M. A. Robb, and H. B. Shlegel, Chem. Phys. Lett. 89, L. Cederbaum, J. Phys. B 8, W. von Niessen, J. Shirmer, and L. S. Cederbaum, Comput. Phys. Rep., V. G. Zakrzewski and W. von Niessen, J. Comput. Chem. 4, J. V. Ortiz, Int. J. Quantum Chem., Quantum Chem. Symp. 23, ; J. S. Lin and J. V. Ortiz, Chem. Phys. Lett. 7, V. G. Zakrzewski, J. V. Ortiz, J. A. Nihols, D. Heryadi, D. L. Yeager, and J. T. Golab, Int. J. Quantum Chem. 60, R. Bauernshmitt and R. Alrihs, Chem. Phys. Lett. 256, M. E. Casida, C. Jamorski, K. C. Casida, and D. R. Salahub, J. Chem. Phys. 08, M. J. Frish, G. M. Truks, H. B. Shlegel et al., GAUSSIAN 98, Revision A.7, Gaussian, In., Pittsburgh, PA, M. J. Frish, G. M. Truks, H. B. Shlegel et al., GAUSSIAN 03, Revision A., Gaussian, In., Pittsburgh, PA, G. Shaftenaar, MOLDEN3.4, CAOS/CAMM Center, The Netherlands, A. I. Boldyrev and P. v. R. Shleyer, J. Am. Chem. So. 3, J. Sun, W. C. Lu, H. Wang, Z.-S. Li, and C.-C. Sun, J. Phys. Chem. A 0, A. I. Boldyrev, X. Li, and L. S. Wang, J. Phys. Chem. 04, X. Li, L. W. Chen, L. S. Wang, A. I. Boldyrev, and J. Simons, J. Am. Chem. So. 2, A. I. Boldyrev and J. Simons, J. Am. Chem. So. 20, L. S. Wang, A. I. Boldyrev, X. Li, and J. Simons, J. Am. Chem. So. 22, D. Yu. Zubarev and A. I. Boldyrev, J. Chem. Phys. 22, X. Li, H.-F. Zhang, L. S. Wang, G. D. Geske, and A. I. Boldyrev, Angew. Chem., Int. Ed. 39, A. N. Alexandrova, A. I. Boldyrev, H. J. Zhai, and L. S. Wang, J. Phys. Chem. A 08, H. J. Zhai, B. Kiran, J. Li, and L. S. Wang, Nat. Mater. 2, H. W. Sarkas, S. T. Arnold, J. G. Eaton, G. H. Lee, and K. H. Bowen, J. Chem. Phys. 6, Downloaded 20 Mar 2007 to Redistribution subjet to AIP liense or opyright, see